Smart management of commercial building climate control
16/07/21
< Project overview >
Introduction
Optimising energy management in medium-sized commercial and residential buildings can bring significant energy and cost savings, especially when linked to occupancy control. It is estimated that around 70% of the rooms in commercial buildings are heated or cooled despite being empty, so resulting in 40% of a buildings’ overall energy expenditure being wasted.
Modern AI-based control algorithms can develop a model of the building from energy consumption and occupancy data, allowing optimisation of energy consumption via IoT-enabled control (including allowing for occupancy levels): in medium-sized buildings these IoT-based systems offer cheaper path to optimising energy management than a fully wired control system.
The key is to provide the cloud-based algorithms enough data at low cost, so they can build up a model of the building from which to base initial control decisions. This project explored a low-cost method of providing this set-up data.
Project aims
In medium-sized commercial and residential buildings, rendering energy management ‘smart’ requires solutions to two significant challenges:
Uncertainty (due to occupancy, user behaviour, weather or other factors) must be taken into account in the energy management process.
Existing ad-hoc solutions must be replaced by more automated and systematic solutions with guaranteed performance properties.
And of course, a building user must be convinced and see a clear return on investment from an activity that will be non-core business.
In this project Alex Rogers of Computing Science supported a Said Business School spin-out EcoSync to overcome these challenges by developing novel sensors that provided data to blend tools from model-based and data-based modelling to underpin IoT control techniques, approaches that are often treated separately in the energy management literature.
The principle Pitch-In barriers targeted were:
Resource access barriers: additional funding required for hardware to provide additional data points.
Business case: lack of knowledge on how to deploy low cost IoT systems.
Business case: lack of understanding in how IoT can generate value in a given application.
What was done?
The original work programme had four phases, including practically demonstrating the new data collection methods in a building and quantifying how it lowered costs installing an AI-based IoT control system. However, COVID lockdowns severely disrupted the programme, and we dropped the demonstration phase. In the light of this we undertook three phases.
Phase 1: Data logger development and prototype production
USB-based dataloggers to record both temperature and occupancy information at the room level were specified and developed by Prof Alex Rogers of Computer Science. This device included light energy harvesting to power the sensor. Their small size is also a significant benefit.
Phase 2: Data collection
The sensors were deployed in the Dyson Perrins Building, currently the home of the School of Geography. It is this an office-type environment. Lockdown, which prohibited access to the building, occurred during the sensors’ deployment. However, the sensors successfully collected data over an extended time.
Phase 3: Addition of IoT control to optimise heating
Installation of IoT controls to radiators proved impossible to achieve, as the building was in lock-down and University IT staff were swamped with Covid-related mitigation work.
Phase 4: Analysis and dissemination
EcoSync developed a Firebase Database into which data was downloaded and visualised. Analysis showed it is sufficient to develop a model of the building and can provide a low-cost method for IoT building control set-up.
Results
The project developed low-cost sensors to collect temperature and occupancy data in buildings. The sensors included light energy harvesting to provide power.
The sensors successfully collected sufficient data which, when combined with building plans, can underpin the development of a virtual building model using AI-based tools. This model can underpin IoT-based control of heating in the building.
This process can radically reduce the cost of installing an IoT-based control system into a medium-sized commercial or residential building.
Impact
The project is underpinning the development of EcoSync.
It combines artificial intelligence with MIT technologies and applies it to real building heating IoT-based control. EcoSync recently completed a successful funding round in January 2021 to continue its development.
Next steps
EcoSync continues to develop as a company and to promote the results. EcoSync did eventually set up the Dyson Perrin building with remote temperature control, modifying their gateway as the University’s IT Facility did not give WiFi access: it is successfully running on EcoSync’s own network with 4G hotspots.
EcoSync are currently working on a request from another Engineering Department project to use this project’s sensors for data collection and data visualisation. If the project is approved the sensors will be in use again from September 2021.
Professor Alex Rogers continues to develop his low-cost sensor technology, broadening the range of applications (eg marketed as Audiomoth for biodiversity though a spin-out Open Acoustic Devices.
Lessons learned
The involvement of Computer Science, EcoSync and Engineers enabled the fast design and manufacturer of novel sensors that were proven to lower the capital cost of a building heating IoT-based control system.
The light energy harvesting idea for the sensors was only identified after the project started: the flexibility of Pitch-In to accommodate this significantly improved the sensors’ capability.
The Covid lockdown was unexpected and caused huge issues. Firstly, it affected how the demonstration building was used and secondly key personnel (eg Estates IT) were totally swamped with measures to mitigate effects upon building re-opening.
The project originally included an engineering science control professor and the team’s control algorithms, which would have been useful as the sensor data gathering could have been better optimised. However, the professor was in charge of switching all of the department’s lecturers online due to the COVID lockdown and so became unavailable. EcoSync accessed MIT’s algorithms so the work managed to progress.
What has Pitch-In done for you?
Pitch-In has allowed space for the development of novel sensors and shown they can be part of reducing the cost of a cloud-based IoT building heating control system.
The project has also provided valuable learning on the difficulties of connecting to building IT infrastructure (and associated risk that means IT staff are reluctant to support) and developed WiFi-based methods to overcome this
Project lead
Paul Goulart – University of Oxford