Conservation Voltage Reduction Strategy for Autonomous Microgrid with Improved Voltage-CurrentDroop-Based Inverter Control Framework

,


Introduction
Te progression of electrical demand toward the smart and optimum utilization of electrical energy took a sharp turn from the conventional structure of grid network to the emerging technology of microgrid (MG). Microgrid is the competent local entity to incorporate varieties of distributed energy sources with an advanced control to propel the future electrical demand towards smart consumption of energy [1]. Te integration of distributed generation (DG) with the microgrid network demands protection schemes to check proper coordination and maintain reliability in grid feeding and grid forming mode [2]. Te control of MG is designed to improve the resiliency against the occurrence of low probability and high impact events [3]. Te key attention of MG is dependent on developing device-level controllers to infuence the reliability of the system, manage the energy optimization, and make it competent to handle the fow of power between generation and consumption to preserve the standard voltage and frequency [4]. Te important aspect of an islanded microgrid is the power electronics converters that enable green renewable sources to integrate with it and independently control the voltage and current signal generated from it. Te converter-based microgrid provides a rapid time response as compared to the classical alternatorbased grid network. Te inverter provides the option in the MG to act as a grid feeding and grid forming mode. Te inverters in grid forming mode dampens the frequency, whereas grid following mode of microgrid disturbs the frequency characteristics with the integration of DG [5]. Te grid forming MG adopts the control feature of regulating frequency and voltage within the set limits, whereas grid feeding MG is mainly dealing with the control of power as the grid network has the inherent characteristic of maintaining constant voltage and frequency [6]. Te control strategies of MG adopt two widely based schemes i.e., centralized and decentralized strategy [7]. Te centralized scheme requires communication cables to maintain the stability of power, and the decentralized mode tactfully controls the power by considering local measurements only [8]. Te decentralized control method is widely utilized to lead the research towards a robust control action and avoid the communication framework. Among various decentralized control methods, the droop control technique is used for proportional sharing of power among DGs. An improved nonlinear droop structure is adopted using the sliding mode controller-based droop method for distribution of accurate power among the DGs [9]. Te determination of droop coefcients in the droop technique is fxed and it is dependent on the rating of the DGs, which creates problem of perturbation of voltage and current parameter at the DC buses. To overcome this problem, the work in [10] designed droop control which is dependent on demand response approach in coordination with boost converter to estimate the optimal value of droop gains. Te droop technique mainly comprises of interlinking relation among parameters such as active power, frequency, reactive power, and voltage depending upon the capacity of the network. Te appropriate droop method characterizes the setting up of frequency and voltage of DGs to share proportional amount of power among DGs to maintain the generation and consumption. Numerous literature are studied and it reports several demerits of classical droop such as the inability to handle sudden load change as it fails to consider load dynamics and it takes the loading condition of the microgrid into consideration to sustain the frequency range. Hence, an advanced droop technique is developed by defning control laws to improve the primary voltage and enhances the sharing of reactive powers in grid forming inverters [11]. Te virtual oscillator control technique is recently used for improved power sharing, faster synchronization, enhances the dynamic characteristics of the system, and ensures better performance than the droop control [12]. In [13], the droop method known as cascade forward neural network technique is utilized to adopt the nonlinear model of the inverter to track reference power and demand at diferent operating characteristics. A nonlinear droop method is proposed for parallel converters in microgrid which utilizes the probability distribution function of the load current to optimize the droop characteristics [14]. Te autonomous dissemination of power is shaped through a classical decentralized droop control scheme, but power sharing gets disturbed due to dependency on network parameters and line impedance mismatch. To tackle this efect, a consensus-based approach is adopted to eliminate the line impedance mismatch by incorporating the correction term of virtual impedance produced by the PI controller [15]. An improved droop control is designed based on dual ascent algorithm to maintain an optimal allocation of power [16]. Te reported literature discussed communication-less framework which modifed the algorithm to accurately track the voltage and frequency, but the modifed droop technique still dependent on active and reactive power to computes the frequency and voltage parameter and hence, the work reported in [17,18] discussed the V-I droop control method to regulate the voltage and it utilizes the alternative approach which shifted the power sharing approach to current sharing based droop control technique. In addition to it, one of the key characteristics of the V-I droop control described here is that it sets voltage proportional to the feedback current received from the output of the flter.
Decentralized droop control features poor dynamics and therefore, the droop method based on voltage-current characteristics is being introduced to exploit the advantages of inverter's fast dynamics characteristics [18]. Te simpler structure of V-I droop control method leads to problem such as nonsynchronization of VSIs due to elimination of power loop. Hence, the work in [19] devised the synchronization method which is coordinated with the V-I droop control technique to provide better power sharing. Te voltage-current droop is seamlessly adopted to stabilize the dynamic characteristic of the microgrid, and the execution of the droop incorporates curtailing the voltage parameter of the direct and quadrature component with respect to the current component including the droop constant seen by individual DER to the downstream load. It characterizes the drooping of the output voltage of the VSI corresponding to the output current of the inverter [20]. Although various literature [18][19][20] is concentrated on the accurate distribution of power by modifying the V-I droop algorithm but almost none of them discussed the appropriate method to determine the V-I droop gain applicable to multifeeder network.
Te participation of consumers in the electricity market is essential to managing the peak demand. Te authors in [20] proposed a cooperative game algorithm to reduce the peak demand by optimal scheduling of electric vehicle and rooftop solar photovoltaics. Te conservation voltage reduction (CVR) scheme is extensively utilized by the power system operators to bring a reduction in the energy consumption without afecting the consumers. Te concept of the close loop voltage deduction method is exploited by many literature to apply the beneft of CVR. Te work in [21] presented the comprehensive study and modelling to analyse the CVR application and identifed the various verifcation methods. Te important point to consider regarding CVR is the reduction of energy consumption by intentionally curtailing the voltage parameter, while fulflling the acceptable voltage standards [22,23]. Tis paper incorporated the CVR algorithm in standalone microgrid to propel future research towards adaptation of this algorithm to conserve the power usage in the stand-alone microgrid. Moreover, it also devised a new algorithm to compute the V-I droop coefcient for multifeeder network, and the proposed scheme is verifed on 5 bus single feeder and 9 bus multifeeder network. Te contribution of the paper is listed as follows: (1) Designed a modifed computation algorithm for VI droop coefcient to make it applicable for multifeeder network (2) Validated the proposed method on multifeeder 9 bus test system (3) Te performance of CVR execution is analysed by determining the amount of power retained with the reduction of voltage.
Te article aggregates the following points in diferent sections: Section 2 discusses the determination of droop parameters for voltage-current droop through the conventional method. Section 3 explains the proposed scheme of estimating droop parameters for voltage-current droop. Section 4 briefy reviewed the architecture of the local controller to utilize it for validating the proposed scheme. Section 5 illustrates the two-test system for validating the proposed scheme. Section 6 presented the result to prove the adequacy of the proposed scheme. Section 7 shows the capability of the controller with proposed droop estimation in response to most severe 3 phase fault condition and fnally, Section 8 summarizes the conclusion with proper evaluation of the result.

Conventional Voltage-Current Estimation Method
Te droop gain determination constitutes the single feeder multi-DGs microgrid system to estimate the droop parameter shown in Figure 1. Te conventional determination of droop parameter with control scheme considers the impedance aggregation of the network for the individual DG located at a specifc bus. For instance, the impedance Z n is found by adding all the line impedance i.e., Z 1 + Z 2 + Z 3 . Te estimation of droop coefcient is dependent on the measurement of current emanating from the DG, and it assured that voltage developed across all buses to be above 0.9 p.u. Te conventional droop determines the drop in the voltage developed by lumped load at the downstream end and it is determined by measuring the inverter's current and multiplying it with the total network impedance as discussed in Figure 1. Hence, the voltage V t set by DG terminal is found by the product of feedback current I filter and droop gain Z l . Te equation is given as follows: Te calculation of droop constant Z n is smooth for single feeder multi-DG microgrid system. Te conventional droop method takes into account the impedance from one end to the downstream end to compute the resultant impedance and accordingly, it becomes the droop parameter for that DG, which determines the voltage drop across the line parameter and includes it in the voltage-current droop scheme.

Modified Voltage-Current Droop Gain Estimation
Te conventional droop gain estimation is applicable only for single feeder microgrid with multi-DGs confguration [23]. However, the conventional strategy is not competent enough to be applied to multifeeder multi-DG microgrid system and thus, the article is intended to devise z bus building algorithm, which computes the voltage-current droop coefcient for the multifeeder multi-DG network. Te droop gain calculation is solely dependent on sequentially summing the individual impedance of the bus network. Te characteristics of z bus matrix is symmetric, and the size of z bus would be similar to the number of buses or number of DGs in the multifeeder network. Figure 2 depicts the controller framework utilized for validating the estimation of the droop parameter. Te proposed strategy concentrated on fnding droop parameters for the individual DG located at specifc bus which takes into account impedance data of network. Figure 3 shows 3 bus systems with 5 impedances. In this network, there are 5 impedances and 3 buses which provide the information about the size of the z bus matrix. Te voltage-current droop calculation sums up individual impedance at a time which can be represented by a matrix as shown in Figure 4. Te modifed z bus matrix can be formed as follows: (a) Firstly, join the branch impedance from node which is specifed in Figure 3 such as branch impedance Z 1 is connected between node 1 and reference node J and then, a new impedance is added to the node as shown in Figure 3 (b) In addition, a new bus is built to the old one to accommodate the new line impedance branch (c) Again, a branch of new line impedance would be incorporated between the two-prevailing buses (d) After that, the new line impedance branch is integrated with the old bus.
Te Z bus matrix formation is determined in the following ways: Step 1. Te impedance Z 1 is connected between node point 1 and reference node point J, and this connection of impedance Z 1 initiates the current I 1 to fow which can be expressed as follows: and the impedance Z bus can be written as follows: Step 2. Furthermore, the line impedance Z 2 is placed between node 2 and reference point J which generates the following matrix: International    International Transactions on Electrical Energy Systems After the formation of Step 2, the matrix Z bus,NEW can be updated as follows: where Z bus,OLD � [Z 1 ].
Step 3. Te line impedance Z 4 is connected between node point 2 and node point 3, and the following matrix can be observed as follows: Now, the updated Z bus,NEW matrix can be written as follows: Step 4. Te connection of line impedance Z 3 between node point 1 and node point 2 forms loop 12J which caused the current i L to fow in the loop. Te following expression is written by integrating Z 3 between node point 1 and 2: Te size of generated matrix is assumed to be P × P, and the inclusion of new line impedance Z new between the node point "g" and node point "h" consisting of "P" number of nodes generates the following matrix comprising loop current i L as follows: . . . . . . . . . : : . . . . . . . . . : : Furthermore, the previous matrix can be modifed as follows: : International Transactions on Electrical Energy Systems 5 Again, the previous matrix is expressed as follows: Te term "c" describes the diference between column #g and #h, and the updated Z bus,OLD matrix can be expressed as follows. Te interconnection of impedance Z 3 caused the voltage drop and it can be expressed as follows: Te term V g and V h can be expanded as follows: Putting equations (14) and (15) into the equation (13) as follows: Furthermore, the representation of equation (16) can be expressed as follows: Te N dimensional row matrix "y" can be expressed as follows: Te current i L in the loop is determined by the solution of the equation (17), and the representation of voltage V bus can be written as follows: Furthermore, the representation of the equation (19) is as follows: Moreover, the illustration of voltage Te updated Z bus,NEW matrix is represented as follows: Te fnal formation of Z bus,NEW matrix illustrates the interconnection of impedance Z 3 between the node point #g and #h, which in reality demonstrates the node points 1 and 2 in Figure 4.
Step 5. Finally, the connection of impedance Z 5 is executed between node point 1 and node point 3 and it can be represented as follows: Te updated fnal Z bus,NEW matrix can be illustrated as follows: Te fnal Z bus5 matrix would be utilized for calculating the improved voltage and current-based droop and size of the Z bus,new matrix as shown in Figure 4 would be 3 × 3 as it constitutes three number of buses. Tere are two standard microgrid test network which is used to analyze the behavior of the estimated droop coefcient produced from the proposed method. Te V-I droop gain is found by locating the position of the DG in multifeeder network. For instance, the V-I droop gain for the DG 1 would be found by just knowing the Z 11 element of the Z bus,NEW matrix. Hence, it can be inferred that the diagonal element of the Z bus,NEW matrix represents the value of V-I droop gain for the specifed DG.

Controller Modelling
Te control framework is described in Figure 2. Te standalone MG has a high-bandwidth and a low-bandwidth controller for maintaining the steady voltage. To establish the set value provided by the voltage source inverter, each loop has two sets of proportional-integral controllers. Te control architecture uses a voltage-current droop algorithm to generate the voltage needed for DSM, which is then passed to a low-bandwidth voltage controller to provide the desired voltage at the consumer end. Wind energy, solar energy, and microturbines are all examples of distributed energy resources, and they are all connected to the power system network via a voltage source inverter. Te local controller is capable of maintaining standard frequency and voltage conditions. Te voltage and current parameters i.e., V filter,dq and I vsi,dq are tracked to their reference values i.e., V * filter,dq and I * vsi,dq , respectively, with the application of a nested feedback loop. Te voltage and current dependent droop technique is employed to maintain the reference voltage for the lower bandwidth voltage loop, and it employs a proportional-integral controller to keep the voltage at the consumer side steady.
Te power handling is efectively performed by the droop mechanism with the proper choice of droop gain. Te classical power-based droop law is valid for the MV network, and it determines the frequency and voltage parameter according to the predetermined coefcient g p and h Q, respectively, as follows: 6 International Transactions on Electrical Energy Systems (25) Te f * depicts the reference frequency, and f nom represents the actual value of frequency of the stand-alone MG, and V * d and V nom depicts the d-axis output voltage of the inverter. Moreover, P and Q represents the instantaneous active and reactive power carrying low-frequency component emanating from the low pass flter of the power controllers as follows: where P i � V out,d * I out,d + V out,q * I out,q , V out,d , I out,d , V out,q , and I out,q are the d-q parameters of the inverter. ω cut depicts frequency of the flter.

Voltage Controller.
Te dynamics of the outer voltage control loop is slow which facilitate steady voltage at the output side as represented in Figure 5. Assuming, the voltage provided by the LC flter is V 0 , and current i load produced by the dynamics of the flter. Te voltage source converter sets the current i vsi as follows: Te equation (6) represents the current in abc frame and it needs to be converted into d-q frame as follows: Equation (7) can be represented by real and imaginary expression as follows: Te linear design of the plant can be expressed as follows: where where inverter's d-q current is represented by i vsi,d and i vsi,q , respectively, the feedback current i load,d i load,q is given to outer controller and the voltage signals V 0d and V 0q are the steady voltage developed at the load and this voltage would be regulated to execute the function of demand side management.
Te controller of the plant is tracking the reference V * od and V * oq , respectively. Te controller processes the output V od and V oq and fnally, it produces the reference current i * od and i * oq to the inner controller. Te expression for voltage loop can be written as follows:

Current Controller.
Te dynamics of the current control loop shown in Figure 6 is fast compared to the voltage control loop which sets the voltage as maintain by the voltage-current droop at the load end. Assuming that the current i vsi is delivered by the inverter, and the term V filter and V vsi represents the flter and inverter voltage as follows: Te d-axis voltage forms the output voltage generated by the flter as follows: International Transactions on Electrical Energy Systems where the frequency of stand-alone microgrid is represented by ω, and the voltage V q,filter is zero as voltage is directed along the d-axis. Furthermore, equations (17) and (18) depicts the real and imaginary terms which drives the inner controller as follows: where the d-q current and voltage of inverter is depicted by i vsi,d and i vsi,q and V d,vsi and V q,vsi , respectively. Te term V d,filter and V q,filter are the d-q voltage developed at the LC flter.

DSM Algorithm Integrated with DER.
Te consumer's power is curtailed by engaging the strategy of voltage reduction to facilitate the DSM as illustrated in Figure 7. Te power and voltage relationship for real bus system voltage V is depicted as follows:    (38) P demand and Q demand signify the power consumed at voltage V 0 . Z and y represent load exponents which are calculated by analysing the type of load connected to the system. Te voltage regulation to 0.9 p.u. applied to equations (37) and (38) conserves almost 19% of active and reactive power for Z � y � 2. It characterizes the accomplishment of the total kVA deduction of 19%. Hence, the research prospect is very high to retain power in stand-alone microgrid.

System Description
Te proposed scheme is implemented on 5 bus network comprising 3 DGs and 5 impedance in the network. Te case study is used to validate the approach for determining the modifed droop algorithm, which is then applied to a singlefeeder and multifeeder microgrid test system. Te building of z bus matrix discussed in Section 3 implemented through MATLAB code to compute the z bus matrix. Firstly, the implementation of control algorithm with DSM strategy on single-feedermulti-DGs5-bus system is depicted, and results are being analysed. Furthermore, the implementation of local controller with demand side management capability is applied to modifed IEEE multifeeder network to corroborate the determination of the proposed modifed droop scheme. Te suggested droop strategy's results are compared to the traditional voltage and current-based droop and power-based droop using bus data from Table 1 to assess the performance. To justify the adequacy of the control action using the proposed approach of estimating the droop parameter, several loading conditions are studied.

Results and Discussion
To show the efectiveness of the proposed droop computation technique, the performance of the control strategy is validated using IEEE 5 bus [23] and IEEE 9 bus microgrid test systems. Both the microgrid test system and the controller are comparable, as illustrated in Figure 2, and are designed to provide demand side management with the voltage regulation and voltage-current droop support. Te voltage-current droop control includes a droop coefcient, which makes power sharing between DGs easier. Te droop coefcient given by the proposed droop computation scheme is applied to a conventional microgrid test system to validate the functionality of the voltage-current droop scheme in this section. Case 1. Performance assessment for executing the DSM strategy with cascade control on 5 bus system.
Te efectiveness of stand-alone microgrid consisting of a local controller with the potential to adopt the DSM capability is examined through two test systems. First, the controller is implemented on a single-feedermulti-DG test system shown in Figure 8, and results are compared with the conventional method for estimating the voltage-current droop coefcient and power-based droop scheme to demonstrate the efectiveness of the proposed scheme.
Te 5-bus test system comprises 3 DGs and 3 constant impedance loads. Te droop parameter for modifed voltagecurrent droop is computed by taking its diagonal elements. For instance, the droop constant for DG1 in 5 bus test system would be Z 11 element, and in a similar way, droop constant for DG positioned at bus no. 3 would be Z 33 . Te formulation of Z bus5 matrix is described in Section 3. Te voltagecurrent droop coefcient values for DGs located at buses 1, 3, and 5 are shown in appendix-I. Te DSM action is being propagated with the support of the outer and inner loop controllers to establish the desired range of voltage. Te voltage regulation is established through a voltage-current droop scheme, and furthermore, a similar voltage appears at the load end with the support of cascaded control action in the local controller. Moreover, the voltage deduction achieves a consumption of power, and it is interesting to note in case of power based droop technique that the power parameter is linked with the voltage or frequency parameter. For instance, the active power is interrelated with frequency for the MV network. Hence, dependency of one parameter over the other disturbs the nominal parameter such as setting up of nominal voltage with the adoption of QV droop. Te classical droop scheme encounters a reduction in the value of voltage in case of step variation of the load which is very common in practical conditions. So, the voltagecurrent droop algorithm acts as an independent control framework, and it has no dependency on other parameters. Te line parameters need to be accounted in the voltagecurrent droop strategy and hence, an appropriate method of determining the droop coefcient is formulated in the paper. No other parameter afects the operation of voltage-current droop; hence, it is exploited to facilitate the mechanism of demand side management. An interesting fact about the voltage-current droop is the voltage is rising with an increase of the load. Te voltage regulation within the specifed standards performed by the local controller does not show plunging of voltage due to consideration of network parameters in the voltage-current droop and nondependency on other parameters.
Te load variation of 2.15 kVA operating at 0.9 power factor is repeated at time intervals of 0.5, 1, and 1.5 seconds. Te local controller sets the voltage according to the specifcation of the droop scheme. Te constant impedance load 1, 2, and 3 is connected in parallel with other loads at bus 2, 3, and 5. Te rms voltage at the bus 1, 2, 3, 4, and 5 when using modifed voltage-current droop considering proposed droop scheme and conventional method of voltage-current droop scheme as discussed in Section 2, and P-f/QV droop scheme is shown in Figure 1. As shown in Figures 9-13, the proposed strategy for estimating droop parameter and the traditional method for obtaining droop constant for voltagecurrent droop have identical steady state performance. Note that the proposed formulation of droop constant for voltage-International Transactions on Electrical Energy Systems current droop with DSM capability is intended to be applied to a multifeeder multi-DG system so that its beneft can be exploited in practical condition. Te voltage for 5 bus test system is demonstrated in Figure 9 and devised strategy to determine droop parameter always guarantees overestimate of the actual voltage. As seen in Figure 10, lowering the voltage leads in a reduction in power consumption. Furthermore, in the case of the P-f/QV droop scheme, the buses of the microgrid encounter higher voltage values than in the modifed and conventional voltage-current droop estimation schemes. Te power consumption gets increased with the rise in voltage as presented in equations (37) and (38). As seen in equations (37) and (38), increased voltage tends to increase power usage. With step increases of the constant impedance load at 0.5 s, 1 s, and 1.5 s, the voltage propagated by the typical P-f/QV droop sees a decrease in the value of voltage.
With the increase of loads, the MV network's close coordination of the reactive power and voltage results in an increase in an reactive power demand. In the case of modifed voltage-current droop, as shown in Figure 11, the voltage of the buses is regulated in close proximity to 0.9 p.u, resulting in a reduction in active power of the load, which directly infuences the frequency of the stand-alone microgrid to improve. In compared to the proposed voltage-current droop estimation technique, the fndings reported with the power based droop reveal a higher variance in the frequency parameter. A similar discovery is presented in Figure 12 as QV droop sufers voltage deduction proportional to the reactive power. Moreover, the voltage-current droop develops voltage corresponding to the voltage source inverter's feedback current. As a result, voltage rise is noticed in any load variable situation. In all load confgurations, the voltage related to conventional P-f/ QV droop is always higher than the proposed voltagecurrent droop computation technique method.
Te proposed droop computation method incorporated in 5 bus test system provides less active power as demonstrated in Figure 10. Te active power sharing appears in proportion to the predefned droop constant, which is true for both the modifed voltage-current and the classic P-f/QV droop schemes. However, in the QV droop scheme, the active power is distributed as per specifed droop constant, whereas in the modifed voltagecurrent droop scheme, the reactive power depends on the voltage-current droop parameter Z n and the voltage source converter's feedback current I f . Te voltage-current droop scheme is defned as a self-sufcient control action that operates independently, and hence, Figure 12 shows the disparity in the sharing of reactive power among DGs but ultimately, the preservation of power is maintained throughout as encountered in Figure 13. Te power demand shows 10% drop for 50% loading condition. It is interesting to note that the voltage deduction method brings down the consumption of power by   10 International Transactions on Electrical Energy Systems almost 15%-18% with the proposed scheme of modifed voltage-current droop, whereas classical droop shows reduction with the proposed application as compared to P-f/QV droop scheme. Tere is a 4% reduction in the power consumption at a low loading condition. Furthermore, at heavy loading conditions, the improved voltage-current droop retains power close to 13%-16%. Compared to the power-based droop technique, the proposed mechanism shows a larger reduction in the demand for all loading conditions. Te proposed modifed approach to compute the droop keeps the real power consumption between 84 and 85 percent, but the P-f/QV droop keeps it between 86 and 97 percent. Te power saving scenario in 5 bus test system is demonstrated in Table2. Compared to the traditional power based droop method, the variation in active power demand corresponding to a change in the voltage parameter at the buses as indicated by equations (37) and (38) concludes that modifed voltage-current droop considerably improves the voltage profle.
Case II. To validate the efcacy of proposed voltage-current droop algorithm on IEEE 9 bus mesh system. Te standard IEEE 9 bus system shown in Figure 14 is modifed to validate the proposed voltage-current droop technique. Tere are 4 DGs and 6 constant impedance loads which make up the 9 bus test system. Te presented results are compared with classical power based droop control to demonstrate the preservation of power in stand-alone microgrid. Te determination of droop parameter with line data are listed in Table 3 for individual DG is computed International Transactions on Electrical Energy Systems as discussed in Section 3. Te voltage-current droop coefcient values for DGs located at buses 1, 2, 3, and 5 are shown in appendix-I. In both cases, creating the Z bus matrix to estimate the droop coefcient and using the usual method of computing the voltage-current droop technique, the 5 bus test system presented in the preceding section provides similar results. Hence, it validates the procedure to formulate the voltage-current droop parameter. Furthermore, the utilization of the estimated droop coefcient with the support of Z bus matrix is applied to modifed IEEE 9 bus multifeeder multi-DG test system to bolster the application of the proposed scheme. Similar results are presented for the IEEE 9 bus multifeeder multi-DG system to apply the formulated droop procedure for voltage-current droop as observed from Figures 15−20. Te difculty in estimating the DGs' droop coefcient is one of the issues associated with traditional voltage-current droop. Te proposed scheme for calculating DG droop simplifes the process and can be applied to any multifeeder network with multiple DGs.
Te reduction in the voltage is efcaciously achieved with the application of proposed droop computation strategy, and Figure 15 (38). Te voltage deduction directly curtails the power consumption, and hence, deduction in power demand tends to improve the frequency. Compared to powerbased droop method, the modifed droop technique keeps the frequency parameter closer to the nominal frequency as depicted in Figure 17.
Te conventional droop assigns the QV droop to the reactive power, while the proposed droop technique to compute the droop coefcient is dedicated for retaining the reactive power in the improved droop computation technique. Due to the limitations of conventional droop estimate for voltage-current droop, as stated in the preceding section, there is a discrepancy in the sharing of reactive power among DGs, as illustrated in Figure 13. Te proposed droop computation technique enhances the sharing of reactive power to a great extent as exhibited in   Figure 18. Te total power consumption with the capability to achieve DSM is signifcantly reduced in the range of 15%-18% in the case of a 9 bus test system as observed in Figure 19. Table 4 shows the power saving phenomena for 9 bus test system. Te loads are run at various power factors to examine power preservation when the modifed voltage-current droop is used. Te fndings are shown in Figure 20 corresponding buses regulated by modifed voltagecurrent droop control. Terefore, this case study represents the actual consumption of power operated at various power factor conditions.
As can be examined from the result, low power factor condition report greater demand deduction and it is computed that power factor of 0.8 display power preservation of 16%-17%, whereas unity power factor demonstrates 9%-10% of power deduction at 36 kVA rating. Te reactive power demand at various power factor is depicted in Figure 20. Te following observation concludes that reactive power variation is directly correlated with the phenomena of power saving in standalone microgrid.

Evaluating the Performance of Droop Estimation Method with DSM Capability in Case of 3-Phase to Ground Fault Condition
To investigate the capability of the proposed droop estimation method, the DSM enabled control scheme is put to a contingency condition such as 3 phase faults for the time period of 100 ms near to bus 1. Note that the occurrence of fault at bus number 1 disturbs the voltage balancing capability of bus 1 and bus 2 in case of voltage-current and conventional droop as illustrated in Figure 21. It is interesting to observe that the modifed droop estimation does not disturb the voltage of bus 3 and bus 4, whereas conventional droop shows diminution in the voltage level as     shown in Figure 21. Te greater amount of power reserve is maintained by the DG of the stand-alone microgrid. Te frequency variability at the time of fault occurrence is observed in Figure 22 which characterizes maintaining the nominal frequency below 314 radians/sec. Te total power consumption with efect of 3 phase to ground fault on the DG positioned near bus 1 is illustrated in Figure 23, and the power of microgrid get to normal as fault gets cleared.

Conclusions
Te DSM enabled microgrid tends to increase the power reserve by employing voltage-current droop which efcaciously performs the CVR scheme. Te paper proposed the droop estimation technique for voltage-current droop of stand-alone microgrid which exploit the advantage of z bus matrix. Te method adopted for droop parameter estimation successfully applied to the standard 5 bus single-feedermulti-DG system and modifed IEEE 9 bus multifeeder multi-DG system, which validates the proposed strategy with the consolidation of cascaded control and voltage-current droop. Te voltage reduction scheme retains signifcant amount of power providing self-reliant voltage to outer voltage loop, and furthermore, the autonomous voltage is conveyed to the load end with the support of local controller to realize demand side management. Te voltage-current droop performs the voltage reduction and delicacy of the scheme always provides voltage greater than the regulated voltage. Te procedure to determine the droop parameter is applicable to any kind of single-feeder and multifeeder confguration test system and applicability of the proposed method to IEEE 9 bus test system validates the point. Te voltage reduction mode within the specifed standards performed the key role in accomplishing the DSM in stand-alone microgrid. Te voltage-current droop maintains lower voltage as per standard and it provides at the most 18% of the power saving in case of low power factor condition and at least 10% of preservation of power at power factor of unity.

Data Availability
Te data used to support the fndings of this research work are included in the original paper.

Conflicts of Interest
Te authors declare that they have no conficts of interest. International Transactions on Electrical Energy Systems 21