A scalable scientific database for chemistry calculations in reacting flow simulations

Abstract

In many reacting flow simulations the dominant computational cost is the modelling of the underlying chemical processes. An effective approach for improving the efficiency of these simulations is the use of on-line, scientific databases. These databases allow for the approximation of the computationally expensive chemical process through interpolation from previously computed exact values. A sequential software implementation of these database algorithms has proven to be extremely effective in decreasing the running time of complex reacting flow simulations. To take advantage of high-performance, parallel computers, a parallel implementation has been designed and shown to be successful for steady-state problems. However, for transient problems this implementation suffers from poor load balancing. In this paper, we propose a modified algorithm for coordinating the building and the distributed management of the database. We present initial experimental results that illustrate improved load balancing algorithms for this new database design. © Springer-Verlag Berlin Heidelberg 2005.