Modelo computacional para otimização de um sistema de tratamento de líquidos por aplicação de pulsos de alta tensão

Abstract

The search for healthy food products has become more frequent due to consumer awareness of health benefits. In the conventional liquid pasteurization treatment method, the elimination of pathogenic microorganisms is carried out by applying high temperatures for a certain time, but it may produce undesirable effects in the liquid, changing its nutritional and sensorial characteristics. The PEF method for treatment of liquids is an emerging and promising technology with potential application in the industry of juice, milk and other beverages. In comparison with the thermal process, the minimizing of the heat effects on the liquid under treatment is the main attraction of this technology. The microbial inactivation measures the treatment efficiency from a microbiological point of view. The energy density is another parameter of interest, since it is related to the operational cost of the process. The microbial inactivation and energy density of a treatment by electric pulses are dependent on the electric field distribution and treatment time. Considering that the microbial inactivation and energy density have conflicting behaviors in relation to these parameters, a problem of multiobjective optimization is configured. Within this context, this work aims to present a computational methodology to generate solutions in relation to microbial inactivation and energy density for PEF systems. Initially, is presented a preliminary computational model for static treatment chambers, not considering thermal analysis. This model is based on a finite element algorithm in MatLabRM for generating the electric field distribution, which is coupled with NSGA-II multiobjective optimization algorithm to generate Pareto curves for different electrode arrangements (coaxial, plate-parallel and cylindrical). The analysis is done for a generic liquid of conductivity of 2mS/cm containing a homogeneous distribution of bacteria E. coli. The results showed a better set of solutions for the coaxial treatment chamber in relation to the other arrangements. Considering the best results obtained for the coaxial treatment chamber, it was selected for use in dynamic systems, after dimensional adjustments. After preliminary studies, a dynamic model is presented, which couples the electric field distribution, temperature and fluid flow using the COMSOLRM software, integrated with NSGA-II multiobjective optimization algorithm. The simulations in the COMSOLRM were performed for experimental computational design considering as variables the applied voltage and the internal electrode radius of the coaxial geometry. The upper limit of the applied voltage range and the radius range of the internal electrode were specified to limit the electric field in the liquid at 50kV/cm, in order to avoid disruptions within the liquid under treatment. After regression of the results of average electric field obtained from a computational design of experiment, the electric field expression generated for the coaxial geometry was considered in the objective functions (microbial inactivation and energy density) of the multiobjective optimization problem. Based on the thermal analysis, it was possible to limit the number of pulses applied and, consequentelly, the treatment time, considering the upper limit of liquid temperature at 45 C. The frequencies of the applied pulses were defined for each type of juice analyzed. The transit time of the liquid within the treatment chamber was synchronized with the frequency of the pulses for a laminar flow. The analysis of uniformity of the electric field within the liquid, which is relevant to ensure a good uniformity of the treatment, was based on the coefficient of variation obtained from the electric field values in the mesh elements. The methodology was applied to four case studies, considering different types of microorganisms (E. coli and S. aureus), different types of juice (grape and orange), multiple treatment stages and different prediction models of microbial inactivation (Weibull and Hulsheger). Based on the Pareto curves generated by the computational model, it was possible to analyze thel solutions generated in terms of microbial inactivation and energy density, correlating with the variables under study. For example, for the bacterium E. coli in grape juice, the results showed a microbial inactivation of about −2.2 (escala log10) for energy density of approximately 115kJ/kg. Using multiple stages, it is possible to obtain a inactivation of −3.1 with energy density of about 90kJ/kg, selecting specific points on the Pareto curves for each stage. Based on the analysis of the generated Pareto curves, the computational model allows the selection of solutions as a function of the requirements of a process, in relation to microbial inactivation and energy efficiency, adding flexibility and dynamism to the design of treatment chambers.