Characterising energy and eco-efficiency of injection moulding processes

Abstract

In the age of climate awareness, energy and eco-efficiency have become an issue of great importance, particularly in the field of manufacturing. Due to the dynamic nature of manufacturing processes, reducing their energy consumption and the associated environmental impacts require knowledge about energy consumption as a function of shop-floor processes and their process parameters. However, the industrial average data were normally used for the energy estimation or life cycle assessment, which may underestimate the energy consumption 10-fold than the actual consumption in some cases. Alternatively, this paper proposes an empirical approach to characterise the energy efficiency on different injection moulding machine tools. The derived models offer a unique ability to predict energy consumption of a given machine tool with an accuracy of over 90%, which can also improve the quality of current life cycle inventory database. The eco-efficiency of the injection moulding process has also been discussed in this paper. The inverse of specific energy consumption models can be directly used for evaluating the eco-efficiency, since injection moulding processes mainly consume electricity in a dynamic manner. The throughput rate has been found as a decisive factor for the energy and eco-efficiency of injection moulding processes.

Keywords::

• empirical modelling
• energy efficiency
• eco-efficiency
• injection moulding

NomenclatureDoE	=	design of experiments
E	=	total energy consumption
HDPE	=	high density polyethylene
LCA	=	life cycle assessment
LCE	=	life cycle engineering
LCI	=	life cycle inventory
LDPE	=	low density polyethylene
m	=	mass
m˙	=	throughput rate
P	=	power demand
PS	=	polystyrene
Q	=	injection volume
SEC	=	specific energy consumption
T	=	cycle time
tc	=	cooling time
td	=	dwell time between two cycles
ti	=	injection and holding time
w	=	thickness

1. Introduction

In the age of climate awareness, energy and eco-efficiency have become an issue of great importance, having direct implications within the field of manufacturing and life cycle engineering. The issue is compounded due to the trend towards energy-intensive manufacturing processes used for the production of new products with larger demand (Duflou et al. 2012). As the energy requirements of manufacturing industries surge, a similar effect can be related to greenhouse gas emission especially in locations where electricity generation is mainly dependent on fossil fuel. For example, in Australia, manufacturing industries recently comprise 18.6% of energy consumption and 27.7% of emissions (Garnaut 2011). Australian manufacturers also need to contend with additional emission penalties as a result of the carbon taxation system implemented in July 2012. Faced with this operating environment, manufacturers are under more pressure to increase their energy efficiency and to reduce their environmental footprint.

Due to the dynamic nature of manufacturing processes, reducing their energy consumption and the associated environmental impacts requires knowledge about energy consumption as a function of shop-floor processes and their intrinsic parameters. However, traditional methods are primarily dependent upon the use of industrial average data or a theoretical minimum energy requirement for estimating energy consumption of manufacturing processes. The issue is compounded due to the limited disclosure of information on behalf of machine tool manufacturers. Although recent studies regarding the energy efficiency of manufacturing processes have served to improve its transparency, the case of injection moulding processes remains under-investigated. In addition, current database of life cycle inventory (LCI) only provides an industrial average for estimating the electricity consumption of injection moulding 1 kg of plastics. For some cases, the actual energy consumption is 10 times greater than the average value. As a result, the biased estimation is unavoidable as it disregards the process conditions. With this in mind, developing a method for characterising energy consumption of injection moulding processes is essential to reducing the energy consumption as well as improving the environmental performance of injection moulding machine tools. More importantly, the technological performances of a process such as surface finish, mould wear and so on need to be taken into consideration at the same time. In other words, a comprehensive view on the economic value of a process should be applied, including both cost reduction from pursuing energy efficiency and the associated economic impact on technological performances. Subsequently, the eco-efficiency of a unit process needs to be introduced and discussed concurrently.

2. Research background

Injection moulding is one of the most pre-dominant types of plastic shaping process; whereby, plastic pellet or powder materials are melted and forced into various cavities to form desired shapes. This process is favoured by designers and engineers for its inherent ability to manufacture parts using a variety of materials, shapes and sizes. As a result, numerous products such as mobile phones, computers and automobiles contain injection moulded parts (Gosato 2000). An injection moulding machine tool generally consists of four units: a heating unit, clamping unit, injection unit and cooling unit. According to the type of drive system, these machines can be classified into hydraulic, hybrid and all-electric types. In this paper, the hydraulic type machines were selected for investigation.

Following the discovery of the injection moulding process during the late nineteenth century, researchers have since focused on improving its performance. Resulting studies have explored online optimisation methods to improve throughput and surface quality (Kamoun, Jaziri, and Chaabouni 2009); investigated the influence of process parameters on the process efficiency using computerised 3D solid modelling environments (Mattis et al. 1996) and several others have recently proposed formulae to best calculate the cooling times used in cost estimation of injection moulded parts (Liang and Ness 1996; Stelson 2003). However, important to note, this pre-existing body of work has lacked attention in the areas of energy and eco-efficiency of injection moulding processes.

2.1 Related energy efficiency studies

In particular relation to energy consumption, Gutowski, Dahums, and Thiriez (2006) have been the first to study the typical injection moulding process using an exergy concept to develop a theoretical model and assess the energy trends. This model revealed that throughput per hour (or process rate) was a critical component of the specific energy consumption (SEC). However, the approach lacked a clear definition for deriving values of model coefficients and thus did not account for machine-to-machine variation. Thiriez (2006) furthered the use of this theoretical model by making comparisons with averaged industrial data; however, associated regression results suggested neither a trend nor an acceptable correlation.

Others such as Dietmair and Verl (2008) proposed a state-based energy consumption simulation for use in the study of machine tool energy efficiency. A similar screening approach was proposed by Kellens et al. (2012) under the initiative – Cooperative Effort on Process Emissions in Manufacturing (CO2PE!). In this approach, power consumption of a machine tool was recorded under different states such as ramp up, standby, processing, ramp-down and so on. This method emphasised the dynamics of machining processes in the sense of containing discrete operational stages but did not account for variable machine loads. Recently, an empirical approach of developing unit process energy consumption models has highlighted the significant shortcoming of previous work established from theoretical modelling and simulation (Li and Kara 2011; Kara and Li 2011). Empirical modelling has proven to be more accurate and reliable in the prediction of energy consumption in the case of several material removal processes. The reason is primarily due to the mechanistic nature of theoretical modelling and the general over-simplification of manufacturing processes by the simulation approach. In a similar study for milling machine tools, the empirical approach has been further validated (Diaz, Redelsheimer, and Dornfeld 2011). Chien and Dornfeld (2013) have proposed a semi-empirical approach to predict energy consumption of injection moulding processes. The prediction of fixed power consumption is highly dependent on the empirical data. Moreover, the analytic models for projecting the variable energy consumption require a large number of input parameters, and some of the parameters are difficult to estimate (e.g. motor electrical efficiency, hydraulic efficiency and heat loss ratio). Thus, the practicality of the semi-empirical models is relatively low than that of empirical models.

According to the above discussions, an empirical approach is favoured in this paper for characterising the energy consumption of injection moulding processes, discussed in detail in Section 3.

2.2 Related eco-efficiency studies

2.2.1 The general eco-efficiency definitions of manufacturing processes

Solely reducing energy consumption may shift the problem to others, for instance, poor surface finish or excessive tool wear; that is to say, other process performances need to be considered whilst improving energy efficiency of manufacturing processes. Bridging those aspects requires an appropriate platform or evaluation method. This requirement exactly agrees with the original definition of eco-efficiency: create more value with less impact (WBCSD 2000). Primarily, eco-efficiency was applied to evaluate the profitability and environmental responsibility of a corporation or a product throughout its life cycle (Aoe 2007; Sinkin, Wright, and Burnett 2008). Owing to the dynamic nature of unit process, the traditional evaluation methods become infeasible, such as a ratio between life cycle costs and life cycle assessment (LCC/LCA) (Huppes and Ishikawa 2005; Kicherer et al. 2007). Gutowski has initially discussed eco-efficiency for manufacturing processes, where he recommended eco-efficiency is often the reciprocal of some intensity metric, e.g. energy intensity (Gutowski 2010). However, exceptions have been found in the case of some complex processes, such as grinding which consumes coolant and grinding wheels in addition to electricity. More importantly, grinding process is not only measured by the removed volume but also by the achieved surface finish. Therefore, eco-efficiency for unit process has been redefined as shown in Equation (1) and Figure 1 (Li et al. 2012). A unit process can be constructed as three layers; on the upper layer, unit process transforms raw materials into output products by changing material attributes, such as geometric features, surface roughness, hardness and so on; on the bottom layer, it consumes energy and other auxiliary resources inducing environmental impacts; and in the centre, process parameters define the actual performance of the process. In order to evaluate the eco-efficiency of a unit process, it is important to select appropriate physical indicators to measure the value of the process, as well as to consider all types of energy and resources consumption comprehensively. Owing to the dynamic nature of unit processes, different parameters result in different process output and environmental impacts; subsequently, the eco-efficiency can be characterised by considering a wide range of combinations of process parameters.

(1)

Figure 1 Layers of the unit process eco-efficiency (Li et al. 2012).

2.2.2 The value of injection moulding processes

The output of injection moulding processes can be measured by the weight of injected parts. This measurement is found generic as it disregards the material and mould types. It can also be used to indicate some quality aspects as the measured weight can be compared with the weight of a standard part; thus, the major defects such as flash and short shots can be further quantified by the weight difference. Other defects, such as volume shrinkage, flow mark, warpage and so on, are usually qualitatively evaluated by technicians. These defects are normally unacceptable, but can be eliminated by properly configuring process parameters. In addition, the design and production of mould are also highly associated with these defects, as well as with some important design requirements such as surface finish, shape complexity and dimensional accuracy. The importance of the mould design and production has attracted great efforts to develop principles in order to improve their mouldability (Gosato 2000; Rees 2002; Cheng et al. 2008). Nevertheless, the mould can be considered as a given tool when evaluating the eco-efficiency of unit injection moulding processes. Therefore, the value of injection moulding processes can be measured by the weight of injected parts in conjunction with the quality checks for major defects.

2.2.3 The environmental analyses of injection moulding processes

Thiriez (2006) initially conducted environmental analyses for injection moulding processes. Their studies exclusively considered environmental impacts incurred during the production of plastic pellets. As a consequence, the environmental impacts of the injection moulding process itself become relatively insignificant. These results emphasise the importance of material production or the selection of material, but do not reflect the real energy and eco-efficiency of the injection moulding process. This is particularly the case since an averaged SEC of the injection moulding process was used for their analyses.

From the unit process point of view, the environmental impacts (as CO₂ emissions) of the injection moulding process are mainly due to the electricity consumption. Unlike other metal machining processes, there is no cutting tool or other additional resources required during this process. Although the production of mould may remain energy intensive, the mould can be used for a long period with the careful selection of mould material and manufacturing methods. According to the mould standard and classification published by Society of the Plastics Industry (SPI), class 101 mould can produce over one million cycles (SPI 2003). This paper assumes that the energy consumption and the associated environmental impact due to mould production can be shared with a large number of produced parts. Therefore, the energy consumption due to mould production over the life of the mould will be minimal from an injection moulding process point of view. Many injection moulding machines use water in the cooling system. The water flow is supplied by either a pump motor or a centralised water circulating system. In many parts of the world, industrial water use is an important concern due to local or regional water scarcity, e.g. Australia; however, since most injection moulding machines use closed cooling channels, the amount of evaporated water is believed to be negligible. Apart from CO₂ emissions, other substances emitted during the process (e.g. fume of melting plastics) may significantly contribute to other environmental impact categories (e.g. human toxicity potential), discussed in detail in Section 4. However, not every emission shows a dynamic trend in relation to the process parameters.

Owing to the dynamic nature of unit processes, different process parameters result in different process output and energy consumptions. Therefore, to evaluate the energy and eco-efficiency of injection moulding processes requires a characterisation among process parameters, output and energy consumption.

3. Energy efficiency characterisation of injection moulding processes

3.1 Methodology

As mentioned in Section 2.1, the empirical approach was used to characterise the relationship between energy consumption and process parameters based on experimental observations. The methodology has been first proposed and implemented with a case of turning process, which mainly consists of four stages as design of experiments (DoE), physical experiments, statistical analysis and validation (Li and Kara 2011). The detailed results of each stage are presented in the following sections. Three different machine tools in terms of age and capacity were tested in this research, including BOY 15 (15 ton), BOY 15S (22 ton) and Bateenfeld BA500CD (50 ton).

3.2 Design of experiments

DoE aims to screen all the possible variables, to select the relevant ones and to determine the levels of variance for the design factors (Montgomery 2008).

For the case of injection moulding, there are numerous factors affecting the process, which can be grouped into the following categories:

• Type of material: generally related to melting temperatures and density.

• Mould design: i.e. shape, size, number of cavities, wall thickness and so on.

• Plasticising unit: number of heaters, temperatures, decompression pressure and speed, and so on.

• Injection and holding unit: i.e. injection pressure and speed, holding pressure, and so on.

• Clamping unit: i.e. clamping pressure, mould close-and-open speed.

• Cooling unit: cooling water flow rate.

• Ejecting unit: ejection speed and pressure.

• Timers: injection and holding time, cooling time and dwell time.

• Operation environment: i.e. temperature, humidity and so on.

Correspondingly, there are a large number of control options to configure the above-mentioned process parameters. The switches for pressures and speeds allow setting for a certain percentage of the maximum value. An additional timer unit numerically configures the injection and holding time, cooling time and dwell time. For sake of practicality, it is inefficient to change each one during machine operation. Some of the parameters are kept constant in practice, such as ejection speed and pressure, mould open-and-close speed and so on. Those parameters are set at the maximum value in order to minimise the cycle time. Other parameters were processed with a 2_III^7-4 fractional factorial design to test the significance in relation to the energy consumption. The screening results were first published in the conference paper (Qureshi et al. 2012), which suggested that the single factor remains insignificant, but the correlations among the factors showed inconsistency. This was due to the power consumptions of the three heaters, which switch on and off at a different pace. As a result, the accuracy of the analysis was influenced significantly. Alternatively, the speed and pressure parameters were combined as one factor. It was reasonable to assume that the higher speed and pressure may require more energy; therefore, the combined factor would result in even more significant trend. For example, if a higher injection speed or a higher clamping pressure requires more power individually, setting both parameters at the high level would consume much more power.

Different materials feature different melting temperature and heat capacity. From an energy point of view, the energy consumption of heating units would correspond to the use of different materials. Normally, the heaters are set at the lowest possible temperature to melt the plastics while achieving adequate viscosity. Moreover, different heat capacity results in different melting and cooling times for each material. Thus, the material type was initially determined as a design factor to which temperature was also coupled.

The mould design also has a significant impact on this process (Rees 2002). The volume, wall thickness (w) and mould cavity numbers were resolved to limit the selection of pressures, speeds and the cycle time. However, mould making is an expensive and time-consuming process due to the high requirements of surface finish and dimensional accuracy. In order to simplify the experiments, all tested injection moulding machines featured a two-cavity, two-plate mould. Two inserts were designed for this research, as shown in Figure 2. They featured a basic round shape with different diameters and thickness (w). This design allowed the experimenters to change the volume of the injected material. When both inserts were installed normally, the maximum injection volume was achieved. The cavity was designed to be blocked, facilitating a change in the injection volume.

Figure 2 Photograph of the inserts and injected parts.

The cycle time is another important indicator, which can be used to assess the process productivity. It generally consists of three elements: injection and holding time, cooling time and dwell time. Ideally, the total cycle time should be reduced to a minimal level. Thus, dwell time (a pausing period between two cycles) was kept constant at 2 s throughout the experiments. With the addition of another two elements, this comprises the total cycle time. Therefore, injection and holding time and cooling time were considered as design factors.

After the screening of all the possible process variables, the involved factors can be summarised in the following Ishikawa diagram, see Figure 3.

Figure 3 Ishikawa diagram for tested injection moulding processes.

As previously mentioned, plastics are required to be firstly melted and subsequently solidified. The shrinkage during the cooling stage is normal and expected. This drawback has attracted intensive research directed towards predicting and managing this phenomenon (Fischer 2003). Since this research needs observed data for input, it is impractical to measure the volume change. Alternatively, the mass of the processed material (m) is constant throughout plasticisation and solidification. The injected part can be easily weighted. Therefore, the model response of SEC for the injection moulding process is the total energy consumption of producing 1 g of part (kJ/g); see Equation (2).

(2)

Throughput rate is another important indicator for the injection moulding process, which indicates the process productivity. It can be considered comparably similar to the material removal rate which has been found as a decisive factor for material removal processes (Kara and Li 2011). The calculation of throughput rate (m˙, g/s) is straightforward as Equation (3).

(3)

where t_i refers to the injection and holding time, t_c refers to the cooling time and t_d refers to the dwell time between two cycles.

After a few trial runs, the levels of each design factor were determined accordingly for the three tested materials, as shown in Table 1. The largest possible range of each process parameter has been obtained, whereby the injected parts were neither incomplete nor created with excessive flushes. In that case, the major defects have been eliminated prior to further experiments. Notably, the cooling time varies significantly for each tested material. The theoretical cooling time was first calculated as the reference level according to the equation proposed by Liang and Ness (1996); then, trial runs have been used to determine the minimal time for the solidification as the lower level, and to derive the time for the part to reach room temperature as the higher level. Afterwards, the schedule of experiments was generated through the use of MiniTab®.

Table 1 The levels of tested factors for injection moulding processes.

Factors	Variable levels
Volume (cm^3)	6.03			18.45
Material	HDPE			LDPE			PS
Heating unit
Zone 1°C	260			160			120
Zone 2°C	300			200			160
Nozzle (%)	30			70			0
Timer
ti (s)	2	6	9	2	6	9	1	5	9
tc (s)	20	40	60	20	60	100	8	34	68
Pressure and speed (%)
Clamping pressure	50	75	99	65	82	99	99	99	99
Injection pressure	45	72	99	65	82	99	50	75	99
Holding pressure	40	50	60	50	55	60	30	55	80
Decompression pressure	10	55	99	10	25	40	20	30	40
Injection speed	10	55	99	20	60	99	20	60	99
Decompression speed	8	14	20	12	16	20	15	25	35
Dosing speed	2	6	9	2	6	9	2	6	9

3.3 Experiment details

From an empirical modelling point of view, each combination of process variables should be repeated several times. Hence, machines were run in automatic mode, producing at least 10 parts with each process configuration. According to the metering strategies for unit process, the monitoring platform was developed by using a LabView® programing interface in conjunction with a National Instruments® data acquisition system (Kara, Bogdanski, and Li 2011). Figure 4 shows a sample power curve over 11 cycles. Each cycle has then been separated and processed to calculate the SEC according to Equation (2).

Figure 4 Exemplary power curve of the tested injection moulding process.

3.4 Regression and modelling

During this stage, a series of statistical analyses were conducted. First, individual design factors were processed using curve estimation in relation to SEC. Although SPSS® offers 11 different types of models, the R² value was very low for using single design factor to predict SEC. The R² value refers to the proportion of variability in the observed response that is explained by the regression model. Alternatively, the compound factor was selected for curve estimation. The throughput rate with an inverse model has resulted in the highest R² value. The largest F-value also suggested high significance of throughput rate, which agrees with the literature (Gutowski, Dahmus, and Thiriez 2006; Thiriez 2006). Figure 5 presents the model plot and the inverse model summary for the BOY 15S, respectively.

Figure 5 Model plot between throughput and SEC for BOY 15S.

The residual analysis, which plots the errors between predicted and observed SEC against the predicted one, also suggests that the inverse model is a good one; because there is no clear trend observed in the residuals plot, see Figure 6.

Figure 6 Residuals plot of the inverse model for BOY 15S.

The same procedures were repeated on the other two machines, and the results were identical. Based on the above statistical analysis, the SEC model for the injection moulding process can be written in a form as Equation (4). The values of coefficients c₀ and c₁ for the three tested injection moulding machine tools are summarised in Table 2.

(4)

Table 2 Summary of SEC models for tested injection moulding processes.

Machine	R^2	SEC model
BOY 15S	0.982
BOY 15	0.961
Battenfeld BA500CD	0.966

3.5 Model validation

Despite the high R² value, additional runs were conducted on BOY 15S to further validate the derived model. Since the models required throughput rate as an input, the mass of the injected parts was also estimated. This can be achieved by multiplying the volume of cavities with material density. The cycle time can be easily estimated based on the configuration of injection and holding time, cooling time and dwell time. It should be noted that the estimation of throughput contains errors, assuming there is no shrinkage or flash in the injected part and that the machine timing is accurate. For the validation runs, insert A (see Figure 1) was rotated to block one cavity. As a result, volume of the injected part was different from that of model development. The results are listed in Table 3.

Table 3 Validation results of SEC model for BOY 15S.

Material	HDPE	LDPE	PS
Density (g/cm^3)	0.941	0.920	1.05
Injection volume, Q (cm^3)	12.42	12.42	12.42
Cycle time (s)	80	70	25
Mass, m (g)	11.69	11.43	13.04
Throughput rate, (g/s)	0.146	0.163	0.522
Predicted SEC (kJ/g)	16.8	15.4	7.3
Predicted E (kJ)	196.15	175.90	94.65
Measured Ei (kJ)	209.251	184.15	98.25
Difference (%)	6.26	4.48	3.67

3.6 Model discussion

The above results have statistically proven that the derived SEC models can accurately predict energy consumption of injection moulding processes. For a given machine, the SEC can vary from 3 to 30 kJ/g, which is dependent on the throughput rate. In other words, using the derived SEC model can achieve a more reliable estimation of energy consumption comparing to the use of average energy consumption rate. More importantly, the derived SEC models only require limited input information for the prediction. For a given mould design, the weight of the injected part can be estimated by multiplying the volume of the cavities and the density of the material. The injection time can be estimated based on the volume of the cavities and the maximum flow rate (Boothroyd, Dewhurst, and Knight 2002). Liang and Ness (1996) reviewed different methods to calculate the cooling time, and all of them showed that the cooling time is proportional to the square of the maximum part thickness. If the temperature at different zones (e.g. melt temperature, mould temperature, ejection temperature) is also known according to the material type, a more accurate estimation of cooling time can be derived. Based on the above information, the throughput rate can be theoretically estimated; and then, the SEC and the energy consumption for producing one part can be predicted through the derived SEC model. Table 4 presents a sensitivity study for the tested three injection moulding machines and three different throughput rates. Comparing the results in the same column, it shows the magnitude of changes among these machines when processing 1 kg of thermoplastics with the same throughput rate. The differences are mainly due to the machine capacity, age and technology used.

Table 4 Sensitivity analysis of derived SEC models.

	Energy consumption of moulding 1 kg of thermoplastics (kJ)
Machine	= 0.1 g/s	= 0.3 g/s	= 0.5 g/s
BOY 15S	22,882	8385	6313
BOY 15	24,591	7799	5400
Battenfeld BA500CD	39,742	11,122	7033

As suggested by the derived SEC models, the throughput rate plays a key role in determining the energy efficiency of an injection moulding process: the higher the throughput rate, the less the energy consumption for producing the same part. In general, there are two approaches to increase throughput rate based on Equation (3). One approach is to increase the mass of the injected material by fitting more cavities into one mould, so more parts can be produced per injection moulding cycle. Notably, it may increase the cycle time, and the trade-off between the number of cavities and the cycle time requires further research. Alternatively, the other approach is to decrease the cycle time of an injection moulding process. Since the cooling time accounts for the major share of the cycle time, reducing cooling time can significantly increase the throughput as well as the energy efficiency. As mentioned before, the maximum part thickness is a key factor for the cooling time. Thus, a thinner part or a small size part is favoured from the energy efficiency perspective. Besides modifying the mould or part design, the cooling system can also be optimised in order to minimise the cooling time. Different researchers have proposed multiple approaches to focus on specific technological objectives. For example, Dimla, Camilotto and Miani (2005) used Model Master and Moldflow analysis to design and optimise the cooling channels in injection moulding tools; Sundmaeker et al. (2013) proposed a guideline to design a more energy efficient mould. In addition to the proposed measures and guidelines, the derived SEC models can quantify the improvements in terms of energy efficiency, which is important information for determining the mould design, cooling system and the process parameters.

4. Eco-efficiency of the injection moulding process

As discussed in Section 2.2, the eco-efficiency of unit process can be quantified by the ratio between process value and the environmental impacts. For the case of injection moulding, the process value can be measured by the weight of injected parts, the output parts of which are defect free (e.g. no excessive flash, flow mark and so on). On the other hand, the environmental impacts of the injection moulding process are due to electrical energy consumption, the use of mould and cooling water, and the fume emitted from molten plastics. Section 3 has focused on the energy consumption since it is the dominant factor for carbon footprint of injection moulding processes. The derived SEC can certainly contribute to the knowledge of the process dynamics in terms of energy efficiency.

Apart from energy consumption, the environmental impacts due to other factors are less significant. As discussed in Section 2.2, a long mould life is achievable. As a result, when allocating the embodied energy of the mould into 1 kg of injected part, the impacts become insignificant owing to the large production volume. Notably, the mould life span is also dependent on the product life cycle. With the trend towards shorter product life cycles, the challenges for flexible mould design grow, and the environmental impact due to mould production will no longer be negligible. Thus, a separate environmental analysis or LCA of the mould is recommended for further research.

The use of industrial water for cooling purpose has also limited environmental impacts, since the water was circulated through a closed channel for most of the conventional injection moulding machines. However, it may not be the case for the newer machines, for instance, water-assisted injection moulding, which directly applies water on the injected parts. Although this new machine can considerably reduce the cooling time, the water adhered to the injected parts may result in a much higher level of water consumption than the conventional ones. Thus, a future work is needed to investigate water consumptions of both injection moulding machines.

Forrest et al. (1995) conducted a 2-year study on emissions from processing thermoplastics. The real measurements lead to the conclusion that the emissions produced for a given process (e.g. injection moulding, blown film, extrusion) would be mainly dependent on the material. The studied case also suggested a low level of process fume, and all the individual chemical species detected were found to be present at concentrations significantly below the corresponding present occupational exposure limits. So the fume of molten plastics can be temporarily neglected in this research. Further research is recommended to validate the results of the emissions, and to compare the associated environmental impacts with other factors.

Compared with machining processes, injection moulding is material efficient in general. The scrap or waste material coming out of the process is mainly the flush and the runner which are used to connect multiple cavities. The flush can be avoided by applying correct process parameters such as clamping pressure, screw speed and so on. The amount of runner can be reduced by improving the mould design. Moreover, due to the nature of thermoplastics, the waste material can be recycled repeatedly (UNEP 2009). The environmental impact of the waste can be assessed with the help of LCA of thermoplastics. Due to the scope of the work, the waste material is currently excluded in this paper.

Based on the above discussions, the eco-efficiency can be simplified as an extension of energy efficiency for the case of injection moulding, where the energy consumption can be further characterised into different environmental impact categories. The derived SEC models for the injection moulding process represent the actual energy consumption for processing 1 g of plastics. Since the value of the injection moulding process can be quantified by the weight, the inverse of the SEC represents the eco-efficiency of injection moulding processes. In general, the higher the throughput rate, the more the plastics processed during a unit time; hence, the process consumes less energy, resulting in less environmental impacts.

The environmental impacts due to energy consumption can be quantified by using LCI database, Ecoinvent Version 2.2 (Ecoinvent Centre 2010). In order to simplify the analysis, the mid-point analysis was conducted and the CO₂ Fossil Fuel Emissions (CO₂ fossil) was selected for demonstration. Since the source and type of electricity generation determines the CO₂ emissions of 1 kWh used electricity, four different locations were selected for comparison: Australia, the USA, European Union and Brazil; and 1 kJ of electricity consumption equals to 0.000237, 0.000199, 0.000128 and 0.000018 kg fossil fuel CO₂ emissions, respectively. Figure 7 shows the relationship between CO₂ fossil emissions per gram of processed material and throughput rate in different regions. Comparing it with Figure 5, the curves of CO₂ emissions offset the trend of the derived SEC model. Due to the high carbon intensity of electricity generation, Australia results in the highest CO₂ fossil emissions among other countries. Brazil performs better than other major industrial countries due to the use of hydro energy. Notably, if the electrical energy is completely generated from renewable resources, the environmental impact of the injection moulding process would be negligible. However, current electricity generation still relies on coals and other non-renewable resources. Thus, the electricity consumption still dominates the environmental impacts in most of the world. In addition, a critical region can also be identified. Increasing throughput rate in the range from 0 to 0.3 g/s would result in a considerable reduction of CO₂ fossil emissions. In short, the critical region should be avoided if possible.

Figure 7 Environmental impact (CO₂ fossil) against throughput (BOY 15S).

Since some of the LCI database contains information regarding energy consumption of injection moulding processes, it is also worthwhile to compare the values from the LCI database with the real measurements obtained in this research.

According to the Ecoinvent database Version 2.2, the SEC of the injection moulding process is 5.328 kJ/g (or 1.48 kWh for processing 1 kg plastics) (Ecoinvent Centre 2010). However, the observed SEC showed a dynamic trend ranging between 3 to 30 kJ/g. Comparing it with the results in Table 4, Ecoinvent will suggest 5328 kJ, whereas the SEC models indicate a range of energy consumption from 5400 to 39,742 kJ depending on the process configuration. Hence, it is not safe to use an average data to estimate the energy consumption and the associated environmental impact of such a process. Alternatively, the derived SEC model (Equation (4)) can also be used during the design stage since it only requires input information about throughput rate and the mass of injected parts. These inputs can be easily estimated with the mould design. As a result, the energy consumption during the injection moulding period can be predicted in a more accurate way, which offers opportunities to improve the mould design in an early stage.

5. Conclusions

This paper has implemented the proposed empirical approach to characterise the energy efficiency of different injection moulding machine tools. The derived SEC models offer a unique ability to predict energy consumption of a given machine tool with an accuracy of over 90%. Throughput rate has been found as the decisive factor to determine the energy consumption of the unit process. In other words, the energy efficiency of the unit process can be improved by optimising the throughput rate of the process. More importantly, the derived models can certainly improve the quality of the current LCI database, which requires minimal input information to estimate the energy consumption of injection moulding processes. The eco-efficiency of the injection moulding process has also been discussed in this paper. The inverse of SEC models can be directly used for the eco-efficiency evaluation, since injection moulding processes mainly consume electricity in a dynamic manner. Thus, the eco-efficiency can be improved by increasing the throughput rate or by using the electricity generated from renewable resources.

Acknowledgement

The generosity of Sydney TAFE for providing the machine tools is also appreciated.