W5UP4

W5UP4 – A novel method of determining events in combination gas boilers: Assessing the feasibility of a passive acoustic sensor
25th February 2016 Thomas Neeld

A novel method of determining events in combination gas boilers: Assessing the feasibility of a passive acoustic sensor

Thomas Neeld, UCL Energy Institute


Overview

For researchers, having access to tools that monitor energy systems is imperative to understanding the behaviours of energy consuming devices in the home. Such tools may help to analyse the impact of energy-saving interventions, whether they be behavioural or environmental (insulation improvement for example), or to determine if energy systems are fit-for-purpose.

In the UK over 70% of energy consumed in the domestic sector is done so in gas-fired boilers (2013-2014); with the combination gas-fired boiler becoming more and more prevalent (over 60% of all heating systems in 2012). Thus the combination gas-fired boiler is arguably the most important single technology type to consider in regard to monitoring energy systems in the domestic sector. Researchers have however found it challenging to accurately determine consumption of combination gas-fired boilers for space heating use and often relied on estimates from vast datasets or using temperature sensors.

This project looked to develop a method to ‘spy’ on combination gas-fired boilers in order to determine accurate estimates of energy consumption and demand type (hot water or central heating). The method by which to do this ‘spying’ was selected to be acoustics: Noises produced by boilers are available externally to the boiler and can in theory provide a significant amount of ‘free’ data if one can ‘understand’ the language of the system; similar to how recent developments in speech recognition software can ‘understand’ human speech. Due to the complexity of such analysis, among other techniques, application of machine learning methods are required – this is a growing area of computer science which has found a multitude of applications.

Aims

The aims for the project can thus be summarised as follows:

Based on the acoustic data received from combination gas-fired boilers:

  • Develop and test techniques which determine the demand type of operation (hot water or central heating)
  • Develop and test techniques which determine the time of burner ignition
  • Develop and test techniques which determines the intensity of a burn

Results

Images of the setup:

(a) Picture showing the position of Microphone A on the side of combination gas-fired boiler A. (b) Picture of combination gas-fired boiler A with the outer casing removed showing the internal components. Highlighted in (b) are the Pre-mix fan; heat exchanger (housing the burner assembly); diverter valve and circulation-pump.

(a) Picture showing the position of Microphone A on the side of combination gas-fired boiler A. (b) Picture of combination gas-fired boiler A with the outer casing removed showing the internal components. Highlighted in (b) are the Pre-mix fan; heat exchanger (housing the burner assembly); diverter valve and circulation-pump.

Picture of the control panel on C-GFB A. Attached to the control panel is the event detector setup consisting of a photoresistor, Orisen Prime data logger and power supply. As observed in the figure the photoresistor is placed directly over the burner light indicator.

Picture of the control panel on C-GFB A. Attached to the control panel is the event detector setup consisting of a photoresistor, Orisen Prime data logger and power supply. As observed in the figure the photoresistor is placed directly over the burner light indicator.

 

 

 

 

 

 

 

 

 

 

 

Spectrogram showing frequency distribution of sounds produced by a boiler under investigation. Here we can see the noise produced at the point of ignition (infrasound pressure pulsation):

Plot of the spectrogram produced by combination boiler A for a central heating demand event. Frequency is constrained to maximum of 50 Hz and Power Spectral Density (PSD) to -50 dB. The low frequency spike produced at during ignition has been highlighted by the white ellipse.

Plot of the spectrogram produced by combination boiler A for a central heating demand event. Frequency is constrained to maximum of 50 Hz and Power Spectral Density (PSD) to -50 dB. The low frequency spike produced at during ignition has been highlighted by the white ellipse.

Conclusions

During the MRes year the project was successful in meeting the first two aims for the boiler under investigation: acoustically determining demand type and the time of ignition to a high levels of accuracy (100% and 97% respectively). Additionally it was deemed that significant potential existed for more work in this area, thus the project was taken on by the researcher onto a PhD – see W5UP10.

Additional outputs

  • Setting up the start of a PhD project for the researcher with more ambitious aims – see W5UP10.
  • Collaboration developed between Imperial College London, Dept. of Electrical and Electronic Engineering.
  • Interest developed with industry bodies to potentially develop a collaboration.

Project Team

Student
Thomas Neeld

Supervisor(s)
David Shipworth

Outputs


Publication

Publication in Building and Environment based on work done during MRes:

Neeld, T., Eaton, J., Naylor, P. A., & Shipworth, D. (2016). A novel method of determining events in combination gas boilers: Assessing the feasibility of a passive acoustic sensor. Building and Environment, 100, 1-9.