[WIP] Implementation of eN transition model #1765
Conversation
Common/src/CConfig.cpp
Outdated
| switch (Kind_Trans_Model) { | ||
| case TURB_TRANS_MODEL::NONE: break; | ||
| case TURB_TRANS_MODEL::EN: cout << "Low-turbulence Transition model: eN 1 equation model (2014)" << endl; break; | ||
| } |
There was a problem hiding this comment.
Formatting.
Should we also include LM in this switch
There was a problem hiding this comment.
Thanks. Updated.
As for the LM option. There are a lot of locations that have this switch. For now I have left the 'case TURB_TRANS_MODEL::LM :' option out as it will be added to develop when the LM PR will be merged. I have done it in this way as in most locations the LM code is not there of course.
I should add that on these LM/EN switch locations the LM implementation has the same code set-up. The thought was that merging an extra case in the switch list would produce no problems?
trans_conv/diff and soures filled in. eN transition added to the ft2 term inside SA sources. I have added my T3A- mesh and .cfg that I use for testing to the Quickstart folder. Next step is to debug eN itself. Regular run is not yet working.
|
This pull request introduces 1 alert when merging 56b93c1 into 2642f7f - view on LGTM.com new alerts:
|
|
This pull request introduces 1 alert when merging fd51593 into 2642f7f - view on LGTM.com new alerts:
|
|
Don't add meshes to the code repo, we have TestCases for that. |
|
This pull request introduces 1 alert when merging 604aadb into 2642f7f - view on LGTM.com new alerts:
|
| virtual void SetAmplificationFactor(su2double val_amplification_factor_i, su2double val_amplification_factor_j) { | ||
| amplification_factor_i = val_amplification_factor_i; | ||
| amplification_factor_j = val_amplification_factor_j; |
There was a problem hiding this comment.
| virtual void SetAmplificationFactor(su2double val_amplification_factor_i, su2double val_amplification_factor_j) { | |
| amplification_factor_i = val_amplification_factor_i; | |
| amplification_factor_j = val_amplification_factor_j; | |
| void SetAmplificationFactor(su2double val_amplification_factor_i, su2double val_amplification_factor_j) { | |
| amplification_factor_i = val_amplification_factor_i; | |
| amplification_factor_j = val_amplification_factor_j; |
| class CUpwSca_TransEN final : public CUpwScalar<FlowIndices> { | ||
| private: | ||
| using Base = CUpwScalar<FlowIndices>; | ||
| using Base::nDim; |
There was a problem hiding this comment.
| using Base::nDim; |
| /*! | ||
| * \brief Adds any extra variables to AD. | ||
| */ | ||
| void ExtraADPreaccIn() override {} |
| * \param[in] config - Definition of the particular problem. | ||
| */ | ||
| void FinishResidualCalc(const CConfig* config) override { | ||
| Flux[0] = a0*V_i[idx.Density()]*ScalarVar_i[0] + a1*V_j[idx.Density()]*ScalarVar_j[0]; |
There was a problem hiding this comment.
| Flux[0] = a0*V_i[idx.Density()]*ScalarVar_i[0] + a1*V_j[idx.Density()]*ScalarVar_j[0]; | |
| Flux[0] = a0*V_i[idx.Density()]*ScalarVar_i[0] + a1*V_j[idx.Density()]*ScalarVar_j[0]; |
| * \param[in] val_nDim - Number of dimensions of the problem. | ||
| * \param[in] val_nVar - Number of variables of the problem. | ||
| * \param[in] config - Definition of the particular problem. | ||
| */ | ||
| CUpwSca_TransEN(unsigned short val_nDim, unsigned short val_nVar, const CConfig* config) | ||
| : CUpwScalar<FlowIndices>(val_nDim, val_nVar, config) {} |
There was a problem hiding this comment.
| * \param[in] val_nDim - Number of dimensions of the problem. | |
| * \param[in] val_nVar - Number of variables of the problem. | |
| * \param[in] config - Definition of the particular problem. | |
| */ | |
| CUpwSca_TransEN(unsigned short val_nDim, unsigned short val_nVar, const CConfig* config) | |
| : CUpwScalar<FlowIndices>(val_nDim, val_nVar, config) {} | |
| * \param[in] val_nDim - Number of dimensions of the problem. | |
| * \param[in] config - Definition of the particular problem. | |
| */ | |
| CUpwSca_TransEN(unsigned short val_nDim, const CConfig* config) | |
| : CUpwScalar<FlowIndices>(val_nDim, 1, config) {} |
| var.d_ft2 = -2.0 * var.ct4 * var.Ji * var.ft2 * var.d_Ji; | ||
| static void get(const su2double& n_hat, CSAVariables& var) { | ||
| if (var.transEN == false){ | ||
| const su2double xsi2 = pow(var.Ji, 2); |
There was a problem hiding this comment.
Please format the code... this is hard to read like this
| @@ -103,6 +103,18 @@ class CSolverFactory { | |||
| */ | |||
| static CSolver* CreateTurbSolver(TURB_MODEL kindTurbModel, CSolver **solver, CGeometry *geometry, CConfig *config, int iMGLevel, int adjoint); | |||
|
|
|||
| /*! | |||
| * \brief Create a turbulent solver | |||
There was a problem hiding this comment.
turbulent -> transition, check other places please
| /*! | ||
| * \brief Destructor of the class. | ||
| */ | ||
| ~CTransENSolver() = default; |
There was a problem hiding this comment.
| /*! | |
| * \brief Destructor of the class. | |
| */ | |
| ~CTransENSolver() = default; |
SU2_CFD/src/drivers/CDriver.cpp
Outdated
| if ( NONE) { | ||
| SU2_MPI::Error("Invalid convective scheme for the transition equations.", CURRENT_FUNCTION); | ||
| } | ||
| else if (EN) numerics[iMGlevel][TRANS_SOL][conv_term] = new CUpwSca_TransEN<Indices>(nDim, nVar_Trans, config); |
There was a problem hiding this comment.
I don't understand the if (NONE) parts, NONE is a constant and always 0
| void CTransENVariable::SetAmplificationFactor(unsigned long iPoint, su2double val_AmplificationFactor) { | ||
| AD::StartPreacc(); | ||
| AmplificationFactor(iPoint) = val_AmplificationFactor; | ||
| AD::SetPreaccOut(AmplificationFactor(iPoint)); //Still need to be fixed? | ||
| AD::EndPreacc(); | ||
| } |
There was a problem hiding this comment.
This should be done inline and preaccumulation does not need to be used for simple assignments like these
| void CTransENVariable::SetAmplificationFactor(unsigned long iPoint, su2double val_AmplificationFactor) { | |
| AD::StartPreacc(); | |
| AmplificationFactor(iPoint) = val_AmplificationFactor; | |
| AD::SetPreaccOut(AmplificationFactor(iPoint)); //Still need to be fixed? | |
| AD::EndPreacc(); | |
| } |
…_feature_eN into SU2_feature_eN_PR01
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| AD::SetPreaccIn(ScalarVar_Grad_i[0], nDim); | ||
|
|
||
| su2double rho = V_i[idx.Density()]; | ||
| su2double p = V_i[idx.Pressure()]; |
Check notice
Code scanning / CodeQL
Unused local variable Note
|
|
||
| const su2double vel_mag = max(sqrt(vel_u * vel_u + vel_v * vel_v + vel_w * vel_w), 1e-20); | ||
|
|
||
| su2double rhoInf = config->GetDensity_FreeStreamND(); |
Check notice
Code scanning / CodeQL
Unused local variable Note
| const su2double vel_mag = max(sqrt(vel_u * vel_u + vel_v * vel_v + vel_w * vel_w), 1e-20); | ||
|
|
||
| su2double rhoInf = config->GetDensity_FreeStreamND(); | ||
| su2double pInf = config->GetPressure_FreeStreamND(); |
Check notice
Code scanning / CodeQL
Unused local variable Note
| // var.ft2 = var.ct3 * (1 - exp(2*(var.amplification - var.Ncrit)) ) * exp(-var.ct4 * xsi2); | ||
| var.ft2 = var.ct3 * (1 - exp(var.modifiedintermittency) ); | ||
| var.d_ft2 = 0.0; | ||
| // var.d_ft2 = -2.0 * var.ct4 * var.Ji * var.ft2 * var.d_Ji; |
Check notice
Code scanning / CodeQL
Commented-out code Note
| const su2double xsi2 = pow(var.Ji, 2); | ||
|
|
||
| if (var.transEN == true){ | ||
| // var.ft2 = var.ct3 * (1 - exp(2*(var.amplification - var.Ncrit)) ) * exp(-var.ct4 * xsi2); |
Check notice
Code scanning / CodeQL
Commented-out code Note
| // if (config->GetKind_Regime() == ENUM_REGIME::COMPRESSIBLE) var.ct4 = 0.025; | ||
| // else var.ct4 = 0.05; |
Check notice
Code scanning / CodeQL
Commented-out code Note
| // var.Ncrit = -8.43 - 2.4*log(config->GetTurbulenceIntensity_FreeStream()/100); | ||
| // var.amplification = min(amplification_factor_i, var.Ncrit); |
Check notice
Code scanning / CodeQL
Commented-out code Note
| // void FinishResidualCalc(const CConfig* config) override { | ||
| // const bool implicit = config->GetKind_TimeIntScheme() == EULER_IMPLICIT; | ||
| // | ||
| // /*--- Compute mean effective dynamic viscosity ---*/ | ||
| // const su2double diff_i_amplification = (Laminar_Viscosity_i + Eddy_Viscosity_i)/sigma_n; | ||
| // const su2double diff_j_amplification = (Laminar_Viscosity_j + Eddy_Viscosity_j)/sigma_n; | ||
| // | ||
| // const su2double diff_amplification = 0.5*(diff_i_amplification + diff_j_amplification); | ||
| // | ||
| // Flux[0] = diff_amplification*Proj_Mean_GradScalarVar[0]; | ||
| // | ||
| // /*--- For Jacobians -> Use of TSL approx. to compute derivatives of the gradients ---*/ | ||
| // | ||
| // if (implicit) { | ||
| // Jacobian_i[0][0] = (0.5*Proj_Mean_GradScalarVar[0]-diff_amplification*proj_vector_ij); | ||
| // Jacobian_j[0][0] = (0.5*Proj_Mean_GradScalarVar[0]+diff_amplification*proj_vector_ij); | ||
| // } | ||
| // } |
Check notice
Code scanning / CodeQL
Commented-out code Note
| const auto index = counter * Restart_Vars[1] + skipVars; | ||
| for (auto iVar = 0u; iVar < nVar; iVar++) nodes->SetSolution(iPoint_Local, iVar, Restart_Data[index + iVar]); | ||
| nodes ->SetAmplificationFactor(iPoint_Local, Restart_Data[index + 2]); | ||
| // nodes ->SetModifiedIntermittency(iPoint_Local, Restart_Data[index + 3]); |
Check notice
Code scanning / CodeQL
Commented-out code Note
Hello!
Proposed Changes
To continue current work being done on different transition models, I am working at the implementation of the eN model for low-turbulence external aerodynamics as proposed by James G. Goder and Mark D. Maughmer (10.2514/1.J052905). This model incorporates a new transport equation to describe the growth of the maximum Tollmien-Schlichting instability amplitude in the presence of a boundary layer.
Related Work
This work will be set-up in the same way as the LM model is in #1751 (which was also generally speaking how I had set up my own LM transition implementation) to make sure merging in a later stage will be easier.
Roadmap
I am hoping to achieve a working implementation of the eN transition method. This work will be done as finishing part of my master thesis. We will have to see if I have time to completely finish the V&V phase due to time restrictions. I have already done quite some work on implementing and verifying the LM method on a private branch and results show mixed results, especially at higher Re numbers (mentioned also by Menter et all.). Thus the step is made to see how these LM results will compare to eN results.
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.