This article aims to present how to apply Benders Decomposition to the problem of production
planning having uncertain demand within a finite horizon under production constrains that the
application is divided into 2 methods called “Single Step Bender” and “Bi Step Bender”. According to
the experimental results comparing with the result of Integer programming, it is found that both methods
can cause the quality results not being different from the result performed by Integer programming with
the statistic significant level 0.05. For this experiment, Bender 1 Step leads to the different result from
lower bound with the average value 0.0475% , but Bender 2 Step and Integer programming have the
average value 0.0495% and 0.0556% respectively. In part of processing time, if it is not a big problem,
Bender 2 Step will use time lower than Bender 1 Step. On the other hand, if it is a big problem, Bender 1 Step will use time lower than Bender 2 Step. Moreover, Integer programming cannot provide the answer in case of too big problem.
Keywords
Stochastic linear programming, Production planning under the uncertain demand, Large-scale linear programming problems, Benders Decomposition.