Multiobjective Simheuristics for Cold Chain Distribution under Temperature Uncertainty

In this work, we propose the Cold Chain VRPTW with Stochastic Temperature (CCVRPTW-ST), an extension of the vehicle routing problem with time windows applied to cold chain distribution, in which the internal temperature of vehicles is modeled as a random variable. The problem is formulated as a three-objective model, while minimizing logistics cost, loss of quality of perishable products and CO² emissions. Thermal degradation is represented by the Arrhenius equation, while temperature uncertainty is incorporated through Monte Carlo simulation coupled to NSGA-II. 54 benchmark instances were generated and four algorithmic variants were evaluated. The results indicate that the stochastic versions outperform the deterministic variant with statistical significance (p<0.05), with no significant difference between the use of ten and thirty Monte Carlo replications.