1. Introduction
Laser fusion cutting relies on the combined action of a focused laser beam and a commonly coaxially arranged high-pressure gas jet. While the laser beam melts the material, the pressure gradients and the shear stress of the gas jet blow the molten metal out of the cutting zone. As a result, a cut kerf of particular shape and size is generated that separates both edges of the remaining material along the desired cut contour.
Figure 1 shows a schematic drawing of the laser beam fusion cutting process. Because of the use of nitrogen as a cutting gas, laser fusion cutting does not involve any chemical reaction as in oxygen cutting, and the melting process relies on the beam-matter interaction only [
1,
2].
Despite the fact that the operating principle of laser cutting is quite simple, there is still a lack of a profound understanding of the involved physical mechanisms and inherent interactions that eventually determine the technical performance indicators of the cutting process, i.e., the achievable cutting speed, the cut kerf shape and size, the cut edge roughness and the amount of dross attachment. Recent research work in this field was primarily triggered by the advent of high-power and high-brightness solid-state lasers, i.e., fiber and disk lasers. Indeed, the detected differences between the emerging solid-state laser cutting with fiber and disk lasers with emission wavelengths of about 1 µm and the well-established and industrially proven laser cutting with CO2 lasers with an emission wavelength of 10.6 µm brought the study of laser fusion cutting back into the focus of scientific research interest.
One of the first experimental investigations on this topic was performed by Wandera et al. [
3] who compared disk, fiber and CO
2 laser cutting results on stainless steel in a thickness range between 1 and 6 mm. They reported that the solid-state laser sources are capable of cutting thin-section sheets much faster than the CO
2 laser but give rise to a different cut edge topography with higher values of surface roughness in thicker-section sheets. These principal characteristics were also confirmed by several subsequent studies. Himmer et al. [
4] compared quality and performance of CO
2 and fiber laser cutting in a thickness range between 1 and 10 mm. The achieved higher cutting speeds in fiber laser cutting were reasoned by higher absorption rates and better focus ability as a result of the shorter wavelength of fiber laser radiation. Hilton [
5] described a series of experiments of cutting stainless steel plates from 0.6 to 6 mm in thickness using a disk and a CO
2 laser both operating at 5 kW power. It was shown that the disk laser was capable of cutting thin materials at higher speed and with lower edge roughness than the CO
2 laser but produced lower cut quality in terms of surface roughness for 3 and 6 mm thickness material. Scintilla et al. [
6,
7,
8] performed a comparative study on fusion cutting cold work steels with 1, 5 and 8 mm in thickness using disk and CO
2 laser beams of similar geometry in terms of focal diameter and depth of focus (Rayleigh length). The finding of much higher cutting rates for the disk laser in 1 mm thickness sheets was considered to be indicative for the primary effect of an increased absorption at the shorter wavelength. However, this advantage was found to diminish with increased sheet thickness. This peculiar dependence could be theoretically reasoned by the pronounced dependence of the absorptivity on the angle of incidence according to the Fresnel equations [
9,
10,
11,
12,
13,
14]. It became obvious that a relatively low value of the Brewster angle of about 80° for maximized energy absorption in case of 1 µm wavelength might be a drawback for solid-state lasers because corresponding values of the cut front inclination are typically higher than this value for thick-section cutting. In case of 10.6 µm of the CO
2 laser, a theoretical Brewster angle of about 87° seems to be matching perfectly under those conditions. Besides those energetic constraints, it was also argued that the cutting regime in thick-section sheets could be limited not by the energy consumption but by the melt ejection through the narrow cut kerf [
15]. Sparkes et al. [
16,
17] identified two distinct melt-eject failure mechanisms in fiber laser cutting of medium-section stainless steel in the range of 6–10 mm thickness. First, the boundary layer separation of the cutting gas flow from the cutting front resulted in additional melting and dross attachment at the base of the cut. Second, melt generated at the top of the cut could not be forced down through the kerf and caused internal melt circulation and additional melting through increased residence times. Wandera and Kujanpää [
18] theoretically estimated melt removal rates for thick-section stainless steel laser cutting and showed that particularly the assist gas pressure, the nozzle diameter and the focal point position affect the efficiency of melt removal from the cut kerf. Despite the reported difficulties in obtaining full melt ejection through narrow kerfs in thick-section cutting, dross-free cut edges with acceptable surface roughness were achieved by an adequate control of beam, gas and process parameters, e.g., Wandera and Kujanpää [
19] optimized parameters for fiber laser cutting of stainless steel plates with 10 mm in thickness and achieved surface roughness values R
Z,Max ≈ 80 µm for particular parameter combinations. Goppold et al. [
20] applied different optical setups and beam geometries and adjusted gas pressures, focal plane positions and gas nozzle diameters to achieve acceptable cut edge qualities in fiber laser cutting stainless steel in a thickness range between 1 and 15 mm. In case of the 10 mm thick probes, a surface roughness of R
Z,
Max ≈ 70 µm was recorded. Later, cut quality enhancements were reported for fiber laser cutting mild and stainless steels in 12 mm thickness by use of beam oscillation methods [
21]. Pang and Haecker [
22] introduced optical setups with annular intensity distributions for disk laser cutting and achieved burr-free cuts in 10 mm stainless steel. However, corresponding roughness values were not mentioned.
Despite these achievements, burr-free solid-state laser cut edges still clearly differ in visual appearance from those in CO
2 laser fusion cutting and give typically rise to higher values in terms of surface roughness, particularly for inert-gas stainless steel cutting with sheet thicknesses above 6 mm. A key feature of the cut edge topography is melt striations, and the investigation of their origin, shape and size is still within the focus of scientific research and involves both experimental and theoretical studies. Hirano and Fabbro [
23] observed the striation generation process in inert-gas laser cutting of 3 mm thick mild steel sheets and found instabilities of the melt flow in regions of the kerf front and the cut kerf flank. They concluded that the instability in the side region causes the periodic initiation of cut edge striations. Inherent dependencies on laser wavelength as well as processing parameters were also discussed in detail [
24,
25]. Ermolaev et al. [
26] used the trim-cut method to visualize the melt flow in inert-gas cutting stainless steel sheets of 6 mm thickness by CO
2 and fiber laser radiation and found that the type of radiation, i.e., the laser wavelength, influences the liquid melt flow behavior on the cut front. They reported a melt flow destabilization in case of fiber laser cutting. Arntz et al. [
27] also analyzed the melt flow dynamics by using the trim-cut technique and in situ high-speed video-diagnostics and found that the occurrence of unstable melt streams directly correlates with increased surface roughness. They also reported an effect of multiple reflections on the striation pattern in case of laser fusion cutting with 1 µm wavelength [
28]. Petring [
29] reported a correlation of the cut flank roughness with theoretically calculated cutting front curvatures as well as a negative correlation of the dross height with the calculated maximum penetration of the supersonic gas jet into the kerf. Amara et al. [
30] performed a Computational Fluid Dynamics (CFD) study of the molten film dynamics and investigated the effect of wavelength on temperature distribution, cut front shape and kerf formation. They found higher maximum temperatures, steeper cutting fronts and smaller striations in case of the CO
2 laser wavelength. Recently, the same research group developed a comprehensive numerical model of the laser beam cutting process which accounts for laser absorption, phase transition, heat and mass transfer, fluid flow, kerf formation and gas jet flow [
31].
It is important to keep in mind that the gas flow does not only causes melt ejection failures in case of insufficient flow rates but also affects the striation formation and cut edge roughness. The purpose of the gas flow is to blow the molten material out from the kerf, and thus, it is quite obvious to assume a strong impact of gas characteristics on striation formation and resultant cut edge topography. However, due to the impossibility to visualize those interactions directly and the difficulty to consider all of the relevant effects in corresponding models, gas flow aspects were usually investigated separately. A first comprehensive characterization of the cutting gas flow was given by Fieret [
32,
33]. Amongst others, a direct correlation between gas pressure and achievable cutting speeds was found. In a further fundamental work by Petring et al. [
34] gas dynamic effects within kerfs were approximated by means of modelled kerf channels made of transparent materials that enabled a Schlieren analysis of the gas flow. Since then, much research on cutting gas characteristics, including investigations of different nozzle designs and nozzle arrangements, has been conducted, e.g., Brandt et al. [
35,
36] investigated the effects of nozzle orientation on the gas dynamics of inert-gas laser cutting of mild steel with 1–4 mm in thickness and reported a 50% increase in maximum cutting speed at particular inclination angles of the used tilted off-axis nozzle. Chen et al. [
37] studied gas dynamics effects on cut quality in terms of roughness, dross attachment, and recast layer thickness for laser cutting mild steel sheets of 1.6 mm thickness. It was shown that the cut quality varies with gas pressure and nozzle stand-off distance. Man et al. [
38] reported results from investigations on effects of inlet gas pressure, nozzle stand-off distance, cut kerf width and depth upon the gas jet patterns inside cut kerf models by using the shadow graphic technique. A review on published other works relating to the crucial role of the assist gas in laser beam cutting was recently compiled by Riveiro et al. [
39].
Besides these useful insights into the complex inherent mechanisms of laser beam fusion cutting, it becomes particularly important from a practical point of view to elaborate how the controllable beam and gas jet parameters have to be adjusted to get optimized processing results for a given material and thickness. The problem in this kind of endeavor consists in the multitude of influencing variables and the highly expectable presence of interactions between them. Hence, the use of Design-of-Experiments (DoE) methods seems to be indispensable to meet this challenge. These methods offer tailored designs for factor screenings, factor weighting and revealing of factor–factor interactions by use of fractional and full two-level factorials, as well as for regression modeling of nonlinear functional dependencies and optimization by use of response surface methods [
40,
41]. Indeed, DoE methods were already successfully tested in studies on laser beam cutting, e.g., Son and Lee [
42] investigated CO
2 laser cutting of structural and stainless steel with 2 mm in thickness and applied multiple regression analysis to describe correlations between processing parameters and cutting quality. Huehnlein et al. [
43] reported on the optimization of laser cutting thin alumina layers based on factor screening and response surface designs. Eltawahni et al. [
44] applied DoE methods to relate cutting edge quality quantities to process parameters in CO
2 laser cutting of medical grade AISI 316L stainless steel with a thickness of 2 mm. Tahir and Aquida [
45] demonstrated the identification of optimum parameter ranges for CO
2 laser cutting of 22MnB5 boron steel of 1.7 mm thickness by use of a response surface design. Kechagias et al. [
46] applied a full factorial experimental methodology to analyze surface quality characteristics of 3D printed Polyactic Acid (PLA) plates with 4 mm in thickness cut by use of a CO
2 laser. Sharma and Yadava [
47] combined Taguchi-based experimental designs and grey relational analysis (GRA) to optimize thin sheet Neodymium-doped Yttrium Aluminium Garnet (Nd:YAG) laser cutting of Ni-based super-alloy with consideration of multiple performance characteristics. A general review on optimization techniques in metal cutting processes was given by Mukherjee and Ray [
48].
In the present study, two-level factorial designs are applied to reveal the most vital factors in laser beam fusion cutting of AISI 304 stainless steel with 10 mm thickness. By combining work on experimental laser beam cutting and numerical simulation of cutting gas characteristics, also a light is shed upon the role of inherent processing variables. Furthermore, the significance of oscillation parameters in cutting with circular dynamic beam shaping is evaluated in comparison to static beam parameters.
2. Materials and Methods
Cutting experiments were conducted on sheets of AISI 304 stainless steel (Walzwerke Einsal GmbH, Nachrodt, Germany) with a thickness of 10 mm. The detailed chemical composition of the test material is given in
Table 1.
A 4 kW multimode fiber laser IPG YLR 4000 (IPG Laser GmbH, Burbach, Germany) was used as the beam source in combination with a Precitec HP-SSL cutting head (Precitec GmbH & Co. KG, Gaggenau, Germany). The beam coming out of the delivery fiber with a diameter of 100 µm was collimated by a 100 mm lens and then focused by a lens with 125 mm focal length. The resultant beam caustic was measured with the Primes MicroSpotMonitor tool (Primes GmbH, Pfungstadt, Germany). The measurement indicates an actual focal beam radius of about 80 µm and a Rayleigh length or depth of focus of about 1.7 mm, respectively. The use of such a beam with short Rayleigh length and small spot size can be considered as a typical feature of laser beam cutting with beam oscillation, in which high local beam intensities are desirable to achieve an improved process performance in comparison to conventional laser cutting with a static beam and typically larger spot sizes and Rayleigh lengths. The selected optical configuration can be considered as a standard setup for laser beam oscillation cutting. The scanning unit ScanLab IntelliScan 20 FC (SCANLAB GmbH, Puchheim/München, Germany) was used for the purpose of lateral two-dimensional beam oscillation. In combination with the specified optical setup, this device enables oscillation amplitudes of the focal spot in the range of up to 100 µm in a frequency range of up to 4 kHz. The experimental setup as applied for the cutting trials without and with beam oscillation is shown in
Figure 2. Nitrogen was used as a cutting gas in combination with a nozzle of the conical standard type. Considered factors as controllable independent variables of the cutting process included the laser power, the focal plane position (position of the beam waist in relation to the cutting probe), the cutting gas pressure, the nozzle stand-off distance (distance between nozzle outlet and top surface of the cutting probe), the nozzle diameter, the beam oscillation pattern, the oscillation frequency and the oscillation amplitude. Cutting test samples were produced for each of the considered parameter sets by three successive cuts on base material stripes of 100 mm width. The first and the third cut were carried out as cuts over the full width of the stripes and the second one as partial cut over a length of 50 mm. The resultant test samples with a length of 100 mm (i.e., the width of the original base material stripes), a width of 20 mm and a thickness of 10 mm allow for an assessment of the left and right cut edges as well as a measurement of geometrical cut kerf features.
Roughness values as common evaluation criterion of cut edge quality were determined as result of profile (line) measurements with a Jenoptik Hommel-Etamic T100 wave instrument (JENOPTIK AG, Jena, Germany) on both cut edge sides as R
a (arithmetical mean deviation of the assessed profile) and R
z (average distance between the highest peak and lowest valley) values at different vertical positions of the cut edge as indicated on the right-hand-side in
Figure 2. The cut kerf geometries are characterized in terms of kerf width at the top and bottom surface of the test probes measured with a Keyence VHX-5000 digital microscope (KEYENCE DEUTSCHLAND GmbH, Neu-Isenburg, Germany). This device was also used for an evaluation of the topographic cut edge structure as a whole.
The cutting performance is evaluated in terms of the cutting speed limit, i.e., the achievable maximum cutting speed for a complete cut through the 10 mm thick base material. Preparatory cutting trials with a linear increase of cutting speed with beam spot position up to the point of the process breakdown were conducted to determine this value. The detected cutting speed was confirmed in a second step by performing a full length cut, and eventually used to produce the defined test sample for the particular parameter constellation.
The treatment of the cutting speed as a response of particular parameter constellations is of essential importance for the performed study and the assessment of the results. In general, the cutting speed can also be and even is often used as independent process variable to control the cut edge quality. In particular, the attachment of dross at the lower cut edge is found to be affected by the cutting speed. As a rule of thumb, a cutting velocity of about 80% of the cutting speed limit gives rise to the best cut edge quality whereas values above and below this optimum make both the quality worse. Most important is the point that the optimum cutting speed is always related to the cutting speed limit. Consequently, within a parameter field including different combinations of independent factors that are all initially anticipated to have an impact on the cutting speed limit, it is not reasonable to define fixed levels of cutting speed. Therefore, the cutting speed was considered as a response of the experiments. This makes the cut edge qualities in terms of roughness comparable because they can be regarded in any case as the cut edge qualities at the cutting speed limit. With respect to the purpose of this study, this is not a drawback because the primary aim of this study was the identification of most vital factors and possible interactions but not the optimization of the process.
Additional simulations of the cutting gas flow through the cut kerf were conducted to support the interpretation of the experimental data. For that purpose, a corresponding Computational Fluid Dynamics (CFD) model was developed by use of the commercial software package Ansys Fluent (Version R2, 2019, ANSYS, Inc., Canonsburg, PA, USA). The parametrized model allows for variations of gas pressure, nozzle stand-off distance and diameter, and cut kerf geometry to investigate the impact of these factors on resultant shear-stress distributions on cutting front and cut kerf edges. For the sake of simplicity, the cut edges of the geometrical model are approximated by inclined but even planes, and partial shell surfaces of truncated cones are used to model the cut front, both in accordance with the experimentally measured kerf width values at the top and the bottom surface of the sheet. The physical model solves the conservation equations of energy, mass and momentum of a turbulent flow under steady-state conditions taking the viscous Shear Stress Transport (SST) k-ω turbulence approach into account. Nitrogen as used cutting gas is considered as an ideal gas with the cutting gas pressure p
G specified at the nozzle inlet according to the corresponding experiments and an ambient pressure p
∞ of 0.1 MPa. No-slip adiabatic boundary conditions are applied for all solid boundaries. The computational domain of the model is schematically shown in
Figure 3.
The used designs for both the experimental and numerical investigations were prepared and analyzed by use of the Design-of-Experiment (DoE) software tool DesignExpert (Version 11, 2017, STAT-EASE, Inc., Minneapolis, MN, USA) that also allowed for the randomization of the running sequence. In case of the conventional cutting process and the CFD simulation, a regular fractional two-level factorial design of resolution 25−1 for the considered 5 factors with corresponding 16 different tested parameter constellations was applied, properly supplemented by 5 additional runs at the center point of the design space for curvature check and pure error estimation. In case of the cutting process with beam oscillation and only 4 considered factors, the full factorial design, with 24 = 16 possible parameter constellations and again 5 additional runs at the center point, was applied. With respect to the main purpose of the performed analysis, i.e., the unveiling of the most vital factors and first-order interactions, it is not necessary to detail the corresponding regression models as well as the statistical outcomes and characteristics of the analysis of variance (ANOVA). Therefore, the discussion of the results will remain focused on the qualitative analysis of the results from an engineering point of view. For that purpose, factors and relevant factor–factor interactions with significant effect on the considered responses were selected graphically by means of half-normal probability plots. Based on these selections, regression models were built on a statistical methodology to describe the dependencies between effects and factors. In some cases, data transformations including inverse, logarithmic or exponential approaches were considered to improve the quality of the models. The effects of significant models’ terms are discussed by means of the resultant model graphs.
4. Summary
Conventional and beam oscillation laser cutting were investigated by factorial designs to reveal the most vital factors on cut performance, cut kerf geometry and cut edge roughness. In addition to the conducted experimental trails, also gas flow simulations were performed to support the interpretation of the experiments.
Under the particular conditions of the performed study, the focal plane position is the most decisive factor with respect to the results of the conventional cutting process, whereas the addressable parameters of the gas flow only show secondary effects on the bottom kerf width and the roughness in the bottom edge region. What is particularly interesting in the results is the finding of noticeable correlations between the measured cut edge roughness and the aspect ratio of the cut kerf geometry.
By combining the experimental design with an adequately designed plan for corresponding gas flow simulations, it could be demonstrated that also essential characteristics of the cut kerf gas flow (in particular the backward directed component of the shear stress as well as the shear stress ratio) are strongly affected by the cut kerf shape. As a result, the occurrence of different cut edge categories can be consistently explained by peculiarities of the gas flow. Furthermore, a detected high correlation between the measured roughness values and the magnitudes of the backward directed shear stress component indicates an inherent dependency of cut edge features on gas flow characteristics. To our best knowledge, such a relationship between shear stress and cut edge roughness was never revealed before.
Most important for the understanding is the finding that the magnitude of the backward directed shear stress component is much more influenced by the kerf geometry than by addressable gas parameters, such as gas pressure, nozzle stand-off distance and nozzle diameter. It means that a careful control of the cut kerf geometry by appropriate laser beam parameters might be regarded as the key for an optimized gas flow to ensure a high quality of the cut edge.
Detected relationships between controllable parameters and cutting performance criteria, i.e., cutting speed limit and cut edge quality, are schematically depicted in
Figure 23. Within the investigated parameter space of this study, the laser parameters are vital for the achievable cutting speed limit. They also determine the cut kerf geometry. Cut kerf geometry and controllable gas parameters are crucial for the gas flow characteristics inside the cut kerf. Whereas the controllable parameters gas pressure, nozzle diameter and nozzle stand-off distance determine the magnitude of the shear stress acting in axial direction for blowing out the molten material, the most decisive effect for the magnitude of the backward directed shear stress component results from the cut kerf geometry. This backward directed component seems to influence the cut edge roughness directly. New research indicates that the cut edge structure itself might be particularly affected by the hidden structure of the gas flow boundary layer [
51,
52].
Additional control variables to affect the kerf geometry are addressable for the variant of beam oscillation cutting. In case of the exemplarily investigated circular beam oscillation process, the oscillation amplitude has a direct effect on the top kerf width and the kerf aspect ratio as well. However, more important is the finding that the dependency between bottom kerf width and focal plane position is dramatically changed in comparison to the conventional cutting trials. As a result, the bottom kerf width remains almost constant within the investigated parameter space. This result can be regarded as an implication that some hidden physical mechanisms are acting in beam oscillation cutting which are able to change the characteristics of the cutting process. Those mechanisms being able to influence the direction and strength of the melt film flow in consequence of a modified energy deposition and a changed thermal state of the melt film might involve effects of (i) a changed gas-melt film interaction in dependence on the temperature distribution of the processing zone or (ii) changed fluid dynamical properties (surface tension, viscosity).