The large-scale integration of intermittent renewable energy sources, such as wind and solar, presents significant challenges to the stability and reliability of modern power grids. In this context, energy storage battery systems have emerged as a critical enabling technology. Their rapid response and flexible power regulation capabilities make them indispensable for balancing supply and demand, mitigating fluctuations, and providing essential grid services. As the penetration of energy storage battery systems increases, understanding their impact on overall system reliability becomes paramount. However, traditional reliability assessments often treat these systems as perfectly reliable assets, overlooking their inherent degradation and failure mechanisms. This paper addresses this gap by proposing a comprehensive framework for power system operational reliability assessment that explicitly accounts for the multifaceted failure characteristics of energy storage battery systems.
The reliability of an energy storage battery is not static; it degrades over time and is susceptible to catastrophic failures. These failure modes can be broadly categorized into two types: performance degradation and thermal runaway. Performance degradation is a gradual process where the battery’s key characteristics, such as its usable capacity and internal resistance (often inferred from voltage characteristics under load), diminish over cycles and calendar life. Thermal runaway, in contrast, is a rapid, often violent failure triggered by internal short circuits, overcharging, or mechanical abuse, leading to fire or explosion. Both failure modes directly impair the energy storage battery’s ability to deliver or absorb power as scheduled, thereby affecting grid reliability. Ignoring these aspects can lead to overly optimistic reliability estimates, potentially jeopardizing system security.
The core of the proposed methodology lies in constructing a unified model for the available output power of an energy storage battery system that integrates both performance degradation and the risk of thermal runaway. For modeling the multi-state nature of performance degradation, the Universal Generating Function (UGF) technique is employed. This method allows for the probabilistic representation of the energy storage battery’s health state based on key parameters like capacity and voltage characteristics. The capacity and voltage state UGFs are defined as:
$$u_C(z) = \sum_{i=1}^{n_C} p_{C,i} z^{g_{C,i}}$$
$$u_V(z) = \sum_{j=1}^{n_V} p_{V,j} z^{g_{V,j}}$$
where \( p_{C,i} \) and \( p_{V,j} \) are the probabilities of the energy storage battery being in capacity state \( i \) and voltage state \( j \), with performance levels \( g_{C,i} \) and \( g_{V,j} \) respectively. The probability that the performance level falls below a minimum required threshold defines the degradation failure coefficient for each parameter. For instance, the capacity degradation coefficient \( R_C \) is:
$$R_C = P(g_C \leq g_{C,min}) = \sum_{g_{C,i} \leq g_{C,min}} p_{C,i}$$
A similar coefficient \( R_V \) is calculated for voltage characteristics. Concurrently, a data-driven thermal runaway warning model is used to assess the instantaneous risk of catastrophic failure. This model, often based on reconstructing normal operating parameters (voltage, current, temperature, state-of-charge), flags anomalies. The thermal runaway failure coefficient \( R_{TR} \) is a binary variable derived from the warning probability \( P_{thermal} \) and a predefined threshold \( \delta \):
$$
R_{TR} =
\begin{cases}
0, & \text{if } P_{thermal} > \delta \\
1, & \text{if } P_{thermal} \leq \delta
\end{cases}
$$
The overall reliability coefficient \( R_S’ \) for the energy storage battery system is the product of these individual coefficients, reflecting the combined effect of all considered failure modes:
$$ R_S’ = R_{TR} \cdot R_C \cdot R_V $$
This coefficient directly modulates the operational constraints of the energy storage battery in the system model. The conventional power and energy capacity constraints are modified as follows:
$$
\begin{align}
0 \leq P_{discharge}(t) &\leq R_S’ \cdot P_{discharge}^{max} \\
0 \leq P_{charge}(t) &\leq R_S’ \cdot P_{charge}^{max} \\
E_{min} \leq E(t) &\leq R_S’ \cdot E_{max}
\end{align}
$$
where \( E(t) = E(t-1) + (\eta_c P_{charge}(t) – P_{discharge}(t)/\eta_d)\Delta t \). This formulation accurately captures how the failure of the energy storage battery reduces its effective power delivery capability and energy storage ceiling.
To evaluate system-level reliability, a sequential Monte Carlo simulation framework is adopted. This involves randomly sampling component states (generators, lines, transformers), renewable power output, and load levels over time. For each sampled system state, an optimal power flow (OPF)-based load curtailment minimization problem is solved. The objective is to minimize the total cost of load shedding, wind curtailment, and solar curtailment:
$$
\min F = \sum_{i=1}^{N_{load}} \sum_{t=1}^{T} \lambda_{L,i} C_{e,i,t} \Delta t + \sum_{k=1}^{N_{wind}} \sum_{t=1}^{T} \lambda_{W,k} \Delta P_{W,k,t} \Delta t + \sum_{r=1}^{N_{solar}} \sum_{t=1}^{T} \lambda_{S,r} \Delta P_{S,r,t} \Delta t
$$
subject to constraints including the modified energy storage battery model (equations above), power balance, generator limits, line flow limits, and ramping constraints. From numerous simulation trials, standard reliability indices are computed, notably the Loss of Load Probability (LOLP) and the Expected Demand Not Supplied (EDNS):
$$
\begin{align}
\text{LOLP} &= \sum_{x \in \Phi} P(x) \\
\text{EDNS} &= \sum_{x \in \Phi} P(x) \cdot C_e(x)
\end{align}
$$
where \( \Phi \) is the set of system states resulting in load curtailment, \( P(x) \) is the probability of state \( x \), and \( C_e(x) \) is the total load curtailed in that state.
| Failure Mode | Nature | Impact on Energy Storage Battery | Modeling Approach |
|---|---|---|---|
| Performance Degradation | Gradual, Stochastic | Reduction in usable capacity and power capability. | Universal Generating Function (UGF) for multi-state reliability. |
| Thermal Runaway | Sudden, Catastrophic | Complete and immediate loss of function. | Data-driven anomaly detection and warning probability. |
A case study based on the modified IEEE RTS-79 system demonstrates the necessity and impact of the proposed method. The system is integrated with wind farms, solar photovoltaic (PV) stations, and multiple energy storage battery systems at key load buses. Three distinct scenarios are compared:
| Scenario | Description | EDNS (MW) | LOLP (%) |
|---|---|---|---|
| A | System with renewables, without energy storage battery. | 15.19 | 0.0866 |
| B | System with energy storage battery, ignoring its failures. | 13.78 | 0.0804 |
| C | System with energy storage battery, accounting for failures (proposed method). | 14.55 | 0.0839 |
The results clearly show that while integrating an energy storage battery system improves reliability (Scenario B vs. A), neglecting its failure mechanisms leads to an overestimation of this improvement. The proposed model (Scenario C) provides a more conservative and realistic assessment, indicating a higher EDNS and LOLP than the idealistic Scenario B. This underscores the critical importance of incorporating the reliability model of the energy storage battery itself into grid planning studies.
Further sensitivity analyses reveal how the degradation of the energy storage battery parameters influences system reliability. The available energy capacity (\( R_S’ \cdot E_{max} \)) and maximum power rating (\( R_S’ \cdot P^{max} \)) of the energy storage battery are key determinants. The analysis shows diminishing returns: once the capacity and power ratings of the energy storage battery exceed a certain threshold relative to system needs, further increases yield minimal reliability benefits. However, below this threshold, the degradation of the energy storage battery’s effective parameters has a pronounced negative impact on system EDNS. The operating temperature of the energy storage battery is another crucial factor, as higher temperatures accelerate degradation and increase the risk of thermal runaway, thereby reducing the system’s overall reliability as computed by the model.
| Effective Energy Storage Battery Capacity (MWh) | System EDNS (MW) | Observation |
|---|---|---|
| 100 | 15.21 | High sensitivity: Capacity degradation significantly worsens reliability. |
| 500 | 14.76 | |
| 700 | 14.54 | |
| 800 | 14.53 | Low sensitivity: System is less vulnerable to further capacity increases. |
| 1500 | 14.52 |
In conclusion, the secure and reliable integration of energy storage battery systems into the power grid requires a holistic assessment framework that acknowledges their inherent limitations and failure modes. The methodology presented in this work provides a robust tool for this purpose. By combining a multi-state performance degradation model with a thermal runaway risk assessment, it generates a comprehensive reliability coefficient for the energy storage battery. This coefficient is seamlessly integrated into a system-wide Monte Carlo reliability evaluation, producing indices that reflect the true, derated contribution of energy storage battery assets. For system planners and operators, this approach offers a more accurate foundation for determining the required capacity and specifications of energy storage battery systems, ensuring that reliability targets are met without unwarranted optimism. Future work may focus on integrating more detailed electrochemical-thermal models of battery aging and exploring the coordinated operation of fleets of heterogeneous energy storage battery systems with varying health states.