29/04/2013 It looks like the model started on 22/04/2013 seems to have crashed during the solve. I'm not sure if this is one of those random crashed or is due to some change I've made to the meshing. Lets hope it's the former. I knew this model would have a long solve time but I just wish I'd seen it sooner...
22/04/2013 Committed comi code to Launchpad Repository via Bazaar. Revision 23. (Hall Model 79) Note: 14/05/2013. I've just commited to the repository again and this was revision 23 so it is likely that the last commit failed or wasn't actually done. However I definitely have a .comi snapshot from the 22/04/2013
17/04/2013 I have improved the meshing in the North & South Shield Walls. To achieve this the meshing has been switched to tetrahedral, and the layering in the plates acheived by creating layers of cells and joining them together. Using tetrahedral elements requires a bit more work to construct the object but I think I have more control over the code. (Back-layering only works with hexahedral). Note: Hexahedral elements was always a pain to work with in this model due to the lack of symmetry!
12/04/2013 I have improved the meshing in the EMR, beamdump, end of the Linac Wall and the Quads. This is now being run as model 77. Once this model has completed I will compare the ErrB/B plots to see if there has been any improvement in the output in the areas where it was indicated that the meshing was poor. The mesh improvements are controlled by a switch in the master file. Also all hexahedral meshing has been turned off in these components - default is now tetrahedral.
11/04/2013 Committed comi code to Launchpad Repository via Bazaar. Revision 22. (Hall Model 71)
10/04/2013 I've spent some time today resizing the boundary of the model to try and remove the effects of the boundary being to close. The model boundary is now at +/-60m in z and +/-40m in x and y. I have slightly increased the size of the elements in the outer air and the boundary volumes to try and keep the number of elements in the model approximately the same. Hopefully this will not have too large an impact upon the solve time. First trial is model 71, which is model 61 but with extended boundaries as described.
08/04/2013 Over the last couple of weeks I have been running models 61 to 66. These are identical to models 51 to 56 except that they have corrected boundary conditions and the boundary is a little further out. As of today, only model 62 has not yet solved.