How Agent-Based Simulation Can Optimize Queue Management

by Joni Newkirk, CEO, Integrated Insight

Summertime is often synonymous with long lines as Americans begin to travel. Long wait times increase at the airport, amusement parks, and even entrances to beaches and state and national parks. And when summer turns to Fall and Winter, add retail checkout lanes to the mix of queues that need to be optimized.

Psychology dictates how consumers perceive waiting time in lines. As Lavi Industries points out in “The Art and Science of Queuing,” consumers want to feel as though they are in control. They want to start right away, or at least be immediately acknowledged. They need to know how long they will be waiting before deciding to get in line. Sensitivity increases if they feel someone else is cutting in front of them. And consumers will need distractions to make the actual wait time feel like less of a burden.

Setting psychology aside, effectively managing queue lines is the most definitive step you can take to enhance the customer experience. However, it’s not always intuitive – and certainly isn’t easy to optimize just by sight.

Cutting wait times significantly is possible with the right process. At a minimum, a queuing process revolves around two forces:

1. The arrival rate of patrons.
2. The amount of time it takes to serve one customer.

Both of these factors can vary. The added complication is how the service is delivered; primarily, how the servers are arranged and how guests in the waiting lines approach the servers. Through agent-based simulation, it is easier to both see and record the impact of different queue processes.

Case Study – Using Agent-Based Simulation to Optimize Queue Lines of a School Lunch Pickup

In 2020, a free lunch pickup program was initiated at Florida schools. This required schools to quickly determine a distribution plan.

We looked at the distribution process for a local Orlando high school and used agent-based modeling to identify bottlenecks. Initially, the high school used a single line for lunch pickup. At the first stop, guests provided their names and the number of meals being picked up. Next, a monitor directed the driver to one of two stations further down the lane for pickup. These two stations distributed the same meal.

In this queue line, valuable time was lost if the pickup occurring in the second station is slower than the first station, as the next car to pick up is blocked by the car ahead. Under this scenario, the full distribution process can take hours to complete, and parents are consuming time sitting in their cars.

We built the distribution environment in agent-based simulation software to see the impacts that single-lane queuing had on the time parents had to wait in their cars. In the video below, you can see wait times reach higher than 30 minutes as the cars stack up.

Using agent-based simulation, we created a model with parallel fulfillment (using two lanes), to see if lunches could be distributed more efficiently.