Earth Mode Calculations: Assessing the Carbon

 Earth Mode Calculations: Assessing the Carbon Footprint of Internet Activity The goal is to produce a credible, complete, interesting footprint of a user’s internet activity for use in the Johnnie Walker Earth Mode plug-­‐in. There is significant variability in existing studies of internet energy intensity and there are no comprehensive, regulatory models to draw from. In lieu of this, we assessed research in the space and used the best available estimates for each phase of the system. We then applied country level carbon intensity factors to the energy calculation to produce a localized carbon footprint. There is much more one could do to increase the accuracy of this assessment – we’ve identified a few ideas at the end of this paper. However, in partnership with Diageo and the Earth Mode team, we produced a tool that can be used for indicatively assessing and learning about the energy impact of internet activity anywhere in the world and hope that others will use and build on this in their carbon footprinting projects. The resulting algorithm uses a combination of methods to best assess the nature of internet use specifically the combination of fixed and variable assets and energy expenditure based on user activity – method selected for each phase is based on the dominant variable (generally time or throughput) This algorithm and factor set builds on the research of others in two key ways: 1. Energy assessment combines best available estimates for each phase of use – the user data collection via the Earth Mode app enables a personalized estimate 2. Carbon impacts are based on localized power estimates at the country level, better reflecting actual impact based on user location Major Components for Energy & Emissions Assessment To assess the energy and carbon footprint of an individual’s internet use, a model of internet activity was created with four components – selected based on the work of (Coroama) for their distinct use profiles. Collectively, these components sum to the direct energy impact of a user’s internet activity. These components are: Device The user device’s energy use is for the most part a function of the time it is on as network activity has little impact on the total power use. Therefore, the formula for device energy use was modelled as: DEVICEENERGYUSE(kWh) = DEVICE (kW) * TIME (hr) The carbon impact of this energy use is 100% localized to the user’s country grid and is therefore estimated by: DEVICECO2E (kgco2e) = DEVICEENERGYUSE(kWh) * COUNTRYCO2(kgco2e/kWh) (See Variables section for source and parameter of each variable) Local Area Network To reach the internet, the device must communicate with a local network. A typical case would be a wireless router at home or a café which then communicates with local Area switching equipment to the Edge/Metro network and beyond. The energy use of this equipment is largely fixed per user as it remains on all or nearly all day prepared to accept activity. Its power consumption does not appreciably change with the throughput of data. The energy formula is estimated by: ENERGYUSE (kWh) = AREANET (kW) * DAYS * 24 (hr/day) This equipment is located 100% within the local area. The carbon emissions formula is then: AREACO2E (kgco2e) = AREAENERGYUSE (kWh) * COUNTRYCO2 (kgco2e/kWh) Metro / Core Network The Metro and Core Network consist of the bulk of longer range data transmission infrastructure, connecting the user to the server holding the internet resource they request. Again drawing from (COROAMA), the energy consumption of this layer is dominated by the transmission, amplification and processing of data. The estimate is given by: METROCOREUSE (kWh) = DATA(gb) * METROCORE (kWh/gb) The Metro / Core Network power consumption are modeled under one variable, but they occur in different geographies. The Metro can be approximated as a national network and the Core as the global network. So, the power is half country grid, half global grid and the carbon intensity of this energy is reflected as follows: METROCO2E (kgco2e) = METROCOREUSE (kWh) * LOCALFRACTION * COUNTRYCO2 (kgco2e/kWh) + METROCOREUSE (kWh) * (1-­‐LOCALFRACTION) * GLOBALCO2(kgco2e/kWh) Data Centre / Servers To model the Data Centers / Servers, we turned to the work of (NRDC) and utilized their inclusive assessment of data center power consumption and (CISCO) for usage figures to derive a server intensity. More detail can be found in the Variables section. Data is the dominant factor in energy use since most equipment will scale up or down with use, including HVAC and balance of system equipment. The equation for energy use is estimated by: SERVERUSE (kWh) = DATA(gb) * SERVER(kWh/gb) This again will be sometimes local, sometimes global, and we have adjusted the carbon impact to reflect this as given by: SERVERCO2 (kgco2e) = SERVERUSE (kWh) * LOCALFRACTION * COUNTRYCO2 (kgco2e/kWh) + SERVERUSE (kWh) * (1-­‐
LOCALFRACTION) * GLOBALCO2 (kgco2e/kWh) Assumptions & Limitations While we have made a thorough effort to have the Earth Mode plug-­‐in accurately reflect each user’s carbon impact due to internet browsing and have as much as possible followed known reporting and assessment guidelines, the results are still only estimates for educational use and should not be used for reporting or otherwise relied upon without further investigation. We attempted to provide a meaningful advancement of current work on the carbon impact of internet activity by many others and make it available in an easy to use (automatic) package that could provide an indicative measurement from anywhere on the planet. There are many, many assumptions and limitations that we know of and likely more that we don’t, here are a few: -­‐
The variables and models used are broad, reflective of country or global averages and further information about any of these systems could improve the calculation significantly -­‐
-­‐
-­‐
This includes only the direct energy impact of browsing and does not include the very significant amount of energy required to manufacture and service the internet infrastructure. Taken in aggregate, these indirect impacts are often 5-­‐10x the direct impact of a system. These calculations take into account only the broadest variation in use cases, there are myriad factors that an individual situation will contain that will render these calculations much less indicative Does not include mobile networks. Mobile data networks use a significant amount of energy and were not reflected here since the plugin is meant for desktop and laptop use References: (COROAMA) -­‐ Coroama, V.C., Schien, D., Preist, C., Hilty, L.M.: The Energy Intensity of the Internet: Home and Access Networks. In: Hilty, L.M., Aebischer, B. (eds.) ICT Innovations for Sustainability. Advances in Intelligent Systems and Computing 310, Springer International Publishing (2014, in press) (COROAMA) -­‐ Coroama, V. C.; Hilty, L. M.: Assessing Internet Energy Intensity: A Review Of Methods And Results. Environmental Impact Assessment Review 45 (2014) pp.63-­‐-­‐-­‐68DOI:10.1016/j.eiar.2013.12.004 (COROAMA) -­‐ Schien, D., Coroama, V.C., Hilty, L.M., Preist, C.: The Energy Intensity of the Internet: Edge and Core Networks. In: Hilty, L.M., Aebischer, B. (eds.) ICT Innovations for Sustainability. Advances in Intelligent Systems and Computing 310. Springer International Publishing (2014, in press) (NRDC) Natural Resources Defense Council, Issue Paper Scaling Up Energy Efficiency Across the Data Center Industry: Evaluating Key Drivers and Barriers August 2014 IP:14-­‐08-­‐a (DEFRA) -­‐ UK Department for Environment Food & Rural Afairs – Greenhouse Gas Conversion Factor Repository – 2015 Global Electricity Emissions Intensity -­‐ http://www.ukconversionfactorscarbonsmart.co.uk/ -­‐ Accessed 3April2016 (HINTON) -­‐ Kerry Hinton, Jayant Baliga, Michael Feng, Robert Ayre, and Rodney S. Tucker, University of Melbourne. Power Consumption and Energy Efficiency in the Internet. IEEE Network • March/April 2011 (SCHIEN) -­‐ Daniel Schien, Chris Preist, Department of Computer Science, University of Bristol. A Review of Top-­‐Down Models of Internet Network Energy Intensity. Atlantis Press. 2014. (CISCO -­‐ Cisco Visual Networking Index: Forecast and Methodology, 2014–2019. 2015.