Research on the Weak-Bus Voltage Support Effect of Energy Storage Type Intelligent Soft Open Point

Abstract

With the integration of a high proportion of distributed generators (DGs), the imbalance between source and load power intensifies, causing the distribution grid to become ‘weak’ during certain periods, which can easily lead to voltage over-limit for the distribution grid. Due to the coupling of active and reactive power in weak power grids for voltage regulation, the effectiveness of single reactive voltage regulation for voltage over-limit is not satisfactory. This paper uses an Energy Storage type Intelligent Soft Open Point (E-SOP) with grid-forming controlled energy storage to simultaneously adjust the active and reactive power between different grid clusters to suppress voltage over-limit for a weak grid. Firstly, based on the system architecture and bus load characteristics, the distribution grid is divided into clusters, and the cluster Thevenin equivalent model is established. Then, different voltage support schemes are compared and analyzed through the cluster active power-voltage (P-V) curve. Through simulation verification of the IEEE 33-bus system containing DGs, it is concluded that the scheme of installing E-SOP in the distribution grid containing DGs enhances the active power transmission capability of the grid and provides better voltage support compared to traditional reactive power compensation and system capacity expansion schemes.

1. Introduction

With the integration of a high proportion of new energy, the imbalance between power grid sources and loads has intensified, and the strength of the power grid has become ‘weak’ during certain periods, leading to voltage stability issues at weak buses in the power grid, which affects the power supply quality and the safe and stable operation of the distribution grid. As a new type of voltage and frequency support equipment for weak buses in the power grid, the grid type energy storage system has the advantages of fast response speed, high control accuracy, and wide adjustment range [1,2,3]. As an advanced power electronic switch, intelligent soft open point (SOP) can achieve power interconnection between different regional power grids, effectively improving the consumption of new energy, the operational efficiency, and power supply capacity of the power grid [4,5,6]. This paper combines grid-forming (GFM) controlled energy storage (ES) with SOP to support weak-bus voltage, in order to improve the voltage stability of the power grid.

The large-scale integration of distributed power sources with intermittent, random, and fluctuating characteristics into the distribution grid has an impact on the safe and stable operation of the system, including changes in power flow distribution, difficulty in local consumption of distributed new energy, power reverse transmission, increased grid losses, and bus voltage over-limit [7]. ES systems have flexible power regulation abilities and power supply and storage capacity abilities through charging and discharging [8], which can effectively alleviate the temporal mismatch between distributed new energy output and load demand [9], providing a solution for stabilizing the grid with large-scale integrated distribution of new energy. In addition, the division of a distributed grid into clusters can reduce the cost of DGs and improve the flexible control and regulation capabilities of DGs. At present, the most traditional and simple cluster division generally only considers geographical factors [10], that is, dividing similar DGs and loads in the active distribution network (ADN) into clusters according to geographical location. This cluster division method is often unreliable. Afterwards, some research considered the electrical coupling and distance between buses in ADN as the basis for cluster division, which was defined by the electrical parameters and sensitivity matrix of buses [11]. Reference [12] proposed a graph-based genetic algorithm for the optimal cluster partitioning of distributed photovoltaic power plants in order to control and quickly respond to grid regulation. Some research viewed the cluster division as a mathematical optimization problem that can be solved using heuristic optimization algorithms such as particle swarm optimization, genetic algorithm, and simulated annealing algorithm. Reference [13] fully considered the consumption capacity of clusters and proposed a heuristic cluster division method based on genetic algorithms that took into account state changes in interconnection switch, effectively improving the on-site consumption capacity of DGs. SOP can achieve optimized control between clusters by considering the global benefits of the distribution grid [14]. Reference [15] proposes a fast method to use machine learning for long-term evaluation of solar energy resources and photovoltaic power generation prediction in large areas, in order to determine potential locations for constructing photovoltaic power plants. Reference [16] established an ADN expansion planning model considering SOP configuration, achieving the collaborative planning of SOP, DGs, and ES. Reference [17] studied the collaborative planning model of photovoltaic inverters and SOP to enhance the consumption capacity of DGs. The E-SOP combines ES with SOP, which not only has the function of mutual power penetration between clusters, but also achieves power supply and load demand balance between clusters. The installed ES device can further improve the power quality of weak buses [18].

At present, three methods are commonly used to suppress the problem of voltage over-limit in the power grid. One is to use reactive power compensation devices to support the voltage and regulate the grid voltage within the normal range [19]. Magnetic Control Reactor (MCR) and Static Synchronous Compensator (STATCOM) can achieve continuous reactive power and grid voltage regulation [20]. These devices cannot provide active power support, thus limiting their effectiveness in regulating the voltage in a weak grid. The second method to suppress grid voltage fluctuations is to enhance the power supply capacity of the system, making it ‘strong’ to counteract large disturbances [21], such as increasing the short-circuit capacity of the system, increasing the transformer capacity, and reducing the internal impedance of the system. The third method is the emerging grid-forming controlled ES, which can be used to support weak buses in the system [22]. ES under grid-forming control is equivalent to a voltage source, achieving source support for weak buses, it has a direct and better voltage support effect compared to the other two methods, however, the configuration cost is high. E-SOP can also realize the voltage support by injecting active power from another cluster and its own energy storage; it not only improves the new energy consumption, but also reduces the configuration cost of the energy storage. To evaluate the voltage support effect of various schemes, a Thevenin equivalent model of the system can be identified first, and then the power transmission capacity or power supply capacity of the system can be compared through active power-voltage (P-V) curves, short-circuit ratios, etc. The Thevenin equivalent method of the system can be divided into offline identification based on circuit models and online identification based on measured data; online identification methods can obtain real-time dynamic equivalent models of the system. Reference [23] suggested that fluctuations on the load side and grid side are uncorrelated, and transformed the problem of identifying Thevenin parameters into an optimization problem for online identification under continuous disturbances in the grid. However, in weak grid scenarios, it remains to be verified whether the fluctuations on both sides are independent and unrelated. Reference [24] introduced an online identification method for power grid impedance that combines data fluctuation feature screening with linear relationship description of equivalent impedance, which can effectively solve the problem of inaccurate Thevenin equivalent impedance caused by fluctuations.

This paper takes E-SOP as the research object and focuses on the problem of weak-bus voltage support. Firstly, the IEEE 33-bus distribution grid is divided into clusters based on electrical distance and power balance indicators to realize the preliminary improvement in voltage stability and new energy consumption; the E-SOP is configured after dynamic power flow calculation, which can reflect voltage changes caused by variations of load and new energy generation units. The SOP is firstly configured at the bus where the deviation rate of active power sensitivity to voltage has the maximum value due to the active power support function of the SOP. The new indicators proposed in this paper are similar to the bus voltage sensitivity standard deviation indicator, however, they can reflect whether the active or reactive power injection has larger impact on the bus voltage; the energy storage unit can be further configured if the bus voltage still exceeds the limit or has large variation. The energy storage unit is controlled with the GFM strategy to achieve a better voltage support effect under the voltage variation. The real-time control of the E-SOP can be realized, which depends on the accuracy of the load prediction and dynamic power flow calculation. Finally, through simulation analysis, the supporting effects of system capacity expansion, reactive power compensation, and GFM controlled E-SOP on weak-bus voltage are compared and studied, verifying the effectiveness of E-SOP method in improving weak-bus voltage stability.

2. Voltage Over-Limit Problem and Grid Cluster Division Indicator

The integration of DGs may lead to voltage fluctuations and over-limit in the grid. This paper takes the IEEE 33-bus system as the research object. Four photovoltaic power generators (PV1, PV2, PV3 and PV4) and one wind power generator (WT) are added into the grid, as shown in Figure 1, where PV1 is 250 kW and connected to bus 32, PV2 is 500 kVA and connected to bus 7, PV3 is 500 kVA and connected to bus 9, PV4 is 500 kVA and connected to bus 30, and WT is 1.8 MW and connected to bus 5 [25]. The load benchmark value used in this paper is from the general model of IEEE 33 bus system, which are all constant power type loads. The K-means clustering method was used to cluster the annual fluctuation curves of photovoltaic and wind turbine output and load, resulting in five scenarios (Scenario 1–5) and the scenario 1 is selected for the simulation, corresponding to the summer season. All data of PV, wind, and load used are referenced from [26]. The daily normalized power forecast curves for all PVs, WT, and loads at 2–33 buses are shown in Figure 2.

This paper divides the IEEE 33-bus system with DGs into clusters based on indicators of the electrical distance and the balance of active and reactive power [27], as shown in Figure 1. Among them, electrical distance is a structural indicator that measures the coupling tightness of voltage changes between nodes, that is, the correlation of voltage changes between clusters is relatively small; power balance is a functional indicator. After decoupling the active and reactive power in the distribution network, cluster division was carried out based on the supply and demand balance of reactive power within the cluster, as well as the full consumption of new energy power generation by the load. Power and energy management within clusters after cluster division based on two indicators will not only reduce the cross transmission of reactive power between clusters, but also improve the consumption of new energy power, both of them enhance the self-regulation ability and stability of the bus voltages within the cluster. The effective voltage per unit value of each bus can be obtained as shown in Figure 3. Suppose that the allowable range of voltage fluctuation is ±5%, the rated effective value of the IEEE 33-bus system line voltage is 12.66 kV, thus its allowable fluctuation range is [12.03 kV, 13.29 kV].

As shown in Figure 3, all buses in cluster 2 did not exceed the upper and lower limits within 24 h of a day. There are no bus voltages exceeding the lower limit in cluster 1, but bus 4 slightly exceeds the upper limit at 01:00. Voltages of buses 5–9 and 26–32 in cluster 3 have all exceeded the upper limit value, and there are no bus voltages exceeding the lower limit; however, the voltage fluctuation is significant. Voltages of buses 13–18 in cluster 4 exceed the lower limit between 9:00 am and 12:00 pm and 15:00 pm and 20:00 pm.

The Voltage Sensitivity Standard Deviation (VS-SD) value of at each bus is an indicator that reflects the voltage fluctuation and its calculation equation for 24 h is shown in Equation (1).

$$\left\{\begin{array}{c}\stackrel{\_\_\_\_\_\_}{U_{\mathrm{SEN}}}={\displaystyle \frac{1}{24}}{\displaystyle \sum _{\mathrm{i}=1}^{24}{U}_{\mathrm{SEN}\mathrm{i}}}={\displaystyle \frac{1}{24}}{\displaystyle \sum _{\mathrm{i}=1}^{24}(\frac{\left|U_i-\stackrel{\_\_}{U}\right|}{\stackrel{\_\_}{U}}\times 100\%)}\\ {\sigma }_{U_{\mathrm{SEN}}}={\displaystyle \frac{1}{24}}\sqrt{{\displaystyle \sum _{\mathrm{i}=1}^{24}{\left(U_{\mathrm{SEN}\mathrm{i}}-\stackrel{\_\_\_\_\_\_}{U_{\mathrm{SEN}}}\right)}^2}}\end{array}\right.$$

(1)

The standard deviation of voltage sensitivity at each bus for 24 h calculated according to Equation (1) is shown in Figure 4.

From Figure 4, it can be seen that the standard deviation of voltage sensitivities at buses of 18 and 33 are the largest in their respective clusters, indicating that the voltage fluctuations at the two buses are large.

Short-circuit ratio (SCR) is the ratio of the short-circuit capacity of the system to the capacity of electrical equipment (including power electronic equipment), which is a measure of system strength and directly related to the system static voltage stability [28]. For different buses of the grid, SCR is different due to different short-circuit capacity, which is related to different three-phase short-circuit currents at the bus location. The equation of the SCR is shown as follows.

$$SCR={\displaystyle \frac{S_{\mathrm{ac}}}{P_{\mathrm{out}}}}={\displaystyle \frac{V_{\mathrm{c}}^2}{\left|Z\right|P_{\mathrm{out}}}}$$

(2)

In the Equation (2), S_ac is the short-circuit capacity of the system, P_out is the active power provided to the load, V_c is the voltage at the load connection point, and Z is the impedance of the system’s Thevenin equivalent circuit observed from the load. When 2 < SCR < 3, the system is called a weak network; when SCR ≤ 2, the system is called an extremely weak network.

The large-scale integration of new energy will lead to a decrease in the voltage support strength of the power system and a reduction in the short-circuit ratio (SCR). New energy generation units have great randomness and fluctuation and installing them on the load side will cause more voltage fluctuations at each bus. Especially during periods of abundant new energy, it is possible for the bus voltage to exceed the upper limit. Cluster 3 has installed a large amount of new energy, and buses in it exceed the upper limit, which is shown in Figure 3c.

Applying Equation (2) to calculate system SCR according to the system Thevenin equivalent circuit looking at the two weakest buses of 18 and 33, the system SCR at bus 18 is 2.3268 and the SCR at bus 33 is 2.8452, both of them indicate weak networks.

However, it is unknown whether to inject active or reactive power to support the bus voltage. Take the voltage of bus 18 at 18:00 pm, which is the most severe moment for the voltage exceeding the lower limit, as an example to analyze the effect of active power and reactive power supporting. From the voltage rise results of bus 18 in Figure 5, it can be seen that under the same power injection value, the active power P has greater impact on the bus voltage than the reactive power Q.

Using the indicator shown in Figure 5 to see the voltage support effect of active or reactive power to each bus is still too complicated. Therefore, a new indicator of ‘sensitivity deviation rate for active and reactive power is proposed in this paper. The sensitivity of active and reactive power to voltage is shown in Equations (3) and (4), respectively.

$${\displaystyle \frac{\partial P_{\mathrm{j}}}{\partial U_{\mathrm{i}}}}{U}_{\mathrm{i}}=\left\{\begin{array}{c}{U}_{\mathrm{i}}{\displaystyle \sum _{\mathrm{n}=1}^{\min (\mathrm{i},\mathrm{j})}\frac{U_{\mathrm{n}}}{R_{\mathrm{n}}}\cos {\theta }_{\mathrm{ij}}\begin{array}{c}\mathrm{i},\mathrm{j}\mathrm{on}\mathrm{same}\mathrm{branch}\end{array}}\\ U_{\mathrm{i}}{\displaystyle \sum _{\mathrm{n}=1}^{\mathrm{k}}\frac{U_{\mathrm{n}}}{R_{\mathrm{n}}}\cos {\theta }_{\mathrm{ij}}\begin{array}{c}\mathrm{i},\mathrm{j}\mathrm{on}\mathrm{different}\mathrm{branches},\mathrm{intersection}\mathrm{point}\mathrm{is}\mathrm{k}\end{array}}\end{array}\right.$$

(3)

$${\displaystyle \frac{\partial Q_{\mathrm{j}}}{\partial U_{\mathrm{i}}}}{U}_{\mathrm{i}}=\left\{\begin{array}{c}{U}_{\mathrm{i}}{\displaystyle \sum _{\mathrm{n}=1}^{\min (\mathrm{i},\mathrm{j})}\frac{U_{\mathrm{n}}}{X_{\mathrm{n}}}\cos {\theta }_{\mathrm{ij}}\begin{array}{c}\mathrm{i},\mathrm{j}\mathrm{on}\mathrm{same}\mathrm{branch}\end{array}}\\ U_{\mathrm{i}}{\displaystyle \sum _{\mathrm{n}=1}^{\mathrm{k}}\frac{U_{\mathrm{n}}}{X_{\mathrm{n}}}\cos {\theta }_{\mathrm{ij}}\begin{array}{c}\mathrm{i},\mathrm{j}\mathrm{on}\mathrm{different}\mathrm{branches},\mathrm{intersection}\mathrm{point}\mathrm{is}\mathrm{k}\end{array}}\end{array}\right.$$

(4)

In Equations (3) and (4), ${\displaystyle \frac{\partial P_{\mathrm{j}}}{\partial U_{\mathrm{i}}}}{U}_{\mathrm{i}}$ represents the sensitivity of the voltage U_i of bus i to the active power P_j of bus j, ${\displaystyle \frac{\partial Q_{\mathrm{j}}}{\partial U_{\mathrm{i}}}}{U}_{\mathrm{i}}$ represents the sensitivity of the voltage U_i of bus i to the reactive power Q_j of bus j, n represents the number of buses in the network, U_n represents the voltage amplitude of bus n, Rn represents the resistance value between bus n-1 and bus n, X_n represents the reactance value between bus n − 1 and bus n, and cos θ_ij represents the cosine value of the phase difference between bus i and bus j. However, considering the actual operation of the distribution network, the voltage amplitude of each bus is roughly maintained around the standard unit of 1.0 p.u., and at the same time, θ_ij is very small, which can be considered as cosθ_ij = 1.

After calculating the active and reactive power sensitivity of each bus for 24 h a day, the sensitivity of i and j in the same branch and different branches are summed up to obtain the final active and reactive power sensitivity values of each bus for 24 h a day. The average sensitivity value is further calculated to determine the deviation rate of active and reactive power sensitivity based on the maximum and minimum sensitivity values, as shown in Figure 6.

From Figure 6, it can be seen that for each bus, the deviation rate of active power sensitivity is higher than that of reactive power sensitivity. This indicates that for a weak grid containing a high proportion of new energy generation units or heavy loads, active power P has a greater impact on bus voltage than reactive power Q. Therefore, E-SOP is used to realize the active power support for a weak grid in this paper.

An E-SOP device needs to connect cluster 3 and cluster 4 to achieve mutual power support, the location of the E-SOP needs to be selected through a voltage sensitivity or power sensitivity deviation rate indicator. Comparing Figure 4 and Figure 6, we can see both of them achieve same conclusion, which is that buses 18 and 33 are needing to be supported.

Due to the serious issue of voltage exceeding the lower limit for certain buses in cluster 4 and voltage exceeding the upper limit for certain buses in cluster 3, an E-SOP device is used to interconnect clusters 3 and 4, thereby achieving power support from cluster 3 to the cluster 4. Therefore, installing an E-SOP device between buses 18 and 33 is most suitable to suppress voltage fluctuations between the two buses. The installation capacity of the E-SOP device is 148 kVA, and the installation location is shown in Figure 7.

3. E-SOP Topology and Control Strategy

An SOP is a fully controlled power electronic device that flexibly adjusts the active power flow between connecting lines and can also provide the required reactive power support. This paper uses a direct mounted MMC type E-SOP without the voltage step-down need of the transformer. Figure 8 shows the single-phase 2-bridge rectifier module after adding one bridge arm in each phase. Two bridge arms increase the number of voltage levels per phase from 2N + 1 to 4N + 1 through carrier phase shifting technology, thereby reducing the harmonic content to half of that before increasing the bridge arm. In addition, the battery BAT is connected in parallel with the submodule DC capacitor through a DC–DC converter, forming an ES type SOP.

For weak grids, SOP can be used to first perform active mutual power penetration between clusters, improving the power supply capacity of weak clusters, and thus provide voltage support; if further voltage support is needed, it can be achieved through GFM ES, which is equivalent to adding a voltage source at weak buses. The inverter in SOP is controlled by an adaptive GFM type control of Virtual Synchronous Generator (VSG), which adjusts the inertia J and damping D parameters adaptively to achieve better dynamic performance of bus voltage during cluster active power support and voltage regulation. If the voltage needs further support, the ES is GFM controlled to support the voltage, and the Virtual Direct Current Machine (VDCM) control strategy is used. In summary, E-SOP achieves voltage support for weak buses by connecting active power between clusters and the GFM controlled ES.

4. Identification of System Models and Comparison of Voltage Over-Limit Control Schemes

4.1. Identification of System Model

The equivalent impedance and power source of the system can be identified by performing Thevenin equivalence when looking through the load terminal to the system. The equivalent system can be used to evaluate the active power transmission capability of the system, that is, the power supply capability. This paper identifies the equivalent models of the system for 24 h per day based on the results of power flow calculations. For example, Thevenin equivalent circuits of the system can be obtained from the buses of the 18 and 33 with large voltage fluctuations. The corresponding active power-voltage (P-V) curve can be plotted, which is intuitive to showing the maximum transmitted power.

4.2. Comparison of Voltage Over-Limit Regulation Schemes

4.2.1. System Expansion Scheme

Improving the short-circuit capacity of the system can enhance its strength, thereby improving voltage stability. After increasing the short-circuit capacity of the 12.66 kV power supply system from the original 100 MVA to 1000 MVA, simulation analysis was conducted on the effective voltage value. The results show that the voltage of bus 33 still exceeds the upper limit at 01:00, and the voltage of bus 18 still exceeds the lower limit at 18:00. The results are shown in

Table 1. RMS value of bus voltage after increasing the power supply short-circuit capacity.

Time	Bus	100 MVA	1000 MVA
01:00	18	12.639 V	12.816 kV
	33	13.272 kV	13.594 kV
18:00	18	11.642 kV	12.017 kV
	33	12.479 kV	12.861 kV

.

4.2.2. Reactive Power Compensation Scheme

The short-circuit capacity of the 12.66 kV power supply in the system remains at the original 100MVA. Due to the issue of exceeding the lower limit at bus 18, a reactive power compensation device is connected to bus 18. Simulation results also show that this scheme is effective for the problem of voltage exceeding the lower limit, while it cannot solve the problem of voltage exceeding the upper limit. The simulation results of voltage values at buses of 18 and 33 at two over-limit moments are shown in

Table 2. Voltage of bus 18 when connected with reactive power compensation device.

Time	Bus	Voltage (kV) without Comp.	Voltage (kV) with 75 kVar Comp.	Voltage (kV) with 200 kVar Comp.	Voltage (kV) with 700 kVar Comp.
01:00	18	12.639	13.06	13.18	13.27
	33	13.272	13.53	13.58	13.67
18:00	18	11.642	11.73	11.87	12.42
	33	12.479	12.76	12.98	13.21

.

4.2.3. Installation of E-SOP Scheme Between Clusters

Firstly, only the power support of SOP was considered, without considering the supporting effect of ES on bus voltage. After connecting buses 18 and 33, the voltage of each bus is shown in Figure 9.

According to Figure 9, all bus voltages in cluster 1 and cluster 2 do not exceed the limit within 24 h throughout the day, but the voltage fluctuations of each bus in cluster 1 are severe. The voltage of bus 33 in cluster 3 slightly exceeds the lower limit at 18:00, while there are no bus voltages exceeding the higher limit. The voltages of bus 11–12 in cluster 4 exceeded the upper limit at 01:00, and the voltage of bus 18 exceeded the lower limit at 18:00. However, comparing the voltage distribution within each cluster in Figure 3 and Figure 9, it can be seen that after connecting clusters 3 and 4 through the SOP device, the problem of voltage exceeding the lower limit has been greatly improved; however, some buses have not yet met the standard. At the same time, the problem of cluster 3 exceeding the upper limit has been suppressed to a certain degree; however, it shows new issues of some bus voltages exceeding the upper limit in cluster 4. When the battery capacity is configured to 1300 kWh, it can be seen that the problem of the voltage exceeding the upper limit for all buses is solved.

Installing battery ES can make the voltage closer to the standard value and further reduce voltage fluctuations. As shown in Figure 9d, the voltage value of bus 18 in cluster 4 at 18:00 is the lowest, at 12.018 kV, corresponding to a per unit value of 0.9493. When E-SOP is configured with 1100 kWh energy storage, the voltage at bus 18 is raised to 12.524 kV, corresponding to a per unit value of 0.9893. It can be seen that the energy storage unit in E-SOP achieves further regulation of bus voltage.

Take the larger ES capacity value of 1300 kWh for suppressing the upper and lower voltage over-limit as the final configuration capacity. Figure 10 shows the voltage distribution of buses in each cluster after installing an E-SOP device with an energy storage capacity of 1300 kWh. It can be seen that all voltages are controlled within the specified range, and the voltage fluctuation is significantly reduced.

4.3. Comparation of Voltage Over-Limit Suppression Schemes

Figure 11a,b show the P-V curves of the equivalent system viewed from buses 18 and 33 at 01:00 am and 18:00 pm, without any voltage compensation device, after system expansion, after installation of reactive power compensation device at bus 18, after connection using only SOP device, and after connection through an E-SOP device, respectively. It can be seen that the SOP scheme can effectively improve the transmission capacity of the system more compared to system expansion and reactive power compensation schemes. Furthermore, configuring limited capacity ES for SOP can further overcome the issue of voltage over-limit; ES units can further release energy to support the voltage exceeding the lower limit effectively.

5. Conclusions

This paper applies the E-SOP based on MMC topology to suppress issues of bus voltage exceeding limits in the distribution network. In terms of control strategy, both the inverter module and the energy storage device can adopt grid-forming control to achieve voltage support for weak buses.

The comprehensive indicator of electrical distance and power balance is used to divide the improved IEEE 33 bus distribution network containing new energy into clusters firstly, enabling each cluster to achieve a balance of active and reactive power within the cluster. However, some clusters still experience voltage limit issues. In this case, the weak voltage buses in each cluster are evaluated through the standard deviation of voltage sensitivity indicator and then the impact of active and reactive power injection on the bus voltage is analyzed through the newly proposed power sensitivity deviation rate indicator, obtaining the optimal location configuration for E-SOP. Furthermore, energy storage capacity is greatly reduced due to the SOP power support function, it is only needed to suppress the voltage fluctuations. Through simulation verification, the optimized E-SOP device has a better voltage support effect than traditional reactive power compensation and system expansion schemes and also reduces the cost of configuration and system modification. It can be seen that E-SOP is an effective voltage support device for distribution networks, which can effectively solve the voltage limit problem caused by distributed new energy sources, improve the operational efficiency and reliability of the system, and promote the consumption of new energy through the SOP active power support function between clusters, as well as the support of low voltage by GFM energy storage and its consumption function of new energy.

Based on SOP, E-SOP is actually a network reconfiguration method. However, this paper only applied one E-SOP for voltage regulation between the two weakest buses in different clusters without carrying out the optimization configuration of E-SOP throughout the whole distribution network. In addition, this paper only considers the configuration of energy storage based on the voltage fluctuation indicator. In the future, further optimal configuration and operation of SOP and its energy storage should be combined with an economic indicator. Finally, the issue of device protection needs to be considered in practical applications.