In this paper three heuristic techniques based in swarm intelligence are studied, namely its ability to solve non-convex, nondifferentiable and highly restricted optimization problems. The performances of Particle Swarm Optimization, Bee Colony Optimization and Cockroach Swarm Optimization solving a set of benchmark functions will be compared. These techniques will also be applied to an electrical engineering problem, to be precise, the economic dispatch with non-convex cost functions. The results obtained up to now, have been demonstrating that these techniques are able to reach good results in the benchmark functions as well as in the problem of economic dispatch.