First, some background on the operation of the electric grid. For all practical purposes the grid itself has no capability to store electricity in any useful amount. The generation must match, on almost a second-by-second basis, the demand at all times. This matching of supply and demand is critically important to the stability of the grid - if it is done badly, the customers will experience the effects very quickly, in the form of brownouts or worse. Almost needless to say, a steady supply of electricity is one of the fundamentals of an industrialized society; our standard of living takes a very quick drop if the supply isn't available when we need it.

In the olden days the utility companies were responsible for doing this matching. With deregulation, in most jurisdictions, there now exists a quasi-government organization whose job it is to arrange for and then control the different suppliers. Names like ISO (Independent System Operator) or TSO (Transmission System Operator) are used to describe this organization. In Ontario it is named the IESO, or Independent Electrical System Operator. The IESO has a long historical record that allows them to predict quite accurately how much electricity will be demanded at any point in time. With conventional power generation they are able to control the supply quite accurately. By controlling the supply and knowing the demand they could highly optimize the generation to lower costs and emissions.

The advent of wind power introduces a new uncertainty into that operation. Not only could the instantaneous demand change in (hopefully small, on the order of a mw or two per minute) unexpected ways, but now the generation can also change in not-so-small unexpected ways. The wind operators release no information they don't have to, but one Warren Katzenstein, a graduate student at Carnegie Mellon University, managed to get some minute-to-minute data for a presentation and a study. By the way, this is the only minute-to-minute chart I've ever found. I added the vertical 5-minute markers and you can see that the output can go up or down by multiple tens of mw's in a single minute! I expected some variation, but this is extraordinary. For a typical 200-mw Ontario project the changes would be twice what this chart shows.

The remaining generators (coal, gas, hydro and nuclear) must be tweaked not only to adapt to changes in demand, but now also to adapt to changes in wind supply. If our goal is to reduce fossil emissions you would use wind to replace coal and gas plants, and thus you would tweak them first. Tweaking nuclear isn't practical in any event; it runs pretty much at full power constantly, providing "base load". Depending on the facility, tweaking hydro is possible. But using CO2-free generation plants to back each other up doesn't save on CO2; only replacing fossil fuel plants does. Coal plants don't like to be tweaked, and are too slow to respond to the changes that wind can introduce. By default gas plants become the preferred choice. With the current penetration of less than 1% in Ontario this doesn't represent much of a problem, but as the penetration increases it becomes a larger problem. Ontario is currently in the process of building at least 3 new gas plants. Alberta, with a 4% penetration, is in the process of building a fossil-fired backup plant, and even Bonneville, with all of its hydro, is also.

An important consideration of how to provide this backup is determining how quickly the controllable supply can be tweaked to match the demand minus the wind supply. The best circumstance would be to have a unit that consumed no fossil fuel until it was needed and then switch it into full production instantly. This circumstance is, by the way, what the advertised CO2 savings numbers all assume. Unfortunately no unit, even gas, is that quick. All ISO's maintain a "spinning reserve" of generators that are kept in phase with the grid but generating at less than their capacity. Some of that reserve is used to back up potential failures; and now some addition reserve is needed to back up wind. Also unfortunately, when a fossil generator is run at something below its capacity its emissions go up, potentially enough to cancel out the savings from burning less fuel - and as mentioned above there is a surprisingly small reduction in fuel burned to begin with.

There are two types of gas turbines used in power generation - Simple Cycle Gas Turbine(SCGT, or sometimes OCGT, Open Cycle Gas Turbine) and Combined Cycle Gas Turbine(CCGT). CCGT is the most efficient, emitting around 0.35T/mw-hr, as opposed to SCGT, which emits around 0.5T/mw-hr. Unfortunately, CCGT is limited to how quickly it can ramp up or down, with a typical ramp rate limit of 5mw/min, compared with SCGT, which can handle changes of multiple tens of mw's/min. The demand on Ontario's grid seldom, if ever, changes unexpectedly by more than a few mw's/min, so CCGT is entirely adequate for this application. All of Ontario's major gas turbine plants (as of mid 2009) are CCGT's. Unfortunately, the changes in wind output are far beyond the capability of CCGT to handle.

Let us say you wanted to produce 200mw of electricity. Two choices come to mind: (1)install a 200mw CCGT and be done with it, or (2)install a 200mw wind farm and 200mw of SCGT. But that gets very expensive to operate - remember that a 200mw wind project may average only 50mw, leaving an average of 150mw to be produced by SCGT. It also creates more emissions than (1). To save some money and emissions, you'd probably have a mix of CCGT and SCGT, and then have to use a higher proportion of SCGT in the windier months, when the variability of wind is at its greatest. Option 1 looks better and better, doesn't it? Kent Hawkins has analyzed this at length (especially "My Articles", item "D"). Kent has also published on WCO's web site. Using any reasonable set of assumptions, it looks like option 2 ends up producing more emissions than option 1. Lang, 0.2mb and Hewson, 0.1mb have also come to similar conclusions.

If all this weren't bad enough, there's more bad news - truly bizarre. What happens when there's too much wind generation? Generally you sell it off, sometimes paying people to take it as Ontario sometimes does. To make matters even worse (if that were possible), consider what happens when there's even the possibility of too much wind generation, where you end up taking an emission-free nuclear plant offline and replace it with a "25/75" mix of wind and gas. Talk about shooting yourself in the foot.

If you've been doing much research prior to reading this, you probably have heard the argument, here advanced by AWEA that wind power variations are just like other power plant failures, and can be handled by the same techniques. Further, using statistical techniques, the argument goes on to claim that a penetration of 20% can be handled with current backup facilities with no significant extra emissions, due in part to geographical dispersion. This may explain why proponents are also big on the "super grid". This logic sounds spookily similar to that used by AIG as they piled credit swaps on top of each other, including the use of statistics to make the risk appear lower than it really was.

A variant of this is the idea of "net demand", where wind power is treated as a reduction in demand, as opposed to additional generation. Traditional dispatchable generators then "merely" supply the delta, albeit with "just" increased unexpected changes in demand. GE, a major supplier of all types of generation, pushes this variant, perhaps in part because wind power (and solar power too) generates power so randomly their computer generation modeling tool (MARS) isn't able to model it. Ontario's IESO apparently has adopted net demand, 0.5mb, section 3.1.3, and I also find it interesting that Ontario's plan, 0.2mb, tables on page 5 calls for new SCGT generation, in the same amount as the planned future wind generation capacity. Even a system with lots of hydro, like Bonneville, has problems with wind variability.

And while I'm at it, take a closer look at the AWEA link above. On the first page there are 7 bullets with 4 external references all extolling the emissions and fuel savings of wind. The assertions look pretty impressive, unless you take the time to read through to the references. There's no "there" there. NONE of the bullets NOR ANY of the references provide real-life numbers, just more assertions and computer models.

Here's some typical questions.

• Couldn't some conventional generators be shut down completely? Of course they can. But if too many of them are shut down and the winds drops quickly enough the ISO is then forced to start cutting power to users. In 2008 the Texas ISO got caught, with the resulting brownouts and shutdowns making the news. And I think it's safe to say that any sort of interruption is far more costly, in terms of both dollars and emissions, than providing proper backup. I see no way around this - the ISO has to keep extra fossil fuel generation online and spinning, and this produces CO2.
• Couldn't you use hydro as the backup? Of course, within limits. But that assumes you always have it available. Generally hydro is already being used whenever it is available, as the water is "free" - both in dollars and in emissions. What would you in turn replace the hydro with? You'd end up just transferring the generation from hydro to fossil, back where you started.
• Doesn't the wind always blow somewhere? Not as much as you'd think. John Harrison did a study that quantifies the relationship in output. In summary, within 400km the output between wind farms is highly correlated. Tom Adams has done similar research, with similar conclusions.
• Couldn't you control the fossil turbines to minimize the emissions. Of course you could, and I would expect the ISO would do so, within all the other parameters they must follow. But whenever it's running, it's still producing CO2. The only question is "how much?" and that can only be answered with a study, a study that seems to not exist.
• Couldn't you build a super grid and move wind energy (and solar energy as well) around from where it was generated to where it was needed? Of course you could, but at what financial and environmental cost? Remember we started out with just wind farms, then we got into having backup, and now all of a sudden we're into reworking our entire grid. I bet before we're done we'll be talking about some very expensive energy storage projects also. In the meantime, the Chinese et al will be installing coal plants that produce a consistent kw-hr for 3 cents. We already have high wages, and when we add high energy costs, you can guess where the energy-intensive industries will go. Exporting our manufacturing base is certainly one way to cut our emissions; sadly the net effect on the planet is an increase.

So far we've been talking about the requirement to balance the grid on a minute-to-minute basis, and how extra fossil emissions are necessarily generated in order to do so. There are other patterns of demand that operate on different time scales other than minute-to-minute. There are also daily, weekly, seasonal and long term patterns. I hope that by understanding the shorter-term situation the reader can then apply that same reasoning to the longer terms as well. One potential solution for the daily problem would be to use some type of storage that would effectively isolate the turbines from the grid, and allow their output to be used in a controlled manner. Batteries, compressed air, heated salts, electrolysis are typical candidates. Sadly, none of these technologies are even close to being deployable, and may never be. If they were I would be much more positive about the contributions wind turbines could make to our grid. For terms longer than daily wind turbines do not provide much relief either. Their output is "non-dispatchable" - it cannot be commanded to contribute. Thus sufficient fossil plants must exist to meet the weekly and seasonal peaks.

As far as I can tell, no fossil plant anywhere in the world has been shut down due to the advent of wind. Perhaps a few have been shut down due to the advent of the necessary gas plants, but even that isn't clear. Proponents point to Denmark as an example of what wind power can accomplish, but when you look at the actual numbers, a different picture emerges.

• Notice there's no relationship between wind production and emissions.
• Overall, there's been little savings.
• A closer look reveals their shift to CHP plants and exchanges with neighbors explains almost all of what little savings there have been.

For intermediate-term storage (i.e. daily) there does exist one scheme that might make wind more effective. That would be pumping hydro uphill in times of good wind and thus giving hydro power better capacity to handle times when the wind isn't blowing very hard. Hydro storage is reasonably efficient - I've seen numbers in the 75% to 85% range. However, it does require some fairly specific geographical conditions that are not common in Ontario. Lake Erie would be the most obvious storage pool, and is already being used in this fashion. I don't know if Niagara has the needed capacity (especially since the upstate NY wind farms are no doubt thinking the same thing). I also wonder what the total losses would be by the time you transmitted the wind energy to Niagara, pumped the water uphill, ran the water downhill and then sent the power back to the user. There's also the difficulty of balancing competing interests, not just between countries, but among shippers, boaters, shore birds, cottage owners and all the other users of Lake Erie. In any event the costs and losses of this operation are not included in any current wind promotional figures.

One action that an ISO could take to improve the operation of the grid would be to remove as much of the uncertainty as possible, thus allowing for the low-cost and/or low-emission generators to be used at their best. Towards this end ISO's are now requiring suppliers to forecast how much energy they will produce, or are creating their own forecasts. Alas, this letter from Environment Canada, in response to a query, shows there might be problems. Even the WSJ has opined on this topic in a well-written column.

Typically this forecast would be used in the ISO's "day ahead" planning, when they figure who needs to be available when. For wind suppliers, this means a highly tailored weather forecast. And where there is money to be made, someone will appear to make it. The results to a google of "wind energy forecasting" will provide the reader some interesting hits, certainly including better explanations of this whole topic than I have provided. If the suppliers do not meet their forecast some ISO's are fining them, and I assume the fines in some way compensate the ISO for the excess costs or emissions caused by a bad forecast. There are also people employed measuring and statistically analyzing how consistently each supplier meets their commitments so they know whose forecast they can trust. All of this, just to optimize a resource that nobody knows the value of in the first place. Incidentally, the ability to accurately forecast also affects the Capacity Credit, as discussed on my generation plant elimination page.

Even if I'm correct in my assesments of these issues, does it really have any detrimental affect on our plans to cut emissions? There are likely two poor results. The first one is that it overvalues the contribution of wind turbines to solving the CO2 problem, thus potentially directing scarce resources away from more effective solutions. If Ontario thinks they are getting 1.0x reduction for y dollars but in fact they are only getting 0.5x, this might well lead to a situation where another solution could yield .75x but isn't pursued due to its (incorrectly) lower perceived value. The second reason is that overestimating the CO2 reduction of wind leads countries to likely emit more CO2 than, for example, Kyoto allows them to. In addition to breaking treaty obligations, this error could confuse the measuring of, modeling of and responses to the increase in CO2.

Note that I am not saying that there is no CO2 benefit to wind. My primary point is that we don't know, and there are very good reasons why it is likely much less than advertised. It is bothersome that apparently no government seems eager to tell us this. I wouldn't expect the industry to do so, unless it was good news. In addition to all the above problems, which are significant, there's the added unknown of how much energy turbines receive off the grid in normal operation. That number, which could be a significant percentage of their output, seems to be a closely guarded secret.