CMU study evaluates uncertainties in methodologies for calculating the product carbon footprint of a server

13 Dec 2010 -- Carnegie Mellon University (CMU) has completed a case study investigating the uncertainty and variability associated with calculation of the greenhouse gas emissions associated with the life cycle of a rack-mount server product (also referred to as the server's "carbon footprint"). IBM collaborated with CMU in this study by providing CMU with technical details and information regarding the IBM server product evaluated by the study.

The results of the study verified that greenhouse gas emissions associated with the operation of the server throughout its useful life dominated the full life cycle carbon footprint. This "use phase" of the server accounted for an estimated 94% of the total greenhouse gas emissions associated with the product. This finding confirmed the importance of IBM's ongoing efforts to increase the energy efficiency of its server products and the data centers where servers are used.

However, while the study confirmed the importance of server energy efficiency on the product's overall carbon footprint, it also highlighted the large uncertainties in quantifying the server's carbon footprint. The study estimated the overall uncertainty for the carbon footprint of the selected sever product to be +/- 35%, with uncertainties associated with product use of +/- 50%. Carbon footprint uncertainty in the product use phase was highly dependent on the server's projected use profile and the greenhouse gas emissions associated with the electricity grid in the locations where the products were sold. Variability in the electricity mixes of different markets led to vastly different impacts of using the product similarly in different places. Assumptions of the server's expected useful life also significantly affected results.

Uncertainty in the calculation of the carbon footprint associated with production of the server was estimated to be +/-15%. However, estimating uncertainty was hampered by the lack of multiple data sets covering many component types. For integrated circuits, where multiple data sets covering several generations of technology were available, uncertainty was estimated to be much higher (+/- 40%).

Given the increasing interest in product carbon footprints (PCF) in public policy and labeling initiatives, the study highlights variables affecting precision of PCFs that merit greater attention and distinction, and calls into question the validity of carbon footprint labels for complex information technology products.

Additional information about the study is available at