Distribution Grid Restoration by Forming Resiliency-oriented Less- Vulnerable Microgrids

In this paper a new method is introduced to increase the resiliency of a distribution grid by forming robust microgrids after extreme contingencies following disasters. The method tries to serve as much load as possible by forming microgrids with alleviated vulnerability based on the Manchester model and optimal AC power flow. The main feature of the proposed method is that the formed microgrids are less vulnerable to cascading failures. Whenever some lines are heavily loaded in an autonomous microgrid, they are likely to be tripped, which in turn could lead to the increase of power flow in the other lines and further overloads could occur until the system would collapse or lose a significant amount of load. The proposed method seeks for lines with the loadings more than a specified threshold and checks for the consequences of tripping such lines before formation of microgrids. Therefore, the risk of cascading failures is minimized in the formed microgrids. The resiliency is improved by alleviating the overloads that might happen upon disconnection of a line due to the faults. Reconfiguration is well exploited to create optimum topologies with the minimal risk of cascading failures. Reconfiguration methods in a microgrid require simultaneous application of continuous and discrete variables which makes solving the problem difficult. Genetic Algorithm is used to find the optimum microgrid’s configuration with less vulnerable lines. The effectiveness of the proposed method is investigated by using a 4-feeder 1069-node unbalanced test system. INTRODUCTION Although reliability, affordability, flexibility, and efficiency of power delivery for end users in electric distribution systems is well investigated, the severe climate events is not considered in depth. Resiliency is considered as the ability to maintain service to critical load following an extreme event with low-probability and high-impact in a distribution system. Resiliency is improved by utilizing microgrids, applying distribution automation and vulnerability analysis as discussed in [1]. In this paper resiliency of distribution system in confronting extreme events is investigated from vulnerability point of view. The concept of vulnerability in a distribution system infrastructure could be seen as a global system property focuses on three elements [2]: – The severity of loss and damages caused by the impact of extreme event considered as a technical dimension – The severity of the system elements exposure to the extreme event, i.e., the probability that an element is exposed to the risk of suffering loss and damages – Degree of resilience, i.e., the ability of the distribution system to forecast, deal with, endeavor and recover from the impact of an extreme event from technical/social point of view. The vulnerability of distribution systems can be reduced by maintaining the access of power sources to critical loads in extreme events. Microgrids with fixed-generation and intermittent supply by renewable energy resources can be implemented to pick up critical loads. Reconfiguration of the distribution feeders in extreme events could be well exploited to allow the load to be served by a number of electrical islands in order to limit the extent disturbance propagation. In this paper a new method is introduced to increase the resiliency of a distribution grid by forming robust microgrids after extreme contingencies following disasters. The method tries to serve as much load as possible by forming microgrids with alleviated vulnerability based on the Manchester model and optimal AC power flow. The main feature of the proposed method is that after forming microgrids upon disconnection from the upstream grid, different microgrids are formed with the less vulnerability by mitigating the cascading failure possibility. The heavily loaded lines in an autonomous microgrid are likely to be tripped, which could in turn lead to the power flow increase of other lines, this could increase the risk of cascading failures and vulnerability of the system. The proposed method analyses the system for lines with the loadings more than a specified threshold and checks the consequences of the outages of such lines before formation of microgrids. In other words, the microgrids are formed by considering the vulnerability and risk of cascading failures. Hence, the formed microgrids are immune to breakdown caused by severe overloads of the lines that may happen due to the outage of heavily-loaded lines. In other words, the formed microgrids are less vulnerable by alleviating the overloads that might happen upon disconnection of a line due to faults. Reconfiguration is used to create optimum topologies that could mitigate the risk of cascading failures. The usage of the reconfiguration methods in a microgrid requires simultaneous application of continuous and discrete variables which makes solving the problem difficult. The taxonomy R3-12.47-2 which is a 4-feeder 1069-node test system with DGs is used as the sample system to demonstrate the feasibility of the proposed method. PROBLEM FORMULATION It is preferred to alleviate the overloading of the lines in the formed microgrids after extreme events. As fixed and intermittent supplies are considered in the formed microgrids, it is required to consider different time periods and update the supply level accordingly. Next, a random outage is simulated to investigate its consequence on the power flow of the remaining system. The new operating point is determined by solving the optimal power flow (OPF). If the OPF is converged, the algorithm checks another line in the same way as mentioned. If OPF does not CIRED Workshop Helsinki 14-15 June 2016 Paper 0053Paper No 0053 Page 2 / 4 converge, then it is required to reconfigure the formed microgrid by opening/closing the available tie switches. Otherwise load shedding is required until the OPF is converged. The lines with relatively high loadings (more than a specified threshold (70% here) are considered as having the potential for the next outage. It is worth mentioning that the routes in a microgrid are usually limited, so the outages of a few lines could lead to complete system collapse. For each line with the loading more than the specified level, a random number is extracted. If the loading of a line is a value between 70% and 90% and the random number is less than 0.3, it is selected for the next outage. Also, any line with loading over 90% and associated random number less than 0.4 is selected for outage, as the lines with loading over 90% have a higher probability to be disconnected. The microgrid may be split into several autonomous islands after the candidate lines are tripped. The power balance of the formed islands should be resumed as far as possible. SAMPLE NETWORK Taxonomy feeder R3-12.47-2 is developed by the Pacific Northwest National Laboratory (PNNL) and represents a moderately populated urban area. This is composed of single family homes, light commercial loads, and a small amount of light industrial loads. Approximately 33% of the circuit-feet are overhead and 67% underground. It would be expected that this feeder is connected to adjacent feeders through normally open switches. For this reason it would be common to limit the feeder loading to 60% to ensure the ability to transfer load from other feeders, and vice versa. The majority of the load is located relatively near the substation. Climate region 3 is the non-coastal South West of the United States and is characterized by a hot and arid climate. Within climate region 3 there are 3 12.47 kV feeder types. It is 4322 kVA [3]. Table 1: Taxonomy feeder R3-12.47-2 Data Nodes 263 Voltage (kV) 12.47 Load (kW) 4300 Voltage Regulators 1 Reclosers 3 Residential Transformers 0 Commercial Transformers 57 Industrial Transformers 5 Agricultural Transformers 0 Simulations are performed on the Taxonomy “R3–12.47–2” sample distribution network, which is a prototypical unbalanced distribution feeder model for moderate urban areas. It is developed by the Pacific Northwest National Laboratory (PNNL) [3] with nominal voltage equal to 12.47 kV and load, including losses equal to 4.652 MVA. A test system with 4 feeders and 1069 nodes including microgrids is developed in [6] by combining four “R3–12.47–2” feeders, adding seven tie switches and four microgrids. It has 156 normally closed sectionalizing switches. In this paper, the 4-feeder and 1069-node test system is used by replacing the microgrids by Distributed Generators (DGs) with intermittent generation, as shown in Figure 1. The transformer capacity at each feeder is 7.5 MVA. The active and reactive power limits of the DGs are shown in Table 2. Table 2: Maximum Capacity of DGs DG # P (MW) Q (MVar) S (MVA) DG1 5.15 2.25 5.62 DG2 1.65 0.95 1.90 DG3 2.50 1.75 3.05 DG4 1.00 0.55 1.14 Table 3 shows the remarkable power flows in the sample system which is shown in Figure 1 (pre-fault condition, all tie/DG switches open and all sectionalizing switches closed). They are calculated by GridLAB-D, which is a new power distribution system simulation and analysis tool with an advanced algorithm at its core designed for determining the simultaneous state of millions of independent devices [4]. Table 3 shows the loads of the three-phase transformers in each feeder of Figure 1. The advantages of GridLAB-D over traditional finite difference-based simulators are: handling unusual situations much more accurately, handling widely disparate time scales, ranging from sub-seconds to many years; its easy integration with new modules and thirdparty systems, lack of requiring the use of reduced-order models for the aggregate behavior of consumer or electrical systems, which averts the danger of erroneous or misapplied assumptions. Table 3: Power Flows in Different Zones Switch # Zone # Power_in (VA) R3-12-47-2_switch_12 Z24 +4.42e+6+1.408e+6j R3-12-47-2_switch_14 Z22 +4.42e+6+1.407e+6j R3-12-47-2_switch_16 Z4 +4.42e+6+1.407e+6j R3-12-47-2_switch_29 Z30 +4.42e+6+1.407e+6j R3-12-47-2_switch_21 Z8 +4.42e+6+1.407e+6j R3-12-47-2_switch_35 Z18 +4.15e+6+1.326e+6j R3-12-47-2_switch_19 Z5 +4.15e+6+1.321e+6j R3-12-47-2_switch_32 Z23 +4.15e+6+1.320e+6j R3-12-47-2_switch_18 Z6 +4.15e+6+1.319e+6j R3-12-47-2_switch_24 Z7 +4.14e+6+1.307e+6j R3-12-47-2_switch_34 Z19 +3.63e+6+1.139e+6j R3-12-47-2_switch_9 Z2 +3.63e+6+1.138e+6j R3-12-47-2_switch_15 Z14 +3.57e+6+1.116e+6j R3-12-47-2_switch_2 Z16 +3.55e+6+1.097e+6j R3-12-47-2_switch_4 Z9 +3.55e+6+1.096e+6j R3-12-47-2_switch_5 Z10 +3.38e+6+1.034e+6j R3-12-47-2_switch_26 Z11 +85955.5+26009.6j R3-12-47-2_switch_7 Z26 +628543+187133j CIRED Workshop Helsinki 14-15 June 2016 Paper 0053Paper No 0053 Page 3 / 4 S F-a