How many electric vehicles could Australia's grid cope with?

[Vostok]

How often have we heard that one reason Australia cannot shift to EVs is the massive load on the grid it would cause and just think of the cost of upgrading the grid to cope. Meltdown!

So I did some simple calculations to see how many vehicles the national grid could cope with today if they all charged after midnight.

Some simple input assumptions: the average Australian driver travels around 15,000 km per year or 41 km per day. If that was in an EV with efficiency of 150 Wh/km, that's an energy need of 6.2 kW - or less than an hour's charging with a 7 kW charger or 2.6 hours off a 10A wall socket. Let's assume an average charge rate of 3.6 kW.

The average national grid load at midnight is 24.1 GW, declining to a minimum of around 19 GW around 4am, and doesn't get back to 24 GW until around 8am. So provided all vehicles can be charged to restore their average daily usage within those 8 hours (a very safe assumption), then the headroom between the midnight load and total grid capacity would allow a certain number of EVs to be charged without requiring any grid expansion.

Australia's national grid capacity is at least 33 GW (the highest recent peak was 33.035 GW at 6:30pm on 14 July). Based on that, we have grid headroom of at least 9 GW to charge EVs after midnight.

So at 3.6 kW charging, Australia today could support 2.47 million EVs - or about 12% of the current national fleet of around 20 million vehicles. Dial the charge rate down to 2.4 kW and the number would increase to 3.7 million or 18.5%. Dial the charge rate down even further to use the full 8 hour window to charge, and the number increases to 11.6 million or 58% of the fleet. Optimise to the max (e.g. use the variable headroom available midnight to 8am and optimise the charge rate of every EV over that time) and the number gets to around 14 million (70%).

So what to make of this?

First, Australia is a long way from having this many EVs on the road, so this is not an urgent problem to fix. More grid capacity is added every year, so it may organically be able to keep up.

I think the grid definitely needs to get smarter, with utilities being able to use V2G to optimise the charge rate of the EV fleet to manage total grid load. Simply building capacity for more peak load is economic madness, when smarts could eliminate the need to build much at all.

But also as solar penetration increases and more households add batteries, that will enable more EV charging to happen "behind the grid" hence reduce the incremental grid load as well. And EVs might get more efficient than 150 Wh/km. And people might drive less post-Covid with more WFH. Lots of things could change.

So there are some challenges ahead, but it looks absolutely do-able to me over decade-type timeframes, and the advent of EVs will not bring grid Armageddon.

 

[MXLRplus]

Since adoption will be a gradual process, if the government does not assign the grid planning to somebody's brother-in-law who just got out of rehab, the grid will expand at an appropriate rate like it always has.

However, the assumption that a competent manager(s) will be tasked to perform the duty is questionable. Government appointments for technical posts aren't as good as you experience in filling private sector executive positions.

In the long run, all bad management will do is to delay acceptance by the consumers, so it's a self-regulating system.

 

[moa999]

The average national grid load at midnight is 24.1 GW, declining to a minimum of around 19 GW around 4am, and doesn't get back to 24 GW until around 8am. So provided all vehicles can be charged to restore their average daily usage within those 8 hours (a very safe assumption), then the headroom between the midnight load and total grid capacity would allow a certain number of EVs to be charged without requiring any grid expansion.

No solar overnight though which provides a fair chunk of that stated peak.

Still have some wind, but the bulk of night power is coming from coal.

OpenNEM: NEM

 

[paulp]

Over the past weeks there has been a lot of media in SA about household solar being force shut down in the middle of the day to help avoid a grid meltdown. Indeed from next week all new installations must give SAPN full control to do just that. So right now in SA a stack of household batteries, either fixed or on wheels, would be beneficial if charging during the day.
Its interesting that SAPN initially refused my application for a powerwall as I have the maximum 30kw solar system. Eventually they worked out my battery would actually help the grid, not hinder it, and my cars help it way more.

 

[paulp]

How often have we heard that one reason Australia cannot shift to EVs is the massive load on the grid it would cause and just think of the cost of upgrading the grid to cope. Meltdown!

So I did some simple calculations to see how many vehicles the national grid could cope with today if they all charged after midnight.

Some simple input assumptions: the average Australian driver travels around 15,000 km per year or 41 km per day. If that was in an EV with efficiency of 150 Wh/km, that's an energy need of 6.2 kW - or less than an hour's charging with a 7 kW charger or 2.6 hours off a 10A wall socket. Let's assume an average charge rate of 3.6 kW.

The average national grid load at midnight is 24.1 GW, declining to a minimum of around 19 GW around 4am, and doesn't get back to 24 GW until around 8am. So provided all vehicles can be charged to restore their average daily usage within those 8 hours (a very safe assumption), then the headroom between the midnight load and total grid capacity would allow a certain number of EVs to be charged without requiring any grid expansion.

Australia's national grid capacity is at least 33 GW (the highest recent peak was 33.035 GW at 6:30pm on 14 July). Based on that, we have grid headroom of at least 9 GW to charge EVs after midnight.

So at 3.6 kW charging, Australia today could support 2.47 million EVs - or about 12% of the current national fleet of around 20 million vehicles. Dial the charge rate down to 2.4 kW and the number would increase to 3.7 million or 18.5%. Dial the charge rate down even further to use the full 8 hour window to charge, and the number increases to 11.6 million or 58% of the fleet. Optimise to the max (e.g. use the variable headroom available midnight to 8am and optimise the charge rate of every EV over that time) and the number gets to around 14 million (70%).

So what to make of this?

First, Australia is a long way from having this many EVs on the road, so this is not an urgent problem to fix. More grid capacity is added every year, so it may organically be able to keep up.

I think the grid definitely needs to get smarter, with utilities being able to use V2G to optimise the charge rate of the EV fleet to manage total grid load. Simply building capacity for more peak load is economic madness, when smarts could eliminate the need to build much at all.

But also as solar penetration increases and more households add batteries, that will enable more EV charging to happen "behind the grid" hence reduce the incremental grid load as well. And EVs might get more efficient than 150 Wh/km. And people might drive less post-Covid with more WFH. Lots of things could change.

So there are some challenges ahead, but it looks absolutely do-able to me over decade-type timeframes, and the advent of EVs will not bring grid Armageddon.

We dont yet really know what other changes mass uptake would create. At 12% for example, would we loose 12% of servo’s and 12% of mechanics. The servo’s are quite power hungry these days with their convenience sections. Not all would charge at night, so say half charged during the day at work or home when solar is available, so now its 24%. So even more servo’s close.
Solar is also rapidly improving in efficiency, which will also have an impact.

 

[Furfy]

At 12% for example, would we loose 12% of servo’s and 12% of mechanics

I'm not so sure of that. It's feasible that Servo's would transition to providing DC fastchargers instead of pumping fuel. This way they can keep their most profitable product turning over (the food/drinks/snacks they peddle you whilst you are 'filling up'). In fact, due to it taking longer to 'fill up' with an EV, one would be stuck at the servo for longer, meaning it's likely they will actually sell more of their most profitable product to EV owners?

Also, even if every car out there somehow became EV's, mechanics will still have their place in society, although not as prevalent as they are now. Vehicles, regardless of EV or Combustion Engine, will still need bushes/bearings/suspension/steering/lights/brakes/tyres/wheel alignments/power steering/etc fixed and replaced. Even electric motors can fail and require replacement. The people who would perform this work would be mechanics of sorts.

 

[Vostok]

No solar overnight though which provides a fair chunk of that stated peak. Still have some wind, but the bulk of night power is coming from coal.

My analysis was about Australia’s current generation and grid capacity to charge EVs, not the composition of that generation, which is a different matter.

But that is the next important point. We want incremental generation to be “clean”. As @paulp noted, EVs could be a very useful tool in absorbing excess solar injected into the grid. Which is why we need a smarter grid.

The renewable situation is not necessarily dire at night - there’s still wind power and pumped hydro. Wind is obviously variable but some nights it generates a lot of power. Also, renewables are often shaded in order to manage the operating point of the fossils. So just because wind is only generating “X” at night doesn’t mean it is capable of only generating “X”. Some turbines might be braked, and if there was more load, they wouldn't be braked.

Snowy 2.0 will be able to dispatch 2GW into the grid - 550k EVs charging at 3.6 kW or 830k EVs charging at 2.4 kW.

At the moment, my wife’s EV is charged at 2.4 kW from midnight to 4am because she drives it to work, and my Model 3 (Slartibartfast) is charged at 4.4 kW from solar from 10am because I mostly don’t need to drive it during the day (either WFH, or going into the office via public transport).

During winter, Slartibartfast‘s target charge is set to 50%. But on sunny days when PW2 gets full early on, I manually adjust its target charge on the App to to take up the excess. Conversely, on cloudy/rainy days, I dial down the set point so that it doesn’t charge at all. The target is currently at 70% and I expect it will creep up to 90% as we head into summer.

I’m hoping that the mooted Tesla PW2/Vehicle integration for charging with excess solar will soon become a thing and I won’t need to manually tweak it anymore.

 

[Vostok]

Over the past weeks there has been a lot of media in SA about household solar being force shut down in the middle of the day to help avoid a grid meltdown. Indeed from next week all new installations must give SAPN full control to do just that. So right now in SA a stack of household batteries, either fixed or on wheels, would be beneficial if charging during the day.

Yep this is critical. Freewheeling household / small scale solar is OK when it’s a relatively small proportion of grid generation. But it becomes an increasing problem as the proportion grows because it (currently) can’t be managed in the same way as grid assets are (e.g. a grid solar farm).

If solar can’t be dumped anywhere (e.g. into household batteries, community batteries, grid batteries, pumped hydro or EVs) and the grid can’t take it because demand is already sated, then the utility needs to have the ability to turn down your inverter. Sounds “mean” but it isn’t. It’s also a business case for utilities to subsidise battery installations, since excess solar otherwise triggers negative wholesale rates (utilities pay others to take their power!).

Hopefully as the grid becomes smarter and storage becomes more pervasive this will be less of an issue.

 

[Shallowal]

Nice work Vostok.
But just to add to the postulate.
In Australia there are roughly 19 million "driving age" (15-80 age) people (National, state and territory population, March 2020).
So to hazard a guess as to how many of people drive a car frequently, lets say 60%. So that makes it less then 12 million people drive a vehicle frequently which fits in nicely with your grid capacity figure.

 

[paulp]

No solar overnight though which provides a fair chunk of that stated peak.

Still have some wind, but the bulk of night power is coming from coal.

OpenNEM: NEM

We dont do coal anymore in SA, although technically we support brown coal on the occasions that we import from Victoria

 

[moa999]

My analysis was about Australia’s current generation and grid capacity to charge EVs, not the composition of that generation, which is a different matter.

Yes but you were pointing at a grid peak of 33GW.
On a sunny day up to 15GW is being provided by solar, so the night capacity is a lot lower.
(And a lot less green)

 

[PC__LoadLetter]

Wind turbines are located where they give reliable power at 3pm. In southern NSW the cool afternoon ocean easterly that overwhelms the hot desert northwesterly arrives around 2pm-ish in Braidwood, 3pm-ish around Goulburn, 9pm-ish around Canberra, and 3am around Cobar. It’s fairly reliable. Accordingly, there are multiple wind farms just south and west of Goulburn.

If there’s a need for reliable power at 9pm, build wind farms on both sides of Canberra (Googong and the Brindabellas).

 

[cafz]

Dial the charge rate down even further to use the full 8 hour window to charge, and the number increases to 11.6 million or 58% of the fleet.

Note that you don't even have to be particularly clever to do this: Just have every car pick a random start time during the window, and overall you'll spread the charging evenly over the time period.

 

[Chuq]

Note that you don't even have to be particularly clever to do this: Just have every car pick a random start time during the window, and overall you'll spread the charging evenly over the time period.

I'm pretty sure that's what the Origin home charging trial (announced a month or so ago) is all about - coordinating the charging between cars so that if 10000 people with EVs say at 9pm that they want their car to be 90% full by 6am, Say each car needs between 1-3 hours, the chargers communicate with the grid and schedule all the cars so that they start at just the right time to smooth out the load equally over the night.

 

[cafz]

I'm pretty sure that's what the Origin home charging trial (announced a month or so ago) is all about - coordinating the charging between cars so that if 10000 people with EVs say at 9pm that they want their car to be 90% full by 6am, Say each car needs between 1-3 hours, the chargers communicate with the grid and schedule all the cars so that they start at just the right time to smooth out the load equally over the night.

Perhaps, but my point was that there's no need for central coördination to do this - it's enough for each of those 10,000 cars to pick a random point between 9pm and (6am minus estimate charge time) to start on their own. With 10,000 cars, the law of large numbers does a lot of heavy lifting for you.

 

[Chuq]

Perhaps, but my point was that there's no need for central coördination to do this - it's enough for each of those 10,000 cars to pick a random point between 9pm and (6am minus estimate charge time) to start on their own. With 10,000 cars, the law of large numbers does a lot of heavy lifting for you.

That's possible, if everyone gets how it works. At the moment, retailers/DNSPs use off-peak time to help to avoid the peak demand occurring as soon as people get home from work. However some limited testing has found that this can potentially be even worse - everyone sets their timers for 9pm (or whenever peak time starts) and the spike of them all starting at once can be worse that people plugging in at different times such as when they get home (staggered between 5pm and 7pm, for example).

I personally set my charging to begin at 2:20am, instead of on the hour or half-hour, for this reason! (Not that I think it's much of an issue at the moment).

The orchestration can work in specific environments, such as an apartment building car park. Have to share a single 22 kW between 12 cars? Probably best done with a centralised system. Everyone plugs in when they get home and it just rotates through 2-3 cars at a time until they're all done.

 

[cafz]

That's possible, if everyone gets how it works. At the moment, retailers/DNSPs use off-peak time to help to avoid the peak demand occurring as soon as people get home from work. However some limited testing has found that this can potentially be even worse - everyone sets their timers for 9pm (or whenever peak time starts) and the spike of them all starting at once can be worse that people plugging in at different times such as when they get home (staggered between 5pm and 7pm, for example).

Yes, it would work best if baked into the "finish charging by <time>" setting of the cars.

I personally set my charging to begin at 2:20am, instead of on the hour or half-hour, for this reason! (Not that I think it's much of an issue at the moment).

Mine tends to be different every day at the moment, because I noticed over winter that scheduling the charge to finish in the morning seems to leave the battery warmer. So I've been doing a rough mental estimate of what time I need to start charging to finish around 6:30.

The orchestration can work in specific environments, such as an apartment building car park. Have to share a single 22 kW between 12 cars? Probably best done with a centralised system. Everyone plugs in when they get home and it just rotates through 2-3 cars at a time until they're all done.

Yes, certainly if you have 10 cars rather than 10,000 that is a different matter!

 

[Vostok]

Nice work Vostok.

Thanks!

But just to add to the postulate. In Australia there are roughly 19 million "driving age" (15-80 age) people (National, state and territory population, March 2020).
So to hazard a guess as to how many of people drive a car frequently, lets say 60%. So that makes it less then 12 million people drive a vehicle frequently which fits in nicely with your grid capacity figure.

As far as I know, the “average distance per year” driven by vehicles in Australia is calculated by dividing the total number of km driven per year across the fleet (presumably based on some reasonably robust sampling data) by the number of vehicles in the fleet. So the frequency of driving should be irrelevant - some cars will be driven a lot more than 15,000km and some cars will be driven a lot less.

So the total energy need for the fleet should be roughly right, and your factor should already be taken into account.

 

[Vostok]

Perhaps, but my point was that there's no need for central coördination to do this - it's enough for each of those 10,000 cars to pick a random point between 9pm and (6am minus estimate charge time) to start on their own. With 10,000 cars, the law of large numbers does a lot of heavy lifting for you.

I think that would be true only if every owner set their car to charge at the same rate (e.g. 3.6kW). In reality that won’t happen, either because different cars have different but fixed AC charge rate, or the owner sets it to whatever they want. So you’d end up with a more random curve rather than a fixed load.

 

[moa999]

Probably best done with a centralised system. Everyone plugs in when they get home and it just rotates through 2-3 cars at a time until they're all done.

The issue then becomes the cost of designing and installing such a system, and any ongoing system charges.
Could quickly eat into any benefits.