Slashdot Mirror
Tesla's Giant Battery In Australia Reduced Grid Service Cost By 90 Percent (electrek.co)
An anonymous reader quotes a report from Electrek: Tesla's giant Powerpack battery in Australia has been in operation for about 6 months now and we are just starting to discover the magnitude of its impact on the local energy market. A new report now shows that it reduced the cost of the grid service that it performs by 90% and it has already taken a majority share of the market. It is so efficient that it reportedly should have made around $1 million in just a few days in January, but Tesla complained last month that they are not being paid correctly because the system doesn't account for how fast Tesla's Powerpacks start discharging their power into the grid.
The system is basically a victim of its own efficiency, which the Australian Energy Market Operator confirmed is much more rapid, accurate and valuable than a conventional steam turbine in a report published last month. Now McKinsey and Co partner Godart van Gendt presented new data at the Australian Energy Week conference in Melbourne this week and claimed that Tesla's battery has now taken over 55% of the frequency control and ancillary services (FCAS) services and reduced cost by 90%. "In the first four months of operations of the Hornsdale Power Reserve (the official name of the Tesla big battery, owned and operated by Neoen), the frequency ancillary services prices went down by 90 percent, so that's 9-0 per cent," said Gendt via Reneweconomy. "And the 100MW battery has achieved over 55 percent of the FCAS revenues in South Australia. So it's 2 percent of the capacity in South Australia achieving 55 percent of the revenues in South Australia."
The system is basically a victim of its own efficiency, which the Australian Energy Market Operator confirmed is much more rapid, accurate and valuable than a conventional steam turbine in a report published last month. Now McKinsey and Co partner Godart van Gendt presented new data at the Australian Energy Week conference in Melbourne this week and claimed that Tesla's battery has now taken over 55% of the frequency control and ancillary services (FCAS) services and reduced cost by 90%. "In the first four months of operations of the Hornsdale Power Reserve (the official name of the Tesla big battery, owned and operated by Neoen), the frequency ancillary services prices went down by 90 percent, so that's 9-0 per cent," said Gendt via Reneweconomy. "And the 100MW battery has achieved over 55 percent of the FCAS revenues in South Australia. So it's 2 percent of the capacity in South Australia achieving 55 percent of the revenues in South Australia."
Is in what is called "ancillary services".
An ongoing issue with operating and maintaining an electrical grid is how to balance electrical generation with electrical consumption. The two vary throughout the day; for example, solar energy adds a surge of power to the grid during sunlight hours, while peak consumer demand for electricity happens around 7-8pm. If you have five minutes, I suggest you watch this video, produced by Vox, discussing it further.
How do electrical companies then compensate for the differences? Or for contingencies, like when an electrical generator needs to be brought offline for emergencies or maintenance? This is where "ancillary services" plays a vital importance. Utilities are desperate to find an efficient way to store surplus power generated when supply is higher than demand, so that it can then be released when demand is higher than supply. Currently, when supply is too high, it is reduced (ex: solar panels and wind turbines turned off), wasting energy. When supply is too low, expensive generators are brought online to meet demand. But if we can make battery technology cost-efficient to store surplus electricity for peak-demand use, it would save vast sums of money, as this article highlights.
My only real concern is how much battery waste this will lead to. Cells need to be replaced every 3-5 years. Until superconductors or high-energy-plasma devices become reality, the only somewhat-environmentally-safe way to store energy long-term is thermal. Hopefully molten-salt storage technology succeeds in this regard.
The battery's purpose isn't power generation, it's load smoothing, like a capacitor in electronics. It has to be able to provide (or absorb) a lot of power in a very short time (milliseconds to seconds) to keep the grid in spec; solar can't do that, fuel-powered generators respond too slow, etc.
So even if they built a solar/salt power station, they'd still need the battery.
Well it was built to stabilise a nearby wind farm, but yeah I don't think it cares where the power comes from to charge it.
The Australian Energy Market Operator, which operates the grid, is essentially a large integer linear program (CPLEX, I believe). It know what equipment is attached to the distribution grid and what the demand is, and it decides what lines get turned on (and in which direction; the Bass Straight connection can work both ways, for example) and whether storage systems are storing or draining and whether new turbines get turned on. It optimises for overall cost.
The thing that complicates it is that the Hornsdale battery reacts faster than the integer linear program. A pumped hydro system (such as you find in the Snowy Mountains) can't turn from storing to generating anywhere near as fast as the battery can. So while the AEMO is working how how best to balance the grid, Hornsdale has already started doing it.
That's one of the reasons the existing power companies didn't like it: they all realised that they wouldn't get paid as much because by the time AEMO decided who should be pumping energy into the system, Hornsdale would already be doing it.
sub f{($f)=@_;print"$f(q{$f});";}f(q{sub f{($f)=@_;print"$f(q{$f});";}f});
Tesla got paid. It's Neoen who may not be getting paid, because the system is re-optimised in 15 minute increments, and Hornsdale responds much faster than that. They are working on 5-minute settlements now.
sub f{($f)=@_;print"$f(q{$f});";}f(q{sub f{($f)=@_;print"$f(q{$f});";}f});
Not necessarily, because the "benchmark" used here was "conventional steam turbine". Steam turbines indeed do take quite a few (tens) of minutes while to take the load depending on their status.
Gas turbines on the other hand do not, and neither does hydro. Both are commonly used for load balancing specifically for this reason. The comparison is... odd. Having read the paper, I'm assuming that this is some kind of a unique market that didn't actually have access to any common spinning reserve sources. The size of the market, with 30MW being sufficient for all of its load balancing for the time tested appears to confirm it. This seems to be a very localized grid with minimal interconnections with outside world for load balancing purposes. Most of the lucrative markets in the world are large interconnected ones.
My understanding is that the model is based on slow spin-up fossil plants, and doesn't accurately account for a battery that can go from 0 to 100% in a fraction of a second.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
The current method for keeping the frequency stable is lots of plants with heavy turbines and generators spinning at high speed - 3000 or 3600 RPM. If load increases or decreases, it takes time for all this mass to speed up or slow down, and this keeps the frequency stable.
Molten salt plants use these same, heavy steam turbines, and so will act to keep the network stable like traditional plants.
It is when this first system is not enough that batteries and gas turbines come online, to support the network while steam plants ramp up their fuel burn (or molten salts increase their steam generation, which should be faster than coal- or oil-fired plants can). Batteries can also absorb power while plants overproduce if the load unexpectedly drops.
Prediction for end of Universe #42: Fencepost error in Quantum_bogosort.cpp
My understanding is that the model is based on slow spin-up fossil plants, and doesn't accurately account for a battery that can go from 0 to 100% in a fraction of a second.
That is correct. The fastest Frequency Control and Ancillary Services market in Australia is billed in 6 second changes and this is the primary of the 8 FCAS markets that the Tesla battery operates in. This is the same market used by emergency systems such as load-shedding / rapid loading. We used to participate in the latter service where I worked as we had some small gas turbines on site. The AEMO's control system could request setpoint changes every 4 seconds. So if somewhere a power plant tripped off line, it would be several seconds before AEMO knew, several seconds more for them to send us a signal, and then up to a minute for us to add a pathetically small about of power to or from the grid in response, and that's assuming we don't trip our turbines on load as a response to the swinging demand.
https://www.aemo.com.au/-/medi... This report details some of the performance differences compared to conventional FCAS providers. Specifically the two graphs on page 6 are quite telling. As is the following quote:
"The Market Ancillary Services Specification (MASS), which specifies each market ancillary service, and how it is to be quantified, does not address performance requirements for regulation FCAS. All regulation FCAS is essentially considered to be equal and interchangeable, and providers are paid the same price per MW of enabled service, regardless of performance." And that is Tesla's main gripe.
Additionally there is the contingency response. On page 7 of the above report is shown how Tesla's battery added 20MW to correct a frequency event as a result of a coal plant tripping offline in less than 5 seconds. Tesla started correcting the issue before the AEMO would even have sent a signal out that there was a problem. And again the note says they don't get paid for this awesome performance.
The AEMO have been talking about adding a sub 1second market to the FCAS and overhauling the FCAS market since early last year. And so has every other major grid operator around the world as this fast technology comes on to the market. A lot of research has been done into this not only because the likes of Telsa want to get paid to play, but also if more of these services come online and the control system is incapable of reacting fast enough then it could lead to more instabilities than they were trying to solve in the first place.
Imagine five or ten of these in America.
It'd be a real infrastructure project that would benefit people.
Oh wait, not under this Congress.
There are few if any places on the US grid where they have the stability problems that the Australian battery is being used to manage.
Comment removed based on user account deletion
You mean the Samsung Galaxy Note 7. The specific model of Samsung that had significant hardware issues. Unlike nearly every model of iPhone, which have had issues with bending, expanding, burning, exploding poor battery life, poor wifi, shitty antennas, cracking screens, unresponsive touch, throttling the CPU on old phones...
If you're going to shill for Apple, at least get your models right.
It wasn't. It was built to stabilise the grid, the goals are slightly different.
A power grid requires a precise balance between supply and demand to exist at all times. If that balance is disturbed, then there can be rapid collapse. This happened in the major South Australia blackout in 2016; which occurred when a major power line failed causing a supply deficit. The deficit was large and most of the major power plants in the region were shut down to allow wind and solar to operate. Wind and solar farms have no supply response capability, so could not assist. The few fossil fuel plants which were active at the time used all their reserve power to make up the deficit. This should have been enough to stabilise the grid, however, the system continued to deteriorate due to an unexpected problem: the wind farms in the region started shutting down on an undocumented (*) safety system which protects the wind turbines from grid instability; this caused a chain reaction making the grid instability progressively worse, until collapse was assured.
The grid operator AEMO (like grid operators in other countries) pays generators (and other companies) for grid stability services, which means a capability to rapidly increase or decrease demand/supply in the event of a grid imbalance.
Due to the nature of the SA grid, with weak long power lines, low demand and high wind/solar generation without the capability for supply response, the fossil power plants in the region were being paid huge grid stability fees to run their plants at idle, just so that they could step on the gas in the event of a power line failure or power plant failure.
The wind farm operator decided to get in on this stability services market by procuring a battery grid stability system. With the battery, they have secured a long-term contract with AEMO for supply of 30 MW-20 minute stability services. The battery is oversized for this, and allows the battery owner to bid for supply of additional stability services on a day-to-day basis when prevailing grid conditions require additional supply of stability services.
The spare battery capacity when not being reserved for grid stability usage, can be used by the battery owner for price arbitrage - charging using low cost overnight power and discharging at peak times when power costs are high. However, the main business case was income from supply of stability services. The key issue here is that the performance and location of the battery are ideal for grid stability services and its generous supply has greatly reduced the market price of stability services.
(*) - generators connected to the grid have to have "fault ride through" capability - so that if there is a grid voltage anomaly, or a short grid interruption, the generator must not shut down. While the output is allowed to reduce in the event of low or absent grid voltage, it must immediately be restored once normal grid voltage returns. For example, if there is a brownout at 50% of normal voltage, the generator must not shut down for at least 1 second. In SA, the wind turbines officially complied with the ride through capability required and declared to the grid operator. However, the manufacturer included an undocumented setting which limited the number of ride through events in a given time period - once this limit was exceeded the ride through capability was disabled and the turbines would trip immediately on a grid problem. This was not declared to the grid operator and hence not included in their simulations and stability calculations.