A Voltage-Based Controller for an Electric Vehicle ...

A Voltage-Based Controller for an Electric Vehicle Charger Ali T. Al-Awami, Member, IEEE, Eric Sortomme, Member, IEEE, Ghous M. Asim Akhtar, Student Member, IEEE, and Samy Faddel, Student Member, IEEE  Abstract—Electric vehicle (EV) integration into the distribution system has been a topic of great interest lately due to the potential challenges it poses. Previous works have focused on either centralized charge control or distributed charge control to solve these issues. In this paper, an adaptive voltage feedback controller for an onboard EV charger is proposed that, unlike other proposed methods, does not require any real-time communication between the EV and the utility. This controller compares the system voltage at the point of charging with a preset reference voltage. The EV charging is reduced as the system voltage approaches this reference. The reduced charging rate takes into account the EV battery state of charge (SOC) and the owner’s end-of-charge time (ECT) preference. To validate the proposed control structure, extensive simulations are carried out on a distribution system with and without other voltage control mechanisms. The simulation results show that this method can eliminate system voltage violations that would otherwise be caused by EV charging while ensuring fairness among the various EVs even with different system configurations and EV penetration levels. The proposed controller shows a good performance in the presence of other voltage control devices and distributed generation units. Also, it can integrate with Vehicleto-Grid services as a lowest level of hierarchical control. Index Terms—Electric Vehicles, Charge Control, Feedback control, voltage control mechanisms

I. INTRODUCTION electric vehicles (EVs) into the power grid INTEGRATING without any negative impacts is important for their successful adoption in large numbers. EVs have many positive benefits, such as reduced local emissions and petroleum independence. However, their charging can have adverse effects on the grid. While problems on the bulk power system are possible for large numbers of EVs [1], impacts on the distribution system are expected to be significant [2]-[4]. The work of A. T. Al-Awami, Ghous M. Asim Akhtar, and Samy Faddel was supported by King Fahd University of Petroleum & Minerals (KFUPM), Dhahran, Saudi Arabia, under project no. RG1318-1 and RG1318-2. A. T. Al-Awami is with the Department of Electrical Engineering at King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia (email: ([email protected]). Ghous M. Asim Akhtar is with the Pakistan Petroleum Limited, Karachi, Pakistan (email: [email protected]). Samy Faddel is with the Department of Electrical Engineering at King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia (email: ([email protected]). E. Sortomme is with the University of Washington Bothell, Bothell, WA 98011 USA (email: [email protected]).

These impacts include line overloads, under-voltages, increased losses, and sharp peak demands. However, it has been shown that through controlled charging, the negative impacts of EV charging can be significantly reduced [4]. Many methods of EV charge management have focused on centralized scheduling and control [4]-[8]. In [4]-[6], various methods of minimizing distribution feeder losses were explored. These studies revealed that feeder load profile can be flattened, voltage violations can be reduced [4], [6] and transformer life can be extended [5]. All of these methods required feeder load forecasts. In [7], [8] economic based charge control was investigated. It was found that price-based methods can sometimes cause distribution system overloads in the night hours due to low system prices but loss optimization would always flatten the load profile as much as possible. These methods require well-developed communication infrastructure to dispatch the control commands to and from the EVs. Some other methods have focused on decentralized control and optimization of EV charging, which requires reduced communications infrastructure and computational burden. In [9], the coordination of EVs was performed using noncooperative games to minimize generation cost. A distributed algorithm that considers the EV battery state of charge (SOC) was proposed in [10] to level the load at night. The algorithm presented in [11] was based on EVs setting their own charge profiles according to price forecasts. Another decentralized method focused on managing all of the charging within a parking lot while the parking lot was given its own maximum charge rate [12]. In [13], a distributed framework was suggested that aims to have EVs to charge at comparable charging rates without overloading the upstream service transformer. These and other similar decentralized charging methods rely on communications from the utility of some sort, even if there is no communication from the EVs back to the grid. A few communication-free EV charging strategies have been developed to allow autonomous charge control. In [14], voltage-constrained local optimization of EV charging was suggested. Each EV in the system optimizes its own charging aiming to maximize its charging rate while not violating nodal voltage or feeder loading constraints. This paper suggests a localized voltage-aware charge strategy of a competitive nature. This localized optimization relies on a pre-specified sensitivity of the voltage at the point of charging to the load variations at that point. Therefore, this approach is not flexible

Copyright (c) 2015 IEEE. Personal use of this material is permitted. However, permission to use this material for any other purposes must be obtained from the IEEE by sending a request to [email protected].

2

to accommodate normal cyclic variations in loading levels and possible feeder re-configurations. In addition, charging fairness among the connected EVs to the system and the battery SOCs are not considered in this competitive approach. In [15], it was assumed that each charging station is equipped with a fuzzy controller. The controller inputs are the nodal voltage at the charging point and the EV battery SOC. Based on these inputs, the controller decides on the actual power draw of the battery. Despite its novelty, fuzzy controller tuning requires extensive effort. Comparable performance can be obtained using simpler control structures. Lopes et al. [16] introduced a voltage-feedback, frequency-feedback control structure for bi-directional V2G within a microgrid. The results showed the effectiveness of this structure in preventing voltage and frequency violations. A voltage-feedback control structure for EVs in a distribution system was shown in [17]. However, the issue of fairness among EVs connected to different nodes in the system was not addressed in [16], [17]. SOC dependency of charging rate was not considered, either. In this work, a novel voltage feedback controller for an electric vehicle charger is introduced to manage charging in a fair manner. It requires no communications from or to the utility in real-time and no real-time coordination. It ensures that the EV charging will never cause the distribution system voltages to vary outside of the ±5% as required by the standard ANSI C84.1-2006 [18]. The features of SOC dependency of charging rate and the EV owner’s preference of end-of-charging time (ECT) are also included. Simulations on a sample distribution system without and with other voltage control devices show the efficacy of this charge controller. Additional simulations are performed to verify the robustness of the controller with respect to changes in EV penetration, network topology and the number of customers with short ECTs. In all cases, the controller works satisfactorily even as the system changes from its initial conditions. This controller works well in the presence of voltage control devices and distributed generation units. Also, if EV charging is managed by a hierarchical Vehicle-to-Grid (V2G) scheme, the proposed controller can act at the lowest level to effectively mitigate network violations. II. VOLTAGE BASED FEEDBACK CONTROLLER An electric vehicle charger converts the AC current from the grid into a constant DC current to charge the batteries. From the grid, the EV, therefore, is often seen as a constant current source [20]. When connected to the grid through an SAE J1772 charging station, a pilot signal is supplied to the EV from the station that tells what the maximum AC current draw is from that connection point. The EV charges at that current unless the battery management system reduces the maximum current draw to improve battery life near the end of the charging cycle, or if the EV charger cannot handle that high current level. The proposed voltage-based controller adjusts this EV charging current, and therefore the charging load, based on the AC voltage observed at the point of connection. The direct objective of the EV battery charging control is to maintain the distribution system nodal voltages within acceptable limits. This will ensure that the feeder losses are

reduced and overloads are avoided [4]. At a given distribution transformer, the load is the composition of controllable and non-controllable loads. Since the voltage profile of the system is a function of its loading levels, the voltage profile can be significantly enhanced by controlling the load. In this work, the only controllable loads considered are the EVs. In the proposed control structure, the feedback signal that is used as an input for the controller is the voltage at the point of charging (POC). The controller output is the regulated charging rate, or the charger current draw (IDi). Since unidirectional power flow is assumed, the charging current minimum limit is zero and its maximum limit is taken from the EV charger specifications or the maximum rating of the charging station, whichever is lower. For each EV, based on the POC voltage and the EV battery SOC, the controller decides on the regulated charging current. Fig. 1 shows a block diagram for the voltage feedback controller. Notice that there is one controller per EV. In order for the charging current, IDi, to be nonzero for a plugged-in EV whose SOC is not yet full, the voltage at the POC must be within permissible limits. In its simplest form, the proposed controller represents a nonlinear proportional relationship between the EV charging rate and the voltage at the POC. The output of the controller is continuous. Hence, the regulated charging rate, IDi, can take on any value between 0 and ̅̅̅̅ 𝐼𝐷𝑖 over a wide range of nodal voltage levels. As long as the SOCi < BattCapi and Vi > 𝑉𝑟𝑒𝑓,𝑖 , the regulated current charging rate, IDi, can be stated as: IDi = IDmax – (IDmax − IDmin) e−(α(𝑉𝑖 − 𝑉𝑟𝑒𝑓,𝑖))

(1)

Where IDmax and IDmin are the maximum and minimum values of current draw, Vref,i is the reference voltage level for the ith EV in per unit (pu), Vi is the actual real-time voltage in pu at the POC, and α is a constant. Because the system loading is measured in power, not current, it is helpful to refer to the EV power draw, which is merely the current draw IDi, multiplied by the node voltage, as shown in Fig. 1. Therefore, for the rest of this work, only the power draw will be referenced even though it is IDi that is actually directly modulated. Thus, the regulated power charging rate, F(Vi), can be stated as: 𝐹(𝑉𝑖 ) = {

𝑃𝐷𝑚𝑎𝑥 − (𝑃𝐷𝑚𝑎𝑥 − 𝑃𝐷𝑚𝑖𝑛 )e−(𝛼(𝑉𝑖 –𝑉𝑟𝑒𝑓,𝑖 )) 0

𝑖𝑓 𝑉𝑖 > 𝑉𝑟𝑒𝑓,𝑖 𝑖𝑓 𝑉𝑖 ≤ 𝑉𝑟𝑒𝑓,𝑖 (2)

Where PDmax and PDmin are, respectively, the maximum and minimum power draws, i.e. charging rates. From (2), it can be seen that if Vi is close to Vref,i, the charging rate is close to PDmin. However, as Vi increases, the power draw will increase. This ensures that the charging rate increases whenever there is room in the grid for higher power draw. A very important aspect of an EV charging strategy is “fairness”. That is, the contribution of each EV to mitigate voltage violations should be decided upon in a manner that

3

does not consistently charge an EV significantly slower or faster than another EV based on their locations in the network.

Fig. 1. Analog controller block diagram, where F (ui)* = exp (𝟏 − 𝑺𝑶𝑪𝒑𝒖, 𝒊(𝒕))

This fairness can be thought of in two directions: horizontal fairness and vertical fairness. Horizontal fairness corresponds to the fact that EVs charging at about the same voltage levels should be charging at similar charging rates. Since Vi for all these EVs are approximately the same, horizontal fairness can be achieved by simply setting Vref,i of these EVs to be identical. Vertical fairness is related to the level of contribution of EVs connected to POCs at different voltage levels. It is desirable that all EVs connected to the same feeder to have almost equal charging opportunities. That is, it won’t be appropriate or acceptable that EVs connected to downstream, i.e. lower voltage, POCs suffer from much lower regulated charging rates than those connected to upstream, i.e. higher voltage, POCs. Vertical fairness can be improved by using a charging rate function that is not excessively sensitive to the voltage level. Otherwise, EVs connected upstream will have an unfair advantage due to their higher voltage level over EVs connected downstream. It is only when the voltage is considerably high that the charging rate should increase. This is achieved using the proposed controller, which is presented in (2) in its preliminary form. Note that the voltage set points, Vref,i, will be kept constant at 0.955 p.u for all EVs at all POCs. This value satisfies the ANSI C84.1 standards. This voltage reference will be kept constant in all cases regardless of seasonal variations. If the voltage is below this set point at a given POC, the charging rates for all EVs connected to that POC should be set to zero. However, because of the way the charging rate function is set up, 𝐹(𝑉𝑖 ) in (2) has a discontinuity at Vi=Vref,i. In order to ensure smooth transition of the power draw from PDmin to zero, a ramp-rate limiter is applied. A. Charging as a Function of State of Charge An additional property that is added to the control scheme is the dependence of the charging rate on the EV battery SOC. This is included by multiplying 𝐹(𝑉𝑖 ) in (2) by exp(1– SOCpu,i), where exp(.) stands for the exponential function and SOCpu,i = SOCi/BattCapi. This term will bias the effective charging rates more towards the least charged EVs and less towards the most charged EVs. Because SOCpu,i changes over a wide range (between 0 and 100%) and because Vi changes over a limited range in normal operational conditions, the resulting charging rate is more sensitive to changes in SOCpu,i than to changes in Vi. This is a desirable feature.

B. Preferred End-of-Charge Time (ECT) The control scheme is further modified in order to include any possible preference of the ECT for the EV owner. This function is optional; i.e. the distribution system operator may or may not opt to offer it to EV owners. Accommodating the ECT preference is done by limiting the EV power draw to a value that is dependent on the remaining uncharged battery capacity. Thus, the minimum power draw for each EV is defined as the average value required over the remaining charging interval. That is, for an EV with a current state of charge of SOCi(t) and a total battery capacity of BattCapi, the power draw is modified to: 𝑃𝐷𝑖 (𝑉𝑖 , 𝑆𝑂𝐶𝑝𝑢,𝑖 ) = {

𝑃𝐷𝑖∗ 0

𝑖𝑓 𝑉𝑖 > 𝑉𝑟𝑒𝑓,𝑖 𝑖𝑓 𝑉𝑖 ≤ 𝑉𝑟𝑒𝑓,𝑖

(3)

Where 𝑃𝐷𝑖∗ = max{𝐹(𝑉𝑖 )e(1−SOC𝑝𝑢,𝑖(𝑡)) , (𝐵𝑎𝑡𝑡𝐶𝑎𝑝𝑖 − 𝑆𝑂𝐶𝑖 (𝑡))/ (𝑑 − 𝑡)}

(4)

Where d is the preferred total charge time (in hours) set by the EV owner. Note that this additional term cannot guarantee that the EV will charge fully before the ECT. This is because the 𝑃𝐷𝑖∗ term in (3) applies only when the POC voltage is higher than Vref,i. Otherwise, 𝑃𝐷𝑖. will be set to zero. Therefore, in extreme conditions, some of the EV owners with preferred ECT might not be granted their preference. Note that the output of this equation is in pu. Therefore, it is multiplied by the actual maximum charging current ̅̅̅̅ 𝐼𝐷𝑖 and the actual voltage (𝑉𝑖 ∗ 𝑉𝑛𝑜𝑚𝑖𝑛𝑎𝑙 ) to obtain the actual power draw, as shown in Fig. 1. A limiter is used to ensure that 𝑃𝐷𝑖. does not exceed the maximum charging rate of the battery. III. TEST SYSTEM The primary distribution test system used for simulating the EV charging impacts is shown in Fig. 2. This is an unbalanced three phase system with 17 load buses on each phase. This system was originally introduced in [21], and was one of the systems used to study EV charging impacts in [4]. It has also been used in microgrid studies [22]. The primary distribution system operates at a nominal 12.47 kV line-to-line voltage. The conductors are organized in a symmetric geometry with a geometric mean spacing of 4.69 ft. Every load bus has 20 houses connected to each secondary phase. The load profile for each house is based on Residential High Winter Ratio (ResHiWR) load profiles on July 20, 2010 found in the ERCOT system with five minute resolution [19]. To the base load profile, normally distributed random noise is added to model variations in individual usage. The parameters of the distribution system are found in Table I. The secondary distribution system is modeled based on the field site configuration of Utility E in [23], which has several splice boxes as well as houses connected directly to the distribution transformer through triplex lines at a nominal service voltage

4

of 240 V as shown in Fig. 3. The parameters of the secondary distribution network are given in Table II. A small resistance is also added to model the EV charging cable. Voltage regulators and capacitor banks are also added for some of the tests. A single-phase voltage regulator with 32 steps is available [30]. The per-step voltage change is 0.003125 per unit. The voltage set point for the VR is 0.99 pu. The regulator feedback control loop has a bandwidth, i.e. the allowable variation in voltage before the VR changes taps, of 0.003125 per unit. For the shunt capacitors, the total available capacity is assumed to be 1 MVar. Each capacitor step is assumed to be 0.1 MVar, which causes a voltage change of about 0.007 per unit. The capacitor bank is assumed to be controller by an autonomous feedback controller whose voltage set point is 0.99 pu and bandwidth is 0.007 pu. An average of one EV per two houses in the system, i.e. a 50% penetration level, is assumed. Each EV is randomly assigned to a house on the secondary network. This level is chosen because it has been shown to cause significant problems with EV charging [4]. Each EV has a maximum charge rate of 6.6 kW and needs to charge 24 kWh to reach full capacity. This corresponds to a 2013 Nissan Leaf [24]. The average initial state of charge of each EV is assumed to be approximately 40% of the battery’s full capacity. In addition, it is assumed that 10% of the EV owners have preferred ECTs, which range between 4 and 7 hours. These have also been assigned randomly. It is assumed that the system under study is under a time-ofuse (TOU) tariff structure. A lower tariff is applied from 7 pm to 7 am. Therefore, it is expected that the majority of EV owners plug in their EVs at or after 7 pm. To take this into account and the fact that EV plug-in time is expected to be random, the EV plug-in time is assumed to follow a Gaussian distribution centered at 8 pm and with a standard deviation of one hour.

Due to space limitations, only some of the results corresponding to nodes 2 and 6 are presented. The former is the most upstream primary load node with the highest voltage, while the latter is the least voltage primary load node. Voltage results at the primary nodes are normalized by the Nominal System Voltage for a 12470Y/7200 rating and results at the POCs are normalized by the Nominal Utilization Voltage for a 240/120 rating [18]. Since the Nominal Utilization Voltage is only 230/115 for a 240/120 rating, the POCs can have higher per unit voltage values than those of the nodes. TABLE I DISTRIBUTION SYSTEM PARAMETERS

Phase Conductor: Neutral Conductor: Max Amps: Houses

TABLE II SECONDARY NETWORK PARAMETERS Parameter EV Charger Penetration Distribution Service Transformer Secondary Conductor (transformer to splice box) Service Conductor (to the houses) No. of customers

Value 50% 150 kVA, %Z = 1.8 350 Al, 4/0 Al Neutral #2 Al 20

138 kV

12.47 kV 1

17 422' 16

3072'

2448' 10

2699'

2

528'

11

IV. SIMULATION AND RESULTS

3

4224'

1320' 1960'

18

To show the merits of the proposed EV charge control scheme, several simulations are conducted. Fig. 4 shows a basic schematic diagram of the simulation setup designed in the Matlab/Simulink environment for each of these simulations. The outputs of the three-phase power flow block are the voltages at the points of charging, i.e. at the houses. Here, n represents the node number, where n ∈ [2, 18], p is the phase number, where p ∈ [1, 3], and i is the house number, where i ∈ [1, 20]. The voltage of each house is fed back to the EV charge controller at that house. The controller decides on the power draw, PD(n,p,i), which is added to the non-EV load at the house, and the total power, P(n,p,i) is used as an input for the next power flow update. It is assumed that 50% of the houses have EVs. Since the total number of houses connected to each phase of the secondary transformer is 20 houses, each phase of the secondary transformer has 10 EVs connected to it. The EVs at each phase are labeled with numbers from one to ten, i.e. EV1, EV2, …, EV10 are connected to phase a; another set of ten EVs labeled EV1-EV10 are connected to phase b; and a third set EV1-EV10 is connected to phase c. This applies to each of the load nodes (nodes 2 – 18).

ACSR 2 ACSR 4 180 1020

1168'

9 686'

15

475'

14

4

8278'

1326' 12

6

5560'

5

7098'

2650' 6411'

13 7

Fig. 2. The distribution feeder test system. Load buses are 2-18.

Fig. 3. Secondary distribution network topology.

8

5

100 80 60 40 20 05pm

120 Total Load (kW)

Total Load (kW)

Node-6

phase a phase b phase c

phase a phase b phase c

100 80 60 40 20

09pm

12am Time

03am

0 05pm

06am

09pm

12am Time

03am

06am

Fig. 5. Total loads at primary nodes 2 and 6 using opportunistic charging. Node-6

Node-2

1

0.98

0.98

0.96 phase a phase b phase c

0.94 0.92 05pm

09pm

12am Time

03am

Voltage (pu)

1

0.96 phase a phase b phase c

0.94 0.92 05pm

06am

09pm

12am Time

03am

06am

Fig. 6. Voltage profiles at primary nodes 2 and 6 using opportunistic charging. Node-2

Node-6

1.05

1.05

1

1

0.95 EV 1 ,phase a EV 4 ,phase a EV 7 ,phase a EV 4 ,phase c EV 10,phase c

0.9 0.85 05pm

09pm

12am

Time

03am

06am

Voltage (pu)

B. Voltage-Responsive Charging A very important aspect of a “good” charging strategy is “fairness”. In order to ensure the charging fairness, the proposed controller is used. The voltage set point is 0.955 pu. This value satisfies the ANSI C84.1 standards and keeps a reasonable margin of 0.005 pu. The effective gain of the controller is not constant, but varies continuously according to (3). According to the IEC-61851 standard [27], there is a minimum acceptable charging rate below which charging should be stopped. This is assumed to be 30% of the charger’s

Node-2 120

Voltage (pu)

A. Opportunistic Charging In opportunistic charging, the EVs are assumed to be charging at maximum charging rate as soon as they are plugged in. That is, the charging process is not controlled by the proposed controller in this case. This case is presented as a benchmark against which the charging case controlled by the proposed controller is compared with. This case highlights the load response to a voltage change of the EV inverter/chargers. This response is approximately consistent with a constant current ac load for a Nissan LEAF [20]. The charging stations themselves are assumed to not have any voltage response, which is the case for the vast majority of stations, such as [25], [26]. The results of opportunistic charging are shown in Fig. 5– Fig. 8. A significant jump is observed in the total load within the first several hours of charging, hence a significant jump in current is also observed. In addition, a voltage dip, considerably below the permissible limit of 0.95 pu, due to the sudden and un-controlled increase in loading is noticed at a number of primary nodes and secondary POCs. Note that, in general, downstream primary nodes suffer from lower voltage profiles than upstream primary nodes. Therefore, two extreme cases are the POC with the shortest secondary wire length connected to primary node 2 (to be labeled as POC A), and the POC with the longest secondary wire length connected to primary node 6 (to be labeled as POC B). POC A is expected to have a very high POC voltage, while POC B is expected to have a very low POC voltage. In the results that follow, more emphasis is given to these two extreme cases. Note that the bold curves in Fig. 5-Fig. 7 correspond to these two POCs or to the primary nodes that each of them is connected to.

Voltage (pu)

Fig. 4. Simulation setup.

rated power, i.e. PDmin = 0.3 pu. The maximum charging rate is 100% of the charger’s rated power. In this case study, since most EVs are expected to stay plugged in for extended hours, a lower maximum charging rate can be used. This will have the added advantage of further reducing the peak load. Therefore, the value of PDmax is set to 0.7. As will be shown later, higher PDmax will still give satisfactory results, but at slightly higher peak load. An important advantage of using constant set points and adaptive gains is that it will still provide fairness when voltage control devices and distributed generators are in use in the system. Also, it needs no adjustment for seasonal load variations once it is well adjusted at the beginning. Fairness is improved even further by making the adaptive gain a function of the battery SOC. The effective gain exponentially decreases as SOCpu,i increases which ensures more fairness among different EVs with different SOCs and it increases the battery life time since the charging rate greatly decreases near the end due to the exponential dependency. Fig. 9– Fig. 15 show the nodal loads, nodal voltage profiles, POC A’s and B’s voltage profiles and their SOCs, total distribution system load , the average charging rate at node 2 and node 6, and primary currents due to the use of a more fair SOC-dependent adaptive control scheme.

0.95 EV 1 ,phase a EV 4 ,phase a EV 10,phase a EV 4 ,phase b EV 3 ,phase c

0.9 0.85 05pm

09pm

12am Time

03am

06am

Fig. 7. Voltage profiles at several secondary POCs including POCs A and B.

6 1

100 Current(Node-02,Phase-c) Current(Node-06,Phase-c)

0.99

Voltage (pu)

Current(Amps)

80

60 40

Node2,phase-c Node6,phase-c

0.98 0.97 0.96

20

0.95 5pm 12am Time

3am

9pm

6am

12am Time

3am

6am

Fig. 10. Voltage profiles at nodes 2c and 6c using fair, SOC-dependent control.

Fig. 8. Current profiles at primary nodes 2-c and 6-c using opportunistic charging

The figures show no voltage violations even at the most downstream POC. In addition, although node 2 (upstream node) has a higher voltage than node 6 (downstream node), the proposed controller greatly reduces the gap in charging time between the EVs connected to each of the nodes. This is because of SOC-dependency which makes the effective charging rate decrease gradually as SOCpu gets higher. While the focal point of this research is the distribution system, these load profiles indicate that this new scheme also benefits the bulk power system by shaving the evening peak load through delaying some of the EV charging load to night and early morning hours. In addition, the proposed controller reduces the line loading to avoid current overload problems that might happen in case of uncontrolled charging. This can be seen by comparing Fig. 8 to Fig. 15. Table III summarizes the comparison in performance among two voltage feedback control schemes: one with constant controller gain and the one proposed in (3) whose controller gain adapts with the voltage level and battery SOC. It shows the effectiveness of the proposed controller in closing the gap between POCs A’s and B’s times to full charge. While only the results corresponding to POC A and POC B are shown, the EVs at the other POCs follow a similar trend.

1.04 1.02

Voltage(pu)

9pm

EV-10(Node-02,Phase-c) EV-03(Node-06,Phase-c)

1 0.98 0.96 0.94 5pm

9pm

12am Time

3am

6am

Fig. 11. Voltage profiles for POCs A’s and B’s EVs using fair, SOCdependent control. 120 100


80

SOC %

0 5pm

60 40 20

TABLE III COMPARISON IN TERMS OF TIME TO FULL CHARGE IN HOURS – CONSTANT VS. FAIR CONTROL SCHEMES Node-02 Node-06 Difference Gain POC A POC B

0 5pm

9pm

12am 3am 6am Time Fig. 12. SOC for POCs A’s & B’s EVs using fair, SOC-dependent control.

4000 3.05

7.375

4.325

5.6416

7.0583

1.4167

Node-2

Node-6

80

phase-a phase-b phase-c

80

60

40


70 Total Load (KW)

Total Load (KW)

100

9pm

12am Time

3am

6am

2000

60

1000

50 40 30

20 5pm

EV NonEV Aggregated

3000

Load(KW)

Proportional constant controller Proposed controller

20 5pm

0 5pm

9pm

12am Time

3am

Fig. 9. Total loads at nodes 2-c and 6-c using fair, SOC-dependent control.

6am

9pm

12am 3am 6am Time Fig. 13. EV, Loads for the distribution system using fair, SOC-dependent control.

7

results confirm the robustness of the proposed scheme to simple and severe system reconfigurations.

120 (Node-02) (Node-06)

100

TABLE IV COMPARISON IN TERMS OF TIME TO FULL CHARGE (IN HOURS) – SOCDEPENDENT SCHEME AT DIFFERENT EV PENETRATION LEVELS

SOC %

80 60 40

Level (%)

POC A

40

5.175

Node-02 Mean Phase C 4.429

50

5.642

5.344

5.708

7.13

5.966

7.492

60

5.95

5.928

6.1

-

-

-

Penetration

20 0 5pm

9pm

12am Time

3am

6am

Fig. 14. Average charging rate at node 2 and node 6 using fair, SOCdependent control.

Latest

POC B

5.192

5.66

Node-06 Mean Phase C 4.488

Latest

5.825

70 Current(Node-02,Phase-c) Current(Node-06,Phase-c)

Current(Amps)

60 50

TABLE V COMPARISON IN TERMS OF TIME TO FULL CHARGE (IN HOURS) – SOCDEPENDENT SCHEME AFTER DISCONNECTING NODES AND CASE OF LIGHT LOADING

40

Node-02

Node-06

30

POC

Mean

A

Phase

Latest

POC

Mean

B

Phase

Latest

20

Case

10 0 5pm

c 9pm

12am Time

3am

c

6am

Fig. 15. Current profiles at primary nodes 2-c and 6-c using fair, SOCdependent control

V. CONTROL SCHEME PERFORMANCE In this section, the performance of the proposed adaptive, SOC-dependent EV charge controller is further studied. The robustness of the proposed controller with respect to varying levels of EV penetrations is first assessed. Table IV shows a comparison in terms of the time needed to fully charge the EVs at nodes 2 and 6. The average and latest charging times at phase c of the two nodes as well as the charging times for EVs connected to POCs A and B are shown for EV penetration levels of 40%, 50%, and 60%. The results demonstrate reasonable tolerance of this control scheme to different levels of EV penetrations up to 50%. If the penetration depth is increased above this value, POC B will not be able to charge fully before 6 am, which means that there is no enough room in the system for high penetration depth above 50%. This indicates that the system needs to be upgraded and additional voltage control devices, e.g. shunt capacitors, should be added to the system. This should be expected as each system has a maximum load limit, after which it must be upgraded. The robustness of the proposed controller with respect to probable system reconfiguration is also studied. This is carried out for simple and significant node reconfiguration events. Simple reconfiguration is carried out by removing one representative peripheral node at a time from the distribution system to simulate switching that node onto an adjacent feeder during reconfiguration. In this case, nodes 4 and 8 are each removed, one at a time. Significant reconfiguration is carried out by removing nodes 10-18, which represent about one half of the system. Table V shows the corresponding charging times for simple and significant reconfiguration events. These

Node # 04 removed

5.642

5.338

5.7

6.66

5.69

6.917

Node # 08 removed

5.642

5.338

5.7

6.8

5.832

7.45

Nodes # 10-18

5.6

5.291

5.65

6.38

5.51

6.475

5.6

5.296

5.65

6.1

5.283

6.1

removed Light loading

System performance during light load condition is also studied. This is done by reducing the non–EV loads. The results in terms of full time to charge are included in Table V. Fig. 16-Fig. 19 show POC A’s and B’s voltage profiles and their SOCs, total distribution system load and the average charging rate at node 2 and node 6 in case of light loading. It is clear that, since there is a room for charging, the EV at POC B charges faster. Also, Fig. 19 shows that the average charging rate at upstream and downstream nodes are almost the same, indicating complete fairness. An assessment of the performance of the charge controller due to an increase in the numbers of EV owners with preferred ECT is carried out as well. Fig. 20 and Fig. 21 show the test results for different percentages, 0%, 10% (base case), 30%, and 50% of owners simultaneously having preferred ECT. For each EV with a preferred ECT, a random integer number between 4 and 7 hours is assigned. The results show that this control scheme can accommodate a high percentage of EV owners with preferred ECT. Note that all EVs with preferred ECTs are charged fully before the specified ECTs. These tests show that once the parameters are tuned, changes in the system do not require retuning. This reduces the computational burden of implementing such a scheme.

8

4000

1

0.98 0.97 0.96 0.95 5pm

ECT=0% ECT=10% ECT=30% ECT=60%

3500

Load(KW)

Voltage (pu)

0.99


3000 2500 2000 1500

9pm

12am 3am 6am Time Fig. 16. Voltage profiles for POCs A’s and B’s EVs in case of light loading.

1000 5pm

9pm

12am 3am 6am Time Fig. 20. Total load (EV + non-EV) for the distribution system at different percentages of EV owners with preferred ECT.

120 100


1 0.99

Node 2

Voltage (pu)

SOC %

80 60 40 20 0 5pm

0.98 0.97 Node 6

0.96

9pm

12am 3am 6am Time Fig. 17. SOC for POCs A’s & B’s EVs in case of light loading.

0.95 5pm

9pm

12am Time

ECT=0% ECT=10% ECT=30% ECT=60% 3am 6am

Fig. 21. Voltage profiles at nodes 2 and 6 at different percentages of EV owners with preferred ECT.

3500 EV NonEV Aggregated

3000

Load(KW)

2500 2000 1500 1000 500 0 5pm

9pm

12am 3am 6am Time Fig. 18. EV, Non EV and Total loads for the distribution system in case of light loading. 120 100

(Node-02) (Node-06)

SOC %

80 60 40 20 0 5pm

9pm

12am Time

3am

6am

Fig. 19. Average charging rate at node 2 and node 6 in case of light loading.

Also, an assessment of the performance of the charge controller in the presence of voltage control devices in the system is carried out. Fig. 22 - Fig. 24 show the test results for the system in the presence of shunt capacitor unit of 1 MVar at node 5. Node 5 is chosen since it is a downstream node and the support at this point will also help nodes 6, 7, and 8. As can be seen in Fig. 22 and Fig. 23, the voltage at POC B has improved and almost all of the EVs at up- and down-stream nodes charge at almost the same rate. This is mainly due to the increase in voltage at node 6 due to the presence of the shunt capacitor. Fig. 24 shows the number of capacitor steps at that node. Also, the performance of the charge controller in the presence of a voltage regulator at node 5 is investigated. Fig. 25-Fig. 27 show POC A’s and B’s voltage profiles, average charge rate and tap positions for a VR (voltage regulator) connected at node 5. As expected, the voltage significantly improves by adding this VR. This leads to increasing the charging rates of the EVs connected to downstream nodes, such as node 6. These tests show the compatibility and effectiveness of the proposed control scheme in the presence of traditional voltage control devices. To test the controller performance in the presence of distributed generation units, a small wind turbine is installed at a single house to see the effect of feeding the house from its own generation unit and how this will affect the charging rate of the electric vehicle connected to that house. Solar energy is not investigated here due to the assumption that the EVs are charging mostly at night where TOU tariff is low.

9

1.04

100

(Node-02) (Node-06)

80 1

SOC %

Voltage(pu)

1.02

120 EV-10(Node-02,Phase-c) EV-03(Node-06,Phase-c)

0.98

60 40

0.96 0.94 5pm

20 9pm

12am 3am 6am Time Fig. 22. Voltage profiles for POCs A’s and B’s EVs in the presence of shunt capacitors at node 5.

0 5pm

9pm

12am 3am 6am Time Fig. 26. Average charging rate at node 2 and node 6 in the presence of voltage regulator at node 5. Node-5

120 100

15 (Node-02) (Node-06)


No. of Taps

SOC %

80 60 40

10

5

20 0 5pm

9pm

12am 3am 6am Time Fig. 23. Average charging rate at node 2 and node 6 in the presence of shunt capacitors at node 5. Node-5 10 phase-a phase-b phase-c

No. of Steps

8 6 4 2 0 5pm

9pm

12am 3am 6am Time Fig. 24. Number of steps for shunt capacitor at node 5. 1.05

Voltage(pu)


1

0.95 5pm

9pm

12am 3am 6am Time Fig. 25. Voltage profiles for POCs A’s and B’s EVs in the presence of voltage regulator at node 5.

0 5pm

9pm

12am

3am

6am

Time

Fig. 27. Voltage regulator’s tap position at node 5.

For this test the wind turbine is installed at the downstream house at node 6, which suffers from low voltage. The wind data that is used is scaled actual data from Bonneville Power Administration (BPA) [31]. Fig. 28 shows the wind variation data. Fig. 29.-Fig. 30 show POC A’s and B’s voltage profiles and their SOCs in this case. The Figures show clearly that installing a WT at the downstream house at node 6-c considerably improves the voltage profiles. At some points in time, the voltage at that house is higher than that of the upstream node. The improvement in voltage leads to a higher charging rate of the EV connected to that house. This is desirable as the house that own the distributed resource is the one that benefits from it the most in charging its EV. More tests are also carried out with micro wind turbines dispersed randomly over 20% of the houses. The controller performances in all these tests are satisfactory as no voltage violations took place. An interesting general observation is that the houses where micro-wind turbine are connected had their EV charge faster. This is expected and desirable as the local generators help improve the voltage profiles at the secondary nodes they are connected to. Due to space limitation, detailed results are not included in this paper. To test the controller behavior if the controller parameters were changed, another set of parameters is selected and tested. For this case, the values of PDmax and PDmin are chosen to be 0.9 and 0.5. It should be remembered that the maximum value of PDmax is one and PDmin should be greater than zero. Fig. 31Fig. 33 show the nodal voltage profiles, POC A’s and B’s voltage profiles and the system loads for this case. The Figures show that the controller still performs well even with more

10

than 25% increase in the controller parameters. However, this results in a higher peak load due to the higher charging rates, as can be seen by comparing Fig. 33 to Fig. 13.

1.02

Voltage(pu)

2500

2000

Power (Watts)

1.04

0.98 0.96

1000

0.94 5pm

12am 3am 6am Time Fig. 32. Voltage profiles for POCs A’s and B’s EVs using SOC-dependent control for PDmax= 0.9 and PDmin=0.5.

500

0 12am

12pm

4000

Fig. 28. Wind power variations for a sample day.

Load(KW)


1

2000

1000

0 5pm

12am 3am 6am Time Fig. 29. Voltage profiles for POCs A’s and B’s EVs in case of WT at downstream house.

9pm

120 EV-10(Node-02,Phase-c) EV-03(Node-06,Phase-c)

SOC %

80 60 40 20 0 5pm

9pm

12am Time

3am

6am

Fig. 30. SOC for POCs A’s & B’s EVs in case of WT at downstream house. 1

Voltage (pu)

0.99


0.98 0.97 0.96 0.95 5pm

EV NonEV Aggregated

3000

1.05

100

9pm

12am

Time

Voltage(pu)

1

1500

0.95 5pm


9pm

12am Time

3am

6am

Fig. 31. Voltage profiles at nodes 2c and 6c using SOC-dependent control for PDmax= 0.9 and PDmin=0.5.

9pm

12am 3am 6am Time Fig. 33. Loads for the distribution system using SOC-dependent control for PDmax= 0.9 and PDmin=0.5.

VI. CONTROLLER COORDINATION WITH V2G In addition to working without communications, the proposed local controller can also coordinate with unidirectional V2G systems to ensure that the distribution system is not adversely affected by V2G dispatch. This effectively becomes the lowest level in a hierarchical control scheme. Having such autonomous control at the lowest level is advantageous since it reduces the communications traffic to the V2G aggregator. In order to test the controller’s performance within coordination with V2G, the group of EVs is scheduled using the OptComb scheduling algorithm of [28] for July 21-22, 2010 on the ERCOT system using the same plug-in times as in the previous sections. The schedule gives the times that the EVs are available to perform various ancillary services, in this case regulation up, regulation down, and/or responsive reserves. During a time period when capacity is scheduled, the system operator can call upon the aggregator to perform the services by moving its consumption above or below the scheduled levels. This is referred to as ancillary service dispatches. The aggregator, in turn, sends its own dispatch signal out to some or all of the EVs to meet the system dispatch signal. The EVs are dispatched by the aggregator using the discrete dispatch method described in [29] following the historic signal used in that work. Fig. 34 and Fig. 35 show that if the EVs simply follow the aggregator’s dispatch signal, the system suffers from undervoltage problems. This is clearly shown in Fig. 35 where the voltage at the down-stream house degrades to value lower than

11

with controller

without controller

1 0.99

1 Node2,phase-c Node6,phase-c

0.99


Voltage (pu)

Voltage (pu)

0.98 0.98 0.97

0.97 0.96 0.95

0.96 0.95 5pm

0.94 9pm

12am Time

3am

0.93 5pm

6am

9pm

12am Time

3am

6am

Fig. 34. Voltage profiles at nodes 2c and 6c with optimal V2G scheduling and dispatch with the proposed controller on the left and without it on the right. with controller

without controller

1.04

1 0.98

9pm

12am Time

3am

0.85 5pm

6am

System loading with controller

2000

0 5pm

12am Time

3am

6am

SOC % 9pm

12am Time

3am

6am

0 5pm

9pm

12am Time

3am

6am

VIII. REFERENCES

9pm

12am Time

3am

6am

EV NonEV Aggregated

4000

[2]

[3]

[4]

3000 2000

0 5pm

20

An effective, autonomous, voltage-based controller for electric vehicle chargers is proposed in this work. This controller, while requiring no real-time communication, effectively coordinates charging among the EVs connected to the distribution nodes in a fair manner so that voltage violations are avoided. The controller ensures almost complete fairness among the various EVs in the system. In addition to the local voltage level, the proposed controller takes into account the battery SOC and the EV owner’s preference (if any) of end-of-charge time. It requires no real-time communication and will thus not present a significant cost due to the limited communication bandwidth required. Multiple simulations are also run to verify the robustness of the controller once it is tuned. The simulations show that for simple and severe changes in network topology, the tuned controllers still work as designed. Changes in the number of EVs on the network or the customer ECT preferences do not cause the controllers to fail in operation. Simulation tests show the compatibility of the proposed controller with voltage control devices, which means that the controller can perform well in the existing distribution systems. Also, good coordination of the proposed controller with a small distributed generation unit is shown. Tests also show that the proposed controller can be successfully integrated with V2G systems.

1000

9pm

40

VII. CONCLUSION

[5]

1000

60

Fig. 37. Average charging rate at node 2 and node 6 with optimal V2G scheduling and dispatch with the proposed controller on the left and without it on the right.

System loading without controller

Load(KW)

Load(KW)

3000

40

0 5pm

5000

EV NonEV Aggregated

(Node-02) (Node-06)

80

60

20

[1]

Fig. 35. Voltage profiles for POCs A’s and B’s EVs with scheduling based on market signal with the proposed controller on the left and without it on the right. 4000

without controller

100

80

0.95

0.9

0.96 0.94 5pm


1

Voltage(pu)

Voltage(pu)

1.02

1.05 EV-10(Node-02,Phase-c) EV-03(Node-06,Phase-c)

without controller

(Node-02) (Node-06)

100

SOC %

0.9 pu. When the proposed controller is applied, however, no voltage violations occur in the system. Fig. 36 shows the EV loading on the system with and without using the feedback controller. All the EVs manage to fully charge when the proposed controller is not used since they charge as soon as they receive the signal without taking into consideration the system voltage and loading. When the proposed controller is used, some of the EVs charging is delayed to avoid voltage violation and overloading. This can be seen in Fig. 37. While it is clear that using the voltage feedback controller will prevent voltage violations as a result of V2G dispatch, it also changes the aggregator’s ability to fully follow the system dispatch by a non-negligible amount. This could cause penalties for the aggregator as well as problems with the system frequency. This problem can be solved by reformulating the optimal scheduling and dispatch of the EVs to account for the voltage feedback behavior. Such a formulation is beyond the scope of this work. However, due to various complex regulatory and business model issues, commercial-scale V2G aggregation is not currently happening and when it does, it will not immediately be available in all regions from the beginning. Nevertheless, the proposed voltage feedback controller can still work in the absence of these systems to provide immediate benefits to the distribution system while the V2G environment is still being built out.

9pm

12am Time

3am

6am

Fig. 36. Non EV and Total loads for the distribution system with optimal V2G scheduling and dispatch with the proposed controller on the left and without it on the right.

[6]

S. W. Hadley and A. A. Tsvetkova, “Potential Impacts of Plug-in Hybrid Electric Vehicles on Regional Power Generation,” The Electricity Journal, vol. 22, no. 10, pp. 56-68, 2009. J. A. P. Lopes, F. J. Soares, P. M. R. Almeida, “Integration of Electric Vehicles in the Electric Power System,” Proceedings of the IEEE, vol. 99, no. 1, pp. 168-183, 2011. L. Pieltain Fernandez, T. Gomez San Roman, R. Cossent, C. Mateo Domingo, P. Frias, “Assessment of the Impact of Plug-in Electric Vehicles on Distribution Networks,” IEEE Transactions on Power Systems, vol. 26, no. 1, pp. 206 – 213, 2011. E. Sortomme, M. Hindi, S. D. J. MacPherson, S. S. Venkata, “Coordinated Charging of Plug-In Hybrid Electric Vehicles to Minimize Distribution System Losses,” IEEE Transactions on Smart Grid, vol. 2, no. 1, pp. 198 – 206, 2011. M. A. S. Masoum, P. S. Moses, S. Hajforoosh, “Distribution transformer stress in smart grid with coordinated charging of Plug-In Electric Vehicles”, in Proc. Innovative Smart Grid Technologies (ISGT), 2012, pp 1 – 8. P. Zhang, K. Qian, C. Zhou, B. G. Stewart, D. M. Hepburn, “A Methodology for Optimization of Power Systems Demand Due to Electric Vehicle Charging Load,” IEEE Transactions on Power Systems, vol. 27, no 3, pp. 1628 – 1636, 2012.

12 [7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18] [19]

[20]

[21] [22]

[23]

[24]

[25]

[26]

[27] [28]

[29]

[30]

D. Wu, D. C. Aliprantis, L. Ying, “Load Scheduling and Dispatch for Aggregators of Plug-In Electric Vehicles”, IEEE Transactions on Smart Grid, vol. 3, no. 1, pp. 368 – 376, 2012. D. Steen, L. A. Tuan, O. Carlson, L. Bertling, “Assessment of Electric Vehicle Charging Scenarios Based on Demographical Data,” IEEE Transactions on Smart Grid, vol. 3, no. 3, pp. 1457 – 1468, 2012. Z. Ma, D. S. Callaway, I. A. Hiskens, “Decentralized Charging Control of Large Populations of Plug-in Electric Vehicles,” IEEE Transactions on Control Systems Technology, Early Access. C. Ahn, C. Li, H. Peng, “Decentralized charging algorithm for electrified vehicles connected to smart grid,” Proc. of American Control Conference (ACC) 2011, June 29 2011-July 1 2011, pp.3924-3929. L. Gan, U. Topcu, S. Low, “Optimal decentralized protocol for electric vehicle charging,” Proc. of 50th IEEE Conference on Decision and Control and European Control Conference (CDC-ECC), Dec. 2011, pp.5798-5804. W. Su, and M. Y. Chow, “Computational intelligence-based energy management for a large-scale PHEV/PEV enabled municipal parking deck,” Applied Energy, vol. 96, no. 8, pp. 171 – 182, 2011. S. Stüdli, E. Crisostomi, R. Middleton, and R. Shorten, “A flexible distributed framework for realising electric and plug-in hybrid vehicle charging policies,” Int. J. Control, vol. 85, no. 8, pp. 1130–1145, 2012. P. Richardson, D. Flynn, and A. Keane, "Local Versus Centralized Charging Strategies for Electric Vehicles in Low Voltage Distribution Systems," IEEE Transactions on Smart Grid, vol.3, no.2, pp.1020-1028, June 2012 M. Singh, P. Kumar, and I. Kar, "Implementation of Vehicle to Grid Infrastructure Using Fuzzy Logic Controller," IEEE Transactions on Smart Grid, vol.3, no.1, pp.565-577, March 2012. J. A. P. Lopes, S. A. Polenz, C. L. Moreira, R. Cherkaoui, “Identification of control and management strategies for LV unbalanced microgrids with plugged-in electric vehicles,” Electric Power Systems Research, vol. 80, pp. 898-906, 2010. F. Geth, N. Leemput, J. V. Roy, J. Buscher, R. Ponnette, and J. Driesen, “Voltage droop charging of electric vehicles in a residential distribution feeder,” 3rd IEEE PES ISGT Eurpoe, Berlin, Germany, Oct. 2012. American National Standard for Electrical Power Systems and Equipment - Voltage Ratings (60 Hertz), ANSI C84.1-2006. Electric Reliability Council of Texas, “Load Profiling,” August 2010, [Online]. http://www.ercot.com/mktinfo/loadprofile/ [Accessed: Aug. 20, 2010]. E. Sortomme, A. I. Negash, S. S. Venkata, D. S. Kirschen, “Multistate voltage dependent load model of a charging electric vehicle,” IEEE Transportation Electrification Conference and Expo, June 2012, pp. 1-5. Cooper Power Systems, Ed. 3, Electrical Distribution-System Protection, Cooper Industries, 1990. M. A. Zamani, A. Yazdani, and T. S. Sidhu, “A CommunicationAssisted Protection Strategy for Inverter-Based Medium-Voltage Microgrids,” IEEE Trans. Smart Grid, vol. 3, no. 4, pp. 2088–2099, Dec. 2012. .F. Lambert, Secondary Distribution Impacts of Residential Electric Vehicle Charging, 2000 :Public Interest Energy Res. (PIER), California Energy Commission Nissan USA, “2013 Nissan LEAF® Electric Car Specifications,” 2013, [Online]. Available: http://www.nissanusa.com/electriccars/leaf/versions-specs/ [Accessed: May 5, 2013] AeroVironment™ EV Solutions™, “Electric Vehicle Charging Station Turning Garages into Filling Stations of the Future,” 2011, [Online]. Available: http://evsolutions.avinc.com/uploads/products/2_AV_EVSERS_B2B_061110_retail_charging.pdf [Accessed: May 5, 2013]. ChargePoint, “CT2020 Family ChargePoint Networked Charging Stations,” 2013, [Online]. Available: http://www.chargepoint.com/files/CT2020-Data-Sheet.pdf [Accessed: May 5, 2013]. IEC 61851-1, Electric Vehicle Conductive Charging System – Part 1: General Requirements. IEC, 2nd ed. 69/173/FDIS, 2010. E. Sortomme and M. A. El-Sharkawi, “Optimal Combined Bidding of Vehicle-to-Grid Ancillary Services,” IEEE Transactions on Smart Grid, vol. 3, no. 1, pp. 70 – 79, 2012. E. Sortomme and K. W. Cheung, “Intelligent Dispatch of Electric Vehicles Performing Vehicle-to-Grid Regulation,” IEEE International Electric Vehicle Conference, March 2012, pp. 1-6. F. A. Viawan, “Voltage Control and Voltage Stability of Power Distribution Systems in the Presence of Distributed Generation,” Ph.D. Dissertation, Chalmers University of Technology, 2008.

[31]

Bonneville Power Administration (BPA), “Wind Data,” [Online]. http://transmission.bpa.gov/business/operations/wind/baltwg.aspx [Accessed: March 20, 2015].

IX. BIOGRAPHIES Ali T. Al-Awami earned his B.Sc. and M.Sc. degrees in Electrical Engineering from King Fahd University of Petroleum & Minerals, Saudi Arabia, in 2000 and 2004, respectively. He earned his Ph.D. from the University of Washington in 2010. In 2000 he joined the Saudi Electricity Company as a System Operation engineer. In 2002, he joined KFUPM as a Graduate Assistant, where he is currently an Assistant Professor there. He authored and co-authored several papers and book chapters in his research areas. His research interests include power system operation and optimization and the integration of electric vehicles and renewable energy sources into the smart grid. Eric Sortomme (S’08) received the B.Sc. degree magna cum laude in electrical engineering from Brigham Young University, Provo, UT, in 2007 and the Ph.D. degree from the University of Washington (UW), Seattle, in 2011. He has authored or coauthored a plethora of technical publications with a research emphasis is on smart grid technologies. He is currently an affiliate assistant professor with the University of Washington Bothell.

Ghous Muhammad Asim Akhtar earned his BE degree in Electrical Engineering from NED University of Engineering & Technology, Pakistan in 2008 and M.Sc. degree in Electrical Engineering from King Fahd University of Petroleum & Minerals, Saudi Arabia in 2013. In Jan2009 he joined Siemens Pakistan as Trainee Engineer. In Apr 2009 he joined Pakistan Petroleum Limited as Assistant Engineer. He is currently Engineer (E&I) with Pakistan Petroleum Limited. His research interests include power system operation and integration of electric vehicles into smart grid.

Samy Faddel earned his BSc degree in Electrical Engineering from Assiut University, Egypt in 2011. He earned his M.Sc. degrees in Electrical Engineering from King Fahd University of Petroleum & Minerals, Saudi Arabia, in 2015. In 2012, he joined GAEB, Egypt as Electrical Engineer up to 2013 where he worked at KFUPM as a Research Assistant. He is currently pursuing his Ph.D. degree in Electrical Engineering. His research interests include integration of renewable energy and electric vehicles in smart grid, power system operation and control.