Decision and scope

Full-Pitzer barite precipitation in Na–Ba–Cl–SO4 water remains unqualified. This report implements the actionable-report alternative in issue #4351, coordinated under #3144. It does not claim completion of the issue’s independent equilibrium-validation acceptance criteria.

The audit starts from NeqSim 3.25.0 master 459b9eccf22f335ce88a6c3ec85338844b5a95c4. The reported exception is a correct refusal of an unsupported topology, not evidence that the TP-flash algorithm should be bypassed. No binary coefficient, theta, psi, equilibrium constant, solver tolerance, legacy opt-out, or scientific qualification level is changed here. The standalone PITZER_BINARY screening calculation is not full mixed-electrolyte qualification.

Two different parameter inventories

Default selection first considers the pinned PHREEQC catalog. If any required row is absent, it retains the legacy dataset; the latter’s mixed-ion coverage check then refuses calculation. Consequently the exception’s six missing legacy interactions are not the list of missing PHREEQC source rows. Combining rows from these two datasets without a coherent convention and validation record would hide that distinction.

The candidate source is PHREEQC commit b0b3be767158ccc3322d2c816625cf470045e67e, database blob 324f852784be84650b77bd7f07f8316aafd8188b, represented by the bundled resource under src/main/resources/neqsim/thermo/phase/phreeqc/. Its required-row inventory is:

Interaction Pinned PHREEQC source status Required action
Na+/Cl-, Na+/SO4–, Ba++/Cl- Explicit B0, B1 and C0 rows exist Retain the complete source temperature functions and charge normalization
Ba++/SO4– No explicit B0, B1 or C0 row Select and review a coherent binary/association convention; establish whether beta2 and nondefault alpha are required
theta Ba++/Na+ Explicit constant 0.07 Source coverage only; retain its associated binary/model convention
theta Cl-/SO4– Explicit constant 0.03 Source coverage only
psi Na+/Cl-/SO4– Explicit constant zero Present source row; distinguish it from an absent row
psi Ba++/Na+/Cl- Absent Obtain a traceable coefficient or explicitly justified model convention
psi Ba++/Na+/SO4– Absent Obtain a traceable coefficient or explicitly justified model convention
psi Ba++/Cl-/SO4– Absent Obtain a traceable coefficient or explicitly justified model convention

There are six missing required catalog rows: three binary-family rows and three psi rows. The legacy table separately contains a zero-valued Ba++/SO4– binary labelled Monnin1999, but lacks both theta and all four psi interactions. A source label alone does not establish that its binary convention can be combined with this PHREEQC subset. No such combination is adopted.

Read-only preflight API

After constructing a SystemPitzer, adding the complete feed, setting mixing rule classic, and calling init(0), use getMissingPhreeqcPitzerCatalogRows(). The corresponding phase API is PitzerParameterDatasets.getMissingPhreeqcPitzerCatalogRows(phase). It returns an immutable sorted list with family and alphabetically sorted species names. For the issue feed it reports:

This traverses the same requirements as automatic catalog selection, including neutral lambda/zeta rows, and excludes EOS-role hydrocarbons by the same rule. It neither initializes properties nor changes the selected dataset or user parameters. Reinitialize the system with init(0) after a composition change. The audit follows the selector’s active-phase-amount threshold of 1e-20 mol; it is not a new molality-based applicability rule. It audits the aqueous role, not arbitrary feed species that may later be introduced by chemistry.

An empty list means only that the required source rows are present. It does not certify the model, select a dataset, establish a valid range, or include optional beta2/alpha and higher-order neutral-family qualification. Use getPitzerParameterQualification() for selected-dataset evidence and requirePitzerDatasetValidationFor(MINERAL_SATURATION_AND_PRECIPITATION) for the separate publication gate. Existing fail-fast selection and property calculations retain their behavior; the all-row diagnostic is evaluated only when explicitly requested.

Frozen public feed matrix

The machine-readable fixture is barite-qualification-cases.csv. All seven feeds use 1 kg water, 298.15 K and 1.01325 bara. Concentrations below are mol/kg water; water moles are calculated from the runtime water molar mass, not rounded to 55.508 mol. The exact original issue feed with 55.508 mol water is tested separately.

Case Na+ Ba++ Cl- SO4– Initial ionic strength, mol/kg water
very_dilute 0.001002 0.000001 0.001002 0.000001 0.001006
dilute 0.102 0.001 0.102 0.001 0.106
issue_molality_basis 1.0 0.001 1.0 0.001 1.004
brine_3m 3.002 0.001 3.002 0.001 3.006
brine_5m 5.002 0.001 5.002 0.001 5.006
barium_excess 1.0004 0.002 1.004 0.0002 1.0066
sulfate_excess 1.004 0.0002 1.0004 0.002 1.0066

Every feed satisfies $m_{Na}+2m_{Ba}-m_{Cl}-2m_{SO4}=0$. Initial ionic strength is $I=\frac12\sum_i m_i z_i^2$ for these fully dissociated inputs. These are qualification candidates and fail-closed regressions, not a validated operating range. No pressure correction or high-temperature barite evidence is established by this matrix.

Use the dimensionless molality activity $a_i=\gamma_i m_i/m^\circ$, where $m^\circ=1\,\mathrm{mol\,kg^{-1}}$ and gamma is referenced to infinite dilution. For the dissolution reaction $\mathrm{BaSO_4(s)}\rightleftharpoons\mathrm{Ba^{2+}}+\mathrm{SO_4^{2-}}$, $SI=\log_{10}(a_{Ba}a_{SO4}/K_{sp})$ uses unit activity for pure barite. Total dissolved barium or sulfate must not be substituted for free-ion activity when complexes are present. Freeze treatment of BaSO4(aq), NaSO4-, HSO4-, acid/base chemistry and the solid standard state before comparing models. Double-counting ion pairing in both association reactions and fitted Pitzer interactions is not acceptable.

Primary evidence and outstanding provenance

The USGS PHREEQC PITZER documentation defines the coefficient families, temperature expansion and nonsymmetric mixing convention. It warns that parameter changes must preserve thermodynamic consistency. Missing coefficients within an existing temperature row are distinct from an entirely absent interaction row. MacInnes scaling affects reported single-ion coefficients; compare neutral activity products or use an explicitly matched reporting convention.

Dideriksen et al. (2024), Applied Geochemistry 171, 106033 (author-hosted paper), sections 3.1.5–3.1.6, is a relevant candidate model source. It evaluates binary and mixing parameters against barite solubility, discusses Pitzer (1975), Monnin and Galinier (1988), and Paige et al. (1998), and supplies a combined database in its supplementary material. Its reported accuracy depends on temperature, salinity and experimental material. This is a research lead, not validation of NeqSim’s current coefficient combination. The supplementary database, redistribution terms, complete species/reaction conventions and every required row must be retrieved and reviewed before adoption; coefficients are not transcribed or fitted here.

Evidence status and completion procedure

Acceptance item Current evidence Remaining work
Public dilute-to-brine feeds Frozen seven-case CSV and exact original reproducer Retain unchanged when adding supported parameters
Explicit parameter topology Complete required-row audit above Review all six absent rows and optional 2–2 terms under one coherent model
Pre-precipitation SI Blocked by missing interactions Independent matched-model SI for every frozen feed
Equilibrium dissolved Ba/SO4 and solid mass Blocked; no numerical results claimed Compare free ions, total dissolved elements and precipitated BaSO4 with qualified independent evidence
Charge and inventory Balanced input and unchanged component inventory on repeated rejected calls Close elemental and charge balances after successful equilibrium
Nonnegative solids and complementarity Not demonstrated by rejection Require $n_s\ge0$, $SI\le0$ and $n_sSI=0$ within predeclared numerical tolerances
Deterministic repeat and dissolution Repeated diagnostics/rejections only Repeat successful precipitation, then dilute/add undersaturated water and demonstrate dissolution with retained solid inventory
Domain and uncertainty No qualified full-mixture T/P/composition envelope Report calibration versus held-out data, model residuals and measurement uncertainties separately

For the independent comparison, pin executable version, database hash, species mapping, solvent mass and pressure convention. First reproduce the chosen published model in IPhreeqc or another independently qualified implementation, then compare NeqSim with the same activity and solid standard states. Preserve raw SI, free/total concentrations and solid-mass results. A PHREEQC run that relies on omitted interactions is not evidence for filling NeqSim’s missing rows with zero. Use held-out measured solubility to assess accuracy separately from code-to-code agreement. Do not widen a tolerance to disguise an unreviewed parameter or speciation mismatch.

PitzerBariteQualificationTest verifies the current blockers and inventory contract. Existing catalog, neutral, mixed-electrolyte and qualification suites remain the regression controls. The issue should remain open until the independent equilibrium and continuation gates above are fulfilled. The report and preflight API can be reviewed and merged separately from parameter adoption.