The paper represents a new application, motion planning under fault, of their steel casting estimator matched to thick slab caster from the Japanese company JFE.
Bentsman said the paper’s success could be traced to an initial breakthrough in research with Thomas. “He had developed a high fidelity 1D dynamic transverse single cross-section (“single- slice”) model of continuous caster, which nicely yielded a step response to the boundary input. But I needed the step response of the entire caster, i.e. the spatially distributed step response to the entire spatially distributed spray input. This is exactly what we got by assembling 200 cross-sections (200 1D PDEs) moving down the caster into a single 2D dynamic model with the help of our PhD student Kai Zheng through spatial interpolation. It yielded an unusual looking caster step response, and, more generally, a real-time estimate of surface and internal temperature of the solidifying shell along with the solidification front position. Then we added a controller based on this estimate, along with a supervisory monitor loop.”
Their methodology has been implemented at Arcelor/Mittal, the world’s largest steel producer, where it has been in use for more than seven years. It has also been used at Nucor Steel Decatur for at least that long, running in production as a real-time monitor, and has been tested in the closed loop.
The award was presented last summer at the AIST annual conference, a large international event with over 6,000 participants. “Our paper competed with papers on a host of other computer applications, so it was really exciting to see our computer control application win,” said Bentsman.