- Email: [email protected]
Evaluation of data on calorific content values determined during ASTM round robin testing of RDF-3
Resources
and Conservation,
8(1983)
233-252
233
Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands
EVALUATION OF DATA ON CALORIFIC CONTENT VALUES DETERMINED DURING ASTM ROUND ROBIN TESTING OF RDF-3
EUGENE National
S. DOMALSKI Bureau
and STANLEY
of Standards,
Washington,
ABRAMOWITZ D.C.
20234
(U.S.A.)
(Received May 5, 1982; accepted in revised form September
7, 1982)
ABSTRACT The potential application of refuse-derived fuel, in particular RDF-3, as an alternative or supplemental fuel is dependent upon its acceptance as an article of commerce. ASTM Committee E-38 on Resource Recovery, and its Subcommittee on Energy, E-38.01, have been actively engaged in the development of concensus standards for this purpose especially since April 1974; standard procedures for the characterization of RDF-3 are being developed. These procedures are based on those ASTM methods used in coal analysis. The procedures developed will insure a meaningful purchase--sales relationship between the buyer and seller. A variety of chemical and physical test procedures were studied by as many as 12 laboratories. A total of 28 editorial draft standards have been prepared. Of this number, 13 have appeared as accepted methods in the Annual Books of ASTM Standards while the remainder are either under evaluation or under some phase of ASTM balloting. The National Bureau of Standards, in cooperation with ASTM subcommittee E38.01, has undertaken a technical review of a selected group of chemical properties of RDF-3. The property of principal interest is the calorific content and is expressed as the higher heating value. In order to properly characterize this property, critical evaluation of methods to determine total moisture, residual moisture, and ash is also necessary. Intra-laboratory (within-laboratory) and inter-laboratory (between-laboratory) variations in determinations for moisture, ash, and higher heating value are discussed. A comparison of these results with those for round robin data for coal is also made. The result is identification of the levels of precision for intra-laboratory and inter-laboratory agreement. INTRODUCTION
The National Bureau of Standards (NBS) has been a participant in the development of standard test methods for one form of refuse-derived-fuel (RDF-3)* with ASTM (American Society for Testing and Materials) Committee E-38 (Resource Recovery) and its Subcommittee E-38.01 (Energy) since the beginning of the testing program in the summer of 1977. Since that time, two preliminary and three definitive series of round robin RDF-3 samples * RDF-3 is defined as a shredded fuel derived from municipal solid waste (MSW) which has been processed for the removal of metal, glass, and other entrained inorganic materials. Generally, this material has a particle size such that 95 wt% passes through a 5-cm (2-inch) square mesh screen.
0166-3097/83/0000-0000/$03.00
0 1983
Elsevier Scientific Publishing Company
234
have been distributed and examined by various testing laboratories. At present, some 15 test protocols are in the form of editorial draft standards. These standards are undergoing evaluation and are under vote for acceptance by ASTM procedures as standard methods.* Another group of 15 standard methods have been published by ASTM, some of which relate directly to . this study [l-5] In addition to the role as a participating test laboratory, NBS has also studied the RDF-3 test protocols for the purpose of simplifying procedures, attempting to clarify ambiguities, modifying procedures to improve the precision and accuracy of test data, providing reasons for the apparent dispersion of data, developing new procedures when there appeared to be such a need, and suggesting levels of precision for tests. Studies at NBS and elsewhere suggest that the heterogeneity of RDF still poses a problem to the extraction of a representative sample of RDF for analysis. Sampling techniques which will provide representative samples need further development. Poor agreement found for intra- and inter-laboratory data will not improve until advances in the sampling of RDF have been found. The focus of this paper is directed toward the evaluation of data on the calorific content, expressed as the higher heating values (HHV), of RDF-3 determined during the round robin testing program. As part of this evaluation of HHV** data, procedures for moisture and ash content determinations were also studied because of the need to calculate higher heating values on an as-received basis (HHVl), a dry basis (HHVB), and moisture-ash-free basis (HHV3). The initial path taken by Subcommittee E-38.01 toward the development of standard test methods for RDF-3 was to examine the ASTM procedures already established for coal and coke, and see what kinds of changes were necessary to accommodate the testing of RDF-3. Modifications were introduced as they appeared necessary. DEFINITIVE
ROUND ROBIN TESTING
Arrangements were made to begin a series of round robin tests on RDF-3. Samples were prepared at the National Center for Resource Recovery (NCRR), Washington, D.C., and were distributed to the various participating laboratories for the first two preliminary and first two definitive series of round robin tests. The samples for the third definitive series of testing were prepared and distributed by Americology, Milwaukee, Wisconsin. The RDF-3 laboratory sample was composed of 2.5 cm (1 inch) or smaller material. Prior to processing the sample to 0.5 mm particles for HHV, residual moisture, ash, and *Because of the concensus methods used by ASTM, their published standards are immediately adopted by the American National Standards Institute (ANSI), the U.S. representative in the International Standards Organization (ISO). **HHV is defined as the energy released by combustion in oxygen of a unit quantity of RDF at constant volume in a bomb calorimeter under specified conditions such that all water formed as a combustion product is in the liquid phase.
235
other determinations, the sample was air dried in the temperature range from 30” to 40°C so that its moisture content could be brought near to equilibrium with the atmosphere in the test laboratory. The chronology of the round robin tests is outlined in Table 1. Six laboratories participated in the first and second preliminary (zeroth) rounds of testing. The precision of the data did not seem satisfactory from these rounds. It was felt that a period of acclimation was required for laboratory analysts to acquire some experience in processing and analyzing RDF-3 samples. Eight laboratories participated in the first definitive series of round robin tests which took place in June and July of 1978. The mean and standard deviation of measurements on higher heating values (HHV) as calculated for three different bases are shown in Table 2; as-received basis (HHVl), dry basis (HHV2), and moistureash-free basis (HHV3). Corresponding data for total moisture, residual moisture, and ash content determinations are given in Table 3. Each mean value, x, listed in Tables 2-11 is the average of four determinations: two runs carried out on day one and two runs carried out TABLE 1 RDF-3 testing program Round robin test
Period of study
Zeroth round robin (first) Zeroth round robin (second) First round robin Second round robin Third round robin Bituminous coal round robin
Sept., Oct. Dec. 1977, June, July Dec. 1978, May, June Feb., Mar.
1977 Jan. 1978 1978 Jan. 1979 1979 1980
TABLE 2 ASTM round robin testing of RDF-3 (first round); statistical data on higher heating values; RDF-3 sample prepared by NCRR Laboratory
Average Std. dev.
Standard deviation S (MJ/kg)
Mean g (MJ/kg) HHVl
HHVP
HHV3
HHVl
HHV2
HHV3
15.056 15.605 15.182 14.177 14.307 15.887 13.628 14.849
17.489 18.240 17.643 17.273 16.717 17.880 16.117 17.182
22.281 23.758 22.041 22.783 23.018 22.374 20.820 21.943
0.053 0.086 0.151 0.040 0.121 0.030 0.407 0.212
0.056 0.107 0.174 0.051 0.140 0.035 0.451 0.240
0.135 0.154 0.440 0.244 0.070 0.044 0.721 0.454
14.835 0.712
17.317 0.668
22.378 0.830
0.126
0.156
0.284
236 TABLE
3
ASTM round robin testing of RDF-3 (first round); statistical data on total moisture, residual moisture, and ash; RDF-3 sample prepared by NCRR Laboratory
Standard
Mean _? (wt%) Total moist.
Res. moist.
deviation
Ash
Total moist.
S (wt%)
Res. moist.
Ash
1 2 3 4 5 6 7 8
13.91 14.44 13.94 13.59 14.41 11.15 15.45 13.58
4.845 3.865 3.853 4.478 ( 1.525)a 4.475 3.830 3.425
21.51 23.23 19.94 24.18 24.06 20.09 22.57 21.69
0.03 0.11 0.07 0.05 0.05 0.01 0.19 0.10
0.034 0.122 0.080 0.056 (0.054)” 0.180 0.216 0.108
0.24 0.28 1.67 0.97 0.74 0.50 1.31 0.91
Average Std. dev.
13.81 1.23
4.110 0.497
22.16 1.64
0.08
0.7 14
0.83
aData from Lab 5 not used in calculations.
on day two, each being performed in the same laboratory, by the same analyst, using the same apparatus. The standard deviation, S, of the four replicates is provided in Tables 2-11. The classification of days and within-days was ignored in calculating S; however, this statistic is still a rough estimate of the variability within laboratories. Also provided in Tables 2-11 are averages of the means for the first round robin sample for the participating laboratories, the corresponding standard deviations, and the average value of the laboratory standard deviations. Examination of the data on HHVl, HHVB, and HHV3 in Table 2 showed that the dispersion of the data within each laboratory appeared higher than what might be expected and, in most cases, was considerably higher than what ASTM protocols for coal and coke (Method D2015) would allow. The standard deviations of the average of the means for HHVl, HHVB, and HHV3 were five to seven times the dispersion expected. The dispersion of total moisture, residual moisture, and ash content data tabulated in Table 3 also appeared to be disproportionately high. Doubts about the homogeneity of the milled (particle size 0.5 mm) RDF-3 sample were being raised at the ASTM meeting following the first round of testing. Some analysts suggested that sub-division was difficult because segregation of light and heavy particles was sometimes observed in either riffling or coning and quartering of the sample in preparation for analysis. The accountability on non-millable portions of the RDF-3 sample was not easy. The personal and professional judgment of the analyst was required for this decision. It more than likely varied from one laboratory to another. Unequal distribution of nonmillable material in the laboratory samples also contributed to this variation.
237
INTER.-LABORATORY COMPARISON SAMPLES In an effort to ascertain whether improved precision could be obtained for higher heating values, moisture and ash content determinations, and other important properties, an interlaboratory reference or comparison sample was prepared for round robin testing, in addition to the usual sample of RDF-3 for the second and third rounds. The inter-laboratory comparison sample was milled to 0.5 mm particles by a single laboratory prior to distribution to all of the participants. The comparison sample would eliminate possible biases which might have been introduced into the particle size reduction operation at each of the laboratories. Differences of opinion as to how: (1) the RDF-3 sample should be milled, split, and blended; (2) non-millable materials should be identified and extracted; and (3) sampling should be performed - suggested that the decisions made by analysts at the various laboratories were not always equivalent. An inter-laboratory comparison sample, it was hoped, would remove such biases. Tables 4 and 5 provide data on the higher heating values, moisture, and ash content for the RDF-3 sample prepared and distributed by Americology to twelve participating laboratories in the second round of testing. A significant improvement in precision was observed except for the results on residual moisture. The standard deviation of the average of the means for the residual moisture had doubled while the actual average of the means itself had decreased from 4.110 to 2.715 wt%. An examination of the standard deviations of the average of the means* for HHV2 in Tables 2 and 4 shows a decrease from 0.668 to 0.370 MJ/kg. A more modest improvement is found in Tables 3 and 5 for total moisture and ash data when the standard deviations of these averages are inspected. Perhaps the laboratory analysts were becoming more accustomed to handling RDF-3. Results in general were still far from the precisions found for coal and coke. Tables 6 and 7 show the statistical data for the inter-laboratory comparison sample for the second round. The improvement in precision was very encouraging. The standard deviation of the average of the means for HHV2 was 0.140 MJ/kg; this represents about a 60% improvement over the corresponding data for HHV2 which was determined after each laboratory carried out its own milling of their laboratory samples. Similarly, the standard deviation of the average of the means for the ash content data for the interlaboratory comparison sample in Table 7 was about half (0.73 wt%) of that found for the corresponding laboratory sample data shown in Table 5 (1.46 wt%). The standard deviation of the average of the means for the residual moisture of the inter-laboratory sample decreased by half in comparison with the The values are given to three decimal figures so that con*429.9226 BTU (lb)-‘/MJ(kg)-‘. versions to BTU(lb)-’ may be made.
238 TABLE 4 ASTM round robin testing of RDF-3 (second round); statistical data on higher heating values; RDF-3 sample prepared by NCRR Standard deviation S (MJ/kg)
Laboratory
Mean 2 (MJ/kg) HHVl
HHV2
HHV3
HHVl
HHV2
HHV3
1 2 3 4 5 6 7 8 9 10 11 12
15.003 15.403 15.952 14.640 14.710 15.159 15.056 14.917 -
18.047 18.996 18.545 18.378 17.992 18.636 18.703 18.227 -
21.634 21.857 21.634 21.476 20.450 21.467 21.762 21.318 -
0.021 0.066 0.053 0.079 0.040 0.098 0.088 0.151 -
0.026 0.119 0.060 0.105 0.051 0.119 0.102 0.209 -
0.056 0.053 0.086 0.170 0.060 0.149 0.063 0.340 -
14.382 16.522 15.624
18.099 18.373 19.094
21.643 21.606 21.802
0.035 0.007 0.165
0.035 0.012 0.237
0.077 0.072 0.279
Average Std. dev.
15.126 0.577
18.464 0.370
21.513 0.386
0.072
0.098
0.128
TABLE 5 ASTM round robin testing of RDF-3 (second round); statistical data on total moisture, residual moisture, and ash; RDF-3 sample prepared by NCRR Laboratow 1 2 3 4 5 6 7 8 9 10 11 12 Average Std. dev.
Mean 2
Standard deviation S (wt%)
(wt%)
Total moist.
Res. moist.
Ash
Total moist.
Res. moist.
Ash
16.87 18.89 19.26 20.34 18.24 18.66 19.48 18.34 -
2.185 2.845 2.998 2.358 2.413 3.135 2.613 2.158 -
16.58 13.09 14.28 14.43 12.03 13.18 14.04 14.50 -
0.03 0.35 0.15 0.07 0.06 0.07 0.04 0.07 -
0.436 0.451 0.175 0.079 0.071 0.081 0.048 0.288 -
0.22 0.48 0.09 0.31 0.33 0.16 0.38 0.45 -
20.53 ( 10.17)a 18.17
2.190 1.568 5.405
16.41 14.88 12.41
0.06 (0.14)a 0.17
0.070 0.173 0.205
0.38 0.20 0.93
18.91 1.08
2.715 0.996
14.17 1.46
0.11
0.189
0.36
aData from Lab 11 not used in calculations.
___---
239
TABLE
6
ASTM round robin testing for RDF-3 (second round); statistical data on higher heating values; RDF-3 inter-laboratory comparison sample prepared by Wisconsin Electric Laboratory
Mean .?
Standard
(MJ/kg)
HHVl
HHV2
1
-
17.729
2 3 4 5 6 7 8 9 10 11 12
-
17.722 17.610 17.740 17.550 17.703 ( 15.614)a 17.566 17.806 17.796 18.038 17.866
22.220 22.344 21.830 22.169 21.369 22.251 (19.536)a 21.855 22.278 22.027 22.567 22.153
17.738 0.140
22.097 0.321
-
Average Std. dev.
_
HHV3
deviation
HHVl
S (MJ/kg)
HHV2
HHV3
-
0.093
-
0.147 9.135 0.070 0.126 0.065 ( 1.456)a 0.477 0.067 0.063 0.088 0.254
0.137 0.472 0.405 0.135 0.293 0.142 (1.865)a 0.400 0.181 0.116 0.156 0.170
0.144
0.237
-
aData from Lab 7 not used in calculations.
TABLE
7
ASTM round robin testing of RDF-3 (second round); statistical data on residual moisture and ash; RDF-3 inter-laboratory comparison sample prepared by Wisconsin Electric Laboratory
Standard
Mean _? (wt%)
deviation
S
Res. moist.
Ash
Res. moist.
Ash
1 2 3 4 5 6 7 8 9 10 11 12
3.360 3.335 3.278 3.353 2.570 3.415 2.788 3.323 2.645 1.683 3.070 3.378
20.22 20.67 19.32 19.97 17.86 20.44 20.05 19.63 20.08 19.49 20.07 19.34
0.070 0.484 0.182 0.079 0.061 0.785 0.120 0.239 0.065 0.052 0.224 0.303
0.43 1.32 0.90 0.40 1.42 0.76 1.23 1.56 0.41 0.22 0.17 1.33
Average Std. dev.
3.017 0.518
19.76 0.73
0.22
0.85
(wt%)
240
regularly processed second round sample. This improvement suggested that fewer biases were interacting. A third round of testing was carried out during the late spring of 1979. The statistical data calculated from the results of testing on both the regular RDF-3 laboratory sample and another interlaboratory comparison sample are presented in Tables 8-11. These results are somewhat disappointing in that they resemble the first round in their level of precision more than the second round. Comparison of HHVl, HHVZ, and HHV3 in Table 8 to Table 2 shows the standard deviation of the average of the means 10 to 20% lower in the third round than in the first round for the regular RDF-3 laboratory sample. Comparison of this same statistic in Tables 3 and 9 shows that the precision for total moisture, residual moisture, and ash content determination has decreased significantly; from 1.23 to 1.75 wt% for the total moisture, from 0.497 to 1.54 wt% for residual moisture, and from 1.64 to 2.61 wt% for ash, respectively. Precisions were also lower when the standard deviation of the average of the means is examined for the results of the third inter-laboratory comparison sample and compared to the corresponding data in the second round of testing. For example, this statistic is 0.358 MJ/kg for HHV2 for the third round (Table 10) as compared with 0.140 MJ/kg for the second round (Table 6). Similarly, the standard deviation of the average of the means for the ash data for the third round is more than three times (2.50 wt%, Table 11) TABLE
8
ASTM round robin testing of RDF-3 (third round); values; RDF-3 sample prepared by Americology _ Laboratory
Mean 2
statistical
Standard
(MJ/kg)
data on higher heating
deviation
S (MJ/kg)
HHVl
HHVZ
HHV3
HHVl
HHVP
HHV3
1 2 3 4 5 6 7 8 9 10 11 12
12.553 12.621 11.998 11.928 12.607 12.044 15.487 13.721 11.972 11.528 10.939 13.202
17.952 18.252 17.296 17.010 17.366 17.084 17.194 17.715 17.191 16.887 16.501 18.643
22.646 22.913 22.188 21.953 22.374 23.079 22.478 23.918 21.174 22.099 21.455 22.520
0.051 0.049 0.293 0.065 0.026 0.091 0.070 0.868 0.042 0.040 0.014 0.414
0.070 0.040 0.407 0.086 0.037 0.107 0.195 0.112 0.060 0.060 0.016 0.502
0.058 0.088 0.354 0.102 0.044 0.516 0.391 0.184 0.077 0.105 0.033 1.033
Average Std. dev.
12.549 0.579
17.424 0.609
22.399 0.728
0.167
0.142
0.249
-
241
TABLE
9
ASTM round robin testing of RDF-3 (third round); residual moisture, and ash; RDF-3 sample prepared Laboratory
statistical data on total moisture, by Americology
Mean _? (wt% ) Total
moist.
Standard Res. moist.
Ash
Total
deviation
moist.
S (wt% )
Res. moist.
Ash
1
30.08
4.470
20.73
0.10
0.134
0.20
2
30.84
3.485
20.33
0.18
0.249
0.19
3 4 5 6 7 8 9 10 11 12
30.64 29.88 27.59 29.51 (9.93)a 27.54 30.35 31.74 33.70 29.17
3.573 4.796 4.063 4.960 -
0.07 0.04 0.09 0.10 (0.67)a 0.62 0.02 0.06 0.08 0.35
0.097 0.048 0.119 0.132 -
7.410 4.523 3.163 3.780 7.843
22.05 22.52 22.38 25.95 23.50 25.94 18.81 23.58 23.10 17.17
0.798 0.340 0.062 0.117 0.457
0.70 0.31 0.26 1.20 0.57 0.51 0.16 0.29 0.10 1.57
30.09 1.75
4.733 1.54
22.17 2.61
0.16
0.232
0.51
Average Std. dev.
_
aData from Lab 7 not used in calculations.
TABLE
10
ASTM round robin testing of RDF-3 (third round); statistical data on higher heating values; RDF-3 inter-lab comparison sample prepared by ORNL Laboratory
-__ 1 2 3 4 5 6 7 8 9 10 11 12 Average Std. dev.
Mean x
(MJ/kg)
Standard
deviation
S (MJ/kg)
HHVl
HHV2
HHV3
HHVl
HHV2
HHV3
-
17.515 17.143 17.077 16.626 17.173 16.694 16.926 17.068 16.787 17.108 17.894 16.782
21.590 22.162 21.492 21.329 22.027 23.488 21.916 22.441 21.971 21.429 22.697 21.376
-
0.070 0.065 0.233 0.165 0.012 0.098 0.149 0.295 0.063 0.053 0.105 0.661
0.144 0.228 0.333 0.258 0.016 0.614 0.326 0.219 0.107 0.302 0.230 0.889
17.066 0.358
21.992 0.642
0.165
0.305
-
242 TABLE
11
ASTM round robin testing of RDF-3 and ash; RDF-3 inter-lab comparison Laboratory
Mean 2
(third round); statistical data on residual sample prepared at ORNL Standard
(wt%)
deviation
moisture
S (wt%)
Res. moist.
Ash
Res. moist.
Ash
1 2 3 4 5 6 7 8 9 10 11 12
4.605 4.373 3.775 4.434 4.135 3.088 5.418 4.298 3.868 4.128 5.328 4.698
18.87 22.64 20.54 22.05 21.71 28.91 22.75 23.94 23.60 20.27 21.44 21.49
0.03 0.18 0.11 0.07 0.09 0.64 0.41 0.22 0.02 0.16 0.37 0.07
0.35 0.66 0.54 0.54 0.07 1.52 1.56 0.64 0.37 0.82 0.39 0.52
Average Std. dev.
4.346 0.643
22.35 2.50
0.64
0.67
that found for the inter-laboratory sample studied in the second round (0.73 wt%, Table 7). The residual moisture, however, is not all that much different in the two inter-laboratory rounds of testing; compare this statistic in Table 11 (0.643 wt%) with the corresponding value in Table 7 (0.518 wt%). BITUMINOUS
COAL
ROUND
ROBIN
The data obtained for HHV, moisture, and ash content determinations for the three rounds of testing were significantly more variable than had been expected. In order to establish some reassurance that a non-RDF reference sample could be analyzed satisfactorily, a Standard Reference Material bituminous coal sample (SRM 1632a from NBS) was distributed to six laboratories by NBS for HHV, residual moisture, and ash determinations. The results of this series of tests are shown in Tables 12 and 13. In Table 12, it is found that the standard deviations of the average of the means for the coal sample are about the same for HHVl, HHVB, and HHV3, i.e., -0.116 MJ/kg. This value is considerably lower than comparable values of the standard deviation of the average of the means found for higher heating values in any of the three rounds of testing. Similarly, values of this statistic for residual moisture and ash content for the coal sample are much lower than comparable values found in earlier rounds of RDF-3 testing; for the residual moisture, the standard deviation of the average of the means is 0.207 wt% and for ash is 0.15 wt% (Table 13). This series of tests on bituminous coal showed that the participating labora. tories could analyze a coal sample to suitable levels of precision.
243 TABLE
12
ASTM round robin testing of RDF-3; statistical sample (SRM 1632a)’ distributed by NBS
values; coal
duplicate
runs (MJ/kg)
HHVl
HHV2
HHV3
HHVl
HHV2
HHV3
1 3 4 5 7 8
25.947
26.109
26.205 26.014 26.098 25.905 25.967
26.602 26.316 26.465 26.247 26.426
33.834 34.043 33.806 33.988 33.734 33.922
0.000 0.056 0.009 0.037 0.035 0.002
0.002 0.133 0.012 0.037 0.019 0.002
0.074 0.167 0.033 0.49 0.028 0.005
Average Std. dev.
26.023 0.112
26.405 0.126
33.887 0.116
0.023
0.035
0.060
Laboratory
Mean Xof
data on higher heating
aSRM = Standard TABLE
Reference
Standard
deviation
S (MJ/kg)
Material.
13
ASTM round robin testing of RDF-3; coal sample (SRM 1632a) distributed Laboratory
statistical by NBS
data on residual
moisture
deviation
and ash;
Mean _? of duplicate
runs (wt%)
Standard
S (wt%)
Res. Moisture
Ash
Res. Moisture
Ash
1 3 4 5 7 8
1.590 1.485 1.145 1.355 1.305 1.720
22.21 21.80 22.16 22.12 22.20 22.10
0.014 0.276 0.007 0.007 0.064 0.042
0.007 0.000 0.042 0.050 0.007 0.007
Average Std. dev.
1.433 0.207
22.10 0.15
0.068
0.019
REPEATABILITY
AND REPRODUCIBILITY
In order to provide some quantitative estimate of where the current stateof-the-art stands with respect to the analysis of RDF-3 samples in contrast to the analysis of coal, a statistical study of the data accumulated in the first three rounds of testing was conducted using a program based upon the theory of nested (hierarchical) designs [6]. In this case, days were nested within laboratories and replicates within days. There were two days of testing for each laboratory, and on each day two replicate determinations were carried out. The data obtained on the NBS bituminous coal sample (SRM 1632a) were
244
analyzed using standard methods [ 71. In this case, each laboratory carried out two replicate determinations, but only on one day. Two statistics were particularly sought, namely, the repeatability interval, Z(r), which is a measure of within-laboratory agreement, and the reproducibility interval, Z(R), which is a measure of between-laboratory agreement. These statistics could easily be compared with existing values listed in documented ASTM Standard Methods of Test for Coal and Coke for HHV, residual moisture and ash. For example, in ASTM D2015 (Standard Test Method for Gross Calorific Value of Solid Fuel by the Adiabatic Bomb Calorimeter) the precision criteria for judging the acceptability of results at the 95% probability level are: Repeatability.
Duplicate results by the same laboratory on different days, using the same operator and equipment should not be considered suspect unless they differ by more than 0.116 MJ/kg, dry basis.
Reproducibility.
Results submitted by two or more laboratories (different equipment, operators, date of test, and different portions of the same gross sample) should not be considered suspect unless two results differ by more than 0.233 MJ/kg, dry basis. Equations required to calculate the repeatability interval, Z(r), are:
Z(r) = 2.83s(r);
(I
where s(r) is the pooled within-laboratory standard deviation and 2.83 is a factor which converts the repeatability standard deviation to a 95% repeatability interval, and s(r) = [s2 (repl) + s2 (days)] li2,
(2
where s(rep1) is the standard deviation of replicate measurements and s (days) is the standard deviation due to variations in measurements observed between days. Similarly, equations needed to calculate the reproducibility interval, Z(R), are: Z(R) = 2.83s(R),
(3)
where s(R) is the between-laboratory fined as above, and
estimate
s(R) = [s2(repl) + s2(days) + s’(labs)]
‘j2,
of precision
and 2.83 is de(4)
where s(labs) is the standard deviation due to variations in measurements observed between laboratories and s(rep1) and s(days) are as defined above. Tables 14-17 summarize the repeatability and reproducibility intervals calculated for HHVB, total moisture, residual moisture, and ash, respectively. Data on all three RDF-3 rounds of testing are tabulated including the results from testing of the bituminous coal sample, SRM 1632a. The first row
245 TABLE 14 Summary of repeatability and reproducibility intervals for higher heating value HHVB, dry basis, in ASTM round robin testing of RDF-3 Method/Test
ASTM D 2015, coal and coke Coal (SRM 1632a) First round robin Second round robin Third round robin Second round interlab comparison sample Third round interlab comparison sample Adopted values
HHV2 (MJ/kg)
ReprodJRepeat., Z(R)lZ(r)
Average
Repeatability interval, Z(r)a
Reproducibility interval, Z( R)a
-
0.116
0.233
2.00
26.405 17.317 18.464 17.424 17.738
0.160 0.661 0.340 0.579 0.554
0.370 1.956 1.091 1.796 0.605
2.30 2.96 3.21 3.10 1.09
17.066
0.688
1.168
1.70
-
0.582
1.745
3.00
‘95% probability.
of Tables 14,16, and 17 gives the acceptability limits for HHVB, residual moisture, and ash content, respectively, according to the appropriate ASTM protocol for coal and coke. No acceptability limits are available for total moisture in coal and coke protocols. Examination of the HHV2 data in Table 14 shows that the repeatability interval for the coal, 0.160 MJ/kg, is close to the acceptable limit of 0.116 MJ/kg. The reproducibility interval obtained for the coal is 0.370 MJ/kg; this departs somewhat from the acceptable limit of 0.233 MJ/kg yielding a ratio of I(R) to I(r) of 2.30 rather than 2.00. The exact statistical significance of the ratio I(R)/l(r) is not clear. A ratio of one (or close to this number) does, however, indicate that the samples supplied to the laboratories are identical and have been analyzed in an equivalent manner. A ratio greater than one indicates that the samples are not identical and/or the analyses of the laboratories are not being done in an equivalent manner. Among the systematic errors which will tend to make this ratio large are different approaches or methods of field-sample subdivision, different procedures for blending (or the absence of blending), different opinions of what constitutes a non-millable portion of the sample, apparatus and person or operator interactions (heat build-up in the Wiley mill used, errors in the calculation of the corrected temperature rise during a calorimetric run, characteristics of a furnace at the ashing temperature, judgement of the technician in weighing a sample, or whether a sample has reached constant weight during drying at 105°C or ashing at 725’C.) In this paper, this ratio is used to help establish a preliminary basis for comparing the variability of RDF-3 with coal. The repeatability interval for the first three rounds of testing of RDF-3 is
246
between three and six times the 0.116 MJ/kg limit. This includes the interlab comparison samples tested in rounds two and three. The values of I(r) for the three rounds of testing give an average value of 0.526 MJ/kg. The corresponding average value for I(R) is 1.614 MJ/kg. It may be observed that the I(R)/l(r) ratio for HHV2 for coal and coke is approximately 2.00, and for RDF-3 is close to 3.00. For practical purposes, RDF-3 repeatability and reproducibility intervals for HHV2 of 0.582 and 1.745 MJ/kg, respectively, may be adopted. The latter values are much larger than those given for coal and coke, but represent the current state-of-the-art in the repeatability and reproducibility of HHV2 for RDF-3.
Fig. 1. RDF-3 reduced to 0.5 mm; magnification 50 times (each square is 0.25 mm on a side).
247 HETEROGENEITY
OF RDF-3
The heterogeneity of RDF-3 is still present at 0.5 mm particle size. Figs. 1 and 2 show analysis samples of 0.5 mm RDF-3 magnified 50 times under a microscope, The particles are heterogeneous even at this level of subdivision. Some are spherical, some are elongated rods, while others are irregularly shaped. Differences in density are suggested by the black and white contrast shown in the photos.
Fig. 2. RDF-3 reduced to 0.5 mm; magnification 50 times (each square is 0.25 mm on a side)
248
Fig. 3. SRM 1632a bituminous coal sample; magnification 50 times (each square is 0.25 mm on a side).
Fig. 3 is the bituminous coal sample used, magnified 50 times under a microscope. The particles are 0.25 mm (-60 mesh) and have a general homo geneous appearance in comparison to the RDF-3 at the same magnification. Differences in the densities of 0.5 mm plastic, cellulose, iron, aluminum, sand, etc. particles in RDF-3 help to defy the blending process and extraction of representative samples for various tests. The values of I(r) and I(R) for the inter-laboratory comparison samples
249
of rounds two and three show that the second round sample gave better results than the third round. The reason for this is not immediately clear, but the I( R)/l(r) ratio does illustrate the benefits of having only one laboratory perform the processing of 2-3 cm nominal RDF-3 down to 0.5 mm. Biases in the reduction of the size of RDF-3 are an inherent part of the reproducibility interval values and cause I(R) to be so much larger than I(r) for the regular rounds of testing. SUMMARY OF RESULTS
The summary of I(r) and I(R) data for total moisture is given in Table 15. A trend is observed between the average of the means and the repeatability interval I(r). Samples increased in total moisture as testing went from the first, to the second round, and to the third rounds; compare 13.82, 18.91, and 30.09 wt% with I(r) values of 0.37, 0.57, and 0.76 wt%, respectively. Intra-laboratory repeatability lies between 0.4 and 0.8 wt% while inter-laboratory reproducibility lies between 3.4 and 5.0 wt%. The ratio of I(R)/l(r) is poorest for the first round, having a value of 9.21. In Table 16 are given I(r) and I(R) values for residual moisture. The limits imposed on coal and coke samples (ASTM D3173) are related to the moisture content of the samples and are given in the table. For example, if less than 5 wt% moisture is present in the coal or coke sample, then I(r) and I(R) values are not considered suspect unless they are above 0.2 and 0.3 wt%, respectively. The data obtained by the six participating laboratories for the standard coal sample fall within the above acceptability limits since the sample contained 1.43 wt% residual moisture. There appears to be a trend in the I(r) values for the three rounds of testing similar to that observed in Table 15 related to the average of the means of the total moisture. Whether there is a relationship or whether this is merely chance is not clear. The averages of the means of the residual moisture do not parallel the I(r) trend. The ratio of I(R)/l(r) ranges from 3.5 to 4.5. TABLE 15 Summary of repeatability and reproducibility of total moisture in ASTM round robin testing of RDF-3 _ Test
First round robin Second round robin Third round robin a95% probability.
Total moisture (wt%) Average
Repeatability interval, Z(r)a
Reproducibility interval, Z(R)a
13.82 18.91 30.09
0.37 0.57 0.76
3.41 3.24 4.99
Reprod./Repeat., Z(R)lZ(r)
9.21 5.68 6.57
250 TABLE
16
Summary of repeatability robin testing of RDF-3
and reproducibility
Method/Test
ASTM D 3173, coal and coke less than 5% moisture more than 5% moisture Coal (SRM 1632a) First round robin Second round robin Third round robin Second round interlab comparison sample Third round interlab comparison sample Adopted values
for residual moisture in ASTM round
Residual moisture (wt%)
Reprod./ Repeat., Z(R)/Z(r)
Average
Repeatability interval, Z(r)a
Reproducibility interval, Z(R)a
1.42 4.11 2.72 4.73
0.2 0.3 0.12 0.41 0.63 0.99
0.3 0.5 0.23 1.44 2.86 4.42
1.50 1.67 1.92 3.51 4.54 4.46
3.02
0.58
1.55
2.67
4.35
0.84
1.93
2.30
-
0.75
3.00
4.00
a95% probability.
TABLE
17
Summary of repeatability and reproducibility robin testing of RDF-3 Method/test
Ash (dry basis) (wt%) Average
ASTM D 3174. coal and coke no carbonates carbonates more than 12% ash 22.10 Coal (SRM 1632a) First round robin 22.16 14.17 Second round robin Third round robin 22.17 Second round interlab 19.76 comparison sample Third round interlab 22.35 comparison sample Adopted values a95% probability.
intervals for ash content in ASTM round
_
Reprod./ Repeat., Z(R)/Z(r)
Repeatability interval, Z(r)a
Reproducibility interval, Z(R)a
0.2 0.3 0.5 0.08 2.73 1.31 1.88
0.3 0.5 1.0 0.16 5.20 4.26 7.57
1.50 1.67 2.00 2.10 1.90 3.25 4.03
3.09
3.16
1.02
2.54
7.35
2.89
2.5
7.5
3.00
251
Again, for practical considerations, the ratio 4.00 with 0.75 and 3.00 wt% as values for I(r) and I(R), respectively, can be adopted. No limits on the amount of residual moisture in the sample have been set as with coal and coke. Table 17 is the summary of values of I(r) and I(R) for the ash content data. The limits for acceptable data with regard to coal and coke samples (ASTM D3174) which contain (or do not contain) carbonates or more than 12 wt% moisture are given in the first three rows of Table 17. The results of testing the standard coal sample show that the participating laboratories have determined that the ash content is well within the levels of acceptability. Values of 0.08 and 0.16 wt% for I(r) and I(R), respectively, are quite satisfactory. The repeatability and reproducibility intervals for the first three rounds are quite high in comparison with the coal and coke acceptability limits. The same is true for the I(r) and I(R) values obtained for the interlaboratory comparison sample. Although the second round data for I(r) and I(R) are better for both the regular and inter-laboratory samples, the absolute values of the average of the means are significantly different; compare 14.17 wt% for the regular RDF-3 sample with 19.76 wt% for the inter-labora. tory sample. This difference may be due to overall heterogeneity of RDF-3. Other biases may be involved but are not apparent. The suggested values for I(r) and I(R) are 2.5 and 7.5 wt%, respectively. These values are based on an adopted I(R)/l(r) ratio of 3.00. This is much higher than expected, but this is what is observed. ACKNOWLEDGEMENTS
Funding for this work was provided by the NBS Office of Recycled Materials and the U.S. Department of Energy Office of Energy from Municipal Waste. The cooperative efforts of the membership of ASTM Committee E-38 on Resource Recovery are acknowledged, and, in particular, those of ASTM Subcommittee E-38.01 on Energy. Special thanks are expressed to Dr. John Love, Jr., Chairman of ASTM Subcommittee E-38.01, for his strong leadership and perseverance during the RDF-3 round robin testing program, and to the participating laboratories for their responsiveness and cooperation.
REFERENCES 1 ASTM Method E711-81, Standard test method for gross calorific value of refuse-derived fuel (RDF-3) by the bomb calorimeter. 2 ASTM Method E790-81, Standard test method for residual moisture in a refuse-derived fuel analysis sample. 3 ASTM Method E791-81, Standard method for calculating refuse-derived fuel analysis data from as-determined to different bases. 4 ASTM Method E830-81, Standard method of test for ash in the analysis sample of refuse-derived fuel (RDF-3).
252 5 ASTM Method E856-81, Standard definition of terms relating to the physical and chemical characteristics of refuse-derived fuel. 6 Davies, O.T., Statistical Methods in Research and Production, Third Edition, published for Imperial Chemical Industries, Ltd., Oliver and Boyd, London and Edinburgh; Hafner Publishing Co., New York, 1957. 7 ASTM Method E691-79, Standard practice for conducting an interlaboratory test program to determine the precision of test methods.
and Conservation,
8(1983)
233-252
233
Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands
EVALUATION OF DATA ON CALORIFIC CONTENT VALUES DETERMINED DURING ASTM ROUND ROBIN TESTING OF RDF-3
EUGENE National
S. DOMALSKI Bureau
and STANLEY
of Standards,
Washington,
ABRAMOWITZ D.C.
20234
(U.S.A.)
(Received May 5, 1982; accepted in revised form September
7, 1982)
ABSTRACT The potential application of refuse-derived fuel, in particular RDF-3, as an alternative or supplemental fuel is dependent upon its acceptance as an article of commerce. ASTM Committee E-38 on Resource Recovery, and its Subcommittee on Energy, E-38.01, have been actively engaged in the development of concensus standards for this purpose especially since April 1974; standard procedures for the characterization of RDF-3 are being developed. These procedures are based on those ASTM methods used in coal analysis. The procedures developed will insure a meaningful purchase--sales relationship between the buyer and seller. A variety of chemical and physical test procedures were studied by as many as 12 laboratories. A total of 28 editorial draft standards have been prepared. Of this number, 13 have appeared as accepted methods in the Annual Books of ASTM Standards while the remainder are either under evaluation or under some phase of ASTM balloting. The National Bureau of Standards, in cooperation with ASTM subcommittee E38.01, has undertaken a technical review of a selected group of chemical properties of RDF-3. The property of principal interest is the calorific content and is expressed as the higher heating value. In order to properly characterize this property, critical evaluation of methods to determine total moisture, residual moisture, and ash is also necessary. Intra-laboratory (within-laboratory) and inter-laboratory (between-laboratory) variations in determinations for moisture, ash, and higher heating value are discussed. A comparison of these results with those for round robin data for coal is also made. The result is identification of the levels of precision for intra-laboratory and inter-laboratory agreement. INTRODUCTION
The National Bureau of Standards (NBS) has been a participant in the development of standard test methods for one form of refuse-derived-fuel (RDF-3)* with ASTM (American Society for Testing and Materials) Committee E-38 (Resource Recovery) and its Subcommittee E-38.01 (Energy) since the beginning of the testing program in the summer of 1977. Since that time, two preliminary and three definitive series of round robin RDF-3 samples * RDF-3 is defined as a shredded fuel derived from municipal solid waste (MSW) which has been processed for the removal of metal, glass, and other entrained inorganic materials. Generally, this material has a particle size such that 95 wt% passes through a 5-cm (2-inch) square mesh screen.
0166-3097/83/0000-0000/$03.00
0 1983
Elsevier Scientific Publishing Company
234
have been distributed and examined by various testing laboratories. At present, some 15 test protocols are in the form of editorial draft standards. These standards are undergoing evaluation and are under vote for acceptance by ASTM procedures as standard methods.* Another group of 15 standard methods have been published by ASTM, some of which relate directly to . this study [l-5] In addition to the role as a participating test laboratory, NBS has also studied the RDF-3 test protocols for the purpose of simplifying procedures, attempting to clarify ambiguities, modifying procedures to improve the precision and accuracy of test data, providing reasons for the apparent dispersion of data, developing new procedures when there appeared to be such a need, and suggesting levels of precision for tests. Studies at NBS and elsewhere suggest that the heterogeneity of RDF still poses a problem to the extraction of a representative sample of RDF for analysis. Sampling techniques which will provide representative samples need further development. Poor agreement found for intra- and inter-laboratory data will not improve until advances in the sampling of RDF have been found. The focus of this paper is directed toward the evaluation of data on the calorific content, expressed as the higher heating values (HHV), of RDF-3 determined during the round robin testing program. As part of this evaluation of HHV** data, procedures for moisture and ash content determinations were also studied because of the need to calculate higher heating values on an as-received basis (HHVl), a dry basis (HHVB), and moisture-ash-free basis (HHV3). The initial path taken by Subcommittee E-38.01 toward the development of standard test methods for RDF-3 was to examine the ASTM procedures already established for coal and coke, and see what kinds of changes were necessary to accommodate the testing of RDF-3. Modifications were introduced as they appeared necessary. DEFINITIVE
ROUND ROBIN TESTING
Arrangements were made to begin a series of round robin tests on RDF-3. Samples were prepared at the National Center for Resource Recovery (NCRR), Washington, D.C., and were distributed to the various participating laboratories for the first two preliminary and first two definitive series of round robin tests. The samples for the third definitive series of testing were prepared and distributed by Americology, Milwaukee, Wisconsin. The RDF-3 laboratory sample was composed of 2.5 cm (1 inch) or smaller material. Prior to processing the sample to 0.5 mm particles for HHV, residual moisture, ash, and *Because of the concensus methods used by ASTM, their published standards are immediately adopted by the American National Standards Institute (ANSI), the U.S. representative in the International Standards Organization (ISO). **HHV is defined as the energy released by combustion in oxygen of a unit quantity of RDF at constant volume in a bomb calorimeter under specified conditions such that all water formed as a combustion product is in the liquid phase.
235
other determinations, the sample was air dried in the temperature range from 30” to 40°C so that its moisture content could be brought near to equilibrium with the atmosphere in the test laboratory. The chronology of the round robin tests is outlined in Table 1. Six laboratories participated in the first and second preliminary (zeroth) rounds of testing. The precision of the data did not seem satisfactory from these rounds. It was felt that a period of acclimation was required for laboratory analysts to acquire some experience in processing and analyzing RDF-3 samples. Eight laboratories participated in the first definitive series of round robin tests which took place in June and July of 1978. The mean and standard deviation of measurements on higher heating values (HHV) as calculated for three different bases are shown in Table 2; as-received basis (HHVl), dry basis (HHV2), and moistureash-free basis (HHV3). Corresponding data for total moisture, residual moisture, and ash content determinations are given in Table 3. Each mean value, x, listed in Tables 2-11 is the average of four determinations: two runs carried out on day one and two runs carried out TABLE 1 RDF-3 testing program Round robin test
Period of study
Zeroth round robin (first) Zeroth round robin (second) First round robin Second round robin Third round robin Bituminous coal round robin
Sept., Oct. Dec. 1977, June, July Dec. 1978, May, June Feb., Mar.
1977 Jan. 1978 1978 Jan. 1979 1979 1980
TABLE 2 ASTM round robin testing of RDF-3 (first round); statistical data on higher heating values; RDF-3 sample prepared by NCRR Laboratory
Average Std. dev.
Standard deviation S (MJ/kg)
Mean g (MJ/kg) HHVl
HHVP
HHV3
HHVl
HHV2
HHV3
15.056 15.605 15.182 14.177 14.307 15.887 13.628 14.849
17.489 18.240 17.643 17.273 16.717 17.880 16.117 17.182
22.281 23.758 22.041 22.783 23.018 22.374 20.820 21.943
0.053 0.086 0.151 0.040 0.121 0.030 0.407 0.212
0.056 0.107 0.174 0.051 0.140 0.035 0.451 0.240
0.135 0.154 0.440 0.244 0.070 0.044 0.721 0.454
14.835 0.712
17.317 0.668
22.378 0.830
0.126
0.156
0.284
236 TABLE
3
ASTM round robin testing of RDF-3 (first round); statistical data on total moisture, residual moisture, and ash; RDF-3 sample prepared by NCRR Laboratory
Standard
Mean _? (wt%) Total moist.
Res. moist.
deviation
Ash
Total moist.
S (wt%)
Res. moist.
Ash
1 2 3 4 5 6 7 8
13.91 14.44 13.94 13.59 14.41 11.15 15.45 13.58
4.845 3.865 3.853 4.478 ( 1.525)a 4.475 3.830 3.425
21.51 23.23 19.94 24.18 24.06 20.09 22.57 21.69
0.03 0.11 0.07 0.05 0.05 0.01 0.19 0.10
0.034 0.122 0.080 0.056 (0.054)” 0.180 0.216 0.108
0.24 0.28 1.67 0.97 0.74 0.50 1.31 0.91
Average Std. dev.
13.81 1.23
4.110 0.497
22.16 1.64
0.08
0.7 14
0.83
aData from Lab 5 not used in calculations.
on day two, each being performed in the same laboratory, by the same analyst, using the same apparatus. The standard deviation, S, of the four replicates is provided in Tables 2-11. The classification of days and within-days was ignored in calculating S; however, this statistic is still a rough estimate of the variability within laboratories. Also provided in Tables 2-11 are averages of the means for the first round robin sample for the participating laboratories, the corresponding standard deviations, and the average value of the laboratory standard deviations. Examination of the data on HHVl, HHVB, and HHV3 in Table 2 showed that the dispersion of the data within each laboratory appeared higher than what might be expected and, in most cases, was considerably higher than what ASTM protocols for coal and coke (Method D2015) would allow. The standard deviations of the average of the means for HHVl, HHVB, and HHV3 were five to seven times the dispersion expected. The dispersion of total moisture, residual moisture, and ash content data tabulated in Table 3 also appeared to be disproportionately high. Doubts about the homogeneity of the milled (particle size 0.5 mm) RDF-3 sample were being raised at the ASTM meeting following the first round of testing. Some analysts suggested that sub-division was difficult because segregation of light and heavy particles was sometimes observed in either riffling or coning and quartering of the sample in preparation for analysis. The accountability on non-millable portions of the RDF-3 sample was not easy. The personal and professional judgment of the analyst was required for this decision. It more than likely varied from one laboratory to another. Unequal distribution of nonmillable material in the laboratory samples also contributed to this variation.
237
INTER.-LABORATORY COMPARISON SAMPLES In an effort to ascertain whether improved precision could be obtained for higher heating values, moisture and ash content determinations, and other important properties, an interlaboratory reference or comparison sample was prepared for round robin testing, in addition to the usual sample of RDF-3 for the second and third rounds. The inter-laboratory comparison sample was milled to 0.5 mm particles by a single laboratory prior to distribution to all of the participants. The comparison sample would eliminate possible biases which might have been introduced into the particle size reduction operation at each of the laboratories. Differences of opinion as to how: (1) the RDF-3 sample should be milled, split, and blended; (2) non-millable materials should be identified and extracted; and (3) sampling should be performed - suggested that the decisions made by analysts at the various laboratories were not always equivalent. An inter-laboratory comparison sample, it was hoped, would remove such biases. Tables 4 and 5 provide data on the higher heating values, moisture, and ash content for the RDF-3 sample prepared and distributed by Americology to twelve participating laboratories in the second round of testing. A significant improvement in precision was observed except for the results on residual moisture. The standard deviation of the average of the means for the residual moisture had doubled while the actual average of the means itself had decreased from 4.110 to 2.715 wt%. An examination of the standard deviations of the average of the means* for HHV2 in Tables 2 and 4 shows a decrease from 0.668 to 0.370 MJ/kg. A more modest improvement is found in Tables 3 and 5 for total moisture and ash data when the standard deviations of these averages are inspected. Perhaps the laboratory analysts were becoming more accustomed to handling RDF-3. Results in general were still far from the precisions found for coal and coke. Tables 6 and 7 show the statistical data for the inter-laboratory comparison sample for the second round. The improvement in precision was very encouraging. The standard deviation of the average of the means for HHV2 was 0.140 MJ/kg; this represents about a 60% improvement over the corresponding data for HHV2 which was determined after each laboratory carried out its own milling of their laboratory samples. Similarly, the standard deviation of the average of the means for the ash content data for the interlaboratory comparison sample in Table 7 was about half (0.73 wt%) of that found for the corresponding laboratory sample data shown in Table 5 (1.46 wt%). The standard deviation of the average of the means for the residual moisture of the inter-laboratory sample decreased by half in comparison with the The values are given to three decimal figures so that con*429.9226 BTU (lb)-‘/MJ(kg)-‘. versions to BTU(lb)-’ may be made.
238 TABLE 4 ASTM round robin testing of RDF-3 (second round); statistical data on higher heating values; RDF-3 sample prepared by NCRR Standard deviation S (MJ/kg)
Laboratory
Mean 2 (MJ/kg) HHVl
HHV2
HHV3
HHVl
HHV2
HHV3
1 2 3 4 5 6 7 8 9 10 11 12
15.003 15.403 15.952 14.640 14.710 15.159 15.056 14.917 -
18.047 18.996 18.545 18.378 17.992 18.636 18.703 18.227 -
21.634 21.857 21.634 21.476 20.450 21.467 21.762 21.318 -
0.021 0.066 0.053 0.079 0.040 0.098 0.088 0.151 -
0.026 0.119 0.060 0.105 0.051 0.119 0.102 0.209 -
0.056 0.053 0.086 0.170 0.060 0.149 0.063 0.340 -
14.382 16.522 15.624
18.099 18.373 19.094
21.643 21.606 21.802
0.035 0.007 0.165
0.035 0.012 0.237
0.077 0.072 0.279
Average Std. dev.
15.126 0.577
18.464 0.370
21.513 0.386
0.072
0.098
0.128
TABLE 5 ASTM round robin testing of RDF-3 (second round); statistical data on total moisture, residual moisture, and ash; RDF-3 sample prepared by NCRR Laboratow 1 2 3 4 5 6 7 8 9 10 11 12 Average Std. dev.
Mean 2
Standard deviation S (wt%)
(wt%)
Total moist.
Res. moist.
Ash
Total moist.
Res. moist.
Ash
16.87 18.89 19.26 20.34 18.24 18.66 19.48 18.34 -
2.185 2.845 2.998 2.358 2.413 3.135 2.613 2.158 -
16.58 13.09 14.28 14.43 12.03 13.18 14.04 14.50 -
0.03 0.35 0.15 0.07 0.06 0.07 0.04 0.07 -
0.436 0.451 0.175 0.079 0.071 0.081 0.048 0.288 -
0.22 0.48 0.09 0.31 0.33 0.16 0.38 0.45 -
20.53 ( 10.17)a 18.17
2.190 1.568 5.405
16.41 14.88 12.41
0.06 (0.14)a 0.17
0.070 0.173 0.205
0.38 0.20 0.93
18.91 1.08
2.715 0.996
14.17 1.46
0.11
0.189
0.36
aData from Lab 11 not used in calculations.
___---
239
TABLE
6
ASTM round robin testing for RDF-3 (second round); statistical data on higher heating values; RDF-3 inter-laboratory comparison sample prepared by Wisconsin Electric Laboratory
Mean .?
Standard
(MJ/kg)
HHVl
HHV2
1
-
17.729
2 3 4 5 6 7 8 9 10 11 12
-
17.722 17.610 17.740 17.550 17.703 ( 15.614)a 17.566 17.806 17.796 18.038 17.866
22.220 22.344 21.830 22.169 21.369 22.251 (19.536)a 21.855 22.278 22.027 22.567 22.153
17.738 0.140
22.097 0.321
-
Average Std. dev.
_
HHV3
deviation
HHVl
S (MJ/kg)
HHV2
HHV3
-
0.093
-
0.147 9.135 0.070 0.126 0.065 ( 1.456)a 0.477 0.067 0.063 0.088 0.254
0.137 0.472 0.405 0.135 0.293 0.142 (1.865)a 0.400 0.181 0.116 0.156 0.170
0.144
0.237
-
aData from Lab 7 not used in calculations.
TABLE
7
ASTM round robin testing of RDF-3 (second round); statistical data on residual moisture and ash; RDF-3 inter-laboratory comparison sample prepared by Wisconsin Electric Laboratory
Standard
Mean _? (wt%)
deviation
S
Res. moist.
Ash
Res. moist.
Ash
1 2 3 4 5 6 7 8 9 10 11 12
3.360 3.335 3.278 3.353 2.570 3.415 2.788 3.323 2.645 1.683 3.070 3.378
20.22 20.67 19.32 19.97 17.86 20.44 20.05 19.63 20.08 19.49 20.07 19.34
0.070 0.484 0.182 0.079 0.061 0.785 0.120 0.239 0.065 0.052 0.224 0.303
0.43 1.32 0.90 0.40 1.42 0.76 1.23 1.56 0.41 0.22 0.17 1.33
Average Std. dev.
3.017 0.518
19.76 0.73
0.22
0.85
(wt%)
240
regularly processed second round sample. This improvement suggested that fewer biases were interacting. A third round of testing was carried out during the late spring of 1979. The statistical data calculated from the results of testing on both the regular RDF-3 laboratory sample and another interlaboratory comparison sample are presented in Tables 8-11. These results are somewhat disappointing in that they resemble the first round in their level of precision more than the second round. Comparison of HHVl, HHVZ, and HHV3 in Table 8 to Table 2 shows the standard deviation of the average of the means 10 to 20% lower in the third round than in the first round for the regular RDF-3 laboratory sample. Comparison of this same statistic in Tables 3 and 9 shows that the precision for total moisture, residual moisture, and ash content determination has decreased significantly; from 1.23 to 1.75 wt% for the total moisture, from 0.497 to 1.54 wt% for residual moisture, and from 1.64 to 2.61 wt% for ash, respectively. Precisions were also lower when the standard deviation of the average of the means is examined for the results of the third inter-laboratory comparison sample and compared to the corresponding data in the second round of testing. For example, this statistic is 0.358 MJ/kg for HHV2 for the third round (Table 10) as compared with 0.140 MJ/kg for the second round (Table 6). Similarly, the standard deviation of the average of the means for the ash data for the third round is more than three times (2.50 wt%, Table 11) TABLE
8
ASTM round robin testing of RDF-3 (third round); values; RDF-3 sample prepared by Americology _ Laboratory
Mean 2
statistical
Standard
(MJ/kg)
data on higher heating
deviation
S (MJ/kg)
HHVl
HHVZ
HHV3
HHVl
HHVP
HHV3
1 2 3 4 5 6 7 8 9 10 11 12
12.553 12.621 11.998 11.928 12.607 12.044 15.487 13.721 11.972 11.528 10.939 13.202
17.952 18.252 17.296 17.010 17.366 17.084 17.194 17.715 17.191 16.887 16.501 18.643
22.646 22.913 22.188 21.953 22.374 23.079 22.478 23.918 21.174 22.099 21.455 22.520
0.051 0.049 0.293 0.065 0.026 0.091 0.070 0.868 0.042 0.040 0.014 0.414
0.070 0.040 0.407 0.086 0.037 0.107 0.195 0.112 0.060 0.060 0.016 0.502
0.058 0.088 0.354 0.102 0.044 0.516 0.391 0.184 0.077 0.105 0.033 1.033
Average Std. dev.
12.549 0.579
17.424 0.609
22.399 0.728
0.167
0.142
0.249
-
241
TABLE
9
ASTM round robin testing of RDF-3 (third round); residual moisture, and ash; RDF-3 sample prepared Laboratory
statistical data on total moisture, by Americology
Mean _? (wt% ) Total
moist.
Standard Res. moist.
Ash
Total
deviation
moist.
S (wt% )
Res. moist.
Ash
1
30.08
4.470
20.73
0.10
0.134
0.20
2
30.84
3.485
20.33
0.18
0.249
0.19
3 4 5 6 7 8 9 10 11 12
30.64 29.88 27.59 29.51 (9.93)a 27.54 30.35 31.74 33.70 29.17
3.573 4.796 4.063 4.960 -
0.07 0.04 0.09 0.10 (0.67)a 0.62 0.02 0.06 0.08 0.35
0.097 0.048 0.119 0.132 -
7.410 4.523 3.163 3.780 7.843
22.05 22.52 22.38 25.95 23.50 25.94 18.81 23.58 23.10 17.17
0.798 0.340 0.062 0.117 0.457
0.70 0.31 0.26 1.20 0.57 0.51 0.16 0.29 0.10 1.57
30.09 1.75
4.733 1.54
22.17 2.61
0.16
0.232
0.51
Average Std. dev.
_
aData from Lab 7 not used in calculations.
TABLE
10
ASTM round robin testing of RDF-3 (third round); statistical data on higher heating values; RDF-3 inter-lab comparison sample prepared by ORNL Laboratory
-__ 1 2 3 4 5 6 7 8 9 10 11 12 Average Std. dev.
Mean x
(MJ/kg)
Standard
deviation
S (MJ/kg)
HHVl
HHV2
HHV3
HHVl
HHV2
HHV3
-
17.515 17.143 17.077 16.626 17.173 16.694 16.926 17.068 16.787 17.108 17.894 16.782
21.590 22.162 21.492 21.329 22.027 23.488 21.916 22.441 21.971 21.429 22.697 21.376
-
0.070 0.065 0.233 0.165 0.012 0.098 0.149 0.295 0.063 0.053 0.105 0.661
0.144 0.228 0.333 0.258 0.016 0.614 0.326 0.219 0.107 0.302 0.230 0.889
17.066 0.358
21.992 0.642
0.165
0.305
-
242 TABLE
11
ASTM round robin testing of RDF-3 and ash; RDF-3 inter-lab comparison Laboratory
Mean 2
(third round); statistical data on residual sample prepared at ORNL Standard
(wt%)
deviation
moisture
S (wt%)
Res. moist.
Ash
Res. moist.
Ash
1 2 3 4 5 6 7 8 9 10 11 12
4.605 4.373 3.775 4.434 4.135 3.088 5.418 4.298 3.868 4.128 5.328 4.698
18.87 22.64 20.54 22.05 21.71 28.91 22.75 23.94 23.60 20.27 21.44 21.49
0.03 0.18 0.11 0.07 0.09 0.64 0.41 0.22 0.02 0.16 0.37 0.07
0.35 0.66 0.54 0.54 0.07 1.52 1.56 0.64 0.37 0.82 0.39 0.52
Average Std. dev.
4.346 0.643
22.35 2.50
0.64
0.67
that found for the inter-laboratory sample studied in the second round (0.73 wt%, Table 7). The residual moisture, however, is not all that much different in the two inter-laboratory rounds of testing; compare this statistic in Table 11 (0.643 wt%) with the corresponding value in Table 7 (0.518 wt%). BITUMINOUS
COAL
ROUND
ROBIN
The data obtained for HHV, moisture, and ash content determinations for the three rounds of testing were significantly more variable than had been expected. In order to establish some reassurance that a non-RDF reference sample could be analyzed satisfactorily, a Standard Reference Material bituminous coal sample (SRM 1632a from NBS) was distributed to six laboratories by NBS for HHV, residual moisture, and ash determinations. The results of this series of tests are shown in Tables 12 and 13. In Table 12, it is found that the standard deviations of the average of the means for the coal sample are about the same for HHVl, HHVB, and HHV3, i.e., -0.116 MJ/kg. This value is considerably lower than comparable values of the standard deviation of the average of the means found for higher heating values in any of the three rounds of testing. Similarly, values of this statistic for residual moisture and ash content for the coal sample are much lower than comparable values found in earlier rounds of RDF-3 testing; for the residual moisture, the standard deviation of the average of the means is 0.207 wt% and for ash is 0.15 wt% (Table 13). This series of tests on bituminous coal showed that the participating labora. tories could analyze a coal sample to suitable levels of precision.
243 TABLE
12
ASTM round robin testing of RDF-3; statistical sample (SRM 1632a)’ distributed by NBS
values; coal
duplicate
runs (MJ/kg)
HHVl
HHV2
HHV3
HHVl
HHV2
HHV3
1 3 4 5 7 8
25.947
26.109
26.205 26.014 26.098 25.905 25.967
26.602 26.316 26.465 26.247 26.426
33.834 34.043 33.806 33.988 33.734 33.922
0.000 0.056 0.009 0.037 0.035 0.002
0.002 0.133 0.012 0.037 0.019 0.002
0.074 0.167 0.033 0.49 0.028 0.005
Average Std. dev.
26.023 0.112
26.405 0.126
33.887 0.116
0.023
0.035
0.060
Laboratory
Mean Xof
data on higher heating
aSRM = Standard TABLE
Reference
Standard
deviation
S (MJ/kg)
Material.
13
ASTM round robin testing of RDF-3; coal sample (SRM 1632a) distributed Laboratory
statistical by NBS
data on residual
moisture
deviation
and ash;
Mean _? of duplicate
runs (wt%)
Standard
S (wt%)
Res. Moisture
Ash
Res. Moisture
Ash
1 3 4 5 7 8
1.590 1.485 1.145 1.355 1.305 1.720
22.21 21.80 22.16 22.12 22.20 22.10
0.014 0.276 0.007 0.007 0.064 0.042
0.007 0.000 0.042 0.050 0.007 0.007
Average Std. dev.
1.433 0.207
22.10 0.15
0.068
0.019
REPEATABILITY
AND REPRODUCIBILITY
In order to provide some quantitative estimate of where the current stateof-the-art stands with respect to the analysis of RDF-3 samples in contrast to the analysis of coal, a statistical study of the data accumulated in the first three rounds of testing was conducted using a program based upon the theory of nested (hierarchical) designs [6]. In this case, days were nested within laboratories and replicates within days. There were two days of testing for each laboratory, and on each day two replicate determinations were carried out. The data obtained on the NBS bituminous coal sample (SRM 1632a) were
244
analyzed using standard methods [ 71. In this case, each laboratory carried out two replicate determinations, but only on one day. Two statistics were particularly sought, namely, the repeatability interval, Z(r), which is a measure of within-laboratory agreement, and the reproducibility interval, Z(R), which is a measure of between-laboratory agreement. These statistics could easily be compared with existing values listed in documented ASTM Standard Methods of Test for Coal and Coke for HHV, residual moisture and ash. For example, in ASTM D2015 (Standard Test Method for Gross Calorific Value of Solid Fuel by the Adiabatic Bomb Calorimeter) the precision criteria for judging the acceptability of results at the 95% probability level are: Repeatability.
Duplicate results by the same laboratory on different days, using the same operator and equipment should not be considered suspect unless they differ by more than 0.116 MJ/kg, dry basis.
Reproducibility.
Results submitted by two or more laboratories (different equipment, operators, date of test, and different portions of the same gross sample) should not be considered suspect unless two results differ by more than 0.233 MJ/kg, dry basis. Equations required to calculate the repeatability interval, Z(r), are:
Z(r) = 2.83s(r);
(I
where s(r) is the pooled within-laboratory standard deviation and 2.83 is a factor which converts the repeatability standard deviation to a 95% repeatability interval, and s(r) = [s2 (repl) + s2 (days)] li2,
(2
where s(rep1) is the standard deviation of replicate measurements and s (days) is the standard deviation due to variations in measurements observed between days. Similarly, equations needed to calculate the reproducibility interval, Z(R), are: Z(R) = 2.83s(R),
(3)
where s(R) is the between-laboratory fined as above, and
estimate
s(R) = [s2(repl) + s2(days) + s’(labs)]
‘j2,
of precision
and 2.83 is de(4)
where s(labs) is the standard deviation due to variations in measurements observed between laboratories and s(rep1) and s(days) are as defined above. Tables 14-17 summarize the repeatability and reproducibility intervals calculated for HHVB, total moisture, residual moisture, and ash, respectively. Data on all three RDF-3 rounds of testing are tabulated including the results from testing of the bituminous coal sample, SRM 1632a. The first row
245 TABLE 14 Summary of repeatability and reproducibility intervals for higher heating value HHVB, dry basis, in ASTM round robin testing of RDF-3 Method/Test
ASTM D 2015, coal and coke Coal (SRM 1632a) First round robin Second round robin Third round robin Second round interlab comparison sample Third round interlab comparison sample Adopted values
HHV2 (MJ/kg)
ReprodJRepeat., Z(R)lZ(r)
Average
Repeatability interval, Z(r)a
Reproducibility interval, Z( R)a
-
0.116
0.233
2.00
26.405 17.317 18.464 17.424 17.738
0.160 0.661 0.340 0.579 0.554
0.370 1.956 1.091 1.796 0.605
2.30 2.96 3.21 3.10 1.09
17.066
0.688
1.168
1.70
-
0.582
1.745
3.00
‘95% probability.
of Tables 14,16, and 17 gives the acceptability limits for HHVB, residual moisture, and ash content, respectively, according to the appropriate ASTM protocol for coal and coke. No acceptability limits are available for total moisture in coal and coke protocols. Examination of the HHV2 data in Table 14 shows that the repeatability interval for the coal, 0.160 MJ/kg, is close to the acceptable limit of 0.116 MJ/kg. The reproducibility interval obtained for the coal is 0.370 MJ/kg; this departs somewhat from the acceptable limit of 0.233 MJ/kg yielding a ratio of I(R) to I(r) of 2.30 rather than 2.00. The exact statistical significance of the ratio I(R)/l(r) is not clear. A ratio of one (or close to this number) does, however, indicate that the samples supplied to the laboratories are identical and have been analyzed in an equivalent manner. A ratio greater than one indicates that the samples are not identical and/or the analyses of the laboratories are not being done in an equivalent manner. Among the systematic errors which will tend to make this ratio large are different approaches or methods of field-sample subdivision, different procedures for blending (or the absence of blending), different opinions of what constitutes a non-millable portion of the sample, apparatus and person or operator interactions (heat build-up in the Wiley mill used, errors in the calculation of the corrected temperature rise during a calorimetric run, characteristics of a furnace at the ashing temperature, judgement of the technician in weighing a sample, or whether a sample has reached constant weight during drying at 105°C or ashing at 725’C.) In this paper, this ratio is used to help establish a preliminary basis for comparing the variability of RDF-3 with coal. The repeatability interval for the first three rounds of testing of RDF-3 is
246
between three and six times the 0.116 MJ/kg limit. This includes the interlab comparison samples tested in rounds two and three. The values of I(r) for the three rounds of testing give an average value of 0.526 MJ/kg. The corresponding average value for I(R) is 1.614 MJ/kg. It may be observed that the I(R)/l(r) ratio for HHV2 for coal and coke is approximately 2.00, and for RDF-3 is close to 3.00. For practical purposes, RDF-3 repeatability and reproducibility intervals for HHV2 of 0.582 and 1.745 MJ/kg, respectively, may be adopted. The latter values are much larger than those given for coal and coke, but represent the current state-of-the-art in the repeatability and reproducibility of HHV2 for RDF-3.
Fig. 1. RDF-3 reduced to 0.5 mm; magnification 50 times (each square is 0.25 mm on a side).
247 HETEROGENEITY
OF RDF-3
The heterogeneity of RDF-3 is still present at 0.5 mm particle size. Figs. 1 and 2 show analysis samples of 0.5 mm RDF-3 magnified 50 times under a microscope, The particles are heterogeneous even at this level of subdivision. Some are spherical, some are elongated rods, while others are irregularly shaped. Differences in density are suggested by the black and white contrast shown in the photos.
Fig. 2. RDF-3 reduced to 0.5 mm; magnification 50 times (each square is 0.25 mm on a side)
248
Fig. 3. SRM 1632a bituminous coal sample; magnification 50 times (each square is 0.25 mm on a side).
Fig. 3 is the bituminous coal sample used, magnified 50 times under a microscope. The particles are 0.25 mm (-60 mesh) and have a general homo geneous appearance in comparison to the RDF-3 at the same magnification. Differences in the densities of 0.5 mm plastic, cellulose, iron, aluminum, sand, etc. particles in RDF-3 help to defy the blending process and extraction of representative samples for various tests. The values of I(r) and I(R) for the inter-laboratory comparison samples
249
of rounds two and three show that the second round sample gave better results than the third round. The reason for this is not immediately clear, but the I( R)/l(r) ratio does illustrate the benefits of having only one laboratory perform the processing of 2-3 cm nominal RDF-3 down to 0.5 mm. Biases in the reduction of the size of RDF-3 are an inherent part of the reproducibility interval values and cause I(R) to be so much larger than I(r) for the regular rounds of testing. SUMMARY OF RESULTS
The summary of I(r) and I(R) data for total moisture is given in Table 15. A trend is observed between the average of the means and the repeatability interval I(r). Samples increased in total moisture as testing went from the first, to the second round, and to the third rounds; compare 13.82, 18.91, and 30.09 wt% with I(r) values of 0.37, 0.57, and 0.76 wt%, respectively. Intra-laboratory repeatability lies between 0.4 and 0.8 wt% while inter-laboratory reproducibility lies between 3.4 and 5.0 wt%. The ratio of I(R)/l(r) is poorest for the first round, having a value of 9.21. In Table 16 are given I(r) and I(R) values for residual moisture. The limits imposed on coal and coke samples (ASTM D3173) are related to the moisture content of the samples and are given in the table. For example, if less than 5 wt% moisture is present in the coal or coke sample, then I(r) and I(R) values are not considered suspect unless they are above 0.2 and 0.3 wt%, respectively. The data obtained by the six participating laboratories for the standard coal sample fall within the above acceptability limits since the sample contained 1.43 wt% residual moisture. There appears to be a trend in the I(r) values for the three rounds of testing similar to that observed in Table 15 related to the average of the means of the total moisture. Whether there is a relationship or whether this is merely chance is not clear. The averages of the means of the residual moisture do not parallel the I(r) trend. The ratio of I(R)/l(r) ranges from 3.5 to 4.5. TABLE 15 Summary of repeatability and reproducibility of total moisture in ASTM round robin testing of RDF-3 _ Test
First round robin Second round robin Third round robin a95% probability.
Total moisture (wt%) Average
Repeatability interval, Z(r)a
Reproducibility interval, Z(R)a
13.82 18.91 30.09
0.37 0.57 0.76
3.41 3.24 4.99
Reprod./Repeat., Z(R)lZ(r)
9.21 5.68 6.57
250 TABLE
16
Summary of repeatability robin testing of RDF-3
and reproducibility
Method/Test
ASTM D 3173, coal and coke less than 5% moisture more than 5% moisture Coal (SRM 1632a) First round robin Second round robin Third round robin Second round interlab comparison sample Third round interlab comparison sample Adopted values
for residual moisture in ASTM round
Residual moisture (wt%)
Reprod./ Repeat., Z(R)/Z(r)
Average
Repeatability interval, Z(r)a
Reproducibility interval, Z(R)a
1.42 4.11 2.72 4.73
0.2 0.3 0.12 0.41 0.63 0.99
0.3 0.5 0.23 1.44 2.86 4.42
1.50 1.67 1.92 3.51 4.54 4.46
3.02
0.58
1.55
2.67
4.35
0.84
1.93
2.30
-
0.75
3.00
4.00
a95% probability.
TABLE
17
Summary of repeatability and reproducibility robin testing of RDF-3 Method/test
Ash (dry basis) (wt%) Average
ASTM D 3174. coal and coke no carbonates carbonates more than 12% ash 22.10 Coal (SRM 1632a) First round robin 22.16 14.17 Second round robin Third round robin 22.17 Second round interlab 19.76 comparison sample Third round interlab 22.35 comparison sample Adopted values a95% probability.
intervals for ash content in ASTM round
_
Reprod./ Repeat., Z(R)/Z(r)
Repeatability interval, Z(r)a
Reproducibility interval, Z(R)a
0.2 0.3 0.5 0.08 2.73 1.31 1.88
0.3 0.5 1.0 0.16 5.20 4.26 7.57
1.50 1.67 2.00 2.10 1.90 3.25 4.03
3.09
3.16
1.02
2.54
7.35
2.89
2.5
7.5
3.00
251
Again, for practical considerations, the ratio 4.00 with 0.75 and 3.00 wt% as values for I(r) and I(R), respectively, can be adopted. No limits on the amount of residual moisture in the sample have been set as with coal and coke. Table 17 is the summary of values of I(r) and I(R) for the ash content data. The limits for acceptable data with regard to coal and coke samples (ASTM D3174) which contain (or do not contain) carbonates or more than 12 wt% moisture are given in the first three rows of Table 17. The results of testing the standard coal sample show that the participating laboratories have determined that the ash content is well within the levels of acceptability. Values of 0.08 and 0.16 wt% for I(r) and I(R), respectively, are quite satisfactory. The repeatability and reproducibility intervals for the first three rounds are quite high in comparison with the coal and coke acceptability limits. The same is true for the I(r) and I(R) values obtained for the interlaboratory comparison sample. Although the second round data for I(r) and I(R) are better for both the regular and inter-laboratory samples, the absolute values of the average of the means are significantly different; compare 14.17 wt% for the regular RDF-3 sample with 19.76 wt% for the inter-labora. tory sample. This difference may be due to overall heterogeneity of RDF-3. Other biases may be involved but are not apparent. The suggested values for I(r) and I(R) are 2.5 and 7.5 wt%, respectively. These values are based on an adopted I(R)/l(r) ratio of 3.00. This is much higher than expected, but this is what is observed. ACKNOWLEDGEMENTS
Funding for this work was provided by the NBS Office of Recycled Materials and the U.S. Department of Energy Office of Energy from Municipal Waste. The cooperative efforts of the membership of ASTM Committee E-38 on Resource Recovery are acknowledged, and, in particular, those of ASTM Subcommittee E-38.01 on Energy. Special thanks are expressed to Dr. John Love, Jr., Chairman of ASTM Subcommittee E-38.01, for his strong leadership and perseverance during the RDF-3 round robin testing program, and to the participating laboratories for their responsiveness and cooperation.
REFERENCES 1 ASTM Method E711-81, Standard test method for gross calorific value of refuse-derived fuel (RDF-3) by the bomb calorimeter. 2 ASTM Method E790-81, Standard test method for residual moisture in a refuse-derived fuel analysis sample. 3 ASTM Method E791-81, Standard method for calculating refuse-derived fuel analysis data from as-determined to different bases. 4 ASTM Method E830-81, Standard method of test for ash in the analysis sample of refuse-derived fuel (RDF-3).
252 5 ASTM Method E856-81, Standard definition of terms relating to the physical and chemical characteristics of refuse-derived fuel. 6 Davies, O.T., Statistical Methods in Research and Production, Third Edition, published for Imperial Chemical Industries, Ltd., Oliver and Boyd, London and Edinburgh; Hafner Publishing Co., New York, 1957. 7 ASTM Method E691-79, Standard practice for conducting an interlaboratory test program to determine the precision of test methods.