Guest post by Energy Matters’ commentators Alex Terrell and Andy Dawson. Alex Terrell is a business consultant in the area of Vehicle Telematics. He has also consulted in Energy and Manufacturing, and has a degree in Engineering. Andy Dawson is an energy sector systems consultant and former nuclear engineer.
This lengthy post is in three parts and aims to provide greater sophistication to a UK 2050 electricity model than can be achieved using the DECC 2050 calculator. Part 1 (below) presents the demand model. Parts 2 and 3 (to follow) will look at how demand may be met by a high nuclear option and from a renewables option.
There is endless discussion and debate over what the UK’s electricity supply should look like moving forward out to 2050. However, there is less discussion over what demand will be, and how this might vary from day to day and season to season. This article looks at UK electricity demand over the long term – specifically setting a scenario for 2050, which also meets the UK Government’s aim of an 80% reduction in CO2 emissions. Electricity demand in 2050 will vary significantly depending primarily on:
- The temperature and resultant heating or cooling requirements
- Whether it is a weekday or weekend
We have created a demand model based on using a number of technology assumptions for 2050, and a high electrification scenario. The demand model has then been matched to UK daily temperature data from the last 20 years to identify how demand could vary on a daily basis in 2050.
Average demand over the course of the year is expected to be 72GW. However, this will vary from a daily average 43 GW (experienced when the temperature is 18 centigrade and it’s a weekend) to 121 GW (experienced when the average temperature is -7C, as it was on 20 Dec 2010).
In cold weather, the largest element of demand is for heating, leading to greater seasonal variation than at present. However, we assume that homes have sufficient thermal mass that demand can be smoothed out over the course of a day, by making small changes to the time of heating. Electric vehicle charging will also help to smooth daily demand and shift demand to weekends.
Two subsequent articles will apply the demand model to historical daily weather patterns from previous years to identify how demand could be met with two extreme supply scenarios:
- A nuclear power scenario, using some supplemental thermal fuel (gas, diesel, biofuel etc.) backup.
- A wind and solar scenario, using thermal fuel as backup.
We have not modelled scenarios with higher levels of gas usage and carbon capture and storage – in particular, the technology of steam reforming natural gas (and sequestrating the resultant CO2) could help reduce peak winter demand as people choose to burn hydrogen for heat rather than use electric heating.
This report follows on from an article published earlier by Andy Dawson on Energy Matters examining whether conversion of the UK’s current electricity supply to a completely (or near-completely) decarbonised mix of nuclear and pumped storage is feasible. During the discussion following that article, a number of commenters raised the question of (a) was comprehensive decarbonisation an option by that approach, and (b ) what were credible future demand scenarios mostly arising from use of the DECC “Pathways” models.
DECC introduced its pathways scenarios modelling tool in order to allow anyone to enter their inputs and see the effects, with an aim of reducing emissions by 80%. One of the issues with the “pathways” models is that they’re rather blunt instruments (in that the various supply scenario components are a limited menu), and that it’s far from clear if the underlying models take adequate account of variations in demand; as has been pointed out by many on these pages, both diurnal and seasonal variations in demand mean that simple models based on averages are far too simplistic to make a real contribution to the debate. This led us to the conclusion that the only real option was to build our own, based on realistic (and readily adjustable) assumptions about the scale and shapes of the various elements of overall demand.
Current UK Final Energy consumption is approximately 140 Million tons of oil equivalents, which equates to 1,628 TWh/year. Electricity is just under 20% of this. It is widely accepted that if the UK wishes to make deep cuts in its carbon emissions, then much of the remaining 80% needs to be converted to electricity. Please note that this implies significant investment in the national electricity transmission and distribution systems. We have NOT attempted to model this, as it is an implicit requirement in any decarbonisation scenario.
However, some sectors are likely to be extremely difficult to “electrify”:
- There do not seem to be any credible technology paths to the decarbonisation of aviation, with the possible exception of biofuels. The weight and aerodynamic implications of cryogenic storage make hydrogen fuelling unlikely, and fully synthetic fuels seem unlikely to be available in volume by 2050 at acceptable prices.
- While there do seem to be viable options for the decarbonisation of long distance bulk shipping (several small modular reactor (SMR) designs seem to be in the “sweet spot” for large ship engines, and weight/bulk of cryogenic H2 storage appears viable on large vessels), for smaller and mid-sized vessels, achieving a high energy density for the fuel source remains an issue.
- Long distance road haulage will be problematic. Conventional battery approaches seem unlikely to give an adequate range:payload:recharging time relationship to make for attractive economics. Options could include H2 fuelling (with the fuel derived from methane reformation), or standardised “swappable” batteries – in the absence, however, of an obvious contender technology we’ve assumed the continuation of liquid fuelling.
Whilst biofuels could go a long way to fulfilling the above demands, it is likely that the electricity sector will have to significantly expand, and go well beyond an 80% CO2 reduction target in order to offset emissions from these sectors. This raises a number of questions:
- At what efficiency can current demand be converted?
- What is the total of electrical demand?
- What is the seasonal and daily variation in this demand?
- How can this demand be met with low carbon technologies?
The 2050 Demand Model
Using the DECC scenarios:
The DECC pathways calculator allows us to make various assumptions and models average energy flows in 2050. We can use this with a “high electrification scenario” to estimate electricity requirements. Figure 1 shows our pathway, alongside the “no change” pathway and the Atkins pathway.
Figure 1 – various outputs from the DECC pathways. Ours is available here. Our pathway works out to an average of 72GW. That would be straightforward, but as we know, averages are not the most important thing for an electrical network. DECC’s pathway does not appear to take into account either seasonal and daily demand variations, or weather related supply changes, as demonstrated here.
How will demand vary over the course of the week and over the year?
For this we need to look more closely at the major demands for electricity: Heating, Vehicles, Industry, and commercial. The following assumptions are sanity checked against the DECC Pathways outputs.
We modelled how demand might vary according to temperature, specifically with external daily average temperatures of 30C (extremely hot), 18C (pleasant), 12C (typical), 0C (Cold) and -10C (exceptionally cold – it last happened in 1838 according to the Central England Temperature records). The outcome is shown in Figure 2. We see that whilst the average electrical demand is only 1,721 GWh/day according to the Atkins pathway, and 1,740 GWh/day according to our pathway, the actual demand could vary from 1215 GWh/day to around 3,000 GWh/day in exceptionally cold weather.
The assumptions within the categories for:
- Heating and Cooling – Homes
- Road Transport
are given below.
Figure 2 – Daily electrical demand according to pathway averages (left) and modelled demand according to external temperature.
Heating and Cooling Homes
The DECC pathways calculator makes no breakdown of when the heating is applied. To understand this, we need to identify current heating needs and make some big assumptions regarding heating demand in 2050.
There were 27 million households in the UK in 2010, up from 19 million in 1970. Population growth has been a small driver, but a bigger driver has been a reduction in the number of people per household. We have assumed that this trend will not continue and are assuming 30 million households in 2050. Note, however, that even were growth to continue, shifts in the housing mix – particularly an increasing proportion of apartment dwellers – can largely offset this impact.
In 1970, the overall rate of heat loss from a home was, on average, 376W/°C. Forty years later, it had fallen by almost a quarter to 287W/°C according to UK housing facts, page 54. This means that the average 2010 home needs 287W to maintain a 1 degree temperature differential with the outside. Figure 3 shows how this is made up and how it varies over time.
Figure 3: Average heat loss per dwelling (W/°C). This measure of heat loss says that for an average home, if it is 1°C cooler outside than inside, it needs 287 watts (2010 figure) of heating to maintain a stable temperature. The measure is affected by insulation and ventilation losses.
The spike in 2008 was due to a change in modelling of ventilation losses. If we add this back to the 1970 figure, then the fall is from about 420 W/°C to 287 W/°C, a fall of 32%.
If we assume a geometric trend rate of fall, we arrive at 2050 figure of 196W/°C. This figure is highly significant in establishing energy requirements in 2050. It means that if the average temperature is 0C outside, and 20C inside, the average home will require 94KWh of heat in a day. This is very easy for a new-build house to achieve (UK homes built since 2013 in accordance with building regulations easily surpass this level), but very difficult for a 1980s build house. Simple cavity insulation, double glazing, and increased loft insulation won’t be sufficient – it would probably need external wall insulation, which also has the added advantage of providing a very high thermal mass inside the insulated boundary. More challenging would be changes to ventilation, including heat recovery from exhaust air. It’s doubtful this latter can be achieved with much of the “legacy” housing stock, and hence places a limit on attainable reductions.
The UK has a very low rate of new house building – only about 0.5% of the stock is replaced each year. Nevertheless, by 2050, perhaps 20% of homes will be built to post 2013 standards and that will help to bring the average energy consumption down. However the energy is supplied, the UK needs to improve the energy efficiency of its housing stock – both of existing houses and of new build houses. This is perhaps the only area where there is consensus amongst energy experts and environmental groups. Based on this analysis, we would also recommend favouring improvements that boost the thermal mass of homes, even where this increases the “embedded energy” of the build. Key assumptions for the model include:
- There are 30 million “homes” – a mix of flats and houses, up from 27 million in 2010.
- The internal temperature is set at 20C – compared to the current average of 18C. However, this can fall to 19C at 0C external and 17C at -10C external. With better insulated homes, people will want to keep them warmer than they do today.
- At 0C outside temperature, the average home requires, on a daily basis:
– 10.5KWh of domestic water heating. This is provided 70% with heat pumps at a COP of 2, 20% from resistance heating, and 10% from solar and district heating.
– 90KWh of space heating, of which 10KWh comes from lighting and appliances. The remainder is provided 80% with heat pumps at a COP of 3 (a typical house might have two 5KW air source heat pumps), 10% from resistance heating, and 10% from solar and district heating.
- The average home will have sufficient heat capacity so that unheated, the temperature will fall by no more than 2C over 24 hours (at 0C outside temperature). This requires, for example, 40 cubic metres of concrete or brick in the walls and floors, inside the insulating envelope. This effectively means that heat can be delivered within reason at any time of the day, and used to even out electricity supply.
- Domestic water requires 3.7KWh of heat per day. In summer, this is mostly provided by solar heating, but in cold weather by a heat pump with a COP of 2 (due to the high temperature differential).
- Heat pump COP at -10C external temperature falls to an average of 2.5 for space heating and 1.5 for hot water.
- Other technologies will be deployed to reduce overall heating demand, to reach the average 196W/°C assumption. For example:
– In houses with high levels if air-tightness and heat recovery, a heat-pump can remove heat from the extracted air or waste water. This would allow a considerably higher CoP.
– There may be some elements of inter-seasonal heat storage – perhaps using air conditioning and heat transfer in the summer to store heat in the ground, to improve the winter performance of ground source heat pumps.
In practice, there will also be some wood and oil heating. Wood heating – whether logs or pellets – is becoming increasingly common in Germany as an alternative to gas. We would support the use of wood heating in very cold conditions as it would make energy supply easier. However, if homes have wood burners, people will use them throughout much of the year, which will increase local air pollution and reduce the feed for biofuels. Hence we assume no wood heating in our model.
If homes have heat pumps for the winter, these heat pumps will also be used for cooling in the summer. This is not a problem for the supply system, as demand will still be significantly lower in the summer. Indeed, with ground source heat pumps, some air conditioning could be useful to replenish underground heat.
This model shows that heating and cooling of homes will require 243 GWh on a typical per day, which is not far off from the Atkins DECC pathways model. However, this could shoot up to over 1500 GWh in exceptionally cold weather (around -10C)
Figure 4 – National electrical heating requirements according to external temperatures and average internal temperatures.
As home heating can make up over half of electricity demand on a cold day, the “Average Heat loss per dwelling” figure is critical in assessing the overall heating requirement. We have assumed 196W/C. Figure 5 shows how overall demand varies with this figure:
Figure 5 – How overall demand according to average home insulation levels. Overall demand is shown in GWh/day on the left scale, and this is averaged to GW on the right scale.
Commercial demand was estimated based on 2011 data for electrical and non-electrical demand. It is assumed that this demand is:
- Mostly switched to electricity using heat pumps and induction heating
- Reduced through efficiency improvements
- Increased mostly according to population growth (0.3% pa), or for Catering and Other, according to economic growth.
The results, in GWh per year are shown in Figure 6.
Figure 6 – Annual commercial demand for electricity and other energy, extrapolated from 2011 to 2050 assuming extensive electrification.
The daily demand was then calculated based on temperature, and the results shown in Figure 7. We have introduced some curtailment of at -10C, as at these temperatures many people don’t make it to work and many shops will close.
Figure 7 – Daily demand for electricity from commercial
Industrial demand was calculated in the same way as commercial. Most elements were assumed to grow in line with the Economy (2% pa) rather than population (0.3% pa), with an annual efficiency improvement of 1%.
Figure 8 – Annual demand for electrical and non-electrical power in industry.
A large proportion of the industrial load is in fact for heat, with a large proportion of this being supplied at relatively low temperature; here we’ve again assumed a large shift to heat-pumps and/or utilisation of waste heat from generating facilities. For high temperature processes, we assume the majority supply (c. 80%) is via direct resistive or RF methods. In all cases, we assume a significant residual demand which is not readily electrifiable.
Industrial demand is assumed to be progressively curtailed below 0C. This is partly as employees have trouble getting to work, and partly due to Demand Side Reduction coming into play with heavy users of electricity. As a result, industrial demand is reduced in rare, extreme weather scenarios.
Figure 9 – Daily demand for electricity from commercial
We assumed that by 2050, road transport would be predominantly electric, with private vehicles able to travel several hundred kilometres on a charge. As a baseline, we took 2015 DfT mileage date and assumed modest increases in distance travelled.
We assumed a high degree of electrification for all vehicles except for Heavy Goods Vehicles (on the grounds of the particular challenges of meeting the range/payload challenges with foreseeable battery technologies; in this space we suspect that hydrogen based technologies are likely to prove more viable, with fuel supplied via steam reformation of natural gas, and sequestration of the resultant CO2 waste stream, or from nuclear thermal production). However, the choice of technology makes little impact on our electricity demand model.
We also added an up lift for cold and hot weather to provide heating and cooling respectively, and curtailment of car, motorbike and Light Goods Vehicles in very cold weather.
The average car drives about 300 kilometres in a week, which should be well within the electric range of most cars in 2050 (The Chevrolet Bolt is aiming for 200 miles range in 2017). This will require about 50KWh per week, and charging can take place once or twice per week. We have assumed that:
- Charging time of day can be partly influenced – users will plug their car in and request that it’s charged by a certain time
- Vehicles will be more likely to charge over the weekend, when they are less likely to be on the road
(It seems that by 2050, to state that the vehicle drives, rather than is driven for appears increasingly probable!)
Adjusting for Weekends
We assumed some shift in demand between weekends and weekdays, as is currently the case.
Road transport (vehicle charging) and use of home lighting and appliances increase at the weekends. Everything else increases for weekdays.
Figure 10 shows the variation on weekdays and weekends for different temperatures.
Figure 10 – 2050 Weekday (WD) and Weekend (WE) daily electrical demand according to average daily temperature.
We assume, with some control over when heating is applied, and when vehicles are charged, demand can be largely flattened out over the day. So for example at zero C, the following illustrates how the 2,548GWh of daily demand could be smoothed to a power supply of approximately 106GW.
Figure 11 – Electricity demand smoothed over the day. Road transport is mostly charged at night. Heating and cooling is adjusted to be counter-cyclical with Industrial demand.
Experience to date shows that consumers are very insensitive to short term, small scale price signals. They will not be able or willing to vary their demand to such a fine level just on the basis of price. So fine tuning demand will need to be done by embedded systems connected to a central demand management algorithm. Examples could be:
- A user will set their thermostat to 20C. The central system will tell the heating when to operate, allowing the temperature to remain between 19C and 21C. There may be a provision to allow the temperature to drop to 18C in “exceptional circumstances”.
- Car users will plug their cars in, and tell the car when it is needed next. The system will then choose the optimum time to charge.
To encourage users to accept the approach (with its “Big Brother” and “loss of control” connotations), there will need to be some financial incentives. These will have to be on an aggregate basis – perhaps an annual reduction on the electricity bill. Some customers will prefer to pay a little bit more to have their own “control”, but enough will accept remote control to flatten demand. It should be noted that several markets attempting Smart Meter rollouts have already seen significant consumer backlash over “loss of control” and privacy issues, resulting in the Netherlands (for example) abandoning plans to make such meters mandatory.
If the demand management systems are off by a few GW, then some pumped storage or battery storage (of the order of 10s of GWh) can be used to even out demand over the day, and nuclear reactors are able to go over or under their nominal power for short periods.
Note also that this does not require the ability to draw power down from EV batteries with all of the complexities inherent in this approach – for example, requiring inverters to take DC output from the batteries into domestic AC supplies.
The Demand Curve
Extrapolating the values derived above, we can estimate a demand curve.
Figure 12 – Demand Curve according to average daily temperature
Matching to the weather
We have identified a weather and day dependent demand. The next question is how much demand this creates, given the potential weather in 2050.
The HadCRUT temperature database for Central England has recorded daily average, minimum and maximum temperatures since 1772. In practice, the average temperatures across the entire United Kingdom are likely to be warmer on extreme cold days than just those in central England. However, central England is near enough to major population centres, so we will err on the side of caution and use this as an average proxy temperature.
We also note that average winter temperatures have been increasing over the last 50 years at a rate of about 0.16 centigrade per decade. However, 2010 recorded 27 freezing days, a sequence of 11 freezing days in succession, and a minimum temperature of -7C.
Figure 13 – Central England: Number of freezing days, longest sequence of freezing days, average winter temperatures, and minimum temperatures over the last 50 years.
For a future energy supply scenario, we therefore think it’s prudent to model demand assuming the temperature profile of the 20 years up to and including 2015. Based on this, we can model the percentage of days when average electricity demand will exceed a certain value.
Figure 14 – A frequency distribution of electricity demand in 2050. The most common level of demand (averaged over the day) is 60GW. Demand always exceeds 43GW. This chart is based on 20 years of daily temperatures (from 1996 to 2015) – more years of data would smooth the red line (number of days).
The supply needs to cater to extreme events. A zoom of the above chart to show the probability of average demand exceeding high levels is shown in Figure 15.
Figure 15 – The percent of days when demand exceeds high levels. The percentage figure (y-axis) has a logarithmic scale.
A temperature of -7C (averaged over 24 hours) would result in a demand of 121GW, and we can expect this approximately once every 20 years. In reality – given the consequences of a supply shortage in cold weather – availability may have to higher than 121GW, though this could be enabled by diesel generators with almost no impact on emissions.
As well as being able to supply this amount for a day, the system needs to sustain a high level of output over a number of consecutive days. Figure 16 shows an example of this.
Figure 16 – Electricity demand over a cold spell, with 11 consecutive days of freezing temperatures leading to demand of over 100GW for that time period. This temperature sequence actually occurred in 2010.
Following articles will show how this demand can be met with:
- A nuclear powered system, using some supplemental thermal fuel (gas, diesel, biofuel etc) backup.
- A wind and solar scenario, using thermal fuel as backup.