[WIP] Multizone adjoints for turbomachinery #2446

joshkellyjak · 2025-02-18T16:22:06Z

Proposed Changes

This is a cleaned up PR of the fixes needed for multizone adjoints for turbomachinery from the previous PR of @oleburghardt and I's work.

This hopefully is useable, so if you would like to test and report please feel free to contact me on Slack.

TODO:

Fix eddy viscosity tape issue in CNSSolver
Add turbomachinery objective functions + constraints
Include MZ testcase

Related Work

Now closed PR #2317

PR Checklist

Put an X by all that apply. You can fill this out after submitting the PR. If you have any questions, don't hesitate to ask! We want to help. These are a guide for you to know what the reviewers will be looking for in your contribution.

[ X ] I am submitting my contribution to the develop branch.
My contribution generates no new compiler warnings (try with --warnlevel=3 when using meson).
My contribution is commented and consistent with SU2 style (https://su2code.github.io/docs_v7/Style-Guide/).
I used the pre-commit hook to prevent dirty commits and used pre-commit run --all to format old commits.
I have added a test case that demonstrates my contribution, if necessary.
I have updated appropriate documentation (Tutorials, Docs Page, config_template.cpp), if necessary.

pcarruscag · 2025-02-20T08:51:45Z

SU2_CFD/include/solvers/CFVMFlowSolverBase.inl

+    AD::EndPreacc();
+
    /*--- Max is not differentiable, so we not register them for preacc. ---*/


What is the difference between ending the preaccumulation before or after?

pcarruscag · 2025-02-20T08:51:59Z

Common/src/CConfig.cpp

    }
+  } else if (Multizone_Problem && DiscreteAdjoint) {
+      SU2_MPI::Error(string("OUTPUT_WRT_FREQ cannot be specified for this solver "


Suggested change

SU2_MPI::Error(string("OUTPUT_WRT_FREQ cannot be specified for this solver "

SU2_MPI::Error("OUTPUT_WRT_FREQ cannot be specified for this solver "

pcarruscag · 2025-02-20T08:52:47Z

Common/src/CConfig.cpp

+                            "(writing of restart and sensitivity files not possible for multizone discrete adjoint during runtime yet).\n"
+                            "Please remove this option from the config file, output files will be written when solver finalizes.\n"), CURRENT_FUNCTION);


Hmmm I'm pretty sure it is possible

This is one for @oleburghardt

Yes and no, writing the files is fine, but the continuation of the multi-zone discrete adjoint solver is erratic. My guess is that some indices aren't cleared properly before re-recording the tape. (Writing adjoint solution variables to file only, without re-recording at all and without sensitivities, might give us some hints.)
The debug mode could eventually pin down where exactly things go wrong.

pcarruscag · 2025-02-20T08:55:35Z

SU2_CFD/src/solvers/CDiscAdjSolver.cpp

-    BPressure = config->GetPressureOut_BC();
-    Temperature = config->GetTotalTemperatureIn_BC();
-
    if (!reset){
      AD::RegisterInput(BPressure);
      AD::RegisterInput(Temperature);
    }

+    BPressure = config->GetPressureOut_BC();
+    Temperature = config->GetTotalTemperatureIn_BC();
+


Why are these wrong but not the others that follow the same pattern?

This threw a tagging error, and if I remember correctly it's because in the other sections the variables it accesses are directly used in the solver, whereas in the Giles implementation they are not, creating a mismatch in tags.

Can we fix it in a way that keeps a clear pattern for doing this type of thing?

I can give it a go

…ding of objective function calculation

…e_mz_turbomachinery_adjoint

SU2_CFD/src/iteration/CDiscAdjFluidIteration.cpp

+        CGeometry::ComputeWallDistance(config, geometry);
+      }
+      geometry[iZone][MESH_0][INST_0]->InitTurboVertexAdj(geometry, config);
+      for (auto iZone = 0u; iZone < nZone; iZone++) {


To fix the issue, we will rename the local variable iZone on line 493 to a distinct name, such as zoneIndex. This will eliminate the shadowing of the parameter iZone and make the code more readable. The change will only affect the loop declaration and any references to the local variable within the loop body.

SU2_CFD/src/iteration/CTurboIteration.cpp

+    TurbomachineryStagePerformance->ComputePerformanceStage(InState, OutState, config_container[nZone-1]);
+
+    /*--- Set turbomachinery objective function value in each zone ---*/
+    for (auto iBlade = 0u; iBlade < nBladesRow; iBlade++) {


To fix the issue, we will rename the iBlade variable declared on line 101 to a distinct name, such as bladeIndex. This will eliminate the shadowing of the iBlade variable declared on line 72. The change will be localized to the for loop on line 101 and its body, ensuring that the functionality remains unchanged while improving code clarity.

SU2_CFD/src/solvers/CEulerSolver.cpp

+    // case ENTROPY_GENERATION:
+    //   Total_ComboObj+=Weight_ObjFunc*GetTurboObjectiveFunction(Kind_ObjFunc, config->GetnMarker_Turbomachinery());
+    //   break;
+    // case TOTAL_PRESSURE_LOSS:
+    //   Total_ComboObj+=Weight_ObjFunc*GetTurboObjectiveFunction(Kind_ObjFunc, config->GetnMarker_Turbomachinery());
+    //   break;
+    // case KINETIC_ENERGY_LOSS:
+    //   Total_ComboObj+=Weight_ObjFunc*GetTurboObjectiveFunction(Kind_ObjFunc, config->GetnMarker_Turbomachinery());
+    //   break;


The best way to fix the problem is to either remove the commented-out code entirely or reinstate it if it is still relevant and functional. In this case, the commented-out code appears to be related to specific objective functions (ENTROPY_GENERATION, TOTAL_PRESSURE_LOSS, KINETIC_ENERGY_LOSS) that are currently not being used. If these cases are no longer relevant, they should be removed. If they are still relevant, they should be reinstated and properly integrated into the switch statement.

For this fix, we will remove the commented-out code since there is no indication that it is required or will be used in the future.

Fixes from tagging tool

98989f7

joshkellyjak added PR: draft changelog:feature labels Feb 18, 2025

Config catch + fix preacc in ComputeVorticityandStrainMag

6082eac

oleburghardt mentioned this pull request Feb 19, 2025

[WIP] Multizone discrete adjoints for turbomachinery #2317

Closed

pcarruscag reviewed Feb 20, 2025

View reviewed changes

Initial add turbo obj functions

b3267b6

joshkellyjak mentioned this pull request Feb 25, 2025

Update CTurbSSTSolver.cpp to fix bug on calculating MuT #2453

Merged

6 tasks

Joshua Kelly and others added 6 commits March 11, 2025 16:24

Initial refactoring of turboperformance into solver for correct recor…

c459ce9

…ding of objective function calculation

Resolve preaccumulation error in SetRoe_Dissipation_FD.

4963a99

Add RegisterContainer for active vectors.

cc1de15

Compatibility with develop.

48e519f

Add eddy viscosity as recording input/output of RANS solvers.

ddc7ab4

Move registration of muT for RANS to new RegisterComplementary.

f77a773

oleburghardt mentioned this pull request Mar 21, 2025

Fix preaccumulation error, improve recording of RANS solvers #2472

Merged

6 tasks

Joshua Kelly added 2 commits March 21, 2025 13:06

Merge remote-tracking branch 'origin/fix_small_AD_issues' into featur…

4a58b8c

…e_mz_turbomachinery_adjoint

Adding geometry to AD recording for turbo calcs. (NOT WORKING CURRENTLY)

1614784

github-advanced-security bot found potential problems May 1, 2025

View reviewed changes

@@ -492,4 +492,4 @@
                   geometry[iZone][MESH_0][INST_0]->InitTurboVertexAdj(geometry, config);
-                  for (auto iZone = 0u; iZone < nZone; iZone++) {
-                    solver[iZone][INST_0][MESH_0][FLOW_SOL]->InitTurboContainers(geometry[iZone][INST_0][MESH_0],config[iZone]);
+                  for (auto zoneIndex = 0u; zoneIndex < nZone; zoneIndex++) {
+                    solver[zoneIndex][INST_0][MESH_0][FLOW_SOL]->InitTurboContainers(geometry[zoneIndex][INST_0][MESH_0],config[zoneIndex]);
                   }

@@ -100,10 +100,10 @@
                 /*--- Set turbomachinery objective function value in each zone ---*/
-                for (auto iBlade = 0u; iBlade < nBladesRow; iBlade++) {
+                for (auto bladeIndex = 0u; bladeIndex < nBladesRow; bladeIndex++) {
                   // Should we set in ZONE_0 or nZone-1?
-                  auto iBladePerf = TurbomachineryPerformance->GetBladesPerformances().at(iBlade).at(nSpan);
+                  auto iBladePerf = TurbomachineryPerformance->GetBladesPerformances().at(bladeIndex).at(nSpan);
                   InState = iBladePerf->GetInletState();
                   OutState = iBladePerf->GetOutletState();
-                  solver[iBlade][INST_0][MESH_0][FLOW_SOL]->SetTurboObjectiveFunction(ENUM_OBJECTIVE::ENTROPY_GENERATION, iBlade, TurbomachineryStagePerformance->GetNormEntropyGen()*100);
-                  solver[iBlade][INST_0][MESH_0][FLOW_SOL]->SetTurboObjectiveFunction(ENUM_OBJECTIVE::TOTAL_PRESSURE_LOSS, iBlade, iBladePerf->GetTotalPressureLoss());
-                  solver[iBlade][INST_0][MESH_0][FLOW_SOL]->SetTurboObjectiveFunction(ENUM_OBJECTIVE::KINETIC_ENERGY_LOSS, iBlade, iBladePerf->GetKineticEnergyLoss());
+                  solver[bladeIndex][INST_0][MESH_0][FLOW_SOL]->SetTurboObjectiveFunction(ENUM_OBJECTIVE::ENTROPY_GENERATION, bladeIndex, TurbomachineryStagePerformance->GetNormEntropyGen()*100);
+                  solver[bladeIndex][INST_0][MESH_0][FLOW_SOL]->SetTurboObjectiveFunction(ENUM_OBJECTIVE::TOTAL_PRESSURE_LOSS, bladeIndex, iBladePerf->GetTotalPressureLoss());
+                  solver[bladeIndex][INST_0][MESH_0][FLOW_SOL]->SetTurboObjectiveFunction(ENUM_OBJECTIVE::KINETIC_ENERGY_LOSS, bladeIndex, iBladePerf->GetKineticEnergyLoss());
                 }

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[WIP] Multizone adjoints for turbomachinery #2446

[WIP] Multizone adjoints for turbomachinery #2446

joshkellyjak commented Feb 18, 2025

Uh oh!

pcarruscag Feb 20, 2025

Uh oh!

pcarruscag Feb 20, 2025

Uh oh!

pcarruscag Feb 20, 2025

Uh oh!

joshkellyjak Feb 20, 2025

Uh oh!

oleburghardt Mar 25, 2025 •

edited

Loading

Uh oh!

pcarruscag Feb 20, 2025

Uh oh!

joshkellyjak Feb 20, 2025

Uh oh!

pcarruscag Feb 20, 2025

Uh oh!

joshkellyjak Feb 20, 2025

Uh oh!

Check notice

Copilot Autofix

Check notice

Copilot Autofix

Check notice

Copilot Autofix

Uh oh!

		AD::EndPreacc();

		/--- Max is not differentiable, so we not register them for preacc. ---/

	SU2_MPI::Error(string("OUTPUT_WRT_FREQ cannot be specified for this solver "
	SU2_MPI::Error("OUTPUT_WRT_FREQ cannot be specified for this solver "

		"(writing of restart and sensitivity files not possible for multizone discrete adjoint during runtime yet).\n"
		"Please remove this option from the config file, output files will be written when solver finalizes.\n"), CURRENT_FUNCTION);

[WIP] Multizone adjoints for turbomachinery #2446

Are you sure you want to change the base?

[WIP] Multizone adjoints for turbomachinery #2446

Conversation

joshkellyjak commented Feb 18, 2025

Proposed Changes

Related Work

PR Checklist

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

oleburghardt Mar 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Check notice

Uh oh!

Copilot Autofix

Check notice

Uh oh!

Copilot Autofix

Check notice

Copilot Autofix

Uh oh!

oleburghardt Mar 25, 2025 •

edited

Loading