Airline network inventory control policy : an application of Markov Decision Processes and Simulation
In this thesis an airline network inventory control policy is developed. The inventory problem being faced differs from traditional hierarchical inventory problems. In this problem the inventory system is composed of a network of warehouses where demand is observed between warehouses and inventor...
| Main Author: | |
|---|---|
| Language: | English |
| Published: |
2009
|
| Online Access: | http://hdl.handle.net/2429/11220 |
| Summary: | In this thesis an airline network inventory control policy is developed. The inventory problem
being faced differs from traditional hierarchical inventory problems. In this problem the
inventory system is composed of a network of warehouses where demand is observed between
warehouses and inventory "consumed" from one warehouse is "provided" to another.
The control policy is developed using two modeling techniques: Dynamic Programming and
Simulation. Dynamic programming, or Markov Decision Processes, is used to help define a
control policy. The limitations imposed by the use of Dynamic Programming are overcome by
implementing the results in a Simulation model.
The policy resulting from this analysis demonstrates the potential to save $390,000 per year with
minimal operational impact to Canadian Airlines International. This represents a 60% reduction
in non-service related equipment movements. |
|---|