Turb Inflow: move support codes to PelePhysics, add documentation

.github/workflows/linux.yml

@@ -162,16 +162,22 @@ jobs:
     steps:
     - uses: actions/checkout@v3
     - name: System Dependencies
-      run: .github/workflows/dependencies/dependencies_gcc10.sh
+      run: |
+        .github/workflows/dependencies/dependencies_gcc10.sh
+         sudo apt-get install -y python3-setuptools
+         python3 -m pip install --user scipy
+         python3 -m pip install --user matplotlib
+         python3 -m pip install --user numpy
     - name: Repo Dependencies
       run: Utils/CloneDeps.sh
     - name: GenerateTurbFile
-      env:
-         AMREX_HOME: ${GITHUB_WORKSPACE}/Submodules/PelePhysics/Submodules/amrex
-      working-directory: ./Exec/RegTests/TurbInflow/TurbFileHIT
+      working-directory: ./Submodules/PelePhysics/Support/TurbFileHIT
       run: |
+        ./gen_hit_ic.py -k0 4 -N 32 -Nk 256
         make -j 2 COMP=gnu
-        ./PeleTurb3d.gnu.ex input hit_file=../../HITDecay/hit_ic_4_32.dat input_ncell=32 amrex.abort_on_unused_inputs=1
+        ./PeleTurb3d.gnu.ex input hit_file=hit_ic_4_32.dat input_ncell=32 amrex.abort_on_unused_inputs=1
+        mkdir ${GITHUB_WORKSPACE}/Exec/RegTests/TurbInflow/TurbFileHIT
+        cp -r TurbTEST ${GITHUB_WORKSPACE}/Exec/RegTests/TurbInflow/TurbFileHIT
     - name: Build
       working-directory: ./Exec/RegTests/TurbInflow/
       run: |

Docs/sphinx/manual/LMeXControls.rst

@@ -36,6 +36,15 @@ Computational domain definition
     peleLM.lo_bc = Interior Interior Inflow
     peleLM.hi_bc = Interior Interior Inflow
 
+If specifying boundaries as ``Inflow``, the bcnormal function must be defined
+in the ``pelelmex_prob.H`` file for the case to define the inflow conditions.
+``Inflow`` boundaries may also be augmented with spatially and temporally
+varying turbulent fluctuations using the ``TurbInflow`` utility from
+PelePhysics. See the ``Exec/RegTests/TurbInflow`` test for an example of how
+to use this capability and the
+`PelePhysics  documentation <https://amrex-combustion.github.io/PelePhysics/Utility.html#turbulent-inflows>`_
+for the relevant input file flags.
+
 Grid/AMR parameters
 -------------------
 
@@ -239,7 +248,7 @@ PeleLMeX algorithm
     peleLM.gravity = 0.0 0.0 -9.81         # [OPT, DEF=Vec{0.0}] Gravity vector [MKS]
     peleLM.gradP0 = 0.0 0.0 10.0           # [OPT, DEF=Vec{0.0}] Average background pressure gradient [Pa/m]
     peleLM.do_periodic_channel = 0         # [OPT, DEF= 0] Add an automatic pressure gradient to maintain initial condition mass flow rate in periodic channel
-    peleLM.periodic_channel_dir = 2        # [OPT, DEF= -1] Required if do_periodic_channel != 0. Direction to apply pressure gradient.    
+    peleLM.periodic_channel_dir = 2        # [OPT, DEF= -1] Required if do_periodic_channel != 0. Direction to apply pressure gradient.
     peleLM.closed_chamber = 0              # [OPT] Override the automatic detection of closed chamber (based on Outflow(s))
     peleLM.floor_species = 0               # [OPT, DEF=0] Crudely enforce mass fraction positivity
     peleLM.deltaT_verbose = 0              # [OPT, DEF=0] Verbose of the deltaT iterative solve algorithm
@@ -258,7 +267,7 @@ Transport coeffs and LES
     peleLM.Prandtl = 0.7                   # [OPT, DEF=0.7] If fixed_Pr or doing LES, specifies the Prandtl number
     peleLM.Schmidt = 0.7                   # [OPT, DEF=0.7] If doing LES, specifies the Schmidt number
     peleLM.Lewis = 1.0                     # [OPT, DEF=1.0] If fixed_Le, specifies the Lewis number
-    
+
     peleLM.les_model = "None"              # [OPT, DEF="None"] Model to compute turbulent viscosity: None, Smagorinsky, WALE, Sigma
     peleLM.les_cs_smag = 0.18              # [OPT, DEF=0.18] If using Smagorinsky LES model, provides model coefficient
     peleLM.les_cm_wale = 0.60              # [OPT, DEF=0.60] If using WALE LES model, provides model coefficient
@@ -280,7 +289,7 @@ Chemistry integrator
     cvode.solve_type = denseAJ_direct           # [OPT, DEF=GMRES] Linear solver employed for CVODE Newton direction
     cvode.max_order  = 4                        # [OPT, DEF=2] Maximum order of the BDF method in CVODE
     cvode.max_substeps = 10000                  # [OPT, DEF=10000] Maximum number of substeps for the linear solver in CVODE
-    
+
 Note that the last five parameters belong to the Reactor class of PelePhysics but are specified here for completeness. In particular, CVODE is the adequate choice of integrator to tackle PeleLMeX large time step sizes. Several linear solvers are available depending on whether or not GPU are employed: on CPU, `dense_direct` is a finite-difference direct solver, `denseAJ_direct` is an analytical-jacobian direct solver (preferred choice), `sparse_direct` is an analytical-jacobian sparse direct solver based on the KLU library and `GMRES` is a matrix-free iterative solver; on GPU `GMRES` is a matrix-free iterative solver (available on all the platforms), `sparse_direct` is a batched block-sparse direct solve based on NVIDIA's cuSparse (only with CUDA), `magma_direct` is a batched block-dense direct solve based on the MAGMA library (available with CUDA and HIP. Different `cvode.solve_type` should be tried before increasing the `cvode.max_substeps`.
 
 Embedded Geometry

Exec/RegTests/HITDecay/README

@@ -3,25 +3,8 @@ originally implemented in PeleC and used to study compressible turbulence in:
 E. Motheau and J. Wakefield, "Investigation of finite-volume methods to capture shocks and turbulence spectra in compressible flows", Commun. in Appl. Math. and Comput. Sci, 15-1 (2020), 1--36.
 https://arxiv.org/abs/1902.06665
 
-The script gen_hit_ic.py can be used to generate an initial solution.
+The script gen_hit_ic.py from PelePhysics/Support/TurbFileHIT can be used to generate an initial solution.
 
 Here, there is no scaling by the turbulent Mach number and the density is kept constant, so that the Taylor-scale Reynolds number is defined by Re=0.5 urms / nu where umrs can be set in the input file (prob.urms0=...) and nu is computed from the mixture composition.
 
-At the beginning of a simulation, a file initialConditions.txt is created to hold the values of urms0, lambda0 and tau, which will be used for post-processing.
-
-The post-processing chain is composed of two major parts: computing the temporal evolution of the turbulent kinetic energy, the square of magnitude of vorticity as well as the square of the divergence. Then the turbulent spectra is computed for the last plotfile found.
-
-IMPORTANT:
-The following tools must be pre-compiled and the executable located at the root of this directory, where the post.sh script is located:
-AmrDeriveSpectrum (https://github.com/AMReX-Astro/AmrDeriveSpectrum)
-AugmentPlotfile (located in amrex/Tools/C_util/AugmentPlotfile)
-
-To proceed to an analysis, please create a directory and copy inside all the plotfiles as well as the initialConditions.txt file. Then execute the post.sh script as in the following example:
-
-./PeleLMeX3d.gnu.MPI.ex inputs.3d
-mkdir ANALYSIS
-cp -r plt_0000* ANALYSIS/
-cp initialConditions.txt ANALYSIS/
-./post.sh ANALYSIS/ 1
-
-The output csv files will be located in the ANALYSIS directory.
+At the beginning of a simulation, a file initialConditions.txt is created to hold the values of urms0, lambda0 and tau, which will be used for post-processing.