How deconstructing energy costs to recognize grid-stability attributes can foster long-duration energy storage and advance the energy transition

The idea: breaking up the kWh compensation

When a coal or a gas plant puts a kWh on the grid, it gets paid a certain dollar value per kWh for the services it provides. Although it is paid for just the kWh of energy, it actually provides five different services that are bundled into that one price it is paid:

Energy (in kWh)
Load following capability (in MW/MW or turndown percent)
Synchronous inertia (in MWs)
Short-circuit current (in short-circuit MVA/MW)
Grid resilience in the form of up and down power regulation (in MW/s)

When a variable renewable energy (VRE) asset like solar photovoltaic (PV) or wind puts a kWh of energy on the grid, it get’s paid the same dollar value per kWh as the coal or gas plant, and yet it provides just one service:

Energy (in kWh)

VRE does not provide the services 2 – 5 above, and yet it is paid the same price per kWh. Long duration energy storage (LDES) systems, specifically synchronous LDES, can provide those services. If this price per kWh can be broken apart into those five components, it will enable long duration energy storage (LDES) deployments to receive compensation for those services that they provide. This in turn can make LDES profitable, which will ultimately enable the further deployment of low-cost VRE, and help achieve a least-cost grid moving forward.

The last 20 years: spending down surplus load-following capability

In 2010, trade organizations like the American Wind Energy Association (AWEA) pushed back on energy storage. The argument was that the grid did not need storage in order to deploy wind, and that wind could lower electricity prices.

At the time, this was true. The grid was coming from a position of every power generating asset being a load-following asset. For example, if every generator is load following (could ramp power up and down), and the difference between user load is only twice as much during the day as it is at night time, then the grid has a 2x surplus of load-following capability. Such a grid can “spend down” this surplus load-following capability by building low-cost VRE. Even though the VRE is not load-following and produced power based on weather, the grid can handle this integration with the result being a slightly lower surplus of load capability. If the VRE produces energy at a lower cost than traditional assets, then this also lowers the net cost to consumers. Win-win.

The VRE penetration limit

“Spending down” surplus load-following capability only works while there is a surplus. As the VRE penetration grows, the percentage of load-following assets drops to be on par with the actual variation in the load on the grid. At this point, two things can happen:

The grid can become unstable as the grid no long has the ability to match generation to load
Build-out of VRE can stall

This is already being seen around the world in large island grids (think Britain or Australia) that have large penetration of renewables. A good example is the Iberian Peninsula (Spain and Portugal), which is only loosely connected across the Pyrenees Mountains to France (and thus is a large island) and has a large penetration of VRE (primary solar). Prices for electricity can hit near-zero for as many as 10+ hours per day, making it economically impossible to build out new solar PV.

LDES as the solution, if recognized

As traditional assets like coal and gas retire and are replaced by energy produced by VRE, the grid is losing stability attributes (items 2 – 5 above). Long-duration energy storage, and specifically “Synchronous LDES”, can provide those services 2 – 5. This means that Synchronous LDES can step in and ensure grid reliability throughout an energy transition.

Of note, batteries are inverter-based resources (IBRs), as are wind and solar PV, and do not provide synchronous inertia or high short-circuit current ratios. These assets provide services 2 and 5, but not 3 and 4. Non-synchronous LDES are IBRs as well. See more in the appendix.

However, the energy markets today do not compensate for any of these services, and LDES projects are not yet economic based on arbitrage (energy shifting) alone.

Deconstructing capacity payments

Similar to the price per kWh being split into the five items listed above, capacity payments can be broken into different classes by timescale.

A. Millisecond timescale, e.g. fast frequency response or other service where an asset puts more or less power on the grid based on its measurement of grid frequency (without waiting for a central command)

B. Sub-minute time scale, or primary frequency response (or spinning reserve, or similar), receiving a centralized signal on 15 second or similar increments to add or remove power from the grid to arrest an instability

C. Sub-hour time scale, or secondary frequency response (or reserves, or non-spinning reserve), receiving central command signal on 15-minute increments, or similar, to restore frequency to baseline.

D. Multi-day resilience capacity. This would be new, and would protect against multi-day weather events (resilience capacity).

E. Seasonal capacity market, as done in some markets today, providing capacity payments to assets to be online and ready during the peak demand (e.g. summer) months.

F. Three-year forward capacity market, as done in some markets, providing a payment to encourage the building of new assets.

How would this be implemented in the current DA and RT markets?

Splitting of the value-adds into constituent components would be implemented partially in the energy market and partially in the capacity markets.

Imagine this end result:

a) an energy asset bids into the DA market which now includes not just capacity (in MW) but also includes in the bid the asset’s synchronous inertia value (in MWs), short-circuit current ratio value (in MVA/MW) and its ramp speed (in MW/s).

b) The market clearing engine would have a load forecast for each hour of the upcoming day (required MWh) and would also have stability requirement forecasts (needed weighted average MWs, MVA/MW, and MW/s). The stability requirements forecast would come partially from the distribution utilities (e.g. node by node MVA/MA) and partially from the ISO/TSOs (e.g. needed interconnection-wide MWs).

c) The market clearing engine, based on the needed attributes (MW, MWs, MVA/MW, MW/s) would close on a separate dollar value for each attribute.

d) The asset, if called upon, would be paid the summation of the price for each attribute times the amount of that attribute that it provides.

Similar to the play between DA and RT markets today, assets bidding and accepted for DA attributes would commit to being available to provide them and would bid that value real time. Other assets could also bid RT only, and all assets that deliver RT would get paid the RT price for each attribute they provide.

A week ahead (WA) market could be created to mirror the DA market—as mentioned in item (D) above—to provide multi-day resilience capacity. A week ahead (WA) capacity market could function similarly to the DA capacity market. It would have assets bid on a 3 – 7 day ahead basis with a requirement to respond within 24 hours. This would not impact the ability of these assets to still bid on the day ahead (DA) and real-time (RT) energy and capacity markets, but would ensure power production in the case of a severe, multi-day weather event. This would recognize the value that multi-day assets provide, whether that is fuel-fired plants with fuel storage on site or new multi-day energy storage technologies like Form Energy, Noon Energy, and Borehole Thermal Energy Storage (BTES).

How would this be implemented in the planning functions?

The three-year forward capacity market (FCM) in places like ISO-NE and MISO can also incorporate this split. The markets can, instead of just bidding out 3-year forward power capacity (MW), could also bid out the 3-year forward need for other attributes (MWs, MVA/MW, MW/s). The thresholds for what is needed for the other attributes would come from reliability modeling of the grid done by NERC and its regional entities. This would help those grids ensure that they remain reliable as they grow and do not have to issue emergency procurements because they did not plan for these stability attributes ahead of time (as happened in Great Britain in 2019).

Conclusion

Reliability attributes are being lost as coal, gas, and other traditional energy generation assets retire, and synchronous long duration energy storage can provide those attributes. Variable renewable energy gets paid full value for the energy it provides but without providing those attributes. Therefore, splitting the value that is paid for a kWh into constituent pieces can allow for storage technologies to get paid for the value they provide, which simultaneously allows for markets to optimize costs, ensuring they have what they need without paying more than what they need to.

Appendix: Example attribute provision

	Energy (kWh)	Load following (turndown %)	Synchronous Inertia (MWs)	Short-Circuit Current (MVA/MW)	Reg up/down (MW/s)	Multi-day Resilience Capacity (MW)	Emissions Free	Fuel risk
Gas	✅	✅	✅	✅	✅	✅	❌	Gas $
VRE	✅	❌	❌	❌	❌	❌	✅	None
Batteries	✅	✅	❌	❌	✅	❌	✅	Charge $
Multi-day energy storage (e.g. Form, Noon)	✅	✅	❌	❌	✅	✅	✅	Charge $
Synchronous 10-20h LDES	✅	✅	✅	✅	✅	❌	✅	Charge $
Load-following renewables (geothermal, concentrated solar)	✅	✅	✅	✅	✅	✅	✅	None

Each storage and generation asset provides a different set of attributes.

Not all kilowatt-hours are created the same