Saturday, August 01, 2009

Western Wind & Solar Integration Study

On Thursday, July 30, NREL hosted the Western Wind & Solar Integration Study (WWIS) Stakeholder Meeting in Denver. Great attendance (probably 80 to 100 or so) and some really interesting updates on this work that began in 2007. If you're not familiar with this study, you can find more information on the NREL WWIS website. Generally, the goal is to assess the feasibility of adding as much as 35 percent wind and solar generation within the WestConnect footprint (WY, CO, NM, AZ, NV, and northern CA). Relying on some gi-normous datasets (e.g. 30,000 30MW wind sites in the footprint with data every 10 minutes for 3 years), this has been an extraordinarily computationally intensive exercise. Unfortunately the solar data has been considerably more sparse.

We'll talk about a few of the observations from this study -- some predictable and others not so much. The bottom line, however, is that having 35 percent renewable energy penetration (30% wind and 5% solar) appears technically feasible although operation of the electric grid would have to be dramatically different than it is now. Essentially, instead of the large number of small control areas that presently exist, it would require a larger, more geographically dispersed control area that could balance the intermittent contributions of wind and solar generators. And, as anyone familiar with this field is aware, transmission (or the lack thereof) is an issue. But the extent to which it is an issue appears to be a function of whether we rely on fewer megaprojects located in the best wind and solar resource areas or a larger number of smaller dispersed projects (albeit with lower capacity factors) sited throughout the footprint. Hence, there is a trade-off between transmission costs and the higher cost of renewable energy from lower quality resources.

One of the very interesting outcomes from the analyses conducted thus far is the relationship between total load and renewable generation. As the General Electric folks conducting the modeling noted, "The bad actor is the wind." There are times in the spring when the wind is high and there may be more total wind and solar on the system than load. This presents great operational difficulties for system operators. Alternatively, there are times in the early morning when load is ramping up sharply just when the wind is ramping down.

These types of issues highlight the importance of forecasting to system operators. From a market perspective, it was found that with a perfect forecast, increasing renewable penetration drives spot prices lower (since there is no fuel cost). But, with an imperfect forecast, the forecast error drives spot prices back up because operators would commit insufficient capacity and have to turn expensive peaking units back on. Furthermore, it was found that at renewable penetrations exceeding 20 percent, coal units would begin to be impacted. However, the greatest impact was found to be on combined cycle gas plants being backed down. Overall, the value of wind energy rose with a perfect forecast and dropped with an imperfect forecast. Not much of a surprise there, I suppose.

The study also found that generator total revenues fell with increasing renewable penetration (as noted above, spot prices decreased). But, the total revenues for nonrenewables fell at a steeper rate for two reasons -- they generated less energy and spot prices fell. There were two other particularly interesting outcomes presented. The first concerned the cost of unserved energy and the important role that demand response could play in mitigating this problem. The second was that the role that hydro, and especially pumped storage hydro (as well as other large scale storage), could play is less than commonly believed.

Researchers found that higher wind penetration also resulted in greater amounts of unserved energy, due largely to over-forecasting of the wind. But, discounting the wind forecast has the effect of driving spot prices down because you're carrying more gas, resulting in less unserved energy. Thus, there was a very high cost to reducing the unserved energy by discounting the wind forecast. It was found to be far more cost effective to get the load to be responsive rather have the system make up the shortfall to the extent that a couple of thousand MW of interruptible load was found to be cost effective.

Lastly, we discussed the operational impacts of increasing renewable penetration on hydro operation. Hydro needed to be scheduled in response to the wind forecast while increasing wind penetration also increased the variation in hydro scheduling. There was found to be a large operating cost increase if you did not shift hydro commitments in response to the wind forecast -- obviously hydro has the flexibility to move while wind does not.

With regard to pumped hydro, it was found that if you have it, the system will use it, but there was no incentive to add more. This seemed counter to the conventional wisdom so much of the ensuing discussion focused on this topic. The researchers reported that increasing pumped storage increased overall costs. As you increased renewable penetration, the storage ran more. As you forced the storage to run more, it drove costs up. It was found that, even with 30 percent wind penetration, the WestConnect footprint has sufficient pumped storage and no more is needed. Exploring this further, the group concluded that the pumped hydro may be more useful in a smaller footprint. In a larger area, it was preferable to use the system as storage. As part of this discussion, one participant from Ireland noted that studies on their system demonstrated a similar result and that pumped storage was not needed until renewable penetration reached as high as 50 percent. There, it was found that the capacity cost of pumped hydro displaced other capacity costs. But, it only provides capacity if you fill it. Thus, you need to reach the higher wind penetration levels before the pumped storage pays out.

So ended a valuable update to this important research initiative. The day concluded with some discussion of next steps and areas of focus as the project moves forward. One shortcoming, as noted earlier, is the dearth of solar data. To model increasing photovoltaic penetration, the project needs more one-minute PV data. Presently, the only one-minute PV data available to the project is from the 4 MW Springerville project in Arizona. Though there are two larger PV projects currently in operation -- notably Nellis AFB and Alamosa, CO -- this data is apparently not being made available to this study. Why?

1 comment:

  1. Somehow I missed this conference. Thanks for the summary.


Please feel free to comment. I welcome your thoughts. However, no anonymous comments. Professional discourse demands that you identify yourself.