UEF//eRepository
DSpace https://erepo.uef.fi
Rinnakkaistallenteet Luonnontieteiden ja metsätieteiden tiedekunta
2017
Evaluation of Real-World
Implementation of Partitioning and
Negative Pressurization for Preventing the Dispersion of Dust From
Renovation Sites
Kokkonen A
Oxford University Press (OUP)
info:eu-repo/semantics/article
© The Authors All rights reserved
http://dx.doi.org/10.1093/annweh/wxx033
https://erepo.uef.fi/handle/123456789/2612
Downloaded from University of Eastern Finland's eRepository
1 This is a pre-copyedited, author-produced version of an article accepted for publication in 1
Annals of work exposures and health following peer review. The version of record Anna 2
Kokkonen, Markku Linnainmaa, Arto Säämänen, Vuokko Lappalainen, Mikko Kolehmainen, 3
Pertti Pasanen; Evaluation of Real-World Implementation of Partitioning and Negative 4
Pressurization for Preventing the Dispersion of Dust From Renovation Sites. Ann Work Expo 5
Health 2017 wxx033. doi: 10.1093/annweh/wxx033 is available online at:
6
https://doi.org/10.1093/annweh/wxx033 7
8 9 10 11 12 13
Evaluation of real-world implementation of partitioning and negative pressurization for 14
preventing the dispersion of dust from renovation sites 15
16 17
1Anna Kokkonen, MSc, 2Markku Linnainmaa, PhD, 2Arto Säämänen, PhD, 1Vuokko 18
Lappalainen, MSc, 1Mikko Kolehmainen, DSc (Tech.), 1Pertti Pasanen, PhD 19
20
1University of Eastern Finland, Department of Environmental and Biological Sciences, 21
P.O.Box 1627, FI-70211 Kuopio Finland, 22
2Finnish Institute of Occupational Health, P.O.Box 486, FI-33101 Tampere Finland 23
24
Address correspondence to Anna Kokkonen, University of Eastern Finland Department of 25
Environmental and Biological Sciences, P.O.Box 1627 FI-70211 Kuopio Finland. E-mail:
26
anna.kokkonen@uef.fi. Tel: +358 40 355 2856 27
28 29 30
2 ABSTRACT
1
2
Objective: The aim of this study was to assess the implementation of partitioning and the 3
negative pressure method in limiting the dispersion of dust to areas adjacent to renovation sites.
4
Methods: The pressure difference between the worksites and adjacent areas and PM10 5
concentrations in the both zones were measured in twelve renovation sites, and the factors 6
affecting the prevention of dispersion of dust were assessed.
7
Results: Poor implementation of partitioning and negative pressurization found in half of the 8
renovation sites lead difficulties in achieving a proper negative pressure, causing dispersion of 9
dust into adjacent areas. Main problems related to flimsy partitioning walls and poor air 10
tightness of the enclosure. Dust concentrations in adjacent areas were substantially lower when 11
natural ventilation in the renovation site was rejected and partitioning walls and their junctions 12
to existing structures were sealed. In case of leaky enclosures, despite the high air exchange 13
rates, a definite negative pressure could not be maintained. Instead, negative pressure minimum 14
of -5 Pa was found to be sufficient for limiting the dispersion of dust from renovation sites.
15
Conclusions: Improvement on implementation of dust controls is required through revising 16
the guidance documents, education and efficient supervision. This study revealed that the 17
current Finnish practice to implement the negative pressurization based on the air exchange 18
rate achieved with the portable exhaust fans alone is not reasonable to assure adequate dust 19
containment. Continuous negative pressure minimum of -5 Pa is suggested, and it should be 20
monitored with alarm devices throughout the renovation processes.
21 22
Key terms: renovation, dust control, partitioning, negative pressure, air tightness 23
24 25 26
3 INTRODUCTION
1 2
Renovation work has its own special features compared to constructing a new building. During 3
renovation work, many hazardous agents in addition to asbestos, such as lead, creosote or 4
microorganisms, can be present together with many other types of dust, e.g. silica and wood 5
dust. The renovation is often done in a section by section manner, while public is still working 6
in adjacent parts of the renovation site. Especially in this kind of cases, effective dust controls 7
are necessary to prevent the spread of contaminants into the adjacent areas and to ensure 8
healthy and comfortable conditions for the workers or occupants. One way to prevent spreading 9
of dust outside a renovation site is to separate the working area with partitioning and applying 10
negative pressure in the working area. Removal and demolition of asbestos containing material 11
is highly regulated (Council Directive 2009/148/EC) while other hazardous materials might 12
easily remain uncontrolled.
13 14
Guidelines for partitioning and the negative pressure method have mainly been issued for 15
asbestos demolition (SLIC 2006, OSHA 2007) or mold repair (EPA 2008). However, these 16
guidelines can also be applied for dust control in other renovation sites. In the partitioning 17
method the renovation site is separated from other spaces and the worksite is equipped with an 18
own ventilation system. Partitioning can be achieved by exploiting existing walls and room 19
division or by erecting temporary wall structures, such as plastic films, plywood or gypsum 20
boards. The enclosure needs to be constructed as air tight as possible with impervious 21
partitioning structures. All lead-throughs, ducts, vents and windows must be sealed off to 22
prevent the leakage of contaminated air outside the renovation site. HVAC systems operating 23
in renovation sites must be shut down, if possible, to avoid the contamination of the systems 24
(SLIC 2006, OSHA 2007).
25 26
Renovation sites have to be negatively pressurized with respect to the adjacent spaces, which 27
can be achieved by installing one or more portable exhaust fan units. Building an airlock to the 28
entrance will limit the spread of contaminants into the adjacent areas by controlling the air flow 29
through doorways (OSHA 2007). The exhaust air should be filtered with a fine filter or HEPA 30
(high efficiency particulate air filter) (SLIC 2006).
31 32
According to OSHA (OSHA 2007), the pressure difference (∆p) between the enclosure and 33
adjacent areas should be at least -5 Pa. In the European guidelines, a negative pressure of 34
between -5 to -20 Pa is recommended (SLIC 2006). A pressure difference of -5 Pa is, however, 35
relatively small, and can be overwhelmed by external factors, such as a strong wind (SLIC 36
2006). In Finland, the minimum negative ∆p of -5 Pa must also be achieved in asbestos 37
demolition (Government Decree 798/2015/MSAH).
38 39
There are no published data evaluating the effectiveness of pressure differences to control dust 40
spread from partitioned rooms in renovation sites. Only a few case-studies (Overberger et al.
41
1995, Rautiala et al. 1998) have reported results of limiting the dispersion of dust outside of 42
renovation sites with partitioning and the negative pressure method, but in these reports the ∆p 43
were not measured. Studies on the contamination control with the negative pressure have 44
mainly been performed in evaluating the effectiveness of airborne infection isolation rooms in 45
hospitals (Hayden et al. 1998, Rydock 2002, Rydock and Eian 2004, Tang et al. 2005, Adams 46
et al. 2011).
47 48
4 The aim of this study was to assess the implementation of partitioning and the negative pressure 1
method in Finland. Further, the study aimed to enhance the current practice for sufficient 2
pressure differences between worksites and adjacent areas to prevent dust dispersion.
3 4 5
METHODS 6
7
Study cases 8
9
The study involved sixteen cases in twelve renovation sites (Table 1) which were under 10
renovation due to moisture damage, plumbing and sanitation repair or modernization. The 11
renovation works were done according to the time schedule and design by contractors’ own 12
preferences, representing real-world implementation of partitioning and negative 13
pressurization. Information on the site operations were gathered by researcher’s observations 14
throughout the measurements. Notes were collected regarding plans for dust control, 15
performance of the partitioning, such as visible air leaks in the partitioning walls, sealing, and 16
whether or not windows and doors were open in the worksite (i.e. natural ventilation).
17 18
Dust control plans were in place in all the renovation sites, but clear criteria for the negative 19
pressurization were missing from four renovation sites (1, 68). The design of the negative 20
pressurization in cases 25 and 912 was based on two different Finnish guidelines for an air 21
exchange rate of 3‒4 h-1 or more than 6 h-1 to be achieved with the portable exhaust fans. The 22
∆p was not monitored in any of the cases.
23 24
Implementation of partitioning varied within the renovation sites (Table 1). The separation 25
walls were constructed either from the plastic films (cases 1, 6, 9 and 12), gypsum board or 26
plywood (cases 23 and 11), or the enclosure was utilizing the existing room division (cases 27
45, 78 and 10). The plastic films were attached with wooden frames. In addition, duct tape 28
was used for sealing the plastic films air tightly in cases 6 and 12, but this was not done in cases 29
1 and 9. Junctions in the gypsum board walls (cases 23) were also taped, but the plywood 30
walls in the renovation site 11 had unsealed junctions.
31 32
The ventilation system operating in the renovated section of the building was shut down in 33
cases 1 and 612. In the other study sites (cases 25), ventilation ducts were plugged since the 34
ventilation system maintained also other parts of the building. The negative pressure between 35
the renovation sites and areas adjacent to them was implemented by installing exhaust fans 36
(Table 1). The exhaust air was filtered with coarse filters (cases 2, 3 and 8), coarse and fine 37
filters (cases 45) or HEPA filters (cases 1 and 9), and led outdoors through windows. In 38
renovation sites 67 and 1012, the exhaust air was not filtered. Make-up air was taken via a 39
supply air terminal device in case 3, from the corridor via a supply air valve in cases 4 and 5, 40
and from the adjacent areas via door gaps or other uncontrolled leakages in the other renovation 41
sites. Windows and doors were left open for the entire workday to improve ventilation through 42
the worksite in eight of sixteen cases (1, 6, 8a and 912). The study sites were classified into 43
either major and minor-leak cases (Table 1) based on whether or not the renovation site had 44
substantial outward leakages. Major leakages (cases 1, 6, 8a and 912) were through the open 45
windows and doors (i.e. natural ventilation). The minor-leak cases (25, 7 and 8b) had either 46
only small (unsealed junctions or visible gaps) or no visible leaking points in the enclosure 47
structures.
48
5 Measurements and data analysis
1 2
The pressure difference and airborne particle measurements were carried out during the dust- 3
producing activities, such as chipping, cutting, and grinding activities. In three renovation sites 4
(2, 89), the assessments were carried out during more than one workday involving different 5
work phases. Therefore, the number of cases was sixteen, although the study included twelve 6
renovation sites.
7 8
The airflow of the portable exhaust fan units was measured with Swema 3000 instrument 9
connected to the hot wire anemometer SWA 31 sensor. The air exchange rate in the partitioned 10
zones [h-1] was calculated by dividing the measured airflow [m3/h] by the partition volume 11
[m3]. The airflow measurements were carried out in working conditions with existing filter 12
loading.
13 14
The ∆p between the renovation site and the adjacent area was continuously monitored with 15
Dwyer Magnesense sensor (logger HOBO U12) with an accuracy of ±1%. Readings were 16
recorded every 30 seconds to 20 minutes intervals (on average 188 seconds) during the entire 17
work days. The pressure-difference sensors were calibrated by the manufacturer before use.
18 19
The indoor air mass concentration [mg/m3] of PM10 particles and its variation were 20
simultaneously measured from the renovation site and the adjacent area with DustTrak 8533 21
(TSI Inc.) to assess the spread of dust. In each study site, the monitors were located stationary, 22
being placed as close as possible to the dust-generating activity in the work area. Outside the 23
renovation site, the monitor located in front of the main entry into the enclosure. Sampling 24
duration ranged from 48 to 550 minutes, with average of 207 minutes. Logging interval was 25
from 10 seconds to two minutes with average of 38 seconds.
26 27
The detection limit of dust monitors was 0.001 mg/m3. The results were corrected by 28
gravimetric sample (Mixed Cellulose Ester Membrane filter with diameter of 37 mm and a 29
pore size of 0.8 µm) collected by DustTrak 8533 (air flow 2.0 l/min). Correction factors varied 30
from 0.25 to 2.33 for worksite and from 0.18 to 2.06 for adjacent areas measurements.
31
Correction factors have to be determined for each measurement in order to obtain reliable 32
results (Park et al. 2009, Watson et al. 2011). The devices were calibrated annually by 33
manufacturer’s services (TSI Inc.) before the measurements.
34 35
Arithmetic mean of ∆p and PM10 particle concentrations were determined during the working 36
activities. The pressure difference data of 30 second to 20 minutes recordings were classified 37
as negative and positive pressure periods. Then the percentage of positive pressure periods ppos
38
[%] was analyzed according to number of the positive pressure recordings Npos [-] compared 39
to the total number of the recordings Nall [-].
40 41
𝑝𝑝𝑜𝑠 =𝑁𝑁𝑝𝑜𝑠
𝑎𝑙𝑙 × 100% (1)
42 43
Protection factors of the enclosures pf [-] were calculated as a ratio of the arithmetic mean 44
PM10 concentration in the renovation site Cr [mg/m3] and the concentration in the adjacent 45
area Ca [mg/m3].
46 47
𝑝𝑓 = 𝐶𝐶𝑟
𝑎 (2)
48 49
6 1
Statistical analysis 2
3
The data was analyzed statistically to determine which parameters displayed a relationship with 4
the performance of partitioning and the negative pressure method. The relationship between 5
mean ∆p, percentage of positive pressure periods, PM10 particle concentration in the worksites 6
and adjacent areas, protection factor for PM10 and air exchange rate was investigated using 7
Pearson and Spearman linear correlation coefficients. Logarithmic or arc sine square root 8
transformation of variables was done to meet the assumption of normal distribution in order to 9
utilize parametric Pearson r. However, no suitable variable transformation was found (non- 10
normal distribution) for ∆p values, thus non-parametric Spearman rho was also adopted.
11 12
A forward stepwise linear regression was conducted to explore the effect of PM10 13
concentration in the renovation site and pressure differences between the worksite and adjacent 14
area on PM10 concentration level in the adjacent area. The predictors were allowed to enter 15
the model using a P value for entry less than 0.05 and for removal greater than 0.10. The 16
differencing of the data was applied to deal with auto correlation issues of the real-time data 17
and to remove trends from the data to accomplish stationary series (Klein Entink et al. 2011).
18
Differenced PM10 concentrations in the worksite and ∆p were delayed based on observed air 19
exchange rates and the dilution of contaminants, having lags until 60 minutes in 3 minutes time 20
resolution. The regression results were analyzed in terms of the sum of the coefficients 21
(Wooldridge 2013) on initial and lagged predictors. More detailed description of the 22
differencing approach and regression analysis is presented in the supplementary materials in 23
online edition.
24 25
P value <0.05 was considered statistically significant. The statistical analyses were carried out 26
with IBM SPSS Statistics for Windows, Version 21.0 (SPSS Inc, Armonk, NY, USA).
27 28 29
RESULTS 30
31
Control of pressure difference 32
33
Mean ∆p between the renovation sites and adjacent areas ranged from a slight positive pressure, 34
0.3 Pa, to a substantial negative pressure, -48 Pa (Table 2). In most of the cases (56%), the 35
negative pressure remained close to neutral pressure (>-1 Pa). The difficulties with achieving 36
a proper negative pressure were mainly related to flimsy partitioning walls and poor air 37
tightness of the enclosure (cases 9 a‒b, 2 a‒c, Figure 1a). In addition, problems were 38
encountered in controlling pressure differences due to windows and doors being left open in 39
the worksites (renovation sites 1, 6, 8a, 912). The mean negative pressure was substantially 40
higher among the minor-leak cases with better air tightness, and they were under positive 41
pressure for only one percent of time (Table 2, Figure 1b). The airlocks in the entrances also 42
aided to achieve higher negative pressure (cases 4‒5, 8b, Table 2, Figure 1b). The importance 43
of airtightness to obtain continuous negative pressure was proved in study site 8. After the first 44
measurements (8a), the control of ∆p was improved by rejecting the natural ventilation and 45
more exhaust fans were installed (8b). Importantly, other leaking points in the enclosure 46
structures were also sealed with plywood boards and duct tape. These improvements resulted 47
in the positive pressure to change into the continuous negative pressure of -6.5 ± 2.2 Pa (Table 48
2).
49 50
7 Percentage of positive pressure periods and air exchange rate exhibited weak, statistically 1
insignificant (P value = 0.461), negative relationship. This means that high frequency of 2
positive pressure periods were observed in cases with low air exchange rate (Table 3).
3
However, pressure difference and the air exchange rate showed weak, statistically insignificant 4
(P value = 0.745), positive relationship with higher ∆p associated with higher air exchange 5
(Table 3). The major-leak cases (1, 8a, 9ab, 11) had nearly three times higher air exchange 6
(11 ± 18 h-1) than that in the minor-leak cases (2ac, 35, 7, 8b) (3.9 ± 3.2 h-1), but the mean 7
pressure difference was considerably worse for major-leaks (-0.1 ± 0.4 Pa) than for minor-leaks 8
(-8.1 ± 16 Pa). Importance of airtightness was most evident in case 1 where the air exchange 9
rate was very high, 43 h-1, yet the ∆p of only -0.1 ± 0.1 Pa was achieved. Similarly, cases 9a-b 10
had relatively high air exchange rate (5.4 h-1) but only minor pressure differences (Table 2).
11
By contrast, air tightly executed enclosures, in cases 5, 7 and 8b, with low ventilation rates of 12
0.32.8 h-1 maintained a moderate negative pressure from -3.0 Pa to -6.5 Pa. Cases 4 and 5 13
were identically partitioned, but case 4 had five times more efficient air exchange rate, which 14
resulted in over ten times higher negative pressure (Table 12).
15 16 17
Dispersion of dust 18
19
In more than half of the cases (56%, cases 1−2, 6, 9−12), dust concentrations in the adjacent 20
areas increased from the level prior the work activity after the renovation work was started 21
inside the enclosure. Protection factors of the enclosures were five times higher among the 22
minor-leak cases, being 180 ± 280, compared those with major leakages, 36 ± 47 (Table 2). In 23
particular, dispersion of dust outside the renovation was related to periods of positive pressure, 24
even though on average the worksite was under negative pressure during the follow-up time 25
(Table 2, Figure 1a). An example is the poorly executed dust control case 9b; during the work 26
phases, PM10 concentration in the adjacent area was six times higher compared to the dust 27
level prior the work was started. At the same time, mean negative pressure was negligible and 28
positive pressure periods occurred during a fourth of the follow-up time (Figure 1a). Door 29
traffic (no airlock) impacted on the dispersion of dust outside the enclosure (0.31 ± 0.18 mg/m3, 30
Table 2). In contrast, in the well-executed case 4, the dust concentration in the adjacent area 31
remained in the same order of magnitude during the follow-up (0.05 ± 0.01 mg/m3, Table 2).
32
Negative pressure remained continuously substantial. Fluctuation of pressure was related to 33
door traffic (Figure 1b).
34 35
Both ∆p and positive pressure periods between the renovation and adjacent areas correlated 36
positively with dust level outside the worksite. This means that high ∆p (positive pressure) and 37
positive pressure occurrence were associated with high dust concentration in the adjacent area 38
(Table 3). However, the correlations were not statistically significant (P values 0.398 and 39
0.103, respectively). Air exchange rate of the renovation site had as well a weak, insignificant, 40
positive correlation with dust concentration in the adjacent area. Cases with major-leaks (1, 6, 41
8a, 9a‒b, 11‒12) had altogether twofold higher air exchange rate than that in the minor-leak 42
cases (2a‒c, 3-5, 7, 8b), but the dust level outside the renovation site was on average tenfold 43
higher (Table 2). In addition, a weak, insignificant, negative correlation was found between the 44
percentage of positive pressure periods and the protection factor (P value = 0.371). Both ∆p 45
and air exchange rate, however, had only very weak, insignificant (P values 0.913 and 0.765) 46
positive relationship with the protection factor (Table 3).
47 48
Table 4 shows the effect of worksite dust level and ∆p on the dust concentration in the adjacent 49
area derived from the forward stepwise regression analysis. Table presents statistically 50
8 significant predictors in each case and the sum of their standardized coefficients (β), index of 1
agreement (IA) and coefficient of determination (R2). Index of agreement describes the degree 2
to which the observed variate is accurately estimated by the predicted variate, being ideal for 3
making cross-comparisons between models (Willmott 1981 and 1982). IA varies between 0.0‒
4
1.0, value 1.0 indicating perfect agreement between the observed and predicted value and 0.0 5
total disagreements. If the IA is above 0.4, the model goodness is better than random. More 6
details of the IA and interpretation of the standardized coefficients with relation to the original 7
data are presented in the supplementary materials in online edition. Also, visualization of the 8
differencing approach (Figures S1-S7) and detailed regression model summary (Table S1) are 9
shown in the supplementary materials.
10 11
When the observed ∆p declines, the change in the differenced data is positive. Thus, the 12
interpretation of the negative standardized regression coefficient (β) of the ∆p indicates that 13
with lower observed pressure differences (i.e. negative pressure), the observed PM10 14
concentration in the adjacent area is also lower (Table 4). Instead, β for worksite PM10 15
concentration is more straightforward, i.e. positive coefficient means that high dust level in the 16
worksite is associated with higher dust level in the adjacent area. Since the purpose of 17
regression analysis conducted here was not to do absolute predictions, but to analyze which 18
factors were related to dust containment, results are not discussed in terms of the exact 19
standardized coefficient values.
20 21
In regression models (Table 4) for air tightly partitioned cases 4 and 8b with a definite, 22
continuous negative pressure (-48 ± 4.9 Pa and -6.5 ± 2.2 Pa, Table 2), only the worksite dust 23
concentration was significantly associated with the dust level in the adjacent area. IA (0.38) in 24
case 4, however, showed that the model goodness was worse than random. Moreover, the 25
negative β-value in case 4 illustrated that with lower worksite dust level the concentration in 26
the adjacent area was actually higher. Model for case 8b included various (yet significant) 27
delayed worksite concentrations (see supplementary Table S1 in online edition), which had a 28
positive relationship with the dust level in the adjacent area. This resulted in high R2 (0.73), 29
therefore, it was suspected that the model was over-predicting the outcome. On the other hand, 30
IA of 0.61 exhibited that the model was possible to interpret. There was a lot of variation in the 31
worksite dust levels, i.e. numerous “peaks” in the data, which may have been related to the 32
high number of significant PM10 lags in the model. Regardless, models of cases 4 and 8b 33
suggested that continuous negative pressure (< -5 Pa) was not significant in explaining the dust 34
dispersion outside the renovation site, that is, implementation of partitioning and negative 35
pressurization in these cases achieved adequate dust containment. As for the minor-leak case 36
2b, which had a slight mean negative pressure (-0.4 ± 0.2 Pa, Table 2), 55% of the dust level 37
in the adjacent area was explained by the ∆p and worksite dust concentration, i.e. high PM10 38
concentration outside the renovation site was related to high dust level in the worksite and 39
positive pressure periods. This case had minor leaking points between the suspended and solid 40
ceiling making it difficult to achieve better negative pressure, and furthermore, to achieve better 41
dust containment.
42 43
Factors related to achievement of better dust containment were well shown with cases 8a and 44
8b (Table 4). Improvements made to partitioning structures and negative pressurization (i.e.
45
rejecting natural ventilation, sealing of leaking points in the enclosure, installation of more 46
exhaust fans in the worksite), resulted in that regression model of case 8a (prior the 47
improvements renovation site was constantly under positive pressure) had ∆p as a significant 48
predictor. On the contrary, pressure difference had no effect on the dust dispersion in the model 49
9 after the improvements. Instead, worksite dust level, being much higher (7.4 ± 5.4 mg/m3, 1
Table 2) in this case, was explaining the dust level in the adjacent area.
2 3
Alike in the model of case 4, the worksite dust level in the minor-leak case 3 had a negative 4
relationship with the dust concentration in the adjacent area, i.e. low worksite concentration 5
were associated with high concentration outside the renovation site. This unexpected result is 6
understandable due to low PM10 concentrations both in the worksite and adjacent areas (0.7 ± 7
0.5 mg/m3 and 0.02 ± 0.17 mg/m3). Similarly in case 8a, the dust concentrations were low (1.3 8
± 0.8 mg/m3 and 0.01 ± 0.001 mg/m3). Both models explained poorly the PM10 concentrations 9
in the adjacent areas.
10 11
Regarding models of cases 9b and 11 with leaky partitioning and problems in controlling the 12
pressure differences, both ∆p and worksite dust concentration were explaining the dispersion 13
of dust in the adjacent area with high positive pressure difference and high concentration in the 14
work site. These models also reached the highest IA-values (0.67 and 0.76).
15 16 17
DISCUSSION 18
19
This study confirmed that airtightness of partitioning structures and the continuous 20
maintenance of negative pressure are important for dust control efficiency to the adjacent areas.
21
In general, partitioning and negative pressurization had been inadequately designed and poorly 22
implemented in half of the studied renovation sites. Problems were related to unsealed 23
installation of the partitioning walls (e.g. plastic films) leading to visible gaps between the 24
enclosure and adjacent areas. The junctions between the temporary and existing structures were 25
found to be the critical points leading to air leakages. Managers were not always aware of the 26
importance of proper sealing and continuous maintenance of negative pressure in the dusty 27
working area. Performance of the partitioning structures was ignored.
28 29
The importance of a properly executed partitioning in order to achieve adequate dust 30
containment was well shown in one study site, in which improving the airtightness of the 31
enclosure led to a constant, moderate negative pressure. Improvements included rejecting 32
natural ventilation in the worksite, sealing of partitioning walls, and making the negative 33
pressurization more effective by installation more exhaust fans. Overall, sealing of the 34
partitioning structures in order to obtain airtight enclosure exerted a considerable effect on the 35
maintaining a continuous negative pressure, and therefore, preventing the dispersion of dust 36
outside the worksite. For example, plastic partitioning walls can result in adequate dust 37
containment as long as their installation and maintenance is done in an airtight manner. In 38
practice, this simply means sealing (with duct tape) all the junctions of the partitioning 39
structures. Rautiala et al. (1998) also reported that the implementation of an enclosure has a 40
significant role in preventing the dispersion of contaminants outside a renovation site.
41
Microorganism concentrations inside the renovation sites were high during demolition, but 42
concentrations in adjacent areas remained at the initial level.
43 44
None of renovation sites had set requirements for negative pressure. The design of negative 45
pressurization was based on the air exchange rate achieved with the portable exhaust fans.
46
Surprisingly, pressure differences were not monitored in any of the cases. Failures in 47
maintaining negative pressurization were more common if the adjusted ∆p was close to zero.
48
The disturbances caused by pressure pulses due to e.g. door openings or piston flows by 49
elevators could ruin the negative pressurization. The opening of windows and doors inside an 50
10 enclosure also made it difficult to achieve and maintain the negative pressure. The mean 1
pressure difference did not alone account for whether or not there was dispersion of dust.
2
Instead, a moderate positive correlation between the dust concentrations in the adjacent areas 3
and the occurrence of positive pressure periods verifies that the more time the renovation site 4
is under positive pressure, the more dust is likely to spread to adjacent areas.
5 6
Despite of the low number of cases with an airlock, our findings give an indication that the 7
presence of an airlock does improve the containment of an enclosure. The airlock helps to 8
maintain the ∆p between the renovation site and adjacent area. Overberger et al. (1995) also 9
observed that a negatively pressurized enclosure combined with an airlock between the 10
renovation site and the adjacent area limited the spread of particles outside the worksite.
11
However, the actual ∆p was not confirmed in that study. The total suspended particulate 12
concentration in the adjacent area ranged from close to the baseline level of 0.1 mg/m3 13
measured before the renovation to a maximum of 0.3 mg/m3 whereas the dust concentration 14
was 2.0 mg/m3 inside the worksite. Similarly, present study showed that PM10 concentrations 15
outside the enclosures with airlocks remained in the same order of magnitude through the 16
workdays. In addition, other studies (Hayden et al. 1998, Adams et al. 2011, Kokkonen et al.
17
2014) dealing with the negatively pressurized airborne infection isolation rooms (AIIRs) of the 18
hospitals, have demonstrated that airlocks substantially limit the dispersion of contaminants 19
into the adjacent areas during door traffic. Adams et al. (2011) stated that the leakage of 20
contaminants outside an AIIR was less than 0.1% whereas Kokkonen et al. (2014) reported a 21
leakage of around 4%. Containment of the AIIR improved at the ∆p under -2.5 Pa with door 22
traffic; however, this effect was not statistically significant (Adams et al. 2011). It is known, 23
that the negative pressure is reversed to the positive for a short time during door traffic, 24
allowing contaminants to spread between the spaces (Hayden et al. 1998, Tang et al. 2005).
25
Based on previous studies and obtained results, it is recommended to build an airlock between 26
the worksite and the adjacent area to control the pressure differentials during door traffic; and 27
furthermore, to limit the dispersion of dust from the renovation site. Moreover, a doormat at 28
the entrance to the worksite aids in the dust control.
29 30
Ventilation rate of the worksite, a considerable factor on dust control, correlated positively with 31
both pressure difference and dust level in the adjacent area, on contrary to expectation and 32
current practice in Finnish renovation sites. This is explained by the high number of those 33
renovation sites, that had major outward leaking points in their enclosures. Regardless, the 34
results indicate that with a leaky enclosure, a continuous, definite negative pressure between 35
the renovation site and adjacent area cannot be achieved. In addition, a high air exchange rate 36
in the worksite does not limit dust dispersion outside the renovation site if there are leaking 37
points in the enclosure. On the other hand, efficient air exchange has a beneficial effect in the 38
renovation site since it causes the dilution of contaminants from the working area. Thus, the 39
contaminant concentrations in the renovation site also contribute to that in the adjacent area.
40 41
Regarding the negative pressurization guidelines, it has been unclear which negative pressure 42
is adequate to prevent the dispersion of dust. Our regression analysis results dealing with 43
airtight enclosures, which had a negative pressure minimum of -5 Pa, support that the negative 44
pressure guideline of -5 Pa (SLIC 2006, OSHA 2007) is sufficient in order to achieve adequate 45
dust containment. Overall, regression model goodness parameters, index of agreement and 46
coefficient of determination, were rather low. This was explained by the fact that only ∆p and 47
worksite dust level were included as predictors for the dust dispersion, but also other factors, 48
such as normal operation and traffic around renovation sites, have likely been influenced the 49
measured dust concentrations in adjacent areas. However, we suggest that the target value for 50
11 a moderate continuous negative pressure should be between -5 to -15 Pa to prevent outward 1
leakage in times of pressure fluctuation due to door openings, elevator traffic or ventilation 2
changes in the adjacent areas. A higher level of negative pressure may weaken the functional 3
properties of temporary partitioning structures, especially those constructed of plastic films.
4 5
Regression analysis results dealing with airtight partitioning also referred that the dispersion 6
of dust outside the renovation site can be limited by allowing low dust levels in the worksite.
7
This means by e.g. adopting working methods, which are producing less dust and using local 8
exhaust ventilation systems to capture dust at source.
9 10
The current Finnish practice to design the negative pressurization for ordinary, non-asbestos 11
renovation sites is based on the air exchange rate achieved with the portable exhaust fans. This 12
study indicate that this approach is not sufficient to assure dust containment. Instead, guidelines 13
should be based on the control of contaminants within the partitioned renovation site and 14
maintaining continuous negative pressure between the renovation and adjacent areas. It is 15
recommended that the pressure difference should be monitored all the time.
16 17 18
CONCLUSIONS 19
20
In conclusion, poor implementation of partitioning and negative pressurization found in half of 21
the studied renovation sites emphasizes the need for revising the guidance documents, as well 22
as the requirement for education. Improvement on implementation of dust controls demands 23
good co-operation between contracting parties. This means e.g. better designing of the 24
partitioning, high-quality installation as well as efficient supervision and continuous control of 25
the negative pressurization. These requirements should be appointed to all contracting parties.
26
It is further concluded that current Finnish partitioning and negative pressurization practices 27
are not sufficient to provide adequate dust containment. Unlike present approach, negative 28
pressurization should not alone be based on the air exchange rate of portable exhaust fans but 29
also control of negative pressure is important. Design and implementation of negative 30
pressurization should be based on requirement minimum of -5 Pa in the working area, which 31
is in agreement with international guidelines. Practical implications are that the condition of 32
the partitioning structures should be checked regularly and all lead-throughs have to be 33
carefully sealed for tight enclosure. Also, an airlock between the worksite and the adjacent area 34
to control the pressure differentials and to limit the dispersion of dust during door traffic is 35
recommended. Importantly, the negative pressurization needs to be monitored with alarm 36
devices throughout the entire renovation processes in order to avoid dust spreading positive 37
pressure events.
38 39 40
SUPPLEMENTARY DATA 41
42
Supplementary data are available at Annals of Work Exposures and Health online.
43 44 45
ACKNOWLEDGMENT 46
47
The work described in this paper was funded by the Finnish Funding Agency for Technology 48
and Innovation (grant number 40239/2010). Emeritus professor Pentti Kalliokoski is 49
12 acknowledged for his collegial review of the manuscript. The authors thank Ewen MacDonald 1
Pharm.D for linguistic editing of this paper.
2 3 4
DECLARATION 5
6
The authors declare no conflict of interest relating to the material presented in this Article. Its 7
contents, including any opinions and/or conclusions expressed, are solely those of the authors.
8 9 10
REFERENCES 11
12
Adams N, Johnson D, Lynch R. (2011) The effect of pressure difference and care provider 13
movement on airborne infectious isolation room containment effectiveness. Am J Infect 14
Control; 39: 9197.
15 16
Council Directive 2009/148/EC of the European Parliament and of the Council of November 17
2009 on the Protection of workers from the risks related to exposure to asbestos at work.
18 19
Hayden C, Johnston O, Hughes R, Jensen P. (1998) Air volume migration from negative 20
pressure isolation rooms during entry/exit. Appl Occup Environ Hyg; 13: 51827.
21 22
Klein Entink R, Fransman W, Brouwer D. (2011) How to statistically analyze nano exposure 23
measurement results: using an ARIMA time series approach. J Nanopart Res; 13:69917004.
24 25
Kokkonen A, Hyttinen M, Holopainen R, Salmi K, Pasanen P. (2014) Performance testing of 26
engineering controls of airborne infection isolation rooms by tracer gas techniques. Indoor 27
Built Environ; 23: 9941001.
28 29
Ministry of Social Affairs and Health, Finland. (2015) Government Degree on the safety of 30
asbestos work. Vol. 798. Finland: Ministry of Social Affairs and Health. Ministry of Social 31
Affairs and Health, Finland.
32 33
Occupational Safety & Health Administration (OSHA). (2007) Safety and health regulations 34
for construction, Subpart Z – Toxic and hazardous substances, Asbestos. 29CFR, Part 35
1926.1101, XVII (7-1-07 Edition). Washington, DC: OSHA. p. 52692.
36 37
Overberger P, Wadowsky R, Schaper M. (1995) Evaluation of airborne particulates and fungi 38
during hospital renovation. AIHAJ; 56: 70612.
39 40
Park JM, Rock J, Wang L, Seo YC, Bhatnagar A, Kim S. (2009) Performance evaluation of 41
six different aerosol samplers in particulate matter generation chamber. Atmos Environ; 43:
42
28089.
43 44
Rautiala S, Reponen T, Nevalainen A, Husman T, Kalliokoski P. (1998) Control of exposure 45
to airborne microorganisms during remediation of moldy buildings; report of three case 46
studies. AIHAJ; 59: 45560.
47 48
Rydock J. (2002) A simple method for tracer containment testing of hospital isolation rooms:
49
Appl Occup Environ Hyg; 17: 48690.
50
13 1
Rydock J, Eian P. (2004) Containment testing of isolation rooms. J Hosp Infect; 57: 22832.
2 3
Senior Labour Inspectors Committee (SLIC). (2006) A practical guide on best practice to 4
prevent or minimize asbestos risks in work that involves (or may involve) asbestos: for the 5
employer, the workers and the labour inspector. Brussels: European Commission.
6 7 8
Tang J, Eames I, Li Y, Taha Y, Wilson P, Bellingan G, Ward K, Breuer J. (2005) Door-opening 9
motion can potentially lead to a transient breakdown in negative-pressure isolation conditions:
10
the importance of vorticity and buoyancy air flows. J Hosp Infect; 61: 28386.
11 12
United States Environmental Protection Agency (EPA). (2008) Mold remediation in schools 13
and commercial buildings. EPA 402-K-01-001. Washington, DC: EPA.
14 15
Watson J, Chow J, Chen L, Wang X, Merrifield T, Fine P, Barker K. (2011) Measurement 16
system evaluation for upwind/downwind sampling of fugitive dust emissions. AAQR; 11:
17
33150.
18 19
Willmott C. (1981) On the validation of models. Phys Geogr; 2: 18494.
20 21
Willmott C. (1982) Some comments on the evaluation of model performance. Bull Amer 22
Meteor Soc; 63: 130913.
23 24
Wooldridge J. (2013) Introductory Econometrics: Basic regression analysis with time series 25
data. Chapter 10. In Introductory Econometrics: A Modern Approach, 5th edn. Joe Sabatino, 26
Michael Worls, eritors. Mason, Ohio, USA: Michigan State University. South-Western 27
Cengage Learning. ISBN-13: 978-1-111-53439-4. p. 33436.
28
14
Table 1. Renovation sites, execution of partitioning, ventilation details and classification of study sites into minor-leak and major-leak cases
Renovation site Enclosure
construction Airlock Make-up air Airflow [m3/h]
Air exchange [h-1]
Filters of the portable
exhaust fan units LEVa Classification
1 Bathrooms in an apartment house plastic no uncontrolled 520 43 HEPA13 no major-leak
2b Hospital, unit 1 gypsum boards no uncontrolled 2530 3.1 coarse filter yes minor-leak
3 Hospital, unit 2 gypsum boards no controlled 3020 5.5 coarse filter yes minor-leak
4 School 1, class room 1 existing walls yesc controlled 2100 11 coarse filter and F7 no minor-leak 5 School 1, class room 2 existing walls yesc controlled 510 2.2 coarse filter and F7 no minor-leak
6 School 2 plastic no uncontrolled 5370 9.4 - no major-leak
7 University 1, office premises 1 existing walls no uncontrolled 540 0.3 - no minor-leak
8a University 1, office premises 2 existing walls yesd uncontrolled 3360 1.1 coarse filters no major-leak 8be University 1, office premises 2 existing walls yesd uncontrolled 8630 2.8 coarse filters no minor-leak
9f University 2, office premises 1 plastic no uncontrolled 4100 5.4 HEPA13 yes major-leak
10 University 2, office premises 2 existing walls no uncontrolled - - - yes major-leak
11 University 2, office premises 3 plywood no uncontrolled 2500 1.7 - yes major-leak
12 University 2, office premises 4 plastic no uncontrolled 1250 0.8 - yes major-leak
a Local exhaust ventilation
b Number of measurements (i.e. cases 2a, 2b, 2c) N=3 involving different work activities
c Sealed airlock made of plywood divided into three sections
d Sealed airlock made of plastic films and wooden frames
e After the first measurements, the enclosure was improved by keeping windows and doors closed and more exhaust fans were installed
f Number of measurements (i.e. cases 9a, 9b) N=2 involving different work activities - No fan or filter
1
15
Table 2. Arithmetic mean (AM) and standard deviation (SD) of pressure difference (∆p), percentage of positive pressure periods (ppos), PM10 concentration in the worksite (PM10w) and adjacent area (PM10a), and protection factor (pf) in each study case.
Case AM ∆p [Pa]
SD ∆p [Pa]
ppos [%]
AM PM10w
[mg/m3]
SD PM10w
[mg/m3]
AM PM10a [mg/m3]
SD PM10a
[mg/m3]
pf [-]
1 -0.1 0.1 6.4 17 10 0.20 0.12 85
2a -0.4 0.2 1.4 4.0 3.0 0.01 0.01 400
2b -0.4 0.2 0.8 3.4 6.0 0.07 0.02 49
2c -0.3 0.4 0 1.1 1.2 0.05 0.01 22
3 -0.7 0.2 0 0.7 0.5 0.02 0.17 35
4 -48 4.9 0 1.2 1.5 0.05 0.01 24
5 -3.0 0.8 0 1.0 0.8 0.08 0.06 13
6 - - - 0.8 0.4 0.50 0.52 1.6
7 -5.4 1.4 0.6 - - 0.03 0.05 -
8a 0.3 0.1 100 1.3 0.8 0.01 0.001 130
8b -6.5 2.2 0 7.4 5.4 0.01 0.01 740
9a 0.2 0.6 63 19 15 1.1 1.0 17
9b -0.4 0.6 24 6.0 4.6 0.31 0.18 19
10 - - - 1.2 1.2 0.17 0.08 7.1
11 -0.6 1.1 26 0.9 1.0 0.03 0.01 30
12 - - - 0.5 0.3 0.63 0.57 0.8
1 2
16
Table 3. Pearson and Spearman linear correlations between ventilation and dust parameters
Parameters 2 3 4 5
Pressure difference * 0.256 0.035 0.100
2 Percentage of positive pressure periods - 0.454 -0.271 -0.215
3 PM10 concentration in the adjacent area - * 0.303
4 Protection factor - 0.088
5 Air exchange -
* Since the parameters were based on the same measurement,correlations were not investigated
1 2
17
Table 4. Significant predictors of differenced PM10 concentration in the adjacent area derived from Forward Stepwise Regression in each study case. Sum of standardized coefficients (β), index of agreement (IA) and coefficient of determination (R2)
Casea βb IA R2
∑ PM10worksite ∑ ∆p
minor-leak
2b 0.909 -0.283 0.55 0.545
3 -0.246 -0.594 0.32 0.136
4 -0.801 0.38 0.426
8b 0.919 0.61 0.732
major-leak
8a -0.323 0.44 0.104
9bc 0.309 -0.706 0.67 0.324
11 1.058 -0.232 0.76 0.567
a The number of lagged observations was sufficient to carry out the regression for cases 2b, 3, 4, 8a-b, 9b and 11.
b Regression model summary (detailed lags of differenced PM10 concentrations in the worksite and differenced pressure differences) and their standardized coefficients are presented in the supplementary materials (Table S1).
c In the initial forward stepwise multiple regression model, the IA was only 0.31 and R2 0.50. This model included two questionable lagged worksite concentrations, which after visualization of the observed and predicted concentration, were reasonable to remove. The model was run again by the Enter –method.
1 2
18 a.
1
2 3
b.
4
5
Figure 1. (a) Well-executed dust control case 4. (b) Poorly executed dust control case 9b.
6
