[Campaign] TwoFluidPipe state-of-the-art R&D gap map

Purpose

Persistent, dependency-ordered gap map for advancing TwoFluidPipe toward a defensible state-of-the-art open implementation. This is a roadmap and evidence ledger, not a claim of OLGA or LedaFlow equivalence.

Every increment must start from current master, search open/recent issues and PRs, freeze a reproducible baseline, state equations/closures/units/validity/provenance, preserve conservation and phase limits, validate against public evidence, and remain one bounded reviewable change.

Baseline refreshed 2026-08-10 at 31d0d9009c3722cbcef7e47fe98802ff6fca0829. The eight commits since 66a866f1e2e82c2fbbd29a7af936e098cd29e849 touch no TwoFluidPipe source or tests; open draft #2913 explicitly defers the stiff dispersed-bubble limitation to #2909.

Prioritized gap map

P0 — foundational correctness and reusable architecture

Conservative gas/oil/water phase-resolved mass transfer and phase identity across appearance/disappearance (TwoFluidPipe flash condensation assigns first aqueous liquid to oil #2762 and merged follow-up evidence).
Closed-boundary, stage-consistent thermal face fluxes and all-cell cooldown (TwoFluidPipe closed cooldown uses inlet-stream flow and skips inlet-cell heat transfer #2792 / fix(twofluid): make closed cooldown boundary-consistent #2796).
Stage-pure, equal-and-opposite virtual-mass acceleration coupling (fix(twofluid): make virtual-mass coupling stage-pure and momentum-conservative #2891 / fix(twofluid): make virtual-mass coupling stage-pure #2892).
Dispersed-bubble interfacial drag must reproduce the declared Schiller-Naumann force without a spurious pipe-diameter scale and use a conservative, dissipative, stage-pure coupled PEA/local-implicit/IMEX treatment of the resulting stiff source (TwoFluidPipe: treat corrected dispersed-bubble drag as a stiff conservative source #2909). The active-phase momentum matrix must remain valid in two- and three-phase limits; the current regime-fraction gas-to-oil/water impulse split is not energy-dissipative for arbitrary oil/water slip. Draft fix(twofluid): correct dispersed-bubble drag force #2908 and simple timestep-dependent force scaling/default virtual mass were rejected by the public severe-slugging benchmark.
Replace the fixed 0.02 N/m bubble-size surface tension and single algebraic diameter assumption with an explicitly configurable, phase-property-aware closure before claiming broad bubbly-flow validity.
Audit every wall/interfacial/oil-water force for dimensions, action/reaction, phase-limit continuity, reverse-slip sign, serialisation, and repeated-RHS purity.
Freeze complete mass, momentum, and energy residual histories for accepted transient steps; fail loudly before advancing invalid states.

P1 — initialization, boundaries, and numerical robustness

Validate steady-to-transient handoff for gas, gas-liquid, and gas-oil-water cases with no artificial first-step impulse.
Add low/zero/reversed-flow boundary models and turndown/minimum-stable-flow tests.
Establish mesh/timestep refinement for pressure, holdup, and phase velocities at nearby operating points and regime transitions.
Characterize positivity, phase-degeneracy, hyperbolicity, CFL/IMEX stability, nonlinear convergence, deterministic repeats, and thread/copy/serialization behavior.
Separate numerical diffusion from any declared physical dispersion model; do not tune either against a single case.

P2 — public physical validation

Curate redistributable public steady pressure-drop/holdup datasets spanning horizontal and inclined stratified, slug, annular, and bubbly regimes.
Validate gas-liquid flow-regime transitions and holdup/pressure trends with uncertainty and documented correlation validity ranges.
Validate three-phase oil-water-gas coupling, oil-water slip, wetting, and terrain accumulation against public data.
Compare NeqSim outputs through the existing benchmark harness with legitimately exported OLGA/LedaFlow cases when available; never infer or copy proprietary closures.

P3 — advanced transients and flow assurance

Hydrodynamic slug initiation, growth, coalescence, translation, and outlet statistics.
Severe-slugging onset, cycle amplitude/period, riser-base pressure, and mitigation response.
Startup, ramp-up, shutdown, cooldown, restart, pigging/line packing, liquid loading, and phase-composition events.
Heat/mass transfer, condensation/evaporation, hydrate/wax-relevant thermal histories, and wall thermal inertia with complete energy accounting.
Minimum stable flow and turndown envelopes with hysteresis and operational uncertainty.

P4 — performance and engineering workflow

Profile only validated cases; improve allocations/parallelism without changing results, stage order, determinism, or thread safety.
Add executable examples only after supporting Java behavior is merged; notebooks must be cleanly executed, output-complete, rendered, and visually inspected.
Report applicability, unresolved limitations, convergence, conservation, mesh/timestep sensitivity, and source provenance in every engineering result.

Evidence context

Public product material is used only to prioritize engineering use cases:

OLGA core publicly describes transient three-phase oil/water/gas flow, dispersion, low-velocity slip, liquid accumulation, slugging, corrosion and hydrate-relevant behavior.
OLGA training includes startup, shutdown, pigging, liquid and hydrate management.
LedaFlow product sheet documents transient multiphase engineering applications.

Relevant open literature includes Enwald et al. (1996), doi:10.1016/S0301-9322(96)00004-X, on Eulerian two-fluid interfacial momentum transfer; Draw et al. (2024), doi:10.1016/j.ijmultiphaseflow.2024.104969, using Schiller-Naumann drag in Euler-Euler bubbly flow; and Hayashi et al. (2025), doi:10.1016/j.ijmultiphaseflow.2025.105173, illustrating the limits of rigid-sphere drag for bubble-interface conditions.

Per-increment ledger requirements

For each PR, record:

exact master and overlap search;
frozen baseline/failing case;
equations, SI units, assumptions, validity and parameter provenance;
independent analytical/correlation or public benchmark evidence;
conservation, physical trend, phase-limit and nearby-point checks;
focused plus proportionate broader tests;
public API, serialization/copy, determinism/thread-safety and performance impact;
documentation-impact statement;
exact-head CI/review/merge-readiness status.

Increment 2026-08-09 — dispersed-bubble drag dimensional correction

Draft PR #2908, exact head 1e4afd45.

Base/current master at publication: 635b9596bf0fa1e304e9282ba8d893a291dd03ca; one commit ahead, zero behind, four intended files.
Frozen failure: for D=0.05 m, alpha_L=0.80, rho_G/rho_L=5/1000 kg/m3, mu_L=0.001 Pa s, and u_r=-1 m/s, master gives -2.5918139392 N/m versus the independent Schiller-Naumann force -53.1625584072 N/m.
Root cause: f_i=C_D d_b/(4D) combined with a_i=6 alpha_G/d_b introduces a spurious d_b/D force factor. Required equivalent shear form is f_i=C_D/4.
Regression: 36 diameter/holdup/forward-reverse-slip points, force agreement to 1e-12 relative, exact zero at phase/slip limits, correct sign.
Local evidence: focused 2/2; TwoFluid package 352 passed with 17 existing skips; broader pipeline selection 139 passed; Java 8 source compilation, Spotless apply/check, documentation search, pre-commit and pre-push all passed.
Documentation impact handled in both TwoFluidPipe model guides; fixed 0.02 N/m bubble-size surface tension and D/5 diameter cap are explicitly retained limitations.
Hosted exact-head Spotless, pre-commit, PaperLab, documentation, CodeQL and Javadoc are green. Java 21 Ubuntu/Windows fast tests and four slow shards remain running. No reviews or review threads.

Do not mark the P0 checklist item complete until #2908 is merged and the exact merged master evidence is recorded.

P0 drag checkpoint: #2909 now records current-master baseline and two rejected numerical remedies. The unscaled Schiller–Naumann closure plus timestep-dependent backward-Euler force scaling passed isolated conservation/stiffness tests through dt/tau=10^3 but failed 3/6 Tengesdal assertions; enabling virtual mass by default also failed 3/6 and underpredicted pressure amplitude. No PR was opened and no benchmark bound was changed. The dependency-ready target is now a coupled PEA/local-implicit/IMEX momentum treatment with manufactured convergence, two-/three-phase conservation, and unchanged public severe-slugging acceptance criteria.

Gap-map checkpoint on master db1c04f681a56f51c3d96621e6b6979d1dc0f4af (2026-08-10):

No open related implementation PR was found for TwoFluidPipe bubble/interfacial drag, stiff-source integration, or partial elimination.
Foundational priority remains TwoFluidPipe: treat corrected dispersed-bubble drag as a stiff conservative source #2909: correct bubble/dispersed-bubble drag plus a conservative stiff momentum coupling. Two post-transport local backward-Euler variants (frozen and fully nonlinear Schiller–Naumann source) passed focused closure/conservation tests but each regressed 2/6 assertions in the public Tengesdal slow benchmark. Detailed results are recorded on TwoFluidPipe: treat corrected dispersed-bubble drag as a stiff conservative source #2909.
Verified benchmark baseline requires -DexcludedTestGroups= because the class is tagged slow; current master passes 6/6 with the override. Commands lacking it execute zero benchmark tests and must not be reported as benchmark evidence.
No branch or PR was published. Next bounded target is a stage-coupled drag operator / momentum solve that preserves TwoFluidPipe: treat corrected dispersed-bubble drag as a stiff conservative source #2909's algebraic guarantees while retaining the unchanged 6/6 Tengesdal envelope.
Other gap-map priorities are unchanged this run; no evidence supports advancing severe-slugging application work ahead of this foundational numerical coupling.

2026-08-11 active-increment checkpoint — draft #2933

The campaign now has one active implementation PR: #2933, exact head 027ea712.

Verified scope and evidence:

the default retains the legacy diameter-scaled response; TwoFluidPipe.setEnableStiffBubbleDrag(true) selects the corrected Schiller–Naumann force and a symmetric pair of local backward-Euler half-steps around transport;
the local active-phase solve uses a mass-proportional pseudo-liquid impulse, preserves oil-water slip, removes exactly absent phases, conserves active-phase momentum to roundoff, and enforces non-increasing kinetic energy;
the 36-point force regression and local stiffness/conservation tests cover dt/tau=10^-3 ... 10^3, sign reversal, zero slip, trace/absent phases, nonlinear refinement, and deterministic repeated calls;
the unchanged compatibility path passes the complete slow-tag Tengesdal suite 6/6; the corrected opt-in diagnostic passes 3/6 and remains outside the published experimental envelope (167.1 kPa minimum swing versus 98 +/- 5 kPa, slug-length ratio 1.164, and no repeated 16-section/0.1 s cycle). This is correctly documented as a remaining model limitation, not quantitative validation;
all six hosted workflows on the exact head completed successfully, the branch is mergeable, and the only review thread was technically assessed and resolved;
documentation impact is handled in both TwoFluid model guides, including closure assumptions, fixed 0.02 N/m bubble-size surface tension, D/5 cap, opt-in migration, and benchmark limitation.

The PR is not engineering-ready to merge yet:

current master is 275b188d; the PR head is 11 commits behind and must be updated without force-push, then all mandatory direct and hosted gates must be rerun on the new exact head;
the published changed tests exercise the closure, local solvers, and TwoFluidConservationEquations serialization, but no committed regression drives the new TwoFluidPipe.setEnableStiffBubbleDrag(true) transient split through the public pipe API. The corrected-path Tengesdal numbers are diagnostic evidence in the PR body, not a reproducible committed test path. A focused pipe-level deterministic/conservative integration regression is needed before the integration wiring can satisfy TwoFluidPipe: treat corrected dispersed-bubble drag as a stiff conservative source #2909's repeated-run, stage-purity, and regression requirements.

Next bounded repair: add that public-API transient regression without changing the frozen physics scope, update the branch from current master, run Spotless apply/check, documentation search, both pre-commit stages when available, focused plus TwoFluid and slow Tengesdal tests, and reassess exact-head CI/reviews. #2909 should remain open; #2933 is an opt-in numerical-foundation milestone and does not complete corrected-default or quantitative bubbly/severe-slugging validation.

No source, branch, or PR mutation was published in this checkpoint.

Merged/helper checkpoint and active integration — #2943 / #2944

Verified current master is a83e7213, which includes merged #2943.

Merged evidence is narrower than the P0 checklist item:

feat(twofluid): add configurable bubble-size closure #2943 adds a serializable, unit-explicit algebraic helper and tests its surface-tension trend, geometry cap, density-order symmetry, equal-density limit, and input rejection.
It does not wire the helper into InterfacialFriction or TwoFluidPipe; production bubble/dispersed-bubble calculations on that master still hard-code 0.02 N/m.
Its default formula also omits the historical production coefficient 2 sqrt(0.725 ...), so the claim that it reproduces the historical diameter is unsupported on merged master.

Draft #2944, exact head 8b74ea37, is the bounded corrective integration:

restores the historical default operation order and D/5 cap;
wires fixed/configured or explicitly selected section-local thermodynamic surface tension through both compatibility and corrected bubble-drag paths;
exposes Java/Python-friendly pipe configuration and preserves it through serialization;
adds helper, production interfacial-area, public-pipe, and serialization regressions plus both model-guide updates.

Exact-head PaperLab, Spotless and pre-commit are green; Java 21 Javadoc is green and the full test matrix, documentation search, and CodeQL remain in progress. This is active PR evidence, not a completed checklist item.

The corrected drag/source path remains opt-in and its published Tengesdal result remains 3/6; no severe-slugging, broad bubbly-flow, or commercial-simulator equivalence claim is advanced.

Merged production-integration checkpoint — #2944

Verified current master 1dee5b51 includes merged #2944.

Merged evidence now supersedes the prior helper-only state:

the historical default bubble-diameter expression and evaluation order, fixed 0.02 N/m surface tension, and D/5 cap are preserved;
the configured closure is wired through production bubble/dispersed-bubble interfacial calculations and exposed through Java/Python-friendly TwoFluidPipe accessors;
opt-in section-local thermodynamic surface tension and serialization/copy behavior have focused regressions;
both TwoFluid model guides document the formula, defaults, configuration, and limitations;
exact PR head 8b74ea37f74a243a67aaa2976d9428df0405fb42 completed Spotless, pre-commit, PaperLab, documentation search, CodeQL, Java 8/21 build/test/Javadoc, and slow shards successfully before merge.

This completes the bounded configurable bubble-size production-integration milestone only. The corrected drag/source path remains opt-in with 3/6 Tengesdal criteria; broad bubbly-flow validation, startup/shutdown/turndown, terrain/transfer qualification, severe slugging, and #2909 remain open. No commercial-simulator equivalence or accountable safety approval is implied.

Flowline hydraulics implementation campaign — user expansion, 11 October 2026

User request: "Hydraulics of flow lines. Implement so neqsim is state of the art for slugging, restart and pigging. If needed integrate it as a campain"

This issue is the umbrella; #3298 remains the owner of two-fluid numerical/severe-slugging qualification, #2911 of shared transactions/clocks, #4228 of canonical SURF topology, and #4281/#4282 of post-pig MEG/water film transfer. Reuse the existing twice-daily #3298 automation, with one shared active implementation PR and explicit file ownership. Do not start a duplicate scheduler or solver.

Current-master audit

Base 6a8c4e5097823caf11217b15edb3a6532f40e9e8. Current source, AGENTS.md, governance, active PRs and the #3298/#4228/#2907 ledgers inspected through GitHub. Three open implementation PRs were found (#4274, #4277, #4282).

TwoFluidPipe has distributed inventory, thermal and boundary APIs; Beggs & Brill remains steady/quasi-steady.
Close the transient/dynamics parity gap: sustained severe slugging, liquid-rich stability, and a published controls benchmark #3298's public sustained severe-slugging and liquid-rich gates remain unfinished. Its latest algebraic probes are prerequisite evidence, not physical qualification.
PLET pigging-facility metadata and flow-assurance advice exist; no dynamic pig force/trajectory model was found in current production source.
feat(pipeline): add experimental non-equilibrium post-pigging film transfer #4282 owns overlapping TwoFluidPipe/conservation/film documentation paths. This increment will not change those paths.

Ordered delivery and qualification contract

Current bounded increment: pressure-driven pig mechanics. Add an immutable, serializable, Java/Python-callable mechanical submodel and scheduled trajectory calculation with explicit SI inputs; exact breakaway, moving friction, stop/restart and reverse-motion events; pressure/gravity work, dissipated friction energy, and mechanical energy balance. Validate analytical constant-force trajectories, threshold equality, gravity, stopping inside a time step, reversal, partition invariance, deterministic reuse and serialization. No fitted/default vendor friction.
Conservative coupled two-fluid foundation (Close the transient/dynamics parity gap: sustained severe slugging, liquid-rich stability, and a published controls benchmark #3298). Resume signed face/phase-presence work once feat(pipeline): add experimental non-equilibrium post-pigging film transfer #4282's ownership releases. Preserve the unchanged five-second matrix before 180/600-second qualification, public amplitude/period limits, liquid-rich/four-hour/CFL and controls criteria. No clipping, hidden floors, fitted donor smearing or private-reference tuning.
Physical shutdown/cooldown/restart. Continue the SAME fluid and wall inventories through closure, terrain drainage/fallback, cooldown and compositional phase change, then pressure/choke ramps and reopening. Test zero/reverse flow, phase disappearance/reappearance, wall thermal inertia, restart pressure and outlet surge, repeated cycles, rejected-step rollback and deterministic checkpoint replay. This means operating restart, not only saved-state deserialization. Hydrate/wax/gel limits must be explicit; restart of gelled oil requires a validated rheology/structure model, not a Newtonian label.
Moving pig/fluid coupling. Couple the mechanical submodel to conservative upstream/downstream cell volumes and shared mass/component/momentum/energy fluxes. Add launcher/receiver events, static/kinetic friction, compressible/choked gas and liquid bypass, slip/leakage, stall and release, slope/diameter changes, liquid swept ahead and receiver surge. Pressure histories imposed on the mechanical submodel alone do not qualify this milestone.
Pigging and restart qualification. Independent public/redistributable data with geometry, fluid, boundary histories, friction provenance and uncertainty; upstream/downstream pressures, pig position/speed/arrival, liquid-slug volume, outlet peak/duration, residual film and energy. Declare acceptance bands before fitting; separate calibration and validation cases. Preserve existing tests, require component/total conservation, mesh/timestep convergence and nearby operating conditions. Single pig first; pig trains only after single-pig qualification.
Canonical field/process delivery. Consume the same [Campaign] State-of-the-art well, subsea gathering, injection and SURF network modelling/design #4228 edge identities, wells, chokes, host and injection boundaries; conservative junctions and receiver/separator capacity; Java/Python examples, diagnostic reports and truthful agent/MCP routing. Publish executed/rendered notebook examples after the core APIs are merged.

First-increment stop boundary and evidence

Only the mechanical pig submodel, its scheduled-input driver, analytical tests, API guide, and capability inventory are in the first PR. It is an experimental mechanics building block, not a coupled hydraulic pigging simulator, field-qualified restart model, or commercial-simulator parity claim. It must expose the difference in documentation and routing. User-specified force parameters remain assumptions.
Public validation leads: Hendrix et al., Experiments and modeling of by-pass pigging under low-pressure conditions, Chemical Engineering Research and Design (2019), PII S0959152418302774; Transient pigging dynamics in gas pipelines: Models, experiments, and simulations, Ocean Engineering (2021), PII S0029801821005618. Bibliographic leads are not imported data or passed validation.

Completion and campaign discipline

Maintain one active PR across this extension and #3298; advance the largest coherent bounded increment, preserve exact heads, and report actual tests, failures, sources, validity and next dependency. This campaign must deliver code and qualification, not repeat status-only probes. If core ownership blocks it, advance a disjoint dependency-ready mechanics/restart/benchmark increment. No partial PR closes #2907 or #3298. Stop recurring implementation after a merged-current-master completion audit; retain failed gates explicitly.

feat(pipeline): add event-resolved experimental pig mechanics

Implemented first flowline-hydraulics increment — PR #4293

Draft PR: [AI-Assisted] feat(pipeline): add event-resolved experimental pig mechanics #4293
Base: 6a8c4e5097823caf11217b15edb3a6532f40e9e8
Published head: faf188f47f0ad821f02cc00c51deb0d425cd88da
Scope: event-resolved experimental pig mechanics under prescribed pressure/inclination, scheduled trajectories, mechanical work diagnostics, analytical tests, documented Java/JPype API and discovery-only capability registration.
Files: PigMotion.java, PigMotionTest.java, docs/process/pig-motion.md, manual/section indices, capability registry and generated coverage; seven remote blob SHAs verified against the tested local files.

Verified results

The documented synthetic 100 kg pig reaches 3 m / 3 m/s after 2 s, stops at 12 m at elapsed 8 s, remains stationary to 12 s, then restarts to 13 m / 2 m/s at 13 s. A reversal case explicitly accounts for both travelled legs in friction dissipation.

All 11 focused JUnit tests pass, zero failures/errors/skips, including a deterministic 200-case force/velocity/inclination matrix comparing a 7 s interval with 70 × 0.1 s intervals, analytical force/work checks, equality at static breakaway, stop/reverse events, gravity, invalid inputs, serialization and exact documentation-snippet execution.

Passed: standalone ECJ Java 8 source/target compilation (major version 52), actual Maven Surefire execution, stable Spotless apply/check, Checkstyle/PMD goals, documentation search (788 Markdown + 1 HTML), coverage generation/freshness, both current pre-commit/pre-push stages and staged diff check. PMD reports no PigMotion findings; its repository-wide report contains pre-existing advisories.

VALIDATION PENDING CI: full local Maven lifecycle was blocked by the JRE-only compiler-wrapper failure; Javadoc executable is absent. These checks are not claimed passed. Hosted exact-head CI is running; engineering coverage and agent-skill cross-reference gates have already passed. The PR is mergeable (no conflicts), remains draft, and has no submitted reviews or inline comments yet.

Engineering boundary and next dependency

This is analytical/numerical verification of mechanical motion, not conservative fluid/pig coupling, pigging experimental validation, or physical flowline restart qualification. It changes neither #4282's film/conservation source nor #3298's physical thresholds. Documentation explicitly records those limits.

The existing automation 6a9098aae9bc8191a7fbbfe85f11e7e1 is now NeqSim Flowline Hydraulics, retaining 08:00/20:00 Europe/Oslo, extended to the finite contract in the preceding scope comment. No duplicate schedule was created.

Shared active PR for the #2907 extension / #3298 campaign: #4293. Next: complete exact-head CI and independent review within its frozen mechanics concept; after merge, resume conservative two-fluid prerequisites when #4282 releases overlapping files, then physical shutdown/cooldown/restart and moving-pig volume/flux/work coupling. Public benchmark and canonical #4228 integration gates remain mandatory. Neither roadmap is complete.

User priority: largest coherent PRs, fewest merges to completion

The user explicitly requests: "Make it build as big and few PRs as possible (so it gets to goal fast)".

The recurring NeqSim Flowline Hydraulics instructions have been updated accordingly:

Bundle the largest dependency-ready, end-to-end capability: prerequisite numerics, physical coupling, events/boundaries, diagnostics, validation, performance safeguards, APIs, docs and examples.
Keep one shared active [Campaign] TwoFluidPipe state-of-the-art R&D gap map #2907/Close the transient/dynamics parity gap: sustained severe slugging, liquid-rich stability, and a published controls benchmark #3298 PR and expand it across runs using ordered commits. Roadmap stages do not imply separate PRs.
[AI-Assisted] feat(pipeline): add event-resolved experimental pig mechanics #4293's previous mechanics-only scope is no longer an automatic ceiling. On the next implementation run, inspect its exact head and actual dependencies, broaden its recorded scope to the largest coherent pigging increment that can be validated, and include hydraulic coupling if its prerequisites and ownership permit. If blocked, name the exact dependency and finish all compatible work.
Split only for a demonstrated scientific/numerical, dependency, provenance, ownership or validation/review boundary, and record the reason. Do not publish a succession of isolated helper/test/docs PRs.
Preserve conservation, numerical/experimental acceptance, independent review and CI. Deliver a completed declared capability promptly; do not prolong a ready PR with unrelated additions.

This supersedes the earlier artificial mechanics-only batching restriction. The existing schedule and single-active-PR policy remain in place. This update changes execution policy; it does not claim additional physics implementation or new test results.

feat(pipeline): add event-resolved experimental pig mechanics (#4293)

2026-10-11 flowline-hydraulics checkpoint — merged mechanics, benchmark provenance, and concrete split boundary

Repository and ownership state

Current master: b37a84f17ae669ddcc8d6c149b0b6813b1b22d35.
#4293 is merged as 39644ff976a85812562748eaa6f25dea4749adcb, preserving its reviewed head faf188f47f0ad821f02cc00c51deb0d425cd88da. All nine hosted workflows on that head completed successfully.
No open PR references [Campaign] TwoFluidPipe state-of-the-art R&D gap map #2907 or Close the transient/dynamics parity gap: sustained severe slugging, liquid-rich stability, and a published controls benchmark #3298. Portfolio occupancy is 4/24.
#4282 remains an active draft at exact head d809313b7141b7d824e65d48d3cabcc4b24b23df and GitHub currently reports it non-mergeable. It owns TwoFluidPipe, TwoFluidConservationEquations, NonEquilibriumFilmTransfer, the flow-assurance skill, the capability registry, generated engineering coverage, process indexes, and the evidence matrix.

The largest next pigging batch would combine route/launcher/receiver event handling with conservative signed-face/phase-presence hydraulics and moving-volume work. Publishing that batch now would overlap #4282's core hydraulic and mandatory capability-generation files. Omitting the registry/generated updates would deliberately fail current quality gates. This is therefore an actual ownership/validation split boundary, not a mechanics-only or PR-size boundary. No branch or competing PR was opened and #4282 was not mutated.

Public benchmark provenance audit

Redistributable event/timing evidence — accepted for a narrow gate. Lima et al., Sensors 17 (2017) 2119, doi:10.3390/s17092119, is CC BY 4.0. The reported run gives:
- pressure-transducer spacing: 5.90 m;
- crossing time: 13.6 s from 272 samples at 50 ms;
- supervisory average speed: 0.43 m/s;
- odometer result: 5.70 m in 12.53 s, reported as 0.45 m/s;
- reported discrepancy: 4.44%.
This source is predeclared only for launcher/receiver and sensor-crossing event timing. It does not validate pressure-force, bypass-flow, liquid-sweep, or coupled hydraulic physics. The acceptance contract is to reproduce event-derived speed/time within the published 50 ms sampling resolution and the paper's reported 4.44% cross-method envelope, without fitting a force closure to that single run.
Low-pressure bypass-pig dynamics — reference identified, artifact not yet redistributable. Hendrix et al., Journal of Process Control 71 (2018) 1–13, doi:10.1016/j.jprocont.2018.08.010, documents 132 air runs in a 52 mm, 62 m loop. The TU Delft thesis reports configuration-level friction evidence (configuration 1: 27.1±6.5, 25.5±3.0, 23.9±5.0, and 26.5±4.5 N at 0/1/2/4% bypass; configuration 2: 40.3±1.5, 44.4±2.9, and 39.5±2.5 N at 0/1/2%). However, the repository record supplies no open raw run data for the article, and the thesis record states restrictive reuse terms with no license. These values may guide a cited validation design, but figures/data will not be copied or digitized into NeqSim without a confirmed reuse basis.
Additional experimental references remain non-artifact evidence. Li et al., Ocean Engineering 232 (2021) 109126, doi:10.1016/j.oceaneng.2021.109126, and Chen et al., Ocean Engineering 198 (2020) 106974, doi:10.1016/j.oceaneng.2020.106974, provide relevant stoppage/compressibility and gas-liquid bypass-pig experiments, respectively, but no redistributable raw dataset was located. They remain literature checks, not executable public benchmark artifacts.

Validation contract and next dependency

Keep analytical PigMotion event/energy verification distinct from the CC BY timing measurement, and both distinct from future coupled hydraulic validation.
Calibrate friction/bypass parameters only on an explicitly designated subset; hold out configuration/bypass cases for validation. Do not use the 2017 event-timing run to calibrate pressure-force physics.
Require conservation and work closure, terminal-event ordering, mesh/timestep refinement, nearby operating points, and the original Close the transient/dynamics parity gap: sustained severe slugging, liquid-rich stability, and a published controls benchmark #3298 five-second matrix before any 180/600-second or coupled-pig qualification claim.
Resume one shared draft PR after feat(pipeline): add experimental non-equilibrium post-pigging film transfer #4282 releases the overlapping hydraulic/capability files, unless a completely disjoint implementation can satisfy all mandatory generated-documentation gates first.

Documentation impact: none. This run changed no Java/API/model documentation; it established provenance, acceptance boundaries, and an explicit ownership blocker. No Maven or formatting command was run, and no test result is claimed.

2026-10-12 flowline-hydraulics checkpoint — #4282 blocker narrowed to metadata integration

Current protected master: 9b1366b6d3f3a8f1b170fbd0c623692fa5252a1a.
Shared active [Campaign] TwoFluidPipe state-of-the-art R&D gap map #2907/Close the transient/dynamics parity gap: sustained severe slugging, liquid-rich stability, and a published controls benchmark #3298 implementation PR: none.
Portfolio occupancy: 3/24 positively identified open implementation PRs (target 20).
feat(pipeline): add experimental non-equilibrium post-pigging film transfer #4282 remains open/draft/non-mergeable at preserved head d809313b7141b7d824e65d48d3cabcc4b24b23df; all ten exact-head workflow groups pass, with no reviews or unresolved threads.

A new read-only three-way audit compared #4282's base 8d52025524c0df5f47dae13838648db8025787ec, its exact head, and current master. Of its 14 changed files, only five overlap the 17 upstream commits:

flow-assurance skill;
capability registry;
reference-manual index;
process index;
generated engineering coverage.

Crucially, upstream did not change #4282's TwoFluidPipe, NonEquilibriumFilmTransfer, TwoFluidComponentTransport, TwoFluidConservationEquations, focused tests, film-drying guide, evidence matrix, or evaporation guide.

The remaining integration is deterministic: retain both disjoint capability IDs (pipeline-pig-mechanics-discovery and pipeline-film-transfer), retain both index rows/sections, retain the disjoint skill additions, and regenerate coverage. The detailed owner handoff is recorded on #4282.

This changes the blocker from a possible core-runtime conflict to a bounded documentation/registry integration owned by #4282. It still prevents safe publication of the next shared batch because conservative signed-face/phase-presence transport and moving-pig integration require the same mandatory registry/generated artifacts. No competing branch, partial PR, rebase, synchronization, workflow retry, threshold change, or peer-head mutation was performed.

Next dependency remains owner/maintainer reconciliation of #4282. Immediately after release, the largest coherent shared PR should combine signed-face/phase-presence transport, route/launcher/receiver events, moving-volume/flux/work coupling, conservation and refinement evidence, the unchanged five-second riser matrix, public timing evidence, API/docs/examples, and mandatory generated coverage.

2026-10-12 public timing gate — shared draft PR #4313

Opened draft PR #4313 as the sole shared #2907/#3298 branch.

Exact base: 7dc323b9bfc917f1fb000655857d316140605975
Exact head: d3cd2a22bc36e2d3acb2cd806d0da4af3dbeac3c
GitHub state: open, draft, mergeable; seven hosted workflow groups are queued/running, with no reported failure, review or inline thread.
Files: new PigMotionPositionEventTest.java; updated docs/process/pig-motion.md.
No production source, capability registry, generated coverage, feat(pipeline): add experimental non-equilibrium post-pigging film transfer #4282-owned file, threshold, clipping, donor smearing or private trace changed.

Executable public evidence

The test reproduces the CC BY 4.0 Lima et al. Sensors 17 (2017) 2119 event values:

5.90 m pressure-sensor spacing;
272 × 50 ms = 13.6 s crossing interval;
0.4338 m/s derived supervisory speed, reported as 0.43 m/s;
5.70 m / 12.53 s = 0.4549 m/s odometer speed, reported as 0.45 m/s;
the publication's 4.44% difference between the rounded speeds.

It advances 272 explicit trajectory intervals, proves the sensor is first reached at sample 272, and compares the endpoint with one 13.6 s interval. The numerical event tolerance is 1e-12 m, over ten orders smaller than one sample's travel.

This is sampled event-timing and partition evidence only—not pressure-force, friction, bypass, swept-liquid, receiver-surge or coupled-hydraulic validation.

Validation

12 focused PigMotion tests pass, zero failures/errors/skips.
Spotless apply/check, documentation search, coverage freshness, Checkstyle, PMD, both task-local pre-commit stages and diff check pass.
Local Javadoc is unavailable because the runtime contains only a Java 17 JRE; no production source or Javadoc changed. Exact-head hosted Javadoc is running.

Expansion boundary

A production position-event API was implemented and tested locally but correctly withheld: it changed mandatory generated coverage, an active #4282-owned conflict path. Publishing it would violate the ownership gate. Once #4282 is reconciled, expand this same shared PR with that event API plus frozen signed-face/phase-presence KKT transport, moving-volume/flux/work coupling, conservation/refinement evidence, the unchanged five-second riser matrix, documentation and generated coverage.

Documentation impact: the existing pig-mechanics guide now records the executable source, license, units, numerical tolerance and explicit non-qualification boundary.

2026-10-13 merged public-timing gate and next coupled-batch boundary

The sole shared #2907/#3298 PR #4313 has merged into protected master.

Reviewed PR head: d3cd2a22bc36e2d3acb2cd806d0da4af3dbeac3c
Merge commit: 4baabddd92432186b8b5bbb8aae7b871a706736e
Current master observed after the merge: d27433c40ba52cd6d647462f3ba388f1a7fc8dce
Files landed: PigMotionPositionEventTest.java and docs/process/pig-motion.md
Exact-head CI: all seven workflow groups passed (build/test/Javadoc, CodeQL, Spotless, pre-commit, documentation search, engineering coverage, production-optimization documentation); no human/Copilot review or unresolved inline thread was present.
Evidence: executable CC BY 4.0 Lima et al. timing arithmetic at 5.90 m, 272 × 50 ms = 13.6 s, 0.4338 m/s versus the reported 0.43 m/s, and the 5.70 m / 12.53 s odometer comparison.
Documentation impact: the pig-motion guide now records provenance, units, numerical tolerance and the explicit timing-only qualification boundary; no public API changed.

This merge is analytical/numerical event-timing evidence only. It does not qualify pressure-force closure, bypass, swept liquid, receiver surge, coupled two-fluid pigging, severe slugging or physical restart.

There is currently no active shared #2907/#3298 implementation PR. Portfolio occupancy is 4/24 positively owned open implementation PRs (target 20). A new shared PR is deliberately not opened because #4282 remains open/draft/non-mergeable at exact head d809313b7141b7d824e65d48d3cabcc4b24b23df, all ten workflows passing, with no reviews or unresolved threads. It still owns the same TwoFluidPipe/conservation runtime and mandatory capability-registry/generated-coverage paths needed for the directional phase-presence and conservative face-transport prerequisite.

Concrete split reason: actual active ownership conflict plus an unavailable prerequisite. Opening or expanding a competing core-solver branch now would violate exact-head ownership and create overlapping runtime/generated changes. The next largest coherent batch remains: directional phase-presence/complementarity and signed donor transport; shutdown/cooldown/restart inventory preservation; launcher/receiver and moving-volume/flux/work coupling; unchanged five-second riser matrix; conservation/refinement and public validation evidence; APIs, documentation, examples and regenerated coverage. That batch starts from then-current master only after #4282 is reconciled and its ownership is released.

Neither #2907 nor #3298 is complete or closed.

2026-11-03 implementation ledger — finite-rate hydrocarbon pool transfer

Roadmap item: conservative dry-hydrocarbon pool-transfer kinetics and complete-cycle numerical qualification, shared by #2907/#3298.

Draft PR: #4838
Exact base: 3204f16e009380d41f143686d71b384b01f2912e
Exact head: af03fb0e0eca628526191eaf1c6dd21a4cc059b5

Delivered in the one shared PR

opt-in HydrocarbonPoolTransfer with explicit per-component coefficients, interface width, donor limits, evaporation enthalpies and available heat;
local-equilibrium inventory target separated from kinetics while preserving existing component, phase-mass, donor-momentum and total-energy ownership;
TwoFluidPipe API/configuration/diagnostics integration;
analytical zero-drive/area, donor/heat-limit, unseeded appearance and 0.2/0.1/0.05 s exponential-refinement evidence;
pipeline mass/component/energy integration coverage;
public synthetic 780 s blowdown/settle/same-gas-refill/hold cycle on 12/24 cells;
source-kernel timing safeguard, durable numerical receipt, process/wiki docs, dynamic/flow-assurance skill guidance and OLGA routing.

Executed evidence

Spotless apply/check: pass.
documentation search audit: pass (828 Markdown pages, 1 HTML page).
focused closure/component/conservative-pipeline tests: pass.
broader component transport, trace inventory, phase appearance, equilibrium source and conservative-energy matrix: pass.
benchmark: pass; warmed source batches 2.665/2.082/1.833 microseconds per cell, diagnostic only.
full cycle: pass on both meshes. Peak oil 69,983.168/70,786.160 kg (1.13%); final oil 11.467/12.198 kg (5.99%).
package with Javadocs skipped: pass. Ordinary Javadoc attachment could not run because this runner has no javadoc executable; no source workaround.
pre-commit executable unavailable locally; equivalent repository Spotless/documentation gates were run directly.

Documentation impact: updated TWOFLUIDPIPE_MODEL.md, two_fluid_model.md, neqsim-dynamic-simulation, neqsim-flow-assurance, OLGA routing and the durable receipt. Controls benchmark documentation: none—no control loop, threshold or controls runtime changed.

Qualification boundary: analytical mechanics/kinetics and numerical conservation are distinct from experimental validation. Coefficients remain explicit and unvalidated; this is not predictive pool depletion, physical restart, severe-slugging validation, field/commercial qualification or hydraulically coupled pigging.

Concrete blocker/split reason: finite 100 W/m complete-cycle and closed-outlet settling sensitivities fail loud when a far-from-equilibrium forced phase changes the conservative-recovery EOS branch. No mass clipping, donor smearing, phase relabeling or threshold weakening was used. Physical finite-heat shutdown/cooldown/restart and moving-pig coupling therefore require conservative forced-phase EOS branch compatibility through that cycle, plus independent public kinetic calibration/validation. #4282 was not mutated.

Next dependency: repair and qualify forced-phase EOS branch continuity through the closed/finite-heat cycle on this same PR if review/ownership permit; only then add physical restart and moving-pig volume/flux/work coupling. #2907 and #3298 remain open.

2026-11-03 stage-3 scope freeze — wall-inventory shutdown/cooldown/restart

Merged prerequisite: PR feat(pipeline): qualify conservative hydrocarbon pool transfer #4838 is merged as f81ac4a2b4fcd0b00da30b4ded7b1af28cda9cb8; current master is b9278492083b9b3d7241739c58d1c010ad13f4d1. The one later master commit changes only DEXPI/P&ID paths. The final feat(pipeline): qualify conservative hydrocarbon pool transfer #4838 head remains under hosted completion audit; pending checks are not counted as passed.
Next useful capability: one coherent stage-3 screening path that retains accepted fluid composition/energy and wall thermal inventory through isolation, terrain drainage, seabed cooldown, phase disappearance/reappearance, controlled refill/reopening and repeated restart cycles.
Compatible prerequisite included: replace the optional fixed-width pool area with an opt-in, volume-preserving hydrostatic cell interface area from the already-merged geometry diagnostic; retain the fixed-width constructor for compatibility and publish per-cell area/rate diagnostics.
API/state contract: validated initial wall-temperature profile, defensive readback, copy/serialization/transaction rollback continuity, and no hidden wall-state reset between accepted event intervals.
Acceptance matrix in the same shared draft PR: analytical hydrostatic-area limits; lumped fluid/wall cooldown energy benchmark; zero/reverse-flow closure; inclined terrain drainage; finite-heat phase appearance/disappearance; pressure/refill/reopen ramp; restart pressure and outlet-surge diagnostics; repeated-cycle conservation; timestep/mesh comparison; rollback and serialized replay; performance guard; Java API/Javadocs; TWOFLUIDPIPE/wiki/dynamic/flow-assurance/routing documentation and an executable example.
Qualification boundary: this is numerical/analytical restart screening. The transfer coefficients, seabed U-value/material data and hydrocarbon kinetics remain user inputs; calibration and validation stay separate. No gelled-oil claim, experimental/field qualification or commercial parity claim.
Concrete split reason for moving-pig hydraulics: moving volume/face remapping and pig pressure-work coupling is a separate high-risk moving-boundary conservation algorithm. It depends on this wall/phase/restart inventory contract and on independently qualified pool/film transport. Folding it in before those gates would make the combined batch impossible to validate responsibly. [AI-Assisted] feat(pipeline): add event-resolved experimental pig mechanics #4293 remains analytical prescribed-pressure mechanics only.
Ownership: this will be the sole shared [Campaign] TwoFluidPipe state-of-the-art R&D gap map #2907/Close the transient/dynamics parity gap: sustained severe slugging, liquid-rich stability, and a published controls benchmark #3298 pipeline implementation branch/PR. It will not duplicate [Campaign] Professional reservoir-to-market dynamic process simulation #2911 scheduling, [Campaign] State-of-the-art well, subsea gathering, injection and SURF network modelling/design #4228 topology, [Campaign] State-of-the-art NeqSim MCP process simulation and troubleshooting #3153 exposure or feat(pipeline): add experimental non-equilibrium post-pigging film transfer #4282 aqueous-film work and will close none of these issues.

2026-11-03 shared implementation continuation — draft #4847

The sole shared #2907/#3298 implementation PR is now #4847 at exact head 6de3c763ca44c83e040c92e33e158eb5efc76a50, based on b9278492083b9b3d7241739c58d1c010ad13f4d1.

The coherent batch advances shutdown/cooldown/refill screening with:

volume-preserving hydrostatic pool interface area and equilibrium appearance without mass seeding;
explicit diagnostics for hydrostatic area and the sub-resolution trace fallback;
persistent wall-temperature initialization, defensive wall diagnostics, physical wall-property validation and transactional carryover;
the complete finite-wall 780 s blowdown–settle–refill–hold path on 12/24 cells;
focused, conservative, transactional, phase-transition and energy tests;
model/API/skill/routing documentation and regenerated engineering coverage.

All declared local focused/proportional/slow tests, Spotless, documentation search, coverage regeneration, whitespace and skip-Javadoc package passed. The normal package is locally blocked only because the environment lacks the javadoc executable; hosted Javadoc is pending. The two meshes report peak/final oil differences of 22.82%/24.29%, so this is broad numerical screening rather than a converged predictive band.

Qualification boundary: this is not severe-slugging experimental validation, physical restart qualification, coupled hydraulic pigging, gelled-oil restart or field/commercial qualification. Moving-pig coupling remains a concrete split because the independently high-risk moving boundary still needs qualified restart/phase-presence and residual-film transport before conservation/public-benchmark review is responsible. No #4282 mutation and no duplication of #2911/#4228/#3153 ownership. #2907 remains open.

2026-11-04 merged audit — restart-screening prerequisite

Shared PR #4847 merged as a55262f178d19a2981cc07595b7c8c374e229c2d from final head fb55d574ea6a918c06e64131dd43c3ddd2be4bca. All 9 final-head workflow groups passed. Current master eb672b7a3d2836393f54326a50d19426d2b7659e retains all pipeline runtime/test/model/skill/evidence files; current-master full Java 21 Ubuntu/Windows suites are still running, so the merged-default audit remains pending those two jobs.

The next single shared implementation batch will complete #4674 end-to-end: explicit reversed gas/liquid donors, full boundary mass/component/phase/energy/momentum accounting, rollback/replay, complete blowdown–settle–refill–hold events, pressure-ringing limits, nearby cases and space/time refinement, plus API/docs/evidence/performance gates. Moving-pig hydraulics remains split only at its real high-risk moving-boundary conservation dependency. #2907 remains open.

This advances the prerequisite flowline boundary/conservation path for restart and later hydraulic pig coupling; it is not reported as coupled pigging.

2026-11-04 shared flowline ledger — draft PR #4865

Roadmap/capability: WS3/#4674 signed-boundary donor, component transport, conservative momentum/pressure diagnostics, and transactional auditability. This is the one shared active #2907/#3298 implementation PR.

Exact base/master at publication: e316ffb6d7eb42a176156dd278731a010d9a5196
Exact PR head: 5514a68e3c7533d43aad1aa9157e07cd038206cc
Branch: codex/4674-signed-boundary-restart-qualification
Diff: 14 files, +506/-45. Production changes are confined to TwoFluidPipe, TwoFluidComponentTransport, and TwoFluidConservationEquations; the remainder is focused tests, generated coverage, and the affected model/evidence/skill/router documentation.
Behavior: positive inlet phases remain external donors; negative prescribed inlet phases use cell 0 state/composition and common signed velocity. Conservative mass/component/energy/momentum paths use the same donor convention. New total phase-momentum-flux and pressure-only-traction ledgers are immutable/defensively copied and preserved by save/restore, rollback/replay, and serialization.
Analytical verification: countercurrent face-0 donor/pressure-force balance and independent external gas/oil fallback outlet checks.
Focused tests passed: TwoFluidInletBoundaryTest, TwoFluidPipeInletBoundaryStateTest, TwoFluidPipeComponentTransportTest, TwoFluidTransactionalEvaluationTest.
Broader tests passed: TwoFluidClosedBoundaryMomentumTest, TwoFluidPipeBoundaryConditionTest, TwoFluidPipeConservativeEnergyTest, TwoFluidPipeConservativeQualificationTest, TwoFluidPipeConservativeTransitionTest, TwoFluidPipeTransactionalTest, TwoFluidOutletBoundaryTest. A clean pomJava8.xml clean focused matrix also passed.
Opt-in qualification command: ./mvnw -DexcludedTestGroups= -Dtest=TwoFluidRichGasDragQualificationTest test — 7/7 passed; Surefire 1,622.836 s, 780 s full-cycle method 1,378.09 s.
12/24-cell screening: peak oil 8,330.683/6,429.568 kg; final oil 8,116.001/6,144.303 kg; differences 22.82%/24.29%; minimum heat-transfer scale 0.001089375/0.000809341; maximum effective area 187.875656/101.226681 m²; maximum wall ΔT 43.222741/45.036733 K; one trace fallback each. This is numerical screening, not refinement or experimental restart qualification.
Publication gates passed: python devtools/run_spotless.py check; python devtools/check_documentation_search.py (833 Markdown + 1 HTML); uv run --project devtools --locked devtools/build_engineering_coverage.py --check (3,492 public types, 34 capabilities, 62 anchors); Checkstyle; git diff --check. The Maven preparation helper was run before Maven/Spotless. Local pre-commit and javadoc executables were unavailable; hosted exact-head CI is authoritative for those gates.
Documentation impact: updated TWOFLUIDPIPE_MODEL.md, two_fluid_model.md, the evidence matrix, dynamic-simulation and flow-assurance guidance, router guidance, and generated engineering coverage.
Remaining limitations: no claim of physical restart qualification, severe-slugging experimental acceptance, hydraulic pig coupling, pig swept-volume/flux/work conservation, public experimental validation, or field/commercial parity.
Concrete split reason: moving-pig coupling is an independently high-risk moving-boundary conservation/event algorithm and still depends on physical restart/refinement closure plus the residual-film/public-benchmark basis coordinated with feat(pipeline): add experimental non-equilibrium post-pigging film transfer #4282. It is not bundled as unvalidated physics.
Next dependency: finish exact-head CI/review follow-up; then pressure-ringing/refinement and shutdown/cooldown/restart inventory closure on this same PR where dependency-compatible, before qualified moving-pig coupling.

The PR remains draft. No issue is closed by this increment.

2026-11-04 exact-head refinement addendum — PR #4865

A read-only follow-up reused the exact runFiniteRatePoolCycle(12, dt) fixture at unchanged head 5514a68e3c7533d43aad1aa9157e07cd038206cc with dt = 0.25 s (halved from the published 0.5 s outer interval). It completed in 289.014 s with all embedded component, phase-mass and conservative-energy assertions intact:

peak oil: 8,330.743515 kg versus 8,330.683379 kg (0.00072% change);
final oil: 8,114.879386 kg versus 8,116.000871 kg (0.0138% change);
minimum heat scale 0.0010878023, maximum interface area 187.950021 m², maximum wall ΔT 43.183151 K, trace-area fallback observed.

This clears outer-step sensitivity for the 12-cell full cycle at the tested pair. It does not clear spatial refinement: the previously recorded 12/24-cell final-oil difference remains 24.29%, so terrain/pool spatial resolution is now the dominant numerical blocker, alongside pressure-ringing and independent physical validation. No source, threshold or PR-head change was made.

Exact-head hosted status at this receipt: 8/9 workflow groups green; [Core] Build, Test, and Javadoc still running. The PR remains mergeable draft with 0 reviews and 0 unresolved threads. No physical restart, experimental slugging or hydraulic-pigging qualification is claimed.

Pigging benchmark/provenance checkpoint — 2026-11-05

Current master was re-read as d6dd0cd190ac32f04a08627fafdeff7f745566e5; PR #4865 is merged and no shared #2907/#3298 implementation PR is open.

The detailed acceptance contract is recorded on #3298 in this ledger comment. The strongest public target found is Hendrix et al. (2018), DOI 10.1016/j.jprocont.2018.08.010: 62 m horizontal, 52 mm-ID, low-pressure air, 0/1/2/4% bypass, pressure and pig-speed transients, friction and stall evidence.

Scope distinction is explicit:

usable for gas-side compressibility, bypass loss, dry friction, launcher/receiver travel, stall/release and velocity-excursion qualification;
not evidence for two-phase swept-liquid slug, residual film, terrain drainage, restart, receiver capacity or field/commercial parity.

Calibration is frozen to configuration 1 steady 0/1/2% runs; validation is configuration 2 0/1/2%, withheld 4% stall and the published local 0% transient. Predeclared bands are 2σ for steady differential pressure/friction, ±10% mean velocity, ±20% maximum velocity and dominant period, exact stall classification and trends, plus <5% mesh/timestep changes while retaining all existing #3298 conservation gates.

Concrete split reason: the detailed dissertation is publicly readable but all-rights-reserved and does not provide a redistributable machine-readable trace dataset. No extracted table/plot fixture may be committed without permission or an independently open source. This provenance gate does not block compatible restart/refinement or gas-side event implementation, but it blocks an experimental-qualification claim.

Changed files/tests/documentation: none; evidence/ledger only. No branch, PR, threshold or issue state was changed. #2907 remains open.

2026-11-05 continuation scope — physical restart dependency

Current master and the implementation base were re-read as f4a980b2168303e9012a21bb4046ce41ad1e8a7a. PR #4865 and the terrain-pool reporting PR #4871 are merged; no shared #2907/#3298 implementation PR is open.

The next shared branch is frozen to the largest dependency-ready restart capability:

replace the long-inclined-cell trace-area fallback with a volume-preserving hydrostatic reconstruction that retains the leading-edge depth without subtractive loss;
preserve every published 3,060-point geometry reference and add analytical reversal, partition/refinement, extreme-trace and full/empty limits;
rerun the unchanged 780 s blowdown–settle–same-gas refill–hold matrix with phase mass, component, thermodynamic-energy and wall-energy ledgers intact;
add low-point inventory/area/heat/fallback and pressure-ringing/refinement evidence, performance safeguards, generated capability coverage and affected model/reporting/skill/router documentation.

Admission gate for gas-side moving-pig coupling on the same branch: the restart matrix must first remove normal-operation trace fallback and meet the predeclared spatial/timestep bands without relaxed tolerances. If it does, the branch may expand to conservative swept-volume/flux/work, launcher/receiver, bypass, friction and stall/release. If it does not, the demonstrated restart spatial mismatch is a numerical incompatibility that blocks responsible moving-boundary coupling; the public Hendrix trace licensing gate independently blocks an experimental-qualification claim.

This is scope/acceptance recording before edits. No branch, PR, threshold or issue state has yet changed; #2907 and #3298 remain open.

Pre-result numerical acceptance amendment — 2026-11-05

Master remains f4a980b2168303e9012a21bb4046ce41ad1e8a7a; no shared #2907/#3298 PR is open. Before observing any modified 780 s result, the restart dependency gates are frozen as follows:

unchanged 12/24-cell, 0.5 s, 780 s blowdown–settle–same-gas refill–hold sequence;
peak and final oil inventories must each change by <10% between 12 and 24 cells (the existing broad <100% screening gate is not sufficient for this continuation);
normal-operation accepted trace wedges must use the volume-preserving hydrostatic reconstruction with zero reported trace-area fallbacks on both meshes;
every existing phase/total mass, named-component, conservative-energy, wall-inventory, boundary-donor and EOS-admissibility gate remains unchanged;
refill inlet pressure is recorded without filtering at the external interval; direction reversals use an explicit 100 Pa adjacent-sample gate. Extrema, peak-to-peak pressure, maximum adjacent-sample rate, total/excess variation and ringing ratio are diagnostics only—no physical surge acceptance band is claimed without independent evidence.

Gas-side moving-pig volume/flux/work and event coupling may enter the same branch only after these dependency gates pass. No threshold will be relaxed after seeing the results. No repository branch, PR, issue state or default branch was changed by this ledger entry.

2026-11-05 continuation ledger — shared draft #4881

Roadmap item: restart prerequisite: representable trace-pool geometry, unfiltered refill-pressure diagnostics and unchanged full-cycle acceptance receipt.

Exact continuity

base: f4a980b2168303e9012a21bb4046ce41ad1e8a7a
head: 90f97fd929e3d2f75351db7ae6447fd5d2b6ed31
shared branch: codex/2907-restart-pig-coupling
draft PR: fix(twofluid): preserve restart transfer area and trace volumes #4881
final diff: 18 files, +480/-91; mergeable; CI in progress; no reviews or unresolved threads at inspection.

Capability advanced

stable leading-edge hydrostatic reconstruction removes centre/range cancellation for representable inclined trace wedges without floors, seeding, clipping or fitted donor smearing;
public TwoFluidPipeReport.analyzePressureTrace(...) exports unfiltered event extrema/times, pressure range/rate, total/excess variation, ringing ratio and thresholded reversals;
API discovery/generated coverage, analytical tests, persisted full-cycle receipt, user docs and dynamic/flow-assurance routing are included in the same PR.

Validation

PASS: 55 focused/proportionate Java 17 tests across analytical geometry, pool transfer, reporting, conservative transition/energy and transactions.
PASS: 3,060 public hydrostatic reference rows plus the independent 1e-30 inclined-trace control.
PASS: Spotless check, documentation-search audit, engineering-coverage generation/check, capability search, Checkstyle (0 violations), JSON and diff checks.
EXPECTED OPEN GATE: both 12/24-cell 780 s cycles complete with zero trace fallback, then the unchanged <10% gate fails: peak oil 22.82%, final oil 24.29%. Pressure range differs 0.45%, but maximum rate/excess variation/ringing ratio differ about 45–48%; 100 Pa reversals are 8/12.
ENVIRONMENT LIMIT: Java 8 and aggregate Javadocs could not resolve uncached Maven plugin dependencies because the runner proxy refused the connection; no source failure was inferred.

Documentation impact: updated TWOFLUIDPIPE_MODEL.md, two_fluid_model.md, reporting guide, evidence matrix, both relevant skill guides, public receipts, capability registry/generated MCP coverage and changelog. No olga-simulation-agent file exists in this repository.

Remaining limitation / concrete stop reason: physical restart, severe slugging and coupled hydraulic pigging remain unqualified. Moving-pig volume/face-flux/pressure-work coupling is withheld because it would inherit the demonstrated unconverged restart state and is independently high-risk moving-boundary conservation work. Hendrix trace provenance/licensing separately blocks experimental pigging qualification. No #4282 changes were made.

Next dependency on this same PR: make terrain-pool area/source coupling mesh insensitive, rerun spatial/time/nearby-point gates, then admit gas-side pig coupling only if the frozen restart gates pass. Neither #2907 nor #3298 is closed.

2026-11-06 continuation — merged restart audit and next shared branch

Merged baseline: master 5fa68e8c67f21ebaaea0d3e7c63be7c0ea58f92a includes #4881 (final head 9bda46693ebae21ad48ce21f50f48bc12384673e). Its declared flowing 12/24-cell 780 s restart screen now passes the unchanged oil-inventory gates (2.478% peak, 2.576% final) with conservation and geometry audit evidence. This is numerical/synthetic restart evidence only; pressure-rate/ringing refinement, physical restart, severe slugging and hydraulic pigging remain unqualified.

Single shared continuation branch: codex/3298-gas-pig-hydraulics-20261106, created from that exact master. No other open #2907/#3298 implementation PR exists.

Largest dependency-ready capability mapped for this branch: a transaction-safe, gas-side coupled single-pig model that reuses PigMotion and the existing strict ideal-gas short-orifice release model. The declared batch includes moving upstream/downstream control volumes; conservative gas mass and energy inventories; pressure/displacement work coupling; bidirectional compressible/choked bypass; breakaway/stall/release; exact launcher/receiver contact accounting; mass/energy/event diagnostics; adaptive performance safeguards; analytical, conservation, timestep/refinement and nearby-point tests; API discovery; documentation and executable example.

Acceptance boundary: this branch may claim coupled ideal-gas control-volume pig mechanics/hydraulics only after its conservation and refinement gates pass. It will not claim distributed TwoFluidPipe coupling, swept-liquid slug, residual-film prediction, severe-slugging qualification, physical restart, field/commercial parity, or experimental validation.

Concrete split reasons: (1) the Hendrix et al. public geometry/provenance supports model selection, but the redistributable pressure/trajectory trace needed for independent experimental qualification remains unavailable; (2) swept-liquid/residual-film coupling depends on qualified multiphase moving-boundary transport and must not overlap #4282/#4281; (3) whole-network transactions remain owned by #2911/#4228. Those items stay explicit rather than being represented by fitted or private traces.

Next publication will remain a draft PR and will record exact commands, evidence, documentation impact and residual limitations.

2026-11-06 pigging continuation ledger — coupled ideal-gas single-pig draft

Draft PR #4888 materially expands the merged prescribed-pressure mechanics baseline into the largest currently dependency-ready gas-side coupled single-pig capability: #4888

Exact master / base: 5fa68e8c67f21ebaaea0d3e7c63be7c0ea58f92a
Exact validated published head: 3c106735a82c75eebdd67df967be03d12c9030ac
Remote integrity: 13/13 published blob SHAs equal the locally validated files; diff is 1,377 additions / 2 deletions.
State: draft; no merge, auto-merge, ready transition, force push, master merge/rebase or closing keyword.

End-to-end capability in this head

The new IdealGasPiggingModel transactionally couples PigMotion to two moving-volume ideal-gas chambers. It conserves chamber mass and internal energy, applies equal-and-opposite pig/gas pressure work, transfers signed donor mass/enthalpy through the existing strict bidirectional subcritical/choked bypass model, resolves launcher retention/release, stop/reversal and receiver-contact/impact events, and limits accepted steps by time, donor mass and pig travel. Public Java/JPype discovery, executable documentation, diagnostics, evidence receipt, performance safeguards and routing documentation are included in the same PR.

Validation

47 focused tests passed with zero failures/errors/skips under ./mvnw -q -Dskip spotless:check -Dtest='PigMotion*,IdealGasPiggingModelTest,ReleaseFlow*Test' test checkstyle:check.
Spotless apply/check, targeted SpotBugs, engineering-coverage regeneration, documentation search (834 Markdown + 1 HTML), JSON syntax and diff whitespace checks passed.
Hendrix-geometry refinement at 20/10/5 ms:
- position: 21.576661 / 21.440220 / 21.372485 m
- upstream pressure: 127143.737 / 127725.321 / 128014.347 Pa
- downstream pressure: 220540.759 / 219938.572 / 219641.258 Pa
Nearby bypass diameter at 0.2 s: 2.5 mm = 3.50417e-4 kg; 3.5 mm = 6.85556e-4 kg.
Sealed mass/energy closure, forward choked and reverse/subcritical bypass, breakaway/stall, receiver impact, serialization, external/internal transaction equivalence and invalid-input failure are covered.

Claim boundary and documentation

This is analytical mechanics plus reduced-order numerical conservation/refinement evidence. It is not experimental validation, field/commercial qualification, distributed hydraulic pigging, swept-liquid slug/residual-film prediction or physical restart qualification.

Updated: pig-motion.md, TWOFLUIDPIPE_MODEL.md, two_fluid_model.md, evidence matrix, dynamic/flow-assurance skills, capability registry/generated MCP coverage and changelog. Controls and OLGA routing are unchanged because this API does not alter those runtimes.

Concrete split reasons / next dependency

The selected Hendrix source lacks a redistributable raw pressure/trajectory trace, so experimental qualification cannot be claimed or bundled without resolving provenance/licensing.
Distributed multiphase moving-boundary, swept-liquid and residual-film work depends on the still-failing two-fluid pressure/rate/ringing refinement gate and must not overlap Add transient stationary wall-film drying after pigging with CPA MEG/water transfer #4281/feat(pipeline): add experimental non-equilibrium post-pigging film transfer #4282.
Network transactions/topology remain owned by [Campaign] Professional reservoir-to-market dynamic process simulation #2911/[Campaign] State-of-the-art well, subsea gathering, injection and SURF network modelling/design #4228.
Hosted CI must supply full-JDK Javadoc and repository-wide gates unavailable in the 536 MB/JRE-only runner.

Next: exact-head CI/review repair on this shared PR, followed by the remaining two-fluid refinement prerequisite and then distributed hydraulic pig coupling when that dependency is qualified.

2026-11-06 flowline extension ledger — PR #4894 expanded and published

Capability now on the shared draft: a conservative signed pressure-reservoir boundary and event-continuous closed-to-ramp restart screen, combined with #4894's detached pressure/donor transaction primitive.

Exact heads

base/master: de81c1f81c226988a2326c27ed1873094ed86e11
pre-expansion: 5ad86aca5a1da1a3e000a18bc67c80754ba977f5
published: 8ce7480e3b5c7c374529c26d16ff2e3ef025ab0c
feat(twofluid): couple detached pressure trials and restart ramp #4894 remains draft and the sole shared [Campaign] TwoFluidPipe state-of-the-art R&D gap map #2907/Close the transient/dynamics parity gap: sustained severe slugging, liquid-rich stability, and a published controls benchmark #3298 implementation PR.
Fast-forward publication used exact-head lease checks; all 17 remote blobs match the validated tree.

Implementation

signed phase face speeds from linear acoustic compatibility;
flashed external inflow donor and first-cell reversed-flow donor carried consistently through phase, component and conservative-energy ledgers;
isolated-reservoir flash cache with public diagnostic, but no cache for live connected streams;
mechanical inlet-face pressure API for continuous shutdown-to-pressure-ramp events;
detached trials now verify actual signed admission/release as well as owner isolation, stream identity, stale-token rejection and rollback.

Validation receipt

combined focused/broader matrix: 76/76 passed;
complete 780 s blowdown–settle–ramp–hold matrix: passed at 12/24 cells and 0.5/0.25 s, 502.717 s wall-test time;
spatial peak/final oil differences: 0.7910% / 0.7910%;
temporal peak/final oil differences: 0.0912% / 0.0949%;
temporal pressure amplitude/rate differences: 0.0985% / 0.1224%;
phase/total mass, named-component and conservative-energy assertions pass for all accepted intervals with zero trace-area fallback;
raw 0.25 s trace retains one 880.882 Pa opening dip (1761.765 Pa excess variation);
formatting, Checkstyle, docs search, generated coverage, JSON, diff, pre-commit/pre-push and focused doclint passed; targeted SpotBugs completed without suppressing longstanding findings;
exact-head CI is pending; no review or inline thread existed immediately post-push.

Documentation impact
Updated model/reference docs, reporting guidance, cookbook, numerical evidence, dynamics/flow-assurance routing, SURF integration and capability coverage. No controls runtime changed, so controls docs are unaffected. This repository has no olga-simulation-agent skill to edit.

Qualification status and split reason
This is numerical restart screening, not public experimental restart/surge qualification and not hydraulic pigging. It does not claim severe-slugging, gelled-oil, swept-liquid/residual-film or field/commercial qualification. The next split is justified by unavailable public restart/surge provenance, one retained opening reversal, missing repeated-cycle/five-second-riser acceptance, and the independently high-risk distributed moving-pig volume/flux/work algorithm requiring dedicated conservation/refinement/public-experiment qualification. #2907 remains open.

2026-11-07 merged-master audit — next shared capability boundary

feat(twofluid): couple detached pressure trials and restart ramp #4894 is merged (source 520c5a1ae720524bf9aa13411344cd792c6005f2, merge 97b98348071f9c1147ee026c2d59a7e8618c164b); audited current master is d8c9354ab490564c3c2220b8f735c645cd575c79. No open shared [Campaign] TwoFluidPipe state-of-the-art R&D gap map #2907/Close the transient/dynamics parity gap: sustained severe slugging, liquid-rich stability, and a published controls benchmark #3298 implementation PR exists.
PASS on current master: the unchanged 12/24-cell, 0.5/0.25 s, 780 s blowdown–settle–ramp–hold matrix (418.547 s suite time). It reproduces 0.7910%/0.7910% spatial peak/final oil differences, 0.0912%/0.0949% temporal oil differences, 0.0985%/0.1224% temporal pressure-amplitude/rate differences and zero trace-area fallback. This is synthetic numerical restart screening, not experimental validation.
FAIL on current master: the unchanged opt-in five-second Tengesdal preparation matrix. Six short handoffs pass; all six five-second lanes fail. Midpoint failure times are 0.8328125/0.826171875/0.625 s; backward-Euler pressure-interpolated times are 1.4/1.8125/0.975 s, at terminal attempted steps of 1.953125e-4–3.90625e-4 s.
Enabling the merged signed-face complementarity path does not clear any lane. A larger-iteration diagnostic and bounded non-monotone Armijo prototype also fail; the prototype was completely discarded. No source, tolerance, conservation threshold, fixture, branch or PR was published.

Next largest coherent branch: repair the reusable unsplit nonlinear transition so the unchanged six-lane matrix passes at 1e-9 nonlinear and 1e-8 independent conservation tolerances, with residual/active-set diagnostics, mesh/timestep/nearby-point evidence, deterministic performance safeguards and documentation. If that prerequisite passes on the same branch, absorb distributed moving-pig volume/face-flux/pressure-work coupling plus launcher/receiver/events/conservation/API/example coverage there rather than opening another PR.

Concrete current split: distributed multiphase pig coupling cannot responsibly inherit a solver that exhausts bounded subdivision before 1.813 s in every public-riser lane. Experimental qualification is independently blocked by the absence of a redistributable raw pig pressure/trajectory trace with resolved provenance/licensing. #4282 and #2911/#4228-owned film/network work were not touched or duplicated.

Documentation impact: none; no runtime/API behavior or supported-envelope claim changed. #2907 and #3298 remain open.

2026-11-07 flowline-extension dependency ledger — phase event before hydraulic pig coupling

Current master is 2f6ea706cde0760adea991fac4e9dd5c4d33c290; no shared #2907/#3298 implementation PR is open. No code or PR was published because the tested changes were neutral and were reverted.

The unchanged five-second public Tengesdal preparation gate still fails at the flowline–riser kink (midpoint 0.8328125 / 0.826171875 / 0.625 s). Exact terminal diagnostics localize the failure to positive trace gas approaching disappearance at cells 8/13. Full Jacobian width, finite-difference step sensitivity, phase-aware column scaling, the existing signed-face complementarity, deeper line search and frozen trial rows do not clear a lane. The detailed numerical evidence and commands are recorded in #3298 comment 6711918150.

Pigging/restart consequence: the existing signed-face block conservatively owns face direction and an exactly absent donor, but not the temporal cell phase-presence event. Moving-pig swept volumes, launcher/receiver events, bypass and restart drainage all require that event to be non-Zeno and conservative. Coupling them now would bundle moving-boundary physics onto a solver that cannot cross the prerequisite phase boundary.

Demonstrated split reason: independently high-risk numerical incompatibility—cell disappearance/reappearance needs conservative event localization or equivalent semismooth temporal complementarity before hydraulic pig coupling can be validated. This is not a roadmap/file-count split. The next shared draft should remain one large capability batch: phase event + signed-face coupling + unchanged five-second matrix + conservation/refinement/performance diagnostics + docs/API, then add compatible restart/pig coupling on the same branch once the prerequisite gate passes.

No mass clipping, floor, fitted smearing, threshold weakening, private-trace tuning or claim of coupled hydraulic pigging was made. Documentation impact is none because runtime/API behavior is unchanged. #2907 remains open.

[Campaign] TwoFluidPipe state-of-the-art R&D gap map #2907

Description

Purpose

Prioritized gap map

P0 — foundational correctness and reusable architecture

P1 — initialization, boundaries, and numerical robustness

P2 — public physical validation

P3 — advanced transients and flow assurance

P4 — performance and engineering workflow

Evidence context

Per-increment ledger requirements

Activity

EvenSol commented on Aug 9, 2026

Increment 2026-08-09 — dispersed-bubble drag dimensional correction

EvenSol commented on Aug 10, 2026

EvenSol commented on Aug 10, 2026

EvenSol commented on Aug 11, 2026

2026-08-11 active-increment checkpoint — draft #2933

EvenSol commented on Aug 11, 2026

Merged/helper checkpoint and active integration — #2943 / #2944

EvenSol commented on Aug 11, 2026

Merged production-integration checkpoint — #2944

EvenSol commented on Oct 11, 2026

Flowline hydraulics implementation campaign — user expansion, 11 October 2026

Current-master audit

Ordered delivery and qualification contract

First-increment stop boundary and evidence

Completion and campaign discipline

EvenSol commented on Oct 11, 2026

Implemented first flowline-hydraulics increment — PR #4293

Verified results

Engineering boundary and next dependency

EvenSol commented on Oct 11, 2026

User priority: largest coherent PRs, fewest merges to completion

EvenSol commented on Oct 11, 2026

2026-10-11 flowline-hydraulics checkpoint — merged mechanics, benchmark provenance, and concrete split boundary

Repository and ownership state

Public benchmark provenance audit

Validation contract and next dependency

EvenSol commented on Oct 12, 2026

2026-10-12 flowline-hydraulics checkpoint — #4282 blocker narrowed to metadata integration

EvenSol commented on Oct 12, 2026

2026-10-12 public timing gate — shared draft PR #4313

Executable public evidence

Validation

Expansion boundary

EvenSol commented on Oct 13, 2026

2026-10-13 merged public-timing gate and next coupled-batch boundary

81 remaining items

EvenSol commented on Nov 3, 2026

2026-11-03 implementation ledger — finite-rate hydrocarbon pool transfer

EvenSol commented on Nov 3, 2026

2026-11-03 stage-3 scope freeze — wall-inventory shutdown/cooldown/restart

EvenSol commented on Nov 3, 2026

2026-11-03 shared implementation continuation — draft #4847

EvenSol commented on Nov 4, 2026

2026-11-04 merged audit — restart-screening prerequisite

EvenSol commented on Nov 4, 2026

2026-11-04 shared flowline ledger — draft PR #4865

EvenSol commented on Nov 4, 2026

2026-11-04 exact-head refinement addendum — PR #4865

EvenSol commented on Nov 5, 2026

Pigging benchmark/provenance checkpoint — 2026-11-05

EvenSol commented on Nov 5, 2026

2026-11-05 continuation scope — physical restart dependency

EvenSol commented on Nov 5, 2026

Pre-result numerical acceptance amendment — 2026-11-05

EvenSol commented on Nov 5, 2026

2026-11-05 continuation ledger — shared draft #4881

EvenSol commented on Nov 6, 2026

2026-11-06 continuation — merged restart audit and next shared branch

EvenSol commented on Nov 6, 2026

2026-11-06 pigging continuation ledger — coupled ideal-gas single-pig draft

End-to-end capability in this head

Validation

Claim boundary and documentation

Concrete split reasons / next dependency

EvenSol commented on Nov 6, 2026

2026-11-06 flowline extension ledger — PR #4894 expanded and published

EvenSol commented on Nov 7, 2026