Climate change

May 28, 2010

TCASE 10: Not all capacity factors are made equal (Part 1)

My new goal with TCASE posts is for them to be shorter, more targeted and more regular, with the aim being to break big problems in sustainable energy down into very focused questions (each of the new TCASE posts will be a maximum of 1,000 words — my new self-imposed editorial limit for this series!). Editorially, I like to note that if any regular  readers are up for submitting a short TCASE post following this format, please email me and I’ll be happy to discuss your idea. Here’s the first of the new batch.

————————————–

Capacity factor (CF) is the amount of energy a power station generates over time (usually a year) compared to what it could have produced if it had been running at full power for the whole period. (Please read TCASE 2, Energy Primer, for a fuller explanation). The CF for coal-fired and nuclear power stations averages 85-90%, wind farms ~20-35%, solar farms ~15-40% (the higher figure is for CSP with thermal storage). Gas or hydro can be high or low — depending…

Now, it’s very tempting to use these percentages as though they were directly interchangable, and indeed I’ve found that most journalists and bloggers happily do this (or else ignore CF completely and cite ‘peak’ power as though it were the same thing). It turns out, however, that this is a seriously misleading practice, as I’ll detail over the next few TCASE posts.

Consider this.

The Blowagale wind farm on Roaring Forty Peninsula has 50 of the 2.5 MWe (peak) GE 2.5xl turbines (rotor diameter = 100 m, hub height = 75 – 100 m, cut-in windspeed of 3.5 m/s, peak at 12.5 m/s, cut-out at 25 m/s). Its peak power is therefore 50 x 2.5 = 125 MWe. Over a 3-year period, it has delivered 1,115 GWh of energy to the grid. The peak expectation would have been 125 x 8760 x 3 = 3,285 GWh, so the CF is 1115/3285 = 34%.

The Trex coal-fired power station in Smogsville is a 500 MWe unit that’s been chugging away for the last 30 years. Over the last 3 years, it has produced 11,300 GWh (out of a possible 13,140), for a CF of 86%.

Okay, on an energy-for-energy basis, all we have to do is build 11300/1115 = 10 of the Blowagale-sized wind farms to replace Trex, right? Actually, that’s dead wrong — at least in the real world — for many reasons, which I’ll explore in the next few TCASE posts. Yet, that’s the impression that’s often given by ‘advocates’ (to use a euphemism).

First, let’s briefly consider what has determined these two CFs.

For Blowagale, the following issues are important: (1) how many turbines are operating (not out for maintenance), (2) the amount of time the wind speed is above the cut-in minimum (3.5 m/s) and below the cut-out maximum (25 m/s), (3) choice of generator size per turbine (I’ll explain this in a future TCASE), (4) whether the electricity is bought by the grid, and (5) wind speed and duration when within operating bounds. When the wind is blowing at 6 m/s, for instance, the power per turbine will be ~0.5 MWe, compared to the 2.5 MWe peak output in 12.5 m/s winds (between 12.5 – 25 m/s, the output remains 2.5 MWe, as dictated by the fitted generator and gearing, etc.).

As you can see, the CF of a wind farm is very much hostage to the variable (and uncontrollable) nature of the wind resource. For Trex, things are rather different. I’ll refer to an earlier comment I made (BWB), and an expansion by Gene Preston (GP):

———————————————-

BWB: A key thing to remember is that, in simple terms, capacity factor = availability factory x demand. For coal, nuclear and gas, their availability factor is determined predominantly by scheduled outages, for standard equipment maintenance, refuelling etc. For plants operated in baseload, demand is nearly constant (i.e. that coefficient is close to 1). A capacity factor for nuclear power might be 90% over a year, assuming 1 month out of each year for refuelling and scheduled maintenance. For the other 11 months, its availability factor is closer to 99% — SCRAMs are what takes this below 100%.

Wind is quite different. It’s availability factor is determined by when the wind is blowing, in addition to scheduled maintenance and, for a wind farm of many turbines in total, the occasional failure of an individual turbine. The engineering availability factor might be in the order of 99% for wind too, but the wind ‘fuel’ is quite a different matter. Sometimes it will be blowing strong enough to deliver near 100% of nameplate capacity, other times it will be 50%, or 20% or whatever. Sometimes, when it is becalmed or too windy (such that the turbines are shut off to avoid damage), it will be 0%. On average, over a year, it will be about 35% in good sites. But this power is not ‘dispatchable’ — it cannot be guaranteed (without energy storage), since the wind is fickle.

What the WWS study says is that for a widely geographically dispersed set of wind farms, you can guarantee, to the equivalent of an 85% availability, a ‘capacity credit’ of about 12%. So, in rough terms, the 12% capacity credit for wind is the equivalent of the 85% capacity factor of a coal-fired power station.

———————————————-

GP:  We need to make a distinction about annual peak, weekends, and low load periods of the year. The coal plant reliability is determined mostly by the forced outage rate (FOR), which could be as good as 95% (5% FOR) when the coal plant is needed most, during the peak load periods. Many maintenance problems can be deferred to the weekend when the load is less. This type of problem usually does not greatly affect the reliability. Then scheduled maintenance is scheduled for light load periods of the year when the plant is not needed. When you do a loss of load probability study, you will find that the greatest loss of load is during the peak load periods, not the lighter load periods. A plant failure during the lighter load periods usually has little consequence, provided the network is electrically stable for the loss of the largest generation within a geographic region.

The annual capacity factor is mostly determined by demand for a coal plant. Coal can go into load following frequently and is dispatched after natural gas and before nuclear, which is even more base loaded than coal. Wind generation can cause gas and coal plants to be backed off because wind had a lower incremental energy cost than either gas or coal. Therefore adding more wind to a region will cause the capacity factor of coal to drop a little, especially when the wind runs during light load periods, which is does frequently. However because coal plants are difficult to dispatch they cannot be run back very far to accomodate wind. Because of the unpredictable nature of wind there must be kept on line a certain amount of gas and coal in the event wind is not sufficient. But there is only a certain amount you can swing gas and coal generators. Therefore as more and more wind is added it becomes more difficult to dispatch the total set of generators.

It’s possible to have some stablity problems with the network as wind is swinging from low to high levels. As you keep adding more and more wind you will reach a point where wind has to be dumped even if there are no transmission limitations. This is because the gas and coal generators cannot be swung enough to accomodate all the wind. Therefore wind is going to have an upper limit, probably no more than about 30% of the total energy. The only way to simulate the network to see how it work is in an hourly simulation model. That model can also be a montecarlo model considering random failures of both generators and line and even wind variability. Every once in a while the hourly model will run into difficulties that require dumping load.

This is the only correct way to model the system.

———————————————-

So, if Trex was operating as a full baseload plant, its CF might be as high as 95%. If it was given an intermediate load-following mode, it might be as low as 60 – 70%. They key is, Trex’s CF is flexible and determined by the grid requirements. Blowagale’s CF is fickle, determined by the wind characteristics, and is (mostly) independent of the grid requirements.

That’s enough for now — more on this CF conflation problem in the next few TCASE posts…

Advertisements

Leave a Comment »

No comments yet.

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Blog at WordPress.com.

%d bloggers like this: