Tesla Big Battery Outsmarts Lumbering Coal Units After Loy Yang Trips (reneweconomy.com.au)

The Tesla big battery is having a crucial impact on Australia's electricity market, far beyond the South Australia grid where it was expected to time shift a small amount of wind energy and provide network services and emergency back-up in case of a major problem. From a report: Last Thursday, one of the biggest coal units in Australia, Loy Yang A 3, tripped without warning at 1.59am, with the sudden loss of 560MW and causing a slump in frequency on the network. What happened next has stunned electricity industry insiders and given food for thought over the near to medium term future of the grid, such was the rapid response of the Tesla big battery to an event that happened nearly 1,000km away. Even before the Loy Yang A unit had finished tripping, the 100MW/129MWh had responded, injecting 7.3MW into the network to help arrest a slump in frequency that had fallen below 49.80Hertz.

Comments

AC frequency

[ebcdic]

For the benefit of Americans reading: the nominal AC frequency in Australia is 50Hz, not 60Hz.

Re:AC frequency

[Thelasko]

For those that don't know. The frequency of AC power is an indicator of the supply and demand status of the grid. The frequency is determined by the speed of the generators at the power station. If there is too much load on the generators, they slow down, and the grid frequency drops.

The Australian grid is targeting 50Hz, and had dropped to 49.8Hz.

Re:AC frequency

[Anonymous Coward]

More demand / less supply > generators have to work harder > greater force needed to spin them > turbines slow down > frequency drops.

There's not really anywhere on an electricity grid where one can connect a meter and say "we need more power" so they monitor frequency instead.

Re:AC frequency

[xvan]

Synchronous Generators are fed by a torque sources, if you have a generator outage, and the other generators don't have enough power reserve there are 2 consequences that balance the power consumption. A drop in voltage and a drop in the network frequency (P=T* w, so with the same torque available, less frequency means less power injected to the network )

Re:AC frequency

[AK Marc]

Electricity grids are stacks of eggs balanced on their point. A single snow event taking out a single line in the US took out millions of people for days, in a very populated and "modern" area.

The Australian grid may have failed. The current was out of spec. If protection circuits activate, they'd shut down the grid. Tesla didn't fill all the missing need, but injected enough power in an "our of spec" event to ensure the grid couldn't fail from that event.

That small boost may have saved a major catastrophe. We may never know. But that it could is a great proof of concept. Battery-based storage can react faster than anything else on the grid, to smooth grid failures to prevent cascades. Now we know, we need them all over the US, before the next snowstorm in the North East.

Re:AC frequency

[Orne]

Not quite. The original coal plant tripped, so the power that it was injecting ceased to be. In the very short term (tens of cycles), the energy demand on the system outweighs the supply, and frequency begins to drop. The remaining synchronized generating resources next engage "primary frequency response", which is an automated (governor) response that temporarily increases the output of the generators. By governor, there is a device in the generator controller that regulates the steam pressure to keep the rotation constant, so the energy imbalance creates mechanical drag that the governor attempts to correct. Each generator twitches up a tiny amount, the frequency decline is arrested, and the system stabilizes. You then have secondary systems that engage that drive the system back to a pre-loss state.

The battery in this contributed primary frequency response, as a direct response to the observed low frequency. In the United States, Energy Storage devices are not required to provide primary frequency response, since almost all frequency response is provided by steam units. As more coal plants are retired and replaced by Wind and Solar (inverter-based units), the US grid will need to adapt and modify its requirements.

Re:AC frequency

[mspohr]

First of all, OP isn't talking about injecting frequency.

I didn't say he did. The summary talks about injecting 7.3MW to "arrest a slump in frequency".

When the system is overdrawn on power the high load slows down the turbine generators and the frequency drops. The solution is to add power to the system. The battery added 7.3 MW of power to the system which helped to bring the frequency back up to nominal 50 Hz.

Re:AC frequency

[Pseudonym]

I've worked a little bit with data from the AEMO. I'm not a power distribution engineer, but I did to learn enough to be able to explain it badly. So here goes...

One way to think of it is that all of the equipment on a given segment of the network synchronises to the frequency of the network, but tries to nudge it ever so slightly closer to 50Hz. If every piece of equipment on the network does this, the network as a whole trends towards the correct frequency. The system can tolerate some drift, so each piece of equipment acting independently can force the network as a whole to keep to 50Hz as long as it isn't overloaded.

A typical alternator that you may find in a generation plant is designed so that it will produce 50Hz when fully loaded, that is, whenever the amount of power that it's designed to generate is being drawn. When none of the power is being drawn, it physically turns around 4-5% faster, so it might run at 52 Hz if you did nothing. So if the full power output of the generator is not being used, you need to physically slow it down.

That's easy, but of course the specific technique depends on how the alternator is being physically turned. If you can turn down the amount of fuel (trivial for hydro, almost as easy for a gas turbine), you do that, or you might use a mechanical or electromechanical governor on a coal plant.

The problem happens when the network is overloaded. When you draw more power from an alternator than it is rated to produce, this acts like an electromechanical brake, and it will run slower than 50Hz. You can't force an overloaded alternator to run faster, so any attempt to increase the frequency won't work. The only fix is to not overload it by adding more power to the system or reducing demand.

One of the key reasons why the South Australian government wanted to build the Tesla battery was because the AEMO couldn't get a generator turned on in time and so had to shed load by deliberately causing blackouts in South Australia. The amusing thing about TFA is that we may have just discovered that the Tesla big battery may be designed to protect the SA grid from the AEMO.

Just for completeness, I'm using the word "network" here to refer to a region for which the frequency is synchronised. I believe this is true for most of the NEM; TFA seems to indicate that Hornsdale (SA) and Gladstone (QLD) are synchronised. However, I seem to recall from the data that Tasmania's connection is via a HVDC link which can work in either direction, so presumably Tasmania's frequency doesn't need to be synchronised to that of the mainland.

The AEMO, by the way, is essentially a big integer linear program plus some human intervention in the case of emergencies. The ILP represents the network constraints (e.g. the capability of every generator, the maximum current of every distribution line, a squillion contract clauses) and tries to minimise dollars per kWh.

Re:AC frequency

[viperidaenz]

It didn't respond too rapidly. The frequency was below the absolute minimum of 49.85Hz for normal operation.

Just putting out a wild idea: they configured the battery to kick in at 49.80Hz on purpose.

We should have batteries at every substation.

[Snorlax]

The resiliency of the power grid would be vastly improved if we put a battery pack (the size of a normal intermodal container) at each substation. These could act like your home UPS, fixing blackouts of a few minutes, when they occur. This also would make the grid much more able to use wind and solar sources, without so much need for standby diesel systems currently in place.

Re:We should have batteries at every substation.

[ArhcAngel]

Until VERY recently this was not at all economically viable because the cost to store the electricity was higher than the cost to generate it. I work for the largest owner of wind energy in North America and for years they would routinely short their windmills electricity production directly to ground because the grid from their location in West Texas to Dallas where it was needed was saturated. If they had stored the electricity the cost of generation + storage would have meant they would have to sell it at a loss. Now that storage costs can beat peak rates you'll see large companies invest in electricity time shifting so they can charge the battery when electricity is cheapest and switch to battery during peak usage to save money. This will benefit the grid directly since it will lower the stress during peak usage overall.

It is getting a little old

[NEDHead]

When is Musk going to stop making big promises and then following through?

He sure is a bad politician.

It'll never work....

[Charcharodon]

....and yet it does.

Trying to remember why it wouldn't have worked. Because it might steal their market share? Yeah pretty sure that was their reason they didn't think it would.

Re:A slump in what?

[HornWumpus]

They are both affected. But power companies will let the voltage drop while holding frequency as close to theoretical as they can. They even run 0.1 Hz high or low at the end of the day to get the correct number of cycles for the period.

If you've ever designed a power supply, you'd see that you must accept low/high voltages, but should expect the frequency to be fairly steady.

Re:A slump in what?

[Thelasko]

If a power source goes offline, wouldn't you see a slump in voltage? Why the decrease in frequency?

In DC, yes. AC is a different animal. The AC frequency is determined by the speed of the generators. When demand outstrips the supply, the generators slow down. Therefore, the frequency drops.

You would likely see a drop in voltage too. However, AC voltage is difficult to measure. Frequency is a much more precise way to measure the status of the grid.

Re:I don't see how it stopped an outage

[HornWumpus]

The grid worked as designed. News at 11.

Steam plants don't come online in 6 seconds, they just don't.

First the UPSs, then load curtailment, hydro and combustion turbines, finally the steam plants and steam parts of combined cycle plants.

The real point (beyond the usual /. 'Ol Musky' blowing) is that apparently Australia was in spinning reserve violation when this happened. Your supposed to have enough power spinning to cover you single biggest unit/transmission line falling over (as they say in Australia).

Re:A slump in what?

[RobinH]

Good question. A simple way to view it is that the grid is powered by generators. The generators are built to run at a fixed speed, and are wound so that the fixed speed outputs (in this case) 50Hz at a fixed voltage. The voltage output of the generator is a sine wave and it will lead (since it's generating) the grid voltage by a small amount (lead means same frequency, slightly ahead of phase). The amount it leads determines the load, and the generator has a limit to how much load it can handle, so if you tried to speed it up by turning it faster, it would start to lead slightly more and the load would increase (more current, but more resistance to the prime mover turning the generator) so the speed stays close to 50 Hz and it only speeds up a very small amount very briefly. When you drop a bunch of generation offline, the rest of the generators see a bunch more load suddenly, which is felt as a physical torque, so the generator gets harder to turn. The prime movers (turbines typically) can't produce more power instantaneously so the generators start to decelerate slightly. That's why you see the grid frequency drop slightly until the turbines increase power to take up the load. That's assuming the remaining generation can handle it. What they're saying here is that the Tesla system, since it uses inverters, can respond faster than the turbines generating power (duh). I'm not sure why it's described as shocking. Near where I live, in Canada, they installed a few MW of magnetic bearing sealed-vacuum flywheel energy storage specifically for frequency regulation due to all the new windmills they installed. The flywheels are spinning at synchronous speed and can absorb and deliver energy to the grid as needed, similar to the Tesla battery system.

Re:na

[Mr D from 63]

...where it was expected to time shift a small amount of wind energy and provide network services and emergency back-up in case of a major problem.

No, the primary purpose of the battery was to help the grid ride through transients just as the one described, not for time shifting. Who is writing this stuff?

What? Who? How?

[Ancient_Hacker]

The narrative and conclusions are a big dodgy. Everybody knew beforehand that batteries can jump in immediately to supply power. And the batteries did not stop a complete collapse, electrical networks are thoroughly analyzed and simulated and braced against major consequences if any one unit trips out. Major outages are quite rare over the decades, and all done without a single battery. Gas turbines can come on-line within 60 seconds and other interconnected plants often have enough reserve capacity to tide over small outages. Batteries are welcome as an immediate source, but they are still awfully expensive and awfully small in GWH.

Re:na

[jofas]

And "What happened next" did not stun electricity industry insiders. It was engineered to do the very thing it did.

Re:Rather Anti-Climatic?

[torkus]

Grid level power management is utterly unlike your home UPS.

I think the article is overstating a bit given the scale, but the macro implications are impressive. Grid-scale generators are slow to ramp up and down - minutes to hours (or even days for startup of nuclear plants). Small, less efficient generators handle the small peaks (oddly enough, called peaking generator) that go beyond baseline generation and any under-utilization goes to waste so it's a careful balancing act. And even the peaking generators aren't instant response whereas the Tesla Battery IS essentially able to go from 0-100MW in moments (they should advertise this along with the Tesla speed records). This allows highly efficient supply of peak-demand (or, in this case, unexpected demand) which is pretty much unheard of.

Having 500MW go offline suddenly does Bad Things to the overall grid. Remember when one plant tripped offline ... I think in upstate NY and blacked out most of the northeast in a cascade failure several years back? Having something able to take a near-instantaneous load, even for a few minutes, is a massive benefit.

Re:I don't see how it stopped an outage

[dj245]

The coal plant that failed was producing close to 600MW. The max output from the graph in the article showed the battery system inject less than 10MW max into the grid. Who pickup up the other 500+ MW? The other coal plant that came online within 6 secs. Basically all the batteries did was reduce the size of the brownout.

The "spinning reserve" generally picks up the demand. "Spinning reserve" consists of machines which are on the grid but not at full load. The spinning reserve should be a minimum of the sum of the largest individual generator + the maximum estimated demand change that could happen in around 10 minutes (the time it takes for a gas turbine to start up). Generally, all that is necessary to change spinning reserve into real power is for a valve to be opened further. For combustion or steam turbines, this can occur in less than a second, and is automatically controlled by the generator controller - the generator demand signal will increase as grid frequency decreases. Spread across many generators, the increase in output is not a significant shock to any individual generator.

In this case, it seems that the Australian grid did not have adequate spinning reserve, which is why the frequency dropped. Many power stations are set to shut down in the case of large frequency variations (for machine protection), which caused the coal power station to shut down.

Frequency drop indicates overload

[raymorris]

The drop in frequency itself isn't the big problem, it's a gauge, an indicator.

The frequency tells you how fast the generators are turning. They are automatically throttled to try to spin at the right speed to produce 50Hz. If they aren't producing 50Hz, that means they are full wide open throttle and still can't keep up. It means they can't produce enough power.

Re:na

[Rei]

Actually, no, it wasnt engineered to back up a power plant in Victoria, it was engineered to back up power in South Australia. There was an entirely different coal power plant that was supposed to back up Loy Yang (which is one of Australia's largest) - a plant that ratepayers have to pay to keep running on standby, which is supposed to hold the grid up until downed power plants can be brought back up and/or more baseload elsewhere ramped up. But from nearly 1000km away, the Tesla battery did the standby plant's job for it during its 4-second wakeup time - stopping and reversing the decline in grid frequency so that there wasn't even a meaningful blink in power quality.

This is not what the Tesla battery was designed to do. It was designed to deal with situations with downed lines / plants in South Australia, to keep the lights on there. It wasn't supposed to take over the work from standby plants halfway across the country. That it technically can should surprise nobody. But that's not what it was purchased to do.

Re:na

[ChumpusRex2003]

It is likely a linear power response to frequency with a small dead band.

In the UK, battery backed frequency response is an important contributor to frequency stability, and is operated with a dead band of 0.015 Hz. The power injection is required to be proportional to the frequency deviation from outside the dead band, reaching 100% rated power at 0.5 Hz deviation from nominal. Response time is a maximum of 1 s.

Additionally, in the UK, the requirement is that the frequency response is symmetrical. If frequency rises, then the system must absorb power - up to 100% of maximum rated power at 50.5 Hz, for a minimum of 15 minutes.