Licentiatavhandlingar_et

Permanent URI for this collection

https://hdl.handle.net/20.500.12703/6597

Browse

Browsing Licentiatavhandlingar_et by Subject "forecasting models"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Food waste in the food service sector - Quantities, risk factors and reduction strategies
    (Department of Energy and Technology, Swedish University of Agricultural Sciences, 2021) Malefors, Christopher
    An estimated one-third of all food produced is wasted, meaning that much of the negative environmental impact caused by food production is in vain. Global ambitions to reduce food waste include halving the levels by 2030, while the new EU food strategy views reducing food waste as a key issue in achieving a sustainable food system.  This thesis presents detailed information on the volumes of food waste, where it occurs, why it occurs and what can be done to reduce it. The information originated from 1189 kitchens operating in establishments such as canteens, care homes, hotels, hospitals, preschools, schools and restaurants throughout Sweden, Norway, Finland and Germany. The results indicated that approximately 20% of food served in the catering sector is wasted, although there is large variation, with canteens reporting 50±9.4 g/portion of food waste and restaurants 190±30 g/portion. To identify risk factors and reasons for food waste, a more detailed subset of data on Swedish preschools and schools was analysed. Some of the risk factors identified related to kitchen infrastructure and guest age, which could be difficult or expensive to tackle as a first option. The main risk factor was the amount of food prepared relative to the number of guests attending, an issue that kitchens can tackle by forecasting. This thesis demonstrated the potential of forecasting attendance as a tool in planning catering operations. The current business-as-usual scenario, where food is prepared for all pupils enrolled, results in a mean error of 20-40%, whereas the best forecasting case, using neural network models, resulted in a mean error of 2-3%. However, forecasts can underestimate demand, creating shortages, so some margin must be added in practical use. Providing kitchens with information about roughly how many guests will attend a meal, plus a sufficient margin, and encouraging them to serve food from a backup stock in cases of forecast underestimation would overcome the problems of shortages, reduce food waste and contribute to a sustainable food system.