The development in computational simulation techniques has brought significant progress in the realization of computational alloy design. The advantages are most significant when developing novel materials of which the research & development cycles are particularly time and energy (and hence cost) consuming, such as high-temperature alloys. In our former research, a computational alloy design approach coupling thermodynamics, kinetics, metal physics and genetic algorithm has been developed. By applying this new approach, novel heat resistant steels have been successfully designed with different microstructural features, which manage to nicely outperform existing commercial alloys. In this thesis, we follow the same approach while more focus has been given on adjusting the alloying level of different elements to solve specific issues, such as the high cost issue caused by a high Cobalt level, and the high microstructural instability caused by a high Chromium concentration. Extended application has been made to design novel heat resistant steels by introducing the concept of self-healing mechanisms. The creep damage (grain boundary cavities) of newly-developed steels during service are expected to be automatically filled by the special-designed Laves phase of which their formation kinetics was adjusted.