Annals of Operations Research 63(1996)397-414 397
A combined branch-and-bound and genetic algorithm
based approach for a flowshop scheduling problem
Amit Nagar
School of Business, University of Windsor, Windsor, Ontario, Canada
Sunderesh S. Heragu and Jorge Haddock
Decision Sciences and Engineering Systems Department,
Rensselaer Polytechnic Institute, Troy, NY 12180, USA
In this paper, we study the applicationof a meta-heuristicto a two-machineflowshop
scheduling problem. The meta-heuristicuses a branch-and-boundprocedure to generate
some information, which in turn is used to guide a genetic algorithm's search for
optimal and near-optimal solutions. The criteria considered are makespan and average
job flowtime. The problem has applications in flowshop environmentswhere management
is interested in reducing turn-around and job idle times simultaneously. We develop the
combined branch-and-bound and genetic algorithm based procedure and two modified
versions of it. Their performanceis compared with that of three algorithms: pure branch-
and-bound, pure genetic algorithm,and a heuristic.The results indicatethat the combined
approach and its modified versions are better than either of the pure strategies as well
as the heuristic algorithm.
Keywords: Genetic algorithm, scheduling, branch-and-bound.
1. Introduction and brief literature review
We study a two-machine bicriteria flowshop scheduling problem in which the
objective is to minimize the weighted sum of schedule makespan and average job
flowtime. The problem has applications in industries where each job must undergo
two basic processes in the same sequence. For example, in the steel industry, each
job undergoes wire-drawing first and annealing next. At the aggregate planning or
macro level in general manufacturing industries, each job must undergo fabrication
first and assembly next. The two-machine bicriteria flowshop scheduling problem
has been studied in [1-3]. All the previous approaches to this problem have relied
on conventional optimization techniques such as goal programming and branch-and-
bound. Selen and Hott [ 1] developed a goal programming formulation for the bicriteria
© J.C. Baltzer AG, Science Publishers