isothermal and adiabatic wall BCs for reacting flows#1604
Open
DimAdam-01 wants to merge 58 commits into
Open
Conversation
# Conflicts: # toolchain/mfc/test/cases.py
- s_convert_conservative_to_primitive_variables read optional t_step/stage without present() guards; m_time_steppers/m_data_output call it without them (UB on every 2D/3D chemistry run). Hoist a present()-guarded apply_ib_ghost_pres flag, gated on chemistry.and.ib.and.n>0. - s_interpolate_image_point wrote Ys_IP/T_IP (non-present optionals) on the bubbles/qbmm call paths; guard under chemistry. Guard caller T_GP/E_GP compute under chemistry; drop unused MW_GP. - Only allocate ghost_points_index/pressure_ghost_point when ib; push their host init to the device (ACC_SETUP_SFs is Cray-only, @:ALLOCATE creates uninitialized device memory on nvfortran) and reset the per-step ghost-point flag on-device instead of a host array assignment. - Remove dead #ifndef/#else duplicate declaration block.
The IB spanned the full domain width, so its top corners sat on the domain x-boundaries. Corner ghost points then had diagonal-normal image-point stencils that reflected into solid/ghost/buffer cells instead of clean fluid; interpolating temperature there is ill-conditioned and amplified inherent CPU/GPU differences (ghost-loop order + chemistry FP) into a ~9% energy divergence, failing the GPU golden. Widen the IB in x past the domain (length_x 0.05->0.06 in the test, Lx->1.2*Lx in the example) so its vertical faces sit outside [0, x_end]. Inside the domain the only IB surface is the horizontal top (normal +y), so every ghost point reflects straight into fluid. CPU vs GPU now agree to ~1e-11 relative (was 9.4%); regenerated golden. CPU and GPU tests both pass.
…over # Conflicts: # src/simulation/m_ibm.fpp # toolchain/mfc/test/cases.py
…st reset - The ghost-point image interpolation reaches s_interpolate_image_point via a bubbles/qbmm call path that omits the chemistry T_IP/Ys_IP optionals, so chemistry+ib+(bubbles_euler|qbmm) dereferenced non-present optionals. Prohibit the unsupported combination in the simulation checker (explicit named abort); chemistry then always takes the else call path, so the interpolation's chemistry guards are provably safe. - Only clear ghost_points_index over the whole domain for moving IBs. Static IBs keep the same ghost set every RK sub-stage and the device copy is zero-initialized, so the per-substage full-grid pass was pure overhead.
Contributor
There was a problem hiding this comment.
Pull request overview
Adds support for immersed boundary (IBM) walls in reacting-flow simulations, including isothermal and adiabatic wall behavior, by plumbing new IB thermal parameters through the toolchain, validation, MPI distribution, and the simulation-side ghost-point enforcement/conversion logic.
Changes:
- Introduces
patch_ib(*)%isothermalandpatch_ib(*)%twallparameters (toolchain definitions/descriptions + validation) and propagates them via MPI. - Extends IBM ghost-point correction to carry chemistry state (Ys/T) at image points and impose isothermal/adiabatic thermal behavior at ghost points, with pressure re-imposition during cons→prim conversion.
- Adds a regression test case (
CF9AFA00) and an example case for an IBM isothermal wall thermal flat plate.
Reviewed changes
Copilot reviewed 18 out of 19 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| toolchain/mfc/test/cases.py | Adds the IBM isothermal-wall chemistry regression case definition. |
| toolchain/mfc/params/validate.py | Tweaks unknown-attribute error truncation threshold. |
| toolchain/mfc/params/descriptions.py | Adds help-text patterns for patch_ib%isothermal and patch_ib%twall. |
| toolchain/mfc/params/definitions.py | Adds isothermal/twall to IB patch parameter definitions. |
| toolchain/mfc/case_validator.py | Validates patch_ib%isothermal/twall constraints under chemistry. |
| tests/CF9AFA00/golden-metadata.txt | Adds golden metadata for the new regression UUID. |
| src/simulation/m_rhs.fpp | Passes t_step/stage into cons→prim conversion from RHS path. |
| src/simulation/m_mpi_proxy.fpp | Broadcasts new IB thermal parameters in simulation. |
| src/simulation/m_ibm.fpp | Implements chemistry-aware IBM interpolation and thermal wall behavior; stores ghost pressure for later conversion. |
| src/simulation/m_global_parameters.fpp | Initializes new IB patch fields with defaults. |
| src/simulation/m_checker.fpp | Prohibits unsupported chemistry+IBM+bubbles/QBMM combination. |
| src/pre_process/m_mpi_proxy.fpp | Broadcasts new IB thermal parameters in pre_process. |
| src/pre_process/m_global_parameters.fpp | Initializes new IB patch fields with defaults. |
| src/pre_process/m_check_ib_patches.fpp | Adds pre_process-time twall validation for isothermal IB patches. |
| src/common/m_variables_conversion.fpp | Adds ghost-point bookkeeping fields and re-imposes ghost pressure during cons→prim for chemistry IBM. |
| src/common/m_derived_types.fpp | Extends IB patch derived type with isothermal and Twall. |
| src/common/include/2dHardcodedIC.fpp | Adjusts HCID=291 IC to set zero streamwise velocity below the wall. |
| examples/2D_ibm_Thermal_Flatplate/case.py | Adds a runnable example case for IBM isothermal wall with chemistry enabled. |
…ry-only fields Copilot review: - pressure_ghost_point was stored only on the moving_ibm<=1 branch, but ghost_points_index is tagged unconditionally, so for chemistry + force-driven moving IB (moving_ibm>1) the cons->prim override re-imposed a stale/zero pressure. Store the finalized ghost pressure once, after both branches. - ghost_points_index/pressure_ghost_point are only read when chemistry, so allocate/deallocate them under (ib .and. chemistry) and guard their writes (tag, per-step reset, pressure store) on chemistry — no full-grid integer+real fields or device updates for non-reacting IBM runs. Verified: CPU and GPU pass for the chemistry-IBM case and two non-chemistry IBM cases (the now-unallocated path).
sbryngelson
approved these changes
Jul 22, 2026
sbryngelson
previously approved these changes
Jul 22, 2026
…HYSICS_DOCS - case.md: document patch_ib(i)%isothermal and %twall (immersed-boundary isothermal/adiabatic walls), including the chemistry+diffusion requirement and the guidance to not place the patch flush with a domain boundary. - case_validator.py: add the missing PHYSICS_DOCS entry for check_chemistry so the auto-generated physics-constraints docs cover the num_fluids=1 and isothermal-wall (domain + IB) temperature requirements.
sbryngelson
approved these changes
Jul 22, 2026
Member
There was a problem hiding this comment.
Approving. Isothermal/adiabatic IB walls for reacting flows; CF9AFA00 passes CPU + GPU (V100) to 1e-9.
Takeover: fixed the optional-arg/device-init correctness bugs, prohibited the unsupported chemistry+IB+bubbles/QBMM combo, root-caused and fixed the CPU/GPU divergence (IB was flush with the domain boundary → widened it), resolved the Copilot review, and added docs. precheck clean.
sbryngelson
approved these changes
Jul 22, 2026
sbryngelson
left a comment
sbryngelson
left a comment
Member
There was a problem hiding this comment.
Approving. Verified: CF9AFA00 passes CPU + GPU (V100) to 1e-9; precheck clean; review resolved.
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## master #1604 +/- ##
==========================================
+ Coverage 59.62% 60.03% +0.40%
==========================================
Files 83 83
Lines 21246 21258 +12
Branches 3145 3143 -2
==========================================
+ Hits 12668 12762 +94
+ Misses 6467 6424 -43
+ Partials 2111 2072 -39 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Immersed boundaries now support chemistry, with isothermal and adiabatic wall boundary conditions at immersed surfaces (
patch_ib(i)%isothermal,patch_ib(i)%twall).Testing
2D thermal flat plate, bottom wall isothermal at 600 K:
Regression test
CF9AFA00(2D -> Chemistry -> IBM Isothermal Wall) passes on CPU and NVIDIA GPU (V100, OpenACC) to1e-9.Maintainer takeover
Finished by the maintainers on top of the original feature:
present()-guarded the new optionalt_step/stageargs in the cons→prim conversion; guarded theYs_IP/T_IPwrites on the bubbles/QBMM paths and prohibited the unsupportedchemistry + ib + (bubbles_euler | qbmm)combo; device-initializeghost_points_index/pressure_ghost_point(uninitialized on nvfortran,ACC_SETUP_SFsis Cray-only), allocate them only underib .and. chemistry, and cover the moving-IB pressure store.1e-11); golden regenerated.case.md+check_chemistryPHYSICS_DOCS) added; Copilot review resolved.Checklist
GPU changes