A Logic Programming Approach to Global Logistics in a Co-Design Environment

TL;DR

This paper presents a logic programming approach using Answer Set Programming to optimize global aircraft manufacturing logistics within a co-design environment, aiming to improve resilience and efficiency.

Contribution

It introduces a novel ASP-based method for modeling and optimizing complex industrial logistics in co-design settings, integrating knowledge graph extraction and visualization.

Findings

Promising initial results in logistics optimization

Effective extraction and translation from knowledge graphs

Open research questions for future improvements

Abstract

In a co-design environment changes need to be integrated quickly and in an automated manner. This paper considers the challenge of creating and optimizing a global logistics system for the construction of a passenger aircraft within a co-design approach with respect to key performance indicators (like cost, time or resilience). The product in question is an aircraft, comprised of multiple components, manufactured at multiple sites worldwide. The goal is to find an optimal way to build the aircraft taking into consideration the requirements for its industrial system. The main motivation for approaching this challenge is to develop the industrial system in tandem with the product and making it more resilient against unforeseen events, reducing the risks of bottlenecks in the supply chain. This risk reduction ensures continued efficiency and operational success. To address this challenging and complex task we have chosen Answer Set Programming (ASP) as the modeling language, formalizing the relevant requirements of the investigated industrial system. The approach presented in this paper covers three main aspects: the extraction of the relevant information from a knowledge graph, the translation into logic programs and the computation of existing configurations guided by optimization criteria. Finally we visualize the results for an effortless evaluation of these models. Internal results seem promising and yielded several new research questions for future improvements of the discussed use case.