2010 Elsevier. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional

Transcription

1 2010 Elsevier. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from Elsevier.

2 Energy Policy 38 (2010) Contents lists available at ScienceDirect Energy Policy journal homepage: Plug-in hybrid electric vehicles as regulating power providers: Case studies of Sweden and Germany Andersson S.-L. a,n, Elofsson A.K. b, Galus M.D. b,göransson L. a, Karlsson S. a, Johnsson F. a, Andersson G. b a Department of Energy and Environment, Chalmers University of Technology, Göteborg, Sweden b Power Systems Laboratory, ETH Zurich, 8092 Zürich, Switzerland article info Article history: Received 1 September 2009 Accepted 6 January 2010 Available online 1 February 2010 Keywords: PHEV V2G Regulating power markets abstract This study investigates plug-in hybrid electric vehicles (PHEVs) as providers of regulating power in the form of primary, secondary and tertiary frequency control. Previous studies have shown that PHEVs could generate substantial profits while providing ancillary services. This study investigates under what conditions PHEVs can generate revenues using actual market data, i.e. prices and activations of regulating power, from Sweden and Germany from four months in PHEV market participation is modelled for individual vehicles in a fleet subject to a simulated movement pattern. Costs for infrastructure and vehicle-to-grid equipment are not included in the analysis. The simulation results indicate that maximum average profits generated on the German markets are in the range h per vehicle and month whereas the Swedish regulating power markets give no profit. In addition, an analysis is performed to identify strengths, weaknesses, opportunities, and threats (SWOT) of PHEVs as regulating power providers. Based on the simulation results and the SWOT analysis, characteristics for an ideal regulating power market for PHEVs are presented. & 2010 Elsevier Ltd. All rights reserved. 1. Introduction Plug-in hybrid electric vehicles (PHEVs) have both an internal combustion engine and an electric motor with onboard electric energy storage (i.e. a battery), which can be charged from the electric grid. PHEVs are considered as promising options to achieve a costefficient and stringent control of CO 2 emissions in the transportation sector (Grahn et al., 2009; Hedenus et al., 2009). The flexibility in charging and the possibilities to charge the vehicle during hours of low electric load, for instance during the night, may decrease the costs of electricity generation (for a given load) and facilitate the integration of intermittent renewable electricity generation (Lund and Kempton, 2008; Göransson, 2008; Göransson et al., 2009). It has also been proposed to let PHEVs participate in grid ancillary services such as frequency control. This would have two large advantages: it would make it less expensive to own a PHEV (Kempton and Tomic, 2005) and more regulating power would allow for a more efficient integration of renewable intermittent power production in the system (Larsen et al., 2008). Real time frequency control is needed in all power systems. When more electricity is produced than consumed, the frequency will go up and regulation down is needed. Regulation down means either an increase in consumption of electricity or reduced production; a PHEV could provide this simply by charging its n Corresponding author. Present address: Vattenfall Power Consultant AB, Box 475, Göteborg, Sweden. Tel.: ; fax: address: sara-linnea.andersson@vattenfall.com (Andersson). battery as decided by the transmission system operator (TSO). When the consumption of electricity is higher than the production, the frequency in the system will decrease and there is a need for regulation up. This is provided through decreasing load or increasing production (Rebours, 2007). A PHEV could offer regulation up by discharging its battery into the grid, if it is equipped for vehicle to grid (V2G) power flows. If a PHEV is charging at a moment when regulation up is needed, the service could also be provided by stopping the charging operation. The value of the regulating power market of which the PHEVs could potentially have a share of is substantial. The total annual market value and market size of the German regulating power markets (here defined as the markets where real time frequency control is traded) is 770 million h and MW (cited from the German TSOs internet platform for publication of regulating power data, Regelleistung, 2008). Corresponding numbers for Sweden are 37.5 million h (using exchange rate 1 SEK=0.1 h) and 1000 MW (Hjalmarsson and Söder, 2001). It is of interest to estimate the number of PHEVs required to provide the corresponding regulating power. It can first be noted that an average personal vehicle is parked for more than 20 h a day (Brooks and Gage, 2001), and even during rush hour, more than 80% of all cars are parked (Lund and Kempton, 2008). To estimate the total number of PHEVs needed to cover the demand for regulating power, three assumptions are made: (a) that the required infrastructure is available so that 80% of the PHEV fleet is connected to the grid at any time; (b) that each PHEV is connected to the grid with a 3.5 kw single phase connection, hereafter called connection capacity; and (c) that a negligible /$ - see front matter & 2010 Elsevier Ltd. All rights reserved. doi: /j.enpol

3 2752 AnderssonS.-L. et al. / Energy Policy 38 (2010) amount of the connected vehicles have empty or full batteries at any time. These assumptions give that the total demand for regulating power would be met if 2.25 million PHEVs in Germany and PHEVs in Sweden participated as regulating power providers. This corresponds to 5.5% of the car fleet in Germany, and 4.2% in Sweden. Each year around 3.2 million new cars are registered in Germany and in Sweden (Anfac, 2009). As an order of magnitude estimation, this implies that if all new cars sold in the two countries were PHEVs with V2G opportunities, within one year these PHEVs could deliver all regulating power needed in Germany and Sweden. To keep the frequency stable, three types of control are used: primary, secondary and tertiary frequency control, as defined by the Union for the Co-ordination of Transmission of Electricity (UCTE, 2004). This study investigates PHEV participation in all these types of real time frequency control. Primary frequency control is the first type of control to respond to a frequency deviation. This is done automatically within seconds after a disturbance or if an unbalance between the load and generation occurs. Its purpose is to stabilize the frequency, possibly at another level than its nominal value. In Sweden, there are two types of primary control: frequency controlled normal operation reserve (FNR) and frequency controlled disturbance reserve (FDR). They are activated in different frequency intervals; while FNR is utilized between 49.9 and 50.1 Hz, FDR is active in the intervals and Hz (Kristiansen, 2007). When the primary control has stabilized the frequency, secondary control comes in to restore the frequency to its nominal value of 50 Hz, to restore the tie-line power between control areas, and to relieve the units participating in the primary frequency control. Secondary control is activated automatically in the UCTE system. However, for this type of regulating power it is important where the deviation from the planned load or production occurred as the TSO in the area of disturbance is responsible for delivering the required regulating power (UCTE, 2004). Secondary control in this meaning does not exist in Sweden (Kristiansen, 2007). Tertiary frequency control is defined as manually activated regulation by the TSO (UCTE, 2004). Normally activation is done by a phone call to a power plant operator (Verbic and Gubina, 2004). The purpose of the tertiary control is to restore the secondary control reserve (UCTE, 2004). All types of control are offered on markets where the TSO contracts regulating power. An actor (defined as a provider of regulating power, e.g. a power plant operator) can offer a specific capacity to any of the regulating power markets, but only to one market at the time. However, the same capacity can simultaneously be offered as regulation up and regulation down. A provider of frequency control can be paid for its services in either or both of the following two ways. A capacity price can be paid to an actor that is ready to deliver regulating power with short notice. An energy price can be paid for the regulating energy that is actually delivered. In the case of regulation up, the energy price is paid to the actor that delivers energy. In the case of regulation down, the actor pays for the energy it extracts from the grid as regulation down, but it pays a lower price than what would be the case if buying the energy on the normal power market. The possible profitability of PHEVs or other electric drive and/or fuel cell vehicles acting as regulating power has been treated in a number of previous studies. Kempton and Tomic (2005) estimated profits for battery and fuel cell vehicles providing different types of ancillary services: peak load power, spinning reserves and regulation (approximately corresponding to primary and secondary control). They concluded that providing regulation is highly profitable for electric vehicles; a net profit in the range h/month and vehicle (using exchange rate 1$=0.776 h) providing a regulating power of kw. In another study they have looked at profitability for EV fleets in four different US ancillary service markets for different years, and found a net profit for different years and markets in the range 6 45 h/month and vehicle providing kw of regulation down (Tomic and Kempton, 2007). They also concluded that adding regulation up usually results in much lower net profit. The Kempton and Tomic regulating power market studies apply estimated average values for both prices and activation frequency of the different types of regulating power. Williams and Kurani (2007) usedthesamemethodtoinvestigate anumberof vehicle types and maximum connection capacities. A large span of net profits is found, but more conservative than Kempton and Tomic: from a loss of 1.5 h/month and vehicle for spinning reserves, to a profit of 82 h/month and vehicle for regulation. Brooks (2002) estimated the value created by PHEVs on a regulating power market corresponding to primary or secondary control in study. The results lie in the range h/month, and represent gross value generated, which means that battery degradation and infrastructure costs need to be subtracted from the values. All of the above studies are made for the US, which implies that the price levels on the regulating power markets could differ significantly from those of the European countries investigated in this work. There are few estimates for Europe. Larsen et al. (2008) estimated profits for electric drive vehicles (EVs) providing secondary and tertiary control in Denmark. They obtained returns in the range h/month and vehicle (using exchange rate 1 DKK=0.135 h), depending on the market and the connection capacity. A calculation for the Portuguese market using average prices arrived at a net revenue of around 20 h/month and vehicle providing 3.5 kw of regulation services (Camus et al., 2009). The purpose of this study is to investigate if there are possibilities for PHEVs to generate profits on the other European regulating power markets, and to use actual regulating power market data in order to simulate vehicle behavior and possible profits generated on typical regulating power markets. Sweden and Germany are chosen as study cases since data were readily available for these two countries, but also because they have significantly different electricity generation portfolios, with hydro and nuclear power dominating in Sweden, while thermal power is the predominant power source of the German power system. In addition, the study aims at investigating the market structures of these regulating power markets and their implications on PHEVs as regulating power providers. The research questions are thus: From an economic perspective: with the current market prices, to what extent is it of interest for PHEVs to act as regulating power providers on the Swedish and German markets? From a market perspective: with the current market designs, to what extent is it possible for PHEVs to act as regulating power providers on the Swedish and German markets? How would a regulating power market be designed to best suit PHEVs? Below, the model that was developed to simulate PHEV behavior on the regulating power markets is described, and results from the simulations with the model are presented. A sensitivity analysis is also presented, varying the cost of battery degradation, the size of the connection capacity per vehicle, and the market prices of regulating power. Thereafter, a SWOT analysis is made of PHEVs as regulating power providers. This forms a basis for suggesting a regulating power market ideal for PHEVs, as well as for the discussion. 2. Simulation of PHEVs delivering regulating power 2.1. Model description In order to estimate the profits that could be generated by PHEVs on the present regulating power markets in Sweden and

4 AnderssonS.-L. et al. / Energy Policy 38 (2010) Germany, a simulation model was developed in Matlab. The model simulates the behavior of 500 individual PHEVs on the regulating power markets in Sweden or in Germany during January, April, July and October In the model, the cars are assumed to drive 2 35 min each day, supposedly from home to work and from work to home. No weekly or seasonal variations in the driving pattern have been included. The remaining time the cars are parked and during this time they are assumed to be plugged in and can participate on the regulating power market. While parked, the model assumes for each minute the car to perform one of the four actions optimizing its economics: 1. Deliver regulation down 2. Deliver regulation up 3. Charge (without participating in regulating power market) 4. Wait. Which action is chosen is determined by a number of conditions (if-clauses): for a car to deliver regulation up or down during minute A, regulation up or down, respectively, must have been activated on the market investigated during minute A. Additionally, the PHEVs bid must be activated (further details on this below). Moreover, the car must have a battery that is neither empty nor full, respectively, during minute A 1. Any regulation up delivered is thus compensated by charged electricity and not by extra use of fuel. The car will charge if the time left to its departure is shorter than the time needed to charge up the battery, so that the car always leaves with a fully charged battery. In a real-life situation, this would only be possible if the driver always knows when she or he will depart and shares this information with the aggregator/tso, which is rather optimistic with regard to user acceptance. Yet, this is one of the possible agreements that could be made between the driver and the TSO. Other possible agreements or contracts, not further studied here, could include that battery SOC may never go below 50%, or contracts without any constraints on battery SOC. The output from the model is the maximum profit per day for each vehicle under the assumptions made. The profit consists of three parts: Capacity payments: Defined as the income the PHEVs have for being available. This is counted as pure profit; no costs for infrastructure and V2G equipment have been included in this study. Profit from selling regulation down: Defined as the savings made from charging the battery at the regulation down price instead of the standard price for charging (defined below). Profit from selling regulation up: Defined as the price received for selling regulation up minus the price paid for buying the energy sold as regulation up plus costs for additional degradation of the battery (all of which are defined below.) Real market prices and demands of regulating power for each minute of the four months given above are used as input to the model. For the Swedish case, data on both prices and activated amounts of regulating power were given by Ripstrand (2008). For the German case, data on contracted amounts and prices were taken from Regelleistung (2008), whereas data on activation of regulating power were taken from Vattenfall (2008c). Thus, the profits generated and energy transactions made by the PHEVs are in principle profits that could actually have been made if the PHEVs had been on the market in Furthermore, the model simulates a best case scenario with respect to the profit of the PHEVs, implying that the results are the highest possible profits from acting on the reserve power market in This is motivated by the necessity to investigate if there is any economic justification for implementing PHEVs as regulating power providers. Moreover, the profits presented here are average maximum profits generated by one vehicle during a month, which possibly would need to be shared between several parties. These parties could be the vehicle owner, the aggregator (defined as a company that organizes the V2G-services of many vehicles being the interface between the TSO and the vehicle owners, see e.g. Brooks, 2002) and perhaps other entities, while also covering additional equipment and infrastructure costs. Primary control is activated automatically, triggered by a frequency deviation. In the model, the cars thus deliver regulation up when the frequency is below 50 Hz, and regulation down when the frequency is above 50 Hz. The regulation characteristics applied in the model are straightforward; as soon as the frequency drops below or goes above Hz, the maximum power (3,5 kw) is delivered from or to the vehicles, respectively. The frequency used is the actual frequency in the Nordel and UCTE system during the modeled months of (2008)(Thor, 2008; Biernacka, 2008). Fig. 1 presents the frequency in the two countries during 2 July 2008 as an example. The black line Power system frequencies on 2 July 2008 Sweden Germany Frequency (Hz) Time (h) Fig. 1. Example of frequency variations in the Swedish and German power systems (Nordel and UCTE) during 2 July 2008.

5 2754 AnderssonS.-L. et al. / Energy Policy 38 (2010) represents frequency in Germany and the grey line represents frequency in Sweden. On this day frequency deviations were larger in Sweden. To participate on the primary regulating market, an actor makes a bid for capacity that is accepted or rejected by the TSO. All accepted actors then deliver primary control whenever the frequency deviates from 50 Hz, irrespective of the price for actually delivering the demanded energy. Actors that deliver primary control are given a capacity payment in both Sweden and Germany, according to a pay-as-bid system (each accepted actor receives the price of their bid). In Germany, the bid the PHEVs make in the model is the maximum bid accepted during that time period, known from Regelleistung (2008). In Sweden, the capacity bid made by the vehicles is assumed to be a representative average bid per hour according to the Swedish TSO (Bäck, 2008). Additionally, energy is paid in primary control in Sweden, at the same price set for tertiary control energy during that hour (Bäck, 2008). Energy delivered in primary control in Germany is not paid (Schulz, 2008). Table 1 gives the bids made by the PHEVs in the model, and actual price intervals and mean prices on the markets during the simulated months. Table 1 also provides the share of time that different types of control were activated during the simulated months, to enable comparison, e.g. that tertiary control is activated much more in Sweden than in Germany. For secondary and tertiary control, the activation is based on bids, so that when a demand for regulation up arises the actor with the lowest bid is activated first, the second lowest bid activated second, and so forth (inversely for regulation down, i.e. the actor who is willing to pay the highest price for energy bought as regulation down is activated first). Both capacity and energy payments are given to actors on German secondary and tertiary markets according to a pay-as-bid system. Actors on the Swedish tertiary market receive only an energy payment, which on the other hand is priced according to marginal cost principle (meaning that all activated actors are paid the bid of the last activated actor). In the model, the PHEVs are always accepted for capacity, at the highest bid that was accepted during that time period, in accordance with the best case scenario. An indication of the price levels on the different markets is given in Table 1. For regulation up energy, the vehicles make a bid that covers the costs they have for delivering the energy plus a 5% profit margin. The cars will be activated for regulation up if this bid is lower than bids that were actually activated during each time period. For regulation down energy, the bids the cars make are such that they are always the first to be activated (in this ideal best case scenario, the bids of the competitors are assumed to be known). Because delivering regulation down only means charging at a lower price, this is the most profitable strategy for the cars. In the model, each PHEV has a battery capacity of 10 kwh with a maximum depth of discharge at 20% of full capacity (i.e. maximum 8 kwh can be utilized). Charging and discharging efficiencies are set at 94% (Söderbom, 2008). The data set containing arrival and departure times of the individual vehicles and corresponding states of charge, based on actual driving cycle composition, is taken as output from the model described by Galus and Andersson (2008, 2009). The vehicles are assumed to be connected to the grid with a connection capacity of 3.5 kw, corresponding to a European standard of household plugs (230 V, 16 A). This is significantly lower than what several previous studies have used, but it is assumed to be the rating for a first generation of PHEVs with slow charging (Söderbom, 2008). The cost for battery degradation is difficult to estimate, because the Li-ion technology is fairly new and rapidly developing. A Li-ion battery lasts for deep cycles (Markel and Simpson, 2006), and estimated future battery investment cost in the case of mass production lies in the range $ per kwh of battery (Williams and Kurani, 2007; Bergman, 2008; Söderbom, 2008). The cost for battery degradation is non-linear with respect to cycle depth. Assuming for simplicity that only full discharge cycles are used, the cost of battery degradation per unit of energy throughput in the battery can be calculated as B D f ¼ ð1þ C f E max where D f is the cost of battery degradation per unit of energy throughput at full discharge cycles, B is the battery investment cost per unit battery size, C f is the number of possible full cycles during a battery lifetime, and E max is the usable share of a unit of battery (e.g. if a battery can be discharged down to a depth of SOC=0.2, then E max =0.8). C f E max therefore represents the total amount of energy throughput that a battery unit can have during its lifetime. This is a simplified equation that does not take the complexities of battery degradation into account, and it is used only to get an estimate of what battery degradation could cost. Using Eq. (1) with a battery investment cost of 300 $/kwh, a lifetime of 3000 cycles and a depth of discharge of 80%, result in a battery degradation cost of 100 h/mwh (assuming 1$E0.8 h). This figure is used for tertiary control, which is assumed to have deep discharge cycles. The battery cycles are likely to be less deep in primary and secondary than in tertiary control, which results in a battery degradation that is around 1/3 of that at deep discharge cycles (Kempton and Tomic, 2005). Here, a degradation cost of 30 h/mwh has been used for primary and secondary control. Charging is assumed to be done at a fixed price, e.g. through the car owner s normal electricity supplier. The Swedish electricity price is 73.9 h/mwh, taken from Vattenfall s 1-year fixed price contract per 2008 for private consumers, with taxes subtracted (Vattenfall, 2008a). The corresponding price in Germany is h/mwh (Vattenfall, 2008b). Taxes, other fees, and subsidies have been excluded, since it would complicate the study considerably without necessarily giving additional insights. Moreover, excluding taxes and fees gives an estimate of the potential profits without interference of tax and subsidy systems, which may change rapidly PHEV profit on current German and Swedish markets Fig. 2 presents the average monthly profit generated per PHEV in all types of regulating markets for both countries (upper plot), and the corresponding average energy transactions (lower plot), that is energy that has been delivered as regulation up and down or charged without regulating power market participation. It can be seen that even under a maximum profit simulation like the one at hand, none of the Swedish regulating power markets are profitable for PHEVs to act on. Average profits per vehicle range from a loss of 19 h/month for primary FNR to 7 h/month profit for tertiary control, with primary FDR yielding 0.5 h/month profit. A PHEV providing control in Germany can generate a fairly high maximum profit on all three markets: 50 h/month when acting as primary control, 80 h/month as secondary control and 30 h/month as tertiary control. The higher profits in Germany can be explained by higher price levels on both capacity and energy for regulating power. In Sweden, there is a capacity payment on the primary market, but it is small compared with German price levels (see Table 1). Moreover, prices for control energy in Sweden are normally close to the spot price, whereas in Germany there is a two orders of magnitude difference between the prices for regulation up and down (as seen in Table 1). As seen in Fig. 2, the PHEVs are activated for regulation up on the tertiary market in Germany, which implies that the PHEVs bid is low enough to be activated

6 AnderssonS.-L. et al. / Energy Policy 38 (2010) Table 1 The bids the PHEVs make on the different markets in the model, and actual prices occurring on the markets during 2008, given as a min-max interval of accepted bids, and a mean accepted bid in brackets. PHEV bid Actual prices: Minimummaximum (mean) accepted bids during simulated months Capacity Up and Down ( /MW/h) Energy Up ( /MWh) Energy Down ( /MWh) Capacity Up ( /MW/h) Capacity Down ( /MW/h) Energy Up ( /MWh) Energy Down ( /MWh) Share of time that control was active during simulated months (%) Primary Secondary Tertiary Sweden 1 Germany 2 Germany 3 Sweden 4 Germany for FNR and Highest accepted Highest accepted bid. Highest 0.36 for FDR bid during each during each period accepted bid during the day 6 and period (one (peak/off peak time during each 0.42 (FNR) and 1.7 month). during one month) period (4h) (FDR) 7 during the night 8 No bidding Unknown. PEHV bids above are representative averages Same as in tertiary control. FNR: 99% FDR: 2% 15 98% 16 Highest accepted bid during each period (19) Peak time: (22) Off-peak time: 6-8 (7) 12 Peak time: 3-5 (4) Off-peak time: 5-19 (8) No energy price. Peak time: (157) Off peak time: (112). Peak time: 0-26 (8) Off peak time: 0-9 (1) Spot price No capacity price Highest accepted bid during each period (6) (2) (55) (405) 0-95 (48) 0-8 (1) 99% 17 76% 18 4% 19 a Bäck (2008). b Regelleistung.net (2008). c Regelleistung.net (2008). d Ripstrand (2008). e Regelleistung.net (2008). f g Bids originally made for frequency response, and are 3.5SEK/MW/Hz/h for both FNR and FDR during daytime and 20SEK/MW/Hz/h during nighttime. The Swedish TSO then adds 1SEK to all accepted bids, so calculations here are based on 4.5 and 21SEK/MW/Hz. Frequency bands are 0.8 Hz for FDR and 0.2 Hz for FNR. h i (106.4 h/mwh (price for charging)+30 h/mwh (cost of battery degradation))n1.05 (for profit) j (73.9 h/mwh (price for charging)+100 h/mwh (cost of battery degradation) )n1.05 (for profit) k (106.4 h/mwh (price for charging)+100 h/mwh (cost of battery degradation))n1.05 (for profit) l Bids originally made per month, here assumed 276 h peak time (weekdays ) and 468 h off peak time (all other time) per month, assuming 31 days/month and 4 weekends/month. m Bids originally made per 4 h. n With one extreme incident of a price at 960 h/mwh removed. o Thor (2008). p Biernacka (2008). q Vattenfall (2008c). r Ripstrand (2008). s Vattenfall (2008c). on this market (Table 1 shows that the PHEV s bid is 217 h/mwh compared with an average bid for regulation up of 405 h/mwh). In Sweden, the PHEVs bid for regulation up (183 h/mwh) is not within the span of activated bids during 2008 (2 125 h/mwh), and is thus never activated. For regulation down, the PHEVs are always the first to be activated in the model. This implies that since tertiary control is active most of the time in Sweden (see Table 1), the cars almost always fill their batteries on regulation down energy when delivering tertiary control in Sweden. However, since the energy price the cars pay for the regulation down energy is similar to the price of the energy they would normally have charged, the profit is still small. The explanation for the difference in price levels on the Swedish and German regulating power markets is that regulating power in Sweden is generally delivered by hydro power plants, which have a significantly lower cost for providing this service than the coalfired power plants that prevalently provide regulating power in Germany (Hjalmarsson and Söder, 2001). The German secondary regulating power market is the most attractive for PHEVs because it combines two advantageous factors: capacity payments are high on the secondary regulating power market and the PHEVs acting on this market are often activated for regulation down (as seen in Fig. 2) at a very attractive price (mean peak price 8 h/mwh, mean off peak price 1 h/mwh as seen in Table 1). The reason that PHEVs delivering Swedish FNR generate a negative profit is related to activation. FNR is almost constantly active whereas FDR is seldom activated, as seen in Table 1 and in

7 2756 AnderssonS.-L. et al. / Energy Policy 38 (2010) Average monthly profit generated per vehicle Euro/car Swe: FNR Swe: FDR Swe: tertiary Ger: primary Ger: secondary Ger: tertiary Average monthly energy transactions made per vehicle Regulation up Regulation down Charge kwh/car Swe: FNR Swe: FDR Swe: tertiary Ger: primary Ger: secondary Ger: tertiary Fig. 2. Upper graph: Average profit generated by one PHEV during one month on the different markets. Lower graph: Average energy transactions (regulation up, regulation down, and charging without regulating power market participation) made by one PHEV during one month on the different markets. the lower plot in Fig. 2. This has a great impact on the results since for the primary regulating power market in Sweden, the energy payment is defined without relation to the cost of the actor (as described in the Methodology section above). Hence, PHEVs deliver regulation up at a loss. Fig. 3 visualizes the profits generated on all markets divided in a number of different categories in order to show from where the profit originates. In the upper plot, the profit generated by one vehicle during the time parked at home is compared with the profit generated during the time parked at work. In the middle plot, the profit coming from capacity payments is compared with the profit (or loss) originating from energy payments. In the bottom plot, profit from providing regulation up is compared with the profit from providing regulation down. In all cases shown in Fig. 3, the sum of the two categories in each market is the total profit. In the upper plot of Fig. 3, it can be seen that profit from the time parked at home is, on all markets where profit is generated, larger than the profit from the time parked at work. This is sometimes due to the longer time parked at home, and sometimes due to higher prices during the night. Irrespectively, this implies that if PHEVs are to make profit by providing regulating power, it would be better to install the required infrastructure for nighttime connection (e.g. at people s homes) than daytime connection (e.g. workplace parking lots for staff). From the middle plot in Fig. 3, it becomes obvious that capacity payments make very important contributions to the total profit on all German markets. If only energy would have been paid for on the German markets, the profit would range from a loss of 7 h/month to a profit of 20 h/month on the different markets. The highest in the range, 20 h/month, is not likely to be enough to motivate PHEV participation on the market. Since the capacity payment on the present Swedish primary market is low, capacity payments do not make a major contribution to the profit on any of the Swedish markets. The lower graph in Fig. 3 gives a comparison of the profit contributions from delivering regulation up and down. Generally, delivering regulation down will always be interesting to PHEVs, as it implies that they can charge at a lower cost. Delivering regulation up is more risky for the PEHVs as the simulation results indicate that this is done at a loss on the primary regulating power markets. However, if capacity is paid for, this income can cover the losses made in providing regulation up (see German primary control in middle plot in Fig. 3). The results indicate that if PHEVs are providing regulating power it may be wise to only offer regulation down. Thus, this would avoid the risk of a loss. The results indicate that also the Swedish primary FNR could generate a profit of 20 h/ month if only offering regulation down. Simulations give that profits in Germany range from around 60 h/month for primary control, to around 40 h/month for secondary control, down to less than 10 h/month for the tertiary market. Although these profits are significantly lower than when offering both up and down regulation, the simpler equipment and infrastructure required to implement PHEVs as regulation down (i.e. less costly with no V2G possibility) may offset some of the lower profits. Delivering regulation down only requires a remote controlled charger Sensitivity analysis Fig. 4 summarizes the results from the sensitivity analysis. The analysis comprised simulations in which three important input parameters were varied: cost of battery degradation, connection

8 AnderssonS.-L. et al. / Energy Policy 38 (2010) Euro/car Home Work 50 Swe: FNR Swe: FDR Swe: tertiary Ger: primary Ger: secondary Ger: tertiary Euro/car Capacity Energy 50 Swe: FNR Swe: FDR Swe: tertiary Ger: primary Ger: secondary Ger: tertiary Euro/car Up Down 50 Swe: FNR Swe: FDR Swe: tertiary Ger: primary Ger: secondary Ger: tertiary Fig. 3. Upper graph: Comparison of average profit generated by one PHEV one month from the time parked at home and the time parked at work. Middle graph: Comparison of average profit generated by one PHEV one month originating from capacity and energy payments. Lower graph: Comparison of average profit generated by one PHEV one month by providing regulation up and down. capacity and market prices of regulating power. The simulations are performed for data from July 2008, which means that the original profits in Fig. 4 differ slightly from the average profits in Fig. 2. The cost of battery degradation was lowered to 1/100th of its original value (1 h/mwh for tertiary control and 0.3 h/mwh for primary and secondary control) with the intention to illustrate how profits could be affected in a possible future where significant advances in battery technology have taken place, i.e. where batteries are less expensive and have longer life lengths than today. Yet, the cost for battery degradation is given per unit throughput of energy, not per unit battery size. Contrary to what one might expect, this parameter is the one that affects the results the least of the parameters varied. The reason is that the battery degradation cost only becomes important when the vehicles sell significant amounts of regulation up, which only happens on the primary markets. FNR is the market that is the most affected by a lower battery degradation cost with an increase in average monthly profit from a loss of around 19 h/month to a profit of around 6 h per vehicle. A less expensive battery also implies that the PHEVs are sometimes, although rarely, activated for regulation on the secondary regulating power market in Germany and the tertiary regulating power market in Sweden. For the original case the simulation gives that PHEVs are never activated for regulation up. The connection capacity was increased from 3.5 to 15 kw, with 15 kw being closer to what previous studies have used (10 15 kw by Kempton and Tomic (2005), kw by Williams and Kurani (2007), 7 and 18 kw by Brooks (2002) and 2 and 20 kw by Larsen et al. (2008). The increase in connection capacity to 15 kw has the largest effect on the results. In German primary and secondary control, the profits increase from 60 to 260 h/month and from 110 to 390 h/month, respectively. The reason is the large capacity payment, which of course increases when the connection capacity increases. However, as discussed in the introduction, a first estimate indicates that the regulating power markets could be saturated already at modest market shares of PHEVs, even with 3.5 kw connection. With an increased connection capacity, the profit per participating vehicle could be substantially larger, but the number of participating vehicles would decrease correspondingly. Interestingly, in Swedish FNR, the loss increases with an increased connection capacity (increased loss from 19 to 54 h/month). Here, the increased capacity payment is not large enough to compensate the increased amount of regulation up sold at a loss. In Swedish FDR, the profit does increase, but only from 0.5 to 1.2 h/month and vehicle. Market prices for regulating power have been varied both up and down, in each case by 20%. To simulate increased regulating power prices, capacity prices and bids for regulation up were increased with 20%, while bids for regulation down were decreased by 20%. These variations intend to simulate a possible future with a higher demand for regulating power, such as would be the case in a system with increased fraction of intermittent power supply. An increase in regulating power prices influences all markets to some extent, making the profits somewhat higher. On all markets except for primary FNR, increased regulating power prices by 20% raise the profits more than having a battery degradation cost of 1/100th of the original. However, a 20% raise

9 2758 AnderssonS.-L. et al. / Energy Policy 38 (2010) Original Battery degradation cost 1Euro/MWh Connection capacity 15 kw Regulating power prices 20% higher Regulating power prices 20% lower 200 Euro/car Swe: FNR Swe: FDR Swe: tertiary Ger: primary Ger: secondary Ger: tertiary Fig. 4. Sensitivity analysis of the average profit generated by one PHEV during one month for variations in battery degradation, connection capacity and regulating power price. in regulating power prices changes the FNR profit to positive values (compared to a loss of 19 h/month in the original case). The case in which capacity and energy prices are lowered by 20% illustrates a case when PHEVs have entered the market at a fairly large scale, and have therefore influenced the price for regulating power. Naturally, all profits per PHEV decrease in this case. Also in this case FNR is the most sensitive, with an increase in loss from 19 to 35 h/month. 3. SWOT analysis of PHEVs as regulating power providers and the ideal regulating power market for PHEVs A SWOT analysis was made to identify the Strengths, Weaknesses, Opportunities and Threats of PHEVs as regulating power providers. The first two factors are internal to the PHEVs, meaning that they concern positive (strengths) and negative (weaknesses) features and properties of the PHEVs as regulating power providers. The last two factors are external and are features of the regulating power market, and other actors on these markets, which will have an influence on the ability of PHEVs to serve as regulating power providers. A more thorough description of the SWOT analysis as a tool is given by Ritsch (2004). Table 2 summarizes the result from the SWOT analysis of PHEVs as providing regulating power with the aim to provide a first basis for design of a regulating power market, which would be tailored for the characteristics of PHEVs. Some of the opportunities and threats from the SWOT analysis are treated further in the Discussion section. As can be seen from Table 2, a major strength of the PHEV as regulating power resource is that it can deliver electricity from its battery to the grid or start charging within seconds. This enables short activation times and makes PHEVs fast enough for all types of frequency control. An obvious weakness is that the energy storage in the battery is limited, which put constraints on the length of the periods that PHEVs can be activated for frequency control (energy is limited). When the battery is full, no more regulation down can be delivered. The connection capacity of the plug through which the PHEV is connected will obviously limit the regulation power provided by each PHEV. This power is very low compared to that of a power station, which is another weakness of PHEVs acting on the regulating power markets of today, where large bids are required. This makes activation of PHEV regulating power more complicated, as many small units need to be activated. Thus, a possibility to pool a certain number of vehicles, e.g. through an aggregator, is needed. A possible problem with PHEVs as regulating power is that they are mobile (cf. Table 2), which means that the TSO cannot know for certain if a specific vehicle is available at a specific time. Considering the importance of a stable and secure power system, this is of course a major weakness when considering individual vehicles. However, with a large number of vehicles aggregated and offered to the market as one unit, this weakness will be significantly less problematic, since the probability that all aggregated vehicles are driving simultaneously is small. As noted before, even during rush hour, more than 80% of all cars are parked (Lund and Kempton, 2008). When a PHEV delivers regulation up there is a significant energy loss, as energy is first charged and then discharged from the battery (94% charging and discharging efficiency is used in the present simulations, with other studies applying both higher or

10 AnderssonS.-L. et al. / Energy Policy 38 (2010) Table 2 Summary of the qualitative SWOT analysis made for PHEVs as regulating power. Internal External Positive Strengths No cost for being available. Regulation down is charging of the battery and can be provided cost efficiently. Fast delivery/acceptance of energy, i.e. short activation time. Many PHEVs in a system could be seen as a guarantee of stable capacity. Compared to EVs, PHEVs have the possibility to use gasoline to provide regulation up when battery is empty. Opportunities Increased demand for regulating power due to increased intermittency (renewable energy) in the power system. Political will to introduce PHEV and to support electrification of the car fleet. Liberalization of regulating power market political pressure to lower entrance barrier. Other actors also put pressure on the regulating power market to change. Potential economic incentive for PHEVs to participate is substantial on some markets. Negative Weaknesses Limited connection capacity. Limited energy storage in the battery (cannot provide regulation down when battery is full). Mobile devices difficult to guarantee constant capacity if a small number of cars participates. Human factor: people can forget to plug in their cars or feel uncomfortable about the TSO controlling their charging operation. Regulation up has a high cost. The cost for delivering up- and down-regulation differs significantly. Regulation up implies energy losses in charging and discharging of the batteries. Threats Conservative and inflexible market. Oligopoly market structure. High security and delivery requirements. Large bid sizes. Long market time frames. Long contract times. Infrastructure development needed. Standardization issues. Co-operational challenges automotive industry and power suppliers. Lower prices on regulating power due to PHEVs on the regulating power market. Symmetric bids and package prices which makes the bidding system inflexible, and does not allow offering only regulation up/down, etc. TSOs may not trust PHEVs as regulating power providers. Risk of technology lock-in, i.e. if charging infrastructure initially is installed without V2G possibilities. lower charging efficiencies (Galus and Andersson, 2009). This can be compared with a few percent efficiency loss for hydro and coal power delivering regulating power (Hjalmarsson and Söder, 2001), although pumped hydro has higher losses. Thus, this is a weakness for PHEVs as regulating power providers (cf. Table 2) since costs increase and there will be an increased overall energy loss in the system. In summary, based on Table 2 it can be concluded that PHEVs have some features that make them suited as regulating power providers in certain situations. One of the most important strengths is that a PHEV s flexible cost of being available for regulating power is zero, that is, there is no extra cost for keeping the vehicle plugged in once the fixed infrastructure and equipment costs are paid for. Another major strength is the cost efficient provision of regulation down. These strengths could be an asset to society, if the regulating power market is re-designed so that the weaknesses identified will be less significant. Table 3 proposes a regulating power market suited for PHEVs and compares this market to the Swedish and German markets of today. The ideal market is described and explained below. From the PHEV point of view, markets with both energy and capacity payments are preferred. Since the PHEV s cost for keeping the capacity ready is zero or close to zero, capacity payments quickly become profitable. Hence, PHEVs have an advantageous position for competing with other actors on a market with capacity payments, since the PHEVs can make low capacity bids. Concerning energy, a PHEV delivering regulation down simply corresponds to charging its battery. It is of course an attractive deal for PHEVs to be given a capacity price for just being plugged in, and then charge the battery with cheap or even free electricity. However, the conditions differ considerably for regulation up. Then the costs for supplying the grid with electricity correspond to the cost for battery degradation due to extra cycling and the electricity cost for the energy stored in the battery, both of which are high in comparison to expected profit. This has two implications. Firstly, an energy payment will make the PHEVs compete on fair terms with other actors on the market, because then bids can be made in correspondence with the costs for PHEVs providing regulating power (cf. primary regulating power markets where energy is not paid or paid without relation to the costs, and where PHEVs have to sell regulation up at a loss). Secondly, it is difficult for PHEVs to compete on markets where bids are made symmetrically for both up and down regulation. Here, PHEVs are not able to take advantage of their competitive edge (very low costs for regulation down) as they are forced to place bids that cover their high costs for regulation up. From a regulating power market point of view, this is also unfortunate, as it could rise the prices and prevent the most cost efficient solution, where each actor does only what it does most cost efficiently. Hence, it is ideal for PHEV to act with asymmetric bids, i.e. bids on either up or down regulation. The contract time is defined as the period during which an actor guarantees regulating power. In order to utilize the pluggedin capacity the aggregator needs to vary the size of its bid in accordance with the amount of PHEVs plugged in. Ideal for PHEVs would thus be a short contract time such as one hour. The market time frame is defined as the time period between market closure and delivery. This is important to PHEVs because an aggregator that offers the PHEV-capacity from one parking lot has difficulties in knowing long before how many cars will be parked during a specific time. These unpredicted variations suggest that PHEVs need a short market time frame, i.e. a bidding process close to delivery, to be able to adjust the bids to the amount of PHEVs plugged in. Ideal for PHEVs would be to make their bids just a few minutes before delivery. However,