Thermal effect on clad dimension for laser deposited Inconel 718

Thermal effect on clad dimension for laser deposited Inconel 718

G Model JMP-644; No. of Pages 8 ARTICLE IN PRESS Journal of Manufacturing Processes xxx (2017) xxx–xxx Contents lists available at ScienceDirect Jo...

2MB Sizes 1 Downloads 57 Views

G Model JMP-644; No. of Pages 8

ARTICLE IN PRESS Journal of Manufacturing Processes xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Journal of Manufacturing Processes journal homepage: www.elsevier.com/locate/manpro

Thermal effect on clad dimension for laser deposited Inconel 718夽 Jennifer L. Bennett a,b,∗ , Sarah J. Wolff a,∗∗ , Gregory Hyatt b , Kornel Ehmann a , Jian Cao a a b

Department of Mechanical Engineering, Northwestern University, Evanston, IL 60208, USA DMG MORI, Hoffman Estates, IL 60192, USA

a r t i c l e

i n f o

Article history: Received 18 November 2016 Received in revised form 10 February 2017 Accepted 3 March 2017 Available online xxx Keywords: Additive manufacturing Direct energy deposition INCONEL 718 Cooling rate Volume change

a b s t r a c t Additive manufacturing of components made of nickel-based and high strength materials that endure extreme environments, such as Inconel 718, is gaining traction in aerospace and automotive industries. However, one of the remaining challenges of laser deposited alloys is the volume change of the clad during a build, leading to warping, compromised dimensional integrity of the final part, and an increase in surface roughness. In addition, there has been no work in the prediction and control for volume change of localized areas within a laser deposited component. The dimensional integrity of a completed laser deposited structure is dependent on the uniformity of each individual clad track, with high variability in thermal history and clad height. The approach in this paper is the use of an in-situ infrared camera to capture the thermal history and determine the unique solidification rate of each localized point of each clad. Clad height measurements various points of the clads relative to the tool path were used to establish a relationship between process parameters, solidification rate and the volume change of the clad that verify analytical thermal models in the literature. Expanding these relationships to more complex build geometries, different laser deposited materials and a wider variety of processing conditions will allow for a better understanding, and therefore control, of the laser deposition process for more ubiquity of additive manufacturing in industry. © 2017 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.

1. Introduction Direct Energy Deposition (DED) is an additive manufacturing (AM) process that is rapidly growing as a process to repair and build fully dense and high strength nickel-based components that can endure extreme environments for aerospace and automotive industries. Although DED is growing in ubiquity, one of the remaining challenges in the process is that many components undergo volume change and anisotropic cooling rates. Blown powder laser deposition, a form of DED, is an AM technique possessing unique potential for the creation of complex geometries that cannot be created by conventional manufacturing methods, including parts with internal cavities, the repair or modification of existing components, and the creation of bi-metallic or gradient material structures [1,2]. In the laser deposition process,

夽 45th SME North American Manufacturing Research Conference, NAMRC 45, LA, USA. ∗ Corresponding author at: Department of Mechanical Engineering, Northwestern University, Evanston, IL 60208, USA and DMG MORI, Hoffman Estates, IL, 60192, USA. ∗∗ Corresponding author. E-mail addresses: [email protected] (J.L. Bennett), [email protected] (S.J. Wolff).

powder is pneumatically conveyed via an inert gas into a molten pool, which is generated by a laser beam on an existing substrate, as seen in Fig. 1. As the laser moves relative to the part at some process feed rate, the molten pool solidifies to form a solid material track. A fully dense 3D geometry can be created by overlapping these tracks side-by-side and depositing consecutive layers on top of the previous layers. The dimensional consistency of the completed structure relies on the uniformity and repeatability of the width and height of each individual clad track. If the heights of the individual clad tracks are lower than the programmed layer height in the motion system, the build will move below the focal point of the powder and laser leading to a decrease and eventual cease of deposition. This phenomenon is referred to as “under-building”. Likewise, if the heights of the individual clad tracks are higher than the programmed layer height in the motion system the build will move above the focal point of the powder and laser and could potentially collide with the deposition nozzle. This phenomenon is referred to as “over-building.” Deposition dimensions depend on process parameters such as laser power, process feed rate, powder flow rate, and substrate temperature [3]. Varying powder flow rates at constant laser power and process feed rate result in differing single-track cross sections with three categories of cross-section shape [4,5]. These cross sections are

http://dx.doi.org/10.1016/j.jmapro.2017.04.024 1526-6125/© 2017 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.

Please cite this article in press as: Bennett JL, et al. Thermal effect on clad dimension for laser deposited Inconel 718. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.04.024

G Model JMP-644; No. of Pages 8 2

ARTICLE IN PRESS J.L. Bennett et al. / Journal of Manufacturing Processes xxx (2017) xxx–xxx

characterized by the angle between the substrate and the tangent of the single-track cross-section edge, as seen in Fig. 2. Low powder flow rate results in high dilution, low clad height and large clad angle, whereas the opposite trends occur at high powder flow rates. Dilution can be used to quantify the bond between the clad and the substrate as well as the laser power efficiency during the build [6]. High dilution indicates an inefficient use of laser power where there is excessive melting of the substrate. Low dilution indicates poor bonding and possible loss of adhesion. Several studies have investigated process maps to determine optimal combinations of laser power and powder flow rate by using theoretical laser deposition models [4]. A model that included the mixing of powder in the melt pool showed that, in the ideal case where dilution is minimal, the clad thickness varies linearly with laser power and process feed rate [7]. Olivera et al. [3] correlated powder flow rate, process feed rate and laser power with the clad height, clad width, clad area, molten area, and clad angle. Zhang et al. [8] discussed the influence of process parameters in terms of specific energy, calculated by dividing actual power by beam diameter and scanning velocity. They found that the height of the cladding layer first increases, then decreases with increasing specific energy, concluding that specific energy affects size, shape and formation rate of the clad. Liu and Li [9,10] modelled a single clad track and a thin wall created using a low power laser, concluding that the clad shape is dominated by the powder concentration at any point for the duration that the point is molten. The above-cited work assumes that with constant parameters the dimensions of the clad track will remain uniform. While this approach may be an acceptable assumption for laser deposition coating where only a few layers are deposited onto a substrate with a much larger thermal mass, the thermal profile of the build radically changes during laser deposition of a 3D component. As new layers of material are successively deposited, heat is conducted into the substrate, which acts as a heat sink [11]. As the process progresses, the thermal gradient decreases, such that with constant process parameters (e.g., laser power, process feed rate, powder flow rate), the melt pool size subsequently increases and the cooling rate decreases, leading to inconsistent track morphology and mechanical properties [12]. Slight changes in process parameters can result in large variations in volume change and cooling rate, which, in turn, influence

Fig. 1. Coaxial laser deposition process.

mechanical properties and the type of post heat-treating or level of post-machining of the surface required to achieve the desired results. In the case of IN718, rapid shrinkage occurs immediately as cooling begins along with unique phase transformations during solidification, depending on solidification rate [16]. This provides motivation for future work in investigating microstructure evolution. Optimal deposition parameters have been investigated by many researchers for several materials and geometries [1–8,17–19]. These studies have investigated the influence of parameters such as laser power, powder flow rate, process feed rate, and powder size. However, these parameters are generally only valid for the specific geometry the researchers optimized, since the thermal history of the component through the deposition process is highly geometry dependent. By tying final component properties to solidification rate, process maps and control parameters to give specified clad dimensions can be made that are transferable across machine platforms and varying part geometries. Consequently, the purpose of this study is to relate process parameters to thermal history and thermal history to clad dimensions in such a way that dimensional changes can be quantified in terms of thermal history and parameter interactions. Such relationships, introduce flexibility and control of process parameters for various build geometries and temperature conditions. 2. Methods 2.1. Laser deposition process For this experimental study a DMG MORI LaserTec 65 3D, a hybrid additive and subtractive five-axis machine tool, was used. The machine includes a direct diode laser with a maximum power of 2500 W at a wavelength of 1020 nm. The beam is focused utilizing a lens with a 200 mm focal length to a 3 mm spot size. Gas atomized super alloy INCONEL 718 powder of particle size 50–150 ␮m was used. Due to its high temperature yield strength and corrosion resistance, IN718 has numerous applications in aerospace and nuclear industries, particularly in turbine components. Argon gas with a flow rate of 7 L/min was used as the shield gas and the conveying gas to deliver the powder coaxially to the melt pool. The experimental set up can be seen in Fig. 1. Single line IN718 clads were deposited on a 1045 medium carbon steel disks 89 mm in diameter and 8.9 mm thick. The clads were 50 mm in length. Ten clad tracks were deposited on each substrate disk with 7 mm spacing between clads, as shown in Fig. 3. The laser power was held constant on each substrate disk while the powder flow rate was incremented from 3.5 to 27.1 g/min in 3.5 g/min increments. The laser power was incremented from 1000 to 2000 W in 200 W increments. The feed rate for all depositions was 1000 mm/min. To allow for cooling of the previous clads before subsequent clads were applied, the machine was dwelled 120 s between the deposition of each clad. Three locations in each clad were studied in detail to understand the effect of cooling rate. The points were located at 10 mm, 25 mm, and 40 mm from the clad’s start point. All three points were in the constant velocity region of the clad, where the acceleration was zero, and were expected to present similar characteristics. The thermal data and clad morphology were analysed at each location. 2.2. Infrared (IR) thermal measurements

Fig. 2. Single clad cross section profiles. (A) Clad with relatively low powder flow rate and high dilution; (B) Clad with ideal powder flow rate; (C) Clad with relatively high powder flow rate and low dilution.

In situ temperature measurements were performed during the deposition using a digital infrared camera (FLIR). The camera’s resolution was 640 × 480 pixels with a spectral range from 7.5 to 14.0 ␮m and an accuracy of ±2 ◦ C. The camera recorded the infrared

Please cite this article in press as: Bennett JL, et al. Thermal effect on clad dimension for laser deposited Inconel 718. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.04.024

G Model

ARTICLE IN PRESS

JMP-644; No. of Pages 8

J.L. Bennett et al. / Journal of Manufacturing Processes xxx (2017) xxx–xxx

3

Fig. 3. Clad tracks on substrate disk.

the liquidus and solidus temperatures of each clad and determines its resulting structure and properties. The slope of the linear regression fit in the solidification region corresponded to the solidification rate in ◦ C/s. 2.4. Clad dimensional analysis

Fig. 4. IR image of the process.

radiation emitted by the laser deposition of clads and recorded the emission of temperatures ranging from 300 ◦ C to 2000 ◦ C. Absolute temperature measurement via IR requires knowledge of the emissivity at the measurement point. The emissivity of most materials is a function of surface condition, temperature, and wavelength of measurement. Since areas of interest in this study were undergoing a phase change from powder to molten to solid track, the emissivity values could not be accurately determined. Therefore, an emissivity value of 0.1 was chosen for all measurements. Fig. 4 shows the thermal image of a clad deposited at 1600 W at a powder flow rate of 27.2 g/min. The clad initiated at the bottom left and terminated at the upper right. The reflection of the thermal radiation can be seen on the nozzle directly above the melt pool. 2.3. Thermal data analysis The thermal history at each point was analysed to determine the solidification rate during deposition. Initially, a thresholding operation was used to determine the location and size of the melt pool during deposition at a given location in the clad. The melt pool region was defined as every pixel whose temperature was above the solidus temperature as the deposition occurred at the location of interest. The temperature of each of the pixels within the melt pool region was extracted for the entire deposition process at every frame of the infrared camera video. The average temperature of the melt pool region of interest and its standard deviation were calculated for each frame. The solidification rate was calculated by fitting a linear function to the area-averaged temperature of the melt pool region in the frames where the temperature average was between the liquidus and solidus temperatures of IN718 at 1260 ◦ C and 1336 ◦ C, respectively. Solidification occurs between

An Alicona Infinite Focus, a white light optical measurement instrument that relies on 3D focus variation, was used to measure the geometric profiles of each clad relative to the substrate surface. Measurements were taken at the 10 mm, 25 mm, and 40 mm points of each clad. The vertical resolution of the measurements was 500 nm and the horizontal resolution was 8 ␮m. Some of the laser clads exhibited un-melted powder particles on the surface of the clad and the adjacent substrate surface. Three measurements were probed at each location on the clad. The raw data determined maximum clad height, and clad width. The height profiles were fitted to parabolic curves to obtain smooth curves to serve as the mean line for surface roughness definition. The difference between the raw data and the mean line was used to determine the surface roughness (Ra), for each clad at each of the locations of interest. 3. Results and discussion 3.1. Solidification rate of clads based on process parameters Determining the solidification rates at localized areas of an AM component is crucial to the control of the process and of the component’s mechanical behavior. Thermal history was recorded via IR data at the specific points of a clad located at 10 mm, 25 mm, and 40 mm from the clad’s start point. The solidification rate demonstrates the influence of the powder flow and laser power parameters on the microstructure of each clad. Understanding the relationship between process parameters and solidification rate can lead to greater insight into the numerical relationships between clad height and solidification rate. Fig. 5 reveals that larger cooling rates occur at lower powder flow rates and at lower laser powers. This result can be attributed a lesser amount of energy being made available to the material as well as the greater surface to volume ratio of melt pools created with lower powder flow rates, leading to greater radiation and convection during a build. Furthermore, as seen in Fig. 5a, the solidification rate of the clad is more sensitive to changes in laser power at lower powder flow rates. An ANOVA analysis of the data reveals that the location of the clad has negligible influence on the solidification rate relative to the high sensitivity of cooling to pow-

Please cite this article in press as: Bennett JL, et al. Thermal effect on clad dimension for laser deposited Inconel 718. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.04.024

G Model JMP-644; No. of Pages 8

ARTICLE IN PRESS J.L. Bennett et al. / Journal of Manufacturing Processes xxx (2017) xxx–xxx

4

Fig. 5. (A) Solidification cooling rate vs. powder flow rate at tested laser powers, (B) Solidification cooling rate vs. laser power at tested powder flow rates.

der flow and laser power. The standard deviation, or coefficient of variation of solidification rate due to laser power in Fig. 5b is 42.7% at a powder flow rate of 3.5 g/min, whereas the variance of solidification rate due to laser power is 31.4% at a powder flow rate of 27.2 g/min. Based on literature on thermal modeling of the laser deposition process, the solidification rate is proportional to powder flow [20]: dTliq→sol dtliq→sol where

1 ∝ √ 

dTliq→sol dtliq→sol

(1)

is the solidification rate and  is the density of the

material within the melt pool, assuming the density is linearly proportional to the powder flow rate into the melt pool at constant scan speeds. Fig. 5a. reflects this relationship as: dTliq→sol dtliq→sol

a1 = √ + a2 p

(2)

where p is the powder flow in g/min and a1 and a2 are variables proportional to (laser power)−3 . By curve fitting this relationship, the goodness of fit was as high as R-square of 0.82, SSE of 1.4e7 and RMSE of 714. The standard deviation, or coefficient of variation of the solidification rate due to powder flow in Fig. 6b is 66.1% at a laser power of 1000 W, whereas the variance of solidification rate due to powder flow is only 25.4% at a laser power of 2000 W. In addition to powder flow, the solidification rate is proportional to the length of the melt pool [20]: log

dTliq→sol dtliq→sol

= b1 ∗ log l + b2

(3)

where b1 and b2 are variables, with b1 assumed to be −1 and b2 varies with powder flow, l is the length of the melt pool and has been shown to vary with laser power, scan speed, beam diameter and the material properties of the deposited powder [21]: l = d(1 + (DQ + E)v)

(4)

where d is the beam diameter, or the width of the melt pool in mm, D and E are material constants, and in this case, are 0.127 and 2.91 respectively [21], Q is the laser power in W, and v is the scan speed in mm/s. Based on Eqs. (3) and (4), the relationship of solidification rate and laser power is: dTliq→sol dtliq→sol

= −c1 [d(1 + (DQ + E)v] + c2

(5)

where c1 and c2 are variables that vary with powder flow and are proportional to ep . This relationship reveals a linear relationship between the solidification rate and laser power, which is consistent with literature [20] as seen in Fig. 5b.

Since material microstructure is dependent on solidification rate, a process that is unaffected by small fluctuations in powder flow would produce a more uniform clad microstructure. Since small fluctuations in powder flow are common, these results suggest that a process with higher laser power would be less sensitive to such powder fluctuations. Future work will confirm this assumption. Fig. 6 illustrates the change in cooling rate for each area of a clad corresponding with the area’s energy density per unit mass, which is dependent on laser processing power and powder flow rate. The scatter plot represents the experimental data with the dotted lines corresponding to the numerically determined fitted model based on Eqs. (2) and (5). The energy density in J/kg is: J=

100Q 6p

(6)

By determining the relationship between powder flow and laser power based on Eqs. (2) and (5), an overall relationship between the solidification rate and energy density at a constant laser power is determined and plotted in Fig. 6a with the experimental data: dTliq→sol dtliq→sol

= d1 exp(d2 J) + d3 exp(d4 J)

(7)

where d1 –d4 are variables that change with laser power. The solidification rate as an expression of energy density at a constant powder flow rate can be expressed as a linear relationship [20] and is seen in Fig. 6b: dTliq→sol dtliq→sol

= −f1 J + f2

(8)

where f1 and f2 are variables that change with powder flow rate. The combined sensitivity of laser power and powder flow, which are both represented in the expression of energy density in Eq. (6), has an ANOVA two-factor interaction p-value of 1.3e−5 , in comparison to the p-values of 0 for the independent influence of laser power and powder flow rate. With a lower powder flow rate, a greater surface to volume ratio of the heated powders leads to greater energy per unit mass and a higher cooling rate. Future work will include thermal processing of additional clads and multi-layer builds processed with different laser powers and powder flow rates. In addition, thermal results will provide inputs into thermal models that capture the volume change in laser deposition. 3.2. Clad dimensions based on process parameters The interaction of laser processing power, powder flow rate and process feed rate contributes to the clad’s geometry, surface roughness and the likelihood of any un-melted powders adhering to the clad surface. The Alicona metrology tool provides geometry profiles

Please cite this article in press as: Bennett JL, et al. Thermal effect on clad dimension for laser deposited Inconel 718. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.04.024

G Model JMP-644; No. of Pages 8

ARTICLE IN PRESS J.L. Bennett et al. / Journal of Manufacturing Processes xxx (2017) xxx–xxx

5

Fig. 6. Solidification cooling rate vs. process parameter-determined energy per unit mass.

Fig. 7. Examples of the clad surfaces 40 mm from start.

and micrographs of the clad’s top surface, as shown in Fig. 7. At the lower laser powers, seen in Fig. 7a and b, and higher powder flow rates in Fig. 7b and d, a greater roughness indicates a higher incidence of un-melted powders. This roughness can also be attributed to melt pool interface lines which can also be attributed to the melt pool flow lines which are seen on the surface of the clads in Fig. 7a and c. but for the purposes of this study, the surface roughness seen on the clads was assumed to be primarily the result of unmelted powders. Un-melted powders can occur because of a low melt pool temperature without sufficient energy to melt all the incident powder or because of ricochet powder fusing to the heated track. As shown in Fig. 8, on average, clad heights decrease with increasing laser power. This can be attributed to the increased dilution of the IN718 powders at increased laser power. These results are consistent with the results of Fan et al. [6] which show that dilution depth increases linearly with laser power. As laser power increases, more energy is made available to the substrate, resulting in a higher level of dilution. Clad height increases with an increase in powder flow. The vertical height of the clads processed with 27.2 g/min powder flow are between 8.0 and 13.5 times greater the heights of the clads processed with 3.5 g/min at constant laser powers. However, laser power has a much less significant impact on clad height. The height of the clads processed at 1000 W laser power are only between 1.2 and 1.4 times greater than the heights

of the clads deposited at 2000 W at the same powder flow rate. This is consistent with the findings of Oliveira et al. [3] who concluded that clad height is primarily dependent on the powder deposited per unit area and that laser power only has a small impact on clad height. The height of the clad exhibits a linear relationship that is dependent on powder flow rate and laser power, with greater weight on the dependence of powder flow rate. Based on Fig. 9, the results show that average roughness increases with increasing powder flow and decreases with increasing laser power. At high powder flows and low laser powers, there may not be enough thermal energy within the melt pool to fully melt all the powder. Furthermore, at high powder flows, there is more opportunity for powder ricochet to adhere to the recently solidified track. Like the relationship between clad height and process parameters, surface roughness also exhibits a linear dependence on laser power and powder flow rate. The relationship between clad height and energy per unit mass exhibits an exponential decay, indicating that at lower ratios of laser power to powder flow rate the height of the clad is more sensitive to changes in laser power or powder flow rate, as seen in Fig. 10a. This exponential decay can be explained by the dependence of shrinkage on the change of Gibbs free energy and enthalpy during phase transformations of a metallic alloy [13]. Because the clad height is approximately the same at energy densities above 20 J/kg, these results indicate that utilizing a higher energy per unit

Please cite this article in press as: Bennett JL, et al. Thermal effect on clad dimension for laser deposited Inconel 718. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.04.024

G Model JMP-644; No. of Pages 8 6

ARTICLE IN PRESS J.L. Bennett et al. / Journal of Manufacturing Processes xxx (2017) xxx–xxx

Fig. 8. (A) Z-height of clad vs. powder flow rate at tested laser powers, (B) Z-height of clad vs, laser power at tested powder flow rates.

Fig. 9. Surface roughness vs. (A) powder flow rates, (B) laser power.

mass during deposition could lead to greater dimensional consistency of a deposited component. However, since the layer heights at these high-energy densities are relatively low, this increase in uniform morphology could be at the expense of cycle time. The relationship between surface roughness and energy per unit mass also exhibits an exponential decay, as seen in Fig. 10b. Even at very high energy densities there is still an average Ra of 0.01 mm indicating that this contribution of the surface roughness is unrelated to energy density but rather is associated with powder ricochet adhering to the recently solidified track. It can be assumed that for this material at an energy density above 12 J/kg the energy available in the melt pool is sufficient to melt all the incident powder. 3.3. Relationship between solidification cooling and volume change Since relationships between specific process parameters and the final deposited clad tend to be machine, geometry, and time dependent, relating the final clad geometry to these parameters cannot

provide a general, process-independent, numerical model. For this reason, relating final clad characteristics to features of the thermal history such as the solidification rate is necessary. Preliminary data show that as solidification rate decreases, the variation of the Zheight within a clad also increases, as shown in Fig. 11, where the dotted lines represent the numerical fits. As Fig. 11a demonstrates, the linear relationship between clad height and powder flow in Fig. 8a can be divided by the relationship between solidification rate and powder flow in Eq. (2) and Fig. 5a to determine the numerical relationship between clad height and solidification rate at a constant powder flow rate, i.e.:

Z = g1 [

dTliq→sol dtliq→sol

−3⁄2 ]

+ g2

(9)

where Z is the clad height and g1 and g2 are variables that linearly depend on laser power. Fig. 11b suggests that for localized areas of clads processed with larger powder flow rates, dimensional consistency is highly dependent on slight changes of the solidification cooling rate throughout the clad due to shrinkage. Increasing

Fig. 10. Relationship between (A) clad height and (B) surface roughness vs. energy per unit mass.

Please cite this article in press as: Bennett JL, et al. Thermal effect on clad dimension for laser deposited Inconel 718. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.04.024

G Model JMP-644; No. of Pages 8

ARTICLE IN PRESS J.L. Bennett et al. / Journal of Manufacturing Processes xxx (2017) xxx–xxx

7

Fig. 11. (A) Z-height of the clad vs. solidification rate at tested laser powers, (B) Z-height of the clad vs. solidification rate at tested powder flow rates.

solidification rates also contribute to decreasing volume change. This aligns with observations of IN718 and other metallic alloys for which rapid solidification results in unique phase transformations that lead to eutectic phases and result in minimal changes in the Gibbs free energy [13] and therefore, minimal changes in volume. The dotted lines in Fig. 11b demonstrate the linear relationship when the linear relationship between clad height and laser power in Fig. 8b is divided by the relationship between solidification rate and laser power in Eq. (5) and Fig. 5b to determine the numerical relationship between clad height and solidification rate at a constant laser power: Z=h

dTliq→sol dtliq→sol

(10)

where h is a variable that changes exponentially with powder flow rate. These relationships are consistent with analytical thermal models in the literature [20,21] and allow for geometric control of an additive build based on localized solidification rates. 4. Conclusion Overall, this study experimentally investigated the influence of powder flow and laser power on the thermal history and dimensions of laser deposited IN718 clads. The relationship between thermal history and final clad height was also examined. Based on the preliminary results, the key findings are: • Increasing laser power decreases variability in solidification rate between clad tracks deposited at various powder flow rates, which could decrease variability in microstructure. • Powder flow has a more significant impact on clad height than laser power. • Increasing energy density decreases the surface roughness to a minimum value. • Increasing energy density and solidification rate decreases the clad Z-height sensitivity to cooling rate, which could increase dimensional consistency. • The numerical relationships between solidification rate and clad height on process parameters match thermal models and leads to a numerical relationship between solidification rate and clad height. Future work includes analysis of melt pool size and solidification for multiple layers and more complex geometries; optical microscopy to investigate solidification structure, grain size, orientation and presence of unique phases; micro-hardness testing of the clads to link solidification to mechanical behavior; and thermal modelling that incorporates finite element methods to predict

and control for volume change during laser deposition. The results from this study and future work can minimize post-processing of AM components, maximize the efficiency of laser deposition, and aid the design of complex geometries with high structural integrity. Acknowledgements The authors would like to thank the Digital Manufacturing and Design Innovation Institute (DMDII) for their support through award number 15-07 and the U.S. Department of Commence National Institute of Standards and Technology’s Center for Hierarchical Materials Design (CHiMaD) under grant No. 70NANB14H012. The authors would also like to thank Nico Martinez Prieto and Hao Wu for their help on thresholding the IR camera data and measuring the clad geometry under the Alicona. This work made use of facilities at DMG MORI and Northwestern University. References [1] Costa L, Vilar R. Laser powder deposition. Compr Mater Process 2009;15(4):264–79. [2] Bennett J, Dudas R, Cao J, Ehmann K, Hyatt G. Control of heating and cooling for direct laser deposition repair of cast iron components. Int Symp Flex Autom 2016:229–36. [3] de Oliveira U, Ocelík V, De Hosson JTM. Analysis of coaxial laser cladding processing conditions. Surf Coat Technol 2005;197(2–3):127–36. [4] Weerasinghe VM, Steen WM. Computer simulation model for laser cladding. Transp Phenom Mater Process 1983:15–23. [5] Weerasinghe VM, Steen WM. Laser cladding with pneumatic powder delivery. Appl Laser Tool 1987:183–211. [6] Z, Fan TE, Sparks F, Liou A, Jambunathan Y, Bao, et al. Numerical simulation of the evolution of solidification microstructure in laser deposition. Proc. 18th Annu. Solid Free. Fabr. Symp 2007:256–65. [7] Hoadley AFA, Rappaz M. A thermal model of laser cladding by powder injection. Metall Trans B 1992;23(5):631–42. [8] Zhang QL, Yao JH, Mazumder J. Laser direct metal deposition technology and microstructure and composition segregation of Inconel 718 superalloy. J Iron Steel Res Int 2011;18(4):73–8. [9] Liu J, Li L. Study on cross-section clad profile in coaxial single-pass cladding with a low-power laser. Opt Laser Technol 2005;37(6):478–82. [10] Li L, Liu J. Effects of process variables on laser direct formation of thin wall. Opt Laser Technol 2007;39(2):231–6. [11] Dinda GP, Dasgupta AK, Mazumder J. Laser aided direct metal deposition of Inconel 625 superalloy: microstructural evolution and thermal stability. Mater Sci Eng A 2009;509(1–2):98–104. [12] Salehi D, Brandt M. Melt pool temperature control using LabVIEW in Nd:YAG laser blown powder cladding process. Int J Adv Manuf Technol 2006;29(3):273–8. [13] Griffith ML, Schlienger ME, Harwell LD, Oliver MS, Baldwin MD, Ensz MT, et al. Understanding thermal behavior in the LENS process. Mater Des 1999;3(23):107–13. [16] Valencia JJ, Spirko J, Schmees R. Sintering effect on the microstructure and mechanical properties of alloy 718 processed by powder injection molding. Superalloys 1997;718(625):753–62, 706 Var. Deriv. [17] Zhang K, Liu W, Shang X. Research on the processing experiments of laser metal deposition shaping. Opt Laser Technol 2007;39(3):549–57. [18] Wolff SJ, Lin S, Faierson EJ, Liu WK, Wagner GL, Cao J. A framework to link localized cooling and properties of directed energy deposition (DED)-processed Ti-

Please cite this article in press as: Bennett JL, et al. Thermal effect on clad dimension for laser deposited Inconel 718. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.04.024

G Model JMP-644; No. of Pages 8 8

ARTICLE IN PRESS J.L. Bennett et al. / Journal of Manufacturing Processes xxx (2017) xxx–xxx

6Al-4V. Acta Materialia 2017, http://dx.doi.org/10.1016/j.actamat.2017.04.027 (in press). [19] Wolff S, Liou H, Cao J, Ehmann K, Balogun O. Preliminary study on the influence of process parameters on the porosity of LENS-processed 316L stainless steel. ISCIE/ASME 2014 Int. Sym on Flexible Automation 2014.

[20] Hofmeister W, Griffith M. Solidification in direct metal deposition by LENS processing. JOM 2001;53(9):30–4. [21] Pinkerton AJ, Li L. Modelling the geometry of a moving laser melt pool and deposition track via energy and mass balances. J Phys D Appl Phys 2004;37(14):1885–95.

Please cite this article in press as: Bennett JL, et al. Thermal effect on clad dimension for laser deposited Inconel 718. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.04.024