Comparative Evaluation of the Dustiness of Industrial Minerals According to European Standard EN 15051, 2006

Transcription

1 Ann. Occup. Hyg., Vol. 54, No. 2, pp , 2010 Ó The Author Published by Oxford University Press on behalf of the British Occupational Hygiene Society doi: /annhyg/mep077 Comparative Evaluation of the Dustiness of Industrial Minerals According to European Standard EN 15051, 2006 INGEBORG PENSIS 1 *, JOYCE MAREELS 2, DIRK DAHMANN 3 and DAVID MARK 4 1 Ankerpoort NV, Op de Bos 300, NL-6223 EP Maastricht, The Netherlands; 2 Redco NV, Kuiermansstraat 1, B-1880 Kapelle-op-den-Bos, Belgium; 3 Institut für Gefahrstoff-Forschung, Waldring 97, D Bochum, Germany; 4 Health and Safety Laboratory, Harpur Hill, Buxton, Derbyshire SK17 9JN, UK Received 6 May 2009; in final form 23 October 2009; published online 2 December 2009 A range of industrial minerals was tested using the rotating drum and the continuous drop methods, the two methods proposed by the published European standard EN [CEN. (2006) EN Workplace atmospheres measurement of the dustiness of bulk materials requirements and test methods. Brussels, Belgium: European Committee for Standardization], to evaluate and compare their dustiness. The assessment of bulk materials dustiness can help to develop less dusty products and to reduce dust exposure to the workers by improving the processing of minerals. The European standard EN (CEN, 2006) proposes a classification system that was developed with the intention to assist in the labelling of products in the future. This paper presents a comparison of both test methods in classifying industrial minerals. The correlation between the dustiness measured by the two methods for the inhalable and respirable fractions is given. The results show there is no unambiguous dependence of the dustiness on the grain size of an industrial mineral. Although dustiness can significantly be affected by product moisture, the influence of this parameter is not studied in detail as the industrial minerals were tested in the conditions they are sold, as the standard requires. Especially, the classification of substances with respect to different classes of dustiness was found to be problematic, as the two methods are by no means yielding identical classification groups for all the substances. In any use of the standard (EN 15051; CEN, 2006) for labelling purposes, a revision of the present classification system provided in the standard is required for industrial minerals. Keywords: dustiness classification; EN 15051; industrial minerals INTRODUCTION *Author to whom correspondence should be addressed. Tel: þ ; fax: þ ; i.pensis@ankerpoort.com The dustiness of a bulk product reflects the tendency of a material to produce airborne dust during handling. The extent of dust generation depends on a large number of parameters, including the type of processing, handling, size distribution, and humidity. Different dustiness testing methods based on different measurement principles are currently used in industry to determine the dustiness of materials. The European standard EN (CEN, 2006) was produced to ensure reproducibility of the dustiness measurement. Two standardized methods (i.e. the rotating drum and the continuous drop methods) are described in which materials are tested under uniform test conditions and dustiness results are reported in terms of health-related fractions. EN (CEN, 2006) presents a classification system to express the dustiness of a product as very low, 204

2 Comparative evaluation of the dustiness of industrial minerals 205 low, medium, or high. Dustiness tests may allow comparison of the relative dust exposure potential of different products (Heitbrink et al., 1990). In the industrial minerals sector, dust emission may pose health risks to workers. A variety of conditions during processing and handling of minerals, often in large quantities, might result in the generation of high levels of airborne dust. Grinding and milling of industrial minerals to produce powders are common practices. Depending on the type of mineral the inhalable, thoracic or respirable fractions (EN 481; CEN, 1993) may represent a potential health risk, e.g. crystalline silica is hazardous in the respirable fraction. Industrial mineral products are used in a wide variety of applications. They are essential raw materials in most manufacturing activities, including paint, electronics, metal casting and foundry, paper, plastics, glass, ceramics, detergents, pharmaceuticals, cosmetics, construction materials, and agriculture. They are also used as processing aids for the food and feed industries and have an increasingly important role in environmental engineering (e.g. water treatment). In view of the development and publication of the EN (CEN, 2006) standard, the Industrial Minerals Association of Europe (IMA-Europe) launched a large industrial project to analyse the dustiness of different industrial minerals using the two dustiness methods presented in the EN (CEN, 2006) standard. The goal of the project was to gain knowledge of the dustiness of mineral products, to compare the classification of the mineral products by the two standardized dustiness methods, and to study the potential value of dustiness in labelling of minerals as well as risk assessment of industrial workplaces. METHODS Samples A series of different industrial minerals from several member companies of IMA-Europe were chosen: quartz, cristobalite, feldspar, ball clay, kaolin, calcined clay, talc, marble, and chalk. The grain size distribution was the primary selection criterion for each industrial mineral. Samples of each of these minerals were collected from different geographical locations. These covered a range of different mineral product grades, i.e. with different grain sizes from coarse to fine. The samples were analysed for their dustiness and their respective size distributions were documented. Table 1 presents the main characteristics per mineral. The moisture content was not taken into account for sample selection. It was intended to perform the analyses on the mineral products as sold as EN (CEN, 2006) requires. The moisture content of all but one of the mineral product samples varied between 0 and 1.1%. For each industrial mineral product, two 5 kg samples were prepared from a homogeneous batch. Care was taken to seal the products to avoid changes in humidity and external contamination and to assure identical sample conditions. In total, 43 samples were sent to two independent laboratories directly by the respective member companies of IMA- Europe, each laboratory being equipped for one of the testing methods. Participants provided general information on the type of mineral and the chemical composition to the analytical laboratory. Each sample was assigned a reference number for anonymity by IMA-Europe. IMA-Europe asked the two laboratories to test the samples according to EN (CEN, 2006), which implies the measurement of moisture content, bulk density, and dustiness. Methodology The Institut für Gefahrstoff-Forschung (IGF, Bochum, Germany) determined the dustiness using the continuous drop method (Dahmann and Möcklinghoff, 2000). The continuous drop method consists of continuously pouring product downwards from a sample tank with the help of a metering device into a vertical pipe. Generally, test material is dropped at a flow rate of 8 g min 1 ( 2 g min 1 ) during 10 min into a slow vertical air current. Through the bottom of the backflow pipe, air is drawn inwards at a flow rate of 53 l min 1 (lift velocity 0.05 m s 1 at a pipe diameter of 150 mm). This upstream airflow carries the dust released from the dropping material to the sampling section. The inhalable and the respirable fractions are sampled simultaneously at a flow rate of 2 l min 1 on, respectively, a filter thimble mm and a membrane filter with 37 mm diameter. Both samplers comply with EN (CEN, 2001). The final result is an average of five tests. With this technique, only the inhalable and respirable fractions are collected. The thoracic dustiness mass fraction is not measured. In this project, a total weight of at least 500 g to 2 kg was used for the five runs together. Several factors rule the required amount of material to be tested. It depends on the grain size of the sample, the density of the material, and the dustiness itself. The detection limit of the gravimetric analysis of the filters

3 Table 1. Main characteristics of the samples by mineral Mineral Feldspar Quartz Cristobalite Talc Calcined clay Clay Kaolin Marble Chalk Number of samples Bulk density (kg m 3 ) Mean (SD) 1126 (243) 1348 (386) 982 (268) 451 (241) 1245 (250) 754 (335) 764 (202) 962 (289) 1268 (321) Minimum maximum Moisture (%) Mean (SD) 0.10 (0.13) 0.02 (0.02) 0.02 (0.03) 0.14 (0.14) 0.07 (0.06) 1.03 (0.12) 0.87 (0.23) 0.07 (0.06) 0.03 (0.01) Minimum maximum Median particle size a (lm) Mean d 50 (SD) 113 (194) 149 (152) 97 (131) 5.9 (5.7) 90 (130) 0.8 (0.3) 1.7 (0.9) 12.6 (15.2) 162 (250) Minimum maximum d 97 /d 10 ratio Mean (SD) 17.9 (21.7) 44.3 (71.5) 24.5 (21.4) 17.1 (8.5) 83.8 (27.6) 18.5 (9.2) 36.0 (17.2) 29.6 (13.4) 48.7 (55.5) Minimum maximum Specific surface area b (m 2 g 1 ) N n/a Mean (SD) 2.2 (2.5) 1.2 (2.1) 0.4 (0.5) 8.4 (6.2) (1.8) 2.1 (1.7) 1.5 Minimum maximum Quartz content (%) Mean (SD) 15.5 (11.7) 98.9 (1.0) 0.6 (0.2) 0.5 (0.2) 40 ( ) 2.6 (3.4) 0.9 (0.7) 0.1 ( ) 0.2 ( ) Minimum maximum I. Pensis et al. BET, Brunauer-Emmett-Teller; n/a, not analysed. a The d 50 is determined by laser diffraction (feldspar, quartz, cristobalite, marble, and chalk), sedigraph (talc, calcined clay, clay, and kaolin), or sieve analysis(coarse products). b The specific surface area is determined by BET if analytically feasible.

4 Comparative evaluation of the dustiness of industrial minerals 207 determines the amount of material to be dropped. Higher volumes are required for less dusty products, which may even lead to the necessity to extend the duration of the dustiness measurement. As the result of the experiments is principally independent of the duration of the measurements, this is explicitly allowed in the standard. The Health and Safety Laboratory (HSL, Buxton, UK) used the rotating drum method (Burdett et al., 2000). The rotating drum consists of a 300-mm stainless steel drum that is equipped with eight internal longitudinal vanes. Test material with a standard volume of 35 cm 3 is spread evenly along the bottom of the drum. The drum rotates at 4 r.p.m. lifting and tumbling the test material continuously by the vanes during 1 min. A slow winnowing air current with mean velocity of 0.1 m s 1 (provided by a pump at 38 l min 1 ) conducts the dust released from the dropping material from the drum to a three-stage dust sampling system. The sampling system comprises two particle size selective metal foam stages followed by a backup filter to enable the determination of the inhalable, thoracic, and respirable fractions of the released dust (in accordance with EN 481). The first stage comprises two 20 pores per inch (p.p.i.) foams of 10 mm thickness, while the second stage uses a single 12-mm thick 80 p.p.i. foam. The filter media used depends on subsequent analysis and the filters and foams have a diameter of 80 mm. The inhalable fraction is determined by adding together the dust collected on the two foam stages and the filter. The thoracic fraction is determined by adding the dust collected on the second foam stage to that on the filter, while the respirable fraction is determined from the dust collected on the filter. The dustiness determination comprises at least three replicate runs. The drum is cleaned between successive runs and fresh foams and filters are used. Representative samples are prepared from the bulk material according to DIN (DIN, 2006) by IGF and BS (BS, 1986) by HSL. Dustiness determination of the tested material is performed by weighing the filters and/or foams before and after the test. The dustiness mass fraction for each health-related fraction, expressed in milligrams per kilogram, is then calculated by dividing the mass of dust collected from each health-related fraction in milligrams by the mass in kilograms of the material placed in the drum or sample tank for each replicate test. At the HSL, the average value and coefficients of variation of the three replicate tests are determined. The coefficient of variation must be,10%. More runs are carried out if this value is exceeded. At the IGF, the mean value and relative standard deviation of five runs are calculated. Statistical analysis The geometric mean (GM) and geometric standard deviation (GSD) of the inhalable, thoracic, and respirable dustiness mass fractions for both continuous drop (five replicate runs) and rotating drum (three replicate runs) methods are determined for each individual sample. The GSD has values between 1.01 and 1.75 for continuous drop method for inhalable and respirable mass fractions and between 1.00 and 1.32 for rotating drum method considering all health-related mass fractions. An overview of the ranges of dustiness mass fractions is presented in Table 2 per mineral, health-related mass fraction, and dustiness method. Note that the GM of the dustiness measurements by mineral should not be considered as an intrinsic characteristic of this mineral as it is only representative for the samples selected for the project. Three samples (respirable fraction) and one sample (inhalable fraction) have dustiness values below the limit of detection (LOD) when measured by the continuous drop method. All samples have dustiness values higher than the LOD for the rotating drum method. The dustiness values below the LOD were replaced by half of the LOD value. Further details on the dependence of dustiness for each mineral in terms of the mineral supplier or particle size are intentionally not discussed here on account of competition between the different industrial mineral suppliers. A statistical analysis was performed with Stata- Corp statistical software (Stata Release 10, 2007). Since both the inhalable and the respirable dustiness mass fractions show highly skewed distributions, all data were log-transformed. In a first stage, the factors that may have a statistically significant influence on the dustiness were studied. A multiple general linear procedure was used to describe the statistical relationship between the dependent variable (either respirable or inhalable dustiness as measured by continuous drop or rotating drum method) and the independent variables (d 50, d 97 /d 10, specific surface area, bulk quartz content, and type of mineral). The particle size of each sample is expressed by the d 50, i.e. the particle diameter at which 50% by volume of the mineral product is smaller than this value. The d 97 /d 10 ratio represents the width of the particle size distribution. A d 97 /d 10 ratio 5 1 means that the material consists of particles with almost the same size and is characterized by

5 Table 2. Dustiness mass fractions of the samples by mineral according to the rotating drum and continuous drop methods Mineral Feldspar Quartz Cristobalite Talc Calcined clay Clay Kaolin Marble Chalk Number of samples Rotating drum Inhalable fraction (mg kg 1 ) GM (GSD) 3037 (2.4) 130 (9.8) 750 (4.4) 2330 (2.9) 1829 (2.8) 825 (2.0) 276 (3.0) 444 (5.2) 881 (3.1) Minimum maximum Thoracic fraction (mg kg 1 ) GM (GSD) 1231 (1.7) 69 (8.5) 450 (4.2) 858 (2.0) 969 (2.6) 262 (1.9) 120 (2.7) 211 (3.3) 513 (3.2) Minimum maximum Respirable fraction (mg kg 1 ) GM (GSD) 339 (1.8) 28 (6.7) 134 (6.9) 187 (1.6) 306 (2.2) 56 (1.9) 43 (1.5) 92 (2.7) 163 (2.6) Minimum maximum Continuous drop Inhalable fraction (mg kg 1 ) GM (GSD) (2.6) 415 (51) 2666 (8.1) (2.9) (1.4) 6054 (2.2) 5072 (1.1) 7605 (5.0) 3893 (1.2) Minimum maximum Respirable fraction (mg kg 1 ) GM (GSD) 123 (1.6) 21 (5.0) 54 (7.8) 743 (3.2) 272 (1.2) 59 (2.4) 52 (1.4) 119 (4.8) 88 (1.5) Minimum maximum Ratio of dustiness mass fractions (continuous drop/rotating drum) Inhalable fraction GM (GSD) 3.75 (2.46) 3.20 (9.51) 3.55 (1.86) 8.91 (1.96) 8.99 (2.77) 7.34 (4.04) 10.2 (1.4) 17.1 (1.4) 4.42 (3.72) Minimum maximum Respirable fraction GM (GSD) 0.36 (2.32) 0.77 (2.21) 0.40 (1.44) 3.97 (2.84) 0.89 (2.51) 1.05 (1.31) 0.97 (1.15) 1.29 (2.24) 0.54 (1.76) Minimum maximum I. Pensis et al.

6 Comparative evaluation of the dustiness of industrial minerals 209 a very narrow particle size distribution. As the difference between d 97 and d 10 values increases, the particle size distribution becomes broader and the d 97 /d 10 ratio will strongly increase. Few samples are characterized by a very high d 97 /d 10 ratio, which is mainly due to a very small denominator. As a result, the ratios of d 97 to d 10 tend to show a skewed distribution. The logarithm of the d 97 /d 10 ratio is therefore used in the model. The d 50 and specific surface area data do not show a skewed distribution to the same extent. No transformation is applied on these variables. First, all independent variables were included in the model (all factors). Subsequently, the modelling is repeated removing each time the factor having the highest P-value until all factors have a P-value,10% in the model (selected factors). The statistical fit of the modelling is checked by examining the normality and stability of the model residuals. In a second stage, the dustiness classification system is addressed. Three aspects are considered. The main question is to what extent the methods are comparable. Correlation coefficients for the different dustiness fractions (respirable and inhalable) are determined to evaluate the correspondence between the continuous drop and rotating drum methods. The Cohen s Kappa is calculated and is an indication for the degree of agreement of dustiness classification by the two dustiness methods. Secondly, it is investigated if the limits between the different classes can be transposed to obtain a better agreement between the two methods for the different classes. Finally, the last question of interest addresses whether the classifications given for the two methods are influenced by different factors. RESULTS Factors influencing the dustiness mass fractions The results of the modelling are presented in Tables 3 6. When modelling the inhalable and respirable mass fractions as measured by continuous drop and rotating drum methods considering all factors, similar results are obtained. Statistically significant factors are the log(d 97 /d 10 ) ratio and the type of mineral, when all samples are included in the model. However, it is found that the influence of the width of the particle size distribution log(d 97 /d 10 ) is important due to the seven samples having a d lm. The inhalable dustiness increases with increasing log(d 97 /d 10 ) in the group of samples with a d lm (Fig. 1a). In contrast, there is no correlation at all between these parameters in the finer materials (d 50, 100 lm) (Fig. 1b). The same trend is observed for inhalable and respirable mass fractions regardless of the dustiness method used. The results were therefore examined in more detail considering only the samples having a d 50, 100 lm. The group of samples with a d lm is not further analysed since the number of samples (seven) is considered too low. When restricting the analysis to samples with d 50, 100 lm, the inhalable mass fractions are significantly influenced by particle size d 50, specific surface area, and mineral type. The dustiness increases significantly with the particle size (continuous drop: P , þ1.6% per lm; rotating drum: P , þ1.7% per lm) and decreases significantly with the specific surface area (continuous drop: P , 3.5% per m 2 g 1 ; rotating drum: P , 4.2% per m 2 g 1 ). Similar results are obtained for the respirable mass fraction. Considering the samples with a d 50, 100 lm, the dustiness decreases with increasing specific surface area (continuous drop: P , 8.6% per m 2 g 1 ; rotating drum: P. 0.10) and increases with the particle size (continuous drop: P , þ3.2% per lm; rotating drum: P , þ3.7% per lm). Classification of the dustiness results Comparisons of the dustiness mass fractions results obtained by the continuous drop and rotating drum methods are shown for the respirable and inhalable fractions in Fig. 2a,b. The classification system according to EN (CEN, 2006) is also plotted. Samples within the bold rectangles fall within the same classification when tested with the two dustiness methods. Samples outside the bold rectangles have a different dustiness classification. The GSD can be neglected because the spread of results is insignificant when compared with the width of a class. The continuous drop method gives a lower classification of dustiness for the respirable fraction. For the inhalable fraction, the rotating drum tends to give a lower classification for many industrial minerals product grades. The correlation coefficient between the two methods is not very good on a linear scale. The correlation coefficient (r) has values of 0.05 and 0.59 for, respectively, respirable and inhalable mass fractions. However, when considering log-transformed data, the correlation is much better with r and r for, respectively, respirable and inhalable mass fractions. According to EN (CEN, 2006), the dustiness of materials is classified as very low, low,

7 210 I. Pensis et al. Table 3. Results of multiple general linear model for inhalable fraction measured by continuous drop method All factors Selected factors Degrees of freedom Coefficient P-value Degrees of freedom Coefficient P-value All samples Log(d 97 /d 10 ) Bulk quartz (%) d 50 (lm) SSA (m 2 g 1 ) Minerals N R Samples with d 50, 100 lm Log(d 97 /d 10 ) Bulk quartz (%) d 50 (lm) SSA (m 2 g 1 ) Minerals N R SSA, specific surface area. Table 4. Results of multiple general linear model for inhalable fraction measured by rotating drum method All factors Selected factors Degrees of freedom Coefficient P-value Degrees of freedom Coefficient P-value All samples Log(d 97 /d 10 ) Bulk quartz (%) d 50 (lm) SSA (m 2 g 1 ) Minerals N R Samples with d 50, 100 lm Log(d 97 /d 10 ) Bulk quartz (%) d 50 (lm) SSA (m 2 g 1 ) Minerals N R moderate, or high. To evaluate to which extent the materials are classified similarly, the number of samples are counted for each combination of categories for the two dustiness methods for the inhalable and respirable mass fractions (Table 7). The Cohen s Kappa coefficient is calculated and has a value of 0.20 for the inhalable and 0.23 for the respirable mass fractions. It confirms there is almost no agreement in classification by the two dustiness methods. For the inhalable fraction, the samples falling in the category low according to the rotating drum ( mg kg 1 ) are almost not classified as low by the continuous drop method. Ten of 12 samples are classified as moderate (

8 Comparative evaluation of the dustiness of industrial minerals 211 Table 5. Results of multiple general linear model for respirable fraction measured by continuous drop method All factors Selected factors Degrees of freedom Coefficient P-value Degrees of freedom Coefficient P-value All samples Log(d 97 /d 10 ) Bulk quartz (%) d 50 (lm) SSA (m 2 g 1 ) Minerals N R Samples with d 50, 100 lm Log(d 97 /d 10 ) Bulk quartz (%) d 50 (lm) SSA (m 2 g 1 ) Minerals N R Table 6. Results of multiple general linear model for respirable fraction measured by rotating drum method All factors Selected factors Degrees of freedom Coefficient P-value Degrees of freedom Coefficient P-value All samples Log(d 97 /d 10 ) Bulk quartz (%) d 50 (lm) SSA (m 2 g 1 ) Minerals N R Samples with d 50, 100 lm Log(d 97 /d 10 ) Bulk quartz (%) d 50 (lm) SSA (m 2 g 1 ) Minerals N R mg kg 1 ) and one sample even as high ( mg kg 1 ). Similarly, 10 of 21 samples classified as moderate according to the rotating drum ( mg kg 1 ) are classified as high using the continuous drop method. This is in line with the high ratios of the two methods shown in Table 2. Considering the mass fraction limit at 1000 mg kg 1 from the rotating drum classification scheme, the best corresponding mass fraction limit for the continuous drop method could be 8500 mg kg 1 at which 86% of the samples could be classified correctly based on the rotating drum method as either very low/low or moderate/high. This correspondence is only valid when considering two categories instead of four. Conversely, if one takes the continuous drop mass fraction limit at mg kg 1 between very low/low/moderate and high (only two categories), the best corresponding mass fraction

9 212 I. Pensis et al. Fig. 1. Width of the particle size distribution (d 97 /d 10 ) versus the inhalable mass fraction measured by the rotating drum for samples with a particle size (a) d lm and (b) d 50, 100 lm. limit for the rotating drum would be 2200 mg kg 1, at which 74% of the samples could be classified the same using the continuous drop method. With respect to the respirable fraction, the agreement between the two methods is better as reflected by the ratios of the two methods which fluctuates around 1 (Table 2). However, the continuous drop method does not separate well between the categories moderate and high of the rotating drum (Fig. 2). If one takes the 250 mg kg 1 cut point for the rotating drum, separating moderate and high, the best corresponding cut point for the continuous drop method would be 120 mg kg 1, at which 66% of the samples would be classified correctly (according to the rotating drum method) as either very low/low/ moderate or high. Conversely, if one takes the continuous drop 125 mg kg 1 cut point between very low/low and

10 Comparative evaluation of the dustiness of industrial minerals 213 Fig. 2. Dustiness as measured by the rotating drum versus the continuous drop methods for (a) inhalable and (b) respirable mass fractions. The classification limits, as published in EN (CEN, 2006), are shown with the respective absolute numbers. The bold rectangles show the areas where both measurement methods give the same classification. moderate/high, the best corresponding cut point for the rotating drum would be 110 mg kg 1, at which cut point 86% of the samples would be classified correctly (according to the continuous drop) method as either very low/low or moderate/high. In order to investigate whether the two dustiness measurement methods depend on different factors, the log ratio of the continuous drop to the rotating drum dustiness mass fractions is modelled using the multiple general linear model. Considering the inhalable dustiness, the GM of the ratios of the two methods is 1.30 and 8.85 in the groups of samples with a d 50 of, respectively,.100 and,100 lm. In the group of finer materials (d 50, 100 lm), the type of mineral is identified as an insignificant though borderline factor (P )

11 214 I. Pensis et al. Table 7. Cross-tabulation presenting the number of samples in the classification schemes of the rotating drum and continuous drop methods for (a) inhalable and (b) respirable mass fractions Rotating drum Total Very low resulting in a different response of the two dustiness methods. A similar modelling of the log ratios of the respirable dustiness mass fractions of the continuous drop to the rotating drum methods did not reveal any significant differently influencing factors except the type of mineral (P, 0.001). The mean of the ratios of the two measurement methods has values of 0.43 and 1.37 in the groups of samples with a d 50 of, respectively,.100 and,100 lm. DISCUSSION Low Moderate High (a) Inhalable dustiness Continuous drop Very low Low Moderate High Total (b) Respirable dustiness Continuous drop Very low Low Moderate High Total The behaviour of materials with respect to dustiness seems to be determined by the nature of the material and the fineness. The statistical analysis reveals that the influencing factors are different when considering samples with a particle size (d 50 ),100 or.100 lm. Different physical properties seem to determine the dustiness of materials. For example, electrostatic forces tend to stick fine particles together reducing the liberation of dust. These forces do not play a significant role in coarser materials. Therefore, the assumption that finer materials will show higher dustiness values, which is currently still reported in several publications (Bundesanstalt für Arbeitsschutz und Arbeitsmedizin, 2006), is not supported by these findings. Furthermore, the situation is much more complicated since, for many minerals, with increasing grain size the dustiness initially increases and then decreases. For example, the dustiness of quartz and cristobalite increases for samples with d 50 up to 150 lm and then suddenly decreases in coarser mineral product grades. However, feldspar displays a similar dustiness regardless of the grain size of the material. These findings are generally in line with earlier results (Upton et al., 1990; Dahmann and Möcklinghoff, 2000). EN (CEN, 2006) proposes two methods to measure the dustiness of bulk materials. The aim of the standard is to classify each mineral product grade according to their tendency to produce dust during handling. Ideally, regardless of the method used, the same classification should be obtained for each individual mineral product grade. Dustiness results according to the two test methods are expected to be different. Therefore, when defining dustiness classes (in terms of very low, low, moderate, and high) the limits between these classes are different for the two methods. The different absolute dustiness mass fraction values, which define the various classes, are a result of the different generation techniques and the different energy supplied in the two laboratory methods. It should be noted that the dustiness mass fraction limit values are based on a comparison of only 12 different materials in an European Unionfunded project SMT4-CT (Burdett et al., 2000). The IMA-Europe dustiness project shows, however, that 50% of the 42 tested samples are differently classified for the respirable fraction. For the inhalable fraction even more of the samples, 60%, enter a different dustiness category. The project demonstrates that, when applied to a number of industrial minerals, the two methods often give a different classification and therefore the aims of the standard are not currently achieved in this respect; see Fig. 2a,b. The arbitrary cut-off of the categories as defined in EN (CEN, 2006), based on a limited number of reference samples, poses problems. At this stage, the results do not explain the factors that may cause a single sample to fall within two different classifications. Based on the 42 samples measured by the continuous drop and rotating drum methods, a shift of one of the classification limits for one of both methods results in a better agreement of the classification by the two methods. Nevertheless, this is only achieved when considering two categories instead of four, e.g. very low/low and moderate/high. Moreover, it is strongly questionable whether a correlation of as low as 66% of the samples is acceptable. Both methods simulate different handling scenarios. The rotating drum tester is basically a device that lifts and drops the same material many times and so

12 Comparative evaluation of the dustiness of industrial minerals 215 it can simulate both dropping and mixing. The continuous drop uses fresh material all through the 10- min run and can therefore simulate the loading of trucks by a conveyor belt. The relatively weak correlation between both methods for the respirable and inhalable fractions frequently indicates a dissimilar ranking of the mineral products. It is questionable whether a wellfitting correlation between both methods exists (Hamelmann and Schmidt, 2003; Lidén, 2006; Bach and Schmidt, 2008). The different classification of many materials by the rotating drum and continuous drop methods emphasizes that dust exposure is strongly related to the type of handling. In principle, the dustiness might be a very valuable tool for predicting the potential exposure during a specific type of handling (Brouwer et al., 2006). Considering one single industrial scenario, the dustiness could be very helpful to discriminate between more and less dusty mineral product grades. Nevertheless, one should not overlook the fact that in industrial processes there are many factors that affect occupational exposure. Therefore, dustiness values must not be used in isolation and the selection of the test method (rotating drum or continuous drop) should be done with a clear view of the materials and their intended use. Following the results of this study, the use of dustiness for labelling mineral products is contentious. It is not possible to use a single label to reflect all industrial scenarios. This may mean that the manufacturer needs to carry out a range of dustiness tests depending upon which specific process the material is being subjected to. The relationship between the dustiness and the moisture content of the samples, and any variation during the determination, could not be investigated. The moisture content of all samples provided in this study was very low. The standard deliberately demands that samples are to be analysed as they are used, which means without an artificial drying or conditioning step. This decision was taken as for industrial workplaces the dust generated will always result from unconditioned materials. In addition, depending on the nature of the investigated products even a freshly conditioned material will change its moisture content during the analytical investigation. Furthermore, the dustiness of a given industrial material depends on a large number of additional factors that will cause a certain measurement uncertainty in itself (Plinke et al., 1992, 1995; Hjemsted and Schneider, 1996). Therefore, the fixation of arbitrary classification limits will influence the findings of the test quite intensively. For example, a material with a dustiness in the middle of one of the classification fields will be classified unambiguously and identically by both methods, whereas a different material with a dustiness closer to one of the limits of the classification fields will be prone to different classifications by the two methods. As the whole procedure is intended to facilitate the risk assessment of users and in consequence to improve working conditions by making use of market mechanisms, this situation needs to be improved. CONCLUSION The measurement of dustiness by the rotating drum and the continuous drop methods for a large range of industrial mineral products shows that a unique classification is not obtained for many mineral products. Clearly, when the standard was established, a limited number of reference samples of minerals were considered. The difference in classification by the two dustiness methods shows that the type of handling has a strong influence on the dust generation. Dustiness is therefore a very valuable tool for risk assessment in industrial workplaces, as the methods of EN do yield relevant information for an estimation of dust exposure by handling a variety of industrial products. In addition, the existence of different methods allows a better approach of the different risks at the workplace. However, the classification system as provided in the present standard (EN 15051; CEN, 2006) is inadequate for classification and labelling of industrial minerals. The responsible standardization committee has therefore agreed to take these considerations into account. FUNDING Industrial Minerals Association Europe, Brussels, Belgium. Acknowledgements The authors thank the IMA Metrology Working Group for their support and advices and acknowledge the help of Dr Pascal Wild PW Statistical Consulting in the statistical analysis of the data. REFERENCES Bach S, Schmidt E. (2008) Determining the dustiness of powders a comparison of three measuring devices. Ann Occup Hyg; 52: Brouwer DH, Links IHM, De Vreede SAF et al. (2006) Size selective dustiness and exposure; simulated workplace comparisons. Ann Occup Hyg; 50: BS. (1986) BS Methods for determination of particle size distribution. Guide to powder sampling. Bundesanstalt für Arbeitsschutz und Arbeitsmedizin. (2006) Einfaches Maßnahmenkonzept Gefahrstoffe. Available at

13 216 I. Pensis et al. EMKG.html nnn5true. Accessed 17 January Burdett GJ, Chung KYK, Mark D et al. (2000) Development of a method for dustiness testing, EU contract SMT4-CT HSE report IR/L/MF/00/11. Sheffield, UK: Health and Safety Laboratory. CEN. (1993) EN 481 Workplace atmospheres: specification for conventions for measurement of suspended matter in workplace atmospheres. Brussels, Belgium: European Committee for Standardization. CEN. (2001) EN Workplace atmospheres. Assessment of performance of instruments for measurement of airborne particle concentrations. Brussels, Belgium: European Committee for Standardization. CEN. (2006) EN Workplace atmospheres measurement of the dustiness of bulk materials requirements and test methods. Brussels, Belgium: European Committee for Standardization. Dahmann D, Möcklinghoff K. (2000) Das Staubungsverhalten quarzfeinstaubhaltiger Produkte. Gefahrst Reinhalt Luft; 60: DIN. (2006) DIN Testing of solid fuels - Sampling and sample preparation - Part 3: Sample preparation. Hamelmann F, Schmidt E. (2003) Methods of estimating the dustiness of industrial powders a review. KONA; 21: Heitbrink WA, Todd WF, Cooper TC et al. (1990) The application of dustiness tests to the prediction of worker dust exposure. Am Ind Hyg Assoc J; 51: Hjemsted K, Schneider T. (1996) Documentation of a dustiness drum test. Ann Occup Hyg; 40: Lidén G. (2006) Dustiness testing of materials handled at workplaces. Ann Occup Hyg; 50: Plinke MAE, Leith D, Boundy MG et al. (1995) Dust generation from handling powders in industry. Am Ind Hyg Assoc J; 56: Plinke MAE, Maus R, Leith D. (1992) Experimental examination of factors that affect dust generation by using Heubach and MRI testers. Am Ind Hyg Assoc J; 53: Upton SL, Hall DJ, Marsland GW. (1990) Some experiments on material dustiness. Proceedings of the Aerosol Society Annual Conference. Surrey, UK, April 1990.