REFIT: Personalised Retrofit Decision Support Tools for UK Homes using Smart Home Technology

Lead Research Organisation: University of Strathclyde
Department Name: Electronic and Electrical Engineering

Abstract

Thermal efficiency retrofit options, appliance upgrades and on-site renewables represent a significant opportunity to deliver energy demand reductions to UK homes. The potential to reduce thermal heat losses through insulation and airtightness (in particular in pre-1980s housing), upgrade the household appliance stock (using the latest energy saving models) and integrated on-site renewables and microgeneration (developing a 'prosumer' culture and reducing energy bills) still remains largely unrealised. There are a number of challenges in providing advice for retrofit solutions to consumers which will promote behaviour change and influence purchasing decisions. Currently consumer information is based on standardised methodologies for nominal house types and the resulting predictions of energy savings have minimal resemblance to reality where the thermal efficiency of the dwelling, efficiency of heating system and appliances, occupancy, user behaviour and preferences will have a significant impact on the effectiveness and uptake of retrofit measures. One solution is to provide consumers with personalised, accurate and trustworthy predictions of energy saving measures which are calibrated and tailored to their dwelling and living patterns, presented in a format to engage and promote action.

This proposal will facilitate a widespread uptake of retrofit measures in UK homes by implementing a holistic approach to providing consumers with personalised, tailored retrofit advice delivered using methods to maximise consumer engagement. Smart Home technology provides a unique opportunity to use real-time measurements, advanced data analytics, digital signal processing and communications techniques, novel visualisation, semantic web and cloud computing technologies to generate advice at different levels of abstraction for informed and justified decision making. The Smart Home concept is currently gaining significant momentum and new developments in open systems, simple use and installation features (ie plug and play), mobile access (ie Smart Phones) and connectivity have brought the concept to the attention of energy companies, ICT companies and appliance manufacturers. The IBM vision of a Smart(er) Home gives three characteristics: 1) Instrumented (sensors and automation of household activities); 2) Interconnected (communication between devices and wider networks - allowing remote access and control of devices); and 3) Intelligent ('the ability to make decisions based on data, leading to better outcomes'). Smart Homes provide consumers with more control over their homes and energy systems and, importantly, how their energy demand and costs can be reduced through interventions.

This proposal brings together a multi-disciplinary team of building, ICT, energy, design and user experts to develop a personalised decision support platform for building envelope retrofits, heating system and appliance replacement purchases, and on-site renewables integration. This will deliver a step-change in the provision and accuracy of retrofit advice to UK householders leading to a low-energy and low-carbon future housing stock. The outcomes will be of benefit to: energy, ICT, embedded systems and telecommunication companies developing technology and business models for Smart Home services; consumers to lower their energy bills and improve the safety, security and comfort of their homes; building component, boiler and appliance manufacturers developing the next generation of low-energy products; and policy makers for new insights into innovative approaches to meeting the security, affordability and carbon reduction aspirations of the UK energy system.
 
Description The team has successfully tackled the challenges of energy disaggregation, termed Non-Intrusive appliance Load Monitoring (NILM) of low-granularity, one dimensional smart-meter data via the development of a range of low complexity algorithms for a large range of appliances. These NILM algorithms formed the basis of further analytical work developed in the group, namely informing appliance retrofit/upgrade decisions, predicting demand from appliances, finding opportunities for load shifting and developing activity recognition algorithms that map appliances or technologies to activities in order to understand households' daily routines. The latter challenge of how to interpret real-time energy data in terms of activities from smart meter data is tackled in conjunction with our social science partner, showing the effectiveness of jointly considering quantitative and qualitative data through mixed methods. In summary, we have demonstrated that, from smart meter data alone, we can generate itemised billing down to appliance level or down to activity level and extracting the time-use profile and energy-profile of any particular activity, such as cooking, laundering or washing, in a particular household and across households and what are the cost implications.
Exploitation Route Our findings answer the following questions:
• How can signal information processing turn low resolution smart meter data into meaningful information?
• Can we develop consistently accurate and practical non-intrusive disaggregation solutions from low-resolution smart meter one-dimensional data?
• How much do activities, such as cooking and energy services like refrigeration, account for in a household's total electricity consumption?,
• Can we generalise when energy-intensive activities occur within households of similar occupancy?
• Which are the largest energy-consuming activities, and what are the implications for demand management and feedback?
Load disaggregation is seen as the next step towards providing effective energy feedback. Load disaggregation providers supply energy disaggregation through a combination of hardware submetering and software analysis. However, these solutions are currently limited to disaggregating high loads and industry is keen to adopt approaches that can operate at Smart Meter data rates, are practical, simple, accurate, and robust for a range of training periods. Our new algorithms are good candidates for future industry uptake.
Our activity recognition method provides an alternative, less expensive option that does not depend on the traditional approach of time diaries, for generating daily time-use profiles of a subset of energy using activities, which can then be compared with national time-use statistics to identify variability, or to segment households. This is attractive for service providers in the Smart Home market to provide value-added feedback to users.
Finally, automatic activity recognition is an important enabler of home automation and effective home energy management systems, as well as Assisted Daily Living with further implications for remote healthcare.
Sectors Communities and Social Services/Policy,Digital/Communication/Information Technologies (including Software),Energy,Environment

URL http://dx.doi.org/10.15129/31da3ece-f902-4e95-a093-e0a9536983c4
 
Description The research supported by this award led to a unique dataset of time-stamped power load measurements, at household and appliance levels. The dataset contains readings from 20 houses in England monitored for a continuous period of about two years as the households went about their daily lives. This is the only such UK dataset at this longitudinal scale with a sampling rate below 1min, that is, sampling is carried out at 8 second resolution which is similar to the load measurements from the UK smart meters being rolled out, via the Consumer Access Device. The dataset is now publicly available, and the data collection, cleaning of the data and overall description are described in Nature Scientific Data Journal paper. The dataset has already attracted the attention of many academic research groups as well as the energy and smart home industry. Additionally, the dataset, non-intrusive appliance load monitoring (NILM) and appliance mining methods developed have yielded additional research directions, e.g., activity recognition where energy consumption is quantified through the lens of activities, load-shifting (exploiting flexibility in time-of-use of appliances to manage peak demand), retrofit advice and smart home automation.
First Year Of Impact 2015
Sector Digital/Communication/Information Technologies (including Software),Energy
Impact Types Societal

 
Description Appliance modelling and usage analysis
Amount £37,115 (GBP)
Organisation NESTEC, Inc. 
Sector Private
Country United States
Start 08/2016 
End 03/2017
 
Description MSCA-RISE - Marie Sklodowska-Curie Research and Innovation Staff Exchange (RISE)
Amount € 859,500 (EUR)
Funding ID H2020-MSCA-RISE-2016 734331 
Organisation European Commission 
Sector Public
Country European Union (EU)
Start 01/2017 
End 12/2020
 
Description Personalised ICT-tools for the Active Engagement of Consumers Towards Sustainable Energy
Amount € 196,414,538 (EUR)
Funding ID 767625 
Organisation European Commission H2020 
Sector Public
Country Belgium
Start 10/2017 
End 12/2020
 
Title REFIT: Electrical Load Measurements 
Description The REFIT Electrical Load Measurements dataset includes electrical consumption data in Watts for 20 households at aggregate and appliance level, timestamped and sampled at 6-8 second intervals. This dataset is intended to be used for consumption statistics, load disaggregation algorithm development and testing, time and energy use profiles and time use statistics. This data repository is hosted at the University of Strathclyde, and has been populated from data obtained from a remote monitoring platform, whose server was at the University of Strathclyde, capturing data in real-time from households in the Loughborough area of the UK. 
Type Of Material Database/Collection of data 
Year Produced 2015 
Provided To Others? Yes  
Impact To date, the REFIT dataset is the largest UK electrical measurements repository, containing measurements sampled at under 1 minute and gathered continuously over a period of 2 years. As such, it is very valuable for designing and testing non-intrusive load monitoring (NILM) algorithms. To this effect, the creators of NILMTK, which is an open source toolkit designed specifically to enable the comparison of energy disaggregation algorithms in a reproducible manner, have included a REFIT data converter so that algorithms from anyone can be tested. 
URL https://pure.strath.ac.uk/portal/en/datasets/refit-electrical-load-measurements(31da3ece-f902-4e95-a...
 
Title REFIT: Electrical Load Measurements (Cleaned) 
Description The REFIT Electrical Load Measurements dataset includes cleaned electrical consumption data in Watts for 20 households at aggregate and appliance level, timestamped and sampled at 8 second intervals. This dataset is intended to be used for research into energy conservation and advanced energy services, ranging from non-intrusive appliance load monitoring, demand response measures, tailored energy and retrofit advice, appliance usage analysis, consumption and time-use statistics and smart home/building automation. When using this dataset please cite the following paper in Scientific Data, http://dx.doi.org/10.1038/sdata.2016.122 This version of the dataset has been cleaned in the following ways: - Timestamp duplicates have been merged. - IAM (Individual Appliance Monitor) readings set to 0 Watts if above 4000 Watts (above the rated limit of the sensor). - Each IAM has been processed to ensure that it only shows readings for one appliance, where possible. - The ReadMe file has been updated with information about monitored appliance changes. - NaN values have been forward filled (< 2 minute gaps) or zeroed (> 2 minute gaps). This work has been carried out as part of the REFIT project (`Personalised Retrofit Decision Support Tools for UK Homes using Smart Home Technology', Grant Reference EP/K002368/1/1). REFIT is a consortium of three universities - Loughborough, Strathclyde and East Anglia - and ten industry stakeholders funded by the Engineering and Physical Sciences Research Council (EPSRC) under the Transforming Energy Demand in Buildings through Digital Innovation (BuildTEDDI) funding programme. A raw data version of this dataset (deposited 23/09/2015) is also available from the Data Sets link below. 
Type Of Material Database/Collection of data 
Year Produced 2016 
Provided To Others? Yes  
Impact n/a 
 
Description Contribution to the EPSRC-funded METER workshop at the Environmental Change Institute, University of Oxford 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact The Expert workshop, an event from the EPSRC METER project, gathered academics working in Energy Demand and Time-use research to ensure that the project aligned with research priorities of the invited experts and produced relevant data. What was relevant to this workshop was REFIT's work on activity recognition work to understand and quantify energy intensity of activities in a household. Discussions generated new ideas and potential collaboration with Dr Grunewald and others,
Year(s) Of Engagement Activity 2015
 
Description DECC Workshop: Specifying and Costing Monitoring Equipment for a Longitudinal Energy Study 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Policymakers/politicians
Results and Impact DECC Workshop: Specifying and Costing Monitoring Equipment for a Longitudinal Energy Study

Invitation to a workshop organised by DECC, where a number of key stakeholders across Government, Devolved Administrations, research councils and the academic community were invited to contribute their ideas to development a longitudinal survey on energy use in the domestic sector to explore evidence needs for the potential survey and to gather views on the most useful applications for data that could be collected by this survey.
Year(s) Of Engagement Activity 2015
 
Description Invited talk on REFIT public dataset 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact The talk described the REFIT Electrical Loads Measurement dataset, released to the public in Sept 2015 under a creative commons licence. Specifically, we discussed the collection methodology, data cleaning, data formatting and data quality assessment, as well as how the dataset differs from similar public datasets.
The talk sparked further questions and debate, which helped shape the 'Nature Scientific Data' as an open access journal article. We also appreciated the usefulness to the academic research groups and companies of this dataset and how it was being used, and how to design metadata.
Year(s) Of Engagement Activity 2016
URL http://www.nilm.eu/nilm-workshop-2016/
 
Description Invited talk on domestic energy feedback algorithms @ University of Edinburgh 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Postgraduate students
Results and Impact Invited talk for ANC/DTC Seminar at the Institute for Adaptive and Neural Computation at the School of Informatics, University of Edinburgh
Attended by researchers and academics in energy monitoring and feedback. Sparked debate on the possibilities and limitations of data monitoring for making useful inferences from smart meter data. This has enabled further opportunities for collaboration in the area of data monitoring and load disaggregation.
Year(s) Of Engagement Activity 2015
URL http://www.anc.ed.ac.uk/events/lina-stankovic-2015
 
Description Reasoning daily activities in elderly homes based on energy monitoring presented at OB-14 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Talk generated interest in the developed data monitoring and processing methods and the conclusions from the study.

Increased visibility of the project and its results
Year(s) Of Engagement Activity 2014
 
Description Scientific contributor to workshop on 21st century standards and labelling programmes 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Third sector organisations
Results and Impact Scientific contributor to workshop on 21st century standards and labelling programmes

Invited to present and lead discussions in the workshop to explore how technology innovations can be used to improve appliance and equipment energy efficiency policies and programmes. The workshop was organised by International Energy Agency (IEA) and the IEA Implementing Agreement for Energy Efficient End-use Equipment (4E) the Super-efficient Equipment and Appliance Deployment (SEAD) initiative of the Clean Energy Ministerial and the International Partnership for Energy Efficiency.

The presentation was entitled "Analytical tools for understanding appliance usage patterns and the potential for energy savings." Audience included European Energy Agencies, DECC and policy markers.
Year(s) Of Engagement Activity 2015
URL https://www.iea.org/workshops/21st-century-energy-efficiency-standards-and-labelling-programmes.html