[PH-Flash] Add mixture enthalpy models with cited caloric data and baseline flash tests by arturcastiel · Pull Request #5297 · OPM/opm-common

arturcastiel · 2026-08-17T12:11:40Z

The compositional path in opm-common cannot evaluate the enthalpy of a flashed mixture. This PR adds that capability — the groundwork for an isenthalpic (P–H) compositional flash, which follows in a separate PR.

What this adds

ComponentCp — cubic ideal-gas heat-capacity polynomial and its closed-form enthalpy integral. The coefficients live on the component classes (C1, C10, SimpleCO2) as cited identity cards, each fitted to the species' reference EoS (Setzmann & Wagner 1991; Lemmon & Span 2006; Span & Wagner 1996) over 250–600 K, with cross-checked triple points.
IdealGasCaloricData — the shared enthalpy datum (H = 0 at 298.15 K) and a name-keyed lookup of the cards. Unknown components throw instead of silently receiving another species' data.
MixtureEnthalpy — molar enthalpy of a flashed state under two models: caloric (ideal-gas; independent of the phase split by material balance) and eos_departure (adds the per-phase residual H_res = -R T² Σ w_i ∂lnφ_i/∂T). The temperature derivative is computed by automatic differentiation: temperature is seeded as a one-variable DenseAd::Evaluation and ∂lnφ_i/∂T is read from the fluid system's own fugacityCoefficient — no hand-derived EoS calculus, so every EoS variant the fluid system supports is covered. The AD is internal to this evaluation; results are plain values.
Baseline tests for the existing isothermal flash on a new synthetic two-component (C1/nC10) fixture: phase labels, material balance, equal fugacity, K-consistency, and ssi/newton/ssi+newton parity.

Testing

About 65% of the diff is tests:

every heat-capacity card is pinned against tabulated reference-EoS values, so a corrupt coefficient row fails loudly instead of passing self-consistent checks;
the AD derivative is verified against an independent finite difference of lnφ, per phase and per component (the seam phaseDLnPhiDT exists for exactly this);
ideal-gas limit of the residual, split-cancellation and monotonicity of the caloric model, the reference datum, and the model-name parser.

Additive only — no existing behavior changes.

Introduce a test-local two-component (methane/n-decane) Peng-Robinson fluid system and a data-driven case harness (FlashCase/runFlash) that applies the full PTFlash input contract, and pin the isothermal flash on it: single-phase labels, two-phase invariants (material balance, equal fugacity, K-consistency) and iteration-method parity (ssi, newton, ssi+newton). A ternary case reuses the in-tree ThreeComponentFluidSystem.

ComponentCp holds the species-blind caloric equations: a cubic ideal-gas heat-capacity polynomial and its closed-form enthalpy integral, zero at the reference datum. The coefficients live on the component classes (C1, C10, SimpleCO2) next to the species' other constants, each fitted to the ideal-gas part of the cited reference EoS over 250-600 K, with CoolProp-cross-checked triple points added alongside. A pin suite guards every card against tabulated reference-EoS values, so a corrupt coefficient row fails loudly instead of passing self-consistent tests.

…arture IdealGasCaloricData owns the shared enthalpy reference datum and the deck-style name lookup over the component caloric cards; unknown components fail loudly rather than receive somebody else's heat capacity. MixtureEnthalpy evaluates the molar enthalpy of a flashed compositional state behind a two-model seam: the caloric (ideal-gas) model, where the phase split cancels exactly by material balance, and the EoS-consistent departure model, where the residual enthalpy is obtained per phase from AD temperature-derivatives of the fluid system's own fugacity coefficients — no hand-derived EoS calculus. Tests cover monotonicity, the reference datum, analytic cp against a finite difference, split-cancellation, the AD derivative against a finite difference of ln(phi), the ideal-gas limit, split coupling of the departure model, and the shared model-name parser.

arturcastiel · 2026-08-17T12:13:58Z

jenkins build this please

akva2 · 2026-08-17T12:41:58Z

jenkins build this please

akva2 · 2026-08-18T04:46:32Z

jenkins build this please

akva2 · 2026-08-18T06:05:56Z

jenkins build this please

arturcastiel · 2026-08-18T06:43:03Z

@akva2 thank you.

Copilot

Pull request overview

Adds compositional mixture-enthalpy evaluation as groundwork for P–H flash support.

Changes:

Adds ideal-gas heat-capacity data and enthalpy integrals.
Implements caloric and EoS-departure mixture enthalpy models.
Adds binary flash fixtures and baseline tests.

Reviewed changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
`CMakeLists_files.cmake`	Registers headers and tests.
`opm/material/components/C1.hpp`	Adds methane caloric and triple-point data.
`opm/material/components/C10.hpp`	Adds decane caloric and triple-point data.
`opm/material/components/ComponentCp.hpp`	Defines heat-capacity polynomial integration.
`opm/material/components/SimpleCO2.hpp`	Adds CO₂ caloric data.
`opm/material/constraintsolvers/IdealGasCaloricData.hpp`	Provides caloric presets and name lookup.
`opm/material/constraintsolvers/MixtureEnthalpy.hpp`	Implements mixture enthalpy models.
`tests/material/flashTestFixtures.hpp`	Adds shared compositional flash fixtures.
`tests/material/test_mixture_enthalpy.cpp`	Tests caloric and departure behavior.
`tests/material/test_twocomponents_ptflash.cpp`	Adds P–T flash baseline tests.

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

arturcastiel · 2026-08-20T07:44:09Z

+     * there is deliberately NO silent fallback — an unknown component must
+     * fail loudly rather than receive somebody else's heat capacity.
+     */
+    static ComponentCp<Scalar> byName(const std::string_view name)


Added ByNameAliasesAndUnknownName to test_mixture_enthalpy.cpp: every documented alias is pinned coefficient-wise against its card (mixed-case spellings included), and unknown/empty names are checked to throw. In 2e64598.

arturcastiel · 2026-08-20T07:44:12Z

+            FluidSystemF1::ParameterCache<double> paramCache(EOSType::PR);
+            paramCache.updatePhase(fs, phaseIdx);
+            return std::log(FluidSystemF1::fugacityCoefficient(fs, paramCache, phaseIdx, compIdx));
+        };
+
+        const double T = Opm::FlashTest::f1Temperature;
+        for (int compIdx = 0; compIdx < numComponentsF1; ++compIdx) {
+            const double ad = EnthalpyF1::phaseDLnPhiDT(outcome.state, phaseIdx, compIdx, EOSType::PR);


Done — the AD-vs-FD check now loops over PR, PRCORR, SRK and RK, passing the EoS type through both the scalar finite-difference path and phaseDLnPhiDT. The flashed composition is only a probe point (the AD == FD identity holds at any state), so the one state serves every EoS. In 2e64598.

arturcastiel · 2026-08-20T07:44:11Z

+    //! C1/nC10 Peng-Robinson binary interaction parameter — the value this
+    //! library's compositional tests use (cf. ThreeComponentFluidSystem). The
+    //! unequal-index shortcut in interactionCoefficient() is valid only
+    //! because the system is binary.


You're right — interactionCoefficient() there returns 0.0 for every pair and cannot establish this value. Reworded: 0.0411 is a pinned fixture value within the range of published C1/nC10 PR coefficients (roughly 0.04-0.05, source-dependent); the two-phase baselines built on it were cross-validated against an independent PR implementation (CoolProp) using this same value. Value unchanged. In 55a10ee.

…alpy tests Review follow-up on the mixture-enthalpy tests. byName() promises case-insensitive deck-style aliases and a loud throw on unknown components, but neither promise was exercised; and the AD-vs-finite- difference derivative check ran only under PR while the alpha-function temperature dependence branches per EoS type. Pin every documented alias coefficient-wise plus the unknown-name throw, and run the derivative identity over PR, PRCORR, SRK and RK — the flashed composition is only a probe point, so the one state serves every EoS.

The comment pointed at ThreeComponentFluidSystem::interactionCoefficient(), which returns zero for every pair and so cannot establish 0.0411. State the actual provenance instead: a pinned fixture value within the range of published C1/nC10 PR coefficients, cross-validated against an independent PR implementation with this same value. Value unchanged.

arturcastiel · 2026-08-20T07:44:14Z

Addressed all three review findings: alias + unknown-name coverage for the caloric-data lookup and the AD temperature-derivative check now running over all four supported EoS types (2e64598), plus a corrected provenance comment on the C1/nC10 interaction coefficient (55a10ee). Test-only changes.

arturcastiel · 2026-08-20T07:46:10Z

jenkins build this please

arturcastiel · 2026-08-20T08:06:36Z

@akva2 is there a reason why I cannot call jenkins?

akva2 · 2026-08-20T08:09:13Z

github is being troublesome and queues up my api calls and stalls for 30m-1h at times. it will likely trigger, it just takes time.

arturcastiel added 3 commits August 17, 2026 11:37

arturcastiel added the enhancement label Aug 17, 2026

arturcastiel marked this pull request as draft August 17, 2026 12:13

arturcastiel added manual:enhancement This is an enhancement/improvent that needs to be documented in the manual and removed enhancement labels Aug 17, 2026

arturcastiel marked this pull request as ready for review August 18, 2026 09:45

arturcastiel requested a review from GitPaean August 18, 2026 09:45

GitPaean requested a balanced review from Copilot August 19, 2026 07:34

Copilot started reviewing on behalf of GitPaean August 19, 2026 07:35 View session

Copilot AI reviewed Aug 19, 2026

View reviewed changes

arturcastiel added 2 commits August 20, 2026 07:43

Navigation Menu

Uh oh!

[PH-Flash] Add mixture enthalpy models with cited caloric data and baseline flash tests - #5297

[PH-Flash] Add mixture enthalpy models with cited caloric data and baseline flash tests#5297
arturcastiel wants to merge 5 commits into
OPM:masterfrom
arturcastiel:mixture-enthalpy-foundation

arturcastiel commented Aug 17, 2026

Uh oh!

arturcastiel commented Aug 17, 2026

Uh oh!

akva2 commented Aug 17, 2026

Uh oh!

akva2 commented Aug 18, 2026

Uh oh!

akva2 commented Aug 18, 2026

Uh oh!

arturcastiel commented Aug 18, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

arturcastiel Aug 20, 2026

Uh oh!

arturcastiel Aug 20, 2026

Uh oh!

arturcastiel Aug 20, 2026

Uh oh!

arturcastiel commented Aug 20, 2026

Uh oh!

arturcastiel commented Aug 20, 2026

Uh oh!

arturcastiel commented Aug 20, 2026

Uh oh!

akva2 commented Aug 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

arturcastiel commented Aug 17, 2026

What this adds

Testing

Uh oh!

arturcastiel commented Aug 17, 2026

Uh oh!

akva2 commented Aug 17, 2026

Uh oh!

akva2 commented Aug 18, 2026

Uh oh!

akva2 commented Aug 18, 2026

Uh oh!

arturcastiel commented Aug 18, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

arturcastiel Aug 20, 2026

Choose a reason for hiding this comment

Uh oh!

arturcastiel Aug 20, 2026

Choose a reason for hiding this comment

Uh oh!

arturcastiel Aug 20, 2026

Choose a reason for hiding this comment

Uh oh!

arturcastiel commented Aug 20, 2026

Uh oh!

arturcastiel commented Aug 20, 2026

Uh oh!

arturcastiel commented Aug 20, 2026

Uh oh!

akva2 commented Aug 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants