LUT Metal Technology
BK10A0401 Bachelor’s thesis and seminar
Benchmarking of laser additive manufacturing process
Lappeenranta 03.10.2012 Ville Matilainen
TABLE OF CONTENTS
LITERATURE REVIEW ... 2
INTRODUCTION ... 2
1.1 PRINCIPLES OF LASER ADDITIVE MANUFACTURING (LAM) ... 3
1.2 Basics of CAD-processing ... 3
1.3 Basics of laser additive manufacturing ... 4
2 BENCHMARK MODEL ... 5
3 TEST METHODS ... 10
4 EXPERIMENTED PROCESSES ... 10
5 RESULTS AND DISCUSSION ... 13
5.1 Geometrical features and dimensional analysis ... 13
5.2 Mechanical properties ... 20
5.3 Reliability and economical aspects ... 24
6 AIM AND PURPOSE OF THIS STUDY ... 28
7 EXPERIMENTAL PROCEDURE ... 28
7.1 Materials used in this study... 29
7.2 Lasers used in this study ... 29
7.3 LAM machine used in this study ... 30
7.4 Analysis methods used in this study ... 31
8 RESULTS AND DISCUSSION ... 32
8.1 Geometrical features and dimensional analysis ... 35
9 CONCLUSIONS... 44
10 FURTHER STUDIES ... 45
REFERENCES ... 46 Appendix I Measurement data
Appendix II Diagrams of measurement results
LITERATURE REVIEW
INTRODUCTION
The laser additive manufacturing (LAM) has gained considerable interest in past few years. The biggest issue on this interest is that the quality of the laser additive manufactured parts is on such high level that the parts can be used in different industrial fields as functional parts. Also the possibility to create geometrically complex parts and working prototypes has raised the interest.
Laser additive manufacturing is a powder based process that allows designer to create parts that are hard or impossible to manufacture with conventional methods. Also the possibility to optimize part weight and strength is an advantage of this technology. Parts are built layer by layer as the laser beam melts the next layer on top of the previous one.
In this study the accuracy and quality of built parts is evaluated and measured. In literature review, the part accuracy and quality is studied from four different cases and the measurement and quality evaluation methods are applied in experimental part. In experimental part three sets of test pieces are manufactured from stainless steel. The manufactured parts are evaluated and measured in order to clarify the capabilities and resolution of state of the art LAM machine.
1.1 PRINCIPLES OF LASER ADDITIVE MANUFACTURING (LAM)
Laser additive manufacturing (LAM) process is layer-wise material addition technique which allows manufacturing of complex 3D parts by selective solidification of consecutive layers of powder material on top of each other. Solid structure is achieved by thermal energy supplied by focused and computer guided laser beam. The process produces almost full dense parts and they do not usually need any other post-processing than surface finishing. (Kruth et al., 2010, p. 1)
The advantages of laser additive manufacturing are geometrical freedom, mass customization and material flexibility. The laser additive manufacturing can be used for building visual concept models, customized medical parts and also tooling moulds, tooling inserts and functional parts with long term consistency. This manufacturing process can provide consistency over the products entire lifeline. It is also important that processes accuracy and ability to manufacture complex geometrical structures are on a good level. Also the process reliability, performance and economical aspects like production time and cost will play in a big role in order to use this manufacturing technique in mass production. (Kruth et al., 2005, p.1)
Of course this technology also has challenges what comes to the part building. First of all, the designer should know the limitations what this process have, for example that building overhanging features is very difficult. Also creating thin walls and small features is a challenge.
(Gibson et al. 2010 p. 705) Post processing of the parts should be considered when designing the geometry of the parts. For example, when building tooling inserts with internal cooling channels, the channels should be designed in that way, that it is possible to get the loose powder out from the channels. (Ilyas et al. 2009, p. 430)
1.2 Basics of CAD-processing
LAM manufacturing process starts with the creation of a 3D CAD-model of a desired object.
Then the 3D CAD-model is converted to STL-file. The STL-file defines optimal building direction of the physical object. The STL-model is based on small triangles, which determine the accuracy and contours of the whole object. When the STL-file is created and manipulated to correct orientation or size, the created model is sliced. Sliced model is then transferred to additive manufacturing machine, which now have the information how to build each layer.
Figure 1 shows the basics of model generation in laser additive manufacturing process. (Gibson, Rosen, Stucker, 2009, p.4-5, Gebhardt, 2003, p. 31)
Figure 1. Model generation in laser additive manufacturing process (Bineli, Peres, Jardin, Filho, 2011, p.3)
1.3 Basics of laser additive manufacturing
A thin layer of powder is spread across the build area using a powder-leveling blade in laser additive manufacturing process. The part building process is performed inside a chamber filled with nitrogen gas to minimize degradation and oxidation of the powder material. The powder and the building platform are preheated to minimize the required laser energy and to prevent warping and internal stresses caused by uneven heat distribution during the build. Figure 2 presents the principles of laser additive manufacturing process. (Gebhardt, 2003, p.31)
Figure 2. Principles of laser additive manufacturing process (Bineli, Peres, Jardin, Filho, 2011, p.2)
When a thin layer of preheated powder has been formed, a focused laser beam is directed onto the powder bed and moved by using galvanometric mirrors in scanner so that it melts the material and form a cross-section of the build part. Once one layer is completed, the build platform is lowered by one layer thickness and a new powder layer is applied by using the recoater blade. After that laser beam scans the next slice of the cross-section. This process is repeated until the whole part is built. Finally, the finished parts are removed from the build platform, loose powder is cleaned off and other finishing operations are performed, if necessary.
Finishing operations can include machining, sandblasting, sanding and polishing. (Gibson, Rosen, Stucker, 2009, p.103-105)
2 BENCHMARK MODEL
The benchmark models found from literature were usually designed to analyze the process accuracy and limitations. Process parameters can be optimized with assist of benchmark studies in order to manufacture parts with higher quality and accuracy. (Kruth et al. 2005, p. 3)
Kruth et al. studies benchmark models with mechanical features such as density, hardness, yield stress and Young’s modulus. Also the dimensional accuracy of the build features is studied. The
benchmark model is shown in figure 3. Dimensions of the benchmark model are 50 mm x 50 mm x 9 mm (Kruth et al., 2005, p. 3-4)
Figure 3. Benchmark model (Kruth et al., 2005, p.2)
The benchmark model consists of sloping plane and rounded corner (see figure 3), where the stair effect can be verified. It also have 2 mm thin plane where thermal distortions and warpage can be studied. The precision and resolution of the used process can be tested with small holes and cylinders (ranging from 0.5 to 5 mm diameters) and with thin walls (ranging from 0.25 to 1 mm thickness) as shown in figure 3. The sharp edges of the benchmark model can indicate the influence of heat accumulation and possible scanning errors. The overhanging surfaces can prove the ability to manufacture overhanging structures without support structures. All of the geometrical features also measure the accuracy of the process in x, y and z-direction. The solid half of the model is used for mechanical testing to figure out the mechanical properties of each benchmark model. (Kruth et al., 2005, p.2)
Another benchmark model is presented by Castillo. This benchmark model is designed and created specifically to study geometrical features and accuracy of different additive manufacturing processes. Benchmark model and its measurements are shown in Figure 4.
(Castillo, 2005, p.6)
Figure 4. Benchmark model (Castillo, 2005, p.6)
Main goal of benchmark model of Castillo is geometrical and mechanical properties. Castillo concentrates on the overhang features, heat distortions, building angles, and resolution in narrow and high details and also in small holes and thin walls. The mechanical properties of the manufactured models are also studied. Like Kruth et al, benchmark models of Castillo are manufactured by different additive manufacturing methods and compared to each other.
(Castillo, 2005, p.6-9)
In article by Vandenbroucke and Kruth, the benchmark pieces were manufactured to evaluate the suitability of laser additive manufactured parts in use in medical field. The benchmark pieces shown in figures 5 and 6, are designed for evaluating the process accuracy and the suitability to manufacture small details. (Vandenbroucke, Kruth, 2007, p. 198-199)
Figure 5. Benchmark piece for process accuracy evaluation (Vandenbroucke, Kruth, 2007, p. 199)
Figure 6. Benchmark piece for evaluating feasibility to build small features (Vandenbroucke, Kruth, 2007, p. 199)
Vandenbroucke and Kruth also studied surface roughness of manufactured pieces. In this study, the laser additive manufactured pieces mechanical properties, such as density, hardness and tensile strength were compared to bulk materials mechanical properties. (Vandenbroucke, Kruth, 2007, p. 199-200)
Geometrical accuracy is studied in article by A.L. Cooke and J.A. Soons. In this article, the benchmark pieces are manufactured with electron beam and laser additive manufacturing. The
test part is Aerospace Industries Association (AIA), National Aerospace Standard, NAS 979 standardized test piece. The test piece consist circle-diamond-square features, with inverted cone. The test piece is shown in figure 7 and 8. (Cooke, Soons, 2010, p. 2)
Figure 7. Geometrical features of benchmark piece (Cooke, Soons, 2010, p. 3)
Figure 8. Top view of the benchmark piece (Cooke, Soons, 2010, p. 3)
This test piece is originally designed for testing CNC-machines performance and machining quality. This piece is measured for its features size, flatness, squareness, parallelism and surface
finish. The test part is not designed to highlight errors of AM systems, but it is used for determine what kind of geometrical errors AM parts have. (Cooke, Soons, 2010, p. 2)
3 TEST METHODS
The dimensional accuracy was tested with digital caliper and the measurements were averaged after five times measurements. Also the build quality of the geometrical features, such as thin walls, small holes, small cylinders, overhanging features and build angles are evaluated visually.
Surface roughness is measured with surface roughness meter and solid part of the model is mechanically tested. In article by Castillo, the benchmark pieces are also scanned and the scanned data is compared to original STL-file. (Kruth et al., 2005, p. 2, Castillo, 2005, p. 19) The benchmark pieces in article by Cooke and Soons are measured by using coordinate measuring machine. The measurement data is compared to test pieces original STL-file. (Cooke, Soons, 2010, p. 3-4)
4 EXPERIMENTED PROCESSES
The benchmark pieces were manufactured by using five different SLS/SLM machine in study of Kruth et al. Machines differ in respect of laser source, optics, powder deposition, scanning equipment and environmental control system. The process parameters are optimized for each process depending on the applied binding mechanism and used powder material. Table 1 shows specifications of the used SLS/SLM processes. (Kruth, et al. 2005, p.2)
Table 1. Specifications of experimented processes. (Kruth et al. 2005, p.2)
Nr Equipment Binding mechanism Material Laser
power
Layer thickness
1 3D Systems DTM
Liquid phase sintering (polymer binder)
Polymer coated
stainless steel 10 W 80 µm 2 Concept Laser Full melting Hot work tool steel 100 W 30 µm
3 Trumpf Full melting Stainless steel 316L ≤ 200 W 50 µm
4 MCP-HEK Full melting Stainless steel 316L 100 W 50 µm
5 EOS Partial melting Bronze based 221 W 20 µm
The equipment and materials used by Castillo are shown in Table 2.
Table 2. Specifications of experimented processes (Castillo, 2005, p.10)
Nr Equipment Binding mechanism Material Layer thickness
1 Concept Laser M3 Linear Full melting Stainless steel 316L 30 µm
2 MCP Realizer SLM Full melting Stainless steel 316L 75 µm
3 ProMetal R2 Liquid phase sintering
(resin binder) S4 (60% Stainless steel+40% bronze) 100 µm
4 Trumpf Trumaform 250 Full melting TiAl 6 V4 50 µm
Studied processes vary by binding mechanism of the used powder (see table 1 and 2). Binding mechanisms are: liquid phase sintering, full melting, and partial melting. Polymer or resin coating of the powder grains act as a binder in liquid phase sintering. This coating liquefies by the laser beam and binds the stainless steel grains. As an additional step, this process needs a furnace cycle in where the polymer is burnt out and the part is infiltrated with molten bronze by using the capillary effect to reach full density. The second binding method is partial melting, which does not have a clear distinction between the binding material and the structural material.
Molten and non-molten areas can be noticed after fusing the powder. The third binding mechanism is full melting where the metal powder is melted completely with laser beam. This creates nearly full dense structures and minimizes the lengthy post processing steps. (Kruth et al.
2005, p. 3, Castillo, 2005, p. 4-5)
The benchmark pieces in article by Vandenbroucke and Kruth, were manufactured with M3 Linear machine by Concept Laser GmbH. The system was equipped with Nd:YAG laser with maximum power of 95 Watts. The materials which were used were titanium alloy and cobalt- chromium alloy. The titanium material was commercial powder, but the cobalt-chromium was self-made with induction melting gas atomization process. Table 3 illustrates the optimized process parameters for these materials. (Vandenbroucke, Kruth, 2007, p. 197)
Table 3. Optimized process parameters (Vandenbouck, Kruth, 2007, p. 200)
Material Ti-6Al-4V Co-Cr-Mo Melting temperature 1650 °C 1330 °C
Laser power 95 W 95 W
Layer thickness 30 µm 40 µm Scan speed 125 mm/s 200 mm/s
Overlap 35 % 30 %
Energy density 195 J/mm3 85 J/mm3 Build rate 1.8 cm/h 4.0 cm/h Part density >99.8 % >99.9 %
Since the titanium powder is highly reactive with oxygen, nitrogen and hydrogen, the building process was carried out with continuous flushes with argon gas. The cobalt-chromium powder was processed in nitrogen environment. Because of the physical properties of the cobalt- chromium material, the process was easier to control. That is also why the cobalt-chromium material has higher density and build rate. (Vandenbrouck, Kruth, 2007, p. 196,199)
The fourth benchmark piece is manufactured by laser additive manufacturing and with electron beam additive manufacturing. Table 3 introduces the specifications of experimented processes.
(Cooke, Soons, 2010, p. 3)
Table 4. Specification of experimented processes (Cooke, Soons, 2010, p. 3)
Process Test Part Material Number of Parts Layer thickness
E-Beam
A1 Ti-6Al-4V 1 100 µm
A2 Ti-6Al-4V 1 100 µm
A3 Ti-6Al-4V 1 100 µm
A4 Ti-6Al-4V 2 100 µm
Laser
B1 17-4 Stainless steel 1 40 µm
B2 17-4 Stainless steel 2 30 µm
B3 17-4 Stainless steel 1 20 µm
B4 17-4 Stainless steel 2 20 µm
Five test pieces were manufactured with electron beam additive manufacturing and other five were manufactured with laser additive manufacturing in article by Cooke and Soons. All the
parts were manufactured by service providers who were instructed not to use any post processing. The test part number B4 was solid piece and the others were built as hollow shells with wall thickness of 3 mm. All the test parts were removed from the building platform and therefore they might include some deformation caused by removal. (Cooke, Soons, 2010, p. 3)
5 RESULTS AND DISCUSSION
Main goal in studies represented in this paper has been to understand the possibilities and limitations of laser additive manufacturing processes and to find potential manufacturing applications, not comparing the techniques to each other. The main objectives of these studies are to test build quality of geometrical features like overhangs, small holes and thin walls, mechanical properties like hardness, tensile strength and densities of the parts. The dimensional accuracy is also studied as well as the economic aspects such as production time and costs of the processes. (Kruth et al. 2005, p. 3, Castillo, 2005, p.1, Cooke, Soons, 2010)
5.1 Geometrical features and dimensional analysis
The requirements of process accuracy increases all the time and it is a demand to build high quality geometrical features. Process parameters have an effect straight on the build quality and accuracy of the parts. (Kruth et al. 2005, p. 3, Castillo, 2005, p. 19)
Kruth et al. found that geometrical resolution with small hole diameter was 0.5 mm. That means that it is not possible to build holes with diameter of 0.5 mm or less. This is because of the loose powder is also melting by the surrounding heat. Minimum wall thickness is dependent on laser spot size, which means that the thinnest wall that can be built has thickness equal to the diameter of laser spot. The cylindrical features smaller than 0.5 mm cannot be built because of the strength of small features are insufficient to resist forces of the powder spreading recoater.
Figure 9 shows different geometrical features of the benchmark pieces. (Kruth et al., 2005, p. 3)
Figure 9. Examples of geometrical features of benchmark pieces (Kruth et al., 2005, p. 3)
As it can be seen from figure 9 section a, the upper picture shows good build quality of cylindrical features and sloping features, whereas the lower shows bad build quality. Section b of figure 9 reveals good build quality of thin walls and holes. Section c shows the good and bad build quality of sharp corners, where section d shows good and bad quality of overhang features.
(Kruth et al., 2005, p. 3)
The dimensional analysis was made to study the accuracy of the used processes. The dimensional analysis is shown in table 5. (Kruth et al., 2005, p. 3)
EOS
Trumpf
Trumpf
Trumpf
Concept Laser
3D Systems
3D Systems
MCP- HEK
Table 5. Dimensional analysis of benchmark pieces (Kruth et al. 2005, p. 3)
Nominal
dimension 3D Systems DTM
Concept
Laser Trumpf MCP-HEK EOS
Length 50 mm
50.59 50.08 50.12 50.78 50.16
Width
50 mm 50.22 50.09 50.11 50.73 50.18
Height
7 mm 7.05 6.96 7.12 7.12 7.03
Hole 1
Ø 5 mm 5.03 4.87 4.84 4.67 4.83
Hole 2
Ø 2 mm 1.96 1.97 1.95 1.72 1.77
Hole 3
Ø 1 mm Badly built Badly built 0.95 Badly built 0.90 Hole 4
Ø 0.5 mm Not built Not built Not built Not built Not built Cylinder 1
Ø 5 mm 4.95 5.03 4.96 5.35 5.12
Cylinder 2
Ø 2 mm 1.90 2.05 1.97 2.23 2.10
Cylinder 3
Ø 1 mm 0.97 1.06 1.02 1.25 1.12
Cylinder 4
Ø 0.5 mm Not built Not built Badly built 0.64 0.63 Wall 1
1 mm 0.97 1.23 1.04 1.34 1.16
Wall 2
0.5 mm 0.71 Not built 0.55 0.76 0.68
Wall 3
0.25 mm Not built Not built 0.33 0.47 0.47 Stair
effect Bad Good Bad Bad Good
Curling Good Good Bad Good Good
Sharp
corners Too Short Good Too short Too short Good Overhangs Good Badly built Badly built Badly built Badly built
As it can be observed from table 5, the features like holes and cylinders are very difficult to build if the diameter is less than 1 mm. The stair effect on sloping planes and rounded corners varies with different manufacturing processes. The stair effect generates from the layer thickness which the process uses. A solution to decrease the stair effect is to decrease also the layer thickness.
Layer thickness is dependent on powder particle size which also gives some restrictions to part resolution. (Kruth et al. 2005, p. 4)
The thermal stresses and curling of thin sections of the benchmark piece is avoided by using full supports for supporting and attaching the part to the building platform. Only one experimented process suffers from warpage and it is because the support structure was fine grid-like support instead of full dense support. Also building of sharp corners is a challenge. The correct scan path of the laser beam is crucial in order to create sharp corners. Otherwise the heat accumulates to the tips of the corners and the corners build up too short. The overhanging features are difficult to build in high quality. Near horizontal overhanging surfaces are very difficult to build without support structures. Since the melted powder material have no support from below, the molten material flows down and the bottom surfaces are not finished well as seen in figure 9 section d.
(Kruth et al. 2005, p. 4)
In benchmark study by Castillo, the same kinds of building defects were found. The overhanging features were usually badly built and most of the benchmark pieces included support structures in overhanging features. In figure 10 and 11 is shown details of one benchmark piece. (Castillo, 2005, p. 18)
Figure 10. Overhanging features with support structures (Castillo, 2005, p. 18)
Figure 11. Details of benchmark piece (Castillo, 2005, p. 18)
As it can be seen from figure 10 the overhanging features need support structures in order to be built. Even though this benchmark piece includes the support structures, the overhanging features and features that are built in angle suffer thickness defects and warpage. The figure 11 presents that the smallest features like cylinders and square tubes are missing. It can also be seen that the smallest holes are not built. (Castillo, 2005, p. 17-18)
In article by Vandenbroucke and Kruth the feasibility to build small and complex features was similar than what presented earlier. In this study was also found that the overhanging features were built badly when they are not supported. The two main reasons for that are that the laser beam scans in loose powder instead of solid material. Therefore the thermal conductivity decreases and the temperature increases which leads to instable melt pools. The second reason is that the instable melt pools form stalactite patterns because the molten material sinks in the loose powder by the effect of gravity. This effect is always in building process where horizontal holes are tried to manufacture. Also the small holes of diameter 0.5 mm were not built because the enclosed loose powder melts because of the surrounding heat. Otherwise the build quality of the manufactured pieces was in a good level and it was concluded that the parts made by this method
were accurate enough to use in most medical and dental applications. (Vandenbroucke, Kruth, 2007, p. 200-202)
In article by Cooke and Soons the dimensional analysis is performed with coordinate measuring machine and the measurements are compared to the STL-file of the piece. The first comparison is made with checking the circularity and size of the circle feature. Figure 12 shows the feature that is compared. (Cooke, Soons, 2010, p. 4)
Figure 12. The circular feature is measured and compared to STL-file (Cooke, Soons, 2010, p. 4)
The circular feature shown in figure 12 was measured three times at different heights of the feature. Results of these measurements are shown in table 6. (Cooke, Soons, 2010, p. 4)
Table 6. Mean radius error and circularity of built feature (Cooke, Soons, 2010, p. 4) Process Test Part Mean Radius
Error (mm)
Circularity Peak to Valley (mm)
E-Beam
A1 -0.130 0.322
A2 0.157 0.249
A3 -0.602 0.214
A4 -0.095 0.352
Laser
B1 -0.084 0.156
B2 0.043 0.094
B3 0.006 0.095
B4 -0.010 0.121
As it can be observed from table 6 the mean radius error is much smaller range in laser additive manufacturing than electron-beam additive manufacturing process. Since these two manufacturing processes are not comparable, it can be said that both of these include errors. The build errors can be seen better from figure 13, which shows the form and size errors of two parts made with same manufacturing process and same process parameters. (Cooke, Soons, 2010, p. 4- 5)
Figure 13. The average of three error traces around the circular feature of two parts made by the same system. (a) Electron-beam system and (b) Laser additive manufacturing system (Cooke, Soons, 2010, p. 5)
As it can be seen from figure 13, the error traces show that the error is repeating itself on both measured parts. This repeatability gives chances to compensate the parameters and optimize the building process in order to achieve better build quality and accuracy. (Cooke, Soons, 2010, p. 5)
5.2 Mechanical properties
When discussing about the mechanical properties of laser additive manufactured parts, the following issues should be taken to account. First of all the densities of the manufactured parts, the surface quality of the parts and of course the mechanical properties which include for example hardness, yield strength and tensile strength. (Kruth et al., 2010, p. 2-3)
The density issue, when using laser additive manufacturing method is the first and perhaps the most important thing is this process. The density of the manufactured part determines the mechanical properties of the finished part. This of course has an effect on the components performance when it is used as a functional part. Of course the objective is to produce 100%
dense parts, but this is not so simple to achieve since there is no mechanical pressure which would press the powder layers against each other, only gravity, temperature effects and capillary force are affecting into the build part. When the laser beam has melted the powder material the
solidification starts, and since the solidification happens rapidly, gas bubbles may get entrapped in the material. These entrapped gas bubbles forms pores to the build structure which may cause a failure when the part is under stress. In figure 14 is shown cross-section of laser additive manufactured part. (Kruth et al., 2010, p. 2)
Figure 14. Micrographs of cross-section of laser additive manufactured part with 4x magnification and 10 x magnification (Kruth et al. 2010, p. 2)
As it can be observed from figure 14, the small black spots throughout the part are pores which are formed during the manufacturing. The gas bubbles during melting and solidification is not the only factor that can cause low density into manufactured parts. The insufficient surface quality also effects on density. If the surface is very rough, the roughness peaks and valleys prevent the recoater to distribute even, homogenous powder layer. This causes variation in layer thickness and since the thickness is not even, the used laser power may not be high enough to melt the new layer completely. (Kruth et al., 2010, p. 2)
In study by Kruth et al. (2005) the mechanical properties of the manufactured benchmark pieces were measured and the measured values were compared to manufacturers stated values. The
10x magnification 4x magnification
densities of these parts were measured with Archimedes principle by weighing the samples in air and subsequently in ethanol to measure and calculate the densities. Table 7 shows the mechanical analysis of the benchmark pieces. (Kruth et al., 2005, p. 2, 4)
Table 7. Mechanical analysis of benchmark pieces (Kruth et al., 2005, p. 4) Mechanical property 3D Systems
DTM
Concept
Laser Trumpf MCP-
HEK EOS
Density (kg/m3) measured 7750 8025 7870 7900 7650
µ-hardness (HV0.100)
material data sheet value 187 420 202 - 117
µ-hardness (HV0.100)
measured 176 ± 10 398 ± 12 251 ± 6 233 ± 5 185 ±
20 Young’s modulus (GPa)
material data sheet value 137 - - - 80
Young’s modulus (GPa)
measured 37 62 49 54 30
Yield strength (MPa)
material data sheet value 305 1000 500 - 400
Yield strength (MPa)
measured 218 1410 535 598 320
As it can be seen from table 7, all these additive manufacturing processes can produce near full dense parts. The micro-hardness is rather high for all processes since the melt pool cools down very rapidly after laser beam has passed the area. The Young’s modulus is measured in this article by bending test and that is the possible reason why measured values are much lower than stated ones. The measured yield strength values differ slightly from the stated values, but they still fulfill the strength requirements for manufacturing. (Kruth et al., 2005, p. 4)
The mechanical properties such as hardness and tensile strength were also measured in article by Castillo. The results of these measurements are shown in table 8. The Vickers hardness tests were performed according to EN-ISO 6507-1/98 standard. (Castillo, 2005, p. 28-29)
Table 8. Mechanical properties of benchmark pieces (Castillo, 2005, p. 29) Mechanical property Concept Laser
Laser Cusing
MCP SLM
ProMetal 3D Printing
Trumpf Laser Forming Hardness (HV)
measured 232.6 212.4 256.8 415.6
Hardness (HV)
material data sheet value 230 237 260-300 420
Tensile strength (MPa)
measured 648.76 626.82 358.73 1165.42
Tensile strength (MPa)
material data sheet value 650 627 682 1080-1090
Elongation at break
measured 30.52 20.84 0.56 3.59
Elongation at break
material data sheet value 25 24 2.3 5.8-6.2
As it can be observed from table 8 the measured values does not differ much from the manufacturers material data sheets. The higher tensile strength and hardness of Trumpf Laser Forming is because the material used with this process was titanium alloy. The tensile strength of the Pro Metal manufactured part is quite low compared to other processes because the manufacturing process differs from full melting processes. In this article, the densities of manufactured pieces were not measured. (Castillo, 2005, p. 29-30)
In article by Vandenbroucke and Kruth, the mechanical properties of the titanium alloy manufactured pieces were compared to similar bulk materials. The bulk materials were Ti-6Al- 4V alloys, where the other one was annealed and the other was solution treated aged (STA). The laser additive manufactured part titanium alloy material links up best to the STA treated titanium alloy. In table 9 is presented comparison of mechanical properties between laser additive manufactured parts and bulk materials. (Vandenbroucke, Kruth, 2007, p. 200)
Table 9. Comparison of mechanical properties between LAM manufactured part and bulk materials (Vandenbroucke, Kruth, 2007, p. 201)
Mechanical property SLM Ti-6Al-4V
Annealed Ti-6Al-4V (Bulk material)
Solution treated aged Ti-6Al-4V (Bulk material)
Density (kg/m3) 4420 4430 4430
Hardness (HV) 410 (micro)
400 (macro) 350 395
Young’s modulus
(GPa) 94 110 110
Tensile yield strength
(MPa) 1125 920 1100
Ultimate tensile strength (MPa) 1250 1000 1200
Elongation at rupture
(%) 6 12 10
Young’s modulus (bending) (GPa) 93 110 110
Bending yield strength (MPa) 1900 1500 1800
Ultimate bending strength (MPa) 2000 1900 2050
As it can be seen from table 9, the mechanical properties of laser additive manufactured parts can achieve similar mechanical properties than bulk materials. The hardness of laser additive manufactured part is higher because during the build process the melt pool solidificates and cools down very rapidly after the laser beam has passed. The elongation at rupture differs from bulk materials because the ductility of LAM parts is lower than bulk materials. This is because there encounters slight embrittlement due to the laser melting. (Vandenbroucke, Kruth, 2007, p.
200)
5.3 Reliability and economical aspects
The possibility to use lased additive manufacturing in industrial applications is not only dependent on geometrical and mechanical properties of the parts, but also on reliability to produce different parts and minimizing the production time and costs. As it can be seen from all
of previously presented studies, the geometrical and mechanical properties of manufactured parts are on such level that the parts can be used in different industrial fields. (Kruth et al. 2005, p. 5) Production time of laser additive manufacturing includes the machine and file preparation and actual part build time. The actual part build time consists of powder deposition and the laser scan time. The total build time is highly dependent on build height and the layer thickness. The higher the part is the longer the build time is. The laser scan time is increased when there is geometrical complexity and large areas to expose. The building times, of benchmark pieces shown in figure 3, are presented in table 10. (Kruth et al. 2005, p. 5)
Table 10. Building time of benchmark pieces (Kruth et al. 2005, p. 5)
3D Systems DTM Concept Laser Trumpf MCP-HEK EOS
27 h 9 h 4.5 h 8.5 h 4.5 h
The build times seems long compared to conventionally manufactured parts, but the freedom to build complex parts with internal structures gives advantages to this manufacturing method.
Since those manufacturing times does not require manpower the costs are mainly determined by machine costs. Also it must be remembered that if, more parts are manufactured during the same build session, the production cost will decrease for each part. (Kruth et al. 2005, p. 5)
In article by Castillo, the building time of benchmark pieces was considered as the total amount of machine hours and secondary operations, not including finishing processes. Post processes such as support removal and removal from the building platform is not considered either.
Building time of manufactured parts is shown in table 11. (Castillo, 2005, p. 27)
Table 11. Building time of benchmark pieces (Castillo, 2005, p. 27) ProMetal MCP Concept Laser Trumpf
65 h 38 h 60 h 39 h
As it can be observed from the table 11, it takes much more time to build higher and complicated benchmark model compared to one from article by Kruth et al. If the post processing times were included, the total manufacturing time of Castillo’s benchmark pieces would increase dramatically.
In article by Vandenbroucke and Kruth, the building time of benchmark pieces was not supported but the build time of dental prostheses framework was presented. The dental prostheses framework is shown in figure 15. These frameworks are built from either titanium or cobalt-chromium powder. (Vandenbroucke, Kruth, 2007, p. 202)
Figure 15. Dental prostheses framework (Vandenbroucke, Kruth, 2007, p. 203)
The building time per prostheses framework versus number of produced frameworks are presented in figure 16. (Vandenbroucke, Kruth, 2007, p. 203)
Figure 16. Build time of prostheses frameworks (Vandenbroucke, Kruth, 2007, p. 203) As it can be observed from figure 16, the build time per part is decreasing when the number of frameworks is increasing during the same build. This is because the powder depositing time is
spread over all frameworks. Figure 16 shows that if eight frameworks are built during one production run, the total build time is 16 hours or 2 hours per framework. This is almost half of the time compared to building only one framework. As it can be seen from figure 16, it is more economical to build as many parts as possible during on production run. (Vandenbroucke, Kruth, 2007, p. 202)
EXPERIMENTAL PART
6 AIM AND PURPOSE OF THIS STUDY
Aim of this study was to determine the limitations of laser additive manufacturing process resolution and to determine the quality of parts build with state of art LAM machine. It was decided to manufacture test pieces with different features such as thin walls, small gaps, and holes which create overhangs. Test pieces were manufactured with prototype LAM machine, similar to EOS EOSINT M-series machines. The laser used in this study was IPG 200 W continuous wave fiber laser. All of the test pieces were manufactured from stainless steel powder.
7 EXPERIMENTAL PROCEDURE
The built pieces included different features and all the pieces were manufactured in the way that the pieces were placed parallel and in 45º angle against the recoater. The features of different parts were measured with microscope software and the measurements were compared to 3D models dimensions. All the parts were built with 5 mm solid supports, which were removed from the parts when sawed off from the building platform. All of test pieces were manufactured with building parameters optimized by machine manufacturer. In figure 17 is shown 3D models of built parts.
Figure 17. 3D models of test pieces
7.1 Materials used in this study
Material used in this study was EOS StainlessSteel PH1 Stainless steel powder. The chemical composition of the material is shown in table 12.
Table 12. Material composition of EOS PH1 (EOS Material data sheet)
Material Concentration
Iron < 80 % Manganese < 1 % Molybdenum < 0.5 %
Nickel < 6 % Niobium < 0.5 %
Silicon < 1%
Carbon < 0.1 % Chromium < 20 % Copper < 6 %
This fine powder is a pre-alloyed stainless steel and it is characterized by having good corrosion resistant and excellent mechanical properties. This type of steel material is widely used in engineering applications where high hardness, strength and corrosion resistance is required.
Suitable applications can be found for example from medical or aerospace industries. Also this material can be used for building functional prototypes, small series products and individually designed products and spare parts. With standard processing parameters it is possible to use layer thicknesses 20 µm and 40 µm. Mechanical properties of built parts are fairly uniform in all directions when using standard process parameters. (EOS Material data sheet)
7.2 Lasers used in this study
The used laser in this study was 200 W IPG YLS-200-SM-CW fiber laser. The laser beam is transferred from the laser source to the galvanometric scanner via optical fiber. The used laser produces 200 W of power at a wavelength of 1070 nm and the focal length is 400 mm. The spot size of the laser beam is 70-100 µm.
7.3 LAM machine used in this study
Laser additive manufacturing machine used in this study consists of laser unit, process chamber and process control computer. It is possible to do adjustments to the process with computer. The LAM machine itself is a prototype machine similar to commercial EOSINT M-series equipment.
The LAM equipment is presented in figure 18.
Figure 18. Prototype LAM equipment, similar to EOSINT M-series equipment
The building process takes place in the process chamber and the process is controlled by controller computer. The building chamber is divided into three platforms, where the middle one is the platform where the parts are built. The building chamber is presented in figure 19.
Figure 19. Process chamber of the LAM machine
The other two platforms serve as powder dispenser platform and as collector platform where the extra powder is collected. The powder spreading is done with recoater which spreads the powder evenly to the building platform. The building platform is heated with thermo elements. The chamber is filled with nitrogen gas to decrease the oxygen level of the chamber atmosphere. The level of nitrogen is 99.8 % during the process. Nitrogen works as a protective gas and it helps to avoid oxidation of the stainless steel parts.
7.4 Analysis methods used in this study
The manufactured pieces were analyzed and evaluated visually. Dimensional analyses are performed to check process accuracy. The dimensional analysis is made with Axio Vision microscopy software. Measurements were taken multiple times and averaged. Geometrical features such as thin walls, narrow gaps and small holes were evaluated visually.
8 RESULTS AND DISCUSSION
In this study, the evaluation of geometrical features and process accuracy are the main goals.
Specially building small features like holes, walls and gaps are investigated. This study is also meant to give limitations and directions of what kind of parts or features can be built. All measurement data is presented in appendix. In figure 20 is presented test pieces 1A and 1B.
Figure 20. Test pieces 1A and 1B with various wall thicknesses. Arrow shows the direction of the recoater movement.
As it can be seen from figure 20, the direction of recoater movement has very big impact on part build quality. The recoater does the recoating with such big power that the thinnest walls collapse if the part is oriented in 90º angle against the recoater. Otherwise the build quality of these parts does not differ much when visually analyzed from the top side. For more accurate analysis the test pieces were visually analyzed from the bottom side also. Bottom side analysis is quite difficult since the parts are sawed off from the building platform and the sawing might effect on the small and fragile features. In figure 21 is presented the test pieces 1A and 1B from bottom side. The test piece 1A (the left side) has also studs in different angles for testing the minimum build angle.
Figure 21. Test pieces 1A and 1B bottom view in same position as in figure 20
As it can be seen from figure 21, the building process of the thinnest walls is not successful. It can be also seen that the removing of the parts from building platform may break some of the features.
Test piece 2A and 2B can be seen from figure 22. These test pieces include gaps with different widths. Also these pieces were oriented to the building platform in such way that they were 90°
and 45° angle against the recoater movement.
Figure 22. Test pieces 2A and 2B. The arrow shows the recoating direction
In these pieces the visual analysis was much more difficult since the looks of the parts does not differ that much. The three stripes which are raised from the surface of test piece 2B are built because the recoater blade is worn and got some dents in it. Because of that the powder is not spread evenly and it causes these stripes.
The third test piece set includes parts with different size holes. The test pieces 3A and 3B are shown in figures 23 and 24.
Figure 23. Test piece 3A sides A and B. The arrow shows the building direction
Figure 24. Test piece 3B sides A and B. The arrow shows the building direction
These pieces were also built in the way that test piece 3A was oriented parallel against the recoating direction and the 3B was oriented in 45° angle against the recoating direction. The building direction can be seen from the figures 23 and 24 for these pieces. Building of holes is challenging in this direction since big amount of overhangs are created. The marks and cuts shown in test piece 3B B-side came from detaching the piece from the building platform.
8.1 Geometrical features and dimensional analysis
Because laser additive manufacturing gives possibilities to manufacture complex shapes and features to the parts, it was decided to investigate the accuracy of this process with previously presented benchmark pieces. Measurement data of test pieces 1A and 1B is presented in table 13.
Table 13. Measurement data of test pieces 1A and 1B
Wall 3D model dimension [µm]
1A averaged measurements [µm]
Difference [%]
1B averaged measurements [µm]
Difference [%]
1 100 255.70 155.70 215.24 115.24
2 200 286.13 43.07 253.85 26.92
3 300 420.25 40.08 423.08 41.03
4 400 506.33 26.58 530.80 32.70
5 500 577.24 15.45 666.67 33.33
6 600 670.89 11.81 764.10 27.35
7 700 762.05 8.86 884.64 26.38
8 800 860.76 7.59 979.49 22.44
9 900 974.68 8.30 1012.82 12.54
10 1000 1068.37 6.84 1120.51 12.05
11 1100 1164.56 5.87 1230.77 11.89
12 1200 1281.01 6.75 1317.97 9.83
13 1300 1298.75 -0.10 1400.01 7.69
14 1400 1410.13 0.72 1494.88 6.78
small 1 100 broken - 228.21 128.21
small 2 200 broken - 323.12 61.56
small 3 300 broken - 556.44 85.48
small 4 400 broken - 602.56 50.64
small 5 500 582.30 16.46 694.90 38.98
small 6 600 660.76 10.13 751.30 25.22
small 7 700 688.63 -1.62 866.76 23.82
small 8 800 764.60 -4.43 974.36 21.79
small 9 900 911.43 1.27 1097.45 21.94
As it can be observed from the table 13, the wall thicknesses of the test pieces are usually thicker than in the 3D model. The test piece 1A which was oriented parallel to the recoater has more accurate wall dimensions than test piece 1B, which was oriented 45° angle against recoater. But then again the small walls are collapsed in test piece 1A when the wall thickness is less than 500 µm. Small walls of test pieces 1A and 1B are shown in figure 25.
Figure 25. Small walls of test piece 1A and 1B
The walls of 1A are broken because of the recoater movement. When part is oriented in 45°
angle against the recoater, it is possible to build thin features. But as the figure 21 shows, the thinnest walls of test piece 1B are not built completely. The first decently built small wall is number 4 and the thickness of it should be 400 µm. But as it can be seen from table 13 the real thickness of that wall is 602.56 µm. The dimensional comparison of the test pieces 1A and 1B can be seen from figure 26.
Figure 26. Wall thickness comparison between test pieces 1A and 1B
y = 1.1317x R² = 0.9689
y = 1.0592x R² = 0.9683 y = x R² = 1
0.00 200.00 400.00 600.00 800.00 1000.00 1200.00 1400.00 1600.00 1800.00
0 200 400 600 800 1000 1200 1400 1600
Measurements [µm]
Model dimensions [µm]
1B 1A Nominal Linear (1B) Linear (1A) Linear (Nominal)
As it can be observed from figure 26, the dimensions of test piece 1A are closer to the nominal values of the 3D model. Wall thicknesses of both test pieces are thicker than in the 3D model.
This happens because the laser melts the cross section of the part, and the loose powder around the cross section melts also and attaches to the contours. The difference of nominal wall thickness and wall thickness of test piece 1A becomes less than 10% when built wall thickness is 700 µm. The same difference is achieved with test piece 1B in wall thickness 1200 µm. It seems that these kinds of features are more accurate when the features are parallel to the powder spreading direction. Comparison of the small wall accuracy is presented in figure 27.
Figure 27. Wall thickness comparison between test pieces 1A and 1B small walls
As it can be seen from figure 27, also the dimensions of small walls are closer to 3D models nominal values in test piece 1A. It should be also noticed that the thinnest walls are not built well in either of the test pieces. In case of test piece 1A it was not possible to measure the thinnest walls since they were broken, as it can be seen from figure 25. Also the smallest walls were badly built in test piece 1B as it can be seen from figure 28 where bottom sides of test pieces 1A and 1B are presented.
y = 1.2869x R² = 0.9053
y = 0.9128x R² = 0.7618
y = x R² = 1
0.00 200.00 400.00 600.00 800.00 1000.00 1200.00 1400.00
0 100 200 300 400 500 600 700 800 900 1000
Measurements [µm]
Model dimensions [µm]
1B 1A Nominal Linear (1B) Linear (1A) Linear (Nominal)
Figure 28. Bottom sides of test pieces 1A and 1B
As it can be seen from figure 28 the smallest walls are built badly and some of the thinnest walls are broken during the detaching from the platform.
The widths of small gaps are analyzed in test pieces 2A and 2B. The measurement data is presented in table 14.
Table 14. Measurement data of test pieces 2A and 2B
3D model gap width [µm]
2A averaged measurements [µm]
Difference [%]
2B averaged measurements [µm]
Difference [%]
100 92.31 -7.69 167.74 61.29
200 193.85 -3.08 241.94 20.97
300 301.54 0.51 329.12 -3.08
400 353.85 -11.54 383.87 -7.26
500 427.69 -14.46 474.19 3.23
600 507.69 -15.38 429.03 -32.80
700 600.00 -14.29 548.39 -21.66
800 664.62 -16.92 612.90 -25.40
900 781.54 -13.16 729.03 -26.52
1000 892.33 -10.77 822.58 -17.74
As it can be seen from the table 14, most of the gaps are smaller than in the 3D model. This is because of the same reason as mentioned earlier. The enclosed loose powder is melted because of the surrounding heat and that causes the smaller gaps. It also seems that the most accurate gap widths can be achieved when the gap width is close to 300 µm. The comparison of the gap widths is presented in figure 29.
Figure 29. Gap width comparison between test pieces 2A and 2B
As it can be seen from table 14 and figure 29 the gap widths are more accurate when the gaps are parallel to the powder spreading direction. All of the gaps are built but the thinnest ones might not be open all the way. Top sides of the test pieces 2A and 2B are shown in figure 30. The bottom sides of the gaps can close up when the pieces are sawed off from the building platform.
The bottom sides of the test pieces 2A and 2B are shown in figure 31.
Figure 30. Top sides of the test pieces 2A and 2B
y = 0.8691x R² = 0.9934 y = 0.822x R² = 0.9125
y = x R² = 1
0.00 200.00 400.00 600.00 800.00 1000.00 1200.00
0 200 400 600 800 1000 1200
Measurements [µm]
Model dimensions [µm]
2A 2B Nominal Linear (2A) Linear (2B) Linear (Nominal)
Figure 31. Bottom sides of the test pieces 2A and 2B
As it can be seen from figure 31, sawing off the pieces from building platform will cause the clogging of the gaps. As it can be seen from figures 30 and 31 the gap building itself is not a problem with this technology but since the accuracy of the gaps varies widely it would be necessary to compensate the variation of the gap width somehow.
The third test set included two test pieces where the accuracy of holes with overhangs is studied.
The test pieces included different size holes from 100 µm diameter to 6000 µm diameter. The challenge of these test pieces was that they were built in the way that the holes created overhangs which are difficult to build. It was known that the accuracy of the holes is not very good when they are built in this way. These test pieces were measured from sides A and B. The A and B sides of the test pieces are presented in figures 32 and 33. The measurement data of test pieces 3A and 3B can be found in appendix.
Figure 32. A-sides of test pieces 3A and 3B
Figure 33. B-sides of test pieces 3A and 3B
As it can be observed from figures 32 and 33 the quality of holes is better in test piece 3B than in test piece 3A. Holes seem to be much more circular in test piece 3B than in 3A. It can be also seen that the smallest holes are not built since the tops of the holes collapses when built in this direction. Also the heat melts the powder inside of the holes and the smallest ones are filled with this powder. But as it can be seen from figures 32 and 33, the smaller diameter holes are built better in test piece 3B than on 3A. The comparison of average hole diameters between test pieces 3A and 3B are presented in figures 34 and 35.
Figure 34. Average hole diameter comparison between test pieces 3A and 3B A-sides
As it can be observed from figure 34, the test piece 3B is more accurate than test piece 3A on A- side of the pieces. The test piece 3B was oriented in the way that it was 45° angle against the recoating direction. It seems that the part orientation has big influence on build quality and accuracy in these kind of features.
y = 0.8258x R² = 0.9562 y = 0.8753x R² = 0.9883
y = x R² = 1
0 1000 2000 3000 4000 5000 6000 7000
0 1000 2000 3000 4000 5000 6000 7000
Measurements [µm]
Model dimensions [µm]
3A 3B Nominal Linear (3A) Linear (3B) Linear (Nominal)
Figure 35. Average hole diameter comparison between test pieces 3A and 3B B-sides
Also the B-sides of the test pieces 3A and 3B differ from the nominal dimensions. As it can be seen from figure 35, the test piece 3B is more accurate than the test piece 3A. Since the build parameters are same in both pieces, the only distinctive factor is the build orientation in these pieces. As mentioned earlier the build orientation is a big factor when these kinds of parts and features are built. As it can be seen from the figures 34 and 35 the smallest holes that can be built are close to 500 µm. When building holes in this way that the holes form overhangs it is impossible to build absolutely round holes without support structures.
9 CONCLUSIONS
Aim and purpose of this study was to clarify the resolution and capabilities of laser additive manufacturing. It was also important to study how the quality of laser additive manufactured pieces are evaluated and measured. Purpose of the experimental part was to study the resolution and quality of stainless steel test pieces manufactured with modern laser additive manufacturing machine.
In literature review it was concluded that the laser additive manufacturing is very promising technology for manufacturing parts for various field of industry. The possibility to create parts
y = 0.8082x R² = 0.9717 y = 0.8623x R² = 0.9923 y = x R² = 1
0 1000 2000 3000 4000 5000 6000 7000
0 1000 2000 3000 4000 5000 6000 7000
Measurements [µm]
Model dimensions[µm]
3A 3B Nominal Linear (3A) Linear (3B) Linear (Nominal)
