Power engineers worry that, as more renewables are added to the grid, replacing old coal, gas and nuclear plants, we will lose lock-step AC synchronous system stability, since the latter had large heavy rotating turbo-generators that provided system inertia against frequency perturbations. The big plants’ rotational inertia acts as a buffer to grid frequency changes, and to varying supply and inductive loads. However, PV solar has no rotational inertia, and wind turbines not much, though direct drive machines can provide some. With more renewables on the grid it will become more of an issue.

So what can be done? Grid power is supplied at 50 Hz, but this frequency can be allowed to vary slightly, as can the voltage, compensating for some variations in supply and load/demand. However, as renewables like solar and, to a lesser extent, wind expand, more frequency support will be needed. There has been talk of adding “synthetic inertia” – i.e. frequency support provided by other means. But there are disagreements about whether batteries or other storage systems can be used to do this well, or at all. At the very least they will need some fancy electronics. Angular momentum is hard to beat. But big combined-heat-and-power plants do have that. So do big flywheels. Though not that much – they can’t be run for long without losing power. The turbines in tidal barrages and tidal lagoons would be more effective, smaller tidal stream turbines less so.

However, all that is some way off. As a stop gap, for rotational inertia, use is sometimes made of so-called “synchronous condensers“. Basically these are like the back end of the turbo-generator part of a power plant, spun using grid power, but not generating power. Old power station units can be used in this way.  It is also possible to run live power plants with no power being produced by their generators just coupled to and freewheeling against the load, to provide some frequency stabilizing inertial load.

In theory then, we could leave some large old coal (or even nuclear) plant turbines on the grid just to provide rotational inertia in this way, without generating power. More likely, power output from smaller gas-fired plant will continue to be used for grid balancing and that ensures frequency stabilization – they have some rotational inertia.

Wind turbines provide less rotational inertia, and the way some operate makes them even worse in terms of grid frequency stabilization. Some are not directly coupled to the grid, but operate asynchronously, generating variable DC power which is then converted to AC in an inverter. That allows the turbines to rotate at optimum speed and maximum output, varying with the wind, but their output is not frequency stable, and that has to be dealt with by the inverter. Some newer wind turbines are however directly coupled and run synchronously at fixed grid-defined rotation speeds, so that there is better frequency stability, although less total energy output. So the grid matching problem can be dealt with to some degree, either by direct AC coupling or via synchronous inverter compensators/condensers. Evidently around 35% of all turbines installed recently had synchronous generators, 70% permanent-magnets such as Goldwind, 30% electrically excited such as Enercon. Interestingly, in Germany, new wind farms are now mandated to provide “synthetic inertia” to help with grid integration (see Pulse blog post).

A move to synthetic inertia is also being looked at by National Grid in the UK, with the output from fast response storage being suitably synchronised as one option (see this article for an overview of some of the fearsome analytic issues).

As can be seen, there are a complex set of issues, with as yet uncertainties as to how best to deal with them. Batteries with invertors can provide some synthetic inertia, but for how long? Could PV systems  with batteries  present a synchronized load to the grid even when they are not generating ? There are lots of unknowns, but lots of possibilities being explored too, such as new synchronous invertor systems for use with solar projects.

How urgent is it? The UKERC study of renewable of integration said “analyses of the impact of reducing system inertia resulting from adding variable renewable generation (and so replacing some synchronous plant that would otherwise be providing inertia) have to date tended to focus on the technical challenges that this may pose”, but it says in terms of its impact and costs “of those studies that do address this issue, the typical conclusion is that it is likely to only become significant at high penetrations of variable renewables, i.e. greater than 50% on an instantaneous basis (although it should be recognized that some systems have already reached this level on occasion). Nevertheless, the analyses which consider penetration levels above 50% do generally conclude that even at these very high penetration levels, sufficient inertia-like resilience could be provided, typically through a combination of very fast response frequency control systems and synthetic inertia”.

So there should be time to sort it- although not too much. The UKERC is relatively sanguine. It points to the possible use of very fast-response assets like batteries and also say that “there is considerable inertia in the rotating mass of wind turbines”. It found studies suggesting that in concert they might provide as much inertia as a large fossil generator of the same rated power, albeit with additional control systems being needed. That might make it possible to go beyond a 50% contribution from wind without loss of frequency stability. Indeed, one study suggested that up to 80% might be possible. Well, we will see.

To round things off more speculatively, grid defectors may argue that, if we all went off grid, then we could avoid all these problems- no need then for synchronous matching.  It is true that not all renewable generators need to feed their output into the grid. However, unless you have a lot of off-grid storage, grid linked systems are vital for top-up imports when local wind and/or sun power is not available, and to allow for exports of any surplus, thus helping to balance the variability of renewables. Nevertheless, some generators might opt out of the grid. For example, in what could be one of the most important new wind-application ideas to emerge recently, a container package-scale system has been developed for using wind-derived electricity to make fertilisers by extracting nitrogen from the the air, using the Birkeland-Eyde plasma arc process. It’s initially been thought of as using surplus wind electricity from the grid, but could also spread off-grid, to local centres in rural areas across the world, to meet local farm needs: a new “power to food” option.

However, there is still obviously value in grid linking, and a recent study suggests that, even with a small mostly isolated grid, as in the Republic of Ireland, high levels of useful output can be attained without major curtailment losses, if the synchronous constraints can be relaxed a bit. Though that may have its issues. Yes, grid frequency-run clocks can drift.

Nevertheless, as we have seen there may be solutions, and smart grid demand management may be an option for avoiding some problems. There is certainly a big literature on that. Although see my separate article on smart meters and Blockchain. There can be problems with some new smart power integration and trading systems.

It can all get very complex for a non-expert. I’m certainly not one, so apologies to any electrical engineers if I’ve garbled some things. But if you want more technical details, here’s more than you may ever want to know, and on wind.

And, finally, if you include nuclear plants, then it’s not just complex: it can also get controversial. As noted above, their big turbo-generators clearly have a lot of rotational inertia to offer for frequency support. But that’s only part of the grid balancing problem. Can nuclear plants load-follow and help maintain overall grid reliability when there is a lot of variable renewables on the grid with potentially no output at times? See my next post.