Native CCE, CVD and DLE laboratory PVT handoff

Closed-mass CCE without an inferred saturation reference

Use DepletionPvtHistoryGenerator.replayNative with Path.CCE for a complete dry closed-mass constant-composition expansion (also called CME). The selected original or tuned EOS is restored independently; its complete component inventory remains in the cell at every pressure. Gas is not removed. Both native equilibrium and retained snapshots describe that same whole cell. hasWithdrawal(row) is false; getWithdrawalProducts(row) throws because no laboratory withdrawal exists.

This route invokes ConstantMassExpansion.runNativeExpansion, which TP-flashes the closed cell on the declared pressure path and returns immutable snapshots. It does not search for saturation or populate the legacy CME relative-volume, Y-factor, formation-factor or transport arrays. Inspect actual phase presence and call initProperties() on restored states before using properties. Absolute cell volumes are in m3; any reported relative volume must name an actual reference row. Do not relabel that reference as a measured or predicted saturation point.

Supply the same immutable sample/model, complete experiment identity, temperature in K, decreasing pressures in bara and fixed PvtSeparatorPath as the native depletion route below. Pressure points must remain above the first separator. getRetainedProducts(row) evaluates the whole cell through that train; these products are a declared surface process, not fluids withdrawn during the CCE. Named closed-cell component closure is checked at 1e-10 * max(1, initial mol). Numerical closure error is not converted into invented CCE production. A failed flash propagates its original cause and pressure context; no successful result or partial snapshot array is returned.

Oil-only, gas-only and nondegenerate gas/oil CCE states are represented without invented absent-phase properties. The qualified synthetic closed-mass gas/oil/gas path preserves all material through phase appearance and disappearance. This is different from gas-withdrawal CVD, whose later single-phase volume-restoration tail is separately qualified below. Closed-mass wet CCE support is described below; wet withdrawal protocols and critical-degenerate splits remain rejected.

NativeCceHandoffTest verifies provenance/ownership, both EOS selections, no-flow and phase limits, failures, named cell/separator conservation, independent pure SRK pressure-equation roots and six binary SRK coexistence states. The bounded independent reference contains explicit inputs and outputs. These are analytical software checks, not measured laboratory calibration/holdouts or an accepted laboratory-to-field chain. The native CVD saturation-reference refinement below remains separate from closed-mass CCE.

Native compositional handoff through phase appearance

Use DepletionPvtHistoryGenerator.replayNative when a dry CVD/DLE path includes an absent reservoir phase. It reuses the existing solver and immutable DepletionSnapshot records, returning actual equilibrium/retained EOS states and separator products. It does not assign Bo/Bg/Rs/Rv/Bw or transport properties to an absent phase, and does not produce a generalized black-oil history.

Supply independently frozen original/tuned models, the selected version (false original / true tuned), complete experiment identity, temperature in K, at least two decreasing pressures in bara, and an immutable PvtSeparatorPath. Depletion pressures must exceed the first separator pressure. Its final temperature/pressure define the product-volume basis; other bases are retained explicitly rather than relabeled as 15 C volumes.

NativeResult retains both immutable checkpoints, the selected EOS, experiment identity, fixed separator basis and actual pressure-ordered snapshots. Restore a snapshot, inspect hasPhaseType("gas") / hasPhaseType("oil"), then call initProperties() before reading density or transport properties. Returned EOS states are independent copies.

getRetainedProducts(row) processes the whole retained cell through the separator train. hasWithdrawal(row) is false if no fluid was withdrawn; then getWithdrawalProducts(row) throws. Otherwise it returns interval, not cumulative, products. Reservoir phase absence and zero withdrawal are distinct. A completed separator can physically have zero volume of one product phase; this does not define a missing reservoir phase’s formation factor.

Native chemical component closure is checked against the original named sample, including sample-to-first-row withdrawals, at 1e-10 * max(1, initial component mol). Separator component closure is checked by PvtSeparatorPath. No conservation of surface-volume pseudo-components is assumed. These are laboratory protocol outputs, not measured field production. A DLE cell changes volume and cannot be relabeled as a fixed-pore-volume reservoir.

The following synthetic example is exercised by NativeDepletionHandoffTest. Identical checkpoints demonstrate API use; they do not establish independently calibrated laboratory accuracy. Real workflows preserve separate calibration and whole-experiment holdouts, distinct from reservoir pressure/GOR/composition holdouts.

SystemInterface original = new SystemSrkEos(323.15, 500);
original.addComponent("methane", 0.7);
original.addComponent("n-decane", 0.3);
original.setMixingRule("classic");
original.useVolumeCorrection(false);
original.init(0);
CalibratedPvtModel model = new CalibratedPvtModel(original, original.clone(),
    "synthetic native sample", "original SRK v1", "accepted SRK v1",
    "synthetic laboratory protocol, not measurements", new String[] {"CCE-1"});
PvtSeparatorPath separator = new PvtSeparatorPath(new double[] {10, 1.0133},
    new double[] {303.15, 288.71},
    "synthetic two-stage train, final 15.56 C / 101.33 kPa absolute");
DepletionPvtHistoryGenerator.NativeResult result =
    DepletionPvtHistoryGenerator.replayNative(model, true, MaterialBalancePvtHistory.Path.DLE,
        "whole synthetic DLE-1", 323.15, new double[] {500, 300, 200, 100, 50}, separator);
boolean initialWithdrawal = result.hasWithdrawal(0); // false: closed oil-only state
SystemInterface initialCell = result.getSnapshots()[0].getRetainedFluid();
double[] retainedProducts = result.getRetainedProducts(0).getVolumesM3();

DLE equilibrium, post-saturation and final stock-tank flash failures now propagate their original cause and stage context. Failed runs cannot return a successful native handoff. Saturation-scan failures also propagate; successful numerical paths are unchanged.

Qualification covers phase appearance, closed-oil limits, selected original/tuned EOS parameters, named component and separator closure, non-default product bases, immutability/serialization and injected flash failures. A pure n-decane test independently bisects the SRK pressure equation using the declared critical data, Soave alpha and library EOS gas constant, with 1e-8 relative volume acceptance. This is numerical EOS consistency, not measured laboratory-to-field accuracy, wet/critical qualification or recoverable-reserves acceptance.

Independent binary qualification and the CVD saturation reference

The reproducible binary SRK audit adds independent phase-equilibrium, lever-rule and gas-withdrawal predictions for nine synthetic DLE/CVD states. The historical legacy phase-count scan selects 359.241739 bara versus the independent upper dew of 359.245071 bara; its CVD cell volume misses the frozen 1e-6 relative bound by 6.530e-6.

Native replayNative now enables ConstantVolumeDepletion.setRefineSaturationConditions(true). The scan first selects a coarse upper gas/oil-to-gas bracket below 1000 bara. A cloned two-phase warm state then feeds the existing constant-duty pressure solver with an incipient liquid fraction of 1e-10. Acceptance requires the same coarse branch (1e-5 bara numerical allowance), normalized nonnegative phase compositions and component reconstruction within 1e-8, nontrivial composition distance above 1e-8, and maximum component log-fugacity mismatch at most 1e-8. The accepted pressure is reflashed with the original whole-cell inventory before defining the CVD reference volume. No EOS input is fitted. Missing/unsupported brackets and failed equilibrium gates throw; there is no fallback to the scan or an unrelated lower-pressure dew root.

The full independent CVD check now passes: maximum relative retained-volume error 4.730e-11, inventory/withdrawal error 4.070e-11 mol, phase mole-fraction error 4.530e-14, and component closure 5.552e-17 mol for the one-mole reference. The original 1e-6 volume/composition/inventory and 1e-10 mol closure bounds are unchanged. Three sample scales also have regression coverage. DLE remains within 1.066e-13 relative volume error.

The direct CVD default and modified-black-oil generate route retain their legacy scan and its precision limitation. The opt-in refinement supports dry, nondegenerate upper dew boundaries; pure fluids, wet paths, oil bubble boundaries, unbracketed/1000-bara domains remain outside its qualified domain. These are analytical synthetic consistency checks, not measured laboratory/field accuracy or full phase-boundary qualification.

Bounded modified-black-oil conversion

DepletionPvtHistoryGenerator.generate evaluates pressure-aligned Rs, Rv, Bo, Bg and Bw records from a frozen CalibratedPvtModel. It runs the existing constant-volume-depletion or differential-liberation solver. The retained composition changes as gas is withdrawn. Independent flashes of the original composition are not used as depletion states.

This is a bounded modified-black-oil interpretation. It does not make the two surface pseudo-components exact conserved chemical components. Keep the complete result and inspect the conversion error before using its MaterialBalancePvtHistory with measured reservoir production.

Inputs and supported domain

Supply the calibration bundle, Path.CVD or Path.DLE, protocol/software reference, reservoir temperature in K, and at least three strictly decreasing pressures in bara. The accepted depletion sample must contain no water. Each interpreted pressure must have both hydrocarbon oil and gas. An absent phase’s properties are rejected rather than filled by an arbitrary value or extrapolated to an incipient phase. Consequently this version cannot generate a complete above-saturation-to-depletion table. Near-critical degenerate Rs*Rv splits are also rejected by the generalized-balance checks.

The fixed separator stages use pressure in bara and temperature in K. Pressures must decrease and end at 1.01325 bara and 288.15 K. Every reservoir pressure must exceed the first separator pressure. Liquid advances to the next stage; each stage’s offgas is separately flashed to standard conditions. Both dry gas and any condensate formed by that offgas are accounted for. All stages and outputs must conserve every native EOS component.

Bw requires a separately accepted pure-water EOS and its provenance reference. Its native checkpoint is retained in Result.getWaterFluid(). It must contain water only and remain one liquid phase at reservoir and standard conditions. Bw is the standard-to-reservoir density ratio. This reference does not model brine, dissolved gas, water exchange with hydrocarbons or a water-containing depletion experiment. Do not pass an unqualified water model merely to complete the argument list.

Evidence and acceptance

Both depletion solvers can opt into native recording with setRecordDepletionSnapshots(true). The default remains false to preserve their normal memory and runtime cost. getDepletionSnapshots() returns immutable per-pressure evidence. Restored fluids and amount arrays are independent copies. CVD records its equilibrated volume-restored cell. DLE records equilibrium before gas removal and retained liquid afterwards. Component withdrawals are differences between consecutive retained inventories, including the original sample for the first step. Solver results are unchanged when recording is used.

The separator calculation is shared with PvtSeparatorPath; see phase-selective production and surface accounting. The modified-black-oil generate method retains its existing dry/two-hydrocarbon-phase restrictions.

The generator interprets oil and gas from each actual equilibrium snapshot through the same surface process. Reservoir and surface volumes consistently include the EOS volume correction. Formation factors are reservoir m3/Sm3, Rs and Rv are Sm3/Sm3, and viscosities are Pa.s.

For each surface component, the reported relative residual is

\[r_k=\frac{S_0-S_{p,k}-S_{r,k}}{S_0}\]

Here $S_0$ is the separator-evaluated first retained cell, $S_{p,k}$ is cumulative separator-evaluated experiment withdrawal after that point, and $S_{r,k}$ is the remaining cell processed by the same path. Both surface oil and surface gas must be defined at the origin. Native chemical-component conservation is checked separately before this approximation error is evaluated.

The caller supplies a maximum relative error strictly between zero and one, plus an engineering justification reference. The generator rejects a handoff when either surface-component error exceeds it. There is no default field acceptance tolerance. Result retains the bound, justification, actual maximum, signed residual arrays, experiment withdrawals, snapshots and the accepted PVT handoff. The bound is neither a confidence interval nor an estimate of PVT measurement uncertainty.

The first retained cell defines zero cumulative experiment withdrawals in the result. Earlier sample-to-first-grid withdrawal remains visible in snapshot zero. These laboratory withdrawals are not measured field production. A DLE cell changes volume, so its withdrawal history must not be treated as a fixed-pore-volume reservoir depletion case.

Qualification and remaining work

DepletionPvtHistoryGeneratorTest checks native component balances, existing CVD/DLE solver agreement, changed retained composition, independent single-stage phase flashes, independent water-density evaluation, separator-path effects, input/output ownership, serialization and explicit rejection gates. These are synthetic and numerical integration checks. They do not establish laboratory or field accuracy, above-saturation extrapolation, critical closure or the validity of the supplied EOS and acceptance tolerance.

The phase-flash interpretation follows Whitson and Torp (1983), SPE 10067-PA, available from the author’s NTNU course archive. The Whitson modified-black-oil documentation also explains the fixed surface-process and surface-density approximation.

See reservoir material balance for the measured history workflow and issue #4740 for the broader qualification contract.

Gas-only CVD continuation

A CVD cell remains at its selected saturation volume after the final liquid phase evaporates. At a lower pressure, withdraw the present gas in its actual component proportions until the cell returns to that volume. In a single phase at fixed temperature, pressure and composition, volume is extensive in moles; therefore the remaining fraction is the target volume divided by pre-withdrawal volume. The existing bounded gas-withdrawal loop performs this operation and reflashes/checks the cell. Above the saturation reference pressure, single-phase rows remain closed expansion states with zero withdrawal.

This correction also applies to direct legacy CVD only after a gas/oil stage has been sampled in the pressure schedule. An unbracketed or entirely single-phase schedule must not trigger tail withdrawal from the legacy scan’s arbitrary reference. This guard does not qualify that reference for pure or supercritical samples. The direct route retains its selected scan reference and precision limitation. Native CVD uses the qualified refined upper dew. It does not assign absent oil density, Bo or Rs. The modified black-oil route still requires both phases and rejects these rows.

The dry methane/n-butane275K synthetic reference covers500/200/100/50/20/5/2bara, including two-phase depletion and a gas-only tail. Independent SRK cubic roots, fugacity equality, two-phase algebraic gas removal and single-phase extensivity provide numerical references. This is model consistency, not measured lab/field accuracy, critical behavior, water support or recoverable reserves.

Closed-mass wet CCE

replayNative accepts gas, oil and aqueous phases for CCE only. The complete cell and every named water/hydrocarbon component remain at each pressure; protocol withdrawal is exactly zero. Existing PvtSeparatorPath processing retains all three product inventories and the explicit final pressure/temperature basis. An absent cell or surface phase has no invented composition or PVT.

Supply an independently accepted wet EOS and its appropriate multiphase-check configuration in each original/tuned checkpoint. The handoff does not select or retune that model, infer water density, estimate saturation, or calibrate a reservoir. Every present phase pair must have composition L1 distance greater than1e-8; this is an unchanged critical-degeneracy rejection threshold, not an accuracy tolerance. Solids/hydrates and other phase types remain unsupported. Wet CVD and modified-black-oil water conversion require separate qualification.

WetCceHandoffTest checks original/tuned serialization, aqueous separator closure, pure-water and single-phase wet-gas limits, nine independently solved gas/aqueous states and three independent gas/oil/aqueous states. The references use unshifted classic SRK and shared explicit component parameters, not measured water/hydrocarbon data. Volume relative and phase-composition/fraction absolute bounds are1e-8; cell component closure is1e-10mol and separator closure1e-8mol for the one-mole fixtures. These bounds establish numerical consistency within the stated synthetic cases, not measured-fluid accuracy or near-critical validity.

Full equations, probe inputs, predictions and provenance are retained in src/test/resources/neqsim/pvtsimulation/native-wet-cce.

Native wet differential liberation

DifferentialLiberation.runNativeLiberation is a separate absolute-state route. The supplied sample is copied; completed immutable snapshots are published only after the complete decreasing pressure schedule succeeds. Legacy normalized PVT arrays, saturation estimation and default runCalc behavior are unchanged. DepletionPvtHistoryGenerator.replayNative uses this route for dry and wet DLE.

At each pressure the retained sample is equilibrated, then the entire physically present gas phase is removed. Both oil and aqueous phases stay in the cell. Vaporized water leaves with the gas; dissolved gas stays with the liquids. An oil/aqueous state without gas has exactly zero protocol withdrawal and does not trigger a phase-count saturation scan. Oil must remain present; unsupported phase types and, in the handoff, critical-degenerate splits fail explicitly. A removed phase changes the sample total: phase fractions and overall composition are recomputed from retained inventories before properties or separation are read.

WetDleHandoffTest compares a methane/n-decane/water SRK path at 350 K and 200/100/50 bara against six independent fugacity equations and three component balances, followed by complete gas removal. Inputs, equations and complete reference outputs are in src/test/resources/neqsim/pvtsimulation/native-wet-dle. Frozen relative volume and absolute phase-composition/amount bounds are 1e-8; whole-cell component closure is 1e-10 mol for the one-mole fixture. Tests also check two-liquid no-gas behavior, checkpoint ownership, repeat replay, separator closure and rejected schedules/oil absence. These are synthetic consistency checks, not measured wet-fluid calibration or laboratory-to-field accuracy.

Keep original/tuned model, whole experiment and separator provenance distinct. This route supplies no absent-phase black-oil properties, no wet CVD reference, no critical accuracy claim and no reservoir production or recoverable reserves.

Explicit initial occupied-volume reference

DepletionPvtHistoryGenerator.replayNativeFixedCell is a separate opt-in for an experiment whose reference is the accepted EOS sample’s occupied volume at the first supplied pressure. It preserves selected original/tuned checkpoints, whole-experiment/sample provenance and one immutable PvtSeparatorPath. The returned NativeResult.hasFixedCellReference() identifies this convention; getFixedCellReference() supplies reference pressure in bara, temperature in K, actual occupied cell volume in m3, and immutable completed snapshots. It throws when called on the existing laboratory routes, which have no such reference.

At the first pressure withdrawal is zero. At each later pressure equilibrium gas alone is removed to restore that occupied volume; all nongas phases remain. The bounded route accepts unchanged gas, gas/oil, gas/aqueous or gas/oil/aqueous phase families. It rejects an absent gas, phase-family changes, critical-degenerate splits, unclosed equilibrium and unreachable volume restoration. It does not assign properties to absent oil or water. The entire run uses independent copies; a late failed trial cannot mutate the caller’s sample or publish partial results. Repeated calls restart from identical input. The core API is ConstantVolumeDepletion.runFixedCellDepletion(temperatureK, pressuresBara).

The opt-in route refines already accepted phase endpoints with the existing multiphase beta solver to a dimensionless log-fugacity residual of 1e-12 against the reference phase and a relative cell-volume residual of 1e-12. It permits at most 12 beta-refinement calls at each endpoint. Refinement must preserve the named phase family, component balance and Gibbs-energy acceptance; failure rejects the whole transaction. These internal stopping criteria improve absolute phase inventory precision at large sample amounts. They are distinct from the frozen independent benchmark bounds and do not guarantee absolute accuracy at arbitrary sample scales or prove global phase stability. Shared flash defaults and legacy CVD stopping criteria remain unchanged.

The Gibbs upper bound is evaluated at the same stored component inventory as the refined state. An approximate flash can have small phase-summed inventory errors, making its raw energy an invalid constrained reference. The precision route reconstructs a nonnegative, component-feasible allocation on copied homogeneous phases with the same EOS, temperature, pressure and phase branches. It verifies reconstruction to floating-point roundoff and retains the existing energy allowance, final component/fugacity checks and phase-family restriction. It does not fit the EOS, change the laboratory protocol or prove global stability. FixedCellFeasibleGibbsTest retains the three public ternary regressions at 300/325 K and 10 bara, with three extensive scales and immutable caller inputs. Its108 additional reference-only controls cover classic SRK/PR, volume correction off/on and nearby T/P. Analytic extensivity and fresh homogeneous constructions verify this energy calculation; they do not qualify a new depletion path or an independently calibrated or measured fluid. See the frozen EOS-control scope in src/test/resources/neqsim/pvtsimulation/fixed-cell-feasible-gibbs/.

This explicit initial-volume convention does not infer a wet upper dew point, normalize measured PVT to saturation, replace replayNative’s upper-dew protocol, or extend modified-black-oil conversion. Existing replayNative(..., Path.CVD, ...) still rejects water-bearing CVD. Choose the route from the actual experiment protocol; an attractive numerical match cannot change that protocol’s identity. Keep original/tuned laboratory calibration and whole-experiment holdouts separate from reservoir-volume matching and held-out field pressures/GOR/composition.

The independent reference and reproduction instructions cover the original216synthetic states across72complete pressure paths and480 additional states across96five-pressure paths, including wet gas without oil, nearby temperatures and three sample sizes. Both checkpoint selections and all separator products are tested. This is bounded numerical consistency, not measured laboratory/field accuracy, critical or global phase-stability acceptance.

Dry phase-changing fixed-cell experiment

For an independently declared dry initial-volume experiment, select DepletionPvtHistoryGenerator.replayNativeDryPhaseChangingFixedCell. Its core API is ConstantVolumeDepletion.runDryPhaseChangingFixedCellDepletion. Both preserve the explicit first occupied volume and permit oil to appear or disappear while gas remains present. FixedCellResult.allowsDryPhaseChanges() records this opt-in in the immutable, serializable result. The existing replayNativeFixedCell and runFixedCellDepletion retain their unchanged-family contracts.

At each pressure, flash the retained sample and remove equilibrium vapor to restore the first occupied volume. The remaining liquids stay in the cell; the next endpoint starts from those retained component amounts. A phase may appear or disappear during this withdrawal. Every accepted local equilibrium still passes the same bounded refinement, component, Gibbs and volume checks. There is no invented liquid composition when oil is absent. Water-containing samples, aqueous/other phase types, absent gas, critical-degenerate splits and unreachable volume restoration fail without mutating the input or publishing a partial path. This does not infer a wet upper dew or qualify near-critical gas condensates.

The following one-mole synthetic example is compiled and executed directly from this guide by DryFixedCellPhaseTransitionTest. The two checkpoints are separately labeled numerical controls, not an experimental EOS tune. The whole-experiment identity includes the initial-volume convention; separator volumes remain actual volumes at the final stated pressure and temperature.

import neqsim.pvtsimulation.util.CalibratedPvtModel;
import neqsim.pvtsimulation.util.DepletionPvtHistoryGenerator;
import neqsim.pvtsimulation.util.PvtSeparatorPath;
import neqsim.thermo.system.SystemInterface;
import neqsim.thermo.system.SystemSrkEos;

/**
 * Executable synthetic fixed-cell laboratory example.
 *
 * @author Even Solbraa
 * @version 1.0
 */
public class DryFixedCellGuideExample {
  /**
   * Runs one declared experiment from immutable original/tuned checkpoints.
   *
   * @return native retained states and separately attributed separator products
   */
  public static DepletionPvtHistoryGenerator.NativeResult run() {
    SystemInterface sample = new SystemSrkEos(275, 100);
    sample.addComponent("methane", .95);
    sample.addComponent("n-butane", .05);
    sample.setMixingRule("classic");
    sample.setMultiPhaseCheck(true);
    sample.useVolumeCorrection(false);
    sample.init(0);
    CalibratedPvtModel model = new CalibratedPvtModel(sample, sample.clone(),
        "synthetic dry sample", "original SRK", "separate identical consistency control",
        "synthetic laboratory protocol; no field fit", new String[] {"whole-experiment-control"});
    PvtSeparatorPath separator = new PvtSeparatorPath(new double[] {1.5, 1.01325},
        new double[] {303.15, 298.15}, "final actual product volumes at1.01325bara298.15K");
    return DepletionPvtHistoryGenerator.replayNativeDryPhaseChangingFixedCell(model, false,
        "synthetic dry CVD; initial100bara occupied volume; public numerical control",
        275, new double[] {100, 60, 50, 30, 20, 5, 2}, separator);
  }
}

For this example, the independently predicted endpoint phase sequence is gas at 100 bara, gas/oil at 60 and 50 bara, then gas at 30/20/5/2 bara. Every endpoint has the same cell volume and named initial = cumulative withdrawn + retained amounts. Retained and withdrawn separator products are available separately, with no withdrawn product at the initial state. These are laboratory protocol predictions, not independently predicted field production or recoverable reserves.

The complete independent reference covers 27 experiments / 243 states, 272.5/275/277.5 K, methane mole fractions 0.94/0.95/0.96, coarse/refined schedules and initially gas/oil cells. Tests use 0.001/1/1000 mol and both frozen checkpoint selections. Cubic roots, binary common tangents, lever rules and endpoint vapor removal are solved without NeqSim. A bounded binary tangent-plane search checks the predicted endpoints. This is independent numerical agreement using declared common EOS inputs; it is neither independent measured accuracy nor a general stability proof. Changing two-phase pressure steps changes the experiment and its component withdrawals; each schedule has its own independent prediction. Refining only an already gas-only tail preserves the final inventory and composition. Laboratory calibration/whole-experiment holdouts and reservoir calibration/held-out pressures, GOR and compositions remain separate acceptance requirements in #4740.

In the qualified binary experiments, liquid amount is unchanged during withdrawal at a fixed pressure because the coexistence compositions form a fixed tie line. The tested topology changes occur between pressure endpoints. Accuracy for a multicomponent topology change during withdrawal requires separate qualification; the endpoint checks alone do not establish it.