Apresentação do Prof. Paulo Seleghim Jr. no projeto temático EMSO, no CTBE, em Campinas.
The substitution of fossil fuels by biofuels is pioneered by Brasil since the release of the Proalcool in the mid seventies. Today, around 48% of the Brazilian energy matrix is based on renewable sources: bio-ethanol, bio-diesel, hydro and bio-electricity. Despite this favorable scenario, further increases in biofuels production is strongly limited by technological bottlenecks, mainly associated to the up-scaling of new production methods developed at laboratory scales. The efficiency of biomass conversion in large scale bioreactors (hundreds or thousands of cubic meters) is a consensual issue among industrials. In a sugar cane juice fermentation tank for example, the distribution of residence time affects the efficiency of biochemical reactions: very little time result in an incomplete conversion (high sugar content in wine) and excessively long time results in loss of the ethanol by evaporation. The numerical optimization of this type of equipment produces an inverse problem which is, therefore, intrinsically ill conditioned. The solution of this problem by minimizing an error functional, although extremely interesting because there is no need for rough simplifications, is not possible through traditional numerical methods. Furthermore, the computational effort involved is extremely huge, if not impracticable, because of the complexity of the fluid and the geometry involved as well as the scale of the problem. The development of techniques capable to arrive to an optimization with sufficient accuracy in reasonable computational time will have a major impact on the design and operation of high performance industrial bioreactors and, consequently, increasing their operational efficiency. The main objective of this research project is to contribute to the development of integrated optimization methods, combining the characteristics of deterministic and random search strategies. Particularly, multiphase flow and bio-chemical reaction equations are numerically modeled and iteratively solved within a multi-objective optimization scheme combining a genetic algorithm and Newton’s method. With this we intend to develop the bases of a new methodology for optimal design and operation of industrial scale bioreactors.