See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/288386911

Statistical testing of the measurement accuracy of mobile roughness

measuring instruments

Article · January 2012

CITATIONS READS

3 330

1 author:

Krzysztof Stepien
Politechnika Świętokrzyska
68 PUBLICATIONS   503 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Theoretical and experimental problems of integrated 3D measurements of elements' surfaces View project

All content following this page was uploaded by Krzysztof Stepien on 15 November 2018.

The user has requested enhancement of the downloaded file.

 XX IMEKO World Congress
Metrology for Green Growth
September 914, 2012, Busan, Republic of Korea

STATISTICAL TESTING OF THE MEASUREMENT ACCURACY OF MOBILE

ROUGHNESS MEASURING INSTRUMENTS

K. Stępień

Kielce University of Technology, Poland, [email protected]

Abstract: Surface roughness is a very important factor larger than their depth. Surface roughness is usually
influencing overall quality of products of modern industry. related to tool paths, for example, turning, grinding
There are numerous methods of measurements of surface or finishing cuts [43].
roughness. A very important group of roughness  waviness, which is observed when the distance
measurements instruments are mobile ones. They are between irregularities is about 100  1000 times
particularly useful in measurements under industrial larger than their depth; in some sources the lower
conditions. However, precision and accuracy of results limit is 40 or 50. Waviness is usually a result of a
obtained with the use of mobile instruments can depend on disturbed manufacturing process [46], an example
an operator. Therefore, the question is how accurate are of which can be vibrations between the workpiece
measurements carried out with mobile instruments. The and the grinding wheel.
paper describes problems relating to measurements of  form deviations are assumed to occur when the
roughness with mobile instruments. It also presents results distance between irregularities is equal or more
of statistical testing of the precision and accuracy of a than 1000:1. Shape deviations are caused, for
typical portable instrument. example, by guideway errors, errors of rotary
elements, or by thermal expansion of elements.
Keywords: roughness, precision, measurement, mobile It should be noted that cracks may also be considered a
instrument. type of irregularity, for which the ratio between the distance
and the depth is less than or equal to 5.
1. INTRODUCTION Surface roughness influences very significantly
functional properties of machine parts [3]. Therefore, its
When designed, a machine element is assumed to be measurement and evaluation is a very important field of
ideal in shape and texture. Thus, the geometry and modern metrology. Some researchers investigate influence
dimensions of a workpiece need to be exactly the same as of stylus tip and other factors on measurement results [4, 5].
those specified in the design [1]. The surface too must be Zawada-Tomkiewicz in [6] evaluates surface roughness
ideally smooth. However, numerous factors cause that the parameters on the basis of the image of the machined
geometrical surface structure of a workpiece is characterized surface. Grzesik in [7] investigates influence of the tool
by certain irregularities. These irregularities lead to wear on roughness of surfaces turned with differently
differences between the real and the nominal workpiece. shaped ceramic tools. Zhengkai Zhang et al in [8] propose a
Figure 1 shows the irregularity types. new approach to the analysis of surface topography that is
based on empirical decomposition of the profile.
Quinsat and Tournier in [9] describe a novel non-contact
system allowing in-situ measurements of roughness of
elements that due to their dimensions and mass cannot be
measured at laboratory. Nowadays, due to development of
3-dimensional analysis of geometrical surface structure
some researchers undertake efforts aiming at digitization of
measurement results [10] and at the definition of parameters
well describing 3-dimensional surface topography [11]
Generally, methods of measurements of surface
roughness may be divided into two main groups: contact and
Fig. 1. Types of surface irregularities [2] non-contact ones. Mobile roughness measuring instruments
are usually contact ones. They are compact, robust and easy
They include: to apply. The main advantage of this type of instruments is
 surface roughness, which occurs when the distance that they allow in-situ measurements under industrial
between irregularities is about 5  150 times conditions and the main drawback is that they are not
isolated from harmful environmental conditions. One of the
possible sources of the errors in measurements with use of e2  Ra 2  Ratrue (5)
mobile roughness measuring instruments are human errors.
In this paper are discussed problems of statistical testing of where Ratrue is the real value of parameter Ra of tested
the precision and accuracy of typical mobile roughness specimen.
measuring instruments. For further consideration this value of the error is taken,
whose absolute value is higher.
2. PROCEDURE 7. To evaluate accuracy of the instrument from the
equation:
The instrument, which was used in the experiments is the e U
 100% (6)
typical mobile roughness measuring instrument, equipped Wmax max
with a inductive probe-head and the tip with a skid. The
specimen to be measured was a roughness standard with where e is given by equations (4) and (5), U is given by the
precisely defined parameter Ra. equation (2) and Wmax is the measuring range of the
The plan of the experiment was following: instrument [12].
1. To carry out two series of 50 roughness measurements
of the specimen, whose roughness parameters are 3. EXPERIMENT
known. The parameter to be investigated is Ra and both
measurement series should be conducted under The experiment involved measurements of an irregular
different measurement conditions. roughness standard produced by Hommelwerke Company.
2. To calculate mean values, standard deviations and The value of the parameter Ra for the standard is equal to
ranges for results of both measurement series. After 0,67 μm. The photograph of the standard is shown in Fig. 2.
calculations we obtain following parameters:
- mean value of the parameter Ra from the first
series of measurements,
- mean value of the parameter Ra from the second
series of measurements,
s1 – standard deviation of the measurement results in
the first series of measurements,
s2 – standard deviation of the measurement results in
the second series of measurements, Fig. 2. Roughness standard used in the experiment
R1 – the range of the measurement results in the first
series of measurements, Measuring instrument used in the experiment was a portable
R2 – the range of the measurement results in the second roughness measuring instrument Hommel Tester T1000 by
series of measurements. Hommelwerke. Hommel Tester T1000 is a mobile device
3. To compare mean values and by the statistical allowing performing in-situ roughness measurements. The
test and evaluate if the difference between them is instrument is shown in Fig. 3.
statistically significant [12]. The test will be conducted
with use of the following equation:
Ra1  Ra 2
t (1)
s12  s2 2
n
where n is the number of measurements in each series.
(in the experiment n=50). Assumed level of significance for
the test is 0.01.
4. To evaluate repeatability of measurement results. As a
measure of the repeatability to take value
(2)
5. To evaluate expanded uncertainty of measurements
from the equation
U  us (3)
where U is the expanded measurement uncertainty, u is Fig. 3. Roughness measuring instrument used in the
the coverage factor and s is the standard deviation experiment – Hommel Tester T1000 by Homelwerke
given by the equation (1). The probability level
considered in the experiment is P = 0.95. In the table 1 there are given basic technical specification of
6. To evaluate measurement bias in both measurement the instrument Hommel Tester T1000.
series from equations:
e1  Ra1  Ratrue (4)
Table 1. Basic technical specifications of the Hommel
Tester T1000
Measurement type Skid measurement
Filter Gauss digital filter
Traverse length lt [mm] 0.48/1.5/4.8/15; max. 16
Sampling length ln [mm] 1.25/4.0/12.5
Cut off λ [mm] 0.08/0.25/0.8/2.5/8.0
Measuring range 80 μm
Diamond stylus tip 5 μm 90°
Skid radius lengthways 30mm/tranversal 1.9
mm
There were two series of 50 measurements carried out under
reproducibility condition of measurements. In the described Fig. 6. Histogram of the results obtained in the first
experiment both series of measurements performed by measurement series
different operators on different days. The first operator was
not skilled at performing measurements in order to establish
if operator’s experience can influence results. Both series
were carried out with the use of the same traverse and cut
off length (traverse length lt = 4.8 mm and cut off length λ
= 0.8 mm). In the experiment the value of the roughness
parameter Ra.

4. RESULTS AND DISCUSSION

Measurement results obtained in the experiments are shown

in Fig. 4 and Fig. 5.
Ra m Fig. 7. Histogram of the results obtained in the second
measurement series
0.76
Standard deviations s in both series of measurements are as
0.74 follows:
 for the first series: s1  0.0344 μm,
0.72
 for the second series s2  0.0338 μm.
0.70 Ranges of measurement results R in both series of
measurements are:
Meas .number
10 20 30 40 50  for the first series: R1  0.1 μm,
Fig. 4. Ra values obtained in the first measurement series  for the second series R2  0.13 μm.
In order to establish if difference of Ra values obtained in
Ra m
two measurements series is significant a statistical test was
0.74
conducted with the use of the equation (1). The value t
0.72 calculated from the equation (3) is t  0.0293 whereas the
0.70 critical value tα taken from a statistical table is equal to
0.68 t  1.96 . Since t  t the difference between the values
0.66 Ra1 and Ra 2 is statistically negligible.
The next step of the analysis of obtained results was an
0.64
evaluation of the measurement repeatability. As a measure
Meas .number of the repeatability a value of a standard deviation was
10 20 30 40 50
taken. According to the procedure described in the previous
Fig. 5. Ra values obtained in the second measurement series
section the measurement repeatability s is equal to the
standard deviation of results obtained in the first
Mean values of the parameter Ra from obtained in both
measurements series, i.e. s  0.0344 μm.
series of measurements are following:
The value of the measurement repeatability was then used
 for the first series: Ra1  0.695 μm,
to calculate an expanded uncertainty of measurements
 for the second series Ra 2  0.6948 μm. according to the equation (3). The expanded uncertainty U
In order to present a distribution of the results histograms for the probability level P  0.95 (for which a coverage
were generated that are shown in Fig. 6 and Fig. 7. factor u  1.96 ) is equal to U  0.0674 μm.
Relatively large number of measurements allowed to inexperienced operator can perform accurate roughness
evaluate measurement bias. Assuming that mean value of measurements with the use of the investigated instrument.
the parameter Ra from both series is equal to 0.695 μm the An analysis of measurement results showed that a slight
measurement bias e calculated with the use of the equations systematic error occurs. Its measure, the bias, was equal to
(4) and (5) is: e  0.025 μm. about 3% of the true value. In order to investigate the if the
Finally, the accuracy class δ of the instrument was value of this error is constant or it can change, the next
evaluated with the use of the equation (6). It was calculated experiment involving measurements of different standards
taking into account values of the instrument repeatability should be carried out. In the next experiment the accuracy of
and the measurement bias. Assuming probability level P = the instrument can be also evaluated with the use of other
procedures, for example, those described in [13] and [14].
0.95 the accuracy class of the instrument Hommel Tester
Generally, obtained results show that the instrument is very
T1000 is equal to   0.12%
accurate. The calculated value accuracy class
An analysis of experimental data shows that there are not proves that the investigated instrument can be even regarded
any significant differences between results obtained in two as a reference one in the laboratory. Considering the fact the
measurements series. The mean values of Ra in both series instrument is dedicated mainly to the in-situ measurements
are very similar. This observation has been confirmed by under industrial conditions it is a very good result.
the result of the statistical test. The repeatability error and
the range in both series are quite similar, too. 5. REFERENCES
The analysis showed that a slight systematic error may
occur. Its measure – the bias – is equal to +0.025 μm. The [1] Humienny Z. et al: Geometrical Product Specifications–
bias was similar in both measurement series. However, it course for Technical Universities, Warsaw University of
should be noted that its value is relatively low (it is a little Technology Publishings, Warsaw, 2001.
less than 3 %). The value of measurement uncertainty [2] Wieczorowski M et al.: Characteristics of surface roughness
- guide. Poznań University of Technology, Poznań, 1996.
reached about 0.07 μm and it is relatively low, too.
[3] Mathia T.G. et al.: “Recent trends in surface metrology”,
Obtained results show that the accuracy of the instrument is Wear, vol. 271, no. 3-4, , pp. 494-508, 2011.
quite high. This observation has been confirmed by the [4] Demircioglu P., Durakbasa M.N.: “Investigations on
evaluation of the accuracy class of the applied Hommel machined metal surfaces through the stylus type and optical
Tester T1000. Calculated value allows to 3D instruments and their mathematical modeling with the
include this instrument to the laboratory devices and it can help of statistical techniques”, Measurement, vol. 44, no. 4,
pp. 611-619, 2011.
be considered as a reference one.
[5] Durakbasa M.N. et al.: “The factors affecting surface
roughness measurements of the machined flat and spherical
6. SUMMARY surface structures – The geometry and the precision of the
surface”, Measurement, vol. 44, no. 10, pp. 1986-1999,
It is estimated that about 90 % of all failures of machine 2011.
parts are initiated due to surface damages such as fatigue or [6] Zawada-Tomkiewicz A: “Estimation of surface roughness
stress corrosion cracking, wear, erosion, etc. Therefore parameter based on machined surface image”, Metrology
measurement of surface roughness is a very important and Measurement Systems, vol. 17, no. 3, pp. 493-504,
2010.
branch of metrology of geometrical quantities. Surface
[7] Grzesik W.: “Influence of tool wear on surface roughness in
roughness measurements may be performed with the use of hard turning using differently shaped ceramic tools”, Wear,
numerous methods. These methods may be generally vol. 265, no. 3-4, pp. 327-335, 2008.
divided into two main groups: contacting techniques and [8] Zhengkai Zhang, Youyun Zhang, Yongsheng Zhu: “A new
non-contacting ones. In the group of mobile devices that approach to analysis of surface topography”, Precision
allow performing in-situ measurements contacting methods Engineering, vol. 34, no. 4, pp. 807-810, 2010.
are dominant. In the paper a contact roughness measuring [9] Yann Quinsat, Christophe Tournier: “In situ non-contact
instrument Hommel Tester T1000 by Hommelwerke was measurements of surface roughness”, Precision Engineering,
investigated. The instrument is equipped with a inductive vol. 36, no. 1, pp. 97-103, 2012.
[10] Pawlus P.: “Digitisation of surface topography measurement
measuring head and a skid. It is mobile, compact and easy to
results”, Measurement, vol. 40, no. 6, pp. 672-686, 2007.
apply, but the question was it measurement results obtained [11] Pfeifer T. et al.: “Derivation of parameters of global form
with the use of this instrument are reliable and comparable. deviations for 3-dimensional surfaces in actual manufacturig
Therefore the experiment described in the section 3 was processes”, Measurement, vol. 29, no. 3, pp. 179-200, 2001.
conducted. It involved performing two series of [12] Adamczak S., Makieła W.: “Fundamentals of metrology and
measurements of a roughness standard. There were fifty quality assurance for engineers” (in Polish), Wydawnictwa
measurements in each series and both series were conducted Naukowo-Techniczne, Warsaw, 2010.
by different operators at different days. The measuring head [13] Adamczak S., Janusiewicz A., Makieła W., Stępień K.:
was hand-held during the measurement. One of the “Statistical validation of the method for measuring radius
variations of components on the machine tool”, Metrology
operators was a novice and one an expert in roughness
and Measurement Systems, Vol. 18 no. 1, pp. 35−46, 2011.
measurements in order to establish if operator’s experience [14] Janusiewicz A., Adamczak S., Makieła W., Stępień K.:
influences significantly measurement results. It turned out Determining the theoretical method error during an on-
that there are not any significant differences between the machine roundness measurement, Measurement, vol. 44/9
results obtained by the operators. It proves that even (2011), pp. 1761-1767.

View publication stats