This paper describes the development of a new method to solve mixed-discrete optimization problems. The method is a two phase approach similar to Tunnel Methods developed for global optimization of continuous problems. It uses a SQP optimization solver in the first phase and an efficient rounding procedure to find discrete solutions in the second phase. All components utilized in this heuristic method are implemented with an emphasis on efficiency. The method was implemented in MATLAB and the solutions of three classical design problems are given. The results show the new method is very robust in finding high quality solutions which are consistently as good or better than past published results.