Production Planning in Glass Container Industries Under Uncertainty: A Probability-Constrained Approach

This article addresses the planning of the production of jars and ampoules in a Brazilian glass container industry, considering uncertainties associated with demand, scrap, productive yield and ramp-up losses. A mathematical formulation for the Stochastic Problem of Pot and Ampoule Production (PEPPA) is proposed, based on probabilistic constraints and risk allocation, aiming to limit the probability of violation of operational requirements. The problem is solved using exact method (Branch-and-Cut), Genetic Algorithm (AG) and Multipopulation Genetic Algorithm (AGMP). Experiments with 200 instances indicate that meta-heuristics obtain viable solutions in all evaluated cases and have competitive computational performance in larger instances. The Monte Carlo simulation shows that stochastic approaches achieve average success rates close to 99%, higher than the deterministic formulation of the problem.