This is an edited version of a post submitted to the power-globe listserv in June.
Most people who work with the tools of mathematical optimisation are familiar with the idea of complementary slackness – a binding primal constraint is associated with a positive price (dual variable), while non-binding constraints have price zero.
With that in mind, let’s switch to electricity markets. In Australia’s National Electricity Market (the NEM, our east-cost interconnected market), we have frequency control ancillary services (FCAS) reserve markets – 8 of them, up and down for regulation and primary, secondary and tertiary contingency response – which is nothing special. Reserve markets are used here and elsewhere because you can’t run an auction for a response that is required within seconds. Indeed, FCAS is always required in a power system, which is another way of saying that the FCAS capacity constraint is always binding. Hence, FCAS capacity is almost always given a positive price, regardless of how it is bought or supplied. In the NEM, like for many ISOs, this reserve capacity is priced by running a reverse auction for it.
Moreover, the required level of FCAS capacity in the NEM is dynamically adjusted according to system conditions, including active generation sets, forecast demand, transmission network congestion, scheduled line outages, weather, and so on. In practice, the engineers at AEMO (the NEM operator) have rules or algorithms for determining the required level of FCAS capacity given system and environmental conditions. When FCAS capacity is procured by auction in this way, the level of capacity is determined by sound engineering judgement, while the cost of the service is determined by the participants’ bids. Additionally, the AEMO currently co-optimises generation and FCAS for least-cost system dispatch, so this is not a strictly energy-only dispatch rule (this approach is also used in many other electricity markets).
Importantly, the FCAS markets are oblivious to the technology that is used to provide the services – it could be a synchronous generator, a large load or an aggregator of many small generators and/or loads. This makes the Australian FCAS markets very flexible, and able to respond to both the future growth of renewable generation and greater controllability of devices embedded in the distribution network. On this topic, UNSW have done some nice work comparing various FCAS market structures and rules in the face of increasing renewable generation (e.g. http://ceem.unsw.edu.au/sites/default/files/documents/WIW13_Riesz-FCAS-2013-09-02a.pdf).
Now consider inertia.
The inertia of the system has always been a non-binding constraint on the power system, as it came part-and-parcel with synchronous generators. As such, it is associated with zero price. Given this, what ISO would bother constructing a market for a service that they knew was going to be supplied for free? Of course none have.
However, as renewable generation, or indeed any converter-interfaced device, increase their proportion of generation capacity and load in a system, there will come a point where additional inertial capacity is required, supplementary to that provided by synchronous generators and motors. This is the point at which the inertia requirement becomes binding, and inertia is given a positive price. Some systems are closer to this constraint binding than others, but in all cases, the level of inertia required for a adequate system stability can be determined by sound engineering judgement.
As such, the principles applied to FCAS capacity markets are well suited here too. But there is a catch: unlike FCAS, synchronous generators and induction motors cannot withhold their inertia. There is a strong argument that if you decide to run an auction for inertia, the contribution from these sources should be still be compensated, even though they cannot help but provide the service.
As such, this becomes effectively another service payment to synchronous generators, in addition to any FCAS payments they receive, as long as they are among the cheapest suppliers of power and services (i.e. part of the least-cost system dispatch mix). All other things held constant, if the number of these generators drops, and the price at which alternative/synthetic inertia is supplied is greater than zero, then the payments that the remaining synchronous generators receive for their inertia should increase.
But more importantly, again, the market is oblivious to the technology that supplies the inertia, so (assuming it is all of the same quality) it doesn’t matter if the inertia comes from a SMES, the kinetic energy of a wind turbine, software-defined inertia of a PV or battery converter, or just an old thermal generation plant converted into a rotating synchronous condenser.
In short, there doesn’t appear to be any theoretical impediment to introducing auctions for inertia into the current market-based arrangements for operating power systems. Doing so would provide a form of capacity payment to synchronous generators, thereby compensating them for a service they currently provide for free, while over the long run opening the market to alternative providers of inertia and letting innovation rip.