456
O. Ogorodnyk et al.
32.1 Introduction
In the last thirty years, the popularity of the injection molding (IM) process had an
increasing growth due to new applications in the fields of appliances, packaging and
automotive industry (Fernandes 2018). As a result, today injection molding is one of
the most frequently used processes for the production of thermoplastic parts in high
volume and low cost.
Injection molding includes only four main phases: plasticization, injection,
cooling and ejection, however, it is a rather complicated process due to the presence
of non-linearities (Chen 2008). At the same time, due to its use for mass production,
high process repeatability is extremely important (Ogorodnyk and Martinsen 2018)
and keeping quality of the products as high and as similar as possible is a “must”.
Depending on the application of the manufactured product, its quality can be
defined in different ways. Usually, the good quality of the part means desired mechanical performance, dimensional consistency and proper appearance (Fernandes 2018).
“Dimensional consistency is a critical attribute for injection molded part quality
and is highly dependent on various processing parameters” (Panchal and Kazmer
2010). Product dimensions can be one of the criteria for accepting or declining a
thermoplastic product.
The part quality, including dimensional accuracy, can be influenced by a significant number of factors such as general condition of the injection molding machine
(IMM), mold that is in use, input material condition, drifting of the process parameter
settings or the machine operator’s fatigue (Kozjek et al. 2019). Some of the process
parameters that may cause product quality variations are melt temperature, mold
temperature, holding pressure, cooling time, etc. (Xu and Yang 2015). Because of
the faulty setting of some of those parameters, such defects as warpage, sink mark, air
traps, and weld lines might occur. As a result, prediction of final parts quality using
certain process parameter values and “optimal setting of injection molding process
variables plays a very important role in controlling the quality of the injection molded
products” Mathivanan et al. 2010; Wortberg and Schiffers 2006).
The optimal process parameter settings were used to be determined by engineers
and IMM operators based on their experience, intuition and trial-and-error (Guo
2012; Shi 2003). However, due to the increasing quality requirements new approaches
for optimization of the plastics injection molding are being continuously developed.
“Researchers introduced the design of experiment (DOE), Taguchi orthogonal array
and flow analysis software such as Moldflow Plastic Insight” (Cheng et al. 2013).
Apart from statistical process control, design of experiments and Taguchi
approach, machine learning methods show all of the necessary capabilities for the
development of predictive models for the quality of the injection molded products
(Cheng et al. 2013). Moreover, they have been proven to be better at dealing with nonlinearities in comparison to conventional statistical methods such as linear regression
(Ogorodnyk and Martinsen 2018).
Quality requirements are not the only challenge that the thermoplastics injection molding industry faces today. Another important issue is plastic pollution,
O. Ogorodnyk et al.
32.1 Introduction
In the last thirty years, the popularity of the injection molding (IM) process had an
increasing growth due to new applications in the fields of appliances, packaging and
automotive industry (Fernandes 2018). As a result, today injection molding is one of
the most frequently used processes for the production of thermoplastic parts in high
volume and low cost.
Injection molding includes only four main phases: plasticization, injection,
cooling and ejection, however, it is a rather complicated process due to the presence
of non-linearities (Chen 2008). At the same time, due to its use for mass production,
high process repeatability is extremely important (Ogorodnyk and Martinsen 2018)
and keeping quality of the products as high and as similar as possible is a “must”.
Depending on the application of the manufactured product, its quality can be
defined in different ways. Usually, the good quality of the part means desired mechanical performance, dimensional consistency and proper appearance (Fernandes 2018).
“Dimensional consistency is a critical attribute for injection molded part quality
and is highly dependent on various processing parameters” (Panchal and Kazmer
2010). Product dimensions can be one of the criteria for accepting or declining a
thermoplastic product.
The part quality, including dimensional accuracy, can be influenced by a significant number of factors such as general condition of the injection molding machine
(IMM), mold that is in use, input material condition, drifting of the process parameter
settings or the machine operator’s fatigue (Kozjek et al. 2019). Some of the process
parameters that may cause product quality variations are melt temperature, mold
temperature, holding pressure, cooling time, etc. (Xu and Yang 2015). Because of
the faulty setting of some of those parameters, such defects as warpage, sink mark, air
traps, and weld lines might occur. As a result, prediction of final parts quality using
certain process parameter values and “optimal setting of injection molding process
variables plays a very important role in controlling the quality of the injection molded
products” Mathivanan et al. 2010; Wortberg and Schiffers 2006).
The optimal process parameter settings were used to be determined by engineers
and IMM operators based on their experience, intuition and trial-and-error (Guo
2012; Shi 2003). However, due to the increasing quality requirements new approaches
for optimization of the plastics injection molding are being continuously developed.
“Researchers introduced the design of experiment (DOE), Taguchi orthogonal array
and flow analysis software such as Moldflow Plastic Insight” (Cheng et al. 2013).
Apart from statistical process control, design of experiments and Taguchi
approach, machine learning methods show all of the necessary capabilities for the
development of predictive models for the quality of the injection molded products
(Cheng et al. 2013). Moreover, they have been proven to be better at dealing with nonlinearities in comparison to conventional statistical methods such as linear regression
(Ogorodnyk and Martinsen 2018).
Quality requirements are not the only challenge that the thermoplastics injection molding industry faces today. Another important issue is plastic pollution,