DESIGN RULES TO PREVENT THERMAL DISTORTION OF LARGE AREA COMPONENTS MANUFACTURED BY
SELECTIVE LASER MELTING
Lisa Riedel1, Sabrina Koss2, André Edelmann3, Ralf Hellmann4
1 B. Eng., Master student
2 M. Sc., research associate
3 Dr.-Ing., Head of centre of additive manufacturing
4 Prof. Dr., Head of applied laser and photonics group
1,2,3,4 Applied laser and photonics group, University of Applied Sciences Aschaffenburg, 63743 Aschaffenburg, Germany
ABSTRACT
Selective Laser Melting is an additive manufacturing technology which allows building nearly full dense components by using a metal powder. It is applicable in many industrial sectors, e.g. in medical engineering and aerospace. For the fabrication process, it is necessary to consider various influencing factors regarding the scanning strategy. Within the scope of this study, thermal distortion is discussed. Therefore, multiple aluminium alloy AlSi10Mg0.5 based specimens with the same dimensions yet having different orientations regarding the building platform are produced while applying different scanning strategies. The selectively laser melted parts are digitalised by a three-dimensional scanner and compared to the construction CAD-file. We find that fabrication with a part orientation being defined by the smallest area on the x-y-plane, the layer-plane, leads to the best conformity between generated and constructed part.
MOTIVATION
In additive manufacturing the material properties and the geometry of the building element are produced simultaneously which is possible due to the layer- wise build-up of volume. In the selective laser melting (SLM) process a thin layer of metal powder (20-100 µm) is applied on the whole platform where a laser beam selectively melts the powder particles together. Subsequently, the building platform is lowered by the layer thickness and a new powder layer is applied. This process repeats until the last layer of the component is completed [1]. One of the biggest advantages of using SLM is the possibility to design components with a high degree of geometrical complexity and a high relative density of more than 99 % [2,3]. In general, the SLM process is time-consuming but for highly individual products like implants SLM is faster than conventional assembly methods [4].
MultiScience - XXXI. microCAD International Multidisciplinary Scientific Conference University of Miskolc, Hungary, 20-21 April 2017
ISBN 978-963-358-132-2
DOI: 10.26649/musci.2017.021
This offers the possibility to build highly individual products within a short amount of time and without creating additional costs as it would be by using conventional assembly methods. In addition, due to the obtained geometrical freedom, entirely new components which are not producible by using conventional methods can be realised. All these benefits are applicable in industrial sectors like aerospace and medical engineering, e.g. for light-weight structures and implants [5].
One of the challenges in SLM fabrication is to overcome thermal distortion. For analysing this effect a large area component should be manufactured and compared to its construction file to see also minimal deviations. One occurring phenomena is that the corners of the components bend up. This is related to a very high temperature input into the workpiece during the SLM process while at the same time different cooling rates may occur within one particular material layer. For instance, the impact of thermal diffusion will be different for areas surrounded by molten material and those surrounded by adjacent powder. In addition, thermal diffusion will also be different into subjacent layers. As a result, different temperature gradients will occur layer-wise over geometry of a SLM generated part, particularly in areas with higher energy input. Outboard regions may cool down more efficiently and therefore contract faster than the inner regions, in turn leading to high residual stress and to wrapping effects [6].
EXPERIMENTAL PROCEDURE
To study the behaviour of the above described phenomena, specimens in different orientations and with different scanning strategies are produced. The dimensions of the specimens are 40 x 12 x 2 mm3 in cuboid shape. One of the specimens is oriented vertically (see Figure 1), two horizontally, with the vertically and one horizontally oriented having a line scanning strategy.
Figure 1
Arrangement of the specimens on the building platform Building platform Wiper
Horizontally oriented specimen with island strategy
Vertically oriented specimen with line strategy
Horizontally oriented specimen with line strategy
y x
Figure 2
Line (a) and island scanning strategy (b)
Line scanning strategy means, that the scanning happens in lines, as shown in Figure 2a. The second horizontally oriented specimen was divided in small squares of 4 x 4 mm2 which were scanned one by one. This is called island strategy and is shown in Figure 2b. Between these squares it is possible to choose an overlap, in such a way that the laser melts the edges of the squares twice. This is intended as to guarantee a sufficient connection of the edges. However, this also increases the energy input in these regions, in turn negatively influencing thermal distortion.
According to previous finding, that squares melt even without overlap, also for this study no overlap was adjusted. In addition, after every layer the scanning direction is rotated by 90°, i.e. the melting tracks change their direction (cf. Figure 2). This ensures that the melting tracks do not burn in, in turn allowing smoother surfaces.
The specimens were arranged on the building platform as shown in Figure 1.
The arrangement is important as spatters of the molten material from one specimen may affect the others. Therefore, the specimens were arranged with a high distance and displaced in y-direction (x-y-plane is the layer-area and the z-direction the building direction). The machine used to perform the experiments was a Realizer SLM 300i with a fibre laser (l= 1064 nm) and a maximum output power of Pl,max =1000 W. The powder used was the aluminium alloy AlSi10Mg0.5. In addition, the following parameters were applied.
Table 1 Applied parameters
Parameter Setting
laser current il 2200 mA layer thickness zlayer 50 µm exposure time tex 160 µs point distance xpoint 100 µm gas flow rargon Argon; 500 l/min
1st Scan 2nd Scan 1st Scan 2nd Scan
a) b)
To evaluate the SLM generated parts with respect to thermal distortion, the parts are digitalized using stripe light projection technology and compared to the original CAD-file via ATOS Professional V8 SR1 software.
RESULTS AND DISCUSSION
In general, all parts built show some effects of thermal distortion. However, those parts with horizontal orientation and island scanning strategy are affected most. Figure 3 summarizes the results of the optical 3D scanning and the comparison to the original CAD data set for the different orientations of the specimen and the scanning strategies. In particular, the right section of Figure 3 shows x-axis cuts of the specimens to help identifying the differences in height of the specimens. Apparently, the horizontally oriented specimen scanned by island strategy has the biggest difference in height of h2 = 1.91 mm, followed by the horizontally oriented specimen with line scanning strategy with h1 = 0.71 mm and, finally, the vertically oriented specimen with h3 = 0.35 mm. We attribute the large deviation of 1.91mm for the island scanning strategy to the significantly higher energy input due to the size of the scanned squares. In case of the line scanning strategy for the horizontally oriented specimen it appears that the longer cooling time between the individual tracks reduces the distortion.
In addition, it is found that those particular corners of the horizontal oriented parts with island scanning strategy are severely distorted where the laser has its starting position for every layer. This can be assigned to the deceasing amount of molten material for decreasing hatch distance [8]. The previously molten material can act as a heat sink for laser energy, because the rescanned material absorbs less incident laser energy [9], i.e. a small hatch distance leads to a high energy and thus thermal input which is yet also cooled down quickly. This in turn leads to residual stresses and ends up in thermal distortion. Hence, for a further optimization it appears to be preferential to vary the laser’s starting position after every layer.
Figure 3
Results of 3D-measurements and comparison with CAD file
Furthermore, it could be determined that the areas where the squares aren’t full- sized due to the size of the specimen – which is at the top and at the right side of the specimen – were also strongly arched towards the top. This indicates that smaller squares at the same laser power are more affected by thermal distortion than bigger ones. As it can be seen in Figure 3 also for the horizontally oriented specimen with line scanning strategy, the typical and expected effect of thermal distortion occurs.
A further similarity is that the corner where the laser started to scan is bent up the most. For the vertically oriented specimen, the best results could be recorded even though there was also a distortion observed. But it should be considered that there may be other reasons for the distortion. A thermal distortion is rather unlikely as the molten area was very small. More likely is that the distortion occurred because of the wiper while applying the powder. After every layer, it could slightly move the specimen which leads to a small distortion. This could be prevented by rotating the specimen with one of the corners to the wiper so that there is just a minimal area which is driven over by the wiper [7].
It is worthwhile to mention that the horizontally oriented specimen with line scanning strategy has the shortest building time, followed by the horizontally oriented specimen with island scanning strategy. The longest building time by far is needed for the vertically oriented specimen (Table 2).
Table 2
Building times for different specimens
Specimen Building time [h]
Horizontally oriented – line scanning strategy 00:36:56 Horizontally oriented – island scanning strategy 00:37:24 Vertically oriented – line scanning strategy 03:25:49
These differences in the total building time result from the need of the continuous repetition of the production cycle. For the horizontally oriented specimens, the total height with support structure was htotal1,2 = 7 mm, whereas for the vertically oriented it was htotal,3 = 45 mm. Thus, the horizontally oriented specimens had – by a layer thickness of zlayer = 50 μm – 140 layers, while the vertically oriented specimen had 900 layers.
CONCLUSION
We have studied the impact of component orientation and laser scanning strategy on the thermal distortion occurring during selective laser melting. The distortion has been analyzed by comparing the original CAD data set with the data acquired by 3D optical scanning using stripe light projection. Namely a vertical and a horizontal orientation with respect to the building platform have been chosen while a line scanning and an island scanning strategy were applied. As a result, a vertical orientation with line scanning appears to be preferential for achieving low thermal distortion. However, this orientation and scanning strategy is also associated with the by far longest production time of the selective laser melting process. For small sized components, also the horizontal orientation with line scanning strategy becomes attractive, since thermal distortion is reduced and the shorter production time is favorable.
REFERENCES
[1] GEBHARDT, A. (2013). Generative Fertigungsverfahren: Additive Manufacturing und 3D Drucken für Prototyping; Tooling; Produktion., 4th revised edition, Munich: Carl Hanser Fachbuchverlag.
[2] LU, Y. et al. (2015): Investigation of the microstructure, mechanical property and corrosion behavior of the selective laser melted CoCrW alloy for dental application. Material science & engineering. C, Biomimetic and supramolecular systems 49, 517-525.
[3] KRUTH, J.-P. et al. (2005): Rapid Manufacturing of Dental Prostheses by means of Selective Laser Sintering/Melting. Proceedings of the 2nd International Conference on Advanced Research in Virtual and Rapid Prototyping.
[4] XIN, X. et al. (2012). Corrosion characteristics of a selective laser melted Co-Cr dental alloy under physiological conditions. Journal of Materials Science, Volume 47, 4813-4820.
[5] CASAVOLA, C.; CAMPANELLI, S.L.; PAPPALETTERE, C. (2008):
Experimental analysis of residual stresses in the selective laser melting process. In the Proceedings of the XIth International Congress and Exposition, Orlando (USA).
[6] VAN BELLE, L.; BOYER, J.-C.; VANSTEENKISTE, G. (2013):
Investigation of residual stresses induced during the selective laser melting process. Key Engineering Materials 554-557, 1828-2834.
[7] HUSSEIN et al. (2013): Advanced lattice support structures for metal additive manufacturing. Journal of Materials Processing Technology 213, 1019-1026.
[8] TAYLOR, C.M. (2002): Morphology of direct SLS-processed stainless steel layers. Proceedings Solid Freeform Fabrication Symposium, Leeds, UK, 530-537.
[9] YADROITSEV, I.; SMUROV, I. (2011): Surface Morphology in Selective Laser Melting of Metal Powders. Physics Procedia 12, 264-270.
