Implementation of Simplified LM Transition model #1901

rois1995 · 2023-01-27T18:26:29Z

Proposed Changes

Hi everyone! I am starting to implement the Simplified version of the Langtry-Menter transition model following the work in https://doi.org/10.2514/6.2012-672. More updates will follow.

Related Work

This follows the implementation of the LM model as #1751.

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 have added a test case that demonstrates my contribution, if necessary.
I have updated appropriate documentation (Tutorials, Docs Page, config_template.cpp), if necessary.

rois1995 · 2023-01-27T18:28:50Z

Technically, this model has only one equation, the one for intermittency. Most of the functions are the same as the original LM model. Should I consider the Simplified model (SLM) as an option for the LM model to avoid duplicates? This is how I started, but I am open to discussions and suggestions.

pcarruscag

If the changes to the solver don't become too intrusive this approach sounds good to me

SU2_CFD/src/solvers/CTransLMSolver.cpp

- Added correlations for Simplified LM. - There is a bug on the computation of grad(n*U)*n

rois1995 · 2023-01-31T17:28:58Z

I have a question: for each point in the mesh I am trying to compute the dot product between the velocity vector and the normal to the wall of the nearest point on the wall. How do I access such information? I've found that in the CPoint class I have the ClosestWall_Elem variable which stores the index of the closest element on a wall. However, when I try to assess the information with a number of cores greater than 2, it crashes. Moreover, to recover the normal of the element I perform a mean of the normals on the nodes of that element. Is there a structure that has the normals saved for each element of the primal grid? The part that I am referring to is from line 208 in CTransLMSolver.cpp .

SU2_CFD/include/numerics/turbulent/transition/trans_correlations.hpp

+    switch (options.Correlation_SLM) {
+      case TURB_TRANS_CORRELATION_SLM::MENTER_SLM: {
+
+        /*-- Thwaites parameter ---*/
+        su2double  lambda_theta_local = 7.57e-3 * du_ds * wall_dist * wall_dist * Density / Laminar_Viscosity + 0.0128;
+        lambda_theta_local = min(max(lambda_theta_local, -1.0), 1.0);
+
+        /*-- Function to sensitize the transition onset to the streamwise pressure gradient ---*/
+        su2double FPG = 0.0;
+        const su2double C_PG1 = 14.68;
+        const su2double C_PG1_lim = 1.5;
+        const su2double C_PG2 = -7.34;
+        const su2double C_PG2_lim = 3.0;
+        const su2double C_PG3 = 0.0;
+        if (lambda_theta_local >= 0.0) {
+          FPG = min(1+ C_PG1 * lambda_theta_local, C_PG1_lim);
+        } else {
+          const su2double FirstTerm = C_PG2 * lambda_theta_local;
+          const su2double SecondTerm = C_PG3 * min(lambda_theta_local + 0.0681, 0.0);
+          FPG = min(1 + FirstTerm + SecondTerm, C_PG2_lim);
+        }
+
+        FPG = max(FPG, 0.0);
+
+        const su2double C_TU1 = 100.0;
+        const su2double C_TU2 = 1000.0;
+        const su2double C_TU3 = 1.0;
+        rethetac = C_TU1 + C_TU2 * exp(-C_TU3 * Tu_L * FPG);
+
+        break;
+      } case TURB_TRANS_CORRELATION_SLM::CODER_SLM: {
+
+        /*-- Local pressure gradient parameter ---*/
+        const su2double H_c = max(min(wall_dist * VorticityMag / VelocityMag, 1.1542), 0.3823);
+
+        /*-- Thwaites parameter ---*/
+        su2double lambda_theta_local = 0.0;
+        const su2double H_c_delta = 0.587743 - H_c;
+        if ( H_c >= 0.587743 ) {
+          const su2double FirstTerm = 0.1919 * pow(H_c_delta, 3.0);
+          const su2double SecondTerm = 0.4182 * pow(H_c_delta, 2.0);
+          const su2double ThirdTerm = 0.2959 * H_c_delta;
+          lambda_theta_local = FirstTerm + SecondTerm + ThirdTerm;
+        } else {
+          const su2double FirstTerm = 4.7596 * pow(H_c_delta, 3.0);
+          const su2double SecondTerm = -0.3837 * pow(H_c_delta, 2.0);
+          const su2double ThirdTerm = 0.3575 * H_c_delta;
+          lambda_theta_local = FirstTerm + SecondTerm + ThirdTerm;
+        }
+
+        /*-- Function to sensitize the transition onset to the streamwise pressure gradient ---*/
+        su2double FPG = 0.0;
+        if (lambda_theta_local <= 0.0) {
+          const su2double FirstTerm = -12.986 * lambda_theta_local;
+          const su2double SecondTerm = -123.66 * pow(lambda_theta_local, 2.0);
+          const su2double ThirdTerm = -405.689 * pow(lambda_theta_local, 3.0);
+          FPG = 1 - (FirstTerm + SecondTerm + ThirdTerm) * exp(-pow(Tu_L/1.5,1.5));
+        } else {
+          FPG = 1 + 0.275 * (1 - exp(-35.0 * lambda_theta_local)) * exp(-Tu_L/0.5);
+        }
+
+        // This is not reported in the paper
+        //FPG = max(FPG, 0.0);
+
+        const su2double C_TU1 = 100.0;
+        const su2double C_TU2 = 1000.0;
+        const su2double C_TU3 = 1.0;
+        rethetac = C_TU1 + C_TU2 * exp(-C_TU3 * Tu_L * FPG);
+
+        break;
+      } case TURB_TRANS_CORRELATION_SLM::MOD_EPPLER_SLM: {
+
+        /*-- Local pressure gradient parameter ---*/
+        const su2double H_c = max(min(wall_dist * VorticityMag / VelocityMag, 1.1542), 0.3823);
+
+        /*-- H_32 Shape factor --*/
+        const su2double H_32 = 1.515095 + 0.2041 * pow((1.1542 - H_c), 2.0956);
+
+        rethetac = exp(127.94 * pow((H_32-1.515095), 2.0) + 6.774224);
+
+        break;
+      }
+      case TURB_TRANS_CORRELATION_SLM::DEFAULT:
+        SU2_MPI::Error("Transition correlation for Simplified LM model is set to DEFAULT but no default value has ben set in the code.",
+                       CURRENT_FUNCTION);
+        break;
+    }


SU2_CFD/include/numerics/turbulent/transition/trans_sources.hpp

SU2_CFD/src/solvers/CTransLMSolver.cpp

SU2_CFD/include/numerics/turbulent/transition/trans_correlations.hpp

 */

 #pragma once
+//#include <cmath>


The best fix is to remove the line //#include <cmath> from SU2_CFD/include/numerics/turbulent/transition/trans_correlations.hpp. This eliminates the commented-out code and improves the code’s readability and maintainability. There is no justification provided for retaining the commented-out include, and there is no indication within the provided code that <cmath> is needed. No further imports, methods, or code changes are necessary.

- Modified LM_OPTIONS to include cross-flow effects: from LM2015 to CROSSFLOW

pcarruscag · 2023-01-31T19:25:16Z

See what is done at the bottom of CGeometry.cpp in CGeometry::ComputeWallDistance.
You need to communicate the normals you need first.

github-advanced-security

CodeQL found more than 10 potential problems in the proposed changes. Check the Files changed tab for more details.

WallyMaier · 2023-02-01T05:47:54Z

Common/include/geometry/dual_grid/CPoint.hpp

+   * \brief Get the index of the closest wall element.
+   * \param[in] iPoint - Index of the point.
+   */
+  inline unsigned long GetClosestWall_Elem(unsigned long iPoint) {return ClosestWall_Elem(iPoint);}


Can you add the param[out] for these please

Sure, I'll do it now.

WallyMaier · 2023-02-01T05:50:10Z

Common/include/option_structure.hpp

    SU2_MPI::Error("Two correlations selected for LM_OPTIONS. Please choose only one.", CURRENT_FUNCTION);
  }
+  if (NFoundCorrelations_SLM > 1) {
+    SU2_MPI::Error("Two correlations selected for Simplified model into LM_OPTIONS. Please choose only one.", CURRENT_FUNCTION);


Suggested change

SU2_MPI::Error("Two correlations selected for Simplified model into LM_OPTIONS. Please choose only one.", CURRENT_FUNCTION);

SU2_MPI::Error("Two correlations selected for simplified LM_OPTIONS. Please choose only one.", CURRENT_FUNCTION);

I'd stick with my change since the options are for the simplified model. They are not simplified options.

WallyMaier · 2023-02-01T05:50:50Z

Common/src/CConfig.cpp


-    /*--- Check if problem is 2D and LM2015 has been selected ---*/
-    if (lmParsedOptions.LM2015 && val_nDim == 2) {
-      SU2_MPI::Error("LM2015 is available only for 3D problems", CURRENT_FUNCTION);


is LM2015 gone? or is crossflow the same?

I have changed the option to CROSSFLOW, since I will use it also for the Simplified model.

WallyMaier · 2023-02-01T05:54:18Z

SU2_CFD/include/numerics/turbulent/transition/trans_correlations.hpp

+        // This is not reported in the paper
+        //FPG = max(FPG, 0.0);
+


Suggested change

// This is not reported in the paper

//FPG = max(FPG, 0.0);

I have left it there since I do not know if I have to keep it or not.

bigfooted · 2023-02-10T16:02:42Z

I have a question: for each point in the mesh I am trying to compute the dot product between the velocity vector and the normal to the wall of the nearest point on the wall. How do I access such information? I've found that in the CPoint class I have the ClosestWall_Elem variable which stores the index of the closest element on a wall. However, when I try to assess the information with a number of cores greater than 2, it crashes. Moreover, to recover the normal of the element I perform a mean of the normals on the nodes of that element. Is there a structure that has the normals saved for each element of the primal grid? The part that I am referring to is from line 208 in CTransLMSolver.cpp .

There is the geometry toolbox for dot product and normal:
https://github.com/su2code/SU2/blob/master/Common/include/toolboxes/geometry_toolbox.hpp
just look in the code for examples, in vscode you can search for GeometryToolbox::DotProduct or GeometryToolbox::Norm

rois1995 · 2023-02-14T11:50:53Z

I have a question: for each point in the mesh I am trying to compute the dot product between the velocity vector and the normal to the wall of the nearest point on the wall. How do I access such information? I've found that in the CPoint class I have the ClosestWall_Elem variable which stores the index of the closest element on a wall. However, when I try to assess the information with a number of cores greater than 2, it crashes. Moreover, to recover the normal of the element I perform a mean of the normals on the nodes of that element. Is there a structure that has the normals saved for each element of the primal grid? The part that I am referring to is from line 208 in CTransLMSolver.cpp .

There is the geometry toolbox for dot product and normal: https://github.com/su2code/SU2/blob/master/Common/include/toolboxes/geometry_toolbox.hpp just look in the code for examples, in vscode you can search for GeometryToolbox::DotProduct or GeometryToolbox::Norm

The problem is more related to the finding of the wall-normal for a point within the volume mesh, not to the computations that it will be involved in.

pcarruscag · 2023-02-14T15:05:26Z

The solution I suggested didn’t work?

rois1995 · 2023-02-14T15:46:22Z

The solution I suggested didn’t work?

I still have to check if the implementation is correct but with more than two cores the code breaks. However, I found out that the wall-normal of a volume point can be computed as the normalized gradient of the wall-distance. Does this sound correct to you?

However, there is a problem: I am using the aux variables to compute these gradients, but to compute dot(n, U) I first need n, thus I cannot compute them simultaneously. Since these computations are performed in the Preprocessing of the solvers, I was thinking to compute the normal within the FLOW_SOL preprocessing and the dot(n, U) in the TRANS_SOL preprocessing since the flow solver comes before the trans solver. Is this right?

pcarruscag · 2023-02-14T16:09:19Z

It's not correct. You need to follow the pattern from CGeometry::ComputeWallDistance
First you collect all the normals that are local to the rank
Then you collect the normals across all ranks (using the NDFlatner), then you access this information by (ClosestRank, ClosestMarker, ClosestVertex)
You cannot access the local information directly because ClosestMarker and ClosestVertex may refer to a different rank.

- Added variables only for debug

rois1995 · 2023-02-24T13:15:26Z

I managed to implement the computation of grad(n*U)*n, but at the moment it is located into CTransLMSolver::PreProcessing. It seems to work with a structured mesh on a flat plate. Currently, I am testing with a 2D profile too. However, being into the PreProcessing of the transition solver, the normals are computed at each iteration, thus it is not computationally efficient if non-deforming meshes are used. I am looking into where to put it such that it is computed just if the mesh is updated (and at the first iteration of course).

Plus, I have added a whole lot of variables to the output, but they will be removed in the final version. They are just used as debug.

pcarruscag · 2023-02-24T15:11:00Z

Do what we do with wall roughness and do it in the same place (computewalldistance and store in CPoint)

rois1995 · 2025-03-10T11:52:13Z

Results for the Eppler E367

Aerodynamic coefficients

Here you can see a comparison on the aerodynamic coefficients between the SU2 implementation and the one coming from reference papers. Results are a little bit different. In particular, for the Menter implementation the mesh topology is different as they have simulated a wind-tunnel like domain whereas mine is in free air (O-grid). I will try to compare the SU2 implementation with their meshes in the next days.

$c_P$ and $c_F$ distributions

$\alpha = 5^\circ$

Here, the main comparison is between the Coder implementation and the reference paper. The laminar bubble is slightly smaller in the reference paper but the shape of the pressure coefficient curve is very similar.

$\alpha = 6^\circ$

Here, the main comparison is between the Menter implementation and the reference paper. As said before, the mesh topologies are completely different. However, the same differences in the skin friction coefficient between the SLM and the LM model can be seen in both the reference and the SU2 implementation. On the other hand, the $c_P$ distribution is very different, with the SLM reference implementation predicting a much smoother effect coming from the LSB. The same smoothing is visible in the SU2 implementation, although it is much weaker. The peak of the $c_P$ at the leading edge is not visible in the reference paper as the image is cropped.

UPDATE - Using Menter's meshes. $c_P$ and $c_F$ distributions at $\alpha = 6^\circ$

The pressure coefficient's distribution essentially remains unaltered. On the other hand, the skin friction is more in line with the reference paper. The oscillations are related to the mesh refinement, mostly due to a non-splined underlying database.

rois1995 · 2025-03-10T14:38:31Z

Every testcase that is failing is due to a bug that I fixed in the CFlowIncOutput where the value of MAX_RMS_W was used for the RMS_W output. I should fix in the tutorials. Should I open another PR for that?

bigfooted · 2025-03-10T19:47:17Z

Every testcase that is failing is due to a bug that I fixed in the CFlowIncOutput where the value of MAX_RMS_W was used for the RMS_W output. I should fix in the tutorials. Should I open another PR for that?

The overhead in this PR is not so large, so I'd say keep it in this PR.

rois1995 · 2025-03-10T20:58:44Z

Every testcase that is failing is due to a bug that I fixed in the CFlowIncOutput where the value of MAX_RMS_W was used for the RMS_W output. I should fix in the tutorials. Should I open another PR for that?

The overhead in this PR is not so large, so I'd say keep it in this PR.

Guess it's too late now! Sorry 😐

pcarruscag · 2025-03-10T22:12:27Z

You did well to make a small PR for it.

pcarruscag · 2025-03-10T22:13:46Z

If you are happy with the validation does that mean this PR is ready for review? Can you remove [WIP] if that is the case?

rois1995 · 2025-03-11T08:55:33Z

I think that the Modified Eppler correlations can be removed as they do not perform very well. We can leave there just the Coder and the Menter correlations. Plus, there are a lot of debug quantities that will be removed before the merge. To me I think that the code can be reviewed. Should I clean everything before reviewing?

rois1995 · 2025-03-11T10:03:55Z

I guess I am still missing the cross-flow validation, I will do it in the next days.

pcarruscag · 2025-07-05T22:17:49Z

What about this one @rois1995, is it ready?

rois1995 · 2025-07-07T08:21:14Z

Hi @pcarruscag, yes, it is ready. I actually performed the 3D simulations on the prolate ellipsoid case but I did not share those here.

Comparing simulations with the reference paper (https://doi.org/10.2514/6.2017-3159) shows really poor results. In the paper, only the Menter correlations have been investigated. The cross-flow effects work in promoting transition to turbulence only for the Modified Eppler correlations, whereas the Coder one shows no influence (the onset function related to the cross-flow transition is always smaller than the standard one).

pcarruscag · 2025-07-07T15:55:19Z

I think everyone agrees to disagree when it comes to transition 🤣
Have you encountered a lot of sensitivity to initial conditions?

rois1995 · 2025-07-23T10:36:20Z

That is so true! There is some variability, especially with the turbulence intensity at freestream and the turbulent-to-laminar viscosity ratio at freestream (although it is lower for the latter).

rois1995 added 3 commits January 27, 2023 18:21

- Just copied the CTransLMSolver header and source to commit

a7136d3

- Start of implementation of Simplified LM transition model

e9226db

- Modify number of variables if SLM option is used

d03742a

pr-triage bot added the PR: unreviewed label Jan 27, 2023

pcarruscag changed the title ~~Implementation of Simplified LM Transition model [WIP]~~ [WIP] Implementation of Simplified LM Transition model Jan 27, 2023

pr-triage bot removed the PR: unreviewed label Jan 27, 2023

pcarruscag reviewed Jan 27, 2023

View reviewed changes

github-advanced-security bot found potential problems Jan 27, 2023

View reviewed changes

SU2_CFD/src/solvers/CTransLMSolver.cpp Fixed Show fixed Hide fixed

- Updates on the transition model.

e17872d

- Added correlations for Simplified LM. - There is a bug on the computation of grad(n*U)*n

github-advanced-security bot found potential problems Jan 31, 2023

View reviewed changes

- Added Cross flow effects for SLM

4b9c615

- Modified LM_OPTIONS to include cross-flow effects: from LM2015 to CROSSFLOW

github-advanced-security bot found potential problems Jan 31, 2023

View reviewed changes

WallyMaier reviewed Feb 1, 2023

View reviewed changes

- Fixed Intermittency_Sep computation for SLM.

04cd56d

- Added computation for normals of grid points

e29b4df

- Added variables only for debug

Fixed normal extraction from structure

0319d62

Added computation of wall normals to the CPoint structure

b1edf61

rois1995 added 2 commits March 7, 2025 18:18

- Fix DG problems when normal computation

4134667

Merge remote-tracking branch 'origin/develop' into feature_Trans_SLM

90efb40

Merge remote-tracking branch 'origin/develop' into feature_Trans_SLM

4076c81

Merge remote-tracking branch 'origin/develop' into feature_Trans_SLM

583bdf3

rois1995 changed the title ~~[WIP] Implementation of Simplified LM Transition model~~ Implementation of Simplified LM Transition model Mar 11, 2025

rois1995 added 10 commits March 12, 2025 18:54

- Modify screen output in SU2 header

f717ed5

- Fix bugs in crossflow correction

0f3f412

- Revert to use of dimensional vorticity

1ef2081

- Fix bug from paper in crossflow

bb3b066

- fixed formatting errors

d2791b0

Merge remote-tracking branch 'origin/develop' into feature_Trans_SLM

41291a4

Merge remote-tracking branch 'origin/develop' into feature_Trans_SLM

24785c2

Merge remote-tracking branch 'origin/develop' into feature_Trans_SLM

5fb0cb9

Merge remote-tracking branch 'origin/develop' into feature_Trans_SLM

b1147bd

- Improved notation for freestream conditions on ReThetaT

0f06f3b

Merge remote-tracking branch 'origin/develop' into feature_Trans_SLM

25fb369

rois1995 added 2 commits October 2, 2025 18:06

- Added comments

651af7c

Merge remote-tracking branch 'origin/develop' into feature_Trans_SLM

e8e2284

@@ -25,8 +25,8 @@
              */
             #pragma once
-            //#include <cmath>
             /*!
              * \class TransLMCorrelations
              * \brief Class for LM model's correlation functions.

	SU2_MPI::Error("Two correlations selected for Simplified model into LM_OPTIONS. Please choose only one.", CURRENT_FUNCTION);
	SU2_MPI::Error("Two correlations selected for simplified LM_OPTIONS. Please choose only one.", CURRENT_FUNCTION);

Implementation of Simplified LM Transition model #1901

Are you sure you want to change the base?

Implementation of Simplified LM Transition model #1901

Conversation

rois1995 commented Jan 27, 2023

Proposed Changes

Related Work

Uh oh!

rois1995 commented Jan 27, 2023

Uh oh!

pcarruscag left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

rois1995 commented Jan 31, 2023

Uh oh!

Check notice

Uh oh!

Uh oh!

Uh oh!

Check notice

Copilot Autofix

pcarruscag commented Jan 31, 2023

Uh oh!

github-advanced-security bot left a comment

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!

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!

bigfooted commented Feb 10, 2023

Uh oh!

rois1995 commented Feb 14, 2023

Uh oh!

pcarruscag commented Feb 14, 2023

Uh oh!

rois1995 commented Feb 14, 2023

Uh oh!

pcarruscag commented Feb 14, 2023

Uh oh!

rois1995 commented Feb 24, 2023

Uh oh!

pcarruscag commented Feb 24, 2023

Uh oh!

rois1995 commented Mar 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Results for the Eppler E367

Aerodynamic coefficients

$c_P$ and $c_F$ distributions

$\alpha = 5^\circ$

$\alpha = 6^\circ$

UPDATE - Using Menter's meshes. $c_P$ and $c_F$ distributions at $\alpha = 6^\circ$

Uh oh!

rois1995 commented Mar 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

bigfooted commented Mar 10, 2025

Uh oh!

rois1995 commented Mar 10, 2025

Uh oh!

pcarruscag commented Mar 10, 2025

Uh oh!

pcarruscag commented Mar 10, 2025

Uh oh!

rois1995 commented Mar 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rois1995 commented Mar 10, 2025 •

edited

Loading

rois1995 commented Mar 10, 2025 •

edited

Loading

rois1995 commented Mar 11, 2025 •

edited

Loading