A New Communication-Free Dual Setting Protection Coordination of Microgrid

Protection coordination for the microgrid system in both islanded and grid connected modes is considered a challenge for microgrid operation. The variation in short circuit currents in each mode of operation, limitation of protection curves in the relays, and the need for fast and reliable communication are the main challenges tackled in research. Direction overcurrent relays are considered the main protection instrument for the microgrid. However, one protection setting for both (main and backup) operation does not provide a fast-operating time. On the contrary, the dependence on communication to maintain the proper backup protection operation is not a reliable solution for protection system. Therefore, this paper introduces a fast, reliable, and simple protection scheme for secure microgrid operation in both islanded and grid connected modes. In this context, one set of protection settings is optimized and used in both modes of operation. A protection setting consists of two time-dial settings and one pickup current setting for each relay. The proposed design technique minimizes the total operating time for both main and backup relays. As all protection functions are executed in the forward direction, no communication is needed among relays. The proposed scheme is tested and verified on a modified portion of IEEE 30 bus test system equipped with distributed generation units. A comparative analysis is carried out and the results demonstrate superior performance by the proposed scheme in comparison with the dual setting scheme with communication.


I. INTRODUCTION
he microgrid concept is a potential solution for the growing demand problem [1]. The reliability factor is important for encouraging the investments in microgrid. Stable operation and robust protection system are essential for microgrid's reliability [2]. The microgrid's different modes of operation (grid connected and islanded) create a variety of fault scenarios which affects the protection system's selectivity, sensitivity and speed [3][4][5]. The Direction Over Current Relay (DOCR) is one of the main microgrid protection components [6]. Many optimization methods are introduced to find the optimal setting of the DOCR with or without fault current limiter in different constraints and modes of operation, in order to minimize the relays operating time [7][8][9][10][11][12][13]. The limited number of industrial protection characteristic curves restrains the optimization algorithms ability to find fast operating time for high fault current resulted from grid connected microgrid [14][15][16][17]. On the other hand, the majority of literature assigns a single protection setting for both main and backup operation in a relay. Consequently, it slows the relays' operating time since one optimization constraint at least from several constraints needs to be released [18]. Some studies deal with these two issues. The first issue is the limited number of available standard protection curves. The second issue is the need to release one optimization constraint. The first issue is investigated in [14][15][16][17], and a solution based on a new tripping current characteristic for DOCRs is introduced. In [15], a new Time Current-Voltage (TCV) characteristic for DOCR is introduced considering an additional term in the time current characteristic equation to represent the relationship between the fault's voltage (voltage dip) and the relay operating time. The optimization technique is used to tune the additional term to achieve proper protection coordination accordingly. Although the method reduces the operating time compared to the conventional method, the obtained operating time is long for the studied maximum fault current.
The relation between the high fault current resulting from DG integration and the fixed point of maximum Current Multiplier Setting (CMS) in the industrial relays is discussed in [14]. The relays fault current may increase beyond this point, which results in a fixed operating time for a variant high fault current. The reference proposed a non-standard characteristic curve by adding a new CMS maximum constraint into the optimization problem. The scheme succeeded in different DG location for both grid connected and islanded operation of the microgrid to maintain the coordination between relays. The results show that the maximum fault current experienced by the relays does not exceed the new proposed CMS, leading to improved relays operating time. However, this improvement is insignificant compared to the high fault current resulting from the studied three phase fault. In [16], an auxiliary variable was added to the classical operation time model of each DOCR to reduce the relays operating time. The proposed auxiliary variable shifts the protection curve vertically in order to adjust the relays operating time. After that, the auxiliary variable is added as a constraint in the optimization model, which creates a new Tripping Characteristic Curve (TCC). The protection strategy is tested and proven superior to the conventional optimization method and the user defined coordination method. Despite the obtained tripping time of primary protection relays is short in most cases, it is slow in cases of faults at the far end of a line.
In [17], a new characteristic curve is proposed with the objective of reducing the energy dissipated in the fault. The protection scheme is tested in a radial system, and it is compared with the conventional scheme. However, more investigation for applications in meshed system are required. For the previous investigated method, i.e. new TCC, it is found that employing one protection curve for both primary and backup protection is not optimal. Although it yields fast primary protection operation, the backup protection operation is very slow for the same fault. For the second issue (i.e. the need to relax one optimization constraint), this issue is tackled in [18][19][20][21][22][23]. The authors investigate a dual protection setting for DOCR, where each setting works in different direction: one for primary protection in forward direction and the second one backup protection in reverse direction. This method requires a communication infrastructure for effective backup relays operation.
In [18], a technique is adopted and verified in both islanded and grid connected modes for the microgrid. This method was an extension work for [19]. In [20], the dual setting method has been developed to cope with the radial system topology, and the transient stability for the connected DGs are considered. Ref. [21] introduced a comparison between conventional optimization coordination and a dual setting optimization in case of different relay characteristics from the IEC standard. It is concluded that the dual setting saves total relays operating time in different characteristics. However, it didn't represent the detailed relays operating time. In [22] the dual setting principle considering multiple fault location is represented. Despite the fact that the dual setting protection scheme reduces total relays operating time, this improvement decays with wide-spread implementation of this scheme. At the same time, replacing all DOCRs with dual setting relays incurs additional costs [23]. Ref. [23] proposed a techno-economic distribution for dual setting relays in meshed distribution system in order to minimize total operating time. In addition, non-standard inverse characteristic curve is optimized for that dual setting relays. This technique is considered as the first research that combines both techniques, non-standard inverse curve besides the dual settings. However, the relays operating time in primary and backup modes are relatively slow with respect to the high threephase fault current in some scenarios.
Although the relays operating time is better than that of the conventional method, the reduction in operating time, for both primary and backup protection, however this is insignificant in comparison with the increase in the three-phase fault current level. In addition, the method requires communication among relays, which introduces new risk factors such as communication link failures and cyber intrusions [24].
In this paper, a new and simple dual protection settings is introduced. Both of the dual settings work in the forward direction, and without any communication among relays. The first setting of a relay is the primary protection's TDS, which based on the very inverse curve. The second setting of the same relay is the TDS for backup protection, based on the normal inverse curve. The main contributions of the proposed protection technique are: • A novel approach to the dual setting scheme that provides faster operating time especially for primary protection. • Operating the backup relays work in forward direction, and thus, no communication link is required for proper backup coordination. This increases the protection's reliability and resilience against cyber threats. • The protection curves are taken from the industrial characteristics' curves; very inverse and inverse curve. The method does not require new relay characteristics. • Adopting the very inverse curve for the primary protection setting, which is characterized by fast tripping action. This contributes much faster operation with respect to higher fault currents brought by higher penetration of distributed generation.
• The backup setting employs the normal inverse curve, providing a reasonable backup operating time, especially for a lower fault current value. The settings are optimized and tested for a portion of the IEEE 30 bus test system. Furthermore, the results are compared with the traditional dual setting scheme presented in [18] and the improvement in the total operating time are demonstrated. In addition, the minimum coordination interval between relays is preserved, thus, demonstrating the advantages of the proposed method.

II. PROPOSED PROTECTION COORDINATION ALGORITHM
The objective function of the protection coordination design is to minimize the total relays clearing time for faults. The design problem can be modeled as linear or non-linear optimization problem [18]. Different optimization tools are commonly used to solve the protection coordination problem. The decision variables of the optimization model are the settings of each relay: The Time Dial Setting (TDS) in primary protection mode and backup protection mode, and the pickup current (Ip). Hence, the objective function is expressed as follows: Subject to, the operating time of the inverse curve: where is the short circuit current pass through relay i. The superscript i is the relay index, j is the fault location index, k is the number of backup relays, pm identifies primary protection relays, and b identifies the backup relays. Subscript c distinguishes between faults occurring while the microgrid is in grid-connected mode versus operating in islanded mode. Eq. (1) represents the optimization's objective function, which is minimizing the total operating time for all the primary relays and their backup relays. Eq. (2) is the generic expression of relay operating time for different inverse overcurrent characteristics.
is the relay operating time for relay i and fault location at j. A and B are coefficients of the tripping curve, whose values are given in Table I [24]. The very-inverse tripping curve is used in the primary protection mode in all relays, while the normal-inverse tripping curve is used for the backup protection mode. The objective function should accomplish the following constraints in both modes of operation (grid-connected and islanded).
Where CTI is the minimum coordination interval required to discriminate between primary relay operating time and backup relay operating time for the same fault location. Therefore, CTI must take a value between (0.2-0.5) sec [11]. A CTI value of 0.2s is adopted in this paper.
The pickup current setting of a relay is designed according to Eq. (4), which ensures that it is larger than the maximum load current in its zone. At the same time, the pickup current must be smaller than the minimum fault current seen by relay in backup mode.
is the maximum load current in relay i, p,i is the pickup current setting of relay i in p.u. It is important to note that there is only one pickup current value setting for both primary and backup protection modes.
are the Time Dial Setting of relay i in primary protection mode and backup protection mode, respectively. and in (5), and (6) are the upper and lower limits of a relay's TDS. These limits are the relay manufacturer settings [22], min and max values are selected to be 0.05 and 0.35.
Eq. (7) describes the novel coordination criterion of this paper. That is, the tripping time in primary protection mode for every relay should always be less than the tripping time in backup protection mode. Point (H) is the point of discontinuity in the protection curve, lying at the maximum possible fault current seen in the backup protection mode, and at the minimum possible fault current observed in primary protection mode. If the fault current measured by the relay is equal to or greater than the value H, the relay operates in primary protection mode with the very-inverse curve. Otherwise, the relay operates according to the backup setting with normal-inverse curve. Fig. 1 represents the proposed protection characteristic curve for the direction over current relay for both primary and backup operation mode. Fig. 1 The proposed protection characteristic curves.
In Fig. 1, the proposed protection characteristic curve combines the normal-inverse curve for (backup setting) and very inverse curve for (primary setting) in one operating curve. This hybrid curve works in forward direction. Point (H) defines the end of the normal inverse curve (for backup setting) and the start point of the very inverse curve (for primary setting). Employing the very inverse curve ensures very fast operating time in primary mode, while the normal-inverse curve provides relatively fast operation in backup mode with respect to CTI. Eq. (8) describes the curve for relay i. (10) where is the relay i operating time, ̃ is / , and H is the fault current value in case of minimum phase fault in the primary relay over the pickup current.
is the saturation current of the relay's current transformer.
Eq. (10) represents the effect of current transformer saturation on (9) whereas the current-transformer core applies a cap on the maximum fault current measured by the relay. Hence, a value of 20 for is adopted based on a number of industrial catalogues [14]. Better design and core material of the CT allow measuring higher currents accurately and therefore, faster tripping time and better optimization. Moreover, the upper limits for the primary relay operating time and the backup relay operating time are expressed in (9). The proposed protection coordination optimization algorithm is shown in Fig. 2.
The test system is divided into protection zones and the relays zones are assigned manually. The fault currents and load currents observed by each relay are calculated accordingly, and fed to the optimization software. For each protection relay, the software detects the size of fault current for all types of faults observed by the concerned relay. This includes double phase faults, and fault current in the backup protection zone to determine Ip, while the three phase fault is used to determine TDS primary, and TDS backup. The minimum pickup current for the relay is determined as 110% of the largest load current experienced by the relay. At the same time, the maximum pickup current for the relay is set to be 60% of the smallest fault current observed by the relay. The software verifies that the upper limit of the pickup current is higher than the lower limit: 110% × load ≤ 60% × min{ fault } If this criterion could not be satisfied, the upper limit is relaxed to 80% of the lowest fault current. The software proceeds to start the optimization algorithm using the PSO algorithm to find the optimal relays settings; Ip, TDS primary, and TDS backup according to constraints in (4), (5), and (6). The optimization problem is repeated 50 times in order to find the best possible solution, and eliminate local optima.

III. DESCRIPTION OF THE TEST SYSTEM
The proposed protection settings algorithm is tested on a distribution portion of the IEEE 30 Bus test system. The test system is modeled based on the parameters given in the appendix according to [25] generation are distributed at some feeders as shown in Fig. 3. The Distributed Generators (DGs) rating are selected so that it can supply the local load in case of islanded operation mode. The DGs operate step up transformers 6.6/33 kV. The distributed generators and the step-up transformers parameters are given in Appendix.  Fig. 3. The distribution portion of the IEEE 30 bus system [18][19], [25].
The optimal coordination is executed to determine optimal relay settings valid for both modes of operation (gridconnected and islanded). In the grid-connected mode, the test system is supplied through the substation at buses 2, 8 and 12 and the DGs. In case of islanded mode of operation, the DGs can feed all the local loads. The protection coordination algorithm includes multi subroutine, which executed in MATLAB. The first one determines the load current in case of islanded and grid-connected mode of operation. The second one determines the fault current for both primary and backup relays in islanded and grid-connected modes of operation. In the last subroutine for optimization, particle swarm optimization (PSO) algorithm is used to produce the protection settings. PSO is one of the heuristic optimization algorithms that deals with a continuous solution space. Each decision variable is represented by one coordinate in a vector. The position and speed are updated for each dimension independently; that is, without interference from any other dimensions [26]. Therefore, the authors believe that PSO is the more suitable tool to solve this problem.

IV. SIMULATION RESULTS AND ANALYSIS
The single point coordination strategy which is adopted in [18] is used here in both methods; the proposed method and the benchmark method [18]. In this section the optimized relays setting, and total operating time for faults at midpoint {F15 -F30} are discussed. The dual-setting design method introduced in [18] is applied on the modified test system shown in Fig.3 for the comparative analysis and evaluation. The settings of both  design techniques are tested in both grid-connected and islanded modes of operation as follows: A) Performance based on single point coordination strategy The same portion of the test system presented in [18] is used in this paper. In order to hold an objective and fair comparison between the two protection design techniques, the proposed optimization protection technique shown in Fig.2 is compared with the dual setting method in ref [18] for both grid-connected and islanded modes of operation with DGs connection. 1) The Dual Setting optimization method in ref [18] The dual setting optimization protection method in ref [18] involves four protection settings ( , , , ) for the DOCRs, which are valid for (grid connected and islanded) operation modes. Each relay has a TDS for primary protection which operates in the forward direction and another TDS for backup protection which operates in reverse direction. The used objective function is the same one in Eq. (1), the result operating time for the protection relays will follow Eq. (2). The used constraints in [18] are: TDSmax and TDSmin are the upper and lower limits of a relay's TDS. A TDSmin value of 0.1s is adopted here, in accordance with the design parameters of [18], and TDS max= 3.2.
The protection curve used in [18] is the normal inverse curve for both primary and backup settings. The corresponding values for parameters A and B in Eq. (2) are 0.14, and 0.02, respectively. The minimum and maximum operating times are 0.1s, and 2.5s, respectively. The optimum protection settings obtained are implemented on the selected test system and tested also with a different set of scenarios for faults occurring at the midpoint of transmission lines, as depicted by Fig. 2. Tables II and III represent the detailed results of primary and backup relays operating time in case of faults {F15-F30} in islanded mode and grid connected mode respectively. It can be seen from the results; the coordination is preserved between primary and backup relays with respect to CTI. While the minimum and maximum limits for the operating time are maintained also.
Eq. (10) is used here, for value = 20; to demonstrate the validity of the obtained results in a real-world industrial application. The primary relays operating time is almost 0.2267s. This is because the fault current in the primary relays is relatively high (If/Ip ≥ 20). As a result, the operating time cannot be less than 0.2267s with the lowest TDS = 0.1. It is difficult to improve the operating time for the primary relays beyond these values. For backup relays operating time, the smallest value theoretically possible is around 0.5287s, which corresponds to a tripping time of 0.2267 for the primary protection relay, and a CTI of 0.3s. The actual results for backup relays operating time vary between 0.5287s and 1.0191s. These values vary between islanded to grid-connected modes, as well. This is because the fault current observed by the relays vary between the two modes of operation. For example, when fault F22 occurs in islanded operation mode, the backup relay R4 operation time is 0.8612s.  This time increases to 1.0191s in case of grid-connected operation mode because the fault current observed by relay R4 is much lower in grid-connected mode. At the same time, other relays demonstrate faster tripping times in grid-connected mode due to higher fault current. For example, when fault F16 occurs in islanded mode, the backup relay R9 operating time is 0.9177. This operating time decreases to 0.6693 in grid-connected operation mode. The optimum relay setting for this method is given in Table IV.

2) Proposed dual setting optimization method
In this section, the results from the proposed protection coordination method will be represented, in addition to comparison between it and the protection method results for [18]. The proposed protection technique produces three protection settings ( , , Ip). The obtained results show that the total relays operating time from the proposed method is reduced significantly from 65.24sfor the method in [18] to 27.98s, which represents a 57% improvement. Table V presents the relays total operating time in both of the gridconnected and islanded modes along with the improvement margin over the design technique of [18]. The following figures; from Fig.4 to Fig.6 represent coordination charts between primary relay and backup relays for F15. In this case the primary relay is R5, while the backup relays are R12 and R9. For the three phase fault at midpoint, the IF/Ip for the primary relay R5= 28.96, which gives operating time =0.0355 sec, as indicated in Fig.4 for the same fault. The backup relay R12, for the same fault, IF/Ip= 12.86, which gives time= 0.2789 sec in Fig.5 while for the other backup relay R9, IF/Ip= 6.4 gives time= 0.257 sec, as represented in Fig.5. It can be concluded that, in these relays set, the coordination time interval (CTI) is maintained.  Furthermore, Table VI illustrates the detailed relays operating time in case of islanded mode. It can be observed that, the minimum coordination time between backup and primary relays is maintained in all fault locations. Moreover, the minimum coordination time interval between primary and backup relays is preserved, along with the maximum operating time constraints enforced upon primary and backup relays. The proposed protection coordination technique demonstrates a significant advantage over the results illustrated in Table II. The majority of primary relays have an operating time of 0.0355s in Table VI compared with 0.2267s in Table II. It can be seen that adopting the very inverse curve in primary relays setting improves the primary relays operating time, despite implementing the CT max constraint. Before comparing the backup protection tripping times for the proposed and the benchmark methods, the following points must be kept in mind: backup protection tripping times for the proposed and the benchmark methods, the following points must be kept in mind: The backup relays operation for the proposed protection technique is in the forward direction, while in Tables II and III the backup relays are in the reverse direction. The minimum backup relays time in Tables II and III is 0.5287s, which is the best time achieved by the benchmark method [18]. Therefore, the comparison between the two methods will consider the values above this one. It can be noticed from the results in Table VI Table II the backup  R4 took long time to operate 0.7646sec, 0.6867sec, and 0.8612 respectively in Table II. • For F25, R25 is one of backup relays in case of R23 does not operate in 0.2586 s, instead of R24 in the reverse direction with slower operating time 0.5990 s in Table II. • For further results analysis, in case of F26, if R11 fails to operate, R6 and R16 will operate in 0.3459 s, and 0.2383s, respectively instead of R9 and R10 which operated slower in 0.7266s, 0.7410s, respectively in Table II. For the same fault, if R22 fails to operate, the backup relays R4, R21, and R25 will operate in time 0.3950s, 0.2375s, and 0.2911s, instead of R23, and R24, which operate in 0.5419s, and 0.7557s in Table II. • For F27, if R24 does not operate, the backup R4, R11, and R21, will operate fast in 0.4664s, 0.3138s, and 0.2375s, respectively, instead of R22, and R23 with slower operating time 0.6889s, and 0.5897s in Table II. • For F28, R24 operate fast to isolate fault in case of R26 fails in 0.2538s, instead of R25 in reverse direction which operate slower in 0.5837s in Table II. • For F29, R29 operate fast to isolate fault in case of R27 fails in 0.2502s, instead of R26 in reverse direction which operate slower in 0.5522s in Table II. • For F30, R26 operate fast to isolate fault in case of R28 fails in 0.2402s, instead of R29 in reverse direction which operate slower in 0.5415s in Table II.
For the grid-connected operation mode, Table VII presents the detailed relays operating time for the proposed method. It can be observed that the coordination between backup and primary relays is maintained in all fault locations. The maximum operating time conditions are kept for the primary and backup relays. The primary relays operating time in Table  VII are much enhanced than in Table III. The primary relays operating times values are 0.0355s, while in Table III, it was 0.2267s. All backup relays operating times are enhanced compared to those in Table III as follows: • In case of F15, R9, and R12 are the backup relays for R5 with operating time 0.3137s, and 0.3152s, instead of R6, and R7 in reverse direction protection in Table III, which have slow operating time 0.6845s, and 0.7016s. • For F16, in case the primary relay R8 does not work, the backup relays R6, R16, and R22 work fast to isolate the fault in 0.2619s, 0.2375s, and 0.2814s, respectively. While the backup relays in reverse direction in Table III, R9, R10, and R11 operated slower to clear the fault in 0.6693s, 0.6525s, and 0.6445s, respectively. • For F17, R12 operate fast in 0.2503s, in case of R6 fails, instead of R7 in Table II which operate slower in 0.5520 s. • For F18, R22 is one of the backup relays for R10, which operates in 0.2591s, instead of waiting long time for R11 in Table III to operate in 0.6192s. • For F19, R9 is the responsible relay to clear the fault in case the main relay R7 has mal-operation in time 0.2618s, instead of waiting time 0.6153s for R6 in Table III to operate. • In case of F22, the backup relays R15, R20, and R23 for R2 will operate in time 0.4389s, 0.4946s, and 0.4205s, respectively, instead of R1, R3 and R4 in the reverse direction in Table III with very long operating times 0.8331s, 0.7786s, and 1.0191s, respectively.  For F29, R29 will operate as a backup to isolate fault at in case of R27 fails to trip in time 0.2636s compared to backup relay R26 in Table III, which operated in 0.5735s. Table VIII gives the proposed relays settings, which achieve the reduction in relays operation time and maintain the proper coordination between backup and primary relays. Table IX represents the setting of H points in all relays.

B) Performance based on three-point coordination strategy
In this section the proposed dual setting optimization method will be implemented through the three-point coordination strategy on the test system. The three-point coordination strategy is more accurate and reliable more than the single point coordination strategy [16-22-23]. The relays coordination in this strategy will be based on three points for faults on the protected lines; near end fault, midpoint fault, and far end fault. The three-point coordination strategy is also implemented for the benchmark method [18], in order to compare the results. Table X shows the new optimization relays settings for the proposed method in case of three-point coordination strategy. Table XI represents the new optimization relays settings for the benchmark method presented in ref [18] in case of three-point coordination strategy. The new settings have been tested successfully in case of near end fault, midpoint fault, and far end fault. Table XII presents a comparison for the total operating time between the proposed method, and benchmark method [18] in case of near end, midpoint, and far end faults. It is clear from Table XII that the proposed method performs better at all fault locations. In fact, the proposed method reduces the total operating time for all fault location by 57%.

V. CONCLUSION
This paper proposes a novel protection coordination method for the direction overcurrent relays settings in microgrids with distributed generation units. The proposed method utilizes a double protection setting for primary and backup operation modes in a single relay. The proposed method is suitable for applications with both islanded and grid-connected operation modes with faster operating time in both modes. The proposed dual setting contributes significant improvement to system reliability, as it eliminates the need for communication among relays. The proposed technique is tested on a modified IEEE test system and the results are discussed. The optimization is executed in case of single point coordination strategy and in case of three-point coordination strategy. The results demonstrate superior performance of the proposed method over the traditional dual setting method by reducing the total operating time of 57% for all fault locations while employing the three-point coordination strategy.