Estimating the Value of Electricity Storage in Chile Through Planning Models with Stylized Operation: How Wrong Can It Be?

In light of the increased renewables penetration in power systems around the world, policy-makers, regulators, planners, and investors are significantly interested in determining the participation of energy storage in prospective scenarios of future generation capacity. In this context, this paper demonstrates the numerical errors associated with electricity planning models with stylized operation, which are of common use nowadays. We particularly focus on errors when quantifying the benefits of pumped hydro storage (PHS). The latest research identifies important distortions in the results of infrastructure expansion planning problems originated due to a stylized representation of power system operation. These distortions have been particularly emphasized in power systems with increased penetration of renewables generation that necessitate higher levels of flexibility to deal with variability and uncertainty. Apart from providing a comprehensive literature review in this subject, we provide additional and novel quantitative evidence focusing on the impacts of additional PHS capacity in power systems. Thus, we compare the outputs from two models: (i) a planning model with a stylized operation that ignores operational details in long-term investment analysis, approximating operational costs through a discretized version of the load curve (i.e., time slice representation), and (ii) a state-of-the-art, advanced planning model that recognizes operational details, including hourly resolution and technical limitations of generation plants (through the so-called unit commitment variables and constraints). Both models co-optimize generation and transmission capacity by minimizing total system investment and operational costs. Through several case studies on the Chilean power network by 2025, it is demonstrated that the benefits in terms of cost savings from PHS are significantly underestimated by the stylized model that ignores operational details. In effect, the stylized model undermines both peaking generation capacity and network capacity deferred by storage as well as the operational cost savings due to reserves and flexibility provisions from PHS. Moreover, it is shown that while CO2 emissions are reduced in the advanced model (as expected), these are increased in the stylized model, which corresponds to a remarkable misleading result. Finally, revenue projections of PHS by using primal and dual information are calculated from both optimization approaches, demonstrating that the stylized approach is biased and erroneously diminishes the PHS revenue in the case of a bulk, transmission-connected PHS in Chile. These conclusions are of particular interest for policy-makers, regulators, planners, and investors in Chile who seek to identify both PHS projects that are socially optimal (minimizing overall system costs) and privately profitable (whose revenues exceed costs).