Reducing Carbon Footprint using Persuasive Technologies and Online Sensing P ROPOSAL FOR VISITING RESEARCH STUDENT November 20, 2008 Dr. Adrian Friday 1 and Dr. Kim Kaivanto 2 1 Computing Department, Lancaster University. 2 Lancaster University Management School, Lancaster University. Address: Computing Department, InfoLab 21, South Drive, Lancaster University, Lancaster, LA1 4WA Contact: phone: 01524-510326, email: adrian@comp.lancs.ac.uk, k.kaivanto@lancaster.ac.uk 1 Motivation UK universities and higher education institutes emit 3.2 million tonnes of carbon dioxide every year. Sector wide we spend an estimated 200 million on energy per annum (on average, 2.5% of the University’s annual budget is spent on energy) 1 . At Lancaster, this amounts to approximately £3m per annum—there is clearly strong motivation to reduce our energy consumption for ethical, competitive and financial reasons. We believe that education of the campus population (increased awareness of carbon impact of behaviour) combined with sensing and actuation to provide us with a smarter infrastructure that both meets user needs while saving energy, can help us address these challenges. We wish to explore how sensing driven ‘persua- sive technologies’ (c.f. [3]) and linking into online communities 2 may help us in exposing hidden behaviour regarding energy usage and lead to positive behavioural change. 2 Specific research objectives The aim of this project is to help lay the groundwork for an empirical exploration of this hypothesis with subjects drawn from the campus population. We are applying for a student intern to work with us during the early phases of this project. The broad scope and potential impact of the work would allow us to offer a range of possible focuses depending on the aptitudes and interests of the student: 1. Carbon footprint is currently calculated from a constant conversion factor that we hypothesise does not accurately account for diurnal and seasonal variation. Calculation and analysis of time-varying CO 2 kg/kWh that accounts for this variation from (supplied) historical UK time series data will enable us to test this hypothesis and more accurately predict carbon output; 2. Development of a flexible infrastructure for gathering and publishing real-time ‘feeds’ of the time- varying CO 2 kg/kWh based on energy monitoring sensor data; 3. Comparatively evaluate streaming database/ middleware approaches and associated query languages for programming with time varying event stream data [2, 1], enabling user level applications to be concisely and efficiently described; 1 Source: Tom Cumberlege, Public Sector Manager at the Carbon Trust, quoted in LU Press release, June 2008. 2 http://www.facebook.com, http://www.pachube.com, http://www.amee.cc