Integrated Volt/Var Control Method for Voltage Regulation and Voltage Unbalance Reduction in Active Distribution Networks

Abstract

The emergence of distributed generation such as solar systems has introduced new challenges in distribution networks that are becoming more apparent with increasing penetration levels. The time mismatch between peak load and peak generation can make voltage levels in distribution networks swing towards extreme limits during a day. Distribution network service providers are struggling to cater for new distributed generation installations while ensuring that the quality of steady state supply voltage meets stipulated requirements. The segregation between medium and low voltage networks in control strategies may result in unnecessary or worse, opposing control actions leading to voltage issues and control on one side of the distribution network influencing the other. By developing an efficient volt/var control method, the steady state voltage level and voltage unbalance in a comprehensive distribution network can be controlled simultaneously. This paper analyses voltage issues and volt/var control in medium and low voltage networks as a whole, unabridged problem. This paper proposes a pragmatic and effective volt/var control method that addresses voltage regulation and voltage unbalance simultaneously using existing infrastructure. The proposed control method is implemented on three types of representative Australian distribution networks and results obtained demonstrate that the proposed volt/var control can simultaneously manage voltage level and voltage unbalance whilst reducing the number of tap change operations and maximizing solar penetration.

1. Introduction

In recent years, awareness of the impacts of fossil-fuel-based generation has led to an increased penetration of distributed generation (DG), especially solar photovoltaic (PV) systems in medium voltage (MV) and low voltage (LV) networks. Renewable DG units are appealing being sustainable, promise a satisfactory return of investments and are also closer to consumers, thus reducing the losses along the feeder [1]. This progress is in fact desirable, as it helps to reduce the carbon footprint, however, the widespread use of these devices introduce new challenges in distribution networks.

The traditional power flow in distribution networks is unidirectional; from the upstream source to the consumer. DG introduces multi-sources and bi-directional power flow in distribution networks. This increased penetration of DG causes voltage rise in distribution feeders, specifically in LV networks [2,3]. Voltage rise occurs when the power generation from DG is greater than the load demand, for example, during a sunny midday when power generated by PV systems exceeds the load consumption at predominantly empty residential areas. This in turn causes the power to flow in a reverse direction since the generation at the consumer location is higher [4].

Several methods have been proposed to mitigate voltage rise in distribution networks. These methods include grid reinforcement, transformer tap change operations and demand side management strategies such as incorporation of energy storage [5]. Among the earliest methods that have been proposed by many is active power curtailment [5,6,7]. With active power curtailment, once the voltage reaches the preliminary limit, the active power generated by the DG system would be curtailed so that the voltage at the point of common coupling (PCC) stays within statutory limits. This is especially true in cases where PV systems are oversized to gain the maximum benefit of installation [8]. This method requires minimal modification of the control logic of PV inverters, hence making it very attractive [4]. Nonetheless, this method is undesirable, as PV owners cannot get as much revenue from their investments [9]. However, when PV inverters draw reactive power, voltage level at the terminal can be reduced, consequently allowing proportional active power to be injected, subject to the capacity of the inverters.

Inverters are versatile voltage regulation devices being capable of supplying and drawing reactive power simply by changing the phase shift between the voltage and current. Such an ability can be used for localized voltage support, especially during high PV penetration and high load demand in LV networks. The revision of the Australian standard for PV inverter requirements in AS 4777.2 has included the addition of volt/var control capability for PV inverters in distribution networks [10]. Though the volt/var control function should be available, its operation for voltage support is yet to be made compulsory in Australia.

LV networks are usually not explicitly modeled in comparison to high voltage (HV) and MV networks. LV networks are also typically not well monitored, hence, the impacts of emerging technology are not well understood [11]. Distribution network service providers (DNSPs) usually rely on customer complaints to recognize voltage issues in their networks, and subsequently, resolve the issues as required. This is not ideal as active prevention of potential voltage issues might not happen while other unreported issues might escape the attention of DNSPs.

Previous voltage regulation practices are mostly carried out on the MV side of distribution networks only [12,13]. The only voltage regulating device in LV networks is the distribution transformer with off-load tap changer, which in Australia is often set to cater the peak load. Given the installation trend and pricing of DG, the DG energy sector is most likely to grow, which may consequently increase the number of DG-related voltage issues [14].

Voltage regulation methods in LV networks used to be isolated from the associated MV networks, with the assumption that the MV networks are not affected by the control methods [15]. However, voltage issues as well as voltage control actions on one side of the distribution network might have significant effects on the other. With increased penetration of PV causing voltage issues in both MV and LV networks, the segregation of voltage regulation methods between MV and LV networks is outdated in the current world.

References [16,17,18] for example, only elaborate strategies about regulating the voltage in the MV network while [5,19,20] aim at regulating the voltage in the LV network. While reported literature can successfully achieve the target voltage regulation, due to the separation between MV and LV networks, the proposed strategies are not sufficient in the constantly evolving active distribution networks. Therefore, a voltage regulation method that addresses challenges in MV and LV networks as a whole while efficiently utilizing all available resources in the network is required to bridge the gaps in current voltage regulation practices.

In addition to maintaining the steady state voltage magnitude in distribution networks, DNSPs are also responsible to ensure that the voltage unbalance level is regulated. According to Australian standard AS 61000.2.2, the long-term voltage unbalance level must not be more than 2% or 3% in areas where large connections of single-phase loads are allowed [21].

Voltage unbalance is generally a concern in LV networks compared to HV and MV networks. While HV transmission conductors are transposed to keep an approximately equal phase impedance, thereby minimizing voltage unbalance, LV conductors are usually shorter, thus untransposed. In addition, loads connected in LV networks can be a combination of single-phase, two-phase or three-phase, which contributes to more complexity in terms of voltage unbalance in LV networks. Though DNSPs endeavor to distribute loads equally across all phases, load demands are irregular and rely heavily on weather, season and time of the day. Adding this together with the aforementioned network structure, some level of inherent voltage unbalance in LV networks is inevitable.

The proliferation of PV system installation in distribution networks can worsen voltage unbalance in LV networks, especially with unequal distribution of single-phase rooftop PV systems [17]. PV penetration is highly stochastic in nature, rendering load and PV rebalancing impractical. Though PV systems may exacerbate voltage unbalance, with appropriate strategies, these systems can be utilized effectively to counteract the negative impact [22].

Reference [23] proposes a new concept for volt/var control that utilizes distributed hybrid power electronic devices at the grid edge for voltage regulation. In addition to regulating the voltage level, the authors observed that the control method unintentionally improved the voltage unbalance level in the tested network. Reference [24] proposes a three-phase optimal power flow formulation to determine the optimal volt/var curve settings for PV inverters to regulate the network voltage level closer to specified set-points, so that the voltage unbalance level can be minimized. The proposed method is tested on a real LV distribution network with favorable results on both voltage regulation and voltage unbalance reduction. This shows that volt/var control can be properly designed to regulate voltage level as well as unbalance level, and the application can be extended to include a more comprehensive MV-LV distribution network, which [24] does not cover.

Due to its inherent nature, voltage unbalance cannot be completely eliminated from distribution networks, however it can be kept under control with proper mitigation strategies. Mitigation strategies for voltage unbalance include load balancing, redistribution of single-phase loads and the use of single-phase devices such as regulators and distribution static compensators (DSTATCOMs) [25]. This paper aims at utilizing single-phase volt/var devices available in the network in terms of residential solar PV inverters to simultaneously regulate steady state voltage and reduce voltage unbalance.

This paper extends the analytical formulation proposed by the authors in [15] to devise a volt/var control method that regulates the voltage level across a combined MV-LV network while reducing voltage unbalance. While the proposed control in [15] has remained within a unity power factor operation of PV inverters, non-unity power factor operation of PV inverters not bounded by AS 4777.2 will mainly be considered in this paper. Active participation of PV inverters in voltage regulation will also be taken into account with the presumption that the inverters are capable of operating at full range of power factor conditions.

The rest of the paper is organized as follows. Section 2 presents the development of the proposed volt/var control method for voltage regulation. Section 3 presents the development of the proposed volt/var control method for simultaneous voltage regulation and voltage unbalance reduction. Section 4 describes the network models used to validate the proposed control method. Section 5 presents the simulation process, results obtained and relevant analysis. Section 6 concludes the paper and discusses possible future developments.

2. Proposed Volt/Var Control in Active Distribution Networks for Voltage Regulation

The operation of residential PV inverters in Australia is restricted to a displacement power factor of 0.95 leading or lagging for a number of reasons including the possibility of negative interaction between multiple PV inverters in close proximity and lack of efficient market strategies for reactive power [10]. Nevertheless, if these concerns are properly addressed, the operation of PV inverters at non-unity power factor might bring more advantages than disadvantages to both DNSPs and customers.

When PV systems are generating a significant amount of active power during low load conditions, some active power might have to be curtailed due to voltage constraints. This is especially true in cases where solar PV panels are oversized to gain the maximum benefit of PV system installations. This is undesirable as PV owners cannot get as much revenue from their investments. However, when PV inverters draw reactive power, this will help to reduce the voltage level at the PCC, consequently allowing proportional active power to be injected, subject to the capacity of those inverters.

DNSPs are responsible for maintaining acceptable power factors in the networks. Accordingly, reactive power devices such as capacitor banks and DSTATCOMs are utilized. This investment along with the maintenance of var devices can be lowered when customer-owned PV systems are allowed to actively correct the power factor in accordance to specific DNSP requirements. In a nutshell, the collective operation of PV systems to actively regulate voltage and power factor can benefit the customers and DNSPs.

With the focus on devising a practical volt/var control strategy that will be applicable for a whole distribution network, this section describes the associated analytical formulation. This is a continuation from the analytical formulation proposed in [15].

2.1. Integrated Volt/Var Control Formulation

Based on the analytical formulation described in [15], this section extends the formulation to consider a compensation-based volt/var control strategy. This formulation considers all available volt/var devices in a network such as OLTCs, capacitor banks and PV inverters.

Figure 1 represents a per unit equivalent circuit for a whole distribution network. In Figure 1, substation transformer and distribution lines are represented as impedances (Z), while loads, capacitors and PV systems are represented as current injections predominantly for one phase with the other two phases shown for completion only. When the voltage at bus i, $V_i$, violates the pre-set voltage limit, a new target voltage, $V_{\mathrm{target}}$, which is within the allowable range, will be set. The difference between these two voltages, $\Delta V_{\mathrm{imp}}$, will be:

$$\Delta V_{\mathrm{imp}}=V_{\mathrm{target}}-V_i$$

(1)

To achieve $V_{\mathrm{target}}$, the corresponding current injection required, $I_{\mathrm{imp}}$, is:

$$I_{\mathrm{imp}}=Z_{\mathrm{ZS},\mathrm{PCC}}^{-1}\Delta V_{\mathrm{imp}}$$

(2)

where $Z_{\mathrm{ZS},\mathrm{PCC}}$ is the matrix of line impedance from the zone substation (ZS) to the PCC between bus i and the volt/var device location.

Assuming that all other current injections remain the same, the new total current injection at the bus i, $I_{i\mathrm{new}}$, is:

$$I_{i\mathrm{new}}=I_i-I_{\mathrm{imp}}$$

(3)

Using $I_{i\mathrm{new}}$, the new $\Delta V$ at bus i can be calculated using Equation (4) [15].

$$\begin{array}{r}\Delta V_i=(R_{\mathrm{ZS},i}+j{X}_{\mathrm{ZS},i})(I_{i,n}^{\mathrm{re}}+j{I}_{i,n}^{\mathrm{im}})\end{array}$$

(4a)

$$\begin{array}{c}\hfill \Delta V_i=R_{\mathrm{ZS},i}{I}_{i,n}^{\mathrm{re}}-X_{\mathrm{ZS},i}{I}_{i,n}^{\mathrm{im}}+j(X_{\mathrm{ZS},i}{I}_{i,n}^{\mathrm{re}}+R_{\mathrm{ZS},i}{I}_{i,n}^{\mathrm{im}})\end{array}$$

(4b)

where $R_{\mathrm{ZS},i}$ and $X_{\mathrm{ZS},i}$ are the line resistance and reactance between ZS bus and bus i (as part of impedance $Z_1$) and $I_{i,n}^{\mathrm{re}}$ and $I_{i,n}^{\mathrm{im}}$ are the real and imaginary current injections from bus i to bus n.

The approximate new voltage at bus i, $V_{i\mathrm{new}}$, will be:

$$V_{i\mathrm{new}}=V_{\mathrm{ZS}}-\Delta V_i$$

(5)

Using the values of $V_{i\mathrm{new}}$ and $I_{\mathrm{imp}}$, the rating, $S_{\mathrm{device}}$, of the volt/var device required to correct the voltage level is:

$$S_{\mathrm{device}}=V_{i\mathrm{new}}{I}_{\mathrm{imp}}^{*}$$

(6)

$S_{\mathrm{device}}$ consists of the values of active and reactive power injections required at the volt/var device bus to achieve $V_{\mathrm{target}}$ at bus i.

If the volt/var device provides or draws only reactive power, the angle of required current injection, $I_{\mathrm{imp}}$, at the device bus must be close to ±90°. For example, when calculating how much capacitive current should be injected to increase the voltage in a network, the angle of $I_{\mathrm{imp}}$ must be close to −90°. If the calculated angle of $I_{\mathrm{imp}}$ is bigger than this, the angle of $V_{\mathrm{target}}$ must be reduced, whereas if the calculated angle of $I_{\mathrm{imp}}$ is smaller than −90°, the angle of $V_{\mathrm{target}}$ must be increased.

When doing the calculations for phases B and C, the angle of $I_{\mathrm{imp}}$ can be moved to the same quadrant as phase A by multiplying the current phasor by 1∠120° and 1∠−120°, respectively (considering phase A as the reference). With this, the angle of $I_{\mathrm{imp}}$ can be compared to ±90° as expected from a purely inductive or capacitive load.

2.2. Development of Integrated Volt/Var Control Methodology

The proposed volt/var control strategy considers all available volt/var devices in the network as well as non-unity power factor operation of PV inverters. Using the proposed volt/var control strategy, PV inverters will be used at night (i.e., when PV systems are not generating any active power) to provide reactive power for local voltage support. This is in order to relieve the network capacity to meet the peak load demand.

$$\left|V_{\mathrm{minimum}}\right|\le \left|V_{\mathrm{warn}-\min }\right|\le \left|V_i\right|\le \left|V_{\mathrm{warn}-\max }\right|\le \left|V_{\mathrm{maximum}}\right|$$

(7)

The volt/var control algorithm continuously monitors voltage magnitudes across the entire network. When the voltage level at a bus violates the specified limit as shown in (7), the control algorithm finds how many volt/var devices are available to carry out the corrective action. Then, the control algorithm calculates how much $I_{\mathrm{imp}}$ and subsequent $S_{\mathrm{device}}$ are required to correct the voltage level. When only reactive power is demanded by the volt/var device to regulate the voltage level, the corresponding $I_{\mathrm{imp}}$ must have either +90° or −90° angle. In this case, the control algorithm will run for multiple iterations using new angle for $V_{\mathrm{target}}$ until a satisfactory angle for $I_{\mathrm{imp}}$ is achieved. If the algorithm fails to converge to a satisfactory angle, a tap change or active power curtailment will be activated.

In addition, these following approaches are adopted for capacitor and PV inverter operations:

• Capacitor operation
  The control algorithm calculates the amount of reactive power required to bring the voltage level closer to 1 p.u. The calculated reactive power is then rounded to a whole number and the closest switching position of the capacitor will be selected, if available. Otherwise, a tap change operation will be carried out.
• PV inverter operation
  To reduce the voltage level during high PV penetration, the control algorithm will start by calculating the amount of reactive power required to be drawn from the network. If this is not possible, then the algorithm will calculate the amount of active power to be curtailed and reactive power to be drawn at the same time. On the other hand, to raise the voltage level when active power generation is unavailable, the control algorithm will calculate how much reactive power injection is required.

In the proposed control algorithm, the lower and upper voltage limits ($V_{\mathrm{minimum}}$ and $V_{\mathrm{maximum}}$) for LV networks are set to 0.95 p.u. and 1.09 p.u., respectively, to ensure that the voltage level across the entire network does not stray outside the established limits in Australian standard AS 60038 [26]. The voltage limits for MV networks are set to ±9% of 1 p.u. As shown in (7), the control algorithm continuously checks if any voltage level violates $V_{\mathrm{warn}-\min }$ and $V_{\mathrm{warn}-\max }$ to ensure that the selected corrective action will not exacerbate the voltage level across the entire network. $V_{\mathrm{warn}-\min }$ and $V_{\mathrm{warn}-\max }$ for LV networks are set to 0.96 p.u. and 1.08 p.u. respectively and ±8% for MV networks.

The control algorithm will cap PV system injection during peak PV penetration to avoid overvoltage situation. This is determined using $S_{\mathrm{device}}$, calculated using (6). For example, when the voltage level at bus i violates the maximum voltage limit, $V_{\mathrm{maximum}}$, the control algorithm calculates $S_{\mathrm{device}}$ required to correct the voltage level. In this case, $S_{\mathrm{device}}$ refers to apparent power required by the PV system at bus i. The total PV system injection, $S_{i,\mathrm{Pnew}}$, calculated using (8) will be set as the new maximum acceptable PV system output.

$$S_{i,\mathrm{Pnew}}=S_{i,\mathrm{P}}-S_{\mathrm{device}}$$

(8)

More active power injection will not be allowed unless more reactive power can be drawn, as restricted by the PV inverter capacity, to ensure that overvoltage situations can be prevented. This also ensures no voltage excursions will happen when PV systems try to reinject full active power generation after the voltage level has been corrected. The control algorithm will recalculate $S_{\mathrm{device}}$ when a significant change in voltage level is observed. With this, more active power injection will be allowed once a significant increase in load demand happens. Using the proposed volt/var control strategy, the operating power factor of the PV systems is not restricted by the standard AS 4777.2 [10], however, it is governed by the calculated $S_{\mathrm{device}}$ and size of PV inverters.

When there are several volt/var devices, hence multiple possibilities of control operations, the control algorithm will select a device operation based on hierarchical priorities as follows:

1.

The voltage level across the entire network must be within allowable range.

2.

The number of tap changes by the OLTC is minimized.

3.

The active power injection from PV systems is maximized.

4.

The number of consecutive device operations is minimized.

The control algorithm will initially select a local device operation that will improve the voltage level at the target bus. However, if a single integrated action can improve the voltage level at multiple buses, this action will be selected instead of multiple, localized device operations. Any device operation that will correct the voltage level at a bus while deteriorating the voltage level at other buses will be avoided. For example, if there is a voltage rise in the LV network at the same time as a voltage drop in the MV network, the best course of action will be selected based on the hierarchical priorities and device availabilities.

Figure 2 summarizes the proposed volt/var control strategy. Based on the calculation of $S_{\mathrm{device}}$, adequate amount of injection to compensate for a voltage drop or rise in a network can be found. With this, excessive compensation can be avoided and the most favorable operation of volt/var devices can be achieved. The hierarchical conditions for the volt/var control can be customized based on DNSPs and specific network requirements. With this, volt/var control operations that best suited a particular DNSP or distribution network can be prioritized.

3. Proposed Volt/Var Control for Voltage Regulation and Voltage Unbalance Reduction

This section presents the analytical background of the proposed volt/var control method for the purpose of simultaneous voltage regulation and voltage unbalance reduction in a combined MV-LV distribution network. The proposed volt/var control method is inspired and derived from previous proposed volt/var control methods from [15,27]. The proposed volt/var control method consists of a voltage regulation scheme from Section 2 and a newly developed voltage unbalance reduction scheme.

3.1. Voltage Unbalance Reduction Scheme

According to Australian standard AS 61000.2.2, the long-term voltage unbalance level must not be more than 2% or 3% in areas where large connections of single-phase loads is allowed [21]. This standard specifies voltage unbalance factor (VUF) as the ratio of negative sequence component to the positive sequence component, as described by (9).

$$\%\mathrm{VUF}={\displaystyle \frac{\mathrm{negative}\mathrm{sequence}\mathrm{voltage}}{\mathrm{positive}\mathrm{sequence}\mathrm{voltage}}}\times 100$$

(9)

The control algorithm for this voltage unbalance reduction scheme continuously monitors the VUF level in the three-phase LV network buses. When the VUF level violates a pre-set limit, the control algorithm selects the phase voltage(s) to be corrected to reduce the VUF level. The phase voltage selection is derived from the definition of phase voltage unbalance ratio (PVUR) shown in (10).

$$\%\mathrm{PVUR}={\displaystyle \frac{\mathrm{maximum}\mathrm{deviation}\mathrm{from}\mathrm{average}\mathrm{phase}\mathrm{voltage}}{\mathrm{average}\mathrm{phase}\mathrm{voltage}}}\times 100$$

(10)

Although (10) defines voltage unbalance differently, by reducing the maximum deviation of phase voltage from the average three-phase voltage, this will be reflected in the reduction of VUF level.

The phase voltage correction is selected based on the level of VUF violation from the pre-set limit [22]. The phase voltage selection is based on two different cases:

• When the VUF level is between $VU{F}_1$ and $VU{F}_2$ as shown in (11), the control algorithm calculates the average phase voltage, $V_{\mathrm{avg}}$, using (12).

  $$VU{F}_1\le VUF<VU{F}_2$$

  (11)

  $$V_{\mathrm{avg}}={\displaystyle \frac{V_{\mathrm{A}}+V_{\mathrm{B}}+V_{\mathrm{C}}}{3}}$$

  (12)
  A float voltage of 1 p.u. is established as $V_f$. The difference between each phase voltage, $V_{\mathrm{phase}}$, from $V_f$, and the difference between $V_{\mathrm{avg}}$ from $V_f$ are calculated using (13) and (14), respectively.

  $$d{V}_{\mathrm{phase}}=|V_{\mathrm{phase}}-V_f|$$

  (13)

  $$d{V}_{\mathrm{avg}}=|V_{\mathrm{avg}}-V_f|$$

  (14)
  If $d{V}_{\mathrm{phase}}$ is greater than $d{V}_{\mathrm{avg}}$, then $V_{\mathrm{phase}}$ will be corrected whereas if $d{V}_{\mathrm{phase}}$ is less than $d{V}_{\mathrm{avg}}$, the voltage on the other two phases will be brought closer to $V_f$ because the voltage on this phase is the main cause for the high level of VUF that prompted the voltage unbalance reduction scheme.
• When the VUF level is higher than $VU{F}_2$ as shown in (15), the voltage on all three phases will be brought closer to $V_f$. This collective operation is to ensure that the correction action will reduce the VUF level at the monitored bus.

  $$VUF\ge VU{F}_2$$

  (15)

The proposed voltage unbalance reduction scheme aims to increase/decrease the phase voltage level to $V_f$ instead of $V_{\mathrm{avg}}$ to ensure a smaller voltage variance from $V_f$. In addition, since PV systems in this proposed voltage unbalance reduction scheme are not bounded by any power factor operation limit, higher level of voltage correction can be achieved.

In this proposed voltage unbalance reduction scheme, the devices utilized are single-phase PV systems in the LV network. Single-phase PV systems connected at the impacted bus will operate to bring the VUF level to an acceptable value. If there are no PV systems connected at the impacted bus, PV systems located at the shortest electrical distance from this bus will operate.

The phase voltage correction is calculated using the compensation based method proposed in Section 2.1. This is to avoid excessive compensation and ensure maximum active power penetration from PV systems. The amount of current injection required by a PV system to correct the phase voltage level is determined using (2). In this case, $V_{\mathrm{target}}$ is $V_f$. Subsequent required PV system injection is determined using (6) and (8).

Once the voltage unbalance reduction scheme is activated and new $S_{i,\mathrm{Pnew}}$ for associated PV systems are calculated, all associated PV systems will operate at this new injection until the VUF level is less than pre-set limits. For case 1, the VUF level must be less than or equal to $VU{F}_3$ before the voltage unbalance reduction scheme is deactivated, as shown in (16).

$$VUF\le VU{F}_3$$

(16)

For case 2, the VUF level must be less than or equal to $VU{F}_4$ before the voltage unbalance reduction scheme is deactivated, as shown in (17).

$$VUF\le VU{F}_4$$

(17)

The purpose of (16) and (17) is to avoid the ‘hunting’ of PV system operation in the process of reducing VUF level. This will also directly prevents voltage excursions due to PV system operation. The VUF limits $VU{F}_1$, $VU{F}_2$, $VU{F}_3$ and $VU{F}_4$ can be set to values that suit the requirements and peculiarities of DNSPs and distribution networks. Figure 3 summarizes the proposed voltage unbalance reduction scheme while a summary of variables is presented in Appendix A.

3.2. Final Control Algorithm

The final control algorithm for the proposed volt/var control method includes the previously proposed integrated volt/var control strategy from Section 2 as the voltage regulation scheme and the voltage unbalance reduction scheme discussed in Section 3.1. In addition to monitoring the voltage level across the entire controlled network, the proposed volt/var control method monitors the VUF level at three-phase buses in the LV network. When any monitored variables violate their respective limits, corresponding control schemes will be activated.

Voltage unbalance level violation typically happens when the three-phase voltage levels are within allowable bounds, since if the voltage levels violate the minimum or maximum limits, they will be immediately corrected. Therefore, though this proposed volt/var control method have both voltage regulation and voltage unbalance reduction schemes, the voltage regulation scheme takes precedence over the latter.

In descending importance, the proposed volt/var control method aims to ensure that:

1.

The voltage level across the entire network is within allowable limits.

2.

The VUF level at three-phase LV buses is less than the maximum allowable limit.

3.

The number of tap changes by the OLTC is minimized.

4.

The active power injection from PV systems is maximized.

5.

The number of consecutive device operations is minimized.

Compared to the previous aims from Section 2.2, the final proposed volt/var control algorithm has an additional aim. The proposed volt/var control method will select device action(s) based on the hierarchical priorities. For example, if a certain PV system injection will cause a voltage level or VUF level violation, the action will not be carried out, and alternative corrective action(s) will be determined. The hierarchical priorities of the proposed volt/var control method can be changed to suit a particular distribution network or DNSP.

The proposed volt/var control method performs voltage regulation and reduce voltage unbalance simultaneously using the same volt/var devices available in the network. For networks that are already equipped with advanced metering infrastructures (AMIs), the implementation of this proposed volt/var control method does not require any additional investment. Therefore, the implementation of the newly proposed volt/var control can not only ensure voltage regulation and voltage unbalance reduction, but is also financially beneficial, as DNSPs can defer new investments while prolonging the lifespan of network assets such as OLTCs.

4. Network Models

The proposed volt/var control is applied on three types of representative MV networks combined with an LV network extension introduced in [27].

1.

Figure 4 shows Network 1, which is a short MV network consisting of 2.2 km underground cable and an average load of 2.6 MVA/km. An LV section is connected at the LV bus bar location.

2.

Figure 5 shows the second representative network, Network 2. This network consists of combined underground cables and overhead conductors with a total length of 7.5 km. An LV section is connected at either location 1 or 2.

3.

Figure 6 shows Network 3, which has a majority of overhead conductors with a total length of 143 km and an average load of 100 kVA/km. An LV section is connected at either location 1 or 2.

The LV network section connected to each MV representative network is developed from a realistic LV network in Australia. As shown in Figure 7, the LV network section is connected through a distribution transformer at the distribution substation (DS). Each line section between LV buses is a three-phase overhead line, and single-phase service lines are connected from the buses to supply single-phase loads and PV systems. The line lengths are modified based on the specific type of MV network that this LV network section is connected to.

5. Simulation and Results

For simulation purposes, the proposed volt/var control method is implemented in five different cases, which are:

1.

Network 1

2.

Network 2 with the LV section at 1;

3.

Network 2 with the LV section at 2;

4.

Network 3 with the LV section at 1;

5.

Network 3 with the LV section at 2.

The representative networks are modeled in OpenDSS while the proposed volt/var control algorithm is implemented in MATLAB. The simulation is run for a 24 h period in which the load and PV profiles vary. The load and PV profiles used for this simulation are depicted in Figure 8. All other network conditions such as the tap positions at the ZS and DS, connected loads and PV systems across the network are as follows:

• The off-load tap changer at the DS is set to 1.06 p.u.
• The OLTC at the ZS is initially set to 1.03 p.u.
• Each LV customer is equipped with a PV system.
• The total aggregated LV load is 300 kVA with 0.9 lagging power factor to ensure uniformity for all cases.
• The distribution of load and PV system across each phase is depicted in Table 1.
• The PV penetration is set to 200% of the PV system distribution.

Table 1. Distribution of load and PV injection for each phase.

Criteria	Phase
	A	B	C
Load (kVA)	15	10	12.5
Load (pf)	0.85	0.95	0.9
PV (kW)	6.25	18.75	12.5

Table 1. Distribution of load and PV injection for each phase.

Criteria	Phase
	A	B	C
Load (kVA)	15	10	12.5
Load (pf)	0.85	0.95	0.9
PV (kW)	6.25	18.75	12.5

Using the newly proposed volt/var control method, the limits $VU{F}_1$, $VU{F}_2$, $VU{F}_3$ and $VU{F}_4$ are specified as shown in

Table 2. VUF limits for the newly proposed volt/var control method.

Label	Value (%)
$VU{F}_1$	1.6
$VU{F}_2$	1.8
$VU{F}_3$	1.4
$VU{F}_4$	1.2

. The VUF limits are set less than the maximum allowable VUF level of 2% stipulated by Australian standard to ensure that voltage unbalance violation will not happen in the tested networks. $VU{F}_3$ and $VU{F}_4$ are set to lower values to ensure that the VUF level has been kept under control before the voltage unbalance reduction scheme is deactivated.

The results obtained are analyzed from voltage regulation and voltage unbalance reduction perspectives. An effective implementation of the proposed control system is reflected in a ‘healthy’ voltage profile across the entire network and VUF level that is below the maximum limit. In addition, the proposed control system should have a minimum number of switching device operation and maximum active power injection from PV systems.

5.1. Voltage Regulation

Subsequent simulation results are presented in 5-min intervals of the simulation period for clarity. To verify the proposed volt/var control strategy, the results obtained will be compared to those using holistic volt/var control (HVVC) method proposed by the authors in [27]. The results obtained for Cases 1 to 4 are described briefly whereas more detailed explanation is for Case 5.

1.

Case 1

Compared to results obtained using HVVC in [27], this proposed volt/var control strategy operates one tap change at the beginning of simulation. This is to reduce the voltage level in the MV network closer to 1 p.u. The minimum voltages for LV and MV networks are 1.045 p.u. and 1 p.u., respectively, while the maximum voltages for LV and MV networks are 1.091 p.u. and 1.025 p.u., respectively. This brings the overall voltage level using this proposed control within a smaller variance to 1 p.u. In contrast to using HVVC, the maximum PV penetration recorded is 200%, which is made possible by the tap operation at the beginning of the simulation.

2.

Case 2

Using the proposed volt/var control strategy, the number of tap change remains at two. The minimum voltages for LV and MV networks show negligible improvements from those using HVVC. The maximum voltage for MV network remains at 1.042 p.u. while the maximum voltage for LV network is reduced to 1.094 p.u. This is due to the reduction of maximum allowable PV penetration to 176%. Similar to Case 1, the voltage level across the network using this proposed control strategy is within a smaller variance to 1 p.u.

3.

Case 3

Using the proposed control strategy, the number of tap change is reduced to one. In this case, the only significant difference in extreme recorded voltage is the maximum voltage in LV network, which has been reduced to 1.089 p.u. The minimum voltage in LV network and the extreme voltages in MV network do not show significant improvement from HVVC in [27]. Similar to using HVVC, the maximum PV penetration is 200%.

4.

Case 4

Compared to using HVVC, the proposed volt/var strategy operates one tap change to increase the voltage level during peak load. This brings the minimum voltages in LV and MV networks to 1.011 p.u. and 0.948 p.u., respectively. The maximum voltages in LV and MV networks are 1.093 p.u. and 1.042 p.u., respectively. In this case, the voltage level in MV network is within a wider variance compared to using HVVC in [27]. This is due to corrective actions taking place at a later interval compared to the capacitor switching operations in [27]. The extra tap operation also increases the maximum recorded voltage in the MV network compared to that using HVVC. On the other hand, the voltage level in the LV network is kept within a closer band to 1 p.u. compared to that using HVVC. Additionally, due to non-unity power factor operation, the maximum allowable PV penetration is increased to 200%. In this case, PV systems on phases B and C are operating at lagging power factor while injecting a capped active power, whereas PV systems on phase A are drawing reactive power, allowing more active power penetration compared to using HVVC.

5.

Case 5

With the proposed volt/var control strategy, PV systems on phases B and C are operating at a capped active power while drawing reactive power whereas PV systems on phase A are injecting full PV generation. The maximum voltage in the LV network is reduced to 1.09 p.u. as illustrated in Figure 9. In addition, due to the voltage drop during peak load, PV systems on phases A and C start injecting reactive power, giving a required voltage boost to local buses. Accordingly, the minimum recorded voltage in LV network is increased to 0.946 p.u. as seen in Figure 9.

Figure 10 shows the voltage profile on phase A at Feeder 1. As seen in Figure 10, the voltage drop increases with the increasing load demand. Due to this, the capacitor at four of the six distribution feeders switch in to respective feeders at the 211th interval to replace a tap change. The capacitors in the remaining two feeders do not switch in as their voltage levels are still within the acceptable range. However, due to further voltage drop, the OLTC operates to increase the voltage level across the entire network at the 226th interval. The minimum and maximum recorded voltages in the MV network are 0.948 p.u. and 1.042 p.u., respectively.

Using the proposed volt/var control strategy, the number of tap change operation performed by the OLTC at the ZS is reduced to one. The difference is due to the lack of stepping down operation once the load demand decreases after the 270th interval. While the voltage profile in the MV network shows negligible differences from that obtained using HVVC, the voltage profile in the LV network improves significantly with the proposed volt/var control. This is mainly due to the non-unity power factor operation of PV systems during periods of increasing PV penetration and load.

Table 3. Summary of results.

(a) Before using the proposed volt/var control method
Criteria		Case
		1	2	3	4	5
Minimum voltage (p.u.)	MV	1.016	0.955	0.951	0.958	0.948
	LV	1.061	1.037	0.945	0.997	0.921
Maximum voltage (p.u.)	MV	1.027	1.042	1.042	1.028	1.04
	LV	1.086	1.093	1.079	1.081	1.23
Maximum PV penetration (200%)		80	80	200	112	192
Number of tap change operation		0	2	2	0	2
(b) After using the proposed volt/var control method
Criteria		Case
		1	2	3	4	5
Minimum voltage (p.u.)	MV	1.000	0.955	0.954	0.948	0.948
	LV	1.045	1.037	0.945	1.011	0.946
Maximum voltage (p.u.)	MV	1.025	1.042	1.042	1.042	1.042
	LV	1.091	1.094	1.089	1.093	1.090
Maximum PV penetration (200%)		200	176	200	200	200
Number of tap change operation		1	2	1	1	1

summarizes the main results from all five cases. In general, the implementation of the proposed strategy on all representative networks has improved the voltage profiles in the combined network, particularly in LV networks. Interestingly, the calculated $S_{\mathrm{device}}$ for PV systems yields power factors that do not violate 0.95 lagging during high PV penetration. Nevertheless, PV systems keep drawing reactive power while operating at a capped injection instead of curtailing active power at a unity power factor. This proves to be beneficial as the PV systems are allowed to inject more active power when compared to using HVVC. In addition, in contrast to the droop curve method employed by HVVC, the proposed volt/var control strategy calculates the adequate amount of active power and reactive power injection required to improve the voltage level at affected buses. Hence, the collective PV system operation in the LV network can efficiently achieve said objective.

One thing to note, during the simulation, the maximum allowable PV system injection is not recalculated as no significant voltage drop is observed. Therefore, the total allowable injection for associated PV systems remains the same after it is first established during the simulation period.

One of the main benefits of the proposed volt/var control strategy is the increased voltage level during peak load period when PV systems are not generating active power. This is demonstrated in Case 5 (as illustrated by Figure 9) where the voltage drop experienced at bus LV4 is the highest since it is geographically and electrically farthest from the ZS. In [27], the HVVC cannot regulate the voltage during this period and has to rely on the manual operation of off-load tap changer at the DS to improve the voltage level, as usually practiced by DNSPs. However, due to the non-unity power factor operation of PV systems at all times, this limitation can be overcome. This is not only beneficial to the network by helping to improve the voltage profile, but this approach can also relieve the network capacity by providing the necessary reactive power. If reactive power support is fully and efficiently utilized during peak load, the off-load tap changer at the DS can be set to produce a lower voltage. This in turn will allow more active power to be imported by PV systems during the day.

Compared to using HVVC, the number of tap change operations is reduced for Cases 3 and 5, increased for Cases 1 and 4, and remains the same for Case 2. The implementation of the proposed volt/var control has definitely reduced the number of tap change operations in comparison to using a traditional volt/var control method; however, it has not certainly reduced the tap change operations when compared to that using HVVC.

5.2. VUF Reduction

Table 4. Maximum VUF before using the proposed volt/var control method.

Criteria	Maximum VUF
Case 1	0.68
Case 2	0.89
Case 3	1.37
Case 4	1.39
Case 5	2.58

shows the maximum VUF level in the LV network before the proposed volt/var control for voltage regulation and voltage unbalance reduction is applied on the five different cases. As shown in Table 4, voltage unbalance is not a concern in cases 1 until 4, however, the VUF level for case 5, in which the LV network is the farthest from the ZS, violates the maximum allowable VUF level defined in AS 61000.2.2.

Figure 11 illustrates the VUF level at LV buses in the LV network before the implementation of the newly proposed volt/var control method for case 5. As shown in Figure 11, the unbalanced distribution of loads and PV systems in the LV network causes the voltage unbalance level to propagate along the growing distance of the LV feeder. From Figure 11, instances of VUF level violation happen towards the end of the LV feeder at the 105th and 221st intervals.

Using the proposed volt/var control method, the voltage unbalance reduction scheme will be activated every time the measured VUF is greater than $VU{F}_1$ or $VU{F}_2$. Figure 12 shows the VUF level in the LV network using the proposed volt/var control method. As seen in Figure 12, the VUF level at the end of the LV feeder raises with increasing PV penetration, and eventually at the 85th interval, the VUF level becomes greater than $VU{F}_1$. Since there are no PV systems directly connected to the three-phase LV bus, the PV system(s) connected through the service mains from bus LV4 will operate to reduce the VUF level. In this instance, the PV system at phase B must operate (active power curtailment and lagging power factor operation) to reduce the voltage level and subsequently, the VUF level. As the VUF level increases, the PV system at phase B, which is connected to bus LV3 also operates at the 91st interval. When the VUF level starts to become higher than $VU{F}_2$ at the 91st interval, PV systems on all phases operate to bring the voltage level closer to $V_f$. This time, the PV systems on phases B and C curtail their active power injection while operating at a lagging power factor while PV systems on phase A operate at a leading power factor.

When the PV penetration starts causing voltage rise, the voltage regulation scheme of the proposed volt/var control method takes over, causing PV systems on phases B and C to operate at capped active power injections. This subsequently lowers the VUF level and at the 126th interval, the voltage unbalance reduction is deactivated.

As shown in Figure 8, starting around the 200th interval, the PV system generation decreases while the load demand increases. The reduced PV generation brings the three-phase voltage level to a ‘healthy’ zone, rendering voltage regulation scheme of the proposed volt/var control method inactive. However, the VUF level in the LV network starts to increase again as shown in Figure 12 and ultimately at the 215th interval, the voltage unbalance reduction scheme is activated. This time, since the VUF level is greater than $VU{F}_2$, PV systems installed at all three phases operate to reduce the VUF level. When the VUF level is under control again at the 226th interval, the voltage unbalance reduction scheme is deactivated.

To show the operation of the voltage unbalance reduction scheme, Figure 13 shows the three-phase voltage at bus LV4, which experienced the worst voltage unbalance, before and after the implementation of the proposed volt/var control method. From Figure 13, the change in voltage magnitudes on phases A and C is not significant, however, a slight decrease of voltage magnitude on phase B is evident. As shown in Figure 13, the implementation of the proposed volt/var control method causes a slight decrease in voltage magnitude on phase B at the 85th and 216th intervals, subsequently reducing the VUF level below the maximum allowable limit.

Table 5. Maximum VUF after using the proposed volt/var control method.

Criteria	Maximum VUF
Case 1	0.68
Case 2	0.89
Case 3	1.37
Case 4	1.39
Case 5	1.96

shows the maximum VUF level in the LV network after the implementation of the proposed volt/var control method for all five cases. Since the maximum recorded VUF levels for cases 1 to 4 are well below pre-set $VU{F}_1$ and $VU{F}_2$, the voltage unbalance reduction scheme is not activated during these four cases. The maximum VUF level for case 5 is reduced to 1.96% using the proposed volt/var control, which shows a significant improvement.

The proposed volt/var control method demonstrates the veracity in carrying out voltage regulation and voltage unbalance reduction in a combined MV-LV distribution network at the same time. The proposed volt/var control method efficiently utilizes volt/var resources available in the network to ensure minimum number of tap change operations at the ZS and maximum PV penetration. As this proposed volt/var control methods foregoes complicated optimization strategies, it can be tailored to suit different networks and DNSPs with minimal efforts. This pragmatic approach makes the proposed volt/var control method universal and versatile.

6. Conclusions

This paper presented the analytical formulation for a compensation-based volt/var control strategy, which is a continuation from the analytical formulation presented by the authors in [15]. The formulation is used to derive a practical integrated volt/var control method that is applicable on combined MV-LV distribution networks. The proposed volt/var control method simultaneously regulates the voltage level across the whole distribution network while keeping the voltage unbalance level below the statutory limit. In this paper, PV systems are considered to be operational at different power factor conditions. The pragmatic approach of the proposed volt/var control strategy makes it easily usable and customizable on different types of networks. The proposed volt/var control strategy is tested on three different realistic Australian distribution networks and results obtained using the proposed volt/var control strategy are compared to those using HVVC proposed by the authors in [27]. Simulation results obtained show that the proposed volt/var control method can effectively manage both steady state voltage and voltage unbalance levels whilst reducing the number of tap change operations and maximizing PV penetration.

The addition of voltage unbalance reduction scheme causes PV systems connected further from the ZS to operate differently (active power curtailment and/or leading/lagging power factor operation) compared to the PV systems connected upstream in order to reduce the VUF level at neighboring three-phase buses. From a PV owner’s perspective, this might not be appealing as active power curtailment equates to a reduced revenue whereas leading/lagging power factor operation does not bring any revenues. A strategic market structure is necessary in order to attract the interests of all stakeholders in distribution networks.