Int J Mater Form (2010) Vol.

3 Suppl 1:531–534
DOI 10.1007/s12289-010-0824-y
© Springer-Verlag France 2010

PROCESS CAPABILITY ANALYSIS IN THE MANUFACTURING

OF PET BOTTLES

Maja Rujnić-Sokele*, Mladen Šercer, Damir Godec

Faculty of Mechanical Engineering and Naval Architecture, University of Zagreb

ABSTRACT: Modern process control is based on the use of statistical methods to improve and ensure quality, and thus
reduce the process cost due to waste and rejects. The paper deals with the process capability analysis in the
manufacturing of poly(ethylene-terephthalate) bottles for packaging of table oil. PET bottles are manufactured by the
procedure of injection stretch blow moulding. The preforms made by injection moulding are biaxially extruded in the
blow mould, with blow parameters affecting the mechanical properties, the barrier properties and the transparency of
the manufactured bottle. The performance of blow moulding machine was monitored by measuring the volumes of
blown bottles. Based on the collected data, the capability analysis of the blow moulding process was conducted.

KEYWORDS: poly(ethylene-terephthalate) bottles, process capability, injection stretch blow moulding

1 INTRODUCTION Since an empty PET bottle reaches its final dimensions

after 72 hours, i.e. it shrinks during this period, the
Poly(ethylene terephtalate) (PET) is a packaging filling line consists of bottles of various volumes,
material that is commonly used to package various depending on the storage time.
beverages, mineral water, oil and vinegar, beer and more
recently wine as well.
PET bottles are manufactured by the procedure of
2 TESTING MATERIAL AND
injection stretch blow moulding. The preforms made by
injection moulding are biaxially extruded in the blow EQUIPMENT
mould, with blow parameters affecting the mechanical For the purposes of this work, the bottles that were used,
properties, the barrier properties and the transparency of were manufactured out of the preform of 26g mass, with
the manufactured bottle. PET properties presented in Table 1.
The statistical process control (SPC) can be applied to
many different processes. The goals of SPC are to Table 1: PET properties
improve and ensure quality, and thus, reduce the process
cost due to waste as result of rejects. While the aim of Property Value
traditional product inspection is to detect defects, the aim
Intrinsic viscosity 0.81  0.02dl/g
of SPC is to prevent them. The process is considered
Acetaldehyde content < 1ppm
capable if the actual product variability is less than the
specification width. If the product variability is larger Moisture content  0.2%
than the specification width, the process is considered Melting temperature 245±2ºC
not capable. [1] Density 1.41g/cm3
The experiment was carried out in an oil manufacturing
factory, and the task was to determine the capability of The blow moulding machine, of type SBO 6/10, of the
the blow moulding procedure. The performance of the Sidel company, Octeville sur Mer, France, was used. The
blow moulding machine was monitored during the temperature regulation of mould cavities was carried out
period of four months by measuring the volumes of by the Vulcatherm device, manufactured by Vulcanic,
blown bottles. France. The temperature regulation medium is water
PET bottles can be filled with oil directly after blowing which can be heated to a maximum temperature of 90ºC.
or after several days of storage in the silo. The The water cycle consists of the cycle with cold water for
productivity of the filling line is 11,000 bottles/h, and the bottleneck and mould bottom cooling and a cycle with
productivity of the blow moulding machine is 6500 warm water for heating the mould body. For this type of
bottles/h. For this reason, all the bottles cannot be filled product the temperature of the cold water is 10 to 12ºC,
with oil directly after blowing, so a certain amount of and of warm water about 50ºC.
bottles are stored in the silo.
____________________
* Corresponding author: I. Lučića 5, HR-10000 Zagreb, Croatia, phone +38516168191, fax +38516150081, mrujnic@fsb.hr
532

3 EXPERIMENT
Over a period of four months the work of a blow
moulding machine was monitored during the production
of bottles for oil from the preforms of 26g mass. The
monitoring was based on:
a) production of bottles by stretch moulding, filling of
bottles by table oil, and weighing of the filled bottles
(measuring volume);
b) production of bottles by stretch blow moulding,
storing of empty bottles >72h (resulting in shrinking, i.e.
volume reduction), filling of bottles by table oil, and
weighing of filled bottles (measuring volume).
The weighing, i.e. measuring of bottles volume was done Figure 2: Data on volume of table oil in bottles after
storage >72h
two to four times during a single shift. Table 2 shows the
values of arithmetic means and standard deviations.
The data on the volume of oil in the bottles were used to
analyze the blowing process capabilities. The process is
Table 2: Values of arithmetic means and standard said to be capable if the output is approximately 100%
deviations in monitoring the volume of oil in PET bottles
(99.73%) of products within the limits of allowed
deviations. The process analysis analyzes primarily the
Bottle process dispersion, i.e. a relation is established between
Bottle (stored 0h) the determined dispersion and the limits of the allowed
(stored >72h)
Number of 261 406 deviations. A capable process can output also rejects if
measurements not centred (the process centre is shifted in relation to the
Arithmetic 1002.31 998.76 centre of the field of allowed deviations). The strongest
mean, ml argument of quality improvement is reflected in the
constant reduction of dispersion [1].
Standard 1.93 2.18
The analysis of the process capability yields numerical
deviation, ml
amounts of a larger number of statistical parameters, so-
Range (max- 10.4 13.2
called indices of process capability, and the indices
min)
based on the process dissipation have the highest
“weight” [1].
It can be noted that the requested average value of
Among the indices, indices Cp and Cpk have the highest
1000ml was filled in the case when the bottles are filled
importance and the widest application, and their amounts
immediately after blowing; however, when the bottles
are obtained on the basis of the following expressions
that had been stored for at least 72 hours were filled, the
[1]:
arithmetic mean of the values of volumes is lower than
the required value. U L
Figures 1 and 2 show the data about the volume of table Cp  (1)
6 s
oil in the bottles, in the case when these were filled
immediately after blowing and after the bottles had been zmin
stored for at least 72 hours (average bottle storage time C pk  (2)
was 96 h). 3

zmin=min(zU, zL) (2a)

U x
zU  (2b)
s

xL
zL  (2c)
s
where U – upper limit of allowed deviations, L – bottom
limit of allowed deviations, x - arithmetic mean of
individual data set, s – estimate of standard deviation:
Figure 1: Data on the volume of table oil in bottles
 xi  x 
2
immediately after blowing
s (3)
n 1
 533

The value of index Cp shows whether the process is Then only Cpk can be calculated, which in case of filling
capable. The higher the index value, the lower the the bottles immediately after blowing amounts to:
process dissipation. Theoretically, the process is capable
if Cp is greater than or equals 1. However, as a rule, the 1002.31  995.00
C pk  C pL   1.27
minimum value of index Cp is required to be 1.33 (U- 3  1.92
L=8s). During the analysis of the process, based on the
The value of index Cpk shows whether rejects are measuring of volumes of stored bottles, the capability
produced during the process, i.e. not produced if the index amounts to:
index value is greater than or equals 1. If the process is
properly centred the values of indices Cp and Cpk are 998.76  995.00
equal [1]. C pk  C pL   0.57
3  2.18
While Cp relates the spread of the process relative to the
specification width, it does not address how well the Histograms of spread of the bottle volume values
process average x is centred to the target value. Cp is measured immediately after blowing and after storage
often referred to as process "potential". Cpk measures not >72h are presented in Figures 3 and 4.
only the process variation with respect to allowable
specifications, it also considers the location of the
process average. [2]
In analyzing the blowing process based on the
monitoring of data on the volumes of table oil in bottles
there are several approaches, during which completely
different results are obtained. In principle, the limits of
allowed deviations have not been clearly defined, i.e.
only the required nominal value of 1000ml is known.
Depending on the set limits of allowed deviations,
completely different values of Cpk. are obtained. The
objective is the production of bottles of maximum
possible volume (with the aim of fulfilling the
requirements of average volume of 1000ml according to
the Regulations on metrology requirements for
packaging), i.e. minimum possible post-shrinkage in
order to have bottles with minimal differences in volume Figure 3: Analysis of the blowing process, volume
measurements immediately after blowing
on the filling line.

3.1 Example 1
Because of the mentioned reasons, the upper limit is set
at 1005ml, and the bottom limit at 995ml. In this case the
values of capability indices Cp and Cpk in filling the
bottles directly after blowing are:
1005.00  995.00
Cp   0.87
6  1.92

1005.00  1002.31
Cpk  CpU   0.47
3  1.92
and in case of filling the bottles after a storage of at least
72h:
Figure 4: Analysis of the blowing process, volume
1005.00  995.00 measurements after storage of >72 h
Cp   0.76
6  2.18
3.3 Discussion about results of blowing capability
998.76  995.00 analysis
C pk  C pL   0.57
3  2.18 According to the capability indices calculated on the
basis of the set limits of allowed deviations at 995, i.e.
1005ml, the blowing process is not capable either in the
3.2 Example 2 case of measuring the volume of new or stored bottles.
Another approach is also possible, in which only the From the value Cpk, (which amounts to 0.47 when
bottom limit of allowed deviations, e.g. 995ml, is set. measuring the volume of new bottles, and 0.57 in
measuring the volume of stored bottles) it could be
534

concluded that the more capable process is in case of

filling the bottles after storage. However, bottles filled
with more than 1005ml of oil do not represent in fact
rejects, so that the more acceptable approach is the one
mentioned in Example 2, i.e. when only the bottom limit
of the permitted deviations is stipulated.
In case of measuring the volume of bottles immediately
after blowing, with the bottom limit of allowed
deviations set at 995ml, Cpk is greater than 1, so that the
blowing process may be considered capable. The
analysis of the process based on the measuring of the
volume of stored bottles, Cpk is 0.57, i.e. the process
cannot be considered capable.

4 CONCLUSION
At the oil production factory the differences in the
volume of oil in the filled bottles were noticed,
depending on whether the bottles arrived to the filling
line immediately after the blowing process or after
having been stored for a certain time. An empty PET
bottle acquires its final dimensions after 72 hours, during
which period the bottle undergoes shrinkage so that the
bottles arriving to the production line have different
volumes depending on the storage period. The possible
solution for resolving the problem would be acquisition
of new blow molding machine, in which case the bottles
could be filled directly after blowing, and the problem of
post-shrinkage and filling of unequal oil quantities
would disappear.
The application of methods of statistical management of
the process enables control of the quality of the PET
bottles blowing process. The quality control shows the
tendency towards the control directly on the production
line; however, the majority of properties can be defined
only after the production, in a testing laboratory. The
work shows the implementation of statistical
management of the process providing the example of
PET bottle blowing.

ACKNOWLEDGEMENT
This work is part of the research included in the project
Increasing Efficiency in Polymeric Products and
Processing Development that is included in program
Rapid Production – From Idea to Reality supported by
the Ministry of Science, Education and Sports of the
Republic of Croatia. The authors would like to thank the
Ministry for the financing of this project.

REFERENCES
[1] Rauwendaal C: SPC – statistical process control in
extrusion. Carl Hanser Verlag, Munich Vienna New
York Barcelona, 1995
[2] Mudronja V: Sposobnost procesa – teorija i praksa.
In CIM'95 Proceedings, Zagreb, 10-17, 1995
[3] Crow K: Process Capability and Product Design.
www.npd-solutions.com/proccap.html