This paper shows a heterogeneous and parallel computing (PHC) methodology applied to the Security Constraint Optimal Power Flow problem (SCOPF). The methodology used an original decomposition of base and contingency cases for the SCOPF and solved the problem through constraint handling, droop control and PV/PQ switching. PHC architecture was based on GPU and CPU to accelerate power flow calculations and parallelize contingency evaluations. The methodology was tested on power grids with sizes from 500 to 20,000 buses with promising results. The methodology allows power system optimization while guaranteeing power system security in N-1 scenarios in time frames appropriate for power system operation.