On April 24th, Mark Lewis of Cornell University came to talk to the department about his research, as well as about the Operations Research program at Cornell University. His abstract follows:
In this talk we consider a two station call center with upgrades. Each station has dedicated (Poisson) arrival streams. Station 1 is equipped with cross trained workers that can serve customers from either station while Station 2 has only dedicated servers. This is the classic "N-network" with the following twist: customers that begin at Station 1 are upgraded to Station 2 after an exponential amount of time. With this innovation we show that different stability conditions can be derived depending on the order of the service rates. We then show that optimal allocation policies follow the c-mu rule, but only under one direction of the index AND a similar assumption on the service rates. Finally, via a short numerical study we show that upgrades are useful in both decreasing average costs and average waiting times...sometimes even of both customer classes. This is joint research with Douglas Down from McMaster University.