Monday, October 10, 2022

Will EVs reduce our carbon emissions if we don't change behaviour?


My last post outlined my idea that switching ICEVs (petrol cars) for EVs (electric cars) is not sufficient to prevent climate change. I wanted to find a different way to illustrate this, so I created a simple mind-sized model. You can download the model and play with it yourself, and form your own conclusions. The model shows that if all new cars bought in Australia are EVs, it is not sufficient to prevent serious climate change if we maintain current patterns of car ownership and use.

If you are buying a new car anyway, it will be less harmful if it is an EV and not an ICEV. However, the best outcome is to retain your existing ICEV car and simply drive it less -- make more use of walking, cycling and public transport instead. Get an ebike or a scooter/moped. This has the added benefit of saving you a lot of money and improving your fitness.

The model

You can download the model here. I've saved it in an Excel spreadsheet for maximum compatibility. If you don't have Microsoft Excel (good on you!) you can open it using LibreOffice, which is a free Office program that runs on all computers. The following assumptions outline my thinking in the model -- these numbers are all adjustable in the model, so if you disagree with my assumptions you can try tweaking them. 
  1. There are 20 million cars in Australia, the total number of cars grows at 1% annually
  2. There are 1 million new cars bought in Australia each year
  3. The embodied CO2 emissions in a new ICEV and new EV are 16.3 and 26.9 tons respectively (as described by Volvo in their article)
  4. The average Australian car drives 12000 km / year
  5. An EV emits 20g / km of CO2 at point of use
  6. An ICEV emits 200g / km of CO2 at point of use

Based on these assumptions, I ran two scenarios:

  • Scenario 1: every new car in Australia is an ICEV
  • Scenario 2: every new car in Australia is an EV

I examined how these two scenarios compare out to 2040, in terms of cumulative CO2 emissions. Below is a graph that shows how cumulative emissions compare under the two scenarios:


  • by 2040, the all-EV scenario has CO2 emissions that are 16% lower than the all-ICEV scenario
  • In both scenarios, cumulative CO2 emissions are above 1 Gt (billion tons) 
  • After 2040, in the all-new-EV scenario, almost all cars in Australia would be EVs. If we stopped buying new cars altogether at this point, additional CO2 emissions in the all-EV scenario would be very small
  • our carbon budget is 1.3 gigatons if we want to remain below 1.5℃ of warming. We have pretty much exhausted this budget just on EVs and have nothing left for decarbonising all other aspects of the economy (in other words, we will hugely overshoot our budget, with consequent climate instability)
  • A significant limitation to the model is that not all cars in Australia drive the same amount. If we prioritised changing the highest emitting (furthest driving) cars to EVs, it would have a more significant effect than observed in this model.

Variations to the model

I tried some variants to see what would happen:
  • reducing the EV point-of-use emissions to 0 g/km doesn't significantly change the results
  • increasing the rate of EV adoption to 4 million new EVs / year reduces the total CO2 emissions at 2040 to 0.8 gigatons.
  • setting annual growth in total car numbers to 0 doesn't significantly change the results


  1. We cannot keep driving petrol cars like we do now and maintain a safe climate
  2. EVs are better than petrol cars in that they emit considerably less CO2 at time of use
  3. However, the additional CO2 emitted when manufacturing EVs (compared to ICEVs, which is already considerable) means it takes a long time (approximately 20 years) to achieve emissions-reduction by simply buying new EVs. During that time, we will exceed our carbon budget.
I know this is bad news, and that people won't want to hear it. I'm not happy about it either. 

Monday, September 12, 2022

Does the rise of EVs mean we can maintain car culture?


We must urgently and rapidly decrease our use of internal combustion engine vehicles (ICEVs). It is important for environmental reasons, but also for our security -- we are currently heavily dependent on oil imports which are likely to become much more expensive (and quite possibly less accessible) in future. Burning oil for transport is unsustainable. We don't have time, and don't have the carbon budget, to replace our existing car fleet while remaining within 1.5℃ or probably 2℃ of warming. 

In general, the EV-naysayers seem to think that we can just keep driving petrol cars forever, and the EV-yeasayers think that technology will work it all out so that we don't have to change our lifestyle. 

I think they're both wrong: EVs, while considerably less-bad than ICEVs, are not good enough to prevent catastrophic climate change if we keep the current patterns of private vehicle use. We urgently need to find another way to decarbonise transport.

If you are going to buy a new car anyway, you should definitely buy an EV. However, if you do buy one you should still try and limit car use and share your EV with others -- this will mean that more value is obtained from its embodied carbon, and hopefully you can reduce the likelihood that others will buy new cars. The only way we are likely to meet our climate goals is by sharing EVs, so this aspect is essential.


Currently, Australia has a plan to make transport sustainable. Our plan is something like:

Let's transition all vehicular transport to electric vehicles. These vehicles will be powered by renewable power which will decrease the climate impact and make transport sustainable.

The problem is that we've never really examined this plan to make sure that it will work. The main thing that most people focus on in this plan is the transition to private EVs. For these to work (by which I mean: be sustainable), we need to satisfy the following criteria:

  1. We need to be able to make EVs quickly enough to matter, so that we can quickly swap our existing car fleet over to EVs. 
  2. The process of manufacturing/distributing the electric vehicles needs to remain inside the biophysical limits that we have. The most familiar of these limits is our remaining carbon budget -- however, water consumption and minerals availability are also very relevant.
  3. EVs need to be a large-enough improvement over ICEVs that we meet our sustainability objectives (eg. the total lifetime emissions of EVs need to be sufficiently less than ICEVs that we can get back "under the curve" of biophysical overshoot)
  4. People need to be able to afford them

Let's look at each of these points in turn:

1. Can we make EVs quickly enough?

Let's get a sense of what is needed. There are approximately 900 million cars in the world globally, and about 20 million cars in Australia. It's hard to get precise up-to-date numbers, but there are about 50000 EVs in Australia (about 1 car in 400), and between 20 and 30 million EVs globally (about 1 car in 500). Currently, rates of production are doubling annually. Thus, it seems plausible that we can make enough vehicles in the next decade if this trend roughly holds.

However, currently the average age of a car in Australia is about 10 years. Hence, this process of replacement will quickly slow as we saturate new cars with EVs. From when 100% of new cars are EVs, it will take close to a decade to replace Australia's entire car fleet. So, even though we might be able to make EVs quickly, people can't afford to buy new cars to replace their existing ICEVs. This suggests that a plan that depends on transitioning private transport to EVs is unlikely to work because we can't swap them over quickly enough.
It's also not yet clear how much utility remains in an EV after 10 years' use. People may not want them.

2. How does the process of manufacturing EVs fit within our biophysical limits? How does it compare to an ICEV?

Volvo have released some helpful data about EV vs ICE embodied carbon. Here are the data, which I sourced from, page 25

I'm using these data to generalise to all EVs. We just want a rough estimate, so I think this is reasonable.

Carbon footprint of Volvo ICEV vs EV

In terms of manufacturing-related emissions. The tally looks like this:

  • Volvo XC40 ICEV: 15.7 tons
  • Volvo C40 EV: 26.4 tons

Volvo's report states that: 

Although total emissions from all phases except the use phase of the C40 Recharge are higher than for the XC40 ICE, the C40 Recharge will over the span of its lifetime cause less emissions thanks to lower emissions in the use phase. Where this break-even occurs depends on the difference in GHG emissions from the production of the car, and how carbon intense the electricity mix is in the use phase.

For all three electricity mixes in the LCA [lifecycle analysis], the break- even occurs at 49,000 (100% wind), 77,000 (current EU average) and 110,000km (global average) respectively, all within the assumed life cycle of the vehicle (200,000km).

I often hear this concept of "break-even" used in this way. It is a mental error, let me explain why.

"Break-even" is bogus

When people talk about "break-even" they are making a comparison against something. In this case, the break-even is being calculated versus the purchase of a brand new petrol car. Volvo are essentially saying "assuming that you are buying a new car anyway, if you buy an EV instead of a petrol car, once you have driven 48000 km your emissions will be lower than they would have been if you bought a petrol car". (It's a bit like a clothing shop telling you you've saved $25 when you bought a new pair of jeans that were discounted -- in reality, you've only saved $25 if you were intending to buy the jeans at full price anyway).

Below is the graph showing, over time, how the four different scenarios play out:

Expected total GHG emissions of EV vs ICEV from Volvo. Note that producing a new Volvo X40 (EV) release approximately 26 tons of carbon dioxide, and producing a new Volvo  

The dotted line shows total emissions from a new petrol car, and the three solid lines show EV emissions which very depending on the electricity source from about 40% to 80% of the petrol car's emissions.

There are two scenarios that Volvo haven't considered:

  1. keep the existing petrol car and maintain existing driving habits
  2. keep the existing petrol car and drive less
(I can understand why Volvo don't want to encourage either of these options)

From Volvo's own data, we can estimate how these two scenarios would compare to the scenarios they explore. In the following figure, the first four rows of data are the same as in the image from Volvo's report. The bottom two are estimates, based-on Volvo's data.

If we compare against the scenario of "keep existing petrol car and existing driving patterns", then EV "break-even" looks like follows:

  • global electricity average: 270000 km*
  • EU electricity average: 190000km*
  • wind power only: 125000 km
If we compare against the scenario of "keep existing petrol car and drive half as much", then EV "break-even" occurs like follows:

  • global electricity average: never*
  • EU electricity average: never*
  • wind power only: 250000 km*
(I have put a star next to some of these, which indicates that payback will occur outside the warranty on the battery, which is typically 8 years or 160000 km)

These data are shown in the following exploratory time series model. The first four lines are the same as those in Volvo's figure (above). The additional two lines correspond to the "keep existing car" and "keep existing car, drive half as much" scenarios, as described. It shows the "break-even" between "keep car" and "new EV (wind)" occurring at 125000 km, and the "break-even" between "new EV (wind) and "new XC40" (petrol) occurring at 48000 kms. 
Note that the "keep car, drive less" drives 1/2 the distance of the other scenarios (i.e. where the X-axis is labelled 200000 kms, it has driven 100000 kms). This is the least complex way to show the difference in behaviour in an easy-to-understand figure.

In other words, even if we buy an EV, drive the same, and charge it with 100% windpower, it takes 125000 km of driving until we have "repaid" the carbon emitted in its manufacture. In all other cases, environmental "break-even" will not occur during the warranty period of the battery.

To me, this suggests that two of the criteria listed at the start of this essay are not met by transitioning our private car fleet to EVs -- EVs are too costly (in environmental terms) to manufacture, and they don't help reduce carbon emissions enough (largely because of their environmental cost to manufacture). Because of their significant up-front environmental cost, we can't make enough of them to replace every petrol car with an EV. 
EVs are great, but we just can't replace every petrol car with an EV.

3. How many cars can we replace with EVs?

As mentioned earlier, according to Volvo, the manufacture of each EV releases 26.4 tons of CO2.  
There are 20 million registered cars on the road in Australia. In 2019, we had less than 3.3 gigatons of carbon budget remaining to stay under 2℃ and less than 1.3 gigatons to stay under 1.5℃.
To replace those 20 million registered cars with EVs would release 20M*26.4 = 528M or ~0.5 gigatons of CO2, which is about half of our total carbon budget if we want to stay below 1.5℃ of warming.
That is almost 1/2 of Australia's carbon budget, just to manufacture a new EV to replace each existing ICEV. 
It doesn't allow us to build the extra electricity generation capacity we will need to actually charge the EVs, it also doesn't help us decarbonise any of the following:
  • energy,
  • agriculture, 
  • medicine, 
  • manufacturing, 
  • mining, 
  • construction, 
  • waste disposal, 
  • sewerage treatment.

If we follow this plan, we will spend 1/2 our remaining carbon budget to partially fix a small part of our problem. Transport causes somewhere between 10-20% of Australia's emissions, so to spend half our carbon budget only reducing emissions for part of transport (passenger EVs doesn't fix freight, air travel, rail, busses) doesn't add up.
Again, the problem is not that EVs are not good -- the problem is that we can't replace every petrol car with an EV.


Let's revisit the four criteria I suggested at the start. Below are the requirements that must be met for EVs to be useful at weaning us from oil powered transport.

1. We need to be able to make EVs quickly enough to matter, so that we can quickly swap our existing car fleet over to EVs. 

I think this is questionable. Even if it is possible for us to make them this quickly (which is uncertain) I don't think we can persuade people to buy them unless we have significant government incentives. These incentives would need to be means tested, and target people who are currently buying vehicles in the $500-$10000 range.

2. The process of manufacturing/distributing the electric vehicles needs to remain inside the biophysical limits that we have. Conceptually, the simplest of these limits to consider is our remaining carbon budget, but water consumption, minerals availability are also very relevant.

I think it is clear that we cannot manufacture enough EVs to replace all existing personal ICEVs and also decarbonise the other parts of our economy that we must if we are to keep within 1.5℃ of warming

EVs need to be a large-enough improvement over ICEVs that we meed our sustainability objectives (eg. the total lifetime emissions of EVs need to be sufficiently less than ICEVs that we can get back "under the curve" of biophysical overshoot)

I think it's apparent that the significant manufacture costs (carbon emissions) of EVs means that we don't avoid enough emissions if we swap our ICEVs for EVs and keep everything else the same.

People need to be able to afford them

This is unclear, but I think it unlikely without significant government intervention.


So what should we do? The numbers are unequivocal -- we must drastically reduce the use of the private car, and it doesn't matter much whether the car is an EV or ICEV. We simply lack the technology to make private cars sustainable, and trying to cling to this idea makes meeting our climate goals impossible.

We must replace ICEV cars with EVs, but we can't replace every ICEV with an EV. Therefore, we need to come up with a way to reduce our car dependence (so that fewer trips need a car) and a way to share EVs (so that one EV can serve several people)

Therefore, we need to embrace public transport, walking, and cycling. These need to become our default transport options, saving private motor vehicles for the trips that actually need them. We need to make use of car sharing platforms (such as Car Next Door) to allow a relatively small number of EVs (maybe 1 million EVs -- where each EV is shared by 20 people) to serve all Australians (reducing the number of cars in Australia by 95%).

Sunday, August 7, 2022

In the age of cheap solar PV, does electricity conservation still matter?

Solar PV is an amazing technology. It has low emissions compared with other electricity generation method. Also, it is decentralised and thus can easily be retrofitted.

Today, it is possible to buy a good-quality 10 kW solar PV system from about $7500. Such a system, installed on mainland Australia, will produce an average of up to 20 kWh/day in winter, and close to 60 kWh/day in summer. Given how cheap that system is, and how much electricity it will produce, should we still need to be concerned with reducing energy use?

I think the answer is yes, for the following reasons:

1. For most residential solar PV installations in Australia, self-consumed power is black (not green) power

When you buy a solar PV system in Australia, the cost of the installation is offset by STCs. An STC represents 1 MWh of renewable electricity from a small-scale generator (eg. a domestic rooftop solar PV array). A new solar PV system creates a number of STCs equivalent to the renewable electricity it will generate over its expected lifetime. To make a solar PV system cheaper, people usually sell its STCs.

To whom are the STCs sold? People buy STCs when they buy GreenPower or want to offset other polluting activities. By buying STCs, someone is buying GreenPower.

The consequence is: if you have sold your STCs, then your self-consumed solar PV power (power that you use direct from your panels, rather than the grid) is actually black power -- you have in effect sold the renewable power coming from your panels (the STCs), and are instead using power from the coal power station, even if the electrons have come from your panels and not the grid, you have sold their "greenness".

Thus, the idea that you are using clean power direct from your solar PV is not accurate.

If you think "maybe some people do this, but I didn't do it", you're probably wrong. Take a look at your quote or invoice for your solar PV installation (mine is below) -- the vast majority of people (more than 90%) sell their STCs. Below is the price summary from my original PV installation in 2013 -- the STC sale reduced the cost by 1/3 so any power I self-consume is effectively black power.

My 2013 solar quote including STC sale
It's still great to install solar PV, even if you sell the STCs. By doing so, you are investing in a distributed renewable grid in Australia. While we should not let the perfect be the enemy of the good, we should keep in mind the limitations of what we have achieved so far.

2. Australia still gets a fairly small proportion of its electricity from renewable sources

In the last year, Australia's NEM (National Electricity Market) was about 1/3 renewable. We have a long way to go (Fig. 1)

Figure 1: breakdown of NEM generation over the last year

However, although we have a long way to go, we are running out of time to get there. Some scientists think we have already run out of time to prevent some dangerous climate change. This article was written in 2018, and among other things says: 
we are only three years away [2021] from overshooting the 1.5℃ target 
While it may already be too late for Australia to make a fair contribution to keeping global warming at 1.5℃, our results show that we can stay within our share of the carbon budget for 2℃ – provided we have the political will to move fast.
But the overriding message is that time is of the essence, if we want to come anywhere close to limiting dangerous climate change. Our various scenarios suggest that even if we implement a rapid, effective response, we are likely to have to take CO₂ back out of the atmosphere in the future, to compensate for the likely overshoot on our share of the global carbon budget.

Note that there are no confirmed, scalable, affordable, methods for extracting CO₂ from the atmosphere, and many scientists do not consider the 2℃ limit to be safe, as outlined here.

3. Currently, the manufacture of renewable energy technologies releases CO₂ into the atmosphere, using our carbon budget.

While solar PV and wind power are our best (and perhaps only) bet for achieving a sustainable electricity  supply, their production does cause carbon emissions (though vastly less than continued use of coal and gas does). Given the highly constrained carbon budget we now have, even the relatively small contribution of wind turbines and solar PV manufacturing will likely become significant.

carbon footprint of various electricity generation methods. Solar is about 20x better than coal, but it still has an environmental impact. Also note that the environmental impact of solar is "front-loaded" meaning that the impact of its entire lifetime of generation is brought-forward. Regarding "coal with carbon capture" -- note that this has not been commercially viable anywhere in the world.

4. We cannot manufacture wind turbines and solar PV quickly enough to supply all current grid electricity in time to avert serious climate change

Some models suggest Australia's grid will be 50% renewable by 2025 and 69% by 2030, becoming fully renewable in the mid 2030s. Recall the 2019 model suggesting that our carbon budget may already be exhausted -- this suggests that our deployment of renewable electricity will not be rapid enough to remain within the safe levels of carbon.


So, are we stuffed? I think the answer is "not necessarily", because all these analyses miss the thing that is easiest to change -- demand.
Let's consider per-capita energy consumption around the world:
If you examine this graph, you will see that Italians use about 1/2 the energy that Australians use. This suggests that we could halve our energy consumption without making significant changes to our society. The percentage of renewables in our grid would go from 1/3 to 2/3 in that process without having to install any additional capacity. Italy is a nice place -- Australians like to go there for a holiday because it's so nice. 

Let's consider look at Australia's historic consumption:
Australia per-capita electricity consumption (kWh per person)

In 1960, Australians used 1/5 the electricity that we do today. 

I downloaded data on Australia's historic energy consumption and corrected it (fairly roughly) by population to create per-capita relative consumption for each year. To me it looks like per-capita energy consumption hasn't changed a lot. The data only start in 1974. (I couldn't easily find older data)
per-capita energy consumption in Australia, 1974-2020

However, in the 70s, many (most?) consumer items (eg. cars, building materials, clothing) were made in Australia, so the embodied energy of the things we bought is included in the figure. However, by 2020 that was no longer the case and our energy consumption is effectively much higher than shown because of our reliance on off-shore industry. Also note that our domestic energy consumption has remained fairly constant even while we've had huge efficiency dividends because of technological improvements.

To try to illustrate this, I accessed Australia-China trade data, looked at the RMB (Chinese currency) value of Australian imports from China, scaled them by the average energy intensity of the Chinese economy (approximately 200 RMB per kWh), scaled that by the carbon intensity of Chinese electricity (2 MWh per ton CO2 -- this is about 1/2 the intensity of coal power) and scaled that by Australia's population to create a Australian per-capita estimate of Chinese CO2 emissions that arise because of the manufacture of Australian-bought products (and are hence our responsibility). This is obviously prone to error, but gives an indication. Note that these data only started in 2012. 

per-capita energy consumption in Australia, including the Australian equivalent per-capita from Chinese manufacturing of Australia-bound goods. 1974-2020

I think it shows that our domestic per-capita coal and oil consumption has remained consistent, our domestic gas consumption has significantly increased, and our foreign coal consumption (embodied in goods we buy) has gone through the roof.

This graph looks somewhat unbelievable. To sanity-check it, I considered the size of the total Chinese economy to estimate the per-capita emissions on current data only:

This data are similar to what is shown in the graph above, but were arrived at differently, suggesting that the numbers are meaningful.

To be honest, these numbers are quite shocking and should give Australians pause for thought.


The easiest and best way to quickly get closer to our goal of being zero carbon is to use less electricity and buy fewer manufactured products. This is true whether you have solar PV installed or not -- installing solar PV does not absolve you of a responsibility to use less.

If you think that Climate change is a problem to take seriously, I challenge you to use less than 2.5 kWh of electricity per person per day (eg. 10 kWh/day for a family of four). You can party like it's 1965 right now!
The Beatles say: just flick that switch off, baby!

If you want to read more, here's a short essay I wrote in 2014 about the merits of the 1950s -- yes, there was good amongst the bad, and we would do well to learn from those who lived then.

Wednesday, August 26, 2020

Efficient at what?

It is common to hear people talk about the efficiency of this-or-that system. Being "efficient" sounds like an unequivocal good, but the question that must be asked is "efficient at what?"

To illustrate this, let's consider "A tale of two heaters" in an average Australian living room.

Heater A is a split system heat-pump. It draws 2 kW of electricity, and outputs 8 kW of heat (it does this by pumping heat from outside the house to the inside of the house). For every 1 kWh of electricity it uses, it puts 4 kWh of hot air into the living room, and has a coefficient-of-performance (COP) of 4.

Heater B is a far-infrared panel (FIR). It is a resistive heater that emits infrared light, and needs to be directed to the occupant to be effective. As a resistive heater, it outputs 1 kWh of directional heat for each 1 kWh of electricity it consumes -- its COP is 1. Let's say it draws 1 kW.

Many people would say that Heater A is "more efficient", and there is some justification for this: for a given electricity consumption, heater A puts more heat into the room than does heater B.

However, because Heater B is targeting its heat directly at the occupant, it doesn't need to put so much heat into the room for the occupant to feel warm. Hence, one can argue that the amount of energy required to provide the service (making the occupant feel warm) is less and hence the FIR is more efficient because it uses less energy to make the person feel warm.

Hopefully, this example shows that when talking about efficiency, it is very important to be clear about exactly what we are doing efficiently. 

Deep problems with efficiency

But the problem goes deeper than this. 

In the 19th century, Britain was worried about running out of coal. At the time, coal was the fuel that drove the factories and war machine that made Britain the world's preeminent power. Some people thought that, as the machines became more efficient, coal consumption would be reduced.

William Jevons argued the opposite -- that an ability to use the resource more efficiently would actually drive increased demand in the resource. He noted that coal consumption rapidly increased after James Watt's more-efficient steam engine, because Watt's new engine made the exploitation of coal much more profitable.

This has become known as Jevon's Paradox, and the Wikipedia article contains much more detail.

We face the same phenomenon today -- improved technology that allows people to climate-control their houses more cheaply has meant that people perform more climate control. Whereas in the early-to-mid 20th century, an internal house temperature of 13 degrees was considered acceptable, it no longer is. Some people even heat their houses to 24 degrees.

Even if you don't do this, consider what you do with the money you save by spending less on fuel -- if you use that money to take more air travel, or to buy more manufactured goods, then you are a living demonstration of Jevon's Paradox.

In a society where 70% of grid electricity comes from coal, as does probably above 90% of the embodied energy in many manufactured goods (such as PV panels), it is simply not good enough to be "efficient".  Efficiency, by itself, will not reduce carbon emissions -- not even in theory.

The only solution is to focus instead on reducing consumption.

Wednesday, June 5, 2019

FAIL: wicking beds that don't last

I have written several articles about building cheap wicking beds using mixed materials and lined with builders plastic to form a reservoir. Unfortunately, these beds simply do not last. The builders plastic develops a leak, roots get in, and the reservoir fails. I am trialling a bed using pond-linear (which is quite a bit thicker), however experiments with a reed-bed made with pond-liner (it leaked!) suggest that this is unlikely to work.

For now, I am transitioning my leaking wicking beds (I think three of four are leaking) to IBCs, which I am hopeful will have better longevity. I will document this more in future.

Monday, March 18, 2019

The cost of travel


This essay examines the carbon cost of travel using different modes of transport, and compares it to other sources of carbon emissions (eg. electricity). It finds that international air travel is a significant source of carbon emissions, especially for people with an efficient house who fly regularly.
Compared to phasing out coal power (a noble goal), phasing out air travel would have a similar impact per-capita for regular flyers.
If middle-class Australians expect coal communities to make personal sacrifices to reduce the emissions intensity of Australia's electricity, I think it is reasonable to ask middle-class Australians to fly less.


In times past, only the wealthy could afford to travel. This was for the simple reason that most people were so poor that moving any significant distance from their home was unaffordable. And, of course, they had to work all the time.

Also energy, or the ability to do work, was so expensive that to travel took a long time (because to travel quickly takes much more energy than to travel slowly). Let's get a perspective on how much energy travel consumes.

This first figure shows the energy output of a human, an ebike, and a horse
Clearly, the ability of a person to do work is tiny compared to a horse -- it takes approximately 10 people to equal the power output of a horse. Horses are pretty powerful! Let's get a bigger perspective:
Suddenly, the horse doesn't look so powerful anymore! Even a small car can produce the power of many horses. The power output of a car is so great that the human output looks like nothing.

Of course, there is a difference -- the power output of both small and large cars is unsustainable (at least for the vast majority of cars which burn fossil fuels), and is causing changes in the Earth's climate that will be paid for by people in the future. Clearly, if we want efficient transport we should avoid cars.
Here we add air travel to the graph. This makes all of the sustainable transport look like it uses no energy at all. Note that for the cars and the plane, I assume that the car has 4 passengers, and the plane is completely full. Also, per-passenger energy consumption of the plane will be larger for business- and first-class passengers (because they take up more space).
Note that it is possible, at least in principle, to electrify car transport. Electrified air travel is currently science fiction. Note also that, currently, more than 60% of Australia's electricity is generated using fossil fuels.

Distance travelled

The figures so far show the energy consumption of different modes of transport, however, clearly people can travel more quickly by air than by foot or bicycle. However, the ability to travel quickly means that people tend to travel more. For example, one would never try to cycle from Sydney to London for a week or two of travel, but with air travel people do precisely this.

The concentration of vast amounts of power to allow air travel is a way for us to cram even more power consumption into our lives.

Let's look at the carbon emissions for different trips to get some perspective:
As before, buses, cars and planes are full and emissions are per-passenger.

Travel vs coal electricity

To reiterate, these numbers are large -- one person flying from Sydney to London produces more than 1.5 tons of emitted carbon. In comparison, in Australia some people (me included) make a fuss about getting rid of coal power because of the environmental impact. The average Australian household uses about 30 kWh per day (this is a huge amount of power), which is just under 11000 kWh per year. The average household has 3 people in it, so that is 3700 kWh/person. Coal power in Australia emits huge quantities of carbon per unit of electricity produced -- it is about 1 kg per 1 kWh. Thus, annual emissions from an average Australian's electricity use is about 3.7 tons.
A one-way flight from Sydney to London produces are much CO2 as 1/3 of annual electricity emissions, if electricity came from 100% coal.

These numbers become even worse if you consider an efficient house (which should be our goal). If you use 8 kWh per day, in a household of three, then one person's annual emissions under a 100% coal scenario are 8 /3 * 365 * 1 = 970 tons/year. A one way flight from Sydney to London is about two years' worth of emissions for this person.

Air travel as a proportion of total carbon emissions

While air travel comprises a relatively small proportion of Australia's total emissions (about 5%), there are hidden costs that are put into other sectors, such as the provision of fuel. Also, the proportion of carbon emissions from air travel is one of the fastest-increasing sectors, and it is also one of the most discretionary -- air travel can be easily reduced. Also, air-travel is a particularly damaging form of carbon emission because of its location in the upper atmosphere.
I have not been able to find statistics, but I suspect that there are a relatively small number of Australians who engage in a relatively large amount of air travel.


There is an increasing push in Australia to phase out coal power and replace it with renewables. This is a good thing. 
However, there is little awareness that carbon emissions from international air travel are similar to the emissions from 100% coal power for average electricity use in Australia. (This is especially true for wealthier Australians that might fly every year).
If middle-class Australians expect coal communities to make personal sacrifices to reduce the emissions intensity of Australia's electricity, I think it is reasonable to ask middle-class Australians to fly less.


Sunday, February 18, 2018

The (potential) problem of distributed energy storage

Over the last 18 months, there have been significant developments in the debate around renewable energy in South Australia. Here are some:
  1. South Australia had a state-wide blackout. It was variously attributed to a storm, wind turbines, renewable energy, the Australian Energy Market Operator, the phase of the moon, and probably other things.
  2. Load shedding (also known as brownouts, or rolling blackouts) was applied to 90000 households in SA on the afternoon of a particularly hot day. This was apparently triple what was requested by the market operator to maintain grid stability, and occurred at a time when a gas-fired power station at Pelican Point (in Adelaide) was idle.
  3. Elon Musk, the CEO of Tesla, told South Australia that Tesla can fix our energy problems in 100 days or it's free -- and has completed the installation of the "big battery" within that 100 day period
  4. The South Australian government released a comprehensive plan calling for grid level battery storage (100 MW), a government owned gas turbine and other energy initiatives
  5. Malcolm Turnbull has proposed an upgrade to the Snowy Mountains hydro facility that will allow it to operate as a pumped hydro plant (effectively a huge battery)
Whew! On top of this, there have been calls to the SA government that, as well as embracing a grid-level battery storage system, they should pursue distributed residential energy storage using a scheme such as Reposit (AGL is also rolling out such a system now, called the virtual battery). The Liberal party, as part of their platform in the upcoming election, are also proposing to subsidise home battery systems in some sort of virtual battery system. This is an internet-coordinated distributed power company that manages home batteries to release energy to the grid during times when the short-term cost (spot price) is high. Households can make money by doing this.

I think this is a great idea, and it's commendable that such policies have bipartisan support in South Australia. It is good for people in cities to install battery storage systems, as long as they remain connected to the grid. There are a few reasons I think this:
  1. The grid is a sunk cost. It's already built so we should use it
  2. Generally speaking, shared resources are used more efficiently. Let's assume that 5% of houses have a battery installed (in a few years time). We, as a society, will be much better off if those batteries are used to stabilise the grid during times of peak demand, rather than have those households disconnect from the grid. Those households should be paid for the service of course!
  3. Distributed power is a good thing because it lessens the load on the grid and improves resiliance (there is not a central point to fail), it also reduces power transmission losses.

The problem

One of the big benefits of having distributed power supplies is that of resilience. If one (or a few) systems fail or are disconnected it probably doesn't matter for the grid as a whole. It lessens our dependence on the grid as a whole, and so improves resilience.

But... although it decreases reliance on the electricity grid, it greatly increases reliance on the telecommunications grid. The internet is vital to coordinate the distributed "virtual" battery. Without the distributed batteries being able to talk to each other and the central controller, which needs to get data from the grid, it is impossible for them to coordinate their activities and the system ceases to function properly (for example, with no internet, it becomes impossible for the grid to request delivery of power from the distributed storage at a time of need).

Do not believe technical people who tell you a system is secure. It is not secure -- it is only a matter of time until software or hardware vulnerabilities are discovered and exploited.

A mitigation

"Solution" is the wrong word here: it's not possible to remove the consequences of attacks -- we can only reduce their consequence and likelihood of success. The distributed battery system needs to communicate, and that communication is inherently insecure. The system needs to be built so that it is resilient to internet-based attacks at the system level. I am not an expert in internet security, but here are my thoughts:
  • The system should be designed with security in mind. The critical aspect of the system is its ability to respond to legitimate internet instructions (only!) and react accordingly. The designers need to be aware of the inherent risk of exposing such command and control interfaces to the internet. The security team should have oversight across all teams during the development of this system. There needs to be robust testing of the system(s) (at all levels, from low level to high level) to examine the security implications of design and implementation decisions.
  • The system should be heterogenous. There should be many types of devices communicating on an agreed open protocol. Having many types of devices means that, even if some of them are compromised by an attacker, they are unlikely all to fail (they will be based on different hardware and software making it unlikely that a universal security vulnerability exists). This represents security through diversity.
  • The system should fail-safe. The internet should not be the only method of communication between devices. They should second-guess the instructions they receive from the cloud, by performing their own assessments of grid stability. For example, the grid can use frequency modulation to signal whether more generation capacity is needed (eg. if the frequency starts to fall from a nominal 50 Hz to 49 Hz, that is a recognised signal -- this needs to be preserved as a signalling method). In this way, the devices should use the internet as a communications channel where it is available, but not be dependent on it for all aspects of their function.
  • User interface elements should be kept separate from the control systems. User interfaces are almost always less secure than system interface elements. This is because user interfaces have the added constraint of usability, which is often at odds with security requirements. User interfaces should be through a seperate cloud/web portal that has no direct connection to the system interface.

On many occasions, systems have been designed to use internet platforms with the unconscious assumption that the internet is both safe and persistent. The internet is neither of those things. I hope that the designers of distributed storage are mindful of this, so that their system(s) will be resilient in the face of unexpected communications downtime or malicious attacks.
Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.