Applied Acoustics 140 (2018) 100–109

Contents lists available at ScienceDirect

Applied Acoustics
journal homepage: www.elsevier.com/locate/apacoust

Technical note

HVAC noise control using natural materials to improve vehicle interior T

sound quality
⁎
Sneha Singh , A.R. Mohanty
Department of Mechanical Engineering, Indian Institute of Technology Kharagpur, Kharagpur 721302, West Bengal, India

A R T I C LE I N FO A B S T R A C T

Keywords: Heating, Ventilation and Air Conditioning (HVAC) unit is a major noise source in a vehicle’s interior space.
Noise control Reducing this noise will improve the sound quality of the vehicle’s interior space and enhance passengers’
Sound quality experience. However, current noise control techniques are high-cost and hazardous to environment. Therefore,
Jute felt this paper studies the first ever usage of low-cost, biodegradable natural materials for vehicle’s HVAC noise
Natural materials
control. Jute felt and waste cotton were found to have higher sound absorption coefficients than other common
Sound absorption
sound absorbing natural materials, and were chosen as sound absorbers for noise control treatment of a pro-
HVAC
totype HVAC unit. Noise sources in the unit were identified and ranked, and the treatment was applied to these.
The treatment significantly reduced the noise level (by 4 dBA) and loudness level (by 7 sones) due to the unit at
the reference passenger’s ear location, with negligible cost and weight. Sound quality evaluation by 24 parti-
cipants showed that the treatment significantly reduced the annoyance of the vehicle interior soundscape
comprising the HVAC noise. Thus, jute felt and waste cotton are low-cost, light-weight, biodegradable and
recyclable natural materials with high potential for HVAC noise control.

1. Introduction
Sound quality of a vehicle interior space is an important aspect of a
vehicle and demands special NVH (Noise, Vibration and Harshness)
attention. Reducing noise inside a vehicle improves the sound quality of
the vehicle interior space, which in turn leads to a higher passenger
comfort, better driving experience, and lesser driver distraction.
Moreover, reducing noise and improving vehicle interior sound quality
enhances a customer’s perception of the vehicle brand; thereby the
vehicle attracts more customers and gets a competitive advantage in the
market [1,2].
Research and development over the last two decades have led to
quieter and better sounding engines. As a result, secondary sound
sources located within a vehicle cabin such as heating, ventilation and
air-conditioning (HVAC) system, entertainment systems, and audio
driver assist systems have become more perceptible to passengers [3].
Among the secondary sources, the HVAC system is the most dominant
noise source in a vehicle’s interior space as it operates throughout as
long as the vehicle is running. Additionally, the HVAC and blower fan
noise reaches the interior space without any sound isolation and can
strongly impact passengers’ comfort. In the hot climate of a tropical
country like India, a vehicle’s HVAC system operates continuously at
higher blower speeds and is one of the most important interior noise
sources. Therefore, HVAC noise control is needed to improve the sound
quality of the vehicle interior space, and it is gaining growing attention
from researchers and manufacturers [3–8].
1.1. Conventional methods for HVAC noise control and sound quality
enhancement
The major noise source in the HVAC unit is the aerodynamic blower
noise [9–12]. Therefore, the conventional strategy for reducing HVAC
noise includes design changes to the blower and its blades [9,10,13,14].
However, the unit examined in this paper was already designed and
produced by the manufacturer for optimum performance; hence only
noise control strategies at post-production stage are discussed in this
paper. Existing post-production noise control techniques include active
noise control [4,11,15], and passive noise control using synthetic sound
absorbing materials such as micro-perforates [5,6], fiberglass, glass
wool and polypropylene [16]. These approaches achieve up to 6–10 dB
noise reduction. However, active noise control approaches involve
costly equipment, work well only in low frequencies, and are effective
only in specific zones of the vehicle interior space. Passive noise control
approaches work best at frequencies above ≈500 Hz, and they apply
noise reduction throughout the interior space. However, the synthetic
sound proofing materials are costly as their manufacturing requires

⁎
Corresponding author.
E-mail address: snehasingh.iitkgp@gmail.com (S. Singh).

https://doi.org/10.1016/j.apacoust.2018.05.013
Received 13 February 2018; Received in revised form 7 May 2018; Accepted 9 May 2018
Available online 24 May 2018
0003-682X/ © 2018 Elsevier Ltd. All rights reserved.
S. Singh, A.R. Mohanty Applied Acoustics 140 (2018) 100–109

high-end equipment (for e.g., micro-perforated panels [5,6]) or high- Table 1

temperature extrusion and synthetic chemical processes (for e.g., fi- NRC and densities of 100 mm thick samples of some common natural materials.
berglass, glass wool, polypropylene [16]). For the same reasons, these Sample NRC Density (kg-m−3)
materials have higher carbon footprints, and are neither biodegradable
nor recyclable. Jute felt 0.75 108
Flax 0.70 155
Waste cotton 0.65 137
1.2. Natural materials for noise control Rice straw 0.65 154
Wood curls 0.55 110
Due to the discussed disadvantages of synthetic materials, the recent Coconut Coir 0.45 108
trend in automotive noise control is a shift towards natural materials,
also known as natural-fiber materials. Some common sound absorbing
1.0
natural materials are jute, cotton, flax, kenaf, hemp, coconut coir,
bamboo curls, and banana [16–21]. These materials provide good

Sound absorption coefficient

sound absorption, are low-cost, light-weight and biodegradable.
Therefore, they have begun to be applied for sound proofing auto-
motive components such as floor carpeting [18] and car boot liners
[20].
However, currently there is no research study on the usage and
effectiveness of natural material(s) for noise control of a vehicle HVAC
system. Testing the effectiveness of natural materials for vehicle HVAC
noise control would help in developing low-cost and eco-friendly
techniques to enhance vehicle interior sound quality. Furthermore, the
existing HVAC noise control case studies ([4–6]) are usually limited in
scope as they do not test the effectiveness of the noise control treatment
on actual humans. Therefore, there is a need for a systematic case study
of a vehicle HVAC noise control using natural materials that tests its
effectiveness through human-subject sound quality experiments. Such
case studies could guide future NVH engineers across various industries
to adopt eco-friendly and human-centered noise control practices.
This paper presents the first ever case study of usage of natural
materials, namely, jute felt [22] and waste cotton for noise reduction in
a vehicle’s HVAC unit. The other novelty of this paper is that it validates
the effectiveness of this noise control treatment through human-subject
sound quality evaluation of the HVAC noise and the vehicle interior
soundscape comprising the HVAC noise. It also presents a first ever
comparison of the sound absorption coefficient of jute felt with other
contemporary sound absorbing natural materials. The remaining paper
is organized such that Section 2 describes the materials used for noise
control. Section 3 describes the test HVAC unit, and its noise evaluation
is in Section 4. Section 5 describes the noise source ranking in the unit.
Section 6 describes the treatment done on the unit and its noise eva-
luation. Section 7 describes the sound quality evaluation of the HVAC
unit.
2. Details of the natural materials used
Jute is a naturally occurring plant that is readily grown at very low
cost in the hot tropical and rainy regions of the eastern parts of India.
Jute fiber is a lignin-cellulose fiber composed primarily of cellulose
(major component of plant fiber) and lignin (major component of wood
fiber). Raw jute fibers have high aspect ratio (length/diameter), high
specific modulus, and high tensile strength (see details in [22]). Raw
jute fibers are cleaned and spun into a jute yarn. Stacks of jute yarn laid
in a definite sequence are pressed under pressure, and the resulting
sheets are bonded using natural rubber. The resulting composite is
called a “jute felt” (see details in [22]). Jute felt is biodegradable and
recyclable, and thus environment-friendly. Jute felt has been ex-
tensively tested by Fatima and Mohanty ([22]) and 1% natural rubber
treated jute felt is found to have lower flammability than fiberglass,
wool and cotton. This jute felt has been shown to have very high normal
specific sound absorption coefficient (see details in [22,23]). Recently,
jute felts have been successfully applied to noise control in vacuum
cleaners and domestic dryers, [19] and in building semi-anechoic
chambers [24].
Experiments were done to compare sound absorption coefficient of
a jute felt sample with some of the other existing low-cost sound
0.8
0.6
0.4 Jute felt
Waste cotton
Coconut coir
0.2 Flax
Wood curls
Rice straw
0.0
500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Frequency [Hz]
Fig. 1. Sound absorption coefficients of 100 mm thick samples of some common
natural materials.
absorbing natural materials, namely, coconut coir, flax, cotton, wood
curls, and rice straw. The experiments were done in Acoustics and
Condition Monitoring Laboratory at the Indian Institute of Technology
Kharagpur using the two-microphone method in the B&K4206 im-
pedance tube setup as per ISO 10534-2 [25]. The sound absorption
coefficient will increase with increase in packing density of these ma-
terials. Therefore, for a more valid comparison the samples of these
natural materials had similar packing densities within the range of
≈110–155 kg-m−3. Table 1 shows the noise reduction coefficient
(NRC) values and the densities of these natural materials. Fig. 1 shows
the normal specific sound absorption coefficient of these materials. Jute
felt was found to have the highest sound absorption coefficient and
highest NRC, and was the lightest of the tested materials. Therefore, it
was chosen as the primary material for noise control. Waste cotton is
more flexible to fit into any shape and it also had good sound absorp-
tion, so it was also decided to apply waste cotton, but only in areas such
as cavities where jute felt could not be easily applied.
3. Description of the HVAC unit
The HVAC unit used for this case study is a prototype supplied by a
manufacturer. It is made of ABS (Acrylonitrile butadiene styrene)
plastic and has the following main components – fresh air inlet, re-
circulation inlet, evaporator, heater, and outlet vents (defrost vent,
cabin vent and foot vent). Figs. 2 and 3 show the positioning of the
HVAC unit in a car, and the detailed schematic of the HVAC unit re-
spectively. A standard vehicle HVAC unit can operate in either “fresh
air inlet mode” i.e. fresh air intake from outside the vehicle or “re-
circulation air inlet mode” i.e., air intake from inside the vehicle cabin”
(see Fig. 2). For this case study, the fresh air inlet was blocked and the
HVAC operated in the noisy recirculation mode. The recirculation inlet
houses a 43-vane blower that sucks in the cabin air (Fig. 3). The air
enters the unit through the recirculation inlet vent, passes through the
blower, and then through the evaporator that consists of an evaporator
coil for cooling the incoming air flow and a filter to purify the air. Some

101
S. Singh, A.R. Mohanty Applied Acoustics 140 (2018) 100–109

1= air release by defrost vent

2= air release by cabin vent
3= air release by foot vent
4= air intake by recirculaƟon inlet
5= air intake by fresh air inlet vent
6= dashboard

6 4
2 1
HVAC 5
3 unit

Fig. 2. Positioning of an HVAC unit in a car.

part of the cooled air is then directly released into the cabin through the
cabin vent and towards the windscreen through the defrost vent. The
remaining air flow passes through the heat exchanger that consists of a
heating coil for heating the incoming air. The heated air is isolated from
the cold air flow through the controllable flappers (see Fig. 3(a)). The
heated air is released towards the floor of the car through the foot vent.
The cabin vent is connected to a duct with two long side vents that
release the cold air flow to rear passengers and a central vent at the
dashboard to release the cold air flow to the front passengers. The scope
of this case study was to test the noise at the location of the front
passengers; therefore side ducts were removed from the test HVAC unit.
The HVAC unit operates on four different manufacturer-set conditions
as summarized in Table 2. These operating conditions were used for all
the measurements, and for the sake of convenience these conditions
will be referred to as “low speed”, “medium-low speed”, “medium-high
speed” and “high speed” conditions from here on.
at the Acoustics and Condition Monitoring Laboratory of the
Mechanical Engineering Department at the Indian Institute of
Technology Kharagpur. The overall ambient noise level of the mea-
surement room was 23.7 dBA for all the measurements. Fig. 4 shows a
view of the measurement setup.
Three ½” B&K 4189 free field microphones on tripods were used to
measure the sound pressure level (SPL) of the sound emitted from every
side of the unit. These microphones were placed at positions 2, 3 and 4
of Fig. 4 at 0.01 m from the front, right and left side of the unit at a
height of 0.5 m from the ground. Additionally, a microphone was
placed at position 1 (see Fig. 4), at a distance of 1.0 m away from the
center of the unit and 1.2 m from the ground. This was the reference
position provided by the automotive manufacturer as the location of the
front passenger’s ear. Later, microphone 1 was replaced with a B&K
4200 D Human Audio Torso Simulator (HATS) to simulate the pas-
senger’s ear more accurately and record binaural sounds which were
later used for psychoacoustic analyses and sound quality evaluation.

4. Noise evaluation of the baseline HVAC unit

4.1. Measurement setup
Measurements were done with the prototype HVAC unit at every
operating condition. These measurements were referred to as the
baseline measurements. The measurements were done in a quiet room
4.2. Noise evaluation
The binaural sounds measured at position 1 by the HATS were
analyzed using PULSE Sound Quality to obtain the SPL and the standard
psychoacoustic metrics namely loudness, sharpness, and roughness.
Table 3 shows the acoustic and psychoacoustic metrics of the HVAC

Defrost vent Recirculation

air inlet vent
Cabin vent
Filter Fresh air
inlet vent
was closed
Recirculation
air inlet vent
Foot
vent Fresh air Line of the
inlet vent foot vent
was closed Recirculation
Cavity
inlet
Defrost vent
Cabin vent

Blower

Heating coil Filter

Flappers Evaporator coil Blower
(a) (b)
Fig. 3. Detailed schematic of the HVAC unit. (a) Cut side view of the HVAC unit. (b) Cut top view of the HVAC unit.

102
S. Singh, A.R. Mohanty Applied Acoustics 140 (2018) 100–109

Table 2
Operating conditions of the HVAC unit.
Sl. No. Blower speed indicator Blower motor terminal voltage Blower motor speed Air outlet speed at cabin vent Air inlet speed at recirculation vent (m/
position (V) (RPM) (m/s) s)

1 Low speed 7.0 2442 8.3 3.9

2 Medium-low speed 9.0 2958 10.4 5.1
3 Medium-high speed 10.5 3342 11.4 5.5
4 High speed 12.5 3702 13.2 6.4

1
6 8
7
9
3

4 1,2,3,4 = microphones
2 5 = exciter
6 = vibrometer
7 = defrost vent
8 = cabin vent
5 10 9 = recirculaƟon inlet
10 = foot vent

Fig. 4. Measurement setup.

noise spectrum at the passenger’s left and right ear position. Fig. 5 70
Pressure (dB re 20μPa)

shows a typical baseline HVAC noise spectrum (at right ear) at medium- 60
high blower speed. 50
The manufacturer had set the target SPL at medium-low speed as
40
65 dBA. However, the average baseline SPL at that speed was 68.4 dBA,
which was higher than the accepted level. It was further observed that 30
loudness level of the baseline HVAC unit at medium-high and high 20
speed conditions had crossed 32 sones, which is the threshold for
10
hearing damage during long-term exposure. Thus, there is a need to
0
reduce the noise level and loudness level of the HVAC unit.
100

160

250

400

630

1.6k

2.5k

6.3k
25

40

63

1k

4k

10k

16k
Frequency (Hz)
5. Noise source identification
Fig. 5. A typical baseline HVAC noise spectrum at medium-high blower speed.
As part of the standard noise control strategy, first an experimental
modal analysis was performed to check if the set operating conditions
random excitation was provided at a 45° angle by a B&K 4824 elec-
were exciting the unit at any of its natural frequencies. This was fol-
tromagnetic exciter (Fig. 4, sensor 5) driven by a B&K 2732 power
lowed by noise source ranking of the HVAC unit.
amplifier in the frequency range 0–2 kHz. The excitation force was
measured by a B&K 8002 force transducer and the unit’s vibration re-
5.1. Experimental modal analysis (EMA)
sponse was measured by a laser vibrometer Polytec PDV 100 (Fig. 4,
sensor 6). The natural frequencies of the HVAC unit were found to be
An EMA was performed to find out the natural frequencies of the
157.5 Hz, 295 Hz, 320 Hz and 875 Hz. The major noise in the HVAC
HVAC unit so as to check if the driving frequencies at the set operating
unit is the aerodynamic noise of the air entering the unit through the
conditions corresponded to any of the natural frequencies of the unit. A

Table 3
Acoustic and psychoacoustic metrics of baseline HVAC noise spectrum.
Sl. No. Blower speed indicator position Measurement channel SPL (dB) SPL (dBA) Loudness (sones) Sharpness (acums) Roughness (asper)

1 Low speed Left ear 66.3 63.2 19.3 1.43 1.69

Right ear 67.2 63.9 20.1 1.43 1.42

2 Medium-low speed Left ear 71 68 25.7 1.49 1.45

Right ear 71.9 68.8 26.9 1.48 1.53

3 Medium-high speed Left ear 73.8 71.1 30.8 1.52 1.63

Right ear 74.3 71.5 32 1.52 1.39

4 High speed Left ear 76.1 73.5 35.9 1.56 1.51

Right ear 76.7 74.1 37.4 1.56 1.46

103
S. Singh, A.R. Mohanty Applied Acoustics 140 (2018) 100–109

Table 4 is described in Table 6, Figs. 7 and 8.

Driving frequencies and natural frequencies of the HVAC unit.
Sl. No. Condition Blower Measured Blower Blade pass 6.2. Evaluation of the treated HVAC unit
motor natural motor frequency
speed frequency frequency (Hz) The resultant HVAC noise was measured and analyzed using the
(RPM) (Hz) (Hz) same setup as used for baseline measurements (see Fig. 4). Table 7
1 Low speed 2442 157.5 40.7 1750.1
shows the acoustic and psychoacoustic metrics of the HVAC noise
2 Medium- 2958 295.0 49.3 2119.9 spectrum at the passenger’s ear position after applying the treatment.
low speed Fig. 9 shows the typical HVAC noise spectrum (at right ear) at medium-
3 Medium- 3342 320.0 55.7 2395.1 high blower speed after applying the treatment. Compared to the
high speed
baseline, the treatment resulted in a reduction of SPL by 4 dBA (3.7 dB)
4 High speed 3702 875.0 61.7 2653.1
at the passenger’s ear position (a minimum SPL reduction of 3.6 dBA
per blower speed). Further, the treatment significantly reduced the
inlet vent at the blower motor frequency, and the air flowing inside the HVAC’s loudness level by 7 sones (24% reduction). However, there was
unit after passing through the vanes of the blower at the “blade pass no significant change in sharpness and roughness. In Section 4.2, it was
frequency” [9–12]. The blower motor frequency, fb, can be obtained found that the loudness of the original HVAC system had crossed the
from the blower motor RPM, Nb as: fb = Nb/60. The blade pass fre- threshold for hearing damage during long-term exposure (32 sones) at
quency is fbpf = 43 × Nb/60 (here number of blades = 43). Table 4 medium-high and high speeds. This problem was solved by applying the
shows the driving frequencies of the HVAC unit. It is observed that the proposed treatment, as it resulted in the loudness levels of the HVAC
aerodynamic noise generated in the HVAC unit does not excite any of unit to be less than 30 sones at every blower speed. Thus, the treated
its natural frequencies. Thus, the unit does not demand a frequency HVAC unit was deemed safer for long term sound exposure.
specific noise control or a change in the operating conditions. Now, the An anemometer was used to measure the outlet air speed at the
next step was to find the dominant noise sources in the unit and apply cabin vent after applying the treatment. There was no change in the
noise control treatment to those sources. outlet air speed at every operating condition. The proposed treatment
costed approximately $0.5 and weighed approximately 0.15 kg (≈3%
5.2. Noise source ranking of the total weight of the baseline HVAC unit). Based on the objective
data, the proposed treatment was deemed effective as it reduced the
The dominant noise sources in the HVAC unit were identified using noise and loudness level significantly with negligible cost and weight,
sound intensity measurements. A B&K 2260 two microphone sound and without compromising the air flow performance of the original
intensity probe with 12 mm spacing was used to measure the sound HVAC unit.
intensity all around the unit, by roving the hand held probe (see Fig. 6)
at a perpendicular distance of 0.01 m from the different surfaces of the 7. Experiments for sound quality evaluation of the HVAC unit
unit. Table 5 shows the noise source ranking of the HVAC unit based on
the 30-s-mean sound intensity level radiating from the different noise Finally, this case study tested the effectiveness of the applied
sources. The radiated noise was the highest from the recirculation inlet. treatment on actual humans through a human-subject evaluation of the
From the principles of noise control, the noise must be reduced at the “annoyance” of the baseline and the treated HVAC sounds. The ex-
most noisy source, here the recirculation inlet, to significantly bring periment details are given below:
down the SPL. Thus, the recirculation inlet must be treated. Other noisy
sources were the defrost vent and the cavity near it. Thus, it was 7.1. Participants
decided to use sound absorbing materials at these locations.
Data was collected from 24 adult participants, 9 females and 15
6. Noise evaluation of the treated HVAC unit males, with a mean age of 32 years (SD = 8.47 years). The participants
included 18 institute staff and students and 6 people from outside the
As discussed in Section 2, jute felt and waste cotton were chosen as academia.
materials for noise control. A number of treatments were tried and
tested and the treatment that maximized the noise reduction while 7.2. Sound samples
minimizing the cost and weight was finalized and is described below:
Two sets of experiments were conducted. In the first set of experi-
6.1. Proposed noise control treatment ments, participants had to evaluate 5 s sound samples of the HVAC
noise recorded binaurally at the manufacturer specified reference po-
The final treatment applied to the baseline HVAC unit primarily sition of a passenger’s ear (see Fig. 4, position 1). Total 8 sounds were
consisted of lining recirculation inlet and other air flow paths with jute used for evaluation corresponding to the sounds of the baseline and the
felt, and stuffing the cavities with waste cotton. The detailed treatment treated HVAC unit running at the four different operating speeds.
However, testing the HVAC sounds without any background
soundscape is not sufficient, because in a real car the HVAC unit runs in
conjunction with other noise sources such as engine noise, road-tire
interaction noise, etc. Noises due to the mentioned sources reach ve-
hicle interior space after attenuation by the vehicle body acting as noise
barrier. Therefore, the second set of experiments were conducted. Here,
sound was recorded using binaural headphones placed in the ears of a
front passenger inside a closed BMW 3 series 320 l car, running at the
speed of 35 kph (≈9.72 ms−1) in the Institute campus with HVAC shut
off. A 5 s segment of this recording (53.5 dBA (58 dB)) (see Fig. 10) was
used as the background soundscape. Eight sounds samples were pre-
pared by adding each HVAC sound to this background soundscape by
Fig. 6. Sound intensity measurement for noise source ranking. superimposing their respective wavefronts in MATLAB.

104
S. Singh, A.R. Mohanty Applied Acoustics 140 (2018) 100–109

Table 5
Noise Source ranking of the HVAC unit with blower operating at 2958 rpm.
Rank Description of the source Figure Sound Intensity Level (dBA)

1 Recirculation inlet 80

2 Defrost vent 78

3 Cavities near the defrost vent 78

4 Other sources – ≤69

Table 6 eliminate the order effect. A repeated measures design was chosen for
Changes done as part of the proposed treatment. this experiment. Thus, every participant listened to all 8 sound pairs in
Component Treatment done a pre-determined sequence randomized by 8 × 8 Latin Square method.
This ensured there was no bias due to sequence effect.
Blower outlet Lined with 8 mm sound absorbing jute felt
Recirculation inlet Lined with 8 mm sound absorbing jute felt
Cabin vent Lined with 8 mm sound absorbing jute felt 7.4. Sound quality measures
Defrost vent Lined with 8 mm sound absorbing jute felt
Foot vent Lined with 8 mm sound absorbing jute felt Research shows that effect of HVAC noise exposure or in general
Flappers Lined with 8 mm sound absorbing jute felt
any environmental noise exposure on humans is usually expressed
Cavities near the defrost vent Sealed with waste cotton
through the semantics of “annoyance” and/or “pleasantness”
[3,7,8,26,27]. A pilot study was conducted asking people to rate HVAC
7.3. Experimental design sounds on these attributes. It was found that people found HVAC sounds
synonymous to a machine noise and related “annoyance” but not
This experiment aimed to test if a passenger/person would prefer “pleasantness” as its attribute. Hence, in the presented experiment,
HVAC sounds after the treatment over the baseline sounds and by how participants were asked to judge the “annoyance” due to the sounds
much. Therefore, a paired comparison listening evaluation was deemed from the HVAC unit.
more relevant and was used as the method for data collection.
Moreover, in a pilot evaluation, the paired comparison method was 7.5. Evaluation environment
found easier to use by non-expert assessors whereas, the usual method
of data collection, namely, semantic differential scales, was found dif- Experiments were conducted in an enclosed quiet room (back-
ficult to use and inappropriate for the task at hand. Table 8 lists the ground ≈24 dBA) within the Acoustics Laboratory at IIT Kharagpur.
sound pairs used for comparison. Within each participant every sound The sounds used for the experiments had been recorded binaurally at
pair was repeated twice in the reverse order to check reliability and the front passenger ear location with reference to the HVAC unit

8 mm jute felt waste cotton

Defrost vent Recirculation

air inlet vent
Cabin vent

8 mm jute felt
lining
Recirculation
air inlet vent
Line of the
Foot foot vent
vent
Recirculation
inlet
Defrost vent
Cabin vent

Blower

(a) (b)
Fig. 7. Detailed schematic of the proposed treatment. (a) Cut side view of the treated HVAC unit. (b) Cut top view of the treated HVAC unit.

105
S. Singh, A.R. Mohanty Applied Acoustics 140 (2018) 100–109

cavities stuffed
with waste cotton
jute felt lining

Fig. 8. Treatment done to the HVAC unit.

Table 7
Acoustic and psychoacoustic metrics of HVAC noise spectrum after treatment.
Sl. No. Operating condition Measurement channel SPL (dB) SPL (dBA) Loudness (sones) Sharpness (acums) Roughness (asper)

1 Low speed Left ear 62.8 59.4 13.7 1.42 1.7

Right ear 63.3 59.8 14.1 1.41 1.55

2 Medium-low speed Left ear 67.4 64.4 19.6 1.47 1.62

Right ear 68.0 64.7 19.9 1.47 1.59

3 Medium-high speed Left ear 70.0 67.3 23.7 1.51 1.41

Right ear 70.4 67.5 24.3 1.52 1.52

4 High speed Left ear 72.5 69.8 28.4 1.54 1.56

Right ear 73.0 70.1 29.0 1.56 1.46

baseline 60
Pressure (dB re 20μPa)

after treatment 50
70
40
Pressure (dB re 20μPa)

60
30
50
20
40
10
30
0
20
25

40

63

100

160

250

400

630

1.6k

2.5k

6.3k

10k

16k
1k

4k

10 Frequency (Hz)
0
Fig. 10. Spectrum of the car interior soundscape used for sound quality ex-
25

40

63

100

160

250

400

630

1k

4k
1.6k

2.5k

6.3k

10k

16k

periments.
Frequency (Hz)

Fig. 9. A typical HVAC noise spectrum after treatment at medium-high blower Table 8
speed. Sound pairs used for paired comparison.
HVAC operating condition HVAC sounds being compared
position. Thus, it was assumed that each listener for these experiments
Pair 1 Low speed Baseline Vs Treatment
was a front passenger. Accordingly, every participant sat on a chair and
Pair 2 Medium-low speed Baseline Vs Treatment
listened to the sounds through a binaural headphone. Pair 2 Medium-high speed Baseline Vs Treatment
Pair 4 High speed Baseline Vs Treatment

7.6. Procedure Pair 5 Low speed Treatment Vs Baseline

Pair 6 Medium-low speed Treatment Vs Baseline
Pair 7 Medium-high speed Treatment Vs Baseline
The sounds appeared in pairs on a computer screen (see Fig. 11). Pair 8 High speed Treatment Vs Baseline
Participants listened to the 8 sound pairs one by one via the binaural
headphone. For each pair, they selected the sound they found as “less
annoying”. A free-play paired comparison method [28] was used, i.e.,
participants could listen to the sounds in each pair as many times as

106
S. Singh, A.R. Mohanty Applied Acoustics 140 (2018) 100–109

Fig. 11. Listening evaluation of HVAC sounds.

Table 9
Percentage probability of preferring treated HVAC sounds over baseline HVAC sounds.
Condition Comparisons made with baseline Ntreated > base Nbase > treated Ptreated > base Pbase≥treated Critical difference (Δ95) Test of significance
(%) (%)

Low speed 48 44 4 91.7 8.3 16 Significant

Medium-low speed 48 34 14 70.8 29.2 16 Significant
Medium-high speed 48 34 14 70.8 29.2 16 Significant
High speed 48 42 6 87.5 12.5 16 Significant
Total 192 154 38 80.2 19.8 32 Significant

they wanted and in any order. This is because free play is more accurate 1 Pij
and user-friendly than the conventional fixed play (where participants Mi = ∑ ln ⎜⎛ ⎞⎟
N i≠j
P
⎝ ji ⎠ (2)
can listen to sounds only once in fixed order) [28]. Participants also had
an option to select “no significant difference”, if they found both sounds
in a pair similarly annoying. This ensured that the paired comparison
was not a forced choice task, and only a significant difference in an- In the Bradley-Terry model, the hypothesis that “one sound is sig-
noyance was recorded. The full procedure was repeated for the two sets nificantly better than the other” can be tested at a 95% confidence
of HVAC sound samples (without/ with car interior background interval by calculating the critical difference (Δ95) as follows [30]:
soundscape) at a gap of two months.
NT 0.5
Δ95 = 1.64 ⎛ ⎞ + 0.5
7.7. Analysis ⎝ 2 ⎠ (3)
here T is the total number of comparisons made, which is, the total
In paired comparison method, the quantitative scores for every number of listeners multiplied by the number of times a sound pair
sound sample is calculated using the Bradley-Terry model [29,30]. The appears within each listener evaluation. The test of significant differ-
sound scores are represented by the following terminologies: ence is done as follows [30]:

• Pair probabilities (P ): It is the probability of listeners to pick sound i

ij ⎧ If Ni−Nj > Δ95 ⇒ "“"Sound i is significantly better than sound j"”"
⎪
⎫
⎪
over sound j. If Ni−Nj ⩽ Δ95
⎨ ⎬
No. of times sound i is preferred over sound j ⎪ ⇒ "“"No significant difference between sound i and sound j"”" ⎪
Pij = ⎩ ⎭
No. of times sound i is compared with sound j (1)
(4)
where Ni is the ‘preference count’ for sound i, i.e., the number of times
• Normalized merit scores (M ): The normalized merit score for a sound
i sound i is chosen in total T comparisons; and Nj is the preference count
is calculated using Eq. (2) where, N is the number of sounds being for sound j.
compared [28]. The sum of merit scores of all sounds is zero. The
sounds with negative merit scores are rejected and sounds with
7.8. Results of sound quality evaluation
positive merit scores are selected.
Tables 9 and 10, Figs. 12 and 13 show the percentage pair

Table 10
Percentage probability of preferring treated HVAC sounds over baseline HVAC sounds in the presence of car interior soundscape.
Condition Comparisons made with baseline Ntreated > base Nbase > treated Ptreated > base Pbase≥treated Critical difference (Δ95) Test of significance
(%) (%)

Low speed 48 42 6 87.5 12.5 16 Significant

Medium-low speed 48 34 14 70.8 29.2 16 Significant
Medium-high speed 48 36 12 75 25 16 Significant
High speed 48 42 6 87.5 12.5 16 Significant
Total 192 154 38 80.2 19.8 32 Significant

107
S. Singh, A.R. Mohanty Applied Acoustics 140 (2018) 100–109

HVAC Sound after Treatment

Baseline HVAC Sound Total

Percentage of sound preference

100 1.2 low speed
med-low speed
80 0.8 med-high speed
high speed

Merit Scores
0.4
60

0.0
40 baseline after treatment
-0.4
20
-0.8
0
low med-low med-high high Total -1.2
Operating Condition HVAC Sounds
(a) (b)
Fig. 12. (a) Percentage of preferring an HVAC sound, treatment Vs baseline. (b) Merit Scores of different HVAC sounds.

probabilities for preferring treated HVAC sounds over the baseline were tested to have higher sound absorption coefficients than many of
HVAC sounds, and the merit scores for the baseline and treated HVAC the existing low-cost sound absorbing natural materials. The major
sounds for every operating condition and overall. Tables 9 and 10 also noise source of the HVAC unit was the aerodynamic flow that was
show the result of test of significant difference between baseline and the noisiest at the recirculation inlet. Therefore, a proposed natural mate-
treated HVAC sounds. It is found that a very high percentage of the rial-based noise control technique involves lining jute felts along the
listeners (80.2%) preferred the HVAC sounds after the treatment over recirculation inlet vent, outlet vents, and other locations in the path of
the baseline HVAC sounds, with as well as without the presence of the the air flow. Moreover, cavities unassociated with air flow path should
car interior soundscape. Moreover, the listeners’ preferences of the be sealed with waste cotton.
treated HVAC sounds over baseline sounds are statistically significant This natural material-based treatment was able to reduce HVAC
for every operating condition and overall. Thus, the applied treatment noise level at the passenger ear location by an average of 4 dBA at all
improved the sound quality of a vehicle interior space by causing a blower speeds. This reduction can be called significant as the minimum
significant reduction in the annoyance of the HVAC sound, and also in noise level reduction perceptible by a human ear is 2 dB, and every
the annoyance of the overall vehicle interior soundscape (comprising 3 dBA noise reduction is said to double the permissible human noise
that HVAC sound). This is also supported by the merit scores data exposure duration. In the human-subject sound quality experiments,
where the treated HVAC sounds score positive and the baseline sounds 80.2% people preferred a typical car interior soundscape with the
score negative for every operating condition and overall. treated HVAC sounds over the soundscape with the baseline HVAC
sounds. Both, the treated HVAC sounds and the car interior soundscape
comprising these treated sounds had statistically significant lower an-
8. General discussions and future recommendations noyance (95% confidence interval) at all blower speeds. From these
results, it can be concluded that the treatment has a potential to im-
This case study demonstrates the potential of jute felt and waste prove sound quality of a vehicle interior space. This is further supported
cotton as light-weight, low-cost, biodegradable and recyclable natural by psychoacoustic analyses showing a loudness level reduction of 7
materials for vehicle HVAC noise control. Jute felt and waste cotton

HVAC Sound after Treatment

Baseline HVAC Sound Total
Percentage of sound preference

100
1.2 low speed
med-low speed
80 0.8 med-high speed
high speed
Merit Scores

60 0.4

0.0
40 baseline after treatment
-0.4
20
-0.8
0
low med-low med-high high Total
-1.2
Operating Condition HVAC Sounds
(a) (b)
Fig. 13. (a) Percentage of preferring an HVAC sound, treatment Vs baseline, in the presence of car interior soundscape. (b) Merit Scores of different HVAC sounds in
the presence of car interior soundscape.

108
S. Singh, A.R. Mohanty
sones, and previous studies find direct correlation between loudness
level and annoyance/pleasantness of an HVAC unit [3,7,8].
The current study was conducted on a ductless HVAC prototype
isolated from the rest of the vehicle. The measurements of noise level
and psychoacoustic metrics would change when the HVAC unit is in-
stalled inside the car, but the general trend of noise reduction and
perception of sound quality should remain the same.
9. Conclusions
The paper presents a case study of first ever usage of natural ma-
terials for a vehicle’s HVAC noise control. Jute felt and waste cotton are
shown to have higher sound absorption coefficients than other con-
temporary sound absorbing natural materials, and are chosen for noise
control treatment. The major noise source of the HVAC unit is the
aerodynamic flow that is noisiest at the recirculation inlet. Thus, pro-
posed noise control treatment involves reducing the aerodynamic noise
by lining recirculation inlet vent, outlet vents and other air flow paths
with sound absorbing jute felt and sealing unwanted cavities with waste
cotton. This treatment is very effective as it significantly reduces the
SPL (by 4 dBA) and loudness level (by 7 sones), with only negligible
weight (3% of total unit’s weight) and cost ($0.50), and without re-
ducing the HVAC air flow. Moreover, the treatment results in a sig-
nificantly lesser annoyance and better perceived sound quality of the
vehicle interior space. Thus, jute felt and waste cotton are low-cost,
light-weight, biodegradable and recyclable natural materials with high
potential for HVAC noise control.
Acknowledgements
We would like to thank the students and staff at the Mechanical
Engineering Department, IIT Kharagpur for participating in our sound
quality experiments. We also thank Mr. Manish Kumar Singh for his
suggestions on the experimental design of the sound quality experi-
ment.
References
[1] Jennings PA, Dunne G, Williams R, Giudice S. Tools and techniques for under-
standing the fundamentals of automotive sound quality. Proc Inst Mech Eng Part D J
Automob Eng 2010;224(10):1263–78.
[2] Singh S, Payne SR, Jennings PA. Toward a methodology for assessing electric ve-
hicle exterior sounds. IEEE Trans Intell Transp Syst 2014;15(4):1790–800.
[3] Leite RP, Paul S, Gerges SNY. A sound quality-based investigation of the HVAC
system noise of an automobile model. Appl Acoust Apr. 2009;70(4):636–45.
[4] Devineni N, Panahi I, Kasbekar P. Predictive multi-channel Feedback active noise
control for HVAC systems. In: IEEE International Conference on Electro Information
Technology, 2011.
109
Applied Acoustics 140 (2018) 100–109
[5] Allam S, Åbom M. Fan noise control using microperforated splitter silencers. J Vib
Acoust 2014;136(3):31017.
[6] Allam S, Abom M. Noise control for cooling fans on heavy vehicles. Noise Control
Eng J 2012;60(6):707–15.
[7] Yoon J-H, Yang I-H, Jeong J-E, Park S-G, Oh J-E. Reliability improvement of a sound
quality index for a vehicle HVAC system using a regression and neural network
model. Appl Acoust 2012;73(11):1099–103.
[8] Minard A, Lambourg C, Boussard P. Perceptual evaluation of the sound quality of
car HVAC systems. In: 45th IINTER-NOISE 2016;2016(15):4467–474.
[9] Neise W. Noise reduction in centrifugal fans: a literature survey. J Sound Vib
1976;45(3):375–403.
[10] Neise W. Review of noise reduction methods for centrifugal fans. J Eng Ind
1982;104(2):151–61.
[11] Wu J-D, Bai MR. Application of feed forward adaptive active-noise control for re-
ducing blade passing noise in centrifugal fans. J Sound Vib 2001;239(5):1051–62.
[12] Velarde-Suárez S, Ballesteros-Tajadura R, Pablo Hurtado-Cruz J, Santolaria-Morros
C. Experimental determination of the tonal noise sources in a centrifugal fan. J
Sound Vib 2006;295(3–5):781–96.
[13] Envia E. Fan noise reduction: an overview. Int J Aeroacoustics 2002;1(1):43–64.
[14] Choi M. Noise attenuating device for a heating-ventilation-cooling system of a
motor vehicle, 2012.
[15] Yang IH, Kwon OC, Lee JY, Oh J-E., Sound quality evaluation for the vehicle HVAC
system after active noise control, in: 2008 ASME International Mechanical
Engineering Congress and Exposition, 2008:603–07.
[16] Arenas JP, Crocker MJ. Recent trends in porous sound-absorbing materials. Sound
Vib 2010;44(7):12–7.
[17] Asdrubali F, Schiavoni S, Horoshenkov KV. A review of sustainable materials for
acoustic applications. Build Acoust 2012;19(4).
[18] Parikh DV, Chen Y, Sun L. Reducing automotive interior noise with natural fiber
nonwoven floor covering systems. Text Res J 2006;76(11):813–20.
[19] Fatima S, Mohanty AR. Noise control of home appliances – the green way. Noise Vib
Worldw 2012;43(7):26–34.
[20] Zulkifli R, Zulkarnain, Mohd Nor MJ. Noise control using coconut coir fiber sound
absorber with porous layer backing and perforated panel. Am J Appl Sci
2010;7(2):260–4.
[21] Thilagavathi G, Pradeep E, Kannaian T, Sasikala L. Development of natural fiber
nonwovens for application as car interiors for noise control. J Ind Text
2010;39(3):267–78.
[22] Fatima S, Mohanty AR. Acoustical and fire-retardant properties of jute composite
materials. Appl Acoust 2011;72(2):108–14.
[23] Bansod PV, Mittal T, Mohanty AR. Study on the acoustical properties of natural jute
material by theoretical and experimental methods for building acoustics applica-
tions. Acoust Aust 2016;44(3):457–72.
[24] Singh S, Mohanty AR. Measurement of boiling liquid levels by decomposition of
sound waves in a waveguide. Appl Acoust 2018;129:248–57.
[25] ISO 10534-2, Acoustics-determination of sound absorption coefficient and im-
pedance in impedance tubes—Part 2: transfer-function method. 2009.
[26] Stallen PJM. A theoretical framework for environmental noise annoyance. Noise
Health 1999;1(3):69–79.
[27] Brown AL, Kang J, Gjestland T. Towards standardization in soundscape preference
assessment. Appl Acoust 2011;72(6):387–92.
[28] Baker S, Jennings P, Dunne G, Williams R. Improving the effectiveness of paired
comparison tests for automotive sound quality. In: Proceedings of the 11th
International Congress on Sound and Vibration 2004:5–8.
[29] Otto N, Amman S, Eaton C, Lake S. Guidelines for Jury Evaluations of Automotive
Sounds, SAE Tech. Pap. 1999-01-1822, 1999.
[30] David HA. The method of paired comparisons Vol. 12. London: Charles Griffin &
Co., Ltd.; 1963.