All users, but especially those with more limited hardware capability, could benefit from the ability to split large models up into sections along the direction of flow. Such sections could then be solved sequentially starting at the upstream end. If SimCFD provided the ability for users to define small subsections of the flow field at the downstream end of each analysis section, or ideally worked these out automatically, those subsections could also be integrated into the upstream end of the next analysis section, thus eliminating the need for the application of more general B.C.s which may not be knowable in advance. I have been asked to analyze a device in which there is no opportunity to invoke symmetry. Consequently, at mesh densities with any hope of returning credible results, the model is much too large for my hardware (> > 10 million elements). To complicate the matter in this particular case, the flow field involves multiple parallel paths, and the major objective of the analysis is to study the dispersion of one liquid phase into another. Due to this combination of factors, there is also no hope of applying general B.C.s to the upstream end of each analysis section in a split model which would reflect the true conditions (not so at the downstream end of each section where general B.C.s would be good enough to run the model). In this case, the conditions of interest are all local velocities, and the second phase concentration throughout the flow field. However, the ability to map temperature gradients and other thermal effects could also be very beneficial. It is already possible to map properties back and forth between SimCFD and SimMech, for example, as is required when running some thermal stress analyses. So why not extend this capability within SimCFD itself? Perhaps there is a parallel opportunity to split mechanical models in SimMech where the only other option is approximate the true conditions with general, manually applied B.C.s. I've been in that situation before as well.
Show More