problem calls for a trade-off between risks and rerimis. To
effectively deal with these corisiderable diflculties, we propose
a novel meta-heurisric approach based on a hybrid evol
n t i o n a~a lgorirhrn combined with constructive heuristics
for addressing just-in-rime pmduction and delivery with time
coristrairits on both the earliness and the lateness of siippl)!
Distribution of ready-made concrete is used as a practical
e.raniple. A case study based on industrial data illustrates
the potential of the proposed approach.
Keywords: Supply chain management, genetic algorithms,
meta-heuristics, concrete delivery.
1 Introduction
Recently, production industry is experiencing a strategic
evolution toward the decentralization of many production activities,
increasing the importance of supply chains. Supply
chains can he viewed as dynamic networks of partially independent
production centers that agree to collaborate for pursuing
both individual and collective aims. For instance, independent
companies that are able to provide complementary
services for the production of a given good may take a significant
advantage by synchronizing their activities to reduce
product lead times or costs. The control and optimization
of material, information and financial flows in supply chains
currently constitutes an important research field [6].
From the logistic viewpoint, the management of supply
chains involves a set of complex and interdependent combinatorial
problems such as the acquisition of raw materials,
scheduling of production facilities and routing of transport
vehicles. Even when considered as independent from the
other ones, each of the mentioned logistic problems suffers
from a nearly prohibitive combinatorial complexity. However,
there is also a strong need for approaches that are capable
of finding satisfactory solutions to these complex proh-
*0.’78n3-8~66.7/04/$t0.00 0 2004 IEEE.
Rotterdam, The Netherlands
u.kaymak@ieee.org
lems in short computation times. A class of modem metaheuristic
approaches that seems to be particularly suited for
dealing effectively and efficiently with the complexity in
supply chains is the Genetic Algorithms (GAS). GAS are
heuristic search techniques inspired from the principles of
survival-of-the-fittest in natural evolution and genetics. They
 

/>have been used extensively to solve combinatorial problems
that cannot be handled by exhaustive or exact methods due
to their prohibitive complexity. When properly configured,
GAS are efficient and robust optimization tools, because they
do not explicitly require additional information (such as convexity,
or availability of derivative information) about the ohjective
function to he optimized. However, GAS are generally
slow, they require large numbers of iterations, and suffer
from specific problems that may cause premature convergence
in suboptimal solutions. Therefore, the average time
that a well-configured GA would need to search for a satisfactory
solution of the entire supply-chain problem (with
its many decision variables) is too high for practical use in
a real industrial context, where decision-algorithm must provide
a solution in relatively short times. For this reason, we
propose a novel meta-heuristic approach based on a hybrid
evolutionary algorithm combined with constructive heuristics
for addressing planning, scheduling and routing for justin-
time production and delivery. In this approach, we use a
GA to perform part of the optimization, while the remaining
part of the scheduling problem is handled by consuuctive
heuristic algorithms. This approach leads to a hybrid
evolutionary algorithm in which the GA constitutes the core
of the search strategy, while multiple heuristic rules called
in specific circumstances contribute to reconstruct a feasible
solution that satisfies all the constraints and objectives
of the problem. In this respect, the proposed approach is
significantly different from other recent applications of GAS