The technology for power system PCS has improved substantially across the studies in this review. While some early studies either did not include PCS (e. g., GE Energy, 2008) or did not include accurate transmission system models (e. g.,WindDS in U. S. DOE, 2008), the most recent studies used detailed PCS models that modeled unit commitment, generator ramp rate constraints, and transmission limits. Given that these costs can be important contributors to wind integration costs (Mount et al.,
2012) , including these details is important. Since most PCS is performed on hourly data, the fine time-scale data challenges described in Sections 17.5.1 and 17.5.2 are unlikely to have a significant effect on PCS results.
Going forward, there is a need to improve the ways in which uncertainty and wind forecasts are handled in PCS. The vast majority of the studies used deterministic PCS models, whereas uncertainty, particularly in wind and solar forecast data, become increasingly important to unit commitment decisions as renewable penetration increases. Also, a number of the studies assumed that the entire fleet of existing power plants would continue to be available for unit commitment, even under 20 to 40% wind scenarios. Clearly, at least some economically uncompetitive power plants would be retired in high renewables cases. This retirement process should be simulated in order to understand which plants or types of plants need to be incentivized to remain operational in order to manage the costs of transition to high-renewables-penetration scenarios.