Improving the perfomance of elevator systems using exact reoptimization algorithms

The task of an elevator control is to schedule the elevators of a group such that small average and maximal waiting and travel times for the passengers are obtained. We present a novel exact reoptimization algorithm for this problem. A reoptimization algorithm computes a new optimal schedule for the elevator group each time a new passenger arrives. Our algorithm uses column generation techniques and is, to the best of our knowledge, the first exact reoptimization algorithm for a group of elevators. We use our algorithm to compare the potential performance that can be achieved for conventional (ie up/down buttons) and two variants of destination call systems, where a passenger enters his destination floor when calling an elevator. This research is part of an ongoing project with our industry partner Kollmorgen Steuerungstechnik.