Impact of Low Voltage - Connected low carbon technologies on ...

0 downloads 166 Views 3MB Size Report
Contents. Executive Summary . ..... These technologies present opposing challenges to the distribution network: with PV
Impact of Low Voltage Connected low carbon technologies on network utilisation Low Carbon London Learning Lab

Report B4

ukpowernetworks.co.uk/innovation

Authors Mark Bilton, Nnamdi Ebeanu Chike, Matt Woolf, Predrag Djapic, Martin Wilcox, Goran Strbac

Imperial College London

Report Citation M. Bilton, N. E. Chike, M. Woolf, P. Djapic, M. Wilcox, G. Strbac, “Impact of low voltage – connected low carbon technologies on network utilisation”, Report B4 for the “Low Carbon London” LCNF project: Imperial College London, 2014.

SDRC compliance This report is a contracted deliverable from the Low Carbon London project as set out in the Successful Delivery Reward Criteria (SDRC) sections “Enabling and Integrating Distributed Generation” & “Enabling Electrification of Heat and Transport”.

Report B4 September 2014 © 2014 Imperial College London.

Contents Executive Summary ................................................................................................................... 4 Glossary ..................................................................................................................................... 7 1

2

3

Introduction ....................................................................................................................... 8 1.1

Objectives and scope ................................................................................................. 8

1.2

Context ...................................................................................................................... 9

1.2.1

Heat pumps ....................................................................................................... 9

1.2.2

Photo-voltaic (PV) generation ......................................................................... 12

1.2.3

Factors Affecting PV Output ............................................................................ 13

1.2.4

Power Quality .................................................................................................. 15

Gaps in research .............................................................................................................. 16 2.1

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

2.2

Photo-voltaic ........................................................................................................... 16

LCL PV and heat pump field trial data ............................................................................. 18 3.1

3.1.1

Site 1 (MPAN 7000009339931) ....................................................................... 19

3.1.2

Site 2 (MPAN 7000038631940) ....................................................................... 21

3.2 4

5

Heat pumps ............................................................................................................. 18

Solar photo-voltaics ................................................................................................. 23

Technology scenarios ...................................................................................................... 26 4.1

LV Network choice ................................................................................................... 26

4.2

Heat-pump modelling.............................................................................................. 27

4.3

PV modelling ............................................................................................................ 29

Analysis ............................................................................................................................ 30 5.1

Network model ........................................................................................................ 30

5.2

Heat Pumps ............................................................................................................. 30

5.2.1

Merton secondary 06601: weather scenario 1 (average temperature -4°C) .. 31

5.2.2

Merton secondary 06601: weather scenario 2 (average temperature 0°C) ... 32

5.2.3

Merton secondary 06601: weather scenario 3 (average temperature 4°C) ... 33

5.2.4

Merton secondary 06601: weather scenario 4 (av. temperature 7°C) ........... 33

5.2.5

Summary .......................................................................................................... 34

5.3

Solar PV.................................................................................................................... 35 2

6

5.3.1

Merton secondary 06601: 10% Solar PV penetration ..................................... 35

5.3.2

Merton secondary 06601: 5% Solar PV penetration ....................................... 37

5.3.3

Merton secondary 06601: 25% Solar PV penetration ..................................... 39

5.3.4

Merton secondary 06601: 30% Solar PV penetration ..................................... 40

5.3.5

Summary .......................................................................................................... 41

Conclusion and Recommendations ................................................................................. 42 6.1

Main findings ........................................................................................................... 42

6.2

Recommendations................................................................................................... 42

6.3

Further work ............................................................................................................ 42

7

References ....................................................................................................................... 43

8

Appendix .......................................................................................................................... 44 8.1

Emulation of the summer power flow on a feeder ................................................. 44

8.2

Heuristic rule development ..................................................................................... 46

8.3

Appliance model development ............................................................................... 50

3

Executive Summary This report focuses on the network impact of two key low carbon technologies that are being promoted by UK policy, namely photo-voltaic solar panels (PV) on domestic premises and domestic heat pumps (HP). These technologies present opposing challenges to the distribution network: with PV there is the potential for over-voltage and back-feeding of power from the low voltage (LV) network onto the high voltage (HV) network when local demand is low but PV generation is high; with HPs, the additional load at times of peak demand may cause thermal and voltage limits of infrastructure to be reached. Note that electric vehicles (EV) are sometimes included in the definition of distributed energy resources (DER), however this subject is covered in a companion report on the impact of EV’s on the distribution network. In order to establish the potential impact of these technologies a number of existing domestic PV and domestic HP installations in the London area were monitored for 12 months, the PV’s for power output and the HP’s for power demand. These had been connected to the distribution network under the connection process outlined in the Energy Network Association’s connection guide G83/2 or its earlier variants [1].The demand and generation profiles of these sites are presented and then discussed in terms of impact they might have. The findings from this field data then informed the development of a configurable model which was used to examine the impact on a typical section of distribution network under different uptake scenarios for the two technologies in homes. The modelling exercise aimed to shed light on whether there would be a significant impact on the operation of the network if typical households were to adopt heat pumps and more households fitted PV panels. An example network was selected from the suite of Engineering Instrumentation Zones (EIZs) which form part of UK Power Networks’ London network and whose detailed topologies were assembled by the Low Carbon London project. This network was then used as a test case. The selected network has one of the highest percentages of domestic customers of the EIZs, as this is the type of network most affected by the introduction of both technologies. Identical scenarios were run with different levels of these technologies and the effects on demand profile and voltage perturbations are presented. The findings from the heat pump modelling are summarised in Table 1.

4

Table 1: Heat pump daily peak load increase at different average temperatures

Percentage peak daily load increase over baseline at: 5% 10% 15% 20% penetration penetration penetration penetration Scenario 1 (-4°C average) Scenario 2 (0°C average) Scenario 3 (4°C average) Scenario 4 (7°C average)

19%

33%

48%

72%

9%

18%

32%

39%

9%

14%

18%

25%

5%

7%

11%

15%

For an average outdoor temperature of -4⁰C and a penetration level of 20% of households owning heat pumps, the peak daily load increased by 72% above baseline. As most heat pumps in the network will be working at full capacity in weather this cold, diversity will be greatly reduced. In addition, heat pumps work at reduced efficiency in low temperature conditions. These two factors are responsible for the large increase in peak load. In contrast, at 7⁰C the corresponding peak load increase was just 15%. Here diversity is increased and heat pump efficiency improved. Reducing penetration levels from 20% (to 15%, 10% and 5%), shows approximately proportional reductions in peak loading increase. Neither of UK Power Networks’ or the Department for Energy and Climate Change (DECC)’s forecasts are expected to reach penetration levels above 5% for the next decade, but from these early indications it would appear that tangible additional transformer headroom is required to account for cold conditions in which diversity is much reduced and the Coefficient of Performance (COP) of the heat pumps collapses, even at lower level of 5% uptake amongst domestic customers. Table 2 summarises the findings from modelling the uptake of PV. Higher penetration levels of PVs in the network cause significant levels of back-feeding at the substation. At 30% penetration this reverse power flow was high enough for there to be almost zero net energy demand. Some over-voltage was seen at all penetration levels. At penetration levels which are currently forecast by UK Power Networks to reach around 5% over the next decade, over-voltages of the order of 2 volts may consistently arise unless voltage control regimes are adjusted. Whilst uptake at higher levels is not currently foreseen, it is interesting to examine the nature of the relationship between penetration and instances of over-voltage at higher uptake levels. At 25% penetration the over-voltage more than doubled to 4.3V, but from here onwards it rose steeply to 14V at a 30% penetration level. In practice, inverter cut-outs should prevent over-voltages of more than 9V. Table 2: Summary of findings from PV modelling

Over-voltage Reverse powerflow level

5% 2V None

Penetration level: 10% 25% 3V 4.3V

30% 14V

Low

High

High

5

PVs are in a sense a more benign technology, since if significant over-voltage occurs then they will temporarily cease generating. However, this is detrimental to customers who risk losing the income associated with the Feed-in Tariff. As indicated above, even modest penetrations of 5% PV may require voltage control regimes to be examined and adjusted over the next few years. Heat pumps could present a more serious problem in that existing networks could be driven beyond thermal capacity with only a small percentage of homes using heat pumps for space heating, unless the network has been reinforced. In particularly cold conditions, the maximum demand for a typical home with heat pump technology can reach 4.5kW after diversity, with 3.6kW from the heat pump operating steady state. Above freezing temperatures the heat pump becomes increasingly benign to the system as well as being a lower carbon alternative to gas central heating.

6

Glossary ASHP

Air Source Heat Pump

COP

Coefficient Of Performance

DER

Distributed Energy Resources

DNO

Distribution Network Operator

FIT

Feed-in Tariff

GSHP

Ground Source Heat -Pump

ICT

Information and Communication Technology

LCL

Low Carbon London

PV

Photo-Voltaic

RHI

Renewable Heat Incentive

UKPN UK Power Networks

7

1

Introduction

The work carried out by Imperial College within the Low Carbon London project has analysed the likely impact on the distribution network caused by the greater electrification to meet climate change targets and the increased presence of distributed energy resources (DER) in the network. This report investigates the impact of two of these DER’s: domestic heat pumps and domestic photo-voltaic solar panels (PV). The effects on the distribution network of increased uptake of electric vehicles, another DER, is also of great concern to DNOs and are examined in detail in a companion report.

1.1

Objectives and scope

In common with other Distribution Network Operators (DNOs) in Britain, UK Power Networks is already experiencing and expecting to further support the uptake of domestic micro-generation, principally rooftop solar photo-voltaic panels. UK Power Networks will also ensure that the network is able to support the increased uptake of heat pumps as an alternative to fossil fuel based heating systems. UK Power Networks’ latest planning assumptions1 would expect a total of 343,000 domestic heat pumps on its three licenced networks covering London, East Anglia and the South-East of England by 2023. A total of 299MW of additional heat pumps in commercial premises are expected to be supported on the network by 2023. Finally, 375,000 homes are expected to be generating using one or other form of micro-generation by 2023. Using estimates of the typical sizes of units and planning assumptions for peak demand growth, the micro-generation capacity at full rated output would represent 17% of peak demand. The report investigates the effect of heat pump and PV uptake on the load profile of a typical secondary (11kV/415V) substation and associated feeder voltages. Heat pumps and PV systems will affect the system with significant seasonal variation. Solar PVs are clearly most productive in summer and high summer when residential demand is low. Heat pumps are clearly most active and exhibit least diversity in cold winter conditions. The report begins by reviewing field data from measured Low Carbon London heat pump and PV sites and these data are used to validate simulated models of domestic load. These models are then used to synthesise supply and demand profiles under different system conditions and demonstrate their effect on the distribution network when combined with existing power flows. For heat-pumps, on the modelled network, the primary concern is transformer capacity and a representative substation transformer was chosen for study. The model of domestic load which is used in the studies is introduced in the Appendix. The report aims to identify the thresholds of adoption whereby networks are likely to experience unwanted effects on thermal and voltage limits.

1

http://library.ukpowernetworks.co.uk/library/en/RIIO/Main_Business_Plan_Documents_and_Annexe s/UKPN_Core_scenario.pdf

8

1.2

Context

European and UK policy is promoting heat pumps as a substitute for fossil fuel heating systems and photo-voltaic solar panels to produce zero-carbon electricity. Heat pumps are of particular value when a high proportion of the electricity they consume has been generated from low carbon sources such as nuclear and renewable generation. If adopted in large numbers, heat pumps could present a challenge to the distribution network, as their demand is likely to be highest when electricity demand is already high and distributions systems are already under stress. They will represent an appliance which may not necessarily be of a higher rating than other existing appliances (such as power showers), but will operate for extended periods. Conversely, on a residential network, PV’s generate electricity when demand is at its lowest, with consumers often being at work or on holiday, and this presents the risk of over voltage when supply outweighs demand. The report takes a view in common with the work of the Smart Grid Forum, and the Transform model developed by the GB DNOs, that networks can become fully utilised either when their thermal capacity is reached, or when their voltage headroom has been exhausted [2]. 1.2.1 Heat pumps The UK’s renewable heat incentive (RHI) scheme came into effect on the 9th April 2014. This scheme promotes a range of ‘low carbon’ technologies for use in the residential sector, see Table 3 [3]. Table 3: Low carbon technologies and RHI tariffs Technology

Tariff

Air-source heat pumps

7.3p/kWh

Ground and water-source heat pumps

18.8p/kWh

Biomass-only boilers and biomass pellet stoves with integrated boilers

12.2p/kWh

Solar thermal panels (flat plate and evacuated tube for hot water only)

19.2 p/kWh

The relative tariffs for these technologies reflect estimates of the carbon savings associated with the individual technology. This disparate group of technologies achieve carbon reductions in very different ways. Solar thermal panels use the solar irradiation to heat hot water only, since in the UK their output in winter is insufficient to afford central heating. Biomass technologies rely on a renewable source of fuel, typically wood chips, which can be supplied through sustainable forestry. Heat pumps on the other hand provide heating services by extracting heat from an external source and typically heating water, which in turn may be used to raise internal temperatures [4].

9

Ground and water source heat pumps have the benefit of a large thermal mass, be it a river, lake or a body of earth from which to extract heat. Heat is drawn out by circulating a refrigerant in underground or underwater pipes. Ground source heat pumps are viable in more rural locations, whilst water source heat pumps are likely to be a niche market. The benefit of both ground source and water source heat pumps is that these materials have a higher volumetric heat capacity. Air source heat pumps on the other hand extract heat from air, and this consequently requires ‘warm’ air to be pushed past the heat exchange unit with a fan. Despite the large differential in the RHI for the different heat pump technologies, field trials have produced very similar efficiencies. The RHI tariffs are designed to offer the same rate of return (12%) on the initial investment across the bands. The higher tariff for ground source heat pumps reflects the greater investment needed for these systems rather than differences in efficiency. The efficiency of a heat pump is described by its coefficient of performance (COP). The COP is the amount of heat energy extracted for the electrical energy input. COP’s in well designed systems will be well over 1, otherwise direct electrical heating is a more economic choice. The heat-pump is able to ‘produce’ heat through a process known as the Carnot cycle and this effect poses physical limits upon the performance of a heat pump. While theoretically a heat pump can have a COP in the region of 10, in practice manufacturers of air source heat pumps claim a COP of 4 to 5, but these are often stated for optimal conditions. In practice, field trials of heat pumps, for example the Energy Saving Trusts long-run trial, have reported COPs in the range of 1.2 to 3.6 [5]. These values relate to the whole system performance not just the heat pump itself and a key finding of the EST trial was that system design and installation within the home is critical to achieve good COPs One issue that affects heat-pump performance is unnecessarily rapid cycling between on and off. The behaviour can be caused by radiator valves which are common in UK homes. EA technologies, funded by DECC, investigated this effect with high resolution power and heat monitoring [6]. The tests where conducted in EA Technology’s monitored domestic house. The conclusion of this work was that rapid on-off cycling can effect heat pump performance and that systems should endeavour to have cycle times of over 6 minutes. Another issue affecting heat pumps is that the COP performance drops with lower outside temperatures, or more accurately, with an increase in the differential between the source temperature and the temperature required for the heat demand in the home, supplied by radiators. Figure 1 gives an example of the relationship between source temperature and COP for an ASHP feeding domestic radiators at 50 degrees. This shows a significant reduction in performance as outdoor temperatures drop. The resulting effect of this is that at lower temperatures the heat pump will in effect operate in steady state if it is not able to achieve the desired target room temperature.

10

Coefficient of Performance

6 5

Coefficient of Performance (Water at 50 degs) Interpolated and Extrapolated

4 3

Coefficient of Performance (50 degs)

2 1 0

-20

-10

0

10

20

30

Source Temperature (Degrees Celsius) Figure 1: COP versus temperature relationship of a commercial heat pump [3]

In summary, heat pumps perform best at higher source temperatures and are capable of producing heat energy up to 3.6 times that input as electrical energy, in normal conditions. At low source temperatures efficiency is much reduced. To avoid additional reduced efficiency caused by rapid cycling, cycle times should be over 6 minutes. Compared with gas central heating they provide a low carbon alternative except under cold weather conditions.

11

1.2.2 Photo-voltaic (PV) generation In parallel to the RHI described in the previous section, the government promotes renewable electricity generation through the Feed-in Tariff (FIT) scheme. The FIT promotes the following technologies [6]:     

solar photovoltaic (PV) panels wind turbines water turbines anaerobic digestion (biogas energy) micro combined heat and power (micro-CHP)

In the UK, solar PV accounted for 12% of renewable electricity capacity and 2.9% of renewable electricity generation in 2013 . Of the 2.41GW of solar PV capacity in the UK, 70% is at domestic level, supported under the Feed-in Tariff scheme [1]. This rapid growth is in part due to the generous tariff at the conception of the scheme, with the solar PV feed-in rate set at approximately three times the retail price, but also due to the rapidly falling costs of solar PV. Whilst there is no specific target for PV uptake the government is promoting PV to contribute a significant part of the government’s target of 15% of all energy coming from renewables in 2020 [7]. Photovoltaic (PV) generation involves the conversion of energy from solar irradiation into electrical energy. The fundamental unit of the PV generator is the photo-voltaic, or solar cell. Multiple PV cells are connected in series and parallel to form a PV module. Multiple modules are connected in series and parallel to create a PV array. The array of modules and a power conditioning unit (PCU) forms the PV generator, or PV system (Figure 2). PV generation is characterised by low efficiencies of energy conversion (solar energy to electrical energy). Efficiencies range from 6% to 18% or more across the range of solar cell technologies [8], given ideal conditions. The feed-in tariff for PV generation is 14.38p/kWh at the time of writing (June 2014).

12

Figure 2: Components of a domestic PV system [9](DTI, 2006)

1.2.3 Factors Affecting PV Output There are two major factors which affect the performance and output of PV generators: Solar Irradiance: Solar irradiance is described as the power (flux) from incident rays on a unit area of the solar cell, in W/m2. The output current of the solar cell is greatly affected by the available solar irradiance, with relatively minor changes in operating voltage. Irradiance itself is influenced by a number of factors, including location (latitude), time (season) of the year, and plane orientation. These factors affect irradiance such that solar cells perform better in latitudes approaching the equator (i.e. the tropics) and during the summer months. Another factor is the “air mass”, which is defined as the path length which sunlight takes through the atmosphere, normalised to the shortest possible distance (when the sun is overhead, air mass equals 1). Solar PV potential in Europe can be estimated from EU Joint Research Centre data, see the map in Figure 3: ‘The maps represent the yearly sum of global irradiation on horizontal and optimally inclined surfaces. Over most of the region, the data represent the average of the period 1998-2011, however, north of 58° N, the data represent the 10-years average of the period 1981-1990. All data values are given as kWh/m2. The same colour legend also represents potential solar electricity [kWh/kWp] generated by a 1 kWp system per year with photovoltaic modules mounted at an optimum inclination and assuming a system performance ratio of 0.75.’ [10]

13

Figure 3: Insolation in Europe

Here we can see that in London under ideal conditions we could expect around 1000kWh per year from a 1kW peak system, or around 25% of a typical households’ electricity consumption. Ideal conditions would involve using a tracking device to follow the sun. In practice solar panels typically have fixed mountings, pitched towards the sun. Figure 4 shows us that, in the UK, optimal output for a solar panel is achieved when it has a southerly orientation and is around 30 – 40 degrees from horizontal [11].

14

Figure 4 Effects of orientation upon solar panel efficiency

Solar cells typically lose performance with an increase in their surrounding and operating temperatures. The loss in power output is determined by the cell’s power temperature coefficient, and is evaluated under standard irradiance and at operating temperatures above 25°C (standard testing temperature). 1.2.4 Power Quality The modern electricity network must be capable of providing reliable and uninterrupted power supply to consumers. To achieve this, electricity supply to consumers must be of an acceptable voltage and frequency (i.e. voltage and frequency of supply must be within predetermined limits). Since heat pumps are motor-driven devices, and PVs are fundamentally DC devices which are interfaced to the AC distribution network, there is potential to generate harmonic content. This report focuses on voltage levels and demand profiles. For a detailed examination of harmonics aspects of heat-pumps and PV, see the companion ‘Impact of LV connected DER on power quality’.

15

2

Gaps in research

In this section we outline the gaps in knowledge regarding the impact of heat pumps and PV on the UK distribution network and introduce the modelling described later in the report.

2.1

Heat pumps

Operating at higher COPs of around 3 and above, heat pump technology becomes competitive with a gas boiler in terms of fuel costs with electricity prices in 2014 being a multiple of three of gas prices. In terms of carbon dioxide emissions, at a COP of 3, heat pumps have roughly the same emissions as a modern condensing gas boiler, with grid electricity giving 0.45 Kg/kWh compared to gas at 0.184Kg/kWh [12]. With the additional benefit of the FIT, in economic terms, heat pumps become an attractive alternative to the gas boiler and they may become a commonplace substitute for it. However there are a range of issues associated with heat-pumps that are distinct from those from gas heating systems. From the house-holders perspective, there is a limit to how much heat energy a given heat pump can produce. As the COP diminishes with external temperature, the heat output in turn diminishes; in a situation where the COP multiplied the maximum power input is less than the heat loss per second of the home, the desired temperature in the home will not be maintained. In this situation, there will be increased coincidence of heat-pump loads as thermostat cycling ceases to operate. From the distribution networks perspective, with a demand typically of several kilowatts they represent a significant additional electrical load in the home. Because the ambient temperature and operating regime affect the demand profile of a heat pump so directly it is not possible to produce a single representative profile of a heat pump ‘after diversity’, as we might with other residential loads. In practice the impact of the heat-pumps will vary upon operating conditions and these will in turn influence the impact upon the network. Since we cannot install heat-pumps in a test network without considerable expense and customer opt-in, a modelling approach is an appropriate means to explore network impacts. By using models we are also able to control the environmental variables that affect heat pump demand. A selected network is simulated with different levels of heat pump penetration and under different weather scenarios.

2.2

Photo-voltaic

The growing presence of residential solar PV generation requires quantitative study of their impacts on the distribution network. As identified earlier, high penetrations of PV generation in times of high output may lead to significant over voltage. Voltage limits are in place in order to ensure that all customers’ appliances can operate as intended, with a supply that they have been designed to expect.

16

Considering the national penetration of PV as discussed earlier affords some understanding of low carbon supply at the national level, but in practice, from the DNOs perspective it is likely to be clustering of homes with PV panels that may cause localised issues. In order to understand the impacts of PV’s we also need to understand the loads coincident with PV generation. To this end, a summer demand baseline is required, and this is described in detail in the Appendix. In the following section we examine the trends seen from LCL measured heat pumps and PV’s. Building upon these data, the following section then develops a configurable model for heat pumps upon which different technology adoption scenarios are examined. A similar approach is applied to PV’s. It then reports likely impact of varying heat pump and PV penetration levels upon the selected section of UK Power Networks’ LV network.

17

3

LCL PV and heat pump field trial data

3.1

Heat pumps

Working with EDF Energy, the Low Carbon London project recruited two existing EDF Energy customers with heat pumps onto a year-long monitoring programme. Each heat pump’s power demand (kW) was monitored at 10 minute resolution. Note that a shorter heat pump trial was conducted to ascertain harmonic power flow associated with heat-pumps. For this analysis see the companion ‘Impact of LV connected DER on power quality’. The two sites examined here present very different data, indicating very different building types and heat pump specifications. The following sub-sections characterize these sites, both in terms of annual energy demand profiles and diurnal demand profiles. The temperature data presented is from sensors at London City Airport [13]. The profile of the coldest day from 2014 can be seen in Figure 5.

Temperature [Degrees Celsius]

12

12/01/2014 10 8 6 4 2

0 1

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Hour

Figure 5, Coldest day during heat pump monitoring period 2013/2014: Sunday 12/01/2014

18

30 25 20 15 10 5

Degrees Celsius

45 40 35 30 25 20 15 10 5 0

Average Temperature

26/01/2014

07/12/2013

Date

18/10/2013

29/08/2013

10/07/2013

21/05/2013

0 01/04/2013

kWh

3.1.1 Site 1 (MPAN 7000009339931) Site 1 represents an air-source heat pump installed in a 2-bedroom family home.

Daily energy

Figure 6: Daily energy versus temperature for 12 month period.

7 represents daily energy use and temperature for the 12 month monitoring period for site 1. We see that during the Winter months daily heat energy demand can exceed 35kWh and in part of July the heat pump appears to be switched off completely. 3.5 3

kW

2.5 2 1.5 1 0.5 00:00 01:00 02:00 03:00 04:00 05:00 06:00 07:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00

0

Time Figure 7: Consumption profile for Sunday 12/01/2014

This site has a 24 hour heating regime, and Figure 7 demonstrates this for the selected ‘coldest day’. The cold am period, at around 1 degree C is characterised by wider periods of load compared to the warmer pm period which was over 6 degrees C. This site demonstrates the heat pump behaviour described earlier. At 1 degrees outside temperature the heat pump is able to cope with the home’s heat demand, since it does not run continuously, but there is little additional capacity in colder conditions.

19

Average Max Min

00:00 01:00 02:00 03:00 04:00 05:00 06:00 07:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00

kW

5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0

Time Figure 8: Weekdays winter 10-minutes average data

3.5 3

kW

2.5 2

Average

1.5

Max

1

Min

0.5 00:00 01:00 02:00 03:00 04:00 05:00 06:00 07:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00

0

Time Figure 9: Weekend winter 10-minutes average data

Figure 8 and Figure 9 demonstrate that this site has a heating regime that is relatively constant throughout the week with weekday and weekend minima, maxima and averages being very similar. The heat pump appears to be able to step its maximum demand up and down between around 2.5kW and 3.0kW since the maximum value average is not constant for all periods.

20

70

30

60

25

kWh

50

20

40

15

30

10

20

Average Temperature

26/01/2014

07/12/2013

Date

18/10/2013

29/08/2013

0 10/07/2013

0 21/05/2013

5 01/04/2013

10

Degrees Celsius

3.1.2 Site 2 (MPAN 7000038631940) Site 2 represents an air-source heat pump installed in a 4-bedroom family home.

Daily energy

Figure 10: Daily energy versus temperature for a period of a year

Site 2 is similar to Site 1 in that the energy demand is highly dependant on outdoor temperature. Here we see a maximum energy demand of over 50kWh per day. Our model ‘coldest day’ in Figure 11 shows, however, that unlike Site 1 the system only provides heat during the day, albeit well in to the evening. Heating starts around 6am and ends around 10pm. The system may have more than one temperature setting, on a timer, and this may account for the demand spike in the morning of the cold day, alternatively this could be a dual purpose heat pump also heating the hot water supply.

4.5 4

2.5 2 1.5 1 0.5 0 00:00 01:00 02:00 03:00 04:00 05:00 06:00 07:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00

kW

3.5 3

Time Figure 11, Consumption profile for Sunday 12/01/2014

21

4.5 4 3.5

kW

3 2.5

Average

2 1.5

Max

1

Min

0.5 00:00 01:00 02:00 03:00 04:00 05:00 06:00 07:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00

0

Time Figure 12: Weekdays winter 10-minutes average data

4.5 4 3.5

kW

3 2.5

Average

2

Max

1.5

Min

1 0.5 00:00 01:00 02:00 03:00 04:00 05:00 06:00 07:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00

0

Time Figure 13: Weekend winter 10-minutes average data

Figure 12 and Figure 13 demonstrate that this site has a heating regime that is relatively constant throughout the week with weekday and weekend minima, maxima and averages being very similar. Because this system does not maintain temperature during the night there is a ‘recovery load’ in the morning period when the property temperature is shifted to the daytime target temperature. This can be seen at around 6 am when the heat pump minimum power approaches the average. Heat pump cycling behaviour only commences at around 8.30 am and we can see that as the external temperature rises the ‘duty cycle’ represented by the average trend decreases. The heat pump appears to have an on/off control characteristic due to the maximum load being the same throughout the day. Again the evening ‘needle peaks’ of maximum load may be due to a minimum temperature set point occasionally causing the heat pump to turn on.

22

3.2

Solar photo-voltaics 35

900

30

800

25

600

20

Degree Celsius

700 500 400

15

300

10

200

5

100 0

0 06-01-00 06-02-18 06-04-12 06-06-06 06-08-00 06-09-18 06-11-12 06-13-06 06-15-00 06-16-18 06-18-12 06-20-06 06-22-00 06-23-18 06-25-12 06-27-06 06-29-00 06-30-18 07-02-12 07-04-06 07-06-00 07-07-18 07-09-12 07-11-06 07-13-00 07-14-18 07-16-12 07-18-06 07-20-00 07-21-18 07-23-12 07-25-06 07-27-00 07-28-18 07-30-12

WattHours per square meter

1000

Time DnSol_WsqM

DiffHorz_WsqM

DirNormIr_WsqM

Temperature

2

Figure 14: Temperature and Solar Radiation for June-July 2013 (Copyright© 2013 by Weather Analytics™ LLC)

Figure 14 shows the temperature and solar radiation for July and June 2013. DnSol_WsqM is the downward solar radiation (Global Horizontal Irradiation) (WattHours per square meter). The downward solar radiation is the principal source of energy for PV panels. Also shown are two other parameters: DiffHorz_WsqM is the diffuse horizontal radiation at surface (WattHours per square meter) and DirNormIr_WsqM is the direct normal irradiation at surface (WattHours per square meter). The graph indicates that we can expect significant variations in PV performance day to day due mainly to variations in cloud cover. It should be noted that, for the purposes of this report, we are concerned with the maximum impact on the distribution network, so a cloudless day was assumed in the modelling of section 5. For Low Carbon London a number of PV systems located within the UK Power Networks license area were monitored for over a year. Figure 15 shows the daily energy production of one of the LCL PV systems. Here the energy produced can indeed be seen to follow the downward solar radiation. Annual variations are large, with a peak production of 4.5kWh per day in June and a minimum of about 0.5kWh per day in January. Figure 16 shows the daily active power profiles from the same unit over the two most productive months: June and July. The plot shows a profile for an average of the two months along with profiles for the minimum and maximum output. The variation between minimum and maximum is wide here. As noted above, this is due to the great difference between cloudless and cloudy days. With heavy cloud cover the peak power produced can be as low as 0.5kW, equivalent to a mid-winter day; with a cloudless sky a peak value of 4.5 kW is reached.

2

Temperature location London City Airport; Source: Copyright© 2013 by Weather Analytics™ LLC

23

70 50 40 30 20

kWh per Day

60

10

Total DnSol_kWsqM

02/03/2014

11/01/2014

Time

22/11/2013

03/10/2013

14/08/2013

25/06/2013

0 06/05/2013

Total Dayly Radition [kWh/m2]

9 8 7 6 5 4 3 2 1 0

PV 7050132397370

Figure 15: Daily energy vs Solar radiation for a year period for PV 7050132397370

5 4

kW

3 2

1

-1

00:00 01:00 02:00 03:00 04:00 05:00 06:00 07:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00

0

Average

Time Max

Min

Figure 16: Average active power for PV 7050132397370

Table 4 shows the heights of the three peaks in Figure 16 for nine solar PV systems in the LCL trials. The picture is one of great consistency. Panels with the same capacity (A, B, C and D, for example) produce almost identical values. The spread between minimum and maximum is also quite similar.

24

PV

Peak power output of minimum

Peak power output of average

Peak power output of maximum

A: 7050570000033

0.1kW

0.6kW

1.2kW

B: 7050470000033

0.1kW

0.7kW

1.2kW

C: 7050170000033

0.1kW

0.6kW

1.1kW

D: 7050370000033

0.1kW

0.6kW

1.1kW

E: 7050532397370

1kW

4.3kW

7.9kW

F: 7050432397370

1kW

4.3kW

7.9kW

G: 7050332397370

1.1kW

4.2kW

7.7kW

H: 7050232397370

1.1kW

4.5kW

8kW

I:7050132397370

0.5kW

2.5kW

4.5kW

Table 4: Daily minimum, average and maximum peak power production for June-July 2013

25

4

Technology scenarios

4.1

LV Network choice

In order to understand the effects of heat pumps on an LV network there are a number of parameters that need to be established:  

  

Weather data is required to provide a realistic outside temperatures. Target temperatures and durations are required for the residence to be modelled, and in particular the customers’ expectations (round-the-clock warmth or early morning / late evening heating only). Typical home insulation values. Heat pump power (i.e. its electrical rating). Heat pump performance (i.e. its Coefficient of Performance (COP)).

Firstly for the weather data, a single day temperature profile was modified to produce 4 separate temperature profiles. Figure 17 shows the four variants of the profile with 3 being the original. The top temperature profile reflects the last 10 year’s winter average temperatures at 6.7 degrees. The daily profile has been taken from []. Trend 3 averages at approximately 0 degrees and trend 1 averages at approximately -4 degrees, are to provide more demanding weather conditions.

10

6 4

Temperature profile 1

2

Temperature profile 2 Temperature profile 3

0 -2 -4 -6

04:00:00 04:01:40 04:03:20 04:05:00 04:06:40 04:08:20 04:10:00 04:11:40 04:13:20 04:15:00 04:16:40 04:18:20 04:20:00 04:21:40 04:23:20

degrees centigrade

8

Temperature profile 4

Time

Figure 17: outdoor temperature scenarios.

DECC also report average UK internal temperatures, over the last 10 years the average is 17.9 C. In the simulation houses are set to have a value of internal temperature of between 16.9 C and 18.9 C. Whilst in practice some homes will use timers to modulate the internal temperature, for this specific exercise this was deemed problematic to model in the absence of data about heating timer usage. As such, the report will concentrate on a model closer to 26

Site 1 shown earlier than to Site 2. It may under-estimate the effect of heat pumps if a significant proportion of homes use them similar to Site 2, in which they need to recover the temperature in the home back to comfort levels early each morning. Home energy loss values are estimated by DECC in terms of watts per centigrade of temperature differential (DECC Energy fact file 2103). To model energy demand at the national level then a single figure could be used. With a model with individually modelled homes we require some heterogeneity to avoid homes exhibiting the same behaviour. To this end, a heat loss parameter is randomised to be between 250 and 350 W/c.

4.2

Heat-pump modelling

As described earlier heat pumps have varying efficiency depending on source and output temperature differentials. In order that we understand system performance under different conditions then we need to model the heat pump COP. The red line in Figure 18 shows the estimated COP at different source temperatures (after Element Energy).

Figure 18: COP in relation to source temperature

The air source temperature will of course never naturally be at 50 degrees, but the differential between source and output can fall to zero if the outside temperature is at the same temperature as the water in the radiators. In the simulations to follow, a universal heat pump is modelled with a COP calculated to be 8.0 minus 0.1 for every degree of source output temperature differential. Given these input parameters, the simulation is run with different levels of penetration on a number of demonstration networks. Figure 19 and Figure 20 shows heat pump load profiles derived from this model for two temperature scenarios. As expected on periods are shorter than off periods at the higher temperature of 7⁰C, whereas on periods dominate at 0⁰C3.

3

This is in line with DECC study “Understanding the Balancing Challenge”, 2012

27

4 3.5 3 kW

2.5 2 1.5 kW2

1 0.5 04:00:00 05:00:00 06:00:00 07:00:00 08:00:00 09:00:00 10:00:00 11:00:00 12:00:00 13:00:00 14:00:00 15:00:00 16:00:00 17:00:00 18:00:00 19:00:00 20:00:00 21:00:00 22:00:00 23:00:00 24:00:00 25:00:00 26:00:00 27:00:00

0

Time

Figure 19: Modelled heat pump profile for typical winter day (7 degree Centigrade)

4 3.5 3 kW

2.5 2 1.5 1 0.5 04:00:00 04:49:40 05:39:20 06:29:00 07:18:40 08:08:20 08:58:00 09:47:40 10:37:20 11:27:00 12:16:40 13:06:20 13:56:00 14:45:40 15:35:20 16:25:00 17:14:40 18:04:20 18:54:00 19:43:40 20:33:20 21:23:00 22:12:40 23:02:20 23:52:00 24:41:40 25:31:20 26:21:00 27:10:40

0

Time

Figure 20: Modelled heat pump at 0 degrees Centigrade

In conclusion, we take an existing verified model of existing domestic electricity consumption without either heat pumps or PV generation. We add to it a heat pump with a electrical rating and heat output similar to the larger of the two monitored sites within the Low Carbon London project, and with a heating demand similar to the more benign of the two sites, by aiming to keep a constant temperature throughout the day. The details of the underlying model are in the Appendix. The model takes into account reductions in efficiency as the temperature drops. Individual heat pumps in the model are set to commence their first cycle at randomised times, to reflect one heat pump will not be operating on the same cycle as their neighbour’s, even if both have the same temperature set point.

28

4.3

PV modelling

Solar panels will differ in their orientation and hence the timing of their power output and this will in a sense provide some diversity to the PV generation for a given feeder. Whilst this is perhaps worth considering in more detail, there is also the natural effects such as clouds which on typical British days will contribute both a reduction of output and peaks and troughs in output. However given that DNO planning assumptions necessarily take into account worst-case scenarios, for the purposes of this report the output profile for all panels exhibit an ‘ideal’ output profile for a cloudless day. In practice the model uses a scaled up version of that seen in Figure 21 which is a real panel output profile for a panel in June. In the model this is scaled up to produce a typical 3.6kW maximum system output. This was set to be in the mid-range of the panels that we monitored by the Low Carbon London project. 1200 1000

Watts

800 600 400 200

-200

04:00:00 05:00:00 06:00:00 07:00:00 08:00:00 09:00:00 10:00:00 11:00:00 12:00:00 13:00:00 14:00:00 15:00:00 16:00:00 17:00:00 18:00:00 19:00:00 20:00:00 21:00:00 22:00:00 23:00:00 00:00:00 01:00:00 02:00:00 03:00:00 04:00:00

0

Time

Figure 21: Actual PV profile on clear sunny June day

In conclusion, we take the existing model of domestic load and modify this to reflect summer minimum (i.e. weekday) demand; and then add to this a PV panel which is similar to the mid-range of the panels that were monitored, and exclude cloud cover. This is intended to represent a ‘worst case’ of the current installed fleet which actually consists of both larger and smaller panels than the one modelled.

29

5

Analysis

5.1

Network model

edse002fnn edse002fo8 edse002foa edse002foc edse002fnp edse002foe edse002fnt edse002fns edse002f9q edse002fnx

edse002fnv

edse002fnz

edse002fny edse002f9u

edse002fa1 edse002foq

Merton LV 06601

edse002fc8 rdbd00310s rdbd005ta6 edse002f2c edse002f2v edse002fnm edse002f2s edse002f2m edse002f2p

rdbd002v1o edse002fp7

edse002fng edse002fnk edse008a7q edse002fcg

edse002fpf edse002fpb

edse002fp2 edse002fpe

edse002few

edse002fph

edse002fo3 edse002fco edse002ftc edse002ff1 edse002ff5 edse002fo7

edse002feh

edse002fe1

Figure 22: LV network under Merton 6601

The simulation of the LV network under Merton substation (UK Power Networks reference 6601) (Figure 22) provides a means to measure the average effect of specific interventions as well as providing an example of effects upon power flow on a London LV network. Similar modelling of an adjacent substation in Merton is described in detail in a companion report which looks in particular at the effects of energy efficient appliances. Modelling of this network was carried out for the following sections considering the impact of heat pump and PV uptake on the network. This network was chosen to illustrate affects of heat pump uptake, because the substation supplies a very high percentage of domestic customers (328/339, 97%). The capacity (500kVA) and number of customers (339) are also typical for a substation in the LPN area.

5.2

Heat Pumps

In modelling the impact of heat pump uptake on the distribution network, the heat pump model introduced in section 4.1.1 was used to represent the typical domestic heat pump. The power of this heat pump was derived from the heat pumps in the LCL trials (sections 3.1.1 and 3.1.2). Four penetration levels (5%, 10%, 15% and 20%) were considered for the four different weather scenarios shown in Figure 17. These scenarios are: scenario 1, a 30

weather scenario with an average temperature of -4°C; scenario 2, a weather scenario with an average temperature of 0°C; scenario 3, a weather scenario with an average temperature of 4°C and scenario 4, a weather scenario with average temperature of 7°C. For this model, current day levels of home insulation were assumed, in line with home energy loss values estimated by DECC (DECC Energy fact file 2103). Power flow modelling was carried out on the network served by Merton 6601 to calculate the transformer loading for the four temperature scenarios. Table 5: For four temperature scenarios, the percentage increase in peak loading over the baseline peak value for the Merton 6601 transformer for different heat pump penetration levels

Percentage peak daily load increase over baseline at: 5% 10% 15% 20% penetration penetration penetration penetration Scenario 1 (-4°C average) Scenario 2 (0°C average) Scenario 3 (4°C average) Scenario 4 (7°C average)

Load (kW)

5.2.1

19%

33%

48%

72%

9%

18%

32%

39%

9%

14%

18%

25%

5%

7%

11%

15%

Merton secondary 06601: weather scenario 1 (average temperature -4°C) 500 450 400 350 300 250 200 150 100 50 0 4

8

12

16

20

24

28

Time (h) Baseline

HP 5

HP 10

HP 15

HP 20

Figure 23: Transformer loading under weather scenario 1 conditions (av. temperature, -4°C)

Figure 23 shows Merton secondary (06601) transformer loadings for four different penetration levels: 5%, 10%, 15% and 20%, for weather scenario 1 in which temperature profile 1 from Figure 17 is used (this has an average temperature of -4°C ). In terms of planning, this is the weather scenario that is of most relevance. With a capacity of 500kVA and 315 consumers, Merton could just support the highest penetration of 20% of homes 31

using heat pumps for space heating. In Table 5 figures show the percentage increase in peak loading over the baseline peak value for the different weather scenarios and penetration levels. This presence of heat pumps with a 20% penetration level increases the peak loading over the baseline value by approximately 72% from 270kW to 465kW. This large increase is caused by two factors: the reduced efficiency of heat pumps at low temperatures (see Figure 18) and reductions in diversity as nearly all heat pumps will be running at full capacity in such cold conditions. The input is constraint throughout the day given that we have simulated a preference to keep temperature constant in the home throughout the day. Considering the lowest penetration level of 5%, the peak load is increased from the baseline level of 270kW to 320kW for the cold weather scenario, a 19% increase, which is still significant. Here the effects of reduced heat pump efficiency and diversity will be unchanged, so the much smaller increase is just due to the reduced number of heat pumps.

Load (kW)

5.2.2

Merton secondary 06601: weather scenario 2 (average temperature 0°C) 500 450 400 350 300 250 200 150 100 50 0 4

8

12

16

20

24

28

Time (h) Baseline

HP 5

HP 10

HP 15

HP 20

Figure 24: Transformer loading under weather scenario 2 conditions (av. temperature 0°C)

Figure 24 shows transformer loadings for scenario 2 (temperature profile 2 from Figure 17), corresponding to those for scenario 1 above. Table 3 shows that at 20% penetration daily peak transformer loading is increased by 39% over the baseline value. This compares with a 72% increase for the same penetration level in scenario 1. The difference is due to improved heat pump efficiency at the higher temperature and increased diversity as fewer heat pumps need to work at full capacity in these conditions. At different penetration levels results show the same pattern as before: reductions in the peak level are caused only by the reduced number of heat pumps.

32

Load (kW)

5.2.3

Merton secondary 06601: weather scenario 3 (average temperature 4°C) 500 450 400 350 300 250 200 150 100 50 0 4

8

12

16

20

24

28

Time (h) Baseline

HP 5

HP 10

HP 15

HP 20

Figure 25: Transformer loading under the weather scenario 3 (average temperature 4°C)

Figure 25 shows transformer loadings for the weather scenario 3 (temperature profile 3 from Figure 17, which has average temperature -4°C). Table 5 shows that at 20% penetration daily peak transformer loading is increased by 25% over the baseline value. This follows the progression seen in the previous weather scenarios: as temperatures increase heat pump efficiency improves and diversity increases.

5.2.4

Merton secondary 06601: weather scenario 4 (av. temperature 7°C)

400 350 Load (kW)

300 250 200 150 100 50 0 4

8

12

16

20

24

28

Time (h) Baseline

HP 5

HP 10

HP 15

HP 20

Figure 26: Transformer loading under weather scenario 4 (av. temperature 7°C)

Figure 26 shows transformer loadings for the weather scenario 4 (temperature profile 4 from Figure 17, which has average temperature 7°C). Table 5 shows that at 20% penetration 33

daily peak transformer loading is increased by 15% over the baseline value. Again, this follows the progression seen in the previous weather scenarios: as temperatures increase heat pump efficiency improves and diversity increases. Clearly one would not plan a network containing heat pumps based on a minimum temperature of 7°C, but this particular example illustrates the effects of diversity and heat pump efficiency. For 20% penetration the increase in peak load at an average outdoor temperature of -4°C is 72%: at 7°C it has fallen to a fifth of this. 5.2.5 Summary For an average temperature of -4⁰C and a penetration level of 20% of household owning heat pumps, the peak daily load increases by 72% above baseline, almost reaching the capacity of the substation transformer. As most heat pumps in the network will be working at full capacity in weather this cold, diversity will be greatly reduced. In addition heat pumps work at reduced efficiency in low temperature conditions. These two factors are responsible for the large increase in peak load. In contrast, at 7⁰C the corresponding peak load increase is just 15%. Here diversity is increased and heat pump efficiency improved. Reducing penetration levels from 20% (to 15%, 10% and 5%), show approximately proportional reductions in peak loading increase.

34

5.3

Solar PV

The LV network under Merton substation 6601, described in section 5.1, was also used in modelling the impact of uptake of PV on the distribution network. As for heat pumps, the high level of domestic consumers connected to the network made it a good choice to represent a part of the distribution network likely to be most affected by the increased presence of PV. For modelling purposes a cloudless day is assumed in all cases. 5.3.1 Merton secondary 06601: 10% Solar PV penetration Since we know that residential demand is on average around 300 – 400W in the middle of the day (based on data gathered from Smart Meters within the Low Carbon London project), given a standard 3.6kW PV system, we might expect 10% solar penetration to reduce demand to near zero during this period. At 10% penetration this does occur in the model, (see Figure 27 and Figure 28), we see a small amount of reverse power flow from late morning to late afternoon. 250 200

kW total

150

100

kWTotal kVARTotal

50

-50

04:00:00 05:00:00 06:00:00 07:00:00 08:00:00 09:00:00 10:00:00 11:00:00 12:00:00 13:00:00 14:00:00 15:00:00 16:00:00 17:00:00 18:00:00 19:00:00 20:00:00 21:00:00 22:00:00 23:00:00 24:00:00 25:00:00 26:00:00 27:00:00

0

Time

Figure 27: Merton 06601 Love Lane substation power flow on clear sunny day in June.

35

35 30 25 kW per phase

20 15

kW1

10

kW2

5

kW3

-5 -10 -15

11:30:00 11:33:00 11:36:00 11:39:00 11:42:00 11:45:00 11:48:00 11:51:00 11:54:00 11:57:00 12:00:00 12:03:00 12:06:00 12:09:00 12:12:00 12:15:00 12:18:00 12:21:00 12:24:00 12:27:00

0

Time

Figure 28: Zoomed view of midday showing the three phases

258 256 254 252 250 248 246 244 242 240 238

Vmin Vmax

04:00:00 05:00:00 06:00:00 07:00:00 08:00:00 09:00:00 10:00:00 11:00:00 12:00:00 13:00:00 14:00:00 15:00:00 16:00:00 17:00:00 18:00:00 19:00:00 20:00:00 21:00:00 22:00:00 23:00:00 24:00:00 25:00:00 26:00:00 27:00:00

Volts

These charts confirm that the model captures the imbalances across different phases, which will be relevant for assessing the impact on voltage. Figure 29 shows the voltage and minima and maxima throughout the day. The maximum is the maximum voltage seen by any of the 339 customers on the network in that timeslot; and the mimum is the minimum voltage seen by any of the customers in that timeslot.

Time

Figure 29: Voltage minima and maxima on network

This demonstrates that even with minimal back-feeding voltage rise is likely to occur. This warranted investigation into even lower levels of penetration, as presented in the next section.

36

5.3.2

Merton secondary 06601: 5% Solar PV penetration

At 5% penetration the demand curve for the substation remains positive throughout the day. See Figure 30.

250

kW total

200 150 100

kWTotal 50

04:00:00 05:00:00 06:00:00 07:00:00 08:00:00 09:00:00 10:00:00 11:00:00 12:00:00 13:00:00 14:00:00 15:00:00 16:00:00 17:00:00 18:00:00 19:00:00 20:00:00 21:00:00 22:00:00 23:00:00 24:00:00 25:00:00 26:00:00 27:00:00

0

Time Figure 30: Power flow on clear sunny day in June

Perhaps surprisingly, this situation still gives rise to over-voltage albeit by 2 volts, see Figure 31. 256 254 252 248 246 244

Vmin

242

Vmax

240 238 04:00:00 05:00:00 06:00:00 07:00:00 08:00:00 09:00:00 10:00:00 11:00:00 12:00:00 13:00:00 14:00:00 15:00:00 16:00:00 17:00:00 18:00:00 19:00:00 20:00:00 21:00:00 22:00:00 23:00:00 24:00:00 25:00:00 26:00:00 27:00:00

Volts

250

Time

Figure 31: Voltage minima and maxima

37

Due to phase imbalance we observe that the second phase is closer to zero and is the likely source of the voltage rise, as demonstrated in Figure 32. 120 100

60 kW1 40

kW2 kW3

20 0 04:00:00 05:00:00 06:00:00 07:00:00 08:00:00 09:00:00 10:00:00 11:00:00 12:00:00 13:00:00 14:00:00 15:00:00 16:00:00 17:00:00 18:00:00 19:00:00 20:00:00 21:00:00 22:00:00 23:00:00 24:00:00 25:00:00 26:00:00 27:00:00

kW

80

Time

Figure 32: Imbalance between phases

38

5.3.3

Merton secondary 06601: 25% Solar PV penetration

Since over voltage appears to occur so readily the PV penetration is now raised to 25% to investigate the effect of larger concentrations. See Figure 33.

Figure 33: Power flow at 25% penetration

Voltage rise for this scenario can be seen in Figure 34. Here we see voltage events over 257 volts.

Figure 34: Voltage rise at 25% penetration

39

This is approaching the G83 PV cut-out limit of 262volts, thus finally we increase penetration to 30 to establish if we can identify the likely penetration to cause cut-out losses. 5.3.4 Merton secondary 06601: 30% Solar PV penetration Figure 35 shows our final scenario with significant back-feeding, a profile that is approaching zero net energy demand. This causes voltages to exceed the G83 limit, see Figure 36.

Figure 35: Power flow at 30% penetration

Figure 36: Voltage maxima with 30% penetration

This situation would not arise if the inverter cut-out is activated. In practice however the inverters would cut out until the local voltage rises above 262.

40

5.3.5

Summary

Table 6: Summary of findings from PV modelling

Over-voltage Reverse power flow level

5% 2V None

Penetration level: 10% 25% 3V 4.3V

30% 14V

Low

High

High

Table 6 summarises the findings from PV modelling. Higher penetration levels cause significant levels of back-feeding at the substation. At 30% penetration the result is almost zero net energy demand. Over-voltage is then a clear problem. Even a small penetration of PV’s is found to produce an over-voltage of 2 Volts. Whilst uptake at higher levels is not currently foreseen, it is interesting to examine the nature of the relationship between penetration and instances of over-voltage at higher uptake levels. At 25% penetration the over-voltage more than doubles to 4.3V, but from here onwards it rises steeply to 14V at a 30% penetration level. In practice, inverter cut-outs should prevent over-voltages of more than 9 Volts.

41

6

Conclusion and Recommendations

6.1

Main findings

The two technologies examined in this report, heat pumps and solar panels pose very different and potentially contradictory problems for the distribution network. For heat-pumps, on the modelled network, the primary concern was transformer capacity. The 6601 transformer operates 270kW, well below its capacity of 500kW. This supported 20% heat-pump penetration with no under-voltage. However, the transformer was close to capacity at this level of penetration for the cold weather scenario. From these early indications it would appear that considerable additional transformer headroom is required to account for cold conditions in which diversity is much reduced and COP collapses, if penetration levels above 20% are eventually reached. The modelling results are based on current standards of home insulation. These issues would not arise to the same extent in a population of well insulated homes and this will be an important area for further work. The issues surrounding Solar PV are on the other-hand voltage related. Voltage rise happens well before the onset of back feeding to the high voltage network. A modest 5% population of PVs can cause over-voltage albeit by only a few volts. However, as PV penetration increases, then voltage maxima increase considerably. For the network as modelled the threshold at which G83 devices trip out (262V) due to over-voltage lies somewhere between a PV penetration of 25% and 30%.

6.2

Recommendations

Given the significant effect that heat pumps can have upon an individual home and network demand profiles it is clear that DNOs need to keep abreast of how this market develops and any clustering of installations which occurs. If they become common place then a great deal of DNO assets may need to be renewed. The recording of heat pump installations is now proposed by the Electricity Network Association and this will afford some visibility of likely effects [14]. The most obvious change to mitigate the over-voltage effect of PVs would be to decrease secondary substation output voltages. On the network simulated, even under high heat pump penetrations low voltage did not become a problem. In these networks the substation voltage could be reduced to 240 providing over 10 volts of headroom to PVs before overvoltage.

6.3

Further work

To further this work the following extensions to the modelling are proposed: sensitivity analysis of PV over-voltage to cloudy insolation; modelling the impact of Heat Pumps on the network with different levels of insulation; neutral voltage rise modelling and individual feeder analyses.

42

7

References

[1] “Increasing the use of low-carbon technologies: Renewable Heat Incentive (RHI),” DECC, [Online]. Available: https://www.gov.uk/government/policies/increasing-the-use-oflow-carbon-technologies/supporting-pages/renewable-heat-incentive-rhi. [2] G. Boyle, Renewable Energy, Milton Keynes: The Open University, 2004. [3] “EST Heat Pump briefing for UKPN.,” 2013. [4] E. T. Robert Green, The Effects of Cycling on Heat Pump Performance, Prepared for: Department of Energy and Climate Change (DECC). [5] “/how-to-apply,” 2014. [Online]. Available: https://www.gov.uk/feed-in-tariffs/how-toapply . [6] DECC, “UK Solar PV Strategy: Road map to a brighter future.,” 2013. [7] Evo Energy, “Solar PV cell comparison,” [Online]. Available: http://www.evoenergy.co.uk/solar-pv/our-technology/pv-cell-comparison/. [8] DTI, Photovoltaics in Buildings Guide to the installation of PV systems, DTI, 2006. [9] EC. [Online]. Available: http://re.jrc.ec.europa.eu/pvgis/cmaps/eur.htm. [10] U. S. Energy. [Online]. Available: http://www.uksolarenergy.org.uk/installing-solarpanels.html. [11] Carbon Trust, “Conversion Factors,” 2013. [12] W. Analytics, Weather data, 2014. [13] Energy Networks Association, “Distributed Generation Connection Guide. 3.3rd edition,” Energy Networks Association, 2011.

43

8

Appendix

8.1

Emulation of the summer power flow on a feeder

While with the heat pump analysis, we can re-use the base-lining used in the energy efficiency partner report, to analyse PV effects we need to emulate summer power flow on a feeder. This emulation is based on the modelling described in sections 8.2 and 8.3.

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

Wtr Wd Smr Wd

04:30 05:30 06:30 07:30 08:30 09:30 10:30 11:30 12:30 13:30 14:30 15:30 16:30 17:30 18:30 19:30 20:30 21:30 22:30 23:30 00:30 01:30 02:30 03:30

kW

Electricity demand is on average lower in summer, shows a comparison of the Elexon Profile class 1 trends for winter and summer weekdays.

Time Figure 37: Summer versus winter weekday profile class one demand curve

For the analysis of disaggregated power flow on a feeder, as described in the energy efficiency report, we need to re-construct profile class one from realistic loads in individual homes. Again we can use the HEUS data as guide; Figure 38 shows us the profile for HEUS, as with winter this is higher than the national average. Figure 39 and Figure 40 show profile class one divided into load categories with unknown and without unknown.

44

700 Showers

kW

600

Unknown

500

Other

400

Heating Water heating

300

Washing/drying/dishwasher

200

ICT

100

Audiovisual Lighting

03:00

01:30

00:00

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

0

Time

Cooking Cold Appliances

Figure 38: HEUS summer weekday profile

0.6 Showers

0.5

Unknown Other

0.4 kW

Heating

0.3

Water heating Washing/drying/dishwasher

0.2

ICT

0.1

Audiovisual Lighting

03:00

01:30

00:00

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

0.0

Cooking Cold Appliances

Time Figure 39: Profile class one with HEUS appliance mix

45

0.6 Showers

0.5

Other

kW

0.4

Heating Water heating

0.3

Washing/drying/dishwasher

0.2

ICT

0.1

Audiovisual Lighting

03:00

01:30

00:00

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

0.0

Cooking Cold Appliances

Time Figure 40: Profile class one with HEUS appliance break down (unknown removed)

8.2

Heuristic rule development

As a point of departure to start modelling the summer profile, the same narrative rules and the same ownership statistics as the winter model were applied to modelling framework. The only key inputs to change were:   

The use of summer weekday TUS data describing human activity. The weather data was changed to a bright sunny June day. The narrative rules were modified to eliminate all ‘mood lighting’.

Whilst the first two points are perhaps obvious, the third was based on the assumption that ‘mood lighting’ is more prevalent in winter. Again using the heuristic findings of the energy efficiency report the best fit between the models is with the rescaled HEUS data for lighting and cold appliances. On the first execution of the summer model there were obvious issues with the lighting and wet appliance profiles, see Figure 41 and Figure 42.

46

light_HEUS+ light_mod 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 00:00 01:30 03:00

kW

0.30 0.25 0.20 0.15 0.10 0.05 0.00

Time

Figure 41: Lighting under simulation first cut using winter heuristics

0.15

kW

0.10 wet_HEUS

0.05

wet_mod 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 00:00 01:30 03:00

0.00

Time

Figure 42: Wet appliances under simulation first cut using winter heuristics

The new lighting curve is not ideal, since it contains a base-load of more than 20 Watts above the HEUS model, however a similar base-load was missing from final model fit so for simplicity this was per Figure 41. The model wet appliance demand profile was approaching double the HEUS data. The corrected profile, see Figure 44, uses the same rules but with all tumble dryer usage removed. Whilst this is probably an over simplification of the real world, in that even in summer some tumble dryer usage would be present, it does bring the demand much closer to HEUS (see Figure 40). The morning peak is larger but this may be more representative than the small HEUS sample.

47

0.12 0.10 kW

0.08

0.06

light_HEUS

0.04

kWPerHome

0.02 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 00:00 01:30 03:00

0.00

Time

0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00

wet_HEUS

03:00

01:30

00:00

22:30

21:00

19:30

18:00

16:30

15:00

13:30

12:00

10:30

09:00

07:30

06:00

wet_mod

04:30

kW

Figure 43: Second cut lighting profile

Time Figure 44: Wet appliances with tumble dryers removed

0.2 0.1 0.1 0.1 0.1 0.1 0.0 0.0 0.0

cold_HEUS+

03:00

01:30

00:00

22:30

21:00

19:30

18:00

16:30

15:00

13:30

12:00

10:30

09:00

07:30

06:00

cold_mod

04:30

kW

Figure 45 demonstrates the increased demand of cold appliances during the day and corresponding warm temperatures.

Time Figure 45: Summer cold appliance demand

From this relatively modest number of model parameter changes we can now reproduce the summer weekday profile with some accuracy, see Figure 46.

48

0.6

0.5

0.3

Smr Wd

0.2

Smr_mod

0.1

03:00

01:30

00:00

22:30

21:00

19:30

18:00

16:30

15:00

13:30

12:00

10:30

09:00

07:30

06:00

0 04:30

kW

0.4

Time Figure 47: Comparison of model and Elexon curve summer weekday

49

8.3

Appliance model development

As described in the report C1 on energy efficiency, the modelling framework depends on models of appliances that provide realistic power flow. In the case of lighting a very simply model is defined with an on and off mode, and single values for active and reactive power. In previous work (Bilton 2010) a set of cold and wet appliance models had been developed to mimic the population of 2010. However with the recent advances in appliance efficiency these models has become out dated. Moreover they did not reflect the appliance topologies identified by the Low Carbon London household survey. In order to develop a realistic baseline, both cold and wet appliance model fleets were developed from the 2010 prototypes. In this process, first a set of baseline appliances are developed and then modified to use less energy and approximate the numbers expressed by the label system as described previously. A fundamental aspect of this work was to understand the appliances available for sale today, and to this end a survey was conducted to establish the current market offering. An appliance superstore in London was visited and every appliance label was photographed for later collation. A summary of washing machines available in Spring 2014 are shown in Table 7. Table 7: Available washing machine ratings in spring 2014

Rating

kWh/annum

Litres/annum

load (Kg)

A+++

count 22

180

10621

8.04

kWh 0.82

kWh/Kg 0.10

L/cycle 48

A++

11

205

10793

7.72

0.92

0.12

49

A+

27

201

9664

6.46

0.92

0.14

44

Here we can see that there is a trend to larger capacity machines, and this is consistent with the label ratings being based on load size. In the sample shown here, a partially loaded A++ machine may use more energy than its A+ equivalent. Note also that the larger machines also use more water, this is likely used for more extensive rinsing given that if it was part of the heated wash cycle it would increase energy demand. The same data for dishwashers is presented in Table 8. Table 8: Available dishwasher ratings in spring 2014

Rating

Count

kWh/annum

L/annum

Settings

kWh

kWh/setting

L/cycle

A++

1

262

1680

13

0.935714

0.071978

6

A+

18

272.0556

3177.778

11.5

0.971627

0.085385

11.95536

50

These show us that only A+ and above appliances are now commonly available. Note in this case the water demand of the A++ appliances is half that of the A+ equivalent. This level of water use is so low that it would be very difficult to clean the same dishes manually with less water (6 litres is roughly one full washing up bowl). The review of cold appliances labelling was conducted to establish the size of appliances and their relative energy demand. The following two tables show us that per litre of storage, larger cold appliances use less energy. This is a corollary of the surface area to volume ratio, and hence losses being more favourable to large appliances.

Figure 47: Relationship between refrigerator size and energy consumption

Figure 48: Relationship between freezer size and energy consumption

51

Table 9: Annual energy consumption and internal volume of the different types of cold appliance.

Type

kWh/annum

Larder (l)

Fridge 1

120

119

Fridge 2

120

140

Fridge 3

134

250

Fridge 4

151

348

Freezer 1

175

67

Freezer 2

189

88

Freezer3

219

156

Freezer4

309

236

Table 9 provides us with a means to start designing physical models of cold appliances since we now have information about the relationship between size and kWh. Given the different demand of fridges and freezers, combined fridge-freezers, clearly there is a need to understand the size of each compartment. Figure 49 charts all the available fridgefreezers by freezer volume versus chiller space volume.

Figure 49: Currently available fridge-freezer topologies

Despite the apparent diversity of domestic appliances, the modelling of more complex appliances is mitigated by different brands having very similar operating behaviour thus multiple brands need not be modelled. Moreover the efficiency improvements that have been effected largely by similar means, for a reduction in water usage in the case of dishwashers.

52

Project Overview Low Carbon London, UK Power Networks’ pioneering learning programme funded by Ofgem’s Low Carbon Networks Fund, has used London as a test bed to develop a smarter electricity network that can manage the demands of a low carbon economy and deliver reliable, sustainable electricity to businesses, residents and communities. The trials undertaken as part of LCL comprise a set of separate but inter-related activities, approaches and experiments. They have explored how best to deliver and manage a sustainable, cost-effective electricity network as we move towards a low carbon future. The project established a learning laboratory, based at Imperial College London, to analyse the data from the trials which has informed a comprehensive portfolio of learning reports that integrate LCL’s findings. The structure of these learning reports is shown below:

Summary

Distributed Generation and Demand Side Response

SR DNO Guide to Future Smart Management of Distribution Networks

A1 Residential Demand Side Response for outage management and as an alternative to network reinforcement A2 Residential consumer attitudes to time varying pricing A3 Residential consumer responsiveness to time varying pricing A4 Industrial and Commercial Demand Side Response for outage management and as an alternative to network reinforcement A5 Conflicts and synergies of Demand Side Response A6 Network impacts of supply-following Demand Side Response report A7 Distributed Generation and Demand Side Response services for smart Distribution Networks A8 Distributed Generation addressing security of supply and network reinforcement requirements A9 Facilitating Distributed Generation connections A10 Smart appliances for residential demand response

Electrification of Heat and Transport

B1 B2 B3 B4 B5

Impact and opportunities for wide-scale Electric Vehicle deployment Impact of Electric Vehicles and Heat Pump loads on network demand profiles Impact of Low Voltage – connected low carbon technologies on Power Quality Impact of Low Voltage – connected low carbon technologies on network utilisation Opportunities for smart optimisation of new heat and transport loads

Network Planning and Operation

C1 C2 C3 C4 C5

Use of smart meter information for network planning and operation Impact of energy efficient appliances on network utilisation Network impacts of energy efficiency at scale Network state estimation and optimal sensor placement Accessibility and validity of smart meter data

Future Distribution System Operator

D1 D2 D3 D4 D5 D6

Development of new network design and operation practices DNO Tools and Systems Learning Design and real-time control of smart distribution networks Resilience performance of smart distribution networks Novel commercial arrangements for smart distribution networks Carbon impact of smart distribution networks

Low Carbon London Learning Lab

UK Power Networks Holdings Limited Registered office: Newington House 237 Southwark Bridge Road London SE1 6NP Registered in England and Wales Registered number: 7290590 [email protected] ukpowernetworks.co.uk/innovation