Portability, predictability and performance for parallel computing: BSP in practice

We report on practical experience using the Oxford BSP Library to parallelize a large electromagnetic code, the British Aerospace finite-difference time-domain code EMMA T:FD3D. The Oxford BS Library is one of the first realizations of the Bulk Synchronous Parallel computational model to be targeted at numerically intensive scientific (typically Fortran) computing. The BAe EMMA code is one of the first large-scale applications to be parallelized using this library, and it is an important demonstration of the cost effectiveness of the BSP approach. We illustrate how BSP cost-modelling techniques can be used to predict and optimize performance for single-source programs across different parallel platforms. We provide predicted and observed performance figures for an industrial-strength, single-source parallel code for a variety of real parallel architectures: shared memory multiprocessors, workstation clusters and massively parallel platforms.

Computational fire engineering - do we have what we need?

This presentation will attempt to address the issue of whether the engineering design community has the knowledge, data and tool sets required to undertake advanced evacuation analysis. In discussing this issue I want to draw on examples not only from the building industry but more widely from where ever people come into contact with an environment fashioned by man. Prescriptive design regulations the world over suggest that if we follow a particular set of essentially configurational regulations concerning travel distances, number of exits, exit widths, etc it should be possible to evacuate a structure within a pre-defined acceptable amount of time. In the U.K. for public buildings this turns out to be 2.5 minutes, internationally in the aviation industry this is 90 seconds, in the UK rail industry this is 90 seconds and the international standard adopted by the maritime industry is 60 minutes. The difficulties and short comings of this approach are well known and so I will not repeat them here, save to say that this approach is usually littered with “magic numbers” that do not stand up to scrutiny. As we are focusing on human behaviour issues, it is also worth noting that more generally, the approach fails to take into account how people actually behave, preferring to adopt an engineer’s view of what people should do in order to make their design work. Examples of the failure of this approach are legion and include the; Manchester Boeing 737 fire, Kings Cross underground station fire, Piper Alpha oil platform explosion, Ladbroke Grove Rail crash and fire, Mont Blanc tunnel fire, Scandinavian Star ferry fire and the Station Nightclub fire.

Computational model for prediction of particle degradation during dilute-phase pneumatic conveying: modeling of dilute-phase pneumatic conveying

A complete model of particle impact degradation during dilute-phase pneumatic conveying is developed, which combines a degradation model, based on the experimental determination of breakage matrices, and a physical model of solids and gas flow in the pipeline. The solids flow in a straight pipe element is represented by a model consisting of two zones: a strand-type flow zone immediately downstream of a bend, followed by a fully suspended flow region after dispersion of the strand. The breakage matrices constructed from data on 90° angle single-impact tests are shown to give a good representation of the degradation occurring in a pipe bend of 90° angle. Numerical results are presented for degradation of granulated sugar in a large scale pneumatic conveyor.

Two-machine shop scheduling: Compromise between flexibility and makespan value

In this paper, we consider the problem of providing flexibility to solutions of two-machine shop scheduling problems. We use the concept of group-scheduling to characterize a whole set of schedules so as to provide more choice to the decision-maker at any decision point. A group-schedule is a sequence of groups of permutable operations defined on each machine where each group is such that any permutation of the operations inside the group leads to a feasible schedule. Flexibility of a solution and its makespan are often conflicting, thus we search for a compromise between a low number of groups and a small value of makespan. We resolve the complexity status of the relevant problems for the two-machine flow shop, job shop and open shop. A number of approximation algorithms are developed and their worst-case performance is analyzed. For the flow shop, an effective heuristic algorithm is proposed and the results of computational experiments are reported.

Predicting the reliability of power electronic modules

This paper discusses the reliability of an IGBT power electronics module. This work is part of a major UK funded initiative into the design, packaging and reliability of power electronic modules. The predictive methodology combines numerical modeling techniques with experimentation and accelerated testing to identify failure modes and mechanisms for these type of power electronic module structures. The paper details results for solder joint failure substrate solder. Finite element method modeling techniques have been used to predict the stress and strain distribution within the module structures. Together with accelerated life testing, these results have provided a failure model for these joints which has been used to predict reliability of a rail traction application