Real-time order allocation model by considering available-to-promise reserving, occupying and releasing mechanisms

Based on the available-to-promise (ATP for short) theory, a three-stage model is proposed to help manufacturers manage Assemble-to-Order supply chain with the high and low class of customers. A mixed integer programming model is formulated to find the optimal ATP reserving policy from history data in the first stage. And two other policies commonly seen in the literature are also constructed. One is based on the probability of high class customers and the other is constructed from the percentage of high class customers in history data. Then in the second stage, each reserving policy is applied with the occupying mechanism to deal with the real orders. Numerical results show that though these policies bring a higher acceptance rate of high class customers, the profit of the manufacturer could be hurt. Among the three reserving polices, our policy offers a balanced way for keeping a high acceptance rate with limited loss. Then in the third stage, the releasing mechanism is considered together with the reserving policy. In this case, no reserving policy becomes the worst and the policy we proposed achieves an overall well performance on average. Numerical experiments also show that our policy tends to reserve more than expected amount for high class customers when the probability of high class customers is low, which helps to improve the service level for high class customers, and less than the expected amount when the probability of high class customers is high, which takes the profit from low class customers into account.