Methods for the preparation of the test sample from the laboratory sample (Part 1 Material & Methods) Work Package 6 Task 6.4 Validation of prcen/ts 15413 (WI 343027) Paolo de Zorzi APAT - Italian Environmental Protection Agency, Environmental Metrology Unit Workshop on Classification, characterization and quality management of solid recovered fuels Rome (Italy), 24 October 2007
Contents Objectives Material & Methods Method to be validated Samples tested Experimental design Equipments
Objective Validation of prcen/ts 15413 (WI 343027) Evaluation of the robustness of the method for preparation of test sample, that is its ability to give consistent results under controlled variations of key analytical parameters/conditions Experimental activity performed by ENEL GEM AT Ricerca (Brindisi) and by APAT (Italian Environmental Protection Agency) - Environmental Metrology Unit (Rome)
Content of prcen/ts 15413 (1) Specifies the correct sequence of operations to control and maximize the representativeness of the test portions prepared from the laboratory sample, prior to physical and/or chemical analysis of solid samples. It is largely based on EN 15002 developed by CEN TC 292 for waste samples.
Content of prcen/ts 15413 (2) A flow sheet for the definition of sequence of operations: applied on the laboratory sample, repeated on all sub-samples subsequently obtained, iterative cycle until all analytical requirements are fulfilled; Guidelines for choosing and applying sample treatment techniques: Homogenization and fraction separation, drying, particle size reduction and sub-sampling (Annex A normative); Relationship between minimum amount of sample, particle size and (sub-)sampling error application of a statistical formula (Annex B).
From Laboratory Sample to the measurement result analysis Laboratory sample Sub sample Test sample Test portion A lot of operation can be performed using different devices modifying: Sample size (e.g. splitting, rieffling, cone and quartering) Particle size (comminution) Heterogeneity
Evaluation of robustness Key variables must be selected among the main degrees of freedom of each method; For each key variable, conservative conditions should be met; i.e. when the methods passes the ruggedness testing in those conditions, it will be actually robust in most real life cases. The kind of sample to be assigned to each method should emphasize the influence of the key variable under control. Key variables No of levels Notes Composition 2 Two kind of materials (MSW and wood) Type of particle size reduction system 3 MSW sample 3 Wood sample Total levels 6
Samples chosen for the evaluation of ruggedness municipal solid waste (QR-D): Produced from the combustible fraction of msw, containing small pieces of tyre residues, passing through a 10 cm grid. demolition wood (QR-B): Residual of wood at different size, generally as flashes, blocks, lamellas and flakes, often longer > 10 cm together with wood powder.
Experimental (1) Preliminary tests identification of applicable treatment procedures (tests on milling, freeze-milling and cutting devices with different sieve) evaluation of most interesting chemical parameters to be tested, and their approximate levels (moisture, ash, LOI, minor/major elements) approximate assessment of the actual heterogeneity of the sample with respect to several parameters
Experimental (2) Actual evaluation of ruggedness Application and verification of the statistical formula for the determination of the minimum mass of sample as function of particle size, for a number of analytes (Ash, Moisture, several elements) Evaluation of effect of different particle size reduction devices on: loss of analytes (mostly volatile ones, like moisture, Hg, other volatile elements ), mass recovery, resulting homogeneity on two different kind of samples
The statistical formula (Annex B) Equation, based on the Gy s sampling theory, used for the estimation of the minimum mass of sample to be analysed as function of some characteristics (particle size, shape, etc.) and some errors associated to sampling (fundamental error)
The statistical formula (Annex B) 1 (1 - p) 3 M sam = π ( D95 ) s ρ g 2 6 CV p Msam: the mass of the sample in g; D95: the maximum particle size (defined as the 95-percentile), in cm; s: the shape factor of particles (s = V95 / D95l3); ρ: the average density of the particles in the material, in g/cm3; g: the correction factor for the particle size distribution of the material; p: the fraction of the particles with the property of interest; generally, this value is assumed to be relatively high (p 0.1) for major parameters, that are likely to be homogeneously distributed, and lower for less homogeneously distributed and/or minor components (p 0.01 or less); CV: the desired coefficient of variation caused by the fundamental error. The lower CV (Coefficient of variation) the better representativeness
Application and verification of the statistical formula (MSW sample has been used, ρ = 2 g/cm3 and g = 0.25 are assumed) p 0.1 (major) 0.001 (minor) D95 (cm) 10 10 s 0.013 0.013 CV 0.1 0.1 M (g) 2988 331630 Sample (D95 = 10 cm) The formula applied on original sample (D95 = 10 cm) leads to minimum mass of sample too high; it has been necessary to reduce the particle size (D95 = > 0.95 cm < 0.95 cm Sample (D95 = 0.95 cm) 0.95 cm), in order to allow more practical mass of test portion. A practical mass of test portion has been fixed to about 50 g, the expected CV has been calculated for several cases of p-factor: for major and homogeneously distributed components a CV less than 0.1 (i.e. RSD<10%); for less homogeneously distributed components, a CV of 0.17 or higher (i.e. RSD>17%). Cutting 0.4 cm 0.95 cm sieve p 0.1 0.01 0.005 0.001 D95 (cm) 0.95 0.95 0.95 0.95 s 0.0328 0.0328 0.0328 0.0328 CV 0.0365 0.12 0.17 0.38 M (g) 50 50 50 50
Sub-sampling for verification of statistical formula 125 g 2 125 g 500 g 62.5 g 2 1000 g 187.5 g 62.5 g 16 sub-samples (A.. R) of about each have been prepared from 1 kg of material (D95=0.95 cm). 500 g Five of them (A, D, G, L, O) have been used for the verification of the statistical formula. The others will be used later for evaluation of ruggedness. 812.5 g = 800 g 400 g 400 g 200 g 200 g 100 g 100 g 100 g 200 g 200 g 100 g 100 g 100 g 100 g 100 g A 47.06 g B 48.1 g C 55.4 g D 49.13 g E 35.3 g F 52.27 g G 46.20 g H I 53.15 g L 57.18 g M 51.26 g N 48.28 g O 51.75 g P 50.17 g Q 50.28 g R Samples used for evaluation of the statistical formula
Treatment and analysis (verification of statistical formula) Sample SM 2000 Cutting mill 2 mm ZM-1 centrifugal mill X6 0.5 mm X6 Moisture + ash Elements (WI 343012) (ICP-OES) CHN (WI 343020)
Considerations on statistical formula (1) compromise between the mass of (sub) sample to be analyzed, its main physical characteristics (i. e. particle size and shape), and the expected fundamental error associated with (sub) sampling. Overall variability includes the variability due to sampling, preparation and analysis; Preparation and analysis contribute to the overall estimated variability for about 5% for most of components determined in this study. This must be taken into account in evaluation of analytical results
Considerations on statistical formula (2) For major components (i. e. ash, moisture, C, Ca, K, Mg, Na, Si), the RSD associated with (sub) sampling is about 10% or less, as expected with factor p from 0.1 to 0.01. For other major components (Al, Cu, Fe and Ti), RSD is above 25%, despite their high concentration. Even an application of factor p = 0.01 is inadequate. For minor elements (Mn, Ni, Sr, and W) the RSD is 10% - 30% as expected with p = 0.001; Cr, Zn and mostly Ba show higher RSD. For the application of the statistical formula, criteria for the choice of factor p can not be limited to the relative abundance of the analytes of interest (the higher concentration the higher p), because there might be some dramatic exceptions of heterogeneous major components (like Cu, Al, Fe in this case). Therefore, a good knowledge of the sample and its nature is necessary.
Devices for particle size reduction First step Final step or Retsch SM 2000 Cutting Mill Retsch SR 300 Rotor Beater Mill Retsch ZM-1 Centrifugal Mill Size reduction by cutting and shearing forces Size reduction by hammering, impact and shear effects Size reduction by high speed impact and shear effects Low speed low heating Final particle size depending on the grid/sieve installed; coarse to mid size cutting (order of magnitude: mm) High speed some heating is developed during processing Feed: soft, medium-hard, tough, elastic, fibrous materials Final particle size depending on the grid/sieve installed; coarse to mid size cutting (order of magnitude: mm) High speed heating may be developed during processing, depending on the type of material Low to mid throughput Final particle size depending on the grid/sieve installed; mid to fine size cutting (order of magnitude: below mm) Feed: soft, medium-hard, brittle, fibrous materials Low throughput Feed: soft, medium-hard; low performances on tough, elastic and fibrous materials Mid to high throughput
Evaluation of the effect of particle size reduction devices Low stress 2 mm ZM-1 centrifugal mill 0.5 mm X3 SM 2000 cutting mill High stress Sample X3 SR 300 rotor beater mill 6 mm X3 ZM-1 centrifugal mill 0.5 mm ZM-1 centrifugal mill 0.5 mm Mid stress SR 300 rotor beater mill 2 mm A n a l y s e s