Multiobjective feed formulation for pig: methodological approach and application

Florence Garcia-Launay (1) , Cécile Crolard (1) , E. Teisseire (2) , Mohsen Davoudkhani (1) , Joël Aubin (2)

1 PEGASE - Physiologie, Environnement et Génétique pour l'Animal et les Systèmes d'Elevage [Rennes]
2 SAS - Sol Agro et hydrosystème Spatialisation

Animal production is responsible for several environmental impacts, to which feed production has usually the major contribution. However, the traditional least-cost feed formulation (LCF) method minimizes the cost without consideration of its environmental impacts. Multi-objective feed formulation has already been proposed to reduce the environmental impacts of pig feeds. It includes an objective function which is a weighted sum (WS) of the normalised values of feed cost, and four environmental impacts of the feed (climate change, land occupation, phosphorus demand and cumulated energy demand) calculated by Life Cycle Assessment. Normalised values are calculated dividing them by their reference value (REF) obtained with LCF (method Norm1). An additional factor α, ranging from 0 to 1 to explore the space of optimal solutions, is then used to weight the relative influence of feed cost and environmental impacts. The aim of this study was to explore potential improvements of the multi-objective method, with applications to pig feeds. We compared WS method with the ϵ-constraint method, which consists in minimizing a single objective while setting a maximum constraint for another one. In our case, the objective function was the weighted sum of the environmental impacts and the constraint was applied on feed cost. Then, we compared the behaviour of the model with Norm1 and with a new normalization method (Norm2), which subtracts the minimum criterion value to the criterion calculated and divides it by the difference between the REF and the minimum criterion values. WS and ϵ-constraint methods produced some common optimal solutions but only ϵ-constraint method allowed to obtain all the solutions of the Pareto front of the problem. Norm2 allowed defining solutions, which reduce consistently all environmental impacts whereas Norm1 allowed compensations between the impacts, which can potentially lead to increase of some impacts. Multi-objective feed formulation method can be substantially improved by implementing Norm2 to avoid possible compensations between impacts, and by applying the ϵ-constraint method to have access to the whole set of feasible optimal solutions.

Domaines: Sciences du Vivant [q-bio]

Florence Garcia-Launay, Cécile Crolard, E. Teisseire, Mohsen Davoudkhani, Joël Aubin. Multiobjective feed formulation for pig: methodological approach and application. 70. Annual Meeting of the European Federation of Animal Science (EAAP), Aug 2019, Gand, Belgium. Wageningen Academic Publishers, Annual Meeting of the European Association for Animal Production, 25, 2019, Annual Meeting of the European Association for Animal Production. ⟨hal-02306201⟩