A genetic algorithm and a simulation tool for the design of hospital material flows
Résumé
The Hospital Centre of Tours (France) is reorganizing the logistic services between its hospital units, located at different places in the city. Each hospital unit has to be supplied with clean linen, tray meals, medicines, sterile equipments for operating theaters, medical patient records (paper files), and other various supplies. At the same time, dirty linen and waste have to be collected from each unit. At the moment, these flows are managed separately by specific services which implies a lot of vehicle routing without much coordination and pooling. In addition, one of the hospital units consists of several buildings through which the progression of vehicles is difficult and easily disrupted when a truck is parked too long to perform its delivery (the driver of the vehicle performs the delivery). A project of reorganization of this hospital unit is currently in progress and plans the creation of an independent team of warehousemen. Thanks to this team, the vehicles will load and unload only in one dock, facilitating the traffic. The team will help loading and unloading trucks and will mainly support the delivery and pick up at the buildings of the units as fast as possible, using eventually the Fenwick and underpass. We focus on the vehicle routing problem (VRP) between these units and these hospitals on a week period, and also on the design of a warehousemen team inside a particularly large hospital unit. The design of this team (composition, planning, etc.) is strongly related to the routes of vehicles between hospital units. In addition, the total number of drivers and warehousemen cannot increase since it is not possible to hire a new person. The whole problem can be seen as a two-level pickup and delivery problem with time windows, where two VRP interact. The flows of commodities are important. For example the hospital units use more than 15 tons of linen per day, and they also consume more than 4500 tray meals per day. We focus only on the number of trucks to be used for the delivery or pickup of each kind of commodity. Each truck has a delivery point which is a building or a hospital unit. A delivery point is characterized by a demand for each kind of truck with a quantity and a time window during which the delivery must be performed. Finally, the delivery time depends on the type of truck, the number of trucks, the mode of transport, and the delivery point.