| dc.description.abstract | Public test cases representing large electric power systems at a high level of fidelity and quality are few to non-existent, despite the potential value such cases would have to the power systems research community. Legitimate concern for the security of large, high-voltage power grids has led to tight restrictions on accessing actual critical infrastructure data. To encourage and support innovation, synthetic electric grids are fictional, designed systems that mimic the complexity of actual electric grids but contain no confidential information.
Synthetic grid design is driven by the requirement to match wide variety of metrics derived from statistics of actual grids. The creation approach presented here is a four-stage process which mimics actual power system planning. First, substations are geo-located and internally configured from seed public data on generators and population. The substation placement uses a modified hierarchical clustering to match a realistic distribution of load and generation substations, and the same technique is also used to assign nominal voltage levels to the substations. With buses and transformers built, the next stage constructs a network of transmission lines at each nominal voltage level to connect the synthetic substations with a transmission grid. The transmission planning stage uses a heuristic inspired by simulated annealing to balance the objectives associated with both geographic constraints and contingency reliability, using a linearized dc power flow sensitivity. In order to scale these systems to tens of thousands of buses, robust reactive power planning is needed as a third stage, accounting for power flow convergence issues. The iterative algorithm presented here supplements a synthetic transmission network that has been validated
by a dc power flow with a realistic set of voltage control devices to meet a specified voltage profile, even with the constraints of difficult power flow convergence for large systems.
Validation of the created synthetic grids is crucial to establishing their legitimacy for engineering research. The statistical analysis presented in this dissertation is based on actual grid data obtained from the three major North American interconnects. Metrics are defined and examined for system proportions and structure, element parameters, and complex network graph theory properties.
Several example synthetic grids are shown as examples in this dissertation, up to 100,000 buses. These datasets are available online. The final part of this dissertation discusses these specific grid examples and extensions associated with synthetic grids, in applying them to geomagnetic disturbances, visualization, and engineering education. | en |
