Agent-Based Simulation of SARS-CoV-2 Spread in Supermarket Checkout Areas

Abstract: The outbreak of the coronavirus disease 2019 (COVID-19) has seen the world scramble for effective countermeasures to limit infection spread in society. Understanding how infection spreads in places where strangers meet in relatively high numbers and proximity to one another is especially important. Supermarkets are one such place where strangers inevitably gather in close proximity indoors. In particular, the checkout area where people queue up to pay tends to be densely populated, making it especially hazardous. One approach to understanding the infection spread is to use agent-based computer simulations to model different scenarios. This paper describes one such simulation of a supermarket checkout area using the Unity 3D engine, including the effect of checkout types and quantity, customer load and COVID-19 countermeasures, i.e., masking and distancing, on infection spread. Using the results from one default scenario and eleven variations, the relative impact of aforementioned factors on exposure in the simulation is discussed. Results indicate that for this simulation the most important factor is preventing queue buildup via having sufficient customer throughput capacity, with potent effects also resulting from operating service registers in such a way that the distance between each queue is maximized as well as increasing distances between agents within queues. Including a self-checkout area was found to be a viable approach to reducing queue times and consequently exposure rates. Comparatively, masking did not yield as notable reductions in exposure rates in the simulation. Similarities in exposure patterns to previous work in the context of supermarkets are discussed, as well as limitations of simulations in capturing the real world.