Identification of Special Causes.
Bicova, Katerina ; Bebr, Lukas
Identification of Special Causes.
1. Introduction
In today's industry, especially in the automotive industry, it
is very important to monitor and follow up primary key production
processes. For this purpose a very useful tool is the Statistical
Process Control (SPC). It can be said, that SPC is a simple and
effective instrument for understanding the process. For evaluation of
the devised control charts determining criteria are valid, based on
which we can ascertain whether the process is stable or unstable. The
aim of this paper is to focus attention on the case where the process is
unstable and where it is necessary to identify the causes of
fluctuations beyond the established limits or the occurrence of other
undesirable trends. The identification of special causes is essential
for process control and further process monitoring and evaluation of the
process. Specific or definable causes relate to sources of fluctuation,
that aren't permanently affecting the process and producing
unexpected changes. These can be either harmful or beneficial.
Detecting the presence of special causes is the role of SPC.
Elimination of special causes is then done in the form of:
--local actions (operating staff, operator within the scope of
their authority)
--action within the system (these fall under the responsibility of
the management).
The specific cause of variability needs to be identified,
eliminated and safeguarded, so that it can no longer be repeated.
[1].
As has been said, progressive SPC methods are used for detailed
analysis of the production process. This ensures the identification of
adverse effects and their correlation with the behaviour of the
production process. It is a tool that can be used to monitor the
long-term stability of the given system and to ascertain whether the
monitored process behaves as expected. It is possible to assess whether
the variability of the parameter is due to random fluctuations, or if it
is a special cause. Use is possible everywhere, where the parameter is
progressively monitored in time [2,3]
Control charts help by measuring to explain the results from the
view of the process dispersion (variability) and also location (mean
value of the process). Therefore the control charts are always presented
in pairs. The most common pair of charts are (X, R) charts. These are
the values of the sampling mean of the values obtained from the
subgroups, which is the process position measure, and the range of
values in each subgroup, which is the degree of process dispersion.
These diagrams are suitable tools for measuring process
fluctuations the span can be easily calculated and is substantial for
small ranges of subgroups. Graphic presentation is desirable in the SPC
for its clearness and readability. [4]
2. Example--a specific solution
For a specific example of using SPC and evaluating the
manufacturing process data from serial production of a simple machined
part of a cylindrical shape are used here. The dimensions of the
component are controlled by the innovative measurement technology system
Equator. This innovative measurement system allows for the required 100%
control, because it is a very precise and used for manufactured
components marked in the automotive industry as critical. [9] The basic
principle of the measurement process is based on the comparison of the
measured part and the etalon. Upon customer agreement the diameter of
the component has been determined as the critical feature.
From each day the relevant data are collected from machined parts
inspections. In terms of cost-effectiveness and SPC effectiveness, it is
important to correctly determine the extent of the logical subgroup and
the length of the inspection interval. [5]
The measured data from one shift from 6:45 am to 2:15 pm are
divided into subgroups after 15 minutes. Every 15 minutes 10 samples are
taken. First, the basic statistical characteristics required for the
construction of the control charts for the selected shift respectively
given day are stipulated.
After the basic statistical characteristics are determined, control
charts are construction. Figure 3 shows an example charts for the
average value (mean) and figure 4 for the span.
3. Solution--Evaluation of the control charts
For regulation charts other decisive criteria also apply by which
we can determine, whether the process is stable or unstable. In order to
determine the stability of the process, i.e assessing the need to
intervene in the process, e.g. the so-called non-random group tests or
special groupings of points or the rules for determining special cases
in the charts for the average from ISO 8258: 1998 [6, 7] are used. For
example follow figure 3.
Based on these sources, the built-in charts were evaluated and the
following occurred (figure 4):
1. Points exist outside the regulation limits, namely the 4th
subgroup under the UCL
2. Points exist outside the regulation limits, namely the 5th
subgroup above the UCL.
These phenomena can indicate that the process has changed at this
point, the measuring system has changed, the control threshold is
incorrectly calculated, or the point is poorly plotted in the graph etc.
On the basis of signaling the instability of the process, its
causes were identified. One of the basic tools of quality management,
namely the Ishikawa diagram (figure 5), was used for this purpose. The
most probable causes, such as the used methods, the machine-tool, the
cutting tools and the operators present in the production process, and
others were analyzed.
As a result of this analysis, it was found that the cause of
process fluctuation, i.e. points beyond the boundaries, was the lack of
manufacturing precision of the machining process with respect to the
influence of the human factor.
So this is a local cause where the operator made an unintentional
error in changing of the cutting tool, caused by his inattention. The
cutting tool has a prescribed replacement frequency in the working
control procedure always after a specified number of machined pieces.
This change was not performed by the operator and therefore a deviation
was caused. The change took place later. This deviation was reflected in
the accuracy of several manufactured pieces and is also evident in the
control charts.
On the basis of the analysis and the identified cause, a corrective
action was taken, namely re-training of the attending staff at the
machine. If the cause were to repeat, it was proposed to further take
another precaution in the form of setting the software of the machine
so, that after a certain number of starts of the machining process the
tool replacement is requested automatically from the operator.
Manufactured parts with insufficient manufacturing precision were
blocked and therefore removed from the sample data set and the control
charts were revised. In this case, the process was already stability and
capable.
4. Conclusion
Please improve the conclusion. The conclusion must give clear
answers on questions: what was the problem, how was it solved, what are
the results/solutions, what is solved, what are future plans.
Human beings have a very important and indivisible share in the
entire production process. Quality work in this area is very important.
However, it is important to realize that high-quality work is not
natural, but it is consistent and good-quality preparation, regular
training and responsibility of each individual. In the future, the
research team would like to concentrate on monitoring the manufacturing
accuracy of the machining process with respect to the influence of the
human factor. For example, shift work is common in many professions that
directly affect the health and safety of others. The quality of life,
health, safety and protection at work on shifts and commuting can affect
workers in any field and thus their resulting work, i.e. total
manufacturing process. [8]
This paper presented a way of evaluating the production process by
using data from serial production, where was checked the critical
character determined by the customer. From the measured data were
created control charts. On the basis of this, the instability of the
process was determined and its causes were determined by compiling the
Ishikawa diagram. Based on the results was detected a local cause caused
by a human factor. Corrective action has been taken and in case of
repetition, local measures are proposed directly by setting the software
on the machine. With further data collection, the process was already
stability and capable.
The human factor has a significant share in the level of precision
of production. It is a risk factor that must be remembered in the
planning, production and control of machinery, especially in the
automotive field, where the precision and quality of production is
required on the first place.
DOI: 10.2507/28th.daaam.proceedings.053
5. Acknowledgments
This paper includes results created within the SGS-2016-005:
Research and development for innovations in the field of mechanical
engineering technology--machining technology.
6. References
[1] Jarosova E. a D. Noskievicov. (2015). Pokrocileja metody
statisticke regulace procesu. Grada Publishing a.s. ISBN
978-80-247-5884-8. Praha.
[2] Tabak, John. Probability and statistics. (2004). The science of
uncertainty. Fact On File, Inc. ISBN 0-8160-4956-4. New York.
[3] Hrehova S. (2016). Predictive Model to Evaluation Quality of
the Manufacturing Process Using Matlab Tools, In Procedia Engineering,
Volume 149, 2016, Pages 149-154, ISSN 1877-7058,
https://doi.org/10.1016/j.proeng.2016.06.649.
[4] Horalek, V. (1999). QS--9000 SPC. CSJ. ISBN 80-02-01293-3.
Praha.
[5] Meran, R., John, A., Roenpage, O., Staudter, Ch. (2013). Six
Sigma + Lean Toolset (Mindset for Successful Implementation of
Improvement Projects). Springer. ISBN 978-3-642-35881-4. Berlin.
[6] CSN ISO 8258. (1998). Shewhartovy regulacm diagramy(Shewharts
control charts). Cesky normalizacni institut. Praha.
[7] Nelsonova pravidla pro regulacm diagramy. [Online] [Cit.: 23.
5. 2017] http://upload.wikimedia.org/wikipedia/commons/7/72/Poster_-_Control_Charts_for_Nelson_Rules.svg
[8] Kenneth P. Wright Jr., Richard K. Boganb, James K. Wyatt
(2013). Shift work and the assessment and management of shift work
disorder (SWD), Sleep Medicine Reviews, Volume 17, Issue 1, February
2013, Pages 41-54 http://www.sciencedirect.com/science/article/pii/S1087079212000251
[9] Martin Melichar, Dana Kubatova. (2015). Processing Data from
Automatic Measurement Device, Procedia Engineering, Volume 100, 2015,
Pages 899-906, ISSN 1877-7058,
http://dx.doi.org/10.1016/j.proeng.2015.01.447
Caption: Fig. 1. Control chart for mean
Caption: Fig. 2. Control chart for range
Caption: Fig. 3. Example of Nelson's rules [7]
Caption: Fig. 4. Control chart with marked points outside the
regulation limits
Caption: Fig. 5. A sample search for causes
Table 1. Calculated statistics for control diagrams
Calculated statistics for CD:
[n.sub.j] 10
k (amount of subgrups) 30
x bar bar 24.030180
Me bar 24.030300
R bar 0.005300
s bar 0.001668
[s.sup.2] bar 0.000003
s tot 0.001937
s xbar 0.000894
COPYRIGHT 2018 DAAAM International Vienna
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2018 Gale, Cengage Learning. All rights reserved.