Reduce and re-use: studying retailers' food waste from an Operations Research perspective

In this thesis, the focus is on the reduction of food waste at the retailer level, either by reducing the food waste that is generated, or by re-using food leftovers. The problem is studied with an Operations Research approach, i.e. by the development of simulation and optimization models. The reduction of food waste is important, as the environmental impact of the food system is large, e.g. 24% of the greenhouse gasses in Europe are related to the food system (Westhoek et al., 2016). As global food loss is estimated at 50% of the total production, the environmental impact of the food waste is substantial. Furthermore, the economic impact of waste is estimated on e143 million for the European countries only (Stenmarck et al., 2016). This thesis considers avoidable food waste, and thus does not include food losses or unavoidable food waste. The focus of this thesis is on food waste at retail level. Although waste levels by the retailer are only 5%, there are several reasons why it is important to address the retailers' food waste. The retailers fulfil a key position in many food supply chains, as they sell many products from different food supply chains. Moreover, they are the last actor in the supply chain where logistic interventions can be implemented. Three main causes for avoidable food waste are identified, (i) the consumer in-store behaviour and the consumer demand; (ii) the store management e.g. the replenishment policies and (iii) the product shelf life. In the chapters of this thesis, several interventions are considered in order to reduce or re-use food waste by the retailer. In Chapter 2, both the replenishment strategy and the consumer in-store behaviour are addressed to obtain a food waste reduction. A discount on nearly expired products is provided, to stimulate consumers to buy these products, instead of the fresher ones, which could be sold at a later point in time. Furthermore, a shelf life based on the actual quality of the product (dynamic shelf life) is applied instead of a fixed, printed date. A simulation-based optimization model is created to optimize the replenishment quantity of the retailer combined with the two interventions to reduce food waste. Depending on the discount given, waste reductions up to 60% can be achieved when a discount is applied, although profit levels will be reduced with 11%. A dynamic shelf life can give a waste reduction around 40% on average, without a loss of profit. The replenishment decision for substitutable products is considered in Chapter 3, addressing the in-store behaviour of consumers. The replenishment of perishable products is optimized together, i.e. the replenishment decisions of one product depends on the replenishment decision of the other product. Moreover, the substitution behaviour of consumers between these products is included. For a two-product case, with one-way substitution, a simulation-based optimization model is presented to determine the optimal order-up-to levels. By combining the replenishment decisions for the two products, an average waste reduction of 35% is obtained. Results show high waste reductions when the willingness to substitute of consumers is high. Moreover, large improvements can be made when the two products differ in shelf life or profit margin. Chapter 4 also includes substitution behaviour in the inventory decisions, but for vertically differentiated products with two-way substitution. A heuristic is presented to optimize the replenishment decisions for multiple products under multiple substitution rounds. The results show the importance of including the inventory levels when deciding on the assortment to carry in the store. Furthermore, the incorporation of multiple substitution rounds has a significant effect on the profit. Moreover, due to demand uncertainty, the inclusion of product substitution and thus combining the inventory decisions for a product group increases the expected profit. The demand uncertainty can be shared among the products within the product group, which reduces the relative uncertainty, and profit levels therefore might rise. At the retailer, it will be impossible to reduce the food waste to zero. The re-use of food leftovers is therefore studied in Chapter 5. A mixed integer linear programming (MILP) model is developed for the menu planning of a soup kitchen. The soup kitchen can buy ingredients, but also receives food donations from retailers or other food companies. By minimizing costs, the MILP model provides a menu plan for the soup kitchen with restrictions on meal and ingredient use to provide a varied and healthy meal to the customers of the soup kitchen. Donations either are provided via a contract between the retailer and the soup kitchen, or arrive on a random basis. Results show the cost reduction for the soup kitchen. Furthermore, when good agreements are made between the retailer and the soup kitchen, 50{70% of the food waste obtained at the retailer can be re-used. In Chapter 6 integrated conclusions are presented combined with a discussion and an outlook to further research. All interventions show a great possibility to reduce food waste, and several linkages between the interventions can be found. By combining the interventions, food waste at the retailer can be addressed at several points, which will increase the total effect of the interventions mentioned before.