1、 Abstract The growing penetration level of photovoltaic(PV)and electric vehicles(EV)increase the operational risk of the urban power system.Besides,the unbalanced load distribution at time-spatial scale also impacts the hosting capacity of the PVs and causes severe transmission congestion.Thus this
2、paper proposed a bi-level optimization model to manage the transmission congestion and enhance the PV hosting capacity considering the reconfigurable capability of the high voltage distribution network(HVDN).In the upper level,the optimal topological structure of HVDN is calculated aiming at minimiz
3、ing the total operational cost.In the lower level,the maximum hosting capacity of PVs is achieved by a second-order cone programming model.The proposed method was verified by an urban power system in China.Numerical results demonstrated that reconfiguring the HVDN topological structure provide huge
4、benefits of enhancing PV penetration level and facilitating the integration of the EV.Index Terms congestion mitigation,double-layer optimization,electric vehicle,photovoltaic I.INTRODUCTION HE increasing urbanization level in China enables the urban power grids enjoying an unprecedented expansion.T
5、he 220kV transmission networks has been gradually extending deep into the load center and shaping the backbone structure of the urban power grids.While the growing amount of the electric vehicles(EV)and the penetration level of the photovoltaic(PV)augment the difficulty of holding the balance betwee
6、n the power supply and demand.In the operational time,the unbalanced load distribution is prone to experience transmission congestion problems.Thus this paper resorts to reconfiguring the topological structure of the high voltage distribution network(HVDN)to improve the load distribution at each 220
7、kV substation and achieving transmission congestion mitigation.Some researches had already investigated this issue.Reference 1-5 modeled HVDN by power supply units and their connection relationships.Based on the flexible This work was supported by the Science and Technology Foundation of SGCC(521104
8、170009).topological structure of the HVDN,they proposed different transmission congestion management models.However,they failed to consider the influences of the renewable energy resources(RES).The intermittence and the uncertainty of the RES are easy to cause the frequent manipulations of the HVDN
9、switches.Reference 6-7 proposed the active energy management strategies in an attempt to enhance the hosting capacity of the PVs with the assistance of the energy storage systems(ESS)control schemes.But the simple accommodation of the PVs deteriorates the congestion problems especially under the hig
10、h penetration level of the EVs.Reference 8 utilized the reconfigure technique of the medium voltage distribution network(MVDN)to enhance the consumption amount of the RES.But the topological structure of the HVDN and MVDN has some distinct differences and the reconfigure technique adopted in MVDN ar
11、e not suitable in HVDN.Thus this paper focused on the dynamic characteristics of the output of PVs and the charging load of EVs and established a bi-level transmission congestion management model based on the reconfigurability of the HVDN.In the upper level,the optimal topological structure of the H
12、VDN is determined.In the lower level,the best load curtailment amount and PV shedding strategy are obtained.The proposed method was verified in a real urban power systems in China and numerical results showed the reconfiguration of the HVDN was an effective tool for congestion management and facilit
13、ating the accommodation of the PVs.II.MODELING OF THE DYNAMIC LOAD A.EV Charging Load Model This paper took the PHEV60(EPRI)10 as an example and used the smart charging/discharging model described in 11 to establish the EV load model.Thus the charging load are described as equation(1).,.)1,0(!)(EVEV
14、nEVnnen pEV EV(1)Urban Power Grids Dynamic Control Model with Photovoltaic and Electric Vehicles Qian Xiong1,Fangfang Liu2,Lin Lv2,Youbo Liu2,Yong Li1,Chengzhi Zhu3 1.Tianfu Power Supply Company of the State Grid Sichuan Electric Power Company,Huayang Road No.11,Chengdu,China.2.Sichuan University,Yi
15、huan Road No.24,Chengdu,China.3.State Grid Zhejiang Electric Power Company,Huanglong Road No.8,Hangzhou,China.T 2018 China International Conference on Electricity Distribution Tianjin,17-19 Sep.2018CICED2018 Paper No.201804270000164 Page1/52462Where EV is the expected value of the accessing number o
16、f the EVs and nEV is the possible number of the EVs that connect to the grid.B.Output Model of PV According to reference 12 and 13,the output of the PVs is related to the illumination intensity and its mathematical descriptions is showed in formula(2).()()()PGFW t t t(2)Where the basic function PGF(
17、t)is taken as the expected output of the photovoltaic.(t)is the obstruction effect of the atmosphere on the sunlight.III.BI-LEVEL OPTIMIZATION MATHEMATICAL MODELS A.Dynamic Risk Assessment of Network Topology In order to evaluate the current operational risk of the urban transmission system,the netw
18、ork congestion risk index(NCRI)is defined as(3)and(4).2ZNG G G(3)+=1(4)Where the first term of the equation denotes the global transmission congestion degree while the second term reflects the most serious congestion condition.The last term of the equation denotes the global PV unit accommodation.G
19、and are the branch security risk index vector and the output vector of the PVs.Their mathematical descriptions are detailed in(5)and(6).(),(1)maLm xi kg w e(5)0 00()0m mmmLL LLLLL(6)Where wxi,k is the probability weight.a is a positive number.(Lm)represents the overload amount of the branch m.Lm is
20、the load ratio of branch m.L0 is the specified threshold.M denotes the total branch number.B.Upper-level Model When NCRI beyond the specified range,the reconfiguration of the HVDN will be triggered.The two-layer optimization model described in literature 14 is used in this paper.The upper level mode
21、l mainly optimizes the topology structure of HVDN,and make the network structure adaptable to supply more EVs and increase PV accommodation.The objective function is described in(7).211min Mtti i iiC F f(7)Where the first term of the function denotes the costs of the load curtailment and PV shedding
22、 amount.The second term denotes the HVDN reconfiguration expense.M represents the number of random variables and i represents the probability weight.tiF is calculated by the lower level model which is described by formula(8).The topological status constraints is detailed in reference 5.C.Lower-level
23、 Model Based on the optimal topological structure obtained by the upper level model,the lower-level model evaluates the control cost produced by the network constraints such as power flow constraints in an uncertainty scenario.Then the evaluation results are feed back to the upper-level optimization
24、 model.The optimal object is to maximize the PV hosting capacity which is showed in(8).niti gti wgtiP P F1,)(min(8)Where reflects the environmental effects of the abandoned output of PV.Pt wgi denotes the expected value of PV output power;Pt gi denotes the actual output of PV.The required constraint
25、s are showed in the next following.()2()ni Qi ii i ij ij ij ijjiP P G X G R B Y(9)()2()ni Qi ii i ij ij ij ijjiQ Q B X B R G Y(10)()mt t ts ss s sw wwsP B B(11)22max max22iVVX(12)2 2 2()(2 2 2)ij ij ij i j ijS G B X X R(13)22max ijSS(14)222i j ij ijX X R Y(15)Where(9)and(10)are power flow constraint
26、s of HVDN.(11)is the direct power flow constraints of transmission network.(12)(15)are the relaxed second-order cone constraints.D.Solving Process The particle swarm optimization algorithm(PSO)is used to find the optimal topological state and the second-order cone programming model is used to calcul
27、ate the optimal PV 2018 China International Conference on Electricity Distribution Tianjin,17-19 Sep.2018CICED2018 Paper No.201804270000164 Page2/52463curtailment amount.The flow chart of the proposed model is showed as Fig.1 StartInitialize the parameters of the PSO algorithmInput network dataIniti
28、alize the positions of the particlesEvaluate the particles using lower-level modelConverge?Update the positionsEnd Fig.1Flowchart of the proposed method IV.CASE STUDY A.Parameters Specification The topological structure of HVDN is showed in Fig.2 while its superior system 220kV transmission network
29、is detailed in Fig.3.The simulation platform is Matlab 2016a and the frequency of the computer is 2.5GHz.The parameters of the PSO algorithm is referred from literature 15.P21 P16 P15P17P18P22P20P19S6 P6P5P1 P2 P3P4S3P24P25P23P26S2S1P8P7S4P10P9P12P11P13P14S5P34 P33 P31 P32P30P27 P28 P29Unit Group G4
30、U21 U16 U15U22U17U18U20U19U25U24U23U26U34 U33U31 U32U30U29 U28 U27U6U5U4U1 U2U3U8U7U9U10U12U11U13U14Unit Group G5Unit Group G6Unit Group G1Unit Group G2Unit Group G3transformation unit220kV transformation unitCharging unit Photovoltaic unit Fig.2Topological structure of HVDN L15L01L12L64L35S1S5S6S3S
31、4S2BSL45L04L16L03L02 Fig.3 Topological structure of 220kV transmission network As showed in Fig.3,the transmission system has six substations and ten transmission lines.The HVDN has thirty-four power units which can be divided into 6 unit groups.Where G1,G3 and G6 are mainly composed by the municipa
32、l and residential electricity load.G2 and G4 are mainly composed by the commercial load.G5 is used to power the heavy industrial load.The feasible topological information of the unit group is shown in TABLE I.TABLE I TRANSFORMABLE CELL GROUP FEASIBLE TOPOLOGICAL NUMBER Unit Group Total topology numb
33、er Feasible topology number G1 256 13 G2 32 5 G3 32 5 G4 1024 21 G5 32 5 G6 1024 23 B.Numerical Result Fig.4 shows the PV units accommodation in the early peak load time.During this period,the EVs are mainly charging in the commercial areas.The output of the PVs in residential areas remain low.The t
34、ransmission congestion is very serious and the risk of the system hits nearly 62.73%as showed in Fig.5.2018 China International Conference on Electricity Distribution Tianjin,17-19 Sep.2018CICED2018 Paper No.201804270000164 Page3/52464PV UnitsPVs accommodation(100%)P1 P12 P29 P3210%20%30%50%70%60%40
35、%80%0%17.2%29.79%6.5%7.42%Fig.4 PV accommodation at early peak load time before HVDN reconfiguration.Fig.5 Network safety region before HVDN reconfiguration at early peak load time The percentage of NCRI indicates the degree of risk to the network,and when the value of NCRI exceeds 0.3,it is assumed
36、 that the urban network is likely to have a blocking risk at the moment.The NCRI in figure 5 is 0.6273 beyond the confidence range.That indicates the transmission networks have serious congestion and low penetration level of PV units.After conducting the proposed methodology,the optimal results of t
37、he network operational status are exhibited in Fig.6 and Fig.7.PV UnitsPVs accommodation(100%)P1 P12 P29 P3210%20%30%50%70%60%40%80%0%73.46%57.96%68.64%49.22%Fig.6 PV accommodation after conducting HVDN reconfiguration Fig.7 Network safety region after conducting HVDN reconfiguration Fig.6 manifests
38、 that the HVDN reconfiguration can significantly enhance the PV accommodation and enlarge the security region of the transmission systems.Through changing the connection relationships of the HVDN power units,the remnant PV output in the residential area were transferred to the seriously congested co
39、mmercial area.As showed in Fig.7,the network safety risk is greatly reduced to 5.32%.TABLE II shows the index which represents the absorption of the output of PVs before and after the HVDN reconfiguration from 8:00 to 18:00.The triggering time of the HVDN reconfiguration are 8:00,12:00,15:00,and 18:
40、00 respectively.TABLE II PVS ACCOMMODATION BEFORE AND AFTER RECONFIGURATION Time(hour)Before Reconfiguration Yes/No after reconfiguration 8 15.23%1 62.32%9 58.78%0 58.78%10 53.04%0 53.04%11 50.49%0 50.49%12 17.49%1 78.03%13 39.76%0 39.76%14 34.48%0 34.48%15 16.7%1 59.83%16 43.68%0 43.68%17 42.09%0 4
41、2.09%18 10.11%1 63.57%2018 China International Conference on Electricity Distribution Tianjin,17-19 Sep.2018CICED2018 Paper No.201804270000164 Page4/52465It can be concluded from TABLE II that after enforcing the HVDN reconstruction,the accommodation of PVs can be increased by at least 30%.Take 12:0
42、0 as an example,the load of EVs is discharged with a high probability,and the level of the PV accommodation is less than 20%.The system network congestion area is mainly concentrated in residential areas and industrial areas,and the risk of the system is nearly 51.49%.By triggering the network topol
43、ogy reconfiguration,the load of the industrial areas in heavier residential areas were transferred to the light loaded part thus mitigates the transmission congestion.Besides,the degree of PV accommodation were increased to 78%.The comparison of the network risk is showed in Fig.8.10203040506070808
44、9 10 11 12 13 14 15 16 17 18Before reconfigurationAfter reconfigurationReconfiguration timeRisk(%)Risk changes before and after reconfigurationTime(h)Fig.8 Comparison of the network risk before and after HVDN reconfiguration.As showed in Fig.8,the risk of the network can be considerably decreased by
45、 reconfiguring the HVDN topological structure.V.CONCLUSION The existing researches fail to count the impacts of RES and the probabilistic mobility of EV charging loads thus would cause frequent transfer actions.This paper combined the probabilistic power flow with the second-order cone programming m
46、odel to improve the fitness of the HVDN topological structure and enhance the computation efficiency.The proposed NCRI indicator properly reflects the system security risk and PV accommodation which provide a helpful decision basis for the operators.Besides,numerical results manifested that reconfig
47、uring HVDN topological structures has huge benefits for improving system security level and enhancing PV hosting capacity.REFERENCES 1 F.A.J.Yong,S.B.L.Junyong,T.C.L.Hongwei and F.D.Z.Xi,The bi-level programming model of load transferring strategy based on topological units of high-voltage distribut
48、ion network,2016 China International Conference on Electricity Distribution(CICED),Xian,2016,pp.1-7.2 JIN Yong,LIU Junyong,ZHANG Xi,et al.Bi-level Model Based on Function Units for Optimizing High-level Voltage Distribution NetworkJ.Advanced Engineering Sciences.2017,49(3):179-190.3 Zhang Xi,Lv Lin,
49、Jin Yong,et al.Congestion Mitigation Model and Algorithrm for Urban Power Grids Considering Reconfigurability of High-voltage Distribution Transformer Unit GroupsJ.Proceedings of the CSEE.2016,36(20):5403-5414.4 JIN Yong,LIU Youbo,LIU Junyong,et al.Operation Congestion Management Model for Urban Pow
50、er Grids Based on Load Transfer Logical ConstraintsJ.Automation of Electric Power Systems.2016,40(23):77-85.5 ZHANG Xi,LIU Youbo,LV Lin,et al.Total Supply Capability Analysis of Urban 220 kV Area Power System Considering Load Transfer Capability of HV Distribution NetworkJ.Power System Technology.20
