Gerard M. Campbell

Cyclic Assembly Schedules for Dynamic Demands

Cyclic schedules, where fixed order interval lengths apply for all items, offer practical benefits in terms of ease of planning and control. Cyclic schedules are frequently preferred in practice, even though they result in higher total costs than non-cyclic schedules when demand is not constant. This study investigates the additional cost of cyclic scheduling in multistage assembly systems where demands vary from period to period, as in material requirements planning (MRP) environments. Capacity constraints are not included in the study.A mathematical formulation of the cyclic problem is developed, and an algorithm to solve the problem is presented. Testing is then performed to compare the costs of optimal cyclic and non-cyclic schedules. Coefficient of demand variation is the key factor affecting the additional cost of cyclic schedules. Results show that additional costs are quite moderate under a wide range of conditions.

Cyclical queueing systems

Abstract Research on cyclical queueing systems is broken down into two categories: (1) cycling server models, and (2) cycling customer models. Within each category, details of a basic model are presented and then specialized models are discussed. This review presents a diverse sample of cyclical queueing models, with emphasis on current work and research directed towards applications.

Establishing safety stocks for master production schedules

Abstract The basic master production scheduling problem assumes that periodic demands are known with certainty. Uncertainty in the forecasts arc typically accommodated afterwards by adding safety stocks to a schedule. Two popular methods for establishing safety stocks are: (1) the constant cycle service level method; and (2) the constant safety stock method. This paper outlines these methods and then develops a third method which results in optimal safety stocks. The paper includes an experimental investigation aimed at comparing performances of the three safety stock methods. The constant safety stock method is shown to perform within one or two per cent of optimal, while the constant cycle service level method performs worse under most conditions. Shorter lead times, variable order interval lengths, and time-dependent forecast errors all adversely affect the performances of the non-optimal methods. An operations manager could use these results to evaluate the appropriateness of the methods for his maste...