Design Variables and control levels

Ongoing research focused on studying layer manufacturing methods has described the possible sources of inaccuracies and the noise factors affecting the AM process [49]. The variables used in this experiment have been selected bearing in mind that they can potentially affect to the performance of all AM systems, independently of the type of AM processes category.

The sources of inaccuracies during the AM process are machine, geometry and process dependent. Therefore, the first design variable of this DOE, takes into consideration the machine factor of the manufacturing process. In this case, a selection of commercially available machines and their best suitable material choice was selected in this experiment.

Table 3 - Design Variables of the DOE and control levels

Design Variables Level 1 Level 2 Level 3

P1 Machine and Material M1 M2 M3

P2 Part Orientation Horizontal Vertical Diagonal

P3 Part Location Top Left Center Bottom Right

P4 Digital Quality High Medium Low

The other three design variables of the DOE, take into account the geometry and process dependency of the AM systems. The second and the third variable are related with the part orientation and part location over the build platform. The fourth and last variable focuses on

36 the effect of the quality of the digital data used in the AM process. The Table 3 makes an overview of the design variables and the control levels used during this experiment.

To define the levels of the machine and material design variable (P1), three different ASTM process categories were selected: Vat Photopolimerization process, Material Jetting process and Powder Bed Fusion process. Consequently, three different commercially available machines were included in the experiment.

Table 4 shows the selected AM systems and the comparison of the technical specifications of the machine as well as the basic technology involved in the process. At the same time, a comparison of the materials’ mechanical properties versus the average mechanical properties of an injection molded PC-ABS blend is displayed.

Table 4 – Design variable V1, Machines and materials technical specifications. Adapted from technical data sheets of the machine manufacturers

Machine Specifications M1 M2 M3

Machine supplier 3D Systems Stratasys EOS

Machine Type Viper SI2 Objet 500 Formiga P110

Industrial Process Category SLA Polyjet SLS

ASTM Process Category Vat

Photopolimerization Material Jetting Powder Bed fusion

Net Build size 250x250x250 490x390x201 200x250x330

Typical Accuracy

Material Properties Supplier Material Code

Test

ASTM D638 1920-2960 MPa 2690-3100 MPa 2600-3000 MPa N/A

ISO 527 2000-2810 MPa N/A N/A 1700 MPa

ASTM D790 1880-2750 MPa 1380-1660 MPa 1700-2200 MPa N/A

ISO 178 1990-2770 MPa N/A N/A 1240 MPa

37 The building orientation of the part over the build platform was selected as the second design variable (P2). The part was printed in three different orientations, horizontal build, vertical build and a diagonal building at an angle of 45^o.

Research has demonstrated that the building orientation has a significant impact over the final topology and properties of the manufactured part [10] [50]. Just by looking at the commercially available machine data sheets, it is possible to note that the suppliers report different accuracies depending upon the manufacturing axis, thus the building orientation on the build platform has an influence on the accuracy of the manufactured parts. Table 4, also displays the accuracy parameters of the machines used in this experiment.

The third design variable chosen for the DOE was the part location over the building platform of the AM machines (P3). The printing platform size between machines is different from machine to machine, thus the parts were located on the tray by using the top left corner, the center and the bottom right corner of the build platform of each machine. The location of the part over the building platform has an effect in the topology of the final produced part. The dimensional and geometrical tolerances of the AM parts produced are affected by this variable, especially in certain AM process categories where the effect of temperature gradients in the building chamber can be sources of inaccuracies. [51]

The last input variable included in this experimental approach aims to research the effect of the digital quality of the data in the AM process. The communication between the CAD and AM systems is built by using a file format called STL (Standard Tessellation Language), AM systems preprocessors have their own slicing algorithms which use STL data. This file format is the de facto standard used by CAD software to create triangular facets to approximate the topology of the digital model. The tessellated CAD model represents an approximated polygon mesh which tries to define the original geometry. This approximation is defined by the cordal error or deviation of the tessellation process. [52] [53]

Most common commercially available CAD software includes the possibility to create an STL file of the original model. In some common CAD packages this process is performed automatically without using any special tools, on the contrary more developed CAD solutions allow you to select parameters such as the deviation, error factors or the amount of triangular facets of the polygon mesh among others.

During this DOE, the quality of the digital data was evaluated by selecting three different deviations. The deviation of the STL file is defined as the maximum possible distance from the theoretical surface of CAD model to the created polygon mesh. The deviations were scaled as low, medium and high digital quality and their values were 0.1mm, 0.01mm and 0.001mm respectively.

The set of design variables chosen for this DOE aim to simulate a real manufacturing situation in which the company needs to define the most suitable combination of parameters in order to produce a part by using AM systems. Figure 9 shows a representation of a typical parameter selection process from the perspective of the manufacturing engineer.

38 Figure 9 - The design variables for the DOE and the overview of the process steps of a typical AM process

The process can be considered as a sequential process in which the manufacturing engineer first needs to decide the most suitable AM process and the material for producing the part, second the building orientation also needs to be defined to enhance the desired final properties of the part. After this selection, the positioning of the part over the build platform is decided and finally the digital data is transferred from the machine pre-processor in order to manufacture the part.