In pultrusion process, fibers are impregnated with a polymer resin and are continuously pulled through a die, where the curing reaction occurs. Upon entering in the metal die, which has a cavity where the material is molded, the reaction system must be heated to a certain temperature, typically using electric resistors coupled on its surface. There are two essential parameters in the evaluation of this process: the energy consumption by the heaters and the productivity, influenced by the pulling speed. The objective of this research is to find the optimal heat flux and, additionally, the optimal speed in order to increase the production rate. The optimization strategy is based on combining the numerical capability of ANSYS CFX with optimization algorithms developed in Matlab, based on the Particle Swarm Optimization Method. We have applied this approach to optimize a L-shaped pultruded composite that has never been optimized before.