CFD中文网

its really hard to debug according to these information.

我推了一下，好像是这个$\nabla\cdot(\nabla\cdot(\bfU\bfU'))$，你验证下？

你最好把外面的大的stl弄个洞出来。shm画网格的时候，如果你的外面的stl是封闭的。就识别不到里面小的stl

https://cfd-china.com/topic/8018/免费帮助debug算例 你看看

喷管出口是超音速么

这种没法搞，注定就是深宽比很大

看起来ESI版本可以。多谢分享。

有时候是这样
有时候也没法完全对称 比如瞬态的圆柱绕流用了稳态求解
你要完全对称可以用对称几何

整不了。只能做成3维，然后映射成2维

这个基本没有一针见血的办法，shm生成边界层已经是老大难了

一整个的patch还是patch单独命名了？

这个应该需要对这种流场比较熟悉的大佬看一下。
我对涡的研究不是很专业。

LES的2D一般就是观察。要真实的对比要做3D

# output particle properties
dump dmp all custom/vtk 5000 post/liggghts_init_*.vtk id type x y z vx vy vz      fx fy fz omegax omegay omegaz radius
ERROR: Invalid dump style (/home/dyfluid/CFDEM/LIGGGHTS/src/output.cpp:589)

这个看起来是后处理输出什么变量导致的
# output particle properties

/home/dyfluid/CFDEM/CFDEMcoupling/tutorials/cfdemSolverPiso

我测试了一下这个没问题。你那个有问题我看好像是运行vtk后处理的时候的问题，如下。我不太东liggghts。你看看顺着思路能不能把后处理某个东西关掉之类的

//   run_liggghts_init_DEM   //

/home/dyfluid/CFDEM/CFDEMcoupling/tutorials/cfdemSolverPiso/bubblingFluidizedBed/DEM

LIGGGHTS (Version LIGGGHTS-PFM 24.01, compiled 2025-01-20-15:21:00 by dyfluid, git commit d075b97762ce4302707903718dda170920e630bd, based on LAMMPS 23 Nov 2013)

# define the attributes associated with the particles,
# 'granular' (or 'sphere') style uses diameter, mass and angular velocity
atom_style      granular

# use an array to map particle IDs to local storage index,
atom_modify     map array

# set simulation domain to be fixed in x y z
boundary        f f f

# save communication by turning off Newton's 3rd law for pairwise interaction,
# note: this setting only influences communication between procs, Newton's
# 3rd law is still used for contact force calculations
newton          off

# use a single value for ghost particle cutoff distance and
# enable velocity to be communicated with ghost particles
communicate     single vel yes

# set unit system to SI
units           si

# define the region used as simulation domain (min/max X, min/max Y, min/max Z)
region          domain block 0.0 0.0432 0.0 0.01 0.0 0.1248 units box

# create the simulation domain and 2 material types (with friction/frictionless)
create_box      2 domain
Created orthogonal box = (0 0 0) to (0.0432 0.01 0.1248)
  1 by 1 by 1 MPI processor grid

# specify the skin distance for neighbor list generation
neighbor        0.0003 bin
neigh_modify    delay 0


# define the material properties required for granular pair styles
# type 1 = particles / walls with friction
# type 2 = frictionless walls
fix m1 all property/global youngsModulus peratomtype 1.2e8 1.2e8
fix m2 all property/global poissonsRatio peratomtype 0.3 0.3
fix m3 all property/global coefficientRestitution peratomtypepair 2 0.986 0.986 0.986 0.986
fix m4 all property/global coefficientFriction peratomtypepair 2 0.1 0.0 0.0 0.0

# specify contact model to use
pair_style  gran model hertz tangential history
SURFACE/DEFAULT loaded
HERTZ loaded
TANGENTIAL/HISTORY loaded
pair_coeff  * *

# set the time step
timestep    0.0000025

# apply gravity
fix gravity all gravity 9.81 vector 0.0 0.0 -1.0

# define walls
fix xwall1 all wall/gran model hertz tangential history primitive type 1 xplane 0.0000
SURFACE/DEFAULT loaded
HERTZ loaded
TANGENTIAL/HISTORY loaded
fix xwall2 all wall/gran model hertz tangential history primitive type 1 xplane 0.0432
SURFACE/DEFAULT loaded
HERTZ loaded
TANGENTIAL/HISTORY loaded
fix ywall1 all wall/gran model hertz tangential history primitive type 2 yplane 0.0000
SURFACE/DEFAULT loaded
HERTZ loaded
TANGENTIAL/HISTORY loaded
fix ywall2 all wall/gran model hertz tangential history primitive type 2 yplane 0.0100
SURFACE/DEFAULT loaded
HERTZ loaded
TANGENTIAL/HISTORY loaded
fix zwall1 all wall/gran model hertz tangential history primitive type 1 zplane 0.0000
SURFACE/DEFAULT loaded
HERTZ loaded
TANGENTIAL/HISTORY loaded

# particle distributions and insertion
fix pts1 all particletemplate/sphere 367127 atom_type 1 density constant 1000 radius constant 0.0006
fix pdd1 all particledistribution/discrete 681573 1 pts1 1.0
Fix particledistribution/discrete (id pdd1): distribution based on mass%:
    pts1: d=1.200000e-03 (max. bounding sphere) mass%=100.000000%
Fix particledistribution/discrete (id pdd1): distribution based on number%:
    pts1: d=1.200000e-03 (max. bounding sphere) number%=100.000000%

fix ins all insert/pack seed 4763687 distributiontemplate pdd1 vel constant 0.0 0.0 -0.15     insert_every once overlapcheck yes all_in yes particles_in_region 9240 region domain

# use constant NVE integration to update position, velocity and angular velocity
fix integr all nve/sphere

# output settings, include kinetic energy
thermo_style    custom step atoms ke

# set frequency of output
thermo          1000

# ignore particles leaving the simulation domain,
# do not normalize thermodynamic output values by the number of atoms
thermo_modify   lost ignore norm no
compute_modify  thermo_temp dynamic yes

# insert the first particles so that dump is not empty
run 1
Setting up run ...
INFO: Resetting random generator for region domain
INFO: Particle insertion ins: inserting every 0 steps
Memory usage per processor = 5.00542 Mbytes
Step Atoms KinEng 
       0        0           -0 
insertion: proc 0 at 0 %
insertion: proc 0 at 10 %
insertion: proc 0 at 20 %
insertion: proc 0 at 30 %
insertion: proc 0 at 40 %
insertion: proc 0 at 50 %
insertion: proc 0 at 60 %
insertion: proc 0 at 70 %
insertion: proc 0 at 80 %
insertion: proc 0 at 90 %
INFO: Particle insertion ins: inserted 9240 particle templates (mass 0.008360) at step 1
 - a total of 9240 particle templates (mass 0.008360) inserted so far.
       1     9240 9.4067122e-05 
Loop time of 0.0254776 on 1 procs for 1 steps with 9240 atoms
Max time of 0.0254776 on 1 procs for 1 steps with 9240 atoms

Pair  time (%) = 0.000333961 (1.3108)
Max Pair time (ib) = 0.000333961 (0)
Neigh time (%) = 0.0043867 (17.2179)
Max Neigh time (ib) = 0.0043867 (0)
Comm  time (%) = 5.4832e-05 (0.215217)
Max Comm time (ib) = 5.4832e-05 (0)
Outpt time (%) = 5.5202e-05 (0.216669)
Max Outpt time (ib) = 5.5202e-05 (0)
Other time (%) = 0.0206469 (81.0394)
Max Other time (ib) = 0.0206469 (0)

Nlocal:    9240 ave 9240 max 9240 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Nghost:    0 ave 0 max 0 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Neighs:    7500 ave 7500 max 7500 min
Histogram: 1 0 0 0 0 0 0 0 0 0

Total # of neighbors = 7500
Ave neighs/atom = 0.811688
Neighbor list builds = 1
Dangerous builds = 0

# output particle properties
dump dmp all custom/vtk 5000 post/liggghts_init_*.vtk id type x y z vx vy vz      fx fy fz omegax omegay omegaz radius
ERROR: Invalid dump style (/home/dyfluid/CFDEM/LIGGGHTS/src/output.cpp:589)

哪个算例

是的，一个整体平均值约束。

增加神经元看看能不能把GPU拉上100%

现在发现已经可以用gpu训练了，但是训练速度还不如cpu，慢不少。

看你GPU占用多少。

https://cfd-china.com/topic/8018 可以上传看看

coalChemisryFoam变密度的求解器无法计算常密度的情况吗？

我确定这个可以。http://dyfluid.com/class.html#id4 我在这个算例里面做的就是一个基本常密度的，附加密度的求解器，压力方程跟coalChemistryFoam压力方程类似，。

具体coalChemisryFoam这个我不太清楚了。这求解器没玩过。