Proficiency Test SYKE 6/2013

Transkriptio

1 REPORTS OF FINNISH ENVIRONMENT INSTITUTE Proficiency Test SYKE 6/2013 Gross and net calorific in fuels Mirja Leivuori, Minna Rantanen, Katarina Björklöf, Keijo Tervonen, Sari Lanteri and Markku Ilmakunnas Finnish Environment Institute

2

3

4

5 CONTENT 3 PREFACE /ALKUSANAT 1 INTRODUCTION 6 2 ORGANIZING THE PROFICIENCY TEST Responsibilities Participants Samples and delivery Homogeneity studies 7 2. Feedback from the pro!ciency test Processing of the data Pretesting of the data Assigned values and uncertainties Standard deviation for pro!ciency assessment and z score 9 3 RESULTS AND CONCLUSIONS 3.1 Results 3.2 Analytical methods Gross and net calori!c value Measurement of carbon, hydrogen, nitrogen, sulphur, moisture, ash and volatile 13 matter 3.3. Analytical methods and status to the results Ash, moisture, volatile matter and elemental measurements Gross and net calori!c value Uncertainties of the results Estimation of emission factor 1 4. EVALUATION OF PERFORMANCE 1 SUMMARY 17 6 YHTEENVETO 17 7 REFERENCES 18

6 APPENDICES 4 Appendix 1 Participants in the pro!ciency test 20 Appendix 2 Preparation of the samples 21 Appendix 3 Testing of samples 22 Appendix 4 Feedback from the pro!ciency test 24 Appendix Evaluation of the assigned values and their uncertainties 2 Appendix 6 Terms in the result tables 26 Appendix 7 Results of each participant 27 Appendix 8 Results of participants and their uncertainties 37 Appendix 9 Summary of the z scores 47 Appendix.1 Analytical methods 48 Appendix.2 Signi!cant di$erences in the results reported using di$erent standard 4 methods Appendix.3 Results grouped according to the methods Appendix 11 Measurement uncertainties reported by the laboratories 64 DOCUMENTATION PAGE 71 KUVAILULEHTI 72 PRESENTATIONBLAD 73

7 ALKUSANAT Suomen ympäristökeskus (SYKE) on toiminut ympäristöalan kansallisena vertailulaboratoriona vuodesta 2001 lähtien. Toiminta perustuu ympäristöministeriön määräykseen, mikä on annettu ympäristönsuojelulain (86/2000) nojalla. Vertailulaboratorion tarjoamista palveluista yksi tärkeimmistä on pätevyyskokeiden ja muiden vertailumittausten järjestäminen. SYKEn laboratoriot on FINAS-akkreditointipalvelun akkreditoima testauslaboratorio T003 ja kalibrointilaboratorio K04 (SFS-EN ISO/IEC 1702) sekä vertailumittausten järjestäjä Pro est SYKE PT01 (SFS-EN ISO/IEC 17043, Tämä pätevyyskoe on toteutettu SYKEn vertailulaboratorion pätevyysalueella ja se antaa tietoa osallistujien pätevyyden lisäksi tulosten vertailukelpoisuudesta myös yleisemmällä tasolla. Pätevyyskokeen onnistumisen edellytys on järjestäjän ja osallistujien välinen luottamuksellinen yhteistyö. Parhaat kiitokset yhteistyöstä kaikille osallistujille! PREFACE Finnish Environment Institute (SYKE) is appointed National Reference in the environmental sector by the Ministry of the Environment according to section 24 of the Environment Protection Act (86/2000) since "e duties of the reference laboratory service include providing pro!ciency tests and other interlaboratory comparisons for analytical laboratories and other producers of environmental information. SYKE laboratories has been accredited by the Finnish Accreditation service as the testing laboratory T003 and the calibration laboratory K04 (EN ISO/IEC 1702) and as the pro!ciency testing provider Pro est SYKE PT01 (EN ISO/IEC 17043, "is pro!ciency test has been carried out under the scope of the SYKE reference laboratory and it provides information about performance of the participants as well as comparability of the results at a more general level. "e success of the pro!ciency test requires con!dential co-operation between the provider and participants. "ank you for your co-operation! Helsingissä 4 Helmikuuta 2014 / Helsinki 4 February 2014 Laboratorionjohtaja / Chief of

8 1 INTRODUCTION 6 Pro est SYKE carried out the pro!ciency test for the analysis of the gross and the net calori!c value as well as for content of ash, carbon, hydrogen, nitrogen, sulphur, analytical moisture content and volatile matter in fuels in September "e samples were prepared from peat (B1), wood pellet (B2) and coal (K1). Additionally, the participants were asked to estimate or calculate the emission factor for peat and coal samples. "e pro!ciency test was carried out in accordance with the international guidelines ISO/IEC [1], ISO 1328 [2] and IUPAC Technical report [3]. "e Pro est SYKE has been accredited by the Finnish Accreditation Service as a pro!ciency testing provider (PT01, ISO/IEC 17043, "is test is included in the scope of Pro est SYKE accreditation. 2 ORGANIZING THE PROFICIENCY TEST 2.1 Responsibilities 2.2 Participants Organizing laboratory: Pro est SYKE, Finnish Environment Institute (SYKE), Centre Hakuninmaantie 6, Helsinki tel , fax Subcontractors: "e peat, wood pellet and coal samples were homogenized and divided into subsamples at the laboratory of Water Protection Association of the Kokemäenjoki River in Tampere (Finland, accredited testing laboratory T064 by the Finnish Accreditation Service, "e samples were tested at the laboratory of Ramboll Finland Oy, Ramboll Analytics in Vantaa, Finland and is the accredited testing laboratory T039 by the Finnish Accreditation Service, ( "e responsibilities in organizing the pro!ciency test were as follows: Mirja Leivuori, coordinator Katarina Björklöf, substitute of coordinator Keijo Tervonen, technical assistance Sari Lanteri, technical assistance Markku Ilmakunnas, technical assistance and layout of the report. "e co-operation partner and the analytical expert was: Minna Rantanen, Ramboll Finland Oy, Ramboll Analytics (Vantaa) In this pro!ciency test (PT) in total 4 laboratories participated, from which 16 were from Finland and 29 from other countries (Appendix 1). "e sample testing laboratory has code 11 in the result tables.

9 2.3 Samples and delivery 7 "e peat sample B1 was provided by Labtium (previously Enas Oy) in Jyväskylä, (Finland) and the coal sample K1 was provided by Helsinki Energia (Finland). "e wood pellet sample was produced by Vapo Group (Finland). "e preparation of the samples is presented more detailed in Appendix 2. Sample B1 was a peat sample from the Finnish marshland. "e material was air dried and grounded by the mill with 00 µm sieve before homogenization and sample dividing. "e wood pellet sample B2 was barked so wood (spruce and pine) sawdust and molding shavings. "e material was!rst crushed with a cutting mill and then grounded by the mill with 00 µm sieve before homogenization and sample dividing. "e coal sample (K1) was prepared from a Polish du# coal. "e material was air dried and grounded to particle size < 212 µm before homogenization and sample dividing. "e samples were delivered on 3 September "ey were requested to be analyzed and reported before 26 September "e samples and the requested measurements were as follows: Sample Sample type Measurements B1 Peat Gross and net calorific value, C, S, H, N, moisture content of the analysis sample (M ad ), ash content, volatile matter (V db ) B2 Wood Pellet Gross and net calorific value, C, H, N, moisture content of the analysis sample (M ad ), ash content, volatile matter (V db ) K1 Coal Gross and net calorific value, C, S, H, N, moisture content of the analysis sample (M ad ), ash content, volatile matter (V db ) In the covering letter sent with the samples it was instructed that the moisture content of the analysis (M ad ) had to be measured as the!rst measurement a er storing samples closed for one day in the participant's laboratory. "e samples were instructed to be homogenated before measurements and to be stored in a dry place at room temperature. Additionally, the moisture content of the analysis was guided to be measured on every measuring day. "is was important as it eliminates the in$uence of humidity on the measurements. Also the participants were asked to report the relative humidity (%) of the measuring room as an average of the measuring dates. Additionally, the participants had the possibility to estimate or calculate the emission factor (as received, EF ar ) for peat and coal samples. For this estimation or calculation, the organizer of this pro!ciency test reported in the covering letter of the samples the total moisture contents as received (M ar ) for peat sample B % and for coal sample K1 8.7 %. 2.4 Homogeneity studies Homogeneity of the samples B1, B2 and K1 was tested by analyzing the gross calori!c value and ash content as replicate determinations from ten subsamples (Appendix 3). Moreover volatile matter and nitrogen was tested from six subsamples as replicate measurements, and additionally the content of carbon, hydrogen and sulphur from one to two subsamples were measured. According to the all homogeneity test results, all samples were considered homogenous. Particle size distribution was also tested from one sub sample of peat (B1) and coal (K1). For the peat sample material the requirement of particle size given in the international standards was

10 8 ful!lled, but not in the case of coal (Appendix 3). However, based on the result of this PT this seems not to have in$uenced the performance of the participants analysing the coal sample. 2. Feedback from the pro ciency test Appendix 4 contains the comments sent by the participants. "e comments were mainly relating to the data input protocols in the laboratories. "e provider gives some comments in Appendix 4 to the participants considering mainly the reporting of the results. It is recommended that di#erent post-analytical errors should be reported to the provider a er the preliminary results. "e provider again strongly recommended that the participants should be more careful when reporting the results to assure that the measurement uncertainty is reported correctly and the data reported is correct. 2.6 Processing of the data Pretesting of the data Before the statistical treatment, the data was tested according to the Kolmogorov-Smirnov normality test and outliers were rejected according to the Hampel test for calculation of the mean value (H in the results sheets). Also before the robust calculation some outliers were rejected in case that the results deviated from the robust mean more than 0 %, in case that the result was reported erroneously (e.g. wrong unit), large deviation between the parallel results were observed or anomalous values in the measured element value were used in the calculation. "e replicate results were tested using the Cochran-test (C in the result sheets). If the result was reported < DL (detection limit), it has not been included in calculation of the results (H in the results sheets) Assigned values and uncertainties "e robust mean was basically used as the assigned value for measurement of the samples B1, B2 and K1 (Appendix ). "e robust mean is not a metrological traceable assigned value. Because it was not possible to have a metrological traceable assigned value, the robust means of the results were the best available values to use for the assigned values. "e reliability of the assigned value was statistically tested according to the IUPAC Technical report [3]. In calculation of the robust mean outliers are not normally rejected, but they are iterated before the!nal calculation of the robust mean. However, in this pro!ciency test some extreme results (at most 1-3 results per analyte) had to be rejected because of rather strict requirements for reproducibility given in the standards for analysis described in the covering letter of the samples. In estimation of the assigned values of gross and net calori!c value, the reasons for extreme values were large deviation between parallel results or anomalous value in the measured element value used in the calculation. Also the mean value (a er using the Hampel outlier test) and the median value of the data were calculated, which were quite near with the assigned values (Table 1). "e results of homogeneity testing of the samples were used as background information in estimation of the assigned values. In some cases, the assigned values were compared with the results obtained in the kernel density plots [3]. In cases, when the number of results was low (6), the mean value of participants was used as the assigned value. When using the robust mean of the participant results as the assigned value, the uncertainties of the assigned values for calori!c values varied from 0.1 % to 0. %. For the other measurements the uncertainty varied from 0. % to 7.9 % (Appendix ). "e participants also calculated emission factors (EF) according to the given total moisture contents as received (M ar ) for the peat and coal samples and the results were evaluated as well. In the pro!ciency test very few emission factors were reported (6-13). Additionally, it was noticed that at least laboratory 42 (B1, K1) did not calculate the emission factor as requested. In additional,

11 9 the calculation of emission factor in the wood pellet (B2) sample was not done as it is CO 2 neutral fuel. % er reporting of the preliminary results in September 2013 only the assigned value of hydrogen for the peat sample (B1) had been changed from.47 to.3. "e change based on the using mean value as the assigned value a er more detail data treatment. "is assigned value change had no in$uence to the total number of satisfactory hydrogen results in the peat sample (B1) Standard deviation for pro ciency assessment and z score For the total standard deviation for this pro!ciency assessment used in the calculation of the z score the target value for reproducibility recommended in the international standards for measurement of calori!c values and other determinants was used [4,, 6, 7, 8, 9,, 11]. "e reproducibility recommended in the standards was mainly ful!lled for the gross calori!c values (± 300 J g -1 ). For the net calori!c value, the reproducibility was ± 37 J g -1 for sample B1, ± 340 J g -1 for sample B2 and ± 346 Jg -1 for sample K1. Higher reproducibility for the net calori!c values were allowed due to the variability of the results and the missing of clear reproducibility information for the net caloric value in the standard methods [e.g., 12, 13]. "ere are more uncertainty sources in calculation of net calori!c value than in calculation of gross calori!c value. For some other measured parameters (i.e. C, H, N, S and ash) the total standard deviation for pro!ciency assessment had to be increased from the reproducibility presented in the standard methods. "e performance evaluation was carried out by using z scores (Appendix 6). "e performance evaluation of participants using calculated z scores are presented in Appendix 7.!e reliability of the assigned value was tested according to the criterion: u/s p & 0.3, where u is the standard uncertainty of the assigned value (the expanded uncertainty of the assigned value (U) divided by 2) and s p the standard deviation for pro!ciency assessment (total standard deviation divided by 2). In the testing of the reliability of the assigned value the criterion was not met in every case, which is indicated by the high uncertainty of the assigned values in the following cases: Sample B1 Measurement H In the above case the standard deviation of the reported results was high and the number of results for the calculation of the assigned value was too low for the reliable performance evaluation (Table 1).!e reliability of the target value for the total deviation and the reliability of the corresponding z score were estimated by comparing the deviation for pro!ciency assessment (s p ) with the robust standard deviation of the reported results (s rob ).!e criterion s rob < 1.2*s p was ful lled fairly and the evaluation of performance is reliable for this pro ciency test. "e results of analysis moisture (M ad ) have not been evaluated, but the assigned values are presented. "e variability in the results was almost in the same range than in the previous test in PT 6/2012 [14]. If the participant likes to estimate the own results by the z scoring, it is possible by using %

12 as the total standard deviation for pro!ciency assessment. "e formula is available from the document Guide for participating laboratories in SYKE pro!ciency testing schemes in the Pro est website est/en. "e results of nitrogen in the peat and the wood pellet sample have not been evaluated, but the assigned values are presented based on mean value of the reported results. "e number of nitrogen results was too low for the performance evaluation in peat sample. In the wood pellet sample, the nitrogen concentration was very low and before reliable performance evaluation, more detail information of the detection limit and its evaluation technique from the participants is needed. "is will be included in future pro!ciency tests of wood pellet. % er reporting of the preliminary results in September 2013 no corrections had been done to the standard deviations for the pro!ciency assessment. 3 RESULTS AND CONCLUSIONS 3.1 Results "e summary of the results is presented in Table1. Explanations to terms used in the result tables are presented in Appendix 6. "e results and the performance of each laboratory are presented in Appendix 7. "e results of participants and their uncertainties are presented graphically in Appendix 8. "e summary of z scores is shown in Appendix 9. "e results grouped according to di#erent methods used are presented in Appendix.3. "e measurement uncertainties reported by the laboratories grouped according to the evaluation procedure is reported in Appendix 11. "e robust standard deviation of results was lower than 2 % for 64 % of the results and it was lower than 6 % for 82 % of the results (Table 1). For moisture (M ad ) in the sample K1 the robust standard deviation was 6.1 %. "e highest robust standard deviation was lower than 1 % (B2, ash). "e standard deviations of the results in this pro!ciency test were nearly in the same range than in the previous respective PT SYKE 6/2012 [14], where the deviations varied mainly from 0.3 % to more than over 18 %. In this pro!ciency test the participants were requested to report the replicate results for all measurements. "e results of the replicate determinations based on the ANOVA statistical handling are presented in Table 2. "e international standards or technical speci!cations relating to measurements in fuels recommend the targets for the repeatability.

13 11 Table 1. Summary of the result in the pro!ciency test 6/2013. Ass. Val.- the assigned value, Mean- the mean value, Mean rob- robust mean, Md - the median value, SD %-the standard deviation as percent, SD rob - the robust standard deviation, SD rob % - the robust standard deviation as percents, Num of Labs - the number of participants, 2*Targ. SD% - the total standard deviation for pro!ciency assessment at 9 % con!dence level (2*s p ), Accepted z-val% - the satisfactory z scores: the results (%), where z 2. In particular, in measurement of the calori!c values, the requirement for the repeatability is ± 120 J/g. In this pro!ciency test the requirements for the repeatability in measurement of the gross calori!c value were 0.7 % for the sample B1, 0.60 % for the sample B2 and 0.40 % for the sample K1 and in measurement of the net calori!c value 0.62 %, 0.64 % and 0.40 %, respectively. In each case, the obtained repeatability in measurement of the gross calori!c value and the net calori!c value was lower than the repeatability requirement (Table 2, the column s w %). "e repeatability was mainly acceptable for carbon C in the elemental measurements and for volatile matter (Table 2, the column s w %). "e estimation of the robustness of the methods could be done by the ratio s b /s w. "e ratio s b /s w should be 2-3for robust methods. However, in many cases the robustness exceeded the value 3 (Table 2). For the gross calori!c value, the ratio s b /s w, was 3.3 % (the sample B1), 3.8 % (the sample B2) and 1.8 % (the sample K1), for the net calori!c values 4.9 %,.3 % and 2.4 %, respectively. For the net calori!c values for the peat and coal samples the ratio s b /s w was in the same range than in the previous PT SYKE 6/2012 [14].

14 12 Table 2. Summary of repeatability on the basis of duplicate determinations (ANOVA statistics). 3.2 Analytical methods Gross and net calori c value "e analytical methods based on di#erent standard methods were used for the measurements in the pro!ciency test. "e used analytical methods of the participants are shown in more detail in Appendix.1. Mostly (89 %), standard methods were used for measurement of calori!c value (EN [4], ISO 1928 [], DIN 1900 [6], ASTM D 86-13[13]). A few laboratories used some national or other standards (e.g. lab 18, 36, 38, 43). "e participants used mostly g of sample for measurement of the calori!c value. "e measurements of calori!c value were mainly done by IKA, PARR or LECO equipment (Appendix.1). In the calculation of gross calori!c value (q-v,gr,d), various correction methods were used. Fuse wire, ignition, acid, moisture, nitrogen and sulphur corrections were most commonly used in several di#erent combinations (Appendix.1). For the calculation of net calori!c value (q-p,net,d) di#erent combinations of correction factors were used as well (Appendix.1). Mainly, the calculated/!xed hydrogen content was used for corrections.

15 Measurement of carbon, hydrogen, nitrogen, sulphur, moisture, ash and volatile matter In the pro!ciency test several standard methods or technical speci!cations were used mainly for measurement of di#erent parameters as follows: Parameter Method C, H and N ISO 2941 [7], ASTM D 373 [8], EN 14 [13] S EN 1289 [9], ASTM D 4239 [1], ISO 334 [16] Analytical moisture content EN [17], ISO 89 [18], DIN 1718 [19], ASTM D 142 [20] Ash content ISO 1171 [], ASTM D 782 [20], EN 1477 [21], DIN 1719 [11] Volatile matter EN 1148 [22], ISO 62 [23] However, in some cases other international standards or national standards were used. For example, in moisture measurements for general analysis standard ISO (lab, 17, 21, 2, 28) was also used. "is was pointed out by participant and this standard will be included in the method list in next pro!ciency test. Moisture content was determined mainly in air gravimetrically by heating at the temperature -7 C. Moisture content was measured also using TGA at the temperatures 7-7 C. N 2 atmosphere was mainly used for determining moisture content for coal samples, but also in a few cases for wood and peat samples (Appendix.1). Ash content was determined mainly gravimetrically by heating at the temperature 0 C (Samples B1, B2) or 81 C (Sample K1). Ash content was measured also using TGA for samples at the temperatures 0 C, 81 C or 70 C (Appendix.1). 3.3 Analytical methods and status to the results "e di#erences between the average concentrations of elements measured by the di#erent standard methods were tested using the t-test. "e results of the t-test are shown in Appendix.2. In Appendix.3 is presented the results of participated laboratories grouped based on used di#erent standard methods Ash, moisture, volatile matter and elemental measurements In measurement of the ash content di#erent methods were used. Statistically signi!cant di#erence was found between the used standard methods DIN 1719 and ASTM D 782. Noticeably is, however, the low number of the results (Appendix.2). "e analysis moisture (M ad ) was measured using di#erent standard methods (Appendices.1 and.3). Statistically signi!cant di#erences were found between the standards methods ISO 89 and e. g. ISO 11722, ISO 331, ISO 068 and di#erent internal methods. On reason behind the di#erence could be that N 2 atmosphere was mainly used for determining moisture content for the coal samples and air atmosphere for the peat and wood pellet samples. "e analysis of volatile matter (V db ) was measured using di#erent standard methods and no statistically di#erence between results was noticed. In measurement of carbon, nitrogen, hydrogen and sulphur di#erent standard methods were used (Appendices.1 and.3). Statistically signi!cant di#erences were found for carbon measurements between the standards methods ASTM D 373 and the other methods. "e other methods used were e. g. ASTM D 373 & ASTM D 291, ISO 609, ISO/TS and di#erent internal methods. For measurements of nitrogen, hydrogen and sulphur no statistically signi!cant di#erence between the results was noticed.

16 3.3.2 Gross and net calori c value 14 For the results of the gross calori!c values for the samples B1, B2 and K1 (Appendices.1 and.3) no clear di#erence between the net and gross calori!c values obtained using the di#erent standard methods and no statistically signi!cant di#erence between the results were found. "ere most important factors, which have to be taken into account measurement of calori!c value are: Sample should be mixed well before analyses are carried out. Analytical moisture and calori!c value should be measured at a same time. 3.4 Uncertainties of the results Several approaches were used for estimating of measurement uncertainty (Appendix 11). "e most used approach based on using data obtained in method validation (Meth 4), using the modelling approach (GUM, Meth ) or using IQO data from the synthetic sample with/without the results from the pro!ciency tests (Meth 3 and 2). From 3 to 23 laboratories reported the expanded measurement uncertainties with their results (Table 3, Appendix 11). "e estimated uncertainties varied greatly for all the tested measurements. Especially, very low uncertainties can be considered as questionable. It is evident, that some uncertainties have been wrongly reported for the calori!c values, not as relative values as the provider of this pro!ciency test had requested (Table 3, Appendix 11). In many other cases, the reported measurement uncertainties did not meet the requirements presented in the standard methods for the repeatability of the method [, 6]. Generally, the approach for estimating measurement uncertainty has not made a de!nite impact on the uncertainty estimates. Despite this, harmonization in the estimating of uncertainties should be continued. One possibility to harmonize measurement uncertainty is to use a new so ware tool Mukit (measurement uncertainty kit), which based on the Nordtest method [24]. "is free so ware is available in the webpage of the calibration laboratory (ENVICAL) of SYKE: www. syke.!/envical/en. Table 3. "e reported range of measurement uncertainties in the PT 6/2013. Measurement Uncertainty B1,% Uncertainty B2, % Uncertainty K1, % Ash C H N q-p,net,d q-v,gr,d S V db

17 3. Estimation of emission factor 1 Additionally, the laboratories were asked to estimate the emission factors for the peat and coal samples distributed in the pro!ciency test by taking into account their own net calori!c values and the total moisture values as received, which was informed in the covering letter of the samples. "e calculation of emission factor in the wood pellet (B2) sample was not done as it is a CO 2 neutral fuel. Very few laboratories (6-13) reported the emission factor in this pro!ciency test (Table 1, Appendix 7). Additionally, at least laboratory 42 (B1, K1) did not calculate the emission factor as requested. Due the possible calculation errors, the performance evaluation of emission factor in the coal sample (K1) is only approximate. "e number of acceptable emission factor results for the peat sample is very low, thus the performance evaluation is not available. "e participants were asked to calculate EF-values using the equation presented in the EC directive 2007/89/EC [2]. Many of the participants also informed that the calculation of EF-value was based on the EC directive 2007/89/EC (Appendix.1). However, the results show that there is some dissimilarity in the calculation of EF values. "is EU directive have been replaced by a new EU directive, EU601/2012 [26], which is recommended to take into use in future calculations. In Finland the Energy Market Authority has made a guideline for the calculation of emission factor for fossile fuels ( pdf). "is is presented in the Appendix.1. 4 EVALUATION OF PERFORMANCE "e total number of laboratories participating in this PT was 4. "e evaluation of the participants was based on z scores, which were calculated using the assigned and the estimated target values for the total deviation (Appendix 6).!e z scores were interpreted as follows: Criteria Performance z 2 Satisfactory 2 < z < 3 Questionable z 3 Unsatisfactory In total, 87 % from the results were satisfactory when deviations of 1 30 % from the assigned values were accepted. About 70 % of the participants used the accredited methods and 8 % of their results were satisfactory. Pro est SYKE arranged a similar pro!ciency test in 2012 and then 93 % of the results were satisfactory [14]. "e satisfactory results varied between 84 % and 92 % for the tested sample types (Table 4). "e criteria for performance had been mainly set according to the target value for reproducibility recommended in international standards or technical speci!cations for measurement of the calori!c values and other determinants. "e reproducibility required in the standards was ful!lled for the gross calori!c values. For the net calori!c value was used increased reproducibility from the value for the gross caloric value. "ere was no criterion for reproducibility for the net calori!c value in standards methods.

18 16 Table 4. Summary of the performance evaluation in the pro!ciency test 6/2013. Sample Satisfactory results (%) Accepted deviation from the assigned value (%) Remarks Peat, B High uncertainty in the assigned value for H. The assigned value of H had been changed from.47 to.3 after the preliminary results. This change had no influence on the total number of satisfactory hydrogen results in the peat sample (B1). For N the number of reported results was low and no performance evaluation was done. For EF the number of reported results was low and no performance evaluation was done. Wood pellet, B The concentration of N was low and no performance evaluation was done. Coal, K Weakened performance evaluation for EF due to the calculation errors of two laboratories (results had not been reported based on the total moisture as received). Peat In the previous pro!ciency test PT 6/2012 totally satisfactory results of the peat sample (B1) were 94 % [14], thus the performance in this PT was in the same range (92 %, Table 4). In the measurement of hydrogen and net calori!c value, 0 % of the results were satisfactory (Table 1). In the measurement of gross calori!c values, 88 % of results were satisfactory when accepting the deviations 1.4 % from the assigned values. In this pro!ciency test the number of satisfactory results of the gross values for the peat sample was in the same range and net calori!c higher than in the previous PT 6/2012 (86 % and 82 % respectively) [14]. In this PT the number of acceptable nitrogen and emission factor results for the peat sample was too low, thus the performance evaluation was not given done. Also the results of analysis moisture (M ad ) have not been evaluated, but the assigned values are presented (Table 1). Wood pellet In the previous pro!ciency test PT 6/2012 totally satisfactory results of the wood pellet sample (B2) were 88 % [14], thus the performance in this PT was in the same range (84 %, Table 4). "e satisfactory results varied between 72 % (gross calori!c value) and 89 % (ash) for the wood pellet sample (Table 1). "e number of nitrogen result was too low for the performance evaluation in peat sample (B2, Table 1). Also there was variability between the results of participants. Before reliable performance evaluation for nitrogen in wood pellet sample could be made, more detail information of the detection limits and their evaluation technique would be needed from the participants. In the measurement of gross and net calori!c values, 72 % and 79 %, respectively, were satisfactory when accepting deviations of 1.4 % and 1.8 % from the assigned values (Table 1). "e number of satisfactory results of the gross and net calori!c values for wood pellet were lower than in the previous pro!ciency test PT 6/2012 (88 % and 80 % respectively) [14]. In this PT, the estimation of EF was not done as it is a CO 2 neutral fuel. Also the results of analysis moisture (M ad ) have not been evaluated, but the assigned values are presented (Table 1).

19 Coal 17 SUMMARY In the previous pro!ciency test PT 6/2012 totally satisfactory results of the coal sample (K1) were 97 % [14], thus the performance in this PT was lower (87 %, Table 4). In the measurement of gross and net calori!c values 8 % and 87 % of results were satisfactory, when accepting the deviations of 1 and 1.2 % from the assigned values (Table 1). In this pro!ciency test the number of satisfactory result of the gross and net calori!c values were lower than in the previous PT 6/2012 (96 % and 0 % respectively) [14]. "e reliability of the performance evaluation of EF for the coal sample (K1) was weak due to the errors in the calculation of the emission factor (EF) and the performance evaluation is only approximate. "e results of analysis moisture (M ad ) have not been evaluated, but the assigned values are presented (Table 1). Pro est SYKE carried out the pro!ciency test for the analysis of the gross and the net calori!c value as well as for content of ash, carbon, hydrogen, nitrogen, sulphur, analytical moisture content and volatile matter in fuels in September One peat sample (B1), one wood pellet (B2) and one coal (K1) sample were delivered to the laboratories for the analysis. In total, 4 laboratories participated in the pro!ciency test. Additionally, the participants were asked to estimateor calculate the emission factor for peat and coal samples. "e robust means or mean (n<6) of the results reported by the participants were used as the assigned values for measurements. "e uncertainties of the calculated assigned values were less than 0.6 % for calori!c values and at maximum 8 % for the other measurements. "e evaluation of performance was based on the z scores, which was calculated using the standard deviation for pro!ciency assessment at 9 % con!dence level. "e evaluation of performance was not done for the measurement of Mad in the all samples, N in the peat and wood pellet samples and EF in the peat and wood pellet samples. In total, 87 % of the participating laboratories reported the satisfactory results when the deviations of 1 30 % from the assigned values were accepted. About 70 % of the participants used the accredited methods and 8 % of their results were satisfactory. In measurement of the gross calori!c value from the peat sample 88 %, from the wood pellet sample 72 % and from the coal sample 8 % of the results were satisfactory. In measurement of the net calori!c value from the peat sample 0 %, from the wood pellet 79 % and from the coal sample 87 % of the results were satisfactory. In general the results were in the same range as in the previous Pro est SYKE test in 2012 [14], but the performance was somewhat lower for wood pellet and coal samples in the present PT. "e reliability of the performance evaluation of EF for the coal sample (K1) was weak due to the errors in the calculation of the emission factor (EF). 6 YHTEENVETO Pro est SYKE järjesti syyskuussa 2013 pätevyyskokeen kalorimetrisen ja tehollisen lämpöarvon sekä tuhkan, vedyn, typen, rikin, kosteuden ja haihtuvien yhdisteiden määrittämiseksi turpeesta, puupelletistä ja kivihiilestä. Lisäksi osallistujilla oli mahdollisuus laskea päästökerroin turpeelle ja kivihiilelle.. Pätevyyskokeeseen osallistui yhteensä 4 laboratoriota. Laboratorioiden pätevyyden arviointi tehtiin z-arvon avulla ja sen laskemisessa käytetyn kokonaishajonnan tavoitearvot olivat määrityksestä riippuen välillä 1 30 %. Mittaussuureen vertailuarvona käytettiin osallistujien ilmoittamien tulosten robustia keskiarvoa tai keskiarvoa, jos tuloksia oli vähän (n<6). Tavoitearvon epävarmuus oli lämpöarvon määrityksissä alhaisempi kuin 0,60 % ja muiden määritysten osalta korkeintaan 8 %. Tulosten arviointia ei tehty testinäytteiden kosteuspitoisuuden määritykselle, typen määritykselle turpeesta ja puupelletistä eikä päästökertoimen laskennalle turpeesta. Arviointi on jonkin verran

20 18 epävarma hiilen päästökertoimelle, koska kaikki laboratoriot eivät olleet laskeneet arvoa tulokosteutta kohti. Koko tulosaineistossa hyväksyttäviä tuloksia oli 87 %, kun vertailuarvosta sallittiin 1 30 % poikkeama. Noin 70 % osallistujista käytti akkreditoituja määritysmenetelmiä ja näistä tuloksista oli hyväksyttäviä 8 %. Kalorimetrisen lämpöarvon tuloksista oli hyväksyttäviä 88 % (turve), 72 % (puupelletti) ja 8 % (kivihiili). Tehollisen lämpöarvon tuloksille vastaavat hyväksyttävien tulosten osuudet olivat 0 % (turve), 79 % (puupelletti) ja 87 % (kivihiili). 7 REFERENCES 1. ISO/IEC 17043, 20. Conformity assessment General requirements for pro!ciency testing. 2. ISO 1328, 200. Statistical methods for use in pro!ciency testing by interlaboratory comparisons. 3. "ompson, M., Ellison, S.L. R., Wood, R., "e International Harmonized Protocol for the Pro!ciency Testing of Analytical Chemistry laboratories (IUPAC Technical report). Pure Appl. Chem. 78: ( 4. EN 14918, 20. Solid Biofuels. Method for the determination of calori!c value.. ISO 1928, Solid mineral fuels- Determination of gross calori!c value by a bomb calorimetric method, and calculation of net calori!c value. 6. DIN 1900, Testing of solid and liquid fuels(- Determination of gross calori!c value by the bomb calorimeter and calculation of net calori!c value. 7. ISO 2941, 20. Solid mineral fuels - Determination of total carbon, hydrogen and nitrogen content - Instrumental methods. 8. ASTM D 373, Standard Test Methods for Instrumental Determination of Carbon, Hydrogen, and Nitrogen in Samples of Coal and Coke. 9. EN 1289, 2011 Solid biofuels - Determination of total content of sulphur and chlorine.. ISO 1171, 20. Solid mineral fuels. Determination of ash. 11. DIN Determination of ash in solid mineral fuels. 12. ASTM D 86-13, Standard Test Method for Gross Calori!c Value of Coal and Coke. 13. EN 14, Solid biofuels. Determination of total content of carbon, hydrogen and nitrogen. Instrumental methods. 14. Leivuori, M., Rantanen, M., Björklöf, K., Tervonen, K., Lanteri, S. and Ilmakunnas, M SYKE Pro!ciency Test 6/2012. Gross and net calori!c value in fuels. Reports of Finnish Environment Institute 4/2013. (

21 19 1. ASTM D 4239, Standard Test Methods for Sulphur in the Analysis Sample of Coal and Coke using High - Temperature Combustion and Infrared Absorption. 16. ISO 334, Solid mineral fuels - Determination of total sulfur - Eschka method. 17. EN , 20. Solid biofuels. Methods for the determination of moisture content. Oven dry method. Part 3: Moisture in general analysis sample. 18. ISO 89, Hard coal - Determination of total moisture. 19. DIN Determining the moisture content of solid fuels. 20. ASTM D Standard Test Methods for Proximate Analysis of Coal and Coke by Macro ermogravimetric Analysis. 21. EN 1477, 20. Solid biofuels. Determination of ash content. 22. EN 1148, 20. Biofuels, Solid fuels, Biomass, Fuels, Chemical analysis and testing, Volatile matter determination, Gravimetric analysis. 23. ISO 62, 20. Hard coal and coke - Determination of volatile matter 24. Näykki, T., Virtanen, A. and Leito, I So"ware support for the Nordtest method of measurement uncertainty evaluation. Accred. Qual. Assur. 17: ( /89/EC ( ). Commission Decision of 18# July 2007 establishing guidelines for the monitoring and reporting of greenhouse gas emissions pursuant to Directive 2003/87/EC of the European Parliament and of the Council. 26. Commission Regulation (EU) No#601/2012 of 21#June 2012 on the monitoring and re p orting of greenhouse gas emissions pursuant to Directive 2003/87/EC of the European Parliament and of the Council Text with EEA relevance (

22 APPENDIX 1 PARTICIPANTS Belgium Bulgaria Croatia Denmark Estonia France Finland Italy Latvia Lithuania Portugal Republic of Ireland Serbia South Korea Spain Sweden 20 Inspectorate Ghent Belgium, Ghent AES-3C Maritza East I Eood testing laboratory, Galabovo Chemical laboratory, ContourGlobal operations, Mednikarovo Coal at Troyanovo 1 Mine, Radnevo Coal at Troyanovo 3 Mine, Radnevo Coal at Troyanovo North Mine, Radnevo Complex Testing (KIL) at Tec Varna Ead, Varna Region Solid Fuel Test laboratory at Brikel EAD, Galabovo Testing laboratory Waters & Fuels in AMEES Ltd, Radnevo TPP Bobov dol Coal laboratory, Kyustendil Cetralni kemijsko- tehnoloski laboratorij, Zagreb FORCE Technology, Brøndby Eesti Energia Ölitööstus AS Chemical laboratory, Auvere küla Inspectorate Estonia AS, Viimsi vald Eurofins Ascal Hydrologie, Forbach Cedex SOCOR, Dechy Ahma Ympäristö Oy, Oulu Ekokem Oy Ab, Riihimäki Finnsementti Oy, Parainen Helsingin Energia, Voimalaitoskemia, Helsinki KCL Kymen Laboratorio Oy, Kuusankoski Kymenlaakson ammattikorkeakoulu, Kotka Labtium Oy, Jyväskylä Labtium Oy, Kuopio Metla, Kannus Oulun yliopisto, Prosessi- ja ympäristötekniikan osasto, Oulu PVO-Lämpövoima, Kristiinankaupunki Ramboll Finland Oy, Ramboll Analytics, Vantaa Ruukki Metals Oy, Raahe Teknologiakeskus KETEK Oy, Kokkola Vaskiluodon Voima Oy, Seinäjoki Vaskiluodon Voima Oy, Vaasa C.T.G s.p.a, Bergamo Inspectorate Latvia Ltd, Venetspils AB ''Šiauliu energija'' chemical laboratory, Šiauliai Axis Industries Biofuel research laboratory, Vilnius Pegop-Energoa Eléctrica, Pego Edenderry Power Operations Ltd, Edenberry Kontrola kvaliteta uglja, Laboratorija TAMNAVA, Stepojevac Lazarevac of testing hard mineral fuels and mineral raw material, Belgrade Intertek KIMSCO Ulsan Testing Center, Ulsan Applus Norcontrol Slu, Madrid LECEM-EP, Madrid Hjortens Laboratorium AB, Östersund SP Technical Research Institute of Sweden, Borås

23 PREPARATION OF THE SAMPLES 21 APPENDIX 2 Sample B1, peat Sample B1 was prepared from peat taken from a Finnish marshland. The peat was air-dried (3 ºC) and grounded in a mill with a 00 µm sieve at the laboratory of Enas Oy. The dried and sieved sample was mixed by a mechanized sample mixer and distributed to subsamples of 0 g using a rotary sample divider equipped with a vibratory sample feeder at the laboratory of Water Protection Association of the Kokemäenjoki River. The particle size distribution of peat was measured by the laboratory of Enas using laser diffraction (Malvern). Sample B2, wood pellets Sample B2 was prepared from barked softwood (spruce and pine) sawdust and molding shavings. The wood pellets were first crushed with a cutting mill and then grounded by the mill with 00 µm sieve at the laboratory of Enas Oy. The sieved sample was mixed by a mechanized sample mixer and distributed to subsamples of 0 g using a rotary sample divider equipped with a vibratory sample feeder at the laboratory of Water Protection Association of the Kokemäenjoki River. Sample K1, steam coal fuel Sample K1 was a Polish duff coal. The coal was dried at room temperature and grounded to particle size < 212 µm at the Helsinki Energy. The dried and sieved sample was mixed by a mechanized sample mixer and distributed into subsamples of 0 g using a rotary sample divider equipped with a vibratory sample feeder at the laboratory the laboratory of Water Protection Association of the Kokemäenjoki River. The particle size distribution of coal was measured by the Helsinki Energia, Power Plant Chemistry using laser diffraction (Malvern).

24 APPENDIX 3/1 TESTING OF SAMPLES 22 Homogeneity Homogeneity was tested from duplicate measurements of calorific value and ash content in ten samples, which were homogenised before sampling (Table 1). Additionally, volatile compounds and nitrogen from six samples was tested. The analytical variation s an and the sampling variation s sam was calculated using one-way variance analysis. For this proficiency test, the analytical results were statistically handled according to the IUPAC guidelines for the treatment of homogeneity testing data and the total standard deviation for proficiency assessment [4]. Table 1. Results from the homogeneity testing of the peat B1, pellet B2 and coal K1 samples. Measurements Mean value s h % s p % s h s a s a /s h Is s a /s p< 0.? s bb s bb 2 c Is s bb 2 <c? Peat (B1) Gross calorific value, J/g yes yes Ash, w-% yes yes Pellet (B2) Gross calorific value, J/g yes yes Ash, w-% yes yes Coal (K1) Gross calorific value, J/g yes yes Ash, w-% yes yes where, s p % s h %, s h s a s bb standard deviation for proficiency assessment as percent, (total standard deviation divided by 2) standard deviation for testing of homogeneity analytical deviation, standard deviation of results in a sub sample between-sample deviation, standard deviation of results between sub samples c = F1 s all 2 + F2 s a 2 where: s all 2 = (0.3 s p ) 2 F1 = 1.88 when the number of sub samples is, F2 = 1.01 when the number of sub samples is Conclusion: In each case, the criteria were fulfilled. Also the results of the volatile matter and nitrogen in the all tested fuel samples support the homogeneity of samples. Thus, all the samples could be regarded as homogenous.

25 23 APPENDIX 3/2 Particle size To test the particle size of peat (B1) and coal (K1) samples tested using laser diffraction (Malvern). Figure 1 is showing the distribution of particle size for the samples B1 and K1. For peat sample B1 the mean size of particles was 30. µm and ca. 99 % of the particles were smaller than 0 µm. For coal sample K1 the mean size of particles was 40.6 µm and 98.4 % of the particles were smaller than 212 µm. For the peat sample material the requirement of particle size given in the international standards was fulfilled, but not in the case of coal [, 6]. However, based on the result of this PT this seems not to be influenced to the performance of the participants. a) The particle size distribution of peat B1. b) The particle size distribution of coal K1. Figure 1. The particle size distribution of the fuel samples.

26 APPENDIX 4 24 FEEDBACK FROM THE PROFICIENCY TEST COMMENTS SENT BY THE PARTICIPANTS Lab Comment to the samples / PT Action/Proftest 17 The laboratory reported ISO standard as one possible method for moisture measurement in coal samples. The ISO will be include as one possible method for moisture measurement in coal samples in future PTs. Lab Comment to the results Action/Proftest 19 The laboratory informed that their ash result for coal sample (K1) wasn t right in the preliminary result list. 26 The laboratory asked for more detailed information of the Cochran and Hampel test informed in the report of the result of laboratory in the preliminary results. 40 The laboratory reported the result of volatile compounds (V db ) for wood pellet sample (B2) erroneously. The corrected results were: and The laboratory reported the results of the calorific values for samples B2 and K1 in the wrong unit. The data of the laboratory had to re-written to the Excel sheet by the provider. Unfortunately, there was input error in the replicate result for ash in the sample K1. The data was corrected in the final report. The correction did not affect to the assign value of ash in K1. The result of the laboratory was satisfactory. The provider clarified the meaning of column outlier test OK (Cochran or Hampel test) in the report of the result of the laboratory. The result of volatile compounds was outliers in the statistical treatment, and so it has not affected the performance evaluation. If the result had been reported rightly, it would have been satisfactory. The participant can re-calculate z scores according to the guide for participating laboratories in Proftest proficiency testing schemes ( proftest/ en > Running proficiency test). The results of calorific values were outliers in the statistical treatment, and so they have not affected the performance evaluation. If the results had been reported rightly, it would have been satisfactory. The participant can re-calculate z scores according to the guide for participating laboratories in Proftest proficiency testing schemes ( proftest/ en > Running proficiency test). The participant asked for zeta values in the preliminary report. The participant did not report expanded measurement uncertainties for all their measurement, thus the results were missing in the zeta value tables. COMMENTS TO THE PARTICIPANTS 3, 9, 20, 23, 43 Comments on results The laboratories are accredited but they did not report their measurement uncertainties with the reported results. Laboratories should have determined the measurement uncertainties for all accredited methods. 22, 24 The laboratories reported more replicate results than required. The two first results were taken to the data evaluation. 42 The laboratory reported the measuring uncertainties in the wrong unit. The provider strongly recommended that the participants are more carefully to report the uncertainty correctly. The laboratory did not calculate the emission factor as requested for the samples B1 and K1.