Project Name Document Name Frequency Sensitive Electric Vehicle and Heat Pump Power Consumption Final report for National Grid July 17th 2015 Element Energy Limited Terrington House 13-15 Hills Road Cambridge CB2 1NL Tel: 01223 852499 Fax: 01223 353475 1 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Contents 1 2 Executive Summary........................................................................................................ 1 1.1 Potential - Electric Vehicles ....................................................................................... 3 1.2 Potential - Heat Pumps ............................................................................................. 4 1.3 Development path ..................................................................................................... 5 Introduction ..................................................................................................................... 7 2.1 Background ............................................................................................................... 7 2.2 Report structure ......................................................................................................... 7 PART 1 - Method .................................................................................................................... 9 3 4 Description of archetypes ............................................................................................. 10 3.1 Electric Vehicles ...................................................................................................... 10 3.2 Heat Pumps ............................................................................................................. 16 National Grid Service Requirements ............................................................................ 23 4.1 Future frequency response requirements ............................................................... 24 4.2 Seasonal and diurnal variation in response requirement ........................................ 25 4.3 Modelling method .................................................................................................... 25 PART 2 - Results .................................................................................................................. 27 5 6 7 Potential Frequency Response Contribution over Time............................................... 28 5.1 Potential future response from EVs ........................................................................ 28 5.2 Potential future response from heat pumps ............................................................ 33 Commercial potential .................................................................................................... 37 6.1 Costs of frequency response provision ................................................................... 37 6.2 Revenues per archetype ......................................................................................... 40 6.3 Cost benefit analysis ............................................................................................... 41 6.4 CO2 emissions impact ............................................................................................. 42 6.5 Commercial models ................................................................................................. 43 Technical barriers and challenges ................................................................................ 51 7.1 Electric Vehicles ...................................................................................................... 52 7.2 Heat Pumps ............................................................................................................. 53 7.3 Verification and telemetry ........................................................................................ 54 7.4 Network impacts and system operation .................................................................. 55 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 8 Roll Out ......................................................................................................................... 56 8.1 Summary of key barriers for the development of EV and heat pump frequency response ........................................................................................................................... 57 9 8.2 Comparison of commercial procurement and mandatory provision ........................ 59 8.3 Next steps for EV and heat pump frequency response roll out ............................... 62 Conclusion .................................................................................................................... 64 Acknowledgements .............................................................................................................. 65 Glossary ............................................................................................................................... 66 References ........................................................................................................................... 68 Authors For comments or queries please contact: joris.besseling@element-energy.co.uk rebecca.feeney-barry@element-energy.co.uk shane.slater@element-energy.co.uk Transmission.ukfes@nationalgrid.com Frequency Sensitive EV and Heat Pump Power Consumption Final Report 1 Executive Summary National Grid has a licence obligation to control electricity system frequency within statutory limits around 50Hz. System frequency is determined and controlled by the real time balance between system demand and total generation. Historically, flexible electricity generation is used to provide frequency response. The electricity industry and the System Operator face new challenges as Great Britain decarbonises. National Grid analysis for the 2014 System Operability Framework (SOF) showed increasing challenges to managing system frequency within statutory limits beyond 2020/2025. This is driven by increasing levels of intermittent renewable generation and a consequent reduction in synchronised conventional generation. At the same time, the uptake of new technologies such as electric vehicles (EVs) and heat pumps is expected to increase. The corresponding increase in electricity demand may pose further challenges to balancing the electricity system. One of the four options identified in the SOF to address this challenge is to provide frequency response by demand side management; using a large number of small loads that have the potential to provide in aggregate a similar response capacity to a few large generation plants. Different small loads could provide frequency response, and loads with inherent storage or inertia are particularly suited, for example EVs and heat pumps. In response, National Grid commissioned Element Energy to carry out this Network Innovation Allowance (NIA) project to assess the potential for EVs and heat pumps to contribute to frequency response. A model was developed to assess the technical potential of EVs and heat pumps to contribute to frequency response. The model schedules EV charging and heat pump consumption for different power consumption archetypes, using historic traces for frequency response requirements. Charging and consumption scheduling is constrained to guarantee there is no adverse impact on the quality of service delivered to the customer. In this study EV charging flexibility is provided by interrupting charging or changing charge rates, but EV supply of electricity back to the grid (vehicle to grid) is excluded. The projection of frequency response potential over time is further informed by National Grid and external scenarios for the uptake of EVs and heat pumps. The study then determines potential costs to technically enable frequency response and operate a frequency response business, summarises the key technical and operational challenges, reviews the commercial models that could support deployment, and identifies possible next steps for the roll out of EV and heat pump frequency response service. As part of the study, interviews were carried out with 22 stakeholders, including EV and heat pump manufacturers, EV charge point manufacturers and operators, EV fleet operators, electricity suppliers, DNOs and demand side aggregators. 1 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Key results for 2030: combined potential of EVs & heat pumps 1200MW combined annual average low frequency response potential Corresponding to an average 82% of projected 2030 requirement Representing £100 million per year revenues based on current prices The analysis shows there is a significant technical potential for both EVs and heat pumps to contribute to frequency response by 2030. The combined annual average potential may reach 1200MW by 2030 in the medium uptake scenario (excluding residential heat pumps). This corresponds to an average 82% of National Grid’s projected frequency response requirement in 2030. The overall demand side potential may be significantly higher when other loads, for example other electric heating technologies like storage heaters or white goods like fridges, are also taken into account. There are no fundamental technical limitations that prevent the implementation of frequency responsive electricity consumption for either EVs or heat pumps. However, further technical development is required for both technologies to realise practical frequency response capability. Most heat pumps require further design iterations to meet dynamic frequency response requirements in a safe and reliable manner. Key development is also required in the control and communication systems, both between EVs and charge points and heat pumps and Energy Management Systems, as well as with demand side response (DSR) management platforms. The annual average potential of EVs and heat pumps in the 2030 medium scenario corresponds to approximately £100 million in annual revenues, based on current tendered prices. This value could potentially be higher if the higher effectiveness of fast responding demand assets in restoring system frequency deviations is taken into account in pricing. While the overall cost of providing frequency response may be very competitive compared to conventional thermal plants, the potential net benefit per household may be relatively limited. This may risk limiting the participation of EV and heat pump owners in frequency response schemes. Therefore, deployment will need to be supported by very efficient commercial models or mandated. Commercial models may support combining frequency response with other DSR uses, increasing the value of customer propositions. Moreover equitable remuneration could be provided for fast and accurate responding demand assets which may provide more effective frequency response management. Alternatively frequency response provision could be mandated. However, this could inhibit the development of innovative commercial models, charging and consumption management (e.g. bidirectional charging for EVs) and the effective combination of different DSM services. Based on the modelling and analysis conducted for this report, the key findings are detailed below. 2 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Key results: EVs Key results: heat pumps 754MW annual average low frequency response potential Corresponding to 52% of projected 2030 requirement Representing £66 million per year based on current prices Able to provide firm response availability over the night when combined with controlled smart charging No fundamental technical limitations to develop frequency responsive capability Potential revenues of £45/EV/year corresponding to a net annual benefit of £26/year for a home charging EV and £35/year for a fleet EV Potential to avoid around 1 tonne CO2 per year per EV 447MW annual average low frequency response potential from commercial and industrial heat pumps Corresponding to 30% of projected 2030 requirement Representing £39 million per year based on current prices Strong diurnal profile with limited response availability over the night No fundamental technical limitations to develop frequency responsive capability Potential revenues of £70/year for a residential heat pump and £23/kW/year for a commercial heat pump, corresponding to net annual benefits of £51/year and £21/kW/year respectively Potential to avoid 0.6 tonnes CO2 per year per kW electric for a commercial heat pump and 2 tonnes per year for a residential heat pump 1.1 Potential - Electric Vehicles There is significant potential to provide frequency response via interrupted charging because electric vehicles are projected to typically be available for charging for 8 hours per day, but only require charging for 3 hours (for the home charging archetype in the analysis) EV batteries and battery management systems are typically capable of meeting frequency response service requirements, with very high response rates and accuracy. This near instantaneous response capability represents a further opportunity, potentially reducing the overall required volume of frequency response. This also implies that the contribution of EVs may be higher than 52% of the total frequency response requirement in the medium uptake scenario, if their fast response capability is utilised in frequency response management. Current charge points are however not typically designed with all of the required control, instrumentation and communication capabilities for providing frequency response. There are no technical barriers to adapt standard chargers (mode 3 and 4) in design iterations, and additional hardware costs that meet specifications suited for frequency response are estimated to be around £10. The impact of dynamic frequency response on charging management and e.g. safety measures has also not yet been tested in the field. 3 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Total annual costs for frequency provision by a home charging EV are estimated to be £19, including hardware, operational and overhead business costs. The total 1 for a fleet EV is lower, at £10, because the overhead cost is spread over the fleet . This gives a net annual benefit of £26 for a home charging EV and £35 for a fleet EV. This also demonstrates that average EV frequency response costs may be lower than opportunity costs of conventional FR providers. Currently there is no standardised protocol for communication between charge points and aggregator management systems that are interoperable and support frequency response requirements 1.2 Potential - Heat Pumps There is a significant potential to provide frequency response by interrupting consumption of heat pumps that are installed with thermal storage (e.g. a hot water tank). The storage enables maintaining temperatures within customer 2 settings. Alternative opportunities include hybrid heat pumps that can switch to natural gas, or to use the thermal inertia in the dwelling fabric. The potential additional contribution of residential heat pumps may be very high (approximately 2GW in the medium scenario in 2030). However this is very dependent on the uptake of heat pumps in the residential sector, especially of retrofits with space heating hot water storage. The analysis is very sensitive to the extent to which heat pumps are installed with thermal storage to provide flexibility, or the thermal inertia of dwellings, which is very dwelling specific. There is significant seasonal variation in heat pump response, as the response available in January is nearly five times the response available in July. Not all current heat pumps are able to meet ramp up speed of response requirements for frequency response, while ramp down rates may be limited for control and safety reasons. These issues can be addressed in the periodic redesign of heat pumps, but require a driver for manufacturers to do so. Currently there is no standardised protocol for communication between heat pumps and aggregator management systems that are interoperable and support frequency response requirements Additional costs to adapt controls and enable metering may be higher than for EVs. Total annual costs for frequency provision by a residential heat pump are estimated to be £20, including hardware, operational and overhead business 3 costs. The total for a 10kW commercial heat pump is £24 . This gives a net annual benefit of £51 for a residential heat pump and £206 for a 10kW electric commercial heat pump. This also demonstrates that average heat pump frequency response costs may be lower than opportunity costs of conventional FR providers. The residential heat pump archetype has higher energy consumption than the EV home charging archetype, and slightly higher uptake numbers. Consequently the overall frequency response potential, as well as individual asset revenues, is higher for the heat pump archetype. However EVs may be able to provide faster response, providing more effective frequency management and require less technical development to enable frequency 1 Assumes a fleet of 10 EVs. Annual costs are calculated over 15 years. 2responsive consumption. EVs provide moreover more flexibility in charging than heat Not modelled explicitly in this study 3pump consumption. This enables EVs, when frequency sensitive consumption is Annual costs calculated over 15 years. combined with controlled smart charging, to provide more firm response throughout the night period when National Grid requirements are highest. 4 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 1.3 Development path Large fleets of kW scale EVs and heat pumps have different characteristics and capabilities compared to conventional synchronised generators. Distributed loads that are able to respond faster and more accurate than conventional generators, may become increasingly valuable in helping to manage system frequency. A review of contractual terms and conditions may be required to utilise the potentially more effective system frequency management capabilities, and accommodate e.g. specific reliability and availability characteristics. Stakeholders interviewed in this study identified that they need clear indication of National Grid future frequency response requirements and confidence in long term market value. For EVs a commercial procurement model appears to be the most appropriate model to develop. The study shows clear drivers to actively support the development of these opportunities in the near future. For heat pumps both mandating and commercial procurement may be appropriate. In parallel to further technology development the impact and benefits of both may be investigated in more detail. The development of commercial models and technologies requires investments from technology developers and third parties developing technologies and customer propositions. These investments can be de-risked by a dual approach supporting technology and market development. This may provide industry confidence on the long term access, value and market size for these solutions. 1.3.1 Next steps for EV and heat pump frequency response roll out Supporting market development should focus on providing confidence that there will be a market in the long term that is accessible and amenable to frequency response from EVs and heat pumps, and values providers based on their effective contribution to resolving frequency management. 1. Near term – strategic agenda and business case Early developments may be supported by clear statements showing that Frequency Response from a large portfolio of low power (kW scale) assets is being actively explored as part of strategic agendas for system management. Technology developers and aggregators can be supported in overcoming technical barriers and uncertainty around real life business models, by supporting demonstrations that prove the technology and commercial structures, in which; National Grid provides clarity on service requirements (response rates, response lengths, reliability), technical requirements (power metering, verification) and contractual terms. National Grid defines baseline methodology used to identify an asset’s contribution to frequency response 2. Medium term – enabling deployment Provide commercial opportunities for early adopters through commercial bilateral contracts with less stringent technical requirements or novel commercial models. 5 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Collaborate with DNOs to identify impact on diversity. Develop industry rules and processes to create transparency between aggregators, National Grid, DNOs and suppliers on location and utilisation of frequency responsive demand. Develop processes to mitigate or limit adverse local network impacts (thermal overload or voltage stability issues). Collaborate with aggregators and suppliers to identify impact on settlement (system balance) or supplier imbalance if it is not the response aggregator. Depending on the implementation of half hourly settlement, new SSCs may need to be developed for different types of response providers. 3. Long term – market access Develop services which remunerate higher responsiveness that some demand side technologies may provide (fast response capability, accurate tracking of frequency deviations), and take into account challenges and limitations of demand side technologies (short lead times for capacity nomination, profiled bids). Implement industry processes and rules providing transparency between DNOs, TSO and aggregators on location and utilisation of frequency response, and implement processes to mitigate adverse local network impacts 1.3.2 Technology development For both electric vehicles and heat pumps there are no fundamental technical barriers to providing frequency response services. However further technical development is required for both, especially for heat pumps, to realise practical frequency response capability; EV hardware is technically capable of providing frequency response. Next steps in technology development may focus on trialling control, automation and telemetry in real world conditions, prior to commercial deployment. Heat pumps are less technically ready. Next steps in technology development may focus on developing the hardware capability. This would help to guarantee reliable and safe operation for frequency response. Technological developments could then be followed by real world trials focusing on control, automation and telemetry in real world conditions, prior to commercial deployment. The analysis has shown that with increasing uptake of EVs and heat pumps these demand side technologies may be able to contribute significantly towards National Grid’s future frequency response needs, at a competitive cost compared to conventional solutions. There are no fundamental technical limitations to implement frequency responsive electricity consumption for these technologies. However further technology development is required, especially on control and automation, and communication and telemetry. The technical capabilities of these assets may enable more effective frequency management, and reduce the overall need for frequency response. However the value at an individual asset level is diluted, which may limit the development of these opportunities. The development of commercial models and technologies requires investments from technology developers and third parties developing technologies and customer propositions. These investments can be de-risked by a dual approach supporting technology development and supporting commercial markets by National Grid, providing confidence on the long term access, equitable remuneration and realistic market size for these solutions. 6 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 2 Introduction 2.1 Background National Grid has a licence obligation to control system frequency within the limits specified in the ‘Electricity Supply Regulations’; +/- 1% of the nominal frequency of 50.00Hz. System frequency is determined and controlled by the real time balance between system demand and total generation. UK electricity demands are met by a mixture of conventional synchronous generation, which has high levels of mechanical inertia, and zero inertia non-synchronous renewable generation. When there is an event that triggers a frequency deviation (for example, a plant outage), the system inertia provided by conventional generation slows the rate of change of frequency, thus helping ensure system stability. Frequency response providers are then used to arrest the frequency drop and restore system frequency. Historically flexible electricity generation is used to provide frequency response. Over the coming decades, intermittent renewable generation will account for a greater proportion of the UK’s electricity supply. This will reduce the amount of mechanical inertia available in the system. Without mitigating actions, National Grid will require a larger volume of frequency response, while intermittent renewable generation is less flexible and is currently less able to provide frequency response services. In addition, the uptake of low carbon demand technologies such as electric vehicles and heat pumps are projected to increase. These loads have the potential to increase electricity demand, change the demand profile, and, without appropriate measures, these loads could be inflexible. Therefore the way in which the electricity system is balanced will need to change in order to remain economic, efficient and coordinated. One of the options identified in National Grid’s System Operability Framework (SOF) to address this challenge is to provide frequency response by demand side management. This involves using a large number of small loads that have the potential to provide in aggregate a similar response capacity to a few large generation plants. Different small loads could provide frequency response, and loads with inherent storage or inertia are particularly suited, for example electric vehicles and heat pumps. In response, National Grid commissioned Element Energy to carry out this Network Innovation Allowance (NIA) project to assess the potential for electric vehicles and heat pumps to contribute to frequency response. 2.2 Report structure Section 3 – defines the EV and heat pump archetypes used in the analysis Section 4 – summarises National Grid service requirements and projections for future needs Section 5 – summarise the analysis of the potential contribution of EVs and heat pumps to frequency response 7 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Section 6 – outlines the key commercial models that may be used for EV and heat pump frequency response Section 7 – assess the key technical and operational barriers and challenges Section 8 – reviews the next steps in the roll out of EV and heat pump frequency response 8 Frequency Sensitive EV and Heat Pump Power Consumption Final Report PART 1 - Method 9 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 3 Description of archetypes 3.1 Electric Vehicles Sales of electric vehicles (EVs) are increasing and there is the potential for high uptake out to 2030. An EV can be defined as a vehicle propelled (fully or in part) by an electric motor using batteries that are charged by an external power source. This definition covers Battery Electric Vehicles (BEVs), Plug-in Hybrid Electric Vehicles (PHEVs) and Range Extended Electric Vehicles (REEVs) as these all charge their batteries from an external power source. However, it does not cover Hybrid Electric Vehicles (HEVs) that have an electric motor and battery, but do not use an external power source to charge the battery. A BEV, also known as a Pure Electric Vehicle, has an electric motor that relies exclusively on the power supplied by a battery. The battery is charged by plugging the vehicle in to an external power source. A PHEV has both an electric motor and an internal combustion engine, which are both used to provide propulsion. The configuration in which they are used depends on the model of the car, but all can travel a number of miles in electric only mode and have their battery charged by an external power source. Finally, a REEV has an electric motor and an internal combustion engine but gains all of its motive force from the electric motor. The internal combustion engine is used to generate electricity beyond the range of the battery. Of these types of EV, BEVs typically have the largest battery capacity, while PHEVs have smaller capacities. Table 1 shows typical battery capacities for different types of EVs. Table 1: Typical EV battery capacities 4 Battery capacity (kWh) 0-4 Small two wheel 4-8 PHEV Large two wheel Quadricycles 8-20 REEV PHEV High performance two wheel Small BEVs 20-50 BEV- cars and small/medium vans REEV 50-100 BEV- high performance cars Larger vans, trucks and buses including BEV, PHEV and REEV 100+ Specialist and very large vehicles 4 A Guide to Electric Vehicle Infrastructure, BEAMA 10 Frequency Sensitive EV and Heat Pump Power Consumption Final Report In this study, we examine the potential for EVs to provide frequency response. The analysis considers both Battery Electric Vehicles (BEVs) and Plug in Hybrid Electric Vehicles (PHEVs). The analysis considers frequency response provision by turning on and off charging or by varying charge power, but excludes Vehicle-to-Grid energy transfer (V2G). This is because V2G presents operational issues that are considered significant by many EV manufacturers. These include concerns on battery degradation (due to the increased number of charging/discharging cycles), energy loss due to round trip efficiency and the challenge in guaranteeing that the vehicle will be fully charged when the customer needs it. EV charging can be broadly divided into two types: AC and DC charging. It should be noted that the EV battery is always charged by DC, and the power supply from the grid is AC. A conversion from AC to DC therefore needs to take place. For an AC charge point, the conversion from AC to DC takes place inside the car using on-board equipment. This equipment is limited in size and weight, and therefore cannot deal with very high charge powers. For a DC charge point, AC is converted to DC inside the charge point. Space and weight considerations are therefore not as important, and higher charge powers can be accommodated. 5 The primary international standard governing electric vehicle charging is IEC 62196 . It defines four charging ‘modes’ as follows: Mode 1. Connection of the EV to a standard AC mains socket outlet. This mode is not recommended in the UK because RCD (residual-current device) protection, which is necessary to ensure shock protection, is not guaranteed for all socket outlets. Mode 2. Connection of the EV to a standard AC mains socket outlet, but with a control box to inform the EV of the charge power that can be drawn and to provide RCD protection. The power is limited to 3kW for domestic socket-outlets and to 7.4kW for industrial socket-outlets. Mode 3. Connection of the EV to a charge point or EVSE (Electric Vehicle Supply Equipment) supplying AC power. The charge point is able to communicate with the EV to inform it of the available power, and there is safety interlocking between the EV and charge point. A single phase Mode 3 system usually operates at 3.7kW or 7.4kW, while a three phase system may operate at a higher rate (typically 11, 22 or 43kW). Mode 4. Connection of the EV to a charge point supplying DC power. There is communication between the EV and charge point to establish safety and define the appropriate charge current. These have a wide range of charging capabilities up to over 100kW. The charging modes relevant for this report are Mode 3 and Mode 4, as these allow smart control of vehicle charging and variation of charge power. Mode 3 charge points are commonly used for home, workplace and public charging. Home charging typically uses charge powers up to 7.4kW as higher powers require a three phase grid connection. Workplace and public charging use a wide range of charge powers, 5 http://www.iec.ch/ 11 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 6 and new installations tend to be higher power. At time of writing, Zap Map , a UK public charge point map, had recorded 2240 slow AC charge points of up to 3kW, 4940 fast AC charge points of power ranging from 7kW to 22kW and 322 rapid AC charge points of power up to 43kW. Mode 3 charge points use a specific protocol for communication between the charge point 5 and the EV, which is compliant with the IEC 61851 standard . It is worth noting that the charge power supplied to the vehicle is limited by either the charge point power, the power capacity of the cable connecting the charge point to the vehicle or the power handling capacity of the on-board charger in the vehicle, whichever is lowest. Mode 4 charge points tend to be larger pieces of equipment than Mode 3 and can provide very high charge powers. They are therefore more suited to public and industrial/commercial applications where the time available for charging is limited. Currently, the most common Mode 4 charge point power is in the 20-50kW range and powers at the 100kW level may be available in the medium term. At the time of writing, Zap Map counts 764 public DC charge points in the UK. The UK government encourages the uptake of EVs and the installation of EV charging infrastructure by offering a number of grants. Grants are available when purchasing a plugin car or van and cover 35% of the cost for a car (up to £5,000) or 20% of the cost for a van (up to £8,000). EV owners can also receive up to 75% off the capital cost of a home charge point and associated installation costs (capped at £700) via a scheme administered by the Office for Low Emission Vehicles (OLEV). If an EV owner or prospective owner does not have off-street parking, they can also request that their local authority install a public charge point on their street. Funding from OLEV is available to local authorities for this purpose. As previously mentioned, dedicated EV charging infrastructure (Mode 3 or Mode 4 charge points) is a pre-requisite for smart management of EV charging. Smart EV charging may 7 bring many benefits, including minimising the increase in peak demand caused by EVs , allowing fleet operators to charge many vehicles at one site without upgrading their grid connection, allowing increased penetration of renewable generation, or, as is explored in this study, providing frequency response services to National Grid. 3.1.1 EV uptake projections and archetypes Three scenarios for the uptake of EVs are used in this study. BEV and PHEV uptake numbers are included. The ‘Medium’ and ‘Low’ scenarios are based on National Grid’s ‘Gone Green’ and ‘Slow Progression’ Future Energy Scenarios (FES) 2015. A ‘High’ scenario was also included, which is based on the base case scenario in previous Element 8 Energy analysis for the Low Carbon Vehicle Partnership (LowCVP) . Figure 1 shows the predicted uptake of EVs for these three scenarios. Figure 2 shows the corresponding predictions for the annual electricity consumption by EVs. All three scenarios use the same assumptions on annual electricity demand per EV, from the National Grid FES 2015. 6 https://www.zap-map.com/ Impact & opportunities for wide scale EV deployment, Low Carbon London Report B1 8 LowCVP, Options and recommendations to meet the RED transport target, Element Energy, 2014 7 12 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Projected number of EVs 6,000,000 Number of EVs 5,000,000 4,000,000 High uptake 3,000,000 Medium uptake 2,000,000 Low uptake 1,000,000 0 2015 2017 2019 2021 2023 2025 2027 2029 Figure 1: EV uptake projections, including BEVs and PHEVs 16,000 14,000 12,000 10,000 8,000 6,000 4,000 2,000 0 High uptake Medium uptake Low uptake 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 GWh Projected annual electricity demand of EVs Figure 2: Projected annual EV electricity demand, including BEVs and PHEVs The uptake of electric vehicles was divided into four distinct charging archetypes: Private vehicles: o Home charging o Work charging o Public charging Fleet vehicles The division of vehicles into private and fleet was based on the Department for Transport 9 (DfT) vehicle licensing statistics for 2014 . Private vehicle charging archetypes were then identified using data from the Ultra Low Carbon Vehicle (ULCV) Demonstrator 10 Programme . The ‘home’, ‘work’ and ‘public’ charging locations are included while ‘other’ and ‘unknown’ charging locations are excluded. The breakdown of EV charging into these four archetypes, shown in Figure 3, is assumed to be constant into the future. For the purposes of modelling ease, it is also assumed that each EV falls exclusively into one of these four archetypes over time (this assumption is 9 https://www.gov.uk/government/statistical-data-sets/veh02-licensed-cars Assessing the viability of EVs in daily life, The Ultra Low Carbon Vehicle Demonstrator Programme, 2013 10 13 Frequency Sensitive EV and Heat Pump Power Consumption Final Report well supported in the case of home charging as the ULCV trial found that drivers with home charging infrastructure used it 97% of the time). The archetype breakdown can therefore be directly applied to the EV projected uptake numbers. Fleet 9% Public 11% Home 47% Work 33% Figure 3: Breakdown of EV uptake by archetype, based on DfT and UCLV data In the analysis presented in this report, public charging is excluded for frequency response provision. This assumption is based on interviews with industry stakeholders, who indicated that most charging at public charge points is opportunity charging, where the time required to charge up the battery is similar or longer than the time the user intends to stay connected for. This means that increasing the charging time by interrupting charging or reducing charge rates for frequency provision would not be acceptable to EV drivers. In order to model the response available from EVs, diurnal charge profiles were obtained for the home, work and fleet archetypes. These were assumed to be the same for PHEVs as for BEVs. For home and work charging, these profiles were derived using charge start 11 time data from the ULCV and Customer-Led Network Revolution (CLNR) EV trials. These profiles were scaled to match BEV and PHEV daily energy requirements from the 12 National Grid FES and assume a charge power of 3kW, typical for a slow charger. The PHEV energy consumption is on average 16% lower than for BEVs, providing a similar, but slightly lower, charging potential for frequency response. Figure 4 and Figure 5 show the resulting diurnal profiles for home and work charging respectively (for March, in the 2030 Medium uptake scenario). For fleet vehicles, a simple profile was constructed based on information from interviews with fleet operators. This charge profile assumed that fleet EVs plugged in to charge between 20:00 and 01:00. The assumptions on daily energy use and charge power were the same as for home and work charging. The resulting aggregate charge profile for fleet vehicles (again for March in the 2030 Medium uptake scenario), is shown in Figure 6. 11 http://www.networkrevolution.co.uk/ In the FES, PHEVs have a 16% lower daily energy requirement than BEVs. This is applied in the analysis. 12 14 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Home Infrastructure Diurnal Profile 1,400 1,200 MW 1,000 800 600 400 200 0 Figure 4: Home charging diurnal profile, March 2030 (medium uptake) Work Infrastructure Diurnal Profile MW 1,600 1,400 1,200 1,000 800 600 400 200 0 Figure 5: Work charging diurnal profile, March 2030 (medium uptake) Fleet Diurnal Profile 600 500 MW 400 300 200 100 0 Figure 6: Fleet charging diurnal profile March 2030 (medium uptake) The EV daily energy requirement varies over the year due to changing drive patterns and the use of in-car heating. This seasonal variation is obtained from CLNR EV trial data and is applied to the analysis of EV response potential. Figure 7 shows the monthly variation in 15 Frequency Sensitive EV and Heat Pump Power Consumption Final Report the daily energy requirement relative to the annual average. As can be seen in this graph, in January, the daily EV energy requirement was 9.8kWh/EV, 35% more than the annual average, while in July it was 6.1kWh/EV, 16% less than the annual average. This is assumed to be the same for PHEVs as for BEVs. Seasonal variation in EV daily energy demand 12 kWh/EV 10 8 6 4 2 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Figure 7: Seasonal variation in daily energy demand for a single EV, 2030 medium scenario 3.1.2 EV charging windows The flexibility in EV charging, essential for EVs to provide frequency response, arises from the fact that EVs can be plugged in for much longer than their required charge time. This flexibility will vary depending on the archetype and on the individual requirements of an EV. In the analysis, home charging EVs are projected to typically be available for charging for 8 hours per day, but only require charging for 3 hours. Private vehicles, charged at home or work, do not have fully predictable patterns, but can offer flexibility as the daily energy requirement is typically quite low and they are parked for a significant proportion of their lifetime. Fleet vehicles have very predictable use patterns but are less flexible as they tend to have higher energy requirements and less time available for charging. For this analysis, the time available for EV charging and the daily energy requirement are assumed to be constant across archetypes. The constraints on EV charging time assumed in the model are as follows: EVs plugged in between 17:00 and 21:00 must be charged by 05:00 EVs plugged in between 21:00 and 01:00 must be charged by 06:00 Otherwise, EVs have 8 hours available for charging. 3.2 Heat Pumps Heat pumps transfer heat from a lower temperature heat source to a higher temperature heat sink. For example, they transfer heat from the lower temperature outside air to the higher temperature indoor air to provide space heating. The principle of operation of a heat pump is shown in Figure 8. The liquid in the evaporator (3) evaporates, absorbing heat from outside. The resulting gas is then passed through the compressor (4), which increases the pressure and causes it to condense in the condenser (1), thus releasing its heat. The compressor is driven by electricity. Heat pumps can provide several times the energy as heat output as they consume in electricity, unlike resistive heaters in which 16 Frequency Sensitive EV and Heat Pump Power Consumption Final Report electricity is converted to heat in a 1:1 ratio, making heat pumps a very efficient form of electric heating. Figure 8: Heat pump principle of operation The heat pump heat source is typically the outside air or ground. Heat pumps that use these heat sources are termed Air Source Heat Pumps (ASHPs) and Ground Source Heat Pumps (GSHPs) respectively. The heat pump efficiency is measured by the Coefficient of Performance (COP), which is the ratio of heat energy delivered to electrical energy consumed. ASHPs can have a COP of up to 4 in mild weather, though this decreases in colder weather as the temperature difference between the inside and outside air increases. Typically, the COP of ASHPs is in the 2-3 range. GSHPs are not susceptible the same efficiency fluctuations as ASHPs but are more expensive to install and require a 13 suitable ground source . The heat pump compressor can be fixed speed, which means that the heat pump can only operate at 100% power, or variable speed, which means that the power to the compressor can be modulated enabling it to speed up or slow down depending on the heating requirements of the building. The use of a variable speed compressor means that frequent on/off cycles can be avoided and the heat output can be more precisely matched to the need. It also means that the heat pump can ramp up when turned on, avoiding high startup currents. For these reasons, variable speed heat pumps are more efficient. New domestic heat pump installations in the UK are typically variable speed, though there are 14 still some fixed speed models on the market . The UK government supports heat pump uptake via the Renewable Heat Incentive (RHI). The RHI rewards the production of renewable heat in retrofit installations via tariffs for each unit of heat produced. Heat pumps can contribute to DSR because of the thermal storage that can be present in the form of a dedicated thermal store for space heating or in the thermal inertia of the building. This means that they can be temporarily switched off or have their power reduced while still keeping the building temperature at comfortable levels. Manufacturers also offer hybrid heat pumps, which can provide heat via a heat pump or via an alternative fuel source (boiler). These also offer possibilities for DSR as the hybrid heat pump can turn off the heat pump and switch to its alternative fuel source when required, while still 13 The Energy Saving Trust, Getting warmer: a field trial of heat pumps, and the Energy Saving Trust website (http://www.energysavingtrust.org.uk/domestic/renewable-heat). 14 Based on interviews with heat pump manufacturers. 17 Frequency Sensitive EV and Heat Pump Power Consumption Final Report maintaining the building temperature at the appropriate level. Hybrid heat pumps are not explicitly looked at in the analysis presented in this report. 3.2.1 Heat pump uptake projections and archetypes Three heat pump uptake scenarios are used in this study. The ‘High’ and ‘Low’ uptake scenarios are from National Grid’s Gone Green and Slow Progression Future Energy Scenarios (FES) 2015. A ‘Medium’ scenario is also included, based on previous Element 15 Energy work for the Committee on Climate Change (CCC) . Two heat pump archetypes, defined in these uptake scenarios, are used in the analysis. These are residential and commercial/industrial. In the low uptake scenario (Slow Progression FES 2015), industrial/commercial heat pumps do not appear, so only residential heat pumps were included in this scenario. Figure 9 shows the projected numbers for residential heat pumps in all three scenarios, while Figure 10 shows the corresponding annual electricity consumption by residential heat pumps. Figure 11 shows the projected electricity consumption by industrial/commercial heat pumps. 15 P1: RHI driven uptake, CCC, Pathways to high penetration of heat pumps, Frontier Economics and Element Energy, 2013 18 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Projected number of residential heat pumps Numver of heat pumps 6,000,000 5,000,000 4,000,000 High uptake 3,000,000 Medium uptake 2,000,000 Low uptake 1,000,000 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 0 Figure 9: Residential heat pump uptake projections 16 Projected annual residential heat pump electricity demand 30,000 25,000 GWh 20,000 High uptake 15,000 Medium uptake 10,000 Low uptake 5,000 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 0 Figure 10: Projected annual residential heat pump electricity demand 8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 0 High uptake Medium uptake 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 GWh Projected annual commercial/industrial heat pump electricity demand Figure 11: Projected annual commercial/industrial heat pump electricity demand 16 The shape of the medium uptake scenario, and the low and high uptake scenario are different, as the medium scenario is the CCC scenario and low and high the National Grid FES. 19 Frequency Sensitive EV and Heat Pump Power Consumption Final Report In order to model the response available from heat pumps, diurnal energy use profiles were obtained for each archetype. The aggregate residential heat pump profile was obtained from CLNR heat pump trial data. This profile was scaled by the residential heat pump daily energy use. The annual average of the residential daily energy use was taken from National Grid’s FES 2015, while the monthly variation in this energy use, shown in 17 Figure 13, was obtained from CLNR data . Figure 12 shows the resulting aggregate residential heat pump profile for March in the 2030 medium uptake scenario. It is assumed in the modelling that each residential heat pump has an electric power demand of 3kW. This tends towards the lower end of the typical range for residential heat pump installations, and is chosen because future heat pump installations are expected to be in well-insulated homes with lower heating energy demands. Heat Pump Residential Diurnal Profile 5,000 MW 4,000 3,000 2,000 1,000 0 Figure 12: Residential heat pump diurnal profile, March 2030 (medium uptake) 18 kWh/heat pump Seasonal variation in residential heat pump daily demand 30 20 10 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Figure 13: Seasonal variation in daily energy use for a single residential heat pump, 2030 medium scenario The aggregate industrial/commercial profile was obtained in a similar way, though in this case no trial data was available. A constructed profile, used in load modelling work for UK 19 Power Networks (UKPN) , which assumes near constant operation during the day and no heat pump use at night, was therefore used. This was then scaled by the daily energy use. 17 http://www.networkrevolution.co.uk/ In summer, heating is mainly used for hot water provision and not for space heating, which may change the summer diurnal profile. This effect is not taken into account here, as the same profile shape from CLNR trial data is used throughout the year. 19 This load modelling work is described in Low Carbon London Report C3, Network impacts of energy efficiency at scale 18 20 Frequency Sensitive EV and Heat Pump Power Consumption Final Report The annual average for the daily energy use was taken from National Grid’s FES 2015, while the monthly variation, shown in Figure 15, was taken from UKPN’s network load 19 modelling work . Figure 14 shows the resulting aggregate industrial/commercial profile for March in the 2030 medium uptake scenario. MW HP Commercial/Industrial Diurnal Profile 1,600 1,400 1,200 1,000 800 600 400 200 0 Figure 14: Commercial/industrial heat pump diurnal profile, March 2030 (medium uptake) Seasonal variation in commercial/industrial heat pump daily demand GWh/day 30 20 10 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Figure 15: Seasonal variation in daily energy use by all commercial/industrial heat pumps, 2030 medium scenario 3.2.2 Potential flexibility of heat pumps In order to provide frequency response heat pumps are required to operate flexibly, interrupting or changing electricity consumption in response to the system frequency, while house temperature levels are maintained within comfort levels. This can be achieved through thermal storage (e.g. hot water tank) for space heating, which is discharged when the heat pumps is interrupted. Alternatively it can be provided by the heat stored in the building’s fabric (thermal inertia). Another option is the use of hybrid heat pumps, which can switch to gas-fired heating to provide flexibility in electricity consumption. Heat pumps 20 are typically, but not always, installed with a thermal storage . For retrofitted heat pumps in existing homes there is not always sufficient space to install a thermal storage tank. In the analysis presented in this report, it is assumed that residential heat pumps are installed with thermal storage in the form of a 180L hot water tank. With the hot water stored at 50ºC, and an indoor temperature of 20ºC, this represents approximately 6.5kWh of thermal storage. Assuming a COP of 3, a heat pump with an electric power of 3kW 20 Based on interviews with heat pump manufacturers. 21 Frequency Sensitive EV and Heat Pump Power Consumption Final Report (typical of a domestic heat pump) would take around 45 minutes to deliver this energy. Further assuming that the thermal storage inherent in the fabric of the building provides another 15 minutes of flexibility, this gives a total thermal storage equivalent to 1 hour of heat pump use, indicating that heat pump use can be shifted by 1 hour without impacting comfort levels. These heat pump flexibility assumptions are also supported by a modelling study by 21 Loughborough University , which found that for a house with high levels of insulation, the time of operation of the heat pump in the property could be shifted by 1-2 hours without the indoor temperature falling below 20ºC. This can be increased to 2-6 hours with the inclusion of a 300L-500L buffer tank for thermal storage. Thus even without thermal storage, the assumption that heat pump use can be shifted by an hour may still hold. Most future heat pump installations are expected to be in insulated homes, but it is uncertain whether significant thermal storage will be installed. The above assumptions on heat pump flexibility are therefore subject to a large degree of uncertainty. This issue was also raised by heat pump manufacturers and other stakeholders in interviews. Due to this uncertainty, the sensitivity of the results to the heat pump shifting ability is examined in the analysis. For transparency, and due to lack of available data, heat pump flexibility is assumed to be constant across archetypes. 21 Assessing Heat Pumps as Flexible Load, Hong et al., Loughborough University, 2013 22 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 4 National Grid Service Requirements National Grid has a license obligation to control the system frequency so that it remains within ±1% of the nominal value of 50Hz, except in abnormal or exceptional circumstances. While this is the statutory limit for deviations, the operational limit that National Grid applies aims to keep the system frequency between 49.8Hz and 50.2Hz. National Grid therefore holds a certain amount of generation and/or demand ready to automatically respond to frequency deviations. The frequency of the electricity system will deviate from the nominal value when there is a mismatch in electricity supply and demand. When supply is greater than demand, the frequency rises, and generation must be turned down or demand increased in response. When supply is less than the demand, the system frequency falls and generation must be ramped up or demand reduced in response. National Grid procures frequency response services from a number of sources. All generators covered by the Grid Code are required to have the capability to provide Mandatory Frequency Response, by automatically changing their active power in response to a frequency change. They can provide three different response services: Primary response: full delivery of response (increase in generation or reduction in demand) within 10 seconds of a frequency event, sustained for a further 20 seconds. Secondary response: full delivery of response (increase in generation or reduction in demand) within 30 seconds of a frequency event, sustained for a further 30 minutes. High frequency response: full delivery of response (decrease in generation or increase in demand) within 10 seconds of a frequency event, sustained indefinitely. National Grid also procures these frequency response services commercially through Firm Frequency Response (FFR), which is open to a wide range of providers, including providers of Mandatory Frequency Response. This creates a route to market for providers whose services may otherwise be inaccessible. Frequency response can be dynamic or non-dynamic (also called static). For dynamic response, the generator or load responds constantly, in proportion to the frequency deviation. For non-dynamic response, the response is provided in full when the frequency deviation exceeds a set point. Dynamic response requires providers to provide a near continuous output modulation, while static response is triggered only infrequently, in the order of 30 events a year. Dynamic response places higher technical requirements on providers and is typically valued higher than static response. To participate in FFR, providers must be able to deliver a minimum 10MW response, though this can be aggregated from multiple sites. National Grid also procures FFR through FFR Bridging and bilateral contracts, which are for providers that do not yet meet the 10MW minimum threshold for FFR or for those who can provide unique frequency response services to National Grid. Dynamic response is typically provided with a dead band of +/-0.015Hz. Technical requirements for FFR providers furthermore include a relay sensitivity tolerance of 0.01Hz. 23 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Providers are also required to communicate response data in real time second by second or minute-by-minute (in which case post event second-by-second data needs to be sent for verification). FFR can be bid for in any time period windows. National Grid also procures frequency response through Frequency Control by Demand Management (FCDM), where providers are prepared to interrupt their demand for 30 minutes 10-30 times a year. 4.1 Future frequency response requirements UK electricity demands are met by a mixture of conventional synchronous generation, which has high levels of mechanical inertia, and zero inertia non-synchronous renewable generation. When there is an event that triggers a frequency deviation (for example, a plant outage), the system inertia provided by conventional generation slows the rate of change of frequency, thus helping ensure system stability, after which frequency response providers are used to arrest the frequency drop and restore system frequency. With higher renewable penetration, the system inertia will decrease. This means that a plant outage will result in a faster frequency deviation. To counteract this, and to keep the grid frequency within operational limits, National Grid will require a larger volume of frequency response, or will need to acquire faster frequency response services. New wind turbine controls are being developed that can provide synthetic inertia. This is another option that may mitigate the loss of inertia from conventional generators and limits the amount of additional frequency response that will be required. Figure 16 shows the projected increase in National Grid’s mean primary frequency response holding requirement to 2030 in both the Gone Green and Slow Progression National Grid FES 2014 (for current service characteristics). The requirement in Gone Green is higher than that in Slow Progression due to the higher proportion of zero-inertia renewable generation in that scenario. In the Gone Green scenario, the 2030 mean requirement is 1464MW while the maximum projected requirement for that year is 2738MW. Mean primary frequency response requirement 1500 MW 1400 1300 Gone Green 1200 Slow progression 1100 1000 2015 2020 2025 2030 Figure 16: Future National Grid primary frequency response holding requirements These projected holding requirements are obtained from illustrative National Grid modelling which, given the mix and volume of electricity generation, outputs the amount of 22 low frequency response needed in case of a major 1800MW plant outage . The 22 It is assumed in the modelling that the minimum level this loss can be reduced to is 600MW in 2015 and 1800MW from 2020 onwards. 24 Frequency Sensitive EV and Heat Pump Power Consumption Final Report projections for the future mix and volume of electricity generation are obtained from National Grid’s Gone Green and Slow Progression Future Energy Scenarios. National Grid is also exploring the provision of faster frequency response services, for example, through a rapid frequency response service. Rapid frequency response (low or high) requires full delivery of response within 5 seconds and can be bid for in current FFR tenders. National Grid is also investigating even faster response and synthetic inertia services in the Enhance Frequency Control Capability Network Innovation Competition project. The predictions of primary response requirement presented above are highly dependent on whether such services, or other novel frequency response services, are developed. 4.2 Seasonal and diurnal variation in response requirement The projected primary response requirements presented above are annual averages. However, the primary response requirement is modelled for each hour in a given year, allowing the diurnal variation in requirement to be obtained. This is presented for the 2030 Gone Green scenario in Figure 17. The diurnal variation shown is the average over the entire year. Diurnal variation in 2030 primary frequency response requirement 2500 MW 2000 1500 1000 500 0 Figure 17: Diurnal variation in average primary frequency response requirement, averaged over the year, 2030 Gone Green scenario The diurnal variation in average primary response requirement is shown in Figure 17. As can be seen, the response requirement is highest overnight, between midnight and 06:00, and is relatively constant through the day. This is due to the fact that the lower levels of demand overnight mean that a single plant outage would have a much larger effect on frequency. 4.3 Modelling method 4.3.1 Modelling method for EVs To find how much response EVs could potentially provide, their charge profiles were adjusted in response to system frequency deviation, based on 2014 half hourly frequency response utilisation. A frequency responsive profile for each day in a given year could thus be obtained, which allowed the calculation of the volume of low or high frequency response provided when there was a signal. It was ensured throughout that vehicles were fully charged by the end of their available charge period. This methodology is demonstrated in an example cartoon profile for a single EV, shown in Figure 18. The EV is 25 Frequency Sensitive EV and Heat Pump Power Consumption Final Report plugged in to charge at 19:00, and must be charged by 06:00, but its charging is interrupted due to low frequency signals. Cartoon charging profile for a single EV 4 Normal charging profile kW 3 Frequency responsive profile 2 Low frequency signal 1 High frequency signal 06:00 05:00 04:00 03:00 02:00 01:00 00:00 23:00 22:00 21:00 20:00 19:00 18:00 0 Figure 18: Cartoon profile for a single EV responding to half hourly frequency signals to demonstrate the modelling methodology The aggregate normal diurnal and frequency responsive profiles for all EVs were calculated for each day in the year for each scenario (each time using the 2014 half hourly frequency signals). This allowed the calculation of the average low frequency and high frequency response for each year and scenario as well as the seasonal and diurnal variation in that average. 4.3.2 Modelling method for heat pumps The method used for calculating heat pump frequency response is the same as for EVs, but the constraints on when a heat pump can respond are different. Similarly to the approach for EVs, the aggregate heat pump energy use profile was adjusted in response to system frequency deviation, based on 2014 half hourly frequency response utilisation. This gave a frequency responsive profile that allowed calculation of the low frequency response and high frequency response provided by heat pumps. The flexibility allowed in the energy use profile was defined by the amount of thermal storage available (as described in Section 3.2.2, it was assumed that the volume of energy storage available was equal to the energy delivered by the heat pump in one hour). 26 Frequency Sensitive EV and Heat Pump Power Consumption Final Report PART 2 - Results 27 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 5 Potential Frequency Response Contribution over Time 5.1 Potential future response from EVs The analysis finds that the potential for frequency response from EVs increases strongly after 2020. By 2030, EVs could contribute an average low frequency response of 754MW in the medium uptake scenario, 52% of the projected average 2030 requirement. This could rise to 1210MW, 83% of the projected requirement, in case of high EV uptake, and could reduce to 377MW, 26% of the projected requirement, if EV uptake is low. The annual average for low and high frequency response was calculated out to 2030 for all EV uptake scenarios. As shown in Figure 19 and Figure 20, the response potential of EVs increases with increased EV uptake, and becomes significant after 2020. Figure 19 shows the potential for low frequency response, which would be achieved by switching off charging or reducing the charge power. Average low frequency response from EVs 1400 1200 MW 1000 800 High uptake 600 Medium uptake 400 Low uptake 200 0 2015 2020 2025 2030 Figure 19: Average low frequency response from EVs to 2030 The modelled potential for high frequency response is slightly lower than that for low frequency response, as can be seen in Figure 20. High frequency response in the context of EVs would be achieved by switching on charging or by increasing charge power. By 2030, this could contribute 577MW in the medium uptake scenario, rising to 926MW in the high uptake scenario, and reducing to 289MW in the low uptake scenario. 28 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Average high frequency response from EVs 1000 MW 800 600 High uptake Medium uptake 400 Low uptake 200 0 2015 2020 2025 2030 Figure 20: Average high frequency response from EVs to 2030 5.1.1 Seasonal variation in response The response from EVs varies over the course of the year as charge patterns and EV daily energy use changes (see section 3.1.1). Figure 21 shows the monthly variation in average low frequency response in 2030 for medium EV uptake. This shows that the seasonal variation in response is small, with variations of approximately ±15% about the annual average. The lowest response availability is in April, with 668MW of low frequency response available in the 2030 medium scenario, while the highest is in January, with 891MW available in this same scenario. Thus, in January, EVs could provide 61% of low frequency response requirements while in April they could provide 46%. Seasonal variation in average low frequency response from EVs, 2030 medium scenario 1,000 600 December November October September August July Home infrastructure June 0 May Work infrastructure April 200 March Fleet vehicles February 400 January MW 800 Figure 21: Monthly variation in average low frequency response from EVs, 2030 medium scenario Figure 21 also shows the breakdown of the available low frequency response by archetype. The bulk of the response is provided by home and work charging with, on average over the year, 53% of the response coming from home charging, 37% from work charging and 10% from fleet vehicles. 29 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 5.1.2 Diurnal variation in response In the charging strategy where EVs start charging when they are plugged in and respond to deviations in system frequency, the response available from EVs varies significantly within a given day. On average, with this charging strategy, EVs can offer very limited frequency response in the early hours of the morning (03:00-04:00), while at the morning and evening peak they could offer around 1500MW of low frequency response in the 2030 medium scenario. This corresponds to twice the average response, and, for the morning peak, could provide over 120% of the projected low frequency response requirement (for 2030 under the Gone Green 2014 scenario). A charging strategy where frequency sensitive consumption is combined with a controlled smart charging strategy, in which the early evening charging peak is spread over the night time, is also examined. Here the response available from home and work charging, which usually peaks soon after they are plugged in to charge, is spread over the entire available charging period, to 06:00. This gives a near constant response availability during the night of around 830MW, approximately 40% of the average night time primary response requirement (for 2030 under the Gone Green 2014 scenario). To demonstrate the diurnal variation in EV response, the 2030 Medium scenario is once again taken as an example. Figure 22 shows the diurnal variation in average low 23 frequency response from EVs, with current charging patterns , while Figure 23 shows the diurnal variation with active charging management to spread home charging and fleet vehicle response over the night. This average is calculated for each half hour over the whole year, and is broken down into response from home, work and fleet archetypes. This shows that with current charge patterns, the response is highest at 09:30 and at 22:30 and lowest at 03:00-04:00. The 09:30 peak mainly arises from cars plugging in to charge after arrival at work while the 22:30 peak is mainly due to home charging. The lack of available response in the early hours of the morning arises because cars are generally either fully charged or obligated to charge without interruption (to ensure that they are fully charged by morning). This dip in response availability at 03:00-04:00 can be prevented through controlled smart charging management, as shown in Figure 23. In this case, the charge start time of some vehicles is delayed so as charging is more spread through the night. Assuming the response provided by home charging and fleet EVs is evenly spread over the entire available charge period results in a near constant response availability of approximately 830MW throughout the night. 23 Current charging patterns: EVs start charging on full power as soon as they are plugged in 30 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Diurnal variation in average low frequency response from EVs, 2030 medium scenario 2000 MW 1500 Fleet vehicles 1000 Work infrastructure 500 Home infrastructure 00:00 01:30 03:00 04:30 06:00 07:30 09:00 10:30 12:00 13:30 15:00 16:30 18:00 19:30 21:00 22:30 0 Figure 22: Diurnal variation in average low frequency response from EVs with current charging patterns, 2030 medium scenario. Error bars show the standard deviation of the total response. 1800 1600 1400 1200 1000 800 600 400 200 0 Fleet vehicles Work infrastructure Home infrastructure 00:00 01:30 03:00 04:30 06:00 07:30 09:00 10:30 12:00 13:30 15:00 16:30 18:00 19:30 21:00 22:30 MW Diurnal variation inaverage low frequency response from EVs with smart charging, 2030 medium scenario Figure 23: Diurnal variation in average low frequency response from EVs with active charging management, 2030 medium scenario. Error bars show the standard deviation of the total response. Figure 17 shows that the diurnal National Grid low frequency response requirement (for 2030 under the Gone Green 2014 scenario), averaged over the year, varies between approximately 1200MW and 2000MW. The requirement is highest during the night, when demand, and consequently the volume of synchronised generation, is lowest. With baseline charge patterns, the available response from EVs is lowest in the early hours of the morning, when the National Grid requirement is high. However, with controlled smart charging management, the response from home charging and fleet EVs may be spread throughout the night, and can provide approximately 40% of the night time 2030 primary response requirement (for 2030 under the Gone Green 2014 scenario). This is shown in Figure 24 and Figure 25. 31 Frequency Sensitive EV and Heat Pump Power Consumption Final Report With large scale uptake of frequency sensitive EV consumption, the diurnal variation in response availability may impact other providers of frequency response. Conventional providers may be required to provide frequency response for shorter more specific windows then they do at the moment. Alternatively, demand side assets with different availability profiles may be aggregated to provide a more constant aggregated availability profile. The other option is to only use the constant baseline EV response capability, and to discard peak availability, so as to obtain a constant level of response. Percentage of requirements satisfied by EVs Diurnal variation in proportion of National Grid requirements satisfied by EVs 140% 120% 100% 80% 60% 40% 20% 0% Figure 24: Diurnal variation in ability of EV response to meet the National Grid requirement (Gone Green scenario) for current charging patterns, 2030 medium scenario Percentage of requirements satisfied by EVs Diurnal variation in proportion of National Grid requirements satisfied by EVs with smart charging 140% 120% 100% 80% 60% 40% 20% 0% Figure 25: Diurnal variation in ability of EV response to meet the National Grid requirement (Gone Green scenario) with smart charging management, 2030 medium scenario 32 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 5.2 Potential future response from heat pumps The potential for frequency response from heat pumps increases steadily from 2015 with increased heat pump uptake. By 2030, commercial and industrial heat pumps could contribute an annual average low frequency response between 447MW and 585MW, based on medium and high heat pump uptake scenarios. These figures respectively correspond to 30% and 40% of the projected average low frequency response requirement for 2030 under the Gone Green 2014 scenario. Including the response potential from residential heat pumps, in the 2030 medium scenario heat pumps could provide 2.5GW of low frequency response. This equates to 170% of the projected response requirement for that year. The annual average for low and high frequency response from heat pumps was calculated out to 2030. Figure 26 and Figure 27 show the response potential increases with heat pump uptake and could become significant before 2020 in both the medium and high uptake scenarios, when residential installations are included. Figure 26 shows the potential for low frequency response from heat pumps in future years, including residential response. In 2030, heat pumps could provide an average low frequency response of 2.5GW in the medium uptake scenario, 170% of the projected primary response requirement (for 2030 under the Gone Green Scenario). This would increase to 3.3GW, 227% of the response requirement, in the high scenario, and decrease to 375MW, 27% of the response requirement if heat pump uptake is low. Average low frequency response from HPs 3500 High uptake (I&C) 3000 Medium uptake (I&C) MW 2500 2000 High uptake (residential and I&C) 1500 1000 Medium uptake (residential and I&C) 500 0 2015 2020 2025 2030 Low uptake (residential and I&C) Figure 26: Average low frequency response from heat pumps to 2030 The response from heat pumps is very dependent on residential heat pump uptake. If only commercial/industrial heat pumps are included, the response provided is 447MW in the 2030 Medium uptake scenario, 30% of the projected average primary response requirement (for 2030 under the Gone Green Scenario). Retrofit installations in existing housing of heat pumps with thermal storage for space heating (assumed in the analysis to provide flexibility) may be challenging. If only heat pumps installed in new builds are included, the response provided by them and industrial/commercial heat pumps in the 2030 medium scenario is 1472MW, slightly over 100% of the projected requirement. 33 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Like for EVs, the potential high frequency response from heat pumps is lower than the potential low frequency response. Figure 27 shows the amount of high frequency response that heat pumps could provide out to 2030 for each of the three uptake scenarios. By 2030, heat pumps could contribute 1.3GW in the medium uptake scenario. This could increase to 1.7GW of high frequency response in the high uptake scenario or fall to 196MW in the low scenario. Again, this is very dependent on residential heat pump uptake, with only 16% of the response coming from commercial/industrial heat pumps in the 2030 medium scenario. Average high frequency response from HPs 1800 1600 High uptake (residential and I&C) 1400 Medium uptake (residential and I&C) MW 1200 1000 800 Low uptake (residential and I&C) 600 High uptake (I&C) 400 Medium uptake (I&C) 200 0 2015 2020 2025 2030 Figure 27: Average high frequency response from heat pumps to 2030 The above results assume that heat pump operation can be shifted by one hour. As outlined in Section 3.2.2, there is significant uncertainty around this assumption. The sensitivity of the results to this assumption was therefore examined. It was found that, for the 2030 medium scenario, reducing this shift time to 30 minutes reduces the low frequency response from heat pumps by 9%. Reducing the shift time further to 15 minutes produces a reduction of 17% relative to the base 1 hour assumption. Thus, even with much lower flexibility, heat pumps still have the technical potential to contribute significantly to frequency response. 5.2.1 Seasonal variation in response The analysis shows a significant seasonal variation in the response available from heat pumps, with variations of approximately ±70% about the annual average. In the 2030 medium scenario, this corresponds to a minimum average low frequency response of 591MW in July and a maximum of 4.3GW in January. This means that in July, heat pumps could meet only 40% of the projected primary response requirement (for 2030 under the Gone Green Scenario) while in January they exceed it threefold. The response from heat pumps changes over the course of the year because of the variation in heat pump energy use (see section 3.2.1). Heat pumps use much less energy in summer than in winter due to lower heating demands. This means that they are turned on less often and at a lower power, which results in lower response availability. 34 Frequency Sensitive EV and Heat Pump Power Consumption Final Report To demonstrate the variation in response available from heat pumps over the course of a given year, the 2030 Medium uptake scenario is taken as an example. Figure 28 shows the monthly variation in average low frequency response from heat pumps in 2030 for medium heat pump uptake. The seasonal change is very significant, with the response available in January nearly five times that available in July. Seasonal variation in average low frequency response from heat pumps, 2030 medium scenario 5,000 MW 4,000 3,000 2,000 Commercial/Industrial 1,000 Residential 0 Figure 28: Monthly variation in average low frequency response from heat pumps, 2030 Medium scenario It is also evident from Figure 28 that the bulk of response is provided by residential heat pumps. On average over the year 82% of low frequency response is provided by residential heat pumps while commercial/industrial heat pumps provide 18%. If residential heat pump installations do not meet projected numbers, perhaps due to retrofit difficulties or lack of consumer engagement, then the available response will be correspondingly lower. 5.2.2 Diurnal variation in response In the analysis, the response available from heat pumps also varies significantly throughout the day. The available response is highest between 05:00 and 22:00, with an early morning peak of 81% above the daily average. It is lowest during the night, at around half the average response value. This means that in the 2030 medium scenario, during the night heat pumps provide 50% of projected primary response requirements of the Gone Green scenario while during the day they provide over 200% of the projected requirement (see Figure 30). To demonstrate this variation, we once again use the 2030 medium uptake scenario as an example. Figure 29 shows the diurnal variation in average low frequency response from heat pumps. This is calculated for each half hour over the whole year, and includes response from residential and commercial/industrial archetypes. It shows that the response peaks at 06:30 and is lowest between 23:00 and 04:00. 35 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Diurnal variation in average low frequency response from heat pumps, 2030 medium scenario 7000 6000 MW 5000 4000 Commercial/Industrial 3000 Residential 2000 1000 00:00 01:30 03:00 04:30 06:00 07:30 09:00 10:30 12:00 13:30 15:00 16:30 18:00 19:30 21:00 22:30 0 Figure 29: Diurnal variation in average low frequency response from heat pumps, 2030 Medium scenario Figure 30 shows the percentage of National Grid’s primary frequency response requirement (shown in Figure 17) that is satisfied on average by heat pumps throughout the day. The diurnal variation in response availability is stronger than for EVs, so similar measures will be needed to provide a constant response. As discussed in section 5.1.2, aggregation with other types of demand side assets could provide a more constant response capability. Alternatively, only baseload availability could be used and peak heat pump response capability could be discarded so as to get a constant level of response over the day. 300% 250% 200% 150% 100% 50% 22:30 21:00 19:30 18:00 16:30 15:00 13:30 12:00 10:30 09:00 07:30 06:00 04:30 03:00 01:30 0% 00:00 Percentage of requirements satisfied by heat pumps Diurnal variation in proportion of National Grid requirements satisfied by heat pumps Figure 30: Diurnal variation in how well heat pump response meets the National Grid Gone Green scenario requirement, 2030 Medium scenario 36 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 6 Commercial potential 6.1 Costs of frequency response provision The total cost to provide frequency response can be broken down into the individual asset cost, including hardware and operational costs, and the business overhead costs for a frequency response aggregator, as outlined in Figure 31. For the analysis it is assumed that an SME is running a mature dedicated EV and heat pump frequency response aggregation service. The analysis moreover assumes that hardware costs are integrated in standardised products, which are manufactured at scale. Figure 31: Breakdown of cost components for frequency response provision The total annualised costs for the different archetypes are summarised in Figure 32. The total cost per EV in the fleet EV archetype is about half the cost for the home charging archetype, as the overhead costs are incurred per customer rather than per EV, and hence these costs are shared between multiple EVs (the archetype assumes 10 EVs in a fleet). For heat pumps most of the costs are incurred per heat pump. A commercial heat pump typically has a significantly higher rating than a residential heat pump and hence the cost per unit of power is 65% lower for the commercial heat pump archetype, compared to the residential archetype (assumes commercial heat pump of size 10kW electric). 37 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Annual costs of frequency response provision £/asset/year 25 20 15 Business overhead cost 10 5 0 Home charging Fleet EV Residential 10kW commercial HP Hardware and operational cost HP Figure 32: Annual costs of frequency response provision by archetype 24 6.1.1 Individual asset cost The cost per individual device includes hardware costs, consisting of a frequency sensor, control equipment and metering and telemetry, and operational costs to cover communications, data processing and maintenance. For industrial facilities, integrating these components and controls into legacy systems can be technically challenging, and the current retrofitted cost of these components for large industrial facilities is high. These costs would be prohibitively high on a per customer basis for kW scale assets. However cost estimates and interviews with aggregators and equipment manufacturers carried out for this study indicate that these costs could be significantly lower for kW scale assets when they are integrated into standardised products that are produced at scale, with specifications to meet minimum requirements for frequency response services. Hardware costs EVs with mode 3 or 4 chargers already contain many of the required functionalities to enable frequency response provision, although some additional hardware is required. The additional required hardware to fully enable frequency responsive consumption consists of frequency sensing, control, metering and telemetry components. If these components are incorporated into standardised charge points that are produced at scale, the additional costs could be approximately £10 (excluding engineering, design and 25 testing costs) per charge point . The specifications of this equipment are lower than for currently used industrial versions, but are compatible with National Grid requirements, providing frequency measurement with an accuracy of 1mHz, power metering with an accuracy of 0.1% over a 2000:1 range of current and communications via WiFi. These costs are per charge point, and hence are incurred fully for both home charging and fleet EV archetypes, as it is assumed that each fleet EV has its own charge point. 24 Assumes fleet of 10 EVs and a commercial heat pump of size 10kW electric. Annual cost calculated over 15 years. 25 Engineering cost estimate carried out for this study, based on quoted design specification (see Appendix) and interviews carried out for this study. Costs for technical components are for orders of 1000 or more. This does not include the engineering cost for design and testing. 38 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Heat pumps may require additional control equipment to be able to provide frequency response. This additional cost is estimated to be £12-£25 per heat pump based on device 26 costs quoted in the Grid Scientific report for the CCC on the costs for residential DSR . This cost assumption was confirmed in interviews with manufacturers. The costs are assumed to scale with the size of the heat pump, and are therefore higher for larger commercial heat pumps. Enabling heat pumps to provide frequency response may furthermore require redesign for most types of heat pumps to be able to provide frequency response in a reliable and safe manner. These development costs are not included in the cost estimate. Operational costs Operational costs include communications, data processing and maintenance. Communication costs are estimated to be approximately £5 per device per year, based on analysis carried out in the Grid Scientific report for the CCC on the cost of domestic DSR 26 infrastructure . In the interviews carried out for this study, aggregators have indicated they expect data processing could be brought down to a few pounds per device per year (£3/device/year is assumed in the analysis). For annual maintenance cost a typical rule of thumb of 3% of the initial hardware costs is assumed. One of the main costs for conventional thermal generation in providing frequency response is the opportunity cost incurred because of operating at part load. This cost includes missed revenues and lower efficiency. The EV and heat pump archetypes considered in this do not incur similar costs, which results in low costs for the provision of frequency response. If EVs were to deliver electricity back to the grid, opportunity costs may be a more significant factor (i.e. round trip efficiency, battery degradation). For other demand response technologies, opportunity costs may also be relevant, e.g. lower efficiency in industrial processes, interrupted production. Business overhead costs The business overhead costs for an aggregator of demand side frequency response include establishing and operating a demand side response platform, billing systems, data processing, legal and other back office staff costs, as well as marketing and sales costs. These costs are to a large extent fixed costs, which scale stepwise with business size, and hence per customer costs may be high for a small or start up business. The overall business overhead costs are estimated to be £10 per year per customer. Over half of this cost covers sales, marketing and customer retention, while 20% goes towards initial set-up for new customers. The other 20% covers base IT systems costs, legal and billing costs. These costs are based on the analysis carried out by Grid Scientific for the CCC on the 26 cost of DSR infrastructure (assuming a small service provider with around 300,000 customers, limited service portfolio, either specialist or simple). This compares to the cost of operating an electric vehicle smart charging service provider of £26/EV/year, as 26 Infrastructure in a low-carbon energy system to 2030: Demand Side Response, Grid Scientific and Element Energy, 2013. Based on scenario for SME with less than 300.000 customers. 39 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 27 estimated in the Green e-Motion project . The overhead business costs are incurred fully by residential EV or heat pump owners. Fleet operators or owners of large commercial or industrial heat pumps incur the same annual costs, but due to the larger size of their asset base, the cost per EV or per kW electric of heat pump power is relatively lower. 6.2 Revenues per archetype National Grid’s total spend on frequency response services was approximately £150m in 28 2014 . As discussed in section 5, EVs and heat pumps have the potential to provide a 29 significant part of this market. Based on the current tendered prices , the annual average potential in the 2030 medium scenario for EVs corresponds to approximately £66 million in annual revenues, and £39 million for heat pumps (commercial and industrial). The average Mandatory Frequency Response holding payments in 2014 were £10/MW/hr for combined low and high frequency response provision, and rose to £20/MW/hr for FFR 28 for some periods . Although this represents relatively high revenues compared to other DSR services, this value is diluted at an individual kW scale asset level. The total revenues for the response provided by the different archetypes in the analysis are summarised in Table 2, for a £10/MW/hr and £20/MW/hr price. These prices do not reflect the fact that the higher responsiveness and accuracy of EV and heat pumps may reduce the overall capacity requirement for frequency response, and hence the relative value of 30 these providers. For instance, in the US PJM pay-for performance frequency regulation market, electricity storage providers are paid a price that is approximately 2-3 times higher 31 than conventional providers to reflect their fast response time . Table 2: Revenues per EV or per heat pump for frequency response provision Revenue per year, Revenue per year, £10/MW/hr payment £20/MW/hr payment Home charging EV £46 £92 Work charging EV £45 £90 Fleet EV £46 £93 Residential heat pump £72 £144 £23 per kW electric £45 per kW electric Industrial/Commercial heat pump For EVs, these revenues are calculated for BEVs and are expected to be slightly less for PHEVs due to their 16% lower daily energy use. Implementing bi-directional charging may increase the available response from an EV. However as discussed in section 3.1, this may also increase battery degradation, increase energy losses due to round trip efficiency 27 Assuming a business with 50,000 customers, double the staff costs of the Spain case example, 20% of staff cost for building and office costs, and 50% of staff costs for IT back end, marketing and customer interaction. Green e-Motion Deliverable 9.4 Part 1, ‘Envisaged EU mobility models, role of involved entities, and Cost Benefit Analysis in the context of the European Clearing House mechanism’, 2014 28 MBSS data and FFR post assessment tender reports 2014 29 Average £10/MW/hr for combined low and high frequency response provision 30 PJM is a Regional Transmission Organisation in the East of the USA. 31 PJM, Performance-based Regulation Market in PJM, July 2014. PJM, monthly average regulation prices after introduction Pay for Performance (Oct 2012), 2011 – 2014 40 Frequency Sensitive EV and Heat Pump Power Consumption Final Report and poses higher operational challenges to ensure quality of service to the customer. For an EV with bi-directional charging and a charge power of 3kW available over an 8 hour charging period, the annual value would be approximately £90-£181/EV (for payments of £10/MW/hr and £20/MW/hr respectively). 6.3 Cost benefit analysis The analysis shows a net benefit in the provision of frequency response for all EV and heat pump archetypes, summarised in Figure 33. While the system opportunity for this type of response may be significant and the cost of providing frequency response may be very competitive compared to conventional thermal plants, the potential net benefit on an individual asset level is diluted significantly. Annualised cost benefit per archetype Home charging EVs Residential heat pump £/yr Fleet EV £/yr £/yr 10kW commercial heat pump £/yr Figure 33: Cost benefit analysis by archetype, all in £/year 32 Figure 33 shows the net annual benefit for different EV and heat pump archetypes. For EVs the net benefit is highest for the fleet EV archetype, as business overhead costs are shared between EVs. The net benefit of a home charging EV is lower at £26 per year; however this represents a saving of approximately 9% of a typical annual EV electricity 32 Assumes a fleet of 10 EVs and a commercial heat pump of size 10kW electric. Annual cost calculated over 15 years. Revenues are for BEVs and are expected to be around 16% lower for PHEVs. 41 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 33 bill . This diluted value will require efficient commercial models for the deployment of EVs in frequency response services. The net benefit of a residential heat pump is almost double that of a home charging EV. 34 This represents a saving of approximately 9% of a typical annual electric heating bill . Similar to EVs, the net benefit per unit is higher for commercial heat pumps due to their larger size. However hardware modification costs for commercial heat pumps may be higher due to the fact they are more integrated in building systems, bespoke and have lower productions (these aspects are not taken into account in the cost estimate). 6.4 CO2 emissions impact Frequency sensitive energy consumption by EVs and heat pumps does not result in additional carbon emissions, as the total energy used remains unchanged. Frequency response from EVs and heat pumps therefore has the potential to avoid carbon emissions by conventional providers of dynamic response. Compared to conventional thermal generation, dynamic frequency response by home charging EVs has the potential to offset around 1 tonne of CO2 per EV per year. Similarly dynamic frequency response by heat pumps has the potential to offset around 0.6 tonnes of CO2 per kW per year for commercial heat pumps and around 2 tonnes of CO2 per year for residential heat pumps. Most dynamic response is currently provided by coal and gas fired power plants, in similar 35 amounts . For these plants, holding capacity available so as to provide frequency response results in lower efficiency and hence an increase in CO2 emissions of 36 approximately 10% . This results in an average increase in emissions of 0.26 tonnes of 37 CO2 per MWh of response held . Table 3 shows the potential CO2 emissions that could be avoided by EV and heat pump archetypes by replacing conventional thermal generation for dynamic frequency response. 33 Assumes EV electricity bill of £300/year (10,000 miles at 3p/mile) Assumes heat pump electricity bill of £550/year (15kWh consumption per day, with electricity price of 10p/kWh) 35 National Grid, BM unit Mandatory response Holding Volumes, 2013 36 When operating at 80% part load, Sources: (i) Short Term Operating Reserve – Carbon Intensity Report (ii) Appendix B, SQSS Review Request GSR007 Amendment Report, National Grid, SP Transmission and SSE, September 2009 37 80% part load, produced at 10% lower efficiency. Gas-fired power plant: 10% additional emissions x 80% part load x 0.4T CO2 per MWhe = 0.16 T CO2 per MWh response holding. Coal power plants: 10% additional emissions x 80%part laod x 0.9T CO2 per MWhe = 0.36 T CO2 per MWh response holding 34 42 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Table 3: CO2 emissions avoided per year per EV or heat pump archetype for dynamic frequency response provision Carbon emissions avoided (Tonnes CO2/year) Home charging EV 1.20 Work charging EV 1.17 Fleet EV 1.20 Residential heat pump 1.87 Industrial/Commercial heat pump 0.60 per kW electric 6.5 Commercial models Commercial models exist to enable third party assets to contribute to National Grid’s management of the electricity system. The commercial models that are used need to be able to deliver sufficient resources to meet National Grid’s needs in a cost effective manner for consumers. Current procurement of frequency response (both mandatory frequency response and 38 FFR) has been developed for a system in which conventional generation facilities provide most of the required response. FFR contracts also provide a route to market for frequency response from distributed demand side resources by aggregators. Moreover, novel products such as the FFR bridging contract, aim to increase market access for demand side participation. Current frequency response products and contractual conditions have however not been developed particularly for novel distributed demand side opportunities. Highly responsive demand side assets may be able to provide more effective response when these conditions are adapted. Moreover current conditions are not set up for the particular requirements of these resources, such as the impact of the large number of small assets on telemetry and verification, diurnal variability, predictability and in the short term minimum capacity. ENTSO-E and European TSOs are exploring the option to mandate frequency sensitive electricity consumption for thermostatically controlled appliances, such as heat pumps. This mandatory requirement would be taken up in Demand Connection Codes (DCC) and implemented through EcoDesign standards. 6.5.1 Key commercial models In the following reviews we distinguish between three high level commercial models, as summarised in Figure 34. This distinction is based on two key aspects. The first aspect is 38 Comprising e.g. large synchronised thermal generation, pumped hydro storage, embedded generators 43 Frequency Sensitive EV and Heat Pump Power Consumption Final Report whether frequency response is provided through a commercial procurement process, or mandated. The second aspect is a distinction between more conventional firm capacity 39 contracts and more dynamic capacity contracts . Firm commercial procurement Current FFR contracts are an example of firm capacity contracts. In this model response capacity is bid well in advance with a fixed capacity over the bid window. Reliability and accuracy conditions are usually defined as static minimum requirements. These contract conditions have typically been developed for response provision by large conventional generation. Dynamic commercial procurement Dynamic capacity contracts may differ from firm capacity contracts on a number of aspects. They may allow profiled bids (with varying capacity nominations for different time periods), short lead times to nomination or ability to adapt capacity nominations close to delivery time, and lower minimum capacity requirements. Another key aspect may be remuneration based on effectiveness and performance, rather than setting minimum requirements. Mandatory frequency response A third option is the introduction of mandatory frequency responsive consumption. ENTSO-E and TSOs are currently reviewing the option to mandate frequency responsive consumption for thermostatically controlled devices. An alternative option is to mandate the technical capability, but implement a commercial procurement model. The following sections will review the key barriers for the development of demand side frequency response services, and compare how these are addressed in the different models. High minimum capacity requirements, fixed capacity bids, long lead times for nomination, minimum accuracy and reliability requirements Low minimum capacity requirements, profiled bids, short lead times to nomination, remuneration based on effectiveness and performance Commercial procurement Firm requirements Dynamic requirements Mandatory Firm commercial procurement Dynamic commercial procurement Mandatory frequency response Figure 34: High level summary of potential commercial models 39 This is separate from the FFR distinction in static and dynamic frequency response. Current FFR services are both classed as “firm capacity contracts” in this review. 44 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 6.5.2 Key barriers and advantages for three commercial models Technical aspects The available EV or heat pump response capacity may vary significantly within a day. 40 Current FFR bids are typically for a fixed capacity per period , while the available response from EVs and heat pumps may vary significantly throughout the day. Firm commercial procurement Aggregators are responsible for providing firm capacity bids to National Grid. As a result not all EV and heat pump potential response capacity may be utilised. Dynamic commercial procurement Could allow more flexible bids, for instance supporting profiled bids for varying capacity throughout the day. As a result the potential response from EVs and heat pumps may be more fully utilised, but increases operational challenge for National Grid. Mandatory model All available response capacity at any one time is utilised by default as first option. National Grid may have limited visibility of the response capacity that will be available. There is also no mechanism to guarantee that the mandatory response is the lowest overall system cost option. 41 A large portfolio of small demand assets provides reliability through diversity , which is different from the reliability provided by individual large assets. When a single large provider of response is unexpectedly not available this has a large impact on the total available response, and hence National Grid places reliability requirements on response providers. However a large fleet of small demand assets provides reliability through diversity. Even when the contracted amount is not provided due to failure of some assets, the total response from that portfolio of assets at any given time may still be close to the total contracted response, limiting the impact on the total response availability. Firm commercial procurement Typically have high static reliability requirements, risk of meeting these lie with aggregator. As a result not all EV and heat pump potential response capacity may be utilised. Dynamic commercial procurement 40 Typically whole day, or 07:00 to 11:00 and 11:00 to 07:00. Bids for any time frame can be placed, however this is not common practice and National Grid will moreover assess each bid individually. 41 This refers to the inherent portfolio reliability due to the fact that failure of assets is likely to be uncorrelated. In other words, a few assets failing at any given time will have a small impact on overall available capacity if they are part of a large portfolio, and so the overall portfolio reliability is likely to be higher than the individual reliabilities of each of the assets. 45 Frequency Sensitive EV and Heat Pump Power Consumption Final Report May be structured to utilise diversified reliability more effectively, using staggered reliability requirements or performance based remuneration. May require National Grid to develop new tools to assess overall response reliability. Mandatory model Diversified reliability can similarly be utilised. Places a larger responsibility on National Grid to assess the mandatory reliability and residual response requirements. Available response capacity from demand assets is more unpredictable the longer the forecast horizon, depending on external influences. The available response capacity from demand assets may depend on external influences, for instance outside temperature for heat pumps. The accuracy of available response forecasts depends on the forecast horizon. This may result in not all EV and heat pump potential response capacity being utilised. Firm commercial procurement Available capacity needs to be forecast well in advance and turn out capacity may differ. As a result not all EV and heat pump potential response capacity may be utilised. Dynamic commercial procurement May be designed to allow shorter lead times for nomination of available capacity or allow bid capacities to be adapted closer to time of delivery. May result in more residual response capacity instructed by National Grid to be available at shorter notice. Mandatory model All response capacity is used as it is available. Requires National Grid to develop tools and processes to forecast mandatory response levels, and more instruction on short time scale to other providers of response to be available. Operational aspects A high overall uptake or local EV and heat pump clustering may require active management of response settings. When demand assets contribute a significant part of total frequency response capacity and their response characteristics are highly correlated, this may result in dynamic frequency stability issues. This is especially the case when assets turn on again simultaneously after a low frequency event. Similarly local distribution network issues, e.g. voltage stability and harmonics, could arise with high local clustered uptake, even when overall uptake is not yet high. This may require active management of the assets to ensure a stable response, or diversification of individual asset response characteristics. Managing and diversifying response characteristic may be more readily done by aggregators in commercial procurement models, than by National Grid in a mandatory 46 Frequency Sensitive EV and Heat Pump Power Consumption Final Report model. Aggregators have a commercial relation with individual providers, and this could be implemented as a precondition to participating. Provision of frequency response by demand assets could change consumption profiles, potentially increasing imbalance. For EVs and HPs, implementing frequency sensitive charging may impact the overall consumption profile. When uptake of demand side frequency response increases without visibility for suppliers, this may result in deviations in the Non Half Hourly (NHH) demand curve from what the suppliers expect. This could pose problems for suppliers in settlement, potentially increasing system imbalance and balancing costs. Although modified settlement configurations can be developed, this is complicated in commercial models if the aggregator is a different entity from the supplier of the consumer. The supplier may not have visibility of third party contracts and impacts or contract changes of a consumer. The impact in future Half Hourly (HH) settlement will depend on how this is implemented (aggregated or individual settlement). Suppliers may experience increased imbalance positions if similarly the aggregator is another third party and the supplier does not have visibility of frequency response contracts and impacts. Reduced diversity and correlated response to frequency deviation events may result in thermal and voltage issues in already strained distribution network areas. Clustered provision of frequency response by demand assets may reduce the diversity of electricity consumption, especially with correlated consumption increase after a frequency deviation event. This may result in thermal constraints and voltage increases in already strained distribution network areas. This may require coordination with DNOs to provide transparency of demand side frequency response providers, and limitations or diversification of response in already strained areas. Commercial procurement May provide additional complexity, needing to develop processes involving the TSO, DNOs and commercial parties (aggregators, suppliers). May reduce the overall required capacity for frequency response and increase incentives for consumer to participate. Mandatory model Can be resolved between the TSO and the DNO, without third parties. Does require National Grid to develop control systems to manage this. Commercial aspects Ability of electrical equipment to provide rapid and accurate response provides higher effectiveness in arresting and restoring frequency deviations than conventional mechanical response providers. Firm commercial procurement 47 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Current firm response contracts do not specifically remunerate fast response capabilities or accuracy of response, beyond minimum reliability requirements. Although providers could tender with higher response characteristics, National Grid does not currently have a process in place to value faster response. Dynamic commercial procurement Could be designed to remunerate on the basis of delivered response speed and accuracy, similar to pay-for-performance frequency regulation in e.g. the US PJM market. May reduce the overall required capacity for frequency response and increase incentives for consumer to participate. Mandatory model Setting of minimum requirements is likely to be required. May not incentivise the development of faster and more accurate response capabilities. Demand assets are suited to provide other DSR uses as well Demand assets could be used for a wide range of other DSR uses in addition to frequency response. Stacking different DSR uses increases the effective use of assets, reduces overall system cost and may increase the value of propositions for consumers, encouraging a greater uptake of DSR. Commercial procurement May support the integration with other DSR uses by aggregators. Mandatory model May reduce the ability of aggregators to bundle different DSR uses and may reduce the overall attractiveness of customer propositions, dis-incentivising the uptake of DSR. This may be addressed by larger local DSR coordination role by DNOs. Frequency response provision may represent only limited value on an individual asset basis The analysis shows there is a significant system benefit for EV and heat pump frequency response. However this opportunity represents a limited value at an individual asset level. Uptake of this opportunity can be supported by efficient commercial models or through mandating. As discussed in section 6.1 a cost effective implementation of frequency response in equipment requires standardised integration of additional components in charge points and heat pumps. Although the cost of doing so at scale may be relatively limited (of the order of £10 per device) initial implementation is likely to be more expensive and also comes with significant development costs. This requires either confidence in a market for technology developers or mandating this capability in design standards or demand connection codes. Commercial procurement 48 Frequency Sensitive EV and Heat Pump Power Consumption Final Report May allow the potential response from demand assets to be utilised more effectively especially, as discussed in the previous paragraphs by; Accommodating the specific characteristics of demand assets and providing equitable remuneration based on performance. Enabling aggregators to combine the provision of frequency response with other DSR uses, lowering the business and operational overhead costs per service and increasing the value of customer propositions, which may increase participation levels Mandatory model Enabling uptake despite diluted value at an individual level may also be supported through mandating frequency response; As discussed in the previous paragraphs, key challenges include that it may not unlock the full potential, and may limit the ability of third parties to also provide other DSR uses using the same assets. A mandatory scheme may include remuneration of consumers, which may depend on whether the costs of administering such remuneration would outweigh benefits to consumers. 6.5.3 Commercial model conclusion The analysis shows a clear system benefit of frequency response from EVs and heat pumps, but the value is diluted on an individual asset level. Unlocking the opportunity that EVs and heat pumps present requires efficient commercial models, or, alternatively, mandatory provision. A qualitative comparison of the different models for EVs and heat pumps is explored in more detail in section 8.2. Commercial procurement models, especially dynamic models, may unlock a high response from individual assets and support the integration by aggregators with other DSR uses, increasing the attractiveness of propositions to consumers. Dynamic commercial procurement moreover most readily allows innovative development of more effective frequency response from demand assets. Commercial procurement models moreover provide the closest fit with current National Grid practices and processes. This however does not provide guarantees on overall uptake. Compared to current Firm Frequency Response contracts, dynamic contracts could support the potential and value of EV and heat pump frequency response by enabling; Profiled bids Adaptation of reliability requirements to reflect diversified reliability Shorter lead times to nomination of capacity Utilisation of fast and accurate response capabilities, and provision of equitable remuneration for these capabilities Integration with other DSR services by assets. A mandatory model on the other hand provides a high guarantee on the overall uptake. However it may not unlock the full potential of individual assets, and may inhibit the integration with other DSR uses and the development of novel deployment models. It would moreover require National Grid to develop new tools and practices to manage 49 Frequency Sensitive EV and Heat Pump Power Consumption Final Report frequency response provision from demand side assets. Difference in individual response constraints and contributions may also lead to issues of fairness. Differences are especially prevalent with EVs where response availability depends on driving and charging practices. In either case, active management of response settings will be required to avoid network issues associated with a reduction in diversity. Also, possible impacts on aggregate demand profiles and hence supplier imbalance must be addressed. 50 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 7 Technical barriers and challenges The literature review and stakeholder interviews for this study indicate that there are no fundamental technical constraints to provide frequency response using EVs or heat pumps. However, both require further technical development in a number of areas, and specific technology characteristics may impact the extent of their capability to provide frequency response. In the analysis we distinguish between four key areas that may present technical challenges and barriers to enabling frequency response from EVs and heat pumps: 1. 2. 3. 4. Hardware capability Control & automation Telemetry and verification Operation and network impacts These are discussed for both EVs and heat pump in the following sections. In this analysis it is assumed that frequency response capability is implemented in a similar way to today’s providers. The components of such a system are outlined in Figure 35. In this setup, assets are equipped with frequency sensors to measure system frequency locally. This signal is used in the local control unit, alongside user inputs (such as desired temperature or time by which charging needs to be complete) and frequency response characteristics (which may be controlled remotely), to manage charge and consumption rates. A third party, for example an aggregator or National Grid, communicates with the control unit to adapt response characteristics and monitor electricity consumption. An alternative approach would be for aggregators to measure system frequency centrally and instruct assets to respond. Such a setup would however increase response times, due to central computation time and delays due to two way communication necessary to instruct assets to respond. Another possibility is for National Grid to broadcast a regulation signal, which is based on the system frequency. This approach is used in the US PJM market. 51 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Figure 35: Components necessary for frequency response provision by EVs and heat pumps with the corresponding data flows 7.1 Electric Vehicles 7.1.1 Hardware capability There are no fundamental technical barriers to EVs providing frequency response and no hardware changes to the EV itself are required. The EV battery provides energy storage and charge power can be reduced or interrupted without significant adverse effects. Battery technology is particularly well suited to providing response because of the speed of the power electronics that regulate charging. The response time of EVs to a control signal is approximately 1 second, making them suitable for provision of very fast frequency response services. The IEC 61851 standards for EV charging require that the car battery responds to a control signal within 5 seconds, while in practice response times are much less. 7.1.2 Control and automation Typically, EV charge points do not currently include the components necessary for frequency response provision and would therefore require a design update. Including frequency responsive capability in the design of a charge point would be relatively straightforward. However, retrofitting this capability in already installed charge points would be challenging due to the difficulty in physically fitting the components required and in integrating the functionality with existing hardware and software. Therefore, early update of charge point design would be necessary to prevent high uptake of unsuitable equipment. Some charge points, mainly higher specification mode 4 DC units, may already have the required frequency measurement, metering and telemetry capability and so may only need 52 Frequency Sensitive EV and Heat Pump Power Consumption Final Report a software update in order to implement frequency response. However, most mode 3 AC charge points do not currently measure frequency or have sufficient power metering. Communications and remote control capability is also currently limited, with low data transfer rates and only crude on/off remote control possible. One practical issue raised by stakeholders is the fact that there is currently no communication standard for the EV to notify the charge point of its state of charge (SOC). Such a standard would need to be developed in order to ensure that the charge point is aware of the EV charging requirements, and can therefore manage the charging without any negative impact on the driver. This is particularly an issue for AC charging, as for DC charging, there is often more communication between the car and the charge point. Also, the impact of frequency response on charging management will not necessarily be well defined in all cases. For example, a sudden reduction in charge power may be interpreted by some EVs as an error condition and result in a shut down. These aspects will need to be examined in trials in order to fully characterise the impact on EVs of varying charge power in response to grid frequency. 7.2 Heat Pumps 7.2.1 Hardware capability There are no fundamental barriers for heat pumps to provide frequency response, though, unlike for EVs, there are hardware issues to resolve. The most fundamental design concerns are around the speed of heat pump response. Heat pumps do not respond as quickly as EVs or other electric devices because in order to make changes to the compressor speed, a mechanical adjustment must be made. While turning off the heat pump would be near instantaneous (on the order of ms), increasing the power draw takes tens of seconds, or even minutes, according to manufacturers. This is because adequate lubrication of the compressor and general system stability must be ensured. Variable speed heat pumps, which represent the majority of models currently on the market, may also ramp down slowly so that control is smooth, though this is not a fundamental need and could be overridden. These issues can be addressed in periodic redesign of heat pumps, especially given that this is still a nascent market, but require a driver for manufacturers to do so. A key requirement for heat pumps providing frequency response is the presence of thermal storage. The volume of thermal storage available is uncertain, though, as discussed in section 5.2, even with limited thermal storage heat pumps could still contribute significantly to frequency response provision. The expected impact of frequency responsive control on heat pump lifetime and efficiency depends on the type of heat pump and the control strategy implemented. The number of 42 heat pump stop/starts should be limited to six per hour , and, in general, according to heat pump manufacturers, varying power is preferable to on/off control. This implies that variable speed heat pumps are more suited to providing response, though issues around the speed of response, discussed above, require real world investigation. 42 This is based on interviews with heat pump manufacturers and on the 2012 EA Technology report, ‘The Effects of Cycling on Heat Pump Performance’. 53 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Large industrial or commercial heat pumps will have more site specific flexibility and hardware characteristics and may therefore need bespoke assessment to see if they can provide frequency response. 7.2.2 Control and automation Generally, heat pumps do not currently have all the components required to provide frequency response. The integration of these features into the complex control system of the heat pump and the existing Building Energy Management System (BEMS) may entail higher costs than for EVs. However, for residential heat pumps, much of the hardware capability, with the exception of the frequency sensor, may become standard in the near term. Heat pumps are expected to have internet connectivity, power metering (which, without a clear driver, may not satisfy National Grid’s requirements) and the ability to respond to external control signals. Other features, which may enhance the ability to provide frequency response, are currently in development. These include self-learning capability, whereby the heat pump would learn the thermal response of the house and the typical user behaviour, thus enhancing predictability and reliability of response, as well as user comfort. For commercial and industrial heat pumps, there is typically already significant monitoring via the BEMS, so these may already have the required metering capability. However, they may not be connected to the internet and may only have crude control capability, in which case additional hardware would still be required to provide frequency response. 7.3 Verification and telemetry A number of stakeholders interviewed for this study indicated that verification requirements, and clarity in how metering and verification is reported to National Grid, are a key issue in the next steps to develop demand frequency response provision. A first aspect is the baseline methodology that is used to determine the amount of response that is provided (response traces compared to counterfactual trace). Developing this methodology early on allows developers to trial under conditions that will be the same in commercial operation and provides transparency of the impact on future business cases. A second aspect is the data evidence that needs to be provided to National Grid. Current FFR providers are required to provide evidence of response provided for each individual asset. This can be done by providing real time second-by-second data or real time minuteby-minute data with subsequent second-by-second data after an event. For demand side participation it will be similarly critical to verify response and guarantee reported response levels are accurate. National Grid is likely to only require real time data on an aggregator level, rather than individual asset level, with individual asset data retained for later verification or auditing. In some trials response verification was carried out based on metering a sample of the participants and extrapolating total response from that, mainly because of high costs. For a business as usual service this is not seen as a likely option for National Grid. This does imply that aggregators need to collect real time data from individual assets. This has implications for potential telemetry options, as discussed below. The real time data exchange and instruction signals requires a telemetry system with high bandwidth and low latency, separate from the smart meter. The most cost effective and 54 Frequency Sensitive EV and Heat Pump Power Consumption Final Report readily available option is to use a broadband connection, possibly combined with local wireless mesh to connect multiple devices for fleet operators. Of the different wireless mesh options WiFi provides a higher bandwidth compared to alternatives (e.g. ZigBee or 43 ZWave ). Another option would be to use the licenced cellular network (e.g. GPRS, 4G, 5G), however this is significantly more expensive than broadband. Other technologies are also emerging that use the unlicensed part of the network, however these have typically low bandwidths. A fourth option may be to use power line communications. Currently there is no standardised protocol for communication between charge points or home and energy management systems, and aggregator management systems. Many proprietary protocols have been developed. Open protocols have also been developed, notably for EVs the Open Charge Point Protocol (OCPP), developed by E-Laad and supported by a range of other stakeholders. A challenge for open protocols is to enable more extensive functionalities, such as required for frequency response. Proprietary protocols can more readily be adapted to enable more extensive functionality. 7.4 Network impacts and system operation Transmission network With large scale uptake, correlated responses from similar demand assets may result in system frequency oscillations. This can be addressed in control design of response settings. In order to provide this, aggregators need to be able to adjust and vary asset response characteristics; including the ability to put assets in and out of frequency sensitive mode, set dead band levels, and adjust dynamic response functions (hysteresis, droop curve). With a large number of kW scale assets, this likely requires management and communication systems between National Grid and aggregators, and between aggregators and assets, with more advanced capabilities than currently employed for instance in aggregating embedded generation. This issue may be especially challenging in a Mandatory Frequency Response model. With low uptake these impacts can be absorbed by other assets, but it may require more active control management when demand assets provide a significant part of the total response. Distribution network Clustered uptake of frequency response by EVs and heat pumps may result in a significant reduction in demand diversity, especially with correlated consumption increase of demand response providers after a frequency event. DNOs interviewed for this study do not expect this to be an issue for most parts of the distribution network. However, for some local network areas that are already strained, this may result in various issues including: Voltage stability (voltage drop and rise); this may happen especially in areas that are net generation. Thermal overloads of transformers; analysis carried out in the Green Motion project for non-diversified ToU tariffs indicates that this may result in overloaded 43 These are protocols to create local networks, often used for applications such as home automation. 55 Frequency Sensitive EV and Heat Pump Power Consumption Final Report transformers in substations in some network areas, while in other network areas 44 such issues will not occur . Harmonic emissions; these could increase (especially if charge rates are modulated in EVs with lower performance inverters), requiring DNOs to take countermeasures (e.g. filters). For modern inverters this is not expected to be a fundamental issue, but should be addressed by OEMs and charge point 45 manufacturers in design of inverters . This is similar to effects that may happen with non-diversified Time of Use (ToU) tariffs in areas of high EV uptake. Addressing these issues requires a process to be in place between DNOs and National Grid to identify where clustered uptake may result in issues on the distribution network. By extension this requires National Grid to have visibility of the location of aggregated response and a system to reject providers in some areas or instruct aggregate response characteristics in collaboration with DNOs. This further increases the complexity of management and communication systems for a large number of kW scale assets contributing to frequency response, as discussed in the previous section. Where EVs or heat pumps providing frequency response are part of a DNO Active Network Management (ANM) system, there is a risk that the ANM system negates frequency response actions. This similarly requires transparency of location of response providers between DNOs and SO. On a very localised level, rapid switching of large loads such as EVs and heat pumps for frequency response may result in transient effects such as flicker. Flicker is caused by rapid fluctuations in the voltage of the power supply and results in visible dimming oflighting, hence its name. 8 Roll Out The previous sections analysed the EV and heat pump technology aspects for enabling frequency response, the potential impact on other stakeholders, the benefits and costs, and outlined the possible commercial models. Figure 36 outlines the main elements in the roll out of EV and heat pump frequency response, and the key barriers for each are summarised in the following sections. Section 8.2 than reviews the relative merit of commercial procurement or mandating for frequency response from EVs and heat pumps. The last section identifies the key next steps that National Grid may take to develop these opportunities and where collaboration with other stakeholders may be required. 44 Green eMotion, Recommendations on grid-supporting opportunities of EVs, 2012 Dansk Energy, Green eMotion grid impact studies of electric vehicles – EV’s impact on Power Quality related to harmonics, 2013 45 56 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Development of EV and heat pump frequency response Commercial market development 1. Strategic agenda & business case 2. Enabling deployment 3. Market access Impact on other stakeholders 1. Power quality (DNO) 2. Thermal overload and voltage stability 3. Settlement and supplier balance Technology and proposition development 1. Hardware capability 2. Control & automation 3. Telemetry and verification 4. Delivery model Figure 36 Key components in frequency response service development: commercial market, impact on other stakeholders and technology and proposition development 8.1 Summary of key barriers for the development of EV and heat pump frequency response 8.1.1 Commercial market development 1. The analysis shows that both EVs and heat pumps may be able to provide a significant part of National Grid’s frequency response needs by 2030, and may be cost effective compared to incumbent providers. The value at an individual asset level is however relatively low. This provides limited incentives to develop these opportunities, which require investments in further development of EVs and heat pumps, communications and telemetry, as well as in business models and management systems. 2. Bilateral contracts may provide a route to market for providers that can’t meet regular FFR requirements and can provide National Grid unique or novel response capabilities. However the technical requirements, service conditions, verification 57 Frequency Sensitive EV and Heat Pump Power Consumption Final Report processes and baseline methodology for aggregated small scale demand providers are not clear. 3. There is currently no market or future framework for the provision of frequency response from large fleets of kW scale EVs and heat pumps, which have different characteristics and capabilities compared to conventional synchronised generators. Fast and accurate responding demand assets may moreover be of increasing value in helping to manage system frequency, reducing the overall need for frequency response. A review of contractual terms and conditions may be required to utilise the potentially more effective system frequency management capabilities, and accommodate specific reliability and availability characteristics. Also, a large number of kW scale assets providing frequency response may have local network and system impacts, as discussed in section 7.4. Mechanisms will need to be put in place to address these impacts. 8.1.2 Impact on other stakeholders 1. Clustered uptake of frequency responsive EVs may increase emissions of harmonics into the distribution network (stakeholders: SO, DNOs, equipment manufacturers, aggregators). Rapid switching of EVs and heat pumps for frequency response may introduce flicker (stakeholders: SO, DNOs, equipment manufacturers, aggregators). 2. Clustered uptake of frequency responsive EVs may result in thermal overload or voltage stability issues in some distribution network areas (stakeholders: SO, DNOs, equipment manufacturers, aggregators). 3. High uptake of frequency responsive EVs and heat pumps may impact the accuracy of SSCs and system imbalance for NHH settlement, if suppliers are not the frequency response aggregator and do not have visibility of response provision impact. Similarly for HH settlement lack of visibility may increase supplier imbalance. (stakeholders: SO, aggregators, suppliers, Elexon) . 8.1.3 Technology development 1. For EVs further technology development may be required, especially interaction with safety measures and emission of harmonics into the distribution network. For heat pumps further technology development is required, especially to support higher ramp rates or stable and safe interruptions. 2. EV charge points (especially AC charge points) and heat pumps do not currently include the components necessary to support frequency response and would therefore require a design update. Moreover there are no widespread standardised communication protocols between vehicles and charge points that support all required functionality. 3. Currently there is no standardised protocol for communication between charge points, heat pumps, and aggregator management systems that are interoperable and support frequency response requirements. 4. Especially for EVs, most stakeholders are currently focused on developing charging infrastructure and basic smart charging management (e.g. peak load reduction to reduce wholesale purchase cost, network and connection charges, Renewable Energy Sources integration). The development of frequency responsive capability and of communication and interoperability standards to support this may not be prioritised by commercial parties. This may mean that technologies (EVs and charge points) are being developed for large scale 58 Frequency Sensitive EV and Heat Pump Power Consumption Final Report deployment, without integrating the capabilities required to provide frequency response, risking that early large scale deployments will not be able to provide such services or only at high retrofit cost. 8.2 Comparison of commercial procurement and mandatory provision National Grid currently procures all frequency response services on a commercial basis. However all generators caught by the requirements of the grid code are mandated to have the capability to provide frequency response. The analysis shows a clear system benefit of EV and heat pump frequency response, but it is unclear if the diluted value at an individual asset level will be sufficient to develop these opportunities. Developing these opportunities can be realised either by supporting early developments in a commercial procurement model or mandating the provision of frequency response. Section 6.5 provides a high level outline for commercial and mandatory models, and discusses how they impact the key barriers for the uptake of frequency response from EVs and heat pumps. The following sections provide a qualitative comparison between commercial and mandatory models for frequency by EVs and heat pumps respectively. 8.2.1 EVs The comparative analysis summarised in Table 4 indicates that a commercial procurement model for EV frequency response may be most appropriate; Although the value at an individual asset level is diluted, third parties are already developing smart charging propositions that may be extended to include frequency response. Combining revenues from different DSR uses and utilising a shared platform may increase the value of a proposition to customers and support uptake of frequency response. Mandated frequency response may limit other DSR opportunities and limit the overall value of propositions. The analysis shows that the system benefit of frequency response from EVs may be higher when combined with controlled smart charging; mandatory frequency response does not readily support this development. The potential and value of frequency response from EVs may potentially be increased by enabling bi-directional charging for frequency response. However enabling this may require significant technology development and mandatory frequency response may reduce the driver to do so. EV availability for frequency response does not only depend on instantaneous parameters, but may depend on consumer future constraints (e.g. required SOC), this may be challenging to accommodate in a mandatory model. However, most stakeholders are currently focused on developing charging infrastructure and basic smart charging management (for example connection capacity management, peak shifting, local renewable integration). The development of frequency responsive capability and of communication and interoperability standards to support this may not be prioritised by commercial parties. This may mean that technologies (EVs and charge points) are being developed for large scale deployment, without integrating the capabilities 59 Frequency Sensitive EV and Heat Pump Power Consumption Final Report required to provide frequency response, risking that early large scale deployments will not be able to provide such services or at high retrofit cost. This may provide a rationale to mandate the capability to provide frequency response. EV deployments currently receive heavy public financial support (subsidy for the installation of charge points and EV subsidy plug-in Car Grant). This provides an opportunity to require supported technologies to be frequency response capable, to implement interoperable communications, to adhere to standards (such as ISO 15118) and consider impacts on power quality in design. This would reduce barriers and limit incompatibility for subsequent development of frequency response services. Table 4 Summary qualitative comparison mandatory and commercial model for EVs Green = Very high compatibility with model Yellow = High compatibility with model Orange = Low compatibility with model Red = Very low compatibility with model EVs Mandatory Commercial System potential System cost effectiveness Individual asset incentive Consumer quality of service impact High system potential Cost effective incumbents compared to Limited individual asset value Challenging to manage Variation in impact across consumers Very different, depending on driving patterns, charge location, preferences and EV Simple compliance tests May be challenging, depending on programming and control settings Impact on development of more effective business models and technology capabilities May inhibit development of more effective frequency response, spreading available response in time using controlled smart charging. May inhibit combining frequency response with other nascent DSR propositions in combined platforms, hence lowering overall overhead costs and increasing value to consumer Dynamic interaction with other May require collaborative actions with DNOs to limit local network 60 Frequency Sensitive EV and Heat Pump Power Consumption Final Report stakeholders impacts in some areas 8.2.2 Heat pumps The comparative analysis summarised in Table 5 does not indicate that either a mandatory or commercial procurement model is preferred. The choice between them will largely depend on whether a sector will develop for this opportunity and on the needs of National Grid. The value at an individual asset level is limited, but higher than for EVs. However there appear less potential advantages than for EVs (lower ramping rates, no potential for bidirectional response, less development to provide other DSR uses). The limited ability of heat pumps to contribute to frequency response at night is more fundamental, and mandating frequency response is not expected to limit the development of solutions for this. Mandatory response may be more straightforward to implement than for EVs, as appliances can be programmed to follow frequency deviations within thermostat settings and response characteristics constraints. Table 5 Summary comparative analysis mandatory and commercial model for heat pumps Green = Very high compatibility with model Yellow = High compatibility with model Heat pumps Mandatory Commercial System potential System cost effectiveness Individual asset incentive High system potential Cost effective incumbents compared to Reasonable individual asset value Consumer quality of service impact Power draw can be managed within thermostat constraints Variation in impact across consumers May depend on availability of storage or type of heat pump Simple compliance tests Compliance tests straightforward Impact on development of more effective business models and technology capabilities may be No active management expected to be developed that may increase the potential as with EVs May inhibit combining frequency response with other nascent DSR propositions in combined platforms, 61 Frequency Sensitive EV and Heat Pump Power Consumption Final Report hence lowering overall overhead costs and increasing value to consumer, however less developments compared to EVs Dynamic interaction with other stakeholders May require collaborative actions with DNOs to limit local network impacts in some areas, but less dynamic interaction compared to EVs 8.3 Next steps for EV and heat pump frequency response roll out The development of business models and technologies requires investments from technology developers and third parties developing technologies and customer propositions. These investments can be de-risked by a dual approach of supporting technology development and creating commercial markets. This section sets out possible next steps to overcome the barriers summarised in section 8.1. As discussed in the previous section, for EVs a commercial procurement model appears to be the most appropriate model to develop. For heat pumps both mandating and commercial procurement may be appropriate, and National Grid may investigate the impact and benefit of each in parallel to further technology development. Supporting commercial market development should focus on providing confidence that there will be a market in the long term that is accessible and amenable to frequency response from demand side assets, and values providers based on their effective contribution to resolving frequency management. The next steps are set out in more detail below. Investigation of mandatory frequency provision should focus on the impact on consumers and other stakeholders (especially DNOs), cost-benefit analysis (including the likelihood that the opportunity will be developed in the absence of mandating) and implications for National Grid operations. These may be carried out in the next few years to inform the ENTSO-E investigation of mandatory frequency response (2020 timeline) for thermostatically controlled appliances in Ecodesign standards and demand connection codes. 1. Near term – strategic agenda and business case Early development may be supported by clear statements showing that frequency response from a large portfolio of low power (kW scale) assets is being actively explored as part of strategic agendas for system management. Carry out technical assessment of the system and local network impacts of large scale or clustered EV and heat pump frequency response uptake, and potential control strategies to mitigate these. 62 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Demonstration projects can support technology developers and aggregators in overcoming technical barriers and the uncertainty around real life business models. Possible options for National Grid to consider include: Providing clarity on service requirements (response rates, response lengths, reliability), technical requirements (power metering, verification) and contractual terms. Engaging with other stakeholders to define what the response characteristics are that may need to be controlled and which will be valued for dynamic control capability with large scale uptake. This allows developers to take this up in technical design. Defining baseline methodology used to identify an asset’s contribution to frequency response. 2. Medium term – enabling deployment Provide commercial opportunities for early adopters through commercial bilateral contracts, with less stringent technical requirements or novel commercial models. High transaction costs and lack of transparency in bilateral contracts would see these phased out as uptake increases. Collaborate with DNOs to identify impact on diversity. Develop industry rules and processes to create transparency between aggregators, National Grid, DNOs and suppliers on location and utilisation of frequency responsive demand. Develop processes to mitigate or limit adverse local network impacts (thermal overload or voltage stability issues). Collaborate with aggregators and suppliers to identify impact on settlement (system balance) or supplier imbalance if it is not the response aggregator. Balancing responsibility on a connection should be clearly defined and consistently metered. There should be no gaps or overlaps in the balancing responsibility of different actors on a connection. Depending on the implementation of half hourly settlement, new SSCs may need to be developed for different types of response providers. 3. Long term – market access Develop services which remunerate higher responsiveness that some demand side technologies may provide (fast response capability, accurate tracking of frequency deviations), and take into account challenges and limitations of demand side technologies (short lead times for capacity nomination, profiled bids). Implement industry processes and rules providing transparency between DNOs, TSO and aggregators on location and utilisation of frequency response, and implement processes to mitigate adverse local network impacts. 63 Frequency Sensitive EV and Heat Pump Power Consumption Final Report 9 Conclusion National Grid commissioned Element Energy to carry out this Network Innovation Allowance (NIA) project to assess the potential for electric vehicles and heat pumps to contribute to frequency response. The analysis shows there is a significant technical potential for both EVs and heat pumps to contribute to frequency response by 2030. The combined annual average potential may reach 1200MW by 2030 in the medium uptake scenario (excluding residential heat pumps). This corresponds to an average 82% of National Grid’s projected frequency response requirement in 2030. There are no fundamental technical limitations for both EVs and heat pumps to implement frequency responsive electricity consumption. However further technical development is required for both to realise practical frequency response capability. Most heat pumps in particular require further design iterations to meet dynamic frequency response requirements in a safe and reliable manner. Key development areas to enable frequency response services are the control and communication systems, both between EVs and charge points and between heat pumps and Energy Management Systems, as well as with DSR management platforms. The annual average potential of EVs and heat pumps in the 2030 medium scenario corresponds to approximately £100 million in annual revenues, based on current tendered prices. This value could potentially be higher if the effectiveness of fast responding demand assets in restoring system frequency deviations is taken into account in pricing. While the overall cost of providing frequency response may be very competitive compared to conventional thermal plants, the potential net benefit per household may be relatively limited. This may risk limiting the participation of EV and heat pump owners in frequency response schemes. Deployment will need to be supported by very efficient commercial models or mandated. The development of business models and technologies requires investments from technology developers and third parties developing technologies and customer propositions. These investments can be de-risked by a dual approach supporting technology development and creating commercial markets. Such commercial markets could provide confidence on the long term access, value and market size for these solutions. The development of these opportunities may be supported in the near term by clear statements that large fleets of kW scale assets are part of strategic agendas for system management, and supporting demonstration projects that prove the technology and commercial structures. This may be followed with bilateral contracts to provide frequency response from EVs and heat pumps that are structured to accommodate their specific characteristics and innovative capabilities. In the longer term services may be adapted similarly to accommodate these resources and benefit from more effective frequency response capabilities. . 64 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Acknowledgements Interviews with a wide range of stakeholders were carried out for this study. We are very grateful to all interviewees for their contribution. The organisations that participated in the consultation are listed below. Daikin Ecotricity Electricity North West E.ON Glen Dimplex Global Energy Systems IBM Mitsubishi Electric Nissan Northern Powergrid OpenEnergi Pod Point Renault Scottish and Southern Energy Power Distribution SSE Star Renewable Energy Transport for London UK Power Networks UPS Upside Energy Western Power Distribution 65 Frequency Sensitive EV and Heat Pump Power Consumption Final Report Glossary AC Alternating Current ASHP Air Source Heat Pump BEMS Building Energy Management System BEV Battery Electric Vehicle DCC Demand Connection Codes DNO Distribution Network Operator CCC Committee on Climate Change CCS Combined Charging System CLNR Customer Led Network Revolution COP Coefficient of Performance DC Direct Current DCC Data Communications Company DfT Department for Transport DSM Demand Side Management DSR Demand Side Response ENTSO-E European Network of Transmission System Operators for Electricity EV Electric Vehicle EVSE Electric Vehicle Supply Equipment FCDM Frequency Control by Demand Management FES Future Energy Scenarios FFR Firm Frequency Response GSHP Ground Source Heat Pump HEV Hybrid Electric Vehicle HH Half Hourly IEC International Electrotechnical Commission ISO International Organisation for Standardisation LowCVP Low Carbon Vehicle Partnership 66 Frequency Sensitive EV and Heat Pump Power Consumption Final Report LCL Low Carbon London LCNF Low Carbon Network Fund MBSS Monthly Balancing Services Summary NHH Non Half Hourly OLEV Office for Low Emission Vehicles PC Profile Class PHEV Plug-in Hybrid Electric Vehicle RCD Residual Current Device RED Renewable Energy Directive REEV Range Extended Electric Vehicle RHI Renewable Heat Incentive SME Small or Medium sized Enterprise SOC State of Charge SOF System Operability Framework SSC Standard Settlement Configuration TSO Transmission System Operator UKPN UK Power Networks ULCV Ultra Low Carbon Vehicle 67 Frequency Sensitive EV and Heat Pump Power Consumption Final Report References Options and recommendations to meet the RED transport target, Element Energy for the LowCVP, 2014 Pathways to high penetration of heat pumps, Frontier Economics and Element Energy for the CCC, 2013 A Guide to Electric Vehicle Infrastructure, BEAMA http://www.iec.ch/ https://www.zap-map.com/ https://www.gov.uk/government/statistical-data-sets/veh02-licensed-cars Assessing the viability of EVs in daily life, The Ultra Low Carbon Vehicle Demonstrator Programme, 2013 http://www.networkrevolution.co.uk/ Low Carbon London Report C3, Network impacts of energy efficiency at scale Assessing Heat Pumps as Flexible Load, Hong et al., Loughborough University, 2013 Impact & opportunities for wide-scale EV deployment, Low Carbon London Report B1 Infrastructure in a low-carbon energy system to 2030: Demand Side Response, Grid Scientific and Element Energy, 2013 Green e-Motion Deliverable 9.4 Part 1, Envisaged EU mobility models, role of involved entities, and Cost Benefit Analysis in the context of the European Clearing House mechanism, 2014 MBSS data and FFR post assessment tender reports 2014 PJM, Performance-based Regulation Market in PJM, July 2014 PJM, monthly average regulation prices after introduction Pay for Performance (October 2012), 2011 – 2014 The Effects of Cycling on Heat Pump Performance, EA Technology report, 2012 68