Low cycle fatigue behavior of direct metal laser sintered Inconel alloy 718

Low cycle fatigue behavior of direct metal laser sintered Inconel alloy 718

Accepted Manuscript Low cycle fatigue behavior of direct metal laser sintered Inconel alloy 718 Sean Gribbin, Jonathan Bicknell, Luke Jorgensen, Igor ...

1MB Sizes 0 Downloads 65 Views

Accepted Manuscript Low cycle fatigue behavior of direct metal laser sintered Inconel alloy 718 Sean Gribbin, Jonathan Bicknell, Luke Jorgensen, Igor Tsukrov, Marko Knezevic PII: DOI: Reference:

S0142-1123(16)30261-4 http://dx.doi.org/10.1016/j.ijfatigue.2016.08.019 JIJF 4059

To appear in:

International Journal of Fatigue

Received Date: Revised Date: Accepted Date:

19 May 2016 20 August 2016 27 August 2016

Please cite this article as: Gribbin, S., Bicknell, J., Jorgensen, L., Tsukrov, I., Knezevic, M., Low cycle fatigue behavior of direct metal laser sintered Inconel alloy 718, International Journal of Fatigue (2016), doi: http:// dx.doi.org/10.1016/j.ijfatigue.2016.08.019

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Low cycle fatigue behavior of direct metal laser sintered Inconel alloy 718 Sean Gribbina, Jonathan Bicknellb, Luke Jorgensenb, Igor Tsukrova, and Marko Knezevica,* a

Department of Mechanical Engineering, University of New Hampshire, Durham, NH 03824, USA b Turbocam Energy Solutions, Turbocam International, Dover, NH 03820, USA

Abstract In this work, we investigate strength and low cycle fatigue (LCF) life of direct metal laser sintered (DMLS) Inconel 718 superalloy at room temperature. To investigate effects of initial microstructure, the material was deposited in two directions. As a result, the axial loading direction was 45° and 90° with respect to the deposition direction. To further investigate effects of initial microstructure, a few samples of the printed material underwent hot isostatic pressing (HIP). As-printed samples as well as the HIP processed samples were further heat treated (HT) according to AMS 5663 after machining the LCF specimens. To have a reference for the LCF behavior of DMLS Inconel 718, a set of wrought Inconel 718 samples in the same HT condition was also made, and the results critically compared against the results for the DMLS materials. Strain controlled LCF tests were conducted under a mean engineering strain of 0.5% and several strain amplitudes ranging from 0.6% to 1.4%. LCF behavior of DMLS HT material was found to be better than that of wrought HT material at lower strain amplitudes. However, the wrought material had longer life at higher strain amplitudes. The results revealed that the role of porosity present in the DMLS specimens is not as significant at low strain amplitudes as it is at high strain amplitudes. Furthermore, we found that the HIP treatment deteriorated the LCF performance of the material, suggesting that a high content of annealing twins present in the HIPed material microstructure has a larger effect on shortening the life of the samples than porosity. Finally, the Coffin-Manson model was fit to extract the strain-life curves for the studied materials. At lower strain amplitudes, where plastic strain is small, the standard Coffin-Manson model deviated from the data. A bilinear Coffin-Manson model for LCF was found to better capture the data. Keywords: Direct metal laser sintering; Inconel 718; Low cycle fatigue; Coffin-Manson model

*

Corresponding author at: University of New Hampshire, Department of Mechanical Engineering, 33 Academic Way, Kingsbury Hall, W119, Durham, New Hampshire 03824, United States. Tel.: 603 862 5179; fax: 603 862 1865. E-mail address: [email protected] (M. Knezevic).

1

1. Introduction Additive manufacturing (AM) is a rapidly growing field. The focus is shifting from prototyping to manufacturing fully functional end-use parts. Direct metal laser sintering (DMLS), as an AM technology, has been increasingly evaluated to complement production of complex shape parts in aerospace industry [1-3]. DMLS has the advantage of eliminating the need for extensive machining and expensive tooling over traditional manufacturing techniques such as forging [1, 4]. It remains to be confirmed that the parts produced using this novel manufacturing technology perform at the same level as wrought or cast counterparts under various monotonic or cyclic loading conditions and temperatures typical for their expected applications. Inconel 718 is the most widely used commercial Ni based superalloy. It is a γ’ and γ’’ precipitate hardened alloy [5]. It contains approximately three weighted percent of γ’ and nine weighted percent of γ’’ [6]. Additionally, δ precipitates are also present but their fraction is minimized as much as possible to less than one weighted percent [6]. The γ’ precipitate is a Ni3(Al,Ti) compound having a face-centered cubic (FCC) crystal structure [7] and a spherical geometry [5]. The diameter of the γ’ precipitate is between 10 nm and 40 nm [8]. The γ’’ precipitate is a body-centered tetragonal (BCT) Ni3Nb compound [7] with a disk shaped geometry [5]. The γ’ and γ’’ precipitates are similar in size in alloy 718. The δ precipitates have a needle-shaped geometry ranging in length between 1-8 µm [8-10]. These precipitates are generally detrimental to mechanical properties unless they form in spherical morphology at the grain boundaries, where they improve the material’s creep behavior by preventing grain boundary sliding as a creep mechanism [9, 11]. In forged billets, grain structure depends on the location within the billets due to dissimilar heat transfer conditions at various locations. The grain size distribution can range between 10 µm to 200 µm in as-forged Inconel 718 [12]. Several studies have been performed to compare the quality of DMLS built and traditionally manufactured samples of Inconel 718, mostly focusing on the material microstructure and quasistatic properties. Zhao et al. [13] compared tensile strength, elongation, and rupture lifetime of deposited vs. wrought and cast Inconel 718 and related the differences to the base powder type and post treatment conditions. After certain heat treatment conditions, the tensile properties of printed and wrought material were comparable. However, ductility and rupture life were reduced in the deposited material compared to wrought and cast material mainly due to the presence of 2

porosity. Amato et al. [14] performed a comprehensive study characterizing the hardness and tensile properties of DMLS built Inconel 718 as well as microstructural features in as-built, heat treated and hot isostatically pressed (HIP) conditions. Liu et al. [15] investigated the relationship between two laser scanning paths and their associated heat transfer mechanisms to describe differences in microstructure formations and associated mechanical properties. The room temperature tensile strength was similar but ductility varied substantially, which was correlated with inhomogeneity in the grain structure. Scott‐Emuakpor et al. [16] investigated the bending high cyclic fatigue behavior of DMLS Inconel 718 using plate specimens under vibration-based testing. The performance of the material compared well with rotating bending fatigue data for cold-rolled Inconel 718. DMLS can introduce a highly inhomogeneous microstructure with grains elongated in the build direction [13-15, 17, 18]. Furthermore, DMLS materials are known to develop porosity, which is detrimental for material strength and failure behavior [19]. The development of porosity has been attributed to a turbulent condition created by the high temperature melt pool [20]. An additional laser pass, referred to as “re-melting” usually helps reduce porosity in the DMLS built specimens [21]. However, this re-melting pass increases production time. A simultaneous heat and pressure treatment, referred to as hot isostatic pressing (HIP), can be applied to close the internal porosity [22, 23]. In Smith et al. [24] the build orientation and post-treatment processes of DMLS fabricated Inconel 718 were correlated with its microstructure and static mechanical properties. Characterization of the material by electron backscattered diffraction (EBSD) revealed elongated grain structure and formation of <001> fiber texture along the build direction (BD). The mechanical response under tension and compression was found to vary with the testing direction by approximately 7%, which was governed by a combination of grain structure and crystallographic texture. Anisotropy of the material flow during compression was observed to be significant, as well as tension-compression asymmetry in the yield stress. Treatment by HIP was found to lower yield stress but improve ductility relative to the annealed and aged material. The HIPed material had equiaxed grain structure containing large grains, a high content of annealing twins, and a random texture. These observations demonstrated that the behavior of DMLS fabricated Inconel 718 is dependent on the build orientation and post processing treatment of the

3

material. Similarly, Edwards and Ramulu [25] reported anisotropy in fatigue behavior between different build orientations for Ti-6Al-4V alloy. In Xiao et al. [26], the microstructure and low cycle fatigue performance of wrought Inconel 718 at room temperature were investigated relative to the concentration of boron (B) in the material. The strain-life fatigue curve was observed to have an inflection point at a plastic strain amplitude of 0.4%. For this reason, a bilinear Coffin-Manson relation was suggested to better describe the fatigue behavior. The effect of minor elements on properties of Ni-based superalloys has been reviewed in [27]. In this paper, we examine tensile strength and LCF life of DMLS Inconel 718 superalloy at room temperature. In particular, we investigate effects of initial microstructure on the LCF performance while keeping the chemical composition and condition of the material constant. The material was built in the form of rods using DMLS. A few printed pieces underwent further processing by HIP. LCF specimen were subsequently machined from the as-printed and HIPed pieces. All samples were then solution heat treated (HT) according to AMS 5663. To have a reference for the LCF behavior of DMLS Inconel 718, a set of wrought Inconel 718 samples in the same HT condition was also tested, and the results critically compared against the results for the DMLS materials. Strain controlled LCF tests were conducted under a mean engineering strain of 0.5% and several strain amplitudes ranging from 0.6% to 1.4%. The results showed that LCF as well as the tensile strength properties of DMLS double-aged material are better than those of the HIPed and wrought material at lower strain amplitudes. Fatigue performance of DMLS 718 compared well with that of wrought 718 at moderate strain amplitudes, with wrought material having the longest life at higher strain amplitudes. The behavior was explained by the initial microstructure and porosity content in the DMLS specimens. Furthermore, we found that the HIP treatment of the DMLS specimens deteriorates the LCF performances of the material suggesting that a high content of annealing twins present in the HIPed microstructure has a larger effect on shortening the life of the samples than porosity. Finally, the Coffin-Manson model was used to describe the strain-life curves for the studied materials. At lower strain amplitudes, where plastic strain is small, the standard Coffin-Manson model was found to deviates from the data. A bilinear Coffin-Manson model for LCF better represented the data. 2. Material and experiments

4

2.1 Direct metal laser sintering of low cycle fatigue specimens DMLS is defined in ASTM F2792-12a [28] as a “powder bed fusion process used to make metal parts directly from metal powders without intermediate “green” or “brown” parts; term denotes metal-based laser sintering system from EOS GmbH-Electro Optical System.” The underlining strategy involves powder particles being distributed across a build area within an inert gas atmosphere. Directed energy from a laser fuses together the powder particles for the particular layer of the part cross section. The build tray is then lowered, another layer of powder particles is spread on the tray, and the next layer is fused. The process continues layer-by-layer until a full three-dimensional part is complete. For the present study, the samples were manufactured in the EOS M280 DMLS machine using the powder produced by gas atomization with a chemical composition meeting the Inconel 718 standard. The chemical composition of the powder was (wt%) 55.5 Ni, 18.2 Cr, 5.5 Nb, 3.3 Mo, 1.0 Co, 0.35 Si, 0.35 Mn, 0.3 Cu, 1.15 Ti, 0.3 Al, 0.08 C, 0.015 P, 0.015S, 0.006 B, and balance Fe. The particle sizes ranged from <1μm to 80μm with a mean of 35μm in diameter. The default values for the scan speed of 960.0 mm/s and laser power of 285.0 W were used during fabrication of the samples. These parameters were suggested for Inconel 718 by EOS [29]. The same set was used in other studies [30]. Laser spot diameter at the metal pool was roughly 100 μm. The melt pool size and shape depend on the superimposition rate of the different layers during the fabrication of the sample. The size (width and depth) mainly depends on the laser power, the distance or spacing between the scan lines (also called hatching distance), and a path of laser movement (also called hatch pattern or scanning strategy) parameters employed. In this study, the default values of 285 W for the laser power, 0.1 mm for the scan spacing, and 67º rotation for the direction of scanning between consecutive layers in the DMLS EOS machine were used [30, 31]. The 67º rotation between the layers was established to lead to a better overlapping of these. The rotation between consecutive layers avoids matching of melt pool lines until multiple layers had passed. It takes 27 layers for the laser path to line up at 360 degrees again. The layerwise building method can introduce a certain anisotropy in properties of the material. The rotation makes the properties of the material obtained more isotropic in comparison with more conventional scanning strategies made of layers with unidirectional vectors [32]. The directional growth of the columnar dendrites is inhibited using the rotation and, as a result, the 5

material is more isotropic [15]. The material was sintered in 40 μm thick layers. Each layer was built by linear passes of the laser. The build chamber was sealed to prevent atmospheric contaminants from entering the build. Ultra-high purity argon was purged into the chamber and constantly flown over the build area to keep an oxygen content lower than 0.1%. The chemical stability of the material benefited from a low oxygen content as it experiences phase changes throughout the building process. To reduce the stress induced by thermal cycles, the build platform was heated to 80ºC. Thermal conduction throughout the component helped with pre-heating before sintering. The selected laser power was enough for melt pool to reach three layers at a time. This allowed for recrystallization of the deeper layers and the sintering of another. After the core of the cross section was sintered, there were two non-linear passes that followed the outside contour. These passes resulted in dimensional accuracy and surface finish of the part. However, due to the melt pool interruption between the inside and outside contours there was a known area of porosity in the external layer of the part, see [24]. The specimens were designed in accordance with ASTM E606 [33] with a modification to the grip section to fit into the existing flat hydraulic grips, as shown in Fig. 1a. Their gage section is 14 mm long and 6.35 mm in diameter. Material for the samples was printed in two orientations with respect to build direction (BD): diagonal at 45o (D), and horizontal (H), as shown in Fig. 1b. Different build orientations result in different microstructure with respect to a loading direction resulting in anisotropy of material properties [24, 25]. The level of anisotropy is important for designers to know when producing parts in the optimal orientation for fatigue performances in a particular application. The material was printed in the form of rods of 14.70 mm in diameter and 147.00 mm in length. As such, these rods were produced via DMLS process with additional stock material. A few of these rods were randomly selected for HIPing. Thus, the HIPing was done before machining LCF specimens. As-built rods as well as those HIPed were then turned down to the final shape per the drawing in Fig. 1a on a lathe. The grip sections were subsequently milled. The average roughness of machined surfaces in the gage section was measured to be 0.9 μm finish. All specimens were heat treated per AMS 5663 after machining.

6

(a)

(b) Fig. 1. (a) Low cycle fatigue standard specimen geometry with dimensions in mm. (b) Orientation of low cycle fatigue samples with respect to the build direction (BD). 2.2 Heat treatment and hot isostatic pressing The DMLS samples after machining underwent HT per AMS 5663. Post-processing HT is typically applied to improve properties of the DMLS parts because DMLS microstructures usually contain non-equilibrium phases as well as very high residual thermal stresses [34]. Vacuum furnace was used for heat treating the samples. The solution HT was performed at 954 °C for 1 h, followed by fan cooling in argon to below 120 °C. Subsequently, the double aging HT was carried out, which consisted of holding at 718 °C for 8 h, followed by furnace cooling at a rate of 50 °C/h to 621 °C and holding at 621 °C for 8 h, and finally air cooled to room temperature. The HT temperatures were not sufficiently high to recrystallize or to fully stress relieve the material [35].

7

A few printed rods underwent HIP at 1163 °C for 4 h under 100 MPa. The HIP temperature was high enough to trigger recrystallization. HIPing is known to entirely stress relieve the alloy microstructure [13]. LCF specimens were then machined and subsequently underwent the same HT per AMS 5663. The wrought specimens were cut from shaped forgings of Inconel 718 with sample loading direction in LCF perpendicular to the original forging load direction. Since the wrought material was used for comparison with the additively manufactured material, the forgings were subjected to the same heat treatment as the DMLS material as prescribed by AMS 5663 before the LCF samples were machined. The material hardness was measured using a Wilson Model 4-OUR Rockwell Hardness Tester. The hardness values were 28 HRC for the as-built material, 47 HRC for the DMLS material after HT, and 44 HRC for both HIPed and wrought material after HT. The HIPed material had a slightly lower hardness than the heat treated diagonal and horizontal builds and is similar the wrought heat treated material. The hardness was found to be similar along the BD and perpendicular to the BD. 2.3 Initial microstructure Figure 2 shows initial microstructure in the samples manufactured using DMLS followed by HT, DMLS followed by HIP and then HT, and forging followed by HT. The initial microstructures of the three sample categories in the same HT condition were characterized by electron backscattered diffraction (EBSD). The individual samples were mechanically prepared using automated grinding and polishing procedures. To this end, a series of SiC papers were used to grind the specimens. After grinding, the samples were polished on a cloth using 5 µm, 0.5 µm, 0.05 µm alumina suspension. Final polishing of the samples was done with 0.02 µm colloidal silica suspension. The automated EBSD data collection was performed using the Pegasus system (Octane Plus SDD detector and Hikari High Speed Camera) attached to a Tescan Lyra (Ga) field emission scanning electron microscope (SEM) at a voltage of 20 kV. The DMLS followed by HT processed samples show elongated grain structure in the build direction with an approximate aspect ratio of four. As a result, the samples machined in the H direction have a different initial microstructure from the samples machined in the D direction. Crystallographic texture is a <100> fiber of a moderate strength aligned with the build direction [24]. Content of annealing twins is very small as revealed by the misorientation plot. The 8

microstructure in terms of the porosity content was characterized using micro X-ray computed tomography (µXCT) in the earlier work [24], and found to contain approximately 0.18% of porosity. The estimate was regarded as the lower bound. The study reported porosity of the DMLS material build in multiple orientations as well as of the wrought material. Both as-built and HT samples were imaged. The measurement found similar content of porosity within as-built and HT samples. No porosity was found in the wrought material. HIPed material was not characterized since it is supposed to have no porosity. Eliminating pores is the entire purpose of the HIP treatment. The DMLS followed by HIP and then HT samples show equiaxed grain structure and a high content of annealing twins. This finding is important as twin boundaries (60º) are known to be detrimental to fatigue performance of the material. A recent study on fatigue cracks in a Ni superalloy revealed that all the cracks were effectively coincident with twin boundaries [36]. The HIPed material is not supposed to contain any porosity. Wrought followed by HT samples show the typical equiaxed grain structure with some fraction of the twin boundaries, which is much less than in the HIPed material. The grain size is also smaller than in the HIPed material. The above three categories of the Inconel 718 samples in the same HT condition are tested in LCF to evaluate the effect of their associate microstructures on fatigue life.

9

(a)

(b)

(c)

10

Fig. 2. Orientation maps showing the microstructure in the samples of Inconel 718 processed differently to the same final condition: (a) DMLS HT, (b) DMLS HIP HT, and (c) wrought HT. The colors in the maps indicate the orientation of the vertical axis with respect to the crystal reference frame according to the inverse pole figure (IPF) triangle. The scale bar of 500 μm is the same for all the maps. Comparison of grain boundary types in the samples is also shown. Twin boundaries for the annealing twin in the maps are highlighted within the maps in black. 2.4 Low cycle fatigue testing To investigate the fatigue properties of Inconel 718, strain controlled low cycle fatigue experiments were conducted. Specimens were subjected to tension – compression cyclic loading at constant strain amplitudes,

, of either 0.6%, or 0.8%, or 1.0%, or 1.2%, or 1.4% and a mean

strain of 0.5% until failure. The number of cycles to failure,

, was recorded. The mean strain

was used in the experiments to help prevent buckling during the compressive stage of cyclic loading. A small mean strain can be used in low cycle fatigue experiments. There is negligible effect on fatigue life, unless the value is so large that it is a significant fraction of the tensile ductility [37]. The value of 0.5% was elected for the reported tests based on preliminary monotonic studies, as described in section 3.1. Experiments were performed on differently manufactured and treated Inconel specimens at several engineering strain amplitudes as shown in Table 1. The same mean strain of 0.5% was used in all experiments. Figure 3 shows the loading profiles. Table 1. Strain amplitudes chosen for the low cycle fatigue testing. The labels indicate category of the samples with HT meaning heat treated and HIP hot isostatically pressed condition. The build orientation, either horizontal (H) or diagonal at 45º (D), is indicated in parentheses. DMLS HT (D) 1.40% 1.20% 1.00% 0.80% 0.60%

DMLS HT (H) 1.40% 1.20% 1.00% 0.80% 0.60%

Wrought HT 1.40% 1.20% 1.00% 0.80% 0.60%

DMLS HIP HT (H) -1.20% 1.00% ---

11

Engineering Strain (%)

2

a=0.6%

1.5 1

a=0.8%

0.5

a=1.0%

0

a=1.2%

-0.5

a=1.4%

-1 -1.5 -2 0

5

10

15

20

25

30

35

Time (s) Fig. 3. Low-cycle fatigue push-pull waveforms used in testing. The applied mean strain is for all and the amplitude strains are indicated in the legend. These tests were performed using the University of New Hampshire MTS Landmark 270 servohydraulic testing system with Flextest software and controller. The load was recorded using a MTS 661.22H-01 250kN load cell with a maximum error of 0.35%. Specimen elongation was measured using a MTS 634.12E-24 0.5” extensometer. LCF tests were performed using displacement control with strain limits controlled using the extensometer. Strain rates slightly varied but were approximately

and frequency approximately 0.15 Hz.

3. Results The research documented herein examined the tensile and LCF properties of Inconel 718 produced via DMLS and compared these properties against those of wrought Inconel 718 in the same condition. 3.1 Monotonic tensile test results Figure 4 shows results of tension tests performed on the LCF specimens for the studied materials. As seen, the samples made by DMLS followed by HT exhibit the highest strength. However wrought and HIPed samples exhibit more elongation before failure. In Smith et al. [24], it was observed that the DMLS Inconel 718 shows tension-compression asymmetry. To test 12

the alignment of the specimen and investigate whether buckling could be observed using the stress-strain response, a group of LCF specimens were compressed and their stress-strain responses compared. Figure 5 shows the test results for the HIPed and wrought specimens. To further ensure alignment of the setup, the procedure using a strain gauged specimen was followed as explained in ASTM E1012-14 [38]. The material response in tension and compression to large plastic strains was reported in [24]. The data was generated with appropriate samples for tensile (ASTM E8 [39]) and compression testing. In present work, the data in tension was measured using the fatigue samples and is in excellent agreement with the previously reported data. However, the data in compression using fatigue specimens could only be reproduced until the sample preserved its integrity, which is approximately up to 1.5%. As seen in Fig. 5, the stress-strain response in compression bands down. This indicates that the specimen has buckled. In Fig. 5b, it can be seen that the material preserves the integrity without buckling up to 1.5% strain in compression. Loss of the sample integrity was also visually monitored during testing. The strain of 1.5% corresponded to a 45 kN load, which was chosen as the critical load for buckling where deviation would occur from the expected material response in compression. A mean strain of 0.5% was selected based on the preliminary experiments in order to reduce the risk of buckling in the compression region. This positive mean strain ensured that the specimen would be under load much below 45kN in compression during LCF testing at large strain amplitudes.

13

True stress (MPa)

2000

1500

1000 Diagonal Horizontal HIP Wrought

500

0 0

0.05

0.1

0.15

0.2

True strain Fig. 4. True stress-true strain response in tension measured using the LCF specimens for the studied materials. (b) 2000

1500

1000 HIP, tension HIP, compression Wrought, tension Wrought, compression

500

0 0

0.05

0.1

True strain

0.15

0.2

True stress (MPa)

True stress (MPa)

(a) 2000

1500

1000

500

0 0

0.005

0.01

0.015

True strain

Fig. 5. (a) Tension-compression asymmetry for the HIPed and wrought specimens. (b) Zoom in into the response up to about 1.5% strain. 3.2 Low cycle fatigue test results

14

The number of cycles to failure for strain amplitudes ranging from 0.6% to 1.4% is shown in Fig. 6. As expected,

decreases with increase of the strain amplitude for each material

category.

3000

2824

Diagonal Horizontal HIP Wrought

2500

Cycles to failure

2176

2000 1666

1500

1335 973

1000

993 638 625

705 561 555

475

500

397

273 272 259 136

0

0.6%

0.8%

1.0%

1.2%

195 230

1.4%

Strain amplitude Fig. 6. Number of cycles to failure for LCF for the samples of Inconel 718 as indicated in the legend. The stress-strain responses for selected cycles over the life of the specimens are shown in Fig. 7 for the 1.2% strain amplitude. As can be seen, for each of these materials, there is initial cyclic hardening of the material in the first several cycles followed by a regime of softening until the fracture. The dislocation structure formation processes underlining this behavior have been reviewed in [40, 41]. The failure of the DMLS samples was always abrupt, while the wrought material exhibited a gradual failure.

15

Diagonal

(a)

1500

Nf=273

True stress (MPa)

True stress (MPa)

1000 500 0 -500 -1000

-1500 -0.02

-0.01

0

0.01

(c)

Nf=272

500 0 -500 -1000 -0.01

0

0.01

0.02

0.01

0.02

True strain HIP

(d)

1500

1500

Nf=475

True stress (MPa)

True stress (MPa)

1000

-1500 -0.02

0.02

True strain Wrought 1000

Horizontal

(b)

1500

500 0 -500 -1000 -1500 -0.02

-0.01

0

True strain

0.01

0.02

1000

Nf=259

500 0 -500 -1000 -1500 -0.02

-0.01

0

True strain

Fig. 7. 1.2% strain amplitude stress-strain hysteresis loops for (a) DMLS HT diagonal (b) DMLS HT horizontal (c) wrought HT and (d) DMLS HIP HT samples. Figure 8 shows the stress-amplitudes for each strain-amplitude tested. Here, it can be seen that the DMLS diagonal sample exhibits the highest amplitude stress over all the other tested samples during the life of the specimens since it exhibits the highest strength. None of the curves shows stable cyclic stress amplitude. The cyclic softening is increasing with the strain amplitude. For most of the samples tested, the specimen underwent a sudden “brittle” failure with rapid crack propagation. Interestingly, failure of the wrought specimens was not so abrupt.

16

(b)

1250

a (MPa)

a (MPa)

1500 1000 750 500 250 0 0 10

(c)

10

1

10

2

10

3

10

(d)

a (MPa)

a (MPa)

1250 1000 750 500 250 10

1

a=1.4% 10

2

10

3

10

4

1000

500

750 250 0 0 10

750

500

10

1

10

2

10

3

10

Cycle a=1.2% 1500 1250 1000 750 500 0 0 10

10

1

10

2

10

3

10

10

1

10

2

10

3

10

4

4

Cycle

8. Stress-amplitudes for DMLS HT diagonal, DMLS HT horizontal, and wrought 500Fig. samples for strain amplitudes of (a) 0.6% (b) 0.8% (c) 1.0% (d) 1.2% and (e) 1.4%.

250

4

Diagonal Horizontal Wrought

1250

1250

750

Cycle

1500 1000

1000

250

Cycle a=1.4%

1500(e)

1250

0 0 10

4

1500

0 0 10

a=0.8% 1500

250

Cycle a=1.0%

a (MPa)

a (MPa)

a=0.6%

(a)

Figure 9 shows the mean stress,

HT

, for each strain-amplitude tested. It can be seen that for

the larger strain amplitudes, the mean stress is almost immediately accommodated by plastic

0deformation 0 2 with continuation of the 4 test. Moreover, it becomes and approaches zero 10 10 10 compressive for the wrought material. This is a direct consequence of the tension-compression

Cycle

asymmetry exhibited by the material with the compressive strength being higher than the tensile strength (Fig. 5). Compressive

is highly desirable for extending the fatigue life. For lower

strain-amplitude tests the mean stress is present and is decreasing throughout the duration of the test.

17

a=0.6%

10

1

750 500 250

2

10

3

10

4

400 300 200 100 0 -100 -200 -300 -400 0 10

(d)

a=1.4%  =1.4%

10

1

10

2

10

3

10

4

Cycle a=1.4% a

a=0.8% 400 300 200 100 0 -100 -200 -300 -400 0 10

10

1

10

2

10

3

10

4

Cycle a=1.2% 400 300 200 100 0 -100 -200 -300 -400 0 10

10

1

10

2

10

3

10

4

Cycle

Diagonal Diagonal 200 300 100 Horizontal 0 Horizontal 200 -100 Wrought -200 100 Wrought -300 -400 010 10 10 10 10 Cycle -100 Fig. 9. Evolution of mean stress during the tests for the materials indicated in the legend under -200 strain amplitudes of: (a) 0.6% (b) 0.8% (c) 1.0% (d) 1.2% and (e) 1.4%. -300 Figure 10 shows the reduction in macroscopic elastic slope from the first cycle to the half-life -400 0 2 4 of the the more reduction there is in 10specimen. Generally, as10the strain amplitude increases, 10 m (MPa)

1000

10

Cycle a=1.0%

400 1500 (e) 400 300

1250

(b)

m (MPa)

400 300 200 100 0 -100 -200 -300 -400 0 10

m (MPa)

m (MPa) m (MPa)

(c)

Mean Stress (MPa)

a

(a)

0

1

2

3

4

0 0elastic slope. It is believedCycle that as a consequence of increase in 2 the accumulation of backstresses 4 10 dislocation density in the10 10 in elastic modulus [42-44]. It is material accounts for this reduction

Cycle

also possible that porosity growth can have an effect on the elastic slope. The dominant mechanism responsible for it warrants further investigation in future efforts.

18

10

Percent reduction

8

Diagonal Horizontal HIP Wrought

220.7 8.02

Initial E (GPa)

6

4

2

0

212.6 5.36

218.1 5.00

215.0 2.32 190.8 1.84 203.2 1.47

0.6%

195.8 206.0 2.09 2.25

203.3 2.60 191.4 1.27

0.8%

201.8 4.17

192.5 3.75

196.6 3.59

216.2 3.44

194.8 4.94

206.6 4.48

1.0%

202.2 4.22

201.3 3.17 195.5 1.97

1.2%

1.4%

Strain Amplitude Fig. 10. Reduction in elastic slope to the half-life of the LCF specimens

4. Discussion The DMLS HT material is stronger in simple tension than the wrought HT material, which is likely due to a finer grain structure. However the DMLS HT material is not as ductile as the wrought HT material. The limited ductility exhibited by the DMLS HT samples is due to the presence of porosity, which is absent in the wrought HT material. The diagonal DMLS HT samples exhibit a slightly higher strength than the horizontal DMLS HT material. This is due to the grain shape effects relative to the loading direction. The mean free path of dislocations moving along the loading direction is longer in the diagonal than in the horizontal sample grain structure. The longer the mean free path of dislocations, the less hardening is accumulated in the material [45-52]. Finally, the curve of the DMLS HIP HT material is below the DMLS HT material curves likely due to coarser grain structure, but ductility of the HIPed material has substantially increased. The loss of ductility in the DMLS HT samples is due to the presence of porosity, which is supposed to be eliminated in the HIPed material. After HIP, the microstructure transforms from being elongated to be equiaxed, and the grain size is much coarser as compared to the grain size of DMLS HT material (Fig. 2). Additionally, the fraction of grain boundaries associated with annealing twin substantially increases after HIP. Consistently with the behavior of the wrought HT material, the number of cycles to failure for the DMLS HT material decreases with the increase of strain amplitude. However, the rate of this decrease for the DMLS HT material is higher than for the wrought HT material. Such 19

behavior is likely due to the presence of porosity, which is absent in the wrought material [24]. For lower strain amplitudes, the porosity in the DMLS HT material does not play as much of a role in the degradation of the material as it does for higher strain amplitudes. Therefore, the deformation of the material at small strains, which is just above the initial yield stress, is not as much influenced by the porosity as it is later in the plastic region, where strain hardening is playing a significant role. As a result, the DMLS HT material, which has a higher strength, also has a longer LCF life than the wrought HT material for low strain amplitudes. The diagonal DMLS HT material has a longer life than the horizontal DMLS HT material since it has a slightly higher strength due to grain structure. As the strain amplitudes increase, the plastic deformation mechanisms and the underlying ductility becomes more important. Therefore the more ductile wrought material exhibits better fatigue performances. Although the HIPed material appears to be as ductile as the wrought material, its LCF behavior is not as good. The shorter life of the HIPed material for the same strain amplitude can be explained by large content of annealing twins in its microstructure. The twin boundaries shorten fatigue life of the material because fatigue cracks in a nickel-based superalloy are known to nucleate and propagate along the annealing twin boundaries [36]. The strain-life data for LCF can be approximated by the Coffin-Manson relation [26, 37]. Here, the strain amplitude

where

is broken up into elastic and plastic components

is the elastic strain amplitude,

50% of the specimen’s life, and

and

,

(1)

,

(2)

are the stress amplitude and elastic modulus at

is the plastic strain amplitude.

The elastic and plastic components of strain amplitude are plotted against

in a log-log

scale. These data points are then fitted with the Coffin-Manson relations [53-55]. ,

(3)

,

(4)

20

where

is the fatigue strength coefficient,

specimens category,

is the elastic modulus during the initial loading per

is the fatigue strength exponent,

is the fatigue ductility coefficient, and

is the fatigue ductility exponent. Since the data is plotted in a log-log scale, the Coffin-Manson model becomes linear, with b and c as the slopes and

and

as the intercepts for the elastic

and plastic curves, respectively. Here, the value of E at 50% life cannot be used since E changes depending on the strain amplitude (see Fig. 10). Note that (3) and (4) are used to describe strain life behavior based on a set of experiments conducted at different strain amplitudes. The measured data for the DMLS HT material is plotted in Fig. 8 along with available literature data for Inconel 718 in the same condition [26] and the strain-life Coffin Manson curves for generic Inconel from Dowling [37]. Note that the literature data was provided for the fully reversed strain cycles.

-1

10

Diagonal Horizontal -- Dowling (2013) Xiao (2005)

Total Elastic Plastic

-2

a

10

-3

10

-4

10

1

10

2

10

3

10

4

10

5

10

Nf Fig. 11. Total, elastic, and plastic strains for horizontal and diagonal DMLS HT specimens tested here compared to the literature data in Dowling [37] and Xiao et al. [26].

21

It can be seen that the Coffin-Manson parameters from Dowling [37] do not accurately describe all the tested material conditions. As shown earlier, elastic and plastic properties vary with the difference in initial microstructure between the H-built and D-built DMLS as well as the wrought material conditions. Therefore, in this work we establish the Coffin-Manson coefficients separately for each condition of the material. The individual strain-life curves are plotted in Figs. 12-14.

Diagonal

-1

10

Total Elastic Plastic Plastic, bilinear

-2

a

10

-3

10

-4

10

2

10

3

10

4

10

Nf Fig. 12. The Coffin-Manson model fit of the measured data for the D-built DMLS Inconel 718 in the HT condition.

22

Horizontal

-1

10

Total Elastic Plastic Plastic, bilinear

-2

a

10

-3

10

-4

10

2

10

3

10

4

10

Nf Fig. 13. The Coffin-Manson model fit of the measured data for the H-built DMLS Inconel 718 in the HT condition.

Wrought

-1

10

Total Elastic Plastic Plastic, bilinear

-2

a

10

-3

10

-4

10

2

10

3

10

4

10

Nf Fig. 14. The Coffin-Manson model fit of the measured data for the wrought Inconel 718 in the HT condition.

23

Figures 12-14 present two choices for the

curve fit. The first choice, shown as a

solid line, corresponds to the traditional formulation of the Coffin-Manson model when results are fitted over the entire range of plastic strain amplitudes. However, as was observed in Xiao at al. [26], there is an inflection point in the

relation for plastic strain amplitudes below

0.4%. For this reason, Xiao at al. [26] proposed an extended Coffin-Manson relation for the curve, which is a bilinear fit in the log-log scale. The same inflection point was observed for all the tested materials in this work, and thus the bilinear fit better describes the data. These bilinear fit curves are plotted as dashed lines in Figs. 12-14. The parameters used to create the Coffin-Manson curves are presented in Table 2 for the full range and Table 3 for the extended bilinear fits. The correction coefficients demonstrate that the data is well represented by the fits.

Table 2. Parameters of the Coffin-Manson model for the DMLS and wrought Inconel 718. E (GPa) Dowling (2013) Diagonal Horizontal Wrought

(MPa)

214

2255

216.51 193.94 203.51

2011.2 2115.7 2463.4

b -0.117

c 1.16

-0.749

-0.0704 0.4014 -0.6670 -0.0819 1.5919 -0.8773 -0.1141 2.2793 -0.8763

Table 3. Parameters of the bilinear Coffin-Manson model for the DMLS and wrought Inconel 718.

Diagonal Horizontal Wrought

≥ 0.40% Correlation c Coefficient 0.0652 -0.3745 0.9999 0.1821 -0.5410 0.9708 0.1458 -0.4614 0.9450

≤ 0.40% c 4.2027 74.454 134.879

-0.9622 -1.3606 -1.4076

Correlation Coefficient 0.9965 0.8660 0.9986

The work herein examined LCF behavior of the Ni alloy Inconel 718 produced via DMLS in multiple build orientations in the as-built followed by HT condition and HIPed followed by HT condition. The comparison with the wrought material in the same HT condition was also made. High cyclic fatigue (HCF) behavior was not pursued as part of the effort. HCF behavior of the 24

same material in multiple build orientations with different surface finishes was examined in a recent study [30]. However the samples were either built to the dimensions or built with additional stock material and subsequently turned down to the final shape on a lathe meaning that they were not in the heat treated condition as the specimens in the present study. The study found that surface roughness resulting from the DMLS process had a significant impact on HCF behavior of the material. In addition, the measured fatigue strength of the horizontally built specimens was greater than that of specimens built in the vertical orientation. The reader is referred to [30] for details.

5. Conclusions DMLS can potentially reduce the cost of manufacturing complex geometry parts. This work showed that Inconel 718 manufactured by DMLS exhibits strength and LCF properties comparable to those of Inconel 718 traditionally manufactured by forging. Specifically, DMLS material lasts longer in LCF than the wrought material in the same condition at low but not at high strain amplitudes. The behavior is rationalized in terms of the DMLS induced porosity present in the microstructure because its influence is not as significant at low strain amplitudes as it is at high strain amplitudes. We find that the HIP treatment deteriorates the LCF performance of the material due to a high content of annealing twins in the microstructure created during the HIP treatment. The presence of annealing twins in the microstructure has larger effect on shortening the life of the material than porosity at moderate to large strain amplitudes. The Coffin-Manson model was fit to delineate the strain-life curves for the studied materials. It was observed that at lower strain amplitudes, where plastic strain is small, the standard CoffinManson model deviated from the data. An extended, bilinear, Coffin-Manson model for LCF was found to better represent the measured data for the considered range of plastic strain amplitudes. In summary, the work showed that porosity and inhomogeneities within grain structure such as non-uniform grain shape and annealing twins are yet to be reduced in the microstructure before DMLS becomes a comprehensively superior manufacturing method for components operating under complex loading.

25

Acknowledgments This work is part of a project supported by Turbocam Energy Solutions and the New Hampshire Innovation Research Center under grant No. 13R217. The DMLS samples of Inconel 718 were manufactured by Turbocam. The authors gratefully acknowledge this support. The EBSD work was performed in the University Instrumentation Center (UIC) at the University of New Hampshire. The authors wish to acknowledge help from Nancy Cherim for assistance with operating the microscope at UIC. References [1] E. Herderick, Progress in Additive Manufacturing, JOM, 67 (2015) 580-581. [2] D. Gu, W. Meiners, K. Wissenbach, R. Poprawe, Laser additive manufacturing of metallic components: materials, processes and mechanisms, International materials reviews, 57 (2012) 133-164. [3] W.E. Frazier, Metal additive manufacturing: a review, J. of Materi Eng and Perform, 23 (2014) 1917-1928. [4] D. Cooper, J. Thornby, N. Blundell, R. Henrys, M.A. Williams, G. Gibbons, Design and manufacture of high performance hollow engine valves by Additive Layer Manufacturing, Materials & Design, 69 (2015) 44-55. [5] G.A. Rao, M. Kumar, M. Srinivas, D. Sarma, Effect of standard heat treatment on the microstructure and mechanical properties of hot isostatically pressed superalloy inconel 718, Materials Science and Engineering: A, 355 (2003) 114-125. [6] C. Wang, R. Li, Effect of double aging treatment on structure in Inconel 718 alloy, J Mater Sci, 39 (2004) 2593-2595. [7] K. Sano, N. Oono, S. Ukai, S. Hayashi, T. Inoue, S. Yamashita, T. Yoshitake, γ″-Ni3Nb precipitate in Fe–Ni base alloy, Journal of Nuclear Materials, 442 (2013) 389-393. [8] M. Burke, M. Miller, Precipitation in alloy 718: A combined Al3M and apfim investigation, in: E.A. Loria (Ed.) Superalloys 718, 625 and various derivatives, The Minerals, Metals & Materials Society, 1991. [9] S. Ghosh, S. Yadav, G. Das, Study of standard heat treatment on mechanical properties of Inconel 718 using ball indentation technique, Materials Letters, 62 (2008) 2619-2622. [10] S. Azadian, L.-Y. Wei, R. Warren, Delta phase precipitation in Inconel 718, Materials Characterization, 53 (2004) 7-16. [11] J.R. Davis, ASM specialty handbook: heat-resistant materials, ASM International, 1997. [12] A. Chamanfar, L. Sarrat, M. Jahazi, M. Asadi, A. Weck, A. Koul, Microstructural characteristics of forged and heat treated Inconel-718 disks, Materials & Design, 52 (2013) 791800. [13] X. Zhao, J. Chen, X. Lin, W. Huang, Study on microstructure and mechanical properties of laser rapid forming Inconel 718, Materials Science and Engineering: A, 478 (2008) 119-124. [14] K. Amato, S. Gaytan, L. Murr, E. Martinez, P. Shindo, J. Hernandez, S. Collins, F. Medina, Microstructures and mechanical behavior of Inconel 718 fabricated by selective laser melting, Acta Materialia, 60 (2012) 2229-2239. [15] F. Liu, X. Lin, C. Huang, M. Song, G. Yang, J. Chen, W. Huang, The effect of laser scanning path on microstructures and mechanical properties of laser solid formed nickel-base superalloy Inconel 718, Journal of Alloys and Compounds, 509 (2011) 4505-4509. 26

[16] O. Scott‐Emuakpor, J. Schwartz, T. George, C. Holycross, C. Cross, J. Slater, Bending fatigue life characterisation of direct metal laser sintering nickel alloy 718, Fatigue & Fracture of Engineering Materials & Structures, 38 (2015) 1105-1117. [17] A. Simchi, H. Pohl, Effects of laser sintering processing parameters on the microstructure and densification of iron powder, Materials Science and Engineering: A, 359 (2003) 119-128. [18] A. Riemer, S. Leuders, M. Thöne, H. Richard, T. Tröster, T. Niendorf, On the fatigue crack growth behavior in 316L stainless steel manufactured by selective laser melting, Engineering Fracture Mechanics, 120 (2014) 15-25. [19] T.M. Pollock, S. Tin, Nickel-based superalloys for advanced turbine engines: chemistry, microstructure and properties, Journal of propulsion and power, 22 (2006) 361-374. [20] L.L. Parimi, G. Ravi, D. Clark, M.M. Attallah, Microstructural and texture development in direct laser fabricated IN718, Materials Characterization, 89 (2014) 102-111. [21] E. Yasa, J.-P. Kruth, Microstructural investigation of Selective Laser Melting 316L stainless steel parts exposed to laser re-melting, Procedia Engineering, 19 (2011) 389-395. [22] S.-H. Chang, S.-C. Lee, T.-P. Tang, H.-H. Ho, Influences of soaking time in hot isostatic pressing on strength of Inconel 718 superalloy, Materials Transactions, 47 (2006) 426-432. [23] S.-C. Lee, S.-H. Chang, T.-P. Tang, H.-H. Ho, J.-K. Chen, Improvement in the microstructure and tensile properties of Inconel 718 superalloy by HIP treatment, Materials Transactions, 47 (2006) 2877-2881. [24] D.H. Smith, J. Bicknell, L. Jorgensen, B.M. Patterson, N.L. Cordes, I. Tsukrov, M. Knezevic, Microstructure and mechanical behavior of direct metal laser sintered Inconel alloy 718, Materials Characterization, 113 (2016) 1-9. [25] P. Edwards, M. Ramulu, Fatigue performance evaluation of selective laser melted Ti–6Al– 4V, Materials Science and Engineering: A, 598 (2014) 327-337. [26] L. Xiao, M. Chaturvedi, D. Chen, Low-cycle fatigue behavior of INCONEL 718 superalloy with different concentrations of boron at room temperature, Metallurgical and materials Transactions A, 36 (2005) 2671-2684. [27] R. Holt, W. Wallace, Impurities and trace elements in nickel-base superalloys, International Metals Reviews, 21 (1976) 1-24. [28] F2792 12a, in: Standard Terminology for Additive Manufacturing Technologies, ASTM International, West Conshohocken, PA, 2012. [29] EOS NickelAlloy IN718 Data Sheet, EOS GmbH - Electro Optical Systems, http://www.eos.info/material-m (2014). [30] P.F. Kelley, A. Saigal, J.K. Vlahakis, A. Carter, Tensile and Fatigue Behavior of Direct Metal Laser Sintered (DMLS) Inconel 718, in: ASME 2015 International Mechanical Engineering Congress and Exposition, American Society of Mechanical Engineers, 2015, pp. V02AT02A001-V002AT002A001. [31] D. Manfredi, E. Ambrosio, F. Calignano, M. Krishnan, R. Canali, S. Biamino, M. Pavese, E. Atzeni, L. Iuliano, P. Fino, Direct Metal Laser Sintering: an additive manufacturing technology ready to produce lightweight structural parts for robotic applications, La metallurgia italiana, 10 (2014) 15-24. [32] D. Manfredi, F. Calignano, M. Krishnan, R. Canali, E.P. Ambrosio, E. Atzeni, From powders to dense metal parts: Characterization of a commercial AlSiMg alloy processed through direct metal laser sintering, Materials, 6 (2013) 856-869. [33] ASTM E606/E606M-12, in: Standard Test Method for Strain-Controlled Fatigue Testing, ASTM International, West Conshohocken, PA, 2012. 27

[34] E. Chlebus, K. Gruber, B. Kuźnicka, J. Kurzac, T. Kurzynowski, Effect of heat treatment on the microstructure and mechanical properties of Inconel 718 processed by selective laser melting, Materials Science and Engineering: A, 639 (2015) 647-655. [35] F. Liu, X. Lin, G. Yang, M. Song, J. Chen, W. Huang, Microstructure and residual stress of laser rapid formed Inconel 718 nickel-base superalloy, Optics & Laser Technology, 43 (2011) 208-213. [36] C.A. Stein, A. Cerrone, T. Ozturk, S. Lee, P. Kenesei, H. Tucker, R. Pokharel, J. Lind, C. Hefferan, R.M. Suter, A.R. Ingraffea, A.D. Rollett, Fatigue crack initiation, slip localization and twin boundaries in a nickel-based superalloy, Current Opinion in Solid State and Materials Science, 18 (2014) 244-252. [37] N.E. Dowling, Mechanical behavior of materials: Engineering methods for deformation, fracture and fatigue 4th ed., Pearson Education, Inc, Upper Saddle River, NJ, 2013. [38] ASTM E1012-14, in: Standard Practice for Verification of Testing Frame and Specimen Alignment Under Tensile and Compressive Axial Force Application, ASTM International, West Conshohocken, PA, 2014. [39] ASTM E8/E8M-15a, in: Standard Test Methods for Tension Testing of Metallic Materials, ASTM International, West Conshohocken, PA, 2015. [40] S.D. Antolovich, Microstructural aspects of fatigue in Ni-base superalloys, Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 373 (2015) 20140128. [41] D.D. Krueger, S.D. Antolovich, R.H.V. Stone, Effects of grain size and precipitate size on the fatigue crack growth behavior of alloy 718 at 427 °C, MTA, 18 (1987) 1431-1449. [42] M. Zecevic, M. Knezevic, A dislocation density based elasto-plastic self-consistent model for the prediction of cyclic deformation: Application to Al6022-T4, International Journal of Plasticity, 72 (2015) 200-217. [43] J.A. Wollmershauser, B. Clausen, S.R. Agnew, A slip system-based kinematic hardening model application to in situ neutron diffraction of cyclic deformation of austenitic stainless steel, International Journal of Fatigue, 36 (2012) 181-193. [44] M. Zecevic, Y.P. Korkolis, T. Kuwabara, M. Knezevic, Dual-phase steel sheets under cyclic tension–compression to large strains: Experiments and crystal plasticity modeling, Journal of the Mechanics and Physics of Solids, 96 (2016) 65-87. [45] H. Mecking, U.F. Kocks, Kinetics of flow and strain-hardening., Acta Metallurgica et Materialia, 29 (1981) 1865-1875. [46] M. Knezevic, J.S. Carpenter, M.L. Lovato, R.J. McCabe, Deformation behavior of the cobalt-based superalloy Haynes 25: Experimental characterization and crystal plasticity modeling, Acta Materialia, 63 (2014) 162-168. [47] M. Zecevic, R.J. McCabe, M. Knezevic, Spectral database solutions to elasto-viscoplasticity within finite elements: Application to a cobalt-based FCC superalloy, International Journal of Plasticity, 70 (2015) 151-165. [48] M. Knezevic, M. Zecevic, I.J. Beyerlein, J.F. Bingert, R.J. McCabe, Strain rate and temperature effects on the selection of primary and secondary slip and twinning systems in HCP Zr, Acta Materialia, 88 (2015) 55-73. [49] M. Knezevic, I.J. Beyerlein, D.W. Brown, T.A. Sisneros, C.N. Tomé, A polycrystal plasticity model for predicting mechanical response and texture evolution during strain-path changes: Application to beryllium, International Journal of Plasticity, 49 (2013) 185-198.

28

[50] M. Knezevic, L. Capolungo, C.N. Tomé, R.A. Lebensohn, D.J. Alexander, B. Mihaila, R.J. McCabe, Anisotropic stress-strain response and microstructure evolution of textured -uranium, Acta Materialia, 60 (2012) 702-715. [51] M. Zecevic, M. Knezevic, I.J. Beyerlein, C.N. Tomé, An elasto-plastic self-consistent model with hardening based on dislocation density, twinning and de-twinning: Application to strain path changes in HCP metals, Materials Science and Engineering: A, 638 (2015) 262-274. [52] M. Knezevic, A. Levinson, R. Harris, R.K. Mishra, R.D. Doherty, S.R. Kalidindi, Deformation twinning in AZ31: Influence on strain hardening and texture evolution, Acta Materialia, 58 (2010) 6230-6242. [53] O. Basquin, The exponential law of endurance tests, in: Proceedings of the American Society for Testing and Materials 1910, pp. 625-630. [54] L.F. Coffin Jr, A study of the effects of cyclic thermal stresses on a ductile metal, Transactions of the ASME, 76 (1954) 931-950. [55] P.P. Gillis, Manson-Coffin fatigue, Acta Metallurgica, 14 (1966) 1673-1676.

29

Highlights LCF of alloy 718 in a given heat treated state is studied in function of initial microstructure: Asbuilt, HIPed, and wrought DMLS induced porosity and content of annealing twins are major microstructural features influencing LCF life of the material DMLS material lasts longer in LCF than the wrought material in the same condition at low but not at high strain amplitudes Coefficients of extended Coffin-Manson LCF model are established to fit the data for each condition of the material

30