A Metric-Based Validation Process to Assess the Realism of Synthetic Power Grids

Abstract

Public power system test cases that are of high quality benefit the power systems research community with expanded resources for testing, demonstrating, and cross-validating new innovations. Building synthetic grid models for this purpose is a relatively new problem, for which a challenge is to show that created cases are sufficiently realistic. This paper puts forth a validation process based on a set of metrics observed from actual power system cases. These metrics follow the structure, proportions, and parameters of key power system elements, which can be used in assessing and validating the quality of synthetic power grids. Though wide diversity exists in the characteristics of power systems, the paper focuses on an initial set of common quantitative metrics to capture the distribution of typical values from real power systems. The process is applied to two new public test cases, which are shown to meet the criteria specified in the metrics of this paper.

Power Flow Convergence and Reactive Power Planning in the Creation of Large Synthetic Grids

“…Extensions of these methods are given in [17] for economic studies and [18] for dynamics. Additional developments in validation and extension are found in [19]- [21], including two more synthetic grids with 200 and 500 buses. The present paper builds upon the approach of [16] and is the first study that has scaled network synthesis of full power grids to ten thousand buses, with a focus on reactive power planning and power flow convergence.…”

“…Included in this case are 300 three-winding transformers, divided into three main types: GSUs, load step-down, and modified network transformer with a low-voltage tertiary. Parameters are calculated analogously to two-winding transformers, as in the metrics of [19], then converted to an equivalent set of three branches with a star bus, as they are typically modeled in commercial power flow solvers.…”

Building Synthetic Power Transmission Networks of Many Voltage Levels, Spanning Multiple Areas

“…After the topology was automatically generated, additional parameters and modifications were manually made to meet ac power flow goals and contingency behavior. Various statistics and validation metrics, as specified in [21], are shown in Table III and Figs 5-6. These validation metrics are based on analysis of the actual grid, with metrics taken from the Eastern Interconnect, WECC, and various subset cases of these.…”

“…In [16], a large test case is developed, however, it is only dc-solvable and approximates the actual grid network, rather than being fully synthetic. Since the publication of [1]- [2], additions to the base synthesis algorithm include generator cost curves for economic and optimal power flow (OPF) analysis [17], initial transient stability models for dynamic simulations [18], more detailed transmission line parameters [19]- [20], and a more robust set of validation criteria, including two additional cases with size 200 and 500 buses [21]. In [22]- [24], further developments in alternative methods for network topology generation are given, not including solvable circuits.…”

Interdependence of Transmission Branch Parameters on the Voltage Levels

“…The collaboration of researchers from five universities has resulted to publishing three fully synthetic power networks called ACTIVSg200, 500, and 2000 cases [21], [24]. Later, they published a set of topological and electrical validation metrics in [28] to assess the realism of the developed synthetic power grids. The authors will continue to augment those cases by adding additional complexities and verification and tuning of the parameters.…”