the influence of laser drilled recast layer thickness on the fatigue performance of Procedia

This paper introduces a novel approach to fatigue life prediction modelling considering the laser drilling effect on film cooling holes of turbine vanes. The methodology proposed is based on a stress-life model such as the Basquin law and the introduction of manufacturing damage effect. The proposed empirical model gives a unique versatility compared to other stress-life models by considering surface damage such as the recast layer produced by the laser drilling process. The proposed empirical model has been thoroughly tested and validated using existing fatigue data. The statistical analysis shows that the proposed model is adequate for estimating the fatigue life of laser drilled specimens considering the recast layer thicknesses effect. The proposed model also can estimate the life of untested specimens even when only a small sample of fatigue data is available, thereby reducing the required testing data.


Introduction
Laser trepan drilling is one of the preferable methods of drilling shaped and complex effusion cooling film hole features in gas turbine engine blades, vanes and combustor liners [1]. However, the fatigue life of such features is highly dependent on the geometrical and metallurgical quality. The quality characteristics of drilled holes are dependent on the laser processing parameters used during the drilling operation. One critical aspect of laser drilling is the formation of a recast layer thickness. The recast layer is the remaining molten material not vaporised or ejected away during the laser drilling process, which re-solidifies and remains attached to the sidewall of the holes [2]. This layer has poor surface roughness, microsegregated microstructure and high tensile residual stresses, often leading to surface cracks, thus, having a detrimental effect on the fatigue performance.

Nomenclature
Process parameters such as power density, pulse duration, trepan speed and associated interaction time are known to influence the recast layer thickness formation in cooling holes produced by laser trepan drilling process [3,4]. Past studies linked desirable geometrical and metallurgical quality characteristics to laser processing parameters according to specific hole shape and dimension, using experiments, statistical modelling or a combination of both [5,6]. However, no studies have modelled fatigue life considering the influence of laser drilling induced recast layer thickness. As such, this paper focuses on developing a methodology and validation of fatigue life prediction model considering the RLT effect on film cooling holes of turbine vane.

Material and preparation of specimens
A second-generation SC nickel-based CMSX-4  superalloy is investigated in this work with a chemical composition presented in Table 1. CMSX-4 is widely used in gas turbines due to its inherent high resistance to creep and fatigue. The material was provided in the form of cast plates of 65mm x 8mm x 2mm in size for laser drilling trials. Fatigue specimens in the form of cast cylindrical bars of 9 mm diameter with the (001) orientation along the axis were produced. The cast plates and bars were supplied from a solution treated, precipitation hardened and hot isostatically pressed material batch in accordance with the collaborating company practices.
Prior to laser drilling , plates were cut, machined, ground, polished and coated with anti-spatter solution as detailed in Ref [7]. The processing parameters of the LTD are shown in Table 2. The preparation of samples for metallography analysis, for example, or the recast layer thickness measurements were conducted according to the standard metallography procedures.
For the fatigue test specimens, cylindrical bars were machined according to dimensions shown in Figure 1. The laser drilling was conducted on a PowerDrill 80 CNC Nd:YAG laser drilling machine at the facilities of the Manufacturing Technology Centre, Ansty. The drilling was performed at an angle of 30 degrees to the fatigue specimen surface and the holes produced were representative of an elliptical and angled shaped effusion cooling film holes. Three levels of laser trepan speed as presented in Table 3 were used to drill the fatigue specimens and model the influence of different RLT on the fatigue performance.

Surface damage measurements
The extent of surface damage produced by laser trepan drilling process was evaluated by measuring the recast layer thickness via SEM. Each trial sample consisted of four holes, in which the RLT was obtained by mean of maximum of 16x4 reading points each equally spaced along the leading and trailing edge side of the holes. Detailed procedure can be found in Ref [7].

Fatigue test
A uniaxial load control type rig with environmental chamber was used to for fatigue tests. Specimens were sprayed with standard salt solution prior to testing and then tested at temperature of 850 o C in an air+SO2 environment to mimic extreme environmental conditions experienced in the internal walls of the cooling holes during service. A load ratio of R = 0 with trapezoidal 1-1-1-1 second waveform resulting in a frequency of 0.25Hz was used for all tests. A total of 24 specimens were tested under LCF regime with a 'run-out' declared at 120,000 cycles.

Fatigue life prediction
Fatigue life prediction methodology using the Basquin's function has been developed, and it is mostly used for fitting the observed stress-life data within the finite region of S-N curve [11]. Basquin's function has been validated for several materials, both notch specimens and smooth specimens [12]. The stress-life curve can be expressed as Where A and B are fitting parameters on the log-log scale S-N curve. The Sa is cyclic applied stress, and Nf is the number of cycles to failure. Several models have been proposed based on the Basquin's function to include the stress ratio and mean stress effects, stress concentration and notch sensitivity, such as Walker equation and the Smith, Watson and Topper's (SWT) model [12]. However, stress ratio effects and notch sensitivity study are beyond the scope of this study. A modified function is used to include the machining surface damage effects into proposed model as 1 1 ( In equation (2), SD is the surface damage due to machining process. The exponent on the SD is the fitting parameter and can be obtained using the test data under different cyclic stresses. Further, to determine each parameter in equation (2), a logarithm form can be expressed as Where A, B and are material constants obtained by the experimental SN data (see Table 4). A simple function has been proposed to determine the material constants by Ref [12] as 11 2 2 y mx m x c   In equation (4), y is equal to log of Nf, x1 is equal to log Sa, and x2 is equal to SDRLT. While, remaining parameters as in Table 5. The obtained material constants for this study are presented in the Table 6.

Modelling the influence of the recast layer
Laser drilling process is known to introduce surface damage such as recast layer and the heat-affected zone around sidewalls of the drilled hole. Latter surface damage is beyond the scope of this study. The thickness of the recast layer is highly influenced by the processing parameters used during the laser drilling operations, such as power density, pulse duration, pulse frequency and trepan speed based on the past studies [4,9]. The recast layers on the laser drilled holes not only introduce tensile stresses but also act as a stress raiser, which is detrimental for the fatigue performance of laser drilled specimens. To include the recast layer thicknesses effects, a polynomial function has been proposed to quantitate the surface damage induced by laser drilling process as In equation (5), 0, 1 and 2 are coefficients of the function. SDRLT is the surface damage recast layer thickness, IP is the power density, and TS is the trepanning speed. The coefficient values ( 0 = 3.14E-4, 1 = 0.22 and 2 = 0.53) are calculated from the experimental data published by authors in Ref [7]. Equations (2) and (5) are combined to predict fatigue life considering the recast layer thickness effects on laser drilled specimens and can be written as The empirical model proposed in equation (6) considers the initial surface damage caused by the key laser drilling process parameters and subsequent fatigue damage caused by the cyclic loading. The temperature, sea salt concentration and sulphur flux inputs were not variable. Therefore, environmental effect was not considered in the proposed model. The laser drilling parameters such as IP and TS values investigated were within a range of values typically used to generate film cooling holes in turbine blades and vanes.  Table 7 indicate that the proposed model (Equation 6) can acceptably predict fatigue life of laser drilled specimens within 95% confidence limits.   In spite of the model predictions provide a good response against the observed data, there is relatively small difference with that of observed data. This stems from the actual fatigue data, where specimens from the same laser drilling conditions yield relatively different RLT's, hence, some scatter on the observed data.

Conclusions
In this study, a methodology for modelling fatigue life prediction of laser drilling specimens using stress-life approach considering the recast layer thickness is proposed. The RLT can be predicted using a polynomial function to experimental data. The proposed model can give reliable predictions under the conditions tested. The main conclusions are as follows: 1. The effect of key laser drilling parameters such as power density and trepanning speed can be modelled by the proposed methodology considering the recast layer thickness effect. 2. The thicker recast layer can significantly lead to high stress concentration and greatly reduce the fatigue strength of cooling holes. This is because thicker recast layer formed after laser drilling operation are brittle, micro segregated, and in some cases with micro cracks. The RLT can act as the main driver for crack nucleation in film cooling holes. Therefore, reducing the fatigue life of laser drilled specimens. 3. The proposed model is suitable for laser trepan drilling of CMSX-4 acute angled cooling holes. Other effusion shaped cooling holes, such as fan shaped, need further investigation because the laser parameters and hole geometry influence on the recast layer thickness might be different.
Overall, the proposed model provides satisfactory prediction responses based on laboratory testing and type of cooling hole shape investigated. It can be applied for understanding of the correlation between the laser drilling parameters, recast layer thickness, and fatigue performance. Also, the proposed model could be applied for parameters optimisation where an increase in productivity demands an increase in the drilling time.