Project Overview
As information infrastructures, such as SCM (Supply Chain Management),
have been spreading, the need for optimization technologies that can solve
wide range of supply chain has been increasing.
There are already many algorithms for specific area of supply chain,
used for problems such as route planning and warehouse location.
The purpose of this project is solving a wide range of optimization problems by
using combinations of several existing algorithms.
As a result, we can build optimization solutions more easily and more
quickly.
The Warehouse Location and Transportation Problem
We defined the warehouse location and transportation
problem (WLTP), which is a problem that combines
the warehouse location problem (WLP) with the modal-shift
transportation problem (MSTP).
We developed a bilevel iterative algorithm for the WLTP that
leverages two existing algorithms for the WLP and
the MSTP as the upper and the lower level problems, respectively.
The distribution network.
Trade-offs in the WLTP
In the WLPT, we can consider several cost factors and their various
trade-offs.
The cost factors that can be considered in the WLTP are:
- The transportation cost from the warehouses to the stores,
- The fixed cost of the warehouses,
- The inventory cost at the warehouses, and
- The transportation cost from the plants to the warehouses.
If we reduce the number of warehouses, for example, the
transportation cost from warehouses to the stores will decrease, but the fixed
cost of the warehouses will increase.
The ordering interval by the warehouses to the plants influences a trade-off
between the inventory cost in the warehouses and the transportation cost
from the plants to the warehouses.
While it is important to consider various trade-offs,
it is difficult to find out optimal policies for wide area optimization.
The trade-off by #warehouses and their location
The trade-off between the inventory cost and the transportation cost.
An algorithm for the WLTP
We designed an algorithm that leverages two existing algorithms. The
combined algorithm iteratively solves the WLP and the MSTP by linking them
using shared parameters.
By generalizing this approach, we can solve wide range
of optimization problems using existing algorithms.
The algorithm framework of the WLTP
Reference
Takayuki Yoshizumi, Hiroyuki Okano,
"The Warehouse Location and Transportation Problem and Its Algorithm Based
on a Decomposed Formulation", Scheduling Symposium 2004 (in Japanese)
|
|
