CrossMark

Available online at www.sciencedirect.com

ScienceDirect

Energy Procedía 105 (2017) 2472 - 2477

The 8th International Conference on Applied Energy - ICAE2016

Coordinated Dispatch of Active Power Distribution Network Containing Large-scaled Distributed Generation and Electric

Vehicle

GAO Fei *, SONG Xiaohui, Zhang Yu, Li Jianfang, Zhao Shanshan

_Electric Power Research Institute, Haidian District, Beijing 100192, China_

Abstract

Regarding the coordinated dispatch of active power distribution network containing large-scaled distributed generations (DGs) and electric vehicles (EVs), this paper proposes a coordinated dispatch method based on voltage iteration. A model for coordinated dispatch that can optimize the node charging power of EVs and the output of DGs is established. Then the network loss in the optimization objective is simplified by the static voltage which obtained in the previous optimization iteration. Meanwhile, the nonlinear restrictions between power and voltage are simplified by the sensitivity equation, also based on the static voltage. Finally, an iteration method is proposed to correct the static voltage used in the simplified model. The coordinated dispatch method proposed in this paper is verified by the IEEE33 node test system.

©2017 The Authors.PublishedbyElsevierLtd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of the scientific committee of the 8th International Conference on Applied Energy. Keywords: active power distribution network; coordinated dispatch; static voltage; sensitivity equation; iteration

1. Introduction

Renewable energy power generation and EV have incomparable advantages in alleviating energy crisis and reducing people's reliance on conventional fossil energy. However, integration of large-scaled DGs to the radial distribution network and the load flow changes will inevitably affect the operating mode in distribution network [1,2]. It is worthy to note that reasonable charging control of EVs not only shifts the peak load [3,4], but also absorbs surplus energies of DGs during peak outputs and hence increases the allowable capacity [5,6] of DGs in distribution network. Therefore, it is highly significant to study coordinated dispatch of active power distribution network containing large-scaled DGs and EVs.

* Corresponding author. Tel.: +86-010- 82812194-8819; fax: +86-010- 82812163. E-mail address: gaofei@epri.sgcc.com.cn

1876-6102 © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of the scientific committee of the 8th International Conference on Applied Energy. doi:10.1016/j.egypro.2017.03.715

At present, the coordinated dispatch research mainly focuses on the economic dispatch in regional power grid. A coordinated strategy between the wind participants and EV owners to increase their revenues and incentives is proposed in [7]. A model is established in [8] to analyse CO2 emission reduction both in the power grid and transport sector. A multi-time synergistic dispatch model for the PEVs and wind power is studied in [9], aiming at minimizing the load variance. A business model based on coordination among EV users, aggregators, electric companies and wind farms is proposed in [10]. However, the coordinated dispatch researches mostly analyse the power balance problem from the perspective of transmission network. It is worth noting that reasonable strategy of the EV charging and DG output to achieve the economic operation in the distribution network system, is important to the dispatch center for the further power generation plan.

This paper proposes an optimized dispatch method based on voltage iteration. A model for coordinated dispatch of active power distribution network is established. Then the network loss and the nonlinear restrictions are simplified to obtain a linearly constrained convex quadratic programming model. Finally, an iteration method is proposed to correct the static voltage to obtain the optimal solution. The coordinated dispatch method proposed in this paper is verified by the IEEE33 node test system.

2. Coordinated dispatch model of DGs and EVs

In this section, the optimal charging of EVs and the active power curtailment ability of DGs are utilized to establish a model for coordinated dispatch in active distribution network. This model aims to provide optimized strategy which satisfies the routine demands of EVs, absorb DGs output and reduce the operating costs.

2.1. Optimization Objective

In a distribution network containing large-scaled DGs and EVs, it is necessary to fully explore the dispatch ability and establish a model for the operational economy and safety. The control variables of the coordinated dispatch model are the charging load of nodes which provides this service for EVs and the active power of DGs at various periods within a day. The optimization objective primarily includes two parts, as shown in (1), network loss and dispatch cost of DGs. Paper [11] proves that load variance and network loss are closely related. Optimizing network loss is equivalent to optimization the load variance which can achieve the peak load shifting.

Where, AT is the time period, Nt is the number of dispatch periods, CloSS(t) and Pos(t) are the unit loss cost and total loss during period t; Ndg is the number of DGs; CtpG(i) and PipG(t) are the unit power cost and restricted power of ith DG during period t.

2.2. Equation constraints of EV charging energy

Considering the driving peak of EVs always around 8 o'clock, in this model 90% of EVs should finish the charging before 8 o'clock in the morning under the optimal charging mode

Where, P0ev(O and Pi,Ey(f) are the charging power of node i during period t before and after coordinated dispatch.

At£ P,EV (t) = AT£ Pi (t) X 90%

Meanwhile, the charging load at various nodes should satisfy the basic charging needs of EVs.

NT N ev

ATE P, (t) = Z E, {SOCj c - SOCj,) (3)

>=i j=i

Where, N,ev is the number of EVs at node i; j is the battery capacity of EVj; SOCj,c and SOQc are the state of charge when EV j leave and return.

2.3. Inequation constraints

(1) Safety. Inequation restriction here primarily refers to the constraints of node voltage and branch flow required for safe operation.

(2) DG output. The photovoltaic generation represents DG in this paper. And its control variable is set as the ratio of the practical output to the maximum output, following the constraint shown below.

(3) EV charging power. Charging power PiEv(t) of node i with period t should not exceed the sum of maximum possible charging load of EV parking at that time.

3. Optimization method based on iteration

The coordinated dispatch model as mentioned above includes nonlinear load flow and voltage constrains so that the model becomes a nonlinear optimization problem characterized by multi-period coupling. It is hard to find a solution for this type of issues. This paper proposes a method based on iteration for simplification of network loss and the nonlinear flow & voltage constrains. Then the static voltage is iteratively corrected after each optimization.

3.1. Optimization objective simplification

According to the node power balancing equation, the injection currents of each node are as follows:

Ii (0- jii i=(p (0+jQi ('))/ K (0+jU (0) (4)

Where I° (t)+jlb (t), P° (t)+jQb (t) and U (t)+jUb (t) are the injection current, injection power and node voltage respectively at node i under static coordinate system during period t.

It is worth noting that the node voltage here is the static voltage obtained by the previous optimization iteration. As the optimization converges, the error caused by the simplification is eliminated. If the loss is ignored, the corresponding line currents are acquired as follows:

i; (t)+jIb (t ) = n? (t)+jTIb (t) (5)

Where I; (t)+jIb (t) and I; (t)+/Tf (t) are the vector expression of current at node i and line l during period t. T is the associated matrix. Therefore, network loss can be expressed as follows:

pIOs (o=[ i; (O! Ri [ I; +[ I ? (' )]T Ri [ II )] (6)

Where, Ri is the diagonal matrix of line resistance.

Substituting the equation above into the coordinated dispatch model, the convex quadratic programming model is obtained, which takes the P,Ev(t) and a(t) as the control variables.

3.2. Constrains simplification

This paper proposed to use the sensitivity relations between voltage amplitude and node injection power to simplify the nonlinear constrains. The sensitivity matrix calculation method in this paper is

adopted as [12] described. Based on the deduction in [12], voltage constrains can be simplified as the linear equations of control variables:

Uk = Uk-1 + Mp AP + MQ AQ (7)

Where, U and U"n are the voltages of node i at iteration k and iteration k-1 respectively. Mp and Mq are the sensitivity matrix of node i to the active and reactive powers of all nodes respectively; AP and AQ are the differential net power of all nodes at iteration k and iteration k-1.

Finally, it is possible to use Kuhn-Tucker conditions to find the global best solution for the linearly constrained convex quadratic programming model proposed in this paper.

3.3. Correction by iteration

As the objective and constrains both need the static voltage, it is necessary to reduce the error in voltage amplitude by iterative correction. The iteration process is as follows:

Figure 1 Flowchart of Optimization Algorithm Based on Iteration

4. Case verification

4.1. Test system

In this paper, IEEE33 node test system is adopted to verify the proposed coordinated dispatch method. Three photovoltaic generation units with the same parameters is integrated at nodes 6, 12 and 28, with the single unit area 3000 m2 and photoelectric conversion efficiency 14%. In this paper, the load parameters given by [13] are taken as the peak load. In test region, the maximum irradiance is set as 1000W/m2. The upper limit of voltage is 1.05 pu and the lower limit is 0.95 pu. Meanwhile, nodes 9, 20 and 31 can provide optimal charging service. The type of EV is Beijing Motors EV200 with a battery capacity of 30.4kWh and a mean charging power of 5kW. In this case, the network loss cost and curtailment cost are both 0.5 yuan/kWh. The iteration convergence precision is set to be 0.001.

The case is set up two scenarios of light load and heavy load to verify the effectiveness of the proposed coordinated dispatch model and algorithm. The base value of load in light load scenario is set to 1/3 of the peak load, and each EV charging node provides charging service for 90 vehicles. While the base value of load in heavy load scenario is set to be the peak load, and each EV charging node provides charging service for 120 vehicles.

4.2. Light Load scenario

In the light load scenario, the DG output consumptive effect of EV is verified first. If the IEEE 33 system is only integrated with DG, the voltages in the time period of 12, 13, 15 and 16 are beyond the upper limit. So the DGs must be curtailed and the output ratios are respectively shown in table 1.

Table 1 DG output ratios when curtailment

Time Period 12 13 15 16

DG Output Ratio 0.77 0.96 0.74 0.96

Active Power of curtailment/W 51.07 7.89 47.69 5.70

When there are EV charging load in the test system, the daily load curve of each charging node is calculated by the random sampling method as [14]. Because of the consumptive effect of EV, the voltages do not exceed the limits even without the coordinated dispatch. But the stochastic charging mode can not balance the peak and valley load, leading to an uneconomic operation with larger network loss. The network loss comparison under the two charging modes is shown in table 2.

Considering the three scenarios of without EVs, stochastic charging mode and optimal charging mode with EVs, the figure 2 gives the corresponding voltage distribution. It can be seen that the optimal charging mode can effectively regulate the voltage amplitude within 6 iterations.

V v- K & w\™ Y\v,

—wsmrâïnr

. 200 I_Optim;

12 16 20 24 4

Node 20

12 16 20 24 4 8

Node 31

»_-./<"•••

8 12 16 20 24 4

Time / h

TO Ö .

V V v..

Figure 2 (a) Voltage distribution comparison in light load scenario; (b) Charging powers comparison in heavy load scenario; (c) Voltage distribution comparison in heavy load scenario

Node 9

9 1 2 1 5 1 8 21 24 27 30 33

3 6 9 12 15 18 21 24 2/ 30 33

4.3. Heavy Load scenario

In the heavy load scenario, the EVs leads to a higher load peak and makes the voltage exceed the lower limits. While the optimal charging mode can effectively shift the load peak, improve the power quality and reduce the network loss. Figure 3 shows the node charging power comparison under two charging mode. It can be seen that with the optimal charging mode, the EVs peak load has been shifted. Table 2 shows the network loss comparison under the two charging modes. Figure 4 presents the network voltage distribution under the two charging modes during the period 21. The voltage amplitude achieves convergence within 7 iterations.

Table 2 Comparison of network losses under two charging modes

Loss under stochastic charging /kWh Loss under optimal charging /kWh Loss reduction ratio Light load 207.88 173.45 16.56%

Heavy load 1849.86 1703.28 7.92%

5. Conclusions

This paper proposes an optimized dispatch method based on voltage iteration. A model for coordinated dispatch of active power distribution network is established. Then the network loss and the nonlinear restrictions are simplified to obtain a linearly constrained convex quadratic programming model. Finally, an iteration method is proposed to correct the static voltage to obtain the optimal solution. The IEEE 33 test system verified that the coordinated dispatch method proposed in this paper can effectively reflects the energy and power constrains according to the actual driving habit of EV users. And the optimal solution can shift the peak load, consume the renewable energy and reduce the network loss. Meanwhile, the iteration algorithm for voltage correction can quickly and accurately solve the simplified model.

References

[1] Eltantawy A B, Salama M M A. Management Scheme for Increasing the Connectivity of Small-Scale Renewable DG. IEEE Transactions on Sustainable Energy, 2014; 5: 1108-1115.

[2] Muttaqi K M, Le An D T, Negnevitsky M, Ledwich G. An Algebraic Approach for Determination of DG Parameters to Support Voltage Profiles in Radial Distribution Networks. IEEE Transactions on Smart Grid, 2014;5: 1351-1360.

[3] Gan L, Topcu U, Low S. Optimal decentralized protocol for electric vehicle charging. IEEE Transactions on Power Systems, 2013; 28: 940-951.

[4] Ma Z, Callaway D S, Hiskens I A. Decentralized charging control of large populations of plug-in electric vehicles. IEEE Transactions on Control Systems Technology, 2013; 21: 67-78.

[5] Antúnez C S, Franco J F, Rider M J, Romero R. A New Methodology for the Optimal Charging Coordination of Electric Vehicles Considering Vehicle-to-Grid Technology. IEEE Transactions on Sustainable Energy, 2016; 7: 596-607.

[6] Carpinelli G, Mottola F, Proto D. Optimal scheduling of a microgrid with demand response resources. IET Generation, Transmission & Distribution, 2016; 8: 1891-1899.

[7] Ghofrani M, Arabali A, Etezadi-Amoli M, Fadali M S. Smart scheduling and cost-benefit analysis of grid-enabled electric vehicles for wind power integration. IEEE Transactions on Smart Grid, 2014; 5: 2306-2313.

[8] Yu D Y, Huang H L, Lei M, et al. CO2 reduction benefit by coordinated dispatch of electric vehicle charging and wind power. Automation of electric power systems, 2012; 36: 14-18.

[9] Yu D Y, Song S G, Zhang B, et al. Synergistic dispatch of PEVs charging and wind power in Chinese regional power grids. Automation of electric power systems, 2011; 35: 24-29.

[10] Xiang D, Hu Z C, Song Y H, Ding H J. Business Model and Day-ahead Dispatch Strategy to Reduce Wind Power Curtailment Through Vehicle-to-Grid. Proceedings of the CSEE, 2015; 35: 6293-6303.

[11] Sortomme E, Hindi M M, MacPherson S D J, et al. Coordinated charging of plug-in hybrid electric vehicles to minimize distribution system losses. IEEE Trans. on Smart Grid, 2011; 2: 198-205.

[12] Brenna M, Berardinis E D, Carpini L D, et al. Automatic distributed voltage control algorithm in smart grids applications. IEEE Trans. on Smart Grid, 2013; 4: 877-885.

[13] Baran M E, Wu F F. Network reconfiguration in distribution systems for loss reduction and load balancing. IEEE Trans. on Power Delivery, 1989; 4: 1401-1407.

[14] Taylor M, J Alexander A. Evaluation of the impact of plug-in electric vehicle loading on distribution system operations. Power & Energy Society General Meeting, 2009.

Biography

GAO Fei received the Ph.D. degree in electrical engineering from Tainjin University in 2012. She is an engineer with China Electric Power Research Institute in Beijing, China. Her current research interest is operation optimization of active power distribution network.