What does the data say?
This is the second of two pieces on Cogeneration or CHP. The first explored the topic, this one will explore the RoI of technology proven for other industries as applied to data centers.
As the data center industry continued to consolidate and competitiveness becomes more intense, IT professionals understand the pressure on both capital and operating budgets. They are torn by two competing forces, faster and more reliable vs. low cost and now. IT equipment improvements are continuously and the desire to update always calls. Reliability has become the mantra of hosted application and cloud customers and although electrical grid failures are not considered “failures against uptime guarantees” for some, businesses affected by outages feel the pain all the same. And if there are solutions, management pressure to implement them quickly and at low cost is always a factor.
Cogeneration is typically neither fast nor cheap, but it does offer an alternate path to reliability and uptime. As in all major investments that require sizable capital and space, the best time to consider cogeneration is during data center construction. That being said, data centers operating today are not going any place soon, so retrofit upgrade paths are also a consideration, especially in areas where electric power reliability from the local utility has become less reliable over time. So when should data center professionals consider cogeneration or CHP? Fortunately there are studies available on public websites that help provide answers.
University of Syracuse data center exterior; Microturbines in utility area (Link to Source)
One such study is an installation at the University of Syracuse. Opened in 2009, the 12,000 ft2 (1100 m2) data center with a peak load of 780 KW employs cogeneration and other green technologies to squeeze every ounce of energy out of the system. (Link to Source) The site’s 12 natural gas fueled microturbines generate electricity. The microturbine’s hot exhaust is piped to the chiller room, where it is used to generate cooling for the servers and both heat and cooling for an adjacent office building. Technologies such as adsorption chillers to turn heat into cooling, reusing waste heat in nearby buildings and rear door server rack cooling that eliminates the need for server fans completes what IBM calls its Greenest Data Center yet.
Left: Heat exchanger used in winter months to capture waste microturbine heat for use in nearby buildings; Right: IBM “Cool Blue” server rack heat exchangers employ chilled water piped under floor.
This is certainly an aggressive project, but can the cost be justified with a reasonable Return on Investment? Fortunately data has recently been released to quantify the energy conservation benefits. PUE performance measured during 2012 was presented at an October 2013 conference and show a steady PUE between 1.25 and 1.30 during the period, a value that compares very favorably when compared to the typical data center PUE of 2.0. Uptime Institute self reporting average PUE is 1.65 with qualifications, Digital Realty Trust survey of 300 IT professionals with annual revenues of at least $1 Billion and 5,000 employees revealed PUE of 2.9. (Link to Sources: Uptime Institute Digital Realty Trust)
IBM/SU Green Data Center 2009 Goals (Link to Source); 2012 Actual Performance (Link to Source)
So how can we calculate the actual RoI and compare it to the projected goals. First, the goals stated in the table on the left show savings of $500,000+ per year. Another presentation by the microturbine supplier shows a $300,000 per year goal, quite a bit different. So how do we know what the savings is? We don’t since there is no reference site where the data center is identical and in an identical location without the CHP. So we can use the 2.0 average PUE and calculate the energy savings, but that’s not a real answer. And we also need to take into account the fact that tax incentives and grants such as the $5 Million for the Syracuse University project needs to be reviewed to determine the cost to non-subsidized projects. Hopefully project managers will provide more information to help data center operators better understand the actual savings as the project matures.
CHP for data centers is presented with an array of benefits including improved reliability through less dependence on grid power, lower power costs, reduced carbon footprint. NetApps installed CHP in their Silicon Valley data center to reduce their reliance on grid power due to frequent rolling brownouts and the uncertainties of the power market costs. Their experience is not as instructive due to the site’s reduced need for cooling due to use of direct air cooling. As a result the CHP system is used only when the utility is strained. It is difficult to find quantitative data for modern installations. While the data seems encouraging, actual energy cost savings are not provided. We will watch the progress at this and other projects over the next several months to see if CHP costs yield an acceptable RoI via reduced energy costs. Stay tuned.