In recent years, the increasing frequency of extreme events, such as severe weather and cyber-attacks, has posed significant threats to the safe and stable operation of distribution networks (DNs). These events often lead to widespread power outages, highlighting the urgent need to enhance the resilience of electrical grids. At the same time, the rapid adoption of electric vehicles (EVs) and energy storage systems presents new opportunities for improving the emergency support capabilities of DNs. In China, the electric vehicle market has experienced explosive growth, with EVs becoming a common sight on roads and in urban infrastructure. This trend is driven by government policies and technological advancements, making China a global leader in EV adoption. The development of EVs, particularly in the context of China EV initiatives, offers a unique chance to leverage their resources for grid resilience. For instance, EVs can serve as mobile energy storage units or provide emergency power through innovative schemes like battery swapping. In this paper, we explore how electric vehicle replaceable batteries (EV-RBs) can be utilized as emergency power sources in post-disaster scenarios, and we propose a bi-level optimization model for configuring multiple types of emergency resources in DNs, including line reinforcement, mobile energy storage systems (MESSs), and EV-RBs. Our approach aims to minimize economic investment while maximizing resilience, leveraging the complementary advantages of these resources. Through case studies based on an improved IEEE 33-node system, we demonstrate the effectiveness of our strategy in reducing load loss and enhancing grid reliability. This work contributes to the growing body of research on smart grid technologies and underscores the potential of electric vehicles in transforming energy systems, especially in the context of China’s ambitious goals for a sustainable and resilient power infrastructure.
The concept of using electric vehicles as flexible resources in power systems has gained traction worldwide, with China EV policies accelerating this integration. Electric vehicles are not merely means of transportation; they can act as distributed energy storage devices, providing services like peak shaving, frequency regulation, and emergency power supply. In particular, the battery swapping technology for electric vehicles has matured significantly, allowing for quick replacement of depleted batteries with fully charged ones. This technology, championed by companies like NIO and CATL in China, enables EVs to function as mobile power sources or as part of stationary storage systems. We propose a novel scheme where EV-RBs are pre-positioned in key locations, such as large parking lots or commercial centers, to serve as emergency power sources during grid outages. When a disaster occurs, these EV-RBs can be combined with idle EVs and charging infrastructure to form power supply loops, supporting critical loads through vehicle-to-grid (V2G) or vehicle-to-load (V2L) operations. This approach leverages the ubiquity of electric vehicles in urban areas, particularly in China, where EV adoption is supported by extensive charging networks and standardization efforts.
To model the use of EV-RBs as emergency sources, we consider various cost components and operational constraints. The total cost for deploying EV-RBs includes installation costs, replacement costs, maintenance costs, and operational losses. The maintenance cost, for example, incorporates expenses for leasing EVs and hiring personnel for battery swapping, which we simplify as a fixed annual value. The operational loss cost is tied to the charging and discharging power of the EV-RBs, expressed as:
$$C_{op,t}^{RB} = \sum_{b \in B} w_b (P_{b,t}^c + P_{b,t}^d)$$
where \(C_{op,t}^{RB}\) is the operational loss cost at time \(t\), \(B\) is the set of EV-RBs, \(w_b\) is the loss cost coefficient for EV-RB \(b\), and \(P_{b,t}^c\) and \(P_{b,t}^d\) are the charging and discharging powers, respectively. The operational constraints for EV-RBs are defined to ensure safe and efficient operation. For instance, the charging and discharging powers are limited by the number of available EVs and charging piles at each location:
$$0 \leq P_{b,t}^c \leq \xi_{b,t}^c \min(\Omega_b^{CP} P_{b}^{CP,\max}, \Omega_b^{EV} P_{b}^{EV,\max})$$
$$0 \leq P_{b,t}^d \leq \xi_{b,t}^d \min(\Omega_b^{CP} P_{b}^{CP,\max}, \Omega_b^{EV} P_{b}^{EV,\max})$$
$$\xi_{b,t}^c + \xi_{b,t}^d \leq 1, \quad 0 \leq Q_{b,t} \leq \min(\Omega_b^{CP} Q_{b}^{CP,\max}, \Omega_b^{EV} Q_{b}^{EV,\max})$$
$$E_{b,t+1} = E_{b,t} + \eta_b P_{b,t}^c \Delta t – \frac{P_{b,t}^d \Delta t}{\eta_b}$$
$$S_{b}^{\min} E_{b}^{BP,\text{rate}} \leq E_{b,t} \leq S_{b}^{\max} E_{b}^{BP,\text{rate}}$$
Here, \(\Omega_b^{CP}\), \(\Omega_b^{EV}\), and \(\Omega_b^{BP}\) represent the number of charging piles, available EVs, and pre-positioned battery packs at location \(b\), respectively. \(P_{b}^{CP,\max}\) and \(Q_{b}^{CP,\max}\) are the maximum active and reactive power limits of the charging piles, while \(P_{b}^{EV,\max}\) is the maximum power output of the EVs. \(E_{b,t}\) is the energy stored in EV-RB \(b\) at time \(t\), \(\eta_b\) is the efficiency, and \(S_{b}^{\min}\) and \(S_{b}^{\max}\) are the state-of-charge limits. Additionally, the power factor constraint ensures that reactive power support does not exceed permissible levels:
$$Q_{b,t} \leq (P_{b,t}^c + P_{b,t}^d) \tan \phi$$
where \(\phi\) is the power factor angle. These constraints make EV-RBs a flexible and reliable resource for emergency power, aligning with the capabilities of modern electric vehicles in China and beyond.
Building on this EV-RB model, we develop a bi-level optimization framework for configuring multiple emergency resources in DNs. The upper-level model focuses on planning decisions, aiming to minimize the total economic investment over the planning period. The decision variables include the number and locations of line reinforcements, the number and capacities of MESSs, and the number and capacities of EV-RBs pre-positioned in key areas. The objective function combines installation costs, replacement costs, residual value recovery, annual maintenance costs, and the annual operational cost from the lower-level model. Mathematically, the objective is expressed as:
$$\min \left( C_{\text{ins}} + C_{\text{rep}} – C_{\text{rec}} + C_{\text{om}} + \frac{(1+\zeta)^{T}-1}{\zeta(1+\zeta)^{T}} C_{\text{op}} \right)$$
where \(\zeta\) is the discount rate, \(T\) is the planning horizon, and \(C_{\text{ins}}\), \(C_{\text{rep}}\), \(C_{\text{rec}}\), \(C_{\text{om}}\), and \(C_{\text{op}}\) represent installation, replacement, residual recovery, maintenance, and operational costs, respectively. The installation cost, for example, is calculated as:
$$C_{\text{ins}} = C_{\text{ins}}^{LR} \Omega^{LR} + \sum_{a \in A} C_{\text{ins},a}^{MESS} \Omega_a^{MESS} + \sum_{k \in K} C_{\text{ins},k}^{RB} \Omega_k^{RB}$$
where \(\Omega^{LR}\), \(\Omega_a^{MESS}\), and \(\Omega_k^{RB}\) are the numbers of line reinforcements, MESSs of type \(a\), and EV-RBs of type \(k\), respectively. Constraints in the upper level ensure that the number of resources remains within practical limits, such as:
$$\Omega_{\min}^{LR} \leq \Omega^{LR} \leq \Omega_{\max}^{LR}$$
$$\Omega_{\min}^{MESS} \leq \sum_{a \in A} \Omega_a^{MESS} \leq \Omega_{\max}^{MESS}$$
$$\Omega_{\min}^{RB} \leq \sum_{k \in K} \Omega_k^{RB} \leq \Omega_{\max}^{RB}$$
These constraints prevent over-investment while ensuring sufficient resources for resilience enhancement. The lower-level model, on the other hand, optimizes the operation of the DN under typical disaster scenarios, minimizing operational costs by scheduling resources like gas turbines, distributed generators (DGs), stationary energy storage systems (SESSs), MESSs, and EV-RBs. The objective function includes penalties for load shedding, DG curtailment, costs for purchasing power from the main grid, fuel costs for gas turbines, operational losses for EV-RBs, and costs for energy storage operation and network losses:
$$\min \sum_{z \in Z} \kappa_z \left[ \sum_{t \in T} \left( \sum_{n \in N} \delta_n PL_{n,t}^{\text{shed}} + \chi \sum_{j \in J} (P_{j,t}^{G,f} – P_{j,t}^{G}) + \lambda_t^{\text{grid}} P_t^{\text{grid}} + \sum_{i \in I} \alpha_i P_{i,t}^{G} + C_{op,t}^{RB} + \sum_{m \in M} w_m (P_{m,t}^c + P_{m,t}^d) + \sum_{s \in S} w_s (P_{s,t}^c + P_{s,t}^d) + \beta \sum_{l \in L} Pf_{l,t} \right) \right]$$
where \(\kappa_z\) is the probability of scenario \(z\), \(PL_{n,t}^{\text{shed}}\) is the shed load at node \(n\) and time \(t\), \(P_{j,t}^{G,f}\) and \(P_{j,t}^{G}\) are the forecasted and actual DG outputs, and \(Pf_{l,t}\) is the power flow on line \(l\). Operational constraints include linearized AC power flow equations, storage dynamics, and network topology constraints. For example, the power flow constraints are given by:
$$Pf_{l,t} = G_l (V_{I(l),t} – V_{E(l),t}) – B_l (\theta_{I(l),t} – \theta_{E(l),t})$$
$$Qf_{l,t} = -B_l (V_{I(l),t} – V_{E(l),t}) – G_l (\theta_{I(l),t} – \theta_{E(l),t})$$
$$-u_{l,t} Pf_{l}^{\max} \leq Pf_{l,t} \leq u_{l,t} Pf_{l}^{\max}$$
$$-u_{l,t} Qf_{l}^{\max} \leq Qf_{l,t} \leq u_{l,t} Qf_{l}^{\max}$$
where \(u_{l,t}\) is the health status of line \(l\) (1 for operational, 0 for damaged), and \(V_{n,t}\) and \(\theta_{n,t}\) are voltage magnitude and angle at node \(n\). The storage constraints for MESSs and SESSs ensure that charging/discharging powers and state-of-charge remain within limits, and for MESSs, we include constraints on mobility, such as the number of moves allowed:
$$\sum_{t \in T} \sum_{n \in N} v_{m,t} \leq \Omega_{\text{trans}}^{\max}$$
where \(v_{m,t}\) is a binary variable indicating if MESS \(m\) moves at time \(t\). This bi-level model is solved using an improved particle swarm optimization (PSO) algorithm for the upper level and a mixed-integer linear programming (MILP) solver for the lower level, ensuring efficient and robust solutions.
To validate our approach, we conduct a case study on an improved IEEE 33-node DN, which includes gas turbines, DGs, SESSs, and potential locations for MESSs and EV-RBs. The system is subjected to typical disaster scenarios generated from historical data using k-means clustering, considering uncertainties in load demand and DG output. We compare four configuration methods: Method 1 (our proposed approach with line reinforcement, MESSs, and EV-RBs), Method 2 (only MESSs and EV-RBs), Method 3 (only line reinforcement and EV-RBs), and Method 4 (only line reinforcement and MESSs). The configuration results and system costs are summarized in the following tables.
| Resource | Category | Method 1 | Method 2 | Method 3 | Method 4 |
|---|---|---|---|---|---|
| Line Reinforcement | — | 1-2, 27-28, 30-31 | — | 1-2, 27-28, 30-31 | 1-2, 27-28, 30-31 |
| MESSs #1 | 0 | 1 | — | 1 | |
| MESSs #2 | 2 | 2 | — | 2 | |
| MESSs #3 | 1 | 1 | — | 1 | |
| EV-RBs | #1 | 0,0,0,0,6 | 6,6,6,6,6 | 0,0,0,0,6 | — |
| #2 | 0,0,0,0,4 | 0,3,0,1,4 | 0,0,0,0,4 | — |
| Cost Category | Method 1 | Method 2 | Method 3 | Method 4 |
|---|---|---|---|---|
| Installation Cost | 5.2000×10^6 | 8.2000×10^6 | 2.6000×10^6 | 4.4400×10^6 |
| Replacement Cost | 2.2035×10^6 | 3.7595×10^6 | 4.4134×10^5 | 2.1959×10^6 |
| Annual Maintenance Cost | 2.6000×10^4 | 8.2000×10^4 | 2.1500×10^4 | 7.5000×10^3 |
| Annual Operational Cost | 3.0601×10^6 | 1.4491×10^7 | 6.5874×10^6 | 3.3063×10^6 |
| Total Comprehensive Cost | 3.7437×10^6 | 1.5636×10^7 | 6.8790×10^6 | 3.9032×10^6 |
As shown in Table 1, Method 1 results in the reinforcement of lines 1-2, 27-28, and 30-31, which are critical for maintaining connectivity and reducing load loss. The allocation of MESSs and EV-RBs is optimized to complement these reinforcements, with EV-RBs concentrated at node 31 to address local power shortages. In contrast, Method 2, which omits line reinforcement, requires more MESSs and EV-RBs to compensate, leading to higher costs. Table 2 reveals that Method 1 achieves the lowest total comprehensive cost, demonstrating the economic efficiency of integrating all three resource types. The annual operational cost is significantly lower in Method 1 due to reduced load shedding and better resource coordination. For example, in a typical disaster scenario where the DN is divided into isolated islands, EV-RBs provide immediate power support to critical loads, while MESSs facilitate energy transfer between regions. This synergy enhances resilience without excessive investment, underscoring the value of our multi-resource approach.
We further analyze the impact of the number of available EVs on the configuration results, as EV-RBs rely on idle EVs and charging infrastructure. Three scenarios are considered: Scenario 1 (baseline with 5,6,5,7,4 available EVs per node), Scenario 2 (reduced availability with 1,3,2,3,1 EVs), and Scenario 3 (increased availability with 8,9,7,10,10 EVs). The results, summarized in Table 3, show that changes in EV availability have a minor effect on the optimal configuration of line reinforcements and MESSs, but influence EV-RB capacity and system costs. In Scenario 2, the reduced EV availability limits the power output of EV-RBs, leading to a slight decrease in EV-RB capacity but an increase in operational cost due to higher load shedding. Conversely, Scenario 3 sees a marginal improvement in costs, but the configuration remains similar to Scenario 1, indicating that the model is robust to variations in EV availability. This resilience is crucial for practical implementations, especially in regions like China where EV adoption is growing but infrastructure may vary.
| Category | Scenario 1 | Scenario 2 | Scenario 3 |
|---|---|---|---|
| EV-RBs Capacity (kWh) | 850 | 750 | 850 |
| Installation Cost (CNY) | 5.2000×10^6 | 5.0300×10^6 | 5.2000×10^6 |
| Replacement Cost (CNY) | 2.2035×10^6 | 2.1499×10^6 | 2.2035×10^6 |
| Annual Maintenance Cost (CNY) | 2.6000×10^4 | 2.4000×10^4 | 2.6000×10^4 |
| Annual Operational Cost (CNY) | 3.0601×10^6 | 3.1579×10^6 | 3.0599×10^6 |
| Total Comprehensive Cost (CNY) | 3.7437×10^6 | 3.8196×10^6 | 3.7435×10^6 |
In conclusion, our proposed strategy effectively leverages electric vehicle replaceable batteries as emergency power sources, combined with line reinforcement and mobile energy storage systems, to enhance the resilience of distribution networks. The bi-level optimization model ensures cost-effective resource allocation while minimizing load loss during extreme events. The case study demonstrates that integrating multiple resource types outperforms partial approaches, reducing total costs by up to 76% compared to methods without line reinforcement. The sensitivity analysis confirms that the strategy is adaptable to variations in EV availability, which is essential for real-world applications in evolving markets like China. As the adoption of electric vehicles continues to rise, particularly in China with supportive policies and technological innovations, the potential for EVs to contribute to grid resilience will only grow. Future work could explore dynamic scheduling strategies for disaster prevention and black-start restoration, further leveraging the flexibility of electric vehicles and energy storage systems. This research highlights the transformative role of electric vehicles in building sustainable and resilient power infrastructures, aligning with global trends and China’s leadership in the EV industry.
Overall, the integration of electric vehicles into power system resilience planning represents a paradigm shift, where transportation and energy sectors converge to address common challenges. In China, the rapid deployment of EVs and charging infrastructure provides a fertile ground for implementing such strategies. We believe that our approach can be scaled to larger networks and adapted to other regions, contributing to the global effort to create smarter and more resilient grids. The continued innovation in battery technology and V2G capabilities will further enhance the value of electric vehicles in emergency scenarios, making them indispensable assets for future energy systems. As we move forward, collaboration between industry, government, and academia will be key to realizing the full potential of electric vehicles in enhancing grid resilience and supporting sustainable development.