A Multi-Objective Voltage Optimization Technique in Distribution Feed- ers with High Photovoltaic Penetration

For environmental reasons, there has been a consistent increase in global installation of photovoltaic (PV) systems. This is further facilitated by the declining cost of PV modules over the years [1–5]. The increase is PV installations is evident as the financial investments in PV installation globally is also on the rise. The increase in levels of PV penetration have lead to several challenges which include reverse power flow, voltage control issues, power quality issues [6–9] (voltage and current harmonics, flicker, momentaries), protection coordination problems, and possible increase in system losses. [3, 10–12]. Of these issues, voltage control is one of the most important and of concern to utility companies. Traditionally, utility companies uses the legacy devices such as reactors, voltage regulators (VR), ON/OFF load tap changers (OLTC), and switched capacitor bank for voltage control [13, 14]. With the integration of intermittent renewable energy resources such as PV systems, the need for fast and more effective voltage control becomes imperative. According to IEEE 1547-2018 [15, 16], inverter-based DERs such as PVs are allowed to participate in feeder voltage regulation. According to [11, 17, 18], the use of smart inverters (SIs) could provide a more effective and economical way of voltage control especially at high levels of penetration of inverter-based DERs. Voltage optimization can be carried out with the use of several components in a distribution network. Several devices such as voltage regulators, OLTC, switched capacitor banks, reactors, energy storages and recently smart inverters (SI). SIs can be used to make significant impact on the voltage profile of the distribution system using the power factor control mode or the Volt-VAR/Watt control mode. With all these devices simultaneously present in the distribution network, there is an obvious need for optimizing their operation and also their interaction in the system. Therefore the optimal coordination of the control operations of the smart inverter and these existing legacy devices becomes necessary. This paper presents a multi-objective voltage control an optimization algorithm which is an extension of a previous work done by the authors in [19]. In order to take the advantage of the reactive power injection by the smart inverter, the reactive power injection by the switched capacitor banks are minimized as one of the optimization objectives. The smart inverters are set to constant power factor mode with the optimal value selected *Corresponding Author: Arif Sarwat, Department of Electrical and Computer Engineering, Florida International University, Miami, FL 33174, United States, Email: asarwat@fiu.edu www.astesj.com https://dx.doi.org/10.25046/aj040648 377 T.O. Olowu et al. / Advances in Science, Technology and Engineering Systems Journal Vol. 4, No. 6, 377-385 (2019) between 0.6 − 1 to allow for reactive power injection. The VR and OLTC switching constraints are set within ±16. The other objective functions are to flatten the nodal voltage profile towards the lower threshold of 0.95pu and minimize the overall active power losses in the network. The rest of the paper is organized as follows: Section 2 presents a brief review of existing voltage control and optimization techniques; the proposed optimization algorithm and its formulation is presented in Section 3; Section 4 describes the feeder used as case study and the simulation set up; Section 5 presents the simulation results and their analyses; while Section 6 concludes the paper. 2 Voltage Control and Optimization Methods Several voltage control techniques have been proposed and reported in literature. Active distribution networks allow for bidirectional flow of active and reactive power due to the integration of distributed renewable energy generators. An evolutionary-based algorithm for LTC and shunt capacitor Scheduling was proposed by authors of [20]. The study was done to minimize the active power loss of the system while keeping the operation of the voltage control devices and the total harmonic distortion within limits. An NREL report and publication by Ding et al [21, 22] developed an iterative optimization algorithm that coordinates the use of SI, LTCs and capacitor banks for optimal voltage regulation while implementing conservative voltage reduction technique as a means of mini zing the total system losses. The impact of the proposed optimization algorithm on the system’s power quality was also analyzed. Reference [23] developed three fuzzy logic controllers that control the OLTC switching, the reactive power sharing (to enable a relaxation of the OLTC switching operation) and reduce the active power curtailment from the integrated distributed generation. Similar to [23], authors of [24] proposed a Volt-VAR optimization (VVO) algorithm that relax the voltage regulators and LTC operations as well as minimizing the amount of PV generation curtailed. The proposed VVO multistage algorithm was compared to the multi-objective VVO. Most of the previously published papers did not optimally choose the SI setting such as its power factor, minimization of reactive power injection by the shunt capacitor banks by allowing more reactive power injection by the SIs, and flattening the system’s voltage profile and implementing conservation voltage reduction by pushing the nodal values to 0.95 pu. To address these research gaps, this paper proposes a multi-objective optimization algorithm that flattens the node voltage profile, minimizes the power loss in the network as well as minimizing the reactive power injection by the capacitor banks by optimally selecting the power factors of the SIs , the tap position of the voltage regulators and LTC as and the amount of reactive power to be injected by the capacitor banks. 3 Formulation of Proposed Optimization Algorithm Objective functions The first objective OF1 of the proposed optimization algorithm is to flatten the voltage profile of the network as much as possible and also to make the nodal voltages close to the lower boundary of the ANSI C84.1 voltage range of 0.95pu. The benefit of this approach is to take the advantage of conservative voltage reduction which allows the total load on the network to be reduced by simply lowering the nodal voltages.


Introduction
For environmental reasons, there has been a consistent increase in global installation of photovoltaic (PV) systems. This is further facilitated by the declining cost of PV modules over the years [1][2][3][4][5]. The increase is PV installations is evident as the financial investments in PV installation globally is also on the rise. The increase in levels of PV penetration have lead to several challenges which include reverse power flow, voltage control issues, power quality issues [6][7][8][9] (voltage and current harmonics, flicker, momentaries), protection coordination problems, and possible increase in system losses. [3,[10][11][12]. Of these issues, voltage control is one of the most important and of concern to utility companies. Traditionally, utility companies uses the legacy devices such as reactors, voltage regulators (VR), ON/OFF load tap changers (OLTC), and switched capacitor bank for voltage control [13,14]. With the integration of intermittent renewable energy resources such as PV systems, the need for fast and more effective voltage control becomes imperative. According to IEEE 1547-2018 [15,16], inverter-based DERs such as PVs are allowed to participate in feeder voltage regulation. Ac-cording to [11,17,18], the use of smart inverters (SIs) could provide a more effective and economical way of voltage control especially at high levels of penetration of inverter-based DERs. Voltage optimization can be carried out with the use of several components in a distribution network. Several devices such as voltage regulators, OLTC, switched capacitor banks, reactors, energy storages and recently smart inverters (SI). SIs can be used to make significant impact on the voltage profile of the distribution system using the power factor control mode or the Volt-VAR/Watt control mode. With all these devices simultaneously present in the distribution network, there is an obvious need for optimizing their operation and also their interaction in the system. Therefore the optimal coordination of the control operations of the smart inverter and these existing legacy devices becomes necessary. This paper presents a multi-objective voltage control an optimization algorithm which is an extension of a previous work done by the authors in [19]. In order to take the advantage of the reactive power injection by the smart inverter, the reactive power injection by the switched capacitor banks are minimized as one of the optimization objectives. The smart inverters are set to constant power factor mode with the optimal value selected between 0.6 − 1 to allow for reactive power injection. The VR and OLTC switching constraints are set within ±16. The other objective functions are to flatten the nodal voltage profile towards the lower threshold of 0.95pu and minimize the overall active power losses in the network. The rest of the paper is organized as follows: Section 2 presents a brief review of existing voltage control and optimization techniques; the proposed optimization algorithm and its formulation is presented in Section 3; Section 4 describes the feeder used as case study and the simulation set up; Section 5 presents the simulation results and their analyses; while Section 6 concludes the paper.

Voltage Control and Optimization Methods
Several voltage control techniques have been proposed and reported in literature. Active distribution networks allow for bidirectional flow of active and reactive power due to the integration of distributed renewable energy generators. An evolutionary-based algorithm for LTC and shunt capacitor Scheduling was proposed by authors of [20]. The study was done to minimize the active power loss of the system while keeping the operation of the voltage control devices and the total harmonic distortion within limits. An NREL report and publication by Ding et al [21,22] developed an iterative optimization algorithm that coordinates the use of SI, LTCs and capacitor banks for optimal voltage regulation while implementing conservative voltage reduction technique as a means of mini zing the total system losses. The impact of the proposed optimization algorithm on the system's power quality was also analyzed. Reference [23] developed three fuzzy logic controllers that control the OLTC switching, the reactive power sharing (to enable a relaxation of the OLTC switching operation) and reduce the active power curtailment from the integrated distributed generation. Similar to [23], authors of [24] proposed a Volt-VAR optimization (VVO) algorithm that relax the voltage regulators and LTC operations as well as minimizing the amount of PV generation curtailed. The proposed VVO multistage algorithm was compared to the multi-objective VVO.
Most of the previously published papers did not optimally choose the SI setting such as its power factor, minimization of reactive power injection by the shunt capacitor banks by allowing more reactive power injection by the SIs, and flattening the system's voltage profile and implementing conservation voltage reduction by pushing the nodal values to 0.95 pu. To address these research gaps, this paper proposes a multi-objective optimization algorithm that flattens the node voltage profile, minimizes the power loss in the network as well as minimizing the reactive power injection by the capacitor banks by optimally selecting the power factors of the SIs , the tap position of the voltage regulators and LTC as and the amount of reactive power to be injected by the capacitor banks.

Formulation of Proposed Optimization Algorithm
Objective functions The first objective OF 1 of the proposed optimization algorithm is to flatten the voltage profile of the network as much as possible and also to make the nodal voltages close to the lower boundary of the ANSI C84.1 voltage range of 0.95pu. The benefit of this approach is to take the advantage of conservative voltage reduction which allows the total load on the network to be reduced by simply lowering the nodal voltages.
where V min j and V max j are the minimum and maximum allowable nodal voltages respectively as expressed in equation (4), while α is a factor which is used to make the nodal voltage V j,t close to V min j . Objective 2 (OF 2 ) is minimizes the total active power losses in the network. P total loss = P trans loss + P lines loss OF 2 = min P total loss (2) where = P trans loss is the sum of the active power loss in the transformers and P lines loss is the sum of the active power loss in the lines. The third objective (OF 3 ) of this work is to minimize the total amount of reactive power injected by the two capacitor banks connected to buses 844 and 848.
Where Q cb is the capacitor bank reactive power injection and N c is the total number of capacitor banks in the network.

Constraints Formulation
Bus voltage constrains is set to be within the ANSI C84.1 standards.
Capacitor bank VAR and switching constrains is given as equation (5). The maximum reactive power injection by these capacitors are 300kV AR and 450kV AR according to the values in the feeder parameters.
Typical VRs and OLTCs have 32 tap steps. For each tap step, the change in voltage is typically 0.625%.
To enable the SIs inject some reactive power into the network in order to allow for some voltage control by the SI, the PF values of the SI are set as variable for the optimization in other to get their optimal values. 0.6 ≤ PV PF ≤ 1 (7) www.astesj.com 378

Feeder Test Case and Simulation
In order to validate the effectiveness of the proposed multi-objective optimization technique, six PVs with specifications as shown in table 1 are integrated to an IEEE 34 node distribution feeder (as shown in figure 1). For each of the PVs, a minute resolution actual global horizontal irradiance and temperature data (as shown in figures 2 and 3 respectively) are used as input to estimate the power output of the six PVs. The PV locations were selected a little farther from the substation transformer where there is a potential for some voltage regulation issues. The voltage profile on these buses clearly show the effectiveness of the proposed optimization algorithm. The bus voltage profiles were adequately reduced without going below the ANSI C84.1 standard voltage level of 0.95pu. As mentioned earlier, this was done to implement the benefits of Conservative voltage reduction which can be effectively used to reduce the total load on the network as well as minimize the total system power loss. Also, as shown in figure 11, the un-optimized voltage profile of bus 890 was increased from less than the minimum 0.95pu to values above the minimum 0.95pu. Figure 12 show the un-optimized and optimized total system losses. The figure clearly shows the there is consistent reduction in the overall system losses after optimization. Figure 13 shows the capacitor reactive power with and without the use of the proposed optimization algorithm. The value of the kV AR injection by the capacitor banks were consistently reduced which further shows the effectiveness of the proposed optimization algorithm.

Conclusions
With the consistent rise in PV penetration on distribution feeders, effective voltage regulation and optimization using Volt-VAR control becomes very important. This paper presented a multi-objective optimization using the existing network legacy devices (switched   capacitor banks, voltage regulators and LTCs) and respective power injection from smart inverters. The proposed algorithm minimizes the voltage fluctuation (by implementing conservative voltage reduction), the total system losses, and the amount of reactive power injection by the switched capacitor banks. The Pareto optimization results showed that by selecting the objective of highest priority, the values of the voltage regulators' tap position, reactive power injection from the capacitor banks and the power factor value of the smart inverters can be effectively selected. Using one of the Pareto optimal solutions, the bus voltage profiles, system power loss and the capacitor bank reactive power injection were simulated. The results showed the effectiveness of the proposed multi-objective algorithm.