There is a vast and growing number of industrial applications for computer simulation of real life problems in fluid dynamics and structural mechanics. While computer programs for performing such simulations exist on the market, their usage is limited due to lack of computer power. Only parallel computers can realistically satisfy future industrial demand for complex computer simulations, but the development of parallel commercial software is frequently seen as being too costly. The parallelization of the PHOENICS code is being used to demonstrate how the enhanced performance and efficiency of parallel architectures can offer significant cost benefits. PHOENICS is one of the well-known commercial, general-purpose fluid dynamics codes capable of simulating fluid flow, heat transfer, combustion and chemical reaction. The code is widely used and has been exemplified and validated across a broad range of industrial sectors including aerospace, automotive, railway, civil engineering, chemical, pharmaceutical and food processing. Many industrial applications now stand to benefit from the parallel, distributed memory version of PHOENICS (based on sequential Version 2.1). The resulting parallel code will be benchmarked on the basis of real industrial applications to enable end-users to judge for themselves the potential benefits of parallel technology. Parallel PHOENICS is, for example, being used to simulate viscous and turbulent flow around a ship's hull, showing that, in this case, the use of parallel computers is more cost-effective than the use of a standard workstation by a factor of around 2.5.
N3S is a computational fluid dynamics package used for the simulation of complex industrial flows. It is widely used in various industries including the automotive, aerospace, defence, energy, oil, environmental and medical sectors. Typical examples of flows simulated by N3S include external flows around aircraft, cars and buildings, and internal flows inside cylinder heads, turbomachinery and combustion chambers, as well as natural convection flows. Migration of the code for use on parallel computers has now been achieved. The resulting parallel code, HPCN3S, promises significant gains in productivity in all its applications.