Niclas Norman Henrichs¹, Jan T. Sehrt², Thomas Brinkmann² and Daniel Michael Berger²

¹Impetus Plastics Engineering GmbH, Germany
²Ruhr-Universität Bochum, Germany

Additive manufacturing using material extrusion enables both the fast and cost-effective production of functional prototypes (rapid prototyping) and ready-to-use plastic products (rapid manufacturing). In addition to the processing of plastic wires, also known as filament, the direct processing of plastic granulate using the Fused Granular Fabrication process is also a practicable method. In this case, the cost- intensive production of filament can be dispensed with, and an even wider range of materials can be processed. The design of the extrusion unit is similar to that of standard screw extruders, although a wide variety of different screw geometries can be found on the market. The motivation for the investigation in this paper is the design, development and construction of an extrusion unit with subsequent benchmarking of different geometries. A standard three-zone screw is used, which should make it possible to process the widest possible range of plastics. In addition to standard plastics (e.g. polypropylene), technical plastics (e.g. polyamide 6) are also used to analyse their processability and thus draw conclusions about the usability of the extrusion unit. The aim of the investigations is to determine the production quality depending on the material to be processed. In addition to the general appearance, the focus is particular on the stringing behaviour of the materials to be examined. The results obtained will be used to determine the extent to which a standard three-zone screw is suitable for additive manufacturing using the Fused Granular Fabrication process. It can be determined that materials such as high impact polystyrene (HIPS), polylactide (PLA), acrylonitrile butadiene acrylic ester (ASA) and polypropylene (PP) exhibit little to no stringing, while polyamide 6 (PA) and polyethylene terephthalate glycol (PET-G) show higher stringing.

Keywords: Additive Manufacturing; Mechanical Simulation; Fused Granular Fabrication; Fused Filament Fabrication; Finite Element Method; Screw Extruder

The additive manufacturing of plastics using the Fused Layer Modelling process (FLM according to VDI 3405) belongs to the process category Material extrusion with thermal reaction bonding of polymers (MEX-TRB/P according to DIN EN ISO ASTM 52900). It is a versatile process to produce prototypes (rapid prototyping) and for small series production (rapid manufacturing). Due to the easy availability of a wide range of thermoplastic materials in the form of thin plastic wires (so-called filaments) the FLM process is the most widely used additive manufacturing process [1,2].

In addition to the Fused Layer Modelling process, the Fused Granular Fabrication process (FGF process for short) is also part of material extrusion in accordance with DIN EN ISO ASTM 52900. In this process, plastic granules are processed directly using a single-screw extruder instead of filaments [3]. This allows an even wider range of materials to be used than in the FLM process, as the user does not have to resort to alternative materials in filament form but can process desired materials directly.

Another disadvantage in the processing of filaments are the mostly undeclared commercial plastic types and additives that manufacturers use for the production of filaments [4]. This poses a particular problem when the specification of the exact materials is critical for authorization, such as in the case of medical authorization [5]. Medically authorized filaments are available on the market. However, at a price of approx. €50/kg, these are around 10-20 times more expensive than comparable medical plastics in granulate form [6]. The cost factor mentioned can also be found in standard plastics. For example, filaments made of ABS are sold for approx. €30/kg, while the material is offered in granulate form for around one-tenth of that price.

A various number of FGF machines are already available on the market. In addition to larger manufacturers such as Arburg (Freeformer), KraussMaffei (powerPrint), HAGE3D (PEX), Metrom (SEAMHex), Pollen (o2) or AIM3D (ExAM), smaller suppliers such as Fablab-München or Dyze Design have already made extrusion units or even entire systems available for the mass market. In this publication, a clear distinction will be made between the machines based on the output quantity per time unit. While large systems with an extrusion nozzle diameter > 1 mm can produce models the size of a chair or other furniture, compact systems with an extrusion nozzle diameter < 1 mm can also be used to precisely mould smaller plastic components.

The subject of this scientific study is FGF processes on compact systems. Large-scale systems serve a completely different market and are not in competition with the production of the detailed components of compact systems. For this reason, they will not be considered further in this study.
Since FLM machines are already available from around €150 (e.g. Creality Ender 3 V2) and good printing results can be achieved even in this price range, FGF machines with similar specifications such as material output and installation space, but also in terms of component quality, have to compete with lower-priced FLM machines.

The screw extruders on the market take different approaches to the selected screw geometry. Scientific approaches that deal with the design of screw extruders can be found, for example, in the work of J.-W. Tseng et al. [7] for the targeted processing of PEEK, Albert Curmi et al. [8] in the investigation of a particularly short extrusion screw for the processing of ABS material and Freitas et al. for the processing of PA12 and ε- PCL (ε-Caprolactam) powders [9]. In these studies, different screw geometries are investigated depending on the materials to be processed.

Although the three-zone screw is an industrial standard in the extrusion of plastics, many other different screw geometries are used in additive manufacturing within the FGF process [10]. Each of the screw geometries used has advantages and disadvantages and therefore offers different possibilities for processing various materials. A standardized guideline for selecting the screw geometry in the FGF process does not exist yet.

One of the difficulties of the FGF process is the controlled management of the melt discharge, which can lead to unsightly stringing during interruptions in extrusion. To explain this phenomenon, this section describes the difference between the retraction function of the FLM and FGF processes. In the FLM process, the filament wire is fed to the extruder with the associated hotend via a driven pair of friction rollers in the form of two drive wheels. These convey the not plasticized filament towards the heated hotend. In this process, the polymer first melts just above the extrusion nozzle and is then deposited onto the build platform in the form of a melt strand. If the path between two walls or components is bridged during the build process and no material is to be discharged, further material feeding can be avoided by turning the drive rollers backwards. This allows a stringing-free component to be produced for a variety of materials with a suitable set of parameters. In the FGF process, on the other hand, the plastic granulate is already melted in the screw chamber due to the high temperature and is plasticised in large quantities, i.e. in melt form. During the build process of the FGF process, the material is applied to the build platform in layers in the same way as the FLM process. However, with a travel path without material feed, it is not possible to withdraw the melt using the friction roller pair. For this reason, turning back the extrusion screw has become established in practice in order to prevent the plastic melt from running out unnecessarily or to reduce the amount discharged [11]. Nevertheless, this only prevents stringing for a limited number of materials.

The devices on the market all advertise with a list of the granulates to be processed. Some advertise micro (grain diameter < 1 mm) and mini granules (grain diameter between 1 mm and 2 mm), original extrusion or injection moulding granules (grain diameter > 2 mm). As it is therefore not possible to make a generally valid comparison between these devices, this publication will use a specially developed system to investigate the limits of processing original plastic pellets when using a cost-effective standard three-zone screw. The aim is to process a wide range of materials from original plastic granulate and to investigate which material types can be processed qualitatively using the FGF process and whether stringing can be effectively prevented during the process. A statement is also to be made as to whether and to what extent a standard three-zone screw is suitable for the FGF process. Finally, possible measures to improve component quality will be identified.

 

The development described in this paper is based on the VDI 2221 guideline [12]. In addition to the selection and design of a screw geometry and the design of the entire structure, a mechanical design of the entire extrusion unit is also carried out. Different drive arrangements and assembly options for the screw extruder are compared using the finite element method (FEM for short) with ANSYS. Particular attention will be paid to the feed area of the unit, as the jamming of pellets can exert high mechanical forces on the extrusion screw and the screw barrel. The extrusion unit is then mounted on a kinematic test platform and put into operation. In a final benchmarking test, original plastic pellets are benchmarked and the usability of the system is analyzed.

Boundary conditions of the extrusion unit

At the beginning of the project, the requirements for the design of the FGF system are defined in a specification sheet. The important criteria include:

• Low weight of the plasticizing unit
• Small installation space of the plasticizing unit
• Processing of original granulate
• Homogeneous melt formation and output
• Reproducibility of the process
• Easy to clean and maintain

The criteria mentioned are transferred into a list of requirements and then, where possible, given numerical values. These result from an analysis of comparable FLM systems. In particular, a nozzle diameter of 0.4 mm and a printing speed of 150 mm/s are important process parameters that must be defined in advance for the development. To enable subsequent research work to investigate larger nozzle diameters, the theoretically possible output rate of the extrusion unit to be developed is set correspondingly higher at 150 g/h. In comparison, the maximum output of a Prusa i3 using a 0.4 mm nozzle is approx. 54 g/h [13].

Screws with a diameter of approx. 14 mm are used in laboratory extruders [14]. The advantages of using a three- zone screw are high extrusion pressure, good degassing of the plastic melt and precise control of the melt flow by varying the screw speed [15]. A possible mechanical weak point in the design of extrusion screws is the feed zone. This zone has a reduced core diameter, as the screw flights must have a minimum size of the grain diameter of the plastic granulate to be processed. Care must also be taken to minimise the melt volume within the screw. Material damage can be caused by degradation of the plastic as a result of a too long throughput time of the melt [16].

An outer diameter of 13.8 mm with a core diameter of 6 mm is initially assumed for the extrusion screw. This results in a flight depth of 4 mm in the feed area of the screw. The L/D ratio plays an important role in determining the screw length. This ratio describes the relationship between the screw length and the screw diameter. In the literature, L/D ratios of approx. 18:1 can be found for small-scale extrusion screws [17]. Material manufacturers provide recommendations for their materials and their processability. However, these values refer to laboratory extruders or industrial extrusion/injection moulding systems with significantly higher output rates. Due to the low mass flow of max. 150 g/h, an L/D ratio of 13.9 is selected for the extrusion screw. A list of the remaining screw parameters can be found in Table 1.

Screw length	192 mm
Pitch t	14 mm
Pipe Spacing λ	100 µm
Flank thickness s	3 mm
Helix Angle ψ	15°
Core diameter metering zone	11.2 mm
Specific thread depth hF	3.9 mm
Specific thread depth hM	1.3 mm
Length of feeding zone	6.2 D
Length of compression zone	3.5 D
Length of metering zone	4.2 D

Table 1: Further parameters of the screw design.

Using the values from Table 1, the compression ratio Epsilon can be described as follows:

The formula results in a compression ratio of ε = 2.38. This is within the range of 2 to 3 suggested in the literature [7]. Figure 1 shows the extrusion screw.

 

 

 

Figure 1: Illustration of the used extrusion screw.

Construction and mechanical design of the extrusion unit:

Once the screw has been designed, a basic structure is developed for assembly. This should accommodate the screw with the associated cylinder and drive unit. Five different assembly concepts are compared for the basic structure using FEM in ANSYS. The aim is to compare which structure represents the best solution in terms of modularity and rigidity.

 

Figure 2: Mechanical setup for the FEM analysis.

The force F determined by the simulation is approx. 3 kN and is applied between the extrusion screw and the barrel in the subsequent load cases. It therefore acts linearly along the Z-axis. With this comparative model, the calculation time per concept can be reduced from approx. 30 h to less than 2 h per iteration. The expected quality of the results is considered sufficient, as the focus is on a comparative analysis of different assembly systems.

 

 

Figure 5: Comparison of the design (on the left) and the fully assembly (on the right) of concept e) Double-sided mounting of aluminum plates.

Kinematics and software used to operate the extrusion unit

The extrusion unit is mounted on an existing kinematic system. This works in a cartesian system and moves the extruder horizontally in the X and Y directions over the build platform, which in turn can move in the Z direction. A modified version of the Reprap firmware is used to control the production machine. It offers the option of controlling the temperature of the three 250 watt heating sleeves along the cylinder and the nozzle separately. Servomotors (Nema 23 size) are used for the X- and Y-axes. Nema 23 stepper motors are used for the Z-axis.

On the software side, the programme CURA from Ultimaker is used as a slicer. As this programme originates from the FLM area, the material flow is calculated in steps/ mm of filament. Due to the use of a screw extrusion unit, the value must be determined as a function of the material for the FGF extrusion unit.

Materials used in the experiments

Different thermoplastics are used for the benchmark tests. For an initial overview, a range of generic materials are obtained from Filament2Print [19]. The disadvantage of this platform is that the exact commercial grade is not defined and no information is given about the origin. However, this is considered sufficient for an initial feasibility analysis of the extrusion unit. In addition, a PP from Lyondell Basell (commercial grade: Purell RP378T) is used. The materials analysed and the processing temperatures used can be found in Table 2.

Material	Upper heating zone	Centre heating zone	Lower heating zone	Nozzle
Acrylonitrile- styrene-acrylic ester (ASA)	100°C	225°C	235°C	245°C
High-impact polystyrene (HIPS)	90°C	210°C	220°C	230°C
Polyethylene terephthalate-glycol (PET- G)	90°C	210°C	220°C	230°C
Polyamide-6 (PA6)	90°C	200°C	210°C	220°C
Polylactide (PLA)	90°C	180°C	190°C	200°C
Polypropylene (PP), Purell RP378T	90 °C	210 °C	220 °C	230 °C

Table 2: Materials used and their processing temperatures [20].

Analyzed benchmark geometries

The different materials mentioned are tested using two components. The first component is a plate with two towers (see Figure 6, left) to deliberately assess stringing due to insufficient material dosing. The test is used to determine the retraction length of the extrusion unit. The parameters obtained in this way are verified by manufacturing a small boat (called Benchy) [21], which, in addition to different overhangs, also tests the spanning of bridges (see Figure 6, right). This boat is a frequently used benchmark in material extrusion and other Additive Manufacturing processes. Due to its geometric properties, is well suited to making far-reaching statements about production quality [22,23]. The two components are covered with a thin film (so-called brim, see blue area in Figure 6) for better adhesion to the building platform across all materials. It can be removed if necessary and is not part of the assessment.

 

 

 

 

 

 

 

 

 

Figure 6: Illustration of the test components, plate with two towers on the left, ship with different overhangs and bridges on the right [20].

Determination of the extrusion parameters

Before the benchmark tests can be carried out, the material flow of the extrusion unit must be calibrated. The starting parameter for each material is set to 400 steps/mm on the screw extrusion unit. Then, using software from the FLM range, a fictional filament length of 500 mm must be extruded on the screw extrusion unit and then weighed using a precision balance and an average value calculated [Table 3]. The mean value is compared with the weight of 500 mm filament of the same polymer type and a new steps/mm value is then determined using the rule of three. The value calculated in this way serves as the starting point for fine adjustment of the material flow.

Measuring 1	Measuring 2	Measuring 3	Measuring 4	Measuring 5	Average	Standard deviation
1.69 g	1.71 g	1.68 g	1.68 g	1.69 g	1.69 g	0.012 g

Table 3: Discharge measurement and averaging of ASA [20].

Now a cube (edge length 20 mm) with only one wall is produced [Figure 7]. This means that under-extrusions (holes) and over-extrusions (irregularities in the outer wall structure) can be easily recognised, and fine adjustments can be made. The procedure is shown below as an example for the ASA material. In the further course of the work, only the steps/mm value determined for each material is stated.

With a weight of 1.28 g for 500 mm ASA filament and an average value of 1.69 g for the equivalent extrusion length of the ASA granulate, this results in a value of 308 steps/mm for the further fine calibration of the ASA granulate. At this value, there are clear discontinuities in the wall, as can be seen in Figure 7 on the left. By iteratively increasing the steps/mm, a value of 350 steps/mm can be determined [Figure 7, right].

Figure 7: Single-walled cube for determining under- or over-extrusion of ASA [20].

Description of the production parameters

In the following section of this publication, various production parameters are explained, depending on the material used. Table 4 shows the initial parameters for the different

Parameter	Value
Layer height	0.2 mm
Wall thickness	0.8 mm
Top and button layer	0.8 mm
Infill density	20%
Build plate temperature	90°C
Retraction length	10 mm
Printing speed	50 mm/s
Build plate adhesion	Brim
Part cooling	100%

Table 4: Initial parameters for the different benchmarks [20].

The process parameters shown in Table 4 serve as the initial state of the various benchmarks. If process parameters are adjusted, this is noted for the respective material.

An evaluation matrix is used to analyze and compare the different benchmarks. In the matrix, the production results achieved are evaluated in four different categories using a points system from 1 to 3 (1 = poor, 2 = moderate, 3 = good). The evaluation criteria are shown in Table 5.

Score	Stringing	Warping	Surface quality	Accuracy of the overhangs
3	No stringing	No warping	Good	No slack
2	Little stringing	Little warping	Moderate	Little slack
1	Heavy stringing	Heavy warping	Bad	Heavy slack

Table 5: Rating table for the benchmarks [20].

Acrylonitrile-styrene-acrylic ester (ASA)

The printing speed and the building platform temperature are adjusted for processing the ASA granulate [Table 6].

Process parameters	Value
Printing speed	60 mm/s
Build plate temperature	120°C
Steps/mm for the extrusion unit	350 Steps/mm

Table 6: Parameter adjustment for the processing of acrylonitrile-styrene-acrylic ester (ASA) for the Benchy [20].

Figure 8: Benchmarks with acrylonitrile- styrene-acrylic ester (ASA), a) Stringing test, b) Benchy, c) Benchy, focus on stringing [20].

The components produced show only a few defects [Figure 8]. Irregularities can be seen on the surface of both towers, which cannot be reduced by adjusting the process parameters. The Benchy only shows very slight stringing around the deck. The assessment by points can be found in Table 7.

Criteria	Score
Stringing	2
Warping	3
Surface quality	2
Accuracy of the overhangs	3

Table 7: Evaluation of the Benchy made of acrylonitrile-styrene-acrylic ester (ASA) [20].

High-Impact Polystyrene (HIPS)

For processing the HIPS granulate, the printing speed and part cooling are adjusted according to the findings from the preliminary tests [Table 8].

Process parameters	Value
Printing speed	60 mm/s
Part cooling	50%
Steps/mm for the extrusion unit	290 Steps/mm

Table 8: Parameter adjustment for processing high- impact polystyrene (HIPS) for the Benchy [20].

Figure 9: Benchmarks with high impact polystyrene (HIPS), a) Stringing test, b) Benchy, c) Benchy, focus on stringing [20].

The benchmarks in Figure 9 do not reveal any defects in the processing of the HIPS granulate. Neither the stringing test nor the Benchy show any stringing, nor are there any material sags on the walls or outer surfaces. No distortion of the component can be detected either. The overhangs and bends do not sag and are well mapped. The assessment by points can be found in Table 9.

Criteria	Score
Stringing	3
Warping	3
Surface quality	3
Accuracy of the overhangs	3

Table 9: Evaluation of the Benchy made of High Impact Polystyrene (HIPS) [20].

Polyethylene terephthalate-glycol (PET-G)

When processing PET-G, no better results were achieved than the following despite parameter changes. The best results were achieved by reducing the build platform temperature and increasing the feed length [Table 10].

Process parameters	Value
Build plate temperature	60°C
Retraction length	15 mm
Steps/mm for the extrusion unit	300 Steps/mm

Table 10: Parameter adjustment for the processing of Polyethylene terephthalate-glycol (PET- G) for the Benchy [20].