HPC will help red hot ideas at Pontypridd

They are full of red hot ideas at the University of Glamorgan in Pontypridd – and the launch of the HPC Wales super computer project will enable them to fire up even more innovations.
Steven Wilcox, the university’s Professor of Intelligent Systems Engineering, is thrilled at the benefits HPC will bring.
“Quite simply, work which could take us up to a week currently could be done in less than an hour,” he said.
“The difference in computer power is that dramatic and it will open up a whole host of new horizons for us.”
The Department of Engineering (www.engineering.research.glam.ac.uk) at the Treforest Campus is currently studying industrial furnaces in collaboration with EU industrial partners such as Corus.
Professor Wilcox explained: “We produce computer models to replicate the behaviour of glass furnaces which produce molten bottle glass. We can use the models to study the placing of auxiliary burners, how to use fuel more efficiently and how to reduce emissions.
“Computational fluid dynamic modelling enables us to see what happens to the natural gas and the air, how the furnace burns and what heat is released. We can produce optimum conditions in a model to ensure we produce the right quality and quantity of glass in combustion conditions which will ensure we are using fuel efficiently and reducing emissions.”
Glass furnaces melt the glass at temperatures of about 1500°C for subsequent processing into glass sheets for buildings, glass containers and bottles.
The University of Glamorgan is working in conjunction with Global Combustion Systems, a British burner manufacturer, and GdF Suez, a French gas supplier, to improve the efficiency of their furnaces and reduce their pollutant emissions.
The works requires a large number of simulations of alternative designs and conditions so the work is greatly facilitated by the use of high performance computing systems.
Professor Wilcox added: “Glass producers tend to run their furnaces for periods of anything up to 15 years, so it is very difficult to get practical time tinkering with how furnaces are set up.
“Computer modelling comes into its own in cases like this. We can use long computational runs to develop algorithms which can then be put to a practical use.
“You need thousands of these runs to train up an algorithm that can be of use on the actual glass furnace process.
“We are working closely with partners in the glass industry, including Pilkington Glass, to develop our work and we are sure HPC will bring huge benefits in this field.”
The team at the University of Glamorgan is also working on other types of furnaces, including steel reheating furnaces.
High temperature gas or oil fired furnaces are widely used in a range of industrial processes including the production of steel and other metals, glass, pottery, bricks and chemicals.
These systems are a major contributor to overall energy use in the industrial sector so the design and control of more efficient furnaces is receiving increasing attention.
A range of computer based techniques and models are available to simulate the thermal behaviour of industrial furnaces.
However these models are generally restricted to steady-state behaviour whereas many practical furnaces operate in a transient manner because of frequent changes in operating and production conditions.
Simulation of the transient performance of a large furnace such as those used in the steel industry is computationally intensive so that increasingly there is a need to use larger high performance computing systems.
The University of Glamorgan is currently working with Corus to develop a model which can potentially be used to simulate the dynamic control of one of their large furnaces heating steel bars to a temperature of about 1250°C.
This requires a model which can operate in at least “real-time” so high performance computer facilities are vital.
Professor Wilcox added: “Undoubtedly, we are excited by the opportunities opened up by the HPC project. It will be a hugely valuable tool in developing our collaborations with industry.”
See previous posts starting at -
http://sirgarblog.blogspot.com/2010/07/new-40m-super-computer-project-for.html