Outage Avoidance and Amelioration using Battery Energy Storage Systems

Transcription

1 Outage Avoidance and Amelioration using Battery Energy Storage Systems Tan Zhang, SIEEE, Stephen Cialdea, MIEEE, John A. Orr, LFIEEE, Alexander E. Emanuel, LFIEEE Department of Electrical & Computer Engineering Worcester Polytechnic Institute Worcester, MA USA Abstract-Planned and unplanned outages on distribution networks may cause major economic loss to the customers, and can result in substantial reparation payments by the utilities. This paper analyzes the use of a Battery Energy Storage System (BESS) to partially or completely avoid these outages. The analysis is conducted with a multi-day high demand scenario where the load demand at peak times exceeds the current capacity of the line. This would result in an outage unless the current from the substation can be reduced. Three basic scenarios representative of different degrees of sophistication of feeder automation were studied: A conventional feeder controlled by one or more reclosing breakers; a modern distribution feeder equipped with a limited number of remote controlled breakers that will disconnect non-vital loads belonging to residential, commercial and industrial customers and a modern distribution feeder equipped with numerous remote controlled breakers that will selectively disconnect commercial, industrial, and residential loads, based on their degree of criticality. This study concluded that the total or partial avoidance of outages by way of BESS usage is feasible; however, the use of "smart grid" control technology can provide more benefit than BESSs in the scenarios studied. Index Terms- Demand Shifting. Energy Storage. Outage. Smart Grid. I. INTRODUCTION Outages may have severe repercussions at sites where public safety and the environment are at risk such as sewage pumping or treatment facilities, mines, certain industrial loads, transportation, and computer server facilities [1, 2]. Such systems are generally equipped with standby generators or uninterruptable power supplies; however these devices may fail when needed. The most devastating form of power outage is the blackout. Depending on the topology of the network and the conditions that led to outage, the blackout may last only a few minutes, hours, or even weeks [3, 4]. The financial losses caused by outages are generally quite minor to residential customers, but may be significant to commercial and industrial (C&I) customers [5]. The mitigation of power outages is a major source of concern for the utilities. This paper addresses the use of Battery Energy Storage System (BESS) as a source of excess power available to the grid. The main role of the battery is to enable the shifting of energy purchasing from on-peak times. However, the BESS can also help provide total or partial ride-through a time interval when the demand on a transmission line, feeder or substation is excessive and may cause inadvisable thermal stress on transformers or cables. This paper also helps to demonstrate that for the designers of Smart Grids there is a major challenge and opportunity to use instrumentation that responds to developing power outages with selective disconnection of non-vital loads. II. BACKGROUND In a radial distribution system, the power is delivered from substation to the end users through dedicated feeders. Each feeder has a recommended apparent power limitation, S L. This limit is dictated mainly by the substation transformers; their type of cooling and thermal time constant has a strong effect on ability of transformers to withstand overloading for a determined time without a significant loss of life. In this study, a 15 kv class cable with an ampacity limit of 400 A, SL = 10 MVA, was considered. If the feeder current exceeds 400 A for a critical time, potential damages to substation equipment will occur [6, 7]. Under conditions of excessive demand, it is necessary to protect the substation and the feeder by implementing planned outages. In this paper, the Marginal Cost gives the relation between the electric energy price and demand as shown in Fig. 1. Fig. la presents values of Marginal Cost (MC) of electric energy, $/MWh, versus the demand power, MW, supplied to a region with a total load varying from 1600 MW to 7250 MW. Fig. Ib presents an analytical MC curve with similar price tendency for the energy supplied by one feeder described by equation Kw = ' 10-8 p ($jmwh) (1) where K w is the Marginal Cost of Electric Energy. p is the total active power (MW) supplied by the feeder. Financial support for this work was provided by Premium Power Corporation, National Grid, and the US Department of Energy /$ IEEE

2 :; , ,-----,-----y-----,. (a) ;;, >. t>.o... c:.> ! ,.;: , '" o U L. 5 C1l... - '" o U o o Power (MW) S, r ,.-----,---,-- ---,----,-----, "...,---- (b) r The PBmax is the maximum power sustained by the BESS for the duration T B without damage to its inverter/charger or a significant reduction of life span. The discharge capacity is: QB = PBmaxT B (2) The round-trip efficiency is: W BDCh 11B = (3) WBCh where W BCh= energy supplied to the BESS during the charging time, W BDCh= amount of energy delivered by the BESS under round-trip condition. III. LOAD VARIATION DATA PREPARATION The 24 hour demand curves considered in this study are represented in Fig. 3 with the following expression: pet) = Pm in + (Pmax - Pmin) {1- exp (1- (t:)2)} (4) Where the parameter b = 4 when 0 t 16 and 28 while 16 t 24. The curves are constrained within the extremes P min and P max' 11.O,----,----, ,..,.p----,---- mal o Power (MW) Figure I. Electric energy marginal cost vs. supplied power: (a) actual measurements [8] for a large grid with S< 9000 MV A, (b) best fit curves applied to a S<IO MVA feeder. Typical 24 hour demand curves are shown in Fig. 2 [9]. One learns from here that the load demand is varying hourly, daily and annually changing with the seasons. =. :ir '"' IOO 97" " 7"0 67" 600 "2" ""0 :>7" :>00 22" '''0 I 9'5..::,:.... Peok 98& M_b. POOk loot M_ I ---- N..... K""-. 7" 0.2.-*2-:>.. 567B :>-4"67 B9-. 0.LJ Figure 2. Load Demand Curves [9]. A.M. I P. M Time of day Although there are many parameters that help describe battery performance, four of them deserve special attention: PBmax (MW)=Rated charging/discharging maximum power, T B (h)=maximum discharging time under the rated power PBmax, QB (MWh)=Maximum discharging capacity, 11B=Round-trip efficiency.. ' ;,----;:ll-----f:16---1:':-o----.j Time (h) Figure hour Load Curves, SL = 10 MVA,PmaxlSL = 1.05 WIVA, PmlnlSL = 0.75 W IVA. The instantaneous load profile for a typical year is shown in Fig. 4. Each day has a load curve variation characterized by Pm in and Pmax. The peak load for the nth day was assumed to follow a uniform distribution based on the expression Pnmax = Pn'i[O O. 35rand(O, 1)] (5) where P n 'i defines the upper boundary of P n max' It is a curve that peaks during the summer and winter times, Fig. 4. One can see the maximum peak value for the 24 hour load curve is varying randomly in the range: (6) The yearly maximwn value of Pn'i occurs in July and is termed PM' P n min is computed in the same manner: O. 65Pn'i Pnmax Pn'i Pnmin = Pn::x[O.65 + O.35rand(O, 1)] (7) where P n ::x defines the upper boundary of P n min' In this study, it was assumed PM=lO.25 MW, and the max(pn::x)=8.5 MW. The weeks with the highest demand occur during the end of June and beginning of July with consequent possible planned outages.

3 By correctly using a BESS dedicated to a group of affected loads, depending on the size of the BESS, it is possible to ride through all or part of the high demand times as well as gain Differential Cost of Energy (DCE) in the meantime [10]. Assuming that the BESS nominal values are PBmax=0.5 W, T B=5 h and llb=85%, it is feasible to "ride through" during the entire Thursday outage period (Fig. 5) with an excess of energy estimated to be 0.5 Wh over the limit of 10 V A (Fig. 6). In this case there is enough energy stored in the BESS to supply the needed additional energy ==-----, Figure 4. Annual Load Variation. Note the potential outage condition during end of June beginning of July. 1 O.Of :a(IIiI_" IV. ETHODOLOGY There are three potential options that respond to outage conditions. A. The Conventional Approach The feeder exposed to overloading is disconnected using the main circuit breaker located at the substation bus. In this case, all the customers supplied by the feeder will experience the outage. If more than one circuit breaker is included in the radial feeder topology, then the outage can be limited to a group of customers with mixed types of loads. For example if the feeder has three circuit breakers, each controlling about 1/3 of the customers, it will be possible to disconnect only 1/3 of the customers at one time in order to avoid the outage. B. Smart Grid Method 1 If more circuit breakers are included, the system capability of outage limitation is greatly improved. On such case, it is possible to disconnect non-vital loads, belonging to all three types of customers and limit the reparation costs to an affordable value. C. Smart Grid Method 11 With the advancement of Smart Grid technologies, it will be possible to implement the remote control of low priority loads, owned mainly by residential customers and limit the reparation cost to a minimum value. V. SIMULATED EXAMPLE Zooming in the week of July 1st (Fig. 4), one will observe that for two days in row the maximum demand exceeds the limit SL =1OVA (Fig. 5) Time(h) Figure 6. Thursday Outage Avoidance Case For Friday's load demand (Fig. 5), the BESS's energy is insufficient to cover the entire excessive demand. Depending on the feeder's energy management method, two possible strategies are available and described in Fig. 7a and 7b that lead to a planned partial outage. A. Conventionally, the BESS will charge with the energy A = 2.94 MWh, and discharge A; = llba = 2.50 MWh avoiding outage and riding through during the time 9:30 A. to 5:50 P. (Fig. 7a). A complete outage will be experienced for the remaining time 5:50 P. to 9:40 P. when BESS cannot supply the demand A'3 when the main circuit breaker of the feeder is tripped. B. If a modem, Smart Feeder is exposed to the above condition, the BESS will charge with A = 2.94 MWh and discharge A; = llba = 2.50 MWh over the entire peak demand (10:00 A. to 9:20 P.). If the utility that operates a Smart Grid has the ability to remote control loads such as compressors, welders, heaters, as well as residential loads like air-conditioners devices, and swimming pumps. Such condition helps ride through the planned outages while minimizing the reparation costs as is illustrated in Fig. 7b. 6 5 M T w Th F Sat Sun Figure 5. Load for week at the beginning of July.

4 10.r-----.=w----'-----'-----'----' Time(h) (a) , (b) "n",,, "... "A"',... ; i ; Time(h) Figure 7. Friday Partial Outage Avoidance Cases. (a) Conventional Approach. (b) Smart Grid Method. In this example, the feeder's load was assumed mixed: 50% residential, 25% small commercial and industrial and 25% large commercial and industrial loads. Table I summarizes typical outage reparation costs. It is learned that for the residential customer, the reparation costs are relatively low when compared to the outage reparation costs for large or small C&I customers [5]. TABLEl AVERAGE OUTAGE REPARATION COST [5] S/kW Res SmallC&1 Large C&I 15 Min. 30 Min. 1 Hour 2 Hours 4 Hours 8 Hours $0.05 $0.60 $2.60 $3.95 $5.30 $5.60 $8.65 $16.01 $23.37 $48.91 $ $ $4.79 $7.46 $10.12 $17.96 $36.94 $68.36 Assuming that it is fair to impose on all customers the same burdens of bearing the inconveniences of the outage, then the outage will affect loads belonging to all three groups in proportion with their proliferation. The second Smart Grid Method limits the outage only to residential customers. The outage reparation costs can be calculated and compared using Table I. The following equation helps compute the total costs of the outage: ($) (8) where P av g = expected average outage demand power. Ko=average outage reparation costs ($/kw) for the unavoided outage time (h). The avoided/unavoided outage reparation costs as well as the daily DCE are summarized in Table II, III and IV assuming that the outage reparation costs remain constant for 8 h or a longer period of time. TABLE II A VOIDED OUT AGE REP ARA TION COSTS (THURSDAY) Feeder Type Resid. 50% Small Com.llnd. 25% Large Com.llnd. 25% DCE TOTAL ($) ($) ($) ($) ($) Conventional 28, , ,053 3, ,442 (One Circuit Breakers) Conventional 14, ,518 86,527 3, ,222 (Two Circuit Breakers) Conventional 9, ,679 57,684 3, ,814 (Three Circuit Breakers) Smart Grid Method I 227 3,832 1,384 3,000 8,443 Smart Grid Method II ,000 3,454

5 TABLE III AVOIDED OUTAGE REPARATION COSTS (FRIDAY) Feeder Type Resid. 50% Small Com.!lnd. 25% Large Com.!lnd. 25% DCE TOTAL ($) ($) ($) ($) ($) Conventional 1, ,736 81,801 2, ,223 (One Circuit Breakers) Conventional ,368 40,901 2, ,076 (Two Circuit Breakers) Conventional ,245 27,267 2,928 93,026 (Three Circuit Breakers) Smart Grid Method I 286 4,825 1,743 3,304 10,158 Smart Grid Method II ,304 3,875 TABLE IV UNA VOIDED OUTAGE REPARATION COSTS (FRIDAY) Feeder Type Resid. 50% Small Com.!Ind. 25% Large Com.!lnd. 25% TOTAL ($) ($) Conventional 28, ,261 (One Circuit Breakers-No BESS) Conventional 27, ,524 (One Circuit Breakers-With BESS) Conventional 14, ,131 ($) ($) 176, ,546 94, ,250 88, ,774 (Two Circuit Breakers-No BESS) Conventional 13, ,763 (Two Circuit Breakers-With BESS) Conventional 9, ,754 (Three Circuit Breakers-No BESS) Conventional 9, ,509 (Three Circuit Breakers-With BESS 47, ,626 58, ,182 31, ,084 Smart Grid Method I (No BESS) 504 Smart Grid Method I 218 Smart Grid Method II (No BESS) 1,008 Smart Grid Method II 437 8,515 3, ,076 12,095 1,333 5, , If the general outage cannot be avoided (as happened Thursday), and the substation main circuit breaker must be tripped, then the reparation costs will be a significant $680,442. If the feeder and some loads are equipped with dedicated circuit breakers or contactors, the outage can be limited to residential loads as well as the low priority customers only, in which case the Smart Grid method I will avoid the reparation costs of $5,443. The Smart Grid method II helps to avoid $454. In addition, $3,000 DCE can be obtained by shifting energy purchase for each method. Friday's outage can only be partially avoided due to the BESS capacity limitation. The avoided reparation costs are summarized in Table III and the unavoided costs are summarized in Table IV. From Table IV one can realize that there will be $693,546 costs by traditional outage method with one circuit breaker. Table III shows that $270,295 can be saved in addition to $2,928 DCE by using BESS. With two circuit breakers, the avoided reparation cost is dropped to $135,148. It will further decrease to $90,098 while increasing the number of circuit breaker to

6 three. If a modern smart feeder is exposed to outage conditions, the reparation costs are partially avoided. The avoided reparation costs for the above examples are in the range of $ 6,854 (for Method I) and $ 571 (for Method II). [10] Tan Zhang, Stephen Cialdea, Alexander E. Emanuel and John A. Orr "Electric Energy Cost Reduction by Shifting Energy Purchases from On-Peak Times," 2013 IEEE Electrical Power & Energy Conference (EPEC) VI. CONCLUSIONS The results indicate that under outage conditions caused by excessive peak feeder loads, substantial economic benefits may be obtained by use of a BESS. This benefit is due to the reduction in reparations costs payable by the utility with the elimination or reduction of the outage time. Of course the customer also benefits with the elimination or shortening of the outage. This approach is especially attractive for the case of a conventional feeder, since from Table II and III, one learns that for the studied case, the use of the BESS can avoid about one million dollars in reparation costs for two outage days. Also, depending on the BESS size, a substantial economic benefit can be obtained by combining the outage avoidance application with the energy purchase shifting application [10]. From this study, the traditional method of tripping the feeder circuit breaker at the substation is operationally and economically less advantageous than modern approaches using remote-controlled circuit breakers that can shed load intelligently, thereby maintaining service to critical loads and savings substantial reparation costs. It is also concluded that in the case of a sophisticated smart grid, the BESS will not provide significant benefits. ACKNOWLEDGMENT The authors gratefully acknowledge the contributions of Clayton Burns and Justin Woodard of National Grid for their assistance in accessing feeder data, Stephen Rourke and staff of ISO New England for their insight into the various aspects of the cost of power. REFERENCES [I] M. 1. Sullivan, B. Noland Suddeth, T. Vardell and A. Vojdani. Interruption costs, customer satisfaction and expectations for service reliability. Power Systems, IEEE Transactions On 11(2), pp DOL: / [2] R. Ghajar, R. Billinton and E. Chan. Distributed nature of residential customer outage costs. Power Systems, IEEE Transactions On 11(3), pp DOL: 10.1l09/ [3] Kristina Hamachi LaCommare, Joseph H. Eto. Understanding the cost of power interruptions to U.S. electricity consumers. Energy Analysis Department Ernest Orlando Lawrence Berkeley National Laboratory. University of California Berkeley Berkeley, California Available: [4] Michael J. Sullivan, Matthew Mercurio, Josh Schellenberg, Sullivan & Co., "Estimated Value of Service Reliability for Electric Utility Customers in the United States," [5] "Electric Energy Storage Technology Options: A White Paper Primer on Applications, Costs, and Benefits," EPRI, Palo Alto, CA, 2010, [6] IEEE Standard for General Requirements for Liquid-Immersed Distribution, Power, and Regulating Transformers, IEEE Std C [7] IEEE Standard General Requirements for Dry-Type Distribution and Power Transformers, Including Those with Solid-Cast and/or Resin Encapsulated Windings, IEEE Std C [8] info/hourly/index.html. [9] Zaborsky and J. W. Rittenhouse, Electric Power Transmission, Rensselaer Bookstore, 1969, p. 582.