TEG Dehydration Modeling
Use this skill to build a complete, converging triethylene-glycol (TEG) gas
dehydration plant in NeqSim. It captures a validated flowsheet topology and the
non-obvious settings (CPA fluid setup, regeneration-column tolerances, recycle
priorities, condenser/preheat energy coupling, makeup calculator) needed for the
process to solve as a closed loop. It returns a ProcessSystem plus the key
streams an agent needs for water dew point, lean-TEG purity, and still-vent
emission classification.
This is a real NeqSim build, not a public correlation placeholder. For a quick
"is dehydration required?" triage without building a plant, use
neqsim-water-dewpoint-dehydration-screening instead.
When to Use
- When a user asks to model a TEG dehydration unit and read water dew point, lean
TEG purity, or regeneration still-vent emissions.
- When an agent needs a runnable TEG flowsheet to study stripping gas, reboiler
temperature, TEG circulation rate, or once-through vs recirculated stripping gas.
- When emissions (NMVOC, methane, benzene, CO2, water) from the regeneration
still vent must be quantified and classified for an environmental screening.
Inputs
The build_teg_plant(...) function is fully parameterized. Key inputs:
feed_fractions: list of mole fractions (or percentages) aligned with
GAS_COMPONENTS (nitrogen, CO2, methane, ... , benzene).
feed_flow_MSm3_day: wet/dry feed gas flow rate in MSm3/day.
feed_temp_C, feed_pressure_bara: feed gas conditions.
absorber_pressure_bara, absorber_temp_C: contactor conditions.
teg_flow_kg_hr, teg_feed_temp_C, lean_teg_purity: lean TEG circulation.
flash_drum_pressure_bara, reboiler_temp_C, stripping_gas_Sm3_hr:
regeneration conditions.
n_absorber_stages, stage_efficiency: absorber model resolution.
water_mode: 'saturated' (water-saturate the feed) or 'specified'
(set an inlet water content in mol-ppm via water_content_ppm_mol).
saturation_temp_C, saturation_pressure_bara: optional saturation TP setter.
recirculate_stripping_gas: True builds a true closed-loop stripping-gas
recycle (split dried overhead, blower, heater) to cut still-vent emissions.
Outputs
build_teg_plant(...) returns (process, streams):
process: a NeqSim ProcessSystem ready to run.
streams: a dict of handles — dehydratedGas, flashGas, stillVent,
leanTEGtoAbs, waterDewAnalyser, strippingGas, recircBlower.
Helper functions produce the engineering results:
comp_mass_flows_kg_hr(stream): per-component mass flow (kg/hr).
teg_mass_fraction(stream): TEG weight percent of a stream.
classify_emissions(stream): dict with NMVOC, methane, CO2, water,
TEG, benzene, and total mass flows (kg/hr).
Engineering Method
The flowsheet follows the standard TEG dehydration topology:
- Fluid —
SystemSrkCPAstatoil (CPA EOS) with setMixingRule(10) and
setMultiPhaseCheck(False). TEG is added last so the lean-TEG composition
maps to leanComp[-1] = lean_teg_purity and leanComp[-2] = 1 - lean_teg_purity.
- Feed conditioning — optional saturation TP setter +
StreamSaturatorUtil
(saturated mode) or an inlet water component (specified mode), then a heater to
absorber TP.
- Absorber —
SimpleTEGAbsorber with gas-in and lean-TEG solvent-in streams.
- Rich TEG let-down and pre-heat — HP flash valve, rich preheater,
lean/rich exchangers (
HeatExchanger with UA values), degassing separator,
fine filter, LP flash valve.
- Regeneration —
DistillationColumn (reboiler + condenser) with loosened
tolerances (temperature 5e-2, mass 2e-1, enthalpy 2e-1) so the glycol/water
split converges; stripping gas fed to the reboiler tray; WaterStripperColumn
for deep lean-TEG purity.
- Recycles — lean-TEG recycle (priority 200,
setDownstreamProperty('flow rate'))
closes the solvent loop; the internal stripping-gas recycle returns stripper
overhead to the reboiler. With recirculate_stripping_gas=True, a second
makeup recycle recirculates dried overhead as stripping gas. Keep both coupled
stripping-gas recycles at the same controller priority (currently 100); an
intermediate priority may be skipped by the current RecycleController sequence.
- Makeup and energy coupling — a
Calculator sizes TEG makeup from TEG lost
in the dry gas, flash gas, still vent, and water draws; the rich preheater is
driven by the condenser energy stream:
richPreheat.setEnergyStream(column.getCondenser().getEnergyStream()).
- Run —
process.runAsThread() then thr.join(...) (worker thread avoids
blocking and tolerates the recycle iterations). A completed worker thread or
successful unit statuses do not prove closed-loop convergence: require
process.solved() and inspect process.getConvergenceDiagnostics() before
accepting water or TEG closure.
Python Usage Pattern
from teg_dehydration_modeling import (
DEFAULT_FEED,
build_teg_plant,
classify_emissions,
teg_mass_fraction,
)
process, streams = build_teg_plant(
feed_fractions=DEFAULT_FEED,
feed_flow_MSm3_day=4.65,
feed_temp_C=25.0,
feed_pressure_bara=70.0,
absorber_pressure_bara=85.0,
absorber_temp_C=35.0,
teg_flow_kg_hr=5500.0,
teg_feed_temp_C=48.5,
lean_teg_purity=0.97,
flash_drum_pressure_bara=4.8,
reboiler_temp_C=197.5,
stripping_gas_Sm3_hr=180.0,
n_absorber_stages=4,
stage_efficiency=0.7,
)
thr = process.runAsThread()
thr.join(300000) # ~15 s once-through, ~40 s with stripping-gas recirculation
water_dew_C = float(streams["waterDewAnalyser"].getMeasuredValue("C"))
lean_teg_wt = teg_mass_fraction(streams["leanTEGtoAbs"])
still_vent = classify_emissions(streams["stillVent"])
print(water_dew_C, lean_teg_wt, still_vent["NMVOC"])
Regeneration-Pressure Sensitivity (reboiler floating on the LP flare)
Use this when the question is "does a few tens of mbar more reboiler pressure matter?" rather
than a full flowsheet. A reboiler/still connected to an LP flare header has no pressure control;
its pressure is the header pressure plus the pressure drop of the overhead path. Lean-TEG water
content at fixed temperature and vapour composition scales with total pressure, so the
effect is proportional (verified 2026-10-08, Aasta Hansteen, SRK-CPA, mixingRule 10):
p_w(x_water, T) = y_water * P_bubble from bubblePointPressureFlash(False) on a
TEG/water binary (fl.getPhase(0) is the gas); solve x_water so that
p_w = y_steam * P_reboiler by bisection (y_steam = steam / (steam + stripping gas), about
0.85 to 0.92 for 76 Sm3/h stripping gas and 690 kg/h water).
- Benchmark: 98.90 wt% TEG at 204 C and 1 atm without stripping gas (textbook value about 98.7).
- 0.200 to 0.215 barg at 198 C: lean-TEG water +1.29 % relative (about -0.005 wt%-points at the
design 99.6 wt%), equal to about 0.6 C of reboiler temperature, and +0.16 C on the equilibrium
dry-gas dew point at 69 barg. Reaching the 0.4 barg H alarm costs about +17 % water.
- Dry-gas equilibrium: flash gas +
water + TEG with 5 mol TEG solution per mol gas so the
liquid composition stays fixed; dewPointTemperatureFlash() for the dew point.
- Water load: saturate the gas at the WET-GAS-SCRUBBER temperature (e.g. 22 C), not the
contactor lean-TEG inlet temperature (35 C); the two differ by a factor of two in kg/h.
- In
neqsim_dev_setup task scripts use ns.JClass("neqsim.thermo.system.SystemSrkCPAstatoil")
and the component name TEG.
Drizo / solvent stripping (stripping_mode="drizo")
Drizo regenerates TEG with a closed loop of recovered hydrocarbon solvent (BTEX) instead of
fuel gas. Oseberg D documents it (system 24 manual 11-4A-AV-C87-24001 pp. 15-16): flash vapour
cooled to 20 C and the condensate sent back to the still feed; still overhead (93 C) cooled to
20 C into a three-phase receiver; solvent vaporised and superheated to 220 C and fed to a packed
stripping column; excess solvent to the closed drain; reboiler 196-204 C at 0.35 barg.
build_teg_plant(..., stripping_mode="drizo", drizo_solvent_kg_hr=..., still_pressure_bara=1.36)
builds exactly this loop. Use components=[...] (with feed_fractions aligned) to add toluene
and m-Xylene so the solvent carries real BTEX, and drizo_solvent_comp={"m-Xylene": 0.4, ...}.
- The solvent circulation rate is a set-point (Splitter
setFlowRates([rate, -1.0], "kg/hr")),
not a result; the stripped purity depends on it, so scan it against the measured lean-TEG
concentration. It is the main calibration knob for troubleshooting dehydration efficiency.
- Three recycles are closed by the loop (solvent recovery, solvent loop, stripping vapour to
the reboiler) in addition to the lean-TEG recycle; a run takes several minutes, so use
runAsThread() with a join of 900 s or more, check process.solved(), and if it is false call
process.run() again (warm restart, up to 3 times; the stripping-vapour recycle often stops at
its iteration cap on the first pass with errComp about 2e-2 and converges on the second).
- The still receiver must be a
ThreePhaseSeparator (water out, oil out = solvent, gas out =
vent); a two-phase separator mixes the solvent into the water draw.
- Without stripping, TEG at 204 C and 1 atm is about 98.9 wt%; a lean purity near that value means
the solvent loop is not stripping (rate too low, or the loop has not converged), not that
the model is right.
- Report lean quality with
glycol_purity_wt() (TEG/(TEG+water)). WaterStripperColumn lets
heavy solvent dissolve in the glycol, so teg_mass_fraction() falls as solvent rate rises
(97.95 wt% of liquid at 1500 kg/h) while the glycol/water purity rises (99.8 wt%); mixing the
two bases hides the stripping effect (measured 2026-10-28, 11.75 t/h semi-lean 97.8 wt%, 204 C,
1.36 bara, 2 vs 4 stages).
- Stages matter more than rate: 4 stages with 300 Sm3/h dry gas gives 99.95 wt% where 2 stages
give 99.79. Use 4 stages for a packed stripping column unless vendor data says otherwise.
- Convergence evidence: the Drizo loop was validated on the Oseberg D basis (12.3 MSm3/d feed,
11.75 t/h TEG, 4 stripper stages, 1000 kg/h solvent): converged in about 100 s after one warm
restart to lean 99.95 wt%, reboiler 0.83 MW (design 1.35 MW), dew point -49 C at 69 barg. The
generic default case of this skill (4.65 MSm3/d, 5.5 t/h TEG) did NOT converge in 8 warm
restarts with 300 or 600 kg/h solvent (lean purity oscillating 98.2-99.2 wt%); tune circulation,
solvent rate and stripper stages for your own case and check
process.solved() before use.
- The same-solvent-rate effect is non-monotonic if you read the wrong basis (see above).
Validation Checklist
Common Mistakes
| Symptom |
Cause |
Fix |
| Regeneration column will not converge |
Default tolerances too tight for glycol/water split |
Loosen setTemperatureTolerance(5e-2), setMassBalanceTolerance(2e-1), setEnthalpyBalanceTolerance(2e-1) |
| Lean TEG purity wrong / NaN |
TEG not added last, or composition indices off |
Add water then TEG last; set leanComp[-1]=purity, leanComp[-2]=1-purity |
| Lean-TEG loop never closes |
Recycle priority/property not set |
Lean TEG recycle priority 200 with setDownstreamProperty('flow rate') |
| Stripping-gas makeup recycle never executes |
Coupled stripping-gas recycles use different priorities and an intermediate level is skipped |
Assign the makeup recycle the same priority as the internal stripping-gas recycle; verify both with convergence diagnostics |
| Class not found |
Wrong CPA class name |
Use SystemSrkCPAstatoil (lowercase statoil) |
| Run blocks or stalls |
Running inline instead of on a thread |
Use process.runAsThread() + thr.join(timeout_ms) |
| TEG inventory drifts |
No makeup |
Add the Calculator makeup sized from TEG in dry gas, flash gas, still vent, water draws |
| Water balance looks open |
Closed-loop convergence was not established, or a water outlet was omitted |
First require process.solved() and all recycle diagnostics solved; then include still vent + regen water/HC draw + degassing flash gas in the water accounting |
| Plant has solvent recovery / 220 C superheated stripping vapour but is modelled with fuel gas |
The unit is a Drizo plant |
Use stripping_mode="drizo"; read the system manual for the solvent loop before choosing the stripping scheme |
Regeneration column reports FALLBACK_PRODUCTS / reboiler duty about 0 |
Feed on tray 1 with stripping gas on tray 1 and a high condenser temperature |
Feed at tray 0, stripping gas through getReboiler().addStream(...), condenser near the real overhead temperature, as in the TEGdehydrationProcessDistillationJS example |
Limitations
- Inputs and examples are public and synthetic; no proprietary plant data,
vendor glycol packages, or company emission factors are included.
- Emission classification is a component-bucketing of stream mass flows
(NMVOC/methane/benzene/CO2/water); it is not a regulatory emission inventory.
- The absorber uses
SimpleTEGAbsorber with stage efficiency, not a rate-based
contactor model; deep dew-point design needs validated review.
- Convergence depends on reasonable inputs; extreme TEG rates, stripping gas, or
reboiler temperatures may require tolerance or guess-temperature tuning.
Related NeqSim Functionality
neqsim.thermo.system.SystemSrkCPAstatoil — CPA equation of state for
glycol/water/hydrocarbon systems (reached via from neqsim import jneqsim).
neqsim.process.equipment.absorber.SimpleTEGAbsorber,
neqsim.process.equipment.absorber.WaterStripperColumn — TEG contactor and stripper.
neqsim.process.equipment.distillation.DistillationColumn — TEG regeneration column.
neqsim.process.equipment.util.Recycle, Calculator — solvent/stripping-gas
recycles and TEG makeup sizing.
neqsim.process.measurementdevice.WaterDewPointAnalyser — dry-gas water dew point.
- For a screening-only water-content triage before building a plant, use the
neqsim-water-dewpoint-dehydration-screening community skill.
References
1---2name: neqsim-teg-dehydration-modeling-23calculation_basis: "neqsim-java"4version: "0.1.0"5description: "Build a validated, closed-loop TEG (triethylene glycol) gas dehydration plant in NeqSim, including absorber, flash drum, regeneration column, stripper, lean-TEG recycle, and still-vent emission classification. USE WHEN: a task needs a runnable NeqSim TEG dehydration flowsheet (water dew point, lean-TEG purity, NMVOC/methane/benzene still-vent emissions) rather than a screening-level water-content estimate."6last_verified: "2026-06-26"7requires:8 python_packages: [neqsim]9 java_packages: [neqsim]10 env: []11 network: []12metadata:13 source: https://skillmd.com/skills/equinor/neqsim-teg-dehydration-modeling-214---1516# TEG Dehydration Modeling1718Use this skill to build a complete, converging triethylene-glycol (TEG) gas19dehydration plant in NeqSim. It captures a validated flowsheet topology and the20non-obvious settings (CPA fluid setup, regeneration-column tolerances, recycle21priorities, condenser/preheat energy coupling, makeup calculator) needed for the22process to solve as a closed loop. It returns a `ProcessSystem` plus the key23streams an agent needs for water dew point, lean-TEG purity, and still-vent24emission classification.2526This is a real NeqSim build, not a public correlation placeholder. For a quick27"is dehydration required?" triage without building a plant, use28`neqsim-water-dewpoint-dehydration-screening` instead.2930## When to Use3132- When a user asks to model a TEG dehydration unit and read water dew point, lean33 TEG purity, or regeneration still-vent emissions.34- When an agent needs a runnable TEG flowsheet to study stripping gas, reboiler35 temperature, TEG circulation rate, or once-through vs recirculated stripping gas.36- When emissions (NMVOC, methane, benzene, CO2, water) from the regeneration37 still vent must be quantified and classified for an environmental screening.3839## Inputs4041The `build_teg_plant(...)` function is fully parameterized. Key inputs:4243- `feed_fractions`: list of mole fractions (or percentages) aligned with44 `GAS_COMPONENTS` (nitrogen, CO2, methane, ... , benzene).45- `feed_flow_MSm3_day`: wet/dry feed gas flow rate in MSm3/day.46- `feed_temp_C`, `feed_pressure_bara`: feed gas conditions.47- `absorber_pressure_bara`, `absorber_temp_C`: contactor conditions.48- `teg_flow_kg_hr`, `teg_feed_temp_C`, `lean_teg_purity`: lean TEG circulation.49- `flash_drum_pressure_bara`, `reboiler_temp_C`, `stripping_gas_Sm3_hr`:50 regeneration conditions.51- `n_absorber_stages`, `stage_efficiency`: absorber model resolution.52- `water_mode`: `'saturated'` (water-saturate the feed) or `'specified'`53 (set an inlet water content in mol-ppm via `water_content_ppm_mol`).54- `saturation_temp_C`, `saturation_pressure_bara`: optional saturation TP setter.55- `recirculate_stripping_gas`: `True` builds a true closed-loop stripping-gas56 recycle (split dried overhead, blower, heater) to cut still-vent emissions.5758## Outputs5960`build_teg_plant(...)` returns `(process, streams)`:6162- `process`: a NeqSim `ProcessSystem` ready to run.63- `streams`: a dict of handles — `dehydratedGas`, `flashGas`, `stillVent`,64 `leanTEGtoAbs`, `waterDewAnalyser`, `strippingGas`, `recircBlower`.6566Helper functions produce the engineering results:6768- `comp_mass_flows_kg_hr(stream)`: per-component mass flow (kg/hr).69- `teg_mass_fraction(stream)`: TEG weight percent of a stream.70- `classify_emissions(stream)`: dict with `NMVOC`, `methane`, `CO2`, `water`,71 `TEG`, `benzene`, and `total` mass flows (kg/hr).7273## Engineering Method7475The flowsheet follows the standard TEG dehydration topology:76771. **Fluid** — `SystemSrkCPAstatoil` (CPA EOS) with `setMixingRule(10)` and78 `setMultiPhaseCheck(False)`. TEG is added **last** so the lean-TEG composition79 maps to `leanComp[-1] = lean_teg_purity` and `leanComp[-2] = 1 - lean_teg_purity`.802. **Feed conditioning** — optional saturation TP setter + `StreamSaturatorUtil`81 (saturated mode) or an inlet water component (specified mode), then a heater to82 absorber TP.833. **Absorber** — `SimpleTEGAbsorber` with gas-in and lean-TEG solvent-in streams.844. **Rich TEG let-down and pre-heat** — HP flash valve, rich preheater,85 lean/rich exchangers (`HeatExchanger` with UA values), degassing separator,86 fine filter, LP flash valve.875. **Regeneration** — `DistillationColumn` (reboiler + condenser) with **loosened88 tolerances** (temperature 5e-2, mass 2e-1, enthalpy 2e-1) so the glycol/water89 split converges; stripping gas fed to the reboiler tray; `WaterStripperColumn`90 for deep lean-TEG purity.916. **Recycles** — lean-TEG recycle (priority 200, `setDownstreamProperty('flow rate')`)92 closes the solvent loop; the internal stripping-gas recycle returns stripper93 overhead to the reboiler. With `recirculate_stripping_gas=True`, a second94 makeup recycle recirculates dried overhead as stripping gas. Keep both coupled95 stripping-gas recycles at the same controller priority (currently 100); an96 intermediate priority may be skipped by the current `RecycleController` sequence.977. **Makeup and energy coupling** — a `Calculator` sizes TEG makeup from TEG lost98 in the dry gas, flash gas, still vent, and water draws; the rich preheater is99 driven by the condenser energy stream:100 `richPreheat.setEnergyStream(column.getCondenser().getEnergyStream())`.1018. **Run** — `process.runAsThread()` then `thr.join(...)` (worker thread avoids102 blocking and tolerates the recycle iterations). A completed worker thread or103 successful unit statuses do not prove closed-loop convergence: require104 `process.solved()` and inspect `process.getConvergenceDiagnostics()` before105 accepting water or TEG closure.106107## Python Usage Pattern108109```python110from teg_dehydration_modeling import (111 DEFAULT_FEED,112 build_teg_plant,113 classify_emissions,114 teg_mass_fraction,115)116117process, streams = build_teg_plant(118 feed_fractions=DEFAULT_FEED,119 feed_flow_MSm3_day=4.65,120 feed_temp_C=25.0,121 feed_pressure_bara=70.0,122 absorber_pressure_bara=85.0,123 absorber_temp_C=35.0,124 teg_flow_kg_hr=5500.0,125 teg_feed_temp_C=48.5,126 lean_teg_purity=0.97,127 flash_drum_pressure_bara=4.8,128 reboiler_temp_C=197.5,129 stripping_gas_Sm3_hr=180.0,130 n_absorber_stages=4,131 stage_efficiency=0.7,132)133134thr = process.runAsThread()135thr.join(300000) # ~15 s once-through, ~40 s with stripping-gas recirculation136137water_dew_C = float(streams["waterDewAnalyser"].getMeasuredValue("C"))138lean_teg_wt = teg_mass_fraction(streams["leanTEGtoAbs"])139still_vent = classify_emissions(streams["stillVent"])140141print(water_dew_C, lean_teg_wt, still_vent["NMVOC"])142```143144## Regeneration-Pressure Sensitivity (reboiler floating on the LP flare)145146Use this when the question is "does a few tens of mbar more reboiler pressure matter?" rather147than a full flowsheet. A reboiler/still connected to an LP flare header has no pressure control;148its pressure is the header pressure plus the pressure drop of the overhead path. Lean-TEG water149content at fixed temperature and vapour composition scales with total pressure, so the150effect is proportional (verified 2026-10-08, Aasta Hansteen, SRK-CPA, `mixingRule 10`):151152- `p_w(x_water, T)` = `y_water * P_bubble` from `bubblePointPressureFlash(False)` on a153 `TEG`/`water` binary (`fl.getPhase(0)` is the gas); solve `x_water` so that154 `p_w = y_steam * P_reboiler` by bisection (`y_steam` = steam / (steam + stripping gas), about155 0.85 to 0.92 for 76 Sm3/h stripping gas and 690 kg/h water).156- Benchmark: 98.90 wt% TEG at 204 C and 1 atm without stripping gas (textbook value about 98.7).157- 0.200 to 0.215 barg at 198 C: lean-TEG water +1.29 % relative (about -0.005 wt%-points at the158 design 99.6 wt%), equal to about 0.6 C of reboiler temperature, and +0.16 C on the equilibrium159 dry-gas dew point at 69 barg. Reaching the 0.4 barg H alarm costs about +17 % water.160- Dry-gas equilibrium: flash gas + `water` + `TEG` with 5 mol TEG solution per mol gas so the161 liquid composition stays fixed; `dewPointTemperatureFlash()` for the dew point.162- Water load: saturate the gas at the WET-GAS-SCRUBBER temperature (e.g. 22 C), not the163 contactor lean-TEG inlet temperature (35 C); the two differ by a factor of two in kg/h.164- In `neqsim_dev_setup` task scripts use `ns.JClass("neqsim.thermo.system.SystemSrkCPAstatoil")`165 and the component name `TEG`.166167## Drizo / solvent stripping (`stripping_mode="drizo"`)168169Drizo regenerates TEG with a closed loop of recovered hydrocarbon solvent (BTEX) instead of170fuel gas. Oseberg D documents it (system 24 manual 11-4A-AV-C87-24001 pp. 15-16): flash vapour171cooled to 20 C and the condensate sent back to the still feed; still overhead (93 C) cooled to17220 C into a three-phase receiver; solvent vaporised and superheated to 220 C and fed to a packed173stripping column; excess solvent to the closed drain; reboiler 196-204 C at 0.35 barg.174`build_teg_plant(..., stripping_mode="drizo", drizo_solvent_kg_hr=..., still_pressure_bara=1.36)`175builds exactly this loop. Use `components=[...]` (with `feed_fractions` aligned) to add toluene176and m-Xylene so the solvent carries real BTEX, and `drizo_solvent_comp={"m-Xylene": 0.4, ...}`.177178- The solvent circulation rate is a set-point (Splitter `setFlowRates([rate, -1.0], "kg/hr")`),179 not a result; the stripped purity depends on it, so scan it against the measured lean-TEG180 concentration. It is the main calibration knob for troubleshooting dehydration efficiency.181- Three recycles are closed by the loop (solvent recovery, solvent loop, stripping vapour to182 the reboiler) in addition to the lean-TEG recycle; a run takes several minutes, so use183 `runAsThread()` with a join of 900 s or more, check `process.solved()`, and if it is false call184 `process.run()` again (warm restart, up to 3 times; the stripping-vapour recycle often stops at185 its iteration cap on the first pass with `errComp` about 2e-2 and converges on the second).186- The still receiver must be a `ThreePhaseSeparator` (water out, oil out = solvent, gas out =187 vent); a two-phase separator mixes the solvent into the water draw.188- Without stripping, TEG at 204 C and 1 atm is about 98.9 wt%; a lean purity near that value means189 the solvent loop is not stripping (rate too low, or the loop has not converged), not that190 the model is right.191- Report lean quality with `glycol_purity_wt()` (TEG/(TEG+water)). `WaterStripperColumn` lets192 heavy solvent dissolve in the glycol, so `teg_mass_fraction()` falls as solvent rate rises193 (97.95 wt% of liquid at 1500 kg/h) while the glycol/water purity rises (99.8 wt%); mixing the194 two bases hides the stripping effect (measured 2026-10-28, 11.75 t/h semi-lean 97.8 wt%, 204 C,195 1.36 bara, 2 vs 4 stages).196- Stages matter more than rate: 4 stages with 300 Sm3/h dry gas gives 99.95 wt% where 2 stages197 give 99.79. Use 4 stages for a packed stripping column unless vendor data says otherwise.198- Convergence evidence: the Drizo loop was validated on the Oseberg D basis (12.3 MSm3/d feed,199 11.75 t/h TEG, 4 stripper stages, 1000 kg/h solvent): converged in about 100 s after one warm200 restart to lean 99.95 wt%, reboiler 0.83 MW (design 1.35 MW), dew point -49 C at 69 barg. The201 generic default case of this skill (4.65 MSm3/d, 5.5 t/h TEG) did NOT converge in 8 warm202 restarts with 300 or 600 kg/h solvent (lean purity oscillating 98.2-99.2 wt%); tune circulation,203 solvent rate and stripper stages for your own case and check `process.solved()` before use.204- The same-solvent-rate effect is non-monotonic if you read the wrong basis (see above).205206## Validation Checklist207208- [ ] Water dew point of the dry gas is well below 0 C at the reference pressure.209- [ ] Lean TEG purity (`glycol_purity_wt`, TEG/(TEG+water)) is in the range of the real plant:210 about 99.9 wt% (0.1 wt% water) for a stripped or Drizo unit, 98.7-99 wt% only when there211 is no stripping. A value of 90-97 wt% means the regeneration is not representative.212- [ ] Still-vent NMVOC and methane mass flows are finite and non-negative.213- [ ] With `recirculate_stripping_gas=True`, still-vent NMVOC and methane do not214 increase versus the once-through case.215- [ ] The process is run on a worker thread and the join timeout is large enough216 for the recycles to converge.217- [ ] `process.solved()` is true and every active recycle reports `solved() == True`;218 record priority, iterations, and residuals for any unsolved recycle.219- [ ] Water and TEG closure are interpreted only after the closed loop converges.220- [ ] Example inputs are public and synthetic.221222## Common Mistakes223224| Symptom | Cause | Fix |225| --- | --- | --- |226| Regeneration column will not converge | Default tolerances too tight for glycol/water split | Loosen `setTemperatureTolerance(5e-2)`, `setMassBalanceTolerance(2e-1)`, `setEnthalpyBalanceTolerance(2e-1)` |227| Lean TEG purity wrong / NaN | TEG not added last, or composition indices off | Add `water` then `TEG` last; set `leanComp[-1]=purity`, `leanComp[-2]=1-purity` |228| Lean-TEG loop never closes | Recycle priority/property not set | Lean TEG recycle priority 200 with `setDownstreamProperty('flow rate')` |229| Stripping-gas makeup recycle never executes | Coupled stripping-gas recycles use different priorities and an intermediate level is skipped | Assign the makeup recycle the same priority as the internal stripping-gas recycle; verify both with convergence diagnostics |230| Class not found | Wrong CPA class name | Use `SystemSrkCPAstatoil` (lowercase `statoil`) |231| Run blocks or stalls | Running inline instead of on a thread | Use `process.runAsThread()` + `thr.join(timeout_ms)` |232| TEG inventory drifts | No makeup | Add the `Calculator` makeup sized from TEG in dry gas, flash gas, still vent, water draws |233| Water balance looks open | Closed-loop convergence was not established, or a water outlet was omitted | First require `process.solved()` and all recycle diagnostics solved; then include **still vent + regen water/HC draw + degassing flash gas** in the water accounting |234| Plant has solvent recovery / 220 C superheated stripping vapour but is modelled with fuel gas | The unit is a Drizo plant | Use `stripping_mode="drizo"`; read the system manual for the solvent loop before choosing the stripping scheme |235| Regeneration column reports `FALLBACK_PRODUCTS` / reboiler duty about 0 | Feed on tray 1 with stripping gas on tray 1 and a high condenser temperature | Feed at tray 0, stripping gas through `getReboiler().addStream(...)`, condenser near the real overhead temperature, as in the `TEGdehydrationProcessDistillationJS` example |236237## Limitations238239- Inputs and examples are public and synthetic; no proprietary plant data,240 vendor glycol packages, or company emission factors are included.241- Emission classification is a component-bucketing of stream mass flows242 (NMVOC/methane/benzene/CO2/water); it is not a regulatory emission inventory.243- The absorber uses `SimpleTEGAbsorber` with stage efficiency, not a rate-based244 contactor model; deep dew-point design needs validated review.245- Convergence depends on reasonable inputs; extreme TEG rates, stripping gas, or246 reboiler temperatures may require tolerance or guess-temperature tuning.247248## Related NeqSim Functionality249250- `neqsim.thermo.system.SystemSrkCPAstatoil` — CPA equation of state for251 glycol/water/hydrocarbon systems (reached via `from neqsim import jneqsim`).252- `neqsim.process.equipment.absorber.SimpleTEGAbsorber`,253 `neqsim.process.equipment.absorber.WaterStripperColumn` — TEG contactor and stripper.254- `neqsim.process.equipment.distillation.DistillationColumn` — TEG regeneration column.255- `neqsim.process.equipment.util.Recycle`, `Calculator` — solvent/stripping-gas256 recycles and TEG makeup sizing.257- `neqsim.process.measurementdevice.WaterDewPointAnalyser` — dry-gas water dew point.258- For a screening-only water-content triage before building a plant, use the259 `neqsim-water-dewpoint-dehydration-screening` community skill.260261## References262263- NeqSim repository: https://github.com/equinor/neqsim264- NeqSim TEG dehydration emissions notebook:265 https://github.com/equinor/neqsim/blob/master/examples/notebooks/teg_dehydration_emissions.ipynb266- NeqSim Skills Guide:267 https://github.com/equinor/neqsim/blob/master/docs/integration/skills_guide.md