Grid Computing was born of power grids and known as a hardware and software infrastructure that provides reliable, consistent and economical access to high computing power, is increasingly important in research to be a form of virtualize computing resources by Internet in large‐scale distributed applications. Because different types of resources, it is important for the organization management and interoperability, which is why they have designed models and algorithms for effective selection of resources that allow the execution of the tasks required by users of the Grid. Because planning seeks control over the combinatorial explosion in assignments on available resources, this research developed an approach using task decomposition in literal syntax for defining logical representation of the problem, through the specification of states using the action description language, first addressed the overall problem and then developed problems with sub goals by applying the principle of reducing negative interactions with allowing the use of causal links by eliminating the overestimation of searches on multiple targets , was achieved optimize runtime evidenced by 20% in 10 different simulation scenarios.