Guest User

Aerodynamics on Demand - Hibou Chat

a guest
Feb 13th, 2014
130
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 8.24 KB | None | 0 0
  1. Julien de Charentenay
  2. Hi.
  3. Julien • 47 mins
  4. Hi
  5. 47 mins
  6. Richard Brook
  7. Julien de Charentenay
  8. Let's walk through what to do with the scenario_0.zip.
  9. I am just copying it somewhere.
  10. And extract it.
  11. Are you windows or linux?
  12. Julien • 46 mins
  13. windows
  14. Im considering a linux partition so I can run swak4foam
  15. 45 mins
  16. Richard Brook
  17. Julien de Charentenay
  18. Do you use blueCFD, blueCFD-SingleCore or another?
  19. Julien • 45 mins
  20. if I can figure out how to compile it that is!
  21. yes I use single core
  22. 45 mins
  23. Richard Brook
  24. Julien de Charentenay
  25. Re Linux: you can try with an emulator first, but partition will be best. Compilation of swak4foam is reasonably straightforward - in my opinion.
  26. Do you have the scenario_0.zip extracted?
  27. Julien • 44 mins
  28. yes and I have put the file you sent me in there too
  29. 44 mins
  30. Richard Brook
  31. Julien de Charentenay
  32. Cool.
  33. Open a blueCFD single core command prompt.
  34. Julien • 43 mins
  35. via the start menu options?
  36. I have:
  37. Command Line with mingw-w32 DP
  38. Command Line with mingw-w32 SP
  39. Command Line with mingw-w64 DP
  40. 42 mins
  41. Richard Brook
  42. Julien de Charentenay
  43. The last one. It uses 64bit and double precision.
  44. Julien • 42 mins
  45. ok done
  46. 42 mins
  47. Richard Brook
  48. Julien de Charentenay
  49. In the command prompt, change the directory to where the case is located. In my case, I typed: cd C:\Documents and Settings\Dell 330\My Documents\test\scenario_0
  50. Julien • 41 mins
  51. hmm, it wont change directory
  52. 39 mins
  53. Richard Brook
  54. Julien de Charentenay
  55. What is your directory?
  56. or what is the path to your case?
  57. Julien • 38 mins
  58. D:\scenario_0
  59. 38 mins
  60. Richard Brook
  61. Julien de Charentenay
  62. Try
  63. d:
  64. cd scenario_0
  65. Julien • 38 mins
  66. ah just d: switched it
  67. 38 mins
  68. Richard Brook
  69. Julien de Charentenay
  70. The good old DOS world!!!
  71. I sent you a file via email. Have you received it?
  72. Julien • 37 mins
  73. i have
  74. 37 mins
  75. Richard Brook
  76. Julien de Charentenay
  77. In D:\scenario_0\system there is an existing file snappyHexMeshDict.
  78. Can you make a copy of the existing one and copy the one I sent over in the directory?
  79. Julien • 36 mins
  80. done
  81. 36 mins
  82. Richard Brook
  83. Julien de Charentenay
  84. The difference between the two file is that the mesh generated by the "new" one will be coarser and will take less time to mesh & run. Once we go through it, you can try again with the "saved" one. ok?
  85. Julien • 35 mins
  86. ok, understood. I just added -original to the end of the name of the old one so no probs
  87. 34 mins
  88. Richard Brook
  89. Julien de Charentenay
  90. Back to the command prompt, type:
  91. blockMesh
  92. Julien • 33 mins
  93. done
  94. 33 mins
  95. Richard Brook
  96. Julien de Charentenay
  97. surfaceFeatureExtract -includedAngle 150 constant/triSurface/setup.stl setup
  98. Let me know if there is any error message...
  99. Julien • 32 mins
  100.  
  101. 31 mins
  102. Richard Brook
  103. Julien de Charentenay
  104. ok. Complaining about missing swak4Foam...
  105. But it looks like it is going through.
  106. Do you have a file D:\scenario_0\constant\triSurface\setup.eMesh ?
  107. If yes, it is "working" so far.
  108. Julien • 28 mins
  109. yes
  110. 28 mins
  111. Richard Brook
  112. Julien de Charentenay
  113. Moving on:
  114. copy system\decomposeParDict.snappyHexMesh system\decomposeParDict
  115. snappyHexMesh -overwrite
  116. Julien • 27 mins
  117. its working away
  118. 27 mins
  119. Richard Brook
  120. Julien de Charentenay
  121. Me too. I have an old computer, so you will probably get there before me.
  122. Julien • 26 mins
  123. mines about 6 years old
  124. 26 mins
  125. Richard Brook
  126. Julien de Charentenay
  127. Once it is complete, you should be able to load the mesh in ParaView (no solution yet, but you can check the mesh)...
  128. I stand a chance then...
  129. Julien • 25 mins
  130. just finished.
  131. 24 mins
  132. Richard Brook
  133. Julien de Charentenay
  134. Still waiting....
  135. Julien • 24 mins
  136. mine is a quad core Q6600 with 6gb ram.
  137. 24 mins
  138. Richard Brook
  139. Julien de Charentenay
  140. I give up. I have a Pentium 4 with 1GB ram...
  141. Julien • 23 mins
  142. ouch
  143. 23 mins
  144. Richard Brook
  145. Julien de Charentenay
  146. Mine is still going, but it may catch up later...
  147. renumberMesh
  148. Julien • 22 mins
  149. done
  150. 22 mins
  151. Richard Brook
  152. Julien de Charentenay
  153. checkMesh
  154. Julien • 22 mins
  155. done
  156. 22 mins
  157. Richard Brook
  158. Julien de Charentenay
  159. simpleFoam
  160. Julien • 21 mins
  161. ok, so I assume this is it running the test?
  162. 20 mins
  163. Richard Brook
  164. Julien de Charentenay
  165. From then, monitor the simulation to check that it converges and when it converges.
  166. Julien • 19 mins
  167. how will I know it has converged?
  168. 19 mins
  169. Richard Brook
  170. Julien de Charentenay
  171. And revert to the full mesh using the original snappyHexMeshDict.
  172. My snappyHexMesh has finished...
  173. Julien • 18 mins
  174. I assume consistent figures under forces and moment?
  175. 18 mins
  176. Richard Brook
  177. Julien de Charentenay
  178. In short there are two main ways:
  179. 1) Look at the residuals - in Khamsin I was able to show them as a graph. In this approach, they are shown as the screen outputs:
  180. Time = 5
  181.  
  182. DILUPBiCG: Solving for Ux, Initial residual = 0.151697, Final residual = 3.27574e-008, No Iterations 5
  183. DILUPBiCG: Solving for Uy, Initial residual = 0.0653019, Final residual = 8.91276e-008, No Iterations 4
  184. DILUPBiCG: Solving for Uz, Initial residual = 0.0984954, Final residual = 6.35355e-007, No Iterations 4
  185. GAMG: Solving for p, Initial residual = 0.634272, Final residual = 4.83282e-008, No Iterations 16
  186. In the above the residuals that matter are the "Initial residual". Until they drop below 1e-3, the simulation is not getting near convergence...
  187. 2) Once the initial residuals are going below 1e-3, look at flattening out of forces and moments. They should be output on the screen as well as in a file (look at D:\scenario_0\forces_....\0\forces.dat )
  188. Can you confirm that you can get the above?
  189. Julien • 13 mins
  190. i have the .dat file
  191. 12 mins
  192. Richard Brook
  193. Julien de Charentenay
  194. It is just a text file that you load (probably in excel as well). Have a play around and let me know if you need help.
  195. Julien • 11 mins
  196. ok
  197. 10 mins
  198. Richard Brook
  199. Julien de Charentenay
  200. On another note: your simulation will not include the following: wheel rotations (you would need swak4Foam), and mesh refinement of the front and rear wings.
  201. Julien • 10 mins
  202. thats fine
  203. will the simulation stop at a certain point? currently at Time = 103
  204. 9 mins
  205. Richard Brook
  206. Julien de Charentenay
  207. To achieve the later (mesh refinement), in the SketchUp model, create groups for the front and rear wings (separate groups are fine) and give the group a name that including "wing" such as "frontwing", "rear-wing".
  208. Simulation stopping. It is set to run for 1500 iteration (time = 1500), but feel free to stop it whenever you believe that it is complete using either Ctrl+C in the command prompt or kill it through the task manager. The "cleaner" way is to edit the file D:\scenario_0\system\controlDict and change the entry under "endTime" to a value less than the current Time.
  209. Julien • 7 mins
  210. thanks for all your help
  211. 6 mins
  212. Richard Brook
  213. Julien de Charentenay
  214. No drama.
  215. You may find that the meshing and running will take a lot longer with the "original" snappyHexMesh.
  216. Julien • 5 mins
  217. I will probably just stick to the usual way of running simulations through Khamsin as I have no problems with setting that up and getting a result. I will let Dan know about how this went though as he has trouble with the whole CFD installation and running for some reason even though I have shared all the info you have given me in the past
  218. 4 mins
  219. Richard Brook
  220. Julien de Charentenay
  221. One of the limitation of blueCFD-SingleCore is its inability of running on multiple computers. Linux would be a good alternative to all you to make the most of your resource.
  222. Let him know that I am available to have a chat online and go through a screencast exercise to set him up.
  223. Julien • 3 mins
  224. im definitely going to look into it. I have a number of computers and laptops to hand that I could create a cluster
  225. will do, thanks again Julien, I'll go have a play and see how things go
  226. 2 mins
  227. Richard Brook
  228. Julien de Charentenay
  229. If you are moving in that direction, you will probably exceed the current possibilities of Khamsin...
  230. Julien • 2 mins
  231. its just an idea 😀
  232. actually getting there will probably be a different story!
  233. 1 min
  234. Richard Brook
  235. Julien de Charentenay
  236. Always looking for bigger cases...
  237. I have some progress margin with my Pentium 4...
  238. Good luck with it and if you manage to create stunning visual, please share...
  239. Julien • Now
  240. Julien de Charentenay
  241. i will!
Advertisement
Add Comment
Please, Sign In to add comment