3.1. Macro- and Microstructure
A comprehensive depiction of the cross-section is provided in
Figure 4, encompassing both a broader perspective and a more intricate examination of the macrostructure of both the WAAM and substrate materials. Notably, this cross-section was extracted from the central region of the WAAM structure as shown in
Figure 1. Looking at the overview in
Figure 4, a difference in the microstructure between the substrate and the WAAM material can be observed. The substrate material reveals a more refined microstructure, specifically evident in its grain size, when compared to the WAAM structure itself. This contrast is highlighted by the comparison between the yellow rectangle (representing the substrate) and the red rectangle (representing the WAAM material) in the overview of
Figure 4. Within the highlighted yellow area in
Figure 4, the heat-affected zone (HAZ) is discernible, wherein a gradual grain coarsening is observable from the lower to upper regions. Furthermore, the overview shows a columnar grain structure of the WAAM material, with columnar grains grown perpendicular to the deposition in the Z-direction.
In the detail of the material’s macrostructure in
Figure 4, the columnar grains can also be seen in more detail. In general, the columnar grain size in WAAM material is larger than the microstructure of materials manufactured with other additive manufacturing processes, such as selective laser melting [
24,
25,
26]. The columnar grains in WAAM materials span a few millimeters across multiple layers, which can be attributed to a significantly lower temperature gradient during solidification and cooling in comparison to the other processes [
27]. This lower solidification velocity is attributed to the higher deposition rates employed in the WAAM process. The higher deposition rates necessitate more power and lead to a wider region of the previous layer being either reheated or even liquefied again. In other words, the large heat input results in a lower temperature gradient. One characteristic of the WAAM macrostructure is the presence of layer bands, as seen in
Figure 4. These layer bands are caused by the layer-by-layer manufacturing process and appear similar to a heat-affected zone. The previously deposited layers are repeatedly heated, leading to changes in the microstructure. In the region of these layer bands, the material experiences a temperature below the
transus temperature, resulting in a finer
lamellar transformation structure.
To compare the microstructure of the substrate with WAAM material, SEM images of different specimen regions were taken. The representative figure of the substrate material (
Figure 5a) has been extracted directly from the unprocessed plate, thereby ensuring the absence of any influence stemming from the WAAM process. Furthermore, the representative figure illustrating the WAAM material (
Figure 5b) has been sourced from the central region of the WAAM structure, specifically within the red rectangular region delineated in
Figure 4, precisely within a columnar grain.
Figure 5a shows the SEM image of the substrate, which exhibits a bimodal microstructure consisting of a distribution of equiaxed primary
grains and lamellar
and
phases, whereby these phases are transformed
. In comparison,
Figure 5b shows the microstructure of the WAAM material, which consists of a fully lamellar (
) microstructure. This fully lamellar microstructure is characteristic for WAAM-manufactured parts, whereas SLM-manufactured Ti-6Al-4V shows a martensitic transformation due to high cooling rates [
27].
3.2. Quasi-Static Test Results
Strain-controlled tensile tests were conducted to investigate the elasto-plastic material behavior and determine the initial strain amplitude levels for the LCF tests. As a first step, a comparative analysis was conducted on the stress-strain curves for both material conditions. This process encompassed the presentation of experimentally measured strain data, acquired from the experimental extensometer, in relation to the corresponding measured stress, as elucidated in
Section 2.
Figure 6 shows two representative stress-strain test curves for the substrate and the WAAM material. When comparing both curves, it is evident that the WAAM material exhibits significantly lower yield strength, ultimate tensile strength, and elongation than the substrate material. However, when comparing the Young’s modulus of both material conditions, it is observed that the slope of the elastic regime and the Young’s modulus of the substrate material are approximately 7% higher than those of the WAAM material.
Experimental data from the conducted quasi-static tensile tests were statistically evaluated and the resulting mechanical properties for both material conditions are presented in
Table 2. As previously described, the WAAM material exhibits about −15% lower yield and ultimate tensile strength, a −26% lower elongation, and it is apparent that these material conditions exhibit a slightly higher degree of scatter. These differences in mechanical properties can be attributed to differences in microstructure. Specifically, the substrate exhibits a finer microstructure with smaller grain size than the WAAM material, as shown in
Figure 5. In addition, the lower elongation of the WAAM material can be attributed to the presence of internal defects, such as pores, or the influence of a texture in the material [
28].
When comparing the experimental results from this study to the minimum requirements for forged and selectively laser melted Ti-6Al-4V as specified in ASTM B381 [
29] and ASTM F2924 [
30], respectively, it can be seen that all properties meet or exceed the minimum requirements. When compared with the values in the literature of other AM processes, it can be generally observed that SLM [
24,
31] and cold metal transfer-wire arc additive manufacturing (CMT-WAAM) [
8,
32] exhibit higher mechanical properties than the WAAM process used in this study. This is primarily due to the finer microstructure resulting from a higher temperature gradient in these processes.
3.3. Low-Cycle Fatigue Properties
In order to give an overview and to compare the LCF behavior, the number of load cycles to failure at different total strain amplitudes was compared and is presented in
Figure 7. By comparing the LCF results of the two investigated material conditions, it can be observed that the number of cycles to failure is consistently lower for the WAAM material at all applied strain amplitudes. This is likely due to the presence of larger grains and potential defects, such as pores, in the WAAM material when compared to the substrate material. The presence of pores tends to result in a lower elongation compared to defect-free material and, as a result, in a lower number of load cycles until failure. The difference between the substrate and the WAAM material in terms of cycles to failure becomes in general smaller with increasing amplitudes, see
Figure 7.
During the experimental LCF tests, specimens were cyclically loaded, and depending on the applied total strain amplitude, plastic strain occurred in addition to the elastic part of the total strain. Due to the characteristic cyclic loading in tension and compression, stress-strain hysteresis could be observed. In this regard, strain measurements were acquired via the extensometer, while stress values were computed by dividing the measured force by the diameter of the specimen, as presented in
Section 2.2.
Figure 8 presents the cyclic stress-strain hysteresis for both material conditions and selected strain amplitudes. In detail, the cyclic response with total strain amplitudes of 0.7%, 1.0%, and 1.4% for the substrate is depicted in
Figure 8a,c,e, while the corresponding response for the WAAM material is shown in
Figure 8b,d,f. As materials in general exhibit either cyclic softening or cyclic hardening, it is interesting to investigate whether the stress decreases or increases with each cycle. To study the cyclic behavior regarding softening or hardening, the first and the stabilized cycle at half the number of load cycles to failure are shown in
Figure 8. Additionally, the stress-strain response for 500, 50, and 10 load cycles is shown for the related total strain amplitudes of 0.7%, 1.0%, and 1.4%, respectively.
The presented stress-strain curves allow for technical observations to be made by comparing the material conditions and different strain amplitudes. The occurring minimum and maximum stress increases with higher strain amplitudes for both materials, with the WAAM material generally exhibiting lower stresses at a defined strain amplitude compared to the substrate. Furthermore, when the total strain amplitude is low (
), the maximum stress lies below the yield strength and no plastic strain occurs, which means that the hysteresis is unincisive, as seen in
Figure 8a,b. With an increasing strain amplitude, more plastic strain and an increasing hysteresis can be observed with increasing maximum stress response. Cyclic softening behavior occurs for the investigated material conditions when looking at the stress-strain hysteresis in general.
In order to achieve a more detailed look at the cyclic softening behavior, the evolution of stress for different strain amplitudes over the number of load cycles is presented in
Figure 9. According to [
12,
33], if the ratio between ultimate and yield strength (
) is below 1.2, the likelihood for the occurrence of cyclic softening is high. The observed ratio for the substrate is 1.05 and for the WAAM material is 1.07, indicating that both investigated materials have a tendency to exhibit cyclic softening under cyclic loading. Softening behavior is indicated by a detailed examination of the evolution of minimal and maximal stress in
Figure 9 for different strain amplitudes. No softening is observed for both material conditions for the small strain amplitude of 0.7%. However, for higher strain amplitudes, softening occurs and increases with increasing strain amplitude. Comparing the cyclic softening observed for the two highest strain levels (1.0% and 1.4%), the maximum stress falls in a similar range for both materials. At the same strain amplitudes, the maximum stress for the WAAM material is lower compared to the substrate, as observed in the quasi-static tensile test results.
To quantify the degree of cyclic softening during the LCF tests, the cyclic softening ratio (CSR) can be calculated for both material conditions and different strain amplitudes, as proposed in previous studies such as [
12,
34,
35]. The cyclic softening ratio is calculated using Equation (
6):
where
is the maximum stress range and
is the stress range at half of the reached cycles (
). The CSR, estimated for all tested strain amplitudes and both material states, is presented in
Figure 10. The comparison of softening ratios between the materials shows that only marginal softening occurs below a strain amplitude of 0.75%. However, for strain amplitudes above 0.8%, the softening ratio increases to 15% for the WAAM and 18% for the substrate and reaches a nearly constant value. It is evident that the softening of the substrate material occurs at a later stage compared to that of the WAAM material. Additionally, the substrate exhibits higher cyclic softening compared to the WAAM material, which can be attributed to its higher maximum stress.
In contrast to the stress-strain response observed in monotonic tensile tests, cyclic loading can result in either hardening or softening, as previously mentioned. The cyclic stress-strain curve provides this additional cyclic information. It is determined by the stabilized hysteresis at half of the load cycles to failure (
) of LCF tests at different strain amplitudes. The experimental data of both material conditions are fitted to the previously described Ramberg–Osgood model, and the determined model parameters are presented in
Table 3. In addition,
Figure 11 shows the experimentally determined, stabilized data points for different strain amplitudes and the fitted Ramberg–Osgood stress-strain curves for both material conditions. It can be observed that the substrate exhibits higher values for the corresponding stress amplitude compared to the WAAM material. Specifically, the cyclic hardening coefficient is higher for the substrate while the hardening exponent is slightly smaller.
The parameters of the previously described MCB equation involves four parameters (
), which were determined using the least-squares fitting method by fitting the elastic and plastic strain components measured directly from the experimental LCF data.
Table 4 presents the determined MCB parameters, and
Figure 12 shows the LCF test results for the model fit and the modeled MCB curves. Towards the relationship of total strain and cycles to failure, the elastic and plastic parts are separately shown by straight lines in the figure. From
Figure 12, it can be observed that there is a small gap between the two material conditions for the total strain life curves. The gap between the conditions is increasing with an increasing number of load cycles, corresponding to higher plasticity in the WAAM material. The substrate material has a slightly higher transition point than the WAAM material, implying that plastic deformation has a more significant impact on the fatigue life of the WAAM material than the substrate. This phenomenon could arise from the presence of defects or disparities in the microstructure.