Abstract

We model, analyze and study the effects of considering condition-based replacement of parts within an integrated Service Parts Logistics (SPL) system, where geographically dispersed customers’ products are serviced with new parts from network facilities. Conventional SPL models consider replacing the parts upon failure. This is true even for the latest models in which facility locations and their part stock levels are jointly optimized. Taking advantage of the increasingly affordable, continuous, and accurate collection of part condition data (via sensors and Internet-of-Things devices), we develop a new integrated model in which optimal conditions to replace the parts are decided along with facility locations and stock levels. We capture the part degradation, replacement and failure process using a Continuous Time Markov Chain (CTMC) and embed this into the integrated location and inventory model. The resulting formulation is a mixed-integer optimization model with quadratic constraints and is solved with a state-of-the-art second-order cone programming solver. Our extensive comparison with the traditional failure-based replacement model shows that optimizing replacement conditions in this integrated framework can provide significant cost savings (network, inventory, transportation and downtime costs) leading to different facility location, allocation and inventory decisions. We also study the effects of several important parameters on the condition-based replacement model, including facility costs, shipment speeds, replacement costs, part degradation parameters, and holding costs.

摘要

我们在集成的“服务零件物流”（SPL）系统中建模，分析和研究考虑按条件更换零件的影响，在该系统中，通过网络设施中的新零件为地理上分散的客户产品提供服务。传统的SPL模型考虑在发生故障时更换零件。即使对于将设备位置及其零件库存水平共同优化的最新型号，也是如此。利用价格越来越便宜，持续且准确的零件状态数据收集（通过传感器和物联网设备），我们开发了一种新的集成模型，在该模型中，确定更换零件的最佳条件以及设施位置和库存水平。我们捕捉到零件退化，使用连续时间马尔可夫链（CTMC）进行更换和故障处理，并将其嵌入到集成的位置和库存模型中。生成的公式是具有二次约束的混合整数优化模型，并使用最新的二阶锥规划求解器进行求解。我们与传统的基于故障的替换模型的广泛比较表明，在此集成框架中优化替换条件可以节省大量成本（网络，库存，运输和停机成本），从而导致不同的设施位置，分配和库存决策。我们还研究了几个重要参数对基于条件的替换模型的影响，包括设施成本，运输速度，替换成本，零件降级参数和持有成本。