Mercury Removal Guard Beds

Overview

The MercuryRemovalBed class (neqsim.process.equipment.adsorber) models fixed-bed chemisorption of elemental mercury (Hg$^0$) onto metal-sulphide sorbents — the standard mercury removal technology in LNG pre-treatment and gas processing plants.

Location: neqsim.process.equipment.adsorber.MercuryRemovalBed

Related classes:

Class Package Purpose
MercuryRemovalBed process.equipment.adsorber Unit operation — process simulation
MercuryRemovalMechanicalDesign process.mechanicaldesign.adsorber Pressure vessel sizing and weight breakdown
MercuryRemovalCostEstimate process.costestimation.adsorber CAPEX and OPEX estimation

Why Mercury Removal Matters

Mercury must be removed from natural gas before cryogenic processing because:

  1. Aluminium embrittlement — Even ng/Nm$^3$ levels of Hg can cause catastrophic failure of brazed aluminium heat exchangers (BAHX) in LNG and NGL plants
  2. Catalyst poisoning — Pd/Pt catalysts in downstream hydrogenation and reforming are sensitive to mercury
  3. Environmental compliance — Product specifications typically require Hg < 0.01 µg/Nm$^3$
  4. Health & safety — TLV-TWA for Hg$^0$ vapour is 0.025 mg/m$^3$

Typical Inlet Concentrations

Source Hg Content (µg/Nm$^3$)
North Sea gas 0.01 – 10
SE Asian gas 10 – 300
Algerian gas 50 – 200
Australian gas 1 – 100

Technology Background

Chemisorption Mechanism

Mercury guard beds use metal sulphide sorbents (CuS, ZnS, FeS) supported on alumina or carbon carriers. The key reaction is an irreversible sulphide exchange:

\[\text{Hg}^0_{(g)} + \text{CuS}_{(s)} \rightarrow \text{HgS}_{(s)} + \text{Cu}_{(s)}\]

This differs fundamentally from physical adsorption:

Property Physical Adsorption Mercury Chemisorption
Bonding Van der Waals Chemical (covalent)
Reversibility Reversible (TSA/PSA) Irreversible
Regeneration Yes No — bed replaced when spent
Capacity Lower Higher (up to 25 wt%)
Selectivity Low–moderate High for Hg$^0$

Commercial Sorbents

Product Vendor Active Phase Typical Capacity
PuraSpec 1156 Johnson Matthey CuS/Al$_2$O$_3$ 10–25 wt% Hg
MRU Various CuS or CuS/ZnS 8–15 wt% Hg
HgSIV UOP Silver-exchanged zeolite Lower (~5 wt%)

Mathematical Model

Kinetics

The model uses an irreversible Langmuir-Hinshelwood rate law:

\[r = k_{eff}(T) \cdot C_{Hg} \cdot (1 - \theta)\]

where:

Temperature Dependence

The rate constant follows Arrhenius behaviour:

\[k_{eff}(T) = k_{ref} \cdot \exp\left[-\frac{E_a}{R}\left(\frac{1}{T} - \frac{1}{T_{ref}}\right)\right]\]

where $E_a$ is the activation energy (default 25 kJ/mol) and $T_{ref}$ is the reference temperature for $k_{ref}$.

Steady-State Mode

In steady state, the bed is treated as a single NTU unit:

\[\eta = (1 - f_{bypass}) \cdot (1 - e^{-NTU})\]

where $NTU = k_{eff} \cdot \tau_{res} \cdot F_{deg}$ with $\tau_{res}$ being the gas residence time in the void volume, and $F_{deg}$ the degradation factor.

Transient Mode

For time-dependent simulation, the bed is discretised into $N$ axial cells. Each time step consists of:

  1. Convective transport (upwind scheme): gas-phase Hg moves from cell to cell
  2. Chemisorption reaction: mercury is consumed from the gas phase and accumulated on the sorbent in each cell
  3. CFL-limited sub-stepping: automatically chosen for numerical stability

The cell-level mass balances are:

\[\frac{\partial C_{Hg}}{\partial t} + u_{int}\frac{\partial C_{Hg}}{\partial z} = -k_{eff} \cdot C_{Hg} \cdot (1 - \theta)\] \[\frac{\partial q}{\partial t} = k_{eff} \cdot C_{Hg} \cdot (1 - \theta) \cdot \frac{V_{void}}{m_{sorbent}} \cdot 10^{-3}\]

Each reaction substep limits uptake to both the mercury available in the cell gas and the remaining sorbent capacity. The same transferred mercury mass is removed from the gas and added to the sorbent, including when a large kinetic step reaches either limit.

Pressure Drop

The Ergun equation provides the bed pressure drop:

\[\frac{\Delta P}{L} = 150 \frac{\mu \cdot u_s \cdot (1-\varepsilon)^2}{\varepsilon^3 \cdot d_p^2} + 1.75 \frac{\rho \cdot u_s^2 \cdot (1-\varepsilon)}{\varepsilon^3 \cdot d_p}\]

Degradation Model

Two parameters model degraded column internals:

Parameter Effect Typical Cause
degradationFactor (0–1) Reduces effective capacity $q_{max}$ and steady-state removal rate Sorbent fouling, liquid carry-over
bypassFraction (0–1) Gas bypasses the sorbent entirely Channelling from damaged bed support, wall gaps

The combined effect reduces removal efficiency significantly. For example, a bed with degradationFactor = 0.6 and bypassFraction = 0.1 will have both lower capacity and 10% of the gas untreated.

In transient operation, reducing effective capacity preserves mercury already captured on the sorbent. If the retained loading equals or exceeds the reduced capacity, that cell stops adsorbing additional mercury. A fully blocked bed has zero effective capacity and transports gas without further adsorption; after the initial gas inventory has flushed through, mercury passes through the bed. Blockage does not release or erase previously captured mercury.

Condensable-contaminant screening

After running the bed, assessContaminant(componentName, saturationMoleFraction) screens a contaminant present in the last calculated bed fluid. Supply the bulk gas saturation mole fraction at the same temperature and pressure, calculated separately using an appropriate equilibrium model. The screening uses relative saturation $a = y/y_{sat}$ and a representative pore radius:

Pore radius Filling model Reported maxAllowableMoleFraction
At least 2 nm Kelvin onset; representative pore is either open or filled $y_{sat}$ multiplied by Kelvin onset
Below 2 nm Dubinin–Radushkevich continuous volume filling Concentration giving 5% pore filling

For the micropore branch, $W/W_0 = \exp[-(RT\ln(1/a)/(\beta E_0))^2]$. The corresponding screening concentration is $y_{lim} = y_{sat}\exp[-\beta E_0\sqrt{-\ln(0.05)}/(RT)]$. condensationExpected is true above 5% filling for micropores and at or above Kelvin onset for larger pores. kelvinOnset remains a diagnostic value in the micropore branch; use the reported concentration limit for the selected mechanism. The 2 nm switch and 5% criterion are screening conventions, not a calibrated sorbent performance specification.

Configure the radius with setSorbentPoreRadius (nm), pore volume with setSorbentPoreVolume (cm³/g), and geometry with setSorbentPoreType. Defaults are 6 nm, 0.35 cm³/g, and cylindrical pores. Micropore defaults are $E_0 = 8000$ J/mol and $\beta = 1$; fit these to measured isotherms for the actual contaminant and sorbent using setDubininCharacteristicEnergy and setDubininAffinityCoefficient. Pore volume does not change the fractional filling criterion.

getContaminantAssessmentJson includes the mechanism, saturation, filling fraction, and concentration limit in mole fraction and ppmv. applyContaminantDegradation replaces the bed’s degradation factor with one minus the assessed filling fraction. Run the bed again to apply that factor to removal performance. Assessments of several contaminants do not automatically combine their effects; each application replaces the previous factor. This is a pore-blocking screening approximation and does not predict competitive adsorption or contaminant transport.

API Reference

Construction

// Name only (configure later)
MercuryRemovalBed bed = new MercuryRemovalBed("Hg Guard");

// Name + inlet stream
MercuryRemovalBed bed = new MercuryRemovalBed("Hg Guard", feedStream);

Bed Geometry

Method Parameter Unit Default
setBedDiameter(d) Internal diameter m 1.5
setBedLength(L) Packed section height m 4.0
setVoidFraction(eps) Inter-particle porosity — 0.40
setParticleDiameter(dp) Sorbent pellet diameter m 0.004

Sorbent Properties

Method Parameter Unit Default
setSorbentType(name) Trade name / type — “PuraSpec”
setSorbentBulkDensity(rho) Bulk density kg/m$^3$ 1100
setMaxMercuryCapacity(q) Max Hg loading mg Hg/kg sorbent 100,000

Kinetics

Method Parameter Unit Default
setReactionRateConstant(k) LDF rate constant 1/s 0.05
setActivationEnergy(Ea) Arrhenius activation energy J/mol 25,000
setReferenceTemperature(T) Reference temperature for $k$ K 298.15

Degradation

Method Parameter Range Default
setDegradationFactor(f) Capacity/rate multiplier 0–1 1.0
setBypassFraction(f) Gas bypass fraction 0–0.99 0.0

Transient Simulation

Method Description
setCalculateSteadyState(false) Enable transient mode
setNumberOfCells(n) Set axial discretisation (default 50)
initialiseTransientGrid() Reset grid to empty bed
runTransient(dt, id) Advance by dt seconds
preloadBed(fraction) Pre-load to simulate spent bed
resetBed() Reset to fresh bed
setBreakthroughThreshold(r) Set C/C$_0$ ratio for breakthrough detection

Results Queries

Method Returns Unit
getRemovalEfficiency() Mercury removal fraction 0–1
getAverageLoading() Mean sorbent loading mg/kg
getBedUtilisation() Average loading / capacity 0–1
getLoadingProfile() Per-cell loading array mg/kg
getConcentrationProfile() Per-cell gas Hg concentration µg/Nm$^3$
getMassTransferZoneLength() MTZ length m
estimateBedLifetime() Estimated time to exhaustion hours
getPressureDrop() / getPressureDrop("bar") Bed pressure drop Pa or bar
isBreakthroughOccurred() Whether breakthrough detected boolean
getBreakthroughTimeHours() Time of breakthrough hours
getSorbentMass() Total sorbent in bed kg
getBedVolume() Packed bed volume m$^3$
toJson() Full JSON report String

Mechanical Design

The MercuryRemovalMechanicalDesign class sizes the pressure vessel:

MercuryRemovalMechanicalDesign design = bed.getMechanicalDesign();
design.setMaxOperationPressure(60.0);       // bara
design.setMaxOperationTemperature(353.15);  // K
design.calcDesign();

System.out.println("Wall thickness: " + design.getWallThickness() + " mm");
System.out.println("Total weight:   " + design.getWeightTotal() + " kg");
System.out.println(design.toJson());

Design Outputs

Output Method Unit
Wall thickness getWallThickness() mm
Inner/outer diameter innerDiameter / getOuterDiameter() m
Tan-tan length tantanLength m
Vessel shell weight getWeigthVesselShell() kg
Internals weight getInternalsWeight() kg
Sorbent charge weight getSorbentChargeWeight() kg
Total skid weight getWeightTotal() kg
Module footprint getModuleWidth/Length/Height() m
Bill of materials generateBillOfMaterials() List

Wall Thickness Calculation

Uses the Barlow / hoop-stress formula per ASME VIII Div 1 (simplified):

\[t = \frac{P_d \cdot D}{2 \cdot S \cdot E - P_d}\]

where $P_d$ is design pressure (110% of max operating), $D$ is inner diameter, $S$ is allowable stress (137.9 MPa for SA-516-70), and $E$ is joint efficiency (0.85). A minimum thickness of 6 mm is enforced.

Cost Estimation

The MercuryRemovalCostEstimate class provides CAPEX/OPEX:

MercuryRemovalCostEstimate cost = design.getCostEstimate();
cost.setSorbentUnitPrice(25.0);  // USD/kg for PuraSpec
cost.calculateCostEstimate();

System.out.println("Total module cost: " + cost.getTotalModuleCost());
System.out.println("Sorbent replacement: " + cost.getSorbentReplacementCost());
System.out.println(cost.toJson());

Cost Structure

Element Code Calculation
Vessel steel cost PEC (part) (shell + internals + nozzles weight) × steel rate × material factor
Sorbent charge PEC (part) sorbent mass × unit price
Bare module BMC PEC × bare module factor
Total module TMC BMC × contingency factor
Grassroots GRC TMC × site factor
Sorbent replacement OPEX sorbent mass × unit price × 1.25 (labour)
Annual maintenance OPEX TMC × 3%

Adjustable Cost Parameters

Parameter Default Method
Sorbent unit price 25 USD/kg setSorbentUnitPrice()
Steel fabrication cost 8 USD/kg setSteelCostPerKg()
Installation factor 1.5 setInstallationFactor()
Maintenance factor 3% (internal)

Usage Examples

Steady-State Example (Java)

SystemInterface gas = new SystemSrkEos(273.15 + 30.0, 60.0);
gas.addComponent("methane", 0.85);
gas.addComponent("ethane", 0.07);
gas.addComponent("propane", 0.03);
gas.addComponent("nitrogen", 0.04);
gas.addComponent("mercury", 1.0e-9);
gas.createDatabase(true);
gas.setMixingRule(2);
gas.init(0);

Stream feed = new Stream("feed", gas);
feed.setFlowRate(50000.0, "kg/hr");
feed.run();

MercuryRemovalBed bed = new MercuryRemovalBed("HgGuard", feed);
bed.setBedDiameter(1.5);
bed.setBedLength(4.0);
bed.setSorbentType("PuraSpec");
bed.run(UUID.randomUUID());

System.out.println("Efficiency: " + bed.getRemovalEfficiency());
System.out.println("Pressure drop: " + bed.getPressureDrop("bar") + " bar");

Transient Breakthrough Example (Java)

bed.setCalculateSteadyState(false);
bed.setNumberOfCells(50);
bed.initialiseTransientGrid();

UUID id = UUID.randomUUID();
double dt = 3600.0; // 1-hour steps

for (int hour = 0; hour < 10000; hour++) {
    bed.runTransient(dt, id);
    if (bed.isBreakthroughOccurred()) {
        System.out.println("Breakthrough at " + bed.getBreakthroughTimeHours() + " hours");
        break;
    }
}

System.out.println("Average loading: " + bed.getAverageLoading() + " mg/kg");
System.out.println("Utilisation: " + (bed.getBedUtilisation() * 100) + "%");

Python Example

from neqsim import jneqsim

MercuryRemovalBed = jneqsim.process.equipment.adsorber.MercuryRemovalBed
# ... (see Jupyter notebook for full example)

Validation

The validateSetup() method checks for common configuration errors:

Check Error if
Bed diameter ≤ 0
Bed length ≤ 0
Void fraction ≤ 0 or ≥ 1
Mercury capacity ≤ 0
Degradation factor < 0 or > 1
Bypass fraction < 0 or ≥ 1

Each error includes a remediation hint for automated recovery.