With the development of grid techniques and the growing complexity of grid applications, reasoning the temporal properties of grid application to ensure its reliability is becoming more and more critical. In this work, two decomposition approaches are proposed to improve the performance of the temporal reasoning of complex grid applications. The proposed approaches are implemented in our GridPiAnalyzer for equipment grid. Results show that our approach can reduce both CPU time and memory cost compared to using traditional formal verification algorithm alone due to the exponential reduction of system state space.