NeqSim Field Development Skill

Comprehensive reference for oil & gas field development using NeqSim's field development framework. Covers the full lifecycle from discovery through decommissioning, with emphasis on concept selection, production forecasting, economics, and risk assessment.

Field Development Lifecycle (Decision Gates)

Gate Packages: Verdict Versus Full Deliverable Set

• A gate verdict is not a gate package. A readiness verdict (ready/review/not_ready) answers "can we go to the gate". A full package answers "here is every product the gate requires, with an owner and a status". When the user asks to "deliver according to DG1" (or any gate), build a product register: one row per required product, a status from a fixed taxonomy (Evidenced / Partly evidenced / Draft for review / Owner action / Not applicable - confirm), the evidence path and the owning role. Never label the result as gate approval; the readiness label stays at most "ready for owner review".
• Build a concept long list before capacity sizing. Development concepts (FPSO, tie-back, subsea-to-shore, phased, joint development) are different from capacity variants of one concept (e.g. 100/150/180 kbbl/d). Screen the long list first, record why concepts are dropped, then size the kept concepts. Capacity variants alone do not satisfy a concept-screening product.
• StudyClass is not an AACE class or gate status. A screening study class (A/B/C) says how deep the work went; the cost estimate carries its own AACE class (5 to 1); the gate carries its own readiness. Report all three separately.
• Placeholder sources stay placeholders. Fiscal, tariff or price inputs marked as examples must be labelled as such in the package and listed as owner actions.
• Company gate product lists and governance live in enterprise skills (for Equinor: enterprise-field-development-decision-gate, gate_package.py); keep them out of this repo.

Workflow Orchestration

FieldDevelopmentWorkflow workflow = new FieldDevelopmentWorkflow("Field Name");
workflow.setStudyPhase(StudyPhase.FEASIBILITY);
workflow.setFidelityLevel(FidelityLevel.SCREENING);

// Set reservoir
ReservoirInput reservoir = new ReservoirInput();
reservoir.setFluidType("gas_condensate");
reservoir.setGIIP(1.5e9);  // Sm3
reservoir.setReservoirPressure(350.0);  // bara
reservoir.setReservoirTemperature(95.0);  // °C
reservoir.setRecoveryFactor(0.65);
workflow.setReservoirInput(reservoir);

// Set wells
WellsInput wells = new WellsInput();
wells.setNumberOfProducers(4);
wells.setWaterDepth(350.0);
wells.setTotalDepth(3800.0);
workflow.setWellsInput(wells);

// Set infrastructure
InfrastructureInput infra = new InfrastructureInput();
infra.setDevelopmentType("subsea_tieback");
infra.setTiebackDistance(25.0);  // km
workflow.setInfrastructureInput(infra);

// Run
WorkflowResult result = workflow.run();

Key Result Classes

Concept Selection (Multi-Concept Comparison)

Define Concepts

// Concept 1: Subsea tieback to existing platform
FieldConcept tieback = new FieldConcept("Subsea Tieback");
tieback.setReservoirInput(reservoir);
tieback.setWellsInput(wells);
InfrastructureInput tiebackInfra = new InfrastructureInput();
tiebackInfra.setDevelopmentType("subsea_tieback");
tiebackInfra.setTiebackDistance(25.0);
tiebackInfra.setHostCapacity(50000.0);  // boe/d
tieback.setInfrastructureInput(tiebackInfra);

// Concept 2: Standalone FPSO
FieldConcept fpso = new FieldConcept("Standalone FPSO");
fpso.setReservoirInput(reservoir);
fpso.setWellsInput(wells);
InfrastructureInput fpsoInfra = new InfrastructureInput();
fpsoInfra.setDevelopmentType("fpso");
fpso.setInfrastructureInput(fpsoInfra);

// Concept 3: Fixed platform
FieldConcept platform = new FieldConcept("Fixed Platform");
// ... configure ...

Batch Evaluation

BatchConceptRunner runner = new BatchConceptRunner();
runner.addConcept(tieback);
runner.addConcept(fpso);
runner.addConcept(platform);
runner.setFluid(tunedEosFluid);
runner.setOilPrice(70.0);    // USD/bbl
runner.setGasPrice(0.30);    // USD/Sm3
runner.setDiscountRate(0.08);

List<ConceptKPIs> results = runner.runAll();

// Rank by NPV
DevelopmentOptionRanker ranker = new DevelopmentOptionRanker();
for (ConceptKPIs kpi : results) {
    ranker.addOption(kpi);
}
List<ConceptKPIs> ranked = ranker.rankByNPV();

Physically Coupled Lifetime Comparison

Use process.fielddevelopment.lifecycle after rapid screening when reservoir pressure, well/SURF hydraulics, processing capacity, gas injection and economics must share one time-dependent state:

FieldLifecycleEvaluator evaluator = new FieldLifecycleEvaluator();
List<FieldLifecycleResult> ranked = evaluator.evaluateAll(
    NorwegianOilFieldCase.createDevelopmentPortfolio());

FieldLifecycleModel accepts a user-built ProcessSystem or a complete multi-area ProcessModel, so screening concepts can be refined without replacing the lifecycle/economic interface. Use existingSurfAndFacility(...) when the new field enters a shared existing SURF area before a host facility; wire the areas with shared streams and map host feeds at their actual subsea or topsides entry points. Whole-model execution, power, sizing and area::equipment bottlenecks then include both SURF and processing. Each concept must own an independent mutable reservoir/process model. Use FacilityLifecycleStrategy.greenfield(...) to auto-size a new detailed processing facility from design cases. Use FacilityLifecycleStrategy.tieback(...) with the existing HostFacility, ProductionProfileSeries, CapacityAllocationPolicy, and HoldbackPolicy types for a producing-host tieback. Connect the model's optional host oil/gas/water feeds to the real shared process so annual FieldLifecycleResult records host load, admitted satellite rate, deliberate holdback, capacity-deferred oil, operating and requested utilization, and their primary bottlenecks. Set FieldLifecycleModel.setProductionPotentialProvider(...) when a detailed NeqSim well/network model or imported reservoir schedule should replace the reference aggregate PI/water-cut potential calculation.

Use AreaDevelopmentPortfolio when one discovery can be routed to several independently modeled producing assets or to a new facility. Add one AreaDevelopmentOption per host/tieback route or greenfield alternative, then use AreaDevelopmentEvaluator to compare route identity, NPV, break-even, recovery, deferment, utilization, bottlenecks and specification compliance. Each option must own its reservoir, SURF and mutable process state.

Attach FieldProductSpecifications to each configuration to check live export-gas CO2/H2S/O2, ISO 6976 GCV/Wobbe, dew points, stabilized-oil RVP/BS&W and treated-water oil-in-water. The built-in evaluator uses NeqSim stream analysers and standards; connect FieldProductQualityProvider where a real facility has a dedicated water-treatment or online-analyser model. Choose REPORT_ONLY for diagnostics or REJECT_OPTION to exclude non-compliant options from the area recommendation.

For brownfield studies, preserve actual equipment limits (autoSizeDetailedProcess(false)). For greenfield studies, run the process at the simultaneous design case, auto-size equipment with the selected design margin, and use explicit oil/gas/water/liquid nameplates for non-coincident component peak cases. Put any debottleneck capacity change in capacityFromYear(...) and its CAPEX in FieldLifecycleConfiguration. Use FacilityModificationPlanner.analyse(result, targetUtilization) to create traceable screening candidates from annual requested utilization and deferred oil. Implement, size and cost each candidate in a cloned detailed process before comparing its rerun lifecycle; the planner's multiplier is not a substitute for that engineering model. See docs/fielddevelopment/FIELD_LIFECYCLE_SIMULATION.md for the connection points and fidelity boundaries.

Reservoir & Well Modeling

Material Balance (SimpleReservoir)

SimpleReservoir reservoir = new SimpleReservoir("Main Reservoir");
reservoir.setReservoirFluid(fluid.clone(), giipSm3, reservoirThickness, reservoirArea);
reservoir.addOilProducer("P1");
reservoir.addWaterInjector("I1");

// Depletion with time steps
for (int year = 0; year < 20; year++) {
    reservoir.setProductionRate(annualRate[year]);
    reservoir.run();
    pressureProfile[year] = reservoir.getReservoirPressure();
}

Inverse Material Balance / Reserves Surveillance

Regress OGIP/OOIP, drive mechanism and aquifer support directly from a measured pressure-vs-cumulative-production history (the "inverse" of the forward SimpleReservoir model). Package neqsim.pvtsimulation.reservoirproperties.materialbalance:

import neqsim.pvtsimulation.reservoirproperties.materialbalance.GasMaterialBalance;
import neqsim.pvtsimulation.reservoirproperties.materialbalance.OilMaterialBalance;
import neqsim.pvtsimulation.reservoirproperties.materialbalance.VanEverdingenHurstAquifer;

// Gas: P/Z straight line → OGIP (+ Cole plot aquifer diagnostic)
GasMaterialBalance.Result g = GasMaterialBalance.fitVolumetric(pressure, z, gp);
double ogip = g.getOgip();

// Oil: Havlena-Odeh depletion / gas-cap / water drive + Pirson drive indices
OilMaterialBalance.Result o = OilMaterialBalance.fitGasCapDrive(f, eo, eg);

// Aquifer: Van Everdingen-Hurst / Carter-Tracy cumulative influx + ECLIPSE AQUTAB
double u = VanEverdingenHurstAquifer.aquiferConstant(phi, ct, h, re, angleDeg);
double[] we = VanEverdingenHurstAquifer.cumulativeInfluxCarterTracy(tD, deltaP, u, reD);

See the neqsim-production-optimization skill for the full surveillance workflow.

For radial input conversion, use inner reservoir radius in metres. Compressibility is 1/Pa for dimensionlessTime and 1/bar for aquiferConstant: ctBar=1e5*ctPa. Physical zeros are distinct from invalid inputs or an unrepresentable positive conversion; do not turn a conversion exception into a zero-support scenario.

For radial support, the infinite-acting pressure/analytic derivative use a qualified mathematical spectral kernel. The finite-radius maximum-policy option remains an unqualified transient approximation, despite sharing the improved infinite branch. Preserve its explicit zero-time/finite-radius limitations in docs/pvtsimulation/reservoir_material_balance.md#aquifer-influx; do not use it as independently validated bounded radial support or equate it with segregated Fetkovich coupling. Carter–Tracy timestep convergence does not remove its intrinsic approximation error. Mathematical references are not measured field acceptance. The current infinite-radial/initialization65/ fixture records initialization-only identity, concurrent first fallback and cold/warm/deferred runtime tradeoffs. Its parent optimization64/ retains the positive-tail correction and complete numerical boundary controls. The current scaling66/ receipt qualifies normalized CT arithmetic across scales, analytic step/ramp limits, signed cumulative projection and explicit invalid/unrepresentable histories. Its bounded numerical controls do not admit finite-annulus physics. Preserve all historical receipts at their exact revisions; current CT values are checked against original independent criteria rather than claimed bit-identical to the earlier recurrence evaluation.

Injection Strategy (Voidage Replacement)

InjectionStrategy strategy = InjectionStrategy.waterInjection(1.0);  // VRR = 1.0
InjectionResult injection = strategy.calculateInjection(
    reservoir, oilRate, gasRate, waterRate
);
double requiredInjection = injection.waterInjectionRate;  // Sm3/d

Well Performance (IPR/VLP Nodal Analysis)

WellSystem well = new WellSystem("Producer-1", reservoirStream);
well.setIPRModel(WellSystem.IPRModel.VOGEL);
well.setVogelParameters(qTest, pwfTest, pRes);
well.setTubingLength(2500.0, "m");
well.setTubingDiameter(4.0, "in");
well.setPressureDropCorrelation(
    TubingPerformance.PressureDropCorrelation.BEGGS_BRILL);
well.setWellheadPressure(50.0, "bara");
well.run();

double operatingRate = well.getOperatingFlowRate("Sm3/day");
double operatingBHP = well.getOperatingBHP("bara");

Production Network

NetworkSolver network = new NetworkSolver("Gathering System");
network.addWell(well1, 3.0);   // 3 km flowline
network.addWell(well2, 5.5);   // 5.5 km flowline
network.setSolutionMode(SolutionMode.FIXED_MANIFOLD_PRESSURE);
network.setManifoldPressure(60.0);
NetworkResult result = network.solve();

Production Forecasting

Decline Curves (Arps)

ProductionProfile profile = new ProductionProfile();
profile.setDeclineModel(ProductionProfile.DeclineModel.EXPONENTIAL);
profile.setPeakRate(25000.0);          // boe/d
profile.setDeclineRate(0.15);          // 15% per year
profile.setPlateauDuration(3);         // years
profile.setProjectLife(25);            // years
double[] annualRates = profile.generateProfile();

Decline-Curve History Matching (Arps + Duong)

Fit decline parameters to a measured rate-time history and estimate EUR with neqsim.pvtsimulation.util.DeclineCurveAnalysis (static, unit-agnostic):

import neqsim.pvtsimulation.util.DeclineCurveAnalysis;

Map<String, Double> arps = DeclineCurveAnalysis.fitArps(t, q);       // qi, di, b, rSquared
double eur = DeclineCurveAnalysis.eurFromFit(arps, qEconLimit);
Map<String, Double> duong = DeclineCurveAnalysis.fitDuong(t, q);     // q1, a, m for tight/shale wells

Production Scheduling

FieldProductionScheduler scheduler = new FieldProductionScheduler();
scheduler.setNumberOfWells(6);
scheduler.setDrillingInterval(6);      // months between wells
scheduler.setFirstOil(2027);
scheduler.setWellProductivity(5000.0); // boe/d/well
scheduler.setDeclineRate(0.12);
scheduler.setFacilityCapacity(28000.0);  // boe/d plateau constraint
double[][] schedule = scheduler.generateSchedule(25);

Well Scheduling

WellScheduler wellScheduler = new WellScheduler();
wellScheduler.setDrillingDaysPerWell(45);
wellScheduler.setCompletionDaysPerWell(25);
wellScheduler.setMobDemobDays(30);
wellScheduler.setRigs(1);
wellScheduler.setWells(6);
wellScheduler.setStartDate(2026, 1);
List<WellSchedule> schedule = wellScheduler.generate();

Facility Design Integration

Building a Process System from Concept

FacilityBuilder builder = new FacilityBuilder();
FacilityConfig config = new FacilityConfig();
config.setInletPressure(85.0);     // bara
config.setExportPressure(150.0);   // bara
config.setDesignRate(50000.0);     // boe/d

// Add processing blocks
config.addBlock(BlockType.INLET_SEPARATION, new BlockConfig()
    .set("stages", 2)
    .set("hp_pressure", 70.0)
    .set("lp_pressure", 5.0));
config.addBlock(BlockType.GAS_COMPRESSION, new BlockConfig()
    .set("stages", 3)
    .set("export_pressure", 150.0));
config.addBlock(BlockType.DEHYDRATION, new BlockConfig()
    .set("type", "TEG"));

ProcessSystem process = builder.build(config, fluid);
process.run();

Bottleneck Analysis

BottleneckAnalyzer analyzer = new BottleneckAnalyzer(process);
analyzer.setReservoirDecline(reservoir);
analyzer.setRateRange(10000, 60000, 5000);  // min, max, step (boe/d)
Map<String, Double> bottlenecks = analyzer.findBottlenecks();

Subsea Production Systems

Tieback Analysis

TiebackAnalyzer analyzer = new TiebackAnalyzer();

// Define host
HostFacility host = new HostFacility("Platform A");
host.setAvailableCapacity(30000.0);  // boe/d
host.setProcessingPressure(70.0);    // bara

// Define tieback options
TiebackOption opt1 = new TiebackOption("Direct Tieback");
opt1.setFlowlineLength(15.0);       // km
opt1.setFlowlineDiameter(10.0);     // inches
opt1.setWaterDepth(350.0);          // m

TiebackOption opt2 = new TiebackOption("Via Manifold");
opt2.setFlowlineLength(25.0);
opt2.setFlowlineDiameter(12.0);
opt2.setWaterDepth(450.0);
opt2.setHasManifold(true);
opt2.setHasBooster(true);

analyzer.setHost(host);
analyzer.addOption(opt1);
analyzer.addOption(opt2);
analyzer.setFluid(reservoirFluid);

TiebackReport report = analyzer.analyze();

Host impact of a tie-in on a benchmarked host model

When the host already has a benchmarked NeqSim ProcessModel, assess the tie-in on a reproducible copy of that model:

1. Record the baseline model version, operating window and constraint basis. If the host package and workspace require incompatible NeqSim JARs, use separate Python processes with explicit classpaths and exchange inputs/results; do not mix those JARs in one JVM.
2. Use the tie-in fluid's own composition and PVT evidence with compatible EOS, mixing rules and component definitions. If only host-fluid data are available, label that fluid as a surrogate and quantify the sensitivity. Recombine oil, gas, water and lift gas on a molar basis; document standard-volume reference conditions, GOR and water-cut definitions, and verify them by flashing the reconstructed fluid. Initialize compositions before reading mole fractions.
3. Connect the additional Stream to the intended inlet Mixer using the host model's supported extension point. Include each external feed exactly once in total and component balances and check the energy balance. An unaccounted feed requires a boundary-accounting correction, not a change that hides a residual. Preserve the baseline for comparison.
4. Run plateau, mid-life and late-life points over the relevant benchmark windows. Report both absolute and incremental compressor power, separator utilisation, export oil and produced water against documented limits. Identify constraints already exceeded in the baseline separately from those introduced by the tie-in; do not assume a universal utilisation threshold.

Evaluate tubing and line back-pressure during depletion

Use the coupled reservoir inflow and tubing/flowline outflow response to determine feasible rates as reservoir pressure, water cut, GOR, gas lift and host pressure change. Bubble point alone does not define a minimum flowing bottom-hole pressure or a recovery ceiling. Establish flow/no-flow conditions from the available reservoir pressure and the required pressure at the same datum.

Reuse the current VFP generation and import/export APIs (see MultiScenarioVFPGenerator and neqsim-subsea-and-wells), checking units, datum depth, axis definitions, feasible points and the reservoir simulator's table format. Compare gas-lift and injection scenarios on the same basis. If a VFP table causes convergence trouble, inspect interpolation, operating branches and control switching. Do not force BHP versus rate to be monotonic just to obtain convergence. The OPM VFP consistency guidance discusses crossing curves at adjacent tubing-head pressures, a different check from monotonicity with rate. Preserve the original table and validate any justified conditioning against the hydraulic model before using it for recovery or economic conclusions.

Subsea System Configuration

SubseaProductionSystem subsea = new SubseaProductionSystem("Subsea System");
subsea.setNumberOfWells(4);
subsea.setWaterDepth(350.0);
subsea.setFlowlineLength(25.0);
subsea.setUmbilicalLength(27.0);
subsea.setRiserType("flexible");
subsea.setTreeType("vertical");
subsea.setHasManifold(true);
subsea.setManifoldWells(4);

Key NeqSim Classes for Field Development

Common Patterns & Pitfalls

Always Verify API Before Use

NeqSim's field development classes are under active development. Before using any class, search the source to confirm:

1. The class exists and compiles
2. Constructor signatures match expected parameters
3. Methods have the expected return types

Temperature Units

• ReservoirInput.setReservoirTemperature() — Celsius (field development classes)
• SystemSrkEos(T, P) constructor — Kelvin
• Always confirm with source which convention each class uses

Production Rate Units

• NeqSim process equipment: setFlowRate(value, "kg/hr") or "Sm3/hr"
• Field development classes: typically boe/d or Sm3/d
• Always specify units explicitly

EOS Selection for Field Development

Fiscal Regime

• Norwegian NCS: 22% corporate + 56% special petroleum tax = 78% marginal rate
• UK UKCS: 30% ring-fence CT + 10% supplementary charge = 40% marginal rate
• Generic: Configurable via GenericTaxModel
• Always use CashFlowEngine with the correct country code

Staged-option VPbo / DG1 screening for an unappraised HPHT discovery (lessons from Afrodite PL293)

• Value the decision as a staged option (stop / commit now / mature to DG2 then develop only if positive); an at-break-even project (NPV about 0 at P50) still has positive EMV when maturation is cheap, so state EMV of maturing, not just P50 NPV.
• Host life is a cost, not a constraint: price the years the profile outlives the host plan (host-alive cost, PV) and report a COP-shift curve; ullage rarely discriminates, host life and cost do.
• Reservoir-model need is a value-of-information question: EVPI on volume vs EVSI of a sidetrack vs its net cost, with the well-count rule (about 45 % of volume over a 4-year plateau).
• Cluster upside: report the share of nearby discovered gas that must be tied in for NPV > 0, not a single NPV.
• Report break-even gas price from the model by bisection, never a rounded literal; keep literals out of narrative strings (derive them from the results dict).
• HPHT dew point: do not trust dewPointPressureFlash (wrong root at 168 C); use a TP-flash bisection and CCE dropout. TiebackAnalyzer rejects shallow hosts through its water-depth gate; screen those hosts with the explicit pipe model.

Qualification output retention

Keep required regression inputs, compact independent references and reproduction scripts in Git. Store bulky generated qualification histories and repeated before/after matrices outside the checkout, preserving full outputs, hashes and public provenance. The radial fixture README documents restoration for PR #4924. Do not re-add its archived run outputs during campaign continuation.