Its sorters combine high accuracy with throughputs of up to 36 tons/hour. Sorted product commands a considerably higher price than unsorted product. Accuracy of sorting reduces waste and increases profit, and so is a fundamental issue across the food industry.

The sorting methodology comprises, first, optical imaging, then pattern recognition, and finally the activation of air jets to remove unwanted product. These jets have a velocity of 20 metres per second and fire in bursts of up to 300 cycles per second. Under the guidance of Prof Frank Smith at University College London, postgraduate KTN Associate Paul Westwood has shown how careful mathematical modelling aids the design of the jet nozzles. The key issue is the interference between adjacent nozzles and consequent lack of sharpness of the jet boundaries. By modelling this interference, we can understand how the nozzle geometry determines the maximum sorting accuracy that can be achieved. This information is crucial when designing new nozzles.