# Fly Me to the Queue

#### by Amanda G. Smith

Several years ago, I took an introductory simulation course that allowed me to participate in a very interesting study of queuing. My team analyzed a unique queue: the waiting area at the baggage carousel at the Dane County Airport. If you’ve ever traveled by plane and checked a bag, you know very well how much traffic can build up at baggage carousels. In order to improve airport efficiency, it would be beneficial to find ways to reduce the amount of time passengers spend in the system. Interestingly, the system departure rates seemed to follow a Weibull distribution, not often seen in day-to-day queues. However, the Weibull distribution makes sense because the time that a customer receives his or her final bag can depend on several different factors such as the time the customer arrives at the baggage claim, how many bags he has, and when his bags arrive on the carousel. The queue is also unique because of the implicit characteristic of a finite calling population; we can know in advance how many customers and how many bags will arrive in the system, since it is exactly what was on the plane before arrival.

Passengers and bags arrive at the system with every plane. After arriving, both passengers and bags must make their way to the baggage carousel. Passengers at the Dane County Airport have a relatively short walk from the terminal to the carousel, though if passengers stop to use the bathroom or get food, the transport time can be significantly increased. Meanwhile, the luggage is loaded onto a cart and driven from the plane to the baggage unloading gate. It is then unloaded onto the carousel, where it remains until retrieved by the passenger to whom it belongs. Passengers wait at the carousel until their bags appear so they can depart the system. We collected our data by observing the carousel area and noting the plane arrival time, baggage arrival time, and passenger and bag departure time. We also kept track of how many bags each passenger retrieved.

Our main goal was to understand which aspect of the system was responsible for the congestion around baggage carousels. Our first idea was to cut the baggage transport time in half. After running our simulation with this new input parameter, the passenger wait time was decreased by a whopping 90%! Our second idea was to reduce the passenger travel time to a small constant, approximately 2 minutes. After testing, we saw a small reduction in overall system time, but wait time actually increased. Since people hate waiting in queues, that particular alternative seemed like a bad idea.

Our conclusion was that a 90% reduction of wait time is a significant enough improvement that it is worth any (reasonable) cost incurred by the airport. However, we also believed the cost for a reduction of baggage transport time could be kept low by performing incremental alterations, such as assigning one more employee to help load bags or increasing cart driving speed. To my knowledge, the Dane Count Airport did not try to implement any of our ideas, but we had a lot of fun experimenting with an unusual type of queue.

~AGS