Francesco Giliberto: Temperature Control of Road Transport 3 October 2024 16:00 - 17:00

Maintaining temperatures within specified ranges is essential for many products to prevent quality degradation and increase shelf life. In this paper, we propose an optimal cooling policy for refrigerated trucks that deliver temperature-sensitive products to customers on a predefined route. Our policy is the first that explicitly accounts for the most important uncertainties such as the duration of door opening times while loading and unloading as well as the initial temperatures of the loaded products. Furthermore, our model features a careful modeling of thermodynamics within the truck, including the heat transfer between the air, the products, the outside environment, and the cooling unit. The resulting problem is cast as a large multi-stage stochastic programming problem that we solve using stochastic dual dynamic programming. In cooperation with industry partners, we set up an extensive battery of test cases in order to examine the quality of our solution. To that end, we benchmark our stochastic policy against a rolling lookahead policy and a myopic practitioner's benchmark out of sample. The results show that our policy clearly outperforms the benchmarks on all routes and in all truck configurations by a large margin, employing a cooling regimen that hedges against warming of the products due to unexpectedly long door opening times. This leads to significantly fewer violations of temperature bounds. In a separate analysis, we show that our policy enables energy savings of up to 25% in cooling unit operation, helping to make energy-intensive cold chains more sustainable.