High performance computer simulations slash costs

Computer simulations are increasingly used to model systems and processes throughout industry to improve productivity and reduce costs.

But until a team led by Brunel academic Dr Simon Taylor got to work, two issues were limiting their use on large-scale systems.

The first was actually building the simulations themselves. Models are costly to develop so being able to re-use elements of one model as the building blocks towards creating larger models would solve that issue, but there were difficult barriers to overcome.

The second major issue is running the simulations themselves. All simulations take time to run, which can vary from seconds to hours depending on size and complexity.

Answering the first required a major global standardisation initiative involving collaborative working across continents, not only between academics and companies but between firms who often found themselves in commercial competition with each other. But by regarding interoperability as a community-wide issue, major strides were made in model element re-use.

Addressing the speed issue required grid computing (using multiple computers to speed up processing an application) but existing grid solutions were not suitable, so Dr Taylor developed a Windows-based desktop computer grid.

This work has led to a step-change in the size of systems that can be successfully modelled. Simulations are now able to run far faster and with more precision, and with added capability to examine a wider variety of options.

Significant cost savings using Dr Taylor’s innovations are now being achieved by Ford in its global engine manufacturing operation, Sellafield in its nuclear waste disposal programme and the NHS in its manpower capability planning.