O presente trabalho aborda o problema de determinação de datas de entrega e o sequenciamento de tarefas com tempos de processamento estocásticos. O ambiente considerado constitui em uma máquina simples e tarefas com custos individuais e distintos de adiantamento e atraso de entrega (earliness e tardiness ou simplesmente E/T). O objetivo é determinar a sequência e as datas de entrega ótimas que simultaneamente minimizam o custo total esperado de E/T. Para a determinação de sequências candidatas, são apresentadas diversas heurísticas construtivas com tempo de execução polinomial baseadas em um método de inserção de tarefas. Considerando tarefas com distribuição normal, experimentos computacionais comprovam a eficácia dos algoritmos para problemas de menor porte, os quais fornecem soluções ótimas em 99,85% dos casos avaliados. Quando aplicadas a um conjunto com uma maior quantidade de tarefas, as heurísticas apresentaram resultados melhores do que o algoritmo disponível na literatura em mais de 80% dos casos. Consideradas tarefas com distribuição lognormal, obteve-se um percentual de otimalidade entre 93,87% e 96,45%, a depender da heurística aplicada. Demonstra-se ainda para o caso com distribuição normal que os métodos propostos são assintoticamente ótimos e, portanto, são indicados para a resolução de problemas de grande porte. / This work addresses the problem of concurrent due-date assignment and sequencing of a set of jobs on a stochastic single-machine environment with distinct job earliness and tardiness penalty costs. It is assumed that the jobs processing times are statistically independent and follow a normal distribution whose mean and variance are provided and they are not necessarily integer values. The objective is to determine the job sequence and the integer due dates which minimize the expected total earliness and tardiness costs. Several efficient insertion-based construction heuristics are proposed in order to find candidates for the optimal sequence with polynomial time complexity. For the normal distribution problem, numerical experiments show that the proposed heuristic methods are able to find the optimal solution in 99,85% when applied to problems with a smaller size. When applied to problems with a bigger size, the solutions found by the proposed heuristics had better costs than the solutions described in the literature in more than 80% of cases. For the lognormal distribution problem, the proposed heuristic methods provided solutions with a percentage of optimality between 93,87% and 96,45%. Furthermore, for the normal distribuition case, it was proven that the heuristics are asymptotically optimal, i.e., it can be used for problems of any size.

O presente trabalho aborda o problema de determinação de datas de entrega e o sequenciamento de tarefas com tempos de processamento estocásticos. O ambiente considerado constitui em uma máquina simples e tarefas com custos individuais e distintos de adiantamento e atraso de entrega (earliness e tardiness ou simplesmente E/T). O objetivo é determinar a sequência e as datas de entrega ótimas que simultaneamente minimizam o custo total esperado de E/T. Para a determinação de sequências candidatas, são apresentadas diversas heurísticas construtivas com tempo de execução polinomial baseadas em um método de inserção de tarefas. Considerando tarefas com distribuição normal, experimentos computacionais comprovam a eficácia dos algoritmos para problemas de menor porte, os quais fornecem soluções ótimas em 99,85% dos casos avaliados. Quando aplicadas a um conjunto com uma maior quantidade de tarefas, as heurísticas apresentaram resultados melhores do que o algoritmo disponível na literatura em mais de 80% dos casos. Consideradas tarefas com distribuição lognormal, obteve-se um percentual de otimalidade entre 93,87% e 96,45%, a depender da heurística aplicada. Demonstra-se ainda para o caso com distribuição normal que os métodos propostos são assintoticamente ótimos e, portanto, são indicados para a resolução de problemas de grande porte. / This work addresses the problem of concurrent due-date assignment and sequencing of a set of jobs on a stochastic single-machine environment with distinct job earliness and tardiness penalty costs. It is assumed that the jobs processing times are statistically independent and follow a normal distribution whose mean and variance are provided and they are not necessarily integer values. The objective is to determine the job sequence and the integer due dates which minimize the expected total earliness and tardiness costs. Several efficient insertion-based construction heuristics are proposed in order to find candidates for the optimal sequence with polynomial time complexity. For the normal distribution problem, numerical experiments show that the proposed heuristic methods are able to find the optimal solution in 99,85% when applied to problems with a smaller size. When applied to problems with a bigger size, the solutions found by the proposed heuristics had better costs than the solutions described in the literature in more than 80% of cases. For the lognormal distribution problem, the proposed heuristic methods provided solutions with a percentage of optimality between 93,87% and 96,45%. Furthermore, for the normal distribuition case, it was proven that the heuristics are asymptotically optimal, i.e., it can be used for problems of any size.

Most research studies in the batch process control problem are focused on optimizing system performance. The methods address the problem by minimizing single criterion such as cycle time and tardiness, or bi-criteria such as cycle time and tardiness, and earliness and tardiness. This research demonstrates the use of Stochastic Utility Evaluation (SUE) function approach to optimize system performance using multiple criteria. In long production cycles, the earliness and tardiness weight (utility) of products vary depending on the time. As the time approaches the due-date, it affects contractual penalties, loss of customer goodwill and the storage period for the completed products. It is necessary to reflect the weight of products for earliness and tardiness at decision epochs to decide on the optimal strategy. This research explores how stochastic utility function using stochastic information can be derived and used to strategically improve existing approaches for the batch process control problem. This research first explores how SUE function can be applied to existing model for bi-objective problem such as cycle time and tardiness. Benchmark strategies using SUE function (NACH-SUE, MBS-SUE, No idle and full batch) are compared to each other. The experimental results show that NACH-SUE effectively improves mean cycle time and tardiness performance respectively than other benchmark strategies. Next, SUE function for earliness and tardiness is used in an existing model to develop a tri-objective problem. Typically, this problem is very complex to solve due to its trade-off relationship. However SUE function makes it relatively easy to solve the tri-objective problem since SUE function can be incorporated in an existing model. It is observed that SUE function can be effectively used for solving a tri-objective problem. Performance improvement for averaged value of cycle time, earliness and tardiness is observed under a comprehensive set of experimental conditions.