MINLP superstructure optimization for protein extraction in the surimi manufacturing process

MARÍA AGUSTINA REINHEIMER; NICOLÁS J. SCENNA

Foz de Iguazú

Congreso; EngOpt 2016 ? 5 th International Conference on Engineering Optimization; 2016

UFRJ

The surimi process starts from holding fish, sorting by size and cleaning. After that, process stages for meat separating are achieved, which are heading and gutting by mechanical fish meat separators, a preliminary washing to remove the blood and adherent particles and then, deboning and mincing.The cyclic washing and rinsing processes of the minced fish, which is also called leaching process, are the central stage. The objective of this stage is to remove soluble compounds resulting in concentrated myofibrillar proteins, which mainly contribute to gel formation.The leaching stage in the manufacturing process of surimi gel requires a large amount of wash water, where more than 65% of the total amount of fresh water required by the entire process is used in the leaching process resulting in high operating cost. Traditionally, the leaching process is achieved in three continuous cycles of washing, where fresh water is added at each one of the three leaching tanks, and the wastewater is subsequently removed from the meat by a rotary sieve before the next washing stage, this is called the conventional arrangement. Other possible configuration is a countercurrent one, in which the fresh water is only supplied to the last tank; then, the wash water is recycled to the previous one. In a previous work, it was proved that the countercurrent configuration entails minimum water consumption in this stage, 56% less than the conventional arrangement for three stages. In this work, a mathematical optimization model for the optimal design of the leaching process is presented, in order to define the optimal number of stages. Precisely, a detailed Mixed Integer Nonlinear Programming (MINLP) model, including operational and geometric constraints, is developed and implemented in GAMS, based on our previous optimization model (NLP model) for water consumption minimization. Discrete decisions associated with the number of stages (leaching tanks and auxiliary equipment units, such as sanitary pumps and rotary sieves) are modelled by using integer variables. Continuous variables are used for process conditions (temperatures, flow-rates, tank volumes, velocities, rate of extraction among others). A maximum of four stages are defined in the model.The optimal results adopt the maximum number of stages (four) and higher volume tanks. The fresh water consumption decreases when the number of stages increases, due to the fact that more recycle streams are available in the countercurrent arrangement, reducing both the fresh water requirement and the liquid waste stream. Despite the fact that the model includes cost aspects, the optimal solution adopts the highest number of stages, because only the operating costs are implicitly contemplated in the objective function. From a computational cost point of view, the model resulted to be enough flexible to perform optimizations and sensitivity analyses.