msec2012-7253 1st revision kc - university of alabama · msec2012-7253 thermal modeling of electron...

1 Copyright © 2012 by ASME

Proceedings of the ASME 2012 International Manufacturing Science and Engineering Conference MSEC2012

June 4-8, 2012, Notre Dame, Indiana, USA

MSEC2012-7253

THERMAL MODELING OF ELECTRON BEAM ADDITIVE MANUFACTURING PROCESS – POWDER SINTERING EFFECTS

Ninggang Shen and Kevin Chou Mechanical Engineering Department

The University of Alabama Tuscaloosa, AL 35487

KEYWORDS

Electron beam additive manufacturing, Finite element analysis, Powder porosity, Transient thermal analysis ABSTRACT

In recently developed Additive Manufacturing (AM) technologies, high-energy sources have been used to fabricate metallic parts, in a layer by layer fashion, by sintering and/or melting metal powders. In particular, Electron Beam Additive Manufacturing (EBAM) utilizes a high-energy electron beam to melt and fuse metal powders to build solid parts. EBAM is one of a few AM technologies capable of making full-density metallic parts and has dramatically extended their applications. Heat transport is the center of the process physics in EBAM, involving a high-intensity, localized moving heat source and rapid self-cooling, and is critically correlated to the part quality and process efficiency.

In this study, a finite element model was developed to simulate the transient heat transfer in a part during EBAM subject to a moving heat source with a Gaussian volumetric distribution. The developed model was first examined against literature data. The model was then used to evaluate the powder porosity and the beam size effects on the high temperature penetration volume (melt pool size). The major findings include the following. (1) For the powder layer case, the melt pool size is larger with a higher maximum temperature compared to a solid layer, indicating the importance of considering powders for the model accuracy. (2) With the increase of the porosity, temperatures are higher in the melt pool and the molten pool sizes increase in the depth, but decrease along the beam moving

direction. Furthermore, both the heating and cooling rates are higher for a lower porosity level. (3) A larger electron-beam diameter will reduce the maximum temperature in the melt pool and temperature gradients could be much smaller, giving a lower cooling rate. However, for the tested electron beam-power level, the beam diameter around 0.4 mm could be an adequate choice.

INTRODUCTION In recent years, Additive Manufacturing (AM) using

commercially available atomized metallic or alloyed powders with a high-energy beam heat source has been developed to fabricate complex shapes and multifunctional, custom designed components. Various types of mechanical components can be built with such rapid fabrication technologies through computer controlled machines [1].

Titanium (Ti) and titanium alloys are materials with outstanding mechanical properties such as low density, high strengths, good chemical resistance and excellent biocompatibility. The combination of these properties in a special structure has many potential applications in the area of medical, aerospace, aeronautics and automotive systems. Generally, fabrications of these materials with conventional techniques are difficult due to their high melting point and extreme chemical affinity to atmospheric gas, especially at elevated temperatures [2].


Electron Beam Additive Manufacturing (EBAM) technology has been developed and commercialized by Arcam AB [3], in which metallic powders are selectively molten by an electron beam, then rapidly self cooled and solidified; the detail of the process can be found in [1]. The electron beam can provide both high energy density and energy efficiency. In addition, product material properties are comparable or even better than parts made by conventional means. Despite extensive advantages over conventional technologies, EBAM still exhibits several process/part deficiencies [4]. Hence, accurate physical models of both heat transfer and heat source are required to investigate thermal phenomena and determine appropriate process parameters based on process variables such as the molten pool size that is potentially correlated to the occurrence of the deficiencies.

However, the simulation of thermal process in EBAM is still a challenging task because of the complex heat transport and physical mechanisms induced by an interaction among thermal, mechanical, and metallurgical phenomena. As the EBAM modeling has not been well investigated, research work about Electron Beam Welding (EBW) [5-9], laser deposition or melting [10, 11], and laser welding [12, 13] were first reviewed. Due to the resultant keyhole from an incident electron or laser beam in such processes, the heat source is generally modeled as a conical volumetric body heat flux under the top workpiece surface [4-8, 10, 12, 13]. The intensity is Gaussian distribution horizontally and decays linearly with the increase of penetration depth [5, 7, 8, 10, 12, 13]. Lankalapalli et al. [12] derived a closed-form solution of temperature distribution with the assumption of a Two Dimensional (2D) heat conduction and a conical keyhole. Then, the penetration depth was numerically predicted in a 3D problem accurately. The mathematical-physical model presented by Lankalapalli et al. was widely used in EBW [5-8], laser deposition [10] and laser welding [13]. The state of the art in this field is the simulation with Level-Set Method, such as Qi and Mazumder [14], Wen and Shin [15]; or Volume-of-Fluid Method, such as Choi et al [16]. These two methods can precisely capture the convection in molten pools and even the free boundary problem. However, due to substantially high computational costs and limited commercial code access, Finite Element Analysis (FEA) is still the most efficient way to simulate the EBAM process. To simulate the effect of convection in FEA, Taylor et al. [17], De and DebRoy [18] increased the thermal conductivity in the molten pool.

Even with similar heat transfer and heat source models, the application of metallic powders distinguishes EBAM from other processes. Figure 1 shows scanning electron micrographs of raw powders used in EBAM, Ti-6Al-4V in this case with powder sizes mainly between 25 and 40 µm. A metal powder layer is formed by a raking mechanism, followed by preheating for powder sintering, prior to each layer deposition. It is well known that thermal properties of metallic powder materials are significantly different from those of the corresponding solid bulk material [19, 20], especially in thermal conductivity. In addition, sintering takes place in the powder bed in EBAM as it is usually preheated to about 700-800 °C before melting the powders. Thus, the effect of sintering should be considered as well. Sih and Barlow [20] developed a porosity dependent

model to predict the emissivity for different material conditions and environments. Various porosity-dependent material thermal conductivity in sintering conditions have been developed in gas surroundings [11, 21, 22], which is not the same as the vacuum condition in this study. Tolochko et al. [23] investigated the mechanisms of selective laser sintering and heat transfer in Ti powders in vacuum conditions. A model was derived to predict the effective thermal conductivity of Ti powder bed in a vacuum. All of the aforementioned researchers consider that metallic powder materials have the same specific heat and latent heat of fusion as the solid bulk materials.

Figure 1. Scanning Electron Micrographs of Raw Ti-6Al-4V Powders Used in EBAM.

For EBAM, Zah and Lutzmann [4] conducted a thermal analysis of EBAM with FEA, taking account the metallic powder in the powder layer. The effects of electron beam power and scanning speed on the molten pool geometry were studied. However, the molten pool size and the temperature values were not clearly illustrated. The powder properties were not fully discussed.

One of the most important factors to the thermal process in EBAM is the raw material, i.e., powders, and thus, associated thermal properties that crucially impact the thermal responses that govern the process performance and part quality. Thus, an accurate thermal model accounting for the powder properties is desired to better simulate the process and for process improvements. In this research, the effects of porosity-dependent metallic powder thermal properties in thermal process of EBAM were investigated with an FEA thermal model developed in ABAQUS software. The model was first validated, and then applied to investigate powder porosity as well as the electron beam size effects in EBAM process.

MATHEMATICAL-PHYSICAL MODEL Heat Transfer Model

With the assumption of negligible molten flow during the solidification process, the governing equation of heat transport during the EBAM process is:


2 2 2, ,

2 2 2

T x y z

s

T T T T

QT T T T Tv

c c t xx y z

,

(1)

where T is temperature, , ,x y zQ is the absorbed heat flux, c is

specific heat capacity, ρ is density, λ is thermal conductivity, vs is the constant speed of the moving heat source in the x direction. All material properties are temperature dependent which will be introduced in a later section.

The latent heat of fusion, Lf, was considered in this model to trace the solid/liquid interface of the molten pool. When the temperature drops between the solidus and liquidus temperatures, TS and , TL, respectively, the latent heat of fusion is modeled as an additional term of the internal thermal energy per unit mass, dU. Hence, the enthalpy is defined as

fH T cdT L f , (2)

where f is the volumetric liquid fraction, which is defined as

f

0

1

S

L S

T T

T T

,

,S

S L

L

T T

T T T

T T

(3)

Heat Source Intensity Modeling

In the EBAM process, a cavity containing metal-ion plasma is created in the powder layer due to the material evaporation resulting from the high energy density of the electron beam. The molten material ahead is removed under the effect of vapor pressure and surface tension. Then, the electron beam penetrates deeper to form the so-called “keyhole” effect. To simulate the heat input distribution, the electron beam is typically modeled as a conical moving heat source with Gaussian distributed intensity at each depth level.

In this study, the heat source was modeled as Eqn. (4), which is slightly modified from Rouquette et al. [8]. The intensity decreases with the increase of the penetration depth,

2 2

2 2

88, , exp

s sb

E E

x x y yUIS x y z f z

,

(4)

with 21

zf z

h h ,

where the parameters are: efficiency coefficient η, voltage: U, current: Ib, penetration: h, and beam diameter: ΦE. xS and yS are the position of the heat source (electron beam) center. Figure 2 shows an example of the horizontal intensity distribution, with η = 1, U = 60 kV, Ib = 2mA, ΦE = 2 mm, and h = 2 mm at z = 0, the top surface.

Figure 2. Heat Source Modeling: Horizontal Intensity Distribution at z = 0.

To apply the heat source in the FEA model, a user subroutine of DFLUX was developed in FORTRAN. The subroutine was to be called at the beginning of each iteration. Then, it reads the simulation time to determine the position of the heat source center, so that the domain of the volumetric heat flux can be determined. The magnitude of the heat flux assigned at each nodule is interpolated with Eqn. (4). Figure 3 is the flow chart of the subroutine. As seen, the subroutine has its first part working as another subroutine UMATHT, to be introduced in a later section, besides the discussed DFLUX.

MATERIAL PROPERTIES Properties of Solid Bulk Materials

The material in this study was Ti-6Al-4V. Temperature dependent material properties were required in this study because a high and wide range of temperatures can be expected. Figure 4 shows the temperature dependent density, thermal conductivity and specific heat [24] below the melting point. The thermal conductivity above the melting point is from Ref.[5] which is almost one time greater than the value below melting point. This increase of the thermal conductivity was applied to simulate the effect of the convection in the molten pool. The other two properties have the same values of those at the melting point. The emissivity of Ti-6Al-4V was estimated from literature; Yang et al. experimentally calibrated the emissivity of Ti-6Al-4V as 0.7 [24].

Properties of Metallic Powder Layers As indicated earlier, thermal properties of metallic powders

are significantly different from those of the corresponding solid bulk material. Many researchers indicated that the specific heat and latent heat of fusion of powders can be considered the same as those of the solid material [4, 11, 19-23]. On the other hand, various studies were conducted about the emissivity and thermal conductivity of metallic powders below the melting point. The emissivity and thermal conductivity in this study were modeled as the following [20].


Figure 3. Flow Chart of User Subroutine in Simulations.

Figure 4. Temperature Dependent Thermal Properties.

Sih and Barlow [20] developed and validated a porosity dependent emissivity model for a metallic powder bed. The overall emissivity, ε, is expressed as

1H H H SA A (5)

with2

2

0.908

1.908 2 1HA

,

2

2

12 3.082

11 3.082 1

S

H

S

,

where εs is the emissivity of the solid bulk material, εH is the emissivity of the gap or gravity between adjacent powder particles, AH is the area fraction of the surface that is occupied by the radiation emitting holes, and φ is the fractional porosity of the powder bed.

As EBAM is conducted in vacuum [2], most thermal conductivity models for metallic powders considered the thermal conduction of the gas filling in the pores. However, these models do not ideally match the condition in this study. Tolochko et al. [23] developed a model of thermal conductivity of Ti powders in a study of selective laser sintering. The model only considered the effective thermal conductivity due to thermal radiation, kr, and effective thermal conductivity due to heat transfer through powder necks. Then it was modeled as

r ck k k (6)

with 316

3rk l T , c bulkk k x ,

where l is the mean photon free path between the scattering events, σ is the Stefan-Boltzmann constant, T is temperature, x = b/R, which is the ratio of neck radius to particle radius, Λ is the normalized contact conductivity for the three packing structures, varying with the porosity. In the case of powders, l is on the order of pore sizes, about the particle diameter. Material State Change

Because powder materials will have the properties of the bulk material once the powders are molten, the material state change was embedded in a subroutine, UMATHT. An internal state variable, working as a material index, is defined to specify


the material as powder or solid at this time frame. The material index can be 0 or 1 (0 – powder, 1 – solid). The change criterion is: the material is molten and under cooling (i.e. T > Tm & DT/Dt < 0). After the subroutine is called in the beginning of every iteration check, it reads the material index of each node to determine the correct material properties to be assigned. As the state change in EBAM can only be one-way (i.e., from powder to solid), the material index will be locked as 1 whenever it becomes solid.

NUMERICAL MODEL IMPLEMENTATION Model Configuration

A 3D FEA thermal model was developed and implemented in ABAQUS, to study the complex thermal process of EBAM. Figure 5 shows the model configuration and boundary conditions. As a scan path is simulated in this model, the base material is modeled as solid, since it has been deposited. A thin layer on the top of the substrate is considered as the latest added powder layer which has the powder material properties. The electron beam heating occurs at the top surface of a powder layer and traverses along the x-axis with a constant speed. Due to vacuum, convection between the powder layer and environment is ignored. Hence, only the radiation was considered in the heat transfer between the powder/part and surroundings. The solid substrate and powder layer were assigned with a uniform temperature distribution of Tpreheat as the thermal initial condition. The temperature of solid substrate bottom is confined as a constant temperature of Tpreheat as the thermal boundary condition.

Figure 5. FEA Model Configuration and Boundary Conditions.

Model Validations Due to lack of EBAM experiment results, the model

validation was conducted by simulating LENS® [25] which has a similar configuration to EBAM. Wang et al. [10] developed an FE model to predict the molten pool size for the LENS® process. The authors simulated the deposition of the last layer (the 10th layer) in the LENS experiment with stainless steel 316 powders by Hofmeister et al [25] to validate their model, using the experiment data as initial conditions for previously built layers. Experiment parameters are an absorbed laser power of 100 W, an initial temperature of 600 °C for the substrate and previously deposited layer, a scan speed of 7.62 mm/s, and a laser beam size of 1 mm. Their case was duplicated by the

authors of this MSEC paper. Figure 6 shows the comparison of the model validation. Figure 6a is the comparison of temperature fields simulated by the authors and Wang et al. Figure 6b is the comparison of the temperature profile along the beam center path (from the beam center to the scanned end). Generally, the results from this study agree with those from Wang et al.. The differences at the center peak temperature may be due to the absence of phase changes, which was considered by Wang et al.

Figure 6. Model Validation: (a) Simulated Temperature Contour; (b) Temperature Profile along Beam Center Scan Pass.

Detailed comparisons were further conducted in selective conditions shown in Figure 7. Both of temperature contours and profiles along the depth direction were compared. Three conditions of 2.5, 7.65, and 20.0 mm/sec were examined. The right-column plots in Figure 7 are comparisons of temperature profiles in the depth direction at a position near the beam center. All the results are fairly close to each other. The slight differences may come from the error of data collections.

Figure 7. Systematic Simulation Result Comparisons in Selected Conditions.


RESULTS AND DISCUSSION Adequate process parameters were determined for EBAM

with Ti-6Al-4V [1], such as the scan speed, the acceleration voltage, the beam current, the layer thickness, and the preheat temperature, etc., all listed in Table 1. Then, the powder layer porosity was varied from solid to 0.6 in order to investigate the effect of powder porosity on thermal responses, i.e. temperature field, temperature history, as well as the heating/cooling rates. The substrate was treated as solidified materials from previous layer depositions. In addition, various beam diameters were simulated to determine an appropriate beam size.

Table 1. Parameter List Used in Simulations. Parameters Values Solidus temperature, TS (°C) 1605 [26] Liquidus temperature, TL (°C) 1665 [26] Latent heat of fusion, Lf (kJ/Kg) 440 [26] Electron beam diameter, Φ (mm) 0.4, 0.7, 1.0 Absorption efficiency, η 0.9 [4] Scan speed, v (mm/sec) 400 [1] Acceleration voltage, U (kV) 60 [1] Beam current, Ib (mA) 0.002 [1] Powder layer thickness, tlayer (mm) 0.1 [4] Porosity, φ 0, 0.3, 0.45,0.6 Beam penetration depth, dP (mm) 0.1 Preheat temperature, Tpreheat (°C) 760 [1] Solid vs. Powder Layer

The comparison is first conducted between a solid (φ = 0%) and a powder for the top layer. The 30% porosity of the powder layer was evaluated for this comparison. Figure 8 shows the simulated temperature contours, molten pool geometries, and temperature profile along the center line of the scan path. The molten pool geometry is shown at the right-lower corner of Figure 8a and 8b. The white regions are where the temperatures are below the melting point. In order to capture the molten pool geometry, the lower temperature limit is changed from 760 °C to 1650 °C which is the melting point. The same modification was applied to the figures illustrated in the later part of the paper.

Figure 8c illustrates the temperature profile along the center line of the scan path from the beam center to the scanned region. For the powder top layer case, the temperature is to drop approximately to the initially preheated temperature of 760 °C about 2.5 mm away from the beam center, while it is about 2.25 mm away from the beam center for the solid top layer. Plateaus are noted from the simulations for both the solid and powder top layers in considering the latent heat of fusion (with Lf). The red dotted line shows the temperature profile without considering the latent heat of fusion. Significant differences can be observed between the conditions of with and without the latent heat of fusion. This demonstrates the user subroutine of UMATHT works well with the feature of latent heat of fusion. As a result of the confined heat in the scan path, the plateau of the powder top layer is longer than that of the solid top layer. Moreover, the temperatures after the fusion boundary are higher than those of the solid top layer.

Due to the drastic decrease of the thermal conductivity from solid to powder and less packing density, the maximum

temperature for the powder-layer case increases to 2744 °C, which is over 400 °C higher than that of the solid layer case. As a result, a wider and deeper melt pool occurs, shown in Table 2. Meanwhile, a longer melt pool is also seen with the application of the powder top layer. This could be attributed to the occurrence of the material state change after the powder is molten and solidified. The solidified material has the properties of the solid bulk material, i.e. thermal conductivity increases to the level of solid materials. Because of a low thermal conductivity on both sides and ahead of the scan path, the heat resistance is much lower along the opposite direction of the scan path. Therefore, the heat dissipation is more dominant along the backward direction, and thus, more heat is confined in the scan path. As a result, the melt pool is elongated.

Figure 8. Temperature Fields and Molten Pool Geometries of Solid vs. Powder Top Layer.

Table 2. Simulated Molten Pool Sizes

Φ (mm) Material Length (µm)

Width (µm)

Depth (µm)

0.4

Solid 750 300 100 φ = 30% 850 400 120 φ = 45% 800 400 127 φ = 60% 750 400 134

0.7 φ = 30%

- - 80 1.0 - - 68


Porosity Level Effects Figure 9 shows temperature contours and molten pool

geometries at various levels of porosity (from 30 to 60%) of the powder layer. Based on Figure 9 and results in Table 2, a higher maximum temperature occurs at a higher porosity. In addition, the molten pool becomes deeper and shorter with the increase of porosity. However, the width of molten pool is almost constant in all cases. This is probably due to a greater thermal resistance around the scan path, so that the heat is driven backward along the scan path and the depth direction.

Cooling rates in a molten pool will affect grain growths during solidification, crack nucleation and residual stresses. Therefore, heating/cooling rates are also examined in this study. Figure 10 shows the temperature and heating/cooling rate history at the center point of the scan path. Whenever the center point reaches the maximum temperature, the time is counted as zero. The horizontal dashed line in Figure 10b indicates the boundary of heating/cooling. The plateau is obtained again in Figure 10a in all conditions considering the latent heat of fusion. It takes place between around 2 to 3 µs, which is generally just the time duration that the cooling rate drops to a very low value approaching the heating/cooling boundary.

Figure 9. Temperature Fields and Melt Pool Geometries for Powder Layer of Various Levels of Porosity.

Between the solid and powder-layer cases, a solid top layer results in a lower heating-cooling rate due to a rather low thermal resistance around the scan path. More heat can dissipate from the heat source, and hence, there is not much heat stored in the fusion zone so that the heating and cooling rates are much lower than those of the powder–layer case. For the powder layer with different levels of porosity, both the heating and cooling rates are higher with the increase of porosity. As the thermal resistance does not vary significantly with the change of porosity, the differences seem to be smaller than the differences between the solid vs. powder layer.

Figure 10. Temperature Histories and Heating/Cooling Rates of Center Point for Various Levels of Porosity.

Beam Diameter Effects Figure 11 shows the temperature contours and molten pool

geometries of different electron beam diameters (from 0.4 to 1.0 mm). The top layer is powders with a porosity of 30% for all conditions. A lower maximum temperature is obtained with the increase of the beam diameter due to the decrease of energy density. Hence, a lower cooling rate could be obtained with a larger beam size, as smaller temperature gradients exist. However, increasing the beam diameter may result in drastic decrease of the heat penetration, which can be characterized by the depth of molten pool. The molten pool depth is compared, listed in Table 2, and it is noted that for a large beam diameter (0.7 mm and above), the molten pool depths are smaller than the powder-layer thickness, 100 µm. Such a result is obviously not practical for the EBAM process requirement. Therefore, 0.4 mm electron beam diameter is an optimal condition at the current power lever tested. However, since a much higher building rate and a lower cooling rate are desired, a larger beam size could be adapted if a greater beam power is available. This could be a consideration to optimize the EBAM process in future.


Figure 11. Temperature Fields and Molten Pool Geometries of Various Beam Sizes.

Half-solid and Half-powder Substrate

Support for overhang structures is one of characteristics different between various additive manufacturing processes; for example, fused deposition modeling will require additional materials for overhang supports, but powder-based processes such as EBAM has the raw material itself (powders) as support for overhang structures. However, as indicated above, powder materials have very different thermal properties and thus responses when subject to thermal input. Hence, though EBAM does not require separate support materials for overhang, the thermal behavior around an overhang structure can be significantly unexpected. As a preliminary evaluation, a model with half-solid, half-powder (divided half way in the x direction in Figure 5) substrate was used to simulate the case. The interface is at the y-z plane and intersected with the x-axis at the middle point of the scan path. The top layer is defined as a powder bed with the porosity of 30%. The electron beam traverses from the solid substrate to the powder substrate. Figure 12 illustrates the simulated molten pool geometries with temperature contours when (a) the beam was close to and (b) after the solid/powder (S/P) interface. With the beam approaching to the S/P interface, the molten pool became slightly longer and deeper. However, the geometry (Figure 12a) does not change much, comparing to that when the beam was far away from the interface in the solid substrate half (Figure 9a). On the other hand, Figure 12b shows the stable molten pool geometry after passing the interface in the powder substrate. The molten pool is deeper and has an extremely long tail in the powder substrate due to the large heat resistance of

the metallic powder bed. The implication is potential overheating and melting around the overhang support area, which may result in dimensional inaccuracy and undesired microstructures.

Figure 12. Molten Pool Geometries of Half-Solid Substrate and Half-Powder Substrate: (a) Solid Substrate side and (b) Powder Substrate.

CONCLUSIONS In this research, an FEA model has been developed to

investigate the effect of powder-sintering effect on the thermal phenomena of EBAM. In EBAM, pre-heating for powder sintering is required, which results in different levels of porosity that significantly affect thermal properties of metallic powders. A user subroutine was developed to take account of the latent heat of fusion, temperature dependent and porosity dependent thermal properties of solid or powder materials, and the material state change. The model was first validated against literature experimental data of a similar process, and then applied to study the porosity level and beam diameter effects. The major findings can be summarized as follows. (1) Because of high thermal resistance of powder materials,

the temperature in the molten pool for the powder-layer case is much higher compared to a solid layer. A longer, wider and deeper molten pool is expected with the powder top layer applied. Although powder materials are changed to solid after solidification, the heat is generally trapped in the scanned region, which results in the tailing effect shown in temperature contours. Because of greater temperature gradients around the molten pool, the powder layer causes the increased cooling rate drastically, which may degrade the integrity of the product.

(2) With the increase of the porosity, temperatures are higher in the molten pool, which increases in the depth direction, but decreases along the beam moving direction. The width of molten pool does not vary with porosity noticeably due to the large thermal resistance on both sides of the scan path. Furthermore, both the heating and cooling rates are higher for lower porosity powders.

(3) A larger electron-beam diameter will reduce the maximum temperature in the molten pool and the temperature gradients could be much smaller, giving a lower cooling rate. However, for the tested electron beam power level, the beam size around 0.4 mm could be an adequate choice. Nevertheless, a larger beam size for higher fabrication rates and slower cooling rates may be achieved with a greater electron beam power. Future work of this research will require temperature

measurements for comprehensive model validations. The initial


and boundary conditions need to be better defined by experiment results. In the current study, a constant preheat temperature, reported in literature, was assigned to the entire base material, and also considered as a boundary condition at the bottom of the build part. However, it was not explicitly indicated if the time is sufficient for the pre-heat temperature to be uniformly distributed in the base material. In addition, further investigations are required to enhance the modeling of heat transfer in molten pool due to convection. Once validated, the model can be extended for a more systematic analysis of different process parameter effects toward process performance and part quality. Because of high temperatures and large temperature gradients (spatial and temporal) in the process, measurements using a thermal camera will be pursued.

ACKNOWLEDGMENT This research is supported by NASA, No. NNX11AM11A,

and is in collaboration with Marshall Space Flight Center (Huntsville, AL), Advanced Manufacturing Team.

REFERENCES [1] Gaytan, S. M., Murr, L. E., Medina, F., Martinez, E., Lopez, M. I., and Wicker, R. B., 2009, "Advanced metal powder based manufacturing of complex components by electron beam melting," Materials Technology, 24, pp. 180-190. [2] Heinl, P., Rottmair, A., Korner, C., and Singer, R. F., 2007, "Cellular titanium by selective electron beam melting," Advanced Engineering Materials, 9(5), pp. 360-364. [3] Available from: http://www.arcam.com/. [4] Zaeh, M. F., and Lutzmann, S., 2010, "Modelling and simulation of electron beam melting," Production Engeering. Research and Development, 4, pp. 15-23. [5] Liu, C., Wu, B., and Zhang, J., 2010, "Numerical Investigation of Residual Stress in Thick Titanium Alloy Plate Joined with Electron Beam Welding," Metallurgical and Materials Transactions B, 41(5), pp. 1129-1138. [6] Luo, Y., Liu, J., and Ye, H., 2010, "An analytical model and tomographic calculation of vacuum electron beam welding heat source," Vacuum, 84(6), pp. 857-863. [7] Lacki, P., and Adamus, K., 2011, "Numerical simulation of the electron beam welding process," Computers & Structures, 89(11-12), pp. 977-985. [8] Rouquette, S., Guo, J., and Le Masson, P., 2007, "Estimation of the parameters of a Gaussian heat source by the Levenberg-Marquardt method: Application to the electron beam welding," International Journal of Thermal Sciences, 46(2), pp. 128-138. [9] Hemmer, H., and Grong, Ø., 1999, "Prediction of penetration depths during electron beam welding," Science and Technology of Welding & Joining, 4(4), pp. 219-225. [10] Wang, L., Felicelli, S., Gooroochurn, Y., Wang, P. T., and Horstemeyer, M. F., 2008, "Optimization of the LENS process for steady molten pool size," Materials Science & Engineering A (Structural Materials: Properties, Microstructure and Processing), 474, pp. 148-156. [11] Roberts, I. A., Wang, C. J., Esterlein, R., Stanford, M., and Mynors, D. J., 2009, "A three-dimensional finite element analysis of the temperature field during laser melting of metal

powders in additive layer manufacturing," International Journal of Machine Tools and Manufacture, 49(12-13), pp. 916-923. [12] Lankalapalli, K., Tu, J. F., and Gartner, M., 1996, "A model for estimating penetration depth of laser welding processes," Journal of Physics D: Applied Physics, 29, p. 1831. [13] Tsirkas, S. A., Papanikos, P., and Kermanidis, T., 2003, "Numerical simulation of the laser welding process in butt-joint specimens," Journal of Materials Processing Technology, 134(1), pp. 59-69. [14] Qi, H., Mazumder, J., and Ki, H., 2006, "Numerical simulation of heat transfer and fluid flow in coaxial laser cladding process for direct metal deposition," Journal of Applied Physics, 100(2). [15] Wen, S., and Shin, Y. C., "Modeling of transport phenomena during the coaxial laser direct deposition process," American Institute of Physics. [16] Choi, J., Han, L., and Hua, Y., 2005, "Modeling and experiments of laser cladding with droplet injection," Journal of Heat Transfer, 127(9), pp. 978-986. [17] Taylor, G. A., Hughes, M., Strusevich, N., and Pericleous, K., 2002, "Finite volume methods applied to the computational modelling of welding phenomena," Applied Mathematical Modelling, 26(2), pp. 311-322. [18] De, A., and DebRoy, T., 2007, "Improving reliability of heat and fluid flow calculation during conduction mode laser spot welding by multivariable optimisation," Science and Technology of Welding and Joining, 11, pp. 143-153. [19] Lin, T. H., Watson, J. S., and Fisher, P. W., 1985, "Thermal conductivity of iron-titanium powders," Journal of Chemical & Engineering Data, 30(4), pp. 369-372. [20] Sih, S. S., and Barlow, J. W., 2004, "The prediction of the emissivity and thermal conductivity of powder beds," Particulate Science and Technology, 22, pp. 291-304. [21] Kolossov, S., Boillat, E., Glardon, R., Fischer, P., and Locher, M., 2004, "3D FE simulation for temperature evolution in the selective laser sintering process," International Journal of Machine Tools and Manufacture, 44(2-3), pp. 117-123. [22] Patil, R. B., and Yadava, V., 2007, "Finite element analysis of temperature distribution in single metallic powder layer during metal laser sintering," International Journal of Machine Tools and Manufacture, 47(7-8), pp. 1069-1080. [23] Tolochko, N. K., Arshinov, M. K., Gusarov, A. V., Titov, V. I., Laoui, T., and Froyen, L., 2003, "Mechanisms of selective laser sintering and heat transfer in Ti powder," Rapid Prototyping Journal, 9, pp. 314-326. [24] Yang, J., Sun, S., Brandt, M., and Yan, W., 2010, "Experimental investigation and 3D finite element prediction of the heat affected zone during laser assisted machining of Ti6Al4V alloy," Journal of Materials Processing Technology, 210(15), pp. 2215-2222. [25] Hofmeister, W., Wert, M., Smugeresky, J., Philliber, J. A., Griffith, M., and Ensz, M. T., 1999, "Invesitigation of solidification in the Laser Engineered Net Shaping (LENS) process," JOM, 51(7). [26] Boyer, R., Welsch, G., and Collings, E. W., 1998, "Materials Properties Handbook: Titanium Alloys," ASM InternationalMaterials Park, OH, USA, pp. 483-636.

msec2012-7253 1st revision kc - university of alabama · msec2012-7253 thermal modeling of electron...

Documents