battery storage and frequency regulation


How does battery storage support frequency regulation on the grid? And how is the battery storage operator compensated?

In: Engineering

Load surge providers, like any power provider, are paid according to the amount of power they provide into the grid. Surge providers are paid more, per KWh, to compensate them for requirements that they respond more quickly than thermal plants. Battery providers get the same sort of deal that hydro surge providers get. The battery is just a new, more geography friendly, way of providing surge to retain balance.

When you connect two (or more) AC sources together it’s really really REALLY important that they by synchronized. Otherwise you induce huge currents in at least one of the systems and will likely damage very large expensive equipment. This happens when you connect two electrical grids, or when you connect an additional generator to a grid.

How fast you can synchronize depends on the generating technology. Large slow mechanical systems might take several minutes to sync up. A battery, because it’s DC going through an electronic DC-to-AC converter, can respond almost instantly. So a battery bank can very quickly kick in, synchronize, and start providing power “instantly” while the main generator takes its time to spin up and synchronize. Batteries make very good bridges for adding power to the grid…they just can’t do it forever. So if a grid needs more power and it’s going to take a few minutes for a new generator to connect, battery storage can provide the buffer to keep the grid up while the generator comes online.

The battery operator, if they’re not the utility themselves, will have a two-way power meter…they pay for power when they take it from the grid, and the utility pays them (or reduces what they owe) when they send power to the grid.

Electricity can not be stored in the wires. When you start using electricity it have to be made at that moment. And the amount of electricity sent to the power grid have to match the amount of electricity used exactly. The beauty of the AC grid is that this is done purely mechanically. The generator anchor and the turbines in the power stations are mechanically linked to the frequency of the grid through the magnetic inductance of the generator. So the turbines move at exactly the same speed as the frequency of the grid. This means that when you start using electricity you load the grid and the energy is taken from the rotation of the turbine. This happens at the speed of light and without any control circuits or human interaction.

The problem we are facing is that a lot of the rotational mass in the power grid is provided by coal and gas power stations. And these are becoming too expensive to operate. They are instead replaced by solar panels and wind turbines which do not have the same rotational mass directly connected to the grid. A lot of places might end up with too little rotational mass as they do not have enough nuclear, geothermal and hydroelectric power stations to provide this rotational mass. One propposed solution is to connect batteries to the power grid. However these can not mechanically be connected to the grid like the turbines but need to go through big complex control logic to supply the exact amount of current that is needed and at the right time.

Electrical load presents itself as physical resistance at the generator. Let’s say we have a generator spinning 3600 RPM, and producing 60hz AC electricity. If we suddenly turned on all the conveyors in a large warehouse, at least for a moment that generator would be slugged with a lot more physical resistance, it might make the generator slow down to say 3400 RPM until it catches back up, and at 3400 RPM it’s only putting out 56.667hz power which is pretty far off the mark and pushes toward damaging sensitive equipment.

But if there’s battery storage, it can take the brunt of sudden increases in demand to keep it at 60hz as well as sucking up the surplus if we suddenly turned all those conveyors back off. Batteries store power as DC though so there will be a VFD or some other mechanism of switching between DC and AC. They can also play a role in interconnecting grids, having the power converted to DC and then back to AC where two different grids meet means the power plants themselves don’t have to do as much work to synchronize their power (they have to sync with others in the same grid but now there’s less worry about plants from other grids not controlled by the operator being out of sync in relation to themselves).

As for compensation I would imagine it’s utility companies operating their own systems, if not then they’ll either have a flat contract for $X per Y time period to provide this service, or use metering to measure overall net use.

In a balanced grid (where demand equals supply) the frequency is constant. Where supply exceeds demand, the excess supply energy builds up as kinetic energy in the rotating machines connected to the grid (i.e. generators and motors start to accelerate) and in turn the frequency of the grid increases. The opposite occurs when demand exceeds supply, and frequency falls.

Grid frequency must be precisely regulated (typically within 0.5% of the specified frequency) to ensure correct operation of everything connected to the grid.

Conventionally, this is by control techniques known as “primary” and “secondary” frequency control.

Primary frequency control is delivered by standing instructions to power generators (and sometimes power consumers or demand aggregators). So, a gas power plant may be contacted by the grid operator and given an instruction “Please generate 100 MW of power. However, if frequency is below 59.9 Hz, increase power by 10 MW for ever 0.1 Hz below 59.9. If frequency is above 60.1 Hz decrease power by 10 MW for every 0.1 Hz above 60.1 Hz”. The plant operator will program that instruction into their controls, and if the frequency changes, a couple of seconds later the plant will automatically adjust to the new power level.

Secondary frequency control is not so automatic. In this the instruction is more like “Please generate 100 MW of power and await further instructions”. The grid operator will monitor the grid and how well primary frequency control is working, and will activate the secondary frequency control as needed, so that the primary frequency control can go back into standby mode.

The problem with primary control is that it has to be immediate – 1 or 2 seconds, which means it has to be automatic, the power plants delivering it have to be very fast to respond, and they have to be fully warmed up and already online. Something like a coal plant can take many minutes to respond to an instruction – coal crushers and converyor belts are slow to accelerate, and furnaces are big and take a long time to change in temperature, so may not be able to offer primary control – something like a hydro plant, as long as it is online, just needs to open its valves a bit wider to respond, so may be able to bid for primary control.

Secondary control can be slower, but it still needs to be reasonably quick (maybe 15-20 minutes) otherwise it doesn’t help in taking the load off primary control. Something like diesel engines or fast-start gas turbines are generally able to start, warm-up and connect to the grid in less than 15 minutes from when the instruction is sent. A coal plant which is already running and warmed up at partial power will also usually be able to make minor adjustments within 15 minutes.

One of the issues with modern grids is that there are lots of things like solar power, which fluctuates wildly in how much energy it can supply. This can cause large frequency shifts, and the grid operators have to buy in lots of frequency control. At the same time, solar generates lots of power in the middle of the day when demand is low – so power plant operators really, really don’t want to leave their plants idling during the day to save the fuel and maintenance. The result is that conventional primary frequency control has become extremely expensive, because there is a lot of need for it, and few power plants willing to provide it.

Batteries have a major advantage in that they can respond quickly enough to provide primary frequency control, but can also provide power for up to an hour or so. They also don’t waste fuel sitting idle. This means that batteries can be quite a cheap method of providing a long-term, reliable combination of both primary and secondary frequency control all in one.

The typical instruction given to a battery frequency control service would be something like – “When frequency is between 59.9 and 60.1 Hz, you can charge and discharge however you like. When frequency drops below 59.9 Hz you must discharge 1 MW into the grid for every 0.1 Hz below 59.9 Hz, up to a maximum of 10 MW. When frequency rises above 60.1 Hz, you must charge 1 MW from the grid for every 0.1 Hz above 60.1 Hz, up to a maximum of 10 MW. The charging or discharging process must continue uninterrupted for a minimum of 30 minutes, unless the frequency returns to between 59.9 and 60.1 Hz. Once frequency returns to the normal range, you must get ready for another charge/discharge cycle immediately.”

Battery frequency response is remunerated in the same way as other frequency response. There is an availability fee – which is paid to the operator guaranteeing that the battery is available – typically priced in $/MW/hr. So a battery with 50 MW response capability, which is guaranteed to be available for 8000 hours per year, might submit an auction bid of $5/MW/hr (or $2 million for a year). There are refunds and penalties if the battery isn’t available. In addition to the availability fee, there are response fees – so, every time the grid requests a charge or discharge response, they get paid for the energy delivered or removed from the grid. The battery operator still has to buy or sell the electricity they charge/discharge – the the response fee means that they get a bonus price when discharging on request, and a special discount when buying on request. It also means that if the battery has spare capacity, then when the frequency is normal, the battery operator could “buy low, sell high”, as long as it doesn’t intefere with the frequency response capability.