UEF//eRepository
DSpace https://erepo.uef.fi
Rinnakkaistallenteet Luonnontieteiden ja metsätieteiden tiedekunta
2019
Control of Dust Dispersion From an Enclosed Renovation Site Into
Adjacent Areas by Using Local Exhaust Ventilation
Kokkonen, A
Oxford University Press (OUP)
Tieteelliset aikakauslehtiartikkelit
© Authors
All rights reserved
http://dx.doi.org/10.1093/annweh/wxz016
https://erepo.uef.fi/handle/123456789/7543
Downloaded from University of Eastern Finland's eRepository
Control of dust dispersion from an enclosed renovation site into adjacent areas by using local 1
exhaust ventilation 2
3
1Anna Kokkonen, PhD, 2Markku Linnainmaa, PhD, 2Arto Säämänen PhD, 2Tomi Kanerva, MSc, 4
1Jouni Sorvari, PhD, 1Mikko Kolehmainen, PhD Tech., 1Vuokko Lappalainen, MSc, 1Pertti 5
Pasanen, PhD 6
7
1University of Eastern Finland, Department of Environmental and Biological Sciences, P.O. Box 8
1627, FI-70211 Kuopio Finland, 9
2Finnish Institute of Occupational Health, FI-33032 Työterveyslaitos, Finland 10
Address correspondence to Anna Kokkonen, University of Eastern Finland, Department of 11
Environmental and Biological Sciences, P.O. Box 1627 FI-70211 Kuopio Finland.
12
E-mail: anna.kokkonen@uef.fi. Tel: +358 40 355 2856 13
14 15 16 17 18
ABSTRACT 1
2
Objective: In real-world applications, implementation of an enclosure and negative pressurization 3
is not always adequate to prevent the dispersion of dust from renovation sites. This study aimed to 4
quantify the effect of local exhaust ventilation (LEV) in controlling the dust concentration within 5
an enclosed renovation site to reduce the dust dispersion into adjacent areas.
6
Methods: The concentrations of inhalable and respirable dust were measured in 16 cases during 7
renovation projects. Filter samples and time-resolved dust concentration data were collected 8
simultaneously from the renovation site and adjacent areas to assess the efficacy of LEV in limiting 9
the dust dispersion.
10
Results: The dispersion of dust outside of the enclosed renovation sites was limited significantly 11
with using LEV. The estimated dust removal efficiency of LEV was 79% for inhalable dust 12
concentration in the renovation site and 62% in the adjacent area. The use of LEV reduced the 13
concentration of respirable dust by 33‒90% in the adjacent area and 80−87% within the renovation 14
site.
15
Conclusions: Using LEV was found to play a substantial role in dust containment, particularly 16
when the enclosure failed to maintain the negative pressure. The study provides data-driven 17
recommendations that are of practical importance as they promote healthier workplaces and policy 18
improvements. In conclusion, dust dispersion into adjacent areas is prevented with an airtight 19
enclosure (including airlocks) and continuous negative pressure. Dust containment was also 20
obtained by having target dust concentration at the enclosed renovation site to below 4 mg/m3 for 21
inhalable dust and below 1 mg/m3 for respirable dust, even though the enclosures not being 22
continuously under negative pressure. The suggested target concentrations are achievable by using 23
on-tool LEV during the most dust-producing tasks.
24 25 26
Keywords: renovation, construction, dust control, local exhaust ventilation, enclosure, negative 27
pressure 28
29
INTRODUCTION 1
2
A unique feature of renovation work is that the other parts of the building might be under normal 3
operation. This situation poses a challenge for effective dust control to protect the other people 4
working in or occupying the adjacent areas from adverse dust exposure. The spread of biologically 5
inert dust outside a renovation site is typically prevented by separating the renovation site using a 6
partitioning enclosure and applying negative pressure. In our previous study (Kokkonen et al., 7
2017), dust containment of an enclosure was ensured by careful sealing and airtight enclosure 8
structures, including an airlock between the renovation site and adjacent area. In addition, 9
verification of the negative pressure below -5 Pa via continuous monitoring was suggested 10
(Kokkonen et al., 2017). There are special requirements for more controlled dust containment 11
when the structures to be removed contain hazardous agents, such as asbestos, lead, creosote or 12
microorganisms (EPA, 2008; Council Directive 2009/148/EC).
13 14
In real-world applications, however, implementation of an enclosure and negative pressurization 15
may not always be adequate to prevent the dispersion of dust from renovation sites: dust dispersion 16
outside of the enclosed renovation site appeared in half of the cases studied (Kokkonen et al., 17
2017). Due to these failures, dust control within the enclosure is emphasized to ensure the dust 18
containment. Although the benefits of using local exhaust ventilation (LEV) as a control measure 19
in construction and renovation sites are well established from the workers’ personal exposure point 20
of view (e.g., Fransman et al., 2008; Meeker et al., 2009; Shepherd et al., 2009; Akbar-Khanzadeh 21
et al., 2010; Carlo et al., 2010), there are no published data assessing the effectiveness of LEV to 22
limit the spread of dust from renovation sites into the adjacent areas to protect the building users 23
from dust exposure. The use of LEV is also beneficial to protect the coworkers (such as electricians 24
performing less dusty work tasks) at the renovation site not to expose to high dust concentrations 25
(Woskie et al., 2002) because the LEV captures the dust at the point of generation. Factors such 26
as the task, practices, material, tooling used, and psychosocial factors have a substantial effect on 27
on-site dust concentrations (Croteau et al., 2004; Tjoe Nij et al., 2004; Flanagan et al., 2006;
28
Beaudry et al., 2013; van Deurssen et al., 2014).
29 30
Evaluation of the effect on dust containment effectiveness by using LEV within enclosures is 31
needed to improve practices in controlling particle dispersion from the renovation sites, and 32
further, to protect persons working in or occupying in the adjacent areas from adverse dust 33
exposure. This work aims to quantify how reducing the dust concentration at an enclosed 34
renovation site using LEV affects the dispersion of dust from the renovation site into adjacent 35
areas.
36 37 38
MATERIALS AND METHODS 39
40
Study cases 41
42
This study involved seven renovation sites that provided 16 cases in which different dust- 43
producing work tasks were performed (Table 1). The cases are labeled according to the same case 44
numbers applied in our previous study dealing with the implementation of enclosure and negative 45
pressurization for dust containment (Kokkonen et al., 2017).Renovation work was conducted 46
according to the planning and time schedule set by the contractors’ own preferences, representing 1
real-world implementation of dust control. The renovation sites were separated from the 2
surrounding spaces (under normal operation) by utilizing the existing room division or 3
constructing temporary partitioning walls using plastic film and either gypsum board or plywood 4
(Table 1). The enclosures were negatively pressurized with respect to the adjacent spaces by 5
installing portable exhaust fan units in the renovation sites (Table 1). The exhaust air was filtered 6
either with coarse filters, coarse and fine filters or HEPA filters and vented to the outdoors. The 7
make-up air was supplied via a supply air valve (case 3), from the surrounding corridor via a supply 8
air valve (cases 4 and 5), and from the adjacent areas via door gaps or other uncontrolled leakages 9
(cases 1a-b, 2a-g, 9a-b, 12). Detailed descriptions of the implementation of the enclosures and 10
negative pressurization, measuring methods of ventilation and pressure differences, as well as the 11
results of control of pressure differences (∆p) between the renovation sites and adjacent areas have 12
been published in our previous article (Kokkonen et al., 2017). Based on that, Table 1 presents the 13
air exchange rates of the renovation sites, pressure differences and percentages of positive pressure 14
periods measured. The air exchange rates of the cases studied ranged from 0.8 to 43 h-1, with an 15
average of 6.9 h-1 (Table 1) (Kokkonen et al., 2017). The mean ∆p varied between a slightly 16
positive pressure of 0.2 Pa to a highly negative pressure of -50 Pa, and the cases were 0−76% of 17
follow-up time under positive pressure (Table 1)(Kokkonen et al., 2017).
18 19
LEV was used in half of the cases (Table 1). The types of LEV used were on-tool extraction, 20
ventilation shroud, and moveable capturing hood (Table 1). In on-tool extraction and ventilation 21
shroud configurations, tools were connected via flexible hoses to vacuum sources. Portable 22
construction site vacuums with either HEPA filters or standard disposable filters were used to 23
provide ventilation airflow for the tools. In the case of moveable capturing hood configurations, 24
inlet hoses of vacuum systems were positioned at the dust generation source. Portable LEV 25
systems recirculated the filtered exhaust air back into the work space (Table 1).
26 27
Table 1. Case details Case Enclosure
construction n [h-1]
∆p [Pa]
ppos
[%]
Tool configuration
Tool LEV type and vacuum source
chipping tiled floor and walls
2d gypsum boards 3.1 0.2 76 jackhammer Hilti TE 706- AVR
- 2e gypsum boards 3.1 -0.1 57 jackhammer Hilti TE 706-
AVR on-tool extraction, portable vacuum Hilti VC 40-U (disposable filter)
9a plastic 5.4 -0.2 39 jackhammers Hilti TE 1000-AVR & Hilti TE 500
- 9b plastic 5.4 -0.4 28 jackhammers Hilti TE
1000-AVR & Hilti TE 500
on-tool extraction, portable vacuum Hilti VC 20-UM (disposable filter)
concrete chipping
1a plastic 43 -0.2 5.9 jackhammer Hilti TE 1000-AVR
-
1b plastic NM NM NM jackhammer Hilti TE
1000-AVR
- concrete grinding
2f gypsum boards 3.1 -0.3 0 floor grinder (unknown) ventilation shroud, portable vacuum Pullman Ermator S26 (HEPA filter)
12 plastic 0.8 NM NM floor grinder Husqvarna PG 680
ventilation shroud, portable vacuum Pullman Ermator T6000 (HEPA filter) dismantling of ceiling, floor and walls
2a gypsum boards 3.1 -0.4 1.2 reciprocating saw Hilti
WSR 36-A moveable capturing hood, portable vacuum Pullman Ermator S26 (HEPA filter) 2b gypsum boards 3.1 -0.4 1.1 reciprocating saw Hilti
WSR 36-A
moveable capturing hood, portable vacuum Pullman Ermator S26 (HEPA filter) 2g gypsum boards 3.1 -0.1 32 reciprocating saw Hilti
WSR 36-A
2 x moveable capturing hood, 2 x portable vacuum Pullman Ermator S26 (HEPA filter)a 4a existing wallsb 11 -50 0 reciprocating saw Makita
JR3070CT -
5a existing wallsb 2.2 -13 0 reciprocating saw Makita JR3070CT
- 5b existing wallsb 5.0 -24 0 reciprocating saw Makita
JR3070CT
- drywall sanding
2c gypsum boards 3.1 -0.1 0 Mirka Deco pole sander on-tool extraction, portable vacuum Nilfisk Panther GD 930 S2 (HEPA filter) installation works
3 gypsum boards 5.5 -0.7 0 mitre saw DeWalt D27111 on-tool extraction, 2 x portable vacuum Pullman Ermator S26 & S13 (HEPA filters),
exhaust fan Onnline KVFU-315B (coarse filter)c
Notes: n: air exchange; ∆p: pressure difference; ppos: percentage of positive pressure periods; -: no LEVs used; NM: not measured
a = LEVs used in both sides of the dismantled wall
b = Including three-stage airlock made of plywood and plastic
c = Inlet hose of an exhaust fan unit positioned at the dust generation source, exhaust air filtered outdoors
1
1
Measurements and data analysis 2
3
Inhalable and respirable dust concentrations at the renovation sites were assessed by collecting 4
dust samples from the stationary sampling locations (a total of 52 samples). Inhalable dust samples 5
were taken from all but one case (excluded case 1b), while respirable sampling was done in 11 of 6
16 cases (excluded cases 2a, 2b, 2g, 3, 4). Stationary samplers were placed approximately 1−2 7
meters from the dust-generating activity in the renovation site. Simultaneously, dust sampling was 8
performed in the adjacent area in front of the main entry into the enclosure to assess the dust 9
dispersion from the renovation site. Samplings were carried out during the dust-producing work 10
activities, i.e. sampling duration was determined by the activity. IOM samplers (SKC Inc., Eighty 11
Four, PA, USA) were used for the inhalable dust, and IOM samplers with MultiDust foam inserts 12
were used for the respirable dust. Dust samples were collected onto cellulose acetate filters 13
(Millipore AAWP, diameter 25 mm, pore size 0.8 µm) by SKC 224 sampling pumps. The flow 14
rate of the pumps was calibrated to 2.0 L/min prior to sampling. After sampling, the flow rate of 15
the pumps were checked, having pre and post-sampling flow rates with 5% for each pump (data 16
not shown). The filters were dried in a desiccator at least 2 days before they were weighed under 17
constant climatic conditions. The filter sampling and analytical methods have been described in 18
detail earlier(Tuomi et al., 2014) and have been accredited by the Finnish Accreditation Service 19
(FINAS).
20 21
In addition, the concentrations of inhalable and respirable dust were monitored from both 22
renovation sites and adjacent areas using Split2 direct reading dust monitors (SKC Inc., Eighty 23
Four, PA, USA) to assess the dispersion of dust. This monitoring was performed in eight cases 24
(2a-b, 2d-f, 5, and 9a-b) for inhalable dust and four cases (1a and 2d-f) for respirable dust. The 25
dust monitors were placed in the same stationary locations in the renovation sites as those of the 26
IOM samplers, and the readings were taken at the same events. Logging interval was 1 minute.
27
The results were corrected by the gravimetric results obtained from simultaneous IOM sampling.
28
The average correction factors were 2.60 for inhalable dust and 0.85 for respirable dust. The dust 29
monitors were calibrated annually by the manufacturer's services before use.
30 31
The values below the limit of detection (LOD) were processed with NDExpo software (Flot 0.8.1, 32
Montreal, Canada), which applies the robust regression on order statistics (ROS) approach of 33
Helsel (2005). The software replaces the <LOD values with exposure estimates based on their rank 34
among the set of detected values. After treatment of <LOD with NDExpo, the geometric mean 35
concentration (GM) and standard deviation (GSD) of each size fraction for the renovation site and 36
adjacent area were determined for the cases of with and without the use of LEV. The efficacy of 37
the LEV was defined by the percent reduction of dust concentration both within the renovation site 38
and within the adjacent area, for both size fractions compared to using no dust control:
39 40
41
% 100
%
LEV no
LEV LEV
no
C C
reduction C (1)
42 43 44
Statistical analysis 1
2
Two statistical analyses were used: linear mixed model analysis was used for the collected dust 3
samples, and regression analysis was used for the direct reading of the dust concentration time 4
series data. For the collected dust samples, two separate linear mixed models were used to explore 5
whether the use of LEV affected the inhalable dust concentration at the renovation site and the 6
dispersion of inhalable dust into the adjacent area. The respirable dust fraction was not analyzed 7
using the linear mixed model because of the small sample size. Shapiro-Wilks test showed 8
concentration data were not normal, but the data were log-normally distributed. Therefore, log- 9
transformed data were used for the mixed models. In the mixed models, the use of LEV was 10
applied as a fixed factor (Equation 2 and 3). Because several measurements from different work 11
tasks and locations within the same renovation sites were conducted, the renovation site was 12
included as a random factor (Equation 2).
13 14 15
ijk j i
yijk (2)
16
yijk = kth observation of dust concentration under the ith fixed effect (LEV) and jth random 17
effect (renovation site) 18
τi = effect of the use or without use of LEV 19
ßj = effect of the jth renovation site 20
ɛijk = associated unobserved error term of the ijkth observation.
21 22 23
The ∆p was used as a covariate in the model for the dispersion of dust into the adjacent area 24
(Equation 3).
25 26
27 yijk i j S
xijk xijk
ijk (3)28
yijk = kth observation of dust concentration under the ith fixed effect (LEV) and jth random 29
effect (renovation site) 30
xijk = kth observation of covariate (pressure difference) under the ith fixed effect (LEV) 31
and jth random effect (renovation site) 32
τi = effect of the use or without use of LEV 33
ßj = effect of the jth renovation site 34
S = slope of the line 35
ɛijk = associated unobserved error term of the ijkth observation.
36 37 38
For the time series concentration data, a forward stepwise linear regression was carried out to 39
examine the effect of the dust concentration at the renovation site and the pressure differences 40
between the renovation site and adjacent area on the dust concentration in the adjacent area. The 41
analysis was performed separately for the inhalable and respirable fractions obtained by Split2 42
direct reading particulate monitors. The predictors were allowed to enter the model using a P value 43
of less than 0.05 for entry and of greater than 0.10 for removal. The differencing of the data was 44
applied to proceed with auto correlation issues of the real-time data and to remove trends from the 45
data to obtain a stationary series (Entink et al., 2011). Differenced dust concentrations in the 1
renovation site and ∆p were delayed based on observed air exchange rates and the dilution of 2
contaminants, i.e., the calculated time when 90% of the contaminants were removed (Kokkonen 3
et al., 2017). The lags were from 1 to 15 and 1 to 45 minutes at a time resolution of 60 seconds.
4
This time resolution was chosen based on the recordings of pressure difference readings. The 5
regression results were analyzed in terms of the sum of the coefficients (Woolridge, 2013) on the 6
initial and lagged predictors. A more detailed description of the differencing approach was 7
provided in our previous study (Kokkonen et al., 2017).
8 9
An alpha <0.05 was considered as statistically significant. The statistical analyses were conducted 10
using the IBM SPSS Statistics for Windows, Version 23.0 (SPSS Inc, Armonk, NY, USA).
11 12 13
RESULTS 14
15
Inhalable and respirable dust concentrations 16
17
The study focuses on quantifying the effect of using on-tool LEV to limit the dust dispersion from 18
enclosed renovation sites, but also the dust removal efficiency by LEV within the renovation site 19
is considered since the dust concentration level at the renovation site potentially contributes to the 20
observed dust concentrations in the adjacent areas. The summary results (filter samples) of the 21
inhalable and respirable dust concentrations at the renovation site and in the adjacent area with and 22
without LEV use at the renovation site are presented in Table 2.
23 24 25
Table 2. Summary results (filter samples) of the inhalable and respirable dust concentrations at the renovation site and in the adjacent area
Dust samplesa GM GSD Min Max N N
[mg/m3] [mg/m3] [mg/m3] [mg/m3] <LOD Inhalable dust, renovation site
no LEV 27 3.6 5.2 110 6 0
LEV 3.6 3.5 0.30 13 9 0
Inhalable dust, adjacent area
no LEV 0.41 5.4 0.06 3.5 7 1
LEV 0.45 1.5 0.30 0.90 9 0
Respirable dust, renovation site
no LEV 4.0 5.5 0.20 12 5 0
LEV 0.25 10.6 0.01 1.8 5 1
Respirable dust, adjacent area
no LEV 0.09 5.3 0.01 0.60 6 0
LEV 0.02 3.5 0.004 0.08 5 2
Notes: GM: geometric mean; GSD: geometric standard deviation; N: number of samples; N
<LOD: Number of samples below limit of quantification; LEV: local exhaust ventilation
a = The sampling duration was 131 minutes on average.
26 27
For respirable dust, 10% of the renovation site samples and 18% of the adjacent area samples were 1
below the LOD when using LEV (Table 2). For all dust-producing tasks, the geometric mean 2
respirable dust concentration at the renovation site was 0.25 ± 10.6 mg/m3 with LEV and 4.0 ± 5.5 3
mg/m3 without LEV (Table 2). Similarly, the dust concentration of the adjacent area remained 4
lower (0.02 ± 3.5 mg/m3) when using LEV than when not using LEV (0.09 ± 5.3 mg/m3) (Table 5
2). Overall, lower adjacent area concentrations of inhalable dust without LEV (Table 2) were most 6
likely related to recognized but uncontrolled factors, such as tightness of the enclosures as found 7
in our previous study (Kokkonen et al. 2017). The three lowest dust concentrations in the adjacent 8
area (data not shown) were observed in cases (4-5), which had no LEV in use but were more 9
airtightly enclosed (sealed airlock divided into three sections) and adjusted with high, continuous 10
negative pressure from -12 Pa to -50 Pa. Adjacent area concentrations were hereby expected to be 11
extremely low in those cases.
12 13
According to the linear mixed model, the estimated marginal means of the inhalable dust 14
concentration at the renovation site (back transformed after log-transformation) was approximately 15
five times lower (F1, 18=6.90, p=0.017) with LEV (3.0 mg/m3, 95% confidence interval 2.0‒6.3 16
mg/m3) than without LEV (14 mg/m3, 95% CI 9.6‒29 mg/m3) (Figure 1). In addition, the dust 17
dispersion into the adjacent area was significantly different (F1,8=7.62, p=0.025) between the cases 18
with and without LEV (Figure 1). The mixed model revealed that the estimated concentrations 19
outside of the renovation site were 2.6 times lower when using LEV. The estimated marginal 20
means of the predicted adjacent area dust concentrations (presented in Figure 1) are as follows:
21
0.53 mg/m3 (95% CI 0.35‒1.1 mg/m3) without LEV and 0.20 mg/m3 (95% CI 0.14‒0.42 mg/m3) 22
with LEV. The estimated marginal means were evaluated at a pressure difference value of -6.4 Pa.
23
The pressure difference did not show a statistically significant effect on the dust dispersion into 24
the adjacent area (F1,6=4.80, p=0.070). The association between the pressure difference and the 25
dust concentration was similar between the cases of LEV and no LEV (interaction term ∆p × LEV 26
treatment: F1,6=2.49, p=0.168). This means higher negative pressure did not increase control of 27
dust according to the mixed model.
28 29 30 31 32
1 Figure 1. Inhalable dust concentrations with no LEV and with LEV: Estimated marginal means 2
derived from separate linear mixed models for the concentrations at the renovation site and in the 3
adjacent area with 95% confidence intervals.
4 5 6
Overall, the estimated dust removal efficiency of LEV (based on back-transformed coefficient of 7
mixed models presented in Table S1, Supplementary material) was 79% for inhalable dust 8
concentration in the renovation site and 62% in the adjacent area. In addition, the efficiency with 9
which LEV could reduce the dust concentrations at the renovation site and in adjacent areas was 10
determined through four pairwise comparisons between no LEV and LEV use in identical working 11
areas (e.g., bathrooms next to each other) within the same renovation site (cases 2d-e, 9a-b). Cases 12
2d-e involved tiled floor chipping, and cases 9a-b involved both tiled floor and wall chipping, 13
where the use of LEV reduced the inhalable dust concentration by 85% (case 2d-e) and 80% (case 14
9a-b) at the renovation site (Table 3). In the adjacent area, the LEV reduction efficiencies were 15
31% and 68%, respectively. Similar dust reduction levels (80‒87%) were achieved at the 16
renovation site with LEV for respirable dust during tiled floor and wall chipping (cases 2d-e and 17
9a-b). Using LEV resulted in a 33% reduction in respirable dust in the adjacent area in case 2d-e.
18
A higher reduction in respirable dust (90%) in the adjacent area was achieved in case 9a-b;
19
however, this reduction was based on the estimate for the LEV samples below the LOD (Table 3).
20 21 22 23
1
Table 3. Pairwise comparison of the dust concentrations with and without LEV at the renovation site and in the adjacent area and the dust reduction efficiency of LEV
Case Inhalable dust [mg/m3] Respirable dust [mg/m3] renovation site adjacent area renovation site adjacent area chipping tiled floor
2d no LEV 89 1.3 8.8 0.12
2e LEV 13 0.9 1.8 0.08
dust reduction [%] 85.4 30.8 79.5 33.3
chipping tiled floor and wall
9a no LEV 59 1.9 9.0 0.20
9b LEV 12 0.6 1.2 0.02a
dust reduction [%] 79.7 68.4 86.7 90.0
a = present value (initially <LOD) obtained from NDExpo software
2 3
Dispersion of dust 4
5
The dispersion of inhalable and respirable dust from the renovation site to the adjacent area was 6
explored using regression analysis for time-resolved dust concentrations to study the relevant 7
factors of dust dispersion. A summary of the results of the regression analysis is presented in Table 8
4 in terms of statistically significant predictors and their standardized coefficients (renovation site 9
dust concentration and pressure difference between the renovation site and adjacent area). The 10
index of agreement (IA) and coefficient of determination (R2) are also provided. The index of 11
agreement describes the degree to which the observed variate is accurately estimated by the 12
predicted variate (Willmott, 1981; Willmott, 1982). More details of the IA were presented in a 13
previous study by Kokkonen et al. (2017). In Table 4, a positive standardized regression coefficient 14
(β) of the dust concentration in the renovation site indicates a true, positive association between 15
the dust concentrations at the renovation site and adjacent area. In other words, a high dust 16
concentration at the renovation site is related to a high dust concentration in the adjacent area. The 17
interpretation of β is more complex for ∆p: when the observed ∆p declines, the change in the 18
differenced data is positive. Therefore, a negative β indicates that the observed dust concentration 19
in the adjacent area is lower with lower observed pressure differences (i.e., negative pressure) 20
(Table 4). Because the purpose of regression analyses conducted here was not to achieve absolute 21
predictions but to study the relevant factors of dust dispersion, the results are not discussed in terms 22
of the exact standardized coefficient values. In addition, the concentration values (AM ± SD) 23
together with regression analysis results are those obtained by the direct reading dust monitors 24
(Table 4). The average sampling duration was 156 minutes.
25 26 27
Table 4. Summary results of the forward stepwise regression analysis of the dust concentration in the adjacent area Model
No. Case LEV use
βa
IA R2 Comment Crenovation site
[mg/m3]
Cadjacent area
[mg/m3]
∑Crenovation site ∑∆p Inhalable dust
I-1 2d no LEV -0.793 0.702 0.54 0.8 Dust dispersion
associated with positive pressure events
65 ± 51 1.4 ± 0.36
I-2 5 no LEV -0.335 - 0.28 0.04 High continuous negative pressure within the enclosed renovation site
12 ± 7.8 0.08 ± 0.02
I-3 9a no LEV 0.356 0.37 0.44 0.27 Dust dispersion
associated with positive pressure events
63 ± 37 1.9 ± 1.2
I-4 2a LEV 0.624 -0.469 0.42 0.35 Low dust concentration within the enclosed
renovation site
3.8 ± 6.0 0.34 ±0.03
I-5 2b LEV 1.26 -0.011 0.73 0.35 Dust dispersion
associated with waste transfer and positive
pressure events
4.3 ± 6.9 0.47 ± 0.05
I-6 2e LEV 0.398 -0.684 0.61 0.41 Dust dispersion
associated with positive pressure events
9.1 ± 10.2 0.89 ± 0.05
I-7 2f LEV 0.53 - 0.41 0.28 Continuous negative
pressure and low dust concentration within the enclosed renovation site
0.85 ± 0.90 0.32 ± 0.02
I-8 9b LEV 0.366 -0.463 0.53 0.47 Dust dispersion
associated with positive pressure events
10 ± 8.1 0.62 ± 0.17
Respirable dust
R-9 1a no LEV - 0.456 0.62 0.21 High air exchange rate (dilution)
12 ± 5.6 0.25 ± 0.03 R-10 2d no LEV 0.927 -0.158 0.47 0.81 Dust dispersion
associated with positive pressure events
6.1 ± 5.0 0.14 ± 0.05
R-11 2e LEV - 0.34 0.38 0.15 Low dust concentration
within the enclosed renovation site
1.2 ± 1.2 0.08 ± 0.008
R-12 2f LEV - -0.64 0.3 0.41 Continuous negative
pressure and low dust concentration within the enclosed renovation site
0.03 ± 0.01 0.005 ± 0.001
Notes: β: standardized coefficients (∑Crenovation site: sum of coefficients for concentration at the renovation site, ∑∆p: sum of coefficient for pressure difference); IA: index of agreement; R2: coefficient of determination; -: insignificant coefficient
a = Regression model summary (detailed lags of the differenced inhalable dust concentrations at the renovation site and differenced pressure differences) and their standardized coefficients are presented in the supplemental online materials (Table S2).
1
1
An IA below 0.4 indicated that the model goodness was worse than random in models I-2, R-11 2
and R-12 (Table 4). In addition, the model goodness (IA 0.41−0.42) was equal to the goodness 3
that would be achieved at random for models I-4 and I-7. Low IA values indicated that the adjacent 4
area concentrations were not related to the dust concentrations at the renovation sites and the 5
pressure differences in these models. Models with continuous negative pressure (I-2 and I-7) had 6
the lowest, constant adjacent area concentrations that were not related to the variation of 7
concentrations at the renovation sites. The fact that dispersion of dust from the renovation site did 8
not occur was particularly expected in case 5 (model I-2). This case had well-implemented dust 9
containment (sealed airlock divided into three sections) compared to the other cases and was 10
adjusted with high, continuous negative pressure (-20 Pa). However, with the use of LEV, the 11
dispersion of dust from the renovation site was not observed in models I-4, I-7, R-11 and R-12.
12
These models had the lowest dust concentrations within the renovation site (inhalable dust I-7:
13
0.85 ± 0.90 mg/m3; I-4: 3.8 ± 6.0 mg/m3; respirable dust R-11: 1.2 ± 1.2 mg/m3;R-12: 0.03 ± 0.01 14
mg/m3). Moreover, LEV was effective in limiting the dust dispersion despite the enclosed 15
renovation site (model R-11) being under positive pressure the majority of the time (65%; ∆p=0.1 16
Pa). A similar finding was obtained for model I-4, which had at times positive pressure periods 17
(1.2%; ∆p=-0.4 Pa).
18 19
However, despite the use of LEV, dispersion of dust was observed and models I-5 and I-6 were 20
possible to interpret (IA=0.73 and 0.61, respectively). The models indicated that the dispersion of 21
dust into adjacent areas was related to the variations in the renovation site concentrations (I-5: 4.3 22
± 6.9 mg/m3; I-6:9.1 ± 10.2 mg/m3) and ∆p (I-5: -0.1 Pa; I-6: 0.1 Pa). These two models also had 23
considerable problems in achieving continuous negative pressure (positive pressure period 24
occurrence I-5: 14%; I-6: 65%). Moreover, in model I-5 dust dispersion was explained by the 25
waste transfer, which was carried out from the enclosure without covering the dusty waste burden 26
in a trolley.
27 28
Uncontrolled pressure differences also explained the dust dispersion in models I-1, I-3, and I-8, 29
with high occurrences of positive pressure periods (from 28% to 66%) and negligible pressure 30
differences (from -0.4 Pa to 0.2 Pa). Nevertheless, pairwise comparison (identical working areas 31
within the same renovation site) between model I-1 with no LEV and model I-6 using LEV; models 32
I-3 and I-8; and models R-10 and R-11, respectively, indicated that the variances of the adjacent 33
area concentrations were significantly higher with no LEV than with LEV (I-1: 1.4 ± 0.36 mg/m3 34
vs. I-6: 0.89 ± 0.05 mg/m3; I-3: 1.9 ± 1.2 mg/m3 vs. I-8: 0.62 ± 0.17 mg/m3; R-10: 0.14 ± 0.05 35
mg/m3 vs. R-11: 0.08 ± 0.008 mg/m3). Table 2 also shows high variations in the observed adjacent 36
area dust concentrations for the cases without LEV. These findings indicate that using LEV 37
eventually limited the dust dispersion outside of the enclosed renovation sites.
38 39
In model R-9, no LEV was used during the chipping of concrete in a small, enclosed renovation 40
area (bathroom). The respirable dust concentration within the enclosed site was high (12 ± 5.6 41
mg/m3), but it was not a significant predictor for the dust concentration in the adjacent area (0.25 42
± 0.03 mg/m3). This result may be related to the notably higher air exchange rate (i.e., effective 43
dilution) in the enclosed site (43 h-1) compared to other models (3.1‒5.4 h-1). The low variance in 44
the adjacent area concentration also suggested that dust from the enclosed site did not disperse into 45
the adjacent area. On the other hand, the IA of 0.62 indicated that the model was possible to 46
interpret even though R2 was low (0.21). The pressure difference appeared as a significant 1
predictor in the model. The mean ∆p was only -0.2 Pa, with a positive pressure period occurrence 2
of 5.9%.
3 4 5
DISCUSSION 6
7
The findings of the present study demonstrated that the use of LEV during the dust-producing 8
tasks of the building renovation plays an important role in the dust containment especially when 9
the enclosure and negative pressurization were inadequately implemented. LEV use significantly 10
limits the dispersion of dust outside of an enclosed renovation site, with overall estimated dust 11
removal efficiency of 62% in the adjacent area. In addition, the significantly lower variances in 12
the dust concentrations in the adjacent area with LEV than with no LEV indicate that LEV is 13
effective in dust containment.
14 15
Task-specific inhalable and respirable dust concentrations were reduced by 31‒68% and 33‒90%, 16
respectively, in the adjacent area with LEV use. These measurements were carried out during the 17
highly dust-producing renovation task of chipping of tiles, in which LEV use reduced the 18
concentrations in the renovation site by 80−87% for respirable dust and by 80−85% for inhalable 19
dust. Similar task-specific LEV efficacies by 80−94% for inhalable dust (Shepherd et al., 2009) 20
and by 72–99% for respirable dust and quartz (Akbar-Khanzadeh et al., 2002; Croteau et al., 2002;
21
Echt et al., 2003; Tjoe Nij et al., 2003; Croteau et al., 2004; Young-Corbett and Nussbaum, 2009;
22
Akbar-Khanzadeh et al., 2010) have been reported during grinding, chipping and cutting of 23
concrete. In the present study, the estimated inhalable dust removal efficiency of LEV was 79%
24
within enclosure by combining all the samples collected together. Our finding support the 25
estimation by Fransman et al. (2008), as the exposure control efficacy library (ECEL) showed the 26
average estimated LEV efficacy by 82% for different types of LEV systems.
27 28
While the concentration level at the renovation site potentially contributes to observed dust 29
concentration in the adjacent areas, the dust dispersion from renovation sites to adjacent areas is 30
also highly affected by the dust containment of the enclosure and success of continuous negative 31
pressure between the renovation site and adjacent areas (Kokkonen et al., 2017). Thus, the three 32
lowest inhalable dust concentrations in the adjacent area were measured in the cases that actually 33
did not have LEV in use but were under high continuous negative pressure with airtight enclosure 34
structures. Smaller particles (respirable fraction) are more easily transferred via air currents 35
through leaks in the enclosure structures. In particular, door traffic between the renovation site and 36
adjacent areas poses a risk for dust dispersion in the absence of an airlock. Without an airlock, dust 37
may transfer from an enclosure into adjacent areas during door traffic since the negative pressure 38
is reversed to the positive for a while during door openings (Hayden et al., 1998; Tang et al., 2005).
39
The prediction by the mixed model indicated that it is beneficial to achieve low dust concentrations 40
(e.g., to adopt less dust-producing work methods or to use LEV) at the renovation site to decrease 41
the spread of dust into adjacent areas. This dust reduction is considered to be important to protect 42
the building users from biologically inert dust exposures, particularly in the case of failures in the 43
enclosure and negative pressurization. If hazardous agents (e.g., asbestos or microorganisms) are 44
present, special requirements for dust containment (EPA, 2008; Council Directive 2009/148/EC;
45
Government Degree 798/2015/MSAH)are necessary to prevent harmful dust exposure for both 1
construction workers and building users.
2 3
Regression analyses support the conclusion that reducing the dust concentration at the renovation 4
site either using LEV or with a high air exchange rate also limits the dispersion of dust into adjacent 5
areas even though the negative pressure does not meet the minimum requirement of -5 Pa 6
suggested by Kokkonen et al.(2017).A high air exchange rate at the renovation site is beneficial 7
because it effectively dilutes contaminants. Regression models indicated that dust dispersion did 8
not occur with either well-implemented dust containment or low dust concentrations at the 9
renovation site. This result was shown with an airtight enclosure and continuous negative pressure.
10
According to regression models with low dust concentrations at the renovation site, it is suggested 11
that dispersion of dust into adjacent areas is adequately prevented by keeping the mean target dust 12
concentrations in the renovation site below 4 mg/m3 for inhalable dust and below 1 mg/m3 for 13
respirable dust. According to studies by e.g., Croteau et al. (2004), Shephard et al. (2009), Meeker 14
et al. (2009), Akbar-Khanzadeh et al. (2010), and Garcia et al. (2014), the suggested target 15
concentrations are achievable by using on-tool LEV during the most dust-producing tasks, such as 16
grinding and sawing. Having the suggested data-driven target concentrations, the mean dust 17
concentration outside of the renovation sites remained below 0.4 mg/m3 for inhalable dust and 18
below 0.1 mg/m3 for respirable dust. These adjacent area concentrations are negligible considering 19
the observed health impacts of biologically inert dust (NIOSH, 2011; DFG, 2012). However, with 20
regard to the occupational health within the renovation site, the suggested values may not 21
sufficiently limit exposure to dust. Thus, appropriate personal protection equipment must be 22
considered at renovation sites.
23 24
There may have been confounding factors that could have affected the concentration levels 25
observed, such as person traffic possibly causing resuspension in the adjacent areas. A limited 26
statistical power due to small sample size was attempted to be taken into account through a 27
thorough interpretation of the case observations. In addition, each study site had its own special 28
features, which were, nevertheless, attempted to be taken into account in the statistical analyses 29
and interpretation of the results. Another common study limitation regarding studies conducted in 30
construction sites is rather small sample sizes, alike in the present study with low concentrations 31
in the adjacent areas. In addition, task-specific samples are usually short-term due to nature of 32
construction tasks. Thus, respirable dust samples are often below limit of detection as was seen in 33
the present study. According to Health and Safety Executive (2014), this challenge could be 34
managed by using a higher flow rate sampler to provide a lower limit of detection for short-term 35
sampling periods or low concentrations. On the other hand, the high flow rate samplers have been 36
reported to overestimate exposure to respirable particles compared to the respirable conventions 37
(Lee et al., 2010).
38 39 40
CONCLUSIONS AND RECOMMENDATIONS 41
42
The use of efficient dust control is critical for promoting the safety and health of both renovation 43
workers and concurrent building users. This study demonstrated that reducing dust concentrations 44
at the enclosed renovation site using LEV plays a critical role in dust containment in case of 45
failures in maintaining an enclosure continuously under negative pressure. Information obtained 46
from the study is beneficial to construction and renovation project management personnel, 1
contracting parties, and field practitioners to improve on implementation and supervision of dust 2
controls. This study provides data-driven recommendations that are of practical importance as they 3
promote healthier workplaces and policy improvements:
4 5
1. Dust dispersion into adjacent areas is prevented with an airtight enclosure (including 6
airlocks) and continuous negative pressure.
7
2. Dust containment is obtained by having target dust concentration at the enclosed 8
renovation site to below 4 mg/m3 for inhalable dust and below 1 mg/m3 for respirable dust.
9
The suggested target concentrations are achievable by using on-tool LEV during the most 10
dust-producing tasks.
11 12 13
ACKNOWLEDGMENTS 14
15
The work described in this paper was funded by the Finnish Funding Agency for Technology and 16
Innovation (grant number 40239/10). The authors thank Juhani Piirainen from the Finnish Institute 17
of Occupational Health for his technical assistance during the measurement campaigns.
18 19 20
DECLARATION 21
22
The authors declare no conflict of interest relating to the material presented in this Article. Its 23
contents, including any opinions and/or conclusions expressed, are solely those of the authors.
24 25 26
REFERENCES 27
28
Akbar-Khanzadeh F, and Brillhart R. (2002) Respirable crystalline quartz dust exposure during 29
concrete finishing (grinding) using hand-held grinders in the construction industry. Am Occup 30
Hyg; 46: 341‒46.
31 32
Akbar-Khanzadeh F, Milz S, Wagner C, Bisesi M, Ames A, Khuder , Susi P., Akbar-Khanzadeh 33
M. (2010) Effectiveness of dust control methods for crystalline quartz and respirable suspeded 34
particulate matter exposure during manual concrete surface grinding. JOEH; 7: 700‒11.
35 36
Beaudry C, Dion C, Gérin M, Perrault G, Bégin D, Lavoué J. (2013) Construction workers’
37
exposure to crystalline silica literature review and analysis. Report R-771. Bibliothèque et 38
Archives nationales du Québec (ed.), ISBN: 978-2-89631-659-5 (PDF). Montréal, Québec, 39
Canada: Institut de recherche Robert-Sauvé en santé et en sécurité du travail (IRSST). pp. 1−108.
40 41
Carlo R, Sheehy J, Feng JH, Sieber W. (2010) Laboratory evaluation to reduce respirable 42
crystalline quartz dust when cutting concrete roofing tiles using a masonry saw. JOEH; 7: 245‒51.
43 44
Croteau G, Guffey S, Flanagan M, Seixas N. (2002) The effect of local exhaust ventilation controls 1
on dust exposures during concrete cutting and grinding activities. Am Ind Hyg Assoc J; 63: 58‒
2 67.
3 4
Croteau G, Flanagan M, Camp J, Seixas N. (2004) The efficacy of local exhaust ventilation for 5
controlling dust exposures during concrete surface grinding. Ann Occup Hyg; 48: 509‒18.
6 7
Council Directive 2009/148/EC of the European Parliament and of the Council of 30 November 8
2009 on the Protection of workers from the risks related to exposure of asbestos at work.
9 10
van Deurssen E, Pronk A, Spaan S, Goede H, Tielemans E, Heederik D, Meijster T. 2014. Quartz 11
and respirable dust in the Dutch construction industry: A baseline exposure assessment as part of 12
a multidimensional intervention approach. Ann Occup Hyg; 58: 724‒38.
13 14
DFG (Deutsche Forschungsgemeinschaft). (2012) Allgemeiner Staubgrenzwert (A-Fraktion) 15
(Granuläre biobeständige Stäube (GBS)). [General limit values for dust.] [MAK Value 16
Documentation in German language, 2012]. The MAK Collection for Occupational Health and 17
Safety. p. 1–78. DOI: 10.1002/3527600418.mb0230stwd0053 (2012). [German]
18 19
Echt A, Sieber K, Jones E, Schill D, Lefkowitz D, Sugar J, Hoffner K. (2003) Control of respirable 20
dust and crystalline quartz from breaking concrete with jackhammer. Appl Occup Environ Hyg;
21
18: 491‒95.
22 23
Flanagan M, Seixas N, Becker P, Brandon T, Camp J. (2006) Silica exposure on construction sites:
24
results of an exposure monitoring data compilation project. JOEH; 3:144‒52.
25 26
Fransman W, Schinkel J, Meijster T, Van Hemmen J, Tielemans E, Goede H. 2008. Development 27
and evaluation of an exposure control efficacy library (ECEL). Ann Occup Hyg; 52: 567‒75.
28 29
Garcia A, Jones E, Echt A, Hall R. (2014) An evaluation of an aftermarket local exhaust ventilation 30
device for suppressing respirable dust and respirable crystalline quartz dust from powered saws.
31
JOEH; 11: 200‒207.
32 33
Hayden C, Johnston O, Hughes R, Jensen P. (1998) Air volume migration from negative pressure 34
isolation rooms during entry/exit. Appl Occup Environ Hyg; 13: 518‒27.
35 36
Health and Safety Executive (HSE). (2014) General methods for sampling and gravimetric 37
analysis of respirable, thoracic and inhalable aerosols. MDHS14/4, p. 1‒13. Available at 38
http://www.hse.gov.uk/pubns/mdhs/pdfs/mdhs14-4.pdf Accessed 16 January 2019.
39 40
Helsel D. (2005) Nondetects and data analysis: Statistics for censored environmental data. Scott 41
M, Barnett V, editors. Hoboken, New Jersey: Wiley Interscience. p. 1−250. ISBN 0-471-67173-8.
42 43
Klein Entink R, Fransman W, Brouwer D. (2011) How to statistically analyze nano exposure 44
measurement results: using an ARIMA time series approach. J Nanopart Res; 13: 69917004.
45 46
Kokkonen A, Linnainmaa M, Säämänen A, Lappalainen V, Kolehmainen M, Pasanen P. (2017) 1
Evaluation of real-world implementation of the partitioning and negative pressurization for 2
preventing the dispersion of dust from renovation sites. Ann Work Expo Health; 61: 68191.
3 4
Lee T, Kim SW, Chisholm WP, Slaven J, Harper M. (2010) Performance of high flow rate 5
samplers for respirable particle collection. Ann Occup Hyg; 54: 697‒709.
6 7
Meeker J, Cooper M, Lefkowitz D, Susi P. (2009) Engineering control technologies to reduce 8
occupational quartz exposures in masonry cutting and tuckpointing. Public Health Rep; 124: 101‒
9 11.
10 11
Ministry of Social Affairs and Health, Finland: Valtioneuvoston asetus asbestityön 12
turvallisuudesta. [Government Degree on the safety of asbestos work]. Vol. 798. (2015). [Finnish]
13 14
National Institute for Occupational Safety and Health (NIOSH). (2011) Current Intelligence 15
Bulletin 64: Coal Mine Dust Exposures and Associated Health Outcomes - A Review of 16
Information Published Since 1995, by DHHS (NIOSH) (Publication #2011-172). April 2011.
17 18
Shepherd S, Woskie S, Holcroft C, Ellenbecker M. (2008) Reducing quartz and dust exposures in 19
construction during use of powered concrete-cutting hand tools: efficacy of local exhaust 20
ventilation on hammer drills. JOEH; 6: 42‒51.
21 22
Tang JW, Eames I, Li Y, Taha YA, Wilson P, Bellingan G, Ward KN, Breuer J. (2005) Door- 23
opening can potentially lead to a transient breakdown in negative-pressure isolation conditions:
24
the importance of vorticity and buoyancy airflows. J Hosp Infec; 61: 283‒86.
25 26
Tjoe Nij E, Hilhorst S, Spee T, Spierings J, Steffens F, Lumens M, Heederik D. (2003) Dust control 27
measures in the construction industry. Ann Occup Hyg; 47: 211‒18.
28 29
Tuomi T, Linnainmaa M, Väänänen V, Reijula K. (2014) Application of good practices as 30
described by the NEPSI agreement coincides with a strong decline in the exposure to respiratory 31
crystalline quartz in Finnish workplaces. Ann Occup Hyg; 58: 806–17.
32 33
United States Environmental Protection Agency (EPA). (2008) Mold remediation in schools and 34
commercial buildings, by EPA (Publication #402-K-01-001). Washington, DC, September 2008.
35 36
U.S. Department of Health and Human Services. (2003) In-depth survey report of control of 37
respirable dust and crystalline quartz from breaking concrete with a jackhammer by A. Echt, K.
38
Sieber, E. Jones, et al. (Report #EPHB 282-11a). Cincinnati, Ohio. February, 2003.
39 40
Willmott, C. (1981) On the validation of models. Phys Geogr; 2: 18494.
41 42
Willmott, C. (1982) Some comments on the evaluation of model performance. Bull Amer Meteor 43
Soc; 63: 130913.
44 45
Wooldridge, J. (2013) Basic Regression Analysis with Time Series Data. Chapter 10. In the 1
Introductory Econometrics: A Modern Approach, 5th Edition. Sabatino J, Worls M, editors.
2
Michigan State University, Mason, Ohio: South-Western Cengage Learning, 2013. p. 33436.
3
ISBN-13: 978-1-111-53439-4.
4 5
Woskie S, Kalil A, Bello D, Virji M. (2002) Exposures to quartz, diesel, dust, and welding fumes 6
during heavy and highway construction. AIHA Journal; 63: 447‒57.
7 8
Young-Corbett D, and Nussbaum M. (2009) Dust control effectiveness of drywall sanding tools.
9
JOEH; 6: 385‒89.
10
