In this study, the theoretical maximum power of a heat engine was investigated by sequential Carnot cycle model, for a low-grade heat source of about $100^{circ}C$. In contrast to conventional approaches, the pattern search algorithm was employed to optimize the two design variables to maximize power. Variations of the maximum power and the optimum values of design variables were investigated for a wide range of UA(overall heat transfer conductance) change. The results show that maximizing heat source utilization does not always maximize power.