Running the examples. Start the source-workspace Python session described in Chapter 1, then run this chapter's Python blocks in reading order. Java blocks form a separate sequence using the same NeqSim build; carry forward objects from preceding Java blocks. The release execution records are in verification/; a successful run establishes API compatibility, while physical validation also requires the checks discussed in the text.

Current NeqSim workflow and reproducibility

This revision uses NeqSim source commit 6cc8026202a5d3f9383c9abd1d97d448993813f9 (12 September 2026). A release date alone cannot identify a simulation: preserve the equation of state, mixing rule, composition, units, component database and calculation settings with the source revision. The source-workspace bootstrap below deliberately loads target/classes; using a previously installed Python package can otherwise hide new Java functionality.[1]

The newer workflow connects three kinds of evidence. Thermodynamic experiments qualify the fluid; hydraulic and equipment models define feasible operation; optimization searches that feasible region. PVTRegression combines laboratory experiment objectives with bounded parameters and separate hold-out validation. ProcessAutomation exposes discoverable input/output addresses and units. Typed energy ports, EnergyBus and EnergyNetworkSolver allocate available power and report unmet demand. These capabilities support traceable decisions only when the model's scope and numerical checks accompany each result.

When reading a worked example, distinguish an input assumption, a calculated result, a regression check and an independent measurement. A generated compressor curve is an assumed teaching map; an automatically sized separator is a preliminary design; a converged optimizer is a numerical result. None of those descriptions by itself establishes vendor qualification or operability outside the conditions actually examined.

Learning Objectives

After reading this chapter, the reader will be able to:

1. Define production optimization formally and state the general constrained optimization problem that underlies all production decisions
2. Distinguish between short-term (daily) and long-term (life-of-field) optimization, and explain why both are needed
3. Describe the major components of a production system — reservoir, wells, subsea, topside, compression, power, export — and explain how they interact as an integrated chain
4. Explain why local equipment optima do not guarantee a global system optimum, using back-pressure coupling as the primary example
5. Articulate the role of process simulation in optimization, including the distinction between steady-state and dynamic simulation
6. Install NeqSim, create a fluid, run a flash calculation, build a process model, and extract results using Python
7. Describe the NeqSim software architecture — ProcessSystem, ProcessModel, and the Automation API — at a conceptual level
8. Outline the three-step optimization workflow (model → calibrate → optimize) and relate it to the structure of this book

---

1.1 What Is Production Optimization?

Production optimization is the systematic process of maximizing the economic value extracted from a hydrocarbon reservoir while respecting safety, environmental, regulatory, and equipment constraints. It encompasses every decision that affects the rate, efficiency, and quality of production — from reservoir management and well operations to subsea transport, topside process control, and export specification compliance.

In its simplest form, production optimization answers a daily question: given the current state of the reservoir, wells, and facilities, what operating set points maximize today's value? In its most comprehensive form, it is a life-of-field discipline that integrates reservoir simulation, well modeling, process simulation, and economic analysis to guide investment decisions and operating strategies over decades.

1.1.1 A Formal Statement

At its mathematical core, production optimization is a constrained optimization problem. Let $u$ denote the vector of decision variables (choke openings, separator pressures, compressor speeds, gas-lift rates, etc.), and let $J(u)$ be an objective function representing economic value — typically net present value (NPV) or daily revenue. The optimization problem is:

where $g_i(u) \leq 0$ are inequality constraints (equipment capacities, safety limits, environmental limits) and $h_j(u) = 0$ are equality constraints (mass balances, energy balances, thermodynamic equilibrium). The decision variables $u$ live in a feasible set $\mathcal{U}$ defined by the physical system.

For example, a daily optimization problem for a platform producing oil and gas might be stated as:

subject to:

where $q_k$ is the total well production rate for well $k$, $p_{\text{oil}}$ and $p_{\text{gas}}$ are commodity prices, $Q_{\text{gas,max}}$, $Q_{\text{water,max}}$, and $Q_{\text{oil,max}}$ are facility capacities, $T_{\text{arrival},k}$ is the arrival temperature at the topside, and $T_{\text{hydrate},k}$ is the hydrate formation temperature for the fluid from well $k$.

Even this simplified formulation reveals the essential character of production optimization: it is a system-level problem. Changing one well's rate affects the pressure in shared manifolds and flowlines, which in turn changes the rates achievable by other wells. Increasing total gas production demands more compression power, which increases fuel gas consumption, which reduces the gas available for export. The coupling is pervasive and often nonlinear.

1.1.2 Short-Term vs. Long-Term Optimization

Production optimization operates on two distinct time scales:

Short-term optimization (hourly to weekly) focuses on maximizing value from the current state of the production system. Decision variables include well choke openings, gas-lift rates, separator pressures, compressor set points, and routing decisions. The reservoir state is treated as given — pressures and compositions change too slowly to be influenced within the optimization horizon. Short-term optimization is often called production allocation or rate optimization.

Long-term optimization (months to decades) focuses on maximizing the total value recovered over the field life. Decision variables include well drilling and completion schedules, water and gas injection strategies, facility capacity investments (debottlenecking, new compression, additional separation stages), and field abandonment timing. Here the reservoir state is a dynamic variable that evolves in response to the production strategy.

The two time scales are linked through the concept of reservoir voidage replacement. A short-term strategy that maximizes today's production by drawing down reservoir pressure faster than it can be maintained through injection may reduce total recovery and long-term value. Conversely, an overly conservative long-term strategy may leave significant value unrealized during periods of high commodity prices.

Throughout this book, we address both time scales. Chapters on well performance, separation, and compression focus primarily on short-term optimization — finding the best operating point for the current conditions. Chapters on field development, multi-scenario optimization, and reservoir coupling address the long-term perspective.

1.1.3 The Production System as an Integrated Chain

A production system is not a collection of independent equipment items — it is an integrated chain where every element constrains and is constrained by the others. Figure 1.1 illustrates this chain schematically.

The reservoir delivers fluid at a rate that depends on the difference between the average reservoir pressure $\bar{p}_R$ and the bottomhole flowing pressure $p_{wf}$. For a simple Productivity Index model:

The bottomhole pressure $p_{wf}$ depends on the wellhead pressure $p_{wh}$ plus the hydrostatic head and friction losses in the tubing:

The wellhead pressure depends on the downstream system: flowline pressure drop, riser pressure drop, manifold pressure, and ultimately the first-stage separator pressure $p_{\text{sep}}$. The separator pressure follows its gas inventory and gas-outlet/compressor controls; the liquid-level controller primarily manipulates liquid withdrawal. These loops interact, but liquid level is not the pressure setpoint. The compressor discharge pressure must overcome the export pipeline back-pressure.

The gain is governed by the coupled pressure response. In the limiting linear-PI example with unchanged tubing losses and an initial drawdown of 50 bar, a 5-bar reduction in bottomhole pressure raises rate by $5/50=10\%$. A separator-pressure change need not propagate one-for-one to bottomhole pressure: friction, multiphase holdup, choke criticality and facility constraints determine the actual response. The network examples therefore solve and verify that response explicitly.

1.1.4 Why Local Optima Are Not Global Optima

The integrated nature of the production system means that optimizing a single piece of equipment in isolation rarely finds the system optimum. Consider a simple example:

Consider a hypothetical operating change from 30 to 50 bara separator pressure. Assume a matched calculation predicts 15% less compressor power and 8% less saleable production. These percentages are assumptions for illustrating the decision, not measured improvements or a prediction for the book's fluid. Accept the change economically only if the value of power/fuel saved exceeds lost product revenue and any other incremental costs. Pressure ratio alone does not establish changes in wear or a field-wide optimum.

This example illustrates a fundamental principle: production optimization must be performed at the system level, not the equipment level. Process simulation — the subject of this book — provides the tool for system-level analysis by modeling the entire production chain in a single integrated model.

1.1.5 The Value of Optimization

The value of optimization must be demonstrated against a specified baseline. Codas et al.'s Urucu study connects calibrated well/network models, pressure constraints and routing decisions; it supports the integrated workflow rather than a universal percentage gain.[2] Table 1.1 identifies quantities that an optimization study should measure.

For an illustrative oil-only calculation, 100,000 bbl/d, a 2% increase, 365 operating days/year and a constant price of US\$60/bbl give US\$43.8 million/year of additional gross revenue. Taxes, royalties, incremental operating cost, outages and gas/oil-equivalent conversion are excluded. A separate US$500-million investment deferred two years at a 10% annual discount rate has a time-zero cost reduction of $500[1-(1.10)^{-2}]=86.8$ million dollars if all other cash flows remain unchanged. This is a transparent timing calculation, not evidence that an actual deferral is technically feasible.

1.1.6 Historical Perspective

The practice of production optimization has evolved significantly over the past fifty years:

1970s–1980s: Trial and error. Operators adjusted well chokes and separator pressures based on experience and intuition. Optimization was performed field-by-field with limited instrumentation and no computer models. Production engineers relied on rules of thumb developed over decades of operational experience.

1990s: Nodal analysis and well models. The widespread adoption of nodal analysis software (PROSPER, PIPESIM) enabled systematic well optimization. Well deliverability could be predicted as a function of operating conditions, but topside processing was still modeled separately. The disconnect between well and facility models meant that system-level optimization remained difficult.

2000s: Integrated production modeling. The concept of integrated asset modeling (IAM) — coupling reservoir, well, network, and process models in a single workflow — gained traction. Commercial tools such as Petex GAP, SPT Group (now Schlumberger), and Petroleum Experts' integrated modeling suite emerged. For the first time, it became practical to optimize the entire production system simultaneously, albeit with simplified thermodynamic models.

2010s: Real-time optimization and digital twins. Increased instrumentation (subsea multiphase meters, topside analyzers, wireless sensors) and compute power enabled real-time optimization and "digital twin" concepts. Cloud computing and data analytics became integral to optimization workflows. Machine learning supplemented physics-based models for rapid screening and anomaly detection.

2020s and beyond: AI-assisted and model-based optimization. The convergence of rigorous process simulation, machine learning, and automation APIs enables a new generation of optimization tools. Open-source simulation engines, accessible via Python and web APIs, lower the barrier to entry and enable rapid prototyping. NeqSim, with its open-source design, Python interface, and automation API, represents this generation — providing rigorous engineering models that can be embedded in automated decision-support workflows.

---

1.2 The Production Value Chain

This section describes each major element of the production chain in detail, establishing the physical and engineering context that subsequent chapters will model in NeqSim.

1.2.1 Reservoir

The reservoir is the source of all production and imposes the ultimate constraint: once the pressure is depleted and the mobile hydrocarbons are swept, production ends regardless of the facility capacity.

Reservoir pressure supplies the pressure potential for production. A normal water-column gradient is approximately 0.433 psi/ft for fresh water and 0.465 psi/ft for a representative saline water, equivalent to 9.8–10.5 kPa/m. The constant-density estimate $p=p_{\mathrm{surface}}+\rho g z$ gives about 295 bara at 3,000 m below a 1.01325-bara datum for $\rho=1000$ kg/m³ and $g=9.80665$ m/s². A 1 psi/ft gradient is not normal hydrostatic pressure. Actual formation pressure also reflects salinity, fluid contacts, overpressure, depletion and the pressure datum; burial depth alone is insufficient.[3] As production proceeds without pressure support, reservoir pressure and deliverability generally decline.

Reservoir temperature increases with depth at a geothermal gradient of approximately 25–35°C per kilometer. For the same 3,000 m reservoir, temperatures of 100–130°C are common. The temperature, together with the pressure and composition, determines the phase behavior of the reservoir fluid — whether it exists as an undersaturated oil, a saturated oil, a gas condensate, a wet gas, or a dry gas.

Fluid composition is the fundamental input to all thermodynamic calculations. Reservoir fluids are mixtures of hundreds of hydrocarbon species plus non-hydrocarbons (nitrogen, CO$_2$, H$_2$S, water). For engineering purposes, the composition is typically lumped into defined components (methane, ethane, propane, butanes, pentanes) and pseudo-components representing the heavier fractions (C$_7$+, C$_{10}$+, C$_{20}$+). The characterization of these heavy fractions — their molecular weight, density, and critical properties — is the subject of Chapter 3.

Drive mechanisms determine how pressure is maintained (or not) as fluids are withdrawn. The primary mechanisms are:

• Depletion drive — pressure declines as fluids expand; recovery factors are typically 10–30% for oil and 60–80% for gas
• Water drive — an active aquifer replaces the withdrawn volume with water influx; recovery factors of 30–60% for oil
• Gas cap drive — an overlying gas cap expands as pressure drops
• Compaction drive — rock compaction provides energy in unconsolidated formations
• Combination drive — most reservoirs exhibit a combination of mechanisms

Decline curves are empirical models that describe how production rate changes over time. The Arps decline model is the most widely used:

where $q_i$ is initial rate, $D_i$ is nominal initial decline with units inverse time, and $b$ is dimensionless. The displayed expression applies for $b>0$; its limit at $b\to0$ is $q(t)=q_i\exp(-D_i t)$, not direct substitution of zero into $1/b$. Classical hyperbolic decline uses $0<b<1$ and harmonic decline uses $b=1$. Forecasting also requires a stated terminal decline or economic limit; the curve is empirical and does not enforce reservoir material balance.

Inflow Performance Relationship (IPR) describes the relationship between bottomhole flowing pressure $p_{wf}$ and flow rate $q$. For single-phase liquid flow above the bubble point, the IPR is linear (the Productivity Index model). Below the bubble point, where two-phase flow develops near the wellbore, the Vogel correlation is commonly used:

where $q_{\max}$ is the absolute open flow (AOF) potential. Chapter 4 develops reservoir engineering and IPR modeling in detail.

1.2.2 Wells

Wells connect the reservoir to the surface facilities. Their design and performance directly affect both the achievable production rate and the ultimate recovery from the field.

Inflow Performance Relationship (IPR) — as introduced above — characterizes the ability of the reservoir to deliver fluid into the wellbore. It depends on reservoir properties (permeability, thickness, skin factor), fluid properties (viscosity, formation volume factor), and the pressure drawdown.

Vertical Flow Performance (VFP) — also called the tubing performance curve — describes the pressure loss in the tubing as a function of flow rate, for given tubing geometry, fluid composition, and wellhead pressure. The VFP accounts for:

• Hydrostatic pressure difference due to the column of fluid (dominant in most wells)
• Frictional pressure loss (significant at high rates in small-diameter tubing)
• Acceleration effects (usually negligible except near critical flow)

The intersection of the IPR and VFP curves on a pressure-rate plot defines the natural operating point of the well. This graphical construction, known as nodal analysis, is the foundation of well performance engineering and is developed in Chapter 5.

Artificial lift methods augment natural flow when the reservoir pressure is insufficient to lift fluids to the surface at economic rates. The principal methods are:

• Gas lift — injecting high-pressure gas into the tubing annulus to reduce the hydrostatic gradient. Gas lift is the most common artificial lift method in offshore production. The optimal gas-lift rate balances the benefit of reduced hydrostatic head against the cost of compressing and injecting the lift gas. Gas lift performance is characterized by the gas-lift performance curve (GLPC), which shows the incremental oil production per unit of injected gas.
• Electric Submersible Pump (ESP) — a downhole centrifugal pump driven by an electric motor, capable of high liquid rates (up to 30,000 bbl/d per pump). ESPs are widely used in high-watercut wells and in onshore fields. Their selection and sizing depend on the required head, flow rate, fluid properties, and power supply.
• Rod pump (sucker rod) — a reciprocating pump driven by a surface beam unit, the most common lift method in onshore U.S. wells. Economical for low-rate wells (under 500 bbl/d) with moderate depths.
• Jet pump, progressive cavity pump (PCP) — niche applications for specific fluid types and rate ranges. PCPs are well-suited for viscous crude oil production.

Well completions — the design of the production zone — critically affect well productivity. Completion types include open-hole, cased and perforated, slotted liners, and gravel packs. The completion efficiency is expressed as the skin factor $S$, where $S = 0$ is an undamaged well, $S > 0$ indicates damage, and $S < 0$ represents a stimulated well (e.g., after hydraulic fracturing).

Sand management — in poorly consolidated formations, producing at high drawdown can mobilize formation sand, leading to erosion, equipment damage, and well failure. Sand screens, gravel packs, and chemical consolidation are used to manage sand production. The allowable drawdown becomes an optimization constraint in Chapter 5.

1.2.3 Subsea Production Systems

In offshore developments, the subsea production system connects the wellheads on the seabed to the surface processing facility. The complexity and cost of subsea systems make them a critical area for optimization.

Subsea trees (Christmas trees) are valve assemblies mounted on the wellhead at the seabed. They control well flow, provide barriers for well intervention, and house production and annulus sensors. Modern subsea trees include pressure and temperature transmitters, multiphase flow meters, and sand detection probes. Subsea trees are rated for water depths up to 3,000 m and pressures up to 15,000 psi (1,035 bara) for high-pressure/high-temperature (HPHT) applications.

Manifolds collect production from multiple wells into a single flowline. A typical subsea manifold serves 4–8 wells and includes valving for individual well isolation. The manifold pressure is a key optimization variable: it is the downstream boundary condition for all connected wells and the upstream boundary for the flowline.

Flowlines transport the multiphase production (oil, gas, water) from the manifold to the riser base. Flowline design must account for pressure drop (minimized by large diameter), heat loss (managed by insulation or pipe-in-pipe systems), and flow assurance threats (hydrates, wax, slugging). Typical flowline lengths range from 5 km for near-platform tiebacks to 150 km or more for long-distance tiebacks.

Risers carry the production from the seabed to the surface facility. Riser systems include steel catenary risers (SCRs), flexible risers, and hybrid riser towers. The riser geometry (vertical height, catenary shape) creates significant hydrostatic pressure loss, particularly for deep-water developments where water depths exceed 1,000 m.

Subsea boosting and processing — for long-distance tiebacks or low-pressure fields, subsea multiphase pumps can boost the wellstream pressure, overcoming flowline losses that would otherwise choke production. Subsea separation and water re-injection allow water to be removed before transport, reducing flowline size and hydrate risk. These systems are modeled in NeqSim as pump and separator equipment within a subsea ProcessSystem (Chapter 7).

Tieback distance has a profound effect on production performance. As the distance between wells and the host facility increases, the pressure lost in the flowline and riser increases, reducing the pressure available at the wellhead and hence the achievable production rate. For a gas-condensate field, a 50 km tieback might require 50–80 bar of flowing pressure at the manifold, compared to 20–30 bar for a 10 km tieback. This difference directly impacts well deliverability and field economics. The pressure budget for a subsea tieback can be expressed as:

Here $\Delta p_{\mathrm{reservoir}}=p_{\mathrm{res}}-p_{wf}$ is reservoir drawdown; each remaining term is a pressure difference along the same flow path and datum. For upward production the hydrostatic contribution is positive. Acceleration and pressure gains from pumps must be added with their appropriate signs when present. Reducing downstream losses can increase available drawdown, subject to the coupled well and network solution.

1.2.4 Topside Processing Facilities

The topside facility — whether a fixed platform, a floating production storage and offloading (FPSO) vessel, or an onshore plant — separates, treats, and conditions the produced fluids for export. Understanding the topside process is essential for production optimization because the facility capacities define the constraints on total field production.

Separation is the first and most important processing step. Production from the wells enters the first-stage (HP) separator, where the three phases — gas, oil, and water — are separated by gravity. Typical HP separator pressures range from 40 to 100 bara, depending on the field pressure and export requirements.

Gas from the HP separator flows to gas processing (dehydration, dew point control) and compression. Liquid from the HP separator flows to the second-stage (MP) separator at a lower pressure (typically 10–30 bara), where additional gas is liberated. A third-stage (LP) separator at 2–5 bara recovers the final flash gas. This multi-stage pressure let-down is the cornerstone of topside process design, and the optimal separator pressures are a key optimization target (Chapter 10).

For $n$ separator pressure levels, a geometric sequence is one possible initial guess, with $n-1$ pressure intervals:

This identity is not a separation optimum. Separator pressures must be optimized against recovered product value, quality and recompression constraints using flashes at each stage. Equal pressure ratios minimize ideal, perfectly intercooled compressor work only under additional assumptions: identical stage inlet temperatures and efficiencies, unchanged gas composition and flow, and negligible interstage pressure losses. Those compressor assumptions usually fail across a separation train, where each stage releases a different amount and composition of gas.

Oil processing downstream of the LP separator includes electrostatic coalescers for water removal, heat exchangers for oil heating (to reduce viscosity and improve separation), and stabilization columns for light-end removal. The oil must meet export specifications — typically less than 0.5% basic sediment and water (BS&W) and a Reid vapor pressure (RVP) below 12 psia for safe tanker transport.

Gas processing treats the separated gas to meet pipeline or LNG specifications. Key processes include:

• Dehydration — removal of water vapor, typically using triethylene glycol (TEG) contactors, to prevent hydrate formation and corrosion in the export pipeline. The water dew point specification is typically $-18°\text{C}$ or lower at the delivery pressure.
• Dew point control — removal of heavy hydrocarbons using Joule-Thomson (JT) valves, turbo-expanders, or mechanical refrigeration to prevent liquid condensation in the export pipeline. The hydrocarbon dew point must meet the contract limit at the specified pressure or across its pressure range. The cricondentherm is the maximum temperature of the hydrocarbon two-phase envelope; it is not the dew point at an arbitrary delivery pressure.
• Acid gas removal — removal of CO$_2$ and H$_2$S using amine absorption (MDEA, DEA), membrane separation, or molecular sieves when concentrations exceed pipeline specifications
• NGL recovery — fractionation of the gas into methane (sales gas), ethane, propane, butanes (LPG), and natural gasoline (C$_5$+) in onshore processing facilities

Water treatment processes produced water to meet discharge or reinjection specifications. OSPAR Recommendation 2001/1, as amended, uses 30 mg/L dispersed oil as a flow-weighted monthly average for produced/displacement-water discharges; it is not a universal instantaneous outlet specification. The applicable permit can impose additional controls.[4] Reinjection quality must be selected from formation-compatibility and injectivity evidence, including oil/solids size distributions, filterability and chemistry; no universal 5 mg/L oil limit establishes protection against formation damage. Produced water volumes typically increase over the field life as the water-oil ratio rises, and water treatment capacity often becomes the binding constraint on total production in the later years of a field's life.

This overview groups equipment functions. A detailed staged separation model must account for every gas and liquid outlet and for pressure matching before mixing streams.

1.2.5 Gas Compression

Compression is typically the most capital-intensive and energy-intensive equipment on a production facility, and it is often the bottleneck that limits total field production. A thorough understanding of compression systems is essential for production optimization.

Recompression handles gas from the MP and LP separators, boosting it from low pressure (2–30 bara) to the HP system pressure (60–100 bara) for export or further processing. Recompression trains typically consist of 2–4 stages with intercoolers and scrubbers at each stage to remove condensed liquids.

Export compression boosts the HP gas to the export pipeline pressure (100–250 bara, depending on the pipeline length and delivery pressure). Export compressors are typically the largest and most expensive machines on the platform, driven by gas turbines of 15–40 MW.

Gas lift compression provides high-pressure gas (150–350 bara) for injection into the tubing annulus of gas-lifted wells. Gas lift compression is dedicated, because the lift gas pressure must exceed the casing head pressure at the wellhead.

Injection compression boosts gas to reservoir pressure (200–500 bara) for pressure maintenance or enhanced recovery by gas injection. Injection compressors operate at the highest pressures on the platform and may require multiple stages of compression.

For a calorically perfect ideal gas, $n_s$ compressor stages with equal pressure ratios, identical suction temperature $T_1$ after perfect intercooling, identical stage isentropic efficiency $\eta_s$ and negligible intercooler pressure loss give the shaft-power estimate:

Here $\gamma=c_p/c_v$ is constant, $R$ is the specific gas constant in J/(kg K), $T_1$ is in K and $\dot m$ is in kg/s, so $W$ is in W. All pressure ratios use absolute pressure. This expression uses isentropic efficiency; polytropic efficiency appears inside the temperature-ratio exponent for a differential-stage ideal-gas model and is not interchangeable with $\eta_s$. Real-gas enthalpy changes and actual intercooling are calculated explicitly in Chapters 14–15.[5]

Why compression is often the bottleneck: As reservoir pressure declines over the field life, the wellhead pressure drops, and more gas is released at lower pressures in the separation train. The recompression duty increases while the total gas volume also increases. Eventually, the installed compression capacity is fully utilized, and total production must be reduced — the field is "compression-constrained." The crossover from "well-constrained" to "facility-constrained" production is a critical transition in the life of every field. Identifying and relieving this bottleneck is one of the most valuable applications of production optimization (Chapter 21).

1.2.6 Power and Utilities

Platform operations require substantial power — typically 20–100 MW for a large offshore facility. The power demand is dominated by gas compression (often 60–80% of total demand), with additional loads from pumps, heat tracing, lighting, drilling, and accommodation.

Gas turbines are the primary power source on most offshore platforms. They burn produced gas (fuel gas) to generate electricity or drive compressors directly through a mechanical coupling. Fuel share depends on duty, efficiency, heating value and available gas rate. For example, 20 MW net output at 35% LHV efficiency with 50 MJ/kg fuel requires $20/(0.35\times50)=1.143$ kg/s of fuel, or 1.14% of a 100 kg/s gas feed. The same duty consumes 11.4% of a 10 kg/s feed. Fuel gas demand therefore competes directly with gas export revenue, and no fixed percentage is transferable between fields. Optimizing compression efficiency therefore has a double benefit: it reduces the power demand and increases the gas available for export.

Simple-cycle gas turbine efficiency is typically 30–38%, depending on the ambient temperature and the turbine model. The thermal efficiency $\eta_{\text{th}}$ is defined as:

where $W_{\text{net}}$ is the net shaft power, $\dot{m}_{\text{fuel}}$ is the fuel gas mass flow rate, and LHV is the lower heating value of the fuel gas.

Waste heat recovery from gas turbine exhaust (typically 450–550°C) can generate steam for heating, power generation (via a steam turbine in a combined cycle), or process use. Combined cycle efficiencies of 45–55% are achievable, compared to 30–38% for simple cycle gas turbines. The economics of waste heat recovery depend on the value of the recovered energy and the weight and space constraints of the facility.

Platform electrical system distributes power from the generators to all consumers. Modern platforms increasingly use variable-speed drives (VSDs) on compressors and pumps, which allow continuous adjustment of speed, improving part-load efficiency and providing an optimization degree of freedom. The interaction between power generation, compression, and production is a system-level optimization problem addressed in Chapter 18.

1.2.7 Export and Metering

The final stage of the production chain delivers products to the buyer at the custody transfer point.

Oil export is accomplished by pipeline (to an onshore terminal) or by shuttle tanker (for FPSOs and remote platforms). Pipeline export requires sufficient pressure to overcome friction and elevation changes. Tanker export requires oil storage on the FPSO (typically 1–2 million barrels capacity) and offloading via a swivel turret or bow loading system. The offloading rate and storage capacity can constrain production during periods of bad weather when tanker operations are suspended.

Gas export is almost exclusively by pipeline from offshore platforms. The export pipeline may be hundreds of kilometers long, and the required inlet pressure depends on the pipeline diameter, length, and delivery pressure. For very long pipelines, the required inlet pressure can be 200 bara or more, placing heavy demands on the export compression system. The pipeline capacity depends on the inlet pressure, the gas composition (density and viscosity), and the delivery pressure — all of which can be modeled in NeqSim.

Fiscal metering is the measurement of oil and gas quantities at the custody transfer point, forming the basis for revenue calculation, royalty payments, and tax assessment. The permissible uncertainty, reference conditions, sampling, and calculation methods depend on the applicable custody-transfer agreement and jurisdiction. ISO 5167, AGA Report No. 9, and NORSOK I-104 address different measurement technologies and system requirements; the applicable editions and project uncertainty budget must be identified before design.

Quality specifications are defined by the receiving network and sales agreement. The following values are illustrative teaching constraints, not universal export limits. The contract must also define pressure, standard volume reference conditions, averaging period, and measurement method.

These limits enter optimization as inequalities, with a suitable operating margin for uncertainty and disturbances. Some become active at the optimum; others remain slack. NeqSim can calculate composition-based heating value and Wobbe index, phase-equilibrium dew points, and tracked component concentrations. Agreement with a fiscal or quality measurement also depends on representative sampling, the EOS and reference basis, water and aerosol entrainment, and the validity of the process removal model. Thermodynamic composition alone does not establish total-sulfur measurement or separator carry-over performance. Chapter 19 develops export specification tracking.

---

1.3 The Role of Process Simulation

Process simulation is the computational backbone of production optimization. A calibrated process model predicts how the production system responds to changes in operating conditions — flow rates, pressures, temperatures, compositions — without the cost and risk of physical experimentation.

1.3.1 What a Process Simulator Does

A process simulator solves the coupled material balance, energy balance, and momentum balance equations for a defined flowsheet of interconnected equipment. At each piece of equipment, the simulator:

1. Receives inlet stream conditions — temperature, pressure, flow rate, composition
2. Applies the equipment model — e.g., adiabatic flash for a separator, polytropic compression for a compressor, pressure drop correlation for a pipe
3. Calculates outlet stream conditions — using thermodynamic equilibrium (flash calculations) and transport property models
4. Iterates if the flowsheet contains recycle streams or specifications that create feedback loops

The thermodynamic calculations are the foundation of everything else. They answer questions such as: given a fluid of known composition at temperature $T$ and pressure $P$, how many phases exist? What is the composition of each phase? What are the densities, viscosities, enthalpies, and heat capacities?

1.3.2 The Simulation Loop

The inner loop of a process simulator can be summarized as:

1. Equation of state (EOS) — a mathematical model relating pressure, volume, temperature, and composition. Common choices include the Soave-Redlich-Kwong (SRK) and Peng-Robinson (PR) cubic equations of state, and the CPA (Cubic Plus Association) model for polar systems. The general form of a cubic EOS is:

where $v$ is the molar volume, $a(T)$ is the attraction parameter (temperature-dependent), $b$ is the co-volume, and $\epsilon$ and $\sigma$ are EOS-specific constants ($\epsilon = 0, \sigma = 1$ for SRK; $\epsilon = 1 - \sqrt{2}, \sigma = 1 + \sqrt{2}$ for PR). The EOS provides fugacity coefficients $\hat{\phi}_i$ that drive phase equilibrium calculations.

1. Flash calculation — given the total composition $z_i$, temperature $T$, and pressure $P$, determine the number and amounts of phases (vapor fraction $\beta$, liquid fraction $1-\beta$) and the composition of each phase ($y_i$ for vapor, $x_i$ for liquid) by solving the Rachford-Rice equation:

where $K_i = y_i / x_i$ is the equilibrium ratio (K-value) for component $i$, determined from the fugacity coefficients: $K_i = \hat{\phi}_i^L / \hat{\phi}_i^V$.

1. Property calculation — once phase compositions and amounts are known, calculate physical properties: density from the EOS, viscosity from correlations such as Lohrenz-Bray-Clark (LBC), thermal conductivity, surface tension, and diffusion coefficients.

1. Equipment model — apply the specific equipment model using the calculated properties (e.g., compressor work from the enthalpy change across the compression, pipe pressure drop from the Beggs and Brill correlation, separator performance from the retention time and K-factor).

1. Iterate — repeat until all recycle streams converge and all specifications are met.

1.3.3 Steady-State vs. Dynamic Simulation

Steady-state simulation calculates a time-independent solution for specified inputs and boundary conditions. Material and energy accumulation vanish, but dynamic stability does not follow from solving those algebraic balances. A steady solution can be unstable; stability requires perturbation/dynamic analysis with the relevant equipment and controls. Steady-state simulation is used for:

• Design studies and equipment sizing
• Capacity checks and equipment rating
• "What-if" scenario analysis (what happens if we increase well rates by 10%?)
• Optimization (find the best operating point among all feasible points)
• Production forecasting (combined with a reservoir decline model)

Dynamic simulation tracks how the process evolves over time, solving the ordinary and partial differential equations that describe accumulation of mass, energy, and momentum in vessels, pipes, and control loops. The governing equation for mass accumulation in a vessel of volume $V$ is:

and for energy:

where $\rho$ is density, $U$ is internal energy, $h$ is specific enthalpy, $\dot{Q}$ is heat transfer, and $\dot{W}$ is work. Dynamic simulation is used for:

• Control system design and PID controller tuning
• Startup, shutdown, and load change procedures
• Slug flow impact assessment (liquid surges from terrain slugging in flowlines)
• Emergency depressurization and blowdown studies
• Compressor anti-surge system verification

NeqSim supports both modes: process.run() performs a steady-state calculation, while process.runTransient(dt) advances the simulation by a time step $dt$ seconds. Chapter 29 develops dynamic simulation in detail.

1.3.4 Calibrated Models vs. Design Models

A design model is built from equipment specifications, design data sheets, and vendor information before the facility is constructed. It predicts how the facility should perform at design conditions and is used for engineering design, procurement, and construction.

A calibrated model (also called an operations model) has been adjusted to match measured operating data — actual separator efficiencies, real compressor performance curves, measured pipeline pressure drops, and tuned IPR curves from well tests. A calibrated model predicts how the facility actually performs and is therefore far more useful for optimization.

The workflow for model calibration is:

1. Collect measured data — flow rates, pressures, temperatures, compositions from process sensors. Data quality is critical: sensor drift, measurement noise, and missing data must be addressed.
2. Compare model predictions to measurements — identify systematic discrepancies (bias) and random scatter.
3. Adjust model parameters — equipment efficiencies, heat transfer coefficients, friction factors, well productivity indices — to minimize the discrepancy, typically in a least-squares sense.
4. Validate the calibrated model against an independent data set (not used in the calibration) to ensure the model generalizes rather than overfitting.

Throughout this book, we emphasize the importance of model calibration and provide practical guidance for tuning NeqSim models to match plant data.

1.3.5 Model Fidelity vs. Computational Cost

There is an inherent trade-off between model fidelity and computational cost. A full-compositional process model with 30+ components and rigorous thermodynamics may take 5–30 seconds to evaluate a single operating point. An optimization algorithm that requires 10,000 function evaluations would then take 14–80 hours — impractical for daily operations.

Strategies for managing this trade-off include:

• Reduced compositions — lumping heavy components to reduce the number of species from 30+ to 8–12, with proportional speedup
• Surrogate models — training machine learning models (neural networks, Gaussian processes, polynomial response surfaces) on simulation outputs and using them as fast approximations in the optimization loop
• Decomposition — optimizing subsystems sequentially (wells first, then topside, then compression) rather than simultaneously
• Gradient-based methods — using derivative information (adjoint methods, finite differences) to reduce the number of function evaluations needed to find the optimum
• Parallel evaluation — running multiple simulation cases simultaneously on multi-core hardware

Measure warm-start and cold-start runtimes for the actual composition, phase region, recycle structure and hardware before setting an optimization budget. The execution records for this book retain elapsed time per example. Near-critical flashes, hydrate calculations and difficult columns can be much more expensive than a single-phase TP flash, so equipment count alone is not a runtime predictor.

---

1.4 Why NeqSim?

NeqSim (Non-Equilibrium Simulator) is an open-source Java library for thermodynamic and process simulation, developed since 2000 at NTNU (Norwegian University of Science and Technology) and Equinor. It is designed from the ground up for rigorous engineering calculations and has been used in production, research, and teaching across the oil and gas industry.

1.4.1 Open Source and Extensible

NeqSim is released under the Apache 2.0 license, making it freely available for commercial and academic use. The source code is publicly hosted on GitHub (github.com/equinor/neqsim), enabling:

• Full transparency — inspect and verify every calculation, from the EOS implementation to the flash algorithm to the property correlations. This is essential for engineering applications where results must be auditable.
• Extensibility — add new equations of state, equipment models, or optimization algorithms by extending the Java classes.
• Reproducibility — share complete simulation models (code, fluid definitions, process configurations) with colleagues and reviewers, ensuring identical results.
• No license cost — democratizes access to rigorous process simulation, particularly for universities, research institutions, and small companies.

1.4.2 Comparison with Commercial Tools

Software selection should follow the required calculation and validation evidence. Table 1.2 gives criteria used in this book rather than an unversioned ranking of simulator products. Commercial HYSYS/UniSim and ProMax workflows depend on purchased options and supported interfaces. Other open-source tools also support automation: DWSIM documents native dynamic simulation, Python/.NET automation and an MCP interface.[6]

NeqSim supplies source access, a Java/Python bridge and process automation used by the worked examples. Their scientific value comes from the verified model, not from the interface or the license alone.

1.4.3 Key Capabilities

The NeqSim capabilities most relevant to this book are:

• Equation of state library — SRK, PR (both with volume translation), CPA (for polar systems like water, methanol, MEG), Electrolyte CPA (for brine chemistry and scale prediction), GERG-2008 (custody transfer reference for natural gas), PC-SAFT, and UMR-PRU
• Flash calculations — TP flash, PH flash, PS flash, TV flash, bubble point and dew point calculations, hydrate equilibrium temperature and pressure
• Transport properties — viscosity (LBC, PFCT, corresponding states), thermal conductivity, surface tension, and diffusion coefficients for all phases
• Process equipment — streams, separators (2-phase, 3-phase), compressors (with performance curves), heat exchangers (shell-and-tube, plate), valves (Cv-based), pipes (Beggs and Brill), mixers, splitters, distillation columns, absorbers, reactors
• Dynamic simulation — transient equipment models with PID controllers, transmitters (PT, TT, LT, FT), and alarm logic
• Multiphase pipe flow — Beggs and Brill correlation with pipeline elevation profile and heat transfer to surroundings
• Reservoir models — simple reservoir with drive mechanism and decline, suitable for integrated production modeling

1.4.4 The NeqSim Ecosystem

NeqSim operates as an ecosystem of interconnected components:

• Java core (neqsim-*.jar) — the main library, containing all thermodynamic and process simulation classes. Can be used directly from Java applications or any JVM-compatible language.
• Python interface (pip install neqsim) — uses jpype to bridge Python and the Java core, enabling use from Jupyter notebooks, scripts, and automation systems. All Java classes are accessible from Python with automatic type conversion.
• Jupyter notebooks — the primary working environment for this book. Each chapter includes companion notebooks with runnable code examples that can be executed, modified, and extended.
• MCP server — a Model Context Protocol server that exposes NeqSim calculations as web services, enabling integration with AI assistants, web applications, and external optimization engines.
• Automation API — the ProcessAutomation class provides string-addressable access to all simulation variables, designed for programmatic access from optimization algorithms, digital twin systems, and real-time advisory tools.

---

1.5 Getting Started with NeqSim

This section walks through the essential first steps: installing NeqSim, creating a fluid, running a flash calculation, building a process, and accessing results programmatically.

1.5.1 Installation

Install the NeqSim Python package from PyPI:

pip install neqsim

The installation includes the Java core library and handles the JVM setup automatically. Verify the installation:

import os
import sys
from pathlib import Path

# Set NEQSIM_PROJECT_ROOT to the NeqSim source clone before starting Python.
PROJECT_ROOT = Path(os.environ["NEQSIM_PROJECT_ROOT"]).resolve()
if not (PROJECT_ROOT / "target" / "classes").is_dir():
    raise RuntimeError("Compile the selected NeqSim source checkout first.")
sys.path.insert(0, str(PROJECT_ROOT / "devtools"))
from neqsim_dev_setup import neqsim_init
neqsim_init(project_root=PROJECT_ROOT, recompile=False, verbose=False)
import jpype
jneqsim = jpype.JPackage("neqsim")
print("NeqSim loaded from", PROJECT_ROOT / "target" / "classes")

For development work or access to the latest features, the Java source can be built from the GitHub repository using Maven:

git clone https://github.com/equinor/neqsim.git
cd neqsim
./mvnw install

1.5.2 Creating a Fluid

A fluid in NeqSim is represented by a SystemInterface object — an instance of an equation-of-state system containing a defined set of components with their mole fractions. The most common choice for oil and gas work is the SRK (Soave-Redlich-Kwong) equation of state:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create an SRK EOS system at 80°C (353.15 K) and 150 bara
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 80.0, 150.0)

# Add components with mole fractions (will be normalized automatically)
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 2.0)
fluid.addComponent("methane", 70.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("n-butane", 3.0)
fluid.addComponent("n-pentane", 2.0)
fluid.addComponent("n-hexane", 1.5)
fluid.addComponent("n-heptane", 3.0)
fluid.addComponent("n-octane", 2.5)
fluid.addComponent("n-nonane", 1.5)
fluid.addComponent("water", 1.0)

# Set the mixing rule — this must ALWAYS be called
fluid.setMixingRule("classic")

# Enable multiphase check (important for three-phase systems)
fluid.setMultiPhaseCheck(True)

Key points to remember:

• The constructor takes temperature in Kelvin and pressure in bara
• Components are specified by name; NeqSim has a database of 100+ pure components with validated critical properties, acentric factors, and binary interaction parameters
• setMixingRule("classic") activates the standard van der Waals one-fluid mixing rule. Never skip this call — without it, the EOS cannot compute mixture fugacities.
• setMultiPhaseCheck(True) enables detection of aqueous, liquid hydrocarbon, and vapor phases simultaneously, which is essential for production system modeling where water is always present

The EOS selection matters for accuracy. Table 1.3 provides guidance:

Chapter 2 develops the thermodynamic foundations in depth.

1.5.3 Running a Flash Calculation

A flash calculation determines the phase equilibrium at given conditions. The most common type is the TP flash — given temperature and pressure, calculate the number of phases, their amounts, and their compositions:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create a gas-condensate fluid at HP separator conditions
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 80.0, 60.0)
fluid.addComponent("methane", 0.80)
fluid.addComponent("ethane", 0.06)
fluid.addComponent("propane", 0.04)
fluid.addComponent("n-butane", 0.03)
fluid.addComponent("n-pentane", 0.02)
fluid.addComponent("n-hexane", 0.01)
fluid.addComponent("n-heptane", 0.02)
fluid.addComponent("n-octane", 0.015)
fluid.addComponent("water", 0.005)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

# Run TP flash
ops = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid)
ops.TPflash()

# CRITICAL: Initialize physical properties AFTER the flash
fluid.initProperties()

# Read results
num_phases = fluid.getNumberOfPhases()
print(f"Number of phases: {num_phases}")

if fluid.hasPhaseType("gas"):
    gas = fluid.getPhase("gas")
    print(f"Gas density: {gas.getDensity('kg/m3'):.2f} kg/m3")
    print(f"Gas viscosity: {gas.getViscosity('kg/msec'):.6f} kg/(m·s)")
    print(f"Gas Cp: {gas.getCp('J/molK'):.2f} J/(mol·K)")
    print(f"Gas Z-factor: {gas.getZ():.4f}")

if fluid.hasPhaseType("oil"):
    oil = fluid.getPhase("oil")
    print(f"Oil density: {oil.getDensity('kg/m3'):.2f} kg/m3")
    print(f"Oil viscosity: {oil.getViscosity('kg/msec'):.6f} kg/(m·s)")

Important: After any flash calculation, you must call fluid.initProperties() before reading transport properties (viscosity, thermal conductivity). The flash alone initializes thermodynamic properties (fugacities, enthalpies), but transport property models require an additional initialization step. Without this call, getViscosity() and getThermalConductivity() may return zero.

1.5.4 Building a Process

NeqSim's process simulation framework models interconnected equipment — streams, separators, compressors, heat exchangers, valves, and pipes — as a sequential flowsheet:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 80.0, 150.0)
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 2.0)
fluid.addComponent("methane", 70.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("n-butane", 3.0)
fluid.addComponent("n-pentane", 2.0)
fluid.addComponent("n-hexane", 1.5)
fluid.addComponent("n-heptane", 3.0)
fluid.addComponent("n-octane", 2.5)
fluid.addComponent("n-nonane", 1.5)
fluid.addComponent("water", 1.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

# Create equipment
Stream = jneqsim.process.equipment.stream.Stream
Separator = jneqsim.process.equipment.separator.Separator
ThrottlingValve = jneqsim.process.equipment.valve.ThrottlingValve
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

# Well stream
feed = Stream("Well Stream", fluid)
feed.setFlowRate(50000.0, "kg/hr")
feed.setTemperature(80.0, "C")
feed.setPressure(80.0, "bara")

# HP Separator
hp_sep = Separator("HP Separator", feed)

# Valve to reduce liquid pressure before MP separator
valve = ThrottlingValve("HP-MP Valve", hp_sep.getLiquidOutStream())
valve.setOutletPressure(25.0, "bara")

# MP Separator
mp_sep = Separator("MP Separator", valve.getOutletStream())

# Build and run
process = ProcessSystem()
process.add(feed)
process.add(hp_sep)
process.add(valve)
process.add(mp_sep)
process.run()

# Read results
hp_gas = hp_sep.getGasOutStream().getFlowRate("MSm3/day")
mp_gas = mp_sep.getGasOutStream().getFlowRate("MSm3/day")
oil_out = mp_sep.getLiquidOutStream().getFlowRate("m3/hr")

print(f"HP separator gas: {hp_gas:.3f} MSm3/day")
print(f"MP separator gas: {mp_gas:.3f} MSm3/day")
print(f"Total gas: {hp_gas + mp_gas:.3f} MSm3/day")
print(f"Oil export: {oil_out:.2f} m3/hr")

This example illustrates the core NeqSim workflow that will be used throughout this book:

1. Define the fluid using an appropriate equation of state and component list
2. Create equipment — each equipment item takes its inlet stream as a constructor argument
3. Connect equipment by using outlet streams from upstream equipment as inlet streams for downstream equipment
4. Build a ProcessSystem by adding equipment in sequence
5. Run the simulation with process.run()
6. Extract results from equipment and stream accessor methods

1.5.5 The ProcessSystem Architecture

NeqSim organizes process simulations around two key classes:

ProcessSystem represents a single process area — a connected set of equipment with a defined execution order. Equipment is added using process.add(equipment), and process.run() evaluates all equipment in sequence. If the flowsheet contains recycles, ProcessSystem handles the iterative convergence automatically.

Key features of ProcessSystem:

• Enforces unique equipment names (duplicate names cause an error)
• Provides getUnit("name") to retrieve equipment by name
• Supports copy() for duplicating equipment configurations
• Handles recycle loops and specification convergence

ProcessModel composes multiple ProcessSystem objects into a multi-area plant model. This is essential for large facilities where the separation train, compression system, gas processing, and water treatment are modeled as separate process areas that share streams at their boundaries:

ProcessModel = jneqsim.process.processmodel.ProcessModel

# Continue from the separator flowsheet created above.
separation_system = process
Compressor = jneqsim.process.equipment.compressor.Compressor
Cooler = jneqsim.process.equipment.heatexchanger.Cooler
compressor = Compressor("Export compressor", hp_sep.getGasOutStream())
compressor.setOutletPressure(120.0)
compressor.setIsentropicEfficiency(0.75)
compression_system = ProcessSystem()
compression_system.add(compressor)
cooler = Cooler("Export cooler", compressor.getOutletStream())
cooler.setOutTemperature(313.15)
gas_processing_system = ProcessSystem()
gas_processing_system.add(cooler)
plant = ProcessModel()
plant.add("Separation", separation_system)
plant.add("Compression", compression_system)
plant.add("Gas Processing", gas_processing_system)
plant.run()  # Iterates all areas until convergence

A ProcessModel executes configured process areas and allows downstream areas to consume updated streams. Feedback across areas requires correctly configured recycle/convergence handling and an explicit boundary-residual check; adding areas is not itself proof of a converged whole-plant solution. This architecture mirrors the physical structure of a production facility, where different process areas are designed and operated by different engineering disciplines but are coupled through shared streams.

1.5.6 The Automation API Preview

For optimization and digital twin applications, NeqSim provides the ProcessAutomation class — a string-addressable interface to all simulation variables. Instead of navigating Java object hierarchies, you can read and write variables by name:

# Get the automation facade
auto = process.getAutomation()

# Discover equipment
units = auto.getUnitList()    # ["Well Stream", "HP Separator", ...]

# List variables for a specific unit
variables = auto.getVariableList("HP Separator")
for v in variables:
    print(f"  {v.getAddress()} ({v.getType()}) [{v.getDefaultUnit()}]")

# Read a variable by address
temp = auto.getVariableValue("HP Separator.gasOutStream.temperature", "C")
pres = auto.getVariableValue("HP Separator.pressure", "bara")

# Write an input variable and re-run
auto.setVariableValue("HP-MP Valve.outletPressure", 20.0, "bara")
process.run()  # Propagate changes through the process

The Automation API includes self-healing diagnostics: if you misspell an equipment name, it uses fuzzy matching to suggest the correct name. This makes it robust for use in automated workflows where exact variable names may not be known in advance. We will use it extensively in the optimization chapters (Part VII).

---

1.6 Optimization Workflow Overview

Production optimization using process simulation follows a three-step workflow that is conceptually simple but practically demanding. This workflow forms the backbone of the applied chapters in this book.

Step 1: Model

Build a process model that represents the production system from reservoir to export. The model should include:

• Reservoir inflow (IPR curves or simple reservoir models with decline)
• Wells (VFP curves or multiphase pipe flow models)
• Subsea systems (flowlines with Beggs and Brill, risers, manifolds)
• Topside processing (separators, compressors, heat exchangers, valves)
• Export systems (pipelines, metering stations)

The model is constructed in NeqSim using the ProcessSystem and ProcessModel framework. Equipment is connected through streams and normally evaluated sequentially; specified recycle loops require iteration. Coupled hydraulic network equations may use a simultaneous solver. The chosen execution and convergence scheme is part of the model specification.

Step 2: Calibrate

Adjust the model parameters to match measured operating data. Key calibration targets include:

• Well productivity indices (matched to well test data)
• Separator efficiencies (matched to composition analyses of gas and liquid outlets)
• Compressor performance (matched to operating points on the manufacturer's performance map)
• Heat exchanger UA values (matched to measured duty and temperature approaches)
• Pipeline friction factors (matched to measured inlet-to-outlet pressure drops)

Calibration transforms a design model into an operations model that accurately reflects the current state of the facility. The quality of the optimization is directly limited by the quality of the calibration.

Step 3: Optimize

Use the calibrated model to find the operating conditions that maximize the objective function while respecting all constraints. The optimization problem is:

where the production rates $q_{\text{oil},k}$ and $q_{\text{gas},k}$ depend on the operating conditions $u$ through the process model, and $C_{\text{fuel}}$ and $C_{\text{chem}}$ are the costs of fuel gas and chemicals.

Decision variables $u$ typically include:

• Individual well choke openings or production rates
• Gas-lift injection rates per well
• Separator pressures (HP, MP, LP)
• Compressor speeds or set points
• Heat exchanger duties
• Routing decisions (which wells produce to which manifold/separator)

Constraints include:

• Equipment capacities (compressor power, separator liquid handling, water treatment)
• Export specifications (heating value, water dew point, H$_2$S content)
• Safety limits (maximum pressures, minimum temperatures for material integrity)
• Flow assurance limits (minimum arrival temperature above hydrate formation)
• Environmental limits (flaring, CO$_2$ emissions, produced water discharge)

The optimization can be performed using gradient-based methods (for smooth, well-behaved problems), evolutionary algorithms (for non-convex problems with discrete variables), or exhaustive search over a discretized decision space (practical when the number of decision variables is small). NeqSim provides the ProductionOptimizer and CapacityConstrainedEquipment classes to support these workflows, developed in Chapter 23.

Calibration data and independent validation evidence are separate inputs to the modeling work. A proposed operating point still requires the physical and constraint acceptance checks developed later in the book.

---

1.7 Book Organization and Road Map

This book is organized into nine parts containing 35 chapters that systematically develop every element of production optimization — from thermodynamic foundations through topside processing to advanced optimization methods and integrated case studies.

Part I: Foundations (Chapters 1–3)

Part I lays the groundwork for all subsequent chapters.

• Chapter 1 — Introduction to Production Optimization (this chapter): Defines production optimization, describes the production value chain, introduces NeqSim, and provides the first working examples.
• Chapter 2 — Thermodynamic Foundations for Process Simulation: Develops the thermodynamic framework underlying all process calculations — equations of state, mixing rules, fugacity, phase equilibrium, and property calculation methods. This chapter provides the theoretical basis for understanding why NeqSim produces the results it does.
• Chapter 3 — Fluid Characterization and PVT Modeling: Covers the practical art of representing reservoir fluids in a process simulator — component selection, C$_7$+ characterization, lumping and delumping strategies, PVT experiments (constant mass expansion, constant volume depletion, differential liberation), and EOS parameter tuning to match laboratory data.

Part II: Reservoir and Wells (Chapters 4–6)

Part II covers the upstream elements of the production chain.

• Chapter 4 — Reservoir Engineering and Inflow Performance: Develops reservoir inflow models (Productivity Index, Vogel, Fetkovich), material balance methods, decline curve analysis, and the NeqSim SimpleReservoir class for integrated modeling.
• Chapter 5 — Well Performance and Tubing Design: Covers vertical flow performance, nodal analysis, tubing sizing, wellhead choke modeling, and the coupling between IPR and VFP that determines the well operating point.
• Chapter 6 — Well Networks, Artificial Lift, and Lift Optimization: Extends to multi-well systems, gas lift design and optimization, ESP selection, and network balancing with shared facility constraints.

Part III: Subsea Systems and Transport (Chapters 7–9)

Part III addresses the transport of multiphase fluids from wellhead to processing facility.

• Chapter 7 — Subsea Production Systems: Covers subsea trees, manifolds, flowlines, risers, subsea boosting and processing equipment, and tieback architecture design.
• Chapter 8 — Flowlines, Risers, and Pipeline Hydraulics: Develops multiphase flow correlations (Beggs and Brill), pipeline sizing, thermal analysis, terrain-induced slugging, and the NeqSim PipeBeggsAndBrills class with elevation profiles.
• Chapter 9 — Flow Assurance: Addresses hydrate prediction and inhibition (MEG, methanol), wax appearance temperature and deposition, asphaltene stability, CO$_2$/H$_2$S corrosion prediction, scale risk, and slugging mitigation.

Part IV: Topside Processing (Chapters 10–13)

Part IV covers the core process engineering of the production facility.

• Chapter 10 — Separation Technology and Equipment Design: Develops separator design (gravity separation theory, retention time, Souders-Brown K-factor), multi-stage separation optimization, three-phase separator modeling, and the NeqSim Separator and ThreePhaseSeparator classes.
• Chapter 11 — Oil Processing and Stabilization: Covers crude oil dehydration, electrostatic coalescers, desalting, stabilization columns, vapor pressure control, and oil export preparation.
• Chapter 12 — Gas Processing and Conditioning: Addresses gas dehydration (TEG absorbers), hydrocarbon dew point control (JT, turbo-expander, mechanical refrigeration), acid gas removal (amine absorption), and NGL recovery (turbo-expander, refrigeration).
• Chapter 13 — Produced Water Treatment: Covers oil-water separation mechanisms, hydrocyclones, flotation units, water injection quality, and the increasing importance of water treatment capacity as a production constraint.

Part V: Compression, Heat Transfer, and Power (Chapters 14–18)

Part V covers the energy-intensive equipment that often constrains production.

• Chapter 14 — Gas Compression Systems: Develops compression thermodynamics, staging optimization, intercooling, anti-surge control, and the NeqSim compressor model with polytropic efficiency.
• Chapter 15 — Compressor Characteristics and Performance Curves: Covers centrifugal compressor maps (head vs. flow at various speeds), surge line, stonewall, reciprocating compressor modeling, and operating point prediction.
• Chapter 16 — Heat Exchangers and Thermal Design: Addresses shell-and-tube and plate heat exchanger design, LMTD and effectiveness-NTU methods, fouling factors, and the NeqSim HeatExchanger class.
• Chapter 17 — Valves, Flow Control, and Pressure Relief: Covers control valve sizing (ISA/IEC Cv method), choke flow modeling (critical and subcritical), and relief valve sizing per API 520/521.
• Chapter 18 — Power Production and Energy Sources: Addresses gas turbines, waste heat recovery, combined cycle systems, fuel gas system management, and platform power optimization.

Part VI: Export, Capacity, and Debottlenecking (Chapters 19–21)

Part VI addresses the constraints that limit total production.

• Chapter 19 — Export Systems and Fiscal Metering: Covers oil and gas export systems, pipeline hydraulics for long-distance transport, fiscal metering technology, and gas quality specification tracking.
• Chapter 20 — Capacity Checks and Equipment Utilization: Develops methods for rating each piece of equipment against its design capacity, calculating utilization factors, and identifying the binding constraint (the "bottleneck") on total production.
• Chapter 21 — Systematic Debottlenecking and Utilization Analysis: Presents a structured approach to identifying and relieving bottlenecks — evaluating options (re-wheeling compressors, adding separation stages, upgrading pumps) and quantifying their production and economic impact.

Part VII: Production Optimization (Chapters 22–28)

Part VII is the heart of the book, applying all preceding theory to optimization.

• Chapter 22 — Production Optimization Theory and Methods: Develops the mathematical foundations — objective functions, constraints, gradient-based methods (sequential quadratic programming), derivative-free methods (Nelder-Mead, pattern search), and evolutionary algorithms (genetic algorithms, particle swarm).
• Chapter 23 — The NeqSim Optimization Framework: Introduces NeqSim's ProductionOptimizer, CapacityConstrainedEquipment, and the optimization workflow classes that connect process models to optimization solvers.
• Chapter 24 — Production Optimization Implementation: Provides complete worked examples of production optimization for realistic field scenarios — separator pressure optimization, gas-lift allocation, compression scheduling.
• Chapter 25 — Real-Time Utilization Monitoring: Covers continuous monitoring of equipment utilization against capacity, automated constraint identification, and advisory systems for operations engineers.
• Chapter 26 — Well and Network Optimization: Addresses multi-well rate allocation (the "well allocation problem"), gas-lift optimization across multiple wells sharing a common gas supply, and network balancing with looped systems.
• Chapter 27 — Multi-Scenario and Stochastic Optimization: Extends optimization to uncertain parameters — uncertain reservoir performance, uncertain commodity prices, and uncertain equipment availability — using Monte Carlo methods and robust optimization.
• Chapter 28 — Field Development Optimization and VFP Tables: Connects facility optimization to field development decisions, including VFP table generation for reservoir simulators, integrated production forecasting, and investment timing optimization.

Part VIII: Dynamic Operations and Advanced Methods (Chapters 29–32)

Part VIII covers time-dependent behavior and advanced computational methods.

• Chapter 29 — Dynamic Simulation and Process Control: Develops transient simulation methodology, PID controller design and tuning, anti-surge control, level and pressure control, and dynamic optimization for startup, shutdown, and load change.
• Chapter 30 — Digital Twins, Automation, and AI-Assisted Optimization: Covers the integration of NeqSim models with real-time data streams, the Automation API for digital twin applications, and machine learning augmentation of physics-based models.
• Chapter 31 — Numerical Methods and Solver Convergence: Addresses the mathematical algorithms underlying flash calculations (successive substitution, Newton-Raphson, inside-out), process convergence (Wegstein, Broyden, direct substitution), and optimization solvers.
• Chapter 32 — Advanced Optimization Methods: Covers surrogate-assisted optimization, multi-objective optimization (Pareto-optimal solutions), stochastic programming, and model predictive control (MPC) for production systems.

Part IX: Applications and Outlook (Chapters 33–35)

Part IX brings everything together with integrated applications.

• Chapter 33 — Onshore Gas Processing Plants: Applies the full methodology to an onshore gas plant with NGL recovery, fractionation, acid gas treatment, and product optimization.
• Chapter 34 — Integrated Case Studies: Presents three complete case studies — a North Sea gas-condensate field with declining reservoir pressure, a deepwater oil field with gas lift and subsea boosting, and a mature field approaching tail production with high watercut.
• Chapter 35 — Future Directions in Production Optimization: Discusses emerging trends — CCS integration (using platforms for CO$_2$ injection), hydrogen production and blending, electrification of offshore installations, autonomous operations, and the expanding role of AI and digital twins.

How to Use This Book

For students taking a production engineering or process simulation course: Read Part I (Chapters 1–3) sequentially to build the foundation, then select chapters from Parts II–V according to the course syllabus. The exercises at the end of each chapter provide graded problems from conceptual to computational. The companion Jupyter notebooks can be run on any laptop with Python installed.

For practicing engineers working on production optimization: Start with this chapter for orientation, then jump directly to the chapters relevant to your current challenge — separator optimization (Chapter 10), compressor debottlenecking (Chapter 21), gas-lift allocation (Chapter 26), or digital twin implementation (Chapter 30). Each chapter is designed to be largely self-contained, with explicit cross-references to prerequisite material.

For researchers exploring new optimization methods or thermodynamic models: Part VII provides the optimization framework, Part VIII covers advanced methods, and the companion Jupyter notebooks provide a starting point for computational experiments. The open-source nature of NeqSim means you can extend the models, implement new algorithms, and contribute improvements back to the community.

---

Reproduced Calculation Results

These examples use the stated fluid recipes and operating assumptions. Curves represent NeqSim calculations unless a caption identifies an analytical illustration, assumed equipment map or synthetic data.

Gas Density spans 8.063–204.1 kg/m³ across the plotted cases.

Compression reduces the volume occupied by a fixed gas mass; real-gas interactions make the density change nonlinear. A fixed mass flow can impose a much larger actual-volume load on low-pressure piping and compressor suction equipment. Use actual inlet density at each operating case when converting mass rate to velocity, vessel gas load and compressor inlet volume.

Compressibility Factor Z spans 0.7503–0.9744 across the plotted cases.

Attractive interactions initially reduce gas volume below the ideal-gas prediction; repulsive effects become more important as pressure increases. A constant Z can distort standard-to-actual volume conversion and the pressure dependence of available throughput. Evaluate Z at the local temperature, pressure and composition along the system rather than applying a single standard-condition value.

Dew branch: pressure spans 1.603–87.5 bara across the plotted cases. Bubble branch: pressure spans 1.105–77.13 bara across the plotted cases.

The saturation boundary identifies states where a second hydrocarbon phase can first appear; it is a thermodynamic boundary, not an operating trajectory. Expansion and cooling can cross into two-phase flow even when the upstream stream is a single gas phase. Overlay the actual operating path and uncertainty envelope, using only finite traced branches and the contract pressure basis for dew-point limits.

Actual volumetric flow spans 446.6–2087 m³/hr across the plotted cases.

The plotted actual volume is the specified mass flow divided by the NeqSim bulk density at each pressure. Lower-pressure gas imposes a greater actual-volume load for the same mass production; this calculation does not predict well deliverability. Use actual inlet volume for equipment capacity checks, and solve the coupled reservoir and tubing model separately for the production rate.

Selected numerical ranges from the plotted cases:

Ranges describe the sampled cases; they are not independent validation tolerances.

1.8 Summary

This chapter has established the foundations for the remainder of the book. The key points are:

1. Production optimization is the discipline of maximizing economic value from hydrocarbon production while respecting safety, environmental, and equipment constraints. It is formally a constrained optimization problem operating on both daily and life-of-field time scales.

1. The production system is an integrated chain — reservoir, wells, subsea, topside, compression, power, export — where every element constrains the others. The coupling through pressure, flow, and composition means that system-level optimization is essential; local equipment optima do not guarantee a global system optimum.

1. Process simulation is the computational backbone of optimization. It solves mass, energy, and momentum balances coupled with thermodynamic equilibrium (flash calculations) to predict system behavior. Both steady-state and dynamic simulation are needed for comprehensive optimization.

1. NeqSim is an open-source Java/Python library for rigorous thermodynamic and process simulation. Its key strengths — multiple equations of state, complete equipment library, Python/Jupyter integration, and automation API — make it well-suited for production optimization workflows.

1. The optimization workflow is: Model → Calibrate → Optimize. Build a process model, calibrate it against measured data, then search for the operating conditions that maximize value within constraints.

1. This book develops every element of the production chain systematically across nine parts — from thermodynamic foundations (Part I) through reservoir and wells (Part II), subsea transport (Part III), topside processing (Part IV), compression and power (Part V), capacity and debottlenecking (Part VI), optimization methods (Part VII), dynamic operations (Part VIII), and integrated applications (Part IX) — with 35 chapters, companion Jupyter notebooks, and exercises.

---

Verification of the worked examples

The pressure budget now separates reservoir drawdown from downstream losses, and the water-column pressure estimate is checked by an independent unit conversion. The separator flowsheet has finite-state, component, mass and enthalpy checks; these checks support its calculated phase allocation, while export limits in the chapter remain illustrative contract inputs.[3]

The calculation and literal-code records are in verification/scientific_revision/ch01_manuscript_physics.json; the chapter scope and code hashes are indexed in foundations_review.json.

Exercises

1. Exercise 1.1 — First simulation. Install NeqSim and run the two-stage separation example from Section 1.5.4. Record the gas and oil production rates. Then change the HP separator pressure from 80 bara to 40 bara and re-run. How do the gas and oil rates change? Explain the physical reason for the change, referencing the flash calculation and the effect of pressure on vapor-liquid equilibrium.

1. Exercise 1.2 — Three-stage separation. Extend the two-stage separation model by adding a third (LP) separator at 5 bara downstream of the MP separator liquid outlet. Calculate the total gas production from all three stages. How does the total gas production compare to the two-stage case? Plot the gas production from each stage as a bar chart using matplotlib.

1. Exercise 1.3 — Bubble point calculation. For the fluid composition in Section 1.5.3, calculate the bubble point pressure at 80°C using NeqSim's ThermodynamicOperations class with bubblePointPressureFlash(). What fraction of the feed is gas at the HP separator conditions (60 bara, 80°C)? At what pressure does the fluid become entirely liquid (single-phase)?

1. Exercise 1.4 — Back-pressure sensitivity. Using the two-stage separation model from Section 1.5.4, vary the HP separator pressure from 20 bara to 100 bara in steps of 10 bara. For each case, record the total gas production from both stages and the oil density from the MP separator. Plot both quantities against separator pressure. At what HP separator pressure is total gas production maximized? Why does increasing the pressure above this point reduce gas production?

1. Exercise 1.5 — System coupling. Consider a system where reducing the HP separator pressure from 60 to 40 bara would increase well production by 10% (due to lower back-pressure). However, the lower separator pressure means the gas must be compressed from 40 bara to 200 bara instead of from 60 bara to 200 bara. Using the ideal compression power formula from Section 1.2.5, calculate the ratio of compression power at 40 bara suction vs. 60 bara suction, assuming $\gamma=1.3$, $R=450$ J/(kg K), $T_1=313$ K, $\eta_s=0.80$ and three perfectly intercooled stages. Use a baseline gas flow of 100 kg/s and a 10% increase in both gas and oil rate. Compare the incremental gas-compression energy cost at US\$0.05/kWh with the incremental gross oil revenue at US\$60/bbl and a 100,000-bbl/d baseline. State the exclusions: fixed efficiency, ideal gas, no pressure losses, unchanged gas/oil ratio, and no taxes or additional equipment/operating costs.

1. Exercise 1.6 — Industry examples. Research and describe three published case studies where production optimization delivered measurable value (use SPE papers, OTC papers, or company reports). For each case, identify: (a) the optimization objective, (b) the key decision variables, (c) the binding constraints, and (d) the reported economic improvement.

1. Exercise 1.7 — Automation API exploration. Using the NeqSim Automation API (process.getAutomation()), list all the variables available on the HP Separator in the example process from Section 1.5.4. Identify which variables are inputs (writable) and which are outputs (read-only). Change one input variable, re-run the process, and report the effect on at least two output variables.

1. Exercise 1.8 — Optimization formulation. Formulate the HP separator pressure optimization from Exercise 1.4 as a formal constrained optimization problem. Define: (a) the decision variable $u$, (b) the objective function $J(u) = R_{\text{oil}}(u) + R_{\text{gas}}(u) - C_{\text{comp}}(u)$ where $R$ is revenue and $C$ is compression cost, and (c) at least three inequality constraints (one equipment capacity, one quality specification, one safety limit). Write the mathematical statement using the notation from Section 1.1.1.

---

1. Beggs, H. D. (2003). Production Optimization Using Nodal Analysis, 2nd edition. OGCI Publications.
2. Golan, M. and Whitson, C. H. (1991). Well Performance, 2nd edition. Prentice Hall.
3. Economides, M. J., Hill, A. D., Ehlig-Economides, C., and Zhu, D. (2013). Petroleum Production Systems, 2nd edition. Prentice Hall.
4. Svalheim, S. and King, D. C. (2003). Life of Field Energy Performance. SPE 83993, presented at the SPE Offshore Europe Conference, Aberdeen, September 2–5.
5. Bieker, H. P., Slupphaug, O., and Johansen, T. A. (2007). Real-Time Production Optimization of Oil and Gas Production Systems: A Technology Survey. SPE Production & Operations, 22(4), 382–391.
6. Codas, A., Campos, S., Camponogara, E., Gunnerud, V., and Sunjerga, S. (2012). Integrated production optimization of oil fields with pressure and routing constraints: The Urucu field. Computers & Chemical Engineering, 46, 178–189. DOI: 10.1016/j.compchemeng.2012.06.016.
7. Gunnerud, V. and Foss, B. (2010). Oil Production Optimization — A Piecewise Linear Model, Solved with Two Decomposition Strategies. Computers & Chemical Engineering, 34(11), 1803–1812.
8. Soave, G. (1972). Equilibrium Constants from a Modified Redlich-Kwong Equation of State. Chemical Engineering Science, 27(6), 1197–1203.
9. Peng, D. Y. and Robinson, D. B. (1976). A New Two-Constant Equation of State. Industrial & Engineering Chemistry Fundamentals, 15(1), 59–64.
10. Beggs, H. D. and Brill, J. P. (1973). A Study of Two-Phase Flow in Inclined Pipes. Journal of Petroleum Technology, 25(5), 607–617.
11. Arps, J. J. (1945). Analysis of Decline Curves. Transactions of the AIME, 160(1), 228–247.
12. Vogel, J. V. (1968). Inflow Performance Relationships for Solution-Gas Drive Wells. Journal of Petroleum Technology, 20(1), 83–92.
13. Rachford, H. H. and Rice, J. D. (1952). Procedure for Use of Electronic Digital Computers in Calculating Flash Vaporization Hydrocarbon Equilibrium. Journal of Petroleum Technology, 4(10), 19–20.
14. Lohrenz, J., Bray, B. G., and Clark, C. R. (1964). Calculating Viscosities of Reservoir Fluids from Their Compositions. Journal of Petroleum Technology, 16(10), 1171–1176.
15. Kontogeorgis, G. M. and Folas, G. K. (2010). Thermodynamic Models for Industrial Applications: From Classical and Advanced Mixing Rules to Association Theories. Wiley.

Running the examples. Start the source-workspace Python session described in Chapter 1, then run this chapter's Python blocks in reading order. Java blocks form a separate sequence using the same NeqSim build; carry forward objects from preceding Java blocks. The release execution records are in verification/; a successful run establishes API compatibility, while physical validation also requires the checks discussed in the text.

Learning Objectives

After reading this chapter, the reader will be able to:

1. Describe the fundamental thermodynamic principles underlying process simulation
2. Explain the mathematical structure of cubic equations of state (SRK, PR) and the CPA model
3. Derive and apply fugacity-based phase equilibrium criteria
4. Understand and implement flash calculation algorithms including Rachford-Rice, successive substitution, and stability analysis
5. Calculate thermodynamic and transport properties from an equation of state
6. Use NeqSim to create fluids with different equations of state, perform flash calculations, and extract all relevant properties
7. Recognize when multi-phase checks (VLLE, three-phase) are necessary

2.1 Introduction

Every process simulation begins with thermodynamics. The accuracy of a production optimization model depends critically on the accuracy of its property predictions — densities that determine hydrostatic heads in risers, enthalpies that govern heat exchanger duties, viscosities that control pressure drops in pipelines, and phase equilibria that dictate how much gas and liquid emerge from a separator.

This chapter develops the thermodynamic framework from first principles through to practical implementation in NeqSim. We begin with the equation of state as the central tool for predicting fluid properties, progress through phase equilibrium theory and flash calculation algorithms, and conclude with the full property calculation chain that supports every simulation in this book.

The reader with a strong background in thermodynamics may skim the theoretical development and focus on the NeqSim implementation sections. The reader new to equation-of-state methods should work through the theory carefully, as every subsequent chapter relies on these foundations.

2.2 Equations of State

An equation of state (EOS) is a mathematical relationship between pressure $P$, molar volume $v$, and temperature $T$ that describes the thermodynamic behavior of a fluid. For process simulation in the oil and gas industry, the equation of state serves as the master model from which all other thermodynamic properties are derived.

2.2.1 The Ideal Gas Law

The simplest equation of state is the ideal gas law:

where $R = 8.314$ J/(mol·K) is the universal gas constant. This equation assumes that molecules have zero volume and no intermolecular interactions. It is accurate only at low pressures and high temperatures — conditions far from those encountered in most oil and gas applications, where pressures of 100–500 bara and temperatures of 50–200°C are common.

The ideal gas law provides a useful reference state against which real gas behavior is measured. The compressibility factor $Z$ quantifies the departure from ideal behavior:

For an ideal gas, $Z = 1$. For real gases at reservoir and process conditions, $Z$ typically ranges from 0.3 to 1.2.

2.2.2 The van der Waals Equation

In 1873, Johannes van der Waals introduced the first equation of state that accounts for molecular volume and intermolecular attractions:

The parameter $a$ represents attractive forces between molecules, and $b$ represents the volume excluded by molecular size. Both are determined from the critical temperature $T_c$ and critical pressure $P_c$ of the pure component:

While the van der Waals equation is rarely used in modern process simulation, its structure — a repulsive term $RT/(v-b)$ and an attractive term $a/v^2$ — established the template for all subsequent cubic equations of state.

2.2.3 The Soave-Redlich-Kwong (SRK) Equation

The Soave-Redlich-Kwong equation of state (Soave, 1972) modified the Redlich-Kwong equation by introducing a temperature-dependent attractive parameter:

The attractive parameter $a(T)$ includes a temperature correction through the alpha function:

The Soave alpha function depends on the acentric factor $\omega$:

where $T_r = T/T_c$ is the reduced temperature. The acentric factor $\omega$ characterizes the non-sphericity of the molecule and is tabulated for all common components.

The SRK equation can be written in the cubic form by introducing the compressibility factor $Z = Pv/(RT)$:

where:

A cubic at fixed temperature, pressure and composition has one or three real roots, with repeated roots at discriminant-zero conditions. When three admissible roots exist, the outer roots are liquid-like and vapor-like candidates; the middle branch is mechanically unstable. Root count is not a mixture phase-stability test: the coexisting phases have different compositions, and a mixture can split even when the cubic at its overall composition has only one admissible root. Fugacity equality plus a Gibbs-energy stability test selects the equilibrium state.[7, 8]

2.2.4 The Peng-Robinson (PR) Equation

The Peng-Robinson equation (Peng and Robinson, 1976) was developed to improve liquid density predictions compared to the SRK equation:

The parameters are:

The cubic form becomes:

Both the SRK and PR equations provide acceptable accuracy for hydrocarbon systems. The PR equation generally gives better liquid densities, while the SRK equation is often preferred for gas-phase properties. In practice, the choice between them is often dictated by the availability of tuned binary interaction parameters for the specific fluid system.

Table 2.1 compares the key parameters of the SRK and PR equations:

2.2.5 Volume Translation

Both SRK and PR equations systematically under-predict liquid densities. Volume translation (Péneloux et al., 1982) corrects this by shifting the molar volume:

where $c$ is a component-specific volume shift parameter. The shift does not affect vapor-liquid equilibrium (VLE) calculations — it corrects only volumetric properties. In NeqSim, volume translation is applied automatically when configured.

2.2.6 The CPA Equation of State

The Cubic-Plus-Association (CPA) equation of state (Kontogeorgis et al., 1996) extends the SRK equation to handle associating compounds — molecules that form hydrogen bonds, such as water, methanol, MEG (mono-ethylene glycol), and organic acids:

The association term $P_{\text{assoc}}$ accounts for hydrogen bonding and is derived from Wertheim's statistical mechanical theory. It introduces two additional parameters per associating site: the association energy $\epsilon^{AB}$ and the association volume $\beta^{AB}$.

The CPA equation is essential for systems involving:

• Hydrate inhibitors (methanol, MEG, DEG)
• Water content prediction in gas streams
• Produced water with dissolved hydrocarbons
• Acid gas systems (H$_2$S, CO$_2$ with water)

In NeqSim, the CPA model is implemented in the SystemSrkCPAstatoil class, which combines the SRK equation for the physical interactions with the CPA association term.

2.2.7 Other Equations of State in NeqSim

NeqSim supports additional equations of state for specialized applications:

The choice of EOS depends on the fluid system and the required accuracy. For most production optimization applications, SRK or PR with appropriate binary interaction parameters provides adequate accuracy.

2.3 Mixing Rules

For mixtures, the EOS parameters $a$ and $b$ must be calculated from the pure-component values using mixing rules.

2.3.1 Classical (van der Waals) Mixing Rules

The classical one-fluid mixing rules express the mixture parameters as composition-weighted averages:

where $x_i$ is the mole fraction of component $i$ and $k_{ij}$ is the binary interaction parameter (BIP) between components $i$ and $j$. The BIPs are the primary tuning knobs for matching experimental VLE data.

For hydrocarbon-hydrocarbon pairs, $k_{ij}$ is typically small (0.0–0.05). For hydrocarbon-CO$_2$ pairs, $k_{ij}$ is larger (0.10–0.15). For hydrocarbon-H$_2$S pairs, $k_{ij}$ ranges from 0.03–0.08. Accurate BIPs are critical for reliable phase equilibrium predictions.

2.3.2 Huron-Vidal Mixing Rules

For highly non-ideal systems — such as water-hydrocarbon or polar-nonpolar mixtures — the classical mixing rules are insufficient. The Huron-Vidal mixing rules (Huron and Vidal, 1979) incorporate an activity coefficient model (typically NRTL or UNIFAC) at infinite pressure:

where $g_{\text{E},\infty}$ is the excess Gibbs energy at infinite pressure from the activity coefficient model, and $C^*$ is positive in this convention: $\ln 2$ for SRK and $\ln[(2+\sqrt 2)/(2-\sqrt 2)]/(2\sqrt 2)$ for PR. Defining a negative constant instead reverses the displayed sign; do not mix conventions. In NeqSim, Huron-Vidal mixing rules are available and can be selected when the classical rules prove inadequate for strongly non-ideal mixtures.

2.3.3 Setting Mixing Rules in NeqSim

The mixing rule must always be set before performing any flash calculation:

import jpype
jneqsim = jpype.JPackage("neqsim")

# SRK with classical mixing rules
fluid_srk = jneqsim.thermo.system.SystemSrkEos(273.15 + 60.0, 100.0)
fluid_srk.addComponent("methane", 0.80)
fluid_srk.addComponent("ethane", 0.10)
fluid_srk.addComponent("propane", 0.05)
fluid_srk.addComponent("n-butane", 0.03)
fluid_srk.addComponent("CO2", 0.02)
fluid_srk.setMixingRule("classic")

# CPA with specialized mixing rules for water-containing systems
fluid_cpa = jneqsim.thermo.system.SystemSrkCPAstatoil(273.15 + 60.0, 100.0)
fluid_cpa.addComponent("methane", 0.80)
fluid_cpa.addComponent("water", 0.15)
fluid_cpa.addComponent("MEG", 0.05)
fluid_cpa.setMixingRule(10)  # CPA mixing rule

2.3.4 Activity Coefficient Models

While equations of state are the dominant approach for hydrocarbon systems, activity coefficient models are sometimes preferred for highly non-ideal liquid mixtures — particularly aqueous systems, electrolyte solutions, and glycol–water systems where the liquid-phase non-ideality is extreme.

In the activity coefficient approach, the fugacity of component $i$ in the liquid phase is expressed as:

where $\gamma_i$ is the activity coefficient and $f_i^{0,L}$ is the fugacity of pure liquid $i$ at the system temperature and pressure. The vapor phase is typically described by an equation of state:

This "gamma-phi" approach decouples the liquid and vapor descriptions, which is advantageous when the liquid phase is highly non-ideal but the vapor phase is nearly ideal (low to moderate pressures).

The Wilson Equation (Wilson, 1964) is the simplest activity coefficient model that accounts for local composition effects:

where $\Lambda_{ij}$ are binary parameters related to molecular interaction energies. The Wilson equation cannot predict liquid–liquid immiscibility (VLLE), which limits its use in water–hydrocarbon systems.

The NRTL Model (Non-Random Two-Liquid, Renon and Prausnitz, 1968) extends the local composition concept with a non-randomness parameter $\alpha_{ij}$:

where $G_{ij} = \exp(-\alpha_{ij} \tau_{ij})$ and $\tau_{ij} = (g_{ij} - g_{jj}) / (RT)$. NRTL can represent both VLE and LLE, making it suitable for water–hydrocarbon and glycol–water systems. The non-randomness parameter $\alpha_{ij}$ is typically set to 0.2–0.47.

The UNIFAC Model (UNIQUAC Functional-group Activity Coefficients, Fredenslund et al., 1975) is a predictive group-contribution method. Instead of requiring binary parameters for every molecular pair, UNIFAC decomposes molecules into functional groups and uses group–group interaction parameters:

where $\ln \gamma_i^C$ is the combinatorial contribution (molecular size and shape) and $\ln \gamma_i^R$ is the residual contribution (group interactions). UNIFAC is particularly valuable when experimental VLE data are unavailable for parameter fitting.

When to use activity coefficient models instead of EOS:

CPA and Huron–Vidal address different model contributions. CPA adds a site-association free-energy term derived from Wertheim theory to a cubic physical term; it is not an NRTL/activity-coefficient association correction. Huron–Vidal is an excess-Gibbs-energy mixing rule for the cubic contribution and can be used where its parameters and implementation support the system. The chosen CPA association scheme, cross-association rules and interaction data must be stated separately. No such model is validated at all pressures merely because it treats both phases with one framework.

2.4 Fugacity and Chemical Potential

2.4.1 Chemical Potential

The chemical potential $\mu_i$ of component $i$ in a mixture is defined as:

where $G$ is the Gibbs energy and $n_i$ is the number of moles of component $i$. A system at equilibrium has equal chemical potentials for each component in every coexisting phase.

2.4.2 Fugacity

For practical calculations with equations of state, fugacity $f_i$ replaces chemical potential. The fugacity of component $i$ in a mixture is related to the chemical potential by:

The fugacity coefficient $\phi_i$ relates the fugacity to the partial pressure:

For the SRK equation, the fugacity coefficient of component $i$ in the mixture is:

For the PR equation:

These expressions are evaluated by NeqSim internally for every flash calculation. The user does not need to implement them, but understanding their structure helps diagnose convergence issues and interpret results.

2.4.3 Phase Equilibrium Criteria

A system of $N_c$ components distributed between $N_p$ phases is at thermodynamic equilibrium when:

1. Thermal equilibrium: Equal temperatures in all phases: $T^{(1)} = T^{(2)} = \ldots = T^{(N_p)}$
2. Mechanical equilibrium: Equal pressures in all phases: $P^{(1)} = P^{(2)} = \ldots = P^{(N_p)}$
3. Chemical equilibrium: Equal fugacities for each component in all phases:

For a two-phase vapor-liquid system, this becomes:

which simplifies to:

where $K_i$ is the equilibrium ratio (K-value) for component $i$. The K-values are the foundation of all flash calculations.

2.5 Flash Calculation Algorithms

Flash calculations determine the phase split and composition of each phase at specified conditions. The most common types are:

2.5.1 The Rachford-Rice Equation

For a TP flash of a mixture with overall composition $z_i$, the phase split is determined by the Rachford-Rice equation. Let $\beta$ be the vapor fraction (moles of vapor / total moles). Then:

This equation must be solved for $\beta$ given the K-values. The compositions are then:

For fixed positive $K_i$ and normalized nonnegative $z_i$, define the left-hand side as $F(\beta)$. On the physical interval $0\le\beta\le1$,

An interior two-phase root exists when $F(0)>0$ and $F(1)<0$, and is unique unless the nonzero-composition components all have $K_i=1$. Otherwise the fixed-K calculation selects a single-phase endpoint; phase stability still needs the EOS test. The poles $1/(1-K_{\max})$ and $1/(1-K_{\min})$ can bound an algebraic search when the K-values straddle unity, but they are not physical vapor-fraction limits. Use safeguarded Newton iteration or bisection on $[0,1]$ for a bracketed two-phase case.[9]

2.5.2 Successive Substitution (SSI)

The standard algorithm for TP flash is successive substitution iteration:

1. Initialize K-values using Wilson's correlation:

1. Solve Rachford-Rice for $\beta$ and calculate $x_i$, $y_i$
2. Calculate fugacity coefficients $\phi_i^L(x)$ and $\phi_i^V(y)$ from the EOS
3. Update K-values:

1. Check convergence: Require small component fugacity residuals $\max_i|\ln(f_i^L/f_i^V)|$ for materially present components, normalized phase compositions, component balances and admissible phase fractions. A small change in $\ln K_i$ alone can reflect damping or stagnation. Recheck stability before accepting a trivial $K_i=1$ solution.

Successive substitution is robust but can be slow near the critical point, where K-values approach unity and convergence becomes first-order.

2.5.3 Newton-Raphson Acceleration

When successive substitution converges slowly, Newton–Raphson methods use the Jacobian, the first derivatives of the residual equations with respect to their unknowns. Fugacity-composition derivatives can themselves involve second derivatives of a thermodynamic potential. The independent variables are typically $\ln K_i$ or the phase compositions, and the Jacobian includes composition derivatives of the fugacity coefficients.

NeqSim implements combined SSI-Newton methods: successive substitution is used for the first few iterations to establish a good starting point, then Newton-Raphson takes over for rapid quadratic convergence.

2.5.4 Stability Analysis (Michelsen's Method)

Before performing a flash calculation, one must determine whether the system is single-phase or multiphase. Michelsen's tangent plane distance (TPD) criterion (Michelsen, 1982) provides this test.

The tangent plane distance function is defined as:

where $w_i$ is a trial composition. If $\text{TPD}(w) \geq 0$ for all possible $w$, the system is stable in a single phase. If $\text{TPD}(w) < 0$ for any $w$, the system is unstable and will split into two or more phases.

In practice, the stability test is performed by minimizing the TPD function from multiple starting points — typically a vapor-like and a liquid-like initial guess. If any minimum has $\text{TPD} < 0$, a flash calculation is performed using that minimum as the initial K-value estimate.

The stability test is particularly important for:

• Systems near the critical point
• Systems with retrograde condensation (gas condensates)
• Three-phase (VLLE) systems where both a VLE and a LLE check are needed

2.5.5 PH and PS Flash Calculations

For adiabatic processes (valves, mixing) and isentropic processes (ideal compression), enthalpy-specified (PH) or entropy-specified (PS) flash calculations are required. These are solved by nesting a TP flash inside an outer loop that adjusts temperature until the enthalpy or entropy constraint is satisfied:

1. Guess T
2. Perform TP flash at $(T, P)$
3. Calculate H(T, P) or S(T, P) from the EOS
4. Compare with specified H or S
5. Update T using Newton-Raphson (with $C_p$ or $C_p/T$ as the derivative)
6. Repeat until convergence

2.5.6 Multi-Phase Flash (VLLE and Three-Phase)

Many production systems contain water as a separate liquid phase in addition to hydrocarbon vapor and liquid. Three-phase (vapor-liquid-liquid equilibrium, VLLE) flash calculations extend the two-phase algorithm by introducing additional phase fractions and composition vectors.

In NeqSim, multi-phase checking is enabled with setMultiPhaseCheck(True). When enabled, the flash algorithm automatically tests for the presence of additional phases using stability analysis and performs the appropriate multi-phase flash if needed.

# Enable multi-phase checking for water-containing systems
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 60.0, 50.0)
fluid.addComponent("methane", 0.70)
fluid.addComponent("ethane", 0.08)
fluid.addComponent("propane", 0.04)
fluid.addComponent("n-heptane", 0.10)
fluid.addComponent("water", 0.08)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

2.5.7 Hydrate Equilibrium Calculations

Gas hydrates are crystalline inclusion compounds where water molecules form a cage structure that traps small gas molecules (methane, ethane, CO$_2$, H$_2$S). Hydrate formation is a major flow assurance concern in subsea pipelines and wet gas systems, and predicting the hydrate equilibrium temperature is critical for production optimization.

The hydrate equilibrium condition requires that the chemical potential of water in the hydrate phase equals the chemical potential of water in the fluid phase:

The van der Waals–Platteeuw statistical mechanical model (1959) describes the hydrate phase by accounting for the occupancy of different cage types by gas molecules. The hydrate equilibrium temperature depends strongly on pressure and gas composition — heavier hydrocarbons and CO$_2$ tend to form hydrates at higher temperatures (they are stronger hydrate formers).

NeqSim calculates hydrate equilibrium through the ThermodynamicOperations class:

# Hydrate equilibrium temperature at a given pressure
fluid_hyd = jneqsim.thermo.system.SystemSrkCPAstatoil(273.15 + 5.0, 100.0)
fluid_hyd.addComponent("methane", 0.85)
fluid_hyd.addComponent("ethane", 0.08)
fluid_hyd.addComponent("propane", 0.04)
fluid_hyd.addComponent("CO2", 0.03)
fluid_hyd.addComponent("water", 0.10)
fluid_hyd.setMixingRule(10)
fluid_hyd.setHydrateCheck(True)

ops_hyd = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid_hyd)
ops_hyd.hydrateFormationTemperature()
print(f"Hydrate equilibrium T at 100 bara: {fluid_hyd.getTemperature('C'):.1f} C")

2.5.8 Wax and Solid Phase Equilibrium

Heavy paraffinic crude oils can precipitate solid wax (n-alkanes with carbon numbers typically above C$_{18}$–C$_{20}$) when cooled below the wax appearance temperature (WAT). Wax deposition in pipelines reduces flow area and increases pressure drop, making WAT prediction essential for flow assurance design.

The solid–liquid equilibrium for wax formation requires:

where $f_i^S$ is the fugacity of component $i$ in the solid (wax) phase, calculated from the pure-component melting properties (melting temperature, heat of fusion) and the solid-solution activity coefficient model.

The wax appearance temperature is the highest temperature at which any solid phase is thermodynamically stable. Below the WAT, the amount of precipitated wax increases as temperature decreases, following the wax precipitation curve (wt% solid vs temperature).

For production optimization, the key design questions are:

• At what temperature does wax first appear? (WAT — determines insulation or heating requirements)
• How much wax precipitates at pipeline arrival temperature? (determines pigging frequency)
• What pour point does the crude have? (determines pumpability limits)

2.6 NeqSim Implementation: Creating Fluids and Running Flash Calculations

2.6.1 Creating a Fluid with Different EOS

The following example demonstrates how to create the same fluid composition using different equations of state and compare the results:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define composition (mole fractions)
components = [
    ("nitrogen", 0.5),
    ("CO2", 2.5),
    ("methane", 75.0),
    ("ethane", 7.0),
    ("propane", 4.0),
    ("i-butane", 1.0),
    ("n-butane", 2.0),
    ("i-pentane", 0.8),
    ("n-pentane", 0.7),
    ("n-hexane", 0.5),
    ("n-heptane", 3.0),
    ("n-octane", 2.0),
    ("n-nonane", 1.0),
]

# Temperature and pressure
T_K = 273.15 + 80.0  # 80°C in Kelvin
P_bara = 150.0

# SRK equation of state
fluid_srk = jneqsim.thermo.system.SystemSrkEos(T_K, P_bara)
for name, moles in components:
    fluid_srk.addComponent(name, moles)
fluid_srk.setMixingRule("classic")

# Peng-Robinson equation of state
fluid_pr = jneqsim.thermo.system.SystemPrEos(T_K, P_bara)
for name, moles in components:
    fluid_pr.addComponent(name, moles)
fluid_pr.setMixingRule("classic")

# Run TP flash for both
ThermodynamicOperations = jneqsim.thermodynamicoperations.ThermodynamicOperations

ops_srk = ThermodynamicOperations(fluid_srk)
ops_srk.TPflash()
fluid_srk.initProperties()

ops_pr = ThermodynamicOperations(fluid_pr)
ops_pr.TPflash()
fluid_pr.initProperties()

# Compare densities
print(f"SRK liquid density: {fluid_srk.getPhase('oil').getDensity('kg/m3'):.1f} kg/m3")
print(f"PR  liquid density: {fluid_pr.getPhase('oil').getDensity('kg/m3'):.1f} kg/m3")
print(f"SRK gas density:    {fluid_srk.getPhase('gas').getDensity('kg/m3'):.1f} kg/m3")
print(f"PR  gas density:    {fluid_pr.getPhase('gas').getDensity('kg/m3'):.1f} kg/m3")

2.6.2 Running Different Types of Flash Calculations

NeqSim supports all standard flash types through the ThermodynamicOperations class:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 80.0, 100.0)
fluid.addComponent("methane", 0.85)
fluid.addComponent("ethane", 0.08)
fluid.addComponent("propane", 0.04)
fluid.addComponent("n-heptane", 0.03)
fluid.setMixingRule("classic")

ops = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid)

# --- TP Flash ---
ops.TPflash()
fluid.initProperties()
print(f"TP Flash: T = {fluid.getTemperature('C'):.1f} C, P = {fluid.getPressure('bara'):.1f} bara")
print(f"  Vapor fraction: {fluid.getBeta():.4f}")
print(f"  Gas density: {fluid.getPhase('gas').getDensity('kg/m3'):.2f} kg/m3")

# Store the enthalpy for PH flash verification
H_total = fluid.getEnthalpy("J")
S_total = fluid.getEntropy("J/K")

# --- PH Flash ---
# Reduce pressure to 50 bara (e.g., across a valve), keep same enthalpy
fluid_ph = fluid.clone()
fluid_ph.setPressure(50.0, "bara")
ops_ph = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid_ph)
ops_ph.PHflash(H_total)
fluid_ph.initProperties()
print(f"\nPH Flash (isenthalpic expansion to 50 bara):")
print(f"  Temperature: {fluid_ph.getTemperature('C'):.1f} C")
print(f"  Vapor fraction: {fluid_ph.getBeta():.4f}")

# --- PS Flash ---
# Isentropic compression to 200 bara
fluid_ps = fluid.clone()
fluid_ps.setPressure(200.0, "bara")
ops_ps = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid_ps)
ops_ps.PSflash(S_total)
fluid_ps.initProperties()
print(f"\nPS Flash (isentropic compression to 200 bara):")
print(f"  Temperature: {fluid_ps.getTemperature('C'):.1f} C")

# --- Bubble Point ---
fluid_bp = fluid.clone()
ops_bp = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid_bp)
ops_bp.bubblePointPressureFlash(False)
print(f"\nBubble point pressure at {fluid_bp.getTemperature('C'):.1f} C: "
      f"{fluid_bp.getPressure('bara'):.1f} bara")

# --- Dew Point ---
fluid_dp = fluid.clone()
ops_dp = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid_dp)
ops_dp.dewPointPressureFlash()
print(f"Dew point pressure at {fluid_dp.getTemperature('C'):.1f} C: "
      f"{fluid_dp.getPressure('bara'):.1f} bara")

2.6.3 Phase Envelope Calculation

The phase envelope — the locus of bubble and dew point curves in $P$-$T$ space — is essential for understanding fluid behavior. NeqSim calculates phase envelopes using a continuation method. The following fragment demonstrates access to raw candidate arrays. It is not accepted as a complete, physically classified envelope: branch identity, finite domains, closure and fresh TP transition brackets must be checked separately. Getter names and finite cricondenbar values alone are insufficient. The independently rebuilt illustration has separate branch-verification evidence.

import jpype
jneqsim = jpype.JPackage("neqsim")

fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 20.0, 50.0)
fluid.addComponent("methane", 0.80)
fluid.addComponent("ethane", 0.08)
fluid.addComponent("propane", 0.05)
fluid.addComponent("n-butane", 0.03)
fluid.addComponent("n-pentane", 0.02)
fluid.addComponent("n-heptane", 0.02)
fluid.setMixingRule("classic")

ops = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid)
ops.calcPTphaseEnvelope()

# Extract data for plotting
temperatures = [t for t in ops.getOperation().get("dewT")]
pressures = [p for p in ops.getOperation().get("dewP")]
bubble_temps = [t for t in ops.getOperation().get("bubT")]
bubble_pressures = [p for p in ops.getOperation().get("bubP")]
cricondenbar_T = ops.getOperation().get("cricondenbar")[0]
cricondenbar_P = ops.getOperation().get("cricondenbar")[1]

Input relative molar amounts are methane 0.8, ethane 0.08, propane 0.05, n-butane 0.03, n-pentane 0.02, n-heptane 0.02, normalized to mole fractions. Fresh TP flashes bracketed each saturation branch at three sampled pressures; phase amounts and density continuity determined physical branch assignment. Component closure and fugacity equality verify these computed states, but do not establish agreement with measured mixture saturation data.

2.7 Property Calculations

Once the flash calculation has determined the phase split and compositions, all thermodynamic and transport properties can be calculated from the equation of state.

2.7.1 Init Levels in NeqSim

NeqSim uses a hierarchical initialization system to calculate properties at different levels of detail:

Critical rule: After any flash calculation, you MUST call fluid.initProperties() before reading physical/transport properties. The init(3) call alone does NOT initialize transport properties.

# CORRECT pattern:
ops.TPflash()
fluid.initProperties()  # MANDATORY
viscosity = fluid.getPhase("gas").getViscosity("kg/msec")  # Now correct

# UNSAFE pattern (properties may be stale or uninitialized):
ops.TPflash()
# Missing initProperties()!
viscosity = fluid.getPhase("gas").getViscosity("kg/msec")  # A getter value alone does not prove initialization

2.7.2 Thermodynamic Properties

After calling initProperties(), the following thermodynamic properties are available:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create and flash a fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 60.0, 80.0)
fluid.addComponent("methane", 0.80)
fluid.addComponent("ethane", 0.10)
fluid.addComponent("propane", 0.05)
fluid.addComponent("n-heptane", 0.05)
fluid.setMixingRule("classic")

ops = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid)
ops.TPflash()
fluid.initProperties()

# --- Overall properties ---
print(f"Temperature: {fluid.getTemperature('C'):.2f} C")
print(f"Pressure: {fluid.getPressure('bara'):.2f} bara")
print(f"Vapor fraction (molar): {fluid.getBeta():.4f}")
print(f"Number of phases: {fluid.getNumberOfPhases()}")

# --- Gas phase properties ---
gas = fluid.getPhase("gas")
print(f"\nGas phase:")
print(f"  Density: {gas.getDensity('kg/m3'):.3f} kg/m3")
print(f"  Molar mass: {gas.getMolarMass('kg/mol') * 1000:.2f} g/mol")
print(f"  Z-factor: {gas.getZ():.4f}")
print(f"  Enthalpy: {gas.getEnthalpy('J/mol'):.1f} J/mol")
print(f"  Entropy: {gas.getEntropy('J/molK'):.3f} J/(mol·K)")
print(f"  Cp: {gas.getCp('J/molK'):.3f} J/(mol·K)")
print(f"  Cv: {gas.getCv('J/molK'):.3f} J/(mol·K)")
print(f"  Cp/Cv ratio: {gas.getGamma():.4f}")
print(f"  Speed of sound: {gas.getSoundSpeed():.1f} m/s")
print(f"  Viscosity: {gas.getViscosity('kg/msec'):.6f} Pa·s")
print(f"  Thermal conductivity: {gas.getThermalConductivity('W/mK'):.5f} W/(m·K)")
print(f"  JT coefficient: {gas.getJouleThomsonCoefficient('K/bar'):.4f} C/bara")

# --- Liquid phase properties ---
if fluid.getNumberOfPhases() > 1:
    oil = fluid.getPhase("oil")
    print(f"\nOil phase:")
    print(f"  Density: {oil.getDensity('kg/m3'):.1f} kg/m3")
    print(f"  Viscosity: {oil.getViscosity('kg/msec'):.6f} Pa·s")
    print(f"  Thermal conductivity: {oil.getThermalConductivity('W/mK'):.5f} W/(m·K)")
    print(f"  Surface tension: {fluid.getInterphaseProperties().getSurfaceTension(0, 1):.6f} N/m")

2.7.3 Density Calculation from the EOS

The density is obtained directly from the molar volume, which is a root of the cubic EOS:

where $M_w$ is the molar mass of the phase and $v$ is the molar volume from solving the cubic equation. For the gas phase, the larger root is used; for the liquid phase, the smaller root.

2.7.4 Enthalpy and Entropy

In this subsection $H$, $S$ and $C_p$ denote molar properties (J/mol, J/(mol K), and J/(mol K)); $v$ is molar volume and composition is fixed. The residual is relative to an ideal gas at the same $T$, $P$ and composition. These unshifted SRK expressions assume temperature-independent $b$ and the stated classical mixing rule. The NeqSim stream API's unqualified enthalpy value instead represents an extensive flow-basis quantity; specify a unit such as J/mol when comparing with these equations.[10] The molar enthalpy departure is:

For the SRK equation, this integral has the analytical result:

The total enthalpy is:

where $H^{\text{ig}}(T)$ is the ideal gas enthalpy, calculated from the ideal gas heat capacity:

The entropy departure is similarly:

2.7.5 Heat Capacity

The constant-pressure heat capacity $C_p$ and constant-volume heat capacity $C_v$ are derived from the EOS through second derivatives:

The ratio $\gamma = C_p / C_v$ is important for compressor calculations and speed of sound.

2.7.6 Transport Properties

Transport properties — viscosity and thermal conductivity — are not directly calculated from the equation of state. Instead, NeqSim uses corresponding-states correlations:

• Viscosity: Lohrenz-Bray-Clark (LBC) correlation for dense phases, or kinetic theory for dilute gas
• Thermal conductivity: Modified Ely-Hanley corresponding states method

These correlations require the density and composition from the EOS, which is why initProperties() (which calls initPhysicalProperties()) must be invoked after the flash.

Table 2.2 summarizes the property calculation chain:

2.7.7 Viscosity Models in Detail

Viscosity is critical for production optimization — it directly affects pressure drop in pipelines, well deliverability, and separator sizing. NeqSim employs several viscosity models depending on the phase and conditions.

The Lohrenz-Bray-Clark (LBC) Correlation (1964) is the standard method for dense-phase hydrocarbon viscosity. It relates the reduced viscosity to reduced density through a fourth-degree polynomial:

where $\mu^*$ is the dilute-gas viscosity, $\xi$ is the viscosity-reducing parameter defined as:

$\rho_r = \rho / \rho_c$ is the reduced density, and $a_0$ through $a_4$ are universal constants. The LBC correlation requires accurate density from the EOS and the mixture pseudo-critical properties.

The Corresponding States Principle (Pedersen et al., 1984) calculates the viscosity of a hydrocarbon mixture by scaling from a reference substance (typically methane):

This method is particularly effective for reservoir fluids with heavy fractions.

2.7.8 Thermal Conductivity

Thermal conductivity governs heat transfer rates in heat exchangers, pipelines (heat loss to surroundings), and wellbores. The Chung et al. correlation (1988) is widely used for gas-phase thermal conductivity:

where $\mu^0$ is the dilute-gas viscosity, $\Psi$ is a correction factor, $M'$ is a modified molecular weight, and $B_6$, $B_7$, $G_2$, $q$, $y$ are functions of the reduced temperature, density, and acentric factor.

For liquid-phase thermal conductivity, NeqSim uses the Ely-Hanley corresponding-states method, which scales liquid thermal conductivity from a reference fluid.

2.7.9 Surface Tension

Surface tension between gas and liquid phases controls droplet formation in separators, mist eliminator performance, and gas-liquid entrainment. NeqSim uses the parachor method (Macleod-Sugden correlation):

where $\sigma$ is the surface tension (N/m), $P_i$ is the parachor of component $i$ (tabulated), $\rho_L$ and $\rho_V$ are the liquid and vapor densities, and $M_L$ and $M_V$ are the phase molar masses. The parachor is an empirical constant specific to each component and is available in standard databases.

Interfacial tension tends to zero as coexisting phases become identical at a critical point. Its pressure dependence away from that point depends on composition and temperature; increasing pressure does not imply a universal increase. Use measured or validated interfacial tension at the separator conditions for droplet/coalescence calculations rather than a generic range.

2.7.10 Component Properties in Each Phase

Individual component properties can also be accessed:

# Mole fractions in gas phase
gas = fluid.getPhase("gas")
for i in range(gas.getNumberOfComponents()):
    comp = gas.getComponent(i)
    print(f"  {comp.getComponentName():15s}: x = {comp.getx():.6f}, "
          f"K = {comp.getK():.4f}, "
          f"fugacity coeff = {comp.getFugacityCoefficient():.6f}")

2.8 The GERG-2008 Equation

GERG-2008 is a Helmholtz-energy mixture model covering21 defined natural-gas components, including gas, liquid and supercritical states in its published domain. Its uncertainty varies with property, composition and temperature/pressure region; the primary paper supplies the applicable ranges. It does not accept arbitrary petroleum pseudo-components. A density property call after an SRK flash does not turn that flash into a GERG phase-equilibrium calculation or establish fiscal compliance.[11]

# Flash with SRK, then evaluate single-phase gas density with Java GERG-2008.
# The legacy SystemGERG2004Eos requires a separate native library.
fluid_gerg = jneqsim.thermo.system.SystemSrkEos(273.15 + 15.0, 50.0)
fluid_gerg.addComponent("methane", 0.90)
fluid_gerg.addComponent("ethane", 0.05)
fluid_gerg.addComponent("propane", 0.02)
fluid_gerg.addComponent("nitrogen", 0.02)
fluid_gerg.addComponent("CO2", 0.01)
fluid_gerg.setMixingRule("classic")

ops_gerg = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid_gerg)
ops_gerg.TPflash()
fluid_gerg.initProperties()

print(f"GERG-2008 gas density at 15°C, 50 bara: "
      f"{fluid_gerg.getPhase('gas').getDensity_GERG2008():.4f} kg/m3")

2.9 Practical Considerations

2.9.1 EOS Selection Guide

The choice of equation of state depends on the application. The following decision tree provides a systematic approach for selecting the appropriate EOS in NeqSim:

Step 1 — Does the system contain associating compounds? Associating compounds are those that form hydrogen bonds: water, methanol, MEG, DEG, TEG, organic acids, amines. If yes, proceed to Step 1a. If no, proceed to Step 2.

Step 1a — Does the system contain ions or electrolytes? If the system includes dissolved salts, mineral ions (Na$^+$, Cl$^-$, Ca$^{2+}$), or brine, use SystemElectrolyteCPAstatoil. If no electrolytes are present but associating compounds exist, use SystemSrkCPAstatoil with mixing rule 10.

Step 2 — Is this a custody transfer or fiscal metering application? Check the contractually specified property method and composition/state range. The example uses the Java GERG-2008 gas-property calculation; the legacy SystemGERG2004Eos has separate native-library requirements. Agreement with a gas-density reference is not automatic ISO conformity of a complete metering system.

Step 3 — Is liquid density accuracy critical? For oil systems where liquid density prediction is important (stock tank calculations, pipeline holdup, separator sizing), prefer Peng-Robinson (SystemPrEos). For gas-dominated systems, SRK (SystemSrkEos) is equally suitable.

Step 4 — Default recommendation. For general-purpose production optimization where none of the above special conditions apply, use SRK with classical mixing rules (setMixingRule("classic")). SRK is well-tested, has extensive BIP databases, and provides reliable results for natural gas and gas condensate systems.

The full selection matrix:

2.9.2 Binary Interaction Parameters

The accuracy of an EOS model depends heavily on the quality of binary interaction parameters (BIPs). NeqSim contains a database of default BIPs for common component pairs, but these should be verified against experimental data for critical applications.

Common BIP ranges for the SRK equation:

2.9.3 Convergence Issues and Remedies

Flash calculations can occasionally fail to converge. Common causes and remedies include:

1. Near the critical point: Use stability analysis to detect single-phase regions; apply damping to successive substitution
2. Very low temperature: Wilson K-value initialization may be poor; use previous converged solution as initial guess
3. Trivial solution: Both trial phases converge to the same composition; this can be numerical stagnation or a true single phase. Resolve it with stability analysis rather than accepting K=1 automatically
4. Three-phase region: Two-phase flash cannot represent the system — enable multi-phase checking
5. Composition-dependent issues: Very dilute components can cause numerical issues — set a minimum mole fraction

NeqSim handles most of these automatically, but understanding the causes helps diagnose residual issues.

2.9.4 Numerical Precision

All flash calculations involve iterative solution of nonlinear equations. NeqSim uses convergence tolerances appropriate for engineering calculations — typically $10^{-8}$ to $10^{-10}$ for fugacity coefficient ratios. For most production optimization applications, this precision is far greater than the uncertainty in the input data.

2.10 Comparison of EOS Predictions

The following example provides a systematic comparison of SRK and PR predictions for a typical North Sea gas condensate:

import jpype
jneqsim = jpype.JPackage("neqsim")
import json

# North Sea gas condensate composition
composition = {
    "nitrogen": 0.4,
    "CO2": 3.4,
    "methane": 74.2,
    "ethane": 7.8,
    "propane": 3.5,
    "i-butane": 0.7,
    "n-butane": 1.3,
    "i-pentane": 0.5,
    "n-pentane": 0.4,
    "n-hexane": 0.3,
    "n-heptane": 2.5,
    "n-octane": 2.0,
    "n-nonane": 1.5,
    "n-decane": 1.0,
    "water": 0.5,
}

results = {}
for eos_name, eos_class in [
    ("SRK", jneqsim.thermo.system.SystemSrkEos),
    ("PR", jneqsim.thermo.system.SystemPrEos),
]:
    fluid = eos_class(273.15 + 100.0, 200.0)
    for comp, frac in composition.items():
        fluid.addComponent(comp, frac)
    fluid.setMixingRule("classic")
    fluid.setMultiPhaseCheck(True)

    ops = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid)
    ops.TPflash()
    fluid.initProperties()

    results[eos_name] = {
        "vapor_fraction": fluid.getBeta(),
        "gas_density_kg_m3": fluid.getPhase("gas").getDensity("kg/m3"),
        "gas_Z": fluid.getPhase("gas").getZ(),
        "gas_viscosity_cP": fluid.getPhase("gas").getViscosity("kg/msec") * 1000,
    }

    if fluid.getNumberOfPhases() > 1:
        results[eos_name]["oil_density_kg_m3"] = fluid.getPhase("oil").getDensity("kg/m3")
        results[eos_name]["oil_viscosity_cP"] = fluid.getPhase("oil").getViscosity("kg/msec") * 1000

print(json.dumps(results, indent=2))

The parity and deviation panels compare 300, 350 and 400 K at 1, 50 and 100 bara. Maximum absolute relative deviations are 1.421 percent for SRK and 0.811 percent for Peng–Robinson, within the declared 3 percent teaching budget. These are NIST reference-EOS values, not invented experimental points or a general mixture-accuracy claim. The separate benchmark notebook retains inputs, raw reference tables, computed values and tolerances.

Reproduced Calculation Results

These examples use the stated fluid recipes and operating assumptions. Curves represent NeqSim calculations unless a caption identifies an analytical illustration, assumed equipment map or synthetic data.

SRK EOS: gas density spans 8.063–204.1 kg/m³ across the plotted cases. PR EOS: gas density spans 8.089–204.8 kg/m³ across the plotted cases.

SRK and Peng–Robinson use different cubic-EOS attraction and covolume expressions, so they predict different real-fluid volumes for the same composition. The difference is a model-form sensitivity; agreement between two models is not independent experimental validation. Compare the alternatives against measured PVT density over the intended envelope and carry the validated model into capacity calculations.

Cp spans 47.79–59.44 J/mol·K across the plotted cases.

Heat capacity measures the enthalpy response to temperature at fixed pressure and varies with molecular degrees of freedom and non-ideal interactions. A constant heat-capacity estimate may misstate heater duty or the temperature rise associated with compressor work. Use stream enthalpy differences for equipment energy balances and treat a constant Cp only as a local screening approximation.

Enthalpy spans -4492–4352 J/mol across the plotted cases.

The slope of an isobaric enthalpy curve is heat capacity; changes in phase distribution can add a latent-heat contribution. Duty is determined by the difference between inlet and outlet enthalpy, not the absolute value of an arbitrarily referenced enthalpy. Keep the same composition, flow basis and reference convention when calculating duty, and examine phase changes before fitting a straight line.

Gas Viscosity spans 11.23–22.81 µPa·s across the plotted cases.

Increasing gas density changes molecular momentum transport and therefore viscosity; the relationship is not the same as liquid-viscosity behavior. Viscosity affects Reynolds number, friction factor and ultimately the distribution of pressure losses. Initialize transport properties and recalculate them at local pipeline states when checking hydraulic capacity.

Joule-Thomson Coefficient spans 0.1379–0.567 °C/bar across the plotted cases.

The Joule–Thomson coefficient is the local temperature derivative with respect to pressure along a constant-enthalpy path. A positive coefficient predicts cooling for a pressure drop, but its changing magnitude prevents applying one inlet value over a large letdown. Use a PH flash for a finite valve pressure drop and inspect condensation and hydrate proximity along the resulting path.

Selected numerical ranges from the plotted cases:

Ranges describe the sampled cases; they are not independent validation tolerances.

2.11 Summary

Key points from this chapter:

• Equations of state (SRK, PR, CPA) are the foundation of all thermodynamic calculations in process simulation. They relate pressure, volume, and temperature and enable the calculation of all derived properties.
• Mixing rules combine pure-component EOS parameters into mixture parameters. Binary interaction parameters ($k_{ij}$) are the primary tuning knobs.
• Fugacity provides the criterion for phase equilibrium: at equilibrium, the fugacity of each component is equal in all phases.
• Flash calculations determine the phase split and compositions. The Rachford-Rice equation, successive substitution, Newton-Raphson acceleration, and Michelsen's stability analysis form the algorithmic toolkit.
• Property calculation follows a hierarchy: flash → initProperties() → read properties. Transport properties (viscosity, thermal conductivity) require initProperties().
• EOS selection depends on the application: SRK/PR for hydrocarbons, CPA for associating compounds, GERG-2008 for custody transfer.
• NeqSim provides a comprehensive implementation of these methods through the SystemInterface and ThermodynamicOperations classes.

Verification of the worked examples

The worked TP flashes are checked for phase-fraction normalization, reconstruction of feed composition and equality of phase fugacities. Hydrate calculations are checked for finite temperatures only; neither a converged flash nor a physically plausible hydrate temperature establishes empirical accuracy. Independent methane reference-EOS comparisons are reported separately, with their own state range and teaching tolerance.[8, 11]

The calculation and literal-code records are in verification/scientific_revision/ch02_manuscript_physics.json; the chapter scope and code hashes are indexed in foundations_review.json.

Exercises

1. Exercise 2.1: Create a natural gas fluid with the following composition (mol%): N$_2$ 1.0, CO$_2$ 2.0, CH$_4$ 85.0, C$_2$H$_6$ 6.0, C$_3$H$_8$ 3.0, iC$_4$ 1.0, nC$_4$ 1.0, nC$_5$ 0.5, nC$_6$ 0.5. Using SRK, calculate the gas density and Z-factor at 50°C and 100 bara. Repeat with PR and compare.

1. Exercise 2.2: For the gas condensate composition in Section 2.10, calculate the phase envelope (bubble and dew point curves). Identify the cricondenbar and cricondentherm. At what pressure does retrograde condensation begin at 100°C?

1. Exercise 2.3: Demonstrate the Joule-Thomson effect by performing a PH flash: start at 50°C and 200 bara, then expand isenthalpically to 80 bara. What is the temperature after expansion? Why does the gas cool?

1. Exercise 2.4: Create a system with methane (80%) and water (20%) at 20°C and 100 bara. Enable multi-phase checking and determine the stable phase count rather than prescribing it. At these conditions methane is above its pure-component critical temperature; expect a methane-rich fluid and an aqueous liquid, not a separate methane liquid merely because three-phase checking is enabled. Check phase compositions and component closure. Report the water content of the gas phase in ppm (molar).

1. Exercise 2.5: Write a Python script that calculates the gas viscosity of pure methane at 50°C for pressures from 1 to 500 bara. Plot the result, identify whether it is monotonic over the sampled range, and relate it to dilute-gas and dense-fluid transport. Do not impose an initial decrease that the selected model or reference data may not show.

1. Exercise 2.6: Compare the SRK and CPA equations for predicting the water content of a natural gas at 40°C and 70 bara. The dry-gas composition is CH$_4$ 90%, C$_2$H$_6$ 5%, C$_3$H$_8$ 3%, CO$_2$ 2%. Add excess liquid water before selecting the mixing rule, enable a multiphase check, and compare the equilibrium gas-phase water mole fractions. A dry composition without a water inventory cannot represent water saturation. Which EOS is more appropriate and what independent data would test the result?

1. Exercise 2.7: For the North Sea gas condensate in Section 2.10, compute and compare the liquid dropout curve (liquid volume fraction vs. pressure at constant temperature of 100°C) using SRK and PR. At which pressure is the maximum liquid dropout predicted?

1. Exercise 2.8 (Advanced): Implement a simple successive substitution flash algorithm in Python (using NeqSim only for fugacity coefficient calculations) and compare the number of iterations required to converge with NeqSim's built-in flash. Test at conditions both far from and near the critical point.

1. Soave, G. (1972). Equilibrium constants from a modified Redlich-Kwong equation of state. Chemical Engineering Science, 27(6), 1197–1203.
2. Peng, D. Y., & Robinson, D. B. (1976). A new two-constant equation of state. Industrial & Engineering Chemistry Fundamentals, 15(1), 59–64.
3. Kontogeorgis, G. M., Voutsas, E. C., Yakoumis, I. V., & Tassios, D. P. (1996). An equation of state for associating fluids. Industrial & Engineering Chemistry Research, 35(11), 4310–4318.
4. Michelsen, M. L. (1982). The isothermal flash problem. Part I. Stability. Fluid Phase Equilibria, 9(1), 1–19.
5. Rachford, H. H., & Rice, J. D. (1952). Procedure for use of electronic digital computers in calculating flash vaporization hydrocarbon equilibrium. Journal of Petroleum Technology, 4(10), 19–23.
6. Péneloux, A., Rauzy, E., & Fréze, R. (1982). A consistent correction for Redlich-Kwong-Soave volumes. Fluid Phase Equilibria, 8(1), 7–23.
7. Huron, M. J., & Vidal, J. (1979). New mixing rules in simple equations of state for representing vapour-liquid equilibria of strongly non-ideal mixtures. Fluid Phase Equilibria, 3(4), 255–271.
8. Kunz, O., & Wagner, W. (2012). The GERG-2008 wide-range equation of state for natural gases and other mixtures. Journal of Chemical & Engineering Data, 57(11), 3032–3091.
9. Wilson, G. M. (1969). A modified Redlich-Kwong equation of state, application to general physical data calculations. Paper presented at the AIChE National Meeting, Cleveland, OH.
10. Lohrenz, J., Bray, B. G., & Clark, C. R. (1964). Calculating viscosities of reservoir fluids from their compositions. Journal of Petroleum Technology, 16(10), 1171–1176.

Running the examples. Start the source-workspace Python session described in Chapter 1, then run this chapter's Python blocks in reading order. Java blocks form a separate sequence using the same NeqSim build; carry forward objects from preceding Java blocks. The release execution records are in verification/; a successful run establishes API compatibility, while physical validation also requires the checks discussed in the text.

Learning Objectives

After reading this chapter, the reader will be able to:

1. Classify reservoir fluids by type (black oil, volatile oil, gas condensate, dry gas, wet gas) and explain their distinguishing characteristics
2. Describe common PVT laboratory experiments and their role in fluid characterization
3. Apply plus fraction characterization methods (Whitson gamma distribution, Pedersen) to extend a fluid analysis into pseudo-components
4. Use NeqSim to create characterized fluids with plus fractions and run PVT simulations
5. Tune EOS parameters to match experimental PVT data
6. Generate and interpret phase envelopes for different fluid types
7. Calculate key PVT properties: GOR, formation volume factor, API gravity, and compressibility
8. Evaluate fluid samples for quality and apply recombination procedures
9. Select between black oil correlations and compositional models for different engineering applications

3.1 Introduction

Accurate fluid characterization is the starting point for every production optimization model. The reservoir fluid — a complex mixture of hundreds or thousands of hydrocarbon species plus non-hydrocarbons — determines the phase behavior, flow properties, and processing requirements of the entire production system. A poor fluid model propagates errors through every subsequent calculation: wrong phase envelopes lead to incorrect separator design, wrong viscosities produce inaccurate pressure drops, and wrong compositions give misleading export specifications.

This chapter covers the complete fluid characterization workflow: from understanding what the laboratory measures, through the mathematical methods for representing the heavy end of the composition, to the practical steps of building and tuning a fluid model in NeqSim.

Keep data used for parameter fitting separate from holdout or independent validation data. Characterization is an inference from the stated sample and assay; this workflow diagram is not evidence that a particular fluid has been calibrated.

3.2 Reservoir Fluid Types

Reservoir fluids are classified based on their initial conditions relative to the phase envelope, the gas-oil ratio (GOR), and the liquid content at surface conditions.

3.2.1 Classification Criteria

The five standard fluid types are:

3.2.2 Phase Envelope Characteristics

The position of the initial reservoir conditions on the phase envelope determines the fluid type:

• Black oil: Initial conditions are well to the left of the critical point; the reservoir is at a temperature below the critical temperature. As pressure drops during depletion, the fluid crosses the bubble point and gas evolves.
• Volatile oil: Initial conditions are near but to the left of the critical point. The fluid exhibits large changes in properties with small changes in pressure.
• Gas condensate: Initial conditions are to the right of the critical point. The reservoir temperature is between the critical temperature and the cricondentherm. As pressure drops below the dew point, retrograde condensation occurs — liquid drops out in the reservoir.
• Wet gas: The reservoir temperature exceeds the cricondentherm. The fluid is always single-phase in the reservoir, but at surface conditions it falls within the two-phase region, yielding some condensate.
• Dry gas: The stated reservoir-to-surface path remains outside the hydrocarbon two-phase region. This classification depends on the composition and operating path; envelope width alone does not determine it.

3.2.3 Fluid Type Identification from Field Data

In practice, fluid type is identified from:

1. Initial producing GOR — measured at the test separator
2. Stock tank oil gravity (API) — from sample analysis
3. Reservoir temperature and pressure — from well logs and DST data
4. C$_{7+}$ fraction — higher C$_{7+}$ content indicates heavier fluid
5. Visual observation — color of separator liquid (clear/straw = condensate; brown/black = oil)

Table 3.1 provides heuristic guidelines:

3.3 Fluid Sampling

3.3.1 Sampling Methods

Reservoir fluid samples are collected by two fundamentally different methods, each with specific advantages and limitations.

Bottomhole sampling uses a wireline-conveyed tool (e.g., MDT, RDT, or RCI) to capture fluid directly at reservoir conditions. The tool is positioned opposite a permeable zone, a drawdown is applied, and a sample chamber is filled at in-situ pressure and temperature. The key requirement is that the bottomhole flowing pressure during sampling must exceed the saturation pressure; otherwise, the sample captures a two-phase mixture that is not representative of the original single-phase reservoir fluid.

Surface recombination sampling collects separator gas and separator liquid samples at the test separator, then recombines them in the laboratory at the measured producing GOR. This method is used when bottomhole sampling is impractical (e.g., gas condensate wells with high GOR, deep offshore wells) or when operating conditions prevent single-phase sampling.

3.3.2 Recombination Calculation

For surface recombination, the laboratory recombines separator gas and liquid at the measured separator GOR. The recombination calculation determines the mole fractions $z_i$ of each component in the reservoir fluid:

where $y_i$ is the mole fraction in the separator gas, $x_i$ is the mole fraction in the separator liquid, and $n_g / n_o$ is the molar gas-to-oil ratio. The molar ratio relates to the volumetric GOR by:

Here the gas standard volume is divided by the separator-liquid volume, and $M_{o,sep},\rho_{o,sep}$ describe that same sampled liquid. A producing GOR per stock-tank oil volume needs the separator-to-stock-tank shrinkage and flash-gas accounting first. Recombination ratios are measured inputs with uncertainty; changing them solely to force a saturation-pressure match can conceal sample loss or inconsistent bases.

3.3.3 Sample Quality Control

Sample quality is critical. Contamination with drilling mud, phase segregation during sampling, or loss of light ends during transfer can invalidate the entire PVT study. The following QC checks are applied:

1. Consistency check: Opening pressure must be compared at a specified temperature after thermal equilibration and with the sampling pressure/volume history accounted for. A cooler bottle can have lower pressure without leaking; unexpected pressure loss at matched conditions warrants a leak or gas-loss investigation.
2. Contamination monitoring: MDT tool measures optical density during sampling to track mud filtrate contamination. An acceptable contamination limit depends on the fluid, contaminant, intended observable and laboratory correction uncertainty.
3. Material balance: The sum of separator gas and separator liquid compositions, recombined at the field GOR, should yield a single-phase fluid at reservoir conditions. If the recombined fluid is two-phase at the reported reservoir pressure, the sample or GOR data may be incorrect.
4. Comparison with offset wells: Saturation pressure, GOR, and C$_{7+}$ content should be consistent with neighboring wells in the same reservoir.
5. Methane content check: A sudden drop in methane content compared to offset data suggests gas loss during sampling.
6. Repeatability: Multiple PVT experiments should give consistent saturation pressures (within ±2–3%).

3.3.4 Choosing the Right Sampling Method

3.4 PVT Laboratory Experiments

PVT experiments are conducted on reservoir fluid samples to measure phase behavior and properties at reservoir and process conditions. These measurements provide the data against which EOS models are tuned.

3.4.1 Constant Composition Expansion (CCE)

The CCE experiment measures the pressure-volume relationship of a reservoir fluid at reservoir temperature:

1. The sample is loaded into a PVT cell at a pressure above the saturation point
2. Pressure is reduced in steps, and the total volume is recorded at each step
3. The saturation pressure (bubble point for oil, dew point for gas condensate) is identified as the pressure where the slope of the $P$-$V$ curve changes
4. Below the saturation pressure, the total volume (gas + liquid) continues to be measured

The relative volume $V_{\text{rel}}$ at each pressure step is defined as:

where $V_t$ is the total volume at pressure $P$ and $V_{\text{sat}}$ is the volume at the saturation pressure. Above the saturation pressure, the fluid is single-phase and the relative volume is governed by isothermal compressibility:

Below the saturation pressure, gas evolves and the relative volume increases sharply. The Y-function provides a linearization useful for smoothing CCE data below the bubble point:

A plot of $Y$ vs. $P$ should approximate a straight line, and deviations indicate data quality issues.

For gas condensates, the CCE also measures the liquid dropout curve — the volume fraction of liquid in the cell as a function of pressure below the dew point. This is the retrograde condensation curve and is critical for reservoir simulation.

3.4.2 Constant Volume Depletion (CVD)

The CVD experiment simulates the depletion of a gas condensate reservoir where retrograde liquid remains immobile in the pore space:

1. The sample starts at the dew point pressure at reservoir temperature
2. Pressure is reduced in a step (typically 25–50 bar increments)
3. Gas is removed from the top of the cell to restore the original cell volume
4. The volume, composition, and Z-factor of the removed gas are measured
5. The liquid dropout (retrograde condensation volume) at each pressure is measured
6. Steps 2–5 are repeated down to abandonment pressure

The key quantities measured at each depletion step $j$ are:

• Liquid dropout $V_{\text{liq}} / V_{\text{cell}}$: Volume fraction of retrograde liquid
• Gas Z-factor from the removed gas: $Z_j = P_j V_{\text{removed}} / (n_{\text{removed}} R T)$
• Cumulative recovery of each component: $R_i = \sum_j y_{i,j} \Delta n_j / (z_i n_{\text{total}})$
• Two-phase Z-factor: $Z_{\mathrm{2ph},j}=P_j V_{\mathrm{cell}}/(n_{\mathrm{remaining},j}RT)$, with $n_{\mathrm{remaining},j}=n_0-\sum_{k\leq j}\Delta n_k$. This is a bulk inventory factor, distinct from the gas-phase Z-factor. Its use in a reservoir material balance must track composition-dependent withdrawal and condensate inventory

CVD data is essential for calibrating the EOS model for gas condensate production forecasting, particularly the liquid dropout curve and the changing gas composition with depletion.

3.4.3 Differential Liberation (DL)

The differential liberation experiment applies to black oil and volatile oil systems:

1. The sample starts at the bubble point at reservoir temperature
2. Pressure is reduced in a step
3. All evolved gas is removed at that pressure, and its volume and composition are measured
4. The remaining oil volume at that pressure is recorded
5. Steps 2–4 are repeated down to atmospheric pressure
6. The residual oil volume at 60°F (15.6°C) and atmospheric pressure is measured

The key outputs calculated from DL data are:

Solution GOR at each pressure:

Oil formation volume factor relative to residual oil:

Gas formation volume factor at each step:

Note that DL values of $B_o$ and $R_s$ must be corrected using the separator test results before use in reservoir simulation, because the DL flashes all gas at each step (which differs from the separator flash path).

3.4.4 Separator Test

The separator test measures the GOR and oil properties at specific separator conditions:

1. A sample at the bubble point is flashed through a multi-stage separator train
2. Gas and liquid volumes are measured at each stage
3. The final stock tank oil volume, API gravity, and GOR are recorded

This test is critical because it defines the reference conditions for reporting field data:

The separator test also provides the correction factor $B_{ob}$ to convert DL-based $B_o$ values to separator-flash-based values:

where subscript $b$ denotes bubble-point values and subscript $d$ denotes DL values.

3.4.5 Swelling Test

The swelling test measures the effect of injecting gas (lean gas, CO$_2$, N$_2$) into a reservoir oil:

1. A reservoir oil sample is loaded at its bubble point
2. Gas is injected in measured amounts (typically 5–20 mol% increments)
3. The new bubble point pressure and swollen volume are measured after each injection
4. The process reveals how much gas can be dissolved and the resulting pressure changes

The swelling factor $S_f$ at each injection step is:

Swelling test data is essential for:

• Gas injection EOR studies — determining the MMP (minimum miscibility pressure) screening
• CO$_2$ flooding feasibility — how much CO$_2$ dissolves in the oil
• Lean gas cycling in gas condensate reservoirs
• Assessing the effect of gas lift gas dissolving in the produced oil

3.4.6 Viscosity Measurements

Viscosity is measured separately, typically using:

• Capillary viscometer for liquid viscosity at moderate pressures
• Falling ball viscometer for high-pressure measurements
• Electromagnetic viscometer (EV) for in-situ measurements
• Rolling ball viscometer for heavy oils at elevated pressures

Viscosity data at multiple pressures and temperatures is essential for tuning the viscosity correlation in the EOS model. Typical measurement protocol includes viscosity at 5–8 pressure steps above the bubble point and 5–8 steps below, all at reservoir temperature.

3.5 Plus Fraction Characterization

3.5.1 The Plus Fraction Problem

A typical gas chromatography (GC) analysis of a reservoir fluid resolves individual components up to C$_6$ or C$_9$, then reports a "plus fraction" — C$_{7+}$ or C$_{10+}$ — that lumps all heavier components together. This plus fraction may contain hundreds of individual species.

For an EOS model, the plus fraction must be split into a manageable number of pseudo-components, each with estimated critical properties ($T_c$, $P_c$, $\omega$) and molecular weight ($M_w$). The quality of this characterization directly affects the accuracy of the phase envelope, density, and viscosity predictions.

3.5.2 Whitson's Gamma Distribution

Whitson (1983) proposed modeling the molar distribution of the plus fraction using a three-parameter gamma distribution:

where:

• $M$ is the molecular weight
• $\eta$ is the minimum molecular weight (typically 84 g/mol for C$_7$+)
• $\alpha$ is the shape parameter (typically 1–5; $\alpha = 1$ gives an exponential distribution)
• $\beta = (\bar{M} - \eta) / \alpha$ where $\bar{M}$ is the average molecular weight of the plus fraction
• $\Gamma(\alpha)$ is the gamma function

The distribution is divided into $N$ pseudo-components (typically 3–10) by splitting the molecular weight range into intervals and calculating the average properties for each interval. The mole fraction of pseudo-component $k$ is:

where $z_{+}$ is the total mole fraction of the plus fraction and the integral boundaries are chosen to give equal-weight or Gaussian quadrature intervals.

3.5.3 Pedersen Method

The Pedersen method (Pedersen et al., 1989) uses carbon number as the independent variable and applies exponential decay for mole fraction:

where $z_n$ is the mole fraction of carbon number $n$, and $A$ and $B$ are fitted from the extended analysis. Molecular weight and density for each carbon number are:

Here $M_n$ is in g/mol and denotes a petroleum pseudo-component estimate, not the exact molar mass of an n-alkane ($14n+2$ approximately). Fit $A_\rho$ and $B_\rho$ in a declared density unit to the measured fractions and plus-fraction density; require positive, plausible extrapolated densities. The Pedersen method is particularly useful when the laboratory provides a partial extended analysis (e.g., to C$_{20}$ or C$_{30}$) because the exponential fit can extrapolate beyond the measured range. However, for heavy crudes with bimodal distributions (e.g., a wax peak), the simple exponential model may not capture the full distribution.

3.5.4 Critical Property Correlations

Each pseudo-component needs molecular weight, density and estimated critical/acentric properties. Lee–Kesler and Twu correlations use boiling point and specific gravity with specific unit systems and fitted reference-fluid relations. An expression ending in an ellipsis, or a reference-paraffin expression without the specific-gravity correction, cannot be used as a complete correlation. Use the documented NeqSim characterization model or the complete cited original method, retain its units and validity range, and check the resulting saturation pressure and density against the PVT sample. The next example performs the implemented characterization directly.

3.5.5 NeqSim Plus Fraction Methods

NeqSim provides two approaches for handling plus fractions:

TBP fraction method (addTBPfraction): Adds individual pseudo-components with specified molecular weight and density. The critical properties are calculated internally using the selected correlation.

Plus fraction characterization (addPlusFraction): Adds a single plus fraction that NeqSim then splits into pseudo-components using the Whitson or Pedersen method:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Using addPlusFraction for automatic splitting
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 90.0, 200.0)
fluid.addComponent("methane", 50.0)
fluid.addComponent("ethane", 7.0)
fluid.addComponent("propane", 4.0)
fluid.addComponent("n-butane", 2.5)
fluid.addComponent("n-pentane", 1.5)
fluid.addComponent("n-hexane", 1.0)

# Add C7+ as a plus fraction: mole%, MW (kg/mol), density (g/cm3)
fluid.addPlusFraction("C7", 34.0, 220.0 / 1000.0, 0.84)

fluid.setMixingRule("classic")

# Configure characterization
fluid.getCharacterization().setLumpingModel("no lumping")
fluid.getCharacterization().setPlusFractionModel("Pedersen")
fluid.getCharacterization().characterisePlusFraction()

print(f"Components after characterization: {fluid.getNumberOfComponents()}")

3.5.6 Watson Characterization Factor

The Watson (or UOP) characterization factor $K_w$ distinguishes paraffinic from naphthenic and aromatic character:

where $T_b$ is the mean average boiling point in Rankine and $S_g$ is the specific gravity at 60°F. Typical values:

3.5.7 API Gravity

API gravity is the industry-standard measure of oil density:

where $S_g$ is the specific gravity at 60°F relative to water. Higher API means lighter oil:

3.6 Black Oil Correlations vs. Compositional Models

3.6.1 When to Use Each Approach

The choice between black oil correlations and compositional (EOS-based) models depends on the fluid type, the engineering application, and the available data.

Black oil correlations treat the fluid as two pseudo-components (oil and gas) with pressure-dependent properties ($B_o$, $R_s$, $B_g$, $\mu_o$, $\mu_g$) described by empirical correlations. They are appropriate when:

• The fluid is a black oil with GOR < 200 Sm³/Sm³
• The process conditions do not approach the critical point
• No compositional tracking is needed (e.g., no injection gas mixing)
• Speed is important (reservoir simulation with millions of grid cells)
• Limited PVT data is available

Compositional models track each component explicitly using an EOS. They are required when:

• The fluid is volatile oil or gas condensate
• Gas injection (CO$_2$, lean gas, N$_2$) is being evaluated
• Retrograde condensation is important
• Compositional grading affects the reservoir description
• Process simulation requires detailed composition (e.g., export gas specifications)

3.6.2 Standing Correlation (Bubble Point)

Standing (1947) developed one of the earliest and most widely used correlations for bubble point pressure:

where $P_b$ is in psia, $R_s$ is in scf/STB, $\gamma_g$ is the gas specific gravity, $T$ is in °F, and API is stock tank oil gravity.

3.6.3 Vasquez-Beggs Correlations

Vasquez and Beggs (1980) developed correlations for oil FVF and solution GOR as functions of pressure, temperature, gas gravity, and API gravity.

Solution GOR:

where the constants $C_1$, $C_2$, $C_3$ depend on whether API $\leq$ 30 or API > 30, and $\gamma_{gs}$ is the gas gravity corrected to a separator pressure of 100 psig.

Oil formation volume factor:

3.6.4 Comparison and Selection Guidelines

An error percentage belongs to a specified validation dataset and metric; the formula name alone does not establish a universal accuracy ranking.

For production optimization, compositional models are strongly preferred because they provide consistent thermodynamic properties across the full range of conditions from reservoir to export, they naturally handle gas-liquid equilibrium at each separator stage, and they are directly coupled with process simulation. Black oil correlations should only be used for quick screening or when running large reservoir simulation models where compositional tracking is computationally prohibitive.

3.6.6 Calibration, validation and the current PVT interface

The September 2026 PVT workflow starts with an immutable laboratory data basis, an untuned fluid and experiments that run independently. Register only measured quantities in the calibration objective. The current regression interface distinguishes the observables shown below; units are part of the data contract. Validate array lengths, finite values and observation ordering before passing the arrays to Java.[1]

Choose a small set of physically relevant RegressionParameter values with documented bounds. runRegression() returns a RegressionResult containing the tuned fluid, parameter values, objectives and uncertainty diagnostics. Library default bounds and a low objective are not acceptance criteria. Inspect structured residuals, parameters at their bounds and correlations between fitted parameters; these reveal whether the data actually constrain the fitted model. Preserve unused experiments for hold-out predictions with the parameter set frozen. Re-fitting after looking at those observations turns them into calibration data.

PVTReportGenerator assembles results using addCCE, addCVD, addDLE and addSeparatorTest, then generateMarkdownReport(). Its reservoir metadata setter uses bara and °C, whereas regression experiment temperatures use K. The handoff should retain raw and prepared data, the base and tuned fluids, parameter bounds, residual plots, hold-out outcomes and the approved operating envelope. The current regression package has no EclipseEOSExporter; use the dedicated black-oil export or E300 import workflow for supported file exchange.

3.7 NeqSim Implementation: Fluid Characterization

3.7.1 Creating a Simple Defined Fluid

For fluids where all components are individually identified:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Simple gas condensate - all components defined
fluid = jneqsim.thermo.system.SystemPrEos(273.15 + 100.0, 250.0)
fluid.addComponent("nitrogen", 0.34)
fluid.addComponent("CO2", 3.59)
fluid.addComponent("methane", 74.16)
fluid.addComponent("ethane", 7.90)
fluid.addComponent("propane", 3.58)
fluid.addComponent("i-butane", 0.71)
fluid.addComponent("n-butane", 1.25)
fluid.addComponent("i-pentane", 0.48)
fluid.addComponent("n-pentane", 0.38)
fluid.addComponent("n-hexane", 0.50)
fluid.addComponent("n-heptane", 2.00)
fluid.addComponent("n-octane", 1.80)
fluid.addComponent("n-nonane", 1.30)
fluid.addComponent("n-decane", 1.01)
fluid.addComponent("water", 1.00)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

3.7.2 Creating a Fluid with Plus Fraction Characterization

For fluids with a reported C$_{7+}$ or higher plus fraction, NeqSim provides built-in characterization:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Black oil with C7+ characterization
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 90.0, 200.0)

# Defined components
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 1.2)
fluid.addComponent("methane", 45.0)
fluid.addComponent("ethane", 6.5)
fluid.addComponent("propane", 4.0)
fluid.addComponent("i-butane", 1.5)
fluid.addComponent("n-butane", 2.5)
fluid.addComponent("i-pentane", 1.2)
fluid.addComponent("n-pentane", 1.0)
fluid.addComponent("n-hexane", 1.5)

# Explicit TBP pseudo-components: amounts in a common mole basis, MW and density
fluid.addTBPfraction("C7", 5.0, 96.0 / 1000.0, 0.738)    # C7: M=96, rho=0.738
fluid.addTBPfraction("C8", 4.5, 107.0 / 1000.0, 0.765)   # C8: M=107, rho=0.765
fluid.addTBPfraction("C9", 3.5, 121.0 / 1000.0, 0.781)   # C9: M=121, rho=0.781
fluid.addTBPfraction("C10", 3.0, 134.0 / 1000.0, 0.792)  # C10: M=134, rho=0.792
fluid.addTBPfraction("C11", 2.5, 147.0 / 1000.0, 0.800)  # C11: M=147, rho=0.800
fluid.addTBPfraction("C12", 2.0, 161.0 / 1000.0, 0.810)  # C12: M=161, rho=0.810
fluid.addTBPfraction("C13", 1.8, 175.0 / 1000.0, 0.820)  # C13: M=175, rho=0.820
fluid.addTBPfraction("C14", 1.5, 190.0 / 1000.0, 0.830)  # C14: M=190, rho=0.830
fluid.addTBPfraction("C15", 1.2, 206.0 / 1000.0, 0.837)  # C15: M=206, rho=0.837
fluid.addTBPfraction("C16", 1.0, 222.0 / 1000.0, 0.843)  # C16: M=222, rho=0.843
fluid.addTBPfraction("C17", 0.8, 237.0 / 1000.0, 0.849)  # C17: M=237, rho=0.849
fluid.addTBPfraction("C18", 0.7, 251.0 / 1000.0, 0.854)  # C18: M=251, rho=0.854
fluid.addTBPfraction("C19", 0.6, 263.0 / 1000.0, 0.859)  # C19: M=263, rho=0.859
fluid.addTBPfraction("C20", 5.0, 450.0 / 1000.0, 0.920)  # C20+: M=450, rho=0.920

# Set mixing rule AFTER adding all components (including TBP fractions)
fluid.setMixingRule("classic")

# These TBP pseudo-components are already defined; no automatic plus split is requested.

Important notes:

• Molecular weight is specified in kg/mol (divide g/mol by 1000)
• Density is specific gravity (relative to water)
• setMixingRule() must be called AFTER adding all components
• characterisePlusFraction() must be called AFTER setMixingRule()
• Do NOT use + in fraction names (e.g., use "C20" not "C20+")

3.7.3 Automatic Fluid Creation

For rapid prototyping and screening studies, NeqSim can create fluids from minimal input:

import jpype
jneqsim = jpype.JPackage("neqsim")

# North Sea gas condensate - typical composition
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 80.0, 150.0)

components = {
    "nitrogen": 0.8, "CO2": 2.5, "methane": 72.0,
    "ethane": 8.5, "propane": 4.2, "i-butane": 1.1,
    "n-butane": 1.8, "i-pentane": 0.7, "n-pentane": 0.5,
    "n-hexane": 0.8
}
for comp, frac in components.items():
    fluid.addComponent(comp, frac)

# Add C7+ as TBP fractions with characterization
fluid.addTBPfraction("C7", 2.5, 96.0 / 1000.0, 0.738)
fluid.addTBPfraction("C10", 2.0, 134.0 / 1000.0, 0.792)
fluid.addTBPfraction("C15", 1.3, 206.0 / 1000.0, 0.837)
fluid.addTBPfraction("C20", 1.3, 350.0 / 1000.0, 0.880)

fluid.setMixingRule("classic")
fluid.getCharacterization().characterisePlusFraction()
fluid.setMultiPhaseCheck(True)

# Quick check: flash at reservoir conditions
ops = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid)
ops.TPflash()
fluid.initProperties()
print(f"Phases: {fluid.getNumberOfPhases()}")
print(f"Density: {fluid.getDensity('kg/m3'):.1f} kg/m3")

3.7.4 Running PVT Simulations

Constant Composition Expansion (CCE)

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create characterized fluid
fluid = jneqsim.thermo.system.SystemPrEos(273.15 + 100.0, 400.0)
fluid.addComponent("methane", 60.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("n-butane", 3.0)
fluid.addComponent("n-hexane", 2.0)
fluid.addTBPfraction("C7", 5.0, 96.0 / 1000.0, 0.738)
fluid.addTBPfraction("C10", 4.0, 134.0 / 1000.0, 0.792)
fluid.addTBPfraction("C15", 3.0, 206.0 / 1000.0, 0.837)
fluid.addTBPfraction("C20", 10.0, 450.0 / 1000.0, 0.920)
fluid.setMixingRule("classic")
fluid.getCharacterization().characterisePlusFraction()

# Perform CCE simulation
pressures = [400, 350, 300, 280, 260, 240, 220, 200, 180, 160, 140, 120, 100]
T_res = 273.15 + 100.0

cce_results = []
for P in pressures:
    fluid_copy = fluid.clone()
    fluid_copy.setTemperature(T_res)
    fluid_copy.setPressure(P, "bara")
    ops = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid_copy)
    ops.TPflash()
    fluid_copy.initProperties()

    result = {
        "pressure_bara": P,
        "number_of_phases": fluid_copy.getNumberOfPhases(),
        "vapor_fraction": fluid_copy.getBeta(),
        "density_kg_m3": fluid_copy.getDensity("kg/m3"),
    }
    cce_results.append(result)

for r in cce_results:
    print(f"P = {r['pressure_bara']:6.0f} bara | "
          f"Phases: {r['number_of_phases']} | "
          f"Beta: {r['vapor_fraction']:.4f} | "
          f"Density: {r['density_kg_m3']:.1f} kg/m3")

Separator Test Simulation

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create oil fluid at bubble point conditions
fluid = jneqsim.thermo.system.SystemPrEos(273.15 + 90.0, 250.0)
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 1.5)
fluid.addComponent("methane", 50.0)
fluid.addComponent("ethane", 7.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("n-butane", 3.5)
fluid.addComponent("n-pentane", 2.0)
fluid.addComponent("n-hexane", 2.5)
fluid.addTBPfraction("C7", 5.0, 96.0 / 1000.0, 0.738)
fluid.addTBPfraction("C10", 5.0, 134.0 / 1000.0, 0.792)
fluid.addTBPfraction("C20", 18.0, 400.0 / 1000.0, 0.910)
fluid.setMixingRule("classic")
fluid.getCharacterization().characterisePlusFraction()

# Build separator train
Stream = jneqsim.process.equipment.stream.Stream
Separator = jneqsim.process.equipment.separator.Separator
ThrottlingValve = jneqsim.process.equipment.valve.ThrottlingValve
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

feed = Stream("Feed", fluid)
feed.setFlowRate(10000.0, "kg/hr")
feed.setTemperature(90.0, "C")
feed.setPressure(70.0, "bara")

hp_sep = Separator("HP Separator", feed)

valve_mp = ThrottlingValve("HP-MP Valve", hp_sep.getLiquidOutStream())
valve_mp.setOutletPressure(15.0, "bara")

mp_sep = Separator("MP Separator", valve_mp.getOutletStream())

valve_lp = ThrottlingValve("MP-LP Valve", mp_sep.getLiquidOutStream())
valve_lp.setOutletPressure(1.5, "bara")

lp_sep = Separator("LP Separator", valve_lp.getOutletStream())

process = ProcessSystem()
process.add(feed)
process.add(hp_sep)
process.add(valve_mp)
process.add(mp_sep)
process.add(valve_lp)
process.add(lp_sep)
process.run()

# Report separator test results
print("=== Separator Test Results ===")
print(f"HP gas rate: {hp_sep.getGasOutStream().getFlowRate('MSm3/day'):.4f} MSm3/day")
print(f"MP gas rate: {mp_sep.getGasOutStream().getFlowRate('MSm3/day'):.4f} MSm3/day")
print(f"LP gas rate: {lp_sep.getGasOutStream().getFlowRate('MSm3/day'):.4f} MSm3/day")
print(f"LP separator liquid rate: {lp_sep.getLiquidOutStream().getFlowRate('m3/hr'):.2f} m3/hr")

# Reflash a product sample at the 60 F API reference temperature.
sample = lp_sep.getLiquidOutStream().getFluid().clone()
sample.setTemperature(288.7056)
sample.setPressure(1.01325)
jneqsim.thermodynamicoperations.ThermodynamicOperations(sample).TPflash()
sample.initProperties()
oil_density = sample.getPhase("oil").getDensity("kg/m3")
sg = oil_density / 999.016
api = 141.5 / sg - 131.5
print(f"Flashed sample API gravity at 60 F: {api:.1f}")

3.7.5 Phase Envelope Generation

This characterized PR/TBP recipe is retained as a numerical-diagnostics example. The current source trace does not provide a verified complete saturation envelope: a missing branch and a candidate that fails an independent TP phase-transition bracket must be rejected. Do not report cricondentherm/cricondenbar as fluid specifications from this trace. The separately labelled defined-compound illustration uses a different fluid basis.

import jpype
jneqsim = jpype.JPackage("neqsim")

# Gas condensate fluid
fluid = jneqsim.thermo.system.SystemPrEos(273.15 + 20.0, 50.0)
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 3.0)
fluid.addComponent("methane", 75.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 4.0)
fluid.addComponent("i-butane", 1.0)
fluid.addComponent("n-butane", 1.5)
fluid.addComponent("i-pentane", 0.5)
fluid.addComponent("n-pentane", 0.4)
fluid.addComponent("n-hexane", 0.6)
fluid.addTBPfraction("C7", 2.0, 96.0 / 1000.0, 0.738)
fluid.addTBPfraction("C10", 1.5, 134.0 / 1000.0, 0.792)
fluid.addTBPfraction("C20", 2.0, 350.0 / 1000.0, 0.880)
fluid.setMixingRule("classic")
fluid.getCharacterization().characterisePlusFraction()

# Calculate phase envelope
ops = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid)
ops.calcPTphaseEnvelope()

# Access results
dew_T = [t for t in ops.getOperation().get("dewT")]
dew_P = [p for p in ops.getOperation().get("dewP")]
bub_T = [t for t in ops.getOperation().get("bubT")]
bub_P = [p for p in ops.getOperation().get("bubP")]

import math
finite_branches = {}
for label, ts, ps in [("getter dew", dew_T, dew_P), ("getter bubble", bub_T, bub_P)]:
    points = [(float(t), float(p)) for t, p in zip(ts, ps)
              if math.isfinite(t) and math.isfinite(p) and t > 0 and p > 0]
    finite_branches[label] = points
    print(label, "finite candidate points:", len(points))
print("REJECTED for specification: complete physical branch validation is absent.")
# Branch names are getter labels, not proof of bubble/dew identity.
# A finite candidate additionally needs a fresh TP bracket at the same composition.

Input relative molar amounts are methane 70, ethane 10, propane 8, n-butane 5, n-pentane 4, n-hexane 3, normalized to mole fractions. Fresh TP flashes bracketed each saturation branch at three sampled pressures; phase amounts and density continuity determined physical branch assignment. Component closure and fugacity equality verify these computed states, but do not establish agreement with measured mixture saturation data. This illustration uses a different defined-compound surrogate because the original TBP-fluid trace did not pass the branch checks; it does not validate that original trace.

3.8 EOS Parameter Tuning

3.8.1 Why Tuning Is Necessary

Default EOS parameters (from generalized correlations) typically give acceptable results for defined components but may be inaccurate for the pseudo-components representing the plus fraction. The critical properties assigned to pseudo-components are estimates, and small errors in $T_c$, $P_c$, or $\omega$ propagate into significant errors in saturation pressure and liquid density. Tuning adjusts selected parameters to match experimental PVT data.

3.8.2 Binary Interaction Parameters ($k_{ij}$)

The binary interaction parameter modifies the attractive term in the EOS mixing rule:

A positive $k_{ij}$ reduces the effective attraction between components $i$ and $j$, generally increasing the saturation pressure and reducing miscibility. The following are illustrative starting ranges, not transferable parameter recommendations. A BIP depends on the EOS, mixing rule, heavy-end definition and fitted data; a positive BIP does not guarantee the same saturation-pressure trend for every mixture. Retain the fitted parameter provenance and validate independent states:

3.8.3 Volume Translation

Cubic EOS liquid-density bias depends on fluid, state and parameterization; neither its sign nor a fixed percentage is universal. The consistent Péneloux translation can improve volumes while preserving the original phase-equilibrium conditions.[12] For a translation with the stated sign convention:

where $c_i$ is the volume shift parameter for component $i$. For pseudo-components, $c_i$ is tuned to match the measured liquid density at reservoir conditions.

3.8.4 Matching Priority and Tolerances

The order of matching priority for production optimization is:

1. Saturation pressure — bubble point or dew point (highest priority)
2. Liquid density / oil FVF — affects volumetric calculations
3. Gas-oil ratio — separator test GOR
4. Liquid dropout — for gas condensates (CVD)
5. Viscosity — affects pressure drop calculations

Illustrative teaching targets for tuned models, to be replaced by project-specific criteria and laboratory uncertainty:

3.8.5 Regression Workflow

A systematic tuning workflow:

1. Start with default parameters and compare against experimental data
2. Constrain plus-fraction molecular weight and density to the measured values and their uncertainties; only vary them within a justified measurement uncertainty, preserving the characterized total inventory
3. Tune $k_{ij}$ between methane and heavy pseudo-components to refine saturation pressure and liquid dropout
4. Adjust volume shift to match liquid density
5. Check separator test GOR and API gravity
6. Verify phase envelope shape and critical point location
7. Iterate until all experimental data are matched within acceptable tolerances

3.8.6 Tuning in NeqSim

The following oil-rich illustrative recipe (30 mol% methane and 52 mol% C20 fraction) demonstrates assigning a BIP and verifying a vapor/liquid saturation boundary. It is not a fitted reservoir sample. The earlier methane-rich recipe returned an 821.1-bara candidate at 100°C, but fresh flashes produced two dense phases labelled oil; that candidate is rejected as an unverified gas/liquid bubble point. The oil-rich example instead gives about 171.2 bara and brackets vapor appearance in fresh TP flashes. The synthetic regression below uses the same oil-rich recipe and recovers an imposed parameter, not a laboratory calibration.

import jpype
jneqsim = jpype.JPackage("neqsim")

# Illustrative oil-rich recipe; impose a BIP, without fitting laboratory data
fluid = jneqsim.thermo.system.SystemPrEos(273.15 + 100.0, 300.0)
fluid.addComponent("methane", 30.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 5.0)
fluid.addTBPfraction("C7", 5.0, 96.0 / 1000.0, 0.738)
fluid.addTBPfraction("C20", 52.0, 400.0 / 1000.0, 0.910)
fluid.setMixingRule("classic")
fluid.getCharacterization().characterisePlusFraction()

# Set custom binary interaction parameter
methane_index = 0
heavy_index = fluid.getPhase(0).getNumberOfComponents() - 1
fluid.setBinaryInteractionParameter(methane_index, heavy_index, 0.04)

# Recalculate to check effect
ops = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid)
ops.bubblePointPressureFlash(False)
print(f"Bubble point after tuning: {fluid.getPressure('bara'):.1f} bara")

# Accept the candidate only when fresh TP flashes bracket vapor appearance.
pb = float(fluid.getPressure("bara"))
gas_fractions = []
for trial_pressure in [pb * 0.998, pb * 1.002]:
    trial = fluid.clone()
    trial.setPressure(trial_pressure)
    jneqsim.thermodynamicoperations.ThermodynamicOperations(trial).TPflash()
    trial.initProperties()
    gas_fractions.append(float(trial.getBeta(trial.getPhaseNumberOfPhase("gas")))
                         if trial.hasPhaseType("gas") else 0.0)
assert gas_fractions[0] > 1e-7 and gas_fractions[1] < 1e-8
print("Fresh TP vapor fractions below/above Pb:", gas_fractions)

Simple Regression Example

A bisection approach to find the $k_{ij}$ that matches a target bubble point:

import jpype
jneqsim = jpype.JPackage("neqsim")


def calc_bubble_point(kij_value):
    fluid = jneqsim.thermo.system.SystemPrEos(273.15 + 100.0, 300.0)
    fluid.addComponent("methane", 30.0)
    fluid.addComponent("ethane", 8.0)
    fluid.addComponent("propane", 5.0)
    fluid.addTBPfraction("C7", 5.0, 96.0 / 1000.0, 0.738)
    fluid.addTBPfraction("C20", 52.0, 400.0 / 1000.0, 0.910)
    fluid.setMixingRule("classic")
    fluid.getCharacterization().characterisePlusFraction()

    idx_c1 = 0
    idx_heavy = fluid.getPhase(0).getNumberOfComponents() - 1
    fluid.setBinaryInteractionParameter(idx_c1, idx_heavy, kij_value)

    ops = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid)
    ops.bubblePointPressureFlash(False)
    return fluid.getPressure("bara")

# A synthetic target generated from a known parameter tests the search.
# Replace this with a laboratory target only after checking the bracket.
target_pb = calc_bubble_point(0.035)
kij_low, kij_high = -0.02, 0.10
f_low = calc_bubble_point(kij_low) - target_pb
f_high = calc_bubble_point(kij_high) - target_pb
assert f_low * f_high <= 0.0, "Target not bracketed by admissible BIP range"
for iteration in range(30):
    kij_mid = (kij_low + kij_high) / 2.0
    pb_calc = calc_bubble_point(kij_mid)
    residual = pb_calc - target_pb
    if abs(residual) < 0.05:
        print(f"Synthetic fit: kij={kij_mid:.5f}; Pb={pb_calc:.2f} bara; "
              f"residual={residual:.3f} bar")
        break
    if f_low * residual <= 0.0:
        kij_high = kij_mid
    else:
        kij_low, f_low = kij_mid, residual
else:
    raise RuntimeError("Bubble-point regression did not meet tolerance")

3.9 Key PVT Properties

3.9.1 Gas-Oil Ratio (GOR)

The gas-oil ratio is the ratio of gas volume to oil volume at standard conditions:

The solution GOR $R_s$ is the amount of gas dissolved in the oil at reservoir conditions. As pressure drops below the bubble point, gas evolves and $R_s$ decreases.

3.9.2 Formation Volume Factor

The oil formation volume factor $B_o$ converts reservoir volumes to surface volumes:

The total formation volume factor $B_t$ accounts for both the oil and its dissolved gas:

The gas formation volume factor:

3.9.3 Compressibility

The isothermal compressibility of oil above the bubble point is:

The effective total compressibility in a reservoir is:

3.9.4 Viscosity Correlations

Dead oil viscosity (Beggs–Robinson; viscosity in cP, temperature $T$ in °F, within the correlation calibration domain):

Live oil viscosity:

where $A$ and $B$ depend on the solution GOR.

3.10 Fluid Characterization Quality Checks

After creating a characterized fluid model, several quality checks should be performed:

1. Composition normalization and conservation: Verify mole fractions sum to 1, then separately check total moles, mass and heavy-fraction molecular weight before and after characterization. Normalization alone is not a mass balance.
2. Phase envelope reasonableness: The critical point and cricondenbar should be physically reasonable
3. Saturation pressure: Compare predicted vs. measured bubble/dew point
4. Density: Compare predicted vs. measured at reservoir conditions
5. GOR: Compare predicted vs. measured at separator conditions
6. Molecular weight: The calculated mixture M$_w$ should match the reported value

# Quality check: print fluid summary
ops = jneqsim.thermodynamicoperations.ThermodynamicOperations(fluid)
ops.TPflash()
fluid.initProperties()

print(f"Number of components: {fluid.getNumberOfComponents()}")
print(f"Number of phases: {fluid.getNumberOfPhases()}")
print(f"Mixture molecular weight: {fluid.getMolarMass('kg/mol') * 1000:.1f} g/mol")
print(f"Overall density: {fluid.getDensity('kg/m3'):.1f} kg/m3")

# Check sum of mole fractions
total_z = 0.0
for i in range(fluid.getNumberOfComponents()):
    total_z += fluid.getPhase(0).getComponent(i).getz()
print(f"Sum of mole fractions: {total_z:.6f}")

3.11 Advanced Topics

3.11.1 Lumping and Delumping

For computational efficiency, the pseudo-components from characterization can be lumped into a smaller number of groups. Common lumping schemes:

• 3-component: Light (C$_1$–C$_3$), intermediate (C$_4$–C$_6$), heavy (C$_7+$)
• 6-component: N$_2$+CO$_2$, C$_1$, C$_2$–C$_3$, C$_4$–C$_6$, C$_7$–C$_{12}$, C$_{13+}$
• Reservoir-specific: Optimized grouping based on K-value behavior

3.11.2 Wax and Asphaltene Characterization

For flow assurance studies, the plus fraction characterization must capture the wax-forming and asphaltene fractions:

• Wax: Primarily normal paraffins in the C$_{16}$–C$_{60+}$ range
• Asphaltenes: Heavy polar components. The CPA or PC-SAFT equation is better suited than cubic EOS.

3.11.3 Compositional Grading

In thick reservoirs, gravity causes compositional variation with depth:

Here $z$ increases downward, $M_i$ is kg/mol, and the column is isothermal and in gravitational equilibrium. A geothermal gradient requires a non-isothermal grading model; substituting a varying temperature into this integrated equation is not sufficient.

Reproduced Calculation Results

These examples use the stated fluid recipes and operating assumptions. Curves represent NeqSim calculations unless a caption identifies an analytical illustration, assumed equipment map or synthetic data.

Molecular Weight spans 16.04–230 g/mol across the plotted cases.

Heavy-end characterization distributes an unresolved fraction into pseudo-components with assigned molecular weights and densities. The assumed tail controls liquid density, vapor–liquid equilibrium and predicted condensate recovery even when the total plus fraction is unchanged. Constrain the distribution with measured plus-fraction molecular weight, density and boiling data; retain the characterization settings with the fluid model.

Traced saturation branch: pressure spans 1.105–42.67 bara across the plotted cases.

Heavy hydrocarbons extend the saturation envelope and make gas-condensate phase behavior sensitive to the uncertain C7+ tail. A traced dew branch can be useful even when the solver cannot produce a complete bubble branch; an absent branch must not be replaced by an invented curve. Validate dew pressure and liquid dropout against laboratory data and distinguish calculated branch coverage from the full physical envelope.

Liquid Volume spans 1.783–100 % across the plotted cases.

At fixed composition, successive equilibrium flashes show how the liquid volume changes as pressure traverses the two-phase region. A fixed-composition pressure sweep does not reproduce a constant-volume depletion experiment, which removes gas and changes the remaining composition. Use the curve as a phase-behavior screen; use NeqSim’s actual depletion experiment model when comparing with CVD laboratory data.

Gas-Oil Ratio spans 0.237–34.01 vol/vol at conditions across the plotted cases.

Pressure reduction transfers volatile components from the hydrocarbon liquid into the equilibrium gas phase. An equilibrium phase gas-to-oil ratio is sensitive to the chosen reference volumes and is not automatically the solution GOR reported by a separator or differential-liberation test. State the volume basis and separation procedure, and match those definitions before comparing with laboratory Rs or field GOR.

Formation Volume Factor Bo spans 1.165–4.036 rb/stb across the plotted cases.

Oil formation volume factor compares reservoir oil volume with the stock-tank oil volume obtained from the same material after a specified surface separation. Using unrelated fluids or inconsistent flash histories in the numerator and denominator produces an apparent Bo with no material-balance meaning. Track the same oil sample through reservoir and stock-tank flashes and report the reference temperature, pressure and separation path.

Selected numerical ranges from the plotted cases:

Ranges describe the sampled cases; they are not independent validation tolerances.

3.12 Summary

Key points from this chapter:

• Reservoir fluids are classified as black oil, volatile oil, gas condensate, wet gas, or dry gas based on their phase behavior and composition
• Fluid sampling requires careful attention to method (bottomhole vs. surface recombination), QC checks, and representative conditions
• PVT experiments (CCE, CVD, DL, separator test, swelling test, viscosity) provide the data for fluid model calibration — each experiment provides specific information for different aspects of the model
• Plus fraction characterization splits the heavy end into pseudo-components with estimated critical properties, using methods such as the Whitson gamma distribution or Pedersen correlations
• Black oil correlations (Standing, Vasquez-Beggs) are useful for quick screening, but compositional models are preferred for production optimization
• EOS parameter tuning adjusts BIPs, critical properties, and volume shifts to match experimental data — the saturation pressure and liquid density are highest priority
• NeqSim provides TBP fraction handling, plus fraction characterization, phase envelope calculation, and separator test simulation for complete PVT workflows
• Quality checks (mass balance, phase envelope, saturation pressure, density, GOR) must be performed before using a fluid model in production optimization

Verification of the worked examples

The characterized-fluid examples retain explicit heavy-end assumptions and distinguish mole normalization from mass conservation. The bubble-pressure regression has a numerical residual test, but it fits a synthetic target and is not laboratory validation. The TBP phase-envelope trace in Section 3.7.5 is explicitly rejected as a complete envelope; finite getter arrays alone do not establish physical branch identity or completeness.[12]

The calculation and literal-code records are in verification/scientific_revision/ch03_manuscript_physics.json; the chapter scope and code hashes are indexed in foundations_review.json.

Exercises

1. Exercise 3.1: Given the following plus fraction data for a gas condensate (C$_{7+}$ = 8.5 mol%, M$_w$ = 155 g/mol, density = 0.795 g/cm³), create a NeqSim fluid model with at least 5 pseudo-components. Calculate the phase envelope and identify the fluid type.

1. Exercise 3.2: For a black oil with bubble point pressure of 180 bara at 95°C, perform a CCE simulation from 300 bara down to 50 bara. Plot the relative volume vs. pressure and identify the bubble point from the change in slope.

1. Exercise 3.3: Set up a three-stage separator test (70/15/1.5 bara) for a typical North Sea black oil. Calculate the GOR at each stage and the total GOR.

1. Exercise 3.4: Create two fluid models for the same composition — one with SRK, one with PR. Compare predicted bubble point pressures and liquid densities.

1. Exercise 3.5: Investigate the effect of C$_{7+}$ molecular weight on the phase envelope. Vary the M$_w$ by ±20% and plot three phase envelopes on the same graph.

1. Exercise 3.6: Calculate the API gravity and Watson K-factor for stock tank oils from three different reservoir fluids using NeqSim separator test simulations.

1. Exercise 3.7: Implement a bisection regression to tune the methane-C$_{7+}$ BIP to match a measured bubble point pressure of 210 bara.

1. Exercise 3.8: For a gas condensate with known CVD data, run a NeqSim CVD simulation and compare the predicted liquid dropout curve against experimental data.

1. Exercise 3.9: Using the Standing and Vasquez-Beggs correlations, estimate $P_b$, GOR, and $B_o$ for a fluid with API = 35, $T$ = 200°F, $\gamma_g$ = 0.75. Compare with a NeqSim compositional model.

1. Exercise 3.10 (Advanced): Collect separator gas and liquid compositions from a NeqSim separator test. Perform the recombination calculation manually to reconstruct the feed composition.

1. Whitson, C. H. (1983). Characterizing hydrocarbon plus fractions. Society of Petroleum Engineers Journal, 23(4), 683–694.
2. Pedersen, K. S., Thomassen, P., & Fredenslund, A. (1985). Thermodynamics of petroleum mixtures containing heavy hydrocarbons. Industrial & Engineering Chemistry Process Design and Development, 24(4), 948–954.
3. Pedersen, K. S., & Christensen, P. L. (2007). Phase Behavior of Petroleum Reservoir Fluids. CRC Press.
4. Danesh, A. (1998). PVT and Phase Behaviour of Petroleum Reservoir Fluids. Elsevier.
5. Lee, B. I., & Kesler, M. G. (1975). A generalized thermodynamic correlation based on three-parameter corresponding states. AIChE Journal, 21(3), 510–527.
6. Twu, C. H. (1984). An internally consistent correlation for predicting the critical properties and molecular weights of petroleum and coal-tar liquids. Fluid Phase Equilibria, 16(2), 137–150.
7. Ahmed, T. (2016). Reservoir Engineering Handbook (5th ed.). Gulf Professional Publishing.
8. McCain, W. D., Spivey, J. P., & Lenn, C. P. (2011). Petroleum Reservoir Fluid Property Correlations. PennWell Books.
9. Beggs, H. D., & Robinson, J. R. (1975). Estimating the viscosity of crude oil systems. Journal of Petroleum Technology, 27(9), 1140–1141.
10. Watson, K. M., & Nelson, E. F. (1933). Improved methods for approximating critical and thermal properties of petroleum fractions. Industrial & Engineering Chemistry, 25(8), 880–887.
11. Standing, M. B. (1947). A pressure-volume-temperature correlation for mixtures of California oils and gases. API Drilling and Production Practice, 275–287.
12. Vasquez, M. E., & Beggs, H. D. (1980). Correlations for fluid physical property prediction. Journal of Petroleum Technology, 32(6), 968–970.
13. Peneloux, A., Rauzy, E., & Fréze, R. (1982). A consistent correction for Redlich-Kwong-Soave volumes. Fluid Phase Equilibria, 8(1), 7–23.
14. Riazi, M. R., & Daubert, T. E. (1987). Characterization parameters for petroleum fractions. Industrial & Engineering Chemistry Research, 26(4), 755–759.

Running the examples. Start the source-workspace Python session described in Chapter 1, then run this chapter's Python blocks in reading order. Java blocks form a separate sequence using the same NeqSim build; carry forward objects from preceding Java blocks. The release execution records are in verification/; a successful run establishes API compatibility, while physical validation also requires the checks discussed in the text.

Learning Objectives

After reading this chapter, the reader will be able to:

1. Apply Darcy's law and the radial flow equation to calculate well productivity
2. Construct inflow performance relationships (IPR) for oil wells (Vogel) and gas wells (back-pressure, LIT equations)
3. Explain reservoir drive mechanisms and their effect on pressure decline and recovery factor
4. Perform material balance calculations for different drive types, including the Havlena-Odeh method
5. Analyze well test data using Horner plots and estimate skin factor from pressure buildup
6. Use decline curve analysis (Arps) to forecast production and estimate EUR
7. Understand the NODAL analysis framework for integrated well-reservoir-facility analysis
8. Use NeqSim's SimpleReservoir class and couple it with wellbore models for production optimization workflows
9. Evaluate the coupling between reservoir simulation and process simulation through VFP tables

4.1 Introduction

Reservoir engineering provides the boundary conditions for production optimization. The reservoir determines how much fluid can be produced, at what rate, and for how long. No amount of topside optimization can overcome the fundamental constraints imposed by the reservoir — its pressure, permeability, fluid properties, and remaining reserves.

This chapter covers the reservoir engineering concepts most relevant to production optimization: inflow performance relationships that define what the well can deliver, reservoir pressure decline that governs the production life cycle, well testing methods that characterize the reservoir, and the system analysis framework that connects reservoir performance to the rest of the production chain.

We focus on practical modeling rather than detailed reservoir simulation. For integrated production optimization, the reservoir is typically represented by IPR curves and decline profiles — simplified models that capture the essential behavior without the computational cost of full reservoir simulation.

4.2 Darcy's Law and Radial Flow

4.2.1 Darcy's Law

Darcy's law describes the flow of a single-phase fluid through a porous medium:

where:

• $q$ = volumetric flow rate [m³/s]
• $k$ = permeability [m² or Darcy; 1 D = 9.869 × 10$^{-13}$ m²]
• $A$ = cross-sectional area [m²]
• $\mu$ = fluid viscosity [Pa·s]
• $dP/dx$ = pressure gradient [Pa/m]

The negative sign indicates that flow is in the direction of decreasing pressure.

4.2.2 Radial Flow to a Vertical Well

For steady-state radial flow to a vertical well in a homogeneous reservoir, integrating Darcy's law in cylindrical coordinates gives:

where:

• $q_o$ = oil flow rate at surface conditions [m³/s]
• $k$ = reservoir permeability [m²]
• $h$ = net pay thickness [m]
• $P_e$ = reservoir pressure at the drainage boundary [Pa]
• $P_{wf}$ = flowing bottomhole pressure [Pa]
• $B_o$ = oil formation volume factor [-]
• $\mu_o$ = oil viscosity at reservoir conditions [Pa·s]
• $r_e$ = drainage radius [m]
• $r_w$ = wellbore radius [m]
• $S$ = skin factor [-]

For pseudo-steady flow in a closed circular drainage area, use average reservoir pressure $\bar P$ rather than the boundary pressure $P_e$. In field units (STB/day, mD, ft, psi, cP):

4.2.3 The Productivity Index

The productivity index (PI or $J$) linearizes the well inflow for undersaturated oil:

The PI has units of m³/s/Pa (or bbl/day/psi in field units). A higher PI means the well produces more for a given drawdown. The PI depends on:

• Rock properties: Permeability $k$ and thickness $h$
• Fluid properties: Viscosity $\mu_o$ and FVF $B_o$ (both pressure-dependent)
• Completion quality: Skin factor $S$ (positive = damaged, negative = stimulated)
• Well geometry: Drainage and wellbore radii

Typical productivity indices:

4.2.4 Skin Factor

The skin factor $S$ accounts for the additional pressure drop (or reduced pressure drop) near the wellbore due to:

The apparent skin $S$ transforms the wellbore radius to an effective radius:

A skin of $S = -4$ is equivalent to increasing the effective wellbore radius by a factor of 55 — the effect of a hydraulic fracture.

4.3 Well Testing

4.3.1 Purpose and Types of Well Tests

Well testing provides the primary source of in-situ reservoir characterization data. By measuring pressure and rate at the wellbore during controlled flow periods, we can determine:

• Permeability-thickness product $kh$
• Skin factor $S$
• Average reservoir pressure $\bar{P}$
• Reservoir boundaries and heterogeneities
• Drainage area and connectivity

The two fundamental well test types are:

• Drawdown test: The well flows at a constant rate from an initially shut-in condition. The pressure decline at the wellbore is analyzed.
• Buildup test: The well is shut in after a period of production. The pressure rise during the shut-in period is analyzed.

4.3.2 Pressure Drawdown Analysis

For a constant-rate drawdown in an infinite-acting reservoir, the wellbore pressure during radial flow follows:

where $\gamma = 0.5772$ is Euler's constant, $\phi$ is porosity, $c_t$ is total compressibility, and $t$ is time.

In field units (psi, bbl/day, mD, ft, cp, hr):

A plot of $P_{wf}$ vs. $\log t$ gives a straight line during the infinite-acting radial flow period. The slope $m$ of this line yields:

And the skin factor:

where $P_{1\text{hr}}$ is the pressure at $t = 1$ hour read from the straight line (not necessarily a measured point).

4.3.3 Pressure Buildup Analysis (Horner Plot)

The Horner method is the most widely used buildup analysis technique. After producing for time $t_p$ at rate $q$, the well is shut in and the buildup pressure $P_{ws}$ is measured as a function of shut-in time $\Delta t$:

The Horner plot is $P_{ws}$ vs. $\log\left(\frac{t_p + \Delta t}{\Delta t}\right)$ — this should yield a straight line during the infinite-acting radial flow period. From this line:

Here $m$ is the positive magnitude of the Horner straight-line slope; the plotted pressure decreases with increasing logarithmic Horner ratio. The skin factor is:

where $P_{ws,1\text{hr}}$ is the buildup pressure at $\Delta t = 1$ hour (from the straight line) and $P_{wf,\text{last}}$ is the flowing pressure just before shut-in.

The extrapolation of the Horner straight line to $\frac{t_p + \Delta t}{\Delta t} = 1$ (i.e., infinite shut-in time) gives the extrapolated $P^*$. Under the constant-rate, infinite-acting homogeneous model it estimates the undisturbed reference pressure; in a depleted bounded reservoir it is not automatically the current volume-average pressure. Boundary corrections and production history are required for that interpretation.

As shut-in time increases, the Horner ratio approaches one and the modeled pressure rises toward 330 bara. The 10 bar-per-decade slope and 100 h preceding production time are specified teaching inputs; no measured well test is represented.

4.3.4 Practical Considerations for Well Testing

Several factors complicate real well test interpretation:

• Wellbore storage: Formation fluid continues to enter and compress the wellbore inventory after surface shut-in of a producer, masking the early-time reservoir response. The wellbore storage coefficient is $C = V_w c_w$ for a liquid-filled wellbore.
• Phase redistribution: In gas-liquid wells, phase segregation in the wellbore after shut-in causes a pressure "hump" that must be identified and excluded from analysis.
• Boundary effects: Faults, gas-oil contacts, or drainage boundaries cause deviations from the infinite-acting straight line at late times.
• Rate history: If the well has not been flowing at a constant rate, superposition in time is needed to account for the variable rate history.

4.3.5 Pressure-Transient Derivative Analysis

Modern well test interpretation uses the Bourdet pressure derivative:

On a log-log plot of $\Delta P$ and $P'$ vs. $\Delta t$:

• Wellbore storage: Both $\Delta P$ and $P'$ lie on a unit-slope line
• Infinite-acting radial flow: $P'$ is constant (horizontal line), from which $kh$ is calculated
• Linear flow (fracture or channel): $P'$ follows a half-slope line
• Boundary effects: $P'$ rises (closed boundary) or drops (constant-pressure boundary)

This diagnostic plot is the first step in any well test interpretation, used to identify flow regimes before applying specific analysis methods.

4.4 Inflow Performance Relationships (IPR)

4.4.1 Linear IPR (Undersaturated Oil)

When the flowing bottomhole pressure $P_{wf}$ remains above the bubble point $P_b$, the oil behaves as a single-phase liquid with approximately constant compressibility, viscosity, and FVF. The IPR is linear:

where $P_r$ is the average reservoir pressure and $J$ is the productivity index. The maximum flow rate occurs when $P_{wf} = 0$:

4.4.2 Vogel's IPR (Saturated Oil)

When $P_{wf}$ falls below the bubble point, gas evolves in the reservoir near the wellbore, reducing the effective permeability to oil. Vogel (1968) developed an empirical correlation for this non-linear behavior:

4.4.3 Composite IPR (Above and Below Bubble Point)

For the common case where the reservoir pressure is above the bubble point but $P_{wf}$ falls below it, the composite IPR combines the linear region with the Vogel region:

For $P_{wf} \geq P_b$:

For $P_{wf} < P_b$:

The total $q_{o,\max}$ for the composite IPR is:

Each equation gives zero rate at 300 bara flowing BHP. At zero BHP the respective absolute-open-flow extrapolations are 6000, 4500 and 5500 m³/d. Different parameters are deliberately used to illustrate the forms; this is not a fit of three models to one well or permission to produce at zero BHP.

4.4.4 Fetkovich Method for Gas Wells

Fetkovich (1973) proposed an alternative to the back-pressure equation that uses an isochronal testing concept. The deliverability equation has the form:

where $C$ and $n$ are determined from multi-rate tests. The Fetkovich method is also applied to oil wells with solution gas drive, using a modified approach that accounts for changes in relative permeability:

where $n$ is the deliverability exponent (typically 0.5–1.0). For $n = 1$, this reduces to a simplified form analogous to the back-pressure equation. For oil wells, $n$ is often close to 1.0 at early times and decreases as the reservoir depletes and the gas saturation increases.

4.4.5 Gas Well IPR: Back-Pressure Equation

For gas wells, the simplified back-pressure equation (Rawlins and Schellhardt, 1935):

where $C$ is the performance coefficient and $n$ is the deliverability exponent (0.5 ≤ $n$ ≤ 1.0).

4.4.6 Gas Well IPR: Laminar-Inertial-Turbulent (LIT) Equation

The more rigorous LIT equation separates laminar and turbulent contributions:

Using pseudo-pressures $m(P)$ for improved accuracy:

The pseudo-pressure is defined as:

which accounts for the variation of gas viscosity and Z-factor with pressure.

4.4.7 Future IPR with Reservoir Depletion

As the reservoir depletes, the IPR curve shifts — the maximum rate decreases. For production forecasting, we need the IPR at future reservoir pressures. The future IPR can be estimated by:

For Vogel's method, the future $q_{o,\max}$ at a new reservoir pressure $P_r'$ is:

In simplified form (assuming the permeability-viscosity-FVF ratio changes slowly):

This produces a family of IPR curves that move down and to the left as the reservoir pressure declines, showing how the well's productive capacity diminishes over time.

4.4.8 Horizontal Well IPR

For horizontal wells, the Joshi (1988) productivity equation:

where $L$ is the horizontal well length and:

4.5 Reservoir Pressure Decline and Material Balance

4.5.1 Drive Mechanisms and Recovery Factors

The energy that drives fluid from the reservoir to the wellbore comes from several mechanisms:

Understanding the dominant drive mechanism is essential for predicting reservoir performance and selecting the correct material balance model.

4.5.2 General Material Balance Equation

The general material balance equation for an oil reservoir (Schilthuis, 1936):

where:

• $N_p$ = cumulative oil production
• $N$ = original oil in place (OOIP)
• $R_p$ = cumulative producing GOR
• $m$ = ratio of gas cap volume to oil zone volume
• $W_e$ = cumulative water influx
• $W_p$ = cumulative water production
• Subscript $i$ = initial conditions

4.5.3 Havlena-Odeh Formulation

Havlena and Odeh (1963) rearranged the material balance equation into a straight-line form that is more convenient for analysis. Defining:

The material balance becomes:

For different reservoir types, this reduces to specific straight-line plots:

• Volumetric undersaturated oil ($m=0$, $W_e=0$, above $P_b$): $F=N(E_o+E_{fw})$, where $E_o=B_o-B_{oi}$; oil expansion must be retained
• Solution gas drive ($m = 0$, $W_e = 0$, below $P_b$): $F = N E_o$
• Gas cap drive ($W_e = 0$): $F / E_o = N + N m E_g / E_o$, plot $F/E_o$ vs. $E_g/E_o$ → slope gives $Nm$, intercept gives $N$
• Water drive: $F / E_o = N + W_e / E_o$, plot $F/E_o$ vs. $W_e/E_o$ → slope gives 1, intercept gives $N$

4.5.4 Drive Index Analysis

The drive index quantifies the relative contribution of each energy source to the total production:

where DDI = depletion drive index, SDI = segregation (gas cap) drive index, WDI = water drive index, and CDI = compressibility drive index. The sum of all indices equals 1.0.

Tracking the drive index over time reveals how the dominant drive mechanism changes as the reservoir depletes — for example, solution gas drive may dominate early in the life of a reservoir, but an expanding gas cap may become the dominant mechanism later.

4.5.5 Water Influx Models

For reservoirs with active aquifer support, the water influx term $W_e$ must be calculated using an appropriate aquifer model.

Schilthuis steady-state model (constant influx rate per unit pressure drop):

This is the simplest model and assumes the aquifer responds instantaneously to pressure changes.

Van Everdingen-Hurst unsteady-state model (the most rigorous analytical model):

where $U$ is the aquifer constant, $\Delta P_j$ is the pressure drop at time step $j$, and $W_D$ is the dimensionless water influx function evaluated at the dimensionless time:

The aquifer constant is:

Use SI consistently: permeability $k_a$ in m², time in s, viscosity in Pa s, compressibility in Pa$^{-1}$, and lengths in m; $U$ then has units m³/Pa. The symbol $k_a$ here is permeability, distinct from the preceding steady-influx coefficient. Here $f$ is the fraction of the aquifer circle (1 for full encirclement, 0.5 for half, etc.).

Carter-Tracy approximation — a practical simplification of the van Everdingen-Hurst model that avoids the superposition summation and is easier to implement in spreadsheet calculations.

4.5.6 Gas Material Balance (P/Z Plot)

For a volumetric gas reservoir (no water influx), the material balance simplifies to:

where $G_p$ is the cumulative gas production and $G$ is the original gas in place (OGIP). A plot of $P/Z$ vs. $G_p$ is a straight line:

• The y-intercept gives $P_i/Z_i$
• The x-intercept gives $G$ (OGIP)
• Under the volumetric, isothermal, closed-tank assumptions, $G-G_p$ is remaining gas in place on the same standard-volume basis; recoverable reserves additionally require an abandonment and commercial-development assessment

For gas-condensate depletion, use a compositional material balance consistent with the evolving reservoir inventory and produced composition. A two-phase volume factor inferred from CVD may support a defined depletion path; simply replacing $Z$ in the dry-gas straight line does not generally restore linearity when composition, phase inventory or produced gas composition changes.

The straight line is an assumed idealized material-balance relation. Its intercept and abandonment cut-off illustrate how resource and recovery estimates differ; real interpretation needs consistent standard-volume data, reservoir-average pressure, Z and the reservoir drive assumptions.

4.6 Recovery Factor

4.6.1 Primary Recovery Mechanisms

The recovery factor — the fraction of original hydrocarbons in place that can be produced — varies dramatically with drive mechanism:

Solution gas drive: As pressure drops below the bubble point, dissolved gas evolves and expands, pushing oil toward the wellbore. Recovery is typically 5–30% of OOIP. This is the least efficient natural drive mechanism because the gas expands throughout the reservoir rather than preferentially displacing oil. The GOR increases rapidly as the gas saturation increases.

Water drive: An active aquifer encroaches into the oil zone as production reduces the reservoir pressure, displacing oil from the pore space. Recovery is typically 30–60% of OOIP. Strong water drive maintains near-constant reservoir pressure, which is beneficial for production rates but eventually leads to high water cuts.

Gas cap drive: An existing or forming gas cap expands as pressure decreases, displacing oil downward. Recovery is typically 20–40% of OOIP. The efficiency depends on the ratio of gas cap to oil zone volume and the degree of gas cap segregation.

Gravity drainage: In steeply dipping or thick reservoirs with good vertical permeability, gravity segregation allows oil to drain downward as gas occupies the upper portion. Recovery can be very high (40–70% OOIP) but at low production rates. This mechanism is particularly effective in fractured reservoirs.

Combination drive: Most reservoirs exhibit a combination of mechanisms, often transitioning from one dominant mechanism to another as depletion progresses.

4.6.2 Typical Recovery Factors by Fluid Type

These ranges illustrate why understanding the drive mechanism is critical for production optimization: the choice of operating strategy (pressure maintenance, gas lift, water injection) fundamentally affects the ultimate recovery.

4.7 Decline Curve Analysis

4.7.1 Arps Decline Equations

Arps (1945) defined three types of production decline:

Exponential decline ($b = 0$):

Hyperbolic decline ($0 < b < 1$):

Harmonic decline ($b = 1$):

where:

• $q_i$ = initial production rate
• $D_i$ = initial decline rate [1/time]
• $b$ = decline exponent (0 ≤ $b$ ≤ 1)
• $N_p(t)$ = cumulative production at time $t$

Typical $b$ values by drive mechanism:

4.7.2 Decline Rate and EUR Estimation

The instantaneous decline rate is:

For exponential decline, $D$ is constant. For hyperbolic decline:

The effective annual decline rate $d$ relates to the nominal decline rate $D$ by:

Estimated Ultimate Recovery (EUR) is calculated by integrating the decline curve to the economic limit rate $q_{\text{el}}$:

For exponential decline:

For hyperbolic decline, the time to reach the economic limit is:

4.7.3 Rate-Cumulative Plots

An alternative diagnostic is the rate-cumulative production plot ($q$ vs. $N_p$):

• Exponential decline: Straight line with slope $-D$
• Hyperbolic decline: Concave-upward curve
• Harmonic decline: Concave-upward curve (more so than hyperbolic)

The rate-cumulative plot is useful because:

1. It does not require time data (useful when production records have gaps)
2. The x-intercept directly gives EUR (when extrapolated to $q = 0$ or $q_{\text{el}}$)
3. Changes in decline behavior (e.g., due to workovers or infill wells) are clearly visible as slope changes

4.8 Reservoir Simulation Coupling

4.8.1 Black Oil vs. Compositional Simulation

Reservoir simulation provides the most detailed prediction of reservoir performance, but the choice of simulation model affects how it couples with process simulation:

Black oil simulation uses pressure-dependent properties ($B_o$, $R_s$, $\mu_o$, $B_g$, $\mu_g$) from PVT tables. It is computationally efficient and suitable for black oils. The output (oil rate, gas rate, water rate, pressure) is passed to the process model as boundary conditions.

Compositional simulation tracks individual component mole fractions and solves the flash problem at each grid cell and time step. It is required for volatile oils, gas condensates, and gas injection processes. The output includes detailed stream compositions that can be directly used in NeqSim process models.

4.8.2 VFP Table Import

The primary interface between reservoir and process simulation is the Vertical Flow Performance (VFP) table. This lookup table provides the relationship between bottomhole pressure and flow rate for different:

• Tubing head pressures
• Water cuts
• Gas-liquid ratios
• Artificial lift parameters (gas lift rate, ESP frequency)

The reservoir simulator uses VFP tables to calculate the flowing bottomhole pressure for each well at each time step, which then determines the production rate through the IPR. VFP tables are typically generated by a wellbore hydraulics model (Chapter 5) and can be created using NeqSim's pipe flow capabilities.

4.8.3 Coupling with NeqSim Wellbore Models

NeqSim provides the PipeBeggsAndBrills class for multiphase wellbore flow calculations. This can be coupled with the SimpleReservoir class to create an integrated reservoir-to-separator model:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create reservoir fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 95.0, 250.0)
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 1.8)
fluid.addComponent("methane", 65.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 4.0)
fluid.addComponent("n-butane", 2.5)
fluid.addComponent("n-pentane", 1.5)
fluid.addComponent("n-hexane", 1.0)
fluid.addComponent("n-heptane", 5.0)
fluid.addComponent("n-octane", 4.0)
fluid.addComponent("n-nonane", 3.0)
fluid.addComponent("n-decane", 2.7)
fluid.addComponent("water", 1.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

# Create well stream at bottomhole conditions
Stream = jneqsim.process.equipment.stream.Stream
well_inflow = Stream("Well Inflow", fluid)
well_inflow.setFlowRate(50000.0, "kg/hr")
well_inflow.setTemperature(95.0, "C")
well_inflow.setPressure(200.0, "bara")

# Model wellbore as a vertical pipe
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
wellbore = PipeBeggsAndBrills("Wellbore", well_inflow)
wellbore.setPipeWallRoughness(2.5e-5)
wellbore.setLength(3000.0)           # 3000 m well depth
wellbore.setElevation(-3000.0)       # Vertical well (negative = upward flow)
wellbore.setDiameter(0.1016)         # 4-inch tubing

# Model flowline to separator
flowline = PipeBeggsAndBrills("Flowline", wellbore.getOutletStream())
flowline.setPipeWallRoughness(5.0e-5)
flowline.setLength(5000.0)           # 5 km flowline
flowline.setElevation(0.0)           # Horizontal
flowline.setDiameter(0.2032)         # 8-inch pipeline

# HP Separator
Separator = jneqsim.process.equipment.separator.Separator
hp_sep = Separator("HP Separator", flowline.getOutletStream())

# Build and run process
ProcessSystem = jneqsim.process.processmodel.ProcessSystem
process = ProcessSystem()
process.add(well_inflow)
process.add(wellbore)
process.add(flowline)
process.add(hp_sep)
process.run()

# Report results
print("=== Integrated Well-to-Separator Results ===")
print(f"Bottomhole pressure: {well_inflow.getPressure('bara'):.1f} bara")
print(f"Wellhead pressure: {wellbore.getOutletStream().getPressure('bara'):.1f} bara")
print(f"Wellhead temperature: {wellbore.getOutletStream().getTemperature('C'):.1f} C")
print(f"Separator inlet pressure: {flowline.getOutletStream().getPressure('bara'):.1f} bara")
print(f"Separator inlet temperature: {flowline.getOutletStream().getTemperature('C'):.1f} C")
print(f"Oil rate: {hp_sep.getLiquidOutStream().getFlowRate('m3/hr'):.1f} m3/hr")
print(f"Gas rate: {hp_sep.getGasOutStream().getFlowRate('MSm3/day'):.4f} MSm3/day")

4.9 System Analysis: NODAL Analysis

4.9.1 Concept

NODAL analysis (first described by Gilbert, 1954; formalized by Mach et al., 1979) is the framework for analyzing the integrated well-reservoir-facility system. The system is divided at a "node" — typically the bottomhole — and two performance curves are plotted:

1. Inflow Performance Relationship (IPR): The rate the reservoir can deliver as a function of bottomhole pressure
2. Vertical Flow Performance (VFP): The bottomhole pressure required to lift the fluid to the surface at each rate

The operating point is where the two curves intersect.

This illustrative incompressible limit holds wellhead plus hydrostatic pressure at 60 bara and friction proportional to rate squared divided by diameter to the fifth power. Larger diameter lowers the required BHP at a fixed rate and raises the intersection rate. Multiphase wells require the coupled model developed in Chapters 5–8.

4.9.2 Applications of NODAL Analysis

4.9.3 Multi-Well Optimization

When multiple wells produce into a common facility, the individual well operating points are coupled through the shared back-pressure. The total production is:

subject to:

• Each well's IPR constraint
• The gathering system pressure drop model
• The facility capacity constraint (separator capacity, compressor capacity)

This optimization problem — maximizing $q_{\text{total}}$ subject to constraints — is the core of production optimization, developed further in Chapter 19.

4.10 NeqSim Implementation: Reservoir Modeling

4.10.1 The SimpleReservoir Class

NeqSim provides the SimpleReservoir class for coupling reservoir performance with process simulation:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create reservoir fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 100.0, 250.0)
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 2.0)
fluid.addComponent("methane", 65.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 4.0)
fluid.addComponent("i-butane", 1.0)
fluid.addComponent("n-butane", 2.0)
fluid.addComponent("i-pentane", 0.8)
fluid.addComponent("n-pentane", 0.6)
fluid.addComponent("n-hexane", 1.0)
fluid.addComponent("n-heptane", 5.0)
fluid.addComponent("n-octane", 4.0)
fluid.addComponent("n-nonane", 3.0)
fluid.addComponent("n-decane", 2.1)
fluid.addComponent("water", 1.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

# Create reservoir
SimpleReservoir = jneqsim.process.equipment.reservoir.SimpleReservoir
reservoir = SimpleReservoir("Main Reservoir")
# Initial in-situ gas, oil and water pore volumes in m3 (illustrative).
reservoir.setReservoirFluid(fluid, 1.0e7, 2.0e7, 5.0e6)

# Add a production well
producer = reservoir.addOilProducer("Producer-1")
producer.setFlowRate(10000.0, "kg/hr")

4.10.2 IPR Curve Generation with NeqSim

Using NeqSim to generate IPR curves by solving the well-reservoir system at different bottomhole pressures:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Reservoir parameters
P_reservoir = 250.0  # bara
T_reservoir = 95.0   # C
PI = 15.0            # Sm3/d/bar (productivity index)
P_bubble = 180.0     # bara (bubble point)

# Calculate IPR using composite Vogel method
pressures_bhp = []
rates = []

for i in range(50):
    P_wf = 10.0 + i * 4.8  # 10 to 250 bara
    pressures_bhp.append(P_wf)

    if P_wf >= P_bubble:
        # Linear region (above bubble point)
        q = PI * (P_reservoir - P_wf)
    else:
        # Vogel region (below bubble point)
        q_at_pb = PI * (P_reservoir - P_bubble)
        q_vogel_max = q_at_pb + PI * P_bubble / 1.8
        q = q_at_pb + (q_vogel_max - q_at_pb) * (
            1.0 - 0.2 * (P_wf / P_bubble) - 0.8 * (P_wf / P_bubble) ** 2
        )
    rates.append(max(0.0, q))

# Print table
print(f"{'P_wf (bara)':>12} {'q_o (Sm3/d)':>12}")
print("-" * 26)
for p, q in zip(pressures_bhp[::5], rates[::5]):
    print(f"{p:12.1f} {q:12.1f}")

4.10.3 Decline Curve Implementation

import math

# Arps decline curve parameters
q_i = 5000.0     # Initial rate, Sm3/d
D_i = 0.001      # Initial decline rate, 1/day (about 30% per year)
b = 0.5           # Hyperbolic exponent

# Forecast for 10 years
time_days = [i * 30 for i in range(121)]  # Monthly steps for 10 years
rates = []
cum_production = []
cum = 0.0

for t in time_days:
    # Hyperbolic decline
    q = q_i / (1.0 + b * D_i * t) ** (1.0 / b)
    rates.append(q)

    if t > 0:
        dt = time_days[1]  # Step size
        cum += q * dt
    cum_production.append(cum)

# EUR calculation
q_el = 50.0  # Economic limit, Sm3/d
t_el = (1.0 / (b * D_i)) * ((q_i / q_el) ** b - 1.0)
print(f"Time to economic limit: {t_el / 365.25:.1f} years")

# Print annual summary
print(f"{'Year':>6} {'Rate (Sm3/d)':>14} {'Cum (MSm3)':>12} {'Annual Decline':>16}")
print("-" * 50)
for yr in range(11):
    idx = yr * 12
    if idx < len(rates):
        annual_decline = (1.0 - rates[idx] / rates[max(0, idx - 12)]) * 100 if yr > 0 else 0.0
        print(f"{yr:6d} {rates[idx]:14.0f} {cum_production[idx] / 1e6:12.3f} {annual_decline:15.1f}%")

The curves compare specified Arps forms and an assumed economic cut-off. They show rate, not cumulative production, and are not history-matched resource or reserve estimates.

4.10.4 Analytical Drive Models for Integrated Production

Alongside the compositional SimpleReservoir, NeqSim provides lightweight analytical drive models (package neqsim.process.fielddevelopment.integrated) that express reservoir pressure as a function of cumulative production. They share the ReservoirDrive interface and plug directly into the integrated reservoir-to-market model of Chapter 28, where thousands of pressure evaluations over a field life would make full compositional simulation prohibitively expensive:

Each model couples a WellDeliverabilityCurve (Chapter 5) to a declining reservoir pressure so that the integrated model can march a production profile forward in time:

integ = jneqsim.process.fielddevelopment.integrated
MaterialBalanceGasDrive = integ.MaterialBalanceGasDrive

initialPressureBara, gasInPlaceSm3, averageZ = 250.0, 1.0e9, 0.90
drive = MaterialBalanceGasDrive(initialPressureBara, gasInPlaceSm3, averageZ)
# pressure falls as cumulative production rises, following p/Z material balance

These drive objects are supplied to IntegratedProductionModel.addWell(name, drive, deliverabilityCurve) (Chapter 28) and to the ReservoirToMarketOptimizer, giving a closed reservoir-to-export model that runs in milliseconds per timestep.

4.10.5 Well Stream Setup

Setting up a well stream from the reservoir for process simulation:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create reservoir fluid at typical conditions
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 95.0, 300.0)
fluid.addComponent("nitrogen", 0.4)
fluid.addComponent("CO2", 1.8)
fluid.addComponent("methane", 68.0)
fluid.addComponent("ethane", 7.5)
fluid.addComponent("propane", 4.2)
fluid.addComponent("n-butane", 2.5)
fluid.addComponent("n-pentane", 1.5)
fluid.addComponent("n-hexane", 1.2)
fluid.addComponent("n-heptane", 3.5)
fluid.addComponent("n-octane", 3.0)
fluid.addComponent("n-nonane", 2.4)
fluid.addComponent("n-decane", 2.0)
fluid.addComponent("water", 2.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

# Create well stream at wellhead conditions
Stream = jneqsim.process.equipment.stream.Stream
well_stream = Stream("Well-1 Stream", fluid)
well_stream.setFlowRate(50000.0, "kg/hr")     # 50 t/hr
well_stream.setTemperature(75.0, "C")          # Wellhead temperature
well_stream.setPressure(70.0, "bara")           # Wellhead pressure
well_stream.run()

# Check the well stream properties
well_fluid = well_stream.getFluid()
print(f"Number of phases: {well_fluid.getNumberOfPhases()}")
print(f"Gas rate: {well_stream.getFlowRate('MSm3/day'):.3f} MSm3/day (total)")
print(f"Temperature: {well_stream.getTemperature('C'):.1f} C")
print(f"Pressure: {well_stream.getPressure('bara'):.1f} bara")

4.11 Gas Condensate Reservoirs

4.11.1 Retrograde Condensation Effect

Gas condensate reservoirs present a unique challenge: as pressure drops below the dew point, liquid condenses in the reservoir pore space. This condensate is typically immobile (trapped by capillary forces) and reduces the gas relative permeability, creating a "condensate bank" near the wellbore.

The productivity reduction depends on condensate saturation, relative permeability, capillary number and the near-well pressure/composition path; no universal 50–80% PI penalty follows from crossing the dew point. A fixed single-phase IPR does not resolve this mechanism. Accurate modeling requires compositional simulation with relative permeability effects.

4.11.2 Mitigation Strategies

• Pressure maintenance: Gas injection to keep reservoir pressure above the dew point
• Lean gas cycling: Inject lean gas (methane) to revaporize condensate
• Hydraulic fracturing: Create large-area flow paths that bypass the condensate bank
• Reduced drawdown: Limit well rate to minimize near-wellbore condensation

4.12 Reservoir Uncertainty and Its Impact on Optimization

4.12.1 Key Uncertain Parameters

4.12.2 Probabilistic Reserves

Reserves are classified probabilistically:

• P90 / low estimate: at least90% probability of equaling or exceeding the stated recoverable quantity; corresponds to 1P (Proved) when all reserves criteria are met.
• P50 / best estimate: at least50% exceedance probability; corresponds to 2P (Proved plus Probable), not Probable alone.
• P10 / high estimate: at least10% exceedance probability; corresponds to 3P (Proved plus Probable plus Possible), not Possible alone.

A simulated quantity distribution does not establish commerciality or reserves status. Apply the complete PRMS project and economic criteria.[13]

4.12.3 Sensitivity to Reservoir Pressure

As the reservoir depletes, the IPR shifts — the maximum rate decreases and the curve changes shape:

• Early life (high $P_r$): Large rate potential; optimization focuses on facility capacity
• Mid life ($P_r$ near $P_b$): Gas-oil ratio increases; compression becomes a constraint
• Late life (low $P_r$): Rate is reservoir-limited; artificial lift becomes necessary

4.13 Multi-Well and Multi-Reservoir Systems

4.13.1 Commingled Production

When multiple reservoir zones produce into a common wellbore:

subject to the constraint that all zones share the same bottomhole pressure $P_{wf}$.

4.13.2 Well Allocation

In a multi-well system with shared facilities, production must be allocated to satisfy:

• Individual well constraints (maximum rate, minimum BHP, maximum GOR, maximum water cut)
• Facility constraints (separator capacity, compressor capacity, water handling capacity)
• Export constraints (pipeline capacity, specification limits)

This allocation problem is the foundation of the short-term production optimization discussed in Chapter 19.

Reproduced Calculation Results

These examples use the stated fluid recipes and operating assumptions. Curves represent NeqSim calculations unless a caption identifies an analytical illustration, assumed equipment map or synthetic data.

PI Model (linear): bottomhole flowing pressure, pwf spans 0–300 bara across the plotted cases. Vogel Model (two-phase): bottomhole flowing pressure, pwf spans 0–300 bara across the plotted cases.

An IPR relates liquid or gas deliverability to drawdown; curvature reflects the selected inflow law and reservoir phase behavior. A high zero-backpressure intercept does not establish a producible operating rate because tubing and facility backpressure still have to be satisfied. Calibrate the appropriate inflow law against well tests and intersect it with a verified tubing relationship on the same rate and pressure basis.

Molar GOR spans 6.001–62.51 gas moles / oil moles across the plotted cases. Gaps retain undefined phase quantities or hydraulic states that fail the stated operating boundary; they are not interpolated.

As pressure falls, dissolved light components evolve into a gas phase and change the gas and liquid volumes. Gas-handling demand can increase relative to oil rate during depletion; a static initial GOR can therefore understate future bottlenecks. Distinguish fixed-composition flashes from depletion with gas withdrawal, and couple the appropriate PVT path to separator and compression capacity.

Across the assumed 0–3000 m depth interval, density spans 159.4–323.9 kg/m³ and temperature spans 50.0–140.0 °C.

Depth changes both hydrostatic pressure and temperature; these effects jointly alter fluid density. A constant-density hydrostatic approximation can bias bottom-hole pressure, particularly for compressible or near-saturation fluids. Integrate density over the pressure–temperature path and compare with measured gradients before using the profile for well control or inflow calibration.

$k_{rw}$ (n_w=3.0): relative permeability, kr spans 0–0.4 across the plotted cases. $k_{ro}$ (n_o=2.5): relative permeability, kr spans 0–1 across the plotted cases.

The Corey curves allocate effective flow capacity between phases according to saturation, residual saturation and fitted exponents. Their endpoints and curvature can control water breakthrough and mobility predictions; the plotted curves are constitutive assumptions rather than EOS results. Replace the illustrative parameters with rock-type-specific core data and examine hysteresis and endpoint uncertainty in reservoir forecasts.

Selected numerical ranges from the plotted cases:

Ranges describe the sampled cases; they are not independent validation tolerances.

4.14 Summary

Key points from this chapter:

• Darcy's law and the radial flow equation provide the physical basis for well productivity. The productivity index $J$ encapsulates rock properties, fluid properties, and completion quality.
• Well testing (drawdown and buildup analysis, Horner plots, pressure derivatives) provides in-situ measurements of $kh$, skin, and average reservoir pressure.
• IPR curves (linear for undersaturated oil, Vogel for saturated oil, Fetkovich and back-pressure/LIT for gas) define the rate that the reservoir can deliver at each bottomhole pressure. Future IPR curves shift as the reservoir depletes.
• Reservoir drive mechanisms (solution gas, gas cap, water drive, gravity drainage) determine the rate of pressure decline, ultimate recovery, and GOR/water cut evolution.
• Material balance (Havlena-Odeh straight-line method, drive index analysis, water influx models) links cumulative production to pressure decline and provides estimates of original hydrocarbons in place.
• Decline curve analysis (Arps equations) provides a simple but effective method for forecasting production and estimating EUR.
• Reservoir simulation coupling through VFP tables provides the interface between detailed reservoir models and process simulation.
• NODAL analysis provides the framework for integrating reservoir performance with well and facility performance.
• NeqSim models the reservoir through SimpleReservoir, wellbore through PipeBeggsAndBrills, and enables integrated reservoir-to-separator modeling.

Verification of the worked examples

The decline cumulative-volume derivative and pressure/unit relations are checked independently. The reservoir objects demonstrate a lumped inventory/API boundary; they do not validate spatial reservoir sweep or commercial reserves. The P/Z interpretation is restricted to its closed, isothermal volumetric assumptions, and the Horner extrapolation is distinguished from average pressure in a bounded depleted reservoir.[13]

The calculation and literal-code records are in verification/scientific_revision/ch04_manuscript_physics.json; the chapter scope and code hashes are indexed in foundations_review.json.

Exercises

1. Exercise 4.1: A vertical well has the following properties: $k = 50$ mD, $h = 20$ m, $r_e = 500$ m, $r_w = 0.108$ m, $S = 2$, $B_o = 1.25$, $\mu_o = 1.5$ cP. Calculate the productivity index in Sm³/d/bar. What is the maximum oil rate if $P_r = 300$ bara?

1. Exercise 4.2: Using Vogel's method, construct an IPR curve for a well with $P_r = 250$ bara, $P_b = 180$ bara, and $J = 20$ Sm³/d/bar (above bubble point). Plot $P_{wf}$ vs. $q_o$ from 0 to 250 bara.

1. Exercise 4.3: A gas well has the following multi-rate test data:

If $P_r = 250$ bara, determine the back-pressure equation coefficients $C$ and $n$.

1. Exercise 4.4: A volumetric gas reservoir has $P_i = 300$ bara, $T = 100$°C, and OGIP = 50 GSm³. Using NeqSim to calculate $Z$ at each pressure, construct a $P/Z$ vs. $G_p$ plot.

1. Exercise 4.5: A well produces 3000 Sm³/d initially with $D_i = 0.0008$/day and $b = 0.5$. Calculate the cumulative production after 5 years and the EUR to an economic limit of 50 Sm³/d.

1. Exercise 4.6: Set up a NODAL analysis in NeqSim for a single well flowing into a separator at 40 bara. Model the tubing using PipeBeggsAndBrills. Find the operating point.

1. Exercise 4.7: A pressure buildup test on a well that was flowing at 500 Sm³/d for 100 hours shows a Horner slope of $m = 15$ bar/cycle. If $B_o = 1.3$, $\mu_o = 2.0$ cP, calculate $kh$. If the pressure at 1-hour shut-in is 185 bara and $P_{wf,\text{last}} = 170$ bara, estimate the skin factor.

1. Exercise 4.8: For a solution gas drive reservoir with $N = 50 \times 10^6$ Sm³ OOIP, use the Havlena-Odeh method to verify the OOIP given production data. Plot $F/E_o$ vs. cumulative production and check that the intercept matches $N$.

1. Exercise 4.9: Build an integrated reservoir-wellbore-separator model in NeqSim. Use PipeBeggsAndBrills for a 3000 m vertical well and 5 km flowline. Compare the wellhead pressure and separator gas rate for tubing sizes of 3.5-inch and 4.5-inch.

1. Exercise 4.10 (Advanced): For a system of three wells producing into a common manifold at 30 bara, each with different IPR parameters, use NeqSim to find the individual well rates that maximize total oil production subject to a total gas handling constraint of 2 MSm³/day.

1. Vogel, J. V. (1968). Inflow performance relationships for solution-gas drive wells. Journal of Petroleum Technology, 20(1), 83–92.
2. Rawlins, E. L., & Schellhardt, M. A. (1935). Backpressure Data on Natural Gas Wells and Their Application to Production Practices. Monograph 7, USBM.
3. Arps, J. J. (1945). Analysis of decline curves. Transactions of the AIME, 160(1), 228–247.
4. Joshi, S. D. (1988). Augmentation of well productivity with slant and horizontal wells. Journal of Petroleum Technology, 40(6), 729–739.
5. Schilthuis, R. J. (1936). Active oil and reservoir energy. Transactions of the AIME, 118(1), 33–52.
6. Gilbert, W. E. (1954). Flowing and gas-lift well performance. API Drilling and Production Practice, 126–157.
7. Mach, J., Proano, E., & Brown, K. E. (1979). A nodal approach for applying systems analysis to the flowing and artificial lift oil or gas well. Paper SPE 8025.
8. Ahmed, T. (2016). Reservoir Engineering Handbook (5th ed.). Gulf Professional Publishing.
9. Dake, L. P. (1978). Fundamentals of Reservoir Engineering. Elsevier.
10. Economides, M. J., Hill, A. D., Ehlig-Economides, C., & Zhu, D. (2013). Petroleum Production Systems (2nd ed.). Prentice Hall.
11. Havlena, D., & Odeh, A. S. (1963). The material balance as an equation of a straight line. Journal of Petroleum Technology, 15(8), 896–900.
12. Fetkovich, M. J. (1973). The isochronal testing of oil wells. Paper SPE 4529, 48th Annual Fall Meeting.
13. Bourdet, D., Whittle, T. M., Douglas, A. A., & Pirard, Y. M. (1983). A new set of type curves simplifies well test analysis. World Oil, 196(6), 95–106.
14. van Everdingen, A. F., & Hurst, W. (1949). The application of the Laplace transformation to flow problems in reservoirs. Transactions of the AIME, 186, 305–324.
15. Carter, R. D., & Tracy, G. W. (1960). An improved method for calculating water influx. Transactions of the AIME, 219, 415–417.

Running the examples. Start the source-workspace Python session described in Chapter 1, then run this chapter's Python blocks in reading order. Java blocks form a separate sequence using the same NeqSim build; carry forward objects from preceding Java blocks. The release execution records are in verification/; a successful run establishes API compatibility, while physical validation also requires the checks discussed in the text.

Learning Objectives

After reading this chapter, the reader will be able to:

1. Calculate pressure traverses in vertical and deviated wellbores using multiphase flow correlations
2. Describe and apply the Beggs and Brill multiphase flow correlation for well tubing performance
3. Analyze wellhead choke performance under critical and subcritical flow conditions
4. Design continuous gas lift systems including valve spacing and injection rate optimization
5. Understand the fundamentals of ESP and rod pump artificial lift methods
6. Interpret well test data (buildup, drawdown) and extract reservoir parameters
7. Generate VFP tables using NeqSim for integration with reservoir simulators and production optimization models
8. Select tubing sizes based on rate, pressure, and flow regime analysis

5.1 Introduction

The well is the conduit between the reservoir and the surface facility. Its performance — specifically, the pressure loss from bottomhole to wellhead — determines how much of the reservoir's delivery potential can actually be captured at the surface. A well with excessive tubing friction, inadequate tubing size, or severe liquid loading may deliver only a fraction of the rate that the IPR would allow.

This chapter develops the theory and NeqSim tools for well performance analysis. We begin with multiphase flow in tubing — the core calculation that determines the tubing pressure profile — then cover choke performance, artificial lift (with emphasis on gas lift), well testing, and VFP table generation.

The link to Chapter 4 is direct: the well's VFP curve intersects the IPR curve to give the operating point (NODAL analysis). The link to Chapter 6 is equally direct: the wellhead pressure must overcome the flowline, riser, and topside back-pressure.

5.2 Multiphase Flow in Wellbores

5.2.1 The Challenge of Multiphase Flow

Flow in production wells is almost always multiphase — oil, gas, and often water flow simultaneously through the tubing. The simultaneous presence of multiple phases creates several complexities not found in single-phase flow:

• Slippage: Gas travels faster than liquid due to buoyancy, creating a difference between the in-situ gas volume fraction and the flowing gas fraction (holdup)
• Flow patterns: The phases arrange themselves in different patterns (bubble, slug, churn, annular) depending on rates and properties
• Changing properties: As pressure decreases up the well, gas evolves from solution, changing the flow rates, densities, and viscosities continuously
• Hydrostatic head: The liquid holdup determines the effective mixture density, which controls the gravity pressure drop — the dominant pressure loss in most wells

5.2.2 Pressure Gradient Components

The total pressure gradient in vertical multiphase flow has three components:

where:

• $s$ = distance along the upward flow path [m]; $\theta>0$ uphill; $f$ is the Darcy friction factor. The displayed expression is a homogeneous-mixture balance. Slip correlations use holdup density for gravity and their own friction/acceleration closures, not one interchangeable density for every term.
• $\rho_m$ = homogeneous-mixture density [kg/m³]
• $g$ = gravitational acceleration [m/s²]
• $\theta$ = pipe inclination from horizontal [degrees]
• $f$ = friction factor [-]
• $v_m$ = mixture velocity [m/s]
• $d$ = pipe internal diameter [m]

In a typical production well:

The dominance of the gravity term means that the liquid holdup — which determines $\rho_m$ — is the most important parameter to predict accurately.

5.2.3 Flow Pattern Maps

The flow pattern in vertical upward flow depends on the superficial velocities of gas and liquid:

The Taitel-Dukler (1980) and Barnea (1987) flow pattern maps provide mechanistic criteria for predicting the transitions between patterns.

5.2.4 Liquid Holdup

The liquid holdup $H_L$ is the fraction of the pipe cross-section occupied by liquid:

The mixture density is then:

The holdup differs from the input liquid fraction (no-slip holdup) because of gas-liquid slippage:

In vertical upward flow, $H_L > \lambda_L$ because gas rises faster than liquid.

5.3 The Beggs and Brill Correlation

5.3.1 Overview

The Beggs and Brill (1973) correlation is one of the most widely used methods for multiphase pressure drop in pipes. It was developed from laboratory data in pipes of 1-inch and 1.5-inch diameter at various inclinations, and it handles all pipe angles from horizontal to vertical.

The correlation follows these steps:

1. Calculate the Froude number and input liquid fraction
2. Determine the flow pattern (segregated, intermittent, distributed)
3. Calculate the liquid holdup using the flow pattern-specific correlation
4. Correct the holdup for pipe inclination
5. Calculate the friction factor with a multiphase correction
6. Sum the gravity and friction pressure gradients

5.3.2 Flow Pattern Determination

The Beggs and Brill flow pattern boundaries depend on two parameters:

where $v_{sL}$ is the superficial liquid velocity, $v_{sg}$ is the superficial gas velocity, and $v_m = v_{sL} + v_{sg}$.

The transition boundaries $L_1$, $L_2$, $L_3$, and $L_4$ are calculated as:

5.3.3 Holdup Calculation

For each flow pattern, the horizontal holdup $H_L(0)$ is calculated from:

where the constants $a$, $b$, $c$ depend on the flow pattern:

The inclination correction factor $\psi$ adjusts for non-horizontal pipe:

where $C$ depends on the flow pattern, liquid velocity number, and Froude number.

5.3.4 Friction Factor

The two-phase friction factor uses the Moody friction factor corrected for multiphase effects:

where $f_n$ is the no-slip friction factor (from the Moody chart or Colebrook equation) and $s$ is an empirical correction.

5.4 NeqSim Implementation: Well Tubing Performance

5.4.1 PipeBeggsAndBrills for Well Flow

NeqSim implements the Beggs and Brill correlation in the PipeBeggsAndBrills class, which can model both horizontal pipelines and vertical/deviated wells:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create reservoir fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 90.0, 200.0)
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 2.0)
fluid.addComponent("methane", 70.0)
fluid.addComponent("ethane", 7.5)
fluid.addComponent("propane", 4.0)
fluid.addComponent("i-butane", 1.0)
fluid.addComponent("n-butane", 2.0)
fluid.addComponent("i-pentane", 0.5)
fluid.addComponent("n-pentane", 0.4)
fluid.addComponent("n-hexane", 0.6)
fluid.addComponent("n-heptane", 3.5)
fluid.addComponent("n-octane", 3.0)
fluid.addComponent("n-nonane", 2.5)
fluid.addComponent("n-decane", 2.0)
fluid.setMixingRule("classic")

# Set up well stream at bottomhole conditions
Stream = jneqsim.process.equipment.stream.Stream
bh_stream = Stream("Bottomhole Stream", fluid)
bh_stream.setFlowRate(20000.0, "kg/hr")  # feasible teaching base; sweep higher rates separately
bh_stream.setTemperature(90.0, "C")
bh_stream.setPressure(200.0, "bara")

# Create vertical well tubing (Beggs and Brill)
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
well_tubing = PipeBeggsAndBrills("Production Tubing", bh_stream)
well_tubing.setPipeWallRoughness(2.5e-5)             # m (smooth tubing)
well_tubing.setLength(3000.0)                         # m (measured depth)
well_tubing.setElevation(3000.0)                     # m (positive = outlet above inlet)
well_tubing.setDiameter(0.1016)                       # m (4-inch tubing)
well_tubing.setNumberOfIncrements(50)                 # Calculation segments

# Build and run process
ProcessSystem = jneqsim.process.processmodel.ProcessSystem
process = ProcessSystem()
process.add(bh_stream)
process.add(well_tubing)
process.run()

# Extract results
outlet = well_tubing.getOutletStream()
print(f"Bottomhole pressure: {bh_stream.getPressure('bara'):.1f} bara")
print(f"Wellhead pressure:   {outlet.getPressure('bara'):.1f} bara")
print(f"Pressure drop:       {bh_stream.getPressure('bara') - outlet.getPressure('bara'):.1f} bar")
print(f"Bottomhole temp:     {bh_stream.getTemperature('C'):.1f} C")
print(f"Wellhead temp:       {outlet.getTemperature('C'):.1f} C")

5.4.2 Pressure Traverse Calculation

The following fixed-bottomhole-pressure sweep reports the wellhead pressure and total pressure drop at each imposed rate. These are endpoint results, not depth profiles. The internal increment data can be examined separately to construct pressure traverses; rates for which the hydraulic calculation fails remain rejected.

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 90.0, 250.0)
fluid.addComponent("methane", 75.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 4.0)
fluid.addComponent("n-butane", 2.5)
fluid.addComponent("n-pentane", 1.0)
fluid.addComponent("n-heptane", 5.0)
fluid.addComponent("n-octane", 3.0)
fluid.addComponent("n-decane", 1.5)
fluid.setMixingRule("classic")

# Generate VFP data for different flow rates
Stream = jneqsim.process.equipment.stream.Stream
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

flow_rates = [20000.0, 40000.0, 60000.0, 80000.0, 100000.0]  # kg/hr

print(f"{'Flow Rate (kg/hr)':>18} {'BHP (bara)':>12} {'WHP (bara)':>12} {'dP (bar)':>10}")
print("-" * 54)

for rate in flow_rates:
    stream = Stream("BH Stream", fluid.clone())
    stream.setFlowRate(rate, "kg/hr")
    stream.setTemperature(90.0, "C")
    stream.setPressure(250.0, "bara")

    tubing = PipeBeggsAndBrills("Tubing", stream)
    tubing.setPipeWallRoughness(2.5e-5)
    tubing.setLength(2500.0)
    tubing.setElevation(2500.0)
    tubing.setDiameter(0.1016)
    tubing.setNumberOfIncrements(40)

    process = ProcessSystem()
    process.add(stream)
    process.add(tubing)
    try:
        process.run()
    except Exception as error:
        print("Infeasible fixed-rate tubing case:", str(error).splitlines()[0])
        continue

    whp = tubing.getOutletStream().getPressure("bara")
    dp = 250.0 - whp
    print(f"{rate:18.0f} {250.0:12.1f} {whp:12.1f} {dp:10.1f}")

5.4.3 Deviated Well Modeling

For deviated wells, the elevation is less than the measured depth. The relationship is:

where $\Delta L_i$ is the measured depth increment and $\alpha_i$ is the local inclination from vertical. In NeqSim:

# Deviated well: 3500 m MD but only 2800 m TVD
well_tubing = PipeBeggsAndBrills("Deviated Tubing", bh_stream)
well_tubing.setLength(3500.0)          # Measured depth
well_tubing.setElevation(2800.0)      # True vertical depth (positive = upward)
well_tubing.setDiameter(0.1016)
well_tubing.setPipeWallRoughness(2.5e-5)
well_tubing.setNumberOfIncrements(50)

5.5 Wellhead Choke Performance

5.5.1 Purpose of Wellhead Chokes

Wellhead chokes serve multiple purposes:

• Rate control: Regulate the production rate to meet allocation targets
• Back-pressure management: Protect downstream equipment from high pressures
• Sand control: Limit drawdown to prevent sand production
• Slugging prevention: Maintain stable flow by keeping downstream pressure stable
• Critical flow: Prevent downstream pressure fluctuations from propagating to the reservoir

5.5.2 Critical vs. Subcritical Flow

When the pressure ratio across a choke reaches a critical value, the flow velocity at the choke throat reaches the local speed of sound, and the flow becomes critical (choked). Under critical flow:

• The flow rate depends only on upstream conditions (pressure, temperature, composition)
• Downstream pressure fluctuations cannot propagate upstream
• The critical pressure ratio is approximately:

For natural gas with $\gamma \approx 1.3$, the critical pressure ratio is approximately 0.55.

5.5.3 Choke Flow Equations

For single-phase gas, the flow through a choke follows:

Here $q_g$ is actual volumetric flow at the upstream state in m³/s, not standard gas volume; $P_1$ is absolute Pa, $\rho_1$ is kg/m³ and $A$ is m². This ideal-gas, constant-$\gamma$ nozzle relation applies above the critical downstream/upstream pressure ratio. Below it, evaluate the bracket at the critical ratio; mass flow then remains capacity-limited. Real control valves require tested flow coefficients and expansion/recovery factors.

For multiphase flow through chokes, the Sachdeva or Perkins correlations are commonly used. NeqSim provides choke modeling through the valve classes:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Choke valve modeling
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 70.0, 80.0)
fluid.addComponent("methane", 80.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 4.0)
fluid.addComponent("n-butane", 2.0)
fluid.addComponent("n-heptane", 3.0)
fluid.addComponent("water", 3.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

Stream = jneqsim.process.equipment.stream.Stream
ThrottlingValve = jneqsim.process.equipment.valve.ThrottlingValve
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

# Well stream upstream of choke
well_stream = Stream("Wellhead Stream", fluid)
well_stream.setFlowRate(50000.0, "kg/hr")
well_stream.setTemperature(70.0, "C")
well_stream.setPressure(80.0, "bara")

# Wellhead choke
choke = ThrottlingValve("WH Choke", well_stream)
choke.setOutletPressure(40.0, "bara")

# Build and run
process = ProcessSystem()
process.add(well_stream)
process.add(choke)
process.run()

# Results
outlet = choke.getOutletStream()
print(f"Upstream pressure:  {well_stream.getPressure('bara'):.1f} bara")
print(f"Downstream pressure: {outlet.getPressure('bara'):.1f} bara")
print(f"Pressure ratio:     {outlet.getPressure('bara') / well_stream.getPressure('bara'):.3f}")
print(f"Downstream temperature: {outlet.getTemperature('C'):.1f} C")
print(f"Temperature drop (JT): {well_stream.getTemperature('C') - outlet.getTemperature('C'):.1f} C")

5.5.4 Joule-Thomson Cooling

An adiabatic valve with negligible net kinetic/potential-energy change is isenthalpic. A positive Joule–Thomson coefficient gives cooling during pressure reduction; the coefficient can change sign. The following local linear estimate is suitable only for a small pressure interval; use a PH flash over a large letdown:

where $\mu_{JT}$ is the Joule-Thomson coefficient [°C/bar]. For natural gas at typical wellhead conditions, $\mu_{JT}$ is approximately 0.3–0.5 °C/bar. This cooling is important for:

• Hydrate risk: The temperature downstream of the choke may fall below the hydrate formation temperature
• Wax deposition: Cooling may promote wax formation
• Phase behavior: Additional gas may condense from the cooled fluid

5.6 Gas Lift

5.6.1 Gas Lift Fundamentals

Gas lift is the most common artificial lift method in offshore production. It works by injecting gas into the tubing at depth to reduce the hydrostatic head of the fluid column, thereby lowering the flowing bottomhole pressure and increasing the production rate.

The injected gas reduces the mixture density in the tubing:

As the gas injection rate increases, $H_L$ decreases, reducing $\rho_m$ and the hydrostatic pressure drop. However, the friction pressure drop increases with the total gas rate. There is an optimum gas injection rate that minimizes the total wellhead pressure required (or maximizes the well production rate).

5.6.2 Gas Lift Performance Curve

The gas lift performance curve shows the well's production rate as a function of gas injection rate. It has a characteristic shape:

1. No gas lift: The natural flow rate (may be zero if the well cannot flow naturally)
2. Increasing injection: Rate increases as hydrostatic head is reduced
3. Optimum injection: Maximum production rate — further injection increases friction more than it reduces hydrostatic pressure
4. Over-injection: Rate decreases due to excessive friction (and compression costs increase)

The optimum injection rate is typically determined by:

where $C_{\text{gas}}$ is the cost of injection gas and $C_{\text{oil}}$ is the value of incremental oil. For a smooth concave response with positive marginal injection cost and an interior optimum, the economic injection rate is below the rate-maximizing injection. Bound-active wells or nonconcave responses require explicit constrained optimization; equal slopes apply only to unconstrained wells.

The declared response has diminishing incremental oil yield. With the stated cost-equivalent slope of 5 m³ oil per thousand standard m³ lift gas, the stationary net-benefit injection is 77.71 thousand standard m³/d. This synthetic trade-off is checked by differentiation; it is not a calibrated lift curve, a current market-price forecast or a native gas-lift allocation result.

5.6.3 Gas Lift Valve Spacing

Gas lift valves are installed at intervals along the tubing string. Their spacing determines the:

• Unloading sequence: Valves are opened sequentially from top to bottom during startup to displace the kill fluid
• Operating depth: The deepest valve (operating valve) is the point where gas enters the tubing during normal production
• Injection pressure: Higher injection depth requires higher surface injection pressure

The spacing calculation uses the available injection pressure and the pressure gradient of the gas lift gas and the production fluid. A typical spacing procedure:

1. Plot the static fluid gradient (tubing filled with kill fluid)
2. Plot the injection gas gradient from the surface
3. The first valve is placed where the gas gradient intersects the fluid gradient minus a margin
4. Each subsequent valve is placed based on the operating envelope and a pressure drop margin (typically 20–35 kPa per valve)

Table 5.1 shows typical gas lift design parameters:

5.6.4 Gas Lift Modeling in NeqSim

NeqSim models gas lift by mixing the lift gas with the production stream at the injection point:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Production fluid (low pressure, needs gas lift)
prod_fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 80.0, 120.0)
prod_fluid.addComponent("nitrogen", 0.3)
prod_fluid.addComponent("CO2", 1.5)
prod_fluid.addComponent("methane", 50.0)
prod_fluid.addComponent("ethane", 6.0)
prod_fluid.addComponent("propane", 4.0)
prod_fluid.addComponent("n-butane", 3.0)
prod_fluid.addComponent("n-pentane", 2.0)
prod_fluid.addComponent("n-hexane", 2.5)
prod_fluid.addComponent("n-heptane", 5.0)
prod_fluid.addComponent("n-octane", 5.0)
prod_fluid.addComponent("n-nonane", 4.0)
prod_fluid.addComponent("n-decane", 4.7)
prod_fluid.addComponent("water", 12.0)
prod_fluid.setMixingRule("classic")
prod_fluid.setMultiPhaseCheck(True)

# Gas lift gas (export gas recycled for lift)
lift_gas_fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 40.0, 120.0)
lift_gas_fluid.addComponent("nitrogen", 1.0)
lift_gas_fluid.addComponent("CO2", 2.0)
lift_gas_fluid.addComponent("methane", 90.0)
lift_gas_fluid.addComponent("ethane", 5.0)
lift_gas_fluid.addComponent("propane", 2.0)
lift_gas_fluid.setMixingRule("classic")

Stream = jneqsim.process.equipment.stream.Stream
Mixer = jneqsim.process.equipment.mixer.Mixer
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

# Production stream at gas lift injection point (downhole)
prod_stream = Stream("Production Stream", prod_fluid)
prod_stream.setFlowRate(15000.0, "kg/hr")  # conservative gas-lift teaching base
prod_stream.setTemperature(80.0, "C")
prod_stream.setPressure(120.0, "bara")

# Gas lift injection stream
lift_stream = Stream("Gas Lift", lift_gas_fluid)
lift_stream.setFlowRate(5000.0, "kg/hr")       # ~50,000 Sm3/d
lift_stream.setTemperature(50.0, "C")
lift_stream.setPressure(120.0, "bara")

# Mix at injection point
mixer = Mixer("GL Injection Point")
mixer.addStream(prod_stream)
mixer.addStream(lift_stream)

# Tubing above injection point
tubing = PipeBeggsAndBrills("Upper Tubing", mixer.getOutletStream())
tubing.setPipeWallRoughness(2.5e-5)
tubing.setLength(2000.0)
tubing.setElevation(2000.0)
tubing.setDiameter(0.1016)
tubing.setNumberOfIncrements(40)

# Build and run
process = ProcessSystem()
process.add(prod_stream)
process.add(lift_stream)
process.add(mixer)
process.add(tubing)
process.run()

# Results
whp = tubing.getOutletStream().getPressure("bara")
wht = tubing.getOutletStream().getTemperature("C")
print(f"Wellhead pressure: {whp:.1f} bara")
print(f"Wellhead temperature: {wht:.1f} C")
print(f"Total flow rate: {tubing.getOutletStream().getFlowRate('kg/hr'):.0f} kg/hr")

5.6.5 Gas Lift Optimization

The gas lift optimization problem involves allocating a limited gas supply among multiple wells to maximize total oil production:

subject to:

This is solved by equalizing the marginal oil gain per unit of injected gas across all wells:

This equal-slope criterion is a classic optimization result that ensures no gas can be reallocated from one well to another to increase total production.

5.7 Electric Submersible Pumps (ESP)

5.7.1 ESP Fundamentals

An ESP is a multistage centrifugal pump installed downhole, powered by an electric motor connected to the surface by an armored cable. Key characteristics:

5.7.2 ESP Performance

The ESP pump performance is described by three curves at constant speed:

1. Head-capacity (H-Q): Head developed vs. flow rate — decreasing curve
2. Efficiency (η-Q): Pump efficiency vs. flow rate — parabolic, with peak at best efficiency point (BEP)
3. Power (P-Q): Power consumption vs. flow rate — generally increasing

The ESP is selected to operate near its BEP for maximum efficiency and minimum wear.

5.7.3 ESP Sizing

The required number of stages is:

where $\Delta P_{\text{required}}$ is the total dynamic head including:

• Friction loss in tubing
• Hydrostatic head from pump depth to wellhead
• Wellhead pressure requirement
• Minus the flowing bottomhole pressure (from IPR)

5.7.4 ESP Considerations for Production Optimization

• At fixed speed an ESP operates at the intersection of its head–flow curve with the system curve; changing backpressure changes rate. A variable-speed drive provides an additional control variable
• Gas slugging can cause ESP shutdowns (gas lock)
• Power consumption is a significant operating cost (especially offshore)
• Workover to replace a failed ESP is expensive — run life prediction is critical

5.8 Rod Pump (Sucker Rod Pump)

5.8.1 Fundamentals

The sucker rod pump is the most common artificial lift method for onshore oil wells. It consists of a surface pumping unit (beam pump), a string of sucker rods, and a downhole positive displacement pump.

Key characteristics:

5.8.2 Rod Pump Performance

The theoretical pump displacement is:

where:

• $d_p$ = plunger diameter [m]
• $S_p$ = effective plunger stroke length [m]
• $N$ = pumping speed [strokes/min]
• 1440 converts min to day

The actual rate is less than theoretical due to gas interference, fluid slippage, and rod stretch:

where $\eta_{\text{vol}}$ is the volumetric efficiency (typically 50–90%).

5.8.3 NeqSim Implementation: SuckerRodPump

NeqSim models the beam pump directly with the SuckerRodPump equipment class (package neqsim.process.equipment.pump). The geometry and operating speed are set with fluent setters and the class returns both theoretical and effective displacement plus the polished-rod load used for surface-unit sizing:

SuckerRodPump = jneqsim.process.equipment.pump.SuckerRodPump

rodpump = SuckerRodPump("rod pump", well_stream)
rodpump.setPlungerDiameter(0.0381)          # 1.5 in plunger [m]
rodpump.setStrokeLength(1.68)               # effective stroke [m]
rodpump.setStrokesPerMinute(8.0)
rodpump.setVolumetricEfficiency(0.80)
rodpump.setPumpDepth(1800.0)                # [m]
rodpump.setFluidDensity(850.0)              # [kg/m3]
rodpump.setRodWeightPerLength(35.0 / 9.80665)  # kg/m from assumed buoyant 35 N/m
rodpump.setDischargePressure(100.0)         # bara, above this 80 bara inlet
rodpump.run()

print("Theoretical:", rodpump.getTheoreticalDisplacement("m3/day"))
print("Actual:", rodpump.getActualDisplacement("m3/day"))
print("Polished rod load:", rodpump.getPolishedRodLoad(), "N")
import math
displacement_m3_day = math.pi * 0.0381**2 / 4.0 * 1.68 * 8.0 * 1440.0
assert abs(rodpump.getTheoreticalDisplacement("m3/day") / displacement_m3_day - 1.0) < 1e-12
assert abs(rodpump.getActualDisplacement("m3/day") / displacement_m3_day - 0.80) < 1e-12
assert rodpump.getPolishedRodLoad() > 0.0

getTheoreticalDisplacement uses actual displacement volume, not a standard-state fluid conversion. The accepted unit strings are m3/sec and m3/day; an unsupported Sm3/day string silently returned m3/s in this source version. The displacement and rod-load checks are geometry/load screening; this one-stream pump object does not establish a complete reservoir-to-surface lift solution. The method evaluates the swept-volume formula above, getActualDisplacement applies the volumetric efficiency, and getPolishedRodLoad combines fluid load and rod weight for the peak surface load.

5.8.4 Hydraulic Jet Pumps in NeqSim

For deviated or high-temperature wells where rod pumps and ESPs are unsuitable, the hydraulic JetPump class (package neqsim.process.equipment.pump) models a nozzle–throat–diffuser ejector driven by a high-pressure power fluid. It is characterised by the nozzle-to-throat area ratio and the power-fluid pressure:

JetPump = jneqsim.process.equipment.pump.JetPump

jet = JetPump("jet pump", prod_stream)
jet.setAreaRatio(0.30)                       # nozzle area / throat area
jet.setPowerFluidPressure(250.0)             # [bara]
jet.setOperatingFlowRatio(0.8)               # produced / power-fluid flow
jet.setPowerFluidDensity(1000.0)             # [kg/m3]
jet.run()

print("Head ratio:", jet.getHeadRatio())
print("Discharge pressure:", jet.getDischargePressure(), "bara")
print("Efficiency:", jet.getEfficiency())
print("Produced rate (actual m3/day):", jet.getProducedRate("m3/day"))
assert prod_stream.getPressure("bara") < jet.getDischargePressure() < 250.0
assert 0.0 < jet.getEfficiency() < 1.0
assert jet.getProducedRate("m3/day") > 0.0

The dimensionless getHeadRatio() (also available pointwise via headRatioAt(M)) and getEfficiency() describe the classic jet-pump performance envelope, while getDischargePressure() and getProducedRate() give the operating point for the surrounding network.

5.9 Well Testing

5.9.1 Pressure Drawdown Test

A drawdown test measures the pressure response when a well is opened at a constant rate after being shut in. The semi-log analysis for radial flow gives:

A plot of $P_{wf}$ vs. $\log t$ gives a straight line with slope:

from which $kh$ (permeability-thickness product) is determined.

5.9.2 Pressure Buildup Test (Horner Analysis)

A buildup test measures the pressure recovery after shutting in a producing well. The Horner analysis plots $P_{ws}$ vs. $\log[(t_p + \Delta t)/\Delta t]$:

where $t_p$ is the producing time before shut-in and $\Delta t$ is the shut-in time. The straight line portion gives:

• Slope $m$: Yields $kh$ and hence permeability
• Intercept: Yields reservoir pressure $P^*$ (extrapolated to infinite shut-in time)
• Skin factor: From the pressure at 1 hour on the straight line:

5.9.3 Drill Stem Test (DST)

A DST is performed during drilling to evaluate the productivity of a formation. It consists of:

1. Opening the well for a flow period
2. Shutting in for a buildup period
3. Opening again for a second flow period
4. Final shut-in and buildup

The multiple flow and shut-in periods provide redundant data for reservoir characterization and fluid sampling.

5.9.4 Pressure Derivative Analysis

The pressure derivative method, introduced by Bourdet et al. (1983), is the primary diagnostic tool for modern well test interpretation. Rather than relying on identifying straight lines on semi-log plots — which can be ambiguous — the pressure derivative reveals flow regimes through characteristic slope patterns on a log-log diagnostic plot.

The pressure derivative is defined as:

On a log-log plot of $\Delta P$ and $\Delta P'$ versus $\Delta t$, various flow regimes appear as characteristic signatures:

The diagnostic workflow is:

1. Plot $\Delta P$ and $\Delta P'$ on a log-log scale
2. Identify flow regimes from the derivative signature
3. Select the appropriate analysis model (radial, fractured, dual-porosity, bounded)
4. Perform the corresponding straight-line or type-curve analysis
5. Verify the interpretation with a simulation match

The radial flow period — identified by a horizontal derivative — is the key segment from which permeability is extracted. The stabilized derivative level $m'$ relates to transmissibility:

where $m'$ is the stabilized derivative value (in field units with pressure in psi and rate in STB/d).

5.9.5 Wellbore Storage and Skin Effects

Wellbore storage occurs because the wellbore itself acts as a fluid container. When the well is shut in at surface, flow from the formation continues because the fluid in the wellbore compresses (or expands) and the liquid level adjusts. This masks the early-time reservoir response.

The wellbore storage coefficient $C$ depends on the dominant storage mechanism:

Compressibility-dominated storage (wells filled with single-phase fluid):

where $V_w$ is the wellbore volume [bbl] and $c_w$ is the wellbore fluid compressibility [psi$^{-1}$].

Liquid-level change (wells with a gas-liquid interface):

where $A_{wb}$ is the wellbore cross-sectional area [ft²] and $\rho_L$ is the liquid density [lb/ft³]. This mechanism gives a much larger storage coefficient and longer storage duration.

During the wellbore storage period, pressure change is linear in time:

On the log-log diagnostic plot, both $\Delta P$ and $\Delta P'$ follow a unit-slope line during storage. The end of the storage period occurs at approximately:

where $S$ is the skin factor. Large skin values (damaged wells) or large storage coefficients significantly extend the storage period, potentially masking the radial flow regime entirely in short tests.

The skin factor $S$ quantifies the additional pressure drop in the near-wellbore region compared to the ideal case. Positive skin indicates damage or flow restriction; negative skin indicates stimulation (e.g., fracturing). Typical values:

The additional pressure drop due to skin is:

This pressure drop acts as a fixed "tax" on the well's deliverability and directly reduces the well's IPR. Reducing skin through stimulation can have a dramatic effect on production rate.

5.9.6 Fitting Deliverability from Test Data in NeqSim

Once a multi-rate or multi-point test is available, NeqSim converts the raw rate/pressure pairs into a usable inflow model with WellTestMatcher (package neqsim.process.fielddevelopment.integrated). The matcher fits either a productivity index or a Vogel inflow and returns a WellDeliverabilityCurve that can be evaluated at any flowing pressure:

integ = jneqsim.process.fielddevelopment.integrated
WellTestMatcher = integ.WellTestMatcher

matcher = WellTestMatcher()
matcher.addTestPoint(1200.0, 180.0)        # rate [Sm3/day], flowing pressure [bara]
matcher.addTestPoint(2100.0, 150.0)
matcher.addTestPoint(2800.0, 120.0)

match = matcher.fitVogel()                 # or matcher.fitProductivityIndex()
print("Reservoir pressure:", match.getReservoirPressure(), "bara")
print("RMS error:", match.getRmsError())

curve = match.getCurve()                    # WellDeliverabilityCurve
print("AOFP:", curve.getAbsoluteOpenFlowPotential(), "Sm3/day")
print("Rate at 100 bara:", curve.rateAt(100.0), "Sm3/day")

A WellDeliverabilityCurve can also be built directly from tabulated points (WellDeliverabilityCurve(pressureBara[], rateSm3PerDay[])) or analytically with WellDeliverabilityCurve.fromVogel(aofpSm3PerDay, shutInPressureBara). The curve exposes rateAt(pBara), slopeAt(pBara), getShutInPressure(), and getAbsoluteOpenFlowPotential(), and is the inflow object consumed by the integrated production model of Chapter 28.

5.10 VFP Table Generation

5.10.1 What Are VFP Tables?

Vertical Flow Performance (VFP) tables are multidimensional lookup tables that describe the relationship between well flow rate and the pressure at a reference depth (typically the wellhead or bottomhole) as a function of:

• Flow rate
• Tubing head pressure (or bottomhole pressure)
• Water cut
• Gas-oil ratio (or gas-liquid ratio)
• Artificial lift parameter (gas lift injection rate, ESP speed)

VFP tables are used by reservoir simulators to model well performance without running a full multiphase flow calculation at every timestep.

5.10.2 VFP Table Structure

A typical VFP table has the following dimensions:

The total number of table entries can be large: 15 × 5 × 5 × 4 × 6 = 9,000 pressure calculations.

5.10.3 VFP Generation with NeqSim

import jpype
jneqsim = jpype.JPackage("neqsim")
import json

# Define fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 90.0, 250.0)
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 2.0)
fluid.addComponent("methane", 70.0)
fluid.addComponent("ethane", 7.0)
fluid.addComponent("propane", 4.0)
fluid.addComponent("n-butane", 2.5)
fluid.addComponent("n-pentane", 1.5)
fluid.addComponent("n-hexane", 1.0)
fluid.addComponent("n-heptane", 4.0)
fluid.addComponent("n-octane", 3.0)
fluid.addComponent("n-decane", 2.5)
fluid.addComponent("water", 2.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

# VFP parameters
well_depth = 2500.0          # m TVD
tubing_id = 0.1016           # m (4-inch)
roughness = 2.5e-5           # m
flow_rates = [10000, 20000, 40000, 60000, 80000, 100000]  # kg/hr
whp_values = [20.0, 30.0, 40.0, 50.0]                     # bara

Stream = jneqsim.process.equipment.stream.Stream
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

vfp_table = []

for whp in whp_values:
    for rate in flow_rates:
        # Set up well from BHP to WHP
        stream = Stream("BH", fluid.clone())
        stream.setFlowRate(rate, "kg/hr")
        stream.setTemperature(90.0, "C")
        stream.setPressure(300.0, "bara")  # Start high, will iterate

        tubing = PipeBeggsAndBrills("Tubing", stream)
        tubing.setPipeWallRoughness(roughness)
        tubing.setLength(well_depth)
        tubing.setElevation(well_depth)
        tubing.setDiameter(tubing_id)
        tubing.setNumberOfIncrements(40)

        process = ProcessSystem()
        process.add(stream)
        process.add(tubing)
        process.run()

        calculated_whp = tubing.getOutletStream().getPressure("bara")
        bhp = stream.getPressure("bara")

        vfp_table.append({
            "flow_rate_kg_hr": rate,
            "whp_target_bara": whp,
            "bhp_bara": bhp,
            "whp_calculated_bara": calculated_whp,
        })

# Print VFP table
print(f"{'Rate (kg/hr)':>14} {'WHP calc (bara)':>16} {'BHP (bara)':>12}")
print("-" * 44)
for entry in vfp_table[:12]:  # First 12 entries
    print(f"{entry['flow_rate_kg_hr']:14.0f} "
          f"{entry['whp_calculated_bara']:16.1f} "
          f"{entry['bhp_bara']:12.1f}")

The parallel curves illustrate an imposed boundary-pressure shift. They are not calculated or measured VFP tables; the executed well examples include hydraulic solution checks and an explicit operating domain.

5.10.4 VFP Tables in Reservoir Simulators

Reservoir simulators (Eclipse, INTERSECT, tNavigator, CMG) use VFP tables as the well model — at each timestep, the simulator looks up the required bottomhole pressure for the current well rate and conditions rather than running a full multiphase flow calculation. This is computationally efficient because:

• A VFP table takes nanoseconds to interpolate, while a Beggs and Brill calculation takes milliseconds
• A field model may have hundreds of wells simulated over thousands of timesteps
• The VFP table encodes the full wellbore hydraulics including tubing geometry, deviation, and artificial lift

The parameters that must be varied to build a complete VFP table depend on the well type:

For a typical oil producer, a well-constructed VFP table might have:

• 10–15 flow rates spanning from near-zero to maximum rate
• 5–8 tubing head pressures covering the operating range
• 5–6 water cut values (0%, 20%, 40%, 60%, 80%, 95%)
• 4–5 GOR values around the expected range

This gives up to 15 × 7 × 6 × 5 = 3,150 entries per table — each requiring a full multiphase flow calculation.

5.10.5 VFP Table Quality and Interpolation

The quality of VFP tables directly affects the accuracy of reservoir simulation results. Key quality considerations include:

Parameter spacing: Table entries should be spaced to capture the nonlinear behavior of multiphase flow. Finer spacing is needed where the VFP curves have high curvature (at low rates for oil wells, near the transition between liquid-loaded and stable flow for gas wells).

Interpolation method: Most reservoir simulators use multi-dimensional linear interpolation in the VFP table. Because multiphase pressure drop is nonlinear with respect to rate, linear interpolation can introduce errors if the rate spacing is too coarse. Some simulators support logarithmic interpolation in rate, which is more accurate.

Extrapolation hazards: If the simulator requests a BHP at conditions outside the VFP table range, it must extrapolate — which can produce physically unreasonable results. Common problems include:

• Rate exceeding the maximum table entry → BHP extrapolated to unrealistically low values
• Water cut exceeding the maximum table entry → incorrect BHP, especially at high water cuts where behavior changes rapidly
• GOR below the minimum → BHP may be overestimated (dead oil assumption)

Table consistency: A physical multiphase VFP curve can be non-monotonic: decreasing hydrostatic head at low rate competes with increasing friction at high rate. Do not force monotonicity to hide this behavior. Check each branch for hydraulic convergence, units and interpolation error; assess dynamic stability separately before choosing an operating point.

Updating VFP tables: As reservoir pressure declines, the fluid composition at the wellbore changes (lower GOR, higher water cut). VFP tables should be regenerated periodically — typically every few years of simulation time — or the table dimensions should cover the full expected range.

5.11 Tubing Size Selection

5.11.1 The Tubing Size Tradeoff

Tubing size selection involves a fundamental tradeoff:

• Large tubing: Lower friction but higher hydrostatic head at low rates (liquid loading risk); also higher capital cost
• Small tubing: Higher friction but lower liquid loading risk at low rates; lower cost

The optimal tubing size depends on the expected production profile:

5.11.2 Liquid Loading and Turner Critical Velocity

Liquid loading is one of the most significant operational problems in gas well production. It occurs when the gas velocity in the tubing falls below the minimum velocity required to continuously transport liquid (water or condensate) droplets to surface. Turner et al. (1969) developed the critical velocity correlation by modeling the balance between aerodynamic drag and gravity on liquid droplets:

where:

• $v_{cr}$ is the critical (minimum) gas velocity [m/s]
• $\sigma$ is the liquid-gas surface tension [N/m]
• $\rho_l$ is the liquid density [kg/m³]
• $\rho_g$ is the gas density [kg/m³]

The coefficient6.56 is the SI conversion of the commonly used1.91 field-unit droplet-screening form, including its approximate20% empirical uplift; without that uplift the coefficient is about 5.46. The field expression uses ft/s, dyn/cm and lb/ft³ and must not be paired with the SI definitions above. For $\sigma=0.06$N/m, $\rho_l=1000$kg/m³ and $\rho_g=50$kg/m³, the uplifted SI expression gives2.55m/s. This is a vertical droplet-screening model, not a general criterion for film reversal, deviated wells or transient unloading.[14]

The corresponding critical gas flow rate for a given tubing size is:

If the production rate drops below this critical rate, the well will begin to load up. The symptoms of liquid loading include:

• Erratic wellhead pressure and flow rate (heading)
• Increasing casing-tubing pressure differential
• Declining production rate
• Intermittent flow (well produces in slugs)
• Eventually, the well dies (stops flowing)

The implication for tubing size selection is clear: smaller tubing maintains higher velocities at lower rates, delaying the onset of liquid loading. However, smaller tubing creates higher friction at peak rates. The designer must consider the full production life:

For wells that will experience a wide range of rates, velocity strings (smaller tubing installed inside existing tubing) or plunger lift can extend the well's flowing life.

A useful rule of thumb: select the tubing size such that the expected minimum flowing rate is 20–30% above the Turner critical rate for that tubing diameter.

5.11.3 Tubing Size Analysis with NeqSim

import jpype
jneqsim = jpype.JPackage("neqsim")

# Compare 3.5-inch vs 4.5-inch tubing for the same well
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 90.0, 200.0)
fluid.addComponent("methane", 70.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 4.0)
fluid.addComponent("n-butane", 2.0)
fluid.addComponent("n-heptane", 8.0)
fluid.addComponent("n-decane", 5.0)
fluid.addComponent("water", 3.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

Stream = jneqsim.process.equipment.stream.Stream
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

tubing_sizes = {
    "3.5-inch (ID=0.076m)": 0.0760,
    "4.5-inch (ID=0.102m)": 0.1016,
}

flow_rate = 60000.0  # kg/hr

print(f"{'Tubing Size':>25} {'WHP (bara)':>12} {'dP (bar)':>10}")
print("-" * 50)

for name, diameter in tubing_sizes.items():
    stream = Stream("BH", fluid.clone())
    stream.setFlowRate(flow_rate, "kg/hr")
    stream.setTemperature(90.0, "C")
    stream.setPressure(200.0, "bara")

    tubing = PipeBeggsAndBrills("Tubing", stream)
    tubing.setPipeWallRoughness(2.5e-5)
    tubing.setLength(2500.0)
    tubing.setElevation(2500.0)
    tubing.setDiameter(diameter)
    tubing.setNumberOfIncrements(40)

    process = ProcessSystem()
    process.add(stream)
    process.add(tubing)
    try:
        process.run()
    except Exception as error:
        print("Infeasible fixed-rate tubing case:", str(error).splitlines()[0])
        continue

    whp = tubing.getOutletStream().getPressure("bara")
    dp = 200.0 - whp
    print(f"{name:>25} {whp:12.1f} {dp:10.1f}")

5.12 Nodal Analysis

5.12.1 Concept

Nodal analysis is the systematic method of determining a well's operating point by decomposing the production system into inflow and outflow components at a chosen solution node — typically the bottomhole. The operating point is the intersection of the IPR curve (inflow from reservoir to node) and the VFP curve (outflow from node to surface).

At the solution node (bottomhole), two pressure relationships must be satisfied simultaneously:

Inflow (IPR): The relationship between flowing bottomhole pressure $P_{wf}$ and rate $q$, determined by reservoir deliverability (Vogel, Darcy, or composite IPR from Chapter 4).

Outflow (VFP): The bottomhole pressure required to lift the fluid to surface at rate $q$, given the tubing geometry, wellhead pressure, and fluid properties (calculated using the Beggs and Brill correlation from Section 5.3).

The operating point is where:

The plotted intersection illustrates the common-node pressure condition. It is not a measured well test or a calibrated production forecast; the chapter notebook solves the declared hydraulic case separately.

5.12.2 Sensitivity Analysis

Nodal analysis is most powerful as a sensitivity tool — changing one variable at a time to quantify its effect on the operating point:

• Tubing size: Larger tubing shifts the VFP curve down (lower required BHP), increasing rate — but only if friction dominates. At low rates, larger tubing may shift the curve up due to liquid holdup.
• Wellhead pressure: Lower THP shifts the VFP curve down, directly increasing rate. This shows the benefit of first-stage separator pressure optimization.
• Water cut: Higher water cut steepens the VFP curve (heavier fluid column) and shifts the IPR curve down (lower PI), reducing rate significantly.
• Skin factor: Higher skin shifts the IPR curve to the left (lower rate for same drawdown). Stimulation shifts it right.
• Gas lift injection: Shifts the VFP curve down by reducing mixture density, increasing rate until the friction penalty dominates.

5.12.3 Multiple Operating Points

In some cases, particularly for high-GOR oil wells or gas-lifted wells, the IPR and VFP curves may intersect at two points. For the reduced scalar dynamic model $d q/dt=K[P_{\mathrm{IPR}}(q)-P_{\mathrm{VFP}}(q)]$ with $K>0$, a root is locally stable only if $d(P_{\mathrm{IPR}}-P_{\mathrm{VFP}})/dq<0$. This often makes a low-rate root unstable and a higher-rate root stable, but the ordering alone is not a proof. Wellbore storage, multiphase transients and controllers can change the stability; startup must be assessed dynamically.

This has practical implications for well startup: the well may need to be kicked off (e.g., with nitrogen or by briefly increasing gas lift) to pass through the unstable region and reach the stable operating point.

5.13 Completion Effects on Well Performance

5.13.1 Skin Factor Decomposition

The total skin factor measured from well tests is a composite of several independent components:

where:

• $S_d$ = Mechanical (damage) skin: Caused by drilling fluid invasion, fines migration, scale, or asphaltene deposition near the wellbore. Typically 0 to +30.
• $S_{pp}$ = Partial penetration skin: When the perforated interval is less than the full reservoir thickness, flow convergence toward the open interval creates additional pressure drop. Can be +5 to +50 for highly partial completions.
• $S_{\theta}$ = Deviation skin: Deviated wellbores expose more reservoir area, which is a negative skin contribution. For high-angle wells, $S_{\theta}$ can be -2 to -5.
• $S_{perf}$ = Perforation skin: Depends on perforation density (shots/ft), penetration depth, phasing angle, and crushed zone. Can range from -1 (ideal perforations) to +20 (poor perforations in damaged zone).
• $S_c$ = Completion skin: Additional pressure drop from screens, gravel packs, or other completion hardware.

Understanding the decomposition is critical for optimization — there is no value in stimulating a well if the skin is dominated by partial penetration (which requires deepening the completion or adding perforations).

5.13.2 Gravel Pack and Frac-Pack Completions

In unconsolidated formations prone to sand production, gravel-pack completions are used to prevent sand while maintaining well productivity.

For an annulus replacing formation material, the incremental skin relative to the original formation is $(k/k_{gp}-1)\ln(r_{gp}/r_w)$. If the pack is an additional serial resistance, use the full pack resistance instead; perforation/convergence damage is a separate contribution.

Gravel pack (GP): A screen is placed across the interval, and sized gravel is packed between the screen and the formation. The skin contribution from the gravel pack depends on gravel permeability ($k_{gp}$, typically 50–200 D) and annular thickness:

Because $k_{gp} \gg k$, the gravel-pack skin is usually small (0.5–3) for a properly designed pack. However, impaired gravel (contaminated during placement) can give much higher skin.

Frac-pack: A hydraulic fracture is created and propped with gravel, connecting the perforation tunnels to the formation with a high-conductivity channel. Frac-packs typically achieve negative total skin (-1 to -4) and are preferred in moderate-permeability formations (10–500 mD) where standard gravel packs would impose unacceptable skin.

The completion's effect on the IPR is direct: the effective skin factor determines the additional drawdown consumed by the completion. For a well producing at 5,000 STB/d with $kh/\mu = 10,000$ mD·ft/cp, each unit of skin consumes $141.2(5000)B/10000=70.6B$ psi, or $4.87B$ bar, with $B$ in reservoir bbl/STB. For $B=1.2$, this is5.84 bar per skin unit.

5.13.3 Impact on IPR

The completion skin modifies the IPR equation. For the Darcy (straight-line) IPR:

The denominator increases with skin, reducing the productivity index. A skin reduction from $S = +10$ to $S = +2$ can increase PI by 30–50% in typical wells, making stimulation (acidizing, hydraulic fracturing) one of the highest-return investments in production optimization.

5.14 Flowing Bottomhole Pressure Surveys

5.14.1 Gradient Surveys

Flowing gradient surveys measure pressure and temperature at multiple depths in a producing well using a wireline or slickline-conveyed gauge. The resulting pressure-depth profile (gradient survey) provides:

• Flowing gradient: The slope $dP/dz$ at each depth, which reflects the local mixture density and friction
• Gas entry points: Sudden gradient changes indicate gas influx from gas lift valves or the formation
• Liquid level: In wells with a gas cap in the annulus, the gradient transition from gas to liquid indicates the dynamic liquid level
• Flow regime identification: Smooth gradients suggest stable flow; erratic gradients suggest slug flow

A flowing gradient survey in a gas-lifted well typically shows:

1. A steep gradient below the operating valve (heavy, unassisted fluid)
2. A gradient change at the gas lift injection depth
3. A lighter gradient above the operating valve (gas-lifted fluid)

5.14.2 Interpretation and Flow Regime Identification

The measured pressure gradient can be compared to theoretical gradients for different flow regimes:

If the measured gradient exceeds the hydrostatic gradient of the wellbore fluid, friction is significant. If it falls below the static liquid gradient, gas is present (reduced holdup).

Gradient surveys taken at different rates provide particularly valuable data for calibrating multiphase flow models — the gradient at each depth varies with rate because the holdup and flow regime change. This data can be used to tune the Beggs and Brill correlation parameters or select between alternative correlations.

5.14.3 Production Logging

Production logging tools (PLT) combine pressure and temperature gauges with flow measurement sensors (spinners, capacitance probes, optical probes) to determine the flow contribution from each zone in a commingled completion. The zonal flow rates are essential for:

• Identifying watered-out zones for water shut-off
• Detecting crossflow between zones
• Quantifying zonal productivity for selective stimulation
• Validating reservoir simulation models

Reproduced Calculation Results

These examples use the stated fluid recipes and operating assumptions. Curves represent NeqSim calculations unless a caption identifies an analytical illustration, assumed equipment map or synthetic data.

Tubing ID = 3.0 inch: wellhead pressure spans 33.92–122 bara across the plotted cases. Tubing ID = 4.0 inch: wellhead pressure spans 33.83–121.7 bara across the plotted cases.

Tubing pressure loss combines hydrostatic head, wall friction and acceleration; changing diameter alters friction and liquid transport. A diameter that reduces friction at high rate may increase liquid loading or change the low-rate operating branch. Compare feasible IPR intersections for each diameter and reject trials with nonpositive outlet pressure or inconsistent upward elevation sign.

The coupled solution gives 59.518 t/hr at 175.60 bara bottomhole pressure. The calculated wellhead pressure is 49.999 bara against a 50 bara target. Gaps retain undefined phase quantities or hydraulic states that fail the stated operating boundary; they are not interpolated.

The operating point is the rate at which reservoir deliverability and the tubing pressure requirement satisfy the same bottom-hole boundary. The root residual quantifies whether the selected rate actually delivers the required wellhead pressure; visual proximity alone is insufficient. Check the pressure residual, BHP greater than WHP for upward production, and sensitivity to IPR and hydraulic-correlation uncertainty.

Oil Production Rate spans 2000–5158 bbl/d across the plotted cases.

Injected gas reduces the hydrostatic density of the wellbore mixture while increasing friction and surface gas handling. Improved lift pressure does not by itself establish incremental oil production; a rate gain requires a coupled inflow solution and sufficient gas capacity. Optimize gas allocation using incremental oil or value per unit injection gas, including compression demand and total gas availability.

0 mol% water: wellhead pressure spans 33.83–121.7 bara across the plotted cases. 10 mol% water: wellhead pressure spans 35.75–119.1 bara across the plotted cases.

Increasing water fraction changes mixture density, liquid holdup and friction while reducing the oil fraction of the produced stream. At a fixed total mass rate, oil production and available lifting margin can both deteriorate as water cut rises. Track oil, gas and water rates separately and recompute the coupled well operating point at each water-cut case.

Selected numerical ranges from the plotted cases:

Ranges describe the sampled cases; they are not independent validation tolerances.

5.15 Summary

Key points from this chapter:

• Multiphase flow in wells is characterized by gravity-dominated pressure loss, with liquid holdup as the critical parameter. The Beggs and Brill correlation handles all pipe inclinations and flow patterns.
• Wellhead chokes regulate production and provide critical flow isolation. The Joule-Thomson cooling across the choke must be considered for hydrate and wax risk assessment.
• Gas lift is the most common offshore artificial lift method. The optimization problem is to allocate limited injection gas among wells to maximize total production — solved by the equal-marginal-gain criterion.
• ESP and rod pumps provide alternatives for wells that cannot flow naturally. ESPs handle high rates; rod pumps are standard for low-rate onshore wells.
• Well testing (drawdown, buildup, DST) provides the permeability, skin, and reservoir pressure needed for IPR construction.
• VFP tables capture the well's multiphase flow performance in a lookup table format for use in reservoir simulators and optimization models.
• Tubing size selection balances friction loss against liquid loading risk, with the optimal size depending on the expected production rate range.
• NeqSim's PipeBeggsAndBrills class provides the core multiphase flow calculation for all well performance modeling.

Verification of the worked examples

The completed wellbore trials are checked for finite positive pressure and temperature, material conservation and the upward pressure-loss direction; infeasible imposed-rate trials remain identified. The rod-pump displacement is checked against swept geometry, and the jet-pump example checks pressure and efficiency domains. These reduced lift models do not constitute a coupled power-fluid, reservoir and surface design; the Turner coefficient is separately checked for its stated unit convention.[14]

The calculation and literal-code records are in verification/scientific_revision/ch05_manuscript_physics.json; the chapter scope and code hashes are indexed in foundations_review.json.

Exercises

1. Exercise 5.1: Using NeqSim, calculate the wellhead pressure for a vertical well producing 50,000 kg/hr through 4-inch tubing from a depth of 3,000 m TVD. The bottomhole pressure is 250 bara and the temperature is 95°C. Use the gas condensate composition from Chapter 2.

1. Exercise 5.2: For the well in Exercise 5.1, generate a complete set of pressure traverse curves for flow rates from 10,000 to 100,000 kg/hr. Plot BHP vs. rate for a fixed WHP of 30 bara to create a VFP curve. On the same plot, overlay the Vogel IPR from Chapter 4 to find the operating point.

1. Exercise 5.3: Model a wellhead choke that reduces pressure from 80 bara to 30 bara. Calculate the Joule-Thomson temperature drop and determine if the downstream temperature is below the hydrate formation temperature for the gas composition.

1. Exercise 5.4: Compare the wellhead pressures for 2-7/8", 3-1/2", and 4-1/2" tubing at flow rates of 20,000, 40,000, 60,000, and 80,000 kg/hr. Identify which tubing size gives the best wellhead pressure at each rate.

1. Exercise 5.5: Design a gas lift system for a well with the following data: well TVD = 2,800 m, reservoir pressure = 180 bara, PI = 20 Sm³/d/bar, tubing ID = 4 inches, available gas injection pressure at surface = 120 bara. Calculate the gas lift performance curve (oil rate vs. gas injection rate) and identify the optimum injection rate.

1. Exercise 5.6: Generate a VFP table with dimensions: 6 flow rates × 4 WHPs × 3 water cuts (0%, 30%, 60%) for a 3,000 m well with 4-inch tubing. Report the table as a formatted data structure suitable for import into a reservoir simulator.

1. Exercise 5.7 (Advanced): For a three-well gas lift system sharing a compressor with 150,000 Sm³/d total injection capacity, each well has a different gas lift performance curve. Determine the optimal allocation of injection gas using the equal-marginal-gain method. Compare with equal allocation and maximum-rate-first allocation.

1. Beggs, H. D., & Brill, J. P. (1973). A study of two-phase flow in inclined pipes. Journal of Petroleum Technology, 25(5), 607–617.
2. Hagedorn, A. R., & Brown, K. E. (1965). Experimental study of pressure gradients occurring during continuous two-phase flow in small-diameter vertical conduits. Journal of Petroleum Technology, 17(4), 475–484.
3. Turner, R. G., Hubbard, M. G., & Dukler, A. E. (1969). Analysis and prediction of minimum flow rate for the continuous removal of liquids from gas wells. Journal of Petroleum Technology, 21(11), 1475–1482.
4. Taitel, Y., & Dukler, A. E. (1980). Modelling flow pattern transitions for steady upward gas-liquid flow in vertical tubes. AIChE Journal, 26(3), 345–354.
5. Brown, K. E. (1984). The Technology of Artificial Lift Methods, Vol. 4: Production Optimization. PennWell Books.
6. Takacs, G. (2009). Gas Lift Manual. PennWell Books.
7. Lea, J. F., Nickens, H. V., & Wells, M. R. (2008). Gas Well Deliquification (2nd ed.). Gulf Professional Publishing.
8. Horne, R. N. (1995). Modern Well Test Analysis: A Computer-Aided Approach (2nd ed.). Petroway Inc.
9. Economides, M. J., Hill, A. D., Ehlig-Economides, C., & Zhu, D. (2013). Petroleum Production Systems (2nd ed.). Prentice Hall.
10. Brill, J. P., & Mukherjee, H. (1999). Multiphase Flow in Wells. SPE Monograph Series, Vol. 17.

Running the examples. Start the source-workspace Python session described in Chapter 1, then run this chapter's Python blocks in reading order. Java blocks form a separate sequence using the same NeqSim build; carry forward objects from preceding Java blocks. The release execution records are in verification/; a successful run establishes API compatibility, while physical validation also requires the checks discussed in the text.

Learning Objectives

After reading this chapter, the reader will be able to:

1. Model production well networks using NeqSim's LoopedPipeNetwork class with multiple wells, chokes, tubing, and gathering flowlines
2. Implement IPR models (Productivity Index, Vogel, Fetkovich) for both oil and gas wells within a network framework
3. Size and model production chokes using IEC 60534-style valve equations with critical flow detection
4. Calculate vertical lift performance (VLP) for well tubing using segmented pressure drop correlations
5. Build complete multi-well gathering networks with source, junction, and sink nodes
6. Model artificial lift systems (gas lift, ESP, jet pump, rod pump) and their effect on network hydraulics
7. Optimize choke settings and production allocation across a well network using NLP and multi-objective methods
8. Track water handling, sand production, corrosion, and GHG emissions within the network model
9. Implement production well networks in Python for rapid prototyping and sensitivity analysis

---

6.1 Introduction

Previous chapters addressed individual well performance — multiphase flow in tubing (Chapter 5), choke behavior (Chapter 15 Valves), and artificial lift methods. In reality, production wells do not operate in isolation. They are connected through a gathering network of flowlines, manifolds, and headers that deliver the combined production to the processing facility. The performance of each well depends on the network back-pressure, which in turn depends on the production rates of all other wells.

This chapter introduces production well network modeling — the discipline of simultaneously solving for flow rates and pressures across an interconnected system of wells, chokes, tubing strings, flowlines, and processing constraints. We present NeqSim's LoopedPipeNetwork class, which implements a Newton–Raphson Global Gradient Algorithm (NR-GGA). Runtime and convergence must be measured for the actual topology, fluid, closures and starting point.

The chapter progresses from individual building blocks (IPR models, choke equations, tubing VLP) to complete network assembly and optimization. We conclude with practical topics: artificial lift integration, water handling, flow assurance, emissions tracking, and Python implementations for rapid engineering studies.

6.1.1 Why Network Modeling Matters

Consider a platform producing from 20 wells through a common manifold at 40 bara. If one well increases production, the manifold pressure rises slightly, reducing the drawdown — and hence the flow rate — of every other well. This coupling means that:

• Individual well optimization is insufficient. Maximizing one well may reduce total platform production.
• Choke settings interact. Opening a choke on Well A changes the back-pressure on Wells B through T.
• Artificial lift allocation is a system problem. Gas lift gas or ESP power allocated to one well affects what is available for others.
• Facility constraints propagate. A compressor limit or separator capacity constraint affects all wells simultaneously.

Network modeling captures these interactions and enables true system-level optimization.

6.1.2 The LoopedPipeNetwork Architecture

NeqSim's LoopedPipeNetwork represents the production system as a directed graph:

• Nodes represent physical locations: reservoir sandface, wellbore, wellhead, manifold, platform arrival
• Elements represent resistance components: IPR (reservoir inflow), tubing, choke, flowline, compressor
• The solver simultaneously determines all nodal pressures and element flow rates

The key innovation is that each "element" can be a different physical model — not just a pipe. The element types supported are:

This generalized approach means that a single network model can represent the complete path from reservoir to export — or any subset thereof.

---

6.2 Production Well Networks in NeqSim

6.2.1 The LoopedPipeNetwork Class

The LoopedPipeNetwork class is the central component for production network modeling in NeqSim. It extends ProcessEquipmentBaseClass and can be embedded within a ProcessSystem for integrated facility simulation.

API units and model scope in the 2026 release

addSourceNode accepts bara and optional kg/hr, whereas NetworkNode.setPressure uses Pa. addWellIPR takes a productivity index in kg/s/Pa for oil or kg/s/Pa² for gas and a Boolean gas flag; reservoir pressure comes from the source node. The examples explicitly convert illustrative standard-volume indices using a stated reference density. getNodePressure returns bara, getPipeFlowRate returns kg/hr, and getTotalSinkFlow returns kg/s. Mixing these interfaces can produce plausible-looking results wrong by factors of 100,000 or 3,600.

Tubing elements use a reduced hydraulic relation; the multiphase-pipe element uses PipeBeggsAndBrills. Gas-lift, ESP and water-cut switches therefore support screening, with detailed well and phase models needed to qualify an operating recommendation. A reported optimizer objective is not a convergence or feasibility certificate. Inspect converged, the residual in Pa, mass balance in kg/s, and the candidate constraint report before comparing production.[1]

import org.apache.logging.log4j.LogManager;
import org.apache.logging.log4j.Logger;
Logger logger = LogManager.getLogger("BookChapter6");
import neqsim.process.equipment.network.LoopedPipeNetwork;
import neqsim.process.equipment.network.LoopedPipeNetwork.*;
import neqsim.thermo.system.SystemInterface;
import neqsim.thermo.system.SystemSrkEos;

// Create a fluid template for the network
SystemInterface gas = new SystemSrkEos(273.15 + 80.0, 200.0);
gas.addComponent("methane", 0.80);
gas.addComponent("ethane", 0.08);
gas.addComponent("propane", 0.04);
gas.addComponent("n-butane", 0.02);
gas.addComponent("n-heptane", 0.04);
gas.addComponent("water", 0.02);
gas.setMixingRule("classic");
gas.setMultiPhaseCheck(true);

// Create the network
LoopedPipeNetwork network = new LoopedPipeNetwork("Production Network");
network.setFluidTemplate(gas);

6.2.2 Node Types

Every network requires at least one source node (where fluid enters) and one sink node (where fluid exits). Junction nodes connect elements internally.

Source nodes represent reservoirs or injection points. They have a fixed pressure (the reservoir pressure) and supply whatever flow rate the IPR allows:

// Reservoir nodes (fixed pressure sources)
network.addSourceNode("Reservoir-A", 250.0, 0.0);  // 250 bar, flow determined by IPR
network.addSourceNode("Reservoir-B", 220.0, 0.0);  // 220 bar

Sink nodes represent delivery points — the manifold, separator inlet, or export pipeline. They may have a fixed pressure (back-pressure constraint) or a fixed demand:

// Platform manifold (fixed back-pressure)
network.addSinkNode("Manifold", 0.0);  // demand determined by network solution

// Alternative: fixed pressure at sink
NetworkNode manifold = network.getNode("Manifold");
manifold.setPressure(40.0e5);        // 40 bara in Pa
manifold.setPressureFixed(true);

Junction nodes are intermediate connection points — wellheads, subsea manifolds, pipeline junctions:

// Wellhead and subsea junction nodes
network.addJunctionNode("WH-A");        // Wellhead of Well A
network.addJunctionNode("WH-B");        // Wellhead of Well B
network.addJunctionNode("BH-A");        // Bottomhole of Well A
network.addJunctionNode("BH-B");        // Bottomhole of Well B
network.addJunctionNode("Subsea-Manifold");

6.2.3 Network Element Types

Each element connects two nodes and has a pressure-flow relationship $\Delta P = f(Q)$ and its derivative $d\Delta P/dQ$. The NR-GGA solver uses these functions to build the Jacobian matrix.

PIPE elements use the Darcy-Weisbach equation for single-phase flow:

// Subsea flowline (single-phase pipe)
network.addPipe("Subsea-Manifold", "Manifold", "Export Flowline",
    15000.0,    // length [m]
    0.254);     // diameter [m] (10-inch)

MULTIPHASE_PIPE elements use the Beggs-Brill correlation for two-phase and three-phase systems, wrapping NeqSim's PipeBeggsAndBrills:

// Multiphase gathering line
NetworkPipe gatherLine = network.addPipe("WH-A", "Subsea-Manifold",
    "Gathering Line A", 5000.0, 0.2032);  // 8-inch, 5 km
gatherLine.setElementType(NetworkElementType.MULTIPHASE_PIPE);
gatherLine.setMultiphaseSegments(20);
gatherLine.setRoughness(4.5e-5);

6.2.4 The Newton-Raphson Global Gradient Algorithm

The NR-GGA solver (also known as the Todini-Pilati method) solves the network equations simultaneously. For a network with $N_n$ nodes and $N_e$ elements, the system is:

Continuity at each node (mass balance):

where $D_i$ is the external demand (positive) or supply (negative) at node $i$.

Momentum equation for each element (pressure-flow relationship):

where $H_i$ is the head (pressure) at node $i$ and $h_k(Q_k)$ is the head loss through element $k$ at flow rate $Q_k$.

The NR-GGA linearizes these equations around the current estimate $(Q^{(n)}, H^{(n)})$ and solves the resulting linear system:

where:

• $A_{11} = \text{diag}(dh_k/dQ_k)$ — diagonal matrix of element derivatives
• $A_{12}$ — incidence matrix (network topology)
• $A_{21} = A_{12}^T$ — transpose of incidence matrix
• $r_1, r_2$ — residuals of momentum and continuity equations

Schur complement reduction eliminates $\Delta Q$ to solve a smaller $N_n \times N_n$ system for nodal pressures:

This is highly efficient because $A_{11}$ is diagonal (trivial to invert), and the reduced system is much smaller than the full system for networks with many more pipes than nodes.

6.2.5 Solver Configuration

// Select the NR-GGA solver
network.setSolverType(SolverType.NEWTON_RAPHSON);

// Convergence settings
network.setTolerance(1e-6);       // Residual tolerance (Pa)
network.setMaxIterations(100);    // Maximum iterations

Adaptive relaxation prevents divergence in stiff networks. The solver automatically reduces the step size when the residual increases:

where $\alpha$ starts at 1.0 and is halved if the residual increases, down to a minimum of 0.1.

Flow initialization sets reasonable starting values for the iterative solver. The network automatically estimates initial flows based on source pressures and pipe resistances:

// The solver handles initialization internally, but you can override:
network.getPipe("Export Flowline").setFlowRate(50.0);  // kg/s

---

6.3 IPR Models Implementation

6.3.1 Productivity Index (PI) Model

The simplest IPR model assumes a linear relationship between flow rate and drawdown:

For oil wells (incompressible flow):

For gas wells (compressible flow — pressure-squared form):

where:

• $Q$ = flow rate [kg/s in network units]
• PI = productivity index [kg/s/Pa for oil, kg/s/Pa² for gas]
• $P_r$ = reservoir pressure [Pa]
• $P_{wf}$ = flowing bottomhole pressure [Pa]

The derivative for the NR-GGA solver is:

NeqSim convenience method:

// Oil well with PI = 15 Sm3/d/bar (converted internally to SI)
network.addWellIPR("Reservoir-A", "BH-A", "Well-A IPR",
    15.0 * 800.0 / 86400.0 / 1e5, // SI PI from 15 Sm3/day/bar and 800 kg/Sm3
    false);    // oil IPR; source node supplies reservoir pressure

// Gas well (set gasIPR flag)
NetworkPipe gasIPR = network.addWellIPR("Reservoir-B", "BH-B", "Well-B IPR",
    0.5 * 0.8 / 86400.0 / 1e10, // kg/s/Pa2 from 0.8 kg/Sm3 gas
    true);     // gas pressure-squared IPR
gasIPR.setGasIPR(true);

6.3.2 Vogel's Equation

Vogel (1968) developed an empirical IPR for solution-gas-drive reservoirs producing below the bubble point:

Solving for the pressure drop as a function of flow rate requires inverting this quadratic:

The derivative $d\Delta P/dQ$ follows from implicit differentiation:

NeqSim convenience method:

// Vogel IPR: Qmax = absolute open flow (AOF) in kg/s
network.addWellIPRVogel("Reservoir-A", "BH-A", "Well-A Vogel", 50.0);       // Qmax (AOF) [kg/s]

6.3.3 Fetkovich Equation

Fetkovich (1973) proposed a generalized backpressure equation for gas and gas-condensate wells:

where:

• $C$ = backpressure coefficient [kg/s/Pa^(2n)]
• $n$ = backpressure exponent (0.5 ≤ $n$ ≤ 1.0)
• $n = 1.0$: Darcy (laminar) flow dominates
• $n = 0.5$: turbulent/non-Darcy flow dominates
• Typical: $n \approx 0.7$–0.9

The derivative:

NeqSim convenience method:

// Fetkovich IPR: C and n from well test analysis
network.addWellIPRFetkovich("Reservoir-B", "BH-B", "Well-B Fetkovich", 1.5e-8,      // C coefficient
    0.85);       // n exponent

6.3.4 Choosing the Right IPR Model

6.3.5 The WellFlow Equipment and Inflow Constraints

The network convenience methods above attach an IPR to a NetworkPipe. For a self-contained well that participates in a ProcessSystem (or the integrated production model of Chapter 28), NeqSim provides the dedicated WellFlow equipment class (package neqsim.process.equipment.reservoir). A single instance can carry any of the inflow laws of Section 6.3:

WellFlow = jneqsim.process.equipment.reservoir.WellFlow

# Standalone oil IPR example, separate from the Java network.
oil_fluid = jneqsim.thermo.system.SystemSrkEos(353.15, 250.0)
oil_fluid.addComponent("methane", 0.20)
oil_fluid.addComponent("n-decane", 0.80)
oil_fluid.setMixingRule("classic")
reservoirStream = jneqsim.process.equipment.stream.Stream("Reservoir", oil_fluid)
reservoirStream.setFlowRate(200.0, "Sm3/day")
reservoirStream.run()
qTest, pwfTest, reservoirP = 200.0, 180.0, 250.0
well = WellFlow("Well-A")
well.setInletStream(reservoirStream)
# Explicit liquid-volume basis (Sm3/day), unlike the legacy MSm3/day setter.
InflowPerformance = jneqsim.process.equipment.reservoir.InflowPerformance
ratio = pwfTest / reservoirP
pi_liquid = qTest * 1.8 / (reservoirP * (1.0 - 0.2*ratio - 0.8*ratio**2))
well.setInflowPerformance(InflowPerformance.vogel(pi_liquid, reservoirP))
well.setLiquidRate(qTest)
# well.setFetkovichParameters(c, n, reservoirP)           # Fetkovich
# well.setBackpressureParameters(a, b, reservoirP)        # Rawlins-Schellhardt
# well.setTableInflow(bhp_array, rate_array)              # tabulated IPR
well.setOutletPressure(120.0, "bara")
well.solveFlowFromOutletPressure(True)
well.run()
assert 200.0 < well.getLiquidRate() < 400.0
print("Liquid rate (Sm3/day):", well.getLiquidRate(),
      "BHP:", well.getBottomHolePressure(),
      "drawdown:", well.getDrawdown())

WellFlow adds inflow constraints that the optimizers of Chapters 21 and 23 honour as capacity limits. They are activated with the fluent useWellConstraints() and bounded with reservoir-management limits:

well.useWellConstraints()
well.setMaxDrawdown(40.0, "bara")           # sand / coning management
well.setMinBottomHolePressure(90.0, "bara")  # lift / stability limit
# Registering a constraint does not make the old 130-bar drawdown feasible.
assert well.getDrawdown() > 40.0  # rejected previous operating point
well.setOutletPressure(210.0, "bara")  # reservoir 250 bara minus allowed 40 bar
well.run()
assert well.getDrawdown() <= 40.0 + 1e-8
assert well.getBottomHolePressure() >= 90.0
assert 0.0 < well.getLiquidRate() < qTest
print("Accepted constrained IPR rate:", well.getLiquidRate(), "Sm3/day")

For commingled completions the class supports multi-layer inflow — addLayer(name, stream, reservoirP, pi), setFlowMode(FlowMode), setTargetZone(...), with getZoneAllocationFractions() and getLayerFlowRates() reporting the crossflow split — and fracture-containment screening via setFracturePressure(...), setBarrierStressContrast(...), isFractureContained(bhp), and getFractureContainmentMargin(). These features make WellFlow the inflow building block for the well-and-network optimization of Chapter 26.

---

6.4 Choke Modeling

6.4.1 IEC 60534 Valve Flow Equation

Production chokes are modeled using a simplified form of the IEC 60534 valve equation:

where:

• $K_v$ = valve flow coefficient [m³/hr/√bar] at full opening
• $\theta$ = fractional opening (0–1, where 1 = fully open)
• $\Delta P$ = pressure drop across choke [bar]
• $\rho$ = fluid density [kg/m³]
• $\rho_{ref}$ = reference density (water = 1000 kg/m³)

For equal percentage characteristics (typical for production chokes):

where $R$ is the rangeability (typically 30–50 for production chokes).

6.4.2 Critical Flow Detection

When the pressure ratio $P_2/P_1$ falls below the critical pressure ratio $x_T$ (typically 0.5–0.7 for choke valves), the flow becomes sonic and no longer increases with further pressure reduction:

The network solver automatically detects critical flow and uses the appropriate equation. This is important because:

• Under critical flow, downstream pressure disturbances cannot propagate upstream
• The well is effectively isolated from manifold pressure fluctuations
• The choke opening becomes the sole control variable for the well's flow rate

6.4.3 Adding Chokes to the Network

// Production choke with Kv = 25 m3/hr/sqrt(bar), 60% open
network.addJunctionNode("Downstream-A");
network.addChoke("WH-A", "Downstream-A", "Choke-A",
    25.0,    // Kv [m3/hr/sqrt(bar)]
    60.0);   // opening [%]

// Adjust choke opening later
NetworkPipe choke = network.getPipe("Choke-A");
choke.setChokeOpening(75.0);  // Open to 75%

6.4.4 Choosing Kv Values

Typical $K_v$ ranges by application:

The relationship between nominal choke size and $K_v$ depends on the manufacturer and trim type (cage, plug, ball). Always use vendor data for accurate sizing.

---

6.5 Tubing VLP Model

6.5.1 Pressure Drop in Vertical Tubing

The tubing element models the vertical (or deviated) flow from bottomhole to wellhead. The pressure drop has three components:

where:

• $\rho_m$ = mixture density (depends on holdup)
• $g$ = gravitational acceleration [9.81 m/s²]
• $L$ = measured depth [m]
• $\alpha$ = inclination from horizontal [degrees]
• $f$ = Moody friction factor
• $D$ = tubing ID [m]
• $v$ = mixture velocity [m/s]

6.5.2 Segmented Approach for Deep Wells

For deep wells (> 1000 m), the fluid properties change significantly with pressure and temperature along the tubing. The tubing model uses a segmented approach:

1. Divide the tubing into $N$ segments of equal length $\Delta L = L/N$
2. At each segment inlet, flash the fluid to get local properties ($\rho_m$, $\mu$, holdup)
3. Calculate the pressure drop across the segment
4. Update pressure and temperature for the next segment
5. Sum all segment pressure drops for the total $\Delta P$

Typical segment counts:

6.5.3 Adding Tubing to the Network

// Vertical tubing: 2500 m depth, 4-inch ID, 20 segments
network.addTubing("BH-A", "WH-A", "Tubing-A",
    2500.0,    // measured depth [m]
    0.1016,    // ID [m] (4-inch)
    90.0);     // inclination from horizontal [degrees] (90 = vertical)

// Deviated well: 3500 m MD, 60° average inclination
network.addTubing("BH-B", "WH-B", "Tubing-B",
    3500.0,    // measured depth [m]
    0.1016,    // ID [m]
    60.0);     // inclination [degrees]

6.5.4 Temperature Profile

The geothermal gradient provides the formation temperature at each depth:

where $G_T$ is the geothermal gradient (typically 25–35 °C/km). The fluid temperature in the tubing is influenced by:

• Heat exchange with the formation through the tubing wall and cement
• Joule-Thomson cooling as pressure decreases
• Frictional heating (usually negligible)

For most production wells, the fluid exits the wellhead at a temperature 20–50 °C below the bottomhole temperature, depending on flow rate, insulation, and well depth.

---

6.6 Building Complete Well Networks

6.6.1 Step-by-Step Network Assembly Pattern

Building a production network follows a consistent pattern:

1. Create fluid template — defines the composition for the entire network
2. Add source nodes — reservoir contact points with fixed pressure
3. Add junction nodes — bottomholes, wellheads, manifolds
4. Add sink nodes — delivery points (platform, FPSO)
5. Add elements — IPR, tubing, chokes, flowlines
6. Configure solver — type, tolerance, max iterations
7. Run and inspect — solve and extract results

6.6.2 Multi-Well Gathering Network Example

Consider a subsea field with three wells tied back to a platform:

      Reservoir A (250 bar)          Reservoir B (220 bar)          Reservoir C (200 bar)
           |                              |                              |
      [IPR: PI=15]                  [IPR: Vogel]                  [IPR: Fetkovich]
           |                              |                              |
         BH-A                           BH-B                           BH-C
           |                              |                              |
      [Tubing 2500m]               [Tubing 3000m]               [Tubing 2000m]
           |                              |                              |
         WH-A                           WH-B                           WH-C
           |                              |                              |
      [Choke Kv=25]                [Choke Kv=30]                [Choke Kv=20]
           |                              |                              |
         Down-A                         Down-B                         Down-C
           |                              |                              |
      [Flowline 5km]              [Flowline 8km]              [Flowline 3km]
           \                              |                              /
            \                             |                             /
             +------------- Subsea Manifold -------------------------+
                                          |
                                  [Riser 1.5km]
                                          |
                                    Platform Inlet

import neqsim.process.equipment.network.LoopedPipeNetwork;
import neqsim.process.equipment.network.LoopedPipeNetwork.*;
import neqsim.thermo.system.SystemInterface;
import neqsim.thermo.system.SystemSrkEos;

// Step 1: Fluid template
SystemInterface fluid = new SystemSrkEos(273.15 + 80.0, 200.0);
fluid.addComponent("nitrogen", 0.5);
fluid.addComponent("CO2", 1.5);
fluid.addComponent("methane", 72.0);
fluid.addComponent("ethane", 8.0);
fluid.addComponent("propane", 4.5);
fluid.addComponent("i-butane", 1.0);
fluid.addComponent("n-butane", 2.0);
fluid.addComponent("n-pentane", 1.5);
fluid.addComponent("n-hexane", 1.0);
fluid.addComponent("n-heptane", 4.0);
fluid.addComponent("n-octane", 2.5);
fluid.addComponent("water", 1.5);
fluid.setMixingRule("classic");
fluid.setMultiPhaseCheck(true);

LoopedPipeNetwork network = new LoopedPipeNetwork("Subsea Field");
network.setFluidTemplate(fluid);

// Step 2: Source nodes (reservoirs)
network.addSourceNode("Res-A", 250.0, 0.0);
network.addSourceNode("Res-B", 220.0, 0.0);
network.addSourceNode("Res-C", 200.0, 0.0);

// Step 3: Junction nodes
network.addJunctionNode("BH-A");
network.addJunctionNode("BH-B");
network.addJunctionNode("BH-C");
network.addJunctionNode("WH-A");
network.addJunctionNode("WH-B");
network.addJunctionNode("WH-C");
network.addJunctionNode("Down-A");
network.addJunctionNode("Down-B");
network.addJunctionNode("Down-C");
network.addJunctionNode("Subsea-Manifold");

// Step 4: Sink node (platform)
network.addSinkNode("Platform", 0.0);
NetworkNode platform = network.getNode("Platform");
platform.setPressure(35.0e5);  // 35 bara back-pressure
platform.setPressureFixed(true);

// Step 5: Elements — IPRs
network.addWellIPR("Res-A", "BH-A", "IPR-A", 15.0 * 800.0 / 86400.0 / 1e5, false);
network.addWellIPRVogel("Res-B", "BH-B", "IPR-B", 80.0);
network.addWellIPRFetkovich("Res-C", "BH-C", "IPR-C", 2.0e-8, 0.80);

// Step 5: Elements — Tubing
network.addTubing("BH-A", "WH-A", "Tubing-A", 2500.0, 0.1016, 90.0);
network.addTubing("BH-B", "WH-B", "Tubing-B", 3000.0, 0.1016, 75.0);
network.addTubing("BH-C", "WH-C", "Tubing-C", 2000.0, 0.1016, 90.0);

// Step 5: Elements — Chokes
network.addChoke("WH-A", "Down-A", "Choke-A", 25.0, 80.0);
network.addChoke("WH-B", "Down-B", "Choke-B", 30.0, 70.0);
network.addChoke("WH-C", "Down-C", "Choke-C", 20.0, 90.0);

// Step 5: Elements — Gathering flowlines (multiphase)
NetworkPipe fl_a = network.addPipe("Down-A", "Subsea-Manifold",
    "Flowline-A", 5000.0, 0.2032);
fl_a.setElementType(NetworkElementType.MULTIPHASE_PIPE);
fl_a.setMultiphaseSegments(15);

NetworkPipe fl_b = network.addPipe("Down-B", "Subsea-Manifold",
    "Flowline-B", 8000.0, 0.2032);
fl_b.setElementType(NetworkElementType.MULTIPHASE_PIPE);
fl_b.setMultiphaseSegments(20);

NetworkPipe fl_c = network.addPipe("Down-C", "Subsea-Manifold",
    "Flowline-C", 3000.0, 0.2032);
fl_c.setElementType(NetworkElementType.MULTIPHASE_PIPE);
fl_c.setMultiphaseSegments(10);

// API assembly only: no elevation profile is supplied for this nominal riser.
// Step 5: Elements — Riser
NetworkPipe riser = network.addPipe("Subsea-Manifold", "Platform",
    "Production Riser", 1500.0, 0.254);
riser.setElementType(NetworkElementType.MULTIPHASE_PIPE);
riser.setMultiphaseSegments(15);

// Step 6: Solver configuration
network.setSolverType(SolverType.NEWTON_RAPHSON);
network.setTolerance(1e-6);
network.setMaxIterations(100);

// Step 7: Solve
network.run();

// Step 7: Results
Map<String, Object> summary = network.getSolutionSummary();
logger.info("Converged: " + summary.get("converged"));
logger.info("Iterations: " + summary.get("iterations"));
logger.info("Total production: {} kg/s", network.getTotalSinkFlow());

6.6.3 Results Inspection

The LoopedPipeNetwork provides several report methods:

Solution summary — key convergence and flow metrics:

Map<String, Object> summary = network.getSolutionSummary();
// Keys: converged, iterations, residualNorm, totalSourceFlow,
//       totalSinkFlow, maxVelocity, maxErosionalRatio

Hydraulic report — detailed pressure and flow for every element:

logger.info("Hydraulic summary: {}", network.getSolutionSummary());

Mass balance report — verifies conservation of mass:

logger.info("Mass balance error: {} kg/s", network.getMassBalanceError());

Individual element results:

// Get pressure at any node
double whPressure = network.getNodePressure("WH-A");  // bara

// Get flow rate through any element
double pipeFlow = network.getPipeFlowRate("Flowline-A") / 3600.0;  // kg/s

// Get element details
NetworkPipe pipe = network.getPipe("Flowline-A");
double velocity = pipe.getVelocity();        // m/s
double reynolds = pipe.getReynoldsNumber();  // dimensionless
double holdup = pipe.getLiquidHoldup();       // fraction

---

6.7 Artificial Lift in NeqSim Networks

6.7.1 Gas Lift

Physical gas lift mixes injected gas with production, changing density, friction, temperature and total mass. In this network's legacy gas-lift switch, however, the pressure benefit is only a capped loss proxy:

It does not add injection gas to the material balance. Use Chapter5's explicit gas mixer/tubing calculation for injection sizing or an economic gas-lift optimum. The API accepts its screening lift-rate input in kg/hr:

// Apply gas lift to Well A: 5000 kg/hr of lift gas
network.setGasLift("Tubing-A", 5000.0);  // kg/hr

// Solve again to see the effect
network.run();

A physical production gain must be established from the coupled IPR, tubing, injection pressure and downstream constraints. Do not interpret a gain from this capped proxy as an empirical50–200% recovery prediction.

6.7.2 Electric Submersible Pump (ESP)

An ESP adds a pressure boost at the pump intake depth, effectively lowering the flowing bottomhole pressure seen by the reservoir:

where the ESP head rise depends on the pump characteristic curve, flow rate, and speed:

In the network:

// ESP on Well B: 150 kW rated power, 55% efficiency
network.setESP("Tubing-B", 150.0, 0.55);  // power [kW], efficiency

6.7.3 Jet Pump

A jet pump (hydraulic lift) uses high-pressure power fluid to create a low-pressure zone that draws in production fluid. In the network model, it is represented as an equivalent pressure boost:

// Jet pump on Well C: 100 kW equivalent, 40% efficiency
network.setJetPump("Tubing-C", 100.0, 0.40);

6.7.4 Rod Pump

Rod pumps are positive displacement devices that maintain a nearly constant flow rate. In the network model, the rod pump provides a pressure boost similar to ESP:

// Rod pump: 30 kW, 50% efficiency
network.setRodPump("Tubing-A", 30.0, 0.50);

6.7.5 Effect on the NR-GGA Solver

Artificial lift modifies the element's $\Delta P(Q)$ function by adding a negative head loss (pressure gain). The solver handles this naturally — the Jacobian simply includes the derivative of the combined gravity + friction + lift term. This means:

• All artificial lift methods are fully coupled with the network solution
• Changing lift parameters on one well affects all other wells through the shared back-pressure
• The optimizer (Section 6.9) can optimize lift allocation across the entire field

---

6.8 Water Handling and Flow Assurance

6.8.1 Water Cut Tracking

Each tubing or flowline element can carry a water cut that affects density, viscosity, holdup, and corrosion calculations:

// Set water cut for Well A's tubing
NetworkPipe tubing_a = network.getPipe("Tubing-A");
tubing_a.setWaterCut(0.30);  // 30% water cut

The water cut affects the multiphase flow calculation through:

• Mixture density: Higher water cut increases liquid density, increasing hydrostatic head
• Viscosity: Oil-water emulsion viscosity can be much higher than pure oil
• Holdup: Water changes the gas-liquid flow pattern transitions

6.8.2 Water Injection

For water injection wells, the network element flow direction is reversed — water flows from surface to the reservoir:

LoopedPipeNetwork variant = new LoopedPipeNetwork("Topology illustration");
variant.setFluidTemplate(fluid);
variant.addSourceNode("Platform", 250.0, 0.0);
// Independent injection topology illustration
variant.addJunctionNode("WI-Wellhead");
variant.addJunctionNode("WI-BH");
variant.addSourceNode("WI-Reservoir", 180.0, 0.0);
variant.addPipe("Platform", "WI-Wellhead", "WI-Flowline", 10000.0, 0.2032);
variant.addTubing("WI-Wellhead", "WI-BH", "WI-Tubing", 3000.0, 0.1778, 90.0);
variant.addWellIPR("WI-BH", "WI-Reservoir", "WI-IPR", 25.0 * 1000.0 / 86400.0 / 1e5, false);

6.8.3 Sand Production Tracking

Sand production causes erosion in chokes, bends, and flowlines. The DNV RP O501 erosion model calculates material loss rate based on particle velocity, impact angle, and particle properties:

where:

• $E$ = erosion rate [mm/year]
• $K$ = material constant
• $F(\alpha)$ = impact angle function
• $v_p$ = particle velocity [m/s]
• $n$ = velocity exponent (typically 2.6 for steel)
• $\dot{m}_p$ = sand mass flow rate [kg/s]
• $\rho_t$ = target material density [kg/m³]
• $A_t$ = target area [m²]

// Set sand rate for a well's tubing
NetworkPipe tubing = network.getPipe("Tubing-A");
tubing.setSandRate(0.001);  // kg/s of sand

6.8.4 Corrosion Models

Two industry-standard CO₂ corrosion models are available:

de Waard-Milliams (1975):

where $v_{corr}$ is the corrosion rate [mm/year], $T$ is temperature [K], and $p_{CO_2}$ is CO₂ partial pressure [bar].

NORSOK M-506:

A temperature- and pH-dependent model that accounts for protective scale formation:

These models are integrated with the network flow solution, using the local temperature, pressure, and CO₂ content at each element.

6.8.5 GHG Emissions Tracking

The network can estimate greenhouse gas emissions associated with production operations:

Map<String, double[]> emissions = network.calculateEmissions();
// Compressor combustion emissions only; flaring, venting and fugitives need separate models.

This is particularly relevant for:

• Gas lift compressor emissions: Fuel gas consumption for compression
• ESP power emissions: Electricity generation for downhole pumps
• Flaring: Gas that cannot be processed
• Fugitive emissions: Leaks from valves, flanges, and seals

---

6.9 Network Optimization

6.9.1 Choke Sensitivity Study

Before running formal optimization, a choke sensitivity study reveals how each well responds to choke changes:

// Small, independently bounded network for the optimization tutorial.
// The detailed four-well multiphase calculation is demonstrated in Python below.
LoopedPipeNetwork study = new LoopedPipeNetwork("Choke optimization tutorial");
SystemSrkEos studyFluid = new SystemSrkEos(303.15, 90.0);
studyFluid.addComponent("methane", 0.95);
studyFluid.addComponent("ethane", 0.05);
studyFluid.setMixingRule("classic");
study.setFluidTemplate(studyFluid);
study.addSourceNode("Source-A", 90.0, 0.0);
study.addSourceNode("Source-B", 85.0, 0.0);
study.addJunctionNode("Down-A");
study.addJunctionNode("Down-B");
study.addSinkNode("Plant", 0.0);
study.getNode("Plant").setPressure(40.0e5);
study.getNode("Plant").setPressureFixed(true);
study.addChoke("Source-A", "Down-A", "Choke-A", 20.0, 80.0);
study.addChoke("Source-B", "Down-B", "Choke-B", 15.0, 80.0);
study.getPipe("Choke-A").setChokeUseValveModel(true);
study.getPipe("Choke-B").setChokeUseValveModel(true);
study.setRelaxationFactor(1.0);
study.addPipe("Down-A", "Plant", "Line-A", 5000.0, 0.2032);
study.addPipe("Down-B", "Plant", "Line-B", 4000.0, 0.2032);
study.setSolverType(SolverType.NEWTON_RAPHSON);
study.setMaxIterations(200);
study.setTolerance(100.0);  // Pa = 0.001 bar for this teaching solve
double[] openings = {40.0, 60.0, 80.0, 100.0};
for (double opening : openings) {
    study.getPipe("Choke-A").setChokeOpening(opening);
    study.run();
    double residualPa = study.getMaxResidual();
    double massErrorKgS = Math.abs(study.getMassBalanceError());
    double sinkKgS = study.getTotalSinkFlow();
    String chokeStatus = study.getPipe("Choke-A").getChokeModelStatus();
    if (!study.isConverged() || !Double.isFinite(residualPa)
        || residualPa > 100.0 || massErrorKgS > 1.0e-6
        || !Double.isFinite(sinkKgS) || sinkKgS <= 0.0
        || !chokeStatus.startsWith("IEC_GAS_")) {
        throw new IllegalStateException("Rejected hydraulic operating point");
    }
    double downAPa = study.getNode("Down-A").getPressure();
    double downBPa = study.getNode("Down-B").getPressure();
    if (!(40.0e5 < downAPa && downAPa < 90.0e5
        && 40.0e5 < downBPa && downBPa < 85.0e5)) {
        throw new IllegalStateException("Invalid passive-network pressure ordering");
    }
    logger.info("Accepted: residual {} Pa, mass error {} kg/s, status {}",
        residualPa, massErrorKgS, chokeStatus);
    logger.info("Opening {}%, flow {} kg/hr, converged {}", opening,
        study.getPipeFlowRate("Line-A"), study.getSolutionSummary().get("converged"));
    logger.info("Hydraulic residual {} Pa", study.getSolutionSummary().get("maxResidual_Pa"));
}

For this fixed-source-pressure example, the gas chokes remain in the supported critical-flow regime. Total plant flow at 40%,60%,80% and 100% opening is6.11035,7.37624,8.64213 and 9.90802 kg/s. The largest hydraulic residual is11.45 Pa and the mass residual is below 1e-6 kg/s. The approximately linear increase is the expected critical-capacity response; this model contains no reservoir IPR and cannot demonstrate a reservoir-limited plateau. The opt-in forward gas-capacity model is essential: the legacy head-loss choke approximation can report hydraulic convergence while flagging unsupported critical flow.

6.9.2 NetworkOptimizer with BOBYQA and CMA-ES

Optimizer termination and hydraulic convergence are separate checks. Reject any candidate whose network residuals fail even if the optimizer reports success. The examples print these statuses explicitly; candidate rates are not accepted operating recommendations.

The NetworkOptimizer class provides formal NLP optimization using derivative-free methods:

BOBYQA (Bound Optimization BY Quadratic Approximation) is ideal for smooth, bounded optimization with 2–20 decision variables:

import neqsim.process.equipment.network.NetworkOptimizer;
NetworkOptimizer optimizer = study.createOptimizer();
optimizer.setMaxEvaluations(50);
NetworkOptimizer.OptimizationResult result = optimizer.optimize();
study.run();
logger.info("Optimizer converged {}, hydraulics converged {}, message {}",
    result.converged, study.getSolutionSummary().get("converged"), result.message);
logger.info("Candidate production {} kg/hr, choke openings {}",
    result.totalProductionKgHr, Arrays.toString(result.chokeOpenings));

CMA-ES (Covariance Matrix Adaptation Evolution Strategy) is a global optimizer for non-convex or multi-modal problems:

NetworkOptimizer optimizer = study.createOptimizer();
optimizer.setAlgorithm(NetworkOptimizer.Algorithm.CMAES);
optimizer.setDeterministicSeed(2026L);
optimizer.setMaxEvaluations(60);
NetworkOptimizer.OptimizationResult result = optimizer.optimize();
logger.info("CMA-ES search converged {}, unvalidated candidate {} kg/hr: {}",
    result.converged, result.totalProductionKgHr, result.message);

6.9.3 Multi-Objective Choke Allocation

Production optimization often involves competing objectives — maximize oil while minimizing water, or maximize rate while minimizing erosion. The optimizeMultiObjective() method evaluates weighted production and compressor-power objectives. Its candidates still need independent feasibility and non-dominance checks. In the unpowered tutorial network below, all compressor powers are zero and the trade-off is degenerate:

NetworkOptimizer optimizer = study.createOptimizer();
optimizer.setParetoPoints(3);
optimizer.setMaxEvaluations(60);
List<NetworkOptimizer.OptimizationResult> pareto = optimizer.optimizeMultiObjective();
for (NetworkOptimizer.OptimizationResult point : pareto) {
    logger.info("Unvalidated candidate {} kg/hr, compressor power {} kW, weight {}",
        point.totalProductionKgHr, point.totalCompressorPowerKW, point.paretoWeight);
}
// This unpowered teaching network has no production/power trade-off.
// Introduce a calibrated compressor and constraints before interpreting a Pareto front.

6.9.4 Large-Scale Network Performance

The NR-GGA solver with Schur complement reduction scales well to large networks:

Solve time depends on fluid flashes, segment count, coupling iterations, topology, hardware, and initialization. Record timing and residuals for the actual study; no sub-second performance guarantee follows from the solver algorithm alone.

The solver's efficiency comes from:

1. Schur complement reduces the system from $(N_e + N_n)$ to $N_n$ unknowns
2. Sparse matrix storage for the incidence matrix
3. Adaptive relaxation can reduce oversized steps; nonsmooth or infeasible cases can still fail
4. Warm starting reuses the previous solution for parameter sweeps

---

6.10 Python Implementation

6.10.1 Building a Multi-Well Network in Python

The following legacy multiphase network is an API and candidate-generation example, not an accepted engineering solution: its default choke relation does not support critical gas capacity, and the pipe named Riser has no elevation profile. The explicitly checked forward gas-choke network in Section6.9.1 is the accepted hydraulic example. Three segments per element keep this API example tractable. Before using it for operating decisions, refine the segments until pressures and rates stabilize, calibrate the IPR and multiphase correlation, and reject non-converged results. The short optimizer budget demonstrates candidate generation rather than guaranteeing an optimum.

import jpype
jneqsim = jpype.JPackage("neqsim")

# Import network classes
LoopedPipeNetwork = jneqsim.process.equipment.network.LoopedPipeNetwork
SolverType = LoopedPipeNetwork.SolverType
NetworkElementType = LoopedPipeNetwork.NetworkElementType

# Create fluid template
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 80.0, 200.0)
fluid.addComponent("methane", 75.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 4.0)
fluid.addComponent("n-butane", 2.0)
fluid.addComponent("n-pentane", 1.0)
fluid.addComponent("n-heptane", 5.0)
fluid.addComponent("n-octane", 3.0)
fluid.addComponent("water", 2.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

# Build network
network = LoopedPipeNetwork("Field Alpha")
network.setFluidTemplate(fluid)

# Define 4-well subsea network
wells = {
    "A": {"Pr": 260.0, "PI": 18.0, "depth": 2800.0, "FL_len": 6000.0, "choke_Kv": 30.0},
    "B": {"Pr": 240.0, "PI": 12.0, "depth": 3200.0, "FL_len": 4000.0, "choke_Kv": 25.0},
    "C": {"Pr": 210.0, "PI": 20.0, "depth": 2200.0, "FL_len": 8000.0, "choke_Kv": 35.0},
    "D": {"Pr": 230.0, "PI": 10.0, "depth": 3500.0, "FL_len": 5000.0, "choke_Kv": 20.0},
}

# Source and sink nodes
for name, w in wells.items():
    network.addSourceNode(f"Res-{name}", w["Pr"], 0.0)

network.addJunctionNode("Manifold")
network.addSinkNode("Platform", 0.0)
platform = network.getNode("Platform")
platform.setPressure(35.0e5)
platform.setPressureFixed(True)

# Build each well: IPR -> Tubing -> Choke -> Flowline
for name, w in wells.items():
    network.addJunctionNode(f"BH-{name}")
    network.addJunctionNode(f"WH-{name}")
    network.addJunctionNode(f"DS-{name}")

    # IPR
    network.addWellIPR(f"Res-{name}", f"BH-{name}", f"IPR-{name}",
                       w["PI"] * 800.0 / 86400.0 / 1e5, False)
    # Tubing
    network.addTubing(f"BH-{name}", f"WH-{name}", f"Tubing-{name}",
                      w["depth"], 0.1016, 90.0)
    network.getPipe(f"Tubing-{name}").setTubingSegments(3)
    # Choke
    network.addChoke(f"WH-{name}", f"DS-{name}", f"Choke-{name}",
                     w["choke_Kv"], 80.0)  # 80% open initially
    # Flowline
    fl = network.addPipe(f"DS-{name}", "Manifold",
                         f"FL-{name}", w["FL_len"], 0.2032)
    fl.setElementType(NetworkElementType.MULTIPHASE_PIPE)
    fl.setMultiphaseSegments(3)

# Riser
riser = network.addPipe("Manifold", "Platform", "Riser", 1200.0, 0.254)
riser.setElementType(NetworkElementType.MULTIPHASE_PIPE)
riser.setMultiphaseSegments(3)

# Solve
network.setSolverType(SolverType.NEWTON_RAPHSON)
network.setTolerance(1e-4)
network.setMaxIterations(100)
network.run()

# Print results
summary = network.getSolutionSummary()
print(f"Converged: {summary.get('converged')}")
print(f"Iterations: {summary.get('iterations')}")
print(f"Total production: {float(network.getTotalSinkFlow()):.2f} kg/s")
print()

print(f"{'Well':>6} {'BHP (bara)':>12} {'WHP (bara)':>12} {'Flow (kg/s)':>12}")
print("-" * 48)
for name in wells:
    bhp = network.getNodePressure(f"BH-{name}")
    whp = network.getNodePressure(f"WH-{name}")
    flow = network.getPipeFlowRate(f"Tubing-{name}") / 3600.0
    print(f"{name:>6} {bhp:12.1f} {whp:12.1f} {flow:12.2f}")

6.10.2 Choke Sensitivity Study with Plots

import matplotlib.pyplot as plt
import numpy as np
import jpype
jneqsim = jpype.JPackage("neqsim")

# (Assume network is already built as above)

# Choke sensitivity for each well
openings = np.array([40.0, 60.0, 80.0, 100.0])  # four teaching points; refine around active constraints
results = {name: [] for name in wells}

for name in wells:
    for opening in openings:
        # Set choke opening
        choke = network.getPipe(f"Choke-{name}")
        choke.setChokeOpening(float(opening))

        # Solve
        network.run()

        # Exclude unconverged candidates from the sensitivity curve.
        converged = bool(network.getSolutionSummary().get("converged"))
        flow = (network.getPipeFlowRate(f"Tubing-{name}") / 3600.0
                if converged else float("nan"))
        results[name].append(flow)

    # Reset to 80%
    network.getPipe(f"Choke-{name}").setChokeOpening(80.0)

# Plot
fig, ax = plt.subplots(figsize=(10, 6))
for name in wells:
    ax.plot(openings, results[name], 'o-', label=f"Well {name}")

ax.set_xlabel("Choke Opening (%)")
ax.set_ylabel("Production Rate (kg/s)")
ax.set_title("Choke Sensitivity Study — 4-Well Subsea Network")
ax.legend()
ax.grid(True, alpha=0.3)
plt.tight_layout()
plt.savefig("figures/choke_sensitivity.png", dpi=150, bbox_inches="tight")
plt.show()

6.10.3 Network Optimization in Python

import jpype
jneqsim = jpype.JPackage("neqsim")

NetworkOptimizer = jneqsim.process.equipment.network.NetworkOptimizer

# Quick optimization: maximize total production by adjusting all choke openings
optimizer = network.createOptimizer()
optimizer.setMaxEvaluations(30)
result = optimizer.optimize()

print(f"Optimizer converged: {result.converged}; {result.message}")
print(f"Candidate total production: {float(result.totalProductionKgHr) / 3600.0:.2f} kg/s")

# Get individual optimal choke settings
optimal_values = result.chokeOpenings
for i, name in enumerate(result.chokeNames):
    print(f"  {name}: {float(optimal_values[i]):.1f}%")

# Re-solve with optimal settings and report
network.run()
print(f"\nPost-optimization well rates:")
for name in wells:
    flow = network.getPipeFlowRate(f"Tubing-{name}") / 3600.0
    print(f"  Well {name}: {flow:.2f} kg/s")

---

6.11 Advanced Network Modeling Topics

6.11.1 Looped Topologies and Redundancy

Real gathering networks often include loops — redundant paths that provide operational flexibility and allow continued production if one flowline is shut for maintenance. The NR-GGA solver handles loops naturally through the simultaneous solution of nodal pressures and element flows.

Consider a subsea field with a ring-main gathering system:

         WH-A                    WH-B
          |                       |
     [Flowline A1]          [Flowline B1]
          |                       |
     Manifold-1 ─── [Crossover] ─── Manifold-2
          |                       |
     [Flowline A2]          [Flowline B2]
          |                       |
         WH-C                    WH-D

The crossover pipe creates a loop. In this configuration:

• If Flowline A1 is shut, Well A can still produce through the crossover to Manifold-2
• The flow distribution depends on the relative resistance of each path
• The solver automatically finds the flow split that satisfies mass balance and momentum equations

LoopedPipeNetwork variant = new LoopedPipeNetwork("Topology illustration");
variant.setFluidTemplate(fluid);
for (String node : Arrays.asList("Down-A", "Down-B", "Down-C")) { variant.addJunctionNode(node); }
// Add the explicitly named junctions before their connections.
for (String node : Arrays.asList("Down-D", "Manifold-1", "Manifold-2")) {
    variant.addJunctionNode(node);
}
variant.addPipe("Down-A", "Manifold-1", "FL-A1", 5000.0, 0.2032);
variant.addPipe("Down-B", "Manifold-2", "FL-B1", 4000.0, 0.2032);
variant.addPipe("Down-C", "Manifold-1", "FL-A2", 6000.0, 0.2032);
variant.addPipe("Down-D", "Manifold-2", "FL-B2", 3500.0, 0.2032);

// Crossover creates the loop
variant.addPipe("Manifold-1", "Manifold-2", "Crossover",
    2000.0, 0.1524);  // 6-inch tie-in

6.11.2 Compressor and Booster Stations

Subsea boosting and topside compression can be modeled as COMPRESSOR elements in the network:

LoopedPipeNetwork variant = new LoopedPipeNetwork("Topology illustration");
variant.setFluidTemplate(fluid);
// Compressor proxy: a liquid booster requires an actual pump model.
variant.addJunctionNode("Manifold");
variant.addJunctionNode("Boosted-Manifold");
NetworkPipe booster = variant.addPipe("Manifold", "Boosted-Manifold",
    "Subsea Booster", 10.0, 0.254);
booster.setElementType(NetworkElementType.COMPRESSOR);
booster.setCompressorEfficiency(0.72);
booster.setCompressorSpeed(4500.0);  // RPM

The compressor element adds energy to the flow (negative head loss), which increases the pressure at downstream nodes and enables higher production rates from distant wells.

6.11.3 Pressure Regulators

Pressure reducing valves (regulators) maintain a fixed downstream pressure regardless of upstream conditions. This is useful for modeling gas distribution networks and pressure let-down stations:

LoopedPipeNetwork variant = new LoopedPipeNetwork("Topology illustration");
variant.setFluidTemplate(fluid);
// Pressure regulator maintaining 20 bar downstream
variant.addSourceNode("HP-Header", 40.0, 0.0);
variant.addJunctionNode("LP-Header");
NetworkPipe reg = variant.addPipe("HP-Header", "LP-Header",
    "PRV-001", 1.0, 0.1016);
reg.setElementType(NetworkElementType.REGULATOR);
reg.setRegulatorSetPoint(20.0e5);  // 20 bara in Pa

6.11.4 Erosional Velocity Monitoring

The API RP 14E erosional velocity limits are checked automatically for each element:

Here the traditional $C$ values 100–150 use velocity in ft/s and density in lb/ft³. For SI inputs use $v_e=1.2193 C/\sqrt{\rho_m}$ with $v_e$ in m/s and $\rho_m$ in kg/m³. Selection of $C$ needs a service-specific basis; this correlation does not cover solids erosion or flow-induced vibration by itself. The network reports the erosional velocity ratio ($v_{actual}/v_{erosional}$) for each element, flagging any that exceed 1.0.

// Check erosional velocity after solving
for (String pipeName : network.getPipeNames()) {
    NetworkPipe pipe = network.getPipe(pipeName);
    double ratio = pipe.getErosionalVelocityRatio();
    if (ratio > 0.8) {
        logger.info(String.format("WARNING: %s erosional ratio = %.2f%n",
            pipeName, ratio));
    }
}

6.11.5 Temperature Tracking

The network tracks fluid temperature through each element. For adiabatic pipes, the temperature changes due to Joule-Thomson cooling (pressure drop) and geothermal heat exchange:

// Set ambient temperature and heat transfer for a subsea flowline
NetworkPipe flowline = network.getPipe("Flowline-A");
flowline.setAmbientTemperature(277.15);  // 4°C seabed
flowline.setOverallHeatTransferCoeff(5.0);  // W/m2K (insulated pipe)
// Overall U represents the selected insulation; geometric sizing is a separate calculation.

Temperature tracking is essential for:

• Hydrate risk assessment: Is the arrival temperature below the hydrate equilibrium temperature?
• Wax deposition: Is the pipe wall temperature below the wax appearance temperature?
• Separator inlet conditions: Does the facility receive fluid at the design temperature?

6.11.6 Handling Non-Convergence

When the NR-GGA solver fails to converge (typically due to infeasible operating conditions), several diagnostic strategies are available:

1. Check for infeasible demands: The total demand may exceed the available supply
2. Reduce initial pressure estimates: Large pressure differences between source and sink can cause divergence
3. Increase maximum iterations: Some stiff networks need 50–100 iterations
4. Use adaptive relaxation: Reduce the step size to stabilize convergence

// Diagnostic: check if the network is physically feasible
network.setMaxIterations(200);
network.setTolerance(1e-4);  // Relaxed tolerance first
network.run();

Map<String, Object> summary = network.getSolutionSummary();
boolean converged = (boolean) summary.get("converged");
if (!converged) {
    logger.info("Residual norm: " + summary.get("maxResidual_Pa"));
    logger.info("Check supply/demand balance and element sizing");
}

6.12 Comparison of Solver Types

The LoopedPipeNetwork provides three solver types, each with different strengths:

6.12.1 Hardy Cross

The original iterative method for looped networks. Corrects flows in each loop sequentially:

Advantages:

• Simple to understand and implement
• Robust for networks with few loops
• Each iteration improves the solution

Disadvantages:

• Slow convergence for large networks (linear convergence rate)
• Does not handle fixed-pressure sinks naturally
• Requires loop identification (DFS spanning tree)

Best for: Small networks (< 20 elements), educational purposes, water distribution networks.

6.12.2 Sequential Modular

Solves each element in sequence, propagating pressures from source to sink:

Advantages:

• Very fast for tree-topology networks (no loops)
• Easy to add new element types
• Matches the physical flow direction

Disadvantages:

• Cannot handle looped topologies
• Sensitive to element ordering
• No guarantee of convergence for complex networks

Best for: Simple well-to-manifold systems without loops or recycles.

6.12.3 Newton-Raphson (NR-GGA)

Simultaneous solution of all nodal pressures and element flows:

Advantages:

• Quadratic convergence rate (much faster for large networks)
• Handles loops, fixed pressures, and mixed boundary conditions naturally
• Schur complement reduction keeps the system small
• Adaptive relaxation can improve robustness but does not guarantee convergence

Disadvantages:

• Requires Jacobian computation (element derivatives)
• More complex implementation
• Single iteration is more expensive (but far fewer iterations needed)

Best for: Production networks (10+ wells), real-time optimization, Monte Carlo studies.

// Comparison example
network.setSolverType(SolverType.HARDY_CROSS);
long t1 = System.nanoTime();
network.run();
long hardyCrossTime = System.nanoTime() - t1;
int hcIter = network.getIterationCount();

network.setSolverType(SolverType.NEWTON_RAPHSON);
long t2 = System.nanoTime();
network.run();
long nrTime = System.nanoTime() - t2;
int nrIter = network.getIterationCount();

logger.info(String.format("Hardy Cross: %d iterations, %.1f ms%n",
    hcIter, hardyCrossTime / 1e6));
logger.info(String.format("Newton-Raphson: %d iterations, %.1f ms%n",
    nrIter, nrTime / 1e6));

---

6.13 Industrial Application Patterns

6.13.1 Daily Production Allocation

Many operators use network models for daily production allocation — distributing the available production capacity among wells to meet sales targets:

# Daily allocation workflow
target_total_mass_rate = 40.0  # kg/s total mixture; phase allocation is a separate calculation

# Step 1: Solve network at current choke settings
network.run()
current_total = float(network.getTotalSinkFlow())

# Step 2: If under-producing, open chokes; if over-producing, close
if current_total < target_total_mass_rate * 0.95:
    # Open most responsive wells first (highest dQ/d_opening)
    optimizer = network.createOptimizer()
    optimizer.setMaxEvaluations(30)
    result = optimizer.optimize()
    print(f"Optimized production: {float(result.totalProductionKgHr) / 3600.0:.1f} kg/s")
elif current_total > target_total_mass_rate * 1.05:
    print("Total-mixture target exceeded; a phase-aware allocation constraint is required.")

6.13.2 What-If Scenarios

Network models enable rapid evaluation of operational scenarios:

• Well shut-in: What happens to other wells if Well B is shut for workover?
• New well tie-in: What production gain from adding Well E?
• Flowline pigging: What is the transient effect of pigging Flowline C?
• Back-pressure increase: How much production is lost if manifold pressure rises 5 bar?

# What-if: shut in Well B
original_flow = network.getPipeFlowRate("Tubing-B") / 3600.0
original_opening = network.getPipe("Choke-B").getChokeOpening()
network.getPipe("Choke-B").setChokeOpening(0.0)  # Close choke
network.run()

# Other wells pick up some rate due to reduced back-pressure
for name in ["A", "C", "D"]:
    new_flow = network.getPipeFlowRate(f"Tubing-{name}") / 3600.0
    print(f"Well {name}: {new_flow:.2f} kg/s")

# Restore
network.getPipe("Choke-B").setChokeOpening(original_opening)
network.run()

6.13.3 Real-Time Production Optimization

For digital twin applications, the network model runs continuously with updated input data from the plant historian:

1. Read current well data: WHP, WHT, choke opening, gas lift rate (from SCADA/PI)
2. Update network model: Set measured values as boundary conditions
3. Solve: Find the current operating point
4. Compare: Model prediction vs. measured rates (model validation)
5. Optimize: Recommend choke adjustments to maximize total production
6. Deploy: Send optimal setpoints back to the DCS

A deployment must benchmark the complete data-to-recommendation cycle and define stale-data, non-convergence, and operator-approval handling before selecting its update interval.

---

Reproduced Calculation Results

These examples use the stated fluid recipes and operating assumptions. Curves represent NeqSim calculations unless a caption identifies an analytical illustration, assumed equipment map or synthetic data.

Choke Outlet P: choke outlet pressure spans 45.61–141.6 bara across the plotted cases. Manifold P: manifold arrival pressure spans 41.41–140.3 bara across the plotted cases. Gaps retain undefined phase quantities or hydraulic states that fail the stated operating boundary; they are not interpolated.

At imposed upstream flow, closing a fixed-Cv choke requires more pressure drop; sufficiently restricted openings may have no feasible positive outlet pressure. This pressure-mode experiment is not a prediction that the same well rate remains deliverable at every opening. Reject infeasible pressure solutions and use the coupled reservoir–tubing–network model when translating a choke opening into a production rate.

Wellhead P: pressure spans 180–280 bara across the plotted cases. Manifold Arrival P: pressure spans 177.9–279 bara across the plotted cases.

Each well’s pressure budget depends on its inlet condition and the resistance of its branch to the common manifold. The well with the highest wellhead pressure is not necessarily the largest contributor once branch losses and common backpressure are included. Compare branch residuals and rates on one basis before adjusting chokes or assigning a debottlenecking priority.

Selected numerical ranges from the plotted cases:

Ranges describe the sampled cases; they are not independent validation tolerances.

6.14 Summary

Key points from this chapter:

• Production well networks cannot be optimized well-by-well — the hydraulic coupling through shared manifolds and back-pressure means that system-level modeling is essential.
• LoopedPipeNetwork provides a generalized graph-based network model where each element can be a pipe, IPR, choke, tubing, multiphase flowline, compressor, or regulator.
• IPR models (PI, Vogel, Fetkovich) capture the reservoir-wellbore flow relationship with increasing fidelity. Choose based on available data and reservoir type.
• Choke modeling using IEC 60534-style equations with critical flow detection enables accurate well rate control simulation.
• Tubing VLP uses a segmented approach for accurate pressure drop calculation in deep, high-temperature wells.
• The NR-GGA solver reduces the network system through a Schur complement. Accept a candidate only after checking hydraulic residuals, nodal mass balance and every element’s supported model status; runtime alone does not establish suitability.
• Artificial lift switches provide screening pressure boosts. The current network gas-lift switch reduces computed loss using a capped gas/produced-mass ratio; it does not inject a material stream or predict the full hydrostatic/friction optimum. Use the explicit lift-gas mixer and tubing workflow of Chapter5 for component and energy balances.
• Water handling, sand, corrosion, and emissions are tracked at the element level, enabling flow assurance and environmental compliance assessment within the production network model.
• NetworkOptimizer provides BOBYQA and CMA-ES algorithms for single- and multi-objective production optimization.

---

Verification of the worked examples

The compact, supported gas-network example supplies the accepted hydraulic case: four choke openings converge with pressure residual below 100 Pa, mass residual below 10⁻⁶ kg/s and valid pressure/flow directions. Its larger legacy multiphase API candidates remain explicitly unaccepted when hydraulics do not converge. The WellFlow example checks the declared drawdown and bottomhole-pressure constraints; neither solver convergence nor a software constraint authorizes operation beyond field limits.[1]

The calculation and literal-code records are in verification/scientific_revision/ch06_manuscript_physics.json; the chapter scope and code hashes are indexed in foundations_review.json.

Exercises

1. Exercise 6.1: Build a 3-well subsea network in NeqSim with the following data: Well A (PI = 20, depth = 2500 m, flowline = 4 km), Well B (PI = 15, depth = 3000 m, flowline = 6 km), Well C (PI = 25, depth = 2000 m, flowline = 5 km). All wells produce to a common manifold at 40 bara. Use the NR-GGA solver to find the steady-state production rates.

1. Exercise 6.2: For the network in Exercise 6.1, perform a choke sensitivity study on Well A (sweep opening from 10% to 100%) while keeping Wells B and C at 80%. Plot the production rate of all three wells vs. Well A's choke opening. Explain why the other wells' rates change.

1. Exercise 6.3: Replace the PI IPR model on Well B with a Vogel IPR ($Q_{max}$ = 100 kg/s). Compare the well's production rate at 60%, 80%, and 100% choke opening with both IPR models. When does the Vogel model predict significantly different results?

1. Exercise 6.4: Add gas lift to Well C at rates of 0, 2000, 4000, 6000, and 8000 kg/hr. Plot the incremental oil production per unit of lift gas. Identify the economic optimum assuming gas costs 0.10 USD/kg and oil sells for 0.50 USD/kg.

1. Exercise 6.5: Use network.optimizeProductionNLP() to find the optimal choke settings for the 3-well network. Compare the optimized total production with: (a) all chokes at 100%, (b) all chokes at 50%, (c) equal production allocation. Report the improvement in %.

1. Exercise 6.6 (Advanced): Build a 10-well network with two manifolds connected by a looped gathering system. Set different reservoir pressures (180–280 bara) and PIs (5–30). Run the NR-GGA solver and compare convergence (iterations, time) with the Hardy Cross solver. At what network size does NR-GGA become clearly superior?

1. Exercise 6.7 (Advanced): Implement a complete production optimization workflow in Python: build the network, run choke sensitivity for all wells, optimize with BOBYQA, generate a Pareto front (production vs. water), and plot all results. Export the optimal choke settings as a JSON file.

---

1. Todini, E., & Pilati, S. (1988). A gradient algorithm for the analysis of pipe networks. In B. Coulbeck & C. H. Orr (Eds.), Computer Applications in Water Supply, Vol. 1: Systems Analysis and Simulation (pp. 1–20). Research Studies Press.
2. Vogel, J. V. (1968). Inflow performance relationships for solution-gas drive wells. Journal of Petroleum Technology, 20(1), 83–92.
3. Fetkovich, M. J. (1973). The isochronal testing of oil wells. SPE Paper 4529, 48th Annual Fall Meeting, Las Vegas.
4. Beggs, H. D., & Brill, J. P. (1973). A study of two-phase flow in inclined pipes. Journal of Petroleum Technology, 25(5), 607–617.
5. IEC 60534-2-1 (2011). Industrial-process control valves — Part 2-1: Flow capacity — Sizing equations for fluid flow under installed conditions.
6. DNV RP O501 (2015). Managing sand production and erosion. Det Norske Veritas.
7. NORSOK M-506 (2017). CO₂ corrosion rate calculation model. Standards Norway.
8. de Waard, C., & Milliams, D. E. (1975). Carbonic acid corrosion of steel. Corrosion, 31(5), 177–181.
9. Powell, M. J. D. (2009). The BOBYQA algorithm for bound constrained optimization without derivatives. Technical Report DAMTP 2009/NA06, University of Cambridge.
10. Hansen, N. (2006). The CMA evolution strategy: A comparing review. In J. A. Lozano et al. (Eds.), Towards a New Evolutionary Computation (pp. 75–102). Springer.
11. Brill, J. P., & Mukherjee, H. (1999). Multiphase Flow in Wells. SPE Monograph Series, Vol. 17.
12. Economides, M. J., Hill, A. D., Ehlig-Economides, C., & Zhu, D. (2013). Petroleum Production Systems (2nd ed.). Prentice Hall.

Running the examples. Start the source-workspace Python session described in Chapter 1, then run this chapter's Python blocks in reading order. Java blocks form a separate sequence using the same NeqSim build; carry forward objects from preceding Java blocks. The release execution records are in verification/; a successful run establishes API compatibility, while physical validation also requires the checks discussed in the text.

Learning Objectives

After reading this chapter, the reader will be able to:

1. Describe the major components of a subsea production system and their functions
2. Explain the pressure and temperature constraints that govern subsea field development
3. Compare subsea vs. dry tree completion strategies for offshore fields
4. Model subsea wells and flowlines using NeqSim's subsea classes
5. Estimate SURF (Subsea, Umbilicals, Risers, Flowlines) costs for concept screening
6. Evaluate subsea processing options including boosting, separation, and compression
7. Optimize field layout by analyzing tieback distances and pressure budgets
8. Select between subsea field architecture configurations for different development scenarios

7.1 Introduction to Subsea Production

Subsea production systems enable the exploitation of offshore hydrocarbon reservoirs by placing wellheads, flow control equipment, and increasingly sophisticated processing equipment on the seabed. Since the first commercial subsea completion was installed in the Gulf of Mexico in 1961, the technology has evolved dramatically — modern subsea systems operate in water depths exceeding 3,000 meters with tieback distances of over 100 kilometers.

The fundamental advantage of subsea production is economic: rather than building a fixed or floating platform above every well cluster, subsea systems tie remote wells back to an existing host facility or an onshore plant. This shared infrastructure model dramatically reduces the capital investment required to develop marginal or remote fields.

However, subsea production introduces engineering challenges that do not arise with conventional dry tree completions:

• Pressure loss in long flowlines and risers reduces the natural driving force for production
• Temperature loss along the seabed exposes the produced fluid to hydrate, wax, and slug flow risks
• Intervention difficulty makes reliability and design margins critical — a failed subsea component may require a vessel costing \$500,000/day to repair
• Flow assurance challenges are more severe due to cold seabed temperatures and long transport distances

This chapter examines the architecture, design, and optimization of subsea production systems, with NeqSim providing the computational tools for pressure-temperature analysis, tieback feasibility studies, and cost estimation.

7.2 Subsea Architecture and Components

7.2.1 Subsea Christmas Trees

The subsea tree (also known as a Christmas tree or wet tree) is the primary well control device installed on the seabed. It sits atop the wellhead and provides:

• Flow control — master valves and wing valves regulate production
• Well shut-in — fail-safe closed valves for emergency and planned shutdowns
• Chemical injection points — for hydrate inhibitor, corrosion inhibitor, and scale inhibitor injection
• Monitoring — pressure and temperature sensors for well surveillance
• Intervention access — through-bore access for wireline or coiled tubing operations

Two main configurations exist:

These are conventional configurations; vendor variants and barrier requirements determine the actual installation and intervention sequence. Water depth, cost and pressure rating alone do not select a tree type.[15]

Tree Valve Functions

A subsea tree incorporates multiple valves in a specific arrangement that provides both flow regulation and well barrier functionality per NORSOK D-010:

• Master valves (upper and lower): Located in the vertical bore, these are the primary well barrier element. The lower master valve (LMV) is actuated hydraulically and is fail-safe closed — it closes automatically on loss of hydraulic supply. The upper master valve (UMV) provides redundancy.
• Wing valves (production and annulus): Located in the horizontal bore, the production wing valve (PWV) provides isolation; normal throttling is performed by the production choke. The annulus wing valve provides access to the annulus for pressure monitoring and chemical injection.
• Swab valve: Located at the top of the vertical bore, the swab valve allows wireline or coiled tubing access for well intervention. It is normally closed during production.
• Crossover valve: Connects the production bore to the annulus bore, enabling circulating operations during well cleanout or stimulation.
• Choke valve: An adjustable restriction downstream of the wing valve that controls the well flow rate. Modern subsea chokes are electrically actuated with position feedback for precise flow control.

The tubing head pressure (THP) — measured at the tree — is the key production surveillance parameter for subsea wells. It represents the pressure available to drive the fluid through the production system. The THP declines over field life as reservoir pressure depletes and water cut increases, and monitoring THP trends is essential for production optimization and well intervention planning:

where $P_{BHP}$ is the bottomhole flowing pressure, $\Delta P_{tubing,friction}$ is the frictional pressure drop in the tubing, $\rho_{fluid}$ is the average fluid density, $g$ is gravitational acceleration, and $h_{TVD}$ is the true vertical depth.

7.2.2 Subsea Manifolds

A subsea manifold collects production from multiple wells and routes it into a common flowline. Key design parameters include:

• Number of slots — typically 4 to 8 well connections
• Pressure rating — matched to the maximum shut-in tubing head pressure (SITHP)
• Bore size — determines flow capacity; typically 4" to 6" production bores
• Valve configuration — individual well isolation and crossover capability
• Weight and footprint — affects installation vessel requirements

Manifolds enable the commingling of production, reducing the number of flowlines required from the subsea field to the host. For example, a field with 12 wells and two 6-slot manifolds requires only 2 production flowlines rather than 12 individual well flowlines.

Header Sizing and Pressure Drop

The manifold production header must be sized to handle the commingled flow from all connected wells without excessive pressure drop. The header diameter is typically determined by limiting the erosional velocity per API RP 14E:

Here a field-unit coefficient such as $C=100$ uses density in lb/ft³ and produces velocity in ft/s; its SI-equivalent coefficient is about 122 when density is kg/m³ and velocity m/s. These coefficients are screening conventions, not a solids-erosion or corrosion model. For a 6-well manifold producing 10,000 bbl/d per well of a 35° API crude with GOR of 500 Sm³/Sm³, the header bore is typically 8"–10" nominal diameter.

The pressure drop through the manifold includes losses from header friction, branch connections, valves, and flow turns. A well-designed manifold contributes 1–3 bar of pressure drop — small relative to the flowline losses but significant when pressure budgets are tight.

Pigging Loops

In multi-well manifolds, pigging loops are incorporated to allow pipeline pigs to pass through the production header for wax removal, corrosion inhibitor distribution, and pipeline inspection. A pigging loop is a U-shaped section of pipe with pig launcher and receiver connections. The loop allows the pig to traverse the full length of the flowline from the manifold to the host, bypassing the manifold internals.

Pigging capability is essential for fields with waxy crudes or long tiebacks where regular wax management is required. Round-trip pigging (launching from the host and returning) requires additional infrastructure but enables intelligent pigging for corrosion assessment per DNV-RP-F116.

Manifold Layout Configurations

The arrangement of wells relative to manifolds defines the field topology:

The choice of manifold configuration depends on the number of wells, phasing of development (wells drilled over several years), seabed topology, and the need for future expansion.

7.2.3 Templates and Foundations

Subsea templates are structural frames installed on the seabed to guide the positioning of wells, trees, and manifolds. Templates provide:

• Well slot positioning — precise spacing for drilling and completion
• Structural support — for trees, manifolds, and jumpers
• Mudmat or suction anchor foundations — resist lateral and vertical loads
• Pipeline end terminations (PLETs) — connection points for flowlines

Template design must account for seabed soil conditions, current loads, installation tolerances, and the weight of installed equipment. A typical 4-slot template weighs 300–600 tonnes.

7.2.4 Jumpers and Connections

Jumpers are short pipe sections that connect subsea trees to manifolds, or manifolds to flowlines. They accommodate installation tolerances and thermal expansion:

• Rigid jumpers — steel pipe, typically 6"–10" diameter, with mechanical connectors
• Flexible jumpers — multi-layered flexible pipe, used where large deflections are expected
• Typical length — 20 to 100 meters

Connection systems include:

7.2.5 Umbilicals and Subsea Control Systems

Umbilicals are composite cables that deliver hydraulic fluid, electrical power, chemical injection, and communication signals from the host facility to subsea equipment. A modern umbilical typically contains:

• Hydraulic lines — for valve actuation (LP and HP supply), typically 2–4 tubes
• Electrical conductors — for power supply and signal transmission
• Chemical injection tubes — for MEG, methanol, corrosion inhibitor, scale inhibitor
• Fiber optic cables — for high-bandwidth communication and distributed sensing

Hydraulic pressure-wave propagation and actuator response are different quantities. The first scales roughly as $L/a$ using the wave speed of the fluid/tube system. Filling, draining, line compliance, fluid viscosity, restrictions and actuator volume govern the much longer stroke/charging time. Local accumulators and subsea control logic can supply the required action without waiting to charge the complete long line. Determine shutdown time from the actual actuator/control circuit and barrier function.

Electro-Hydraulic Control

The subsea control system is the nervous system of the production system. The dominant technology is the multiplexed electro-hydraulic (MUX E/H) system, which combines:

• Hydraulic power — separate pressure circuits serve the qualified actuator functions. Many production valves use hydraulic opening and spring closure; do not assume an HP supply universally powers fail-safe closure. Verify spring/accumulator energy, ambient pressure compensation, leakage and loss-of-supply behavior for each function.
• Electrical power — typically 600V AC or 3.3 kV AC for subsea distribution, powering subsea control modules (SCMs), choke actuators, condition monitoring sensors, and subsea electronic modules (SEMs).
• Signal transmission — digital communication over electrical conductors or fiber optics. Modern systems use fiber optics for high-bandwidth telemetry including condition monitoring data, distributed temperature sensing (DTS), and acoustic sand detection.

Each subsea tree has a subsea control module (SCM) that receives commands from the master control station (MCS) on the host facility and executes valve operations locally. The SCM contains redundant electronics (dual SEM cards) for reliability.

Power Delivery to Subsea Equipment

For subsea processing equipment (pumps, compressors, separators), the power requirements far exceed what conventional umbilicals can deliver. Power delivery options include:

• Medium-voltage AC (up to 36 kV): Suitable for subsea pumps up to ~15 MW at distances up to ~50 km
• High-voltage AC (66–145 kV): Required for large subsea compression systems (20–50 MW)
• High-voltage DC (HVDC): Emerging technology for ultra-long distances (>100 km), minimizing reactive power losses

Variable speed drives (VSDs) are located either topside (with the variable-frequency power transmitted through the umbilical) or subsea (with fixed-frequency power transmitted and converted locally). Subsea VSDs reduce cable losses but add subsea complexity.

Chemical Injection

Chemical injection through umbilical tubes is essential for flow assurance management:

The choice between methanol and MEG for hydrate inhibition is a major design decision. Methanol is simpler (no regeneration required) but consumed continuously, making it expensive for high water-cut or long-distance fields. MEG can be regenerated and recycled but requires a topside MEG reclamation unit.

The wells connect to a common manifold and a trunk line to the host. This is a hub layout, not a daisy-chain topology; line lengths, elevations and pressure-loss models are needed for hydraulic analysis.

7.3 Subsea Field Architectures

The spatial arrangement of wells, manifolds, flowlines, and risers defines the subsea field architecture. The choice of architecture profoundly affects capital cost, operability, reliability, and the ability to phase development over time.

7.3.1 Architecture Types

Satellite wells are individual subsea trees connected directly to the host facility by dedicated flowlines. This is the simplest architecture and is used for developments with 1–3 wells or wells that are widely spaced and cannot be economically grouped.

Cluster manifold architecture groups 4–8 wells around a central manifold, with the manifold connected to the host by one or two production flowlines and an umbilical. This is the most common configuration for medium-sized fields.

Daisy chain architecture connects wells or manifolds in series along a single flowline. Each well or manifold tees into the flowline, which runs from the most distant well to the host. This minimizes flowline length but means that all wells share a common flowline — a blockage or failure in the flowline affects all upstream wells.

Subsea to shore eliminates the offshore host facility entirely, routing subsea production through long-distance flowlines directly to an onshore processing plant. This architecture is used for gas fields near coastlines (e.g., Ormen Lange in Norway, 120 km tieback to Nyhamna).

Template-based architecture uses a drilling template that integrates well slots, manifold functions, and pipeline connections into a single structure. This is common in the North Sea where fields are developed with a large number of closely spaced wells.

7.3.2 Architecture Comparison

7.3.3 Selection Criteria

The choice of architecture is driven by:

• Reservoir geometry: Elongated reservoirs suit daisy-chain; compact reservoirs suit cluster manifolds
• Number of wells and phasing: Template architectures are efficient for large, fully defined developments; modular manifolds suit phased drilling campaigns
• Distance to host: Compare the extra flowlines, shared-capacity limits and intervention strategy; there is no universal 15 km economic cutoff
• Seabed terrain: Rough seabed or steep slopes may dictate flowline routing and manifold placement
• Flow assurance: High wax or hydrate risk may favor architectures with pigging capability (cluster manifold with pigging loop) over daisy chain (difficult to pig)
• Expansion potential: If additional wells may be drilled later, the architecture should accommodate future tie-ins

7.4 Subsea vs. Dry Tree Completions

The choice between subsea (wet tree) and dry tree completions is one of the most consequential decisions in offshore field development. It affects capital cost, operating cost, production efficiency, intervention frequency, and ultimately recovery factor.

7.4.1 Dry Tree Platforms

Dry tree completions place the wellheads on the platform deck, providing direct access for wireline, coiled tubing, and workover operations. Platform types that support dry trees include:

• Fixed platforms — jacket or gravity-based structures in shallow water (< 300 m)
• Compliant towers — for moderate water depths (300–600 m)
• Tension Leg Platforms (TLPs) — taut-moored for deepwater (500–1,500 m)
• Spars — deep-draft cylindrical hulls (500–2,500 m)

The primary advantage of dry trees is well access. Wireline and coiled tubing operations that cost \$200,000 and take 2 days from a platform may cost \$5,000,000 and take 3 weeks from a subsea intervention vessel.

7.4.2 Decision Criteria

7.4.3 Impact on Recovery Factor

Tree architecture can affect intervention access, backpressure and economically available wells. A universal 3–8% recovery uplift is unsupported; assess the same reservoir and development constraints before attributing a recovery difference to tree type. Potential mechanisms include:

• More frequent well intervention enables better reservoir management
• Easier installation of artificial lift (gas lift, ESPs)
• Lower wellhead back-pressure (no long flowlines)
• More production/injection wells can be economically justified

This recovery factor difference must be weighed against the capital cost difference, which often favors subsea for smaller fields or tiebacks to existing infrastructure.

7.5 Pressure and Temperature Constraints

7.5.1 The Pressure Budget

The pressure budget is the fundamental framework for subsea system design. It traces the available pressure from the reservoir to the first-stage separator:

where:

• $P_{res}$ is the average reservoir pressure [bara]
• $P_{sep}$ is the first-stage separator pressure [bara]
• $\Delta P_{IPR}$ is the pressure drop across the reservoir (inflow performance) [bar]
• $\Delta P_{tubing}$ is the tubing pressure loss from bottomhole to tree [bar]
• $\Delta P_{choke}$ is the choke pressure drop [bar]
• $\Delta P_{flowline}$ is the subsea flowline pressure loss [bar]
• $\Delta P_{riser}$ is the riser pressure loss (including hydrostatic head) [bar]

At the start of field life, when reservoir pressure is high, the pressure budget has significant margin. As the reservoir depletes, the available driving pressure decreases, and the flowline and riser pressure drops become an increasing fraction of the total. Eventually, the natural driving force is insufficient and artificial lift or subsea boosting is required.

Rearranging the pressure budget from the wellhead perspective, the pressure available to drive the fluid through the production system is the wellhead pressure minus the separator pressure:

This form is useful for tieback analysis: for a given wellhead pressure (determined by reservoir conditions and well performance), the maximum tieback distance is limited by the sum of flowline, riser, and choke pressure drops. As tieback distance increases, $\Delta P_{flowline}$ grows, leaving less margin for the choke and eventually requiring reduced flow rates or subsea boosting.

7.5.2 Temperature Constraints

Temperature management is equally critical in subsea systems:

• Hydrate formation — if the fluid temperature drops below the hydrate equilibrium temperature at the flowing pressure, solid hydrates can form and block the flowline
• Wax deposition — if the temperature drops below the wax appearance temperature (WAT), wax crystals nucleate and deposit on pipe walls
• Arrival temperature — the topside process requires a minimum arrival temperature for effective separation and hydrate prevention
• Cooldown time — during an unplanned shutdown, the fluid in the flowline cools toward the seabed temperature; the time to reach the hydrate formation temperature is the available time for remediation

The temperature profile along a subsea flowline depends on:

where:

• $T(x)$ is the fluid temperature at distance $x$ along the flowline [K]
• $T_{sea}$ is the seabed temperature [K]
• $T_{inlet}$ is the inlet temperature [K]
• $D$ is the pipe outer diameter [m]
• $U$ is the overall heat transfer coefficient [W/m²·K]
• $\dot{m}$ is the mass flow rate [kg/s]
• $C_p$ is the specific heat capacity [J/kg·K]

7.5.3 NeqSim Pressure-Temperature Analysis

NeqSim's PipeBeggsAndBrills class calculates both pressure and temperature profiles along subsea flowlines and risers:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define a typical subsea production fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 70.0, 200.0)
fluid.addComponent("nitrogen", 0.8)
fluid.addComponent("CO2", 3.5)
fluid.addComponent("methane", 65.0)
fluid.addComponent("ethane", 8.5)
fluid.addComponent("propane", 5.0)
fluid.addComponent("i-butane", 1.0)
fluid.addComponent("n-butane", 2.5)
fluid.addComponent("i-pentane", 1.0)
fluid.addComponent("n-pentane", 1.5)
fluid.addComponent("n-hexane", 2.0)
fluid.addComponent("n-heptane", 4.0)
fluid.addComponent("n-octane", 3.0)
fluid.addComponent("water", 2.2)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

# Create feed stream at wellhead conditions
Stream = jneqsim.process.equipment.stream.Stream
wellstream = Stream("Subsea Wellstream", fluid)
wellstream.setFlowRate(80000.0, "kg/hr")
wellstream.setTemperature(70.0, "C")
wellstream.setPressure(180.0, "bara")

# Model 15 km subsea flowline
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
flowline = PipeBeggsAndBrills("Subsea Flowline", wellstream)
flowline.setPipeWallRoughness(5.0e-5)
flowline.setLength(15000.0)           # m (15 km)
flowline.setElevation(0.0)         # horizontal
flowline.setDiameter(0.254)        # 10-inch ID in meters
flowline.setNumberOfIncrements(50)
flowline.setConstantSurfaceTemperature(277.15)  # 4°C seabed temperature

# Build and run
ProcessSystem = jneqsim.process.processmodel.ProcessSystem
process = ProcessSystem()
process.add(wellstream)
process.add(flowline)
process.run()

# Read arrival conditions
arrival_T = flowline.getOutletStream().getTemperature("C")
arrival_P = flowline.getOutletStream().getPressure("bara")
delta_P = wellstream.getPressure("bara") - arrival_P
print(f"Arrival temperature: {arrival_T:.1f} °C")
print(f"Arrival pressure: {arrival_P:.1f} bara")
print(f"Pressure drop: {delta_P:.1f} bar")

7.5.4 Complete Subsea Tieback Model

A comprehensive tieback model includes the well tubing, choke, subsea flowline, and riser — the full pressure path from bottomhole to topside separator. This NeqSim example models each segment:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define production fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 85.0, 280.0)
fluid.addComponent("nitrogen", 0.6)
fluid.addComponent("CO2", 2.8)
fluid.addComponent("methane", 62.0)
fluid.addComponent("ethane", 7.5)
fluid.addComponent("propane", 5.0)
fluid.addComponent("i-butane", 1.2)
fluid.addComponent("n-butane", 2.8)
fluid.addComponent("i-pentane", 1.0)
fluid.addComponent("n-pentane", 1.5)
fluid.addComponent("n-hexane", 2.5)
fluid.addComponent("n-heptane", 4.5)
fluid.addComponent("n-octane", 3.5)
fluid.addComponent("water", 5.1)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

Stream = jneqsim.process.equipment.stream.Stream
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
ThrottlingValve = jneqsim.process.equipment.valve.ThrottlingValve
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

# --- Bottomhole stream ---
bhp_stream = Stream("Bottomhole", fluid)
bhp_stream.setFlowRate(60000.0, "kg/hr")
bhp_stream.setTemperature(85.0, "C")
bhp_stream.setPressure(280.0, "bara")

# --- Well tubing (3500 m MD, ~3200 m TVD) ---
tubing = PipeBeggsAndBrills("Well Tubing", bhp_stream)
tubing.setPipeWallRoughness(2.5e-5)
tubing.setLength(3500.0)              # 3.5 km measured depth
tubing.setElevation(3200.0)       # positive = outlet above inlet
tubing.setDiameter(0.1143)         # 4.5-inch tubing ID
tubing.setNumberOfIncrements(30)
tubing.setConstantSurfaceTemperature(285.15) # geothermal average ~12 C

# --- Subsea choke ---
choke = ThrottlingValve("Subsea Choke", tubing.getOutletStream())
choke.setOutletPressure(110.0)     # below the calculated upstream wellhead pressure

# --- Subsea flowline (20 km horizontal on seabed) ---
flowline = PipeBeggsAndBrills("Subsea Flowline", choke.getOutletStream())
flowline.setPipeWallRoughness(5.0e-5)
flowline.setLength(20000.0)
flowline.setElevation(0.0)
flowline.setDiameter(0.254)        # 10-inch
flowline.setNumberOfIncrements(50)
flowline.setConstantSurfaceTemperature(277.15)  # 4 C seabed

# --- Riser (400 m water depth) ---
riser = PipeBeggsAndBrills("Production Riser", flowline.getOutletStream())
riser.setPipeWallRoughness(5.0e-5)
riser.setLength(500.0)               # ~500 m riser length
riser.setElevation(400.0)          # 400 m vertical rise
riser.setDiameter(0.254)
riser.setNumberOfIncrements(15)
riser.setConstantSurfaceTemperature(280.15)

riser.setHeatTransferCoefficient(5.0)  # assumed insulated-pipe U, W/(m2 K)
# --- Build and run ---
process = ProcessSystem()
process.add(bhp_stream)
process.add(tubing)
process.add(choke)
process.add(flowline)
process.add(riser)
process.run()

# --- Report pressure budget ---
P_bhp = bhp_stream.getPressure("bara")
P_wh = tubing.getOutletStream().getPressure("bara")
P_choke_out = choke.getOutletStream().getPressure("bara")
P_fl_out = flowline.getOutletStream().getPressure("bara")
P_topside = riser.getOutletStream().getPressure("bara")
T_topside = riser.getOutletStream().getTemperature("C")
assert 0.0 < P_choke_out <= P_wh, "A passive choke cannot add pressure"

print("=== Subsea Tieback Pressure Budget ===")
print(f"Bottomhole pressure:     {P_bhp:.1f} bara")
print(f"Wellhead pressure:       {P_wh:.1f} bara  (tubing dP = {P_bhp - P_wh:.1f} bar)")
print(f"After choke:             {P_choke_out:.1f} bara  (choke dP = {P_wh - P_choke_out:.1f} bar)")
print(f"Flowline outlet:         {P_fl_out:.1f} bara  (flowline dP = {P_choke_out - P_fl_out:.1f} bar)")
print(f"Topside arrival:         {P_topside:.1f} bara  (riser dP = {P_fl_out - P_topside:.1f} bar)")
print(f"Topside arrival temp:    {T_topside:.1f} C")

7.6 Tieback Distance Analysis

7.6.1 Maximum Tieback Distance

The maximum tieback distance is determined by the intersection of three constraints:

1. Pressure constraint — the available pressure drop in the flowline (total pressure budget minus tubing, choke, and riser losses) limits distance
2. Temperature constraint — the arrival temperature must remain above the hydrate equilibrium temperature (with safety margin) or the WAT
3. Cooldown constraint — the time for the fluid to cool below the hydrate temperature during a shutdown must exceed the minimum intervention time

For a given set of operating conditions, the maximum tieback distance is:

The pressure-limited distance can be estimated from the Beggs and Brill correlation or simplified as:

where $\Delta P_{available}$ is the pressure budget allocated to the flowline, $D$ is the diameter, $\lambda$ is the friction factor, $\rho_{mix}$ is the mixture density, and $v$ is the mixture velocity. At fixed velocity this relation is linear in diameter. At fixed volumetric rate, substitution of $v=4Q/(\pi D^2)$ gives a $D^5$ dependence if density and friction factor remain constant. Multiphase holdup, friction-factor changes, acceleration and elevation invalidate treating the resulting factor of 32 as a general design rule.

The temperature-limited distance is derived from the exponential temperature decay equation:

where $T_{min}$ is the minimum acceptable arrival temperature (hydrate equilibrium temperature plus safety margin).

7.6.2 Tieback Feasibility Study with NeqSim

The following tieback screen calculates arrival pressure and temperature across a range of flowline lengths at one fixed feed rate. A feasible tieback also requires a stated receiving-pressure requirement and separate flow-assurance and shutdown criteria; this table alone does not establish a maximum feasible distance.

import jpype
jneqsim = jpype.JPackage("neqsim")
import json

# Fluid definition (as above - reused from Section 7.5.3)
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 70.0, 200.0)
fluid.addComponent("nitrogen", 0.8)
fluid.addComponent("CO2", 3.5)
fluid.addComponent("methane", 65.0)
fluid.addComponent("ethane", 8.5)
fluid.addComponent("propane", 5.0)
fluid.addComponent("i-butane", 1.0)
fluid.addComponent("n-butane", 2.5)
fluid.addComponent("i-pentane", 1.0)
fluid.addComponent("n-pentane", 1.5)
fluid.addComponent("n-hexane", 2.0)
fluid.addComponent("n-heptane", 4.0)
fluid.addComponent("n-octane", 3.0)
fluid.addComponent("water", 2.2)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

Stream = jneqsim.process.equipment.stream.Stream
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

# Parametric study: vary tieback distance
distances_km = [5, 10, 15, 20, 30, 40, 50]
results = []

for dist in distances_km:
    test_fluid = fluid.clone()
    feed = Stream("Feed", test_fluid)
    feed.setFlowRate(80000.0, "kg/hr")
    feed.setTemperature(70.0, "C")
    feed.setPressure(180.0, "bara")

    pipe = PipeBeggsAndBrills("Flowline", feed)
    pipe.setPipeWallRoughness(5.0e-5)
    pipe.setLength(float(dist) * 1000.0)
    pipe.setElevation(0.0)
    pipe.setDiameter(0.254)
    pipe.setNumberOfIncrements(50)
    pipe.setConstantSurfaceTemperature(277.15)

    pipe.setHeatTransferCoefficient(5.0)  # assumed insulated-pipe U, W/(m2 K)
    proc = ProcessSystem()
    proc.add(feed)
    proc.add(pipe)
    proc.run()

    arrival_T = pipe.getOutletStream().getTemperature("C")
    arrival_P = pipe.getOutletStream().getPressure("bara")
    results.append({
        "distance_km": dist,
        "arrival_T_C": round(arrival_T, 1),
        "arrival_P_bara": round(arrival_P, 1),
        "dP_bar": round(180.0 - arrival_P, 1)
    })

# Print results table
print(f"{'Distance (km)':>14} {'Arrival T (°C)':>15} {'Arrival P (bara)':>17} {'dP (bar)':>10}")
for r in results:
    print(f"{r['distance_km']:>14} {r['arrival_T_C']:>15.1f} {r['arrival_P_bara']:>17.1f} {r['dP_bar']:>10.1f}")

7.6.3 Typical Tieback Distances

The following broad ranges are orientation examples, not demonstrated capability or acceptance limits. A particular tieback requires a field-specific thermal, hydraulic and shutdown assessment:

These distances assume typical pipe sizes (8"–14"), seabed temperatures of 2–5°C, and host separator pressures of 20–40 bara.

7.7 Subsea Processing

7.7.1 Motivation for Subsea Processing

Subsea processing addresses the declining pressure budget as reservoirs deplete. By placing processing equipment — pumps, separators, or compressors — on the seabed, the effective tieback distance is extended and production is maintained longer without topside modifications.

The key drivers for subsea processing are:

• Pressure boosting — adds energy to overcome flowline friction and riser hydrostatic head
• Separation — removes water or gas on the seabed to reduce flowline hydraulic load
• Gas compression — compresses gas to maintain delivery pressure as reservoir pressure declines
• Sand management — removes sand before it enters flowlines

7.7.2 Subsea Boosting (Multiphase Pumps)

Subsea multiphase pumps are the most widely deployed form of subsea processing. They boost the total well stream — gas, oil, and water — without prior separation. Key technologies include:

Helico-axial pumps are the dominant technology for subsea boosting. The Framo helico-axial pump uses a combination of an axial impeller and a helical inducer to handle gas–liquid mixtures. The impeller accelerates the fluid, and a diffuser converts kinetic energy to pressure. Multiple stages (typically 4–12) are stacked to achieve the required differential pressure. The key advantage is tolerance of high gas volume fractions (GVF up to 95%) without the slugging and vibration problems that affect conventional centrifugal pumps.

Twin-screw pumps are positive displacement machines that can cover a broad gas-volume-fraction range when designed for it. Sustained operation near 100% gas may require liquid recirculation for sealing and cooling; duration, temperature rise, differential pressure and sand loading must remain within the vendor envelope. Two intermeshing screws trap and transport fluid volumes along the screw axis. They are particularly suited for low-flow, high-differential-pressure applications and viscous fluids. However, they have lower volumetric capacity than helico-axial pumps and are more sensitive to sand erosion.

Vigdis multiphase pumping started in May2021. Åsgard and Gullfaks started different subsea compression technologies in 2015; they should not be listed as equivalent pump deployments.[16, 17]

7.7.3 Subsea Separation and Water Injection

Subsea separation separates the produced fluid into gas and liquid phases on the seabed. The separated streams are transported in dedicated flowlines:

• Gas — flows to the host at higher velocity but lower frictional loss per unit mass
• Liquid — can be boosted by a less complex liquid pump

The Tordis subsea separation system (installed 2007) demonstrated subsea separation and water reinjection, removing water on the seabed and reinjecting it into a disposal well, reducing the hydraulic load on the flowline and topside produced water treatment.

Subsea water separation and reinjection (SSBI) offers substantial benefits for high water-cut fields:

• Reduces the volume of fluid transported through flowlines by 30–60%
• Reduces topside produced water treatment requirements
• Enables reinjection for pressure support close to the producing wells
• Extends field life by reducing the hydraulic load that limits production rate

Design challenges for subsea separators include limited space for gravity settling (compact separators use cyclonic or pipe separator technology), sand handling, and the need for subsea water treatment before reinjection (to avoid reservoir plugging).

7.7.4 Subsea Wet Gas Compression

Åsgard began subsea gas compression in September2015 with gas/liquid separation before compression. Gullfaks began wet-gas compression in October2015 without that upstream phase separation. Their different liquid-handling architectures determine compressor qualification and auxiliary equipment. Åsgard phase2 was completed in 2025.[18, 17, 19]

The subsea compression station at Åsgard consists of two parallel compression trains, each with an inlet liquid knockout drum (subsea scrubber), a centrifugal compressor driven by a high-speed electric motor, an anti-surge recycle system, and a subsea cooler downstream of the compressor.

Wet gas compression differs fundamentally from dry gas compression:

Key design challenges include:

• Liquid handling in compressor impellers
• Anti-surge control without fast-acting recycle valves
• Long-distance electrical power supply (up to 60 MW at 36 kV)
• Reliability requirements (5+ year uninterrupted operation target)

7.7.5 Subsea Power Distribution

Powering subsea processing equipment requires a subsea electrical infrastructure that is itself a major engineering system:

• Subsea transformers: Step voltage up for long-distance transmission, step down for local distribution
• Subsea switchgear: Circuit breakers and contactors for equipment isolation and protection
• Subsea variable speed drives (VSD): Control pump and compressor speed for flow matching
• Subsea power cables: Medium- or high-voltage cables from host to subsea distribution

The total power demand for a subsea processing station can range from 5 MW (single pump) to 60 MW (compression and pumping), making power delivery one of the most challenging aspects of subsea processing.

7.7.6 Technology Readiness and Deployment

Technology qualification applies to a specific design and operating envelope. Use a dated qualification dossier and field-reference list; counts and generic TRLs for an entire technology family can hide differences in pressure, fluid, power, intervention and duty. The dated operator examples above establish deployment history, not blanket qualification for a new field.

7.7.7 Modeling Subsea Boosting with NeqSim

A screening estimate of subsea boosting can insert a pressure increase between the flowline and riser. This example uses a compressor as a thermodynamic pressure-raising proxy for the defined multiphase fluid. It is not a qualified wet-gas compressor or multiphase-pump performance model; detailed design needs phase handling, vendor maps, and an appropriate machine model:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 60.0, 150.0)
fluid.addComponent("methane", 72.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("n-butane", 2.5)
fluid.addComponent("n-pentane", 1.5)
fluid.addComponent("n-hexane", 2.0)
fluid.addComponent("n-heptane", 3.0)
fluid.addComponent("n-octane", 2.5)
fluid.addComponent("water", 3.5)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

Stream = jneqsim.process.equipment.stream.Stream
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
Compressor = jneqsim.process.equipment.compressor.Compressor
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

# Wellhead stream
wellstream = Stream("Wellstream", fluid)
wellstream.setFlowRate(60000.0, "kg/hr")
wellstream.setTemperature(60.0, "C")
wellstream.setPressure(120.0, "bara")

# Subsea flowline (30 km)
flowline = PipeBeggsAndBrills("Subsea Flowline", wellstream)
flowline.setPipeWallRoughness(5.0e-5)
flowline.setLength(30000.0)
flowline.setElevation(0.0)
flowline.setDiameter(0.254)
flowline.setNumberOfIncrements(50)
flowline.setConstantSurfaceTemperature(277.15)

flowline.setHeatTransferCoefficient(5.0)  # assumed insulated-pipe U, W/(m2 K)
# Subsea booster (modeled as a compressor for pressure increase)
booster = Compressor("Subsea Booster", flowline.getOutletStream())
booster.setOutletPressure(150.0)  # pressure increase above the flowline inlet
booster.setPolytropicEfficiency(0.75)
booster.setUsePolytropicCalc(True)

# Riser (500 m water depth, vertical)
riser = PipeBeggsAndBrills("Riser", booster.getOutletStream())
riser.setPipeWallRoughness(5.0e-5)
riser.setLength(600.0)       # ~600 m riser length
riser.setElevation(500.0)  # 500 m water depth
riser.setDiameter(0.254)
riser.setNumberOfIncrements(20)
riser.setConstantSurfaceTemperature(280.15)

riser.setHeatTransferCoefficient(5.0)  # assumed insulated-pipe U, W/(m2 K)
# Build process
process = ProcessSystem()
process.add(wellstream)
process.add(flowline)
process.add(booster)
process.add(riser)
process.run()

# Results
topside_P = riser.getOutletStream().getPressure("bara")
topside_T = riser.getOutletStream().getTemperature("C")
booster_power = booster.getPower("kW")
assert booster_power > 0.0, "A booster must consume positive shaft power"
print(f"Topside arrival pressure: {topside_P:.1f} bara")
print(f"Topside arrival temperature: {topside_T:.1f} °C")
print(f"Subsea booster power: {booster_power:.0f} kW")

7.8 Subsea Well Design with NeqSim

7.8.1 The SubseaWell Class

NeqSim provides the SubseaWell class for modeling subsea well completions, including mechanical design calculations per API 5C3 and NORSOK D-010:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Create a production fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 90.0, 350.0)
fluid.addComponent("methane", 70.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("n-butane", 2.0)
fluid.addComponent("n-pentane", 1.5)
fluid.addComponent("n-heptane", 5.0)
fluid.addComponent("n-octane", 3.5)
fluid.addComponent("water", 5.0)
fluid.setMixingRule("classic")

Stream = jneqsim.process.equipment.stream.Stream
stream = Stream("Well Stream", fluid)
stream.setFlowRate(50000.0, "kg/hr")
stream.setTemperature(90.0, "C")
stream.setPressure(250.0, "bara")
stream.run()

# Create subsea well
SubseaWell = jneqsim.process.equipment.subsea.SubseaWell
well = SubseaWell("Producer-1", stream)
well.setWellType(SubseaWell.WellType.OIL_PRODUCER)
well.setMeasuredDepth(3800.0)           # Total measured depth [m]
well.setWaterDepth(350.0)               # Water depth [m]
well.setMaxWellheadPressure(345.0)      # Maximum WHSP [bara]
well.setReservoirPressure(400.0)        # Initial reservoir pressure [bara]
well.setProductionCasingOD(9.625)       # 9-5/8" casing
well.setProductionCasingDepth(3800.0)   # Casing set depth [m]
well.setTubingOD(5.5)                   # 5-1/2" tubing
well.setTubingWeight(23.0)              # lb/ft
well.setTubingGrade("L80")             # API 5CT grade
well.setHasDHSV(True)                   # Downhole safety valve
well.setPrimaryBarrierElements(3)       # NORSOK D-010
well.setSecondaryBarrierElements(3)     # NORSOK D-010
well.setDrillingDays(45.0)             # Drilling duration
well.setCompletionDays(25.0)           # Completion duration

# Run mechanical design
well.initMechanicalDesign()
WellMechanicalDesign = jneqsim.process.mechanicaldesign.subsea.WellMechanicalDesign
design = well.getMechanicalDesign()
design.calcDesign()
design.calculateCostEstimate()

# Print results
json_output = design.toJson()
print(json_output)

7.8.2 Casing Design per API 5C3

The mechanical design calculation follows API Bull 5C3 for casing strength:

Burst rating:

Collapse rating (for the yield strength collapse regime):

Tension rating:

where:

• $Y_p$ is the minimum yield strength [psi]
• $t$ is the wall thickness [in]
• $D_o$ is the outer diameter [in]
• $A_s$ is the cross-sectional area of the pipe body [in²]

Illustrative screening design-factor inputs are listed below; they have not been traced to a NORSOK D-010 clause and must not be used as a compliance table. Actual casing assessment includes load combinations, temperature, corrosion allowance, connection capacity and the applicable standard edition:

7.9 SURF Cost Estimation

7.9.1 Cost Components

SURF (Subsea, Umbilicals, Risers, Flowlines) costs typically represent 30–50% of the total development cost for a subsea field. The main cost components are:

7.9.2 Cost Estimation Approach

NeqSim's WellMechanicalDesign class includes parametric cost estimation for subsea wells. For the complete SURF cost, a structured estimation approach combines well costs with infrastructure costs:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Subsea field parameters
n_wells = 6
n_manifolds = 2
flowline_length_km = 20.0
umbilical_length_km = 22.0
riser_length_m = 400.0
water_depth_m = 350.0

# Parametric cost estimates (MNOK, 2024 basis)
tree_cost = 25.0      # per tree
manifold_cost = 40.0   # per manifold
flowline_cost_per_km = 15.0   # includes installation
umbilical_cost_per_km = 8.0
riser_cost_per_m = 0.005      # flexible riser

# Calculate SURF cost
surf_cost = (
    n_wells * tree_cost +
    n_manifolds * manifold_cost +
    flowline_length_km * flowline_cost_per_km +
    umbilical_length_km * umbilical_cost_per_km +
    riser_length_m * riser_cost_per_m * 2  # production + gas lift
)

print(f"Subsea trees:   {n_wells * tree_cost:.0f} MNOK")
print(f"Manifolds:      {n_manifolds * manifold_cost:.0f} MNOK")
print(f"Flowlines:      {flowline_length_km * flowline_cost_per_km:.0f} MNOK")
print(f"Umbilicals:     {umbilical_length_km * umbilical_cost_per_km:.0f} MNOK")
print(f"Risers:         {riser_length_m * riser_cost_per_m * 2:.0f} MNOK")
print(f"Total SURF:     {surf_cost:.0f} MNOK")

7.9.3 Cost Sensitivity

The dominant cost drivers for SURF systems are:

• Water depth — affects riser length, installation vessel requirements, and equipment ratings
• Tieback distance — flowline and umbilical costs scale linearly with distance
• Number of wells — each well requires a tree, connections, and control system
• Pipe diameter — larger pipes cost more per meter but may enable higher production
• Insulation — wet insulation (syntactic foam) adds \$500–2,000/m; pipe-in-pipe adds \$2,000–5,000/m

The slices are assigned teaching inputs, not vendor quotations or a cost estimate for a specified development. Use a dated quantity and cost basis, installation scope and uncertainty model for investment analysis.

7.10 Field Layout Optimization

7.10.1 Layout Considerations

The subsea field layout — the spatial arrangement of wells, manifolds, flowlines, and risers — affects both capital cost and production performance:

• Well spacing — too close risks interference; too far increases jumper lengths
• Manifold location — should minimize total jumper and flowline lengths
• Flowline routing — must avoid geohazards, existing infrastructure, and excessive free spans
• Riser connection — the point where the flowline transitions to a riser affects both cost and flow assurance

7.10.2 Optimization Approach

Field layout optimization is typically a multi-objective problem:

where $x$ is the vector of layout variables (well positions, manifold locations, pipe routes), $C_{SURF}$ is the SURF cost, $Q_{prod}$ is the total production, and $R$ is a flow assurance risk metric.

In practice, the optimization is performed in stages:

1. Reservoir-driven well placement — optimized by reservoir engineers for drainage
2. Manifold clustering — group wells into manifold clusters to minimize inter-connections
3. Routing and sizing — optimize pipe routes, diameters, and insulation levels
4. Flow assurance verification — confirm the layout meets hydrate, wax, and slugging criteria

7.11 Subsea Coolers and Subsea Processing Integration

In some fields, particularly those with high reservoir temperatures or gas condensate fluids, a subsea cooler may be installed to reduce the fluid temperature before entering long flowlines. The reduced temperature decreases the specific volume of the gas phase, reducing velocity and friction pressure drop.

However, subsea cooling must be balanced against flow assurance risks — lower temperatures bring the fluid closer to hydrate and wax formation conditions. The optimal subsea cooler exit temperature is typically:

where $\Delta T_{margin}$ is a safety margin, typically 5–10°C above the hydrate equilibrium temperature.

7.12 NeqSim Implementation Summary

The key NeqSim classes used for subsea production system analysis are:

Reproduced Calculation Results

These examples use the stated fluid recipes and operating assumptions. Curves represent NeqSim calculations unless a caption identifies an analytical illustration, assumed equipment map or synthetic data.

Pipeline Pressure: pressure spans 146.5–150 bara across the plotted cases.

Friction and elevation redistribute pressure along the subsea route; changing density and phase holdup alters the local gradient. The arrival pressure determines how much pressure is available for the riser and receiving plant, rather than constituting a free design input. Use the surveyed elevation profile, verify the pressure budget and refine segmentation until the arrival pressure is stable.

Fluid Temperature: temperature remains 70 °C across the plotted cases.

This notebook holds each pipeline segment at the specified 70 degree Celsius temperature; heat transfer to the seabed is not solved. The flat profile is an isothermal boundary assumption and cannot quantify normal-operation cooling or shutdown margin. Enable and calibrate a thermal model before using arrival temperature to select insulation or hydrate protection.

Hydrate Equilibrium Curve: pressure spans 20–200 bara across the plotted cases. Pipeline Operating Conditions: pressure spans 146.5–150 bara across the plotted cases.

The hydrate curve is a separate NeqSim equilibrium calculation; the operating points use the isothermal pipeline assumption. The 70 degree Celsius path is comfortably warmer than equilibrium in this constructed example, but its apparent margin does not account for actual seabed heat loss. Recompute the operating path with heat transfer and evaluate cooldown and free-water availability before selecting protection.

Superficial Liquid Velocity spans 0.001–0.03144 m/s across the plotted cases.

Superficial gas and liquid velocities characterize the competing inertia, buoyancy and phase-continuity mechanisms that form flow regimes. A horizontal regime map is a screening tool and may not describe inclined sections, risers or transient severe slugging. Use regime labels within the correlation’s domain and assess liquid accumulation and transient slugging for the actual route.

The assumed pressure allocation starts at 300 bara reservoir pressure and ends at 176 bara separation pressure. The individual losses are prescribed budget entries rather than a solved network balance.

The total well-to-platform pressure difference is consumed by tubing lift, choke loss, flowline friction and riser elevation. A negative passive choke loss or a booster with negative power exposes an inconsistent model boundary rather than a production benefit. Check each pressure contribution and the sign of equipment work before allocating additional backpressure or evaluating subsea boosting.

Selected numerical ranges from the plotted cases:

Ranges describe the sampled cases; they are not independent validation tolerances.

7.13 Summary

Key points from this chapter:

• Subsea production systems enable the development of offshore fields by placing wellheads and equipment on the seabed with tiebacks to host facilities
• The major components are subsea trees, manifolds, flowlines, risers, and umbilicals
• Subsea tree valve functions (master, wing, swab, choke) provide both flow control and well barrier integrity per NORSOK D-010
• Manifold design includes header sizing for commingled flow, pigging loops for pipeline maintenance, and layout configurations (hub-spoke, daisy chain, dual-header)
• The pressure budget — from reservoir to separator — determines the maximum production rate and tieback distance; the wellhead pressure must overcome flowline, riser, and choke pressure drops
• Temperature management is critical to prevent hydrate and wax formation in subsea flowlines
• Subsea field architectures (satellite, cluster manifold, daisy chain, subsea to shore) each have distinct advantages depending on field size, distance, and phasing
• Subsea processing technologies — multiphase pumping (helico-axial, twin-screw), separation, and wet gas compression — extend field life and tieback distances
• Power delivery to subsea processing equipment is a major design challenge, requiring medium- or high-voltage subsea electrical systems
• Umbilicals deliver hydraulic control, electrical power, and chemical injection (methanol, MEG, scale inhibitor) to subsea equipment
• NeqSim's SubseaWell, PipeBeggsAndBrills, and ProcessSystem classes enable comprehensive subsea system modeling from bottomhole to topside
• SURF costs typically represent 30–50% of total subsea development costs and are dominated by flowline length and water depth
• Field layout optimization balances cost, production, and flow assurance objectives

Verification of the worked examples

The completed subsea pipe/cooler calculations are checked for material conservation and valid pressures and temperatures. The thermal example states its chosen overall heat-transfer coefficient; the audit does not calibrate heat loss, multiphase holdup or vendor equipment performance. Published operator reports distinguish the actual Åsgard, Gullfaks and Vigdis technologies and startup dates; cost and casing factors are explicitly screening inputs.[18, 17, 16]

The calculation and literal-code records are in verification/scientific_revision/ch07_manuscript_physics.json; the chapter scope and code hashes are indexed in foundations_review.json.

Exercises

1. Exercise 7.1: For the reference fluid in Section 7.5.3, calculate the pressure and temperature at the end of a 25 km subsea flowline for pipe diameters of 8", 10", and 12". Plot the results and identify the minimum acceptable diameter if the arrival temperature must be at least 20°C above the hydrate equilibrium temperature.

1. Exercise 7.2: Using the subsea well model in Section 7.8.1, modify the water depth from 350 m to 1,500 m and recalculate the well cost estimate. What is the approximate cost increase per meter of additional water depth?

1. Exercise 7.3: A subsea field has 8 wells producing a gas condensate fluid (GOR = 3,000 Sm³/Sm³). The water depth is 800 m and the distance to the host is 35 km. Using NeqSim, model the complete system (wells → manifold → flowline → riser → separator at 60 bara) and determine: (a) the total pressure drop, (b) the arrival temperature, (c) whether subsea boosting is needed.

1. Exercise 7.4: Perform a SURF cost estimate for the field in Exercise 7.3 using the parametric cost model in Section 7.9.2. Calculate the SURF cost for two alternative layouts: (a) one manifold with all 8 wells, and (b) two manifolds with 4 wells each connected by a gathering flowline.

1. Exercise 7.5: Repeat the tieback distance analysis in Section 7.6.2 with and without subsea boosting (50 bar pressure increase). Plot the additional distance achievable with boosting as a function of flowline diameter.

1. Exercise 7.6: Estimate the hydraulic response time for a subsea control system with a 60 km umbilical. Assume a 3/8" hydraulic tube, 345 bar working pressure, and standard hydraulic fluid. Discuss the implications for emergency shutdown system design.

1. Exercise 7.7: Compare the pressure budgets for a satellite well (10 km direct tieback) and a manifold well (5 km jumper to manifold, then 15 km flowline) for the same wellhead conditions. Which architecture delivers lower topside arrival pressure, and why?

1. Exercise 7.8: A late-life subsea field has declining wellhead pressures (from 180 bara to 80 bara over 10 years). Using the subsea boosting model from Section 7.7.7, determine the year at which boosting becomes necessary to maintain a topside arrival pressure of 30 bara through a 20 km flowline and 400 m riser.

1. Bai, Y. and Bai, Q. (2019). Subsea Engineering Handbook, 2nd Edition. Gulf Professional Publishing.
2. Gudmestad, O.T. (2015). Marine Technology and Operations: Theory & Practice. WIT Press.
3. Zhen, L. and Songhurst, B. (2018). "Subsea processing: The next step in subsea field development." Journal of Petroleum Technology, 70(3), 44–52.
4. NORSOK D-010 (2021). Well Integrity in Drilling and Well Operations, Rev. 5. Standards Norway.
5. API Bull 5C3 (2008). Bulletin on Formulas and Calculations for Casing, Tubing, Drill Pipe and Line Pipe Properties. American Petroleum Institute.
6. DNV-RP-F101 (2019). Corroded Pipelines. Det Norske Veritas.
7. Sangesland, S. (2018). "Subsea well technology." Chapter in Petroleum Production Engineering, Elsevier.
8. Eriksson, K. and Høvik, J. (2017). "Subsea compression — technology development and qualification." OTC, Paper OTC-27893.
9. API RP 14E (1991). Recommended Practice for Design and Installation of Offshore Production Platform Piping Systems, 5th Edition.
10. DNV-RP-F116 (2015). Integrity Management of Submarine Pipeline Systems. Det Norske Veritas.
11. Akers, T. and Amin, A. (2014). "Subsea multiphase pumping: Technology update." SPE, Paper SPE-170238.
12. Gjerdseth, A.C., Faanes, A., and Ramberg, R. (2012). "The world's first subsea compression system — from design to operation." OTC, Paper OTC-23427.

Running the examples. Start the source-workspace Python session described in Chapter 1, then run this chapter's Python blocks in reading order. Java blocks form a separate sequence using the same NeqSim build; carry forward objects from preceding Java blocks. The release execution records are in verification/; a successful run establishes API compatibility, while physical validation also requires the checks discussed in the text.

Hydraulic evidence: geometry, thermal model and feasible points

In the current PipeBeggsAndBrills interface, setLength takes metres and setElevation is outlet elevation minus inlet elevation, also in metres. A rising production riser therefore has a positive elevation change. A 20 km flowline requires setLength(20000.0), and geometric consistency requires $|\Delta z|\le L$. Confusing kilometres and metres can understate friction by three orders of magnitude and make a riser geometrically impossible.

PipingRouteBuilder can translate a line-list route into serial hydraulic segments. Record the measured length, internal diameter, roughness, elevation, fitting treatment, heat-transfer coefficient and ambient temperature for each segment. Recalculate fluid properties along the route and compare pressure and temperature profiles after refining the segmentation. A final pressure alone cannot show where acceleration, liquid accumulation or a thermal pinch controls the result.[1]

A forward calculation fixes inlet pressure and rate. If its pressure becomes nonphysical before the outlet, that rate is infeasible for the stated boundary conditions. Preserve it as a failed point in the sensitivity table; never substitute the last finite pressure. A deliverability calculation instead varies the rate to meet outlet pressure and requires a bracketed solution. Steady-state holdup and flow-regime screening do not establish slug frequency, restart inventory or transient stability.

Learning Objectives

After reading this chapter, the reader will be able to:

1. Apply the Darcy-Weisbach equation with the Colebrook-White friction factor for single-phase pipe flow
2. Identify multiphase flow regimes (stratified, slug, annular, bubble) and predict their occurrence using flow regime maps
3. Apply the Beggs and Brill correlation for multiphase pressure drop and liquid holdup calculations
4. Calculate steady-state temperature profiles in insulated and uninsulated pipelines
5. Model pipeline pressure and temperature profiles using NeqSim's PipeBeggsAndBrills class
6. Analyze riser hydraulics including severe slugging and terrain-induced slugging
7. Size pipelines for single-phase and multiphase flow applications

8.1 Introduction

Flowlines, pipelines, and risers are the arteries of the production system, transporting multiphase hydrocarbon fluids from the wellhead to the processing facility and single-phase products from the facility to market. The design and operation of these transport systems directly affect production rate, energy consumption, and flow assurance.

This chapter covers the fundamentals of pipe flow — from single-phase friction factor correlations to the complexity of multiphase flow with its distinct flow regimes, liquid holdup, and terrain effects. The emphasis is on practical pressure drop and temperature calculations using both analytical methods and NeqSim's numerical tools.

Pipeline hydraulics is central to production optimization because:

• Pressure drop in flowlines and risers determines the back-pressure on wells, directly affecting production rate
• Temperature drop along pipelines governs flow assurance risks (hydrates, wax, slugging)
• Flow regime affects pressure drop magnitude, slug loads on equipment, and liquid inventory in the pipeline
• Pipeline sizing is a critical design decision — too small wastes pressure energy, too large wastes capital

The printed dimensions and inlet pressure are schematic inputs. This block diagram is not a cross-section or an accepted hydraulic sizing result.

8.2 Single-Phase Pipe Flow

8.2.1 The Darcy-Weisbach Equation

The Darcy-Weisbach equation is the fundamental relationship for pressure drop in single-phase pipe flow:

where:

• $\Delta P$ is the frictional pressure drop [Pa]
• $f$ is the Darcy friction factor (dimensionless)
• $L$ is the pipe length [m]
• $D$ is the internal diameter [m]
• $\rho$ is the fluid density [kg/m³]
• $v$ is the mean velocity [m/s]

The total pressure drop in an inclined pipe includes a gravitational (elevation) term:

where $\Delta h$ is the elevation change [m] (positive upward) and $g$ is gravitational acceleration [m/s²].

8.2.2 The Moody Diagram and Friction Factor

The friction factor depends on the Reynolds number and relative pipe roughness:

where $\mu$ is the dynamic viscosity [Pa·s].

The flow regime is classified as:

8.2.3 The Colebrook-White Equation

For turbulent flow, the Colebrook-White equation provides the friction factor:

where $\epsilon$ is the absolute pipe roughness [m]. This implicit equation must be solved iteratively. Common explicit approximations include:

Swamee-Jain (1976):

Haaland (1983):

Typical absolute roughness values:

8.2.4 Single-Phase Pipeline Sizing Example

import jpype
jneqsim = jpype.JPackage("neqsim")
import math

# Define dry gas for export pipeline
gas = jneqsim.thermo.system.SystemSrkEos(273.15 + 25.0, 120.0)
gas.addComponent("methane", 90.0)
gas.addComponent("ethane", 5.0)
gas.addComponent("propane", 2.0)
gas.addComponent("i-butane", 0.5)
gas.addComponent("n-butane", 0.8)
gas.addComponent("CO2", 1.0)
gas.addComponent("nitrogen", 0.7)
gas.setMixingRule("classic")

# Flash to get properties
ThermodynamicOperations = jneqsim.thermodynamicoperations.ThermodynamicOperations
ops = ThermodynamicOperations(gas)
ops.TPflash()
gas.initProperties()

# Gas properties at pipeline conditions
rho = gas.getDensity("kg/m3")
mu = gas.getPhase("gas").getViscosity("kg/msec")
print(f"Gas density: {rho:.2f} kg/m3")
print(f"Gas viscosity: {mu:.6f} Pa.s")

# Pipeline parameters
Q = 15.0e6  # 15 MSm3/day
# NeqSim Sm3 uses 15 C and 1.01325 bara. Convert through the EOS, including Z.
reference = gas.clone()
reference.setTemperature(288.15)
reference.setPressure(1.01325)
ThermodynamicOperations(reference).TPflash()
reference.initProperties()
assert reference.getNumberOfPhases() == gas.getNumberOfPhases() == 1
mass_rate = Q / 86400.0 * reference.getDensity("kg/m3")
Q_actual = mass_rate / rho
D = 0.762  # 30-inch pipe, ~762 mm ID
L = 150000.0  # 150 km
epsilon = 0.045e-3  # new steel, m

A = math.pi * D**2 / 4.0
v = Q_actual / A
Re = rho * v * D / mu

# Swamee-Jain friction factor
f = 0.25 / (math.log10(epsilon / D / 3.7 + 5.74 / Re**0.9))**2

# Pressure drop
dP = f * L / D * rho * v**2 / 2.0
dP_bar = dP / 1.0e5

print(f"Velocity: {v:.2f} m/s")
print(f"Reynolds number: {Re:.0f}")
print(f"Friction factor: {f:.6f}")
print(f"Inlet-property screening drop: {dP_bar:.1f} bar over {L/1000:.0f} km")
assert rho > 0 and mu > 0 and Re > 4000 and 0 < f < 0.1
assert 0 < dP_bar < gas.getPressure("bara")
# Finite pressure changes require distributed compressible-flow integration.
print(f"Screening drop/inlet pressure: {dP_bar / 120.0:.3f}")

8.3 Multiphase Flow Fundamentals

8.3.1 Why Multiphase Flow Is Different

In most offshore production systems, the fluid flowing through flowlines and risers is a multiphase mixture of gas, oil, and water (and sometimes sand). Multiphase flow is fundamentally more complex than single-phase flow because:

• Phase distribution — the phases do not flow uniformly; they distribute according to flow regime
• Slip — gas typically travels faster than liquid due to buoyancy and drag differences
• Liquid holdup — the fraction of the pipe occupied by liquid differs from the input liquid fraction
• Pressure drop mechanisms — friction, gravity, and acceleration all depend on the phase distribution
• Dynamic behavior — multiphase flow can be inherently transient (slug flow)

8.3.2 Key Definitions

Superficial velocity is the velocity each phase would have if it occupied the entire pipe cross-section:

where $Q_G$ and $Q_L$ are the actual volumetric flow rates of gas and liquid at local conditions.

Mixture velocity:

No-slip liquid holdup (input liquid fraction):

Actual liquid holdup $H_L$ is the fraction of the pipe cross-section occupied by liquid. Due to slip, $H_L > \lambda_L$ in most cases (gas flows faster, so liquid accumulates).

Gas void fraction:

8.3.3 Multiphase Flow Regimes

The distribution of phases in the pipe depends on gas and liquid velocities, fluid properties, and pipe geometry. The four principal flow regimes in horizontal pipe are:

Stratified flow — at low gas and liquid velocities, the liquid settles to the bottom of the pipe and the gas flows above. The interface may be smooth (stratified smooth) or wavy (stratified wavy).

Slug flow — at moderate velocities, intermittent slugs of liquid bridge the entire pipe cross-section, separated by gas pockets (Taylor bubbles). Slug flow produces cyclic pressure and flow rate fluctuations.

Annular flow — at high gas velocities, the liquid forms a thin film on the pipe wall and the gas flows through the core, carrying entrained liquid droplets.

Bubble (dispersed) flow — at high liquid velocities and low gas velocities, small gas bubbles are dispersed in the liquid phase.

In vertical upward flow, the regimes are:

• Bubble flow — small bubbles rise through the liquid
• Slug flow — large Taylor bubbles separated by liquid slugs
• Churn flow — chaotic oscillatory regime between slug and annular
• Annular flow — liquid film on walls, gas core with droplets

8.3.4 Flow Regime Maps

Flow regime maps plot the boundaries between regimes as functions of superficial gas and liquid velocities. The most widely used maps are:

Taitel and Dukler (1976) for horizontal flow — uses dimensionless groups based on the equilibrium stratified film model:

where $X$ is the Lockhart-Martinelli parameter.

Taitel, Barnea, and Dukler (1980) for vertical flow — transitions depend on the Kutateladze number and dimensionless gas velocity.

8.4 Pressure Drop Correlations for Multiphase Flow

8.4.1 The Beggs and Brill Correlation

The Beggs and Brill (1973) correlation is one of the most widely used methods for multiphase pressure drop in inclined pipes. It was developed from experimental data covering all inclination angles from horizontal to vertical. The total pressure gradient is:

where:

• $f_{tp}$ is the two-phase friction factor
• $\rho_n = \rho_L \lambda_L + \rho_G (1 - \lambda_L)$ is the no-slip mixture density [kg/m³]
• $\rho_s = \rho_L H_L + \rho_G (1 - H_L)$ is the slip (actual) mixture density [kg/m³]
• $v_m$ is the mixture velocity [m/s]
• $\theta$ is the pipe inclination angle from horizontal [rad]
• $P$ is the absolute pressure [Pa]

The correlation proceeds in four steps:

Step 1: Determine the flow regime using the Froude mixture number:

and the no-slip liquid holdup $\lambda_L$. The transitions are defined by correlations:

Step 2: Calculate the horizontal liquid holdup $H_L(0)$ using regime-specific correlations:

Step 3: Correct for inclination using the Beggs–Brill inclination multiplier. A separately enabled Payne correction is an additional empirical holdup adjustment; it is not the definition of this multiplier:

where $\psi$ depends on the inclination angle, flow regime, and the no-slip liquid velocity number:

Step 4: Calculate the two-phase friction factor:

where $f_n$ is the no-slip friction factor and $S$ is an empirical two-phase correction based on no-slip liquid fraction and liquid holdup; it is not a direct measurement of interfacial roughness.

8.4.2 The Mukherjee and Brill Correlation

Mukherjee–Brill correlates liquid holdup using inclination and dimensionless gas velocity, liquid velocity and viscosity groups, with distinct coefficients and regime conditions. The previously abbreviated four-term expression omitted required velocity dependencies and is not a usable implementation. Use the complete published correlation with its regime rules and consistent units when making a comparison.

8.4.3 Mechanistic Models

Modern pipeline simulators increasingly use mechanistic models that predict the flow regime from first principles (conservation of mass, momentum, and energy for each phase) rather than empirical correlations. Notable mechanistic models include:

• OLGA — the industry-standard dynamic multiphase flow simulator
• LedaFlow — a newer model based on the slug tracking concept
• Zhang et al. (2003) — unified mechanistic model for all pipe inclinations

Mechanistic models can represent additional mechanisms and transitions, but extrapolation beyond their closure-correlation validation range is not automatically more accurate.

8.4.4 Comparison of Correlations

No universal pressure-drop accuracy percentage ranks these methods. Compare them against the same independent measurements, including measurement uncertainty, fluid properties and flow regime; a mechanistic formulation still contains empirical closures.

8.5 Multiphase Pipeline Modeling with NeqSim

8.5.1 The PipeBeggsAndBrills Class

NeqSim implements the Beggs and Brill correlation through the PipeBeggsAndBrills class. This class calculates:

• Pressure profile along the pipeline
• Temperature profile (with heat transfer to surroundings)
• Liquid holdup profile
• Flow regime at each calculation increment

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define a typical oil-gas-water production fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 75.0, 120.0)
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 2.5)
fluid.addComponent("methane", 55.0)
fluid.addComponent("ethane", 7.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("i-butane", 1.5)
fluid.addComponent("n-butane", 3.0)
fluid.addComponent("i-pentane", 1.5)
fluid.addComponent("n-pentane", 2.0)
fluid.addComponent("n-hexane", 3.0)
fluid.addComponent("n-heptane", 5.0)
fluid.addComponent("n-octane", 5.0)
fluid.addComponent("n-nonane", 3.0)
fluid.addComponent("water", 6.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

# Create inlet stream
Stream = jneqsim.process.equipment.stream.Stream
inlet = Stream("Pipeline Inlet", fluid)
inlet.setFlowRate(100000.0, "kg/hr")
inlet.setTemperature(75.0, "C")
inlet.setPressure(120.0, "bara")

# Create pipeline using Beggs and Brill
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
pipeline = PipeBeggsAndBrills("Production Flowline", inlet)
pipeline.setPipeWallRoughness(5.0e-5)
pipeline.setLength(20000.0)            # 20 km
pipeline.setElevation(0.0)          # horizontal
pipeline.setDiameter(0.3048)        # 12-inch (0.3048 m)
pipeline.setNumberOfIncrements(100)
pipeline.setConstantSurfaceTemperature(277.15) # 4°C seabed

# Build and run
ProcessSystem = jneqsim.process.processmodel.ProcessSystem
process = ProcessSystem()
process.add(inlet)
process.add(pipeline)
process.run()

# Read results
outlet_P = pipeline.getOutletStream().getPressure("bara")
outlet_T = pipeline.getOutletStream().getTemperature("C")
print(f"Outlet pressure:    {outlet_P:.1f} bara")
print(f"Outlet temperature: {outlet_T:.1f} °C")
print(f"Pressure drop:      {120.0 - outlet_P:.1f} bar")
print(f"Temperature drop:   {75.0 - outlet_T:.1f} °C")

8.5.2 Pressure and Temperature Profiles

NeqSim divides the pipeline into increments (set by setNumberOfIncrements) and updates pressure, temperature and local phase properties along the route. The following diameter comparison reports only each line's outlet pressure and temperature. These endpoint results must not be read as a plotted profile; the stored increment data can be inspected separately.

import jpype
jneqsim = jpype.JPackage("neqsim")

# Use the same fluid and pipeline setup as Section 8.5.1
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 75.0, 120.0)
fluid.addComponent("methane", 60.0)
fluid.addComponent("ethane", 7.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("n-butane", 3.0)
fluid.addComponent("n-pentane", 2.0)
fluid.addComponent("n-hexane", 3.0)
fluid.addComponent("n-heptane", 6.0)
fluid.addComponent("n-octane", 5.0)
fluid.addComponent("water", 9.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

Stream = jneqsim.process.equipment.stream.Stream
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

# Compare different pipe diameters
diameters_inch = [8, 10, 12, 14]
diameters_m = [d * 0.0254 for d in diameters_inch]

for d_inch, d_m in zip(diameters_inch, diameters_m):
    test_fluid = fluid.clone()
    feed = Stream("Feed", test_fluid)
    feed.setFlowRate(80000.0, "kg/hr")
    feed.setTemperature(75.0, "C")
    feed.setPressure(120.0, "bara")

    pipe = PipeBeggsAndBrills("Pipeline", feed)
    pipe.setPipeWallRoughness(5.0e-5)
    pipe.setLength(20000.0)
    pipe.setElevation(0.0)
    pipe.setDiameter(d_m)
    pipe.setNumberOfIncrements(80)
    pipe.setConstantSurfaceTemperature(277.15)

    pipe.setHeatTransferCoefficient(5.0)  # assumed insulated-pipe U, W/(m2 K)
    proc = ProcessSystem()
    proc.add(feed)
    proc.add(pipe)
    proc.run()

    out_P = pipe.getOutletStream().getPressure("bara")
    out_T = pipe.getOutletStream().getTemperature("C")
    dP = 120.0 - out_P
    print(f"Diameter: {d_inch}\"  dP: {dP:.1f} bar  Outlet T: {out_T:.1f} °C")

8.5.3 Effect of Flow Rate on Pressure Drop

The relationship between flow rate and pressure drop is nonlinear. At low rates, gravitational effects dominate in inclined pipes. At high rates, friction dominates:

import jpype
jneqsim = jpype.JPackage("neqsim")

fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 70.0, 100.0)
fluid.addComponent("methane", 60.0)
fluid.addComponent("ethane", 7.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("n-butane", 3.0)
fluid.addComponent("n-hexane", 4.0)
fluid.addComponent("n-heptane", 8.0)
fluid.addComponent("n-octane", 5.0)
fluid.addComponent("water", 8.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

Stream = jneqsim.process.equipment.stream.Stream
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

flow_rates = [30000, 50000, 80000, 100000, 120000, 150000]  # kg/hr
print(f"{'Flow Rate (kg/hr)':>18} {'dP (bar)':>10} {'Outlet T (°C)':>14}")

for rate in flow_rates:
    test_fluid = fluid.clone()
    feed = Stream("Feed", test_fluid)
    feed.setFlowRate(float(rate), "kg/hr")
    feed.setTemperature(70.0, "C")
    feed.setPressure(100.0, "bara")

    pipe = PipeBeggsAndBrills("Flowline", feed)
    pipe.setPipeWallRoughness(5.0e-5)
    pipe.setLength(15000.0)
    pipe.setElevation(0.0)
    pipe.setDiameter(0.254)
    pipe.setNumberOfIncrements(60)
    pipe.setConstantSurfaceTemperature(277.15)

    pipe.setHeatTransferCoefficient(5.0)  # assumed insulated-pipe U, W/(m2 K)
    proc = ProcessSystem()
    proc.add(feed)
    proc.add(pipe)
    proc.run()

    out_P = pipe.getOutletStream().getPressure("bara")
    out_T = pipe.getOutletStream().getTemperature("C")
    print(f"{rate:>18} {100.0 - out_P:>10.1f} {out_T:>14.1f}")

8.6 Elevation Effects and Terrain

8.6.1 Gravitational Pressure Component

In inclined pipes, the gravitational pressure component can be the dominant contribution to total pressure drop:

For a riser in deepwater, the hydrostatic head of the liquid holdup creates a substantial pressure decrease from the riser base to the topside in upward flow. For example, with $H_L = 0.3$ and typical fluid densities ($\rho_L = 700$ kg/m³, $\rho_G = 80$ kg/m³):

For a 1,000 m riser: $\Delta P_{hydrostatic} = 266 \times 9.81 \times 1000 / 10^5 = 26.1$ bar.

8.6.2 Terrain-Induced Slugging

Undulating terrain in subsea flowlines creates conditions for terrain-induced slugging:

1. Liquid accumulates in low points (valleys)
2. Gas pressure builds upstream of the liquid plug
3. When the gas pressure exceeds the hydrostatic head of the liquid, the gas blows through
4. The liquid plug accelerates as a slug toward the next low point or the riser
5. The cycle repeats

The severity of terrain slugging depends on:

• Depth of the valleys relative to pipe diameter
• Gas-liquid ratio
• Flow velocity (worse at low rates)
• Fluid properties

8.6.3 Modeling Elevation Effects in NeqSim

NeqSim's PipeBeggsAndBrills class handles elevation through the setElevation parameter:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 50.0, 80.0)
fluid.addComponent("methane", 65.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("n-butane", 3.0)
fluid.addComponent("n-hexane", 4.0)
fluid.addComponent("n-heptane", 6.0)
fluid.addComponent("n-octane", 4.0)
fluid.addComponent("water", 5.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

Stream = jneqsim.process.equipment.stream.Stream
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

# Model riser (800 m water depth)
inlet = Stream("Riser Base", fluid)
inlet.setFlowRate(60000.0, "kg/hr")
inlet.setTemperature(50.0, "C")
inlet.setPressure(80.0, "bara")

riser = PipeBeggsAndBrills("Production Riser", inlet)
riser.setPipeWallRoughness(5.0e-5)
riser.setLength(1000.0)            # ~1 km total length (catenary)
riser.setElevation(800.0)       # 800 m vertical rise
riser.setDiameter(0.254)        # 10-inch
riser.setNumberOfIncrements(40)
riser.setConstantSurfaceTemperature(280.15)

riser.setHeatTransferCoefficient(5.0)  # assumed insulated-pipe U, W/(m2 K)
process = ProcessSystem()
process.add(inlet)
process.add(riser)
process.run()

topside_P = riser.getOutletStream().getPressure("bara")
topside_T = riser.getOutletStream().getTemperature("C")
print(f"Topside arrival pressure: {topside_P:.1f} bara")
print(f"Topside arrival temperature: {topside_T:.1f} °C")
print(f"Total riser dP: {80.0 - topside_P:.1f} bar")

8.6.4 Hilly Terrain, Slack Flow, and Pigging Considerations

Onshore pipelines and some seabed flowlines traverse undulating terrain with alternating uphill and downhill sections. This creates unique hydraulic challenges beyond simple elevation effects:

Liquid accumulation in low points: In gas-dominated systems, liquid (condensate or water) accumulates in topographic low points (valleys). At low flow rates, the gas velocity may be insufficient to sweep this liquid up the next incline. The accumulated liquid increases the effective hydrostatic head and reduces throughput capacity. Over time, liquid holdup can grow until the pipeline becomes "waterlogged."

Slack flow (gravity-dominated flow): In downhill sections of liquid-dominated pipelines, the hydrostatic pressure gradient can exceed the frictional pressure gradient, creating a situation where the liquid accelerates due to gravity. If the pipeline is not full (gas pocket at the top of the downslope), "slack flow" occurs — the liquid separates from the upper wall and free-falls. This produces:

• Low pressure at the top of the hill (possibly below bubble point, causing gas breakout)
• Unpredictable two-phase flow regime transitions
• Surge loading at the bottom of the downslope

With inclination positive upward, gravity dominates friction in a descending, full-bore liquid segment when

This comparison alone is not a slack-flow criterion. Slack flow requires a loss of full-bore liquid occupancy, determined by the boundary heads, available inventory and local absolute pressure relative to vapor pressure. A vapor cavity or entrained-gas region requires an appropriate multiphase/transient model.

Pigging in hilly terrain: Pigs traveling through undulating terrain accumulate liquid ahead of them in each downhill section. This creates progressively larger liquid slugs as the pig advances. An inventory screening bound for liquid collected from the listed sections is:

where $N_{valleys}$ is the number of low spots, $H_{L,i}$ is the average holdup in each section, and $L_{downhill,i}$ is the length of each downhill section. Pipeline profile data must be analyzed to estimate pig-generated slug volumes for slug catcher sizing.

Terrain slugging mitigation: Strategies include:

• Maintaining velocity above liquid sweep threshold
• Regular pigging schedules to limit liquid accumulation
• Slug catchers at the pipeline outlet sized for terrain slugs
• Pipeline profile optimization during route selection (minimize elevation changes)

8.7 Heat Transfer in Pipelines

8.7.1 Overall Heat Transfer Coefficient

The rate of heat loss from a pipeline to the surroundings is characterized by the overall heat transfer coefficient $U$:

where:

• $q$ is the heat loss rate [W]
• $U$ is the overall heat transfer coefficient [W/m²·K]
• $A$ is the heat transfer area [m²]
• $T_{fluid}$ is the bulk fluid temperature [K]
• $T_{ambient}$ is the ambient (seabed or air) temperature [K]

For a pipe with multiple layers (steel wall, insulation, coating, concrete), the overall $U$-value is:

where:

• $h_i$ is the internal film coefficient [W/m²·K]
• $D_i$ is the internal diameter [m]
• $D_o$ is the outer diameter [m]
• $k_j$ is the thermal conductivity of layer $j$ [W/m·K]
• $h_o$ is the external film coefficient [W/m²·K]

8.7.2 Typical U-Values

8.7.3 Steady-State Temperature Profile

For a pipeline with constant $U$-value and ambient temperature, the steady-state temperature profile is:

The arrival temperature at the end of a pipeline of length $L$ is:

Key observations:

• Arrival temperature increases with flow rate (less time for heat loss)
• Arrival temperature decreases with pipe length and $U$-value
• At very high flow rates, $T_{arrival} \to T_{in}$ (adiabatic limit)
• At very low flow rates, $T_{arrival} \to T_{amb}$ (thermal equilibrium)

8.7.4 Joule-Thomson Effect

In gas-dominated systems, the Joule-Thomson (JT) effect causes the gas temperature to change as pressure drops. For most natural gases at typical pipeline conditions, the JT coefficient is positive — meaning the gas cools as pressure decreases:

This JT cooling compounds the heat loss to the surroundings and can be significant in long high-pressure gas pipelines. In the example, the pressure–enthalpy flash accounts for real-fluid temperature change as pressure falls, while a specified overall heat-transfer coefficient accounts for external cooling. The optional additional JT-heat and friction-heating switches remain disabled: adding a separate JT heat term to an enthalpy formulation would double count that mechanism. The present checks cover finite profiles, mass conservation and pressure direction; they do not independently validate the pipeline heat-loss closure.

8.7.5 Insulation Types and Burial Effects

The selection of insulation system depends on the required $U$-value, water depth, installation method, and design life. The principal subsea insulation technologies are:

Polyurethane (PU) foam: Applied as a multi-layer external coating (40–120 mm thick). PU foam has low thermal conductivity ($k \approx 0.03$–0.04 W/m·K) and is cost-effective for moderate insulation requirements ($U \approx 3$–8 W/m²·K). Depth-limited to approximately 1,000 m before hydrostatic compression degrades the foam.

Syntactic foam: Glass or polymer microspheres in an epoxy matrix. Higher density and compressive strength than PU foam, suitable for deepwater applications. Thermal conductivity $k \approx 0.10$–0.15 W/m·K, giving $U \approx 2$–5 W/m²·K.

Pipe-in-pipe (PiP): An inner production pipe is surrounded by insulation (aerogel, microporous silica, or vacuum) within an outer carrier pipe. Achieves the lowest passive $U$-values ($U \approx 0.5$–2.0 W/m²·K) but at significantly higher cost. Used for long tiebacks where arrival temperature is critical.

Direct electrical heating (DEH): Electric current is passed through the pipe wall (or a heating cable), actively maintaining the fluid temperature above critical thresholds (hydrate or wax). Used during shutdown and restart rather than steady-state production.

Burial: Trenching and backfilling the pipeline provides thermal insulation from the soil. The effective $U$-value depends on burial depth $H_b$, soil thermal conductivity $k_s$, and pipe diameter:

For typical North Sea clay ($k_s \approx 1.5$ W/m·K) with 1 m burial depth, the burial $U$-value is approximately 3–6 W/m²·K. Burial also provides mechanical protection and on-bottom stability.

The steady-state temperature profile for a pipeline with constant $U$ and ambient temperature is:

For the stated constant-property model, the characteristic cooling length is $L_T=\dot m c_p/(\pi DU)$, not a time constant. With $\dot m=80000/3600$kg/s, $c_p=2200$J/(kgK), $D=0.254$m and $U=5$W/(m²K), $L_T=12.25$km. Use the same diameter/area basis for $U$. This temperature law neglects pressure-work/JT effects, phase changes and axial conduction.

8.8 Pipeline Sizing

8.8.1 Sizing Criteria

Pipeline sizing balances capital cost (larger diameter = higher cost) against operating cost (larger diameter = lower pressure drop = lower compression energy):

The API RP 14E erosional velocity limit is widely used:

where $C$ is typically 100–150 (conservative) or up to 200 (with erosion-resistant materials), and $\rho_m$ is the mixture density in lb/ft³.

8.8.3 Erosional Velocity in Detail

The API RP 14E erosional velocity formula is widely used for preliminary pipeline and piping design:

where $\rho_m$ is the gas-liquid mixture density at flowing conditions [lb/ft³] and $C$ is an empirical constant. The standard recommends $C = 100$ for continuous service, but the appropriate value depends on several factors:

It is important to note that the API RP 14E formula is empirical and applies primarily to erosion by liquid droplets in gas flow. For solid particle (sand) erosion, more rigorous models such as DNV RP O501 should be used, which account for sand rate, particle size, impact angle, and material hardness.

8.8.4 Economic Pipeline Diameter Optimization

The optimal pipeline diameter balances capital expenditure (CAPEX) — which increases with diameter — against operating expenditure (OPEX) — which decreases with diameter because pressure drop and compression/pumping energy decrease.

The CAPEX of a pipeline is approximately proportional to $D^{1.2}$ to $D^{1.5}$ (accounting for steel weight, coating, installation):

The compression or pumping energy cost is proportional to the pressure drop, which scales approximately as $D^{-5}$ for single-phase turbulent flow (from Darcy-Weisbach):

The total annual cost (annualized CAPEX + OPEX) has a minimum at the economic optimum diameter. In practice, the selected diameter must also satisfy:

• Erosional velocity limit (upper bound on velocity)
• Minimum velocity for solids/liquid transport (lower bound)
• Available pipe sizes (standard API sizes)
• Arrival pressure constraint
• Arrival temperature constraint (flow assurance)

The economic optimum velocity for different fluid types:

8.8.5 Sizing Procedure with NeqSim

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 65.0, 100.0)
fluid.addComponent("methane", 60.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("n-butane", 3.0)
fluid.addComponent("n-pentane", 2.0)
fluid.addComponent("n-heptane", 8.0)
fluid.addComponent("n-octane", 6.0)
fluid.addComponent("water", 8.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

Stream = jneqsim.process.equipment.stream.Stream
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

# Design parameters
flow_rate = 80000.0       # kg/hr
max_dP_bar = 30.0         # Maximum allowable pressure drop
pipe_length_km = 25.0     # Flowline length
inlet_P = 100.0           # bara
inlet_T = 65.0            # °C

# Evaluate candidate diameters
candidates = [
    (8, 0.2032), (10, 0.254), (12, 0.3048), (14, 0.3556), (16, 0.4064)
]

print(f"{'Diameter':>10} {'dP (bar)':>10} {'Outlet T (°C)':>14} {'Status':>12}")
for d_inch, d_m in candidates:
    test_fluid = fluid.clone()
    feed = Stream("Feed", test_fluid)
    feed.setFlowRate(flow_rate, "kg/hr")
    feed.setTemperature(inlet_T, "C")
    feed.setPressure(inlet_P, "bara")

    pipe = PipeBeggsAndBrills("Flowline", feed)
    pipe.setPipeWallRoughness(5.0e-5)
    pipe.setLength(pipe_length_km * 1000.0)
    pipe.setElevation(0.0)
    pipe.setDiameter(d_m)
    pipe.setNumberOfIncrements(80)
    pipe.setConstantSurfaceTemperature(277.15)

    pipe.setHeatTransferCoefficient(5.0)  # assumed insulated-pipe U, W/(m2 K)
    proc = ProcessSystem()
    proc.add(feed)
    proc.add(pipe)
    proc.run()

    out_P = pipe.getOutletStream().getPressure("bara")
    dP = inlet_P - out_P
    out_T = pipe.getOutletStream().getTemperature("C")
    status = "OK" if dP <= max_dP_bar else "EXCEEDS"
    print(f"{d_inch:>8}\"  {dP:>10.1f} {out_T:>14.1f} {status:>12}")

8.9 Riser Hydraulics

8.9.1 Riser Types

The riser connects the seabed flowline to the topside facility. Common riser configurations include:

8.9.2 Severe Slugging

Severe slugging is a cyclic flow instability that can occur at the base of a riser when a relatively flat flowline connects to a vertical riser:

1. Slug formation — liquid accumulates at the riser base, blocking gas flow
2. Slug growth — gas pressure builds behind the liquid plug in the flowline
3. Slug production — when the gas pressure exceeds the hydrostatic head, the liquid is pushed up the riser as a slug
4. Gas blowthrough — once the liquid is expelled, gas blows through the riser at high velocity

This cycling produces:

• Large pressure fluctuations (±20–40% of steady-state pressure)
• Liquid surge at the topside separator (can exceed separator capacity)
• Gas rate fluctuations affecting compressor operation
• Fatigue loading on the riser structure

8.9.3 Severe Slugging Criteria

A severe-slugging boundary compares the rate of gas-pressure buildup in the upstream volume with the rate at which a liquid plug raises hydrostatic head in the riser. It therefore requires phase flow rates, compressible gas inventory, pipe/riser geometry, downstream pressure and a closure for liquid holdup. A dimensionless ratio of static pressure to hydrostatic head alone cannot distinguish a stable high-rate case from a low-rate oscillation.

Use a documented transient or linear-stability model and state its closure and boundary conditions. A steady Beggs–Brill profile can supply a base state, but cannot establish slug amplitude, period or a universal stable gas-lift rate.

8.9.4 Severe Slugging Mitigation

8.9.5 Riser Base Gas Lift

Gas lift injection at or near the riser base is one of the most effective methods for preventing severe slugging and improving riser hydraulics. The injected gas:

1. Reduces liquid holdup in the riser, lowering the hydrostatic back-pressure
2. Increases mixture velocity, pushing the flow regime from slug toward annular (stable)
3. Prevents liquid accumulation at the riser base by maintaining continuous gas throughput

Determine the stabilizing gas-lift range with the selected transient/stability model and check the extra friction and compression duty. Confirm the operating range against measured riser behavior; no universal static-pressure ratio guarantees stability.

A typical riser base gas lift system requires:

The gas lift rate must be balanced against compression cost and the impact on topside gas handling capacity. Excessive gas lift also cools the riser fluid (cold injection gas), potentially worsening hydrate risk.

In NeqSim, riser base gas lift is modeled by adding a gas stream at the riser base using a Mixer unit before the riser segment, analogous to the gas lift modeling in Section 5.6.4.

8.10 Pigging Operations

8.10.1 Purpose of Pigging

Pipeline pigs are devices that travel through the pipeline, propelled by the flowing fluid. They serve multiple purposes:

• Liquid sweep — remove liquid holdup from gas pipelines
• Wax removal — scrape wax deposits from pipe walls
• Inspection — intelligent pigs measure pipe wall thickness, corrosion, and geometry
• Batching — separate different products in multi-product pipelines
• Commissioning — dewater and dry new pipelines

8.10.2 Pig-Generated Slugs

When a pig sweeps liquid ahead of it, the accumulated liquid arrives at the pipeline outlet as a slug. The slug volume can be estimated as:

where $H_L$ is the average liquid holdup before pigging. For a 100 km pipeline with 10-inch diameter and $H_L = 0.05$:

This slug volume must be accommodated by the slug catcher at the receiving facility.

8.11 Pipeline Operations and Flow Monitoring

8.11.1 Operational Monitoring

Pipeline operations require continuous monitoring of key parameters to ensure safe, efficient transport:

• Pressure monitoring: Inlet and outlet pressure transmitters track the overall pressure drop. Deviations from the expected $\Delta P$ indicate changes in flow rate, fluid composition, liquid accumulation, or wax/scale buildup.
• Temperature monitoring: Distributed temperature sensing (DTS) fiber optic systems can provide the temperature profile along the entire pipeline length. Hot spots may indicate leaks; cold spots may indicate liquid holdup or insulation damage.
• Flow measurement: Multiphase flow meters at the wellhead and single-phase meters at the outlet track production rates and detect discrepancies that indicate leaks or metering errors.
• Pig tracking: Pig passage indicators (signallers) at pig launcher, receiver, and intermediate points confirm pig location and velocity.

8.11.2 Leak Detection

Pipeline leak detection systems fall into two categories:

Computational methods (internal):

• Mass balance: Compare inlet and outlet flow rates; a persistent imbalance indicates a leak. Sensitivity depends on metering accuracy (typically detects leaks > 1–2% of throughput).
• Pressure-point analysis: Monitor pressure at multiple points along the pipeline. A leak creates a localized pressure depression.
• Real-time transient modeling (RTTM): A full hydraulic model runs in real time, comparing model predictions with measured data. Deviations trigger leak alarms. RTTM can detect smaller leaks (0.5–1%) and estimate leak location.

External methods:

• Fiber optic sensing: Acoustic or temperature anomalies along the fiber indicate leak location
• Hydrocarbon detection: Subsea or atmospheric sensors detect released hydrocarbons
• Intelligent pigging: Wall thickness measurements identify thinning before through-wall failure

8.11.3 Wax Management and Chemical Injection

For waxy crude oil pipelines, regular wax management is essential:

• Wax inhibitor injection: Pour-point depressants or wax crystal modifiers injected at the pipeline inlet reduce deposition rate
• Regular pigging: Wax scraper pigs maintain pipe bore and prevent buildup to unpiggable levels
• Pigging frequency: Determined by wax deposition rate and maximum allowable wax thickness (typically 2–5 mm)
• Hot oiling: Circulating heated oil to melt deposited wax (less common subsea, used onshore)

8.12 Two-Phase Flow Correlation Comparison

8.12.1 Overview of Major Correlations

Multiple empirical correlations and flow regime maps have been developed for multiphase pipe flow. Each has strengths and limitations based on its development database:

Baker (1954): One of the earliest flow regime maps for horizontal two-phase flow. Uses the parameters:

where $\lambda$ and $\psi$ are fluid property correction factors referenced to air-water at standard conditions. The Baker map identifies seven flow regimes (annular, bubble, stratified, wave, slug, plug, dispersed) but was developed from small-diameter data (< 4 inches) and does not handle inclination.

Mandhane, Gregory, and Aziz (1974): Developed from a large database of air-water and air-oil data in horizontal pipes. Uses superficial gas and liquid velocities directly as the axes, making it intuitive. The Mandhane map identifies elongated bubble, slug, stratified, wavy, annular-mist, and dispersed bubble regimes. It is valid primarily for horizontal flow in pipes of 1–6 inches diameter near atmospheric pressure.

Taitel and Dukler (1976): The first mechanistic flow regime prediction model for horizontal flow. Rather than empirical boundaries, it predicts transitions from physical criteria:

• Stratified → slug/annular: Kelvin-Helmholtz interfacial instability criterion
• Slug → dispersed bubble: Turbulent breakup overcoming buoyancy coalescence
• Stratified smooth → stratified wavy: Wind-wave generation criterion

The Taitel-Dukler model uses dimensionless groups derived from the two-fluid model and is more physically based than the empirical maps. It generalizes better to different fluid properties and pipe diameters.

8.12.2 Applicability Guidelines

8.12.3 Practical Selection Guidance

For engineering design, the selection of correlation depends on the application:

1. Horizontal or near-horizontal flowlines (< 10°): The Taitel-Dukler model for flow regime prediction, combined with Beggs and Brill or OLGA for pressure drop calculation.
2. Vertical or highly inclined risers and wells: The Taitel-Barnea-Dukler (1980) vertical flow model, combined with Hagedorn-Brown or Beggs and Brill for pressure drop.
3. Undulating terrain with mixed inclinations: Beggs and Brill handles all angles in a single framework, making it practical for long pipelines with varying profile. Mechanistic models (OLGA, LedaFlow) provide better accuracy.
4. Preliminary sizing and screening: Baker or Mandhane maps provide quick flow regime identification. Beggs and Brill gives adequate pressure drop estimates (±20–30%).
5. Detailed design and transient analysis: Mechanistic simulators (OLGA, LedaFlow) are recommended for final design, especially when slugging dynamics, terrain effects, or operational transients are important.

Regardless of the correlation used, multiphase flow calculations should always be validated against field data or flow loop data when available. NeqSim provides the Beggs and Brill correlation as its built-in method, which gives reliable results for most steady-state engineering calculations.

8.13 Complete Flowline-Riser System Model

The following example models a complete subsea production system — wellhead through flowline and riser to topside separator:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define production fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 80.0, 180.0)
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 2.0)
fluid.addComponent("methane", 60.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("i-butane", 1.0)
fluid.addComponent("n-butane", 2.5)
fluid.addComponent("i-pentane", 1.0)
fluid.addComponent("n-pentane", 1.5)
fluid.addComponent("n-hexane", 2.5)
fluid.addComponent("n-heptane", 5.0)
fluid.addComponent("n-octane", 4.0)
fluid.addComponent("n-nonane", 2.5)
fluid.addComponent("water", 4.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)

Stream = jneqsim.process.equipment.stream.Stream
PipeBeggsAndBrills = jneqsim.process.equipment.pipeline.PipeBeggsAndBrills
Separator = jneqsim.process.equipment.separator.Separator
ProcessSystem = jneqsim.process.processmodel.ProcessSystem

# Wellhead conditions
wellstream = Stream("Wellhead", fluid)
wellstream.setFlowRate(90000.0, "kg/hr")
wellstream.setTemperature(80.0, "C")
wellstream.setPressure(180.0, "bara")

# Subsea flowline: 18 km horizontal
flowline = PipeBeggsAndBrills("Subsea Flowline", wellstream)
flowline.setPipeWallRoughness(5.0e-5)
flowline.setLength(18000.0)
flowline.setElevation(0.0)
flowline.setDiameter(0.254)
flowline.setNumberOfIncrements(60)
flowline.setConstantSurfaceTemperature(277.15)

flowline.setHeatTransferCoefficient(5.0)  # assumed insulated-pipe U, W/(m2 K)
# Riser: 500 m water depth
riser = PipeBeggsAndBrills("Production Riser", flowline.getOutletStream())
riser.setPipeWallRoughness(5.0e-5)
riser.setLength(600.0)
riser.setElevation(500.0)
riser.setDiameter(0.254)
riser.setNumberOfIncrements(20)
riser.setConstantSurfaceTemperature(280.15)

riser.setHeatTransferCoefficient(5.0)  # assumed insulated-pipe U, W/(m2 K)
# HP Separator at topside
hp_sep = Separator("HP Separator", riser.getOutletStream())

# Build and run
process = ProcessSystem()
process.add(wellstream)
process.add(flowline)
process.add(riser)
process.add(hp_sep)
process.run()

# Print results
print("=== System Results ===")
print(f"Wellhead:   P = {wellstream.getPressure('bara'):.1f} bara, T = {wellstream.getTemperature('C'):.1f} °C")

fl_out_P = flowline.getOutletStream().getPressure("bara")
fl_out_T = flowline.getOutletStream().getTemperature("C")
print(f"Flowline end: P = {fl_out_P:.1f} bara, T = {fl_out_T:.1f} °C")

rs_out_P = riser.getOutletStream().getPressure("bara")
rs_out_T = riser.getOutletStream().getTemperature("C")
print(f"Topside:    P = {rs_out_P:.1f} bara, T = {rs_out_T:.1f} °C")

gas_rate = hp_sep.getGasOutStream().getFlowRate("MSm3/day")
oil_rate = hp_sep.getLiquidOutStream().getFlowRate("m3/hr")
print(f"Gas rate:   {gas_rate:.3f} MSm3/day")
print(f"Oil rate:   {oil_rate:.1f} m3/hr")

The negative vertical coordinate denotes elevation below sea level. This is a geometric sketch, not a pressure/temperature profile; it must be supplied to a hydraulic and thermal calculation.

Reproduced Calculation Results

These examples use the stated fluid recipes and operating assumptions. Curves represent NeqSim calculations unless a caption identifies an analytical illustration, assumed equipment map or synthetic data.

Pressure Drop spans 0.2395–75.79 bar across the plotted cases. Gaps retain undefined phase quantities or hydraulic states that fail the stated operating boundary; they are not interpolated.

A larger bore reduces velocity and frictional pressure loss at fixed rate; density, holdup and friction-factor changes modify the simple diameter scaling. A small increase in bore can recover pressure margin, but excessive size may affect low-rate liquid transport and project cost. Select a diameter from the feasible arrival-pressure envelope and check turndown, holdup, erosion and installation constraints.

Pressure Drop spans 0.3547–55.31 bar across the plotted cases.

Raising mass flow increases velocity, friction and possibly acceleration losses as the gas expands along the line. The pressure-drop trend can steepen near the hydraulic limit; a failed forward solve is not a valid high-flow data point. Bracket the feasible flow interval, retain failure diagnostics and solve the arrival-pressure boundary directly when determining capacity.

Average Liquid Holdup spans 0.07955–0.1663 across the plotted cases.

Gas shear and liquid slip compete with gravity to determine the amount of liquid retained in the pipe. Holdup is a local flow-correlation result and differs from the inlet liquid fraction; it affects both hydrostatic loss and liquid inventory. Compare the selected correlation with representative operating data and examine the low-gas-rate regime for liquid accumulation.

The sampled pipe diameters span 6–16 inches. Bisection finds throughputs of 33.42–426.88 t/hr at the 20 bar pressure-loss allowance; the search upper bound was checked to lie beyond the capacity boundary.

The hydraulic capacity is a boundary solution: the maximum feasible rate satisfying the available pressure-drop or arrival-pressure constraint. If the numerical search stops at its rate bound, the resulting point is a bound-limited result rather than a demonstrated pipeline limit. Expand the bracket where needed and report active pressure limits, residuals and the feasibility of the final trial.

Selected numerical ranges from the plotted cases:

Ranges describe the sampled cases; they are not independent validation tolerances.

8.14 Summary

Key points from this chapter:

• Single-phase pipe flow is governed by the Darcy-Weisbach equation with the Colebrook-White friction factor for turbulent flow
• Multiphase flow involves distinct flow regimes (stratified, slug, annular, bubble) that depend on phase velocities, fluid properties, and pipe geometry
• The Beggs and Brill correlation provides a comprehensive method for multiphase pressure drop in inclined pipes, including liquid holdup prediction
• Heat transfer in pipelines is characterized by the overall $U$-value; typical values range from 2 W/m²·K (pipe-in-pipe) to 50 W/m²·K (bare pipe)
• Riser hydraulics involve significant hydrostatic pressure effects and the risk of severe slugging at low flow rates
• Pipeline sizing balances capital cost against pressure drop and velocity constraints
• NeqSim's PipeBeggsAndBrills class handles multiphase pressure drop, temperature profiles, and elevation effects
• The Joule-Thomson effect can contribute significantly to cooling in high-pressure gas systems

Verification of the worked examples

The completed route calculations are checked for mass/component conservation, positive profile states and pressure-loss direction. The fixed-property thermal characteristic length is independently checked by dimensional analysis; the NeqSim thermal closure itself is not independently calibrated here. Correlation accuracy requires matched measurements, and a gravity/friction comparison is not treated as a sufficient slack-flow or slugging criterion.[20]

The calculation and literal-code records are in verification/scientific_revision/ch08_manuscript_physics.json; the chapter scope and code hashes are indexed in foundations_review.json.

Exercises

1. Exercise 8.1: Calculate the friction factor and pressure drop for a 100 km, 36-inch gas export pipeline carrying 20 MSm³/day of dry gas at 150 bara and 25°C. Use NeqSim to determine gas properties and compare the analytical Darcy-Weisbach result with the NeqSim PipeBeggsAndBrills calculation.

1. Exercise 8.2: For the multiphase fluid in Section 8.5.1, plot the pressure drop per unit length as a function of flow rate (10,000–200,000 kg/hr) at constant inlet pressure. Identify the flow rate at which the flow regime transitions from stratified to slug flow.

1. Exercise 8.3: A subsea flowline carries wet gas (GOR = 5,000 Sm³/Sm³) over 30 km at 4°C seabed temperature. The inlet conditions are 150 bara and 60°C. Compare arrival temperatures for $U$ = 3, 10, and 30 W/m²·K. At what flow rate does the arrival temperature fall below 20°C for each $U$-value?

1. Exercise 8.4: Model a riser of 1,200 m height with a 10-inch diameter carrying a production fluid at 60,000 kg/hr. Calculate the pressure drop at the riser for gas-oil ratios of 100, 500, 1,000, and 5,000 Sm³/Sm³. Plot the results and explain the trend.

1. Exercise 8.5: A production flowline has an undulating profile with three 30 m elevation changes over its 15 km length. Model this terrain using NeqSim with three pipe segments and calculate the total pressure drop. Compare with the flat terrain case.

1. Exercise 8.6: For the complete flowline-riser system in Section 8.11, determine the minimum production rate below which the arrival temperature drops below the hydrate equilibrium temperature (calculate the hydrate temperature at the arrival pressure). This is the critical turndown rate.

1. Exercise 8.7: Estimate the pig-generated slug volume for a 50 km, 12-inch flowline with an average liquid holdup of 3%. If the slug catcher can only handle 100 m³, calculate the number of pigging passes needed to clear the line.

1. Beggs, H.D. and Brill, J.P. (1973). "A study of two-phase flow in inclined pipes." Journal of Petroleum Technology, 25(5), 607–617.
2. Taitel, Y. and Dukler, A.E. (1976). "A model for predicting flow regime transitions in horizontal and near horizontal gas-liquid flow." AIChE Journal, 22(1), 47–55.
3. Brill, J.P. and Mukherjee, H. (1999). Multiphase Flow in Wells. SPE Monograph Volume 17. Society of Petroleum Engineers.
4. Bøe, A. (1981). "Severe slugging characteristics." Selected Topics in Two-Phase Flow. NTH Trondheim.
5. API RP 14E (1991). Recommended Practice for Design and Installation of Offshore Production Platform Piping Systems. American Petroleum Institute.
6. Shoham, O. (2006). Mechanistic Modeling of Gas-Liquid Two-Phase Flow in Pipes. SPE.
7. Mokhatab, S., Poe, W.A., and Mak, J.Y. (2019). Handbook of Natural Gas Transmission and Processing, 4th Edition. Gulf Professional Publishing.
8. Guo, B., Song, S., Chacko, J., and Ghalambor, A. (2005). Offshore Pipelines: Design, Installation, and Maintenance. Gulf Professional Publishing.
9. Moody, L.F. (1944). "Friction factors for pipe flow." Transactions of the ASME, 66, 671–684.
10. Colebrook, C.F. (1939). "Turbulent flow in pipes with particular reference to the transition between the smooth and rough pipe laws." Journal of the Institution of Civil Engineers, 11, 133–156.

Running the examples. Start the source-workspace Python session described in Chapter 1, then run this chapter's Python blocks in reading order. Java blocks form a separate sequence using the same NeqSim build; carry forward objects from preceding Java blocks. The release execution records are in verification/; a successful run establishes API compatibility, while physical validation also requires the checks discussed in the text.

Learning Objectives

After reading this chapter, the reader will be able to:

1. Explain the thermodynamic basis for gas hydrate formation and predict hydrate equilibrium conditions
2. Design hydrate prevention strategies using thermodynamic inhibitors (MEG, methanol) and calculate inhibitor dosing rates
3. Assess wax deposition risk using wax appearance temperature calculations and operational strategies
4. Evaluate asphaltene stability using de Boer screening and the colloidal instability index
5. Predict CO$_2$ and H$_2$S corrosion rates using the de Waard-Milliams model
6. Screen for scale formation risk (CaCO$_3$, BaSO$_4$) based on water chemistry
7. Perform comprehensive flow assurance screening using NeqSim to define safe operating envelopes

9.1 Introduction to Flow Assurance

Flow assurance — a term coined by Petrobras in the 1990s (from the Portuguese garantia de fluxo) — encompasses all engineering activities that ensure the uninterrupted transport of hydrocarbon fluids from reservoir to market. The discipline addresses the threats that can restrict or stop production by blocking flowlines, corroding pipe walls, or destabilizing the multiphase flow.

The principal flow assurance threats are:

The economic impact of flow assurance failures is substantial. A single hydrate plug can cost \$1–10 million to remediate and months of lost production. A corrosion-induced leak can result in environmental damage, regulatory penalties, and production shutdown.

9.2 Gas Hydrates

9.2.1 What Are Gas Hydrates?

Gas hydrates (also called clathrate hydrates) are crystalline solid compounds formed when water molecules create cage-like structures that trap small gas molecules. They form at high pressures and low temperatures — conditions commonly encountered in subsea flowlines and deepwater risers.

The water molecules form a hydrogen-bonded lattice with cavities that encage guest molecules (methane, ethane, propane, CO$_2$, H$_2$S, nitrogen). The thermodynamic stability of hydrates depends on:

• Temperature — lower temperatures favor hydrate formation
• Pressure — higher pressures favor hydrate formation
• Gas composition — different guest molecules stabilize different hydrate structures
• Water presence — free water is required for hydrate formation

9.2.2 Hydrate Structures

Three principal hydrate crystal structures are known:

Natural gas typically forms Structure II hydrates because propane and isobutane stabilize the large sII cages. However, gas with very low C3+ content (e.g., biogenic gas or CO$_2$-rich gas) may form Structure I.

9.2.3 Hydrate Equilibrium Thermodynamics

The van der Waals-Platteeuw (vdWP) model describes the thermodynamic stability of hydrate phases. The chemical potential of water in the hydrate phase relative to the empty lattice is:

where:

• $\Delta\mu_w^H=\mu_w^{empty}-\mu_w^{hydrate}$ is the positive empty-lattice minus occupied-lattice reference difference in this equation
• $\nu_i$ is the number of type $i$ cages per water molecule
• $\theta_{ij}$ is the fractional occupancy of cage $i$ by guest $j$

The cage occupancy follows a Langmuir-type model:

where $C_{ij}$ is the Langmuir constant for guest $j$ in cage $i$, and $f_j$ is the fugacity of guest $j$ in the fluid phase.

At equilibrium, the chemical potential of water is equal in all coexisting phases (hydrate, liquid water, ice):

This equilibrium condition defines the hydrate P-T curve — the locus of temperature and pressure at which hydrates form.

9.2.4 Hydrate P-T Curve Calculation with NeqSim

NeqSim calculates hydrate equilibrium using a rigorous implementation of the vdWP model coupled with its equation of state:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define a typical natural gas composition
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 20.0, 100.0)
fluid.addComponent("nitrogen", 1.0)
fluid.addComponent("CO2", 3.5)
fluid.addComponent("methane", 72.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 4.5)
fluid.addComponent("i-butane", 1.0)
fluid.addComponent("n-butane", 2.0)
fluid.addComponent("n-pentane", 1.0)
fluid.addComponent("water", 7.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)
fluid.setHydrateCheck(True)

# Calculate hydrate equilibrium temperature at various pressures
ThermodynamicOperations = jneqsim.thermodynamicoperations.ThermodynamicOperations

pressures_bara = [20, 40, 60, 80, 100, 150, 200, 250, 300]

print(f"{'Pressure (bara)':>16} {'Hydrate T (°C)':>16}")
print("-" * 34)

for P in pressures_bara:
    test_fluid = fluid.clone()
    test_fluid.setPressure(float(P), "bara")
    ops = ThermodynamicOperations(test_fluid)
    try:
        ops.hydrateFormationTemperature()
        T_hyd = test_fluid.getTemperature("C")
        print(f"{P:>16} {T_hyd:>16.1f}")
    except Exception as e:
        print(f"{P:>16} {'N/A':>16}")

9.2.5 Hydrate Phase Envelope

The following calculation tabulates hydrate-equilibrium temperature against pressure for the stated composition. Compare these values with a separately calculated operating path and its uncertainty; this example does not plot an inhibited boundary or establish blockage probability.

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define gas condensate fluid
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 20.0, 100.0)
fluid.addComponent("nitrogen", 0.5)
fluid.addComponent("CO2", 2.0)
fluid.addComponent("methane", 75.0)
fluid.addComponent("ethane", 8.0)
fluid.addComponent("propane", 5.0)
fluid.addComponent("i-butane", 1.0)
fluid.addComponent("n-butane", 2.5)
fluid.addComponent("i-pentane", 0.5)
fluid.addComponent("n-pentane", 1.0)
fluid.addComponent("n-hexane", 1.0)
fluid.addComponent("n-heptane", 1.0)
fluid.addComponent("water", 2.5)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)
fluid.setHydrateCheck(True)

ThermodynamicOperations = jneqsim.thermodynamicoperations.ThermodynamicOperations

# Calculate hydrate curve
pressures = [10, 20, 30, 50, 70, 100, 150, 200, 250, 300]
hydrate_temps = []

for P in pressures:
    test_fluid = fluid.clone()
    test_fluid.setPressure(float(P), "bara")
    ops = ThermodynamicOperations(test_fluid)
    try:
        ops.hydrateFormationTemperature()
        T_hyd = test_fluid.getTemperature("C")
        hydrate_temps.append(T_hyd)
    except Exception:
        hydrate_temps.append(None)

# Print hydrate curve
print("Hydrate Equilibrium Curve:")
print(f"{'P (bara)':>10} {'T_hyd (°C)':>12}")
for P, T in zip(pressures, hydrate_temps):
    if T is not None:
        print(f"{P:>10} {T:>12.1f}")

9.2.6 Hydrate Subcooling and Risk Assessment

Subcooling measures the temperature driving force relative to the calculated hydrate-equilibrium boundary. Blockage risk also depends on free-water availability, kinetics, residence time, flow history and inhibition; subcooling alone does not determine its severity:

The following bins are illustrative screening categories, not validated blockage-risk levels or universal intervention requirements. Select mitigation using the fluid, water and operating-history evidence for the case.

9.3 Hydrate Inhibition

9.3.1 Thermodynamic Inhibitors

Thermodynamic inhibitors (THIs) shift the hydrate equilibrium curve to lower temperatures (or higher pressures), expanding the safe operating envelope. The two primary THIs used in the oil and gas industry are:

Mono-ethylene glycol (MEG):

• Most common offshore inhibitor
• Can be regenerated and recycled (closed-loop system)
• Effective inhibition of 20–40°C depending on concentration
• Typical lean MEG concentration: 80–90 wt%
• Typical rich MEG concentration: 30–50 wt% in the water phase

Methanol:

• Lower cost per unit volume
• Can be recovered by appropriate separation/distillation, but losses and recovery economics often favor once-through injection
• Partitions into gas and hydrocarbon liquid phases (higher losses)
• Flash point concerns for storage and handling

9.3.2 The Hammerschmidt Equation

The classical Hammerschmidt equation estimates the hydrate depression temperature:

where:

• $\Delta T$ is the hydrate temperature depression [°C]
• $K_H$ is the Hammerschmidt constant (1,297 for both MEG and methanol with depression in °C;2,335 belongs to the °F convention)
• $w$ is the inhibitor concentration in the aqueous phase [wt%]
• $M_i$ is the molecular weight of the inhibitor [g/mol] (62.07 for MEG; 32.04 for methanol)

Rearranging gives $w=100\Delta T M_i/(1297+\Delta T M_i)$. For an illustrative10°C depression it gives32.4 wt%MEG and 19.8 wt%methanol. The GPSA SI erratum explicitly specifies1297 for both compounds.[21] This low-order aqueous-phase estimate does not include vapor/hydrocarbon partitioning, salinity, inhibitor concentration limits or freeze/viscosity constraints. The injected concentration and flow must satisfy an overall inhibitor/water balance; use a suitable equilibrium model and data before extrapolating to concentrated solutions.

9.3.3 MEG Dosing Calculation with NeqSim

NeqSim provides rigorous hydrate inhibition calculations by including the inhibitor as a component in the thermodynamic model:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define gas with water and MEG
fluid_with_MEG = jneqsim.thermo.system.SystemSrkCPAstatoil(273.15 + 20.0, 100.0)
fluid_with_MEG.addComponent("methane", 75.0)
fluid_with_MEG.addComponent("ethane", 8.0)
fluid_with_MEG.addComponent("propane", 5.0)
fluid_with_MEG.addComponent("i-butane", 1.0)
fluid_with_MEG.addComponent("n-butane", 2.0)
fluid_with_MEG.addComponent("CO2", 2.0)
fluid_with_MEG.addComponent("water", 5.0)
fluid_with_MEG.addComponent("MEG", 2.0)   # MEG injection
fluid_with_MEG.setMixingRule(10)  # CPA mixing rule
fluid_with_MEG.setMultiPhaseCheck(True)
fluid_with_MEG.setHydrateCheck(True)

ThermodynamicOperations = jneqsim.thermodynamicoperations.ThermodynamicOperations

# Calculate hydrate temperature with MEG
ops = ThermodynamicOperations(fluid_with_MEG)
fluid_with_MEG.setPressure(100.0, "bara")
ops.hydrateFormationTemperature()
T_hyd_with_MEG = fluid_with_MEG.getTemperature("C")
print(f"Hydrate temperature with MEG: {T_hyd_with_MEG:.1f} °C")

# Compare with uninhibited fluid
fluid_no_MEG = jneqsim.thermo.system.SystemSrkCPAstatoil(273.15 + 20.0, 100.0)
fluid_no_MEG.addComponent("methane", 75.0)
fluid_no_MEG.addComponent("ethane", 8.0)
fluid_no_MEG.addComponent("propane", 5.0)
fluid_no_MEG.addComponent("i-butane", 1.0)
fluid_no_MEG.addComponent("n-butane", 2.0)
fluid_no_MEG.addComponent("CO2", 2.0)
fluid_no_MEG.addComponent("water", 7.0)
fluid_no_MEG.setMixingRule(10)
fluid_no_MEG.setMultiPhaseCheck(True)
fluid_no_MEG.setHydrateCheck(True)

ops2 = ThermodynamicOperations(fluid_no_MEG)
fluid_no_MEG.setPressure(100.0, "bara")
ops2.hydrateFormationTemperature()
T_hyd_no_MEG = fluid_no_MEG.getTemperature("C")

print(f"Hydrate temperature without MEG: {T_hyd_no_MEG:.1f} °C")
print(f"Hydrate depression: {T_hyd_no_MEG - T_hyd_with_MEG:.1f} °C")

9.3.4 Low Dosage Hydrate Inhibitors (LDHIs)

In addition to thermodynamic inhibitors, Low Dosage Hydrate Inhibitors (LDHIs) are increasingly used. LDHIs are classified into two distinct categories based on their mechanism of action:

Kinetic Hydrate Inhibitors (KHIs):

KHIs are water-soluble polymers (typically polyvinylpyrrolidone (PVP), polyvinylcaprolactam (PVCap), or copolymers thereof) that delay hydrate nucleation and slow crystal growth without shifting the thermodynamic equilibrium. Key characteristics:

• Polymer-based chemicals that adsorb on the hydrate crystal surface and disrupt nucleation
• Effective subcooling typically 6–14°C (the maximum subcooling below the equilibrium temperature at which the KHI can prevent hydrate formation for the required hold time)
• Dosage: 0.5–3.0 wt% of the water phase (typically 0.5–1.5 wt% for moderate subcooling)
• The hold time — the period over which the KHI can delay hydrate formation — depends on subcooling, polymer type, and concentration. Typical hold times range from 12 to 72 hours
• Advantages: much lower injection volumes than THIs (factor of 10–100 less), no regeneration plant needed, lower CAPEX and OPEX for moderate subcooling applications
• Limitations: time-dependent — effective for hours to days, not indefinitely; limited to subcooling less than approximately 14°C; performance sensitive to brine salinity and production chemical interactions; not suitable for shut-in protection without supplementary strategies

Anti-agglomerants (AAs):

AAs are surfactant-based chemicals (typically quaternary ammonium salts or similar amphiphilic molecules) that do not prevent hydrate formation but instead prevent hydrate crystals from agglomerating into large masses. Hydrate particles remain dispersed as a transportable slurry in the hydrocarbon liquid phase:

• Allow hydrate formation but keep crystals small (typically < 100 µm) and well-dispersed
• The hydrate slurry flows through the pipeline without risk of blockage, provided the hydrate volume fraction remains below approximately 20–30%
• Dosage: 0.5–2.0 wt% of the water phase
• Require an oil-continuous system (limited to < 50–60% watercut); in water-continuous systems, the dispersed hydrate particles can bridge and form plugs
• Advantages: effective at any subcooling (no subcooling limit); provide shut-in protection (hydrate crystals remain dispersed even during extended shutdown)
• Limitations: watercut ceiling; require liquid hydrocarbon phase; potential emulsion stabilization in separators; higher unit cost than KHIs

9.3.5 Comparison of Hydrate Inhibition Strategies

The choice between THI, KHI, and AA depends on the subcooling, watercut, shut-in duration, and economic factors:

For deepwater developments with subcooling > 15°C, THIs (typically MEG) remain the default choice. KHIs are increasingly used for satellite wells and short tiebacks with moderate subcooling. AAs are attractive for oil-dominated systems where the watercut remains below the inversion point.

9.3.6 Hydrate Management Strategies

9.4 Wax Deposition

9.4.1 Wax Formation Mechanism

Wax (paraffin) deposition occurs when dissolved long-chain alkanes (typically $n$-C$_{18}$ to $n$-C$_{60}$) crystallize out of the oil as the temperature decreases. The key temperatures are:

Wax Appearance Temperature (WAT): The temperature at which the first wax crystals form. This is the cloud point of the oil.

Pour Point: The temperature below which the oil ceases to flow due to wax gelation.

The wax deposition process involves:

1. Nucleation — first wax crystals form when $T < WAT$
2. Crystal growth — wax molecules diffuse from the bulk oil to the crystal surface
3. Deposition — wax crystals deposit on the cold pipe wall by molecular diffusion (dominant mechanism), Brownian diffusion, shear dispersion, and gravity settling
4. Aging — the deposited wax layer hardens over time as lighter hydrocarbons diffuse out

9.4.2 Wax Deposition Rate

The Singh et al. (2000) model for wax deposition rate by molecular diffusion is:

where:

• $m_w$ is the deposited wax mass [kg/m²]
• $D_{eff}$ is the effective diffusion coefficient of wax in oil [m²/s]
• $dC/dr$ is the radial concentration gradient of dissolved wax at the pipe wall
• $R$ is the pipe radius [m]

The diffusion coefficient depends on temperature through an Arrhenius relationship:

9.4.3 Wax Management

9.4.4 Wax Calculations with NeqSim

NeqSim can estimate the wax appearance temperature through flash calculations that include the solid wax phase:

import jpype
jneqsim = jpype.JPackage("neqsim")

# Define a waxy crude oil composition (C7+ with heavy tail)
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 60.0, 50.0)
fluid.addComponent("methane", 30.0)
fluid.addComponent("ethane", 5.0)
fluid.addComponent("propane", 4.0)
fluid.addComponent("n-butane", 3.0)
fluid.addComponent("n-pentane", 3.0)
fluid.addComponent("n-hexane", 4.0)
fluid.addComponent("n-heptane", 8.0)
fluid.addComponent("n-octane", 8.0)
fluid.addComponent("n-nonane", 6.0)
fluid.addComponent("n-decane", 5.0)
fluid.addComponent("nC11", 4.0)
fluid.addComponent("nC17", 6.0)
fluid.addComponent("nC20", 5.0)
fluid.addComponent("water", 9.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)
fluid.setHydrateCheck(True)

# Liquid-property cooling screen only: TPflash without a wax model cannot determine WAT.
ThermodynamicOperations = jneqsim.thermodynamicoperations.ThermodynamicOperations

temperatures = [60, 55, 50, 45, 40, 35, 30, 25, 20, 15, 10]
print(f"{'T (°C)':>8} {'Phases':>8} {'Liquid density (kg/m3)':>22}")

for T in temperatures:
    test_fluid = fluid.clone()
    test_fluid.setTemperature(float(T) + 273.15, "K")
    test_fluid.setPressure(50.0, "bara")
    ops = ThermodynamicOperations(test_fluid)
    ops.TPflash()
    test_fluid.initProperties()
    n_phases = test_fluid.getNumberOfPhases()
    rho_oil = test_fluid.getPhase("oil").getDensity("kg/m3")
    print(f"{T:>8} {n_phases:>8} {rho_oil:>22.1f}")

9.5 Asphaltene Stability

9.5.1 Asphaltene Chemistry

Asphaltenes are the heaviest, most polar fraction of crude oil — defined operationally as the fraction insoluble in $n$-heptane but soluble in toluene. They consist of large polyaromatic hydrocarbons with heteroatom-containing functional groups (N, O, S) and have molecular weights in the range 500–2,000 g/mol.

Asphaltenes are normally stabilized in the crude oil by resins (polar aromatics that form a solvation shell around asphaltene particles). Destabilization occurs when:

• Pressure depletion above the bubble point can reduce asphaltene solubility; below the bubble point, gas liberation changes the liquid composition and can promote redissolution
• Composition changes from commingling or CO$_2$/lean gas injection
• Temperature changes (complex effect — can increase or decrease stability)

9.5.2 de Boer Screening

The de Boer et al. (1995) screening method uses the difference between reservoir pressure and bubble point pressure as the primary indicator:

The original screening trend associates greater susceptibility with light, low-density oils that are strongly undersaturated with gas; heavier oils near their bubble point can be less susceptible despite higher asphaltene content. A rectangular low/medium/high table reverses and oversimplifies that trend. Use the original chart and measured live-oil density, saturation pressure and precipitation-onset data; pressure undersaturation alone is not a precipitation criterion.[22]

9.5.3 Colloidal Instability Index (CII)

The CII uses SARA (Saturates, Aromatics, Resins, Asphaltenes) fractionation data:

9.5.4 Asphaltene Onset Pressure

The asphaltene onset pressure (AOP) is the pressure at which asphaltenes first precipitate during isothermal depressurization. It is typically measured by depressurization experiments with near-infrared detection. The AOP is usually above the bubble point:

An upper onset can lie above bubble pressure, while a lower redissolution boundary may lie below it. Bubble pressure is not itself the lower precipitation boundary. Determine both onsets along the actual composition and temperature path, then assess deposition kinetics, transport, residence time and tested mitigation before choosing an operating envelope.

9.6 Corrosion

9.6.1 CO$_2$ Corrosion

CO$_2$ corrosion (sweet corrosion) is the most common form of internal corrosion in oil and gas pipelines. Dissolved CO$_2$ forms carbonic acid in the presence of water:

The carbonic acid attacks the steel surface:

The resulting iron carbonate (FeCO$_3$, siderite) can form a protective scale if conditions are favorable (temperature > 60–80°C, low flow velocity).

9.6.2 The de Waard-Milliams Model

The de Waard and Milliams (1975) model, updated by de Waard, Lotz, and Milliams (1991), is the most widely used empirical CO$_2$ corrosion model:

where:

• $CR$ is the corrosion rate [mm/year]
• $T$ is the temperature [°C]
• $P_{CO_2}$ is the partial pressure of CO$_2$ [bar]

Correction factors are applied for:

9.6.3 CO$_2$ Corrosion Rate Estimation with NeqSim

NeqSim can calculate the CO$_2$ partial pressure and water chemistry needed for corrosion estimation:

import jpype
jneqsim = jpype.JPackage("neqsim")
import math

# Define production fluid with CO2
fluid = jneqsim.thermo.system.SystemSrkEos(273.15 + 60.0, 80.0)
fluid.addComponent("methane", 70.0)
fluid.addComponent("ethane", 7.0)
fluid.addComponent("propane", 4.0)
fluid.addComponent("CO2", 5.0)      # 5 mol% CO2
fluid.addComponent("n-butane", 2.0)
fluid.addComponent("n-hexane", 3.0)
fluid.addComponent("n-heptane", 4.0)
fluid.addComponent("water", 5.0)
fluid.setMixingRule("classic")
fluid.setMultiPhaseCheck(True)
fluid.setHydrateCheck(True)

# Flash to get CO2 partial pressure in gas phase
ThermodynamicOperations = jneqsim.thermodynamicoperations.ThermodynamicOperations
ops = ThermodynamicOperations(fluid)
ops.TPflash()
fluid.initProperties()

# Get CO2 mole fraction in gas phase
y_CO2 = fluid.getPhase("gas").getComponent("CO2").getx()
P_total = fluid.getPressure("bara")
P_CO2 = y_CO2 * P_total

# de Waard-Milliams corrosion rate
T = 60.0  # °C
CR = 10.0**(5.8 - 1710.0 / (T + 273.0) + 0.67 * math.log10(P_CO2))

print(f"Total pressure: {P_total:.1f} bara")
print(f"CO2 mole fraction in gas: {y_CO2:.4f}")
print(f"CO2 partial pressure: {P_CO2:.2f} bar")
print(f"Estimated corrosion rate: {CR:.2f} mm/year")

# Corrosion rate at different temperatures
print(f"\n{'Temperature (°C)':>18} {'CR (mm/year)':>14}")
for T in [20, 40, 60, 80, 100, 120]:
    CR_T = 10.0**(5.8 - 1710.0 / (T + 273.0) + 0.67 * math.log10(P_CO2))
    print(f"{T:>18} {CR_T:>14.2f}")

9.6.4 H$_2$S Corrosion (Sour Corrosion)

H$_2$S corrosion introduces additional mechanisms beyond CO$_2$ corrosion:

• Sulfide stress cracking (SSC) — hydrogen embrittlement of high-strength steels
• Stress-oriented hydrogen-induced cracking (SOHIC) — combined stress and hydrogen effects
• Pitting corrosion — localized attack under iron sulfide (FeS) deposits

The NACE MR0175 / ISO 15156 standard defines material requirements for sour service based on H$_2$S partial pressure and pH:

9.6.5 Corrosion Allowance and Material Selection

The corrosion allowance is the extra wall thickness added to account for metal loss over the design life:

For a typical 25-year design life with $CR = 0.3$ mm/year: $CA = 7.5$ mm. If the corrosion rate exceeds approximately 0.3 mm/year with inhibition, corrosion-resistant alloys (CRAs) are typically selected:

9.7 Scale Prediction

9.7.1 Common Scale Types

Scale deposits form when the produced water becomes supersaturated with dissolved minerals: