This paper discusses the petrol-station replenishment problem with adjustable demands (PSRP-AD). This problem originates from a practical application of fuel delivery, where the main objective is to satisfy all petrol-station demands at a minimal cost. However, the problem includes tanks with adjustable loads while respecting the agreement made between the company and the customers. This agreement allows the company to reduce the delivery up to a given threshold less than the ordered demand. For the PSRP-AD, we first describe the problem and provide the mathematical modelling that distinguishes between the loading and routing phases, followed by an improved tabu search to solve it. Within the framework of the tabu search, we embed the Kolmogorov–Smirnov statistic in the classical moves in order to speed up the search process. The results show that our solution is competitive on the benchmark instances and outperforms the current delivery method used by the company with a significant improvement.

本文讨论了需求可调的加油站补给问题（PSRP-AD）。这个问题源于燃料输送的实际应用，其主要目的是以最小的成本满足所有加油站的需求。但是，问题在于在遵守公司与客户之间达成的协议的同时，可调整负载的储罐。该协议允许公司将交货减少到给定的门槛值以下，而不是订购的需求。对于PSRP-AD，我们首先描述问题并提供区分加载阶段和路由阶段的数学模型，然后进行改进的禁忌搜索以解决该问题。在禁忌搜索的框架内，我们将Kolmogorov–Smirnov统计信息嵌入经典动作中，以加快搜索过程。