Dit project is uitgevoerd met subsidie van het Ministerie van Economische Zaken; regeling Energie Onderzoek Subsidie: Lange Termijn This project is subsidised by the Netherlands Ministry of Economic Affairs; programme Enery Research Subsidy: Long Term

SenterNovem EOS-LT 05024 Intelligent E-transport Management - ITM

WP 1 Report nr. 08-9164

Date: 14 August 2008

Trends in electricity demand and supply and effects on grids in the Netherlands

Synthesis report ITM project WP1

Authors: P. de Boer (KEMA), Y. Boerakker (ECN), M. Bongaerts

(Continuon), A. Brand (ECN), R. Hunik (IWO), M. Mangnus (KEMA), F. Nieuwenhout (ECN), A. Postma (Essent Netbeheer), F. Schulze (KEMA),

A. Seebregts (ECN), P. Vaessen (KEMA), J.J. de Wolff (KEMA)

© ITM projectpartners, Nederland, zijnde: KEMA Nederland B.V., Arnhem; ECN, Petten; St. IWO, Ede en IWO Project b.v., Ede; Essent, Den Bosch; Continuon, Arnhem. Alle rechten voorbehouden.

Dit document bevat vertrouwelijke informatie. Overdracht van de informatie aan derden zonder schriftelijke toestemming van of namens de ITM-partners via penvoerder KEMA Nederland B.V. is verboden. Hetzelfde geldt voor het kopiëren van het document of een gedeelte daarvan. De ITM projectpartners en/of de met hen gelieerde maatschappijen zijn niet aansprakelijk voor enige directe, indirecte, bijkomstige of gevolgschade ontstaan door of bij het gebruik van de informatie of gegevens uit dit document, of door de onmogelijkheid die informatie of gegevens te gebruiken.

Source of the cover photo: http://www.spinnovation.com/

-3- EOSLT05024 08-9164 confidential CONTENTS

page

1 Introduction .......................................................................................................... 4

2 Electricity scenarios ............................................................................................. 5 2.1 Selection of scenarios .......................................................................................... 5 2.2 Generating capacity in SE scenario ..................................................................... 6 2.3 Final electricity demand ....................................................................................... 6 2.4 Renewable electricity ........................................................................................... 7

3 Wind Power in the netherlands ............................................................................ 9

4 Plug-in electric vehicles.......................................................................................12

5 Heat pumps.........................................................................................................16

6 Electricity networks .............................................................................................21

7 Discussion ..........................................................................................................26

-4- EOSLT05024 08-9164 confidential 1 INTRODUCTION The increased application of renewable energy sources (for example wind or solar energy) for electricity production will enhance fluctuations in centralized power generation. Through such fluctuations the stability and quality of power supply becomes more and more an issue. Moreover, due to the variations in demand during the day, more conventional power plants have to act at night as spinning reserve. This results in low efficiencies at relative high emission levels, while for renewable energy sources it is often not possible to turn off during night, resulting in extreme cases in loss of produced electricity. So the increase of renewable energy with large fluctuations in supply and fluctuations in demand makes the management of the electricity network more complex. For the coming decades, the fluctuation in the supply of wind energy is the most relevant, and therefore discussed in more detail in the chapter 3 of this report. For the context of this report, chapter 2 presents the trend in electricity demand and supply. A growing share of intermittent renewables will result in increasing demand for flexibility in the whole electricity system including consumption. Recently, driven by a favourable ‘spark spread’, industrial and horticultural cogeneration owners are aggregating their CHP units into virtual power plants of which the flexibility is sold on power markets such as the APX. These units have a capacity of typically 1 MW or more. At the smaller scale of residential electricity connections there is some limited experience with demand response. A recent study by KEMA and ECN (KEMA, 2007) showed that for an average household theoretically about 200 kWh per year can be shifted to other periods of the day. This is between 5 and 10% of the annual electricity consumption. In the future we expect that this percentage can increase substantially due to the introduction of new appliances which consume or generate larger amounts of electricity than the refrigerators and washing machines which have been analysed in the previously mentioned study. Furthermore some sort of storage of electricity or heat is required. Relevant appliances which can be expected to have increased penetration in the future are micro-cogeneration, air-conditioning, electric heat pumps and electric vehicles. Two of these have been selected to be studied in more detail in the ITM project, namely electric vehicles and electric heat pumps. This selection is based on the large potential of these two applications with respect to CO2- and energy-reductions. These two important trends can contribute to solve the network problems of more fluctuations and differences in electricity demand and supply during day and night: the increase of electric transportation using batteries (electric vehicles), charged at night, and the increase of the amount of electric heat pumps (possibly in combination with thermal storage). Both options offer the possibility to manage the electricity demand over time to compensate for the increased uncertainties in supply due to integration of intermittent renewables. In this

-5- EOSLT05024 08-9164 confidential way a fluctuating supply can be managed by a controlled demand. Plug-in hybrid electric vehicles are discussed in more detail in chapter 4, and electric heat pumps in chapter 5. The drawn developments generate new demands for the management of the system. Network management is necessary to facilitate a maximum contribution of renewable energy sources, while at the same time meets the technical and social-economic requirements. Developments in the grid are discussed in chapter 6. A discussion of the role of electric vehicles and heat pumps in integration of renewables and mitigating imbalances is provided in chapter 7. KEMA 2007, Implicaties vraagrespons bij huishoudens voor netbeheerders, Studie van KEMA en ECN in opdracht van Continuon Netbeheer, Eneco Netbeheer en Essent Netwerk, 30630509-Consulting, 07-1681, 5 September 2007.

2 ELECTRICITY SCENARIOS 2.1 Selection of scenarios In the beginning of the ITM project, two existing and relatively recent long term scenarios have been selected as starting point for further use in the project. The purpose of using these two different scenarios is to show an envelope of the directions into which the electricity production system in the Netherlands could develop. These scenarios are based on a recent long term scenario WLO (‘Welvaart en LeefOmgeving) study (up to 2040) for the Netherlands (CPB et al, 2006) and have also been used as midterm (2020) reference projections (ECN/MNP, 2005), as a basis for Dutch energy and climate policies. In the Strong Europe scenario (SE) stringent international climate policy emerges leading to high CO2 prices (from 11 EUR/ton CO2 in 2012-2020 to about 60 EUR/ton CO2 in 2030). The average annual growth of the electricity demand is 0.9%. In 2007, a policy intensive variant of the SE scenario has been constructed with additional assumptions on policies for energy saving and an even larger role of renewable energy. This ‘Green4sure’ scenario has been quantified up to the year 2030 (CE, 2007; Van Dril et al., 2007). In the Global Economy (GE) scenario, climate and renewable policies will be abolished after 2020. The average annual growth of the electricity demand is 1.6%.

-6- EOSLT05024 08-9164 confidential 2.2 Generating capacity in SE scenario The development of the generating capacity in the Strong Europe scenario is presented in the following figure 1. Conventional nuclear, coal and gas plants remain at a constant absolute level of about 10 GW. Contribution from renewables and CHP increases from less than half in 2000 to more than 75% in 2040.

05

10152025

303540

4550

2000 2005 2010 2015 2020 2025 2030 2035 2040

[GW]

CHP (mainly gas)

Renew able

Nuclear

Gas

Coal (excl. BM co-firing)

Figure 1 Generating capacity, Strong Europe scenario

2.3 Final electricity demand The final electricity demand in the three scenarios is shown in figure 2.

0

20

40

60

80

100

120

140

160

180

200

220

2000 2005 2010 2015 2020 2025 2030 2035 2040

[TWh]

Global Economy

Strong Europe

Green4Sure

Figure 2 Final electricity demand the Netherlands for three long term scenarios

-7- EOSLT05024 08-9164 confidential 2.4 Renewable electricity The renewable production is further detailed out for the Strong Europe scenario (see figure 3). The ‘discontinuity’ with respect to renewable energy makes the Global Economy less plausible given current national and EU policy developments, and less relevant for the ITM project which investigates an electricity system with a (persistent) large role of wind energy.

0

2

4

6

8

10

12

14

16

18

20

2000 2005 2010 2015 2020 2025 2030 2035 2040

[GW]

Solar PV

Hydropow er

Other biomass

Biomass co-firing

Waste

Wind at sea

Wind onshore

Figure 3 Renewable generating capacity, Strong Europe

Wind energy The electricity from wind energy is the main source of renewable electricity in the coming decades. The potential role of wind energy is illustrated in more detail for the Strong Europe (SE) scenario and its Green4Sure variant (G4S) (see table 1). After publication of the WLO study, variants of the Strong Europe scenario up to 2030 have been constructed that includes the effect of additional policy instruments (CE, 2007; Van Dril et al. 2007) or increased shares of offshore wind energy (We@sea). The shares of renewable electricity production become even larger in those variants.

-8- EOSLT05024 08-9164 confidential Table 1 Installed capacity wind energy in the three scenarios Capacity wind energy (MW)

2005 2010 2015 2020 2025 2030 2035 2040Strong Europe (SE)

Wind onshore 1150 1269 1511 1901 2000 2000 2000 2000Wind at sea 0 700 1700 3000 4500 5300 7800 10000

SE-Green4Sure (SE-G4S) Wind onshore 1150 1269 2760 3013 3463 4013

Wind at sea 0 700 1844 3532 6023 9000The electricity produced from wind energy grows to 25% of the final electricity demand in 2040. As percentage of total generating capacity, this amounts to 27% (12 GW) in 2040. In the Green4Sure SE variant wind energy grows much faster than in the original SE scenario, totalling at 33% (production) and 36% (capacity, 13 GW) in 2030.

References and others sources of information

CE (2007): Green4sure - Het Groene Energieplan. Hoofdrapport, CE, Delft, mei 2007.

CPB et al. (2006): Bijlage Energie (MNP/CPB/RPB/ECN) in: Welvaart en Leefomgeving - een scenariostudie voor Nederland in 2040, Achtergronddocument, CBP/MNP/RPB, Den Haag, ISBN 90-6960-150-8. Also at: http://www.ecn.nl/publicaties/default.aspx?nr=ECN-B--06-002

ECN/MNP (2005): Van Dril, A.W.N. & H.E. Elzenga (coörd.), Referentieramingen energie en emissies 2005-2020. ECN-C--05-018, Petten/Bilthoven, maart 2005.

Van Dril, A.W.N; Beurskens, L.W.M.; Boerakker, Y.H.A.; Daniëls, B.W.; Kroon, P.; Seebregts, A.J.; Tigchelaar, C.; Volkers, C.H.. (2007): Effecten op CO2 emissie en energiegebruik van Green4sure. ECN-E--07-034, Petten, maart 2007.

Green4Sure:

http://www.ecn.nl/en/ps/research-programme/energy-scenarios/green4sure/

www.green4sure.nl

WLO: http://www.ecn.nl/en/ps/research-programme/energy-scenarios/wlo/

-9- EOSLT05024 08-9164 confidential

3 WIND POWER IN THE NETHERLANDS The variability and the limited predictability of wind energy are expected to cause extra imbalance in the interconnected power system of the Netherlands. An estimate of the extra balancing power at system and at market participant level comprises the degree of variability and the degree of predictability for wind energy production for a given scenario of wind energy penetration. In this chapter estimates of the extra balancing power due to wind energy are presented for a scenario with 7.8 GW of installed wind power (6.0 GW offshore and 1.8 GW onshore) distributed over 7 market participants (program responsible parties) in the year 2020. These estimates originate from studies performed in the framework of the We@Sea project "System Integration and Balance Control with Large-scale Offshore Wind Energy" [1][2][3]. Wind power variability Total power produced by a mix of 4000MW wind turbines on shore and 6000MW of wind farms off shore is at any moment in time less than 10,000MW. In figure 4, wind power in all 15 minute periods in a year are ranked in decreasing order in a so called power-duration curve. Average yield is 3700MW.

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

0 5000 10000 15000 20000 25000 30000 3500015-minute periods per year

Win

dpo

wer

[MW

]

Figure 4 Power duration curve of 10,000 MW of wind power in the Netherlands

-10- EOSLT05024 08-9164 confidential At the central (=national) level the imbalance due to wind variations across 15-minute intervals is estimated to be equivalent to plus/minus 10%-11% of installed wind capacity, which corresponds to a ramping rate of 0.7% of capacity per minute (table 2). Across longer time intervals the imbalance is found to be larger, but the ramping rate understandably is smaller. Imbalance due to wind variations within 15-minute intervals, on the other hand, is estimated to be equivalent to plus/minus 0.4% of installed wind capacity, corresponding to a ramping rate of 0.5% of capacity per minute. It is expected that these amounts of imbalance can easily be handled by the ramp rate capabilities of the peaking units and the TenneT controller. Table 2 Wind power variability for different time scales

Time scale Variability (99.7% confidence interval)

Absolute [MW]

Relative [% of capacity per minute]

< 15 min. [–35 ... +35] [–0.45 ... +0.45] 15 min. [–795 ... +853] [–0.68 ... +0.73] 30 min. [–1200 ... +1222] [–0.51 ... +0.52] 1 hour [–1808 ... +1744] [–0.39 ... +0.37] 6 hour [–4678 ... +4620] [–0.17 ... +0.16]

From this table 2 it can be concluded that for the integration of wind energy in the Netherlands a substantial amount of flexible power is required. Part of this flexibility can be provided by storage systems, preferably systems that can be charged in a few hours, such as for example the batteries in plug-in hybrid vehicles. Forecasts of wind power At the central level the imbalance due to day-ahead wind forecast errors is estimated to vary between a shortage of –3907 MW (equivalent to 45% of installed wind capacity) and a surplus of +3482 MW (50% of capacity). These amounts fall beyond the ramp rate capabilities of the other units so that dedicated provisions are needed. The power forecast errors are found to be more concentrated around the mean and to have tails that decay faster than in the normal distribution, with the exception of a few isolated outliers (figure 5). The imbalance due to intra-day forecast errors is smaller because the wind forecast error decreases with forecast horizon, and even can be minimized if online production data is employed (figure 6). This shows the importance of continuous forecast updates and online data.

-11- EOSLT05024 08-9164 confidential

-6000 -4000 -2000 0 2000 4000 60000

0.2

0.4

0.6

0.8

1x 10

-3

Aggregated Forecast Error (MW)

Prob

abilit

yD

ensi

ty Sigma: 1013.16Mu: 40.73

Forecast Error: 6000+1800 MWFit to Normal Distribution

Figure 5 Aggregated power forecast error for 7800 GW of installed wind power

Figure 6 Wind power forecast error normalized to installed capacity as a function of forecast horizon and based on persistence (≤ 3h) or numerical weather prediction (> 3 h)

[1] W.W. de Boer, A.J. Brand, M. Gibescu, System balancing with 6 GW offshore wind energy in the Netherlands – Analysis of current methods for balancing (Executive Summary Work Packages 1 and 2 We@Sea Project "System Integration and Balance Control in the case of Large-scale Offshore Wind Energy"), draft, 2006.

-12- EOSLT05024 08-9164 confidential [2] W.W. de Boer, A.J. Brand, M. Gibescu, B.C. Ummels, System Balancing with 6 GW offshore wind energy in the Netherlands – Instruments for balance control (Executive Summary of Work Package 3 We@Sea Project "System Integration and Balance Control with Large-scale Offshore Wind Energy"), draft, 2007. [3] M. Gibescu, A.J. Brand, Estimation of balancing requirements for large-scale wind energy at central and market participant level in the Netherlands, Submitted to Wind Energy Journal, May 2007.

4 PLUG-IN ELECTRIC VEHICLES In the past the application of batteries for transportation has been hampered mainly by low energy density and high internal resistance. Due to recent developments these handicaps have been considerably reduced, resulting in commercially attractive possibilities for electric transportation with (motor-) bicycles, scooters, passenger cars and small till medium sized delivery vans. Such a penetration can be realized by applying batteries in hybrid cars and vans, which can be charged for small (for example commuter) distances. This results in commercially attractive cars and vans with a high public acceptance called plug-in hybrid electric vehicles (PHEVs). The environmental benefits of these PHEVs are considerable. − primary energy reductions of more than a factor two − almost zero local emissions − total well-to-wheels emission reductions of more than a factor 2 for CO2 and higher for

other pollutants − large reductions of brake dust by regenerative braking. Recent developments in the automobile production show an “explosion” of hybrid vehicle models. In these models the conventional drive train (with internal combustion engine and mechanical transmission) is combined with an electrical motor/generator and batteries. With sophisticated motor and battery management this results in higher efficiencies and inherently lower emissions. In the past some Japanese car manufacturers have taken the lead with these models and their success leads to many followers. Broadening on this concept, in which partial electrification of the drive train is introduced to improve the overall efficiency, more improvements are possible by further electrification by applying larger battery packages and electrical motors and adding the possibility of external charging of these batteries. This is the concept of the so called plug-in hybrid vehicles, see figure 7.

-13- EOSLT05024 08-9164 confidential

Figure 7 Plug-in hybrid vehicle It is important to realize that these new concepts only can be realized due to recent developments in battery technology. The most important improvements have been realized in Ni metal hydride (NiMH) and Li ion batteries. Of these technologies the NiMH batteries are mostly applied in hybrid models due to their earlier development resulting in a more reliable product at this moment than Li-ion batteries. However, the best perspectives are expected for the Li ion technology due to their higher energy density and (recently developed) high power density. The required longevity of these batteries is only claimed by a small amount of battery manufacturers and not amply shown. Costs and benefits The higher production costs for a PHEV result in higher purchase costs for a new vehicle. However, driving costs for PHEV’s are much lower than for conventional cars. An example of the profit related to driving PHEV’s is shown in table 3. Assumed car life is 300,000+ km, of which 250,000 km in electric driving mode. In this example the cost difference for driving is calculated based on consumer prices for fuel and electricity of early 2008. Because of this large difference, other cost differences (maintenance, repair, insurance) are ignored in this calculation. As shown at the bottom of the table, the undiscounted total profits over the car life are substantial.

-14- EOSLT05024 08-9164 confidential Table 3 Financial comparison of 9 kWh PHEV with conventional gasoline car for 2010

Ref.gasoline PHEV

Specific fuel consumption 7 L/100km 15 kWh/100 km

Km/cycle 47

Fuel/cycle 3.26 L 8 kWh

Fuel price 2008 € 1.50/L € 0.17/kWh *

Fuel costs/cycle € 4.90 € 1.36

Amount of cycles/car life - 5357**

Fuel costs/250.000 km** € 26250 € 7286

Car purchase cost (2010) € 19560 € 27895

Total profits during car life € 10629 * Typical domestic night tariff ** Assuming 250,000 km electric driving for the PHEV Conclusion Due to the much lower direct driving costs the higher investment costs for buying a PHEV are earned back amply within the life of the vehicle. Market expectations for PHEV EPRI has published in July 2007 a report, in a PHEV market study is included. Figure 8 shows a result of this study. Three distinct market adoption scenarios, each based on PHEVs entering the market in 2010 and achieving maximum new vehicle market share in 2050, are studied. As shown in the table in figure 3.1, PHEVs reach a maximum of 20% new vehicle market share in the Low PHEV scenario, 62% in the Medium PHEV scenario, and 80% in the High PHEV scenario. Comparing the situation concerning PHEVs in the US and the Netherlands there are several striking differences: − The fuel costs in the Netherlands are much higher than in the US. This results in much

more absolute cost reduction for driving in the Netherlands, if conventional cars are replaced by (PH)EVs

− The emissions of power stations for electricity production are the Netherlands much less than in the US, resulting in more environmental benefits by changing to (PH)EVs

− In the Netherlands there are nearly no car producers for conventional cars; a change to a new type of automobiles may challenge the Dutch industry to make (components for) that new type without much harm for existing industry

-15- EOSLT05024 08-9164 confidential − The developments for PHEVs are in the US much further than in Europe (and in the

Netherlands). This will result in a slower start of penetration in the Netherlands. These differences make the Low PHEV scenario for our country less probable with respect to maximum penetration, but a slow start is likely. In this report the assumed penetration for PHEVs in 2020 and 2040 are therefore respectively 0.1 and 0.6.

Figure 8 Expectations of EPRI concerning penetration for PHEVs in the US. On top: table with 3 introduction scenarios for 2050. Bottom figure: introduction in Medium Penetration scenario, the most likely introduction scenario for the Netherlands as expected by the ITM project team

-16- EOSLT05024 08-9164 confidential 5 HEAT PUMPS Introduction in the field of heat pumps Heat pumps can provide year-round climate control for a house or commercial building, by supplying heat in the winter and cooling it in the summer. Some types of heat pumps can also generate hot tap water. This multi-functionality makes the heat pump very attractive. Moreover a heat pump system can have a high energy efficiency and low operational costs. Nevertheless, it is more expensive to buy a complete heat pump system than to buy a conventional heating system. Besides, it is necessary to make some adaptations to the building, like low-temperature radiators and floor/wall heating. Therefore it is mainly attractive for newly built buildings. This explains why often other systems are chosen, despite of all the environmental advantages of a heat pump. In a heat pump heat from the low-temperature heat source with a low temperature (around 10-20 oC) is converted into usable heat with a higher temperature (up to 55 oC). This usable heat can be applied for several applications, e.g. space heating or hot tap water. The operation of the process of a heat pump requires energy, but much less energy than the usable heat that is generated. The heat pump transfers the heat by circulating a refrigerant in a closed system in a cycle of evaporation and condensation. A compressor pumps this refrigerant as a vapour from the low pressure side to the high temperature side. In one heat exchanger coil, the refrigerant is evaporated at low temperature and pressure and absorbs heat from the low temperature heat source. The refrigerant is then compressed and flows to the other heat exchanger, where it condenses at high pressure. At this location, it releases the usable heat at the high temperature. See figure 9.

-17- EOSLT05024 08-9164 confidential

Figure 9 Basic heat pump cycle [1]

Since ground and air outside always contain some heat (as long as T > 0 K or -273 oC), a heat pump can supply heat to a building even on cold winter days. The heat source can be ground water, air or surface water. The current standard however, is to create a doublet ground storage system with a cold and a warm reservoir. When the heat pump has to generate heat, the source heat is taken from the warm reservoir (+/- 20oC) which enables the heat pump to perform with a very good efficiency. The source heat medium leaves the heat pump with a reduced temperature and is fed to the cold reservoir. In summer times when there is a net cooling demand, the medium from the cold reservoir can be used directly, i.e. without interference of the heat pump compressor, for space conditioning purposes. In the process of supplying cold, the medium itself has picked up energy and increased in temperature, and is fed to the warm reservoir. The summer operation requires only very minor fluid transportation pump energy and regenerates the warm reservoir for heat generation the next winter. For one unit of heat from a heat pump, about 0.25 to 0.5 units of energy are needed (electricity, gas or heat). Heat pump scenarios in the WLO Study The WLO study (Welvaart en Leefomgeving) contains four different scenarios for the period 2005-2040. For the ITM project, the penetration of heat pumps in the ‘Strong Europe scenario’ (SE) and the ‘Global Economy High Oil Prices scenario’ (GEHP) are analyzed in more detail. A distinction has been made between heat pumps for space heating and heat pump boilers.

-18- EOSLT05024 08-9164 confidential The ITM project distinguishes heat pumps that are used for space heating with about 1000 – 3000 hours of operation per year and heat pump boilers used for tap water heating1. In Table 4 the electricity input and the heat production of the heat pumps for space heating are shown for both the residential sector and the service sectors for the SE and the GEHP scenarios. Where the SE and GEHP scenarios for the electric heat pumps differ from each other in the residential sector, they are exactly the same in the service sector. Table 4 Electricity input for and heat production due to electric heat pumps for space

heating [7] 2000 2005 2010 2015 2020 2025 2030 2035 2040

Heat pumps households, SEelectricity input PJe 0.0 0.1 0.1 0.2 0.2 0.2 0.3 0.3 0.4 heat production for space heating PJth 0.1 0.4 0.7 1.1 1.0 1.2 1.4 1.7 2.0 Heat pumps services sector, SEelectricity input PJe 0.3 0.9 1.5 2.1 2.6 3.2 3.8 4.3 4.9 heat production PJth 1.0 2.9 4.7 6.4 8.2 10.0 11.8 13.6 15.4 Heat pumps households, GE and GEHPelectricity input PJe 0.0 0.1 0.2 0.2 0.3 0.3 0.3 0.3 0.3 heat production for space heating PJth 0.1 0.4 0.8 1.2 1.7 1.7 1.7 1.7 1.7 Heat pumps services sector, GE and GEHPelectricity input PJe 0.3 0.9 1.5 2.1 2.6 3.2 3.8 4.3 4.9 heat production PJth 1.0 2.9 4.7 6.4 8.2 10.0 11.8 13.6 15.4

The electricity input and heat output for the household sector in table 4 are related to space heating only, as the numbers are corrected for heat pump boilers. In the service sector the number of heat pump boilers is very low in relation to the number of heat pumps for space heating. Therefore, also for the service sector the electricity input and heat output can be considered as related to space heating even though no correction has been made for heat pump boilers. For tap water heating the numbers are given in table 5: Table 5 Contribution of electric heat pumps to heating of tap water (TW) in dwellings

GE en GE (high oil price level) 2000 2005 2010 2015 2020 2025 2030 2035 2040 Heat output EHP tap water (PJth) 0,1 0,2 0,4 0,6 0,8 0,8 0,8 0,8 0,8 Number of dwellings with TW heating with HP 10.000 20.000 40.000 62.000 90.000 89.000 89.000 88.000 87.000

SE 2000 2005 2010 2015 2020 2025 2030 2035 2040 Heat output EHP tap water (PJth ) 0,1 0,2 0,4 0,6 0,6 0,7 0,8 1,0 1,1 Number of dwellings with TW heating with HP 10.000 20.000 41.000 65.000 62.000 78.000 94.000 111.000 127.000

Three different options for the introduction of heat pumps in the Netherlands To predict the introduction of heat pumps in the Netherlands, three different options are defined. Two of these options can be seen as variations to the WLO scenarios. So these

1 1 Operating hours are based on Protocol Monitoring Duurzame Energie

-19- EOSLT05024 08-9164 confidential options are not calculated in the scope of the WLO study. They are just rough indications of possible trends for the further penetration of electric heat pumps in the Netherlands. 1 WLO scenario: A scenario based on the continuation of the current strategy (GEHP and

SE) leads to about 4 PJ of avoided primary energy in 2020, independent on the choice of GE or SE scenario (with about 40.000 heat pumps for space heating and respectively 60.000 to 90.000 for hot tap water generation in 2020).

2 Objective: The optimistic option in which the objectives of the Dutch government are

realized. The Dutch government will define their objectives this year. Based on information of the Platform Duurzame Electriciteit, the objective is estimated at 30 PJ of primary energy in 2020. This requires more than 400.000 heat pumps for space heating in the Netherlands. Of course the exact number depends on the size, COP and the number of heat pump boilers.

3 Transition: The last option is based on the idea that a real transition will be realized to

introduce the heat pumps in the Dutch market. So this option is the real maximum, that assumes that in each situation that a heat pump is feasible, a heat pump will be used. This leads to more than 900.000 heat pumps in 2020.

Number of heat pumps in each option The assumptions for the first option are already described in chapter 2. Option 2 is defined by the numbers given the total amount of heat pumps that are necessary. So if we want to realize the goal of 30 PJ in 2020, then 400.000 heat pumps are required. Of course the exact number depends on the size, COP and the number of heat pump boilers The last option, the transition option, does not set a goal in PJ per year. This option is based on the maximum amount of heat pumps that can be realized in the Netherlands. This paragraph will describe the used assumptions. Table 6 shows the assumptions for the number of heat pumps that can be installed in the Netherlands. Table 6 Possible number of heat pumps per year

Amount Number of buildings with a heat pump Newly built dwellings 70.000/year 50% = 35.000 Replacement dwellings 450.000/year 5% = 22.500 Renovation dwellings 300.000/year 30% = 90.000 Newly built offices 5000/year 50% = 2.500 Replacement offices 50.000/year 10% = 5.000 Total maximum 136.000 /year

-20- EOSLT05024 08-9164 confidential So the maximum amount of heat pumps (for space heating as well as tap water heating) is about 136.000 heat pumps per year until 2020. This is a very optimistic option. Of course we need some time for this large introduction. Therefore in the coming years the introduction will increase, growing in numbers to 130.000/year. This leads to +/- 900.000 heat pumps in the Netherlands in 2020. For the determination of the maximum effects of the heat pumps we assume that all these heat pumps are used for both space heating and tap water heating. Figure 10 shows the total number of heat pumps as determined for the three options.

0

200000

400000

600000

800000

1000000

1200000

1400000

1600000

2000 2010 2020 2030

year

tota

lnum

bero

fhea

tpum

ps

WLO

Objective

Transition

Figure 10 Options for introduction of heat pumps in the Netherlands

-21- EOSLT05024 08-9164 confidential 6 ELECTRICITY NETWORKS2

There is a trend towards unpredictable fluctuations in the amount of power supplied to the grid because of the growing share of wind farms and a demand-side trend towards higher power consumption, particularly associated with electric-powered vehicles. These trends have implications for the existing three-tier grid system, with its HV, MV and LV levels. This chapter examines the present structure and considers how the existing grid is affected by the trends. The present structure of the grid is well established: there have been changes at various points, but no large structural changes have been made to the grid for many years. The present low-voltage grid has not received attention, because its structure is more or less fixed, with the result that little or no optimization has been possible. If there should be any substantial integration of small dispersed generating units (e.g. PV systems) and large dispersed consumers (e.g. heat pumps), this could have a major impact on the performance of the existing low-voltage grid. It is therefore appropriate to consider whether changes to the control arrangements or design of the existing grid – with a view at achieving a better balance between supply of and demand for electricity3 – are necessary, desirable and possible. Existing infrastructure4

Before a detailed analysis of the possible design and performance shortcomings of the existing low-voltage grid can be made, it is necessary to describe the existing three-tier structure, which consists of a high-voltage (HV) grid, medium voltage (MV) grids and low-voltage (LV) grids.

2 This section is based largely on the report Methodology for Designing Low-Voltage Grids with Numerous Small Dispersed Generators KEMA, 2001 3 In the past, the term ‘load’ was generally used to describe power consumed by connected users. This made sense, since power consumption was directly related to the load on the cables and other network components. However, since power generated also contributes to the load placed on the cables, we have chosen to use the word ‘load’ to refer to all current passing along the conductors in the electricity grid, regardless of where the current comes from or goes to. Power consumption by parties connected to the grid is referred to as ‘demand’, and power supplied to the grid as supply. KEMA, 2001. 4 This section is based primarily on the report An Exploratory Study of Hybrid (AC+DC) Medium-Voltage Grids with Dispersed Generation and Storage. KEMA, 1999.

-22- EOSLT05024 08-9164 confidential High-voltage grid The existing HV transmission grid operates at several voltage levels:

- 380 kV and 220 kV is used for the national transmission grid and international interconnections;

- 150 kV and 110 kV (in the north-east Netherlands) are used for the regional transmission grids.

The Medium-voltage grids Most of the Netherlands’ medium-voltage grids operate at 10 kV. The medium-voltage grid has a mesh structure for the most part, but is in general operated radially5. The MV system features a single supply point and centralized control. The medium-voltage stations are not remotely controllable; switching operations have to be performed locally by hand. The Low-voltage grids Most of the existing low-voltage grids are structured and operated radially, although there are some meshed grids in the urban areas. The LV grids operate at 230/400 V. Where necessary, the LV grids are provided with branches. Unlike the MV grids,6 the LV grids are neither reconfigurable nor harmonized. Reconfiguration is possible, however, only in the meshed LV grids. Issues in the LV grid7

Over the years, the grid described above has been affected by various changes in the way it is used, mainly the increase in demand and the increase in dispersed generation. A number of associated problems for the existing infrastructure at the LV level are listed below. 1 Voltage quality. Most electricity companies have experienced voltage quality problems.

Such problems mainly affect predominantly rural regions. Because rural areas tend to be located at the far end of cables, the supply voltage is often lower than it should be and more prone to fluctuation. On the other hand, if a CHP unit is located in such an area, the voltage is apt to be too high. Dispersed generation therefore necessitates a very different approach to rural grid management. Urban regions generally experience such problems to a much lesser extent.

2 (Over)loading. Overloads typically occur because of growth in the demand for electricity and in case of faults. There are no structural overload problems, with a few incidental exceptions. Overload problems can always be avoided by good management and

5 Some 20% is operated meshed with n-1 conditions 6 Some 90 per cent of the existing MV grid is reconfigurable, however not remotely, therefore this is not applied in practice for economic reasons. 7 This section is based largely on the report State-of-the-Art LV Grids. Design, Operation and Problems. KEMA, 1998.

-23- EOSLT05024 08-9164 confidential

anticipation; if a prospective client is expected to cause an unacceptable load increase, the infrastructure is upgraded prior to connection.

3 Asymmetrical network loading. Asymmetry in excess of the Grid Code is not permitted. In an LV grid, asymmetry is measured by direct and/or indicative tests, such as measuring the current per phase in a station, thermographic inspection and performing visual discoloration checks of the connection point indicators (KEMA, 1998).

4 Fault tracing. Fault tracing in LV grids varies from one electricity company to the next. Tracking down faults raises both technical and organizational issues. The technical issues often depend on local circumstances.

5 Environment and safety. Repairs usually have to be performed under non-energized conditions. However, the rules (NEN 50110) do allow repairs to be carried out under low-voltage conditions, subject to certain conditions.

National load developments In figure 11, information is presented on the highest load on the Dutch high-voltage grid, as recorded in December 2006. December is normally the month with the highest load, because the short days and low temperatures drive demand for heat and light.

Figure 11 Highest load on the Dutch HV grid (TenneT, 2006)

Although efforts are made to control the grid load by supply-demand balancing, achieving the desired goal tends to be not easy in practice. Thus far, load controls are applied to the HV grid only. Action on the lower tiers of the grid structure is more complex and harder to implement. The party with program responsibility (PRP) is responsible for supply-demand balancing. Parties connected to the grid are required to indicate to the PRP how much electricity they are likely to produce and consume over time. Ideally, supply will equal demand at any given point in time; in practice, however, this is not the case. There are

-24- EOSLT05024 08-9164 confidential always discrepancies in the load and the production levels (due to producers having technical problems, the weather not being as expected, etc), resulting in imbalance, which is corrected by TenneT. TenneT organizes the market for control capacity for the compensation of imbalances. The graph below (figure 12) shows daily imbalances, which are a measure of the quality of the forecasts and the market response. In an ideal world, there would be no surpluses or shortages. Supply-demand imbalances cost money, so maintaining adherence to the production and demand (e.g. by load control) is financially attractive.

Figure 12 PRPs’ daily imbalances in 2006 (TenneT, 2006)

Load profiles for household connections (E1) For all connections without automatic meter reading categories of similar use have been defined. Load profiles are used to represent an average consumption pattern for all connections in the same category. Most households fall in category E1: connections with a current capacity =< 3x25 amps. Overall consumption is not greatly affected by the day of the week: the amount of electricity consumed is about the same on a weekend day as on a weekday (top left-hand graph, figure 13). On all weekdays, there is a peak in consumption at 23:00: the time that overnight tariffs come into effect (lower left-hand graph). There is no similar peak just before the overnight tariff period ends at 07:00.

-25- EOSLT05024 08-9164 confidential

Figure 13 E1 consumption on the various days of the week (upper left-hand graph); E1 daily consumption patterns (Monday=1; upper right-hand graph); E1 evening consumption pattern (lower left-hand graph); E1 overnight consumption (lower right-hand graph). All profiles exclude public holidays

References − Ecofys, 2003, Profielenmethodiek. − Energeia, 2007, CBS: Totale elektriciteitsproductie daalt in 2006, decentrale productie

stijgt licht. − EnergieNed, 1996, Elektriciteitsdistributienetten. − KEMA, 1998, Hulshorst, W.T.J., Mesland, A.J., State of the art LS-netten. Ontwerp,

Bedrijfsvoering en Knelpunten. − KEMA, 1999, Vaessen, P.T.M., Verkennende studie hybride (AC + DC) midden-

spanningsnetten met decentrale opwekking en opslag.

Median 25%-75% Non-Outlier Range OutliersExtremes1 2 3 4 5 6 7

day (1=monday)

0.000010

0.000015

0.000020

0.000025

0.000030

0.000035

0.000040

0.000045

0.000050

0.000055

0.000060

E1a

start timeE1a

E1a/day: 1E1a/day: 2E1a/day: 3E1a/day: 4E1a/day: 5E1a/day: 6E1a/day: 70.00

1.753.50

5.257.00

8.7510.50

12.2514.00

15.7517.50

19.2521.00

22.75

0.000010

0.000015

0.000020

0.000025

0.000030

0.000035

0.000040

0.000045

0.000050

start time

E1a

E1a/day: 1E1a/day: 2E1a/day: 3E1a/day: 4E1a/day: 5E1a/day: 6E1a/day: 7

18.5 19.0 19.5 20.0 20.5 21.0 21.5 22.0 22.5 23.0 23.50.000025

0.000030

0.000035

0.000040

0.000045

start time

E1a

E1a/day: 1E1a/day: 2E1a/day: 3E1a/day: 4E1a/day: 5E1a/day: 6E1a/day: 7

0.00 0.75 1.50 2.25 3.00 3.75 4.50 5.25 6.00 6.75 7.50

0.000015

0.000020

0.000025

0.000030

-26- EOSLT05024 08-9164 confidential − KEMA, 2001, Remijn, N.J.B.M., Ontwerpmethodiek laagspanningsnetten met veel kleine

decentrale opwekkers − KEMA, 2002, Remijn, N.J.B.M., Grootschalige inpassing van kleine opwekkers in

laagspanningsnetten, Deel 1: Ontwerpmethodiek. − KEMA, 2004, Optimale leveringszekerheid, Methodiek voor het maatschappelijk

optimum, PREGO 14. − KEMA, 2005, The contribution to distribution fault levels from the connection of

distributed generation, DTI (UK) − Tennet, 2006, Transportbalans

7 DISCUSSION In table 7 the introduction scenarios for electric heat pumps and plug-in hybrid vehicles are summarized, based on the most positive scenarios from chapter 4 and 5. When all heat pumps would be operating simultaneously and all electric vehicles are being charged, their total load amounts to 5.4 GW in 2020 and 11.0 GW in 2040. These numbers are in the same order of magnitude as the current difference in load between peak and valley, which was 7 GW in 2006 (see figure 11). Table 7 Summary of plug-in vehicle and heat pump scenarios

2020 2040 Plug-in vehicles: #Plug-in vehicles (millions) 0.9 6.5 Total charging capacity @3kW: (GW) 2.7 6.5 Annual electricity demand (TWh/a) 2.6 19.0

Electric Heat pumps: # heat pumps (millions) 0.9 1.5 Total installed capacity @3kW: (GW) 2.7 4.5 Annual electricity demand (TWh/a) 3.2 5.3

-27- EOSLT05024 08-9164 confidential Contribution of heat pumps and PHEVs to reduce daily variations in electricity demand Electricity demand from heat pumps and electric vehicles amount to about 15% of total final electricity demand in 2040 (see table 8). The load factor (peak load divided by average load) in the Netherlands in 2006 was about 80%. Assuming a similar percentage for 2020 and 2040, it can be shown that heat pumps and electric vehicles combined have the potential to reduce the ‘night through’ by 21% in 2020 and 76% in 2040. This assumes that the heat storage is sufficiently large that all electricity consumption of heat pumps can be shifted to the off-peak hours, and similarly, that all the battery charging takes place in the off-peak hours. This will certainly not be true for 100% of the cases. But this shows that both electric heat pumps and electric vehicles together can provide a substantial contribution to improving stability in future electricity grids. Table 8 Share of plug-in vehicle and heat pump electricity demand as percentage of final

electricity demand in SE scenario, and their contribution to filling the ‘night trough’

2020 2040

Final electricity demand in SE scenario (TWh/a) 137 161 Electricity demand heat pumps + EV (TWh/a) 5.8 24.3 Heat pumps and EV as percentage of SE: (%) 4.2 15.1

Annual electricity demand to create a flat load curve (at load factor of 0.8) (TWh/a)

27 32

Heat pump and EV contribution to a flat load curve (%) 21 76