Comparison of Portable, Real-Time Dust Monitors Sampling Actively, with Size-Selective Adaptors, and Passively
Health and Safety Laboratory, Harpur Hill, Buxton, Derbyshire SK17 9JN, UK
* Author to whom correspondence should be addressed. Tel: +44-1298-218527; fax: +44-1298-218392; e-mail: andrew.thorpe{at}hsl.gov.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
The performance of three, portable, real-time dust monitors was investigated inside a calm air dust chamber for a range of industrial dusts and two sizes of aluminium oxide dust. The instruments tested were the Split 2 (SKC Ltd), Microdust Pro (Casella Ltd) and DataRam (Thermo Electron Ltd), which sampled either passively or actively by connecting a manufacturer-supplied, size-selective adaptor and an air sampling pump to the inlet of the monitor. Two size-selective adaptors were tested with the Split 2: the GS-3 cyclone adaptor and the Institute of Occupational Medicine (IOM) inlet with porous foam inserts. Similarly, two size-selective adaptors were tested with the Microdust Pro: the Higgins–Dewell cyclone adaptor and the conical inhalable sampler (CIS) adaptor with porous foam inserts. The DataRam was tested with a GK 2.05 cyclone adaptor since there was no porous foam adaptor available. The instruments' responses were compared with the reference dust samplers: Casella Higgins–Dewell cyclone for the respirable fraction and IOM sampler for the inhalable fraction. The response of the dust monitors was found to be linear with respirable dust concentration when operated either passively or actively using the cyclone size-selective inlets. Their responses were, however, lower when operated actively with the cyclone adaptors compared to the passive operation and lower still when used with the porous foam inserts. There was also often more scatter in the porous foam measurements, attributable to variable clogging of the foams caused by inconsistent loading with dust. The dust monitor responses were sensitive to changes in particle size when operated passively but much less so in active mode with the cyclone adaptors. The Microdust Higgins–Dewell cyclone adaptor measurements agreed closely with the reference respirable concentration for all dusts, whereas those for the DataRam GK 2.05 and Split 2 GS-3 cyclone adaptors were different to the reference. Concentrations measured with the foam adaptors were considerably lower than both the reference cyclone samplers and the dust monitor cyclone adaptors and increasingly undersampled as they became loaded with dust. Inhalable dust measured with the Split 2 IOM adaptor agreed closely with the reference IOM inhalable samplers, whereas the Microdust CIS adaptor underestimated the inhalable concentration compared to the reference.
Keywords: active sampling • cyclone • inhalable dust • passive sampling • porous foam • real-time dust monitors • respirable dust
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Real-time (direct-reading) dust monitors are used by occupational hygienists for many different applications such as walk-through surveys, background sampling, site dust measurements, assessment of the effectiveness of dust control systems and measurement of indoor air quality (Maynard and Jensen, 2001). They are also used as part of an exposure visualization system (Rosén et al., 2005) to identify high levels of dust generated by poor work practice, in the investigation of dust control techniques, and to generate hygiene training information. The main advantage of dust monitors is that they give an instantaneous measure of airborne dust concentration, thereby reducing considerably the time and effort associated with standard gravimetric methods (HSE, 2000). When properly calibrated, they can give an accurate measure of respirable dust, i.e. that which enters the mouth and nose and passes to the lower regions of the respiratory system (Thorpe and Walsh, 2002). Previous work has also looked at a selection of dust monitors to see if they can be used to measure the inhalable fraction of airborne dust clouds (Thorpe, 2007). They were all found to greatly underestimate the inhalable concentration, although the linearity was good over a wide range of concentrations. Their calibration was found to be particularly sensitive to changes in particle size and monitor response decreased rapidly with increasing particle size. This was not surprising since all the instruments tested were of the light-scattering type, which scattering theory predicts are sensitive to dust particles primarily in the respirable size range.
Careful calibration of real-time dust monitors is therefore very important if accurate, quantitative measurements of airborne dust concentration are required. Typically, the monitors are calibrated initially in the factory using a ‘standard’ test dust and are usually adjusted to agree with respirable dust concentration measurements made using reference methods (HSE, 2000). However, it is highly unlikely that the standard dust will exhibit the same light-scattering characteristics as the dust being measured in practice and so a separate calibration should be carried out each time the dust monitor is exposed to a different dust. During calibration in the field, a reference sampler (e.g. a Higgins–Dewell cyclone sampler for respirable dust concentration measurements) is placed close to the dust monitor and sampling takes place for long enough to collect a weighable sample on the filter. The dust monitor is programmed to concurrently log the concentration at a set sampling rate for the duration of the test. At the end of the calibration, the average concentration measured using the dust monitor is compared to the gravimetric method. This yields a calibration factor that can be used to scale the dust monitor readings to give a ‘true’ result, and in some cases can be programmed into the instrument to automatically adjust subsequent measurements. For some dust monitors, this is the only method available by which they can be calibrated. However, a selection of dust monitors that normally sample the dust passively (by using the natural movement of the surrounding air to introduce dust into the inlet) can be made to sample the dust actively. Here, a small sampling pump is used to draw the air through a size-selective adaptor that fits over the inlet of the dust monitor. The sampled dust then passes through the monitors' detection zone and is collected onto a backup filter, which can be used to calibrate the monitor. The size-selective adaptor usually consists of a cyclone or a porous foam plug to ensure that the correct size fraction is sampled. Three examples are the Microdust Pro (Casella CEL), the Split 2 (SKC Ltd) and the DataRam (Thermo Scientific). The former two can be fitted with cyclone and porous foam inlet adaptors, whereas the latter can only be fitted with a cyclone adaptor.
The main aim of this study was to provide guidance for users of real-time dust monitors when operated both passively and actively, specifically
- to compare the performance of the three real-time dust monitors, operating passively and actively, with gravimetric reference samplers positioned alongside, based on the determination of calibration factors (defined as the ratio of the reference concentration measurement to the monitor response) for each mode of operation and
- to compare the performance of the on-board respirable size-selective sampler (cyclone or porous foams) with a reference Casella Higgins-Dewell cyclone sampler (HSE, 2000).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Real-time monitors
Each of the monitors tested could be adapted to convert from passive to active operation. Only the appropriate manufacturer-supplied converters for the particular monitor were used. For the Split 2 and Microdust Pro dust monitors, this involved either the use of a size-selective cyclone attached to the instrument inlet or the use of calibrated polyurethane porous foams to allow the respirable size fraction into the instruments' light-scattering detection zone. The DataRam uses only a cyclone for size selection. For each instrument, the aerosol is drawn through the size-selective inlet using a small sampling pump and the dust is collected onto a filter for subsequent gravimetric analysis. Each dust monitor uses a different type of cyclone as the size selector, which is designed or can be adjusted to measure the ACGIH–CEN–ISO definition of respirable dust (CEN, 1993).
SKC Split 2.
This instrument can be operated passively so that the natural movement in the surrounding air introduces particles into the sensing chamber, which is detached from the main body of the instrument. It can also be configured to sample actively by combining the sampling head with a personal sampling pump and a cyclone inlet (for sampling the respirable fraction) or IOM inlet (for sampling the inhalable fraction) for concurrent dust monitoring and sampling. The IOM inlet can also be fitted with calibrated polyurethane foam plugs to selectively sample the respirable or thoracic fraction of the dust. The cyclone supplied by SKC is the GS-3—a 10-mm, lightweight, multiple-inlet sampler made from conductive plastic. It is designed to meet the ACGIH–CEN–ISO respirable size-selection curve. It has a 50% cut-point of 4.0 µm when operated at a flow rate of 2.75 l min–1. The GS-3 uses an adaptor to attach it to the IOM inlet of the SPLIT 2's sampling head.
A conducting plastic cassette containing a 25-mm filter is fitted in the rear of the sampling head to capture the particles that pass through the measuring zone and is used to determine the reference gravimetric dust concentration. In previous tests (Thorpe, 2007), the cassette was found to leak dust around the seal so that not all the sampled dust was collected onto the backup filter. This resulted in an underestimation of the reference dust concentration. SKC have since improved the design and now incorporate a screw-on type filter holder based on a standard IOM body and filter cassette which results in a much more secure fit.
The Split 2 is calibrated initially by the manufacturer using SAEJ726 Fine Standard test dust (ISO 12103 Pt 1, 1997). This is largely composed of silica and has a size distribution similar to Arizona road dust (ARD)—another commonly used calibration dust. It is also supplied with a reference calibration element to carry out a single-point check of the factory calibration.
Thermo Scientific DataRam pDR—1000.
The sensing chamber is an integral part of the instrument, which is located underneath a rubber-protected hood through which the dust passes. The rubber hood can be removed and replaced with a size-selective cyclone inlet, which effectively converts the instrument to the 1200 model for active particulate sampling. The cyclone is a BGI model GK 2.05 that has well-defined particle separation characteristics and is one of a family of cyclones (Kenny and Gusmann, 1996). By operating the sampling pump at specific flow rates, the cyclone provides precisely defined particle size cuts. For example, at 4 l min–1, the quoted cut-point of the cyclone is 2.5 µm as required for PM2.5 monitoring. At 2.65 l min–1, the computed cut-point is 4 µm as required for respirable monitoring. Kenny and Gusmann (1996) demonstrated that the GK 1.52X and the GK 2.69 cyclones when operated at 2.2 and 4.2 l min–1, respectively, showed good agreement with the respirable sampling convention (CEN, 1993). Since the GK 2.05 is from the same family of cyclones, it should also agree well with the convention when operated at 2.65 l min–1. Gravimetric calibration of the instrument is accomplished under field conditions using an integral filter located downstream of the photometric sensor. Unlike the Split 2, there is no physical method of adjusting the zero or checking the calibration. Like the Split 2, the DataRam is also calibrated by the manufacturer to measure SAEJ726 Standard test dust.
Casella Microdust Pro.
The sensing head of this monitor is a detachable cylindrical measurement wand. It is a non-pumped monitor that relies on the ambient movement of the surrounding air or the movement of the wand through the air to introduce dust into the sensing zone. The dust enters and leaves through a hole in the side of the probe. The monitor is supplied factory calibrated against total suspended particulate (TSP) when challenged with a test aerosol of SAEJ726 Fine Standard test dust. This is in contrast to other dust monitors (including the Split 2 and DataRam) that are calibrated against the respirable fraction using reference gravimetric cyclone samplers, usually inside a calm air dust box. Like the Split 2 monitor, the ‘factory’ calibration setting can be checked at any time using a calibration insert, which effectively carries out a single-point calibration check.
The Microdust is converted to an active sampler by attaching an in-line adaptor over the wand so that it covers the inlet holes. There are adaptors that will collect the respirable fraction based on the Higgins–Dewell cyclone, and the inhalable fraction based on the conical inhalable sampler (CIS). The dust is drawn into the adaptor using a personal sampling pump set to the correct flow rate for the type of adaptor used. The selected dust then passes through the instrument's detection volume and is finally collected onto a filter placed inside a standard cassette. The CIS adaptor can also be fitted with a cassette containing a combination of porous polyurethane foam filters that allow size-selective sampling of the various health-related size fractions when operated at a flow rate of 3.5 l min–1 (Kenny and Stancliffe, 1997). These include PM10, PM2.5 and respirable size fractions.
The Microdust can also be fitted with a fan-driven aspirating unit for situations where there is insufficient air movement to introduce aerosol into the inlet. This, however, is not size selective and does not incorporate a backup filter to collect the dust. It cannot therefore be used to calibrate the instrument; it merely aids entry of the dust into the inlet.
Methodology
The performance of the three real-time dust monitors when operated passively and actively was compared using a standard method for determining airborne dust concentration. Since the monitors respond primarily to dust of the respirable size fraction, the Higgins–Dewell gravimetric cyclone sampler (manufactured by Casella Ltd) was chosen. It was assumed that the cyclone sampler gave negligible errors and that this was the true measurement of concentration. This is a reasonable assumption to make since such cyclones are usually calibrated in calm air conditions (Liden and Kenny, 1991) and these tests were also carried out inside a calm air dust box. Errors were minimized by frequently checking the sampler flow rates and by ensuring that dust samples of 
1 mg were collected on the filters (HSE, 2000). If <1 mg of dust is collected on the filter, any reduction in filter mass due to inevitable fibre loss during handling may introduce significant errors.
As previously described, the Split 2 and Microdust dust monitors can be used with in-line, inhalable adaptors that contain polyurethane porous foams to select the respirable fraction of airborne dust. By summing the masses of dust collected on the foams and the backup filter, a measure of the inhalable concentration can also be obtained. Reference personal IOM inhalable samplers (Mark and Vincent, 1986) were therefore included for comparison with the foam adaptors. Also, previous work has shown that the Split 2 inhalable adaptor was prone to leakage of dust around seals resulting in a significant underestimation of the inhalable concentration (Thorpe, 2007). SKC have since modified the design of the adaptor and so comparisons with the reference IOM inhalable samplers would determine whether or not the modifications were effective.
Dust samples
Wood dust (sanded beech), plain white flour and chalk dust were selected to represent dusts found in typical industries such as woodworking, bakeries and construction/quarrying. ARD was also chosen as a test dust because it is often used to calibrate instruments of this type. The ARD was supplied by Powder Technology Incorporated and conformed to ISO 12103 Pt 1, 1997. Two grades of aluminium oxide (trade name Aloxite—supplied by Washington Mills UK Ltd) were chosen to investigate the effects of particle size on instrument response. These were 1200 and 800 grades which represent quoted mass median aerodynamic diameters (MMAD) of 6 and 13 µm, respectively. Aloxite is an abrasive dust used mainly in grinding applications but an additional major advantage is the narrow distribution of the particle size for each grade, making it ideal for investigating the effects of particle size on the performance of aerosol measuring devices (Mark et al., 1985). Plain white flour (manufactured by Be-Ro) was bought from a local supermarket. The beech wood dust was produced in the laboratory by sanding a plank of beech using a belt sander fitted with a 40-grade sanding belt. The dust produced was captured by attaching a vacuum cleaner to the dust extraction port on the side of the sander. The chalk dust was supplied by Omya UK Ltd, which was mainly composed of calcium carbonate and had a quoted D50 of 10 µm for the particle size.
Calm air dust chamber
All the tests were carried out inside a calm air chamber since Kenny et al. (1999) demonstrated that random errors in reference concentration, arising from temporal instability, are much easier to control in the smaller space of an aerosol chamber than in a large wind tunnel. The reference inhalable IOM samplers were used in isolation since Kenny et al. (1999) demonstrated that they can be used in isolation rather than on a manikin in calm air conditions without any noticeable difference in performance. Kenny et al. (1999) also found that low wind efficiencies for the isolated IOM and CIS samplers, although slightly higher than measurements made inside a tunnel at a wind speed of 0.5 m s–1, compared well to the average inhalability measured in low wind speeds using a breathing manikin (Aitken et al., 1998).
The chamber was similar in design to that used by the dust monitor manufacturers during initial calibration with Standard test dusts. The experimental set-up was also similar to that used by Chung and Vaughan (1989). The test chamber illustrated in Fig. 1 comprised two galvanized steel sections of dimensions 1 x 1 x 1 m stacked one on top of the other, which was grounded to reduce any effects caused by charge on the aerosol. Typically, the volume flow rate was set to 220 l min–1, which equated to an air velocity through the chamber of 0.4 cm s–1. The monitors were therefore tested well within calm air conditions (assumed to be air movements of <10 cm s–1). Although the temperature and relative humidity inside the chamber were not regulated, they remained fairly constant between 21–23°C and 30–35%, respectively, throughout the tests.
|
The dust was introduced into the chamber by one of two methods, depending on its properties. The first method was the rotating brush generator model RBG 1000 manufactured by PALAS GmbH. The main disadvantage of this method is that the resultant aerosol can have an extremely high charge produced by tribo-electrification effects; however, an ionizing fan on the chamber helped to reduce the charge on the aerosol. The main advantage is that it can produce a very constant and reproducible feed of dust. The ARD, Aloxite dusts and chalk dust were all generated using this method because they all formed stable compacted cakes of dust within the dust feed cylinder. The beech and flour dust, however, did not compact or feed well using the PALAS. The TOPAS GmbH model SAG 410 dust generator was used for these dusts, which is more suited to free-flowing powders. Here, the dust is placed inside a hopper, which sits above a moving toothed belt. An air-driven nozzle removes the dust from the belt and ejects it at a high velocity into the top of the calm air chamber. The hopper was easily refilled during operation without any effects on the dust feed.
Test procedure
All the samplers and dust monitors were placed on a rotating turntable inside the chamber and a typical set-up is shown in Fig. 2. This rotated at one revolution every 2 min and had a reciprocating action. This meant that the sampling pumps could be placed outside the chamber with the air sampling tubes entering the chamber from outside. The tubes were suspended above the samplers and dust monitors so that they did not become entangled during testing. The reference samplers and dust monitors were all placed on the same circumference of the turntable to ensure that they were exposed to the same concentration of aerosol. Testing was carried out in three stages for each dust. The first stage was with the dust monitors operating passively, relying on the movement of the monitors to introduce aerosol into the inlet. During these tests, a cyclone sampler was placed in close proximity to each dust monitor to measure the reference respirable dust concentration. For the second set of tests, the monitors were operated actively by attaching the cyclone inlets and the dusty air was drawn into the inlets using Casella Apex personal sampling pumps. Once again, cyclone samplers were placed in close proximity to each dust monitor to measure the reference concentration. For the third set of tests, the Microdust and Split 2 dust monitors were operated actively and were fitted with the inhalable adaptors containing porous polyurethane foams to select the respirable fraction. The DataRam was also included in the tests, which was operated passively. Cyclone and IOM samplers were placed close to each monitor to measure the reference respirable and inhalable concentrations, respectively. The IOM samplers were included so that the performance of the Microdust and Split 2 adaptors could be assessed as inhalable samplers. Before testing commenced, the reference cyclone samplers, IOM samplers, Split 2 adaptors and Microdust adaptors were loaded with clean 25-mm glass fibre filters. The DataRam GK 2.05 cyclone adaptor was loaded with a 37-mm glass fibre filter. All the filters were conditioned and weighed using a five-place balance before and after exposure. Sampler flows were set to their correct values and were checked before and after each test using a calibrated bubble flow meter with a range of 0–6 l min–1 (Gilian Ltd). Each dust monitor was zeroed prior to each test, in accordance with the manufacturer's instructions and programmed to record a measurement every 5 s. Single-point calibration checks were carried out using the supplied calibration element for the Microdust and Split 2 monitors. The DataRam does not use a calibration element.
|
Tests were carried out for at least three concentrations of each dust. The concentration was varied by changing the piston and belt speeds of the PALAS and TOPAS dust generators, respectively. The test duration was varied according to the dust concentration but was usually between 30 min and 2 h so that a weighable dust sample was obtained (>1 mg) on the filter.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Real-time monitors
Passive mode versus active mode.
The reference cyclone concentrations generally agreed within 5% of each other and so averages of these were used for comparison with the real-time dust monitors. Plots of monitor response versus reference cyclone concentration for each dust type, dust monitor and mode of operation are not shown since there are 18 in total. Instead, calibration factors and a linear fit to the data indicated as R2 values are summarized in Table 1. Mean calibration factors are calculated as the inverse slope of the graphs of monitor concentration plotted against reference cyclone concentration. The inverse slope was determined from a linear regression while constraining the intercept to zero. The calibration factor is therefore the number that the monitor concentration is multiplied by so that it agrees with the reference concentration.
|
Generally speaking, all the dust monitors showed good linearity for most of the dusts tested, with the exception of the active measurements made with the foam adaptors. This is indicated by R2 values close to 1 when the dust monitors were operated passively and actively with the cyclone adaptors attached. When the monitors were operated actively with the porous foam adaptors attached, there was much more scatter in the results, which is indicated by a reduction in the R2 values in most instances. The increased scatter is almost certainly caused by dust-loading effects, i.e. as the foams became loaded with dust their filtration efficiency increased resulting in less dust reaching the dust monitor sensor and backup filter. The sampling efficiency of the reference cyclone samplers should remain relatively unchanged with increasing dust loading as shown by Thorpe (2001). This is explained in more detail in the section Foam dust-loading effects.
For all dusts, the dust monitors showed a lower response when operated actively with the cyclone adaptors attached than when operated passively. This is indicated by the correspondingly higher calibration factors given in Table 1 and is also illustrated in Fig. 3 for chalk dust. This was not unexpected since, with the cyclone adaptor attached, only the respirable fraction of the dust enters the sensing zone of the dust monitor, i.e. particle sizes <12 µm. When operated passively, if there is sufficient air movement, then particles of all sizes will enter the monitor's sensing zone. Even though each dust monitor responds mainly to the respirable fraction of airborne dust, they will continue to partly detect particles up to 
30 µm, defined by the response curve for the particular monitor, resulting in a higher measurement of respirable dust concentration. Willeke and Degarmo (1988) carried out similar comparisons using a MiniRam dust monitor (Mie Inc.), which is a predecessor to the DataRam used in these tests. They used a Dorr–Oliver cyclone to determine the reference concentration. They did not observe any distinguishable differences in passive and active operation, although there was a large degree scatter in their data introduced by zero drift of the instrument.
|
The response of the Microdust and Split 2 dust monitors was lower still when operated actively with the porous foam adaptors attached. This is illustrated in Fig. 3 and Table 1. This was unexpected since the foams are designed to essentially sample the same respirable size fraction as the cyclones. The effect is compounded as the foams start to clog with dust and act as more efficient filters.
Table 1 shows that the Microdust monitor consistently read higher than both the Split 2 and the DataRam whether operated actively or passively. The discrepancy is because the DataRam and Split 2 are calibrated in the factory against the respirable fraction, whereas the Microdust is calibrated against the concentration of TSP, which approximates to the inhalable concentration. Generally, the DataRam measurements were close to the reference cyclone measurements for most dusts when operated passively. This is shown typically in Fig. 4 for beech wood dust, where the data points are close to the solid black line, which represents a 1:1 relationship. The Split 2 nearly always read slightly lower than the DataRam regardless of dust type and whether it was operating passively or actively. This is illustrated in Table 1, where the calibration factors for the Split 2 are often higher than the DataRam. Both the DataRam and Split 2, when operated passively, underestimated the concentration for ARD which itself is used to factory calibrate the monitors. It may have been that the calibration of both instruments had drifted since manufacture. Alternatively, because the air inside the chamber was very still, the rotational movement of the monitors on the turntable might not have been sufficient to effectively introduce the dust into the monitor inlets.
|
Effects of particle size on the performance of the dust monitors.
The effect of particle size on dust monitor response is best illustrated using the results for the two grades of Aloxite dust. Additional factors such as colour and refractive index will affect the instrument response for the other dusts. Particle size effects are summarized in Table 2 for the monitors operating both passively and actively. It can be seen that when operated passively, a relatively small increase in the particle size from 6 to 13 µm resulted in a significant increase in the monitor response (shown by the percentage change in response in Table 2). This was especially noticeable for the Microdust and Split 2 monitors. In contrast, when operated actively with the cyclones attached, the response for each dust monitor was practically unaffected by increasing particle size. This is because the cyclone effectively homogenizes the two sizes of dust by allowing through most of the 6-µm dust particles but removing the largest particles of the 13-µm dust. The size distribution of the two dusts as sampled by the dust monitor sensors is therefore similar. The fact that there is very little scatter in the results, indicated by R2 values close to 1, is a good indication that the particle size effects are real. When the Microdust and Split 2 monitors were operated actively with the foam adaptors attached, an increase in response was observed with increasing particle size. This was probably caused by the different clogging characteristics of the two grades of dust. Coarser dust is less clogging and will, for the same mass load as the finer dust, have a smaller effect on the sampling efficiency of the porous foam, meaning that more dust will penetrate into the monitor's sensing zone.
|
Samplers
Cyclone and respirable porous foam adaptors.
In most cases, the dust monitor cyclone adaptors showed extremely good linearity compared to the reference measurements made with the Higgins–Dewell cyclones. There was a degree more scatter in the flour results, but for the other dusts the R2 values were very close to 1 as shown in Table 3. This shows that the experimental set-up and test methods gave consistent and accurate results. Both the GS-3 and GK 2.05 cyclone adaptors underestimated the respirable concentration measured using the reference Higgins–Dewell cyclones, for most of the dusts tested. This is shown in Table 3 for all the dusts tested and Fig. 5 for ARD. This was somewhat unexpected since they are both designed to sample dust according to the ACGIH–CEN–ISO respirable size-selection curve—like the Higgins–Dewell cyclone. The Microdust cyclone adaptor agreed closely with the reference Higgins–Dewell cyclones, which was not surprising since it is essentially the same design. The reason for the underestimation is unclear. It might be that some of the dust entering the monitor is lost within the sampling head before it reaches the backup filter. This is possible since the dust travels quite a distance from entering the cyclone to reaching the backup filter, especially for the DataRam dust monitor. This does not, however, explain why the Split 2 GS-3 cyclone overestimated the concentration of 13 µm Aloxite dust. It may be that the GS-3 and GK 2.05 monitor cyclone adaptors behave differently to the reference cyclones in calm air conditions. The effects, however, appear to be real as indicated by the high R2 values of the linear fit to the data.
|
|
Measurements made with the porous foam adaptors are lower than the corresponding measurement made with the cyclone adaptor for each dust type and dust monitor, shown by the slopes of the graphs summarized in Table 3. There is, however, a much greater degree of scatter in the foam measurements shown by the lower R2 values, which is almost certainly the result of variable clogging caused by inconsistent loading of the porous foam with dust.
Foam dust-loading effects.
As was observed earlier, there is more scatter in the concentrations measured with the Microdust and Split 2 porous foam adaptors. It is thought that the dust deposited on the foams causes clogging during sampling. As the foams clog, then this will effectively increase their sampling efficiency and decrease the D50 cut-point, resulting in less dust penetrating the foam and less dust being collected by the backup filter. Figure 6 shows the effects of dust loading on sampler performance. The results are plotted for ARD, 1200 grade Aloxite (MMAD 6 µm) and 800 grade Aloxite (MMAD 13 µm), since these had the greatest range of dust mass deposited on the foams. Foam adaptor concentration divided by reference cyclone concentration is plotted as a function of dust collected on the foams. This gives a good indication of how dust loading affects the foam samplers, since the reference cyclone is relatively unaffected by the amount of dust it collects (Thorpe, 2001). It can be clearly seen that the concentration measured by the foam adaptors compared to the reference cyclone concentration decreased significantly with the amount of dust deposited on the foams for the Aloxite 1200 and ARD. For these dusts, even relatively small amounts of dust deposited on the foams significantly affected the measurement of concentration. The effect is further illustrated in Fig. 7, where dust concentration measured by the three dust monitors is plotted against time for ARD. The DataRam was operated passively and the Microdust and Split 2 dust monitors were operated actively with the foam adaptors fitted. The dust concentration inside the test chamber was fairly constant with time throughout the test as shown by the DataRam plot. Figure 7 confirms that as the foams start to clog, the concentration measured by both monitors decreases significantly. It is also evident from Fig. 6 that for the courser 800 grade Aloxite (MMAD 13 µm), the performance of both the CIS and IOM samplers appears to be largely unaffected by dust loading (for a roughly equivalent total dust load). This indicates that clogging of the foams is much more dependant on particle size rather than mass of dust collected. Dust-loading effects were observed with the other dusts but these were not as pronounced, either because the dust was less clogging or because the spread of dust deposits on the foam was not as great.
|
|
Stancliffe and Chung (1997) have also reported changes in the size-selective properties of porous foams when exposed to welding fume. They found that when they exposed a modified IOM sampler containing foam plug size selectors to welding fume, there was an observed decrease in D50 with increasing loading. Clogging of the foam caused it to overselect and resulted in undersampling of the respirable fraction. This is contrary to work carried out by Kenny et al.(2001) who observed little change in the sampling characteristics of an IOM dual fraction dust sampler with dust load when challenged with laboratory generated and general workplace aerosols produced during processes such as metal refining, casting, mineral production, tyre processing, foods and pharmaceuticals. Their laboratory generated aerosols consisted of 800 grade Aloxite, which we also found had little effect on the sampling characteristic of the porous foam, as mentioned previously. They observed only a slight decrease in the D50 cut-point, even with exceptionally high dust loads (>20 mg of dust on the foams). In addition, they found good agreement between the IOM dual fraction sampler and a standard cyclone sampler, which is contrary to many of our findings. This would appear to confirm that different dusts exhibit different clogging characteristics associated with, i.e. particle size and ‘stickiness’.
SKC and Microdust Pro-inhalable adaptors.
Comparisons of the inhalable concentration measured using the Split 2 IOM adaptor and Microdust CIS adaptor with the reference IOM sampler are shown in Fig. 8. Overall, for all the dusts tested, the Split 2 IOM adaptor data correlated very well with the reference IOM sampler data, agreeing within 2%. This confirms that the modified design implemented by SKC has improved the sealing and eliminated dust leakage within the sampling head. The Microdust CIS adaptor data also correlated well with the reference IOM data but underestimated the concentration by 16%. This is consistent with the work carried out by Kenny et al.(1999). They showed that the IOM and CIS samplers have similar efficiency values at smaller particle sizes but for larger particles the CIS undersamples compared to the IOM sampler. This is thought to be due to sedimentation of particles between the inlet and filter plane of the CIS sampler. Examination of exposed CIS samplers showed this effect, especially for large particulate aerosols. The IOM samplers did not exhibit such losses because of the way it is designed. Figure 9 shows the effect of particle size by comparing the IOM and CIS results for the two grades of Aloxite. There appears to be some further evidence, given the high degree of fit, that the CIS increasingly underestimated the IOM inhalable concentration as the particle size increased from 6 to 13 µm. This effect should be even more pronounced for larger particle sizes.
|
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
The responses of the SKC Split 2, Casella Microdust Pro and Thermo Electron DataRam dust monitors were found to be linear with respirable dust concentration when operated passively and actively using cyclone size-selective inlets. However, the response was lower when the monitors were used actively with cyclone samplers attached to their inlets, rather than passively. The monitor responses fell further when operated with the porous foam adaptors attached. The degree of scatter in the results was also greater, which was attributed to variable clogging of the foams caused by inconsistent loading with dust. The extent of clogging, as revealed by the decrease in the response of the real-time monitors and the ratio of foam adaptor concentration to the reference cyclone concentration, depends on the type of dust. The dust monitor responses were sensitive to changes in particle size when operated passively but much less so when operated actively with the cyclone in-line adaptors. However, particle size effects can be complicated by the response of the dust monitors to changes in other physical properties such as refractive index, colour and shape.
The Microdust monitor consistently read higher than both the Split 2 and the DataRam whether operated actively or passively. This is because the Microdust is calibrated against the concentration of TSP rather than respirable. The Split 2 nearly always read slightly lower than the DataRam regardless of dust type and whether it was operating passively or actively.
The DataRam GK 2.05 cyclone adaptor underestimated the reference respirable concentration by 33% on average for all dusts. The Split 2 GS-3 cyclone adaptor underestimated the reference dust concentration by 13% on average for all dusts but overestimated the concentration of 13 µm Aloxite dust. An explanation could be that dust was lost within the sampling head of the monitors before it reached the backup filter, although this does not explain the increase in concentration measured by the Split 2 GS-3 cyclone for 13 µm Aloxite dust. Alternatively, it may be that the GS-3 and GK 2.05 monitor cyclone adaptors behave differently to the reference cyclones in calm air conditions. Nevertheless, this questions the use of the Split 2 and DataRam in-line cyclones for field calibration of the dust monitors. The Microdust Higgins–Dewell in-line cyclone adaptor showed much better agreement with the reference cyclones and therefore could probably be used to calibrate the Microdust dust monitor in the field.
Concentrations measured with the Split 2 IOM and Microdust CIS foam adaptors were considerably lower than both the reference Higgins–Dewell cyclone samplers and the in-line dust monitor cyclone adaptors. This could have been partly caused by dust losses within the sampling head. However, they also increasingly under-sampled the respirable dust concentration as they became loaded with dust. Consequently, the in-line foam adaptors could not be confidently used for field calibration of the dust monitors.
The Split 2 IOM adaptor agreed closely with the reference IOM samplers when used to measure the inhalable fraction of airborne dust. This is the result of a redesigned filter holder by SKC to improve internal seals within the sampler after a previous design was found to leak dust around the backup filter. A periodic leak test on all types of actively sampling instruments would nevertheless be useful as part of the instrument maintenance schedule to check that connections are airtight. The CIS adaptor underestimated the inhalable concentration compared to the reference IOM sampler by 16% on average for all dusts tested.
In summary, accurate field calibration of the Split 2 and DataRam dust monitors should probably be carried out using separate gravimetric cyclone samplers situated close to the dust monitors. The Microdust dust monitor could be calibrated in the same way or with its own in-line cyclone adaptor. A different calibration factor would have to be applied to the monitor concentration depending on whether the monitor is operated passively or actively.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
We thank Paul Roberts [Health and Safety Laboratory (HSL)] for assistance with the measurements and Rodger Clark (HSL) for useful discussions.
Received June 19, 2007; in final form August 21, 2007
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Aitken RJ, Baldwin PEJ, Beaumont GC, et al. Aerosol inhalability in very low winds. J Aerosol Sci (1998) 30:613–26.[Web of Science]
CEN. Workplace atmospheres: size fraction definitions for measurement of airborne particles in the workplace. In: CEN standard EN 481 (1993) Brussels: Comite Européen de Normalisation.
Chung KYT, Vaughan NP. Comparative laboratory trials of two portable direct-reading dust monitors. Ann Occup Hyg (1989) 33:591–606.
HSE. Methods for the determination of hazardous substances. General methods for sampling and gravimetric analysis of respirable and inhalable dust. In: MDHS 14/3 (2000) Sudbury, UK: Health and Safety Executive Books.
Kenny LC, Gusmann RA. Characterization and modelling of a family of cyclone aerosol preseparators. J Aerosol Sci (1996) 28:677–88.[Web of Science]
Kenny LC, Stancliffe JD. Characterisation of porous foam size selectors for the conical inhalable sampler. In: HSE report IR/L/A/97/19 (1997) Buxton: Health & Safety Laboratory.
Kenny LC, Aitken RJ, Baldwin PEJ, et al. The sampling efficiency of personal inhalable aerosol samplers in low air movement environments. J Aerosol Sci (1999) 30:627–38.[CrossRef][Web of Science]
Kenny LC, Chung KYT, Dilworth C, et al. Applications of low-cost, dual fraction dust samplers. Ann Occup Hyg (2001) 45:35–42.
Liden G, Kenny LC. Comparison of measured respirable dust sampler penetration curves with sampling conventions. Ann Occup Hyg (1991) 35:485–504.
Mark D, Vincent JH. A new personal sampler for airborne total dust in workplaces. Ann Occup Hyg (1986) 30:89–102.
Mark D, Vincent JH, Gibson H, et al. Applications of closely-graded powders of fused alumina as test dusts for aerosol studies. J Aerosol Sci (1985) 16:125–31.[CrossRef][Web of Science]
Maynard D, Jensen PA. Aerosol measurement in the workplace, chapter 25. In: Aerosol measurement—principles, techniques, and applications—Baron PA, Willeke K, eds. (2001) New York: John Wiley & Sons Inc. 779–99.
Rosén G, Andersson IM, Walsh PT, et al. A review of video exposure monitoring as an occupational hygiene tool. Ann Occup Hyg (2005) 49:201–17.
Stancliffe JD, Chung KYK. Preparation, handling. Loading and analysis of PUF plugs used for welding fume sampling. In: HSL report IR/L/A/97/09 (1997) Buxton: Health & Safety Laboratory.
Thorpe A. Investigation of the effects of extensive use on the penetration of cyclones. In: HSL report IR/L/EXCON/01/20 (2001) Buxton: Health & Safety Laboratory.
Thorpe A. Assessment of personal direct-reading dust monitors for the measurement of airborne inhalable dust. Ann Occup Hyg (2007) 51:97–112.
Thorpe A, Walsh PT. Performance testing of three portable, direct-reading dust monitors. Ann Occup Hyg (2002) 46:197–207.
Willeke K, Degarmo SJ. Passive versus active aerosol monitoring. Appl Ind Hyg (1988) 3:263–6.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||