## OBSERVED FAILURE The v5 run completed to its declared end label, `Time = 5000s`, and emitted the solver marker `End`. It did **not** satisfy the frozen residual criteria. At time 5000: | Field | Initial residual | Frozen threshold | Status | |---|---:|---:|---| | Ux | 9.23×10⁻⁵ | 1×10⁻⁵ | FAIL | | Uy | 1.07×10⁻⁴ | 1×10⁻⁵ | FAIL | | Uz | 1.15×10⁻⁴ | 1×10⁻⁵ | FAIL | | h | 6.73×10⁻⁵ | 1×10⁻⁵ | FAIL | | e | 1.82×10⁻⁴ | 1×10⁻⁶ | FAIL | | p_rgh | 3.87×10⁻⁵ | 1×10⁻⁴ | PASS | | k | 4.80×10⁻⁵ | 1×10⁻⁴ | PASS | | omega | 3.77×10⁻⁶ | 1×10⁻⁴ | PASS | These are equation initial residuals, not the much smaller final linear-solver residuals. Run completion and numerical convergence are therefore separate: - Solver end marker: observed. - Process exit result: not retained or otherwise verified. - Residual convergence: FAIL. - Design readiness: NOT_ESTABLISHED. ## IMPORTANT EVIDENCE - `attempts/v5/solver_v5.log`, time 5000: - Solid `e` initial residual: `0.00018234749`. - `Ux`, `Uy`, `Uz`: `9.2288626e-05`, `0.00010684136`, `0.00011485148`. - `h`: `6.7264055e-05`. - `p_rgh`: `3.8669461e-05`. - `omega`: `3.7717048e-06`. - `k`: `4.7960261e-05`. - Solid temperature range: 360.34–538.77 K. - Fluid temperature range: 139.8–505.07 K. - End marker: `End`. - `v5_case/system/fluid/fvSolution`, `residualControl`: - `p_rgh 1e-4`, `U 1e-5`, `h 1e-5`, `k 1e-4`, `omega 1e-4`. - One outer corrector; two pressure correctors. - `v5_case/system/solid/fvSolution`, `residualControl`: - `e 1e-6`. - `v5_case/system/controlDict`: - Steady calculation represented by `deltaT 1`, `adjustTimeStep no`, `endTime 5000`. - Only retained function-object diagnostic is `yPlus`. - `v5_case/postProcessing/fluid/yPlus/1260/yPlus.dat`, time 5000: - Minimum 36.22, maximum 289.78, average 100.62. - `attempts/v4/solver_v4.log`: - Fatal termination at label 1277 because `Hs` temperature coordinate reached 400.036 K outside the former 120–400 K table. - `attempts/v5/methane_HEOS_property_table.csv`: - Retained table spans pressures 5–12 MPa and temperatures 120–600 K. - `v5_case/0/fluid/U`: - Prescribed inlet mass flow `0.0113 kg/s`. - `v5_case/0/fluid/T`: - Prescribed inlet temperature `139.8 K`. - `v5_case/constant/fluid/pRef`: - Reference pressure `7.86 MPa`. - `v5_case/constant/fluid/momentumTransport`: - RAS `kOmegaSST`. - `v5_case/constant/solid/physicalProperties`: - Solid conductivity `350 W/(m K)`. - Retained mesh headers: - Fluid: 11,520 cells. - Solid: 58,880 cells. - Combined: 70,400 cells. ## KNOWN FACTS 1. The prescribed inlet mass flow, inlet temperature, outlet/reference pressure, SST model, and solid conductivity correspond to source-method inputs. 2. v5 avoided the earlier 400 K property-table limit by using a table extending to 600 K. 3. Thermal fields were still changing substantially: - Solid maximum temperature: 431.18 K at 1261, 461.59 K at 2000, 492.13 K at 3000, 516.96 K at 4000, 538.77 K at 5000. - Fluid maximum temperature: 399.12, 428.79, 458.84, 483.42, and 505.07 K at the same labels. 4. Solid-energy residual decreased from 3.26×10⁻⁴ at 1261 to 1.82×10⁻⁴ at 5000, but remained approximately 182 times its threshold. 5. Enthalpy residual decreased only from 8.43×10⁻⁵ to 6.73×10⁻⁵ and remained about 6.7 times its threshold. 6. Velocity residuals improved greatly, especially Uy, but all components remained above threshold at 5000. 7. The final y+ distribution is nowhere near the source’s approximate y+≈1 target. ## UNKNOWN/UNVERIFIED ITEMS - Actual process exit code. - Global energy balance: NOT_ESTABLISHED. - Measured inlet/outlet mass flows and mismatch: NOT_ESTABLISHED. A prescribed inlet value and continuity residuals are not an inlet/outlet balance. - Monitor stability: no declared acceptance threshold and no retained engineering monitor histories beyond solver quantities and y+. - Mesh-quality report: NOT_EVALUATED. - Mesh independence: NOT_ESTABLISHED; only one v5 grid is available. - Exact first-cell height and wall-normal refinement distribution. - Suitability of the wall functions for the observed y+ distribution: NOT_ESTABLISHED because no frozen suitability criterion exists. - Roughness implementation in the v5 case. - Eligible experimental result comparison: NOT_EVALUATED. - Measured outlet temperature, inlet pressure, pressure drop, or wall-temperature data. - Whether the reconstructed geometry accurately represents the manufactured lower-corner radius, electrodeposition, 90-degree inlet turn, and half-channel source model. - Validation and model fidelity. Neither follows from solver completion or residual convergence. ## REFERENCE-METHODOLOGY OBSERVATIONS The methodology extract reports: - Structured hexahedral meshes refined toward walls. - An approximate y+ target of 1. - SST-family turbulence modeling. - A half-channel symmetry model with a 90-degree inlet turn. - Measured mass flow and inlet temperature, with measured outlet static pressure. - Real-gas property tables. - Polished-wall equivalent sand roughness of approximately 1.1 μm. - Copper-alloy conductivity of 350 W/(m·K). - Calorimetric and inverse-method heat-flux alternatives. The source methodology does not supply an eligible validation target in this evidence packet. Its setup values cannot establish agreement with experiment. ## RANKED CASE-SPECIFIC HYPOTHESES ### 1. The v5 run stopped during a slow, continuing thermal transient in iteration space **Support** - Solid and fluid maximum temperatures rose monotonically through the retained checkpoints, including increases of about 21.9 K and 21.6 K respectively between 4000 and 5000. - Solid `e` and fluid `h` residuals declined slowly but remained above threshold. - No thermal plateau is evident at 5000. - Velocity residuals changed as the evolving density and temperature fields continued to alter the flow. **Contradiction** - Some fields already meet their thresholds. - Slow residual reduction alone does not prove eventual convergence; a residual floor or non-steady solution remains possible. **Missing evidence** - Heat-rate, outlet-temperature, pressure-drop, and wall-temperature monitor histories. - A predeclared trend/plateau criterion. **Discriminating test** Restart from the preserved v5 state with **all physical and numerical settings unchanged** and run a fixed 200 additional iterations solely as a diagnostic. This is not permission to redefine `endTime` as an acceptance criterion. Freeze before execution: - Record `e`, `h`, all U components, maximum solid/fluid temperature, outlet bulk temperature, and inlet/outlet mass flow every 20 iterations. - Compare the 4800–5000 and 5000–5200 slopes. - Do not alter thresholds. **Prediction if correct** Temperatures and outlet quantities will retain a consistent directional slope, while `e` and `h` continue systematic decay rather than fluctuating around stationary levels. **Prediction if wrong** Thermal quantities will plateau or oscillate without meaningful drift, while failed residuals remain at approximately fixed levels. ### 2. One outer fluid-solid coupling corrector is producing a segregated-coupling residual floor **Support** - `system/fvSolution` declares only one outer corrector. - Thermal coupling is strong: the final fluid and solid maxima exceed 500 K while the inlet is 139.8 K. - Individual linear equations reach small final residuals, yet their next-iteration initial residuals remain above the global criteria. That pattern is consistent with inter-equation or inter-region coupling error dominating individual linear-solver error. - Pressure repeatedly reaches its `maxIter 100` on the first correction near the end, although its initial residual passes the frozen global threshold. **Contradiction** - Residuals and temperatures still exhibit secular trends, which may be explained entirely by Hypothesis 1. - No controlled outer-corrector comparison exists. **Missing evidence** - Interface heat imbalance by iteration. - Outer-corrector residual histories. **Discriminating test** Only after the unchanged continuation test, compare two preserved-state branches over the same fixed 100 iterations: control with one outer corrector versus treatment with two. Change no other setting. **Prediction if correct** Two outer correctors will cause a prompt, repeatable reduction in `e`, `h`, and U initial residuals per iteration and reduce fluid-solid interface imbalance relative to the control. **Prediction if wrong** Both branches will show comparable physical-monitor slopes and residual decay when normalized by work; the added corrector will mainly increase cost. ### 3. The wall mesh and wall treatment are inconsistent with the source-method near-wall approach **Support** - Final interface y+ is 36–290, average about 101. - The source method reports wall-refined structured grids targeting approximately y+=1. - The case uses `omegaWallFunction` and `kqRWallFunction`, whereas the source-method comparison requires resolving or otherwise consistently treating the near-wall region. - Heat transfer is central to this conjugate problem, making near-wall treatment potentially influential. **Contradiction** - High y+ does not by itself cause residual nonconvergence. - No frozen wall-resolution acceptance criterion exists. - The current wall functions may be numerically intended for a wall-modeled regime, although suitability is unverified. **Missing evidence** - First-cell spacing, growth ratio, roughness boundary condition, mesh-quality report, and a wall-normal refinement comparison. **Discriminating test** Perform one predeclared wall-normal refinement study that targets approximately y+=1 while keeping geometry, boundary conditions, properties, turbulence family, and convergence thresholds unchanged. This must be treated as a separate grid, not as a repair that validates the baseline. **Prediction if correct** The refined grid will produce y+ near the source target and materially alter wall heat transfer, wall/fluid temperatures, or pressure drop beyond the predeclared comparison tolerance. **Prediction if wrong** The refined grid will change y+ substantially but leave the engineering observables within the predeclared comparison tolerance. ### 4. The expanded property table removed a crash condition but may only postpone another range failure **Support** - v4 failed immediately after the fluid reached 400.036 K, just outside its property-table temperature range. - v5 reached a fluid maximum of 505.07 K and a solid maximum of 538.77 K while still heating. - The v5 methane table ends at 600 K. **Contradiction** - Fluid temperature remained roughly 95 K below the retained table maximum at 5000. - No v5 out-of-range error occurred. - Solid temperature is not itself the methane-table lookup coordinate. **Missing evidence** - Predicted terminal fluid temperature and local pressure-temperature trajectory. - Independent verification of table accuracy throughout 5–12 MPa and 120–600 K. **Discriminating test** Before any longer production run, predeclare a range-margin diagnostic: continue the unchanged case for the fixed 200-iteration test and track maximum fluid temperature and its slope against the 600 K boundary. **Prediction if correct** Maximum fluid temperature will continue rising at a rate indicating loss of adequate table margin before a steady state is reached. **Prediction if wrong** The temperature slope will decay with substantial margin remaining below 600 K. ## RECOMMENDED FIRST EXPERIMENT Run the **unchanged 200-iteration continuation from the preserved v5 state**, with frozen monitor outputs and predictions as specified under Hypothesis 1. This has the highest information value because it: - Changes no model, boundary condition, material property, mesh, relaxation factor, or acceptance threshold. - Directly distinguishes unfinished secular thermal evolution from a stationary residual floor. - Establishes whether a subsequent outer-corrector experiment is justified. - Quantifies property-table margin without changing the table. - Preserves v5 as a failed baseline. - Is substantially smaller than a mesh-refinement study. Stop after the fixed 200 iterations. Do not continue merely because a residual appears close to passing. Interpret the result against the frozen slope predictions; retain the run regardless of outcome. ## CONFIDENCE AND UNCERTAINTY - **High confidence:** v5 completed its declared iteration range but failed the frozen residual criteria. - **High confidence:** the retained thermal fields had not reached an evident stationary plateau by 5000. - **High confidence:** wall resolution differs strongly from the source’s approximate y+≈1 method. - **Moderate confidence:** incomplete thermal settling is the dominant immediate explanation. - **Low-to-moderate confidence:** one outer corrector is independently limiting convergence; a controlled comparison is required. - **Low confidence:** the evidence supports any validation, design-readiness, or physical-accuracy conclusion. The evidence supports a ranked diagnosis and a controlled first experiment, but it does not yet support a unique root cause.