Document jBma9OJdvDR510B8Y2Kz2pJdy

October 21, 1975 To: V J DelBrooco, Plaquemlne R A Dove, Akron Research 0 LaBrecque, North Chicago R F Mitchell, Niagara Falls i/ Results of the most recent (9^12-75) collaborative tests on the vinyl chloride charcoal tubeB are presented in Table I below* This series of tabes, VC-C-1 thru VC-C-4, contained duplicate samples at the 111.5 microgram and 37*2 microgram level of vinyl chloride. A statistical evaluation of the results is also included to show the accuracy and precision of each Plant* Table I Interplant Vinyl Chloride Results ____ Staqflads_ Niagara Falls North Chicago Plaquemlne Akron Research VC-C-1 VC-C-2 VC-C-A.VC~C-3 99.0 94.0 128*0 122.5 118.0 79.0 111.1 113.5 34*0 32.0 42.5 42,0 25.0 28.0 37.5 37.5 Direct comparison of the results reveals that Niagara Falls and Plaquemlne were generally on the low side of the standards, whereas North Chicago was on the high aide. Akron Research was very close to the actual standards. From these results the percent relative accuracy (percent difference from standard) was calculated and is shown in Table II, Uxarapla: 99.0 - 111.5 x 100 = -11*2$ 111.5 60037152.5 2 Table II Percent Relative Accuracy VC-C-1 VC-C-4 VC-C-2 VC-C-3 Average Niagara Falls North Chicago Plaquemine -11.2 -15.7 14.8 9.9 5.8 -29.1 -8.6 -14.0 14.2 12.9 -32.8 -24.7 -12.4 13.0 -20.2 Akron Research 1.8 0.8 0.8 0.75 The overall accuracy achieved by Akron Research, is outstanding. To analyze the precision within each plant, the range of duplicate samples must be determined from the percentage data in Table II. Example: North Chicago VC-C-1 VC-C-4 14.8$ 9.9$ 14.8 - 9.9 gives a range of 4.9. The smaller the value of the range of duplicates, the greater is the precision vithin a particular lab. After the range of duplicates is daterained for each plant, an average range value for ell the plants, R, ia obtained and an upper control limit calculated. Any range value that exceeds the upper control limit (2tT) is considered to be an outlier, and is omitted from subsequent calculations. Table III gives the values for the ranges of duplicates. Table III Ranees of Duplicates Niagara Falls North Chicago Plaquemine Akron Research VC-C-1 VC-C-4 4.5 4.9 34.9 2.2 Average range, H = 7.66 VC-C-2 VC-C-3 5.4 1.3 8.1 0 60037152.6 3 To calculate a 2.^(95$ confidence) upper control limit for 8 subgroups with 2 observations per subgroup, the following formula ia used: ucl2 = I + 2/3 /"(d 4 I) - Rj = 7.66 + 2/3 /"(4.S7 7.66) - 7.66J7 = 7.66 + 2/3 /"37.30 - 7.66J7 = 7.66 + 2/3 fW.Uj = 7,66 + 19,76 * 27,42 Taking 27.4 as the 2<J" upper control limit, it can be seen that Plaquemine1 s range of duplicates (34.9), exceeds the upper control limit,. By analyzing Plaquemine9 s results, 79 ug can be considered an outlier and omitted from further calculations. 'Plaquemine should review their analytical procedure to determine possible causes for the discrepancy in results at the 111.5 microgrem level. The fact that the ranges of duplicates for the other plants do not significantly approach the upper control limit demonstrates good precision. While the ranges of duplicates gives an indication of the precision each lab (intralab), the standard deviation (0> which is the deviation from the mean, indicates precision among the labs (lntorlab). Table IV gives the standard deviation for the 111.5 microgrnm level. V X* 99.0 94.0 128.0 Table IV Interplant Data For 111.5 Microcram Level yyX (x - X) (x - (X -. X)2 112.3 13.3 176.89 897.45 18.3 334.89 15.7 246.49 n cr 7* 12.23 122,5 118.0 10.2 5.7 104.04 32.49 111.1 1.1 1.21 113.5 1.2 1.44 * Plaquemine outlier of 79 omitted from calculation of"",, 60037152.7 4 For charcoal tubes containing 111.5 micrograms, irtezplar.t analysis found a mean value of 112,3 micrograas uith a standard deviation cf 12.2 miorograjns. Tha % relative standard deviation is 10.9$. X 34.0 32.0 42.5 42.0 25.0 28.0 37.5 37.5 Table V Interplant Data For 37.2 Microgreat Level 1 (X - X) (X - X)2 lt(x - X)2 34.8 0.8 0.64 276.47 2.8 7.84 7.7 59.29 7.2 51.84 9.8 96.04 6.8 46.24 2.7 7.29 2.7 7.29 n 8 cr 6.28 For charcoal tubes containing 37.2 microgram of vinyl chloride, interplant analysis found a mean value cf 34.8 raicrograms with a standard deviation of 6,,3 microgramso The % relative standard deviation is 18.1$. The smaller the relative standard deviation, the greater is the precision among labs. The $ reletivc standard deviation values of If'. 9$ at the 111.5 microgram level and 18 f t the 37.2 microgram level indicate that the precision among labs is not as gtoi as it should be and further collaborative tests will be conducted. The isuits reported by Akron Research pj/ove that excellent precision and accuracy err- obtainable. The help cf h.A. Bedford, Division Manager, Qualify Control, Chemicals, with tin statistical i ark involved in this report is greatly appreciated. Paul W Bankovich cr cc: C A Johnson j, M, D. B W Modrell D H Francis W L Cox L A Bedford G L Hess yCorporate^/Industrial Hygiene 60037152.8