Image Credit BVM-Epelem Ltd.

"Most engineering algorithms have been developed as turbocharged versions of old hand-calculation methods," says Gabor Domokos from the Budapest University of Technology and Economics' Innovation and Knowledge Centre of Information Technology. "We thought, now that we have these enormous computational resources, why don't we think in a different way?"

Pre-stressed reinforced concrete beams, used primarily as the main building structure in highway bridges, can be over 30 meters long and weigh many tons. The reinforcing comes from the insertion of a steel bar inside the concrete, and it is this bar that placed under high tension, or pre-stressed, during construction.

"When the beams are used in bridges, the center of the beam tends to bend down," explains Domokos. "To counterbalance that, high tension is applied to the steel beam to make it permanently bend upward. The forces used to do this are enormous, and very small errors may result in large deformations and permanent damage."

Deformed beams can't be used, wasting tens of thousands of dollars for the company fabricating them. The beams are also incredibly strong in the vertical direction, but very fragile in the horizontal direction, making them very sensitive to any imbalance in horizontal pre-stressing forces. These torsional, asymmetric deformations were not exactly calculated in the past due to the complex mathematics involved.

"Calculating the effect of asymmetric stresses on the beams is very hard, and in the past companies have relied on specialized, highly-skilled engineers and their years of experience to get it right," notes Domokos. "We have adapted a new algorithm to create a tool that will help the engineers, not replace their intuition and experience."

The algorithm allows engineers to pre-test different stressing forces, especially torsional, asymmetric deformations, on a model of a reinforced concrete beam. The engineers use the output to augment their knowledge when carrying out the final pre-stressing. The team's algorithm had been previously applied to DNA, microfibers, and the shape of climbing plants, and they've spent the past two years adapting it for use by the concrete beam industry.

"We used the grid for the computation because we knew that companies can't afford to wait for computation and they don't want to buy the hardware," says Domokos. "The algorithm would use a large amount of resources for a short time, something the grid can easily give them."

Engineers at BVM-Epelem Ltd., the team's industrial partner, are now testing the grid-enabled tool. The engineers interact only with a simple user interface designed around their usual working methods. The tool runs on a small grid site at the Budapest University of Technology and Economics that runs the gLite middleware developed by the EGEE project.

"My goal is to have this tool running by next year," adds Domokos. "I would like to see beams transported out of the factory that have been built aided by our software."