Proceedings of the Institute of Acoustics

 

A comparison of two acoustic characterisation methods of air source heat pumps

 

MA Radivan, University of Salford, Greater Manchester, England
LS Barton,  University of Salford, Greater Manchester, England
AJ Torija Martinez, University of Salford, Greater Manchester, England
JA Hargreaves, University of Salford, Greater Manchester, England
V Acun, University of Salford, Greater Manchester, England
DBC Wong-McSweeney, University of Salford, Greater Manchester, England
SN Graetzer, University of Salford, Greater Manchester, England
DC Waddington, University of Salford, Greater Manchester, England

 

1 INTRODUCTION

 

The transition towards sustainable heating solutions is gaining momentum in the UK, with air source heat pumps (ASHPs) a strong candidate for residential heating in the near future. This move is prompted by the nation’s commitment to achieving net-zero carbon emissions, a goal that necessitates a departure from traditional heating systems. ASHPs have emerged as a compelling alternative to conventional methods of home heating, offering improved efficiency. Despite their benefits, the shift would bring a new source into our homes and communities, leading to concerns over their potential impact on environmental noise.

 

Addressing these concerns, the MCS-020 standard for permitted development rights sets out noise based guidelines based on a single value A-weighted sound power level and a presumed point source directivity. This framework provides a straightforward method for predicting ASHP noise, accessible even to those with limited experience of acoustics. However, this process may overlook factors such as the tonality and directivity of ASHP emissions, both of which can significantly affect human perception⁽¹⁾.

 

This paper explores these high fidelity measurement methods of ASHP noise. From these a sound power level determination will be calculated using BS EN ISO 3746: 2010⁽²⁾ which is a well- established method that quantifies in-situ 1/3 octave band levels; and the multi-panel power ranking method from the Microflown Scan&Paint 3D system. Additionally, capabilities of the Scan&Paint 3D system, which uses state of the art technology to create accurate spatial mappings of noise sources, are explored to inform of the contribution of components inside the ASHP casing and indicate the near-field directivity of the unit. The data from both of these methods will also be used to plot a power spectral density (PSD) indicating the frequency characteristics of an ASHP. Through a comparison of these methods, the inherent challenges and implications of the findings will be discussed. The research aims to assess the validity of the assumptions underpinning the MCS-020 standard, thereby contributing to a more nuanced understanding of ASHP acoustics and their impact on future suburban soundscapes.

 

2 METHODOLOGY

 

2.1 Measurements at Energy House 2.0

 

The measurements discussed in this paper were taken in the Energy House 2.0 facility at the Univer- sity of Salford. It consists of two climate-controlled chambers that can each accommodate two de- tached houses. The ambient conditions are controlled by a Heat, Ventilation, and Air-cooling (HVAC) system which was switched off for the measurement to reduce background noise. The benefit of mea- suring an ASHP at EH 2.0 is the use of an indoor, acoustic environment. Whilst the measurement is feasible outdoors there is a possibility of transient noises disturbing the signal and a higher overall background level.

 

2.2 Measurement overview and conditions

 

The two measurement methods were taken of the York YKF05CNC Monobloc ASHP installed at EH 2.0 on two days with different operating conditions. In the first case, the parallelepiped method of sound power level determination was conducted with an engineer from the manufacturer present in order to control the operating conditions of the unit. Typically, the external ambient temperature and the target heating temperature determines the load of the ASHP cycle but in this case it was independent. The operational parameters; the compressor frequency and fan speed were recorded.

 

On the second occasion, the Scan&Paint 3D system was used with an ambient temperature 5°C with 85% relative humidity, the internal temperature of the house was cooled to 10°C, and the target temperature set at the thermostat was the maximum 30°C. The compressor frequency and fan speed were again recorded. It was found that the fan speed varied between 440-460rpm and the compressor frequency started at 47Hz and dropped to 19Hz over the course of the measurement. The fan speed at a distance of 60cm from the fan grille measured at ∼ 4m/s.

 

2.3 BS EN ISO 3746:2010 parallelepiped method

 

 

Figure 1: The microphone array used for the parallelepiped measurement from BS EN ISO 3746:2010⁽²⁾

 

The main pieces of equipment required for the sound power level determination are class 1 measurement microphones for the enclosing measurement surfaces, in this case two for each surface, and a data acquisition unit which in this case was Dewesoft SIRUSi-8xACC.

 

The measured sound pressure levels from the noise source and the background levels for each enclosing measurement surface of the sound source, as a function of the surface area, are determined to calculate the sound power level. Eight microphones were arranged around the ASHP, as shown in figure 1, following the specifications in BS EN ISO 3746:2010⁽²⁾.

 

In this case, there are two planes adjacent to the sound source, which are assumed to be totally reflective - the ground and the wall. A reading of the background sound pressure level, with the sound source shut off, is taken to make a correction, K1A, if the background level is too high. In this case a correction was not necessary.

 

A correction K2A is made depending on the environment in which the measurement was taken. If the measurement was made outside, a free-field assumption is made and no correction is required; but if the measurement was taken indoors, a correction is made for the reverberation time of the room and a diffuse field assumption is made, as in this case.

 

 

In equation (1) the sound pressure level for each measurement surface is calculated, where K_1A = 0 .

 

 

In equation (2) the sound power level is calculated where S is the area, in square metres, of the measurement surface and S₀ = 1m² a constant defined in BS EN ISO 3746-2010⁽²⁾. From the measured values the sound power level was determined using a Python script developed for this purpose at the University of Salford.

 

2.4 Microflown Scan&Paint 3D

 

 

Figure 2: The Microflown S&P3D system in use with the components identified

 

The Microflown Scan&Paint 3D system is an innovative measurement system designed for precise sound source localisation and visualisation of sound fields from sound sources. The system employs the company’s proprietary ‘Ultimate Sound Probe’ (USP) that captures, in a compact device, both pressure and a 3D cartesian velocity vector via a set of three orthogonal ‘flown’ sensing elements. This is scanned across the defined measurement surfaces, while being tracked in both position and rotation by an infrared camera. From this, the system can produce 3D colour plots, such as vector plots and colourmaps, and sound power level determinations.

 

A 3D model is acquired prior to the measurement by photogrammetry or LiDAR techniques. In figure 2, a breakdown of each component in the system can be seen - the system is comprised of: a companion software for the system called Velo running on a PC ‘A’; a data acquisition unit (DAQ) ‘B’; the ultimate sound probe ‘C’ with an optional windshield ‘D’, which is mounted on an orb ‘E’; with circular retroreflectors to be tracked by an infra-red camera ‘F’ in 3D space, which is also mounted on a remote-control handle which allows functions in the software to be controlled. The sound power level is determined by use of a method called multi-panel ranking which is loosely based on the ISO 9614-3:2009⁽³⁾ .

 

One of the limitations of the system identified in pilot studies is the susceptibility to wind noise⁽⁴⁾. It was found that the maximum wind speed of the ASHP was around 6m/s measured from 3cm away which is the ideal measurement distance for the probe. A study was conducted to understand the conditions under which the probe can tolerate flow noise. It was found that the ideal conditions that the probe should be used under less than 4m/s. Taking this into account, for this measurement a modular structure was used to create a frame that spaced the scanning surface 60cm from the probe, in order to reduce the wind speed. The wind speed at this distance was measured at less than 2m/s.

 

3 RESULTS

 

3.1 BS EN ISO 3746:2010 parallelepiped method

 

 

Figure 3: The L_eq from the parallelepiped method (refer to figure 1 for microphone positions)

 

 

Figure 4: The L_w from the parallelepiped method (refer to figure 1 for microphone positions)

 

In figures 3 and 4, the directivity of the heat pump from the parallelepiped method is indicated from the equivalent sound pressure level and sound power level contributions for each surface for the microphones at each enclosing surface. In figure 1, the microphones are coloured to enable the identification of a rudimentary directivity. The overall sound power level from the parallelepiped method is shown in figure 4.

 

 

Figure 5: A PSD from Mic 1, from the parallelepiped method

 

An unweighted PSD can be seen in figure 5 from microphone 1. The first three expected orders from the fan and the compressor have been plotted, along with calculated tonal peaks and the broadband level excluding tonal components. The signal was acquired with a sampling frequency of 20KHz and processed with an FFT length of 16384 leading to a resolution of 1.22Hz which is sufficient for identification of low frequency tonal components of the signal.

 

3.2 Microflown Scan&Paint 3D

 

 

Figure 6: Overall PSD from Mic 1, from S&P3D method

 

A narrow band, unweighted, PSD can be seen in the figure 6, which indicates the frequency components of the heat pump and two tonal components seen at higher frequencies. There are no lower tonal components present which would be expected from the fan and the compressor. These measurements were taken with an FFT length of 2048 and a sampling rate of 24KHz which gives the frequency plot a resolution 11.7Hz which is insufficient for identifying tones and explains the absence of lower harmonics.

 

 

Figure 7: A full frequency range intensity vector plot of the ASHP

 

The intensity vector plot is shown in figure 7 for the full frequency spectrum, also showing directivity of the ASHP and the components within the enclosure of the unit. The sources of noise corroborate the schematic of the heat pump⁽⁵⁾ - the intake can be seen on the right side, the fan on the centre left and compressor in the lower left-hand side.

 

 

Figure 8: An intensity colour map ranging from 20-70Hz of the ASHP

 

An intensity colour map can be seen in figure 8 for 20–70 Hz. This frequency range is where the first orders from the fan and compressor are expected, based on the schematic.

 

The sound power level for each panel is presented in figure 6 along with the total sound power level from the multi panel ranking method for 1/3 octave bands. The total sound power level is 49.37 dBA which is around 5dB lower than the parallelepiped method. The quoted overall single value sound power level is quoted as 56dBA in what is described as a cooling silent mode⁽⁵⁾. This corroborates the parallelepiped method but is unlikely to be correct as there is no guarantee the operating cycles match in terms of their working parameters.

 

3.3 Discussion

 

The discrepancy between the two methods is for of variety of reasons: a big contributing factor to this is the fact that the measurements are taken in-situ in an uncontrolled environment. Another is the 60cm distance between the front face of the heat pump leading to a reduced measured intensity, which is closer to the noise floor of the probe. Note that being close to the probe noise floor does not necessarily increase the active intensity, as would happen with sound pressure close to the noise floor of a microphone, because the sensor self-noise is incoherent between pressure and particle velocity. But nonetheless, it can compromise measurement accuracy.

 

The shape between the two total sound power level curves seen in figures 5 and 6 from 80 – 8KHz are somewhat similar but the limitations of the multi-panel ranking method can be seen: below 80 there is a cut off, and there is negative sound power levels for every panel except the right hand one. It is unclear as to why this happens, but it will contribute to the total sound power level being lower than the parallelepiped method. The final reason is that it was not possible to ensure that the heat pumps were operating under the same conditions. The first measurement was conducted in a service mode with the help of an engineer from the manufacturer and the fan speed and compressor frequency were static throughout the measurement whereas this was not possible for the Scan&Paint 3D measurement, as discussed in the methodology.

 

 

Figure 9: A plot of the sound power level from S&P3D

 

4 CONCLUSIONS

 

In this paper the findings of two measurements that attempted to measure the frequency and directivity characteristics of an ASHP have been presented. Whilst a comparison of two measurement methods has been conducted, difficulties were encountered due to the dynamic nature of a heat pump cycle have restricted this. The results show the complex nature of heat pump’s operation and demonstrate the differing emission from the various components inside an ASHP. This suggests that directivity cannot be assumed to be a point source, as is currently done in most modelling, including the MCS-020 standard. Research to extend these methods to capture ASHP directivity is under way. The presence of tonal components has shown that the MCS assumption of a single value sound power level may be too limited.

 

These findings also reinforce that an ASHP has a complex acoustic character, with varying cycles, frequency content and directivity. The human response to this is another priority research area.

 

5 REFERENCES

 

1. M Torjussen, J Harvie-Clark, A Lamacraft, and P Rogers. NOISE FROM ASHPS - WHAT DO WE KNOW? In ACOUSTICS 2023 . Institute of Acoustics.
2. BS EN ISO 3746:2010.
3. BS EN ISO 9614-3:2009.
4. Lucy Barton, Jonathan Hargreaves, and Max Radivan. Higher-fidelity analysis of air source heat pump noise via scanning intensity measurement. In Proceedings of INTER-NOISE 2024 .
5. https://brsheatpumps.co.uk/wp-content/uploads/2022/10/YKF-brochure-2022-1.pdf.