A Generic Carbon Footprint Model

We have previously reported the carbon footprint calculations for the National Digital Preservation Services in Finland, e.g. as a short paper in iPRES 2023. We have found out in our calculations that the carbon footprint for the hardware, including both manufacturing, transportation and disposal costs, has a profound impact on the total carbon footprint of our service platform. In our platform, the hardware component carbon footprint is much greater that the service life carbon footprint (electricity consumption). Now it is time to expand our carbon footprint calculations and create a generic model for calculating the footprint in cases where hardware manufacturers do not report the in-body carbon footprint (includes hardware manufacturing, logistics and disposal) figures for their hardware.

The basic assumption is that all servers, regardless of the manufacturer, share components which have an in-body carbon footprint reported by at least one vendor. We will turn this sparse matrix for server component in-body carbon footprint values into a generic server components in-body carbon footprint calculation. This generic calculation is estimated to be fairly accurate and the goal is to develop this into an open model for hardware component carbon footprint calculation. This model would be highly beneficial to the digital preservation community in general, as it could be used by any producer of a digital preservation service to calculate their hardware component carbon footprint, without having to have separate in-depth discussions with their hardware manufacturers.

Since the previous report from 2023, the National Digital Preservation Services have expanded the platform to 7.5 Petabytes and the server manufacturer has been changed from HPE to Kaytus. At the same time the Kaytus hardware lifespan has been extended into 7 years. We have also replaced one data centre, the new data centre having a PUE of 1.05 while the previous data centre had a PUE of 1.66. This poster will use our current digital preservation platform, allowing for up to 7.5 Petabytes of data, as an example when creating a generic carbon footprint calculation using the previous assumptions detailed above.