Main content area

Towards a holistic approach for multi-objective optimization of food processes: a critical review

Madoumier, Martial, Trystram, Gilles, Sébastian, Patrick, Collignan, Antoine
Trends in food science & technology 2019
algorithms, models, multi-criteria decision making, process design, product quality
While Multi-objective Optimization (MOO) has provided many methods and tools for solving design problems, food processes have benefitted little from them. MOO encompasses the identification of performance indicators, process modelling, preference integration, trade-off assessment, and finding the best trade-offs. In this review, the use of these five elements in the design of food processes through MOO is analysed. A number of studies dealing with food processes MOO have been identified. Even though these studies improve the design process, they often approach MOO in a simplified and insufficiently rationalized way. Based on this review, several research issues are identified, related to the improvement of the different models and methods, and to the development of more holistic MOO methods for food processes.The challenge of food process engineering, taking into account the many constraints, is becoming increasingly difficult. multi-objective Optimisation (MOO) has provided many methods and tools to solve design problems, which often entail conflicting objectives (e.g. product quality vs. profit). Thus, only a trade-off can result from the design process, where preferences and weighting of the conflicting objectives has to be included.In this review, MOO methods used in food process design studies are analysed through the framework of the five elements which constitute a MOO method: i) performance indicators to quantify the design objectives; ii) a predictive process model to evaluate the different design solutions; iii) decision-maker and/or expert preferences, expressed as weights of the different objectives and satisfaction levels on indicator values; iv) a selection method to select one or several “best” trade-offs; v) an optimisation algorithm to find the best design solutions among the feasible ones.Key findings and conclusions: The authors highlight the fact that more holistic MOO methods can be used to improve the design process, as numerous methods have been developed to build MOO methods, particularly in the field of Multi-criteria Decision Making (MCDM). The literature reveals that MOO is often approached in a simplified and insufficiently rationalized way, which hinders the use of MCDM methods to improve the design process. Several research issues are also identified, related to the improvement of the different models and methods, which could help develop more holistic MOO methods for food processes.