NeqSim 3.28.0

Status: prepared release candidate; publication pending. Prepared on 2026-11-03 from master 047cb7e4c59101404bcc3aa8f01761630ca1d87f, covering 75 commits after v3.27.0. Only changes in that merged snapshot are included. The preparation PR does not create a release tag or publish artifacts.

Complete change list for this snapshot.

Review before upgrading

Numerical behavior and saved models

Rebaseline representative engineering cases against measurements, material and energy balances. Saturation/PS roots, equilibrium sensitivities, CVD cell-volume restoration, wet laboratory depletion, compressor chart power, TEG absorption and transient pipeline energy/phase transport have changed. Passing a software test does not qualify all fluids or operating ranges.

• Pressure-flash roots now require fugacity closure; PS recovery and bracketed saturation paths have corrected root handling and reduced repeated work (#4784, #4831). Compare phase identity, equilibrium residuals, enthalpy/entropy and calculated boundary conditions.
• Equilibrium sensitivities and batched property gradients have corrections (#4786). Recheck downstream optimization and uncertainty calculations that consume derivatives.
• Compressor chart power is local to each run (#4775); rerun reused equipment objects and compare power, head and pressure after input changes.
• Horner buildup permeability, skin and extrapolated pressure are corrected (#4829, #4835). Reassess existing fitted well-test results rather than carrying old derived values forward.
• SimpleTEGAbsorber handles hydrocarbon liquid, with additional CPA TEG/MEG coverage (#4840). Qualify the actual feed phases, carryover and dehydration duty for each application.

Load representative prior-release .neqsim files and verify retained equipment state, controllers, component inventories, units and rerun results. Added fields and same-version round trips alone do not prove cross-version persistence. Public API and saved-model compatibility review remains a release gate; this preparation does not claim universal backward compatibility. Any incompatible contract must follow governance and the API lifecycle policy before publication.

Retain the 3.27.0 migration notes and, when upgrading from an earlier release, the 3.26.0 migration notes, including the JAMA-to-EJML interface migration and E300 critical-volume units.

Runtime and publishing toolchain

The standard core artifact requires Java 17+, the separate Java 8 artifact requires Java 8+, and the MCP runner requires Java 21+. Core and test source remains Java 8 compatible. Repository tooling, reports and notebooks require Python 3.11+, separately from the Java runtime requirements.

Restore the Maven wrapper to 3.9.16, the publishing baseline used for 3.27.0. #4811 reintroduced Maven 3.10.0 while the Java 8 release profile still uses central-publishing 0.11.0. #4318 records the local-metadata publication-bundle issue with that combination. Retain standard-profile central-publishing 0.12.0 and both profiles’ waitUntil=published settings; qualify both bundles before advancing the wrapper. Other merged dependency updates, including Gson, Enforcer and JaCoCo, remain in this candidate.

Main changes since 3.27.0

DEXPI 2.0.1, PFDs and instrumented P&IDs

Native DEXPI 2.0.1 packages now connect tagged process models, instrumentation and safeguarding proposals, physical projections, drawing registers and steady-state/dynamic simulation bindings (#4836, #4839, #4845). Native PFD and instrumented P&ID review drawings have reconciled topology, auxiliary line identities, physical sensing connections and valve symbols (#4848).

The P&ID generation guide and quality review workflow describe the required evidence. The reviewer checks actual package hashes, native references, registers, drawing tag coverage and layout, then retains separate visual and controlled-standards review records. Regeneration invalidates earlier receipts. Missing design inputs or reviews remain HOLD; inconsistencies remain FAIL.

PASS_FOR_ENGINEERING_REVIEW is a bounded review disposition. It does not assign ISO/IEC/ISA conformance, CAE receiving-tool qualification, construction fitness, SIL or operating approval. Use project-controlled standard editions and actual visual/physical review. A generated drawing or DEXPI export alone is insufficient.

The professional process-overview workflow also adds A3/slide output and repairs small-flow preservation and slide validation (#4777, #4796). Physical isolation-inventory evidence is retained in layout studies (#4790).

Laboratory PVT and reservoir handoff

Native depletion paths preserve actual phases and inventories through phase appearance, closed-mass CCE and wet DLE. CVD upper-dew reference and post-liquid disappearance volume restoration have corrections. Explicit fixed-cell depletion adds transactional execution and refined equilibrium/volume precision (#4781, #4792, #4797, #4806, #4819, #4824, #4832, #4843).

Follow the native PVT handoff guide for path-specific assumptions. Closed-mass CCE, gas-withdrawal CVD, changing-volume DLE and explicit initial-cell volume are distinct laboratory protocols. Absent phases do not acquire invented formation factors. Synthetic analytical checks are not measured laboratory holdouts or a qualified laboratory-to-field chain.

Reservoir work adds phase-selective production and separator outputs, and source-identified bounds on laboratory gas-density handoff (#4771, #4776).

Pipelines and Safety Shutdown: experimental limits remain

Two-fluid updates expose phase-resolved boundary ledgers, improve interior hydrostatic pressure and signed work, couple stratified pressure moments and cross-sectional gravity energy, conserve total energy in the optional Rusanov flux, retain bidirectional phase-face transport, add a gas-only pressure outlet and qualify bounded conservative hydrocarbon pool transfer (#4769, #4768, #4778, #4782, #4791, #4798, #4809, #4818, #4821, #4838).

Review the two-fluid evidence matrix. These bounded repairs do not establish complete rich-gas blowdown, terrain pooling and same-gas repressurization/evaporation qualification. Safety Shutdown acoustics retain their documented fluid, wave-speed, pump and boundary limits; no general multiphase hydraulic-hammer qualification is claimed.

Release-source studies add external heat coupling and finite-rate retained phase inventories (#4807, #4820); consult the coupled release-inventory guide.

Controls, optimization and reusable engineering evidence

Sampled temperature and physical flow transmitter scans gain dynamic qualification, alongside TEG restart/sample-system dynamics, oscillation analysis, compressor load limiting and time-series process replay (#4800, #4825, #4834, #4840, #4841).

Engineering evidence gains revision-controlled scenarios, map-to-driver migration records/timelines linked to lifecycle replay, optimizer termination reporting, discipline-report manifests and a runnable field-acceptance workflow (#4766, #4780, #4789, #4799, #4803, #4804, #4817, #4827, #4837). Evidence remains specific to the executed model, revision, inputs and acceptance criteria.

Huron–Vidal source-root checks, selected-pair cost evidence and retained CO2-MEG and nitrogen-water primary-source receipts advance qualification without turning quarantined or held evaluator proposals into accepted model parameters (#4772, #4783, #4805, #4823, #4842). Structure-H hydrate data retain the 3.27.0 limits; no new executable sH support is claimed. Documentation gains executable examples and corrected API navigation, including literal physical-property examples (#4774, #4844, #4846).

Planned distributions

Distribution	Version or expected asset	Runtime
Core Maven artifact	com.equinor.neqsim:neqsim:3.28.0	Java 17+
Core standalone library	neqsim-3.28.0.jar	Java 17+
Java 8 Maven artifact	com.equinor.neqsim:neqsim:3.28.0-Java8	Java 8+
Java 8 standalone library	neqsim-3.28.0-Java8.jar	Java 8+
MCP runner and checksum	neqsim-mcp-server-3.28.0-runner.jar and .jar.sha256	Java 21+
MCP container	ghcr.io/equinor/neqsim-mcp-server:3.28.0	Container runtime

The MCP module retains its own 1.0.0-SNAPSHOT Maven identity; its core revision/neqsim.version, runner filename and container tag use 3.28.0. Python and plugin-marketplace packages have separate release processes.

Release completion requirements

1. Review the final candidate, API and saved-model compatibility, migration notes and any contract decisions required by governance.
2. Require the configured CI and security gates on the final head: core and slow tests, model specifications, formatting, documentation/Javadoc and packaged Java 21 MCP protocol qualification. Parent results do not qualify this head.
3. Build and inspect Java 17/Java 8 libraries, sources/Javadoc and the MCP runner. Verify embedded versions, class-file targets, pinned Java 8 dependencies, actual supported JVM behavior, checksums and both publication bundles.
4. Merge the reviewed version PR to let the existing default-branch workflow build a draft release. Include these notes with the generated change list.
5. After release-owner approval, publish and verify both Maven Central versions reach PUBLISHED, the GitHub assets/checksums are downloadable and the MCP container exists. Record the final tag, commit and workflow evidence.

The preparation PR records actual local/hosted validation and pending gates. The expected distributions above are not asserted to be published or available.