NeqSim can use chemical-reaction equilibrium together with electrolyte EOS models such as Electrolyte-CPA and electrolyte GE models such as Pitzer. Electrolyte EOS models select the STANDARD mole-fraction reaction-data source, SystemPitzer selects the dedicated PITZER molality-standard-state source, and SystemKentEisenberg selects the dedicated KENT_EISENBERG apparent-constant source.

A shared reaction name or stoichiometry is not evidence that its equilibrium constants, activity convention, or validity range transfer between thermodynamic models. Capture the actual initialized state with ChemicalReactionModelAudit before changing model-specific reactions.

SystemInterface cpa = new SystemElectrolyteCPAstatoil(298.15, 1.01325);
cpa.addComponent("CO2", 0.01);
cpa.addComponent("water", 0.99);
cpa.chemicalReactionInit();

SystemInterface pitzer = new SystemPitzer(298.15, 1.01325);
pitzer.addComponent("CO2", 0.01);
pitzer.addComponent("water", 0.99);
pitzer.chemicalReactionInit();

ChemicalReactionModelAudit.AuditComparison comparison =
    ChemicalReactionModelAudit.compare(cpa, pitzer);

The audit reports the selected typed data source, reaction-quotient concentration basis, deterministic active-reaction list, stoichiometry, stored literature/data reference, reference temperature, equilibrium-constant coefficients, and the validation status declared by that model-specific data row. The comparison separates concentration conventions, reactions present in only one model, and common reactions whose stored parameters or validation declarations differ.

Use audit.getReactionsWithoutValidatedEvidence() to list active rows that are not marked VALIDATED, audit.hasValidatedEvidenceForAllActiveReactions() for a non-throwing check, and audit.requireValidatedEvidenceForAllActiveReactions() for an explicit fail-closed publication gate. Use comparison.requireEquivalent() when a workflow requires two initialized systems to have the same source, concentration basis, active reaction set, stoichiometry, and stored parameters; its exception reports every differing category. Individual reaction snapshots and ChemicalReaction objects expose getValidationStatus(). The status qualifies only the stored equilibrium-constant correlation for the selected source, standard state, and documented range; it does not prove complete finite-salinity speciation, phase equilibrium, or process performance.

The API is deliberately read-only. It requires chemicalReactionInit() to have been called and never initializes reactions implicitly, runs a flash, or changes model state. It therefore adds no work to ordinary neutral PR/SRK/CPA calculations and no work to electrolyte calculations unless the audit is explicitly requested.

For the typed PITZER source, validation is also an initialization contract. After irrelevant and dependent reactions have been removed, chemicalReactionInit() rejects every remaining active reaction row whose status is not VALIDATED, before that row can add its model-specific product species. The current MDEAprot (Huttenhuis2005) and DEAprot (Austgen1989) compatibility rows remain marked USEREACTION=1 and UNVALIDATED; a SystemPitzer feed containing the corresponding neutral amine therefore fails closed instead of silently using an unqualified molality-standard-state correlation. No reaction coefficient or activation flag is changed. STANDARD and KENT_EISENBERG retain their legacy initialization behavior because their distinct standard states require separate qualification.

Pitzer reaction concentration basis

SystemPitzer evaluates solute activities from molality, defined as moles of solute per kilogram of water solvent, while solvent activity retains its mole-fraction convention. This matches the concentration basis of the Pitzer ion-interaction equations. The PITZER source supplies molality-standard-state correlations for CO2/HCO3-, HCO3-/CO3–, and water dissociation; Electrolyte-CPA retains the legacy STANDARD source and its mole-fraction reaction convention. Pitzer’s thermodynamic framework is documented in DOI 10.1021/j100621a026 and the binary-electrolyte formulation in DOI 10.1021/j100638a009.

Carbonate parameter evidence and validity

The three Pitzer correlations reproduce the analytical expressions distributed in the public-domain USGS PHREEQC 3 phreeqc.dat database. The CO2 and carbonate expressions trace to Plummer and Busenberg (1982), DOI 10.1016/0016-7037(82)90056-4; PHREEQC 3 is documented by Parkhurst and Appelo (2013), DOI 10.3133/tm6A43. The authoritative software page marks the source and usage as public domain: https://www.usgs.gov/software/phreeqc-version-3.

NeqSim fits its existing four-coefficient natural-log form, ln(K) = K1 + K2/T + K3 ln(T) + K4 T, to the PHREEQC base-10 analytical expressions. Training temperatures are 0, 10, …, 90 degC. Held-out validation temperatures are 5, 15, …, 85 degC. Maximum held-out absolute errors are 0.000471 log10 units for CO2 dissociation, 0.000157 for bicarbonate dissociation, and 0.000192 for water dissociation; RMS errors are 0.000311, 0.000104, and 0.000127 log10 units, respectively. The declared validity range is 0-90 degC at the infinite-dilution thermodynamic standard state; Pitzer activity coefficients provide the finite-ionic-strength correction.

No Pitzer binary interaction parameter was fitted or changed. REACTIONDATAPITZER.csv marks CO2water, carbonate, waterreac, and the first H2S dissociation water-H2S as VALIDATED. Other compatibility rows remain UNVALIDATED; their existing activation flags are unchanged except that unsupported water-HS is disabled. The declaration is diagnostic; the USEREACTION field separately controls the model-specific active set.

H2S first-dissociation evidence and active-set boundary

For SystemPitzer, water-H2S represents H2S + H2O = HS- + H3O+ on the solute-molality and pure-water activity standard states. Its natural-log correlation is an exact basis conversion of the public-domain PHREEQC 3.9.0-17591 pitzer.dat expression at commit b0b3be7:

log10(K1) = 11.17 - 3279/T - 0.02386 T, with T in kelvin.

The stored NeqSim coefficients are [25.719875488743, -7550.176519927477, 0, -0.054939680319] in ln(K) = K1 + K2/T + K3 ln(T) + K4 T. The mapping is algebraic rather than fitted. It is checked at 5, 15, …, 85 degC against the PHREEQC expression. Independent validation uses Hershey, Pleše, and Millero (1988), DOI 10.1016/0016-7037(88)90183-4, at their experimental temperatures 5, 25, and 45 degC. The maximum absolute difference from their infinite-dilution pK1 correlation is 0.071 pK unit, within a conservative 0.08 gate; no experimental table is copied.

The second dissociation water-HS is inactive for Pitzer. The audited PHREEQC dataset defines no S--/S-2 aqueous species or second-dissociation equilibrium, and the first-dissociation paper reports only pK1. Retaining the legacy unvalidated pK2 as active would create unsupported sulfide speciation. Electrolyte-EOS and Kent-Eisenberg tables remain unchanged because their standard states and validation questions are distinct.

This qualification covers only the aqueous first-dissociation constant from 5 to 45 degC at the infinite-dilution standard state, with the PHREEQC curve checked through 85 degC. It does not qualify H2S gas-liquid equilibrium, pressure corrections, neutral-H2S Pitzer interactions, concentrated-brine speciation, oil partitioning, amine absorption, or the second dissociation.

The Newton chemical-equilibrium solver reads the same typed concentration-basis selection as the reaction diagnostics. For SOLUTE_MOLALITY, it forms solute chemical potentials from ln(moles/kg solvent) + ln(gamma) and retains ln(x) + ln(gamma) for solvent activity. Solvent mass is calculated once per iteration, so this branch adds no work to neutral calculations and no inner-loop component scan. Legacy mole-fraction and Deshmukh-Mather paths retain their existing expressions. A focused H2S-water regression requires maximum absolute ln(Q/K) below 2e-6, maximum element residual below 1e-8 mol, absolute charge below 1e-8 mol, normalized charge below 2e-6, normalized non-negative phase fractions, deterministic repeated execution, and the expected 25-to-45 degC ionization trend.

Legacy STANDARD and KENT_EISENBERG rows currently return UNSPECIFIED. That value means the table has not yet adopted the explicit declaration; it must not be interpreted as evidence that a row is invalid.

Scientific use

Use this diagnostic to freeze model-specific validation questions before changing reaction data:

  1. Compare EOS and GE systems with the same feed species and reaction-init lifecycle.
  2. Record differences in active reactions and stored parameter provenance.
  3. Check the activity/standard-state convention used by each thermodynamic model.
  4. Validate equilibrium constants and speciation against independent public data over a declared temperature, pressure, ionic-strength, and composition range.
  5. Split reaction tables or activate/deactivate model-specific reactions only when the scientific evidence requires it.

Relevant foundations include Plummer and Busenberg (1982), DOI 10.1016/0016-7037(82)90056-4, for carbonate equilibria and the Pitzer electrolyte framework for activity-coefficient treatment. NeqSim does not infer model-specific parameter validity merely because two sources retain the same reaction name.