Variable, Intermittent, Unreliable

One of the most common arguments used against renewable power is that it’s unreliable. Though sometimes the word used is variable. At other times, it’s called intermittent. Without turning this into a post about the semantics of the English language, I do think it’s worth a very brief summary (or at least my opinion) on what these terms mean in practice.

I looked up some dictionary definitions of these adjectives:

Unreliable = “not able to be trusted or believed.”

Variable = “likely to change often.”

Intermittent = “not happening regularly or continuously; stopping and starting repeatedly.”

I’ll introduce a couple of others later on…

As I often point out to people, renewable power is not one homogeneous entity, so the appropriateness of these terms varies.

In the case of geothermal power for example, drill a deep hole to extract hot water and it’s not hot one minute and then cold the next. So geothermal power plants can produce steady (little varied, non-intermittent) output.

By contrast, power generated from burning biomass (or its fuel derivatives such as biogas) is no different from conventionally-fuelled power in terms of our discussion here. So long as you have fuel at the power plant, you can burn it as and when required. In practice it could be steady or it could prove to be variable and intermittent. It depends how it’s used. After all, power demand varies, by large amounts and often. Hence so does the output of some of the conventional, fuel-driven power sources (like gas) used to balance it. I don’t hear people complaining about the variability or intermittency of either power demand or gas generation though…

The reason for that is – obviously – that variable and/or intermittent power supply is not a problem so long as it coincides with or can be controlled to match variable and intermittent power demand!

So I think the term we really need to concern ourselves with is reliability, which in this context simply means the ability to achieve this match between supply and demand.

That brings us to sources like solar power or wind power. We can’t control what the weather decides to do and so are dependent – at least in the absence of other measures – on coincidence between supply and demand instead.

In some markets, where hot sunshine drives demand for air-conditioning, the correlation between changes in solar power output and power demand can be quite good. Even then, there are no guarantees against clouds, dust storms and so on.

In the case of wind, here’s 90 days of wind generation data for the UK (Q1, 2015), averaged at half-hourly intervals. It maxes out at 6,627MW and there’s a minimum of just 30MW. It pretty neatly illustrates the terms “variable” and “intermittent”! The data was sourced from Elexon.


UK wind generation Q1 2015


Here’s a zoom in on the first ten days of that data, plotting wind power generation on the left axis along with the total generation (i.e. to meet total demand) at those same times; on the right axis. We could argue about some specific ups and downs coinciding, but overall this isn’t a picture that tells us we can rely on increases/decreases in wind generation coinciding with increases/decreases in power demand!


UK wind and demand Q1 2015


The second graph also illustrates (as we’ve already alluded to) the variable and intermittent nature of demand. The advantage with demand though is that these properties are – to some extent – predictable.

Predictable = “something that happens in a way or at a time that you know about before it happens.”

For reliability in our power system, it actually doesn’t matter so much if something is variable or intermittent (be that supply or demand); it matters that we can predict when these changes will occur, and hence schedule in a solution that “keeps the lights on”.

Also, the further ahead we can predict, the better able we are to schedule in the most efficient and economical solution. If we have to adjust at short notice, this usually means going to faster-reacting but more expensive solutions.

In the case of wind (or solar) a key to reliability is therefore better weather forecasts – more accurate predictions of the wind available for longer periods ahead of the current time. These have certainly improved for short periods (hours) ahead, but are still highly uncertain.

Uncertain = “not known or fixed”.

Look at days two and three on the second figure above (the second and third red peaks). I don’t have any data, but my guess is that weather forecasters could predict a few days ahead that the winds would drop between those days, though the uncertainty around exactly when and by exactly how much was likely quite large until a few hours (maybe less) prior to the actual event. Other generators would have been ready to take up the load, but the exact speed and degree to which these needed to be deployed could have been finalised at very short notice (tens of minutes, or minutes).

Whatever the situation, the lights didn’t go out: the system remained “reliable”.

(That’s not too surprising. Even though demand curves are broadly predictable day to day, the exact minute-by-minute demand is uncertain: the system operator doesn’t know the precise moment when I fancy a cup of tea or my neighbour decides to microwave his dinner. The system is set up to cope with this short timeframe unpredictability, via various “ancillary” grid services).

The key benefit of better forecasts in this case wouldn’t have been to make the system more reliable, but perhaps it could have made it more economic by lowering the level of uncertainty in the supply forecast. This could perhaps have allowed the system operator to schedule a different – cheaper – generating mix as the wind dropped.

The important point is: it is the whole system we need to judge as to whether or not it is reliable, not the individual generators within it!

If we built our entire system from wind*, then it wouldn’t be reliable, because there’d be nothing to take its place if the wind varied out of sync with demand. (*I’ve not heard anyone of sound mind suggest this though, even wind proponents).

If our system contains other generators able to take up the load when the wind drops it can certainly remain reliable. It also helps that wind generators continue to be better designed and operated to more smoothly ramp up or down.

Whether a system is economic (or at what proportions of wind the economics change) is actually a different argument. However this seems to be the one that people really allude to when they are describing “variable” renewables as “unreliable”. In particular the argument usually boils down to the fact that we have to have other power plants available to back up wind farms when the wind drops.

Comparing days two and three on the chart earlier, they both have the same total demand requirements, but on day three there’s far less wind. So clearly yes, we do need generating capacity running on day three, which was lying idle on day two. We need “backup”, no argument.

The problem is that the economics of “backup” are not simple; they are a system-wide calculation.

For example:

  • Does it actually add anything to our capacity requirement? (Lots of installed capacity already lies idle, at night for example; plus we already operate a substantial reserve, in case a giant conventional plant breaks down – yet we don’t regard these as “unreliable”)
  • Which fuel is being displaced when the wind does blow, and what does it cost? (Every unit of energy generated by wind is quantity of fuel we don’t need to burn; maybe the fuel savings offset the costs of maintaining the reserve capacity)
  • Are we a fuel importer? (Or, as I like to think of it, a “wealth exporter”; what value is placed on energy security or balance of payments?)
  • Do we have access to external capacity? (Interconnectors: importing power may be cheaper)
  • Do we have access to stored energy to fill the gaps? (Pumped storage or, in future, batteries; perhaps primed when the wind was plentiful)
  • Are we able to control demand? (It’s almost certainly cheaper than adding supply)

There are other variables you could add which might determine whether adding wind into the system is economic, if it increases overall system costs then by how much, and what percentage of a mix makes sense for wind or other renewables (environmental “value” considerations, grid costs, for example).

However the point is that it’s perfectly possible to be variable and intermittent and reliable. It’s easier if that variability is more predictable and less uncertain, and more challenging if not (as is unarguably the case with sources like wind). If we worry about variability (and uncertainty) on the supply side, maybe we could expend more effort addressing it on the demand side too?

In the end the way forward will depend on a combination of economics, technology and system design; undoubtedly driven by other social and political goals (sustainability, security, job creation etc.).

Either way, what we increasingly need going forwards is a system – both supply and demand – which is flexible.

Flexible = “able to change or be changed easily according to the situation”.

Like this? Then Share it!

Leave a comment

Your email address will not be published. Required fields are marked *