Abstract
Battery storage is essential for the future smart grid. The inevitable battery degradation renders the battery lifetime volatile and highly dependent on battery dispatch, and thus incurs opportunity cost. This paper rigorously derives the degradation cost function of battery storage for power system dispatch. We first derive the intertemporal optimality condition that guarantees the long-term dispatch optimality when short-term dispatch is made. Then we prove that the solution to a cost-form dispatch model implies this intertemporal optimality condition and thus is also optimal from the long-term perspective. The derived unit degradation cost is time-variant to reflect the time value of money and takes the form of a constant value (the marginal cost of usage) divided by a discount factor. In case studies, we demonstrate the existence of an optimal value of the marginal cost of usage that corresponds to the optimal long-term dispatch outcome. We also show that the optimal marginal cost of usage depends on system and battery parameters, such as the marginal cost of the system and the battery cycle life.
Introduction
In the transition to a smart grid, battery storage is indispensable for balancing the inherent volatility of renewable energy sources. However, electrochemical storage is a finite resource; every charge and discharge cycle induces degradation, affecting the battery's lifetime and economic value. Traditional power system dispatch models often fail to account for this accurately, either ignoring degradation or using a static "Levelized Cost of Degradation" (LCOD). This paper addresses the challenge by deriving a rigorous degradation cost function that treats battery usage as a time-variant opportunity cost.
Methodology: Bridging Short-Term and Long-Term Dispatch
The authors propose a multi-stage intertemporal model to align daily operational decisions with long-term asset health through a Marginal Degradation Cost (MCD) framework.
- Analytical Derivation: The study mathematically proves that a specific time-series of MCDs can be integrated into standard short-term dispatch objectives to guarantee optimal usage over the battery's entire lifespan.
- The MCD Formula: The optimal MCD consists of two primary components: a constant value adjusted by a discount factor (to account for the time value of money) and a term representing "functionality fade"—the impact of current usage on future capacity and efficiency.
- Backward Algorithm: Because current costs depend on future states, the authors developed a "Backward Algorithm." This method starts at the end of the battery's projected life and iterates backward to determine the optimal marginal costs for the present.
Key Insights
The research reveals that the "true" cost of using a battery is highly dynamic and context-dependent.
Dynamic vs. Static Costs
Unlike the static LCOD, the optimal MCD is time-variant. In the case study, the MCD rose from approximately $7/MWh in Year 1 to $13/MWh in Year 15. This upward trend suggests that batteries should be utilized more intensively early in their life cycle to maximize economic returns before significant degradation occurs.
System Dependency
The optimal cost of battery usage is not just a property of the battery itself but is linked to the grid's overall state. For instance, as the marginal operating cost of thermal generators in the system increases, the optimal MCD of the battery also rises, reflecting its increased value as a substitute for expensive thermal generation.
The "LCOD Gap"
Traditional LCOD methods tend to be overly conservative, often setting the usage cost too high. This causes the battery to remain idle during periods that are moderately profitable but fall below the LCOD threshold. The proposed MCD approach captures these opportunities, significantly increasing the total extracted value.
Results and Validation
The model was validated using a simulated microgrid based on 2015 ERCOT load profiles and Texas wind data. The results were striking:
- Economic Performance: The proposed method achieved $9.4 million in total system cost savings over the battery's life.
- Comparative Advantage: The standard LCOD method saved only $4.1 million, meaning the dynamic MCD approach was 2.3 times more effective at extracting value from the storage asset.
- Operational Efficiency: By pre-calculating the MCD, complex long-term optimization is simplified into a series of short-term problems solvable in real-time by grid operators.
Conclusion
This framework provides a mathematically grounded way to incorporate battery degradation into grid operations. While the "Backward Algorithm" introduces some computational overhead and assumes a degree of foresight regarding future load and weather, the economic gains are substantial. Future work may focus on adapting this framework for stochastic environments and exploring more complex, non-linear electrochemical degradation models.
Summary updated