Resumo:
This PhD thesis presents an optimization model for the use of industrial waste as alternative
fuels taking into account targets such as quality, specific heat consumption, and pollutant
emissions constraints, among others. The model takes into account the raw material and fuel
cost, and also the electric power consumption cost requested for grinding which is another
important factor. In the modeling, two optimization techniques were used: (i) the
deterministic approach, through the Quadratic Sequential Programming (SQP) method in a
combination with Monte Carlo’s method and (ii) the heuristic approach, through the Genetic
Algorithm (GA) and Differential Evolution (DE) techniques applying penalty functions in
order to validate the obtained results. Among the employed algorithms, the GAs didn't show a
satisfactory solution, presenting convergence problems due to the need of satisfying all the
imposed restrictions. On the other hand, for the SQP and the DE algorithms, the results were
satisfactory with a small difference between the obtained optimum values. The DE method
demanded a larger computational time when compared with the SQP due to the great number
of objective function evaluations, an inherent characteristic of the DE. The results obtained
from the modelling were quite satisfactory, being possible to evaluate the levels of
substitution of the primary fuel for the alternative fuel derived from the industrial wastes,
taking into account the acceptable limits of pollutant emissions, the cement quality standards
as well as others parameters specific which are for the cement industry.