Abstract
The penetration of renewable energies requires additional storages to deal with intermittency. Accordingly, there is growing interest in evaluating the opportunity cost (usage value) associated with stored energy in large storages, a cost obtained by solving a multistage stochastic optimization problem. Today, to compute usage values under uncertainties, an adequacy resource problem is solved using stochastic dynamic programming assuming a hazard-decision information structure. This modelling assumes complete knowledge of the coming week uncertainties, which is not adapted to the system operation as the intermittency occurs at smaller timescale. We equip the twotimescale problem with a new information structure considering planning and recourse decisions: decision-hazard-decision. This structure is used to decompose the multistage decision-making process into a nonanticipative planning step in which the on/off decisions for the thermal units are made, and a recourse step in which the power modulation decisions are made once the uncertainties have been disclosed. In a numerical case, we illustrate how usage values are sensitive as how the disclosure of information is modelled.
Introduction
As renewable energy integration increases, large-scale energy storage systems, such as pumped hydro, become vital for grid stability. Managing these resources effectively requires calculating a "usage value"—the opportunity cost of saving energy for future use. Current industry standards, such as those used by the French TSO (RTE), often rely on a Hazard-Decision (HD) information structure. This model assumes that at the start of a week, operators have perfect foresight of all uncertainties (demand, renewable output, plant outages) for the next seven days.
However, this "wait-and-see" approach is unrealistic. In practice, "slow" thermal units (e.g., nuclear or coal) must be scheduled well in advance, while "fast" units and storage react to real-time fluctuations. Ignoring these timing constraints leads to an overestimation of system flexibility and biased storage price signals.
The Decision-Hazard-Decision (DHD) Framework
To address these shortcomings, the authors propose a two-timescale approach distinguishing between planning (weekly) and recourse (hourly) decisions. They introduce the Decision-Hazard-Decision (DHD) information structure.
In the ideal non-anticipative model ($DHD^h$), planning occurs at the start of the week, and recourse decisions are made hourly as information is revealed. To maintain computational tractability, the authors propose a relaxation ($DHD$) that groups hourly recourse into a single vector at the end of the week. This captures the inherent rigidity of slow units while remaining solvable via Stochastic Dynamic Programming (SDP).
Unlike the more common Stochastic Dual Dynamic Programming (SDDP), the authors utilize SDP to handle the binary on/off variables of thermal units without relaxation. This prevents the distortion of results that occurs when binary variables are treated as continuous.
Key Insights and Results
The study reveals that information availability acts as a significant constraint on the system. The standard HD model often falls into a "flexibility trap," assuming the system is more agile than it is in reality. Because the HD model "knows" the future, it undervalues storage as a hedging tool against unexpected events.
Numerical evaluations on a representative energy system (comprising base, semi-base, and peak thermal units alongside storage) show that:
- Usage Value Sensitivity: Usage values are lower in the DHD model when storage levels are high, reflecting a different valuation of the resource under realistic information constraints.
- Increased Utilization: The DHD model relies more heavily on storage for balancing because it recognizes the rigidity of thermal units, leading to more frequent cycling (pumping and turbining).
- Higher Expected Costs: The DHD model results in higher expected costs ($B^{HD} \leq B^{DHD}$), which accurately reflects the "cost of reality" where future uncertainties are not known in advance.
Conclusion
The DHD framework provides a more accurate valuation of energy storage by respecting the operational timescales of different power sources. While the current implementation is limited by the "curse of dimensionality" inherent in SDP, it sets the stage for more realistic grid management. Future work will explore spatial decomposition techniques to scale this framework to larger, interconnected systems with multiple storage facilities.
Summary updated