REDUCE PEAK-TIME ENERGY USE BY DEMAND BIDDING PROGRAM IN IRAN

Transcription

1 REDUCE PEAK-TIME ENERGY USE BY DEMAND BIDDING PROGRAM IN IRAN Hossein DELAVARIPOUR Amir KHAZAEE Meran GHASEMPOOR Hossein HOOSHMANDI ABSTRACT Demand bidding program (DBP) is one type of demand response programs tat encourages large consumers to cange teir energy consumption pattern and reduce teir peak to increase system reliability. Wit implementation of advanced metering infrastructure (AMI) system in distribution network, DBP as been adopted in practice by Masad Electric Energy Distribution Company (MEEDC) as a system operator. MEEDC s DBP is a risk-free and internet-based bidding program tat offers te candidate customers for voluntarily reducing power wen a DBP event is called. Tis program is implemented for large consumers wit a demand of 100 kw or greater aims to reduce peak demand of day in summer. Details of te program and te implementation issues are explored. INTRODUCTION In 009, seven sites were selected from seven cities in Iran to run large-scale, pilot projects on te smart grid and smart community. Masad is one of tese cities in wic Masad Electric Energy Distribution Company (MEEDC) as te responsibility to implement smart grid road map. Demand side management (DSM) tecnology is one of te major goals in te smart grid projects. Based on te traditional functions, te smart grid DSM as new contents including automation, demand response [1], smart consume, remote energy efficiency monitor control, energy efficiency, power generation, and so on. Future smart grid integrates demand response programs, were consumers play principle roles in optimizing energy consumption profile. Demand bidding program (DBP) is one type of demand response programs tat offers customers incentives for reducing demand wen called by system operator to increase system adequacy [,3].It is wort noting tat te demand bidding is typically presented for purcase allocation problem in electric energy market [4-6]. In connection wit demand response concept, demand bidding for consumers is relatively new program wic involves no competition and risk-free approac to reduce energy consumption of large consumers. Tese two concept of demand bidding are quite different in terms of features and purposes. Demand bidding program for consumers is also useful for system operator due to improving efficiency of distribution and energy price stability. It as been recently adopted in practice by Soutern California Edison (SCE) [7] and Pacific Gas and Electric Company (PG&E) [3]. One of requirements to implement DBP is AMI system [8] in distribution network. Tis system is implemented in Iran. A new version of te DBP is developed and implemented in practice by MEEDC as a system operator wic will be discussed in details in tis paper. Masad Electric Energy Distribution s Demand Bidding Program (MEEDC s DBP) is a risk-free and Internet-based bidding program tat offers te industrial customers for voluntarily reducing power wen a DBP event is called. If te candidate customer can reduce power on event day wen a DBP event is activated, it will be rewarded. On te oter and, if te candidate fails to reduce te energy according to te requirement, tere is no financial penalty. Tis risk-free feature of DBP attracts most of te large-scale customers. Masad is te second major city of Iran wit more tan 1.4 million of consumers [9,10]. As sown in Fig. 1, industrial consumers (wic contained less tan 1% of wole consumers in Masad) approximately consumed 4% of final energy consumption in 015. Until now, more tan 700 largescale candidate customers wit a demand of 100 kw or greater (during te preceding 1 monts) equipped wit or witout emergency generation units, are participating in MEEDC s DBP. Fig. 1: Electricity consumption in eac consumer groups in Masad Te participation days (DBP events) would be predicted previously based on load forecasting algoritms and would be notified before 14:00 PM of te event day for participating during 4 ours of te day event. Te customers receive te reward if te minimum amount of teir energy reduction meet 15% of teir baseline. In tis paper, we discuss te DBP tat is currently used in practice and formulate te corresponding matematical optimization problem. Specifically, te problem is to determine te optimal bid tat maximizes te expected reward subject to requirement of te operator. CIRED 017 1/5

2 DEMAND BIDDING Te Demand Bidding program is optional for customers wit billed maximum demand of 100 kilowatts (kw) or greater during te past year. In one case, te candidate customer can coose te amount of energy reduction in terms of bidding value for a specific bidding event wit two options. 1) Manually bid after receiving a day-aead event alarm, in oter word, customers may submit only one bid for eac notification and indicate te amount of kw it will reduce for eac our of te event. ) Standing bid wic allows customers to bid at beginning of te participation period and applying it to future events. In anoter case, te biding value is determined by system operators or demand response aggregators. If te actual amount of energy drops to certain requirement, te customer will be rewarded. On te oter and, for te lack of joint participation in reducing te energy in te event no financial penalty is considered. Tis risk-free feature attracts interest of large customers. In DBP, to calculate reward a minimum amount of energy reduction is considered wic can be determined wit different metods [11]. In one metod, te amount of actual energy reduction must drop between certain percentages of te bid amount. In anoter one, a Fix amount per kw is considered as minimum bid requirement during any ours of te event. Event period is generally four ours associated wit peak condition in a single day event. It sould be noted tat cooperation will be investigated only during tis period following a call event. An event is called in te following emergencies: According to forecasts by te network operator of te total demand for te day-aead Prediction a rise in temperature for te day-aead Lack of adequate productive resources for te dayaead Te implementation requirements of te program are as follows: Network sould be equipped wit AMI system wic allows to measure and record data in 15- minute intervals to monitor te consumption beaviour of end-user. Now all large customers in Masad are equipped wit smart meters and consumption data in te form of reports and graps are available for display in a Web application. Messaging system is required to confirm acceptance and rejection of participation in te event. Candidate customers must ave access to te Internet or communication tools in order to send/receive alarm message and determine biding value. It is noteworty, te customer is responsible for correct sending and receiving of messages. In oter word, candidate customers must always ceck te website. Demand Response Providers / Aggregators (DRP/AGG): Tese entities are required to review te existing potentials in rewriting program, adjusting te contracts, registering customers, supporting te implementation of programs and billing. IMPLEMENTATION OF MEEDC S DBP Large customers wit a demand of 100 kilowatts (kw) or greater in any 3 monts during te preceding 1 monts are eligible to enroll and participate in DBP. Customers wo are selected for participation in DBP are encouraged to reduce teir energy consumption during te events tat occur in summer. In tis way, demand response provider (DRP) entity of MEEDC may dispatc at its discretion, to one or more customers, a day-aead event notification by 14:00 PM. DRP will dispatc only one event per day for a minimum of two ours and a maximum of four ours during 11:00 AM to 15:00 PM. Total ours of te event, during te summer (te period of cooperation) will be less tan 00 ours. Customers must reduce energy consumption more tan 15% of teir ourly baseline value. A valid customer specific energy baseline is calculated by averaging maximum load during two mont prior to beginning participation period. For customers wo are equipped wit backup generating resources suc as diesel generator and CHP, MEEDC provides some benefits suc fuel subsidies to make te possibility of biding optimal amount of energy reduction and so more profits. Customers can provide a part of teir base load during event by using tese resources. Tis makes better situation for customers to scedule optimal bidding value. In addition to large industrial customers wo are te best candidates for DBP implementation, otels, city centers, entertainment centers and offices are proper customers for inserting to tis program [11]. Generally tese customers are equipped wit emergency backup suc as diesel generators, CHP and solar generators. Insertion of tese resources in event ours causes reduction of energy consumption wit minimal cange in teir consumption pattern. If te customers aren t equipped wit tese resources, tey can account on managing teir air conditioner, ligtening, eater and cooler systems for participating in DBP. Reward or consumption credit is determined by DRP/AGG based on Rial (R) (Iran's currency) per kw. Tis credit can be differently determined by DRP/AGG discretion according to customer activities, and network condition in te view of reliability and increase of generation cost. For example, in California tis amount sets on 0.5 $/kw by PG&E and SCE companies wic is fixed over participation period for teir load zones. In MEEDC tis reward can be assigned in constant amount 500 R/kW wic is sum of low load, mean load and full load tariffs (TOU tariffs) for day-aead events. DATA MINING As is mentioned previously, one of te requirements for CIRED 017 /5

3 proper implementation of demand side management is AMI system and data mining. In tis regard, MEEDC designs a website-based real-time monitoring system to observe participations of customers carefully. Meanwile, it is possible tat customers can follow teir operations and acquired rewards. Some important duties of te web service are as follow: Notification of participants: based on participation contract, te participants must be notified to te participants before 14:00 o clock of te day before te event day. Tere were 3 types of notification: , SMS and web service. Load data of participants: consumption and demand of electricity data for eac participant was monitored in order to find out te level of participation. Participant portal: eac participant ad a password secured portal and participants are able to get teir ourly load and consummation data. Tis opportunity enables te participant for cost saving planning. Real-time price management: tis web service is also as te ability of real-time calculation of incentives and it is accessible for bot participants and operators. Fig. sows mining tools for observing customer performance. In tis figure, te columns present te amount of demand in event ours of one day. Red colour illustrates te lack of te customer cooperation in specific our. For instance, Fig. 3 sows te pattern of daily consumption of a customer and its response to requirement during a specific event. In tis case, customer as pledged to reduce its consumption to less tan te appointed requirement. As previously mentioned, customers wo equipped wit emergency power generation system (mostly diesel) are able to enter tem during event time and receive rewards. Fig. 4 sows load profile of a otel tat can reduce its energy consumption to less tan requirement wit starting its Fig. : mining tools for observing customer performance diesel generator. In tis participation, otel could auction approximately 350 kw from its consumption. DEMAND BIDDING FORMULATION In anoter study, tis paper addresses a computational approac to elp te customers in reacing maximal Fig. 4: load profile of a otel wit starting its diesel generator during a specific event day benefits subjected to DBP. Tis section presents an optimization problem including maximum reward as fitness function and energy reduction band as problem constraints. By solving tis problem, amount of reduction (bid) will be calculated optimally wic is beneficial for customers in improving expected reward. It is notable tat tis formulation is useful for customers wo can scedule consumption devices in order to cange probability distribution function. In oter words, customers can ave direct control on teir consumption devices computationally. In tis way, consumers can use teir emergency generation resources wic are more controllable. However, tis formulation can also be used non-sceduled loading. Te optimization problem is formulated as follow, Te Etb t.l max ( Et x E t ).r t.f ( x t )dx t (1) tb t. bt t1 l.b b.b t t t were, b t is bid amount at our. f(x t ) is probability distribution function (pdf) of total customer energy consumption x t during our t. Et denotes mean energy consumption during our t witout load sceduling. r t is pecuniary reward per unit of energy reduction. T e is total demand bidding event ours. Respectively, l and are minimum and maximum amount of reduction eligible to receive reward. Tey are defined in terms of a fraction or percentage of te bid. In (1) te parameters, r t, T e, l and are given by te MEEDC/DRP and b t is cosen by te otel. Et and f(x t ) are defined by te beaviour of customer energy usage. Note tat te above optimization can be applied wit and witout load sceduling. Witout load sceduling te amount of bid may not be tuned optimally. It is notably tat b t is independent of f(x t ) wit load sceduling, will be CIRED Fig. 3: 017 pattern of daily consumption of a customer 3/5 and its response to requirement during a specific event day

4 te respective pdf of customer energy consumption wit load sceduling. In practical case, f(x t ) is typically unknown. Terefore, an empirical distribution from measurement migt be used instead. If large enoug data were recorded in te past, we can estimate te required distribution of total energy consumption using a statistical metod. In many cases, te energy consumption in a building can be suitably modelled by a Normal distribution according to te measurement [1]. In particular, Normal distribution provides a good fit during ig load ours [1]. To simplify te optimization is done for eac our of te demand bidding event ours based on defined function () and using numerical derivative metod [13] wic allows us to determine te optimal bid. (b Et b t.l t ) E t t t t t b t. ( E x ).f ( x )dx () d (b t ) / db t b b * 0 t t Wit considering Normal pdf ( x N(En, n ) ) during event our t, te optimal bid in corresponding t is obtained by 4( l ) n Et E n ( Et E n ).ln( ) * b l l t (3) l were: E n is mean or expectation of te distribution. σ n is standard deviation. σ n is variance. * Te expected reward at our t based on bt is calculated as follows, * E t En n v u DB t ( )( erf (u ) erf ( v )) ( e e ) (4) * * Et En b t. Et En b t.l u, v n n erf ( x) ( / ) x t 0 e dt Witout load sceduling, Et E te optimal bid in (3) n reduces to * 4n t b.ln( ) (6) l l In tis situation expected reward at our t is found by, l ln( ) ln( ) * n l l l l t DB ( e e ) (7) As can be seen, te reward in comparison wit (4) can be assessed at a lower value witout sceduling. By adjusting E n and σ n of pdf, customer can acieve to furter reward. CONCLUSION Te demand bidding program is optional for customers wit billed maximum demand of 100 kilowatts (kw) or greater during te past year wo voluntarily commit to reduce teir energy consumption for at least four consecutive ours during an event. In tis paper, te procedure of MEEDC s DBP is presented and its impact on load profile is investigated. In addition, tis paper offers a computational approac of DBP to get te maximum reward to encourage more customers. Te MEEDC s DBP could ave 61 MW reduction in a specific our of an event day in summer. REFERENCES [1] D.S. Kirscen, G. Strbac, P. Cumperayot and D. de Paiva Mendes, 000, Factoring te Elasticity of Demand in Electricity Prices, IEEE transaction on power systems, Vol. 15, No., [] K. Zare, M. Parsa Mogaddam and M.K. seik al eslami, 007, Large Consumer s Decision Making to Cost Reduction in Real Time Power Market, UPEC, England. [3] Pacific Gas and Electric Company, 014, Electric sceduling E-DBP, San Francisco, California. [4] Y. Liu and X. Guan, 003, Purcase allocation and demand bidding in electric power markets IEEE Trans. Power Syst., vol. 18, [5] J. Saebi, J. Moammadi, H. Taeri and S.S. Nayer, 010, Demand Bidding/Buyback Modeling and Its Impact on Market Clearing Price, IEEE International on Energy Conference and Exibition (EnergyCon). [6] G. Strbac, Daniel S. Kirscen, 1999, Assessing te competitiveness of Demand side Bidding IEEE Transaction on Power System, Vol. 14, no. 1, [7] Soutern California Edison, 016, Demand bidding program, California, Available: ttp:// [8] R.R. Moassel, A. Fung, F. Moammadi and K. Raaemifar, 014, A survey on Advanced Metering Infrastructure, Electrical Power and Energy Systems, Vol. 63, [9] P. Tarasak, C.C. Cai, Y.S. Kwok and S. Wa O, 014, Demand Bidding Program and Its Application in Hotel Energy Management, IEEE Transaction on smart grid, Vol. 5, No., [10] A. Kazaee, M. gasempour, H. Hoseinzade and H. Hoosmandi, 015, Demand management program for large industrial consumers by using AMI system, 3rd International Conference on Electricity Distribution, CIRED, Lyon. [11] A. Kazaee, M. gasempour, 013, Distribution loss minimization: a case study in a commercial section in Masad, rd International Conference CIRED 017 4/5

5 on Electricity Distribution, CIRED, Stockolm. [1] L. Pederson, 007, Load modelling of buildings in mixed energy distribution systems, Doctoral Tesis, Norwegian Univ. Sci. Tecnol., Trondeim, Norway. [13] I.S. Gradsteyn and I.M. Ryzik, 007, Table of Integrals, Series, and Products, 7t ed. Burlington, MA, Elsevier, USA. CIRED 017 5/5