The Sacramento Municipal Utility District held it's third annual Data Center energy efficiency course and vendor event yesterday at their energy center/headquarters, with my colleagues Mark Hydeman of Taylor Engineering and Ray Pfeifer of Synapsense sharing course presentation duties with me.
Mark and I significantly improved our respective slide decks, which I think contributed to a much smoother overall session compared to the one we delivered at Pacific Gas and Electric's Energy Center just two weeks ago.
As always, we pored over the course evaluations at the end of the day to see what improvements we could make. Although the scores were uniformly high, we were disappointed in the relative dearth of written comments.
One set did set me back a little. A participant indicated that he thought we had given short shrift, and even outright dismissed, the applicability of on-site generation to serve data centers.
I covered the self-generation, demand response, and load-shifting potential at data centers in all of one slide and five minutes. My central points about self-generation are:
- Data centers have very high load factors, which of course make them an ideal host for cogeneration systems. However, utility rates are lowest for large, high load factor customers, so the economic viability of cogen is problematic.
- The last thing data centers need is a source of heat! That means that the recovered heat from a cogen system would have to be turned into cooling through absorption chillers. For a variety of reasons, including operability, maintenance, and reliability, absorption chillers are a no-go for most facility managers.
- Data center operators are focused on reliability, and will generally prefer utility service at two-nines reliability than a cogen system which doesn't even meet 99% availability. Couple that with finicky absorption chilling, and you just don't have a winner.
- Solar PV and fuel cells may be more attractive options for data centers, though given the energy densities of current construction, lots of space for PV would be needed to make a meaningful impact. Roof space is simply insignificant in the face of the power demands of DCs. PV could potentially be fed to a data center on the DC bus in a UPS, obviating the need for rectification, which could be a real benefit.
Our commenter is absolutely right in saying that in the face of impending environmental regulations, and indeed the need to address climate change, we will have to look much closer at self-generation where it makes sense. I simply point out the current barriers, and don't see a current path to overcome them.
I like to say that if California, with high power rates, resourceful and imaginative companies and people, huge subsidies for self-generation, and cooperative utilities, isn't making any inroads at data center sites, who is?