Archive for the ‘Electricity Generation’ Category

A Watched Pan…

January 18, 2022

Click on Image for larger version.  A vision of domestic bliss in the de Podesta household. Apparatus in use for measuring the rate of heating of 1 litre of water on an induction hob.

In the beginning…

Friends, my very first blog article (written back on 1st January 2008 and re-posted in 2012) was about whether it is better to boil water with an electric kettle or a gas kettle on a gas hob.

Back then, my focus was simply on energy efficiency rather than carbon dioxide emissions. I had wanted to know how much of the primary energy of methane ended up heating the water. I did this by simply timing how long it took to boil 1 litre of water by various methods.

Prior to doing the experiments I had imagined that heating water with gas was more efficient because the fuel was used directly to heat the water. In contrast, even the best gas-fired power stations are only ~50% efficient.

What I learned back then was that gas cookers are terrible at heating kettles & pans! They were so much worse than I had imagined that I later spent many hours with different size pans, burners, and amounts of water just so I could believe my results!

Typically gas burners only transferred between 36% and 56% of the energy of combustion to the water – the exact fraction depending on the size and power of the burner. Heating things faster with a bigger burner was less efficient. Using a small flame and a very large pan, I could achieve an efficiency of 83%, but of course the water heated only very slowly.

This inefficiency was roughly equivalent to or worse than the inefficiency of the power station generating electricity, and so I concluded that electric kettles and gas kettles were similarly inefficient in their use of the primary energy of the gas. But that using electric kettles allowed one to use the correct amount of water more easily, and so avoided heating water that wasn’t used.

14 years later…

After a recent conversation on Twitter (@Protons4B) I thought I would look at this issue again.

Why? Well two things have changed in the last 14 years.

  • Firstly, electricity generation now incorporates dramatically more renewable sources than in 2008 and so using electricity involves ever decreasing amounts of gas-fired generation.
  • Secondly, I am now concerned about emissions of carbon dioxide resulting from lifestyle choices.

Also being a retired person, I now have a bit more time on my hands and access to fancy instruments such as thermometers.

The way I did the experiments is described at the end of the article, but here are the results.

Results#1: Efficiency

The chart below shows estimates for the efficiency with which the electrical energy or the calorific content of the gas is turned into heat in one litre of water. My guess is these figures all have an uncertainty of around ±5%.

  • The kettle was close to 100% efficient:
  • The induction hob was approximately 86% efficient
  • The Microwave oven was approximately 65% efficient

In contrast, heating the water in a pan (with a lid) on a gas hob was only round 38% or 39% efficient.

Click on Image for larger version. Chart showing the efficiency of 5 methods of heating 1 litre of water. 100% efficiency means that all the energy input used resulted in a temperature rise. The two gas results were for heating pans with two different diameters (19 cm and 17 cm).

It was particularly striking that the water heated on the gas burner (~1833 W) took 80% longer to boil than on the Induction hob (~1440 W) despite the heating power being ~20% less on the induction hob.

Click on Image for larger version. Chart showing the rate of heating for each of the 5 methods of heating 1 litre of water. Notice that the water heated on the gas burner (~1833 W) took 80% longer to boil than on the Induction hob (~1440 W) despite the heating power being ~20% less on the induction hob. Notice that up until 40 °C, the microwave oven heats water as fast as the gas hob, despite using half the power!

Results#2: Carbon Dioxide Emissions 

Based on the average carbon intensity of electricity in 2021 (235 g CO2/kWh), boiling a litre of water by any electrical means results in substantially less CO2 emissions than using a pan (with a lid) on a gas burner.

I performed these experiments on 17th January 2021 between 4 p.m. and 7 p.m. when the carbon intensity of electricity was well above averages: ~330 g CO2/kWh. In this case, boiling a litre of water in a kettle or induction hob still gave the lowest emissions, but heating water in a microwave oven resulted in similar emissions to those arising from using a pan (with a lid) on a gas burner.

Click on Image for larger version. Charts showing the amount of carbon dioxide released by heating 1 litre of water from 10 °C to 100 °C using either electrical methods or gas. The gas heating is assumed to have a carbon intensity of 200 gCO2/kWh. The left-hand chart is based on the carbon intensity of 330 gCO2/kWh of electricity which was appropriate at the time the experiments were performed. The right-hand chart is based on the carbon intensity of 235 gCO2/kWh of electricity which was the average value for 2021. Electrical methods of heating result in lower CO2 emissions in almost all circumstances.

Results#3: Cost 

Currently I am paying 3.83 p/kWh for gas and 16.26 p/kWh for electricity i.e. electricity is around four times more expensive than gas.

These prices are likely to rise substantially in the coming months, but it is not clear whether this ratio will change much.

So sadly, despite gas being the slowest way to heat water and the way which releases the most climate damaging gases, it is still the cheapest way to heat water. It’s about 40% cheaper than using an electric kettle.


For the sake of the climate, use an electric kettle if you can.


That was the end of the article and there is no need to read anymore unless you want to know how I made these measurements.


Estimating the power delivered to the water + vessel

  • For electrical measurements I paused the heating typically every 30 seconds, and read the plug-in electrical power meter. This registered electrical energy consumed in kWh to 3 decimal places.
    • I fitted a straight line to the energy versus time graph to estimate power.
  • For gas measurements I read the gas meter before and after each experiment. This reads in m^3 to 3 decimal places and I converted this volume reading to kWh by multiplying 11.19 kWh/m^3.
    • The gas used only amount to 0.025 m^3 so uncertainty is at least 4% from the digitisation.
    • I divided by the time – typically 550 seconds – to estimate the power.

Mass of water

  • I placed the heating vessel (kettle, pan, jug) on the balanced and tired (zeroed) the reading.
  • I then added water until the vessel read within 1 g of 1000 g. Uncertainty is probably around 1% or 10 g.

Heating rate with 100% energy conversion

  • Based on the power consumed, I estimated the ideal heating rate if 100% of the supplied power caused temperature rises in the water by using the equations.

  • I assumed the average specific heat capacity of water of the range from 10 °C to 100 °C was 4187 J/ (kg °C)

Measuring the temperature.

  • For electrical measurements I paused the heating typically every 30 seconds, stirred the liquid with a coffee-stirrer for 2 or three seconds, and then took the temperature using a thermocouple probe..
  • For gas measurements it wasn’t possible to the pause the heating because of the way I was measuring the power. So about 10 seconds before the reading was due I slipped the coffee stirrer under the lid to mix the water.

Estimating the rate of temperature rise.

  • For all measurements I fitted a straight line to the temperature versus time data, using only data points below approximately 80 °C to avoid the effects of increased temperature losses near to the boiling point.

Mass of the ‘addenda’.

  • The applied power heated not only the water but also its container.
  • The heat capacity of the 19 cm stainless steel pan (572 g) was roughly 6% of the heat capacity of the water.
  • I chose not to take account this heat capacity because there was no way to heat water with a container. So the container is a somewhat confounding factor, but allows more meaningful comparison of the results.

Efficiency of boiling

  • I estimated efficiency by comparing the actual warming rate with the ideal warming rate.
  • I then calculated the energy required to heat 1 kg of water from 10 °C to 100 °C, and multiplied this by the efficiency.
  • In this way the result is relevant even if all the measurements did not start and stop at the same temperatures.


  • I heated the water in the microwave in a plastic jug which did not have a tight fitting lid. I am not sure if this had an effect.
  • I did notice that the entire microwave oven was warm to hot at the end of the heating sequence.

I’m dreaming of 2026…

December 3, 2021

Friends, a curious thought occurred to me yesterday:

“What would happen if the Hinckley C nuclear Power Station was completed on time in 2025?”

I know it’s unlikely but it is conceivable.

But whenever it does start operating, it will produce an extra 3.26 GW of electricity 24 hours a day, 7 days a week. This is enough to change the UK electricity generation “market” at a stroke.

 January 2020

To illustrate the effect that Hinkley C will have, I downloaded hourly generation figures for 10 days at the start of 2020.

I picked this range of dates because the data was free – I would need to have paid to obtain more data! But it is sufficient to make my point. Take a look at Gridwatch for monthly and yearly summaries.

I have not plotted all the generating sources or interconnectors bringing us electricity, but instead I have just drawn three curves from the data set.

  • Total Demand
  • Nuclear 
  • Nuclear + Wind

Click Image for a larger version. Generation from the first 10 days of 20 showing electricity from nuclear power stations, the sum of nuclear + wind generation, and total demand. The gap between the black and green curves is filled with electricity from a variety of sources, but mainly with electricity from gas-fired generators.

I have picked these data because of the way the UK grid is run. In simplified form, it works like this:

  • First the grid accepts all the nuclear electricity available. This is because (for technical reasons) nuclear power stations cannot easily change their output.
  • Then the grid accepts whatever renewable energy (solar or wind) that is available. This can be well-predicted a day or two ahead of time.
  • Then, through a complex system of contracts, and “market”-mechanisms, the grid adds electricity from a variety of sources to meet demand. Most of this is usually met by gas-fired generation which emits ~ 450 g CO2/kWh of electricity supplied.

But what would the situation look like if Hinkley C were operating, and wind generation were twice it’s value in 2020?

This situation could hypothetically occur as soon as 2026.

January 2026?

  • IF demand in January 2026 were by chance the same as in 2020, and…
  • IF Hinkley C were generating at full power and…
  • IF wind power was twice what it was in 2020…

…then we would find ourselves in the extraordinary situation depicted in the graph below.

Click Image for a larger version. A hypothetical situation in January 2026 when nuclear supply is supplemented 3.26 GW of generation by Hinkley C, and wind generation is twice what it was in 2020. The gap between the black and green curves is now routinely reversed – indicating a regular ‘oversupply’ of green electricity.

The graph shows that the combination of ‘nuclear + wind would’ regularly exceed demand before considering any other sources of generation.

During these periods demand would be met entirely with low carbon sources and the carbon intensity of UK electricity would fall to pleasingly low values (~50 gCO2/kWh).

And during these periods – shown in blue on the graph above – there would be typically 10 GWh a day of ‘surplus electricity’.

The marginal cost of this ‘surplus’ electricity is debatable, but it is close to zero.

Could this really happen?

Yes. It has already happened briefly this year over the late May bank holiday weekend. And as renewable electricity generation grows, such situations will occur ever more frequently.

And whenever Hinkley C comes on stream – in 2025 or later – such events will inevitably become commonplace.

What are the consequences?

I don’t know!

The electricity “market” operates by complex rules, and as this year’s ‘odd’ May weekend event showed, prices cannot just fall to zero, but actually become negative: i.e. companies will pay you to use their electricity!

But however it is dealt with by the “market” rules, the reality is that the UK will be routinely generating renewable electricity in excess of demand at a cost close to zero. This has consequences at many levels.

  • First of all: from the point of view of investors looking to build renewable generation – solar or wind – they will no longer be able to ‘dump’ electricity onto  “the market” whenever the Sun happens to shine and the wind happens to blow. This will make life more difficult for these investors.
  • Secondly: anyone who can store energy for later re-sale or re-use will have access to very low cost electricity. This represents an opportunity for many nascent industries and technologies.

However the biggest consequence could be a change in conception of how the “market” operates. So far, we have almost always assumed that supply will adjust to meet demand. We have had some incentives to use ‘off-peak’ electricity, but not many.

With an ‘oversupply’ of electricity, there will be opportunities for industries which can adjust demand to meet available supply. If this is implemented well, then the economic singularities arising from a zero price will be avoided – but the energy should still be cheap and it will still be green.


If green hydrogen is ever to play a role in the UK, then this could present an initial opportunity. 10 GWh/night of electricity is sufficient to generate 200 tonnes of hydrogen (@50 kWh/kg).

If EV use grows at the rate that many anticipate, then smart charging at night could help sustain that growth. 10 GWh/night would be enough to deliver 25 kWh (roughly a 50% charge) to 400,000 EVs.

I am sure many more ideas will emerge about how to reap this low-cost, low-carbon harvest.


So the conclusion of my whimsy is that whenever Hinkley C starts generating it will transform the UK electricity supply “market” overnight. And this could happen sooner than I had anticipated.

Whenever it happens, it will significantly reduce the average CO2 emissions from electricity in the UK. But there will also be costs associated with this ‘cheap’ electricity.

You might consider it ironic – and perhaps not a little unfair – that the introduction of some of the most expensive electricity ever put onto the UK grid – EDF are guaranteed ~10.6 p/kWh for all the electricity that Hinkley C produces – will put pressure on generators using wind turbines and solar PV stations – the suppliers of the cheapest energy ever supplied to the grid.

You might consider that this is not how “markets” are supposed to work. That is why I have put every instance of the word “market” in quotation marks. If you know a better word to use, I would love to know what it!

November 2021: Heating and Carbon Emissions

December 1, 2021

Friends, it is December already and I am preparing to hibernate.

But before I curl up and doze, I just thought I would summarise some of the energy statistics for the house in November.

They are actually pretty remarkable: the low-carbon solar-powered world really is here already…


The average November temperature in Teddington was a very typical 8 °C, but the end of the month included several colder days with average temperatures of ~1 °C, and minimum temperatures down to -3 °C.

Click image for a larger version. Daily averages of the internal and external temperatures this November 2021. Also shown is the monthly average (dotted blue line) and the internal temperature measured every 2 minutes (black line connecting the red dots).

However, as the graph above shows, internally our house remained at an extremely stable temperature, barely changing day or night.


Over the month, the house used 609.4 kWh (~20.3 kWh/day) of electricity. This is roughly double our typical use of electricity without heating.

Click image for a larger version. Remarkably, ~21% of the electricity used in November 2021 came from the solar panels. The bulk (~70%) was purchased off-peak (00:30 to 04:30) and stored in the Tesla Powerwall for use during the day. Just 9% was purchased at peak rates after the battery ran out of charge.

This electricity demand was met as follows:

  • 129 kWh (~21%) came from the solar panels.
  • 428.9 kWh (70%) was off-peak electricity from the grid @5p/kWh = £24.02
  • 51.6 kWh (9%) was peak-rate electricity from the grid @16.26p/kWh =£8.39

The shifting of our time-of-use by using the Tesla Powerwall resulted in the average cost of a unit of grid electricity being ~ 6.2p/kWh.

Added to those costs is the standing charge of 25p/day, or £7.50/month.


We are still using gas for cooking and through the month we used:

  • 41.84 kWh (~1.4 kWh/day) @3.83 p/kWh = £1.60

Added to this is the standing charge of 23.85/day, or £7.16/month.

Heating & Domestic Hot Water (DHW)

Over the month, the heat pump used 312.1kWh (~10.4 kWh/day) of electricity – about half of the total 609.4 kWh used through the month.

At an average cost of 6.2 p/kWh of electricity, this cost £19.35.

Using this electricity, the heat pump delivered 1128.6 kWh of heat at an average cost of (£19.35/1128.6) = 1.7 p/kWh.

If a 90% efficient gas boiler had delivered this energy it would have used (1128.6/0.9)= 1254 kWh of gas which would have cost £48.03.

So the heat pump delivered savings of approximately 60% over using a gas boiler.

If we had not used the battery to allow the use of cheap-rate electricity, then 312.1 kWh of electricity would have cost approximately £44.76 – roughly a 7% saving over using gas.

Heat Pump Performance

As the graph at the head of the page makes clear, the 5 kW Vaillant Arotherm plus heat pump performed well, providing heating and DHW uncomplainingly even in the cold weather.

Click image for a larger version. COP data from last 12 days of November 2021. The blue dots are hourly averages and the large yellow dots show daily averages. The data include the domestic hot water and anti-legionalla cycles which heat water above 55 °C. The trend line indicates that COP is typically between 3 and 4 for external temperatures between 0 °C and 11 °C.

The key measure of the performance of a heat pump is its coefficient of performance (COP). This specifies the ratio of the heating energy delivered divided by the electrical energy consumed.

The graph above shows how the COP varied hour-by-hour and day-by-day through the last 12 days of November.

It is clear that the COP falls at lower external temperatures. This is because the heat pump has to work harder to deliver the heat across a bigger temperature difference.

  • Outside the external temperatures are lower
  • Inside the heat pump increases the temperature of the water flowing through the radiators to deliver more heat.

More specifically,

  • When the external temperature is ~ 10 °C, the water flowing through the radiators is at ~ 32 °C, a temperature difference of ~22 °C. With this small temperature difference the heat pump can operate with a COP in excess of 4.
  • When the external temperature is ~ 0 °C, the water flowing through the radiators is at ~ 42 °C, a temperature difference of ~42 °C. With this temperature difference the heat pump can only achieve a COP of ~ 3.

Some hourly readings show COP values much less than the trend line. These are hours in which ‘odd’ events occurred, such as de-frost cycles on the heat pump heat exchanger, or re-heating the hot water tank at 55 °C.

Carbon Dioxide Emissions

Since this is just a monthly analysis, I will consider only the fuel costs and neglecting the embodied carbon in the heat pump etc.. But please be assured, in the fuller analysis this is fully accounted for.

During November MyGrid GB reported the average carbon intensity to be 235 gCO2/kWh whereas Carbon Intensity reported the average to be 191 gCO2/kWh. The graph below shows the hour-by-hour data and the curves look similar but appear to be offset by ~ 45 gCO2/kWh. I don’t know which one is correct but I am going to calculate with the Carbon Intensity figures because they allow me to download half-hourly data for the whole month. The difference between the estimates can be added as a constant. Apologies for the confusion.

Click image for a larger version. Hour-by-hour Carbon intensity data from Carbon Intensity and MyGridGB They appear to differ by a constant additive number of ~ 45 gCO2/kWh. I do not know which one is right.

Analysing the data, I find that the 4 hours of off-peak electricity had an average carbon intensity of 141 gCO2/kWh versus 191 gCO2/kWh for the other 20 hours. (This would be ~ 186 gCO2/kWh versus 246 gCO2/kWh for the MyGridGB estimates.)

Using Carbon Intensity figures I estimate that:

  • 428.9 kWh of off-peak electricity @141 gCO2/kWh released = 60.47 kg CO2
  • 51.6 kWh of off-peak electricity @191 gCO2/kWh released = 9.86 kg CO2

for a total of 70.33 kgCO2 released from electrical use through the month. (92.46 kgCO2 using the MyGridGB data)

The effective carbon intensity would be 146 gCO2/kWh (or 191 gCO2/kWh if the MyGridGB figures were correct)

The heat pump used 312.1kWh of electricity and so released 45.57 kgCO2. Dividing this by the 1128.6 kWh of heat delivered by the heat pump (£45.57/1128.6) = 40 gCO2/kWh. Using the MyGridGB estimate this would increase to 53 gCO2/kWh.

Whichever estimate is correct, this is truly low-carbon heating.

If a 90% efficient gas boiler had delivered this heating it would have released 251 kgCO2.

So the heat pump emitted around 20% of the CO2 which would have been emitted by using a gas boiler.

Even if we had not used the battery to allow the use of cheap-rate electricity, the carbon savings would still be dramatic.


As I wrote the other day, I am relieved to find that all the investments I have made in my home are working – the house is emitting just a small fraction of the CO2 it emitted previously, with absolutely no loss of quality of life.

  • The triple-glazing and external wall insulation reduce heating demand by half
  • The solar panels are still delivering 20% of our electricity demand in November!
  • The battery allowed me to time-shift the use of low-carbon off-peak electricity.
  • And the air source heat pump operated with more than 300% efficiency even on the coldest says.

Time to snuggle up…

The Role of a Battery in Meeting Winter Electricity Demand

November 26, 2021

Click Image for a larger version. The 23rd November 2021 was a glorious day in West London.

Friends, these last few days of November have finally been cold enough for all the parts of my plan for a low-emissions home to come together.

  • The triple glazing and external wall insulation are reducing the heating required to keep the house comfortable.
  • The air source heat pump is reducing the amount of energy required to produce that heat.
  • The battery is charging itself at night on cheap electricity and running the house during the day, only running out of charge in the evening.
  • And on sunny days, the solar panels are topping up the battery during the day!

And so far, nothing major has gone wrong!

In this article I will outline the role of the battery in all of this by looking at how the battery operated during two recent quite chilly days, one sunny and one dull.

Battery: Capacity

The battery is a Tesla Powerwall 2, with 13.5 kWh of storage. This is large for a domestic battery but still only 25% the capacity of a typical mid-size EV battery.

We use about 10 kWh/day of electricity for day-to-day living, and in summer the battery and solar PV allow us to operate off-grid for almost 6 months.

But now we are heating our home electrically, and expect to have to provide around 50 kWh/day of heating – something achieved using around 15 kWh/day of electricity using an air source heat pump.

So winter demand is expected to be around 25 kWh/day of electricity – roughly twice as much electricity as the battery can store.

So in winter the main role of the battery is to allow us to buy cheap rate electricity, and then use it later in the day, minimising the cost of the extra electricity we are using for heating.

Battery: Losses

Intrinsically, battery cells can only store DC electricity, but the Powerwall needs to store and discharge AC electricity.

So the Powerwall, includes an AC to DC converter on its charging input and a DC to AC converter (called an inverter) on its output. Overall, it promises to deliver back 90% of the electric energy stored in it.  Tesla call this ‘90% round trip efficiency’.

Additionally the battery requires power to maintain itself: it needs to keep its internal controllers going, and to operate a heating and cooling system to maintain the battery at a suitable temperature during charging and discharging in order to ensure the longevity of the battery. Tesla guarantee that the battery will retain 80% capacity after 10 years.

These two effects – round trip efficiency and self-consumption – make it difficult to estimate the so-called ‘state of charge’ of the battery (SOC) i.e. how ‘full’ it is. This is because energy stored in the battery seems to slowly disappear and it is not clear quite how to account for that.

So I have crudely approximated both effects by a simple average power loss in the battery, typically between 100 W (2.4 kWh/day) and 150 W (3.6 kWh/day). I have then adjusted this rate to make sure the Powerwall is ’empty’ at approximately the observed time. The Powerwall also reports what ‘it’ thinks it’s internal state of charge is.

Household Demand

The graph below shows that household demand on Tuesday 23rd November 2021 and Wednesday 24th November 2021 was broadly similar.

Click Image for a larger version. Household demand on the 23rd and 24th November. Most of the roughly 1 kW of demand is from the heat pump (~0.6kW).

Normal non-heating household demand is typically ~10 kWh/day but on these days, the house used ~24 kWh of electricity with ~15 kWh being used to operate the heat pump to provide hot water and space heating:

The battery itself seemed to consume about 3 kWh with probably ~1.3 kWh of that being round-trip losses, and 1.7 kWh (~70 W) being self-consumed to maintain its temperature and operating system.

State of Charge

The graph below shows both my estimate of the state of charge of the battery (i.e. how full it is ) through each day. Also shown as red dots is the self-reported state of charge of the battery.

Click Image for a larger version. The estimated state of charge (SOC) of the battery (kWh) during the 23rd and 24th November. The green line is my estimate and the red dots show the battery’s self-reported estimate. Also shown is the solar power (kW) during the day which was substantial on the 23rd – enough to meet household demand and partially re-charge the battery – but not very substantial on the 24th. The green dotted line shows how the battery would discharge if there were no ‘solar boost’

The overall uncertainty in my estimate of the state of charge is quite large – perhaps about 0.5 kWh – as shown by the fact that on the 23rd the battery apparently ‘overfills’ and the 24th it ‘underfills’. However my estimates do coincide reasonably well with the Powerwall’s own estimates.

  • On the 23rd, the battery was initially not quite empty and then charged at approximately 3.6 kW using off-peak electricity available from 0:30 to 04:30. It discharged during the day at around 1 kW, and then after being boosted by 7.4 kWh of solar PV, ran out of charge between 23:00 and 24:00. That evening I was obliged to purchase only ~ 0.5 kWh of full-price electricity.
  • On the 24th, the battery was initially empty and then charged at approximately 3.6 kW using off-peak electricity available from 0:30 to 04:30. Discharging at ~1 kW as on the 23rd, but with just 0.94 kWh of ‘solar boost’, it ran out of charge between 18:00 and 19:00. That evening I was obliged to purchase ~5.4 kWh of full-price electricity!

Because the battery had been fully charged and emptied on the 24th, I could evaluate its round trip efficiency using data reported by the battery itself. It reported that it had received 13.4 kWh of charge and discharged 12.1 kWh – just over 90% of the charge, which is in line with specification. But I do not think this figure includes ‘self-consumption’ which I believe appears as a ‘phantom’ domestic load.

Click Image for a larger version. Grid use during the 23rd was almost exclusively during the cheap rate when the battery was charged while using cheap-rate electricity operate the domestic load, including a dishwasher cycle. On the 24th the battery ran out in the early evening and around 5.5 kWh of full price grid electricity was used.

This behaviour makes sense in general terms. If the battery is full (13.5 kWh) at 4:30 a.m., and demand is around 1 kW, then we would expect the battery to run out 13.5 h later i.e. around 6:00 p.m.. Each kWh of solar generation delays that time by an hour.

Costs & Carbon Emissions.

In the Table below I have summarised as best I can the costs and carbon emissions arising from the house on these two days and compared them with two alternative situations:

  • The first imagines that we had the same solar PV, and heated the house with a heat pump but with no battery.
  • The second imagines that we had the same solar PV, but heated the house with a gas boiler.

Click the image for a larger version of the table. Entries associated with burning gas are coloured in blue. Calaulations are bsed on Electricity prices of 5p/kWh (off-peak) and 16.3 p/kWh (peak) and gas prices of 3.83 p/kWh.

The analysis and evaluation of the alternative scenarios is tedious beyond measure, so allow me to simply summarise.

In terms of money:

  • The battery saves lots of money every day, with or without the ‘solar boost’.
  • It makes the use of an ASHP not only low-carbon, but also very cheap.

In terms of carbon dioxide emissions:

  • The battery reduces ‘my’ carbon dioxide emissions by optimally capturing solar energy.
  • On dull winter days the carbon dioxide emissions would be similar with or without a battery.
  • Using an ASHP – with or without a battery – drastically reduces carbon dioxide emissions compared to using a gas boiler.


As I have commented before, the battery is primarily a financial investment.

In summer:

  • The battery it allows me to fully utilise the solar PV, taking the house essentially off-grid for around 6 months.
  • This saves me hundreds of pounds a year.

In winter:

  • The battery it allows me to fully utilise the small amount of solar PV available, and to time-shift the use of off-peak electricity.
  • This also saves me hundreds of pounds a year.

The apparent savings in carbon dioxide emissions associated with using a battery are illusory and in fact the battery is really an additional electrical item using power and causing further emissions.

The solar PV provides low-carbon electricity, but without a battery, any mismatch between production and our domestic use of electricity is expensive.

  • Overproduction of electricity is exported to the grid and I get 3p/kWh.
  • Underproduction of electricity requires me to import from the grid and I must pay 16.3p/kWh.

So without a battery, the low carbon dioxide electricity is still used and still helps the planet, displacing generation from gas-fired power stations. But I don’t get much benefit for it!

What if… we had taken Climate Change seriously in the 1990’s?

November 21, 2021

Friends, reading comments about COP26, I noticed that many people viewed it as ‘an apocalyptic failure’, while others viewed it as ‘meaningful progress’.

In the face of such different interpretations, I began to wonder what unequivocal ‘success’ at COP26 would have looked like?

And reflecting further, I began to wonder what the world would be like now if humanity had begun taking climate change seriously in the 1990’s.

I picked the 1990’s because the basic reality of the phenomenon became mainstream in 1980’s. And so considering (say) 1995 as a starting point would have given  a clear 10 to 15 years for debate about what actions humanity might best take. Remember the landmark Kyoto Protocol was signed in 1992.

And my conclusion surprised me.

I came to the conclusion, that although I wished profoundly that we had acted differently back then, in fact I don’t think that we would be substantially further along the road to a renewable-energy economy than we are now.


Well it’s all hypothetical and I could be wrong, but I am probably not completely wrong.

Anyway: that’s what this article is about: how much lower might CO2 concentrations be now if we had acted earlier?

Carbon dioxide

The measure of success or failure in our attempts to avoid the most dangerous impacts of anthropogenic global warming is the atmospheric concentration of carbon dioxide (CO2).

In 1995 atmospheric CO2 concentrations were ~360 ppm, and they have risen in the intervening 26 years to roughly 415 ppm.

Click the image for a larger version. The ‘Keeling Curve’ showing the rise in atmospheric concentration of CO2 since 1959 as measured at the Mauna Loa observatory: link. Also shown are three alternative trends that might have happened. The green dotted line shows what would have happened if annual emissions had remained the same as in 1995. The orange dotted line shows what would have happened if annual emissions had fallen since 1995. The red dotted line shows what would have happened if annual emissions increased even faster than they actually did..

If in the years since 1995 we had managed to stop growth in emissions i.e. just emit the same amount of CO2 every year – then the Keeling curve would have continued in roughly a straight line – shown as a green dotted line above.

In that case, the atmospheric CO2 concentrations now would be around 400 ppm. This is 15 ppm lower than the situation now – which would represent a saving of 7 years emissions.

But could we reasonably have hoped for better? Surely if we had started back then we would have hoped for a reduction in annual emissions by now?

I am not sure. The world population has increased from ~5.6 billion to ~7.6 billion over the period since 1995, and it is not obvious to me that we could reasonably have hoped for a significant reduction in annual emissions.

And this is especially the case since many of the technologies which are today commercially available at scale did not exist back in 1995.

Let me give you some examples.

Wind Turbines

I could be wrong, but I think the idea that 9.5 MW wind turbines could built and operated would have been inconceivable to most planners back in the 1990’s. The idea of having vast windfarms at sea would also have seemed deeply unrealistic. No one would have made a national plan that depended on an untried technology.

So early turbines were built and incorporated onto electricity grids, and the cost was subsidised to enable the industries to grow.

And as experience with the technology grew post 2000, we might conceivably have deployed wind power faster than we actually did. But faster deployment would have led to the installation of more lower power turbines which we would now be scrapping.

The massive and ongoing reduction in cost arises from installing monsters 10 times as large as turbines installed a decade ago.

It is probably true that we are behind the best possible wind deployment curve – but we are probably only a small number of years behind that curve. We are certainly not 25 years behind the most optimistic possible curve.


I could be wrong again, but I think the idea that solar PV generation could be built and operated on the present scale and at the present cost would have been inconceivable to most planners back in the 1990’s.

As with wind, initial solar generation in the UK was heavily subsidised and that allowed an industry to develop, both in the UK, but most significantly in China where most panels are made.

In 1995 I am pretty sure that the idea that rooftop solar panels in the UK (!) might generate more than 1 GW on a sunny day would have been inconceivable.

The idea that we would have Solar PV farms generating 9 GW on sunny days is still astounding.

The idea that the cost of solar PV would fall as low and as fast as it has could not have been reasonably predicted by anyone.

And the impact of this in other countries will be more profound than in the UK.

But it is hard to imagine that the industry could have grown and prices fallen substantially faster than they already have.

Electric Vehicles

Mainstream manufacturers of internal combustion engine vehicles (ICEVs) had the technologies to make profitable and useable electric vehicles (EVs) for decades e.g. the Chevrolet EV1, Volt, and Bolt. But for a variety of reasons they resisted the technological shift.

Despite mandates to produce EVs, the legacy automakers were so committed to ICEV technology – which was where most of their profits and commercial advantages lay – that they failed to commit to the change.

So it is possible that if mandates had been maintained, the shift to electric vehicles could have begun much earlier.

But the fact that ICEVs are now becoming obsolete before our eyes is due – in my opinion – to one person: Elon Musk.

Tesla was started in 2003 and Musk joined in 2004 (link) with the aim to “expedite the move to sustainable transport and energy, obtained through electric vehicles and solar power”. The Tesla Roadster (of which only 2450 were sold) was the first production car to use lithium ion batteries. The scale and pace of the change instigated by Tesla exceeds anything that the legacy carmakers have ever achieved. And the shift to EV manufacturing in China is staggering.

And so although an earlier mandate to shift to EV’s might possibly have moved things slightly faster, my guess is that the giga-factory concept which has been invented by Tesla, will ultimately drive the switch to EVs faster than the legacy automakers – or government planners – could ever have conceived.


Batteries will play a key role in a sustainable energy civilisation. And their performance in terms of longevity, capacity and safety, has improved astoundingly in recent years.

It is not obvious to me that it would have been possible to scale up battery production substantially faster than what has actually happened.

Indeed progress has been so rapid that it is perfectly possible that had we started earlier, we might well have started using battery technologies that were barely fit for purpose.


Several aspects of the renewable energy revolution – such as cheap solar panels, EVs, or battery manufacture – rely on sophisticated and world-leading manufacturing in China.

It is not obvious that the manufacturing capacity the world requires could have been made available much faster than it already has.

The Internet & Computers

I find it hard to believe, but the internet barely existed in 1990: today it is ubiquitous

Network speeds are fast enough that video conferencing is now routine, and the speed of the network enables many technological innovations which allow us to cope with the unpredictability of energy generation using renewables.

It is not obvious that advances in these fields could possibly have developed any faster

Nuclear Power

One area where we might conceivably have achieved much more than we have is in the area of nuclear power.

I remember at the time being dismayed when the Blair government’s refused to engage with the issue, and basically let the UK civil nuclear industry die.

I personally would have been happy if the UK had built a few more nuclear power stations over the past 25 years. And it would have been quite conceivable to have built out nuclear power much more widely on an international scale.

I don’t know how much we might reasonably have hoped that this would lower CO2 emissions. Perhaps by 10% per year globally? Perhaps by a more significant fraction for the UK.

But, if we had adopted a very pro-nuclear solution to our energy problems, then there may have been other unintended consequences.

The capital requirements of a large nuclear power station (~£20 billion) are so high – even for governments – that they could easily have deprived wind and solar energy projects of the capital they needed for their initial growth.

As we look at the field now, we can see that perhaps a small number of truly gigantic nuclear power stations are possibly not the best implementation of nuclear power. And although nuclear power does have low CO2 emissions, it is expensive – something like 10 times more expensive than wind energy.

From today’s perspective, it is likely that smaller, modular reactors (e.g. this, or this, or this) will be cheaper, safer, and contribute to the grid more flexibly.

If we had already committed to nuclear power in a big way, it is quite possible that not only would the growth of renewables have been squeezed, but the price of electricity might be substantially higher.


Friends, I would like you to be sure about what I am suggesting, and what I am not suggesting in this article.

  • I am not saying that this is the best of all possible worlds.
  • I am not saying that actually our delay in acting has been ‘for the best’.
  • I am not saying we should be blasé about our progress in addressing Climate Change.

But I am reflecting that:

  • This is not the worst of all possible worlds.
  • Our past delays mean that we have not acted on climate change at the fastest rate possible.
  • But the last 25 years have seen the growth of new industries and technologies at truly astounding rates.

The change in public consciousness has also been profound.

And I am reflecting that although it would obviously have been better to address the challenges of Climate Change sooner, by chance some aspects of recent progress has been so rapid, that we may not be a full 30 years behind where we might have been.

Any delay is regrettable, but even if we had acted sooner, we would already have already committed ourselves to at least a century of climatic challenge.

My conclusion is that it is really is not too late to act on this now – but it is getting later every day…

A weekend away…

November 15, 2021

Click image for a larger version. The view from Winchelsea village is splendid, and in my opinion, all the better for the views of Camber Sands wind farm and Dungeness nuclear power station.

Friends, I was just chatting with my heat pump (HP – he/him) the other day, when HP made an odd request:

HP: “Michael, would you and Stephanie mind going away for a few days?

I was honestly a bit alarmed. This is our home and although we have taken HP in and hope to provide him with a loving home, we really don’t want him to take over! But then HP explained.

HP: “I’ve got this issue with my Weather Compensation. And when you and Steph are inside the house, switching things on and off, cooking, and just generally heating the place with your fleshy metabolisms, I just can’t see if it’s working or not!”

It still seemed a bit previous, but after discussing it together as a family, Steph and I agreed to go away for a long weekend in Winchelsea and to leave HP to look after the house.

When we came back…

On returning, we were worried we might find the house trashed – you know what youngsters are like.

But in fact the house was pristine and HP had kept the temperature stable all weekend – all by himself!

And to cap it all, HP had left us with a whole pile of beautiful data. I was so proud that I told him I was going to share it on my blog.

Weather Compensation

Looking at 72 hours of data without us messing things up, it was clear that HP really did have good weather compensation.

As the external temperature went down, HP responded by increasing the temperature of the water flowing in the radiators.

The effect of HP’s efforts was to keep the internal temperature within ± 0.3 °C of 20.5 °C over most of this period.

And it is also worth noticing how low the flow temperatures were, peaking at just under 34 °C when the external temperature fell to 7°C.

Click image for a larger version. Graph showing the external temperature (°C) and the flow temperature (°C) of water in the radiators over the 72 hours starting at midnight on 10/11 November 2021. The grey dots show the data every 2 minutes and the green curve shows the data averaged over 1 hour. Notice how when the external temperature falls, the flow temperature rises. This is called ‘Weather Compensation’.

One can see the ‘Weather Compensation’ effect more clearly if one plots the hourly-averaged flow temperature versus the external temperature.

Click image for a larger version. Graph showing the average flow temperature (°C) of water in the radiators versus the external temperature (°C) over the 72 hours starting at midnight on 10/11 November 2021. The blue dashed line shows the nominal ‘Weather compensation’ curve that HP was trying to achieve.

The blue dashed line on the graph above shows the nominal ‘Weather Compensation Curve’ that Vaillant had suggested that HP should aim for – a curve which Vaillant helpfully label with the meaningless number ‘0.7’.

It suggests that even when the outside temperature reaches 0 °C, we should be able to stay cosy with a flow temperature in the radiators of around 40 °C.

In different houses, different compensation curves can be chosen to match the heat pump output to the heat losses from the house.

And over the period of 3 days, HP used just 19.8 kWh of electricity (275 W) but delivered a whopping 83.7 kWh of heat (1.16 kW) – an average coefficient of performance of 4.2.


HP really enjoyed his time at home alone – he seemed so proud that he had mastered the Weather Compensation which can be quite tricky even for clever heat pumps like HP.

It really is great having HP in our family, and Stephanie and I enjoyed our time away too.

I really wish everyone could have a heat pump as considerate as HP.

P.S. In case you haven’t been following, HP is a 5 kW Vaillant Arotherm plus heat pump.

Powerwall: Assessment of degradation of storage capacity after 8 months

November 9, 2021

Friends, it is now 8 months (235 days) since we installed the Tesla Powerwall domestic battery (link).

And one of the questions I am most commonly asked concerns its likely lifetime.

All batteries degrade over time, even Tesla’s, but the practical question is “How much degradation occurs and over what period?

By chance last night (8th/9th November) I had the opportunity to assess this degradation and, in case you are short of time and need to do something more important than read this blog, my estimate is that the battery degradation so far is immeasurably small.

For those of you still with me, allow me explain how I made the measurement and what the results suggest regarding the battery lifetime.

Powerwall Control

The Powerwall battery is apparently controlled by an ‘App’ on my phone.

The ‘Tesla App’ has pleasing – almost compulsively enticing – graphics showing power flowing to and fro from the grid; the battery; our house; and the solar panels. It’s a really engaging interface. It also allows detailed data to be downloaded for analysis.

Click image for a larger version. Screenshots from the Tesla ‘App’ showing the charging and discharging of the battery, its state of charge, and the overall operation of the battery system.

However, I say the battery is ‘apparently’ controlled by the App because in reality the battery is controlled and monitored 24/7 over the internet by Tesla. And Tesla give me only limited control via the App.

Currently the battery is set to ‘Time-Based Control‘ which charges the battery from solar PV when available or cheap rate electricity if required.

And Tesla make the choice about how much charge to take overnight based on it’s estimate for how much solar power will be available the following day.

I don’t have the algorithm it uses, but it seems to do a fair job.

The reason I consent to this egregious interference with my liberty is that in return for ceding control, Tesla promised that the battery would retain 80% of its specified 13.5 kWh capacity (i.e. 10.8 kWh) in 10 years time i.e. after a nominal 3650 partial charge cycles.

I think this is a guarantee worth having and so I submit to the Tesla-Brain.

In fact since I bought the Powerwall I believe this guarantee has been degraded to 70% after 10 years – which suggests it really is quite a tough specification.

Expected Battery Degradation

Various reports on the web, and Tesla’s 2020 environmental impact report, indicate that Tesla car batteries seem to retain around 90% of their range after 200,000 miles (320,000 km).

Click image to see a larger version. Excerpt from Tesla’s 2020 Environmental Impact Report showing roughly 10% reduction in EV range after 200,000 miles.

It’s hard to know how that colossal range would translate into the 3650 partial charge and discharge cycles of a domestic battery over 10 years.

In part it depends strongly on the range of the charge-discharge cycles. Charging and discharging over the middle of a battery’s range – between say 10% and 90% – is relatively benign. But rapidly charging and discharging from 0% to 100% degrades battery capacity. This is why Tesla want to have control over the battery.

My thought when I bought the battery, was that domestic service would be generally less stressful than service in a motor car. Why?

  • The Powerwall has it’s own re-circulating fluid temperature control and does not need to operate in the climate extremes of a car battery.
  • A Tesla car battery is about 4 times larger than a Powerwall’s 13.5 kWh, but the maximum EV discharging rates – which can affect battery life – are up to 25 times higher than the Powerwall’s transient maximum of 7 kW.

Other reports (link) suggest that Powerwall degradation might be considerably faster than for EV’s.

However, the general pattern of battery degradation (reflected in the figure above) is that the greatest rate of degradation is at the start of the service life of the battery.

If my Powerwall were to degrade linearly to 80% capacity over 10 years (120 months) then after 8 months I might expect to see 0.18 kWh decrease in capacity. Small, but possibly detectable.

What did I measure?

By chance last night the battery ran out just before midnight, so I knew it had zero ‘state of charge’.

Additionally the Tesla-Brain decided to fully charge the battery over the four hours of cheap rate (5p/kWh) electricity starting at 00:30. So I was able to observe a full charge from empty.

The graphs below (using data downloaded via ‘the App’) show what happened. Note the data only have a time-resolution of 5 minutes.

Click image for a larger version. Graph showing the charging of the battery from 00:30 at approximately 3.6 kW and the discharging of the battery after 04:30 at approximately 300 W to meet domestic demand.

Using the charging rate and the time I can work out the ‘state of charge’ of the battery and compare this with the specified capacity.

Pleasingly, the maximum state of charge appeared to correspond closely with the initially specified capacity.

Click image for a larger version. Graph showing the calculate ‘state of charge’ of the battery during charging from 00:30 to 04:30. Within the (considerable) uncertainties of this measurement, the maximum state of charge is closely in line with its specified original capacity.

Interestingly, while the battery was charging, the Tesla control circuitry also used cheap-rate electricity to run the dishwasher and top up the domestic hot water using the heat pump.

Click image for a larger version. Graph showing the household demand from midnight to 06:00. The battery was charging in the background during these high power events.


First of all, some caveats:

  • All these measurements are self-reported by the Powerwall, and so should rightly be subject to sceptical interpretation.
  • The data have limited resolution both in power and time.

However, when I have been able to check the reported values against independent measurements – e.g. for estimates of the energy reaped from the solar panels each day – I have found them in close agreement at the level of 0.1 kWh.

So taking these measurements at face value, I find no detectable degradation in battery capacity after 8 months or 235 days.

  • If the battery capacity were degrading linearly over time to 80% of initial capacity after 8 years I would have expected to see 0.18 kWh decline in capacity.
  • If the battery capacity were degrading faster than linearly – as it plausibly might – then I would have expected to see perhaps 0.3 kWh or 0.4 kWh degradation.

Obviously I will re-visit this issue at some point in the future, but the fact that there is no detectable degradation so far suggests that the retained capacity after 10 years may indeed exceed 80%.

Which would be nice.

A Year of Solar Energy

November 8, 2021

Friends, it’s just coming up to the anniversary of the installation of solar photo-voltaic (PV) panels at Podesta Towers in Teddington. So I thought it might be interesting to see what a year of generation has brought.

First I’ll describe the installation; then explain what I expected (or at least hoped for); and then outline what has actually happened together with a discussion of the role of our domestic battery.

And finally I’ll remind you – and myself – of how solar panels fit into to my efforts to reduce carbon dioxide.

But in case you’re short of time, here are the salient points.

  • 12 solar panels generated just over 3500 kWh of electricity which is close to last year’s total domestic consumption ~3700 kWh.
  • This will have avoided emission of ~0.8 tonnes of carbon dioxide this year.
  • On the sunniest days the system generated ~25 kWh/day and in mid-winter average generation was ~ 2kWh/day which compares with around 10 kWh/day of non-heating electrical use.
  • When used with a battery, we were substantially off-grid for around 4 months.

1. The Installation

I described the installation in some detail here, but briefly it consists of 12 panels from Q-Cells, each 1.0 m x 1.7 m in size with a nominal generating power of around 340 W. I think this year’s versions are already more powerful!

Click for a larger image. Google Maps view of my home showing the shape and orientation of the available roofs. And photographs before and after the installation.

The installation cost £4,230 pounds which did not include the cost of scaffolding which was already up on the house at that point.

I chose Q-cells panels because I liked their completely black appearance; they seemed adequately specified; they were readily available; and they were cheap – around £130/panel if I recall correctly.

I later found out that they have pleasingly low embodied carbon dioxide, less than 1.6 tonnes for this installation, and so the embodied carbon dioxide payback time will be around 2 years. Q-cells also score highly for the not producing toxic waste.

2. What did I hope for?

The quotation from local installer GreenCap Energy included an estimate of the expected output – 3780 kWh/year.

But rather than trust the installer I downloaded data from an EU project that cleverly allows one to estimate how a particular solar installation with an arbitrary location and orientation would have performed hour-by-hour over the entire period from 2005 to 2016. This data suggested I might reasonably expect 3847 ± 173 kWh/year.

Click for a larger image. Estimates from an EU re-analysis project of what my solar panels WOULD have generated over the years 2005 to 2016. Year-to-Year variability appears to be about 5%.

I then averaged these 11 years of hour-by-hour data to yield my estimate for the expected monthly performance. They are shown as yellow dots on the graph below.

Click for a larger image. Expected generation in kWh per day. The yellow dots are the monthly averages of the estimated generation from 2005 to 2016. The green dotted line is a crude guess based on a ‘sine-squared’ function. The red dotted line shows our typical average daily electricity consumption ~10.5 kWh/day.

One feature of the simulated data is that the peak generation is expected to occur from April to July – a range which is not centrally arranged around the longest day (June 21st).

3. What happened?

Solar generation was broadly in line – but a little lower – than expectations. But the day-to-day and week-to-week variability was much greater than I had appreciated.

This variability makes it hard to plot readable graphs because they look chaotic! So let me introduce the data one stage at a time.

Looking at the monthly averages (below) we see that most months were close to expectations, but April was especially sunny, and August was a bit disappointing. So far, November has been a brighter than normal. Please note, the December data is from December 2020, because the data from December 2021 is not yet available.

Click for a larger image. Comparison of monthly averages of actual solar generation (green dots) with expected generation in kWh per day. (yellow dots). The black error bars show the standard deviation of that months daily data.

Now let’s additionally plot the daily data.

Click for a larger image. Similar to the graph above but now additionally showing the actual daily solar generation (kWh/day)

A few features of the daily data are really quite remarkable.

  • Firstly, the day-to-day variability is large. This means that the solar generation on a given day is almost no indication of the likely generation on the next day.
  • Secondly, the peak generation of around 25 kWh/day can occur in either April or July despite the substantial differences in day length and solar path.
  • Thirdly, even in mid-summer there can be runs of several utterly miserable days with very little solar generation.

If we average the daily data over a week, (see below) then we still see variability – deviations from the nominally-expected generation – which deviate from the expected generation consistently over periods of up to 3 weeks

Click for a larger image. Similar to the graph above but now additionally showing the ±3 day average of the generation as pink or purple lines.

Another way to look at the data which emphasises the trend more strongly than the variability, is to show cumulative generation through the year.

Click for a larger image. Cumulative generation (kWh) shown as a blue line against nominal expected generation.

Actual generation (just over 3500 kWh) is about 8% lower than I expected, but I am not especially surprised.

It could be that 2021 was a ‘a bit dull’ over the summer months when most generation takes place, or because I had incorrectly allowed for losses at the inverter – which converts DC electricity into AC mains electricity.


The PV panels were installed in November last year and to our surprise they immediately made a measurable difference – reducing our use of electricity from the grid by around 2 units per day.

Click image for a larger version. Daily electricity usage (from a smart meter) before and after solar panel installation.

However, it was not till our battery was installed in March 2021 that the transformational power of solar became apparent. Within a couple of days, the household went ‘off-grid’ and remained off-grid for around 80 days.

Click image for a larger version. Daily electricity usage (from a smart meter) since solar panel installation. Install the panels led to a small reduction in grid use. Consumption rose over Christmas. Consumption began to fall as we entered spring, and then fell to zero once we installed the battery. As we enter Autumn and Winter grid consumption is rising because since July we are using electricity to power a heat pump which heats the house and provides hot water. The bold green line shows the daily consumption averaged over ± 1 week.

As we enter autumn and winter, the solar cells still contribute significantly – 48% of our electricity in October was solar. But generation will be just around 2 kWh/day through November, December and January.

Since July, the ASHP installation has been using ~1.5 kWh of electricity a day to provide ~4 kWh of domestic hot water.

Now (November) the ASHP is providing space heating within the house, and in the coldest weather (~0°C) I expect  this will require an additional 15 kWh/day of electricity to provide ~50 kWh/day of heating. Most of this electrical energy will be downloaded at cheap rates overnight.


The PV and battery system has been installed to reduce carbon dioxide emissions from the house. They are the part of a suite of measures to reduce heating demand, eliminate gas use, electrify heating and increase the use of renewable energy.

Together they have dramatically reduced the running costs of the house.

The graph below shows expected household carbon dioxide emissions (not including consumption or travel) over the period up to 2040.

Click image for a larger version. Anticipated household carbon dioxide emissions (not including consumption or travel) over the period up to 2040. The red line shows what would have happened if I had made no changes. The green line shows the expected outcome.

In the short term, all the actions I have taken have made things worse!

The embodied carbon dioxide in the solar panels, insulation, batteries, and heat pump amounts to ~11.5 tonnes, and this ‘debt’ will not be re-paid until the end of 2023.

I would love to add more solar panels, but I am resolved to hold off until my existing carbon dioxide debt is re-paid.

Overall, I hope you can see that the solar panels are central to the plan to reduce anticipated emissions by 60 tonnes by 2040


Re-visiting the “washing-up dilemma”

November 6, 2021

Friends, I can remember a time when the main question about washing up was: “Who is going to do it?”

But dishwashers changed all that. At first there were questions about whether dishwashers or regular washing-up was technically superior.

However nowadays, dishwashers do a very creditable job of cleaning crockery and cutlery – and in what was a devastating blow for the British Tea Towel industry – they also do a fine job of drying.

[Aside: If you would like to see how dishwashers work – check out these Technology Connections videos:

But this then raised the thorny question of whether hand-washing or using a dishwasher was cheaper.

And now of course, the BIG question is which is greener i.e. has the lowest associated carbon dioxide emissions. This article is an attempt to answer this question.

Energy and Carbon Emissions

In her bookEnergy and Carbon Emissions: the way we live today“, Nicola Terry does a fine job of answering such questions.

The book is a gold mine of information on all aspects of our current ‘carbon problem’.

And she looks at this problem in Chapter 7: Should I get a more efficient X?

She concludes that:

“Washing dishes by hand probably generates less carbon [dioxide] emissions if you heat water by gas – even less if you have a solar panel for hot water

And in characteristic and admirable detail, she explains the assumptions underlying her conclusion.

But as His Bobness might have commented… things have changed.

2011 versus 2021

The energy landscape has changed significantly since 2011 when Nicola Terry published her book.

  • Carbon dioxide emissions from electricity are now 235 gCO2/kWh rather than 545 gCO2/kWh in 2011.
  • Carbon dioxide emissions from gas are still roughly 200 gCO2/kWh assuming ~90% boiler efficiency: the same as 2011.
  • In our house,
    • Hot water comes from a heat pump which produces hot water around 3 times more efficiently than a gas boiler.
    • In summer, practically all our electricity comes from solar panels with no proximate emissions: 0 gCO2/kWh
    • In winter, we set the dishwasher to run on cheap-rate electricity which typically has 10% less CO2 emissions than daytime electricity.
  • In other houses,
    • Hot water might come from solar thermal panels.

So given this complexity, can we definitively compare the environmental impact of using a dishwasher and washing up by hand?

Well I have made an attempt – and below I will run through some example calculations and then I’ll collate a more extensive set of calculations at the end.


Taking our recent Bosch model as typical, a dishwashing cycle seems to typically consume around 12 litres of water and around 1 kWh of electricity.

Comparing the consumption for different modes of operation, it is clear that nearly all this energy is used to heat the water. At 70 °C – way more than your hands could stand – fats within food residues soften considerably making them easier to remove in the water spray.

In the winter,

  • Carbon dioxide emissions associated with running a cycle in 2021 on average would have been 235 gCO2 using electricity direct from the grid.
  • At peak demand (4 pm to 7 pm), the emissions might be higher – but if run in the early hours of the morning – emissions might be ~10% lower.

In the summer,

  • Carbon dioxide emissions associated with running the dishwasher are around zero because the electricity is derived from the solar panels.

So assuming a year is 50% summer and 50% winter, carbon dioxide emissions associated with running a single cycle in 2021 on average would be 0.5 x 235 = 118 gCO2 per cycle.

So running the dishwasher 3 times per week would result in emission of around 354 gCO2 per week.

Dishwashers do offer ‘eco’ modes using cooler water which could reduce this.

Washing-up by hand

In general, hand washing probably consumes a bit more water than a dishwasher, but let us assume in the first instance that the same 12 litres of water will suffice, and that the dishes are rinsed in cold water.

Heating 12 litres of water by 40 °C from (say) 10 °C to 50 °C – roughly the hottest water one might reasonably use – requires:

12 litres x 10 °C  x 4200 J/°C/litre = 2,016,000 joules

Or in more familiar units, 0.56 kWh. If this water were heated with a heat pump with a COP of 2.5, this would require 0.224 kWh of electrical energy.

In the winter,

  • Carbon dioxide emissions associated with 0.224 kWh of electricity drawn from the grid would have been 0.224 x 235 gCO2/kWh ~ 53 gCO2

In the summer,

  • Carbon dioxide emissions associated with 0.224 kWh of electricity from the solar PV panels would have been zero.

So assuming a year is 50% summer and 50% winter, then washing up by hand 3 times a week results in weekly emissions of 3 x 27 = 81 g CO2/week

If the water were heated by gas, the equivalent emissions would have very similar in winter, but would not be reduced in summer, resulting in emissions of about 150 gCO2/week.

And if the water were heated by direct electrical heating without a heat pump, three washes per week would result in emissions of about 158 gCO2/week.


I have carried out these calculations for a number of situations using this spreadsheet (Dishwasher Calculations). I compared:

  • Hand washing with water from a heat pump powered by solar energy and a battery for half of the year.
  • Hand washing with water from a heat pump powered by grid electricity.
  • Hand washing with water from a gas boiler.
  • Hand washing with water heated by an immersion heater powered by solar energy and a battery for half of the year.
  • Hand washing with water heated by an immersion heater using grid electricity.
  • A dishwasher powered by solar energy and a battery for half of the year
  • A dishwasher powered by grid electricity.

The figure below summarises my calculations assuming that a family washes up 3 times per week either by hand or with a dishwasher.

Click on the image for a larger version. Summary of calculations showing estimated annual CO2 emissions for a variety of ways to do washing up. See text for details.

It is clear that wash-for-wash, washing-up by hand results in lower CO2 emissions. This is especially striking when using a heat pump to generate the hot water.

The lower emissions arise primarily because the water used for washing-up by hand is cooler than the water used in a dishwasher.

However different households behave in different ways, and wash-for-wash comparisons may be unrealistic.

For example, in a small household, a dishwasher can store soiled dishes out of sight for a day or two, and so operating a dishwasher (say) 3 times a week is practical while not leaving out unsightly piles of unwashed dishes. However, if a similar household were relying on hand-washing, they might wash up once each day. Let’s call this Option#2.

Or for larger households, running a dishwasher once a day is more typically part of a family routine, whereas hand-washing for a larger family would be more likely – in my experience – to use rather more water – perhaps as much as twice as much. Let’s call this Option#3.

The results for Options 2 & 3 are shown below. Note that the vertical scales are different on each of the graphs shown.

Click on the image for a larger version. Summary of calculations for Option#2 showing estimated annual CO2 emissions for a variety of ways to do washing up. See text for details. Note that the vertical is different from the other figures.

Click on the image for a larger version. Summary of calculations for Option#3 showing estimated annual CO2 emissions for a variety of ways to do washing up. See text for details. Note that the vertical is different from the other figures.

The different options have a range of CO2 emissions that vary from ‘a few kilograms per year’ to ‘a few tens of kilograms per year’.

But looking across all the options, hand-washing generally results in reduced emissions, just as Nicola Terry had concluded back in 2011. But now the minimum emissions come with water heated by a heat pump.

Click on the image for a larger version. Summary of calculations for Options #1, #2, and #3 showing estimated annual CO2 emissions for a variety of ways to do washing up. See text for details.


My first conclusion is this: How can anyone be expected to wade through such a calculation to determine how to do the washing up!

My second conclusion is not that “Dishwashers are bad”. Indeed my wife and I own a dishwasher and use it regularly.

Personally, I am convinced of the overwhelming need for society as a whole to reduce carbon dioxide emissions, and I also feel this as an intense personal responsibility too.

Nonetheless, no decision can be entirely one-dimensional. As with many similar decisions, there are other criteria that can also be significant – criteria that have nothing to do with carbon dioxide emissions.

Off the top of my head, it could be that some people just don’t like washing up! Or it could be that some people are busy and using a dishwasher is helpful. Or perhaps that someone’s partner might just want a dishwasher for some reason, and so a dishwasher could be part of a declaration of love – an important human dimension.

But… although any decision about how to live one’s life is multi-dimensional, one of those dimensions is very likely to be the associated carbon dioxide emissions, almost no matter what the actual decision concerns.

In this case, my calculations tell me that:

  • For washing up, the overwhelming contribution to the carbon emissions is caused by heating water – so using less hot water will reduce emissions no matter how the washing up is done.
  • If using a dishwasher, it is a good idea to try to ensure that it is more-or-less full, and to think about using ‘eco’ modes.
  • And when washing up is done by hand, it is a good idea to be mindful of the amount of hot water used – and possibly use cold water to rinse.

Finally, in future years, it is possible many of these emissions may well change again.

It is planned that the carbon intensity of UK electricity should fall to 100 gCO2/kWh by 2030 – and this will reduce the emissions of any appliances which use that electricity. Additionally, it could be that dishwashers will be built which can use hot water heated by a heat pump instead of electrically heating their own water.

Finally this calculation does not include the embodied carbon dioxide emissions from either a dishwasher, a heat pump, a battery or solar panels. My guess – for what it is worth – is that assuming reasonably long lifetimes for these items, the embodied carbon dioxide emissions will not significantly alter these conclusions.


1000 days of data

October 9, 2021

Friends, those of you who work with spreadsheets may know that Excelworks out dates in terms of the number of days since 1st January 1900. Or in other versions of Excel, the number of days since 1st January 1904!

Inspired by this arbitrary choice, sometime ago I chose 1st January 2018 as ‘day 1’ for my measurements of energy use around the house.

And from this arbitrary starting date, I have been measuring for just over 1000 days!

So I thought it would be nice to summarise what has happened in the last three years, and to speculate on what the coming winter might hold.

Heating Demand

Our internal thermostats have been set to 19 °C since ‘Day 1’.

So I calculate the demand for heating as the difference between 19 °C and the average weekly outside temperature.

The heating demand (averaged over ±2 weeks) for the last 3 years is shown below.

Click image for a larger version. Graph showing smoothed temperature demand versus day for since the summer of 2018. Also shown is a projection of what the coming winter has in store if it is the same as last year.

I obtained this data from the weather station in my back garden, but you can use the wonderful Meteostat website if you don’t have a nearby station. I used Meteostat to fill in occasional gaps in the data.

So far, this autumn seems to be several degrees warmer than the equivalent period in 2020 i.e. demand is lower.

Gas Use 

Since just before day 1, I have been reading the gas meter once a week. From this I can work out the average rate at which I am using energy (i.e. average power). In this blog I express this as kWh/day.

[To convert kWh/day to watts, multiply the number by 1000/24 = 41.7 e.g. 50 kWh/day = 2083 W i.e. ~2.1 kW.]

Click image for a larger version. Graph showing smoothed temperature demand versus day for since the summer of 2018 as in the previous graph. Also shown are the dates of various interventions, and smoothed gas consumption (kWh/day) plotted against the right-hand axis.

It is clear that gas consumption roughly follows temperature demand. However the Triple-Glazing and External Wall Insulation (EWI) have reduced the gas used to meet a given temperature demand by about half.

In the summers of 2019 and 2020, gas consumption fell to roughly 5 kWh/day, most of which (around 3.5 kWh) seems to have been for hot water, with the balance being used for cooking.

Since the heat pump installation, in July 2021, the gas is only used for cooking (~1.5 kWh/day) , and this will continue until my wife and I can get our heads around installing an electric oven & hob.

Electricity from the grid 

Since just before day 1, I have been reading the electricity meter once a week. From this I can work out the average rate at which I am using electricity from the grid.

Click image for a larger version. Graph showing smoothed temperature demand versus day for since the summer of 2018 as in the first graph. Also shown are the dates of various interventions, and smoothed electricity consumption (kWh/day) plotted against the right-hand axis. Also shown is the small amount of electricity which was exported to the grid.

It is clear that electricity consumption was – through 2019 and 2020 – roughly independent of temperature demand.

Solar panels were installed at around the same time as the EWI (~day 660) but this did not substantially affect the amount of electricity we drew from the grid until we installed a battery (~day 810). After that, we drew very little electricity from the grid during the period March to September.

After the heat pump installation (~day 930), we began heating the hot water with electricity rather than gas. But heat pumps require only about 30% of the energy which a boiler would use to heat water.

Looking to the winter ahead, we expect solar generation to fall to around 2 kWh/day on average, but electricity use to rise above our normal ~ 10 kWh/day because the heat pump will be used for space heating (varying with the temperature demand) as well as hot water (~1 kWh/day).

Last winter gas consumption peaked at 50 kWh/day. If the heat pump operates with a coefficient of performance of 3 – which seems a safe guess – then this should require around 50/3 ≈17 kWh/day of electricity.

Carbon Dioxide emissions. 

From the data above it is possible to roughly estimate the corresponding carbon dioxide emissions.

I have assumed that:

  • Each kWh of gas consumption results in 0.2 kg of CO2 emissions.
    • This is fixed by the chemistry of methane combustion.
  • Each kWh of electricity imported from the grid results in 0.24 kg of CO2 emissions
    • This is an average of the last three years carbon intensity (Link).

There is some uncertainty in the figures above, but the assumptions are pretty uncontroversial. These represent estimates of actual amounts of CO2 which entered the atmosphere due to ‘my’ actions.

How one should deal with exported solar electricity is more controversial. Some people point out that because of the way electricity is ‘dispatched’, solar generation directly displaces gas-fired generation. Thus each kWh of my solar generation avoids the emission of  0.45 kg of CO2 emissions from a gas-fired station.

One might argue that such exports are therefore equivalent to negative emissions even though no CO2 is actually removed from the atmosphere.

With this assumption the daily carbon emissions are summarised in the graph below.

Click image for a larger version. Graph showing estimated household carbon dioxide emissions per day since the summer of 2018. Also shown are the dates of various interventions, and the expected emissions for the coming year. As discussed in the text, avoided emissions due to exports of electricity are counted as negative emissions.

The graph shows that – subject to the uncertainty of the projection – since 2018:

  • Winter emissions will have fallen from 25 kg/day to 5 kg/day – a 5-fold reduction
  • Summer emissions will have fallen from emitting ~3 kg/day to avoiding ~1 kg/day of someone else’s emissions.

Click image for a larger version. Table shows estimated carbon dioxide emissions in tonnes for the last three years (period July to June) along with a forecast of the emissions in the coming year.

The net effect of all these changes is gas emissions are now negligible. Electricity emissions were roughly halved by installing solar panels and a battery, but in the coming year they will probably return to roughly their previous value because of the electricity used to operate the heat pump.

There are two interesting things to note about the forecast aside from the fact that it’s a forecast and we don’t know what the winter will be like.

Firstly, this assumes the heat pump COP will be 3. My hope is that it will be better than this because the well-insulated house should require such a small amount of heating that I should be able to lower the flow temperature in the radiators to 40 °C. At this temperature the heat pump has a specified performance closer to a COP of 4 even with an external temperature of -5 °C.

[Aside. As of 9th October 2021, I am still waiting excitedly for the weather to get cold enough that the heat pump will switch itself on so I can test this! The insulation appears to be good enough that the internal temperature is still greater than 19 °C (~ 20 °C) without any heating!]

Secondly, 90% of the CO2 emissions now arise from electricity. So as the grid gets greener (we hope) in coming years, these CO2 emissions should naturally reduce. If the target 100 gCO2 emissions per kWh is reached in 2030, the overall household emissions will fall to under 0.5 tonnes.

What else?

There are still one or two things I could do to reduce household CO2 emissions. But at this point my intention is just to measure how these existing interventions perform for a year or two.

In 2018/19, household CO2 emissions comprised the largest category of ‘my’ emissions, and now they are more similar to emissions from other activities: consumption, transport and investments (i.e. my pension)

In the coming year I hope to turn my attention to these much trickier categories.

I’ll let you know how it goes…

%d bloggers like this: