Merge branch 'improvement.DSMC.tempGradient' into 'master.dev'

[improvement.DSMC.tempGradient] Bugfix and extension of temperature gradient calculation

See merge request piclas/piclas!814

CMakeListsMachine.txt

@@ -251,7 +251,7 @@ ELSEIF (CMAKE_Fortran_COMPILER_ID MATCHES "Intel")
   # Compile flags depend on the generator
   IF(NOT "${CMAKE_GENERATOR}" MATCHES "Ninja")
     # add flags only for compiling not linking!
-    SET (PICLAS_COMPILE_FLAGS "-fPIC-fpp -allow nofpp_comments -assume bscc")
+    SET (PICLAS_COMPILE_FLAGS "-fPIC -fpp -allow nofpp_comments -assume bscc")
   ELSE()
     SET (PICLAS_COMPILE_FLAGS "${NINJA_COLOR_DIAGNOSTICS} -allow nofpp_comments -assume bscc")
   ENDIF()

docs/documentation/userguide/features-and-models/BC-particle-solver.md

@@ -70,14 +70,20 @@ The wall velocity will then be superimposed onto the particle velocity.
 ### Linear temperature gradient
 
 A linear temperature gradient across a boundary can be defined by supplying a second wall temperature and the start and end vector
-
-    Part-Boundary2-WallTemp2=500.
-    Part-Boundary2-TemperatureGradientStart=(/0.,0.,0./)
-    Part-Boundary2-TemperatureGradientEnd=(/0.,0.,1./)
-
-Between these two points the temperature will be interpolated, where the start vector corresponds to the first wall temperature,
-whereas the end vector to the second wall temperature. Beyond these position values, the first and second temperature will be used
-as the constant wall temperature, respectively.
+as well as an optional direction to which the gradient shall be limited (default: 0, x = 1, y = 2, z = 3)
+
+    Part-Boundary2-WallTemp2      = 500.
+    Part-Boundary2-TempGradStart  = (/0.,0.,0./)
+    Part-Boundary2-TempGradEnd    = (/1.,0.,1./)
+    Part-Boundary2-TempGradDir    = 0
+
+In the default case of the `TempGradDir = 0`, the temperature will be interpolated between the start and end vector, where the
+start vector corresponds to the first wall temperature `WallTemp`, and the end vector to the second wall temperature `WallTemp2`.
+Position values (which are projected onto the temperature gradient vector) beyond the gradient vector utilize the first (Start) and second temperature (End) as the constant wall temperature, respectively. In the special case of `TempGradDir = 1/2/3`, the temperature gradient will
+only be applied along the chosen the direction. As oppposed to the default case, the positions of the surfaces are not projected
+onto the gradient vector before checking wether they are inside the box spanned by `TempGradStart` and `TempGradEnd`. Outside of
+the defined box, the temperature corresponds to `WallTemp`. The applied surface temperature is output in the `DSMCSurfState` as
+`Wall_Temperature` for verification.
 
 ### Radiative equilibrium
 

docs/documentation/userguide/features-and-models/Bhatnagar-Gross-Krook.md

@@ -84,11 +84,11 @@ An option is available to utilize a moving average for the variables used in the
 
     Particles-BGK-MovingAverage = T
 
-The purpose is to increase the sample size and reduce the noise for steady gas flows. For this, a factor 
-0 < Particles-BGK-MovingAverageFac < 1 must be defined with which the old $$M^n$$ and newly sampled moments $$M$$ are weighted 
-to define the moments for the next time step $$M^{n+1}$$:
-
-$$ M^{n+1}=Particles-BGK-MovingAverageFac*M+(1-Particles-BGK-MovingAverageFac)*M^n$$
+The purpose is to increase the sample size and reduce the noise for steady gas flows. For this, the factor $f$
 
     Particles-BGK-MovingAverageFac = 0.01
 
+between zero and one must be defined with which the old $M^n$ and newly sampled moments $M$ are weighted 
+to define the moments for the next time step $M^{n+1}$:
+
+$$ M^{n+1}=f M+(1-f) M^n.$$

docs/documentation/userguide/features-and-models/features-particle-solver.md

@@ -263,9 +263,9 @@ Currently, only merging based on the number of particles within the cell is impl
 
 Furthermore, the spread or aggressiveness of the merge algorithm can be changed, i.e. how deep the merge extends into the mesh starting from each cell. 0 is the least aggressive merge, 3 the most aggressive merge.
 
-  Part-CellMergeSpread                  = 0
+    Part-CellMergeSpread                = 0
 
 There is also the possibility to define a maximum number of cells that can be merged. In this way, a desired "resolution" of the virtual cells can be achieved.
 
-  Part-MaxNumbCellsMerge                = 5
+    Part-MaxNumbCellsMerge              = 5
 

regressioncheck/CHE_DSMC/BC_DiffuseWall_TempGrad/analyze.ini

@@ -1,5 +1,6 @@
-! compare the PartAnalyze.csv with a reference file
-compare_data_file_name      = PartAnalyze.csv
-compare_data_file_reference = PartAnalyze_ref.csv
-compare_data_file_tolerance = 5e-2
-compare_data_file_tolerance_type = relative
+! absolute comparison of wall temperature gradient due to strong fluctuations in the other variables in DSMCSurfState
+h5diff_file            = BC_TempGrad_DSMCSurfState_000.00000001000000000.h5
+h5diff_reference_file  = BC_TempGrad_DSMCSurfState_000.00000001000000000_ref.h5
+h5diff_data_set        = SurfaceData
+h5diff_tolerance_value = 40             ! corresponds to at most 1% deviation (at 4000K)
+h5diff_tolerance_type  = absolute

regressioncheck/CHE_DSMC/BC_DiffuseWall_TempGrad/hopr.ini

@@ -5,7 +5,7 @@ NVisu        =1
 Mode         =1
 
 Corner   =(/0.0,0.0,-0.25,, 1.0,0.0,-0.25,,1.0,0.1,-0.25 ,, 0.0,0.1,-0.25 ,, 0.0,0.0,0.25 ,, 1.0,0.0,0.25 ,, 1.0,0.1,0.25 ,, 0.0,0.1,0.25 /)
-nElems   =(/20,1,1/)
+nElems   =(/20,1,10/)
 elemtype =108
 
 BCIndex  =(/6 ,4 ,1 ,3 ,2 ,5/)