Guest essay by Eric Worrall
A new report from Lux Research has suggested that Google, Amazon, Facebook and Apple, and other big name tech businesses, corporations which make substantial claims about the “sustainability” of their business models, have significantly underestimated how much their data centres depend on coal generated electricity.
Lux state “Each year, the datacentres that power social media, streaming video, cloud computing, and connected devices use more than 90 billion kilowatt-hours of electricity – enough to power New York City twice over – and their consumption is still growing rapidly. The companies like Google, Amazon, Facebook, and Apple that run them have the most advanced data analytics tools at their disposal, as well as high-minded public commitments to sustainability.
“Our team of data scientists analyzed the North American electric grid, improving the accuracy of carbon reporting by a factor of 80. The results show that many sites are far more reliant on coal than reported – notably, they include many large data centers,” said Ory Zik, Lux Research Vice President of Analytics and the team leader of Lux’s energy benchmarking.
“For example, we found that Google underestimates its dependence on coal in four out of seven datacentres, in particular at its Berkeley County, S.C. location,” he added.
The report can be downloaded here, though you have to provide some personal information to request the download.
Lux also published information about their methodology in Environmental Science and Technology;
Using a complex network framework, the North American electric grid is modeled as a dynamic, equilibrium-based supply chain of more than 100 interconnected power control areas (PCAs) in the contiguous United States, Canada, and Northern Mexico. Monthly generation and yearly inter-PCA exchange data reported by PCAs are used to estimate a directed network topology. Variables including electricity, as well as primary fuels, technologies, and greenhouse gas emissions associated with power generation can be traced through the network, providing energy source composition statistics for power consumers at a given location. Results show opportunities for more precise measurement by consumers of emissions occurring on their behalf at power plants. Specifically, we show a larger range of possible factors (∼0 to 1.3 kgCO2/kWh) as compared to the range provided by the EPA’s eGRID analysis (∼0.4 to 1 kgCO2/kWh). We also show that 66–73% of the variance in PCA-level estimated emissions savings is the result of PCA-to-PCA differences that are not captured by the larger eGRID subregions. The increased precision could bolster development of effective greenhouse gas reporting and mitigation policies. This study also highlights the need for improvements in the consistency and spatiotemporal resolution of PCA-level generation and exchange data reporting.
Lux are not claiming the underestimation of coal usage is deliberate; rather they seem to be suggesting that the US EPA eGRID database, upon which the named companies base their erroneous “sustainability” claims, substantially underestimates coal usage, due to inferior methodology.