Analytical methods in reliability analysis are useful for studying simple problems. For complex networks with cross-linked (non-series/parallel) component configurations, it is difficult to use mathematical reliability analysis. Powerful methods for reliability analysis of such systems have been developed using discrete event simulation. The main drawback of these methods is that they are computer time intensive. In this paper, the main idea behind these methods is further explored and modified in order to reduce the computational loads. The modified approach presented here leads to a great time saving which is very important for reliability analysis of large scale systems. This modified method is then modeled by Petri net, which is a powerful modeling tool.