agentleFS
equinorequinor/neqsim-community-skills/skills/field-development/neqsim-reservoir-facility-value-chain/SKILL.md

neqsim-reservoir-facility-value-chain

Skill4 starsChanged 5 months agoApache-2.0View on GitHub

History-match and forecast a gas chain from reservoir to export: p/z tank from shut-in wellhead pressure, well deliverability, facility capacity from process-model sweeps, separation pressure at the supply/capacity crossing, multi-field export coupling, scenarios, P10/P90. USE WHEN: linking reservoir depletion to a platform or export system, history matching, finding whether wells or facility limit production, or valuing compression and well levers.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Tank: Use shut-in WHP (on-stream hours = 0, WHP above ~20 barg, per-window maximum or upper quartile) as the pressure proxy.

    line 59
  2. Very specific:

    Facility: Run the platform process model over a throughput grid at several separation pressures (throughput_sweep).

    line 69
  3. Very specific:

    Run python -m pytest in the skill folder.

    line 129

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Reservoir-to-Facility Gas Value Chain

Screening workflow that couples a reservoir tank, well deliverability and a facility capacity curve, so one model answers "who limits production now, and when does that change?". It was built for the Troll A/B/C to Kollsnes chain (Troll East, Troll West gas province, Troll B and C oil rims) and is field-agnostic. Use NeqSim process models (ProcessSystem, ProcessModel) for the facility numbers and NeqSim SimpleReservoir or OPM Flow when a 3D reservoir model exists; this skill is the fast layer between them.

When to Use

  • Reservoir depletion must be linked to a platform or export system in one model.
  • A gas field needs a history match from shut-in wellhead pressure and allocated volumes.
  • The question is whether wells or the facility limit production, and when that changes.
  • Compression or well levers must be valued against a forecast.

Inputs

  • Monthly allocated volumes per wellbore and a daily sample with WHP, choke and on-stream hours.
  • Gas composition, reservoir temperature, TVD for the static column.
  • A facility throughput sweep at several separation pressures (utilisation versus rate).
  • Flowline pressure drop at a reference rate, and the export arrival pressure.

Outputs

  • Tank parameters (p/z_i, GIIP) and the fitted well law per well group.
  • Facility capacity versus separation pressure, the best separation pressure and the binding side (SUPPLY or FACILITY) per month.
  • Scenario comparison and P10/P90 from bootstrap and parameter sampling.

Engineering Method

StepPython (this skill)NeqSim Java (neqsim.process.fielddevelopment.integrated)
Shut-in WHP to reservoir pressurestatic_bottomhole_pressureMaterialBalanceHistoryMatch.staticBottomholePressure
p/z line, GIIP, outlier rejectionfit_pz_lineMaterialBalanceHistoryMatch.fitPzLine
Real-gas p/z tankGasTankRealGasMaterialBalanceDrive + DranchukAbouKassemZ
Well law q = C (pr^2 - pwh^2)^nRSDeliverability.fitRawlinsSchellhardtFit, WellDeliverabilityCurve.fromRawlinsSchellhardt
Facility capacity from sweepcapacity_from_sweep, PressureCapacityCurveSupplyCapacityBalance.PressureCapacityCurve
Best separation pressuremaximise_rateSupplyCapacityBalance.maximiseRate

Workflow

  1. Data. Monthly allocated volumes per wellbore and a daily sample (a few days every 6 months) with WHP, choke, on-stream hours. Pull per year with retries and check month coverage per facility: a 5-year query window can time out and silently return zero rows.
  2. Tank. Use shut-in WHP (on-stream hours = 0, WHP above ~20 barg, per-window maximum or upper quartile) as the pressure proxy. Convert with the static column, fit p/z against cumulative gas, reject outliers (single-well, recently shut-in). Check that the intercept is a sane initial pressure and that two compartments with the same gas give similar p/z_i.
  3. Wells. Fit the Rawlins-Schellhardt law on rows with the choke wide open only. Early-life rates are choke-controlled and give nonsense (negative quadratic term, R2 below 0). Keep n between 0.5 and 1.0; a short window can return n above 1, so fix n from a longer-record neighbour and refit C. Always hold out the latest 2-3 years.
  4. Equivalent wells. Wells on stream = sum of on-stream hours / (24 x days in month). Do not use the well count from the register; subsea templates ramp up and wells cycle.
  5. Facility. Run the platform process model over a throughput grid at several separation pressures (throughput_sweep). Capacity is where first-constraint utilisation reaches 1.0 (capacity_from_sweep). Utilisation above 1.0 at every rate means a head or speed limit that does not depend on flow, so that pressure is below the facility floor. Scale capacity with polytropic head when the discharge or export pressure changes.
  6. Balance. Wellhead pressure is the separation pressure plus a flowline drop that scales with rate squared. Solve q = rate(Ps + dp (q/q0)^2) by bisection (flow_balance); fixed-point iteration oscillates and can collapse to zero. Choose Ps with maximise_rate; the binding side (SUPPLY or FACILITY) is the headline result, per month.
  7. Export coupling. Several fields feeding one export line raise the discharge pressure through P_man^2 = P_arr^2 + K q_total^2 (K from one measured point). The arrival pressure is an input; the receiving terminal is not modelled unless asked.
  8. Scenarios and uncertainty. Fixed versus optimised Ps, extra low-pressure compression (extra capacity and a lower Ps floor), extra wells, capacity upgrades. Bootstrap the p/z points and refit (p/z_i and GIIP jointly), sample the deliverability exponent, well performance and capacity.

Python Usage Pattern

from reservoir_facility_value_chain import (
    Gas, GasTank, PressureCapacityCurve, RSDeliverability, maximise_rate, static_bottomhole_pressure)

gas = Gas({"C1": 0.92, "C2": 0.04, "CO2": 0.02, "N2": 0.02})
pr_now = static_bottomhole_pressure(whp_barg=220.0, gas=gas, tvd_m=1500.0, t_mean_k=330.0)
tank = GasTank(gas=gas, t_res_k=345.0, pz_i=300.0, giip=1.0e11)
pr = tank.pressure(cum=2.0e10)                                   # reservoir pressure after 2e10 Sm3
wells = RSDeliverability.fit([320, 300, 280, 260], [200, 180, 160, 140], [9.0, 8.0, 7.5, 6.0])
capacity = PressureCapacityCurve([60, 70, 80], [30.0, 36.0, 41.0])   # from the process-model sweep
ps, rate, binding = maximise_rate(lambda ps: wells.rate(pr, ps + 10.0), capacity, 60.0, 80.0)

The numbers above only show the call shape; use field data and a real sweep for a study.

Common Mistakes (each cost a failed run)

  • Pressure is a difference of two large numbers. Moving GIIP by 8 % with p/z_i fixed put the present reservoir pressure at the wellhead pressure and the calibration became complex. Sample (p/z_i, GIIP) together from a bootstrap of the data, never independently.
  • Oil-rim shut-in WHP is an index, not an absolute pressure (liquid column); report it as such and label the fitted GIIP "effective".
  • Allocated gas of oil-rim wells is largely gas-cap gas; model it by drawdown scaling against the gas-cap pressure, not by GOR times oil.
  • Month 0 of a pull is usually partial. Exclude the current month from rates and totals.
  • Hindcast honestly: tank fitted to 2020, deliverability trained to 2023, tested 2024-2026. Report the bias per year; monthly errors are dominated by outages that on-stream hours do not capture.
  • Comparing options (a well swap, a tie-back, a host change): (1) put every option on one calendar
    • a surrogate that starts both routes in the same year hides the years a template delays the gas (it flipped a separate-compartment NPV from -0.5 to -1.0 GNOK). (2) Decompose a bundled proposal into its parts (an extra well on a shared tank is acceleration worth about +1 GNOK; the tie-back that carries it can cost about -1 GNOK) and value each alone; the sum hides which part pays. (3) A well-count cap on "extra wells" must be read against the base project's own wells; a cap of a few MSm3/d silently removes the whole project. Give capacity as MSm3/d for all wells at that host, sized to the planned wells. (4) Check that the stated base case and the code agree (a notes table said f = 0.8 while the headline run used f = 1.0); run both and say which one the headline uses. (5) Check a flowline dP assumption against a real hydraulic run at the actual well rate and host pressure. For a 320 m riser in wet gas the gas-only head is about 1 bar, but Beggs and Brill adds 2-3 bar of condensate holdup; treat that as an upper bound and cross-check with OLGA.

Validation Checklist

Run python -m pytest in the skill folder. For a real study, validate the tank against held-out shut-in pressures, the wells against a hold-out period, and the facility curve against the process model at two operating points.

  • The p/z intercept gives a sane initial pressure and two compartments with the same gas give similar p/z_i.
  • The deliverability exponent n lies between 0.5 and 1.0 and the fit uses choke-wide-open rows only.
  • The latest 2-3 years are held out and the bias per year is reported.
  • Every compared option uses one calendar and the stated base case matches the code.

Limitations

  • Screening layer: dry-gas Kay's-rule properties, a single tank per compartment and a quadratic flowline drop.
  • Facility numbers come from process-model sweeps; the receiving terminal is not modelled unless asked.
  • Monthly errors are dominated by outages that on-stream hours do not capture.
  • Built from the Troll chain; other fields need their own data checks.

Related

neqsim-integrated-production-and-lifecycle, neqsim-capacity-and-utilization-analysis, neqsim-production-optimization, neqsim-reservoir-model-builder, neqsim-near-well-and-injectivity, neqsim-ncs-value-chain-optimization, neqsim-optimization-and-doe.

References

  • Rawlins, E.L. and Schellhardt, M.A. (1935), Back-Pressure Data on Natural Gas Wells and Their Application to Production Practices, US Bureau of Mines Monograph 7.
  • Dranchuk, P.M. and Abou-Kassem, J.H. (1975), Calculation of Z factors for natural gases using equations of state, J. Can. Pet. Technol. 14(3).
  • NeqSim neqsim.process.fielddevelopment.integrated classes named in the method table.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.