You are on page 1of 5

Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com
ScienceDirect
Available ScienceDirect
Availableonline
onlineatatwww.sciencedirect.com
www.sciencedirect.com
Procedia Manufacturing 00 (2019) 000–000
ScienceDirect
ScienceDirect 
Procedia Manufacturing 00 (2019) 000–000 www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia
Procedia Manufacturing 28 (2019) 102–106
Procedia Manufacturing 00 (2017) 000–000
www.elsevier.com/locate/procedia
International Conference on Changeable, Agile, Reconfigurable and Virtual Production
International Conference on Changeable, Agile, Reconfigurable and Virtual Production
AI Based Injection Molding Process for Consistent Product Quality
AI Based Injection Molding Process for Consistent Product Quality
Manufacturing Hong
Engineering a
Society International b
Conference MESICc*2017, 28-30 June
2017, Kumar
Seok Park a
, Dang Xuan Phuong b
, Saurabh
a,c
Hong Seok Park2017, VigoXuan
, Dang (Pontevedra), , Saurabh Kumarc*
Phuong Spain
Department of Mechanical Engineering, University of Ulsan, 93 Daehak-ro, Namgu, Ulsan 680-749, Republic of Korea
a,c bFaculty of Mechanical Engineering, NhaUniversity
Trang University,
Department of Mechanical Engineering, of Ulsan,293
Nguyen Dinh Chieu,
Daehak-ro, Namgu,Nha Trang
Ulsan City, 650000,
680-749, RepublicVietnam
of Korea
Costing models for capacity optimization in Industry 4.0: Trade-off
b
Faculty of Mechanical Engineering, Nha Trang University, 2 Nguyen Dinh Chieu, Nha Trang City, 650000, Vietnam

between used capacity and operational efficiency


Abstract
Abstract
In manufacturing processes,A. Santana , P.isAfonso
widely used, A. for Zanin
producing, R. Wernke
a a,* b b
Injection Molding plastic components with large lot size. So,
continuous
In manufacturingimprovements in product
processes, Injectionquality
aMoldingconsistency
University isof widely
Minho,is 4800-058
crucial
used for toGuimarães,
maintaining
producing a competitive
plastic
Portugal componentsedge with
in thelarge
injection molding
lot size. So,
industry.
continuous Various optimization
improvements techniques
in product like
quality ANN, GA,89809-000
b consistency
Unochapecó, Iterative method,
is crucialChapecó, andBrazil
to maintaining
SC, simulation based are
a competitive being
edge usedinjection
in the for optimization
molding
of Injection
industry. Molding
Various process and
optimization obtaining
techniques likeoptimal
ANN, GA, processing
Iterativeconditions.
method, andBut still due based
simulation to variation
are being during
used molding cycles,
for optimization
quality failureMolding
of Injection occurs. As manyand
process constituents
obtaininglike process,
optimal Material,conditions.
processing machine together
But stillyields
due product quality.
to variation This molding
during paper is focused
cycles,
on Realfailure
quality time AI basedAscontrol
occurs. many of process parameters
constituents like process, in Material,
injection molding cycle. Process
machine together yields parameters and their
product quality. This interrelationship
paper is focused
Abstract
with
on Real quality
time failure
AI basedhascontrol
been studied and parameters
of process later supposed to be used
in injection to generate
molding algorithm
cycle. Process for compensating
parameters and their the deviation of
interrelationship
process parameters.
with quality failure Pressure
has beenand temperature
studied and later sensor assisted
supposed to monitoring
be used to systemgenerateis algorithm
used to collect data in real time
for compensating theand based on
deviation of
Under
its
process the
comparison concept
parameters. of standard
withPressure
the "Industry 4.0",anproduction
values
and temperature processes
interrelationship
sensor assisted is formed
monitoring willsystem
be pushed
between tocollect
parameters
is used to be and
increasingly
plastic
data interconnected,
in realmaterial
time andproperties.
based on
Algorithm
its comparison
information generates
withon
based new
thea process
standard
real timeparameter
values
basis an values
and, to compensate
interrelationship
necessarily, is the
much deviation
formed
more and machine
between
efficient. In thiscontrol
parameters and follows
context,plastic the same.
material
capacity The entire
properties.
optimization
process
Algorithm
goes is supposed
beyond generates to beprocess
new
the traditional smart
aimand automatic
parameter
of capacity after
values being trained
to compensate
maximization, thewith AI and
deviation
contributing machine
and
also machine
for learning
controltechniques.
organization’s follows the Simulation
same.and
profitability using
Thevalue.
entire
Moldflow is software
supposed and
to bereal industry
smart and collected
automatic data
after has
being been
trained used
withfor
AI understanding
Indeed, lean management and continuous improvement approaches suggest capacity optimization insteadusing
process and machine whole
learning molding
techniques.process establishing
Simulation of
relationship betweenand failure
Moldflow software
maximization. The study of and
real industry parameters.
capacity collected An dataautomotive
optimization has been used
and costing product in real industry
for understanding
models is an important is chosen
whole moldingfor
research data establishing
process
topic thatacquisition,
deserves
implementation
relationship between and validation
failure ofandentire AI based system.
parameters. An automotive product in real industry is chosen for data acquisition,
contributions from both the practical and theoretical perspectives. This paper presents and discusses a mathematical
implementation and validation of entire AI based system.
model
© 2019for Thecapacity
Authors. management based on
Published by Elsevier B.V.different costing models (ABC and TDABC). A generic model has been
developed
©
This2019
© 2019 The
is an
The and
open it was
Authors.
Authors. used to
Published
accessPublished analyze
by
article under idle
Elsevier capacity
B.V.
the CC BY-NC-ND
by Elsevier B.V. and to design
license strategies towards the maximization of organization’s
(https://creativecommons.org/licenses/by-nc-nd/4.0/)
This is an
value. The open access article
trade-off capacityunder the CC BY-NC-ND
maximization vs license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
operational efficiency is highlighted and it isAgile,
shown that capacity
Peer-review
This is an openunder responsibility
access of the
article under thescientific
CC BY-NC-NDcommittee of the(https://creativecommons.org/licenses/by-nc-nd/4.0/)
license International Conference on Changeable, Reconfigurable
Peer-review under responsibility of the scientific committee of the International Conference on Changeable, Agile, Reconfigurable
optimization
and Virtual
Peer-review might
Production. hide operational inefficiency.
under responsibility of the scientific committee of the International Conference on Changeable, Agile, Reconfigurable
and Virtual Production.
©
and2017 TheProduction.
Virtual Authors. Published by Elsevier B.V.
Peer-review
Keywords: Data under responsibility
Driven; of the scientific
Smart manufacturing; Machinecommittee of the Manufacturing
learning; Diagnosis; Injection MoldingEngineering Society International Conference
2017.
Keywords: Data Driven; Smart manufacturing; Machine learning; Diagnosis; Injection Molding

Keywords: Cost Models; ABC; TDABC; Capacity Management; Idle Capacity; Operational Efficiency

*1.Corresponding
Introduction author. Tel.: +82-010-3073-7035; fax: +82-(0)52-259-1680.
* E-mail address:author.
Corresponding saurabh2313@gmail.com
Tel.: +82-010-3073-7035; fax: +82-(0)52-259-1680.
E-mail
The address:
cost of idle capacity is a fundamental information for companies and their management of extreme importance
saurabh2313@gmail.com
2351-9789
in modern©production
2019 The Authors. Published
systems. by Elsevier
In general, it isB.V.
defined as unused capacity or production potential and can be measured
This is an open
2351-9789 © access
2019 Thearticle under
Authors. the CC BY-NC-ND
Published by Elsevier license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
B.V.
in several
Peer-review
ways:
under
tons of
responsibility
production, available
of the scientific committee
hours
of the
of manufacturing, etc. The management of the idle capacity
International Conference on Changeable, Agile, Reconfigurable and Virtual
This is an open access article under CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
* Paulo
Production.Afonso. Tel.: +351 253 510 761; fax: +351 253 604 741
Peer-review under responsibility of the scientific committee of the International Conference on Changeable, Agile, Reconfigurable and Virtual
E-mail address: psafonso@dps.uminho.pt
Production.

2351-9789
2351-9789 ©©2017
2019The
TheAuthors.
Authors. Published by Elsevier
Published B.V. B.V.
by Elsevier
Peer-review underaccess
This is an open responsibility of the scientific
article under committee oflicense
the CC BY-NC-ND the Manufacturing Engineering Society International Conference 2017.
(https://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the scientific committee of the International Conference on Changeable, Agile, Reconfigurable
and Virtual Production.
10.1016/j.promfg.2018.12.017
Hong Seok Park et al. / Procedia Manufacturing 28 (2019) 102–106 103
2 Hong Seok Park/ Procedia Manufacturing 00 (2019) 000–000

Introduction

In manufacturing processes, injection molding has been a premiere and popular technology in the production of
plastic components. With the evolution of hi-tech modern molding machines and simulation technologies, faults are
becoming increasingly infrequent [1]. Increased demand for quality products, puts an extra pressure on
manufacturers in becoming top producers of plastic components. The quality of molded products are the result of
multiple machine, material and process parameters [2]. Improper settings of process variables will produce various
defects in the final product [3,4]. As need for control of the injection molding process is high, the first step in this
case is to precisely design, measure and monitor the process to make the key process variables observable and
controllable [5]. The process of injection molding includes four main stages: plasticization, injection, cooling and
ejection. Among these four, the cooling stage takes from 50% to 80% of the cycle time [6]. Many optimization
techniques and simulation-based studies are being in use for obtaining optimal process parameters for conducting
injection molding process and obtaining product with highest quality. But still there is a question mark on their
practical utility as due to variation during molding cycles, quality failures occur. Process monitoring and control, as
well as use of variotherm technology or conformal cooling/warming channels would benefit from application of
artificial intelligence methods to function in the most optimal way [7].
The goal of this paper is to present a knowledge-based study and implementation of autonomous quality control
system for consistent quality products with injection molding. We are in the age of revolutionized information and
technology, where machines are able to generate the data recorded during production cycle and with Big data
management, we have now the capability to process the data and obtain requisite trouble shooting methods. But still
in the field of injection molding, quality and its failure troubleshooting are dependent on human interference. Real
time process parameter optimization and control without any human interference is still a distant approach due to
complexity of injection molding process and related input and output parameters. The contributions of this paper are
system model for AI based quality control, powered by Big data and its proposed implementation in real industrial
environment using real data of injection molding process for an automotive product. The paper includes state of the
art in this field, followed by research methodology, description, discussion and conclusion.

1. State of the Art

In the recent years, several approaches have been used to monitor real time process parameter values and utilize it
for developing efficient quality control system and reduce some prominent quality defects like warpage, shrinkage
and flashes. These approaches include the Model based classical control, Taguchi techniques, Artificial neural
networks (ANN), Fuzzy logic, Genetic Algorithms, Support Vector Machines, Case Based Reasoning are being in
use for process parameters optimization [8]. Adaptive process control has also been utilized for self-learning control
which utilizes data acquisition, data mining and knowledge-building models [9]. Process monitoring for injection
molding using nozzle-based pressure and temperature sensors has also been explored [10]. Apart from that many
pressure and temperature sensor assisted monitoring system are used by manufacturers to get insight of the molding
process in real time. Still there is lot of scope for the development in AI based control system as with the use of
machine learning and big data, injection molding machines can learn the operations by themselves.

2. Research Methodology

This paper describes about real time monitoring, diagnosis and autonomous quality control system. The
process flow starts with data generation from millions of injection molding cycles with the use of cavity pressure
and temperature sensors as shown in figure 1. Data generated is recorded with their identical cycle id, corresponding
process parameters, real sensor signals and corresponding product quality types as shown in the table 1. Data
collected of millions of cycles is stored in a database for further analysis and reference model generation purpose.
Collected data is used to study the interrelationship between machine input process parameters, real sensor readings
and product quality type. Based on this study a compensatory algorithm is proposed to variate the input parameters
of the machine to get desirable conditions inside the cavity, depicted by sensor readings.
104 Hong Seok Park et al. / Procedia Manufacturing 28 (2019) 102–106
Hong Seok Park/ Procedia Manufacturing 00 (2019) 000–000 3

Table 1: Cycle number and corresponding values Fig. 1. Sensor assisted data acquisition system

After the analysis of real time sensor readings and comparing it with the standard values, system itself generates
new machine input process parameters accordingly. A conceptual flow diagram for the realization of control system
is shown in the figure 2.

Fig. 2. Flow diagram for control system

3. Description and Discussions

In real time monitoring and control system, 2 temperature and 2 pressure sensors are installed inside the cavity of
car door module. Sensors provide real time inputs about cavity temperature and pressure. Product quality depends
upon various process parameters like injection pressure, hold pressure, mold temperature, injection speed etc.
Among these parameters some are controllable directly through the injection molding machine controller and some
of them are dependent on machine input parameters. In order to have change in dependent parameters an
interrelationship is formed between plastic materials pvt behavior and machine variables. Experienced engineers of
injection molding machine can also provide the corresponding machine variable values for an uncontrollable process
parameter. In supervised learning phase of the AI based modelling these cavity temperature and pressure values are
identified based on the corresponding final product quality that what values of cavity temperature and pressure gives
product of bad quality. After analyzing it with comparison with reference, the system calculates the difference and
based on its learning algorithms it generates corresponding input machine parameters to provide required increase or
decrease in it. New process parameter implementation is followed by the machine controller. All the programming
for data acquisition, analysis, comparison and new process parameter estimation is done with the use of python. An
interface has to be generated to integrate developed control system in to the actual injection molding machine
control system. Collected data from millions of cycles is utilized to train this AI based model for its automatic
functioning and improved accuracy. The proposed quality control system if implemented successfully in an injection
molding machine will be able to provide immediate control of machine control values to compensate the fluctuation
in the operating conditions of an ongoing molding process. System architecture for implementation and validation
of this system is shown in the fig 3.
Hong Seok Park et al. / Procedia Manufacturing 28 (2019) 102–106 105
4 Hong Seok Park/ Procedia Manufacturing 00 (2019) 000–000

Fig. 3. Smart Control System

4. Conclusion

Quality consistency and stable processing conditions are a common problem for injection molding process due to
non-uniform variations in the molding machine after several cycles. The quality of final product depends on all the
parameters involved in molding process. Process monitoring can be done with the assistance of sensors and other
methods. Then relation between parameters that can and cannot be measured can be described through application
of modelling and formulas. With use of 2 temperature sensors and pressure sensors data for millions of cycles is
recorded and stored in a database to generate reference model and control algorithms. After building standard value
and establishing interrelationship between sensor signals, and machine input parameters, study is focused on real
time monitoring and control. For that purpose, a learning algorithm is setup with the mindset of generating machine
input parameters for corresponding compensation in sensor readings and hence the product quality. Despite several
existing methods of quality control, use of AI will be more beneficial than the conventional ones. Future works with
this study is to generate learning algorithms and implementation of whole AI based control system for the
manufacturing of car door with the use of use of injection moiling, and validation of whole system with real
experiments.

Acknowledgements

This work is supported by the Ministry of Trade, Industry & Energy (MOITE, Korea) under Industrial
Technology Innovation Program No. 10062677. “For plastic injection molds with 50% improved cooling efficiency,
development of 3D printing equipment & mold steel powder manufacturing technology having 30 µm grade in
diameter”.

References

[1] Dominik Kozjek, Rok Vrabic, David Kralj, Peter Butala. A Data-Driven Holistic Approach to fault Prognostics in a Cyclic manufacturing
Process, Procedia CIRP 63 (2017) 664-669.
[2] Suhash Kulkarni, Robust Process Development and Scientific Molding, Hanser Publication 2 nd Edition.
[3] Guilong W, Guoqun, Huiping, Yanjin G, Analysis of thermal cycling efficiency and optimal design of heating/cooling systems for rapid heat
cycle injection molding process. Materials & Design 31(7); 2010.p 3426-41
[4] Zhao P, Zhou H, He Y, Cai K, Fu J, A nondestructive online method for monitoring the injection molding process by collecting and analyzing
machine running data. The International Journal of Advanced Manufacturing Technology. 72 (5-8). P. 765-77
106 Hong Seok Park et al. / Procedia Manufacturing 28 (2019) 102–106
Hong Seok Park/ Procedia Manufacturing 00 (2019) 000–000 5

[5] Karbasi H, Reiser H, Smart Mold: Real-Time in-Cavity Data Acquisition. First Annual Technical Showcase & Third Annual Workshop,
Canada;2006
[6] Masood S, Trang N N, Thermal analysis of conformal cooling channels in injection moulding. Proceedings of the 3 rd BSME-ASME
International Conference on Thermal Engineering; 2006
[7] Olga Ogorodnyk, Kristian Martinsen, Monitoring and control for thermoplastics injection molding A review. Procedia CIRP; 00(2017) 000-
000

[8] Alberto Tellaeche, Tekniker-IK4, calle Inaki Gipuzkoa, Machine Learning algorithms for quality control in Plastic Molding Industry,
Industrial Practice paper.
[9] Seid Zapcevic, Peter Butala, Adaptive process control based on a self-learning mechanism in autonomous manufacturing systems,
International Journal of Advanced Manufacturing Technology. (2013) 66:1725-1743
[10] Stephen Orzechowski, Alexandre paris and Christopher J. B. Dobbin, A Process Monitoring and Control System for Injection Molding using
Nozzle-based Pressure and Temperature Sensors.

You might also like