Mark Bramfitt - Utilities & IT

Novel Use for Fuel Cells: Backup Generation

Fuel cells and solar PV get all of the attention in the data center industry as alternatives to traditional electric utility service, though there are rumblings for combined heat and power (cogeneration) installs in the background.

Part and parcel of any discussion of primary power sources (including utility service) is backup generation. Sometimes that discussion includes relying on utility service as the backup, but I imagine that the UPS and diesel generator combo has ninety-nine percent plus market share.

Now a company called CommScope is proposing the use of fuel cells as backup, rather than primary power sources for data centers, per this article in Datacenter Dynamics.

I agree with the criticism of diesel generators up to a point - they of course produce emissions of pollutants and greenhouse gases when operated and tested, but with proper maintenance they are soundly reliable. Most backup generation failures that I've heard of (many at hospitals as well as at data centers) occur due to problems in switchgear, not the gen sets themselves.

But there is another reason I think that CommScope may be grasping at straws. They are positing a sixteen hour run time, using hydrogen fuel. From my days promoting alternate fuel vehicles, hydrogen storage was always an achilles heel. Costly. Leaky. Very high pressure.

If fuel cells were to make a go of it as backup generation, I'd talk to Bloom or ClearEdge, and use natural gas service as the fuel source. 

Now utility natural gas service is not 100% reliable either, but it does have the advantage of being able to serve backup gen essentially continuously if needed.  I tried to make the case that diesel gen sets should be converted to natural gas (easy to do; 15% diesel still used as a pilot fuel for ignition) for this reason, and because emissions would be dramatically lower, but have never seen that idea take hold.

This is not to say that there aren't data center operators out there willing to try new approaches. Google put the UPS on the server in the form of a battery, and Microsoft has looked at fuel cells at the rack level.

I'd like to see a head-to-head TCO comparison of diesel gen sets and fuel cells as backups, though regardless of the outcome, an operator is looking at the most expensive power he will ever use!

 

Part 1: eBay's Gas-Fired Data Center in Utah

Yevgeniy Sverdlik at Datacenter Dynamics has a good report that provides some more detail on the novel data center project that eBay has undertaken in Utah.

Some of the facility features are quite interesting - the use of both modular/container and traditional data center spaces, for example.

But of course it's the power equation that fascinates me, and though I don't agree with DCD that this is "glimpse into the future", it certainly poses interesting questions!

First off, the center relies on natural gas fuel cells for primary power, with the utility as backup. The claim is that low natural gas prices, combined with on site conversion of the natural gas to power at a better heat rate than a utility (and without the modest transmission and distribution losses), yield a cost-effective alternative to utility power.

I don't believe that that is at all correct, even in areas with high power prices.

I would hazard a guess that fuel cell power, even with Cheap Frackin' Gas (trademark pending!), runs at maybe just under ten cents a kWh at this scale. And remember, if they are using the grid as backup, the utility will be charging for that, and backup service is not cheap. California utility rates for any customer over a MW are right at 8 and a half cents per kWh.

The connection for me is that eBay can grow the facility in one megawatt increments using fuel cells, and they are also obviating the costs of UPS and back-up generators, which is no small capital cost reduction.

As to the environmental impacts, eBay will be able to claim essentially zero emissions (watch out for any gas leaks though!) but the energy source is of course not renewable. Putting that up against other renewable/clean options isn't a win but it's definitely a place or show.

I see that Microsoft is now looking into fuel cells too, but I'm guessing the scalability advantage, the stripped-down power delivery and conditioning scheme, and the environmental advantages are the primary drivers, not cost savings.

Customer-Side Energy Storage: The Demise of Utilities?!

I had a good exasperated chuckle at this breathless account of customer-side energy storage that appeared in the Atlantic last week as part of a broader series on "The Great Energy Shift".

The report starts with the ominous warning that "trouble lies ahead" for utilities because customers are going to start installing battery storage systems that help them avoid peak pricing rates.

This is a bit of a facile observation given that utilities are increasingly changing their rate structures to reflect the true costs of providing service, particularly at peak times, that they provide incentives for load management, are experimenting with utility-side energy storage systems, and are in some cases set to provide incentives for customers to do the same.

In what way does this constitute "trouble"?

I appreciate that a hotel operator may have difficulty implementing load management strategies. I mean, turning off air conditioning in the summer in a San Francisco hotel isn't doable, not because guests who pay $300 a night won't tolerate it, but because there just isn't that much air conditioning load when it's only 62 degrees outside!

And installing an energy storage system is not "pulling the plug on the utility", it is load management, pure and simple.  And, demand rates have not risen thirty percent in California over the past three years - customers are now seeing peak usage charges for fifty or so hours out of the year.

It makes sense for the utility and/or their customers to manage that peak load. It drives the capacity ratings of the distribution system up, and really kills on the generation side, where marginal costs can spike to well over a dollar a kWh during peak periods.

I'll credit author Todd Woody with at least providing some context around energy storage, including noting that California utilities have been authorised to install 1350 MW of storage by 2020 (the costs of which will be reflected in customer rates, by the way), and that sixty percent of the costs of the system installed at the Intercontinental were covered by incentives (which will also be paid out of rates).

And yes, the system is financed by the maker of the system (Stem), though that is hardly novel, as there isn't a PV array that isn't financed today. (Stem also indicates that they sell utility-side storage systems.)

But I have to say, given all of the regulatory impetus behing real-time utility pricing, orders to install utility-side storage, and subsidies for customer-side storage, I just don't see that the "traditional utility business model, which for over 100 years ago has served its purpose well, has come to an end."

If anything, the utility business model, enabled by the regulatory process, makes the energy storage market viable.

Self Generation For Data Centers: Fuel Cells Gain More Attention

I've been delivering the same story on self generation potential for data centers for some time now, which goes like this:

• Data centers are an attractive load for self gen because of their close to unity load factor, which means you can run some solutions (cogen and fuel cells) around the clock and recover your investment quickly.

• However, utilities reward large, high load factor customers with their lowest commercial/industrial rates, so you've got to be cheap.

• Data centers have very high reliability requirements, so operators are leery of increasing the complexity of their power delivery and conditioning schemes, a hurdle for all types of self gen.

• The last thing a data center has is a need for heat, so cogen schemes have to either find another host for waste heat, or turn to absorption chilling.

• Absorption chilling seems to be losing market share, and besides, new data center designs rely on air economizers with supplemental evaporative cooling, so have no need for cooling anyway.

• Solar PV has the natural advantage of a great green profile, and it delivers power when the utility rates are highest, but you need plenty of space to begin to make a mark against the loads of utility-scale data centers.

• Wind is as green as PV, but is not a self gen player. If you want wind power, you buy it on the market and have your utility deliver it to you.

• Fuel cells look to have a real advantage, in part because they can potentially obviate the need for backup generation and UPS in a data center. (And natural gas prices have never been lower in the US.)

• All of these thoughts relate to new builds.  For existing data centers, invest in energy efficiency before you even think about self gen: you'll get superior results at far less risk.

I continue to watch the market to see if these axioms are holding true, and I think to a large extent they are. For example, Google buys wind power, Apple is installing large-scale solar, and Ebay is using fuel cells.

I did note this report from Data Center Dynamics of a 100 kW fuel cell installation at an Equinix colo in Germany, which is interesting because it looks like  a "traditional" fuel cell, featuring a digester to turn natural gas into hydrogen, with the cell producing both power and heated water. 

The fuel cell company, N2telligence (download English language brochure), makes some interesting claims that frankly puzzle me. For example, they say the fuel cell uses oxygen (yes, anything that oxidizes fuel uses oxygen) so can draw down the O2 content in the data center to support fire suppression. Hmmm?

They also say that other fuel cell technology emits nitrogen (no more than was in the natural gas and air that runs through the reaction, just like theirs...) and that their system produces more environmentally favorable water.  Again.  Hmmmm?

Despite those questions, all due props to Equinix for the trial. Colo providers have only very gingerly dipped their toes into the self gen pool, so this is a leadership move on their part. Do tell us how it goes!

GSMI Green Data Center Conference Presentation Deck

I took a quick trip to the outskirts of Dallas last week to deliver one of two keynote addresses at the GSMI Green Data Center Conference, focusing my remarks on the intersections of the data center and utility industries.

My remarks covered not only the utility and energy factors inherent in making a siting decison for a new data center, but why energy issues will be increasingly important for IT managers even for existing operations.

The concluding theme was that the IT industry and utilities have more to do with each other than ever, and that there are opportunities to partner on energy efficiency and perhaps load management and self generation. And for the big players, a need for them to help guide and influence energy policy and regulations, given the importance of the grid in supporting their business vision.

ASs I noted in an earlier post, it was gratifying to get insightful questions about energy policy in the final few minutes of my time slot, as well as good one-on-one discussions afterwards!

Here's a copy of the deck:

Download GSMI GDCON Dallas 6 2013

The Future of Nuclear Power in the US

Probably a couple of quick posts that react to discussions I had after my presentation at the GSMI Green Data Center Conference in Dallas last week...

You never know what questions you'll get at the end of a presentation, and after delivering a presentation describing how utilities and the data center and IT industries are becoming increasingly intertwined, I was pleased to get more questions about utility regulatory policy than on data center issues.

One of the questions, which generated quite a bit of ongoing discussion, was why the US is not building nuclear power plants as a means of addressing climate change, and energy security and availability.

My short answer is that beyond technical issues and even siting concerns, regulatory policy in the US precludes utilities or merchant generators from developing power plants of all types. Simply put, it's hard to finance what are often multi-billion dollar projects without regulatory assurance that cost recovery in rates will happen.

We've seen that in California, where our regulatory commission is allowing the investor-owned utilities back into the power generation business becasue the private market was abandoning projects mid-stream, and certainly not building enough capacity to meet projected needs.

This article from today's New York Times describes the only current nuclear power plant project in the US, describing the regulatory challenges as well as the headwinds of cheap natural gas and falling prices for renewable generation (notably wind, not solar).

One of the central points of my presentation was that power portfolios in the US are getting cleaner, but that is coming at a cost. For data center developers, understanding the future price for power at a given utility is as important as knowing the current costs. Utilities that have to move to cleaner power portfolios are going to see the largest percentage rate hikes.

Better Than Free Power - No Kidding!

My colleagues in the high tech world may not quite believe this, but electric power on the open market is occassionally priced at less than free - producers will actually pay you to take it.

Here's what happens: utilities or customers enter into power supply contracts including ones that feature "take or pay" clauses for base load capacity. Hydroelectric plants fall in this category because there is a minimum amount of flow through the dam to preserve habitat downstream.

So how do we get to negative pricing? Let's say a utility has 1000 megawatts of take-or-pay base load resources, but in certain times of the year customer load drops below that. Now the utility is faced with paying the generator for power that can't be used. (Easy enough for a hydro operator to back off on production by altering the pelton wheel profile or bypassing the turbines completely.)

Now the utility would be happy to pay a customer to take the power as long as the price was less than what he had to pay the generator (better to pay you a penny a kilowatt-hour than the generators two cents).

In my experience, this is a rare occurence, maybe happening on winter nights when loads are lowest.

But that situation is changing in interesting ways. In the Pacific Northwest, wind producers have contracts that feature take-or-pay clauses, and the operators receive favorable tax treatment based on the actual power delivered to the grid. Bonneville now has something like 30% of their portfolio coming from wind, with much of the rest hydro.

So now, even on some summer nights, customer load isn't enough for the power generated by the hydro and wind plants, and there are significant costs and impacts involved in curtailing the power production. The wind operators in particular don't want to be curtailed because of those tax breaks.

The four state region is grappling with this issue in a positive way - seeking opportunities to engage customers who could use power during these surplus events. They're looking at traditional load shifting strategies like precooling office buildings and refrigerated warehouses, and time shifting electric water heating loads.

And...we're trying to think about how data centers could participate. I personally don't see a pre-cooling strategy being viable, but I keep coming back to thermal energy storage as an exciting solution for data centers. You get load shifting capability (and might even get paid for taking off-peak power sometimes!) along with redundancy in your cooling infrastructure.

I know of only one data center that has implemented a thermal storage system, but I can't help but think this is a larger market opportunity.

Google's Energy Use and "Equivalent Homes"

The New York Times reported today that Google has released information on the energy use of their global operations - dominated of course by data centers but including their offices.

I'm surmising now that Google has a data center footprint somewhere in the 225 MW neighborhood; I was off in some analysis I did for a client that pegged them at 300 MW, but I count that as close!

The article attempts to put the load figure into terms that are understandable to laymen, positing that 260 megawatts could serve several hundred thousand homes. That's a spot-on estimate in my experience.

In California, we often use the rule-of-thumb that a MW of capacity is needed to serve about 750 homes. It turns out that this estimate is really based on usage rather than peak demands. The average home in CA uses 750 or so kWh per month, or about 10,000 kWh per year. If you do the math, a MW of generating capacity at 100% load factor would therefore serve 750 homes or so.

But of course utilities do not have flat load curves. The big California utilities have load factors in the 30% range, meaning that you need something on the order of 3 MW of capacity to serve 750 average homes.

For the Google comparison, I fall on the side of using the energy equivalency rather than the demand, though for the best accuracy, we would compare Google's energy use to average home use, or demand to demand.

NYT: Water-side Data Center Cooling Gaining Traction

Nice article in the New York Times yesterday by James Kanter (in the Green Column of the Global Business Section) describing how some data center site selections are valuing locations with access to water which can be used for water-side cooling.

Notably, Google's next data center in Finland is located in a former paper mill, and will use sea water for cooling.  Eminently doable of course, though more expensive than using freshwater given corrosion and usually higher pumping costs. Liked the statistic that Google's five owned data centers have on average a 16% overhead for cooling (and I suppose power delivery and conditioning).

I'm not sure that I agree with the conclusion that sites with ample water supplies are less suitable for on-site generation - the assumption is that the sites will generally have climate conditions that won't favor solar PV.

First, solar PV isn't the only technology available for on-site generation, though it is particularly apt for peak load shaving. Fuel cell technology may have broader advantages by obviating some power conditioning equipment, for example.

Second, as the Google example shows, if you're using seawater for cooling, that only limits your site selection to coastal locations, and you could certainly select sites with ample PV exposure, balanced against ambient temperature conditions.

The upshot though is that leading data center operators are starting to think like industrial plant developers, carefully evaluating the inputs of their cost structures, and thinking about ways not only to drive up efficiency, but to use "natural" resources like cooling water that can lower costs.

 

 

Rethinking Self Generation for Data Centers

Just because I have had a long utility career doesn't mean that I am inherently against self-generation - in fact I have represented both sides of the opportunity, both promoting co-generation and trying to thwart it.

And I think utilities are now in that gray middle area, wondering if distributed and self-generation can be implemented to their benefit, and even be a business opportunity.

For data centers though, I have consistently minimized the applicability of self-generation for what I think are sound reasons.  Certainly, data centers (especially those without free-cooling) are an attractive load for generation, given an essentially unity load factor.  That means a capital investment in generation equipment can be recouped quickly.

But on the flip side, there is no opportunity to easily use reclaimed heat (absorption chilling, anyone?);utilities offer their lowest rates to large, high load profile customers; and the reliability of rotary generation equipment of any kind doesn't come close to what data centers are looking for.  (Even large-scale, sophisticated cogeneration systems typically require maintenance downtimes stretching into weeks per year).