J. Gondzio and A. Grothey
Abstract
A real-life distribution planning problem is dealt with in this paper.
Modelling of this problem combines the use of graphs to describe
the topology of the distribution network, dynamics to capture
multi-periodicity of the planning, and uncertainty of the future
demands. The combination of these factors causes the (overall)
problem to be non-trivial and defy standard optimization software.
The problem is an example
of Operations Research application which needs a dedicated solution
approach. In this paper the problem has been modelled in such a way
that its complicated structure can be exploited by a specialised
optimization tool: its main difficulty has been converted into
an advantage. The use of structure-exploiting modelling and solution
methodology has enabled solving larger instances of the problem
and has shed new light on the interpretation of results.
The use of OOPS (Object-Oriented Parallel Solver) was crucial
in achieving these goals.
Key words: Distribution Planning Problems, Interior Point Methods, Exploiting Structure.