Developments of renewable energy resources imposes many uncertainties and variabilities in power grids. One of the best approaches to mitigate these stochastic disturbances is thought the use of Battery Energy Storage System (BESS). Besides application of the BESS such as decreasing the disturbances in distribution system, the grid frequency can be controlled in contingencies using the appropriate storage in transmission network to compensate the power shortage. Thus, the optimal siting and sizing of the BESS is important to have the minimum costs and losses. This paper describes a heuristic method to find the optimal location(s) and capacity of a multi-purpose BESS including transmission and distribution parts. In the transmission storage part, a sensitive analysis is performed using Complex-Valued Neural Networks (CVNN) and Time Domain Power Flow (TDPF) in order to detect the optimal BESS location(s). Additionally, running TDPF and Economic Dispatch (ED) leads to the optimal BESS size. In the distribution storage part, the optimal BESS size is calculated to perform distribution grid services such as peak load shaving and load curve smoothing. The proposed method has been applied to a real model (Maui island in Hawai'i-United States) to calculate the optimal results for both transmission and distribution sides. (C) 2016 Elsevier Ltd. All rights reserved.