Dynamic solver selection for an Internet simulation backbone

This current research is to develop automated solver selection capabilities necessary to achieve a simulation backbone in which domain experts can plug in simulation solvers they have developed and have these automatically available for use to designers and researchers whose simulation requirements take them into that discipline. The ultimate goal is a one-button press execution of a simulation performed using the correct simulation solvers on machines distributed over the Internet using software automatically taken from the Internet. In order to achieve this objective, we have started to develop a framework to identify and characterize a system of computer-interpretable rules that describe when a theoretical and computational model is applicable to a design simulation. For example, if a design model comprises some geometric volume and fluid flow conditions are specified, a rule can indicate that the problem is in the discipline of fluid dynamics. Depending on the specific shape and size of the volume and the velocity and viscosity of the fluid, rules can be used to further distinguish the problem as being in the sub-discipline of, for example, laminar, turbulent, or compressible flow. The existence of side effects, such as changes in heat transfer properties or aero-elastic effects, can also be inferred using rules that consider the results from a fluid dynamics simulation. The research scope has been limited to selected disciplines in mechanical engineering