Agentic Workflow Examples

This document shows how NeqSim’s AI agents handle real engineering problems, from simple single-agent interactions to complex multi-agent pipelines. Each example shows the prompt, what happens behind the scenes, and the output you receive.

Prerequisites: Read Introduction to Agentic Engineering for the underlying concepts, and Agents Reference for the full agent catalog.

Example 1: Simple Property Lookup

Difficulty: Beginner Agent: @thermo.fluid Time: seconds

The Prompt

@thermo.fluid What is the density and viscosity of methane at 200 bara and 40°C?

What the Agent Does

  1. Selects EOS — SRK for a simple hydrocarbon
  2. Creates fluid — SystemSrkEos(273.15 + 40.0, 200.0) with methane
  3. Sets mixing rule — "classic"
  4. Runs flash — TPflash() to determine the phase state
  5. Initializes properties — initProperties() (critical step for transport props)
  6. Reads values — density via getDensity("kg/m3"), viscosity via phase methods

Behind the Scenes

The agent loads two skills automatically:

The Output

At 200 bara and 40°C, methane is a supercritical fluid:
- Density: 162.3 kg/m³
- Dynamic viscosity: 2.45 × 10⁻⁵ Pa·s
- Compressibility factor Z: 0.785

Example 2: Process Simulation Notebook

Difficulty: Intermediate Agent: @solve.process Time: minutes

The Prompt

@solve.process Build a 3-stage compression train from 5 to 150 bara with
intercooling to 35°C for a natural gas at 100,000 Sm3/day

What the Agent Does

  1. Plans the process:
    • Compression ratio per stage: $(150/5)^{1/3} \approx 3.11$
    • Stage 1: 5 → 15.6 bara
    • Stage 2: 15.6 → 48.4 bara
    • Stage 3: 48.4 → 150 bara
  2. Creates fluid: 
    fluid = SystemSrkEos(273.15 + 25.0, 5.0)
fluid.addComponent("methane", 0.85)
fluid.addComponent("ethane", 0.08)
fluid.addComponent("propane", 0.04)
fluid.addComponent("n-butane", 0.02)
fluid.addComponent("nitrogen", 0.01)
fluid.setMixingRule("classic")

  3. Builds flowsheet with 3 compressor + cooler pairs

  4. Runs simulation via process.run()

  5. Creates visualization — discharge temperature, power per stage, P-T diagram

Skills Loaded

Output

A complete Jupyter notebook with:

Example 3: Flow Assurance Study

Difficulty: Intermediate Agent: @flow.assurance Time: minutes

The Prompt

@flow.assurance Check if a wet gas pipeline from platform to shore has
hydrate risk. The gas is 90% methane, 5% ethane, 3% propane, 2% CO2
at 120 bara. Pipeline is 80 km subsea at 4°C seabed temperature.
What concentration of MEG is needed for inhibition?

What the Agent Does

  1. Creates CPA fluid — Uses SystemSrkCPAstatoil because water + MEG are polar molecules requiring the CPA equation of state

  2. Calculates hydrate curve — Sweeps pressure from 10 to 200 bara, finding the hydrate formation temperature at each pressure

  3. Compares with pipeline conditions — The pipeline operates at 120 bara with arrival temperature near 4°C. If the hydrate temperature at 120 bara is above 4°C, there is hydrate risk.

  4. Calculates MEG requirement — Adds MEG at increasing concentrations (10%, 20%, 30% weight in aqueous phase) until the hydrate curve shifts below the minimum pipeline temperature

  5. Generates plots:

    • Hydrate curve (P vs T) with pipeline operating point
    • MEG dosage sensitivity
    • Pipeline temperature profile

Skills Loaded

Output

Hydrate Analysis Results:
- Hydrate formation temperature at 120 bara: 18.5°C
- Pipeline minimum temperature: 4.0°C
- Subcooling: 14.5°C — SEVERE HYDRATE RISK

MEG Inhibition:
- 30 wt% MEG: hydrate T drops to 8.2°C — insufficient
- 40 wt% MEG: hydrate T drops to 2.1°C — SAFE (1.9°C margin)
- Recommended: 40 wt% MEG in water phase

Notebook saved with hydrate curve and MEG sensitivity plots.

Example 4: Full Engineering Task with Report

Difficulty: Advanced Agent: @solve.task Time: 30-60 minutes

The Prompt

@solve.task Design a TEG dehydration unit for a 50 MMSCFD wet natural gas
at 70 bara and 35°C. Target water dew point is -18°C per NORSOK P-001.
Include equipment sizing, TEG circulation rate, and reboiler duty.
Compare results against GPSA Engineering Data Book correlations.

What the Agent Does

Step 1 — Scope & Research

  1. Creates task folder: task_solve/2026-XX-XX_teg_dehydration_design/
  2. Fills task_spec.md:
    • Design standard: NORSOK P-001
    • Reference: GPSA Engineering Data Book, 14th Ed.
    • Acceptance criteria: water content < 30 ppm, pressure drop < 0.5 bar
  3. Researches TEG dehydration theory in notes.md

Step 2 — Analysis & Evaluation

  1. Creates main simulation notebook:
    • CPA fluid with natural gas + water
    • TEG absorption column (SimpleTEGAbsorber or DistillationColumn)
    • TEG regeneration (reboiler + stripping gas option)
    • Runs simulation and extracts results
  2. Creates benchmark notebook:
    • Compares water removal efficiency against GPSA Fig. 20-70
    • Validates TEG circulation rate against GPSA correlation
    • Creates parity plot (NeqSim vs GPSA)
  3. Creates uncertainty notebook:
    • Varies inlet water content, temperature, TEG lean purity
    • Monte Carlo with N=200 iterations
    • Tornado diagram showing sensitivity ranking
    • Risk register (equipment, operational, commercial risks)
  4. Saves results.json with all key numbers

Step 3 — Report

  1. Runs generate_report.py to produce:
    • Word document (.docx) with professional formatting
    • HTML document with interactive navigation
    • All figures embedded and numbered
    • References cited
    • Benchmark validation table (PASS/FAIL)
    • Uncertainty P10/P50/P90 table
    • Risk register with color-coded severity

Output Structure

task_solve/2026-XX-XX_teg_dehydration_design/
├── results.json
├── figures/
│   ├── teg_performance_vs_circulation_rate.png
│   ├── water_dew_point_sensitivity.png
│   ├── benchmark_parity_plot.png
│   ├── tornado_diagram.png
│   └── risk_matrix.png
├── step1_scope_and_research/
│   ├── task_spec.md
│   └── notes.md
├── step2_analysis/
│   ├── 01_teg_dehydration_design.ipynb
│   ├── 02_benchmark_validation.ipynb
│   └── 03_uncertainty_risk.ipynb
└── step3_report/
    ├── generate_report.py
    ├── TEG_Dehydration_Design_Report.docx
    └── TEG_Dehydration_Design_Report.html

Example 5: Multi-Agent Composition

Difficulty: Advanced Agent: @neqsim.help (routes to multiple) Time: varies

The Prompt

@neqsim.help I need to design a 20-inch gas export pipeline from an
offshore platform to shore. I need:
1. Steady-state hydraulics (pressure drop over 120 km)
2. Hydrate check and inhibitor requirement
3. Wall thickness per DNV-OS-F101
4. Cost estimation for pipeline and installation

How the Router Composes the Pipeline

The router detects four sub-tasks spanning three disciplines:

Request Analysis:
├── Sub-task 1: Pipeline hydraulics     → @process.model
├── Sub-task 2: Hydrate check           → @flow.assurance
├── Sub-task 3: Wall thickness          → @mechanical.design
└── Sub-task 4: Cost estimation         → @mechanical.design

Dependencies:

The router sequences them and passes results via the neqsim-agent-handoff skill schema.

Execution Sequence

Phase 1: @process.model creates a PipeBeggsAndBrills or AdiabaticPipe model for the 120 km pipeline. Output: pressure and temperature at every point along the pipeline.

Phase 2a: @flow.assurance takes the P,T profile and overlays the hydrate equilibrium curve. Determines if any point along the pipeline enters the hydrate region. Calculates MEG dosage if needed.

Phase 2b: @mechanical.design takes the maximum operating pressure and applies DNV-OS-F101 wall thickness calculation. Selects material grade (API 5L X65), applies design factors, corrosion allowance.

Phase 3: @mechanical.design uses wall thickness, pipeline length, material grade, and installation method to estimate costs using NeqSim’s SURF cost estimation classes.

Combined Output

A unified summary with results from all four sub-tasks, cross-referenced so that dependencies are explicit (e.g., “wall thickness of 22.3 mm based on maximum pressure of 142 bara at inlet from hydraulic simulation”).

Example 6: Capability Gap Discovery

Difficulty: Advanced Agent: @capability.scout Time: minutes

The Prompt

@capability.scout Can NeqSim handle HPHT (high pressure high temperature)
well design for a 280°C, 1200 bara reservoir with H2S and CO2?

What the Scout Returns

Capability Assessment for HPHT Well Design:

✅ FULL COVERAGE:
  - H2S/CO2 phase behavior (CPA EOS, multiple mixing rules)
  - Well casing design (API 5C3, NORSOK D-010)
  - Corrosion assessment (DeWaardMilliams, NACE MR0175)
  - Cost estimation (WellCostEstimator, SURFCostEstimator)

⚠️  PARTIAL COVERAGE:
  - HPHT EOS accuracy: SRK/PR may have reduced accuracy above 200°C
    Recommendation: Use PR-Peneloux with validated BIPs for H2S/CO2
  - Thermal stress: Basic tubing stress via API 5C3, but no thermal
    cycling fatigue model

🔴 GAPS:
  - No dedicated HPHT material selection model (CRA vs carbon steel at 280°C)
    NIP: Create HPHTMaterialSelector class in process.mechanicaldesign

Recommended Agent Pipeline:
  1. @thermo.fluid — Create HPHT fluid with H2S, CO2 at extreme conditions
  2. @flow.assurance — Corrosion assessment at reservoir conditions
  3. @mechanical.design — Casing design with HPHT design factors
  4. @solve.task — Full study with uncertainty and reporting

Example 7: Gas Quality Compliance

Difficulty: Beginner Agent: @gas.quality Time: seconds

The Prompt

@gas.quality Check if this gas meets EN 16726 H-gas specification:
methane 89%, ethane 6%, propane 2.5%, CO2 1.5%, N2 1%

What the Agent Does

  1. Creates fluid with the given composition
  2. Runs ISO 6976 calculations (superior heating value, Wobbe index, relative density)
  3. Compares results against EN 16726 limits for H-gas

Output

EN 16726 H-Gas Compliance Check:

| Property | Calculated | Limit | Status |
|----------|-----------|-------|--------|
| Wobbe Index | 51.2 MJ/m³ | 46.1 – 52.8 | PASS |
| Superior HV | 39.8 MJ/m³ | > 34.95 | PASS |
| Relative Density | 0.604 | < 0.700 | PASS |
| CO₂ | 1.5 mol% | < 2.5 | PASS |
| H₂S | 0 mg/m³ | < 5 | PASS |

Result: GAS MEETS EN 16726 H-GAS SPECIFICATION ✅

Example 8: Field Development Concept Selection

Difficulty: Advanced Agent: @field.development Time: 30-60 minutes

The Prompt

@field.development Evaluate two development concepts for a 25 km subsea
tieback of a lean gas field (85% methane, 8% ethane, 4% propane, 2% CO2,
1% N2) at 350 m water depth:
- Concept A: Direct tieback to existing host platform (20-inch pipeline)
- Concept B: Subsea compression with 16-inch pipeline
Resource estimate: 15 GSm3 (P50). Norwegian NCS fiscal regime.
Gas price 2.0 NOK/Sm³. Discount rate 8%.

What the Agent Does

  1. Creates task folder via devtools/new_task.py
  2. Loads skills: neqsim-field-development, neqsim-field-economics, neqsim-subsea-and-wells, neqsim-production-optimization
  3. Builds both concepts using FieldDevelopmentWorkflow with ConceptDefinition
  4. Runs pipeline hydraulics for each concept (PipeBeggsAndBrills)
  5. Estimates SURF CAPEX using SURFCostEstimator (regional factor: Norway)
  6. Generates production profiles with ProductionProfileGenerator
  7. Runs NPV analysis with NorwegianTaxModel (78% marginal tax rate)
  8. Performs Monte Carlo uncertainty analysis (N=200) varying GIP, gas price, CAPEX multiplier
  9. Compares concepts via BatchConceptRunner and ranks by NPV, IRR, payback
  10. Generates Word + HTML report with figures

Key Output

A concept comparison table with NPV (P10/P50/P90), IRR, breakeven price, CAPEX breakdown, and a recommendation with risk assessment.

Patterns for Effective Prompts

Be Specific About Conditions

# Less effective:
@process.model Simulate a separator

# More effective:
@process.model Simulate a 3-phase separator at 70 bara and 80°C for a
gas-condensate with 95% methane, 3% propane, 1% n-hexane, 1% water
at 200,000 Sm3/day

Mention Standards When Applicable

# Without standard:
@mechanical.design Calculate wall thickness for a 12-inch pipeline at 100 bara

# With standard (triggers deeper analysis):
@mechanical.design Calculate wall thickness for a 12-inch subsea pipeline
at 100 bara per DNV-OS-F101 with Equinor TR requirements

Request Specific Deliverables

# Vague:
@solve.task Study TEG dehydration

# Specific:
@solve.task Design a TEG dehydration unit for 50 MMSCFD at 70 bara.
Target -18°C water dew point per NORSOK P-001. Deliver a notebook with
validation against GPSA and a Word report.

Use the Scout for Unknowns

# Before starting a complex task:
@capability.scout Can NeqSim handle mercury removal from LNG feed gas
with activated carbon adsorption and mercury mass balance?