Enhancements for the fixed CL mode #780

jayantmukho · 2019-08-28T07:13:51Z

Proposed Changes

This is a simple enhancement of the fixed CL mode. Currently the Fixed CL mode only exits when the specified iteration limit is reached. With these changes, the mode is a little more robust and checks for the specified Residual or Cauchy convergence and additionally checks that the CL is converged to the target CL value to within CAUCHY_EPS.

I have currently preserved the Finite Differencing that occurs at the end of the Fixed CL mode. I am not sure why this is required and would like to get rid of it if possible. If there is a reason for the finite differencing to calculate dCL/dAlpha, please let me know.

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.

I am submitting my contribution to the develop branch.
My contribution generates no new compiler warnings (try with the '-Wall -Wextra -Wno-unused-parameter -Wno-empty-body' compiler flags).
My contribution is commented and consistent with SU2 style.
I have added a test case that demonstrates my contribution, if necessary.

pcarruscag

LGTM
Thanks @jayantmukho.

talbring

@jayantmukho Thanks for the improvements! Though I have some comments, see below.

SU2_CFD/src/integration_time.cpp

jayantmukho · 2019-10-15T08:48:50Z

SU2_CFD/src/output/CFlowOutput.cpp

@@ -2820,10 +2820,11 @@ bool CFlowOutput::WriteVolume_Output(CConfig *config, unsigned long Iter){
      return true;
    }
  } else {
-    return ((Iter > 0) && Iter % config->GetVolume_Wrt_Freq() == 0);
+    if (config->GetFixed_CL_Mode() && config->GetFinite_Difference_Mode()) return false;
+    return ((Iter > 0) && Iter % config->GetVolume_Wrt_Freq() == 0) || force_writing;


This isn't the cleanest implementation but is the least intrusive as it is mostly contained in CFlowOutput. @talbring

The other way to do this (which I have a slight preference towards) would be to add an option to disable volume output completely in COutput itself. I would prefer it because it would be a more general function that could possibly be used in other situations as well. But it would be a pretty powerful option that might cause conflicts with other options/solvers if not handled with care.

Unfortunately, the way it is now, there is one state variable (Finite_Difference_Mode) that is stored in the config @pcarruscag . But everything else is contained in the solver

I am ok with that!

jayantmukho · 2019-10-15T08:52:17Z

SU2_CFD/src/output/CFlowCompOutput.cpp

+void CFlowCompOutput::SetAdditionalScreenOutput(CConfig *config){
+
+  if (config->GetFixed_CL_Mode()){
+    SetFixedCLScreenOutput(config);
+  } 
+}
+
+void CFlowCompOutput::SetFixedCLScreenOutput(CConfig *config){
+  PrintingToolbox::CTablePrinter FixedCLSummary(&cout);    
+
+  if (fabs(historyOutput_Map["CL_DRIVER_COMMAND"].value) > 1e-16){
+    FixedCLSummary.AddColumn("Fixed CL Mode", 40); 
+    FixedCLSummary.AddColumn("Value", 30);  


I moved all the output to the CFlowCompOutput. I had to make a few new functions but I tried to copy the way the WriteAdditionalFiles function works. I hope this is what you had in mind @talbring

It prints everytime the angle of attack changes and looks like this:

Yep looks good

jayantmukho · 2019-10-15T08:56:15Z

SU2_CFD/src/solver_direct_mean.cpp

@@ -7320,71 +7310,51 @@ void CEulerSolver::SetActDisk_BCThrust(CGeometry *geometry, CSolver **solver_con
 void CEulerSolver::SetFarfield_AoA(CGeometry *geometry, CSolver **solver_container,


This function got cleaned up, and now it does only SetFarfieldAoA (As opposed to perform all the logic for the Fixed CL mode). Moved all that to the FixedCL_Convergence function. Which now that I say out loud, should be name FixedCL_Driver or something of that sort

talbring

Thanks @jayantmukho. All of that looks good to me.

talbring · 2019-10-18T07:39:04Z

.travis.yml

@@ -72,11 +72,11 @@ install:

 before_script:
    # Get the test cases
-    - git clone --depth=1 -b develop https://github.com/su2code/TestCases.git ./TestData
+    - git clone --depth=1 -b enhancement_fixed_cl https://github.com/su2code/TestCases.git ./TestData


Dont forget to change that

talbring · 2019-10-18T07:41:13Z

SU2_CFD/include/output/CFlowOutput.hpp

   */
-  bool WriteVolume_Output(CConfig *config, unsigned long Iter) override;
+  bool WriteVolume_Output(CConfig *config, unsigned long Iter, bool force_writing) override;


I am fine with that change!

talbring · 2019-10-18T07:41:29Z

SU2_CFD/include/output/CFlowCompOutput.hpp

+   * \brief Write any additional output defined for the current solver.
+   * \param[in] config - Definition of the particular problem per zone.
+   */
+  void SetAdditionalScreenOutput(CConfig *config) override;


Good idea to add a routine like that.

talbring · 2019-10-18T07:41:46Z

SU2_CFD/include/solver_structure.hpp

-  su2double Cauchy_Value,  /*!< \brief Summed value of the convergence indicator. */
-  Cauchy_Func;      /*!< \brief Current value of the convergence indicator at one iteration. */
-  unsigned short Cauchy_Counter;  /*!< \brief Number of elements of the Cauchy serial. */
-  su2double *Cauchy_Serie;      /*!< \brief Complete Cauchy serial. */


Thanks for cleaning that up

talbring · 2019-10-18T07:43:59Z

SU2_CFD/src/output/CFlowOutput.cpp

@@ -2820,10 +2820,11 @@ bool CFlowOutput::WriteVolume_Output(CConfig *config, unsigned long Iter){
      return true;
    }
  } else {
-    return ((Iter > 0) && Iter % config->GetVolume_Wrt_Freq() == 0);
+    if (config->GetFixed_CL_Mode() && config->GetFinite_Difference_Mode()) return false;
+    return ((Iter > 0) && Iter % config->GetVolume_Wrt_Freq() == 0) || force_writing;


I am ok with that!

talbring · 2019-10-18T07:44:11Z

SU2_CFD/src/output/CFlowCompOutput.cpp

+void CFlowCompOutput::SetAdditionalScreenOutput(CConfig *config){
+
+  if (config->GetFixed_CL_Mode()){
+    SetFixedCLScreenOutput(config);
+  } 
+}
+
+void CFlowCompOutput::SetFixedCLScreenOutput(CConfig *config){
+  PrintingToolbox::CTablePrinter FixedCLSummary(&cout);    
+
+  if (fabs(historyOutput_Map["CL_DRIVER_COMMAND"].value) > 1e-16){
+    FixedCLSummary.AddColumn("Fixed CL Mode", 40); 
+    FixedCLSummary.AddColumn("Value", 30);  


Yep looks good

jayantmukho · 2019-10-22T23:50:31Z

In the process of the final merge which involves pulling corresponding TestCase and Tutorial branches into their develops as well. (and changing the travis.yml file)

jayantmukho added 9 commits July 9, 2019 16:39

improving the fixed_cl mode

e5824ba

Merge branch 'fix_set_ffd_script' into enhancement_fixed_cl

695b103

removed AoA update during the FD section

458d123

Merge branch 'develop' into enhancement_fixed_cl

79a93df

removed forcing residual convergence for fixed_cl mode

960c4ef

Merge branch 'develop' into enhancement_fixed_cl

f773594

revert to constant K

5d7a30c

Merge branch 'develop' into enhancement_fixed_cl

21ab4a5

clean up

db9387e

pr-triage bot added the PR: unreviewed label Aug 28, 2019

jayantmukho changed the title ~~Enhancement fixed cl~~ WIP Enhancement fixed cl Aug 28, 2019

pr-triage bot removed the PR: unreviewed label Aug 28, 2019

pcarruscag approved these changes Aug 30, 2019

View reviewed changes

WallyMaier added the PR: reviewed-approved label Aug 30, 2019

talbring requested changes Sep 2, 2019

View reviewed changes

SU2_CFD/src/integration_time.cpp Outdated Show resolved Hide resolved

SU2_CFD/src/integration_time.cpp Outdated Show resolved Hide resolved

pr-triage bot removed the PR: reviewed-approved label Sep 2, 2019

jayantmukho added 14 commits September 9, 2019 17:49

indentation fixes [ci skip]

e295750

Merge branch 'develop' into enhancement_fixed_cl

8e8af6b

fully functional, not pretty yet

da91901

some cleanup, works for normal driver and singleZoneDriver

ab89114

need to fix behavior when iteration limit is reached

4f6975b

fixed segault in testcase, more cleanup

0a4c4ff

New options added in TestCase [ci skip]

c644d0b

Fixed behavior for max iter and for when ITER_DCL_DALPHA=0

aa87eb1

making continuous adjoint implementation consistent

2716ba8

removed fixed cl from incompressible solver

9444f92

some changes to test case config file

af9cfee

added option to config file

50a57b9

changed regression test values

618d6f4

reverted integration_time.cpp back to develop status

db033ec

jayantmukho commented Oct 15, 2019

View reviewed changes

jayantmukho added 6 commits October 15, 2019 01:57

small fix [ci skip]

3ba26f2

Merge branch 'develop' into enhancement_fixed_cl

f4afeee

small fix [ci skip]

593b7b2

update testcase

775597c

reverting makefile

fd4d8c9

changed travis to point to corresponding TestCases branch

530cf67

talbring approved these changes Oct 18, 2019

View reviewed changes

pr-triage bot added PR: reviewed-approved and removed PR: unreviewed labels Oct 18, 2019

multipoint fixes

5600e3d

pr-triage bot added PR: unreviewed and removed PR: reviewed-approved labels Oct 18, 2019

jayantmukho added 7 commits October 21, 2019 13:44

fixing some flow.meta output issues

de09502

fixing regression values

3caeac6

small bug fixes

45d7e37

fix regression test vals

c65812b

Merge branch 'develop' into enhancement_fixed_cl

46da8cd

final clean ups

6a83788

Merge branch 'develop' into enhancement_fixed_cl

6682e40

reverting travis back to develop

388ba7d

jayantmukho merged commit 47c1f13 into develop Oct 23, 2019

jayantmukho deleted the enhancement_fixed_cl branch October 23, 2019 12:22

pr-triage bot added PR: merged and removed PR: unreviewed labels Oct 23, 2019

talbring added the changelog:feature label Nov 6, 2019

talbring changed the title ~~Enhancement fixed cl~~ Enhancements for the fixed CL mode Nov 8, 2019

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Enhancements for the fixed CL mode #780

Enhancements for the fixed CL mode #780

jayantmukho commented Aug 28, 2019

pcarruscag left a comment

talbring left a comment

jayantmukho Oct 15, 2019

talbring Oct 18, 2019

jayantmukho Oct 15, 2019

talbring Oct 18, 2019

jayantmukho Oct 15, 2019

talbring left a comment

talbring Oct 18, 2019

talbring Oct 18, 2019

talbring Oct 18, 2019

talbring Oct 18, 2019

talbring Oct 18, 2019

talbring Oct 18, 2019

jayantmukho commented Oct 22, 2019

		@@ -7320,71 +7310,51 @@ void CEulerSolver::SetActDisk_BCThrust(CGeometry geometry, CSolver *solver_con
		void CEulerSolver::SetFarfield_AoA(CGeometry geometry, CSolver *solver_container,

Enhancements for the fixed CL mode #780

Enhancements for the fixed CL mode #780

Conversation

jayantmukho commented Aug 28, 2019

Proposed Changes

PR Checklist

pcarruscag left a comment

Choose a reason for hiding this comment

talbring left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

talbring left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jayantmukho commented Oct 22, 2019