Intermedium agents play an important role in indirect interplant heat integration. Each of them has a unique performance in heat recovery, however seldom conjunctively employed, which simplifies the problem but limits the extent of heat recovery. Thus, in order to fully exploit the potential of cascade heat recovery across plants, this work presents an optimization-based method to perform the automatic selection and arrangement of intermedium fluids in their integration with each single plant. A total site heat exchanger network super-structure model of possible configurations combining the allocation of medium streams (multiple agents and multiple circles) across plants and the heat exchanges between intermediate fluids and inner-plant process streams is proposed. For single-period concerns, the generalized model is formulated into a mixed-integer nonlinear programming problem to perform the network optimization at minimum total annualized cost target. Beyond this, the multiperiod problem is further studied to accommodate the time-related variations in the long operating horizon of an industrial park, and a robust network is achieved with updated modeling. Finally, a numerical case is illustrated in several scenarios; the significant drop in network cost shows the effectiveness of the method and the extended application of the method in a multiperiod design is also presented.