Decision support with respect to facility location and fleet composition for FoodBank Cape Town

[摘要] ENGLISH ABSTRACT: FoodBank South Africa is an non-profit organisation formed to establish a national network ofcommunity foodbanks in urban and rural areas of South Africa, with all participants workingtowards the common goal of eliminating hunger and food insecurity. FoodBank Cape Townwas the first of these community foodbanks launched in South Africa on 2 March 2009. Theoperations of FoodBank Cape Town include sourcing food and redistributing it to agencies(social services organisations running feeding programmes). Currently the majority of the foodis sourced from the retail sector and then redistributed to approximately two hundred agencies.The logistics involved in both sourcing and distributing food are vital to the efficient functioningof FoodBank Cape Town. Since the costs associated with these logistics operations are very high,streamlining these operations has been identified as a priority area for efficiency improvement.The focus in this thesis is on the distribution logistics involved, specifically focussing on a facilitylocation problem according to which FoodBank Cape Town can establish local distribution depotsto which it delivers food and from which the agencies collect food assigned to them.A mixed-integer programming model is formulated for the above facility location problem andsmall test instances of the problem are solved using different exact and approximate solutionmethods in order to identify a suitable solution methodology for the full (large-scale) FoodBankCape Town facility location problem. The full facility location problem is solved approximatelyby means of a meta-heuristic solution method in the more highly constrained instances, whilean exact method is selected for solving the lesser constrained instances. The problem is firstsolved based on the distances between the warehouse and the depots as well as the distancesbetween the agencies and the depots, for the twenty four instances where 17 to 40 depots arelocated. The model is then developed further to incorporate the cost of distribution. Thiscost-based facility location model is solved with a view to minimise the cost of food distributionfrom the warehouse to the depots and the cost of food distribution incurred by each agencyto collect food from its assigned depot. A basic vehicle routing technique is applied to thecost-based facility location solution and the associated costs of the distribution are updated.This cost-based solution updating process is performed iteratively until the solution converges.Since the cost of food distribution depends on the vehicle fleet composition used, a vehicle fleetcomposition comparison of possible FoodBank Cape Town vehicles is performed to determinethe most desirable vehicle fleet composition to be used for the distribution of food to depots.The results of the FoodBank Cape Town facility location problem and vehicle fleet compositioncomparison are presented and recommendations are made to FoodBank Cape Town regardingthe preferred number of depots, the location of these depots and the preferred vehicle fleetcomposition.

[发布机构] Stellenbosch University