An Intelligent Energy Management System for PHEVs Considering Demand Response

Transcription

1 1 An Intelligent Energy Management System for PHEVs Considering Demand Response Wencong Su 1, Student Member, IEEE, and Mo-Yuen Chow 2, Fellow, IEEE Abstract-- The anticipating large penetrations of PHEVs in the smart grid environments brings up many technical issues (V2G, G2V, energy management, etc.) that need to be addressed. An emerging issue is that a large amount of PHEVs simultaneously plug-in to the grid (e.g. at a municipal parking deck during a ball game) will have a huge potential to threaten the power system power quality and stability. In this paper, the effect of hybrid demand side management and demand response programs will be well presented. Case studies show how the price factor could be fully considered into an intelligent energy management system (iems) for PHEVs. Accordingly, the simulation results show the optimal electricity consumption by PHEVs considering both local load policy (e.g. critical load) and demand response. Index Terms--PHEV, Smart Grid, Intelligent Control, Demand Side Management, Demand Response, Energy Management, Monte Carlo Simulation. T I. INTRODUCTION HE technology, economic and environmental incentives are changing the features of the power systems. Plug-in hybrid electrical vehicles (PHEVs) are receiving the increasing attention due to its low pollution emission and low cost per mileage. At 2.2% of the automobile market share, 3,000 hybrid vehicles were sold in 2007 in US [1]. The Electric Power Research Institute (EPRI) projects that by 20, 62% of the entire U.S. vehicle fleet would consist PHEV (moderate PHEV penetration scenario) [2]. U.S. Department of Energy projects that approximately 1 million PHEVs will be on the road in 2015 and 425,000 PHEVs sold in 2015 alone. At this penetration rate, PHEVs would account for 2.5% of all new vehicle sales in 2015, with PHEV-12s dominating overall PHEV market sales [3]. Accordingly, the emergence of PHEVs may bring many technical issues and potential challenges [4-6]. Powered parking structures can ease the pain of the energy crisis. U.S. parking structures have the opportunity to return the energy to the electric grid through This work was supported in part by the National Science Foundation, Awarumber: EEC and this work is a part of an ongoing project in collaboration of the FREEDM systems centre (Future Renewable Electric Energy Delivery and Management) with ADAC (Advanced Diagnosis Automation and Control) Lab at North Carolina State University and ATEC (Advanced Transportation Energy Center). 1 Wencong Su is with the Department of Electrical and Computer Engineering, North Carolina State University, Raleigh, NC, 276, USA 2 Mo-Yuen Chow is with the Department of Electrical and Computer Engineering, North Carolina State University, Raleigh, NC, 276, USA wsu2@ncsu.edu chow@ncsu.edu solar and wind power, while reducing the need for foreign oil by changing PHEV at a fraction of the cost of gasoline [7]. Their energy consumption patterns will also significantly influence the electricity market. Not only do the PHEVs utilize grid power for charging, they also have the potential to transfer power back to the grid to alleviate peak power demand and provide ancillary services to the grid. This paper is organized as follows: section II will first describe the existing Intelligent Energy Management System (iems) architecture developed by FREEDM/ATEC Center to manage a large amount (e.g. 0) of PHEV charging at a municipal parking deck, then we will use agent concept to define each system components. Section III will then introduce the demand side management programs, and show how it can affect the load demand pattern. Two types of timebased demand side management approaches: TOU (Time-of- Use) and EDRP (Emergency Demand Response Program) will be applied to validate the performance of the demand side management programs. In section IV, case studies show how the energy cost could be factored into the iems for charging PHEVs at a municipal parking deck. Accordingly, the simulation results show the optimal electricity consumption by PHEVs considering both demand response and local load requirement (e.g. critical load). Section V will summarize the paper and briefly discuss the future work. II. IEMS ARCHITECTURE In previous studies [8-9], the foundation of an iems for PHEVs at a municipal parking deck has been developed. The initial iems algorithms work has been validated with the implementation of the algorithm in Matlab/Simulink and Labview. In addition, the communication between the iems and the chargers is achieved via ZigBee communication nodes [9]. The model aims at reflecting the real world scenario under a certain intelligent control philosophy. Fig. 1 illustrates the basic architecture of the entire system, which is composed of the power grid, the energy management system and the PHEVs. An enhanced distributed control topology is under development in order to regulate multiple loads from a cluster of vehicles. The information about price and power level will be periodically updated to the iems. Eventually each agent will be able to meet the desired load demand based only on the local information and a minimum amount of communication. We also consider customer preferences and energy price so as to optimize the energy usage. It will be discussed in the

2 2 following sections. Fig. 1. Intelligent PHEVs Charging System Architecture. [8] III. DEMAND SIDE MANAGEMENT Demand Side Management (DSM) is defined by Department of Energy (DOE) [10] as Changes in electric usage by end-use customers from their normal consumption patterns in response to changes in the price of electricity over time, or incentive payments designed to induce lower electricity use at times of high wholesale market prices or when system reliability is jeopardized. Therefore, the demand side management programs should be incorporated into the existing iems model in order to avoid voltage sag and blackout and to maximize the financial benefits. In addition, this under-utilized capacity could effectively power a national fleet of PHEVs with little need to increase the energy delivery capacity of the existing grid infrastructure [11]. There are many ways to achieve this goal. Some of DSM programs will be introduced as follows. Then the mixed DSM programs can contribute to the optimization of the energy utilization. Real-time Pricing (RTP): It is a dynamic price to reflect the real cost of electric energy cost. Utilities change the retail price hourly based on the usage information received from customers. RTP programs are the most efficient and direct DSM programs. Time-Of-Use (TOU) Program: Generally speaking, the electricity price will be low in the off-peak period, and will be high in the peak period. Thus the customers can adjust their consumption pattern with the various prices. Emergency Demand Response Program (EDRP): Based on historical demand, price data, and short term load forecasting, utilities will pay large customers a significant amount of incentives if they can reduce their electricity consumption in the peak period. As a result, the peak load demand is reduced to relieve the potential electricity shortage. Fig. 2 shows the relationship between electricity market price and demand elasticity. Obviously a small reduction on demand side can induce a huge reduction of electricity market price. The reason behind it is that the generation cost is extremely high during peak load period. Fig. 2. Simplified effect of DSM on electricity price. [12] Elasticity is defined as the demand sensitivity with respect to the price [13]: d d / d0 E (1) p p / p0 Where d and p are changes in demand and price respectively; d0 and p0 present the initial demand quantity and electricity price. The elasticity is an indicator of the effect of demand variation on the electricity price. Basically there are two types of elasticity coefficients: Selfelasticity ( E ) and Cross-elasticity ( E ). Self-elasticity ii ( E ii ) shows the effect of demand change on price in a single period. It is always a negative value. Cross-elasticity ( E ) shows the effect of demand change on the price in a multiperiod. It is always a positive value. E1,1 E1,24 E (2) E24,1 E 24,24 In the simulation part, hourly load profile is selected to model the demand variation on the end-use customers side. Thus the elasticity coefficients can be arranged in a 24 by 24 matrix. The diagonal elements are self-elasticity coefficients and the off-diagonal elements are cross-elasticity coefficients. Let d(i) be the customer demand in i-th hour (h), p(i) be the energy price in i-th hour ($/h), and A(i) be the incentive in i-th hour ($/h). Considering the single period and multi period, the final model is presented by [14] 24 d0( i) E( i) p( i) p0 ( i) Ai ( ) d( i) ( i) E 0( i, j) p( j) p0 ( j) A( j) 1 j1 p0 ( j) p0 ( i) (3) i 1,2,...,24 Equation (3) shows how much should be the customer's demand in order to achieve a maximum benefit over an interval of 24 hours. The detailed model shows how customers benefits can be maximized by changing the demand pattern. EDRP and TOU programs have been run to analyze the effect of demand variation on the electricity price. The DSM programs have been fully implemented in Matlab. Based on the historical data of load profile in summer, the load curve is divided into two intervals as: Peak load period: 7:00 a.m. 10:00 p.m., and Low load period: 10:00 p.m. 7:00 a.m. Initially we considered 10 $/h as an incentive in a jj jj

3 3 DSM program and the potential for DSM programs as %. Also Self-elasticity and Cross-elasticity used is listed as in Table 1. The entire real-time price and load demand data are accessible from a public database established by The Independent Electricity System Operator (IESO) [15-16]. Table 1. Self Elasticity and Cross Elasticity Peak Period Low Period Peak Period Low Perio x x 104 Fig. 5. Load Profile with 5 or 10$/h incentive, x Fig. 3. Load Profile with 10 $/h incentive, x Fig. 4. Load Profile with 10 $/h incentive, The load curve after optimization looks much flatter. The solid one is the original load curve. The curve in dash line is the load profile with DSM programs. The customers have the ability to reschedule the energy consumption. The peak demand has been reduced and loads are transferred from the peak period to low periods. It is obvious from DSM schedule that the customers benefit financially as long as the demand is flexible Fig. 6. Load Profile with 5 or 10$/h incentive, The load curve with a lower incentive is black colored and star marked. It is evident that the peak load reduction is smaller if a lower incentive is paid. So the customers are less interested in participating in this program. The utilities also benefit financially by cutting down the extremely high generation cost in the peak period. In other words, the load demand in the peak period declines significantly so as to reduce the generation cost. Accordingly, the power system reliability will be improved as well. The numerical calculations of this case study are presented in Table 2 and 3. In these cases, the maximum load variations are reduced from 13 to 3969 and from 6102 to 3482 respectively. Table 2. Numerical calculations of load demand with DSM programs on August 26, Before DSM After DSM Maximum Load Demand Minimum Load Demand Maximum Load Variation Maximum Load Variation Average load 35.62% 23.56%

4 4 Table 3. Numerical calculations of load demand with DSM programs on August 31, Before DSM After DSM Maximum Load Demand Minimum Load Demand Maximum Load Variation Maximum Load Variation Average load IV. DEMAND RESPONSE 39.19% 22.51% The anticipating large penetration of PHEV into our societies will add a substantial energy load to power grids, as well as add substantial energy resources that can be utilized. When the PHEV loads are aggregated into thousands at a short period of time (e.g., working day early morning when employees going to work and plug-in their PHEVs into the PHEV parking deck), the loaeeds to be managed carefully to avoid interruption to the grid service. Developing a demand response engine that can best fit the existing iems testbed is a big challenge. Fuzzy logic might be one of the potential approaches that can address this issue. References [17-18] describe the fuzzy control such that by providing an algorithm, it converts the imprecise linguistic control strategy into an automatic control strategy. The fuzzy control is much more robust, especially for the implementation of demand response into a multi-agent system. However we need to look into the scalability of PHEV loads, the load elasticity with respect to price variation, the local energy control policy and the multi-objective functions embedded in iems decision makers in more detail. An expert system has been well described and proposed in [19]. It is much easier to be implemented. Although it is a semi-automated approach to apply demand response program, it can still be used at this moment to validate the algorithms and demonstrate the basic procedure of demand response for charging PHEVs at a municipal parking deck. Assume there are several municipal parking decks providing power to PHEVs. The daily load profile of each parking deck is predicable from short-term point of view. Every minute the utilities inform all the agents, the suggested power level and an incentive rate they can offer. In section III, we assumed that the potential of demand program is %, which means that every load is a responsive load. In fact, some loads cannot be shut down entirely even if the energy price is extremely high. Fig. 5 shows that a parking deck for PHEVs follows demand response programs %. Comparing the total energy cost under demand response with the total energy cost without demand response during next 10 minutes, each agent (parking deck) makes a decision on its own. Also considering the local energy policy such as minimum power supply at that instant, agents are able to decide whether to follow the demand response event or not. Load (% ) Original Load Profile 30 Load (%) Load Profile After DR Fig. 5. Original and Optimal Energy Consumption by PHEVS at Parking Deck A. Some parking decks do not take part in demand response programs at all by comparing the total energy cost under demand response with the total energy cost without demand response during next 10 minutes. Load (% ) Original Load Profile L o a d (% ) Load Profile after DR Fig. 6. Original and Optimal Energy Consumption by PHEVS at Parking Deck B Some parking decks follow demand programs at a specific time only. L oad (% ) Original Load Profile 30 Load (% ) Load Profile After DR Fig. 7. Original and Optimal Energy Consumption by PHEVS at Parking Deck C In order to test the energy allocation to PHEVs in real-time with utility constraints and varying inputs, we also have applied Monte Carlo simulation to model phenomena with predictable uncertainty in inputs [20]. Monte Carlo simulation makes possible to simulate and monitor the real-world parking deck scenarios under a certain energy usage optimizations. In this case, we take parking deck C for example. In initial work [8-9], we have developed a simulator to help develop effective iems algorithms for PHEVs at a municipal parking deck. Fig. 8 shows the simulator for five charger station which has been used for testing and analysis of the proposed algorithms. Fig. 9 shows the PHEVs plug-in time at parking deck C. The PHEVs SOC at plug-in is shown in Fig. 10 and Table 4.

5 Table 4. Range of SOC_In with the percentage anumber of vehicles for Optimal Allocation for SOC Maximization. 5 SOC at Plug-in % Number of vehicles Plot of Vehicle States of Charge at Plug-out 0.3 Fig. 8. Simulink Based Testbed for 5 PHEV Chargers [8] Distribution of Plug-in Time Percentage of Vehicles Percentage of Vehicles Time of Plug-in (hrs) Fig. 9. PHEVs Plug-in Time at Parking Deck C Plot of Vehicle States of Charge at Plug-in State of Charge Fig. 11. PHEVs SOC at Plug-out in Parking Deck C. Table 5. Range of SOC_Out with the percentage anumber of vehicles for Optimal Allocation for SOC Maximization SOC at Plug-out % Number of vehicles Percentage of Vehicles State of Charge Fig. 10. PHEVs SOC at Plug-in in Parking Deck C In optimal allocation for SOC Maximization algorithm results, most PHEV leave a parking deck with SOC 55% - 85%. The simulation results are summarized and listed in Table. 3. V. CONCLUSION AND FUTURE WORK This paper presents the motivation and the implementation of the iems algorithm in Matlab/Simulink. We also discuss the effect of demand side management programs on the energy consumption pattern in order to avoid grid instability and maximize the financial benefits for both customers and utilities. The goal of the further study is to fully implement the demand side management programs into the existing FREEDM testbed. Our ultimate goal is to optimally allocate the power energy to PHEVs at a municipal parking deck over

6 6 a large number of operating scenarios with considering multiobjective optimization and demand side management through local information with a limit amount of communication availability. VI. REFERENCES [1] "Hybrid Sales Soar 38 Percent In U.S.", CBS NEWS, Detroit, April 21, 2008, [Online] Available: 4.shtml [2] M. Duvall and E. Knipping, Environmental Assessment of Plug-in Hybrid Electric Vehicles, EPRI, July [Online] [3] K Sikes, T Gross, Z Lin, J Sullivan, T Cleary, and J Ward, "Plug-In Hybrid Electric Vehicle Market Introduction Study: Final Report ", ORNL/TM-2009/019, U.S. Department of Energy, January 2010, [Online] Available: [4] S. Shao, M. Pipattanasomporn, and S. Rahman, "Challenges of the PHEV Penetration to the Residential Distribution Network", in proceedings, IEEE Power and Energy Society General Meeting, 2009 [5] Michael Kintner-Meyer, Kevin Schneider, and Robert Pratt, Impacts Assessment of Plug-In Hybrid Vehicles on Electric Utilities and Regional U.S. Power Grids Part 1: Technical Analysis, PNNL Report, Nov 2007, [Online]. [6] Stanton W. Hadley, Impact of Plug-in Hybrid Vehicles on the Electric Grid, ORNL Report, Oct 2006 [Online]. [7] Ted O Shea, Ramping Up Efficiency, [8] P. Kulshrestha, L. Wang, M.-Y. Chow, and S. Lukic, "Intelligent Energy Management System Simulator for PHEVs at Municipal Parking Deck in a Smart Grid Environment," in Proceedings of IEEE Power and Energy Society General Meeting, Calgary, Canada, (invited) [9] P. Kulshrestha, K. Swaminathan, M.-Y. Chow, and S. Lukic, "Evaluation of ZigBee Communication Platform for Controlling the Charging of PHEVs at a Municipal Parking Deck," in Proceedings of IEEE Vehicle Power and Propulsion Conference, Dearborn, Michigan, U.S.A, Sept 7-11, [10] U. S. Department of Energy, Energy Policy Act of [11] C Gerkensmeyer, MCW Kintner-Meyer, and JG DeSteese, "Technical Challenges of Plug-In Hybrid Electric Vehicles and Impacts to the US Power System: Distribution System Analysis", PNNL-19165, U.S. Department of Energy, January 2010, [Online] Available: rts/pnnl pdf [12] M. Fahrioglu and F. L. Alvarado, "Designing incentive compatible contracts for effective demand management," Ieee Transactions on Power Systems, vol. 15, pp , Nov [13] A. H. Albadi and E. F. EI-Saadany, "Demand response in electricity markets: An overview," in proceedings, IEEE Power Engineering Society General Meeting, 2007, Vols 1-10, pp [14] D. S. Kirschen and G. Strbac, Fundamentals of power system economics, 1 ed.: John Wiley & Sons Inc., [15] The Independent Electricity System Operator (IESO), ly Ontario Energy Price. [Online] Available: [16] The Independent Electricity System Operator (IESO), Adequacy Report. [Online] Available: [17] Y.H. Song and A.T. Johns, of Fuzzy Logic in Power Systems, Part 2: Comparison and Integration with Expert Systems, Neural Networks and Genetic Algorithms, IEE Power Engineering Journal, pp , Aug [18] C.C. Lee, Fuzzy Logic in Control Systems: Fuzzy Logic Controller, Parts I&II, IEEE Transactions on Systems, Man and Cybernetics, vol. 20, no. 2, pp , Mar./Apr. 19. [19] Q. B. Dam, S. Mohagheghi and J. Stoupis, "Intelligent Demand Response Scheme for Customer Side Load Management," 2008 Ieee Energy 2030 Conference, pp , [20] W. Su, and M.-Y. Chow, "Evaluation on Intelligent Energy Management System for PHEVs Using Monte Carlo Method", 2010 (Forthcoming) VII. BIOGRAPHIES Wencong Su is currently working toward Ph.D. degree in the Department of Electrical and Computer Engineering at North Carolina State University. He received B.S. with distinction in Electrical Engineering from Clarkson University in 2008 followed by a M.S. in Electrical Engineering from Virginia Tech in He also worked as a R&D engineer intern at ABB U.S. Corporate Research Center in Raleigh, NC, from May 2009 to August His current research interests are Microgrid modeling and simulation, distributed control, and Intelligent Energy Management System for Plug-in Hybrid Electric Vehicles. Mo-Yuen Chow received the B.S. degree from the University of Wisconsin, Madison, in 1982 and the M.Eng. and Ph.D. degrees from Cornell University, Ithaca, NY, in 1983 and 1987, respectively. Upon completion of the Ph.D. degree, he joined the Department of Electrical and Computer Engineering, North Carolina State University, Raleigh, and has held the rank of Professor since His core technology is diagnosis and control, artificial neural network, and fuzzy logic with applications to areas, including motors, process control, power systems, and communication systems. He has established the Advanced Diagnosis Automation and Control (ADAC) Laboratory at North Carolina State University.