Retail Pricing and Day-Ahead Demand Response in Smart Distribution Networks

Transcription

1 Intelligence Systems in Electrical Engineering, 4 th year, No. 4, Winter Retail Pricing and Day-Ahead Demand Resonse in Smart Distribution Networks Shaghayegh Yousefi 1 and GholamReza Yousefi 2 1 Faculty of Electrical and Comuter Engineering, Tarbiat Modares University, Tehran, Iran yousefi_sh313@yahoo.com 2 Deartment of Electrical and Comuter Engineering, Isfahan University of Technology, Isfahan, Iran yousefi@cc.iut.ac.ir Abstract : This aer focuses on day-ahead (DA retailing for fixed and Time-of-Use ( rice taker customers and DA real time ricing for active customers who articiate in short-term markets. Customers resonse to the offered hourly rices are modeled using an hourly accetance function which includes decreasing linear robability density functions based on the hourly imum and imum retail rices allowed by market regulators. Furthermore, the retailer offers its active customers to articiate in the DA demand resonse rogram and voluntarily reduce their real time consumtion for offered incentives. Numerical studies reresent the effect of imlementing demand resonse rograms on the total benefit of retailing. Keywords: Day-ahead retail market, Demand resonse, Hourly accetance function, Pricing, Smart ower grid. Nomenclature 1 ( P acc Accetance of the real time rice Benefit resulted from selling electricity to the fixed rice taker B fix B B B ( P EP + DR Fix INC P h th hour ($ customers at Benefit resulted from selling electricity to the rice taker customers at h th hour ($ Benefit resulted from selling electricity to the real time rice taker customers at th Obtained benefit through energy rocurement for active customers and imlementing DADRP ($ Hourly load consumtion of fixed rice taker customers (MWh Retailer s share from the hourly incentives aid by ISO ($ Offered real time tariff at h th hour ($/MWh PF Penetration factor (% ( P h hour ($ Hourly load consumtion of real time rice taker customers as a function of electricity rice (MWh Hourly load consumtion of 2. Submission date: 10, 04., 2012 Accetance date: 15, 10, 2012 Corresonding author: Deartment of Electrical and Comuter Engineering, Isfahan University of Technology, Isfahan, Iran rice taker customers (MWh 1. Introduction Smart ower systems lean on four fundamental factors: Security, Quality, Reliability and Availability (SQRA to meet the electrical needs of digital society based on convergence of energy, communication, internet, and electronic commerce [1]. Equied with Advanced Metering Infrastructure (AMI, the smart ower grid follows demand resonse (DR rograms based on dynamic and time-differentiated retail rices which roduce large amounts of rice resonsive demand and result in significant SQRA and economic benefits such as electricity rice reduction, transmission lines congestion resolving, security enhancement and imrovement of market liquidity [2],[3]. Electricity markets and customers articiation in DR rograms have been focused by many researches in recent years [2]-[10]. References [9]-[17] have studied ricing strategies and subsequent reaction of customers due to changes in the offered retail rices. Modeling customer resonse to the both grous of timebased and incentive-based DR rograms has been resented in [13]. Ref. [14] has roosed a load management strategy that considers the ower-urchase of energy of the consumers in a real-time ricing DR rogram otimizing the negotiation between the consumer and retailer. Considering DR atterns for different consumer tyes, rice elasticity matrices have been

2 Retail Pricing and Day-Ahead Demand Resonse in Smart Distribution Networks 24 develoed in [15] to accurately model consumer behavior. Clustering similar demand atterns of customers has been addressed in [16] using large-scale sets of electric demand rofiles. In a comlete cometitive retail market, customers may have the right to switch to more rofitable ricing rograms offered by other retailers even in a short-term market such as DA one. Reference [10] has focused on fixed ricing attern for monthly bilateral contracts and has modeled the customer resonse using a tye of market share function. Based on the market share function emloyed in the recent aer, an hourly accetance function has been roosed in [17] to model customers demand in resonse to the varying hourly rices. Reference [11] has resented an accetance function based on the accetable energy costs for different clusters of customers. The latter model does not reresent the ercentage of the load, which customers may decrease, or urchase from other retailers according to the offered rices. A linear accetance function has been roosed in [12] secifying the robability of acceting the offered Time-of-Use ( rices by the endusers. Retail energy roviders should have recise evaluations from customers demand as well as the wholesale rice of the electricity in order to run their retailing activities in an economically otimized manner. Consequently, different forecasting aroaches have been resented in many researches [18]-[20]. Ref. [18] has roosed several data ing aroaches for electricity market rice classification. Short term multile load forecasting (STMLF and the use of anthroologic and structural data within STMLF have been resented in [19]. Ref. [20] has focused on short-term gross annual electricity demand by alying fuzzy logic methodology using general information on economical, olitical and electricity market conditions. The restructured ower market rovides divergent classes of customers with the requisite electricity ower at some distinctive rices including redetered fixed and time-of-use rates and also hourly varying cost. This aer models Day-Ahead (DA retailing oerations of a retailer for its customers who adot three ricing atterns i.e. fixed ricing, and Real Time ( ricing rograms. The fixed and rice taker customers urchase energy at redetered rices while rice taker customers ay their electricity use at hourly varying rices which are offered by their resective retailer. Hereinafter, the former two grous of customers are referred to as inactive customers and the customers in the third class are called active customers. The exected cometition among energy roviders in a cometitive retail environment affect the offered retail rices to the active customers who kee the retail market under surveillance and may decrease their electricity urchase from their current retailer through suly all or a art of their electricity demand from other resources in order to manage the corresonding electricity bills. Here, the focus is on DA real time ricing as a time-based DR rogram for active customers and Day-Ahead Demand Resonse Program (DADRP as an incentive-based DR scheme, both from the retailers resonsibilities. This aer does not focus on redicting rocedure. Accordingly, it is considered that the retailer emloys artificial neural networks (ANNs as a forecasting aroach to redict customers demand as well as wholesale electricity rices with u to %10 error of rice forecasting. As in a cometitive retail market, the rocedure addressed here allows these customers to adjust their effective demand according to the real time retail rices and/or change their retail energy rovider even in a daily basis. Consequently, this aer rooses an hourly accetance function. In order to reresent the rational behavior of customers, the roosed accetance function is comosed of decreasing linear robability density functions suorted by imum and imum DA retail rices which are allowed by market regulators. Furthermore, in this aer the retailer lays the role of an enrolling articiant who rooses its active customers to articiate in a DADRP and reduce their energy consumtion. DADRP allows customers to submit an offer concurrent with the DA energy market to curtail electricity consumtion for the following day. The retailer forecasts DA wholesale rices. The redicted rices indicate the aroximate imum values for the rices which would be offered to ISO for the curtailed load by DADRP articiants. The retailer suggests some rices less than the redicted values to articiant customers in order to hel customers in making iment offers of load reduction. The remaining arts of the aer are structured as follows. Section 2 resents the retailer s interactions with other market layers. Section 3 discusses energy rocurement for inactive customers. Section 4 is on the energy

3 Intelligence Systems in Electrical Engineering, 4 th year, No. 4, Winter rocurement and retail ricing for active customers. Section 5 is assigned to the dayahead demand resonse rogram and section 6 reresents numerical studies. Finally, section 7 concludes the aer. 2. Retailer s interactions with other market layers Smart distribution networks suly their customers with electricity of different quality levels and ricing atterns. In a real world cometitive retail market there are some cometitive retailers which try to artake in the retail market as more as ossible. However, in this aer we focus on oeration of one retailer that has an indirect cometition interaction with others via sensitivity of customers to the roosed hourly rice signals. A retailer is an economic enterrise which does marketing, wins electricity customers and encourages them to hold next contracts offering various demand resonse rograms so as to gain rofit. The smart ower grid follows DR rograms to establish bidirectional interrelations with customers in order to imrove the system oeration state, utilizing demand-side otential. DR rograms are conducted in two grous of incentive-based and time-based rograms [8]. One of the key roles of the retailer is to know local consumtion market comletely to be the final winner of the customer absortion game. In order to analyze different grous of customers, we suose the retailer categorizes its customers to residential, commercial and industrial customers into fixed rice takers, and real time rice takers grous. Fig. 1 shows the interactions between a retailer and other market layers in a smart distribution network. Distribution system comany (DISCO is resonsible for calculating and managing SQRA indices in distribution networks and regulating distribution market, to otimize system's erformance. The retailer urchases the demanded energy from the DA wholesale market and serves its customers in cooerating with DISCO. All insection and maintenance oerations of the system are erformed by DISCO, which is aware of distribution system conditions, its strengths, weaknesses and contingency eliation tools such as maneuver oints, etc. Fixed and rice taker customers, here called as inactive customers, wish to access to reliable, qualified, and inexensive electricity but not engaging in rice variations. They do not articiate in short-term contracts and the retailer is obligated for sulying their consumtion according to their contracts. The retailer should have a recise rediction of inactive customers energy consumtion. Other Retailers Real Time Price Takers Cometition Negotiation DISCO Cooeration Retailer Negotiation Negotiation Negotiation Price Takers Energy Resources Fixed Price Takers Fig. 1. The retailer s interactions with other market layers in a smart distribution network Real time rice taker customers usually urchase a ortion of their demand through bilateral contracts and then bid in day-ahead and sot markets to rocure their needed surlus electric ower. They wish to buy reliable and qualified electricity at dynamic rices in order to otimize electricity consumtion and manage their electricity bills. These customers, here called as active customers, are aware of cometitor retailers offers and have a good rediction about their hourly demand. They monitor other rice alternatives and different rovided services and usually articiate in a variety of demand resonse rograms. Retailers lan for their marketing activities based on the wholesale energy rices. In DA energy market, they should have an accetable rediction of DA energy rices, which directly affects the energy rocurement lans and retail ricing for real time rice taker customers. Here, it is considered that the retailer redicts the DA hourly wholesale rices with u to %10 error of rice forecasting. 3. Energy rocurement for inactive customers The retailer is aware of its different classes of customers. Here, the residential, commercial and industrial customers are classified into fixed, and rice takers grous. Let R Fix ( h and R ( h,h = 1, 2,..., 24 be the forecasted hourly load consumtion of residential customers who adot fixed and ricing atterns, resectively. Also, C Fix ( h,c ( h and I ( h demonstrate the forecasted hourly load consumtion of commercial and industrial customers who

4 Retail Pricing and Day-Ahead Demand Resonse in Smart Distribution Networks 26 articiate in fixed/ ricing atterns, resectively. As shown in Eq. 1, Fix ( h includes energy consumtion of customers from residential and commercial sectors. Fix Fix Fix = R + C (1 rice taker customers include customers from trile sectors of consug grous. The hourly load consumtion of these customers is formulated as: h ( = R + C + I (2 Here, it is considered that the retailer has sulied a ortion of reviously redicted consumtion of the inactive customers, tyically 0.9 of it, through long-term contracts. Therefore, the surlus consumtion which is necessary to be sulied in short term D h, can be obtained as: markets, ( = 0. 1 D + ( D D D Where (3 D 1 reresents the inactive customer s consumtion based on retailer s rior forecasts and D 2 is the last forecast of their energy consumtion at h th hour of the next day. The term D reresents each of fixed/ rice taker grous Fix Fix R h,c h,r h,c h,i h in searate demand forecasting rocedures for them. ( ( ( ( ( ( Considering w as the forecasted DA wholesale rice, the cost of energy rocurement for fixed rice takers would be Fix w. If the redetered fixed rice for residential and commercial customers be denoted by R C Fix and Fix, then the exected benefit through energy retailing for fixed-rice taker customers will be obtained as: Fix R B Fix = R Fix + (4 Fix C C h h Fix h h ( ( ( ( Fix The obtained benefit of retailing for rice takers deends on redetered eak, off-eak and valley rices for trile R R R,,, sectors of residential ( w o v C C C commercial (, o, v I I I (,,, and industrial o v customers. Here, it is considered that commercial and industrial customers ay C I the energy at the same rices = C I = and C I =. The forecasted and o o v v DA wholesale rice directly affects the cost of energy rocurement for the customers of ricing class ( w. Consequently, the retailer s benefit from energy rocurement for rice taker customers will be acquired such as: B R = w+ R + + I C ( C where, R v if h R R = o if h R if h C v if h C C = o if h C if h { valley} { off eak} { eak} { valley} { off eak} { eak} (5 The term R reresents redetered rates of electricity at eak, off-eak and valley hours for residential customers, while C includes these rices for commercial customers. 4. Energy Procurement and Retail Pricing for Active Customers This section is devoted to the energy rocurement and real time ricing for active customers. The offered hourly rices ( will be otimized so as to imize the exected obtained benefit through electricity retailing for this class of customers. Active customers urchase a ortion of their energy need in the DA market. In order to suly the active customers with electricity of hourly varying worth, the retail energy rovider offers otimum DA rices and exeriences subsequent demand side reactions. The customer s effective DA demand reflects their resonse to the offered rices. Therefore, as the

5 Intelligence Systems in Electrical Engineering, 4 th year, No. 4, Winter effective demanded loads deend on the offered rices, it is required to suly requisite electricity and detere real time rices for it, simultaneously. Let the demanded load of commercial and industrial customers who adot real time ricing C h and attern be denoted by ( ( I (, resectively. Therefore, the total amount of the effective demand of -rice taker customers is: ( = C ( + I ( A demand value of ( (6 includes the total load of real time rice takers comosed of commercial customers demand and industrial Accetance function The function of = F( has been addressed in [10]-[12]. The robability density function according to which the retailer sets its equilibrium rices is usually defined as the market share or accetance function [10]. Accetance function is a decreasing function in which by increasing the offered rice, the total confirmed demand would be decreased. Several factors such as long-term strategy of the retailer, risk attitude, the behavior of customers and the behavior of the cometitors must be considered in deteration of accetance function [21]. In [12], two decreasing linear functions have been introduced as the models of customers reaction against offered rices. ricing means to value electricity at two rices deended on time of energy consumtion while real time ricing is constituted from offering 24 hourly rices. Here, for each hour of DA real time ricing a linear accetance function is roosed based on the imum and imum retail rices allowed by indeendent system oerator (ISO which directly affect the customers resonses to the offered rices. If the imum and imum hourly retail rices be denoted by and, then a linear accetance function can be assigned to each offered rice such as: ( ( + ε acc h = (7 + ε The roosed accetance function deends on the imum and imum hourly retail rices, which are allowed by the market regulators. Maximum rice accetance by the active customers occurs at =. In other words, selling energy to the customers at the imum hourly rices satisfies them comletely. Fig. 2 shows the roosed linear accetance function for DA real time ricing. According to Eq. 7, acc( + ε = 0 and acc( = 1. It means that is the highest offered rice in which acc( is above zero and customers do not accet the offered rices beyond.minimum retail rices ( are the hourly rices at which the effective demands of active customers are 0 h. equal to their initial demands; ( Fig. 2. The roosed hourly accetance function Therefore, the active customers effective hourly demands will be as follows: h = h acc h = ( ( 0 ( ( ( ( + ε 0 h + ε (8 The offered hourly rice affects customers demand as it is demonstrated by Eq. (8. Offering = attracts the imum electricity demand while selling electricity at > results in loosing demand Benefit function Retailing benefit through sulying the effective demand of -rice taker customers will be obtained as: B = ( ( w (9 Neglecting the ignorable value ε and substituting Eq. 8 in Eq. 9 results in: B h = h h h 1 Probability ( 0 ( ( ( w ( + ε + ε Price ($/MWh ( ε + h (10

6 Retail Pricing and Day-Ahead Demand Resonse in Smart Distribution Networks 28 The hourly benefit function (Eq. 10 would B ( be imized by h which results in the otimum hourly rices as: ( = 2 1 ot + w (11 The robability of acceting otimum hourly rices would be as. ( ( w acc h = 0 5 and consequently, the retailer would receive active customers effective demand as the following: 0. 5 ot ( = 0 w ( (12 According to Eq. (10 and (12, the sequent retail benefit would be as the following: B ot 0 = 4 ( 2 w (13 Therefore, the otimum retail rices and retail benefit will be deendent uon the imum retail rices detered by ISO. 5. Day-Ahead Demand Resonse Program DADRP allows customers to submit an offer concurrent with the DA energy market to curtail electricity consumtion for the following day. The offer would secify a roosed rice in the range of 50 to 1000 $/MWh, the amount of curtailment (at least 100 kw er demand resonse asset, and imum duration (u to four hours over which the retail customer would be willing to reduce consumtion. The offers less than or equal to the DA market hourly clearing rices will be acceted by the ISO. According to the New-England energy market whose demands/rices data and retail regulations are utilized in this study, customers must enroll in DADRP through an enrolling articiant such as local utilities, cometitive electricity retailers and demand resonse roviders. Due to the focus of the aer on the retail rocedures, here the retailer lays the role of a trailblazer enrolling articiant who rooses its active customers to articiate in DADRP and reduce their energy consumtion. It also suggests some rices for the curtailed load to be offered by customers. Each demand resonse asset should decrease at least 100 kw of its average recorded consumtion. The ratio of offer load reduction amount ( L in DADRP to the whole DA demands of active customers is called here as enetration factor ( PF which is deendant to the ercentage of articiating customers and their hourly load reduction (Eq. 14. L PF = (14 0 Regarding the enetration factor, the obtained benefit through sulying not registered loads in DADRP can be udated as: ot 0 ( ( B h = 1 PF 4 2 (15 ( w If the hourly offers are acceted, the retailer will be aid the hourly day-ahead wholesale rice multilied by the offer reduction amount ( 0 PF w. However, it is u to the articiant customers and their retailer to work out an agreement on how the demand resonse ayments will be shared. Here, it is considered that the retailer owns %30 of DADR rogram s ayments. Therefore, the retailer s share from the incentive aid by ISO will be as: = PF 0 3 INC w 0. (16 The obtained benefit through energy rocurement for active customers together with imlementing DADRP will be as: EP+ DR B h = B h INC h (17 ( ( ( + Substituting Eq. (15 and Eq. (16 in Eq. (17 results in the total benefit obtained through energy rocurement for active customers and gaining incentives for load reduction in DADRP as follows: EP+ DR B h = h PF h ( 0( w( ( 1 PF w 4 ( ( 2 (18 6. Numerical Study In this study, the required data is extracted from the day-head of New England, Connecticut in May 2010 [22]. It is considered that the retailer sulies 30% of all Connecticut electricity demand of residential, commercial and industrial sectors which are categorized in three

7 Intelligence Systems in Electrical Engineering, 4 th year, No. 4, Winter grous of customers, i.e. fixed and rice takers as inactive and rice takers as active customers. The target day for the retail rocess is 26 May Table 1 demonstrates articiation share of customers in three ricing strategies. It is considered that 90% of demands in all sectors are sulied through bilateral long term. Therefore, the reder ortion of it (i.e. 10% should be sulied in DA market. Forecasting inactive customers energy consumtion, the retailer estimates the surlus electric energy to be rocured in DA market (Eq. 3. Real time rice taker customers send their demands to the retailer based on their load rediction. Fig. 3 shows requisite electric energy for three grous of customers through the next 24 hours. Table 1. Particiation of customers in three ricing strategies Inactive customers Active customers Particiant sectors Dynamic Pricing Fixed valley Off-eak eak Industrial Commercial Residential Portion (% Price ( /kwh Portion (% Price ( /kwh To be otimized Portion (% Price ( /kwh To be otimized Fig. 3. Requisite electric energy to be rocured for three classes of customers The retailer urchases energy in order to suly its inactive customers requisite electricity at redetered fixed/ rices. Fig. 4 shows the hourly cost of energy rocurement for them and consequent benefits.

8 Retail Pricing and Day-Ahead Demand Resonse in Smart Distribution Networks 2 Fig. 4. Costs and benefits in energy rocurement for the inactive customers (fixed and rice takers Retailing for -rice taker customers includes retail ricing through energy rocurement for them using DA wholesale market. Fig. 5 comares imum/imum DA retail rices with the offered retail rices which are obtained based on the roosed hourly accetance function. Fig. 5. Comarison of DA wholesale rice, imum, imum and offered retail rices Fig. 6. Active customers effective demand, hourly costs and obtained benefits through rocurement PF = 0 electric demands ( The retail energy rovider also offers articiating in DADRP to its active customers and suggests the forecasted DA wholesale rices as the load curtailment incentive rice. Here, it is considered that PF changes from 10% to 50% so as to analyze the benefits obtained by retailer through retailing and erforg DADRP simultaneously. Fig. 7 resents the total hourly benefit in case of different articiating rates. Without offering DADRP to the active customers, ( PF = 0 they would demand electric energy at otimum DA rices. Fig. 6 shows the effective demands as well as subsequent energy rocurement costs and benefits. Fig. 7. The effect of different articiating rates in DADRP on the total hourly benefit

9 Intelligence Systems in Electrical Engineering, 4 th year, No. 4, Winter Fig. 8 dislays daily benefit of DA retailing accomanied with DADRP involving 0-50% of active customers DA demand. Fig. 8. Daily benefit of DA retailing accomanied with DADRP As it can be seen in the figure, the total benefit obtained by retailer is comosed of energy selling benefit and incentives from ISO subsequent to imlementing DADRP. Greater articiating rates ( PF eventuate much more benefit for the retailer (i.e. benefitpf = 0. 5 = benefit PF = 0 7. Conclusion Smart grids are established based on bidirectional interactions between the ower system oerators and customers who may lay as distributed energy resources in form of distributed generators or volunteer demand reducers. Demand resonse rograms refer to mechanisms of encouraging customers to lay active roles in energy markets. In this aer, the retailer categorizes its residential, commercial and industrial customers in grous of fixed rice takers, and real time rice takers. The latter grou includes active customers who urchase a ortion of their requisite energy in short-term markets and articiate actively in demand resonse rograms. This aer discusses the roblem of DA retailing for fixed and rice takers and DA ricing for active customers. Accordingly, it models customers resonse to the otimum hourly rices and offer a articiation in DADRP roosed by retailer. An hourly linear accetance function is roosed in this aer which utilizes imum and imum DA retail rices allowed by market regulators to adjust hourly settings. Numerical studies demonstrate the effect of articiation rate on the retail benefit. According to the obtained results, if the retailer s share of incentives is 30%, its total benefit will grow as more customers articiate in DA demand resonse rogram. However, the roortion of incentive sharing between the retailer and customers, directly affect the resulted benefits and corresonding decisions. References [1] EPRI, Electricity technology roadma: 2003 summary and synthesis, meeting the critical challenges of the 21st century, EPRI reort no , [2] P. Centolella, The integration of rice resonsive demand into regional transmission organization (O wholesale ower markets and system oerations, Energy, Vol. 35, No. 4, , [3] H. A. Aalami, M. Parsa Moghaddam, G. R. Yousefi, Demand resonse modeling considering interrutible/curtailable loads and caacity market rograms, Alied Energy, Vol. 87, No. 1, , [4] A. J. Conejo, J. M. Morales, L. Baringo, Real-time demand resonse model, IEEE Trans. Smart Grid, Vol. 1, No. 3, , [5] L. A. Greening, Demand resonse resources: who is resonsible for imlementation in a deregulated market?, Energy, Vol. 35, No. 4, , [6] K. Zare, M. Parsa Moghaddam, M. K. Sheikh-El-Eslami, Electricity rocurement for large consumers based on Information Ga Decision Theory, Energy Policy, Vol. 38, No. 1, , [7] J. M. Yusta, H. M. Khodr, A. J. Urdaneta, Otimal ricing of default customers in electrical distribution systems: Effect behavior erformance of demand resonse models, Electric Power Systems Research, Vol. 77, No. 5-6, , [8] H. A. Aalami, M. Parsa Moghaddam, G. R. Yousefi, Modeling and rioritizing demand resonse rograms in ower markets, Electric Power System Research, Vol. 80, No. 4, , [9] K. Herter, S. Wayland, Residential resonse to critical-eak ricing of electricity: California evidence, Energy, Vol. 35, No. 4, , 2010.

10 Retail Pricing and Day-Ahead Demand Resonse in Smart Distribution Networks 32 [10] A. R. Hatami, H. Seifi, M. K. Sheikh-El- Eslami, Otimal selling rice and energy rocurement strategies for a retailer in an electricity market, Electric Power Systems Research, Vol. 79, No. 1, , [11] N. Mahmoudi-Kohan, M. Parsa Moghaddam, M. K. Sheikh-El-Eslami, An annual framework for clustering-based ricing for an electricity retailer, Electric Power Systems Research, Vol. 80, No. 9, , [12] S. A. Gabriel, A. J. Conejo, M. A. Plazas, S. Balakrishnan, Otimal rice and quantity deteration for retail electric ower contracts, IEEE Trans. Power Systems, Vol. 21, No. 1, , [13] Sh. Yousefi, M. Parsa Moghaddam, V. Johari Majd, Otimal real time ricing in an agent-based retail market using a comrehensive demand resonse model, Energy, Vol. 36, No. 9, , [14] J. Lujano-Rojas, C. Monteiro, R. Dufo- Lóez, J. Bernal-Agustín, Otimum residential load management strategy for real time ricing (P demand resonse rograms, Energy Policy, Vol. 45, , [15] N. Venkatesan, J. Solanki, S. Khushalani Solanki, Residential Demand Resonse model and imact on voltage rofile and losses of an electric distribution network, Alied energy, Vol. 96, , [16] E. C. Bobric, G. Cartina, G. Grigoras, "Clustering Techniques in Load Profile Analysis for Distribution Stations, Advances in Electrical and Comuter Engineering, Vol. 9, No. 1, , [Online]. Available: htt://dx.doi.org/ /aece [17] Sh. Yousefi, M. Parsa Moghaddam, V. Johari Majd, Agent-Based Modeling of Day- Ahead Real Time Pricing in a Pool-Based Electricity Market, Iranian Journal of Electrical and Electronic Engineering, Vol. 7, No. 3, , [18] D. Huang, H. Zareiour, W.D. Rosehart, N. Amjady, Data Mining for Electricity Price Classification and the Alication to Demand- Side Management, IEEE Transactions on Smart Grid, Vol.3, No.2, , [19] F. Javed, N. Arshad, F. Wallin, I. Vassileva, Erik Dahlquist, Forecasting for demand resonse in smart grids: An analysis on use of anthroologic and structural data and short term multile loads forecasting, Alied Energy, Vol. 96, , [20] S. Kucukali, K. Baris, Turkey s short-term gross annual electricity demand forecast by fuzzy logic aroach, Energy Policy, Vol. 38, No. 5, , [21] P. Kotler, G. Armstrong, Princiles of marketing, Prentice Hall, [22] Indeendent System Oerator (ISO New England. Hourly zonal information, [Accessed Nov. 2010], online: htt://