Performance of High Flow Rate Personal Respirable Samplers When Challenged with Mineral Aerosols of Different Particle Size Distributions

Abstract

It is thought that the performance of respirable samplers may vary when exposed to dust aerosols with different particle sizes and wind speeds. This study investigated the performance of the GK 4.16 (RASCAL), GK 2.69, PPI 8, and FSP 10, high flow rate personal samplers when exposed to aerosols of mineral dust in a wind tunnel at two different wind speeds (1 and 2 m s⁻¹) and orientations (towards and side-on to the source of emission). The mass median aerodynamic diameter of four aerosolized test dusts ranged from 8 to 25 µm with geometric standard deviations from 1.6 to 2 µm. The performance of each sampler type was compared with that of the SIMPEDS (Higgins–Dewell design) sampler. There was slight evidence to suggest that the performance of the FSP 10 is affected by the direction of the inlet relative to the air flow, although this was not significant when most respirable dust concentrations were compared, possibly due to the variability of paired dust concentration results. The GK 2.69, RASCAL, and PPI 8 samplers had similar performances, although the results when side-on to the emission source were generally slightly lower than the SIMPEDS. Despite slight differences between respirable dust concentrations the respirable crystalline silica values were not significantly different from the SIMPEDS. The GK family of cyclones obtained most precise results and more closely matched the SIMPEDS. A comparison with dust concentration results from previous calm air chamber studies (where wind speeds were < 0.4 m s⁻¹) found that the relative performance between samplers was similar to those observed in this work indicating consistent performance relative to the SIMPEDS in both calm and moving air.

INTRODUCTION

Many processes in the workplace generate aerosols containing small particles of dust that can penetrate to the alveoli region of the lung. These particles are termed ‘respirable-sized’. The physiological effect of small particles is currently taken to relate to the potential to penetrate into the alveolar region. The performance of samplers to mimic this potential has been defined through convention by the International Organisation for Standardization/European Centre for Normalization/American Conference of Government Industrial Hygienists (ISO/CEN/ACGIH) (ACGIH, 1985; CEN, 1993; ISO, 1995). The conventional respirable sampler performance refers to a rotational average and is specified for a sampler ideally collecting particles from the ambient air with particle aerodynamic diameter ranging up to about 16 µm and with a 50% cut size (d₅₀) of 4.0 µm. Inhalation of respirable crystalline silica (RCS) causes lung diseases such as silicosis, lung cancer (IARC, 2012), and possibly chronic obstructive pulmonary disease (COPD). Exposure to RCS is also associated with renal disease and autoimmune disorders (NIOSH, 2002). It is estimated that 1 million workers are regularly exposed to this hazardous substance in the United States (US) and 100 000 in the UK (HSE, 2007). One of the primary uses for a respirable sampler is to assess worker exposure to aerosols containing RCS.

Several samplers have, according to EN 13205 (CEN, 2002), a performance that matches the respirable sampling convention within certain parameters (Bartley et al., 1994). However, it is difficult to achieve a perfect match and the particle selection characteristics for most samplers deviate from the sampling convention, which results in a mass concentration bias. Even when samplers closely match the respirable sampling definition the characterization of the particle size selection performance might not take into consideration the retention of particles on internal sampler surfaces and losses of dust from the filter. Significant differences between respirable samplers in terms of the collected dust concentrations have been observed in laboratory studies (Stacey et al., 2013).

It’s important for occupational health professionals to have an idea of the difference in performance between sampler equipment when comparing data. Differences in performance between high flow rate samplers were evaluated in calm air conditions (wind speed < 0.4 m s⁻¹) on three grades of Arizona Road dust (ARD) and coal dust (Stacey and Thorpe 2010; Lee et al., 2012; Stacey et al., 2014) and gave very similar results. Although different grades and types of dusts were used, the size distributions of the aerosolized particles were very similar (Stacey et al., 2014). The mean particle size of an aerosolized powder measured in calm air conditions can be finer than the bulk material from which it originated. This is probably due to the lower settling rates of smaller compared with larger particles, which may cause a build-up of the proportion of finer particles as the heavier ones ‘fall out’ quicker. Stacey et al. (2013) postulated that the performance of samplers is probably more similar when the test aerosol has a very small particle size distribution. What remained unknown is how the performance of these samplers changes when challenged with a range of mineral aerosols with particle size distributions that are potentially more realistic of an individual’s exposure. Wind tunnels can be used to evaluate sampler performance over a range of wind speeds and to assess any bias caused by sampler design and flow rate when positioned away and towards the source of the aerosol emission. This article compares the mass and RCS concentrations collected by each test sampler when exposed to aerosols of mineral dusts in various wind tunnel conditions and compares these results with previous studies conducted in calm air conditions at the Health and Safety Laboratory (Stacey et al., 2010, 2013, 2014).

Currently, X-ray diffraction (XRD) is routinely used at the HSL to measure RCS content. Although the current analytical limit of detection for XRD is calculated to be around 10 µg (3 σ), a sample mass between 20 and 50 µg is needed to obtain a reliable quantitative value of RCS (HSE, 2005). A range is given as the quantifiable mass value is dependent on the sample matrix and the XRD reflection used for analysis. The sample mass collected with low flow rate respirable samplers (~2 l min⁻¹), commonly employed in the UK and the US, is not always sufficient to determine reliable RCS measurements. Therefore, high flow rate samplers (≥ 4 l min⁻¹) will potentially have a significant future role when assessing the exposure of workers to RCS within the 8-h reference sampling period. In addition, there is currently no standard quantitative method to monitor the exposure of workers to RCS for short task-specific periods. This may limit the ability of the occupational hygienist to perform task specific sampling to investigate parts of a working shift that might contribute the highest to worker exposure. Semiquantitative approaches such as video visualization and direct-reading instruments may provide good indicative values; however, they still require a result from an air sampler for calibration. Therefore, it is advantageous to assess if current high flow rate samplers provide results comparable with reference samplers that operate at lower flow rates. This article also reports the first comparison results for two new samplers; the Parallel Plate Impactor (PPI 8) (SKC Ltd, Dorset, UK) and GK 4.162 which has been termed, the Respirable Air Sampling Cyclone Aluminum Large (RASCAL) (BGI Inc, Waltham, MA, USA) sampler.

MATERIALS

The samplers included in this study were the GK 2.69 operating at 4.2 l min⁻¹, the GK 4.162 operating at 9 l min⁻¹ (BGI Inc), the FSP 10 (GSA Messgerätebau, Neuss, Germany) operating at 10 l min⁻¹, and the PPI 8 operating at 8 l min⁻¹ (SKC Ltd). The Safety In Mines Personal Dust Sampler (SIMPEDS) was used for comparison in every test (Casella Measurement Ltd, Bedford, UK), as it is of the Higgins–Dewell design, is commonly used in the UK and has been used as a reference for other work at low air speeds. The GK 2.69 and GK 4.162 samplers are based on the same GK cyclone design developed by Kenny and Gussman (1997), and should have a similar performance at their respective flow rates. The FSP 10 and the SIMPEDS are both based on the Higgins–Dewell cyclone design. The PPI 8 is not a cyclone and uses four impaction plates to select the respirable fraction. All samplers in this study used filters that were either recommended by the manufacturer or by the RCS analysis method. The sampler types, some inlet dimensions, and the types of filter used in each sampler are shown in Table 1. Each sampler was made of conductive materials. The SIMPEDS samplers were used with Apex personal sampling pumps (Casella Ltd, Bedford, UK). The GK 2.69 samplers were used with Buck LP7 personal sampling pumps (A.P. Buck Inc, Orlando, FL, USA) and the RASCAL, FSP 10, and PPI 8 samplers were used with Legacy pumps (SKC Ltd, Dorset, UK). An additional external pulsation damper was applied to all the pumps to ensure that the sampler performance was not potentially affected by high pulsation rates (Lee et al., 2014). Sampler flow rate was checked before each test using a calibrated TSI model 4146 flow calibrator (TSI Inc, Shoreview, MN, USA) and adjusted if necessary so that it was within 0.1 l min⁻¹ of the target value. It was then checked after the test was complete, and the sample was rejected if the flow difference exceeded ±5 % of the nominal flow rate. The challenge dusts used during the tests were chosen such that they contained an appreciable amount of crystalline silica for further analysis by XRD. These were Arizona Road dust (ARD) (coarse grade, 0–180 µm diameter, ISO 1997), Pittsburgh coal dust used in a previous study by Lee et al. (2012) and a refractory mineral dust. ARD typically contains around 70% of crystalline silica. The refractory mineral, comprising of mainly Mullite, contained around 10% crystalline silica. The coal dust was expected to contain around 7% crystalline silica in the respirable fraction (Lee et al., 2012), but on subsequent analysis contained much smaller quantities (~0.5% in the bulk).

Table 1.

Respirable samplers and nominal flow rates used in this study

NA, not available.

METHOD

The performance of the high flow rate samplers at different wind speeds was tested in HSL’s dust tunnel facility. For comparability with their performance in calm air conditions described in previous work (Stacey and Thorpe, 2010, Lee et al., 2012; Stacey et al., 2013, 2014), the GK 2.69 and FSP 10 were maintained at the manufacturer’s recommended flow rates of 4.2 and 10 l min⁻¹ rather than those proposed by other researchers (Lee et al., 2010) in order to meet the ISO/CEN/ACIGH convention more closely. The evaluation method was based on that described in EN 13205 (CEN, 2002).

The HSL dust tunnel

The dust tunnel (shown in Fig. 1) is 12 m long with a 1 m (height) × 1.5 m (width) cross section. A rotating screw powder metering system (March Systems Ltd) was used to feed dust into a venturi dust disperser powered by compressed air, which injected the dust into the tunnel via a length of flexible tubing. Dust was injected at a single point and directed towards a centrally located metal plate for dispersal and mixing. A wooden turbulence grid was positioned ~2 m downstream to reduce large scale turbulence and the sampling point was about 8 m further downstream. The concentration of dust inside the tunnel could be adjusted by changing the size or rotational speed of the screw. Air was blown down the tunnel using a large centrifugal fan to a bank of high efficiency particulate air (HEPA) filters, so that the cleaned air was recirculated. Measurement of the air velocity profile inside the tunnel at the sampling position was very uniform, typically varying by a relative standard deviation < 6% over the area of the tunnel that the samplers occupied. This is within the limits defined in EN 13205 (CEN, 2002). The tunnel air velocity was set using a TSI Velocicalc Plus (model 8388) hot wire anemometer placed centrally inside the tunnel upstream of the manikin. The majority of tests were performed at a wind speed of 1 m s⁻¹. One set of tests was performed with a wind speed of 2 m s⁻¹. These wind speeds were selected based on trials in the preliminary study; at velocities higher than 2 m s⁻¹ the dispersion of particles was less uniform and at velocities lower than 1 m s⁻¹ large particles settled out before reaching the sampling position.

Figure 1.

Schematic diagram of the HSL dust tunnel 50×16mm (300×300 DPI).

A ‘Little Anne’ manikin was selected to support the samplers inside the dust tunnel as it comprises the upper torso and head of a human and was easily accommodated into the dust tunnel giving a blockage of about 13% of the tunnel’s total cross-sectional area, which is less than the maximum tunnel blockage of 15 % quoted in EN 13205 (CEN, 2002). The manikin was attached to a rotating turntable that operated with a reciprocating action. In the majority of tests, the sampler and torso were static and either positioned facing or side on to the source of emission. In two tests, the turntable and manikin were rotated at 1rpm to gain an orientation average. Initial homogeneity of aerosol generation was determined using six SIMPEDS samplers placed equidistant over the torso of the manikin. An aerosol of 1000 grade Aloxite (mass median aerodynamic diameter = 9 µm, Washington Mills Ltd, Manchester, UK) was generated with the manikin positioned facing and side-on to the tunnel air flow of 1 m s⁻¹.

During the tests, six SIMPEDS reference samplers and one each of the high flow rate samplers were positioned on the chest area of the manikin during each sampling run. One SIMPEDS was placed between each of the high flow rate samplers. The pumps were located outside the tunnel so that they did not become exposed to dust and were attached to the samplers using flexible silicone rubber tubing that passed through the side of the tunnel. The tubes were suspended above the manikin so that they did not become entangled when it rotated.

Twelve tests were conducted with ARD at 1 m s⁻¹ (five facing, four side-on, and three rotating), 10 tests with coal dust (three facing, four side-on, and three rotating), eight tests with Mullite powder (four facing and four side-on). Six tests were also carried out with the ARD at a wind speed of 2 m s⁻¹ (three facing and three side-on).

Particle size analysis

An Aerodynamic Particle Inspector instrument [API Aerosizer, (discontinued) TSI Inc] was used to measure the size distribution of the airborne particles. This uses the time of flight (TOF) principle to determine the aerodynamic size distribution of airborne particles in real time in the size range 0.20–200 µm. The API was placed below the tunnel with its inlet connected to a length of copper pipe passing vertically through the tunnel floor and positioned alongside the manikin. Attached to the copper pipe was a sharp-edged isokinetic probe designed to match the wind speed of the tunnel, ensuring the API took a representative sample of the airborne dust. The isokinetic inlet was positioned at the same height as the samplers, parallel to manikin, and at approximately same distance from the emission. The curvature in the tube from the sample inlet to the API was designed to minimize wall losses. The API measures the aerodynamic size distribution of particles. The densities used for calculation of the geometric size distribution for ARD, coal dust, and Mullite were: 2.65, 1.5, and 2.6g cm⁻³, respectively.

Gravimetric analysis

Filters were conditioned in a room with controlled humidity (50±5%) and temperature 20±2°C. They were weighed using an UX6 ultra-microbalance (Mettler Toledo Ltd, Leicester, UK) with the resolution set to 1 µg for these experiments. Air concentrations and sampling times were selected so that 0.15–1.5mg of dust was collected by the SIMPEDS sampler, thus ensuring good weighing precision.

Quartz analysis

A sub set of samples from each candidate high flow rate sampler were selected from three sampling tests from each dust and wind speed. These samples were analyzed for RCS and were compared with the average values obtained from three SIMPEDS samples collected within each sampling test. The samples were selected to obtain one sample at the bottom, middle and top of the gravimetric mass loading range collected by each sampler for each dust and wind speed. Samples on 25mm diameter filters were analyzed directly by XRD following the Method for the Determination of Hazardous Substances (MDHS) 101 (HSE, 2005). The collected dust on the PVC filters larger than 25mm diameter were transferred to 25mm silver filters according to the procedure described in the National Institute for Occupational Safety and Health (NIOSH) 7500 method (NIOSH, 2003).

The filters from the candidate samplers were placed into a platinum or ceramic crucible and heated in a furnace at 450°C. A different sample preparation procedure was used with the 37mm cellulose nitrate filters from the FSP 10 sampler because of the chemical nature of the filter. The filters were first covered with 1,3 butadione (CAS 107-88-0) in a platinum crucible before placing in a furnace and the temperature raised to 450°C. The remains of each filter was then washed into a beaker with isopropanol and placed in an ultrasonic bath for 5–10min. The resultant suspension was then filtered onto 25mm diameter silver filters ready for XRD analysis. The sample deposition area on the analysis filter was ~14mm in diameter. Correction for depth and sample absorption effects was made on all samples that indicted absorbance (generally with a dust loading >1mg) using the intensity of the reflection from the silver filter (NIOSH, 2003) beneath the sample.

Data analysis

The airborne dust and RCS concentrations for each sampler were calculated from the mass measured divided by the volume of air sampled. The variability of the results from the six SIMPEDS samplers included in each run was also reported to give an indication of the homogeneity of the aerosol during the candidate sampler tests. The ratio of dust concentrations collected by the test and the average of the reference SIMPEDS samplers was determined for each condition (change of wind speed and direction) to assess their differences. In addition, trend line relationships between the reference SIMPEDS and candidate sampler air concentrations were determined for each aerosol. The x–y intercept for the trend line relationships was set to zero, when this was not significantly different, so the values for the slope represented the relative difference between the two samplers. The significance of each candidate sampler from the ideal relationship (a trend line with the slope of 1) was calculated from a t test based on the difference and the standard error of the trend line.

RESULTS

Homogeneity tests using six reference SIMPEDS samplers exposed to Aloxite obtained standard deviations for respirable dust concentrations of 0.69 (facing) and 0.66 (side on), from a mean aerosol concentration of about 15mg m⁻³ (coefficient of variation ~ 4.5%). The average standard deviation of the six SIMPEDS sampler results during the candidate sampler test runs when measuring respirable dust is shown in Table 2 and show the dust concentration to be reasonably uniform in the wind tunnel, although the variability increased when the air velocity was increased from 1 to 2 m s⁻¹. The air concentration ratios of each candidate sampler divided by the average obtained by the six SIMPEDS reference samples in each test was determined for each, dust, wind speed and manikin position and are shown in Table 3. Trend line relationships for respirable dust concentrations for each candidate sampler compared to reference sampler are shown Fig. 2. The coefficients of determination range from 0.97 to 0.99 indicating good linear relationships with the SIMPEDS. The differences between the candidate sampler and the SIMPEDS are small but significantly different from the ideal relationship (slope = 1.0) for the GK 2.69 (95 % confidence level = 0.96–0.99), GK 4.162 (95 % confidence level = 1.02–1.04) and FSP 10 (95 % confidence level = 1.08–1.12).

Table 2.

Variability of respirable dust concentrations collected by six reference SIMPEDS samplers

Dust	Wind speed (m s−1)	Percent relative standard deviation of dust concentration values
		Average	Minimum	Maximum
Coal	1	4.2	1.2	5.6
Mullite	1	3.5	2.2	5.1
Coarse ARD	1	4.3	1.9	7.4
Coarse ARD	2	8.0	3.1	14.6

Table 3.

Average ratios of respirable dust air concentrations measured with candidate samplers with reference SIMPEDS sampler

Dust	Wind speed (m s−1)	Orientation of manikin	GK 4.162 (RASCAL)	GK 2.69	FSP 10	PPI 8
Coal	1	Facing to the source of emission	1.09	1.00	1.17	1.06
Mullite	1		1.05	0.98	1.16	1.07
Coarse ARD	1		1.05	0.99	1.11	0.97
Coarse ARD	2		1.03	1.02	1.10	1.09
	Average		1.06	1.00	1.14	1.03
Coal	1	Side-on to the source of emission	1.03	0.95	1.08	1.20
Mullite	1		1.01	0.95	1.13	1.01
Coarse ARD	1		1.04	0.95	1.06	0.92
Coarse ARD	2		1.01	1.02	0.92	1.03
	Average		1.02	0.96	1.06	1.05
Coal	1	Rotating	1.07	0.93	1.09	1.19
Coarse ARD	1		1.02	0.97	1.12	1.06
Combined results
Average			1.04	0.97	1.10	1.06
Percent coefficient of variation (% CV)			3.5	4.6	7.1	8.8

Figure in bold are the average values.

Figure 2.

Comparison of respirable dust concentrations between each high flow rate sampler and the SIMPEDS for all mineral dust aerosols and air speeds.

XRD results for RCS

Trend line relationships for RCS for each candidate sampler compared to the reference sampler are shown in Fig. 3. The equations for the slopes of the trend lines with the SIMPEDS were, GK2.69 = 1.05 (r ² = 0.88), GK 4.16 (RASCAL) = 1.04 (r ² = 0.98), PPI8 = 1.04 (r ² = 0.96) and FSP 10 = 0.97x – 0.032 (r ² = 0.67). None of these samplers provided a slope that was significantly different from the SIMPEDS sampler as the ideal relationship value of 1.00.

Figure 3.

Comparison of respirable crystalline silica concentrations determined with four high flow sampler types and the average of three SIMPEDS reference samplers.

Particle size

The summary statistics of particle size of aerosols measured by the API at the sampling position near the manikin are listed in Table 4 in the order of decreasing particle size (mass distribution). The median particle sizes are larger than those measured in the calm air chamber work (Stacey et al., 2014) and suggest a greater proportion of larger particles were available in the aerosol.

Table 4.

Aerodynamic particle size (µm) mass distribution for the four test aerosols

Dust type	Wind speed (m s−1)	Mean	Geometric standard deviation	Median	Mode
ARD coarse	2	22	1.98	25.5	42.5
ARD coarse	1	17.4	1.86	20.4	24.1
Mullite	1	14.9	1.9	17.4	23.5
Coal dust	1	7.62	1.61	7.85	11.8

Differences for FSP 10

The average respirable dust concentration value for the tests with the FSP 10 facing the emission source was higher (average ratio of candidate to reference sampler = 1.14) compared with those collected when side-on (average ratio of candidate to reference sampler = 1.06). The average ratio value for the side-on tests is influenced by low results obtained from the three repeat tests with the ARD at a wind speed of 2 m s⁻¹. Fig. 4 shows the difference in sampler performance at different orientations to the emission when measuring RCS concentrations. For these FSP 10 RCS concentration trend line equations, comparing side-on and facing tests are significantly different (P < 0.01). The coefficients of determination (r ²) for each trend line for side-on and facing the emission were 0.99 and 0.98, although, the number of RCS ratio tests is more limited (six results when face-on and five when side-on) than those used for assessment of respirable dust concentrations.

Figure 4.

Differences in RCS measurement with FSP 10 sampler orientation 95×92mm (600×600 DPI).

Comparison with previous calm air chamber work

A summary of trend line slope values between high flow rate respirable samplers and SIMPEDS reference sampler from both previous calm air chamber tests and the present wind tunnel testing is listed in Table 5. The coefficient of variation of the slope values in Table 5 for all conditions (calm air and wind tunnel) are 3, 11, and 7 % for the FSP 10, PPI 8, and GK 2.69 samplers respectively, indicating a close agreement. The slope values for the dust concentration ratios for the FSP 10, GK 2.69, and PPI 8 samplers when compared with the SIMPEDS are plotted in Fig. 5. The error bars shown indicate the range of ratio values and are the maximum and minimum ratios obtained for each candidate sampler, dust, and air speed.

Table 5.

Trend line slope values for respirable dust concentration ratios between high flow rate samplers and SIMPEDS reference sampler

Conditions	Source	Trend line slope value			Aerodynamic particle size (µm)	Particle size method	Comment
		FSP 10	PPI 8	GK 2.69
Calm air	Stacey et al. (2013)	1.04	NT	0.89	2.5 mass median	Marple impactor	Ultrafine grade ARD
Calm air	Stacey et al. (2014)	1.07	NT	0.91	2.8 mass median	API – real time	Ultrafine grade ARD
Calm air	Stacey et al. (2013)	1.05	NT	0.79	4.7 mass median	Marple impactor	Medium grade ARD
Calm air	Stacey et al. (2014)	1.07	NT	0.82	4.6 mass median	API – real time	Medium grade ARD
Calm air	Stacey and Thorpe (2010)	1.08	1.02	0.99	4.5 mass mean	Dust was collected with an IOM head onto a filter and then analyzed using the API	Fine grade ARD
Calm air	Stacey and Thorpe (2010)	1.08	0.92	0.91	4.5 mass mean		Powered saw
Calm air	Stacey and Thorpe (2010)	1.09	0.80	0.92	4.0 mass mean		Powdered hand drill
Wind tunnel	This study	1.21	1.04	0.93	7.8 mass median	API – analyzed in real time	Pittsburg coal dust
Wind tunnel	This study	1.10	0.97	0.96	17.4 mass median		Mullite Powder
Wind tunnel	This study	1.16	1.10	0.94	20.4 mass median		Coarse grade ARD Wind speed = 1 m s−1
Wind tunnel	This study	1.01	1.07	1.00	25 mass median		Coarse grade ARD Wind speed = 2 m s−1
Percentage coefficient of variation		3	11	7	For all trend line slope values and all conditions

NT, no test performed.

Figure 5.

Slope values for dust concentration ratios with the SIMPEDS reference sampler (combined results of previous calm air chamber studies and present study).

DISCUSSION

These samplers may provide useful options for occupational hygiene assessment for sampling periods within the 8-h reference period used to assess compliance with a WEL, as they operate at higher flow rates than are routinely used at present (typically ~2 l min⁻¹). All samplers obtained a reasonable comparison with the performance of the SIMPEDS. The GK 4.162 (RASCAL) and GK 2.69 samplers were in close agreement with the SIMPEDS sampler (Higgins–Dewell type design) in terms of accuracy.

Agreement with calm air data

Most laboratory based tests for respirable samplers have historically been conducted in calm air conditions that more typically represent the air speeds found in factories. This study assessed the differences in performance between respirable samplers in air speeds ≥ 0.5 m s⁻¹ that are more commonly found in mines, quarries and construction sites where there is frequent exposure to RCS. Differences between samplers can occur because of their slightly different particle size selection characteristics found for each respirable sampler (Stacey et al., 2013). For example, Lee et al. (2010) show that the sampling efficiency of the GK 2.69 sampler for 10 µm particles was ~9% higher than the respirable convention at the flow rate of 4.2 l min⁻¹. It’s possible that the GK 2.69 would sample a higher proportion of larger particles than other samplers, which may increase the mass concentration sampled relative to the SIMPEDS, as the mass median particle size of the aerosol increases. The GK 4.162 (RASCAL) sampler, is based on the same design as the GK 2.69 sampler (it has the same internal dimensions) and should therefore behave in a similar way. The FSP 10 sampler is based on the Higgins–Dewell design and should have a performance that is similar to the SIMPEDS. Both Higgins–Dewell design samplers tended to oversample the smaller (<4 µm) and undersample the larger (>4 µm) particle sizes relative to the convention. The FSP 10 design is based on a sampler developed by Cossey and Vaughan (1987) for the sampling of vents. The flow rate was originally set at 9.2 l min⁻¹ to meet the British Medical research Council d₅₀ cut-off value of 5 µm. The current flow rate for the FSP 10 sampler (10 l min⁻¹) was recommended by the manufacturer. Lee et al. (2010) obtained a d₅₀ cut-off value of 4.8 µm at this flow rate and found that the FSP 10 oversampled particles between 1 and 5.5 µm. The SIMPEDS has a d₅₀ value that is closer to 4 µm, so will collect fewer smaller sized particles than the FSP 10. As the median particle size of the aerosols increases and the proportion of smaller particles decreases at fixed geometric standard deviation (GSD), the performance of the FSP 10, and SIMPEDS may get closer. The PPI 8 sampler is expected to have a very close match with the ISO/CEN/ACIGH convention according to the work of Trakumas, and Slater (2009).

This work shows that the samplers generally agree closely with the SIMPEDS reference sampler (Fig. 5) for air speeds of 1 and 2 m s⁻¹ when challenged with a range of dust materials and sampling conditions. These data suggest that there is a potential to be able to apply correction factors to improve comparability between data because the average difference between samplers is consistent for different dusts and conditions. However, it might not be necessary to do this in these cases due to their closeness of agreement. Consistently higher slope values were obtained for the FSP 10 sampler and this may be because the flow rate was too low in order to match the ISO/CEN/ACGIH convention more correctly (Lee et al., 2010; Stacey et al., 2014). The GK 2.69 sampler obtained median mass concentration values that were comparable with values obtained with calm air chamber studies and slightly lower than the SIMPEDS. It is apparent from Fig. 5 that the performance of the GK 2.69 compared with the SIMPEDS when sampling different dusts is similar and there was no increasing tendency to collect more mass.

Inlet effects and air concentration differences

Respirable samplers are not expected to have the same issues with orientation or inlet effects compared with those that select the inhalable fraction (mostly less than 100 µm) because the respirable samplers collect finer particles (mostly less than 16 µm) that have lower inertia and are more likely to follow the inlet air streams. However, it has been reported that the Dorr Oliver cyclone under-sampled at 90° and oversampled at 180°, relative to the respirable convention at wind velocities greater than 1 m s⁻¹ (NIOSH, 2003). The efficiency of collecting particles (aspiration efficiently) is affected by many factors including, particle aerodynamic diameter, inlet velocity, inlet shape and dimensions, wind velocity, and wind direction. In these experiments, most inlet velocities for the candidate samplers (Table 1) are greater than the air speeds used and should have high aspiration efficiency in the respirable range. Differences may occur due to the design of the sampler and the presence of a body in the air stream. Air speeds approaching a body face-on will slow and the trajectory of larger particles may change whereas air speeds approaching a body side-on are more laminar and do not slow as much. These differences in air stream may cause mass concentration differences (Witschger et al., 1998). In this work, we compare the candidate samplers with a reference sampler positioned on the manikin.

Although many of the mass concentration values for the cyclone samplers were slightly lower when positioned side-on, the performance of candidate samplers relative to the SIMPEDS were not dependent on orientation to the source of the emission (Table 3). The coefficients of determination for the trend lines including data in all test conditions and orientations (Fig. 2) ranged from 0.99 to 0.97, which indicates a linear relationship with the performance of the SIMPEDS. Probability values for t-tests comparing side-on and facing tests ranged from 0.04 to 0.39, indicating no difference. Among the candidate samplers, the GK 2.69 and GK 4.162 appear to have the least difference with the SIMPEDS reference sampler. All the ratio values for ARD when the inlet of the GK 2.69 sampler was positioned side-on to the source of emission at a wind speed of 1 m s⁻¹ were 0.95 but this is not statistically significant as the range of ratio values includes 1. All the inlets for the SIMPEDS, GK 2.69, and GK 4.162 point down so the sampler collection efficiency which should not be affected by orientation even though the SIMPEDS has an inlet velocity ~67% less than the GK2.69.

It was anticipated that the FSP 10 might be influenced by orientation as it was originally designed with a forward facing protruding rectangular inlet for the sampling of dusts in vents (Table 1). At low air speeds, there should be little difference in the performance of this type from any other cyclone with a similar slot inlet. Like the other cyclone candidate samplers, many of the ratio values with the reference sampler are lower for the FSP 10 when side on to the emission but overall they were not significantly different which is confirmed by the overall trend in Fig. 2 which shows that there is no significant difference. A difference for the FSP 10 was potentially evident for the respirable dust concentration values obtained for the side-on tests of ARD at the highest air speed of 2 m s⁻¹ (Table 3). A simple two-sided t-test for unequal variances provided a low probability value that these results, collected at an air speed of 2 m s⁻¹, are comparable with other side-on orientated FSP 10 results (P < 0.01). This higher air speed may exaggerate the difference for the FSP 10. Analysis of a limited number of samples for RCS for each sampling position (Fig. 4) supports a hypothesis that increased variability might be due to a directional nature of the inlet. In this case air concentration results for the two sampler positions relative to the emission source are different (P < 0.01). In addition to the potential FSP 10 inlet effect for RCS measurement there are practical considerations to consider that aren’t measured in laboratory studies. The inlet on the FSP 10 projects ~10cm from the worker, and so may also be closer to the emission providing a source of bias, especially if the inlet extends beyond the breathing zone boundary.

CONCLUSIONS

All the samplers obtained good agreement with each other for different aerosols, and wind speeds.

A comparison of equivalent samplers with previous results collected with different dusts and different grades of ARD in calm air conditions and where the aerosols had a smaller median aerodynamic particle size, show that the relative collection efficiencies were consistent in various conditions. These relative differences are probably also consistent with workplace aerosols.

The GK 4.162 (RASCAL) and GK 2.69 samplers demonstrated the best agreement with the SIMPEDS (Higgins–Dewell design) at their respective flow rates and should thus provide good comparative values if future results are compared with the current exposure data.

The FSP 10 is also potentially a useful sampler but does show some indication of additional bias for RCS measurement when its inlet is facing an emission source. There is also debate over the best flow rate the FSP 10 should be operated at to ensure that it meets the ISO/CEN/ACGIH respirable convention.

These high flow rate samplers can provide useful options for the occupational hygienist to employ for short duration sampling events such as task specific sampling exercises less than the current 8h reference period and for 8h time weighted average measurements of low concentrations.

DISCLAIMER

This publication and the work it describes were co-funded by the Health and Safety Executive (HSE) in Great Britain and the National Institute for Occupational Safety and Health (NIOSH) in the United States. Its contents, including any opinions and/or conclusions expressed, are those of the authors alone and do not necessarily reflect HSE policy. The findings and conclusions in this presentation have not been formally disseminated by NIOSH and should not be construed to represent any agency determination or policy. Mention of any company or product does not constitute endorsement by the Centers for Disease Control and Prevention or NIOSH.