A network model for the real-time communications of a smart grid prototype

Within a large-scale distributed (or centralized) smart grid, the communication network is designed to connect multiple power management systems and collect data from hundreds or thousands of power sensors over a large geographical area. Due to the ripple effect of inconsistent communication delays, the network performance becomes a major concern to support power system applications. For example, in a large NSF project Future Renewable Electric Energy Delivery and Management (FREEDM), we implement a smart grid prototype, called the FREEDM Hardware-in-the-loop (HIL) testbed. The Distributed Grid Intelligences (DGIs) in the prototype can group specific peers to exchange the power among the power demands and supplies. But the grouping sometimes is not successful due to the inconsistent delays. In this paper, we present a queueing model to describe the performance of the grouping network that supports the communications among the DGIs in our smart grid prototype. First, we develop a queueing model to describe the communication traffic among the DGIs in the network with different topologies. Second, based on the network model, we analyze the delay performance and illustrate that the model can be used to predict the grouping delays. Third, we have collected extensive experimental data which is used to demonstrate the accuracy of the model. Finally, we have implemented the HIL testbed with the capabilities of integrated real-time communication and power exchange. The network model is used to investigate the influence of the communication inconsistency on the total cycle time of power operations so that new grouping protocols can be designed in the future.