J Intell Manuf (2010) 21:843–851
DOI 10.1007/s10845-009-0260-3
Parallel machine scheduling problem to minimize
the earliness/tardiness costs with learning effect
and deteriorating jobs
M. Duran Toksarı · Ertan Güner
Received: 5 May 2008 / Accepted: 2 March 2009 / Published online: 19 March 2009
© Springer Science+Business Media, LLC 2009
Abstract The focus of this work is to analyze parallel
machine earliness/tardiness (ET) scheduling problem with
simultaneous effects of learning and linear deterioration,
sequence-dependent setups, and a common due-date for all
jobs. By the effects of learning and linear deterioration, we
mean that the processing time of a job is defined by an increas-
ing function of its starting time and a decreasing function of
the position in the sequence. We develop a mixed integer
programming formulation for the problem and show that the
optimal sequence is V-shaped: all jobs scheduled before the
shortest jobs and all jobs scheduled after the shortest job are
in a non-increasing and non-decreasing order of processing
times, respectively. The developed model allows sequence-
dependent setups and sequence-dependent early/tardy pen-
alties. The illustrative example with 11 jobs for 2 machines
and 3 machines shows that the model can easily provide the
optimal solution, which is V-shaped, for problem.
Keywords Earliness/tardiness · Learning effect ·
Deterioration jobs · V-shaped property · Parallel machine ·
Sequence-dependent setup time · Common due date
Introduction
In this paper, we have considered parallel machine sched-
uling problem with learning and deterioration effects, and a
M. D. Toksarı (B )
Industrial Engineering Department, Erciyes University,
Kayseri, Turkey
e-mail: dtoksari@erciyes.edu.tr
E. Güner
Industrial Engineering Department, Gazi University, Ankara,
Turkey
common due-date for all jobs where the objective is
minimization of the weighted sum of earliness and tardiness
penalties of jobs. The weighted earliness tardiness problem
arises in a simplified JIT production environment (Mondal
and Sen 2001). In a JIT scheduling environment, a job that
completes early must be held in finished goods inventory
until its due date, while a job that completes after its due date
may cause a customer to shut down operations. Therefore,
an ideal schedule is one in which all jobs finish exactly on
their assigned due dates (Baker 1994).
In the ET literature, the types of penalty cost functions
used in the objective function include linear, nonlinear, uni-
form penalties, and penalty differences among jobs. Penalty
differences between earliness and tardiness are important
assumption. They can be classified on four basic forms as job
dependent (Baker 1990; Bank and Werner 2001; Mosheiov
and Shadmon 2001; Zhu and Heady 2000; Hendel and Sourd
2006), unequal (Zeng et al. 1993; Ventura et al. 2005), equal
and penalty cost with job dependent proportional weight
(Sun and Wang 2003; Arkin and Roundy 1991; Szwarc 1993;
Bauman and Jozefowska 2006). In determining penalty func-
tions, different penalty functions are more appropriate in that
earliness and tardiness may be undesirable at the same pro-
portion.
Almost all ET articles assume that setup time is either
immaterial or sequence-independent. However, sequence-
dependent setups are very important to some industries. For
instance, in the chemical industry, machine setup times vary
considerably as a function of the sequence of processing jobs
(Zhu and Heady 2000). A review of the literature indicates
that only a few ET studies (Zhu and Heady 2000; Colemon
1992; Sourd 2005) address the sequence-dependent setups.
The article published by Zhu and Heady (2000) was focused
on ET problem with multi machines (parallel machines) and
sequence-dependent setups. The main difference between the
123