Converting best manual practice into generic automatable strategies for unstructured mesh parallelisation

The manual effort required to convert sequential computational mechanics programs into a useful, scalable parallel form is considerable. Tools that can assist in the conversion process are clearly required. Computer aided parallelisation tools (CAPTools) have been developed to generate efficient parallel code for real world structured grid application codes such as Computational Fluid Dynamics. Automatable single-program multi-data (SPMD) overlapping domain decomposition (DD) techniques established for structured grid codes have been adapted by the authors to manually parallelise unstructured mesh applications. Inspector loops have been used to provide generic techniques for the run-time support necessary to extend the capabilities of CAPTools to automatic implementation of SPMD DD techniques in the parallelisation of unstructured mesh codes. Copyright © 1999 John Wiley & Sons, Ltd.

Three parallel programming paradigms: Comparisons on an archetypal PDE computation

Three paradigms for distributed-memory parallel computation that free the application programmer from the details of message passing are compared for an archetypal structured scientific computation -- a nonlinear, structured-grid partial differential equation boundary value problem -- using the same algorithm on the same hardware. All of the paradigms -- parallel languages represented by the Portland Group's HPF, (semi-)automated serial-to-parallel source-to-source translation represented by CAP-Tools from the University of Greenwich, and parallel libraries represented by Argonne's PETSc -- are found to be easy to use for this problem class, and all are reasonably effective in exploiting concurrency after a short learning curve. The level of involvement required by the application programmer under any paradigm includes specification of the data partitioning, corresponding to a geometrically simple decomposition of the domain of the PDE. Programming in SPMD style for the PETSc library requires writing only the routines that discretize the PDE and its Jacobian, managing subdomain-to-processor mappings (affine global-to-local index mappings), and interfacing to library solver routines. Programming for HPF requires a complete sequential implementation of the same algorithm as a starting point, introduction of concurrency through subdomain blocking (a task similar to the index mapping), and modest experimentation with rewriting loops to elucidate to the compiler the latent concurrency. Programming with CAPTools involves feeding the same sequential implementation to the CAPTools interactive parallelization system, and guiding the source-to-source code transformation by responding to various queries about quantities knowable only at runtime. Results representative of "the state of the practice" for a scaled sequence of structured grid problems are given on three of the most important contemporary high-performance platforms: the IBM SP, the SGI Origin 2000, and the CRAYY T3E.

Multiphase mesh partitioning

We consider the load-balancing problems which arise from parallel scientific codes containing multiple computational phases, or loops over subsets of the data, which are separated by global synchronisation points. We motivate, derive and describe the implementation of an approach which we refer to as the multiphase mesh partitioning strategy to address such issues. The technique is tested on several examples of meshes, both real and artificial, containing multiple computational phases and it is demonstrated that our method can achieve high quality partitions where a standard mesh partitioning approach fails.

Multiphase mesh partitioning for parallel computational mechanics codes

We consider the load-balancing problems which arise from parallel scientific codes containing multiple computational phases, or loops over subsets of the data, which are separated by global synchronisation points. We motivate, derive and describe the implementation of an approach which we refer to as the multiphase mesh partitioning strategy to address such issues. The technique is tested on example meshes containing multiple computational phases and it is demonstrated that our method can achieve high quality partitions where a standard mesh partitioning approach fails.

Maximising heat transfer efficiency in the cold crucible induction melting process

Induction heating is an efficient method used to melt electrically conductive materials, particularly if melting takes place in a ceramic crucible. This form of melting is particularly good for alloys, as electromagnetic forces set up by the induction coil lead to vigorous stirring of the melt ensuring homogeneity and uniformity in temperature. However, for certain reactive alloys, or where high purity is required, ceramic crucibles cannot be used, but a water-cooled segmented copper crucible is employed instead. Water cooling prevents meltdown or distortion of the metal wall, but much of the energy goes into the coolant. To reduce this loss, the electromagnetic force generated by the coil is used to push the melt away from the walls and so minimise contact with water-cooled surfaces. Even then, heat is lost through the crucible base where contact is inevitable. In a collaborative programme between Greenwich and Birmingham Universities, computer modelling has been used in conjunction with experiments to improve the superheat attainable in the melt for a,number of alloys, especially for y-TiAl intermetallics to cast aeroengine turbine blades. The model solves the discretised form of the turbulent Navier-Stokes, thermal energy conservation and Maxwell equations using a Spectral Collocation technique. The time-varying melt envelope is followed explicitly during the computation using an adaptive mesh. This paper briefly describes the mathematical model used to represent the interaction between the magnetic field, fluid flow, heat transfer and change of phase in the crucible and identifies the proportions of energy used in the melt, lost in the crucible base and in the crucible walls. The role of turbulence is highlighted as important in controlling heat losses and turbulence damping is introduced as a means of improving superheat. Model validation is against experimental results and shows good agreement with measured temperatures and energy losses in the cooling fluid throughout the melting cycle.

On Reynolds number and scaling effects in microchannel flows

This paper presents a numerical study of the Reynolds number and scaling effects in microchannel flows. The configuration includes a rectangular, high-aspect ratio microchannel with heat sinks, similar to an experimental setup. Water at ambient temperature is used as a coolant fluid and the source of heating is introduced via electronic cartridges in the solids. Two channel heights, measuring 0.3 mm and 1 mm are considered at first. The Reynolds number varies in a range of 500-2200, based on the hydraulic diameter. Simulations are focused on the Reynolds number and channel height effects on the Nusselt number. It is found that the Reynolds number has noticeable influences on the local Nusselt number distributions, which are in agreement with other studies. The numerical predictions of the dimensionless temperature of the fluid agree fairly well with experimental measurements; however the dimensionless temperature of the solid does exhibit a significant discrepancy near the channel exit, similar to those reported by other researchers. The present study demonstrates that there is a significant scaling effect at small channel height, typically 0.3 mm, in agreement with experimental observations. This scaling effect has been confirmed by three additional simulations being carried out at channel heights of 0.24 mm, 0.14 mm and 0.1 mm, respectively. A correlation between the channel height and the normalized Nusselt number is thus proposed, which agrees well with results presented.