Global climatic change means that for every part of the power chain from generation, across the grid and for intensive energy users such as data centres, the name of the game is carbon abatement.
A response is needed from the data centre sector. This must align with where data centres of the future will be located – e.g., in data centre parks or within industry campuses. It must also encapsulate how their operation will be integrated with local, metro and national power, heating and cooling infrastructure.
From a data centre perspective, the changes happening in upstream power can appear chaotic. The first question to be raised is: “Can the data centre play a role in reducing its own carbon footprint while supporting greenhouse gas abatement of the grid itself?”
That alone sounds ambitious.
A second consideration is that as well as bringing benefits to the grid, future designs may well have to provide carbon-free cooling and heating for a surrounding campus or to the local public or privately owned built environment.
Yet more complexity arises because data centre operators have their own priorities. Any improvements to the whole power chain and use of cooling and heat must be achieved using existing energy sector technology while improving, or at least maintaining, on-site power reliability and reducing GHG emissions.
Current and future data centre space, power and cooling demands present the industry with new challenges.
Fundamentally the challenge is the need for on-site embedded power generation, based on a sustainable design with a low carbon footprint. Such challenges call for new ways of designing data centres.
One proposed solution is an innovative approach built around Combined Heat and Power (CHP), which includes a list of considerations encompassing decentralisation of energy production; use of renewable energy; small scale energy production (Microgrid); improvement of energy usage and power distribution efficiencies; how power at the site is generated and used; together with how the waste heat harvested on-site is re-used.
The benefits of a design that involves the use of CHP production at the site of the end-user eliminates power transmission losses and enables the capture of heat from the exhaust of a gas turbine, so improving the overall efficiency of the power production process.
Installing co-generation plant at the site will provide all required power, as well as cooling. Also, heating for nearby campus buildings or agricultural use.
This can be achieved because within the data centre itself, power reliability can be improved by multiple on-site power generations sources. The use of natural gas in such a design creates an added environmental benefit in that NOx, SOx and particulate production is reduced dependent upon the overall grid fuel mix emission factor.
Case study – how CHP can work to lower data centre emissions
There follows a sample study of a data centre with an assumed IT power capacity of 10 MW (overall electrical capacity of 11.48 MW) and associated cooling demand of 3,000-ton (10.5 MW). A typical installation would include three turbine engines in an (N+1) redundant configuration. All mechanical cooling equipment is also configured in an (N+1) redundant configuration.
Turbine exhaust gas temperature ranges from approximately 340°C to 540°C. Exhaust gases are diverted through a heat exchanger to produce steam which is used in an absorption chiller to produce chilled water.
Two 5 MW gas turbines have a cumulative exhaust gas flow rate of approximately 150,000 lb/hr – sufficient to produce over 7,000-ton of cooling (24.6 MW).
In the example, an absorption chiller replaces traditional cooling plant including a centrifugal chiller and cooling towers to reject the heat utilising a reversed Carnot cycle process. Typical cooling plant utilises water cooled chilled water plant with a centrifugal compressor, cooling towers and pumps. The range is 0.8 to 1.0 kilowatt per ton of cooling.
For a typical 1.0 kilowatt per ton centrifugal chiller plant, energy usage is approximately 3 MW, leading to total site energy usage of 13 MW (i.e., IT load plus mechanical load). By comparison, the use of an absorption chiller frees 3 MW of power, which is available to provide relief on the electric grid and reduce the overall energy consumption of the facility.
Such a design cuts the carbon footprint of a 11.48 MW total connected load data centre by 50%, while fuel consumption is reduced by 553,431 MMBtu. The subsequent reduction in carbon emissions is equivalent to total annual greenhouse gas emissions generated by, e.g., 20,258 cars or 10,818 homes.
As noted above, a global response to a global problem is needed. A single solution will not fit all circumstances. However, many of the problems to be faced are common to different geographies.
Depending on existing national power strategies and fuel mix, different locations have different dependencies. Countries with easy access to and a high dependency on e.g., coal, may have low-cost power but high kgCO2e/kWh. Some countries, such as Poland, China and Germany, already face criticism from environmental activists for their continued use of coal power. In all territories, whether in advanced or developing economies, how CHP for data centres is deployed must not add to the total or marginal emissions of changing grid infrastructure.
Between now and 2030, how such grids decarbonise may dictate the adoption rate of CHP based on its carbon footprint and return on investment. Nonetheless, the time to consider CHP as one design option is now.
The third paper in a series from the EYPMCF and i3 Solutions and GHG Abatement Group, entitled Towards More Sustainable Data Center Design Using CHP, explores this issue in more depth. It will be available for download soon.