State coordinated voltage control in an active distribution network with on-load tap changers and photovoltaic systems

0 Introduction

Global renewable energy deployment has been in the forefront to reduce environmental impacts over the past two decades.The fastest growing technology in renewable power generation is photovoltaic (PV) systems compared to other renewable peers, such as large hydro and wind [1].The increased demand and non-dispatchable generation may increase the burden on the grid in the future, which may adversely affect grid performance and operation.The mismatch between the generation and demand, that is, the residual load management, is a critical aspect for determining the power system flexibility, which can be achieved by increasing interconnections, demand-side response, and storage.Flexible generation may also be used in high demand cases [2].To ensure the operating voltage is within limits, voltage control in medium- and high-voltage networks can be achieved using on-load tap changers (OLTCs) [3].They can also be used to control remote terminals using line drop compensation with appropriate settings [4].The operational principles of OLTCs, voltage regulators, and auto-transformers are similar.Usually, the tap changer is provided toward the high-voltage side as the current is low, which makes it easier to switch the current following the make before break concept In addition, the availability of a large number of turns helps achieve higher precision regulation [5].To address the voltage issue raised by increasing distributed generation in the grid, a two-stage approach for coordinating the OLTCs and SVCs for voltage regulation was proposed in reference [6].

Considering the dependency on power production, distributed energy resources (DERs) interfaced with the grid at the point of common coupling can be arranged into four categories: direct machine coupling, full power electronics, partial power electronics, and modular or distributed power electronics interfaces [7].The most efficient category is direct machine coupling, as it does not require an intermediate stage.A distributed optimization algorithm for VAR-enabled DERs was developed in reference [8].The communication complexity is reduced by linearizing the power-flow coupling of neighboring nodes.In reference [9], an algorithm for voltage control was proposed for coordinating various regulating devices with DG.This algorithm is flexible in distribution systems without the requirement of a voltage control communication infrastructure.

The control strategies adopted for grid-connected PV inverters can be broadly classified into two categories [10].Autonomous control is generally fast, as it involves controlling the gate pulses of the power electronic switches that are operated at high frequencies.Local control can be performed using V/f, P-Q, or droop controls [10,11].The droop control technique is a grid-supporting function that is among the popular control choices for distributed power converters as they overcome the load sharing problem when operating in parallel within a grid.These can be operated either separately or in combination with the other two methods [12].

To satisfactorily operate the grid, control strategies are applied at the system level by utilizing various equipment connected to the distribution network.Peer-to-peer control, also known as distributed control, possesses the properties of both autonomous and centralized control.The control actions are locally initiated to locally control the bus voltage.However, control actions are communicated through centrallevel controllers when local control cannot regulate the bus voltage.When local control is insufficient, centralized control actions take place [12-13].The distribution system voltage regulation through active power curtailment with PV inverters and solar generation forecasts is presented in reference [14].In reference [15], the authors compared the static and dynamic reactive power supply method using a solar PV system according to German guidelines.The dynamic method is more precise than the static method, as it is operational only when there is deviation from the setpoint value.

Several techniques [16-18] have been proposed to achieve voltage control using a solar PV system and OLTC for optimal grid performance.In reference [19], a case study was presented in which OLTC control is used in an LV distribution grid in the presence of a PV system that does not participate in voltage regulation.In reference [20], optimal tap control of the OLTC was proposed to increase solar PV penetration.However, the solar PV inverter control mechanism was not utilized.In reference [19,20], minimizing the number of tap changes to achieve optimal control has not been emphasized.In reference [21], a multiobjective optimization for volt-VAR control was proposed.It uses capacitor banks and static VAR compensation and does not consider a PV system.

This study focuses on the following objectives:

· Analyzing the OLTC impact in the presence of PV systems with Q-V control.

· Design and implementation of suitable coordinated hierarchical control algorithms and a grid dynamic performance investigation.

· Comparison of coordinated and uncoordinated solar PV system and OLTC control, along with a grid performance evaluation.

1 Modeling of OLTC, Grid Connected PV System, and Sensitivity Analysis

The voltage drops caused by the current flow can be analyzed using a simple single line diagram (Fig.1).Let us consider a simple feeder with impedance through which power flows from the sending end (US) to the receiving end voltage (UR) to supply the load

Fig.1 Single-line diagram of a feeder connected with a PV and a load

The complex line current can be expressed as the ratio of the receiving end apparent power and voltage [22],

where.

Initially, it was assumed that no PV contribution was made.In addition, the voltage drops along the distribution feeder can be expressed as reference [23,24],

It is assumed that shunt admittance is negligible.When a PV system is considered, the current can be expressed as,

where PPV and QPV are the real and reactive powers generated by the PV system, respectively.The reactive power sign convention associated with the converter may change as the inverter control actions take place.The current injected into the grid depends on the complex receiving end voltage and apparent power.

The sensitivity coefficients are often used to associate the static grid scenario to define the control logic using a linearized power flow.The Jacobian submatrices obtained from the Newton-Raphson method can be used to formulate the voltage sensitivity coefficients.For instance, to derive the Q-V sensitivity relationship, we set ΔP = 0, and the relation between ΔQ and ΔV is expressed as reference [25],

Where JPV, JPθ, JQθ and JPV are the Jacobian matrix elements.Fig.2 shows a schematic diagram of the OLTC transformer along with its equivalent circuit.The primary and secondary sides are denoted by the notations p and s, where V, I, Z, and n correspond to the voltage, current, impedance, and turn ratio of the transformer, respectively [22].The minimum functionality requirements in the Danish grid code for various PV systems are listed in Table 1.The classification of the distributed PV system control is shown in Fig.3.The methods listed in Table 1 involve the reactive power control to obtain the desired voltage level.The RMS voltage variation must be within ±10%.The active power is injected independently of the reactive power.The set point for Q can be changed according to the requirements [26,30].

Fig.2 Schematic representation of an OLTC with an equivalent circuit

In constant power factor operation, the reactive power is proportionally controlled to the active power [27].In the voltage control case, the control function regulates the voltage at the desired set point.The reactive power varies as a function of the voltage.An intentional dead band may be provided, which may restrict the reactive power change when the voltage is within the specified bandwidth.In this study, a Q-V droop control is used to regulate the voltage.This type of control requires local voltage control signals to proportionally vary the reactive power [12].

Table 1 Various PV system control strategies [26]

Control strategies Rating of the PV system A B C D(11-50 kW)(0.05-1.5 MW)(1.5- 25 MW) (＞25 MW)Q control Y Y Y Y Power factor Control Y Y Y Y Voltage Control N N Y Y Automatic PF Control Y Y N N

Fig.4 shows a typical PV system Q-V characteristic [28].When the voltage is within the specified limits, no reactive power support is provided by the PV system.When the voltage is less than the minimum specified band voltage, the reactive power is injected into the system based on the operating voltage and droop setting (and vice versa for overvoltage).The specifications provided in [28] are used for calculating droop setting (V3 = 1.02 and V4 = 1.08).

Fig.3 Classification of distributed PV control strategies

Fig.4 Q-V droop characteristic of the PV system [28]

2 Methodology

The rated reactive power delivered by any PV system to the grid is fixed (Qmax), and the parameters are listed in Table 2.The coordinated control algorithm is illustrated in Fig.5.The PV system reactive power availability is checked when the bus voltage is not within the limit.If the voltage of any bus is violated, the PV system reactive availability is used based on its availability.The OLTC taps are changed if the PV reactive power is not available.This process is continuously repeated during the day for 48 equal intervals (every 30 min).The droop settings in the overvoltage case can be calculated as reference [12],

Table 2 Voltage-reactive power parameters [12]

Voltagereactive power parameters Default settings Allowable setting range Min Max Vref VN 0.95VN 1.05VN V and V 1 4 Vref ±0.08 V V ref N -0.18 V V 1 = ref V V V 4 =+0.18 ref N Q and Q 1 4 44% of MVA rating 0 100% of MVA rating V and V 2 3 Vref ±0.02 0 V V V 2 =-ref N 0.03 V V 3 = ref Q and Q 2 3 44% of MVA rating 0 100% of kVA rating

3 Result and Discussions

A practical 20 kV MV network is considered from a rural area of southern Germany, which is derived by the CIGRE task force [29] for the European configuration (see Fig.6).The three-phase feeders are considered as balanced, and the network topology can be changed by connecting the feeders using switches S1, S2, and S3.

The 14-bus system consists of two feeders that are energized using an external grid connected at Bus-0.The R/X ratio is 0.1, with a short circuit capacity of 5000 MVA, which suggests that the network is very strong.Feeder-1 consists of underground cables (UC), whereas Feeder-2 comprises overhead lines (OHLs).These distribution feeders also differ in their lengths, resulting in different impedance between the nodes.The HV/MV transformers are of equal rating (25 MVA, 110/20 kV), which were initially considered without the tap change mechanism.The related data were obtained from reference [29] for modeling.The grid performance is analyzed with all the switches (S1, S2, and S3) open, in essence, a radial configuration.The power flow analysis was performed using the Newton-Raphson method in DigSilent PowerFactory [25].The voltages at various buses during the peak loading condition are shown in Fig.7.

Fig.5 Flowchart for coordination of OLTCs and Q-V droop control

Fig.6 20 kV CIGRE MV benchmark network and different zones in the feeder based on sensitivity

Because the voltage at Bus-11 is minimal without the PV system, an investigation is carried out to determine the PV system impact placed at Bus-11 and the OLTC.In addition, the Q-V sensitivity is higher at buses having large voltage drops and reduces as the bus voltage approaches unity.The daily solar generation and load variations are shown in Figs.8 and 9, respectively.The x-axis in the PV system power output is shown in 30 min intervals of a day.

Fig.7 Voltage variation at different buses

Fig.8 Power output of a PV system (in kW)

Fig.9 Daily load profile for feeder-1 loads

3.1 Operation of OLTC control

In this section, an analysis of the grid voltage behavior with OLTC control is presented.The PV system is operated at a unity power factor and does not take part in voltage control by absorbing or injecting reactive power within the system.Each tap’s voltage change magnitude is generally expressed as a percentage of the winding voltage.The specifications used are provided in reference [29].The tap can be altered by ±10% from its neutral position.The voltage must be regulated within ±10%.As the transformer belongs to the Dyn1 vector, the phase shift provided by the transformer is 30° (lag).The OLTC operation is targeted to control the remote zones/buses voltages, which are categorized in various case scenarios, as listed in Table 3.

Table 3 Investigated cases

Cases Remote locations Case I No OLTC operation Case I I OLTC operation for Zone-1 Case III OLTC operation for Zone-2 Case IV OLTC operation for Bus-8 Case V OLTC operation for Bus-3

Fig.10 represents the voltage variation at Bus-11 (the most sensitive bus) for all case scenarios.The PV system power output is almost zero from 1:00 am to 8:00 am, as the solar radiation is almost negligible during this interval.The voltage varies near 1 p.u., and the voltage at Bus-11 with or without control is the same.No tap change action was initiated during this interval in all cases.

During the daytime, when solar power generation begins, the load consumption begins to decrease.The tap position was altered in each case (except Case I) and was set to -1, as shown in Fig.11.For Case II, an additional tap variation was observed between 9:00 am and 10:00 am, where the tap position was set to -2 at 9:30 hrs.Although there is a sudden drop in PV system penetration, which may be owing to climate variations or other factors, the subsequent change results in a voltage rise.This results in an additional tap change when compared to other cases.It should be noted that the tap change initiated in Case IV is earlier than in Case III, because the Q-V sensitivity affecting Bus-8 is higher than that of Bus-6 (in Zone-1).The voltage variation at Bus-1 for all cases is shown in Fig.12.

Fig.10 Voltage magnitude at Bus-11 in all scenarios

As discussed above, the OLTC operation regulates the low voltage of remote terminals/zones.The undervoltage condition at Bus-1 results in undesirable operation.This suggests that the OLTC operation in each case requires other measures to control the system remote terminal voltage magnitude within ±10%.

Fig.11 Variation of the tap position

Fig.12 Voltage magnitude at Bus-1 for all scenarios

3.2 Q-V droop control of PV inverter

In this subsection, emphasis is placed on the utilization of PV inverter reactive power capability without OLTC control.First, the apparent power rating of the PV system inverter was determined.In this case, it is assumed to be equal to the rated power produced by the PV system, specifically, the inverter rating is 2 MVA.The maximum reactive power injection capability is taken to be 44% of the MVA nameplate rating (i.e., Qrated = 0.968 MVAR).In other words, the power factor at which the PV system can operate should lie within 0.9, leading and lagging, when operating at its rated kVA, as suggested in the grid codes [28].

However, the power factor rating is chosen to be unity, as the consumer wants to extract all the power produced from the PV panel.Thus, the available capacity must be utilized without affecting real power generation.As the voltage magnitude limit is 1.02 p.u.for the overvoltage condition, the voltage control will be initiated prior to OLTC control, which is operated at ±10%.

Fig.13 shows the variation in the voltage magnitude in Zone-1 (at Bus-11), Zone-2 (at Bus-6), Bus-3, and Bus-8.At 7:30 am, when the solar PV generation increases with the sunrise, the Zone-1 voltage magnitudes reach a value above 1.02 p.u.The PV system in Zone-1, as well as the PV system at Bus-8, begins absorbing reactive power to bring the voltage to its set point (except for the PV system at Bus-1).As the generation increases and the load simultaneously decreases, the voltage continues to rise and the reactive power absorption increases.Although the voltage profile shape is approximately the same, the magnitude differs when reactive power absorption starts.The reactive power absorptions of PV systems connected at various buses are shown in Fig.14 (except for the PV system at Bus-1).

3.3 Uncoordinated control of OLTC and Q-V control

Fig.13 Voltage magnitude (in p.u.) at Bus-11 with and without Q-V control (x-axis in time sequence)

Fig.14 Reactive power consumption of the PV system

The effect of OLTC remote control for different cases in the presence of Q-V control performed by the PV system was analyzed.The simultaneous PV system voltage control and OLTC operating at different voltage control set points (voltage ref.is the same, that is, 1 p.u.) has an adverse effect on the system.Fig.15 represents the voltage magnitude variation at Bus-11 for various cases.It can be inferred that the voltage during each operation is quite low in all cases compared to the voltage control by OLTC and Q-V control individually.The OLTC tap position was significantly affected by the presence of PV system voltage control.The variation in the tap position is shown in Fig.16.

Fig.15 Voltage magnitude at Bus-11 for each case

Fig.16 Tap position of the OLTC

The tap position change also affects the transformer secondary terminal voltage.The worst case is observed when the transformer regulates the voltage at Bus-11.A similar behavior was observed when only OLTC was operating.However, in this case, the voltage magnitude drops considerably low to a 0.8 p.u.value.Fig.17 represents the Bus-1 voltage magnitude variation.The voltage magnitude variation at all the other buses will be within the Bus-1 and Bus-11 variation (the least and most sensitive buses, respectively).This voltage tap increase resulting in an undesirable voltage change is due to the participation of solar PV systems in voltage control.In this scenario, the reactive power variation is different from that of the Q-V control.

Notably, the reactive power absorption magnitude was reduced.In addition, PV systems located in Zone-1 and Bus-3 do not take part in voltage control.It is evident from the above discussion that the OLTC control along with the Q-V droop control must be coordinated.The number of taps changes along with the maximum tap position.

Fig.17 Voltage magnitude at Bus-1 for all case scenarios

3.4 Coordinated control of OLTC and Q-V control

This subsection describes the control strategy adopted for maintaining all the Feeder-1 bus voltages within the ±10% voltage range.A distributed control technique is used, having features of both centralized and autonomous control, as discussed in Section 2.In this case, the OLTC is operated as a centralized controller, because the remote-control operation is simultaneously performed with the integrated PV system at all buses.The objective of this study is to minimize the tap change, ensuring voltage remains within the prescribed limits.

The control action is initiated by comparing the actual voltage with the dead band provided in the Q-V control (1.02 p.u.).If the voltage is within the limits, no control is initiated.As the voltage increases beyond the limit, PV systems are required to regulate the bus voltage if the inverter reactive power capacity is available.When there is a further voltage increase, which cannot be regulated by PV systems, a tap change is initiated.However, the tap position must lie within the specified range.As the OLTC regulates the bus voltage within ±10%, the actual voltage is compared with the upper voltage band.If the tap position is not available or the bus voltage is more than 1.05 p.u., the voltage check for Q-V characteristics is initiated.This closed operation allows the controllers to maintain the voltage within a permissible value.

Fig.18 shows the voltage magnitude variation of Bus-1 and Bus-11.It is evident that the voltage magnitude reaches a maximum value of 1.05 p.u.at Bus-11, while Bus-1 voltage reaches up to 0.90 p.u.Note that the voltage at all the buses is within the ±10% range, where the voltage magnitudes lie between Bus-1 and Bus-11.

Fig.18 Voltage magnitude at Bus-1 and Bus-11

The PV inverter at any bus absorbs reactive power as soon as the voltage reaches beyond 1.02 p.u.The Q-V control returns the voltage to its voltage set point value depending on the availability of reactive power capability.When the voltage rise is significant, the Q-V control alone is not sufficient to bring the voltage within the limits.This tap change was initiated by the transformer, as shown in Fig.19.It should be noted that voltage is brought within limits by changing the tap position to -1.During this interval, the PV system power production reaches a maximum value.When the real power injection was reduced, the tap position was returned to the neutral position.

Fig.20 shows the reactive power variation of all the PV systems with Q-V control enabled.Note that all the PV systems are contributing to voltage control by absorbing reactive power.However, the reactive power absorbed at each node by the PV systems is different.This is because all bus Q-V sensitivities are different.

Fig.19 OLTC tap position

Fig.20 Reactive power output of PV systems with Q-V control

4 Conclusion

With the increasing number of solar PV systems in the power grid, effective management using coordinated control is developed using available OLTC and smart inverter-based PV systems’ reactive power capability.The buses, whose sensitivities are different from those of other zone buses, are individually considered.It was observed that when the most sensitive bus was regulated, fewer tap changes were required.The operating PV system voltage bandwidth is lower than the OLTC because of the lower Q-V droop regulation.The PV system loading also increases with a reactive power absorption increase.When the OLTC and Q-V droop are operated simultaneously without coordination, the tap change requirement increases.With the proper as proposed coordinated control, the tap change is minimized, and the voltage magnitudes are also within the permissible range.Furthermore, the impact of other types of autonomous control and its performance evaluation in the presence of an OLTC can also be investigated.

Acknowledgment

This work is financially supported by a project under the scheme entitled “Developing Policies & Adaptation Strategies to Climate Change in the Baltic Sea Region” (ASTRA), Project No.ASTRA6-4 (2014-2020.4.01.16-0032).

Declaration of Competing Interest

We have no conflict of interest to declare.