Sie sind auf Seite 1von 50

Summary Report on the ASTM E-691-99 Interlaboratory Study to Determine the Precision

and Bias of the Coefficient of Friction Test Method Employed by the UWTLP BOT-3000.



The aim of this study is to evaluate the precision and bias of the static and dynamic coefficient of
friction measuring method employed by the UWTLP BOT-3000. This study was carefully planned and a
well-written test method and instructions were provided to the participating laboratories.

The Interlaboratory Study (ILS) included 8 laboratories. Each laboratory tested 4 UWTLP BOT-
3000 devices (hereafter called devices.) The dynamic and static coefficient of friction of each surface was
obtained with 4 replicated measurements. Each laboratory provided 64 data points, for a total of 512
determinations.

Tests were conducted to determine the static and dynamic coefficient of friction of two surfaces:

Formicasurface
Teflontape surface

The Formica surface was tested using a leather sensor. The Teflontape surface was tested using a
Neolitesensor. Formica and Teflon test surfaces were chosen due to their surface stability characteristics,
reducing the influence of test surface variability.

Four main factors may contribute to the variability between test results: (a) operator, (b) equipment
used, (c) calibration of equipment, and (d) environment. The relationship of the test results to the accepted
reference value defines its accuracy. Accuracy is expressed in terms of precision and bias.

The precision of the test method is evaluated in terms of 2 measurement concepts, repeatability and
reproducibility. Repeatability concerns the variability between independent test results obtained within a
single laboratory. The tests performed in each laboratory were conducted by a single operator. The test
procedure was done in a minimum elapsed time in order to maintain the conditions (a) through (d) as
constant as possible and to reduce the variability. Reproducibility refers to the variability between single
test results obtained in different laboratories. It helps determine if any bias is introduced by the (a) through
(d) factors. The reproducibility measure reflects what precision might be expected when different devices
are sent to random laboratories.

To evaluate the data consistency 2 statistics are used: the k-value to examine the consistency of the
within-laboratory results, and the h-value to examine the consistency from laboratory to laboratory.

After preparing separate calculation worksheets for each device, the cell average for each laboratory
( x ), the cell standard deviation (s), the average of the cell averages from all the laboratories ( x *), the cell
deviation for each laboratory (d), and the standard deviation of the cell averages (
x
s
*
) were calculated.
The calculated values of h and k were also included in the worksheets as well as the repeatability standard
deviation (s
r
) and the reproducibility standard deviation (S
R
).



Page 1 of 50 September 12, 2005

Data Consistency

The values of h and k were calculated for each laboratory and for each device, providing a total of 32
values of h and 32 of k for each test. The h and k values were plotted by laboratories and by devices. The
plot of h and k by laboratories did not show any bias. The plot of h and k by device also did not show any
bias. All devices have both positive and negative h values among the laboratories.

Critical Values of the Consistency Statistic:

For 8 laboratories and 4 determinations, the critical values of h and k at the 0.5% significance level
are: h =2.15 and K =1.9 (as defined by ASTM E691-99 Table 5, Critical Values of h and k at the 0.5%
Significance Level.) The h and k values exceeding the corresponding critical values for the different tests
are as follows:

- Leather Dynamic Coefficient of Friction (DCOF Leather): only 1 value of h and 3 values of k
exceeded the 0.5% significance level. The value of h is for Laboratory 1, device serial number
1062 while the k exceeding values were for Laboratory 2, device serial number 1014 and
Laboratory 6, device serial numbers 1033 and 1062.

- Leather Static Coefficient of Friction (SCOF Leather): all h values were within the 0.5%
significance level and only 1 k value was flagged corresponding to Laboratory 6, device serial
number 1026.

- NeoliteDynamic Coefficient of Friction (DCOF Neolite): all h values were within the 0.5%
significance level and only three k values were not. They were Laboratory 1, device serial
number 1033 and 1062 and Laboratory 4, device serial number 1044.

- NeoliteStatic Coeffficient of Friction (SCOF Neolite): all h values were within the 0.5%
significance level and only one k value was not. That was Laboratory 3, device serial number
1026.

The consistency of the test result is demonstrated by the fact that only 1 h value out of a total of 128
and only 8 k values out of a total of 128 were out of the target range. These out of range values are
uniformly distributed between the devices corresponding to 2 k values for each device.

Device Average Plot

The individual determinations of the coefficient of friction have been plotted together with their
average per device. These 4 plots show that the 4 devices are not biased. Furthermore, the average value of
every device is very similar and located well within the data distribution for the other devices.

Range Plots

The plots of the repeatability and reproducibility values, versus the device average, graphically
confirm a high level of statistical consistency.




Page 2 of 50 September 12, 2005

Precision Statement

The analysis of the test result data used statistical tools to evaluate the consistency of the data
pertaining to both within-laboratory repeatability and between-laboratory reproducibility.

The precision is evaluated through the repeatability standard deviation (Sr), the reproducibility
standard deviation (SR), and the 95% repeatability (r) and reproducibility (R) probability limits. These
values are presented in the 4 precision statements corresponding to DCOF Leather, SCOF Leather, DCOF
Neoliteand SCOF Neolite.

In conclusion, this study shows that the determination of the coefficient of friction using the UWTLP
BOT-3000, when following the corresponding test procedure is consistent with the 0.5% significance level
as defined by the ASTM E691-99 and did not show any bias due to operators, devices, or laboratory
environment.

Statistical analysis completed by Dr. Eugenio Tacconi, an Independent Statistician for this study.

















Notes:
1. Formicais a registered trademark of the Formica Corporation. Teflonis a registered trademark
of DuPont De Nemours and Company. Neoliteis a registered trademark of the Goodyear Tire and
Rubber Company.
Page 3 of 50 September 12, 2005
Page 4 of 50 September 12, 2005
Page 5 of 50 September 12, 2005
Page 6 of 50 September 12, 2005
Page 7 of 50 September 12, 2005
Page 8 of 50 September 12, 2005
Page 9 of 50 September 12, 2005
Page 10 of 50 September 12, 2005
Page 11 of 50 September 12, 2005
Page 12 of 50 September 12, 2005
Page 13 of 50 September 12, 2005
Page 14 of 50 September 12, 2005
Page 15 of 50 September 12, 2005
Page 16 of 50 September 12, 2005
Page 17 of 50 September 12, 2005
Page 18 of 50 September 12, 2005
Page 19 of 50 September 12, 2005
Page 20 of 50 September 12, 2005
Page 21 of 50 September 12, 2005
Page 22 of 50 September 12, 2005
Page 23 of 50 September 12, 2005
Page 24 of 50 September 12, 2005
Page 25 of 50 September 12, 2005
Page 26 of 50 September 12, 2005
Page 27 of 50 September 12, 2005
Page 28 of 50 September 12, 2005
Page 29 of 50 September 12, 2005
Page 30 of 50 September 12, 2005
Page 31 of 50 September 12, 2005
Page 32 of 50 September 12, 2005
Page 33 of 50 September 12, 2005
Page 34 of 50 September 12, 2005
Page 35 of 50 September 12, 2005
Page 36 of 50 September 12, 2005
Page 37 of 50 September 12, 2005
Page 38 of 50 September 12, 2005
Page 39 of 50 September 12, 2005
Page 40 of 50 September 12, 2005
Page 41 of 50 September 12, 2005
Page 42 of 50 September 12, 2005
Page 43 of 50 September 12, 2005
Page 44 of 50 September 12, 2005
Page 45 of 50 September 12, 2005
Page 46 of 50 September 12, 2005
Page 47 of 50 September 12, 2005
Page 48 of 50 September 12, 2005
Page 49 of 50 September 12, 2005
Page 50 of 50 September 12, 2005

Das könnte Ihnen auch gefallen