Acoustic comfort evaluation for hypermarket workers

Building and Environment 59 (2013) 369e378

Contents lists available at SciVerse ScienceDirect

Building and Environment journal homepage: www.elsevier.com/locate/buildenv

Acoustic comfort evaluation for hypermarket workers Sabrina Della Crociata*, Antonio Simone, Francesco Martellotta Dipartimento di Scienze dell’Ingegneria Civile e dell’Architettura, Politecnico di Bari, via Orabona 4, I-70125 Bari, Italy

a r t i c l e i n f o

a b s t r a c t

Article history: Received 2 July 2012 Received in revised form 6 September 2012 Accepted 7 September 2012

The paper presents the results of an on-site measurement procedure to assess acoustic comfort for hypermarket workers. The assessment is based on the collection of both objective measurements of environmental parameters as well as subjective ratings of acoustic comfort. The study was carried out in a hypermarket located in Southern Italy, in which four sub-spaces having similar acoustic characteristics were identified. Workers were asked to move to selected measuring points and fill in a questionnaire to rate their subjective perceptions. Measuring points were chosen to represent the typical conditions inside the hypermarket. Factor analysis and linear regression analysis were used to pick up, among the many noise indexes available in the literature, those that could better describe the staff subjective attitude towards the acoustic environment. Finally, by comparisons with subjective “comfort” thresholds, optimal intervals for selected parameters were defined. Analysis showed that the A-weighted equivalent sound pressure level LeqA and the percentile level LA90 could be used to describe subjective auditory sensations in most situations. Ó 2012 Elsevier Ltd. All rights reserved.

Keywords: Acoustic comfort Subjective ratings Noise indexes Hypermarket workers

1. Introduction The concept of “health”, according to the WHO, is the result of psycho-physical and social well-being, rather than simply “lack of illness”. It is extended to include global comfort, which depends on several factors such as thermal environment, indoor air quality, visual and acoustic conditions, and is difficult to characterize. Acoustic comfort evaluation is very important in office-like workplaces, in which risk of hearing impairment is often low but annoyance induced by noise noticeably influences subjective health. Moreover, environmental noise causes mental stress and loss of concentration, which can adversely affect worker’s performance [1]. By means of collection of both objective measurements of environmental parameters and subjective perceptions, a lot of research has been carried out to study disturbance due to office noise [2e10], the effect of acoustic, thermal and visual environments on indoor comfort in offices [11], and acoustic comfort in urban open public spaces and in noisy and quiet roads [12e16]. Acoustic comfort in schools is very important because noise, in combination with thermal and visual comfort, causes annoyance, and impairs communication, learning and concentration [17,18]. Only a few researches studied acoustic comfort in large-scale retail

* Corresponding author. Tel.: þ39 080 5963631; fax: þ39 080 5963419. E-mail address: [email protected] (S. Della Crociata). 0360-1323/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.buildenv.2012.09.002

trade buildings, investigating in particular the acoustics of large public spaces, like atrium spaces and shopping arcades [19]. Large-scale retail trade buildings include supermarkets, hypermarkets, department stores and shopping malls in ascending order of complexity and dimension. They are characterized by a huge range of products and their environment is not homogeneous, consisting of several sub-spaces with different thermal, acoustic and visual conditions which have to be studied individually. The spread of hypermarkets at the expense of small trade retailing means that greater attention must be paid towards the assessment of comfort conditions in these commercial buildings both for customers and staff. In fact, fixed workplaces for the staff and longer stays for the customers require a greater attention to the global comfort conditions. Hypermarkets’ acoustic environment depends on a lot of external and internal acoustic sources, work activities, as well as on customers’ confluence. To evaluate hypermarket’s acoustic comfort thoroughly, objective measurements of environmental parameters were collected together with workers subjective perceptions (as they are regular users and know very well their work environment). The aim of this study is to investigate acoustic comfort conditions for workers in hypermarkets, in particular in different acoustic environments, identifying, among many noise indexes found in literature, those that could better describe the staff subjective attitude towards the acoustic environment and that, by comparison with suitable “comfort” thresholds, might ensure an acoustically satisfactory workplace.

370

S. Della Crociata et al. / Building and Environment 59 (2013) 369e378

2. Method

customary workplace. The questionnaire was anonymous and it was filled in avoiding any influence from members of the survey team, co-workers, or superiors. Most employees have dynamic jobs and they move between many different areas, so it is hard to define a “reference” fixed workplace. Moreover, as the research included thermal environment measurements, using instruments that are characterized by a slow response time, monitoring and evaluation points were kept fixed (Fig. 1). Their location was decided taking into account the hypermarket layout, the nature and organization of each job and the need for characterizing every workplace [21]. For acoustic measurements, a sound level meter with real-time frequency analyzer and digital recorder was used (0.1 dB resolution and 0.7 dB accuracy). The random incidence microphone was placed at a distance of 1.0 m from any reflective surface and 1.5 m above the floor. Before and after every measurement session the microphone was calibrated with a 94 dB acoustic calibrator. Measurements and sound recordings were taken continuously, allowing the determination of Aweighted equivalent continuous sound level (LeqA), C-weighted peak sound pressure level (LCpk), time-history in one-third-octave bands from 20 Hz to 20 kHz, and A-weighted statistical levels (LA99, LA90, LA50, LA10, LA5, LA1). In order to correlate objective parameters with subjective responses, acoustic measurements were averaged considering the 10-min interval before the end of questionnaire filling [9].

2.1. Building description The acoustical environmental survey was part of a research [20] carried out in a hypermarket located in Bari, a city in Southern Italy facing the Adriatic Sea. The shopping center is located in a suburban area, it is situated at road level and consists of a shopping arcade and a hypermarket. The arrangement of different goods on sale is outlined in Fig. 1. The total area was divided into four acoustically homogeneous sub-spaces: warehouse, “quiet” sales area (which includes optician and pharmacy, personal healthcare products, clothing and foot-wear, stationary, TV/video, household appliances and utensils, sports wear), “noisy” sales area (which includes beverages, nonperishable goods, refrigerated decks, fishmonger’s, fruit and vegetables divisions, the frozen food section, full and limited service divisions like butcher’s, delicatessen, bakery and confectioner’s, home cleaning products, brico and seasonal), and checkouts. The checkouts lead into the shopping arcade and hence receive noise from the many cafes, restaurants, non-food stores located nearby. 2.2. Data acquisition Data were collected from February 2010 to July 2011. Continuous physical measurements were carried out on working days, for time intervals from one to 5 h at each measurement point. At the same time, employees were randomly invited to answer a questionnaire, while they were occupied in their work, moving to the measurement station closer to their usual workplace. Anyway, they were asked to answer considering their subjective sensations when and where the questionnaire was filled in, even if it was not their 1 A

2

3

4

5

6

7

8

9

2.3. Noise indexes A large number of noise indexes have been defined by many authors in order to describe various aspects of community noise. From time to time, and in varying contexts, each noise index proved to be appropriate to describe subjective auditory sensations of perceived intensity and annoyance, noise sources and environmental 10

11

12

13

14

15 A

N

B

B

C

C

D

D

E

E

F

F

G

G

H

H

I

I

L

L

M

M

N

N

O

O

Warehouse

“Quiet” sales area

“Noisy” sales area

Checkouts

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

Fig. 1. Hypermarket plan with an alphanumeric reference system to locate surveyed points quickly. Bold squares represent surveyed points. The reference system unit is 10 m.

S. Della Crociata et al. / Building and Environment 59 (2013) 369e378

characteristics. A-weighted equivalent sound pressure level (LeqA) is probably the most widespread noise index, it is simple to measure and correlates well with many psycho-physiologic effects of noise. However, probably because of the complexity of the acoustic environment, combined with the characteristics of the sample (and its dimensions), a preliminary analysis [20] proved that LeqA was scarcely correlated with the subjective perception of the employees. So, in order to investigate whether different descriptors might perform better than LeqA, in this study most of the noise indexes found in the literature were determined. Linear equivalent sound pressure level (Leqlin), A-weighted statistical levels (LA90, LA10, LA5), noise climate (LA10  LA90), noise pollution level (LNP), Zwicker’s loudness level (LLZ), Stevens’ loudness level (LLS), noise criterion (NC), preferred noise criterion (PNC), noise rating (NR), room criterion (RC), balanced noise criterion (NCB), room criterion mark II (RCII), quality assessment index (QAI), office noise index (ONI), combined noise index (CNI), speech interference level LSIL and PSIL were all determined from measured spectra. Detailed description of each of the above mentioned parameters may be found elsewhere [3,5]. 2.4. Questionnaires Questionnaires are considered a more effective means to analyse the comfort parameters than using only unsolicited complaints [22]. The questionnaire layout and questions are based largely on the survey form developed in the HOPE project [23] and on ISO 10551:1995 standard [24]. In order to match the subjective answers to physical measurements, date and time of interview were recorded as well as the name of the surveyed point in the reference system. The survey begins with a section concerning socio-anagraphic data of respondents (gender, age and level of education) and general information about their work (duty or status, entry-time, stay-time near the surveyed point and if this is a customary workplace). Such factors generally showed negligible influence on subjective response to noise in most large-scale surveys [5e7,10]. However, they proved to be of some relevance in studies in which complex sound environments were analysed using the soundscape approach [14e16]. In particular, age and education level seem to play a role in influencing the subjective preference in urban open spaces, so it will be interesting to investigate if similar correlations exist in the space under investigation. A three-point self-evaluation scale about hearing ability, provides additional information that may be useful to explain abnormal responses in the subsequent analysis. The need to keep the format as uniform as possible throughout the whole questionnaire sections implied the use of questions that differed slightly from those used, and validated, in previous studies [2,3,5]. In fact, acoustic comfort was investigated (Table 1) by means of seven-point bipolar scale questions with semantic descriptors at the extremes and numeric values for each scale point. Perceived (API) and desired (ADI) sound intensity was rated using scale from “very weak” to “very loud”, while “acoustic satisfaction” (ASat) was rated using a scale from “very satisfied” to “very unsatisfied”. The typical annoyance scale was not used because of the complexity of the acoustic environment, and because a satisfaction scale was thought more suitable to assess all the comfort aspects in a similar way [20]. Furthermore, previous studies [2] showed good correlations between annoyance Table 1 Acoustic comfort evaluation scales. Acoustic API ADI ASat

Very quiet Very quiet Very unsatisfied

1 1 1

2 2 2

3 3 3

4 4 4

5 5 5

6 6 6

7 7 7

Very loud Very loud Very satisfied

371

and dissatisfaction. Possible sources of acoustic discomfort were studied through a multiple choice question, with an open field to indicate other causes not previously listed. As one of the most important effects of acoustic discomfort is the difficulty to understand spoken messages, a dichotomous question about speech intelligibility was specifically included. A total of 610 questionnaires were collected in the hypermarket, equally distributed between the four sub-spaces and among the different seasons.

2.5. Noise environment The four sub-spaces into which the hypermarket was divided, have different acoustic conditions because of different noise sources. The plot of the average, minimum, and maximum onethird-octave spectra of typical hypermarket noise in the four sub-spaces is shown in Fig. 2. For these sub-spaces, the statistical distribution of levels are shown in Fig. 3a pointing out, in particular, LA10 and LA90 (i.e. the noise levels that are exceeded for 10% and 90% of each sample period), that conventionally represent peak and background noise levels respectively. The warehouse showed the larger fluctuations, and the background noise was about 52 dB due to the presence of background music having a broad, flat spectrum (Fig. 2a). Periodical peaks appeared at 71 dB due to transpallets unloading goods, especially early in the morning. The “quiet” sales area includes sales divisions sparsely attended by customers. In this area acoustic environment was characterized by background music (LA90 is about 54 dB) and use of transpallets coming from warehouse (Fig. 2b). Transpallet movements are less frequent than in warehouse and, even though maximum one-third-octave spectra were similar in both areas, LA10 resulted in about 65 dB (Fig. 3a). The peaks were usually caused by customers passing by with metal shopping trolleys and by voices. In the “noisy” sales area there were lots of noise sources like refrigerated decks, food preparation and packaging machines, people voices, paging announcements by full service counters, which together produced a higher background noise (LA90 is about 65 dB) and lower fluctuations (LA10 is about 73 dB) than in warehouse and in “quiet” sales area (Fig. 3a). It is interesting to notice that the maximum one-third-octave spectrum was similar in both the “quiet” and the “noisy” sales area (Fig. 2c), being determined by the same causes. In the “noisy” sales area the average spectrum also showed a low frequency tone at 50 Hz and 100 Hz due to electric motors of the different appliances. At the checkouts there were lots of noise sources like customers passing with metal shopping trolleys, plastic shopping trolleys and their telescopic handles slamming, “beeps” due to barcode readers, background music, people voices, packaging and paging messages (Fig. 2d). Moreover, the checkout position is close to the shopping arcade, therefore receiving noise and music from the nearby cafes, restaurants, and non-food stores. These noise sources are often superimposed and determined a higher background noise (LA90 is about 66 dB) and lower peaks (LA10 is about 72 dB), exposing checkouts workers to higher levels for more time. Noise criterion curves NCB, RC and RC Mark II can give interesting information about the shape of the sound spectrum, even though they should ideally be used only to evaluate HVAC noise in unoccupied rooms. Most of the spectra were “hissy” (100% RC Mark II, 81% NCB, 49% RC) because of background music and people voices, which are characterized by mid and high frequencies. Moreover many high-frequency sounds appeared at the checkouts: “beeps” due to barcode readers (showing peaks between 1 kHz and 1.6 kHz) (Fig. 2d), telescopic handles of shopping trolley slamming, and customers passing by with metal shopping trolleys.

20

0

10

d

max mean min

90

0

max mean min

100

60

50

50

40

40

30

30

20

20

10

10

0

0

100 Hz 125 Hz 160 Hz 200 Hz 250 Hz 315 Hz 400 Hz 500 Hz 630 Hz 800 Hz 1.00 kHz 1.25 kHz 1.60 kHz 2.00 kHz 2.50 kHz 3.15 kHz 4.00 kHz 5.00 kHz 6.30 kHz 8.00 kHz 10.0 kHz 12.5 kHz 16.0 kHz 20.0 kHz

70

60

20.0 Hz 25.0 Hz 31.5 Hz 40.0 Hz 50.0 Hz 63.0 Hz 80.0 Hz

80

70

20.0 Hz 25.0 Hz 31.5 Hz 40.0 Hz 50.0 Hz 63.0 Hz 80.0 Hz

90

80

SUM(A)

30

10

SUM(LIN)

40

20

SUM(A)

50

30

SUM(LIN)

60

40

SUM(A)

70

50

SUM(LIN)

80

60

100 Hz 125 Hz 160 Hz 200 Hz 250 Hz 315 Hz 400 Hz 500 Hz 630 Hz 800 Hz 1.00 kHz 1.25 kHz 1.60 kHz 2.00 kHz 2.50 kHz 3.15 kHz 4.00 kHz 5.00 kHz 6.30 kHz 8.00 kHz 10.0 kHz 12.5 kHz 16.0 kHz 20.0 kHz

70

20.0 Hz 25.0 Hz 31.5 Hz 40.0 Hz 50.0 Hz 63.0 Hz 80.0 Hz

80

100

max mean min

100 90

SUM(A)

c

b

100 Hz 125 Hz 160 Hz 200 Hz 250 Hz 315 Hz 400 Hz 500 Hz 630 Hz 800 Hz 1.00 kHz 1.25 kHz 1.60 kHz 2.00 kHz 2.50 kHz 3.15 kHz 4.00 kHz 5.00 kHz 6.30 kHz 8.00 kHz 10.0 kHz 12.5 kHz 16.0 kHz 20.0 kHz

90

100 Hz 125 Hz 160 Hz 200 Hz 250 Hz 315 Hz 400 Hz 500 Hz 630 Hz 800 Hz 1.00 kHz 1.25 kHz 1.60 kHz 2.00 kHz 2.50 kHz 3.15 kHz 4.00 kHz 5.00 kHz 6.30 kHz 8.00 kHz 10.0 kHz 12.5 kHz 16.0 kHz 20.0 kHz

max mean min

100

SUM(LIN)

a

S. Della Crociata et al. / Building and Environment 59 (2013) 369e378

20.0 Hz 25.0 Hz 31.5 Hz 40.0 Hz 50.0 Hz 63.0 Hz 80.0 Hz

372

Fig. 2. Plot of the average, minimum and maximum one-third-octave spectra of typical hypermarket noise in warehouse (a), “quiet” sales area (b), “noisy” sales area (c), checkouts (d).

3. Results and analysis 3.1. Subjective evaluations Acoustic comfort was evaluated by means of perceived and desired intensity, acoustic satisfaction, speech intelligibility, and considering any possible cause of acoustic discomfort. A small percentage of respondents (about 3%) judged their hearing ability to be “not good”, therefore the corresponding questionnaires were discarded in order to avoid any possible bias in the statistical analysis, because any hearing impairment could have modified frequencies and sound levels perception. Acoustic comfort is a complex aspect to evaluate. In fact, the presence of a pleasant sound could considerably improve acoustic comfort, even if its sound level is rather high [12]. Therefore perceived intensity needs to be combined with subjective acoustic satisfaction (which is affected by job nature and sound source type) to provide a comprehensive picture of acoustic comfort. The correlation between API and ASat was first investigated by averaging the ASat votes corresponding to each API scale point. High and statistically significant correlations appeared in the whole hypermarket (Fig. 4 a) and in the four sub-spaces (Warehouse R2 ¼ 0.86; “quiet” sales area R2 ¼ 0.90; “noisy” sales area R2 ¼ 0.77; checkouts R2 ¼ 0.82). Regression line showed that, on average, subjects need a variation of two points on the API scale to determine a one-point variation on the ASat scale. To better understand subjective acoustic sensation in each of the four sub-spaces, ASat votes were plotted as a function of API without any averaging (Fig. 4bef). In the four sub-spaces most of the workers expressed API judgments between 2 and 6, while only few workers used the extreme ratings. Only in the quiet warehouse (Fig. 4c), low API ratings corresponded to a significant number of high ASat votes.

Conversely, in the other three areas API votes were more scattered over the whole interval, probably because the acoustic satisfaction was influenced by other factors than just the sound level perceived. In order to investigate this aspect in greater detail, the possible sources of acoustic discomfort were analysed. ASat ratings were consistent with the frequency of complaints about acoustic discomfort sources (Table 2). This suggested that acoustic satisfaction might be influenced by sources of acoustic annoyance [12]. The histogram in Fig. 5 shows that the frequency distribution of ASat responses for employees who indicate at least one cause of acoustic discomfort was shifted towards the lower end of the scale, suggesting higher dissatisfaction rates compared to employees indicating no specific cause of acoustic discomfort. Hypermarkets are characterized by many sound sources, such as HVAC, refrigerated decks, frozen food section, background music, voices, packaging and paging system. Table 3 shows the distribution of annoyance causes referred to the whole sample and to each sub-space. For the whole area the most annoying sources were background music, transpallets and voices. The percentages change slightly in the four areas because of different acoustic environment and specific sources, but they were consistent with objective analysis performed above (Section 2.5). In the warehouse and in the “quiet” sales area workers were mostly annoyed by frequent use of transpallets and by background music. In the “noisy” sales area refrigerators and people were the most annoying acoustic sources. At the checkouts, voices and music coming from both hypermarket and shopping arcade were the most annoying noise sources. It is interesting to observe that music, which was supposed to increase the pleasantness of the environment (which proved to be the most important factor affecting subjective responses in urban spaces [15]), seems to show an opposite effect in hypermarkets (at least for workers). In particular, in the checkout area, the superposition of

S. Della Crociata et al. / Building and Environment 59 (2013) 369e378

a

373

100 Warehouse

90

“Quiet” sales area “Noisy” sales area

80

Checkouts

%

70 60 50 40 30 20 10 0 30

35

40

45

50

55

60

65

70

75

80

85

90

95

720

780

840

dB

b 90

Warehouse Checkouts

85 80

LeqA

75 70 65 60 55 50 45 40 0

60

120

180 8

240

300

360

420

480

540

600

660

900

t (s) Fig. 3. Acoustic distribution functions of typical noise for each of the four sub-spaces, horizontal lines identify percentile levels LA10 and LA90 (a) e a typical time-history measured over a period of 15 min in warehouse and checkouts (b).

music coming from different areas contributed to increase annoyance ratings. Lack of speech intelligibility (due to specific sound sources) might be related to a lower ASat rating. Therefore the percentage of the employees complaining for a lack of speech intelligibility was calculated for those indicating at least one cause of acoustic discomfort and for the others. In the first case 35% of subjects complained for at least one source of acoustic discomfort and for a lack of intelligibility, while the percentage dropped to 5% for the other group, confirming the expected dependence. Finally, correlations between subjective responses to noise and social, demographical and behavioural aspects, were taken into account and some statistically significant differences were found in the responses given by different groups. However, apparently “strange” correlations appeared between age, gender, and job groups. A more detailed analysis showed that some of the jobs were strongly characterized in terms of age and gender. For example, cashiers were women and aged 20e50, while storekeepers were men and mostly under 30 years old. For this reason, considering only social and demographical aspects, it was difficult to understand on which of the factors the observed differences could depend. Anyway, it was interesting to observe that only acoustic satisfaction showed variations as a function of age, gender, educational level, and if the workplace was customary or not (p-

value  0.001). Acoustic perceived level was much less influenced by the above factors. 3.2. Factor analysis between noise indexes As stated above, several objective parameters have been used to investigate possible correlations with subjective perceptions in terms of acoustic satisfaction and acoustic perceived intensity. However, considering that only the criterion numbers were used for NCB, RC and RCII indexes, and that there are many similarities among the different parameters, some sort of data reduction analysis was particularly useful to avoid unnecessary duplications. In fact, correlation analysis between the selected parameters showed that most of noise indexes were well correlated with each other (Appendix A). A factor analysis was then made to reduce the number of significant parameters without affecting their ability to describe different aspect of the sound field (Fig. 6). In factor analysis the number of factors is the number of independent (and uncorrelated) patterns of relationship existing between the variables, each one explaining a part of the whole variance. In this case about 96% of the total variance was explained by the first and the second factor. In particular the first one explained 78% of the variance and was related to sound levels, peaks and subjective loudness, while the second one explained 18%

374

S. Della Crociata et al. / Building and Environment 59 (2013) 369e378

b

a7

7 6

6

ASat

ASat

5

R² = 0.92

5 4

4

3

3

2

2

1 1

2

3

4

5

6

1

7

1

2

3

1

2

3

API

d

6

6

5

5

7

5

6

7

4

4 3

3

2

2 1

1

2

3

4 API

5

6

7

f

7 6

6

5

5

4

4 API

7

ASat

ASat

6

7

1

e

5

ASat

ASat

c7

4 API

4

3

3

2

2 1

1 1

2

3

4 API

5

6

7

1

2

3

4 API

5

6

7

Fig. 4. ASat mean votes as a function of API in the whole hypermarket (a), and ASat votes distributions as a function of API in the whole hypermarket (b), in the warehouse (c), in the “quiet” sales area (d), in the “noisy” sales area (e), in the checkouts (f). Marker area is proportional to the number of questionnaires.

Table 2 Absolute and relative frequency of employees who indicated at least one cause of acoustic discomfort as a function of given ASat vote. Acoustic satisfaction

N

%

1 2 3 4 5 6 7

19 31 102 91 34 34 10

90 89 82 45 41 41 22

of the variance and was related to a mix of independent aspects including background noise, fluctuations and spectral imbalance. At this point, any acoustic parameter well correlated with one of the factors might be used to describe it. So, even though LeqA was poorly correlated with subjective perceptions in the preliminary analysis [20], its widespread usage supported its selection to represent the first factor. Background noise level LA90, the fluctuations LA10  LA90, and the quality assessment index QAI were selected to represent the second factor. In this way the subsequent discussion should lead to clearer results. 3.3. Correlations between objective and subjective parameters In the whole hypermarket and in the four sub-spaces, correlations between objective parameters, selected by factor analysis, and

S. Della Crociata et al. / Building and Environment 59 (2013) 369e378

no cause of acoustic discomfort

50%

375

1,0 LNP

LA10-LA90

ONI

at least one cause of acoustic discomfort

QAI

40% 0,5

Factor2

30%

LA CNI5 0,0

NR LSIL CB valore NSUM(A) RRC CII alore LA 10 PSIL valore Lstevens NC

20%

Lzwicker PNC

Leqlin

10%

- 0,5 LA 90

0% 1

2

3 4 5 Acoustic satisfaction

6

- 1,0

7

- 0,2

0,0

0,2

Table 4 Summary of the performance of noise indexes in correlation with human sensation of dissatisfaction and loudness in hypermarket, warehouse, “quiet” sales area, “noisy” sales area, checkouts. p-Value <0.05 significant *, p-value <0.01 highly significant **, p-value <0.001 extremely significant ***. Slope of regression lines of ASat and API. “Comfort” thresholds. R2 ASat

Slope ASat

“Comfort” threshold

R2 API

Slope API

Global data

LeqA LA90 LA10  LA90 QAI

0.45 0.40 0.31 0.05

*** *** *** ***

0.05 0.03 0.03 0.02

68.9 63.8 1.7 37.9

0.66 0.52 0.43 0.05

*** *** *** ***

0.07 0.04 0.04 0.02

Warehouse

LeqA LA90 LA10  LA90 QAI

0.02 0.14 0.06 0.26

* *** *** ***

0.01 0.03 0.02 0.04

e 58.1 4.0 17.6

0.11 0.48 0.17 0.31

*** *** *** ***

0.04 0.09 0.04 0.06

“Quiet” sales area

LeqA LA90 LA10  LA90 QAI

0.42 0.38 0.01 0.13

*** *** 0.10 ***

0.07 0.08 0.02 0.05

65.3 58.3 e 26.5

0.42 0.33 0.18 0.42

*** *** *** ***

0.09 0.08 0.06 0.10

“Noisy” sales area

LeqA LA90 LA10  LA90 QAI

0.23 0.12 0.01 0.00

*** *** 0.15 0.87

0.06 0.06 0.01 0.00

68.8 66.0 e e

0.61 0.31 0.53 0.36

*** *** *** ***

0.12 0.07 0.12 0.06

Checkouts

LeqA LA90 LA10  LA90 QAI

0.67 0.71 0.47 0.66

*** *** *** ***

0.17 0.15 0.10 0.17

66.9 62.2 5.7 25.7

0.62 0.57 0.69 0.56

*** *** *** ***

0.15 0.15 0.18 0.09

7 6

R² = 0.14

5 4 3 2

Noise sources (%)

Warehouse “Quiet” sales area “Noisy” sales area Checkouts Hypermarket

- 0,4

Fig. 6. Loading plot for first 2 factors in hypermarket.

ASat

Table 3 Summary of the annoyance causes referred to the whole sample and to particular sub-samples.

- 0,6

Factor1

Fig. 5. Histogram of ASat response frequencies of employees who indicate at least one cause of acoustic discomfort and employees indicating no specific cause of acoustic discomfort.

subjective responses, ASat and API, were investigated. Subjective ratings were averaged in order to reduce the typical scatter that characterizes subjective responses (as observed in Fig. 4). For each of the selected objective parameters the corresponding subjective ratings falling within one unit intervals were averaged. In addition, a fuzzy logic approach was used to overcome limits of a discrete values-scale to evaluate individual perception [2]. Finally, in order to avoid an excessive influence of single outliers in small groups, only intervals with a number of observations greater than 5 were considered and the corresponding averages were weighted as a function of the number of answers in each interval [20]. Table 4 gives the coefficients of determination R2 of linear regression analysis with their level of significance expressed in terms of residual probability p, the slopes and “comfort” thresholds assumed as the parameter value below which ASat ratings are equal or higher than four. Plots of the best-correlated parameters are also given in Figs. 7e10. The correlations between objective parameters and ASat were weaker than correlations with API. The higher correlations observed between objective parameters and API could be explained considering that subjects (as seen in Fig. 4) could more easily assess sound intensity. Acoustic satisfaction was probably influenced by other factors and, above all, subjects seemed to remain stuck to the “average” value using (as shown in Fig. 4) comparatively fewer extreme ratings. Correlation between LA90 and ASat in the warehouse was significant but quite poor (R2 ¼ 0.14, p-value <0.001, Fig. 7). The slope of the regression line between LA90 and ASat was mild, suggesting a substantial indifference to background noise. In fact, ASat votes varied by one scale point against a LA90 variation of about 20 dB. Conversely, the background noise was better correlated with API (R2 ¼ 0.48, p-value <0.001). Considering that the “comfort” threshold of LA90 was about 58.1 dB, and that background noise in

- 0,8

Music

Refrigerator

Voice

Transpallet

HVAC system

4.6 7.4 3.3 9.3 24.6

0.3 0.8 5.2 3.8 10.2

2.1 3.8 4.1 7.4 17.4

10.5 5.6 2.0 1.0 19.0

0.2 1.5 2.5 2.0 6.1

1 35

40

45

50

55

60

LA90 Fig. 7. Warehouse: ASat as function of LA90. Marker area is proportional to number of questionnaires.

376

S. Della Crociata et al. / Building and Environment 59 (2013) 369e378

a

b

7

7 6

6 R² = 0.42

R² = 0.38

5 ASat

ASat

5 4

4

3

3

2

2 1

1 55

60

65

45

70

50

55

60

65

LA90

LeqA

Fig. 8. “Quiet” sales area: ASat as function of LeqA (a) and LA90 (b). Marker area is proportional to number of questionnaires.

Fig. 9. “Noisy” sales area: ASat as function of LeqA (a) and API as function of LeqA (b). Marker area is proportional to number of questionnaires.

b

a7

7 6

6

ASat

ASat

R² = 0.67

5

R² = 0.71

5 4

4

3

3

2

2 1

1 55

60

65

60

70

62

64

LA90

c

d

7

68

5

70

7 R² = 0.47

6

6

5

R² = 0.66

ASat

ASat

66 LeqA

4

4

3

3

2

2 1

1 20

25

30 QAI

35

0

2

4

6

8

LA10-LA90

Fig. 10. Checkouts: ASat as function of LA90 (a), LeqA (b), QAI (c) and LA10 LA90 (d). Marker area is proportional to number of questionnaires.

10

S. Della Crociata et al. / Building and Environment 59 (2013) 369e378

warehouse was low and the average was about 45.4 dB (st. dev. ¼ 5.2), it can be concluded that acoustic environment was satisfactory for most of warehouse’s workers. Spectral imbalance (QAI) showed correlations with both ASat and API (respectively, R2 ¼ 0.26, p-value <0.001, and R2 ¼ 0.31, p-value <0.001). However, these correlations suggested that workers may prefer imbalanced spectra which in the warehouse appeared more frequently in presence of background music. In the “quiet” sales area LeqA and LA90 were well correlated with ASat (respectively, R2 ¼ 0.42, p-value <0.001, and R2 ¼ 0.38, p-value <0.001). These objective parameters were well correlated also with API (for LeqA R2 ¼ 0.42, p-value <0.001, for LA90 R2 ¼ 0.33, p-value <0.001). In this area background sound levels (45e65 dB) were louder than in the warehouse (35e60 dB), but the “comfort” threshold for LA90 was similar, about 58.3 dB. As the mean background noise level was about 55.2 dB (st. dev. ¼ 3.8) acoustic environment was satisfactory for most workers in this area. In the “noisy” sales area the worst correlations appeared between objective parameters and ASat ratings, while LeqA was well correlated to API (R2 ¼ 0.61, p-value <0.001). This suggests that workers in this area were aware of the noise levels, but their satisfaction was scarcely related to this aspect [15]. This might be explained by their personal contribution to noise emissions with their own work (e.g. when they activated food preparation and packaging machines), or by their higher threshold of tolerance being accustomed to their noise environment. Finally, at the checkouts the correlations between subjective and objective parameters were high and statistically significant. Background noise LA90, caused by numerous sources, showed the best correlation with ASat (R2 ¼ 0.71, p-value <0.001), but also LeqA performed fairly well (R2 ¼ 0.67, p-value <0.001) as well as QAI (R2 ¼ 0.66, p-value <0.001). The correlation between ASat and QAI showed that higher levels of noise spectral imbalance affected negatively staff’s satisfaction possibly because of the large amount of high-frequency noises, like “beeps” due to barcode readers (which have peaks between 1 kHz and 1.6 kHz) (Fig. 2 d), customers passing with metal shopping trolleys and telescopic handles of plastic shopping trolley slamming. The presence of such impulsive noises might also explain the correlation between LA10  LA90 and ASat (R2 ¼ 0.47, p-value <0.001), even though, on average, noise fluctuations were very small (4e10 dB). Finally the LA90 “comfort” threshold was about 62.2 dB and suggested higher tolerance than in quieter areas. However, this value coincided with the average of measured levels (st. dev. ¼ 2.64), suggesting that a reduction in background levels (e.g. acting on music levels) might be preferable in order to improve the comfort conditions in this area. 4. Conclusions The paper discussed the results of on-site measurement to assess acoustic comfort for hypermarket workers. The measurements were carried out by collecting simultaneously both objective environmental parameters and staff subjective perceptions of acoustic comfort, during normal working time. Hypermarket’s acoustic satisfaction for the whole sample and in the four sub-spaces was highly correlated with acoustic perceived intensity. In particular, in the warehouse low perceived acoustic intensity levels were judged as satisfactory while high perceived levels were undesirable. Acoustic satisfaction proved to be influenced also by others factors, like sources of acoustic annoyance, in fact ASat was generally worse when at least one annoyance acoustic source was identified. A factor analysis was carried out among all noise indexes considered in this research, to reduce their number and identify those capable of describing independent aspects of noise. LeqA, LA90, LA10  LA90 and QAI were consequently selected to investigate their

377

correlations with subjective responses. These indexes proved to well describe subjective auditory sensations in terms of acoustic satisfaction and acoustic perceived intensity, in particular LA90 and LeqA, as in offices and urban open public spaces [4,5,12,15]. In the warehouse and in the “quiet” sales area acoustic environment was satisfactory for most of workers. In the “noisy” sales area acoustic satisfaction was scarcely related to noise levels probably because workers were accustomed to their noise environment, often determined by noise emissions of their own work. At the checkouts background noise and higher levels of noise spectral imbalance affected negatively staff’s satisfaction and a further reduction in background levels might increase acoustic satisfaction. New measurements and subjective data collections are under way, according to the discussed aspects, to evaluate entirely acoustic comfort for hypermarket workers. The results of research were obtained in a specific hypermarket so, to generalize the results, the investigation should be extended further taking into account also other aspects to evaluate acoustic comfort, like job typology. In fact, the effect of noise on working performance, and consequently subjective acoustic evaluation, is strongly influenced by the active or passive attitude of the human being towards that noise.

Acknowledgments This work was supported and financed by the Istituto Nazionale per l’Assicurazione contro gli Infortuni sul Lavoro. The authors would like to acknowledge Michele D’Alba and Leonardo Calderoni for their work and support, and Coop Estense s.c.r.l. that cooperated with the research by allowing access to one of their hypermarkets and involving their staff.

Symbols

ADI API ASat CNI LeqA LA90

acoustic desired intensity acoustic perceived intensity acoustic satisfaction combined noise index A-weighted equivalent sound pressure level (dB) A-weighted equivalent level exceeded for 90% of the time (dB) A-weighted equivalent level exceeded for 10% of the time LA10 (dB) A-weighted equivalent level exceeded for 5% of the time LA5 (dB) LA10  LA90 noise climate (dB) linear equivalent sound pressure level (dB) Leqlin Zwicker’s loudness level (phon) LLZ Stevens’ loudness level (phon) LLS Noise pollution level (dB) LNP LSIL speech interference level, average of levels at 500, 1000, 2000 and 4000 Hz (dB) NC noise criterion NCB balanced noise criterion NR noise rating ONI office noise index (dB) PNC Preferred noise criterion PSIL speech interference level, average of levels at 500, 1000 and 2000 Hz (dB) QAI quality assessment index (dB) RC room criterion RCII room criterion mark II

378

S. Della Crociata et al. / Building and Environment 59 (2013) 369e378

Appendix A Correlation matrix between noise indexes. Bold numbers represent R2 greater than 0.50

Leqlin LeqA LA90 LA10 LA5 LA10  LA90 LNP LLZ LLs NCB RC RCII NC PNC NR QAI LSIL PSIL ONI CNI

Leqlin

LeqA

LA90

LA10

LA5

LA10  LA90

LNP

LLZ

LLs

NCB

RC

RCII

NC

PNC

NR

QAI

LSIL

PSIL

ONI

CNI

1 0.82 0.65 0.82 0.74 0.16 0.08 0.88 0.84 0.80 0.83 0.82 0.80 0.81 0.74 0.01 0.80 0.83 0.01 0.65

1 0.50 0.96 0.94 0.04 0.26 0.98 0.98 0.99 0.99 0.99 0.97 0.92 0.97 0.09 0.99 0.99 0.12 0.95

1 0.52 0.34 0.69 0.06 0.55 0.51 0.47 0.52 0.51 0.51 0.55 0.43 0.05 0.47 0.51 0.15 0.37

1 0.95 0.04 0.24 0.95 0.94 0.95 0.96 0.96 0.93 0.89 0.92 0.06 0.96 0.96 0.13 0.90

1 0.00 0.40 0.91 0.91 0.94 0.93 0.93 0.90 0.84 0.92 0.11 0.95 0.94 0.25 0.92

1 0.54 0.06 0.05 0.03 0.05 0.05 0.05 0.08 0.02 0.26 0.03 0.05 0.71 0.01

1 0.22 0.24 0.28 0.24 0.25 0.23 0.18 0.30 0.42 0.28 0.25 0.95 0.36

1 0.99 0.98 0.97 0.96 0.94 0.91 0.94 0.05 0.98 0.97 0.09 0.90

1 0.98 0.96 0.96 0.94 0.90 0.96 0.07 0.98 0.97 0.10 0.92

1 0.98 0.98 0.95 0.91 0.96 0.10 1.00 0.98 0.13 0.95

1 0.99 0.96 0.93 0.95 0.08 0.98 1.00 0.11 0.95

1 0.96 0.92 0.95 0.08 0.98 1.00 0.11 0.94

1 0.96 0.94 0.08 0.95 0.97 0.10 0.92

1 0.88 0.04 0.91 0.93 0.07 0.86

1 0.13 0.96 0.95 0.15 0.94

1 0.10 0.08 0.38 0.23

1 0.99 0.14 0.96

1 0.11 0.95

1 0.19

1

References [1] Martellotta F, Della Crociata S, Simone A. Laboratory study on the effects of office noise on mental performance. In: Forum acusticum 2011, 27 Junee1 July, Aalborg. [2] Ayr U, Cirillo E, Fato I, Martellotta F. A new approach to assessing the performance of noise indices in buildings. Appl Acoust 2003;64:129e45. [3] Ayr U, Cirillo E, Martellotta F. Further investigations of a new parameter to assess noise annoyance in air-conditioned buildings. Energ Buildings 2002;34: 765e74. [4] Jensen KL, Arens E, Zagreus L. Acoustical quality in office workstations, as assessed by occupant surveys. Indoor Environmental Quality (IEQ); 2005. p. 2401e5. [5] Ayr U, Cirillo E, Martellotta F. An experimental study on noise indices in air conditioned offices. Appl Acoust 2001;62:633e43. [6] Keighley EC. The determination of acceptability criteria for office noise. J Sound Vib 1966;4(1):73e87. [7] Keighley EC. Acceptability criteria for noise in large office. J Sound Vib 1970; 11(1):83e93. [8] Hay B, Kemp MF. Measurements of noise in air conditioned, landscaped offices. J Sound Vib 1972;23(3):363e73. [9] Kjellberg A, Landström U, Tesarz M, Söderberg L, Åkerlund E. The effects of nonphysical noise characteristics, ongoing task and noise sensitivity on annoyance and distraction due to noise at work. J Environ Psychol 1996;16: 123e36. [10] Keighley EC, Parkin PH. Subjective response to sound conditioning in a landscaped office. J Sound Vib 1979;64(3):313e23. [11] Huang L, Zhu Y, Ouyang Q, Cao B. A study on the effects of thermal, luminous, and acoustic environments on indoor environmental comfort in offices. Building Environ 2012;49:304e9.

[12] Yang W, Kang J. Acoustic comfort evaluation in urban open public spaces. Appl Acoust 2005;66:211e29. [13] Paunovi c K, Jakovljevi c B, Belojevi c G. Predictors of noise annoyance in noisy and quiet urban streets. Sci Total Environ 2009;407:3707e11. [14] Yu L, Kang J. Effects of social, demographical and behavioral factors on the sound level evaluation in urban open spaces. J Acoust Soc America 2008; 123(2):772e83. [15] Kang J, Zhang M. Semantic differential analysis of the soundscape in urban open public spaces. Building Environ 2010;45(1):150e7. [16] Yu L, Kang J. Factors influencing the sound preference in urban open spaces. Appl Acoust 2010;71(7):622e33. [17] Kruger EL, Zannin PHT. Acoustic, thermal and luminous comfort in classrooms. Building Environ 2004;39:1055e63. [18] Zannin PHT, Marcon CR. Objective and subjective evaluation of the acoustic comfort in classrooms. Appl Ergon 2007;38:675e80. [19] Chen B, Kang J. Acoustic comfort in shopping mall atrium spaces: a case study in Sheffield Meadowhall. Architectural Sci Rev 2004;47:107e14. [20] Della Crociata S, Martellotta F, Simone A. A measurement procedure to assess indoor environment quality for hypermarket workers. Building Environ 2012; 47:288e99. [21] ANSI/ASHRAE. Standard 55-2004: thermal environmental conditions for human occupancy. Atlanta, Georgia: American Society of Heating Refrigerating and Air-Conditioning Engineers (ASHRAE); 2010. [22] Chan DWT, Burnett J, de Dear RJ, Ng SCH. A large-scale survey of thermal comfort in office premises in Hong Kong. ASHRAE Trans Symposia 1998; 11(3):1172e80. [23] HOPE: Health Optimisation Protocol for Energy-efficient Buildings. Prenormative and socio-economic research to create healthy and energy efficient buildings. http://hope.epfl.ch/. [24] ISO 10551: 1999. Ergonomics of the thermal environment e assessment of the influence of the thermal environment using subjective judgement scales.