Flow shops with reentry: The total weighted completion time objective

TL;DR

This paper studies the scheduling problem in flow shops with reentry, where jobs revisit machines multiple times, aiming to minimize total weighted completion time, and introduces new algorithms and complexity results for this problem.

Contribution

It proves NP-hardness for reentrant flow shops with multiple machines, proposes the LRL and WLRL priority rules, and develops an FPTAS and pseudo-polynomial algorithms for fixed machine counts.

Findings

LRL rule minimizes total unweighted completion time

WLRL rule has a worst-case ratio of about 1.2

FPTAS provides near-optimal solutions efficiently

Abstract

Flow shops are widely studied machine environments in which all jobs must visit all machines in the same order. While conventional flow shops assume that each job traverses the shop only once, many industrial environments require jobs to loop through the shop multiple times before completion. This means that after traversing the shop and completing its processing on the last machine, a job must return to the first machine and traverse the shop again until it has completed all its required loops. Such a setting, referred to as a flow shop with reentry, has numerous applications in industry, e.g., semiconductor manufacturing. The planning problem is to schedule all loops of all jobs while minimizing the total weighted completion time. In this paper, we consider reentrant flow shops with unit processing times. We show that this problem is strongly NP-hard if the number of machines is part of the input. We propose the Least Remaining Loops First (LRL) priority rule and show that it minimizes the total unweighted completion time. Then, we analyze the Weighted Least Remaining Loops First (WLRL) priority rule and show that it has a worst-case performance ratio of $(1+\sqrt{2})/2$ (about 1.2). Additionally, we present a fully polynomial time approximation scheme (FPTAS) and a pseudo-polynomial time algorithm if the number of machines in the flow shop is fixed.