Characterization of Surface Integrity of 3D-Printed Stainless Steel by Successive Grinding and Varied Burnishing Parameters

Abstract

Additive manufacturing (AM)’s ability to produce customized products with reduced material wastage and other advantages helped the technology to gain popularity in many industries. However, its poor surface integrity is its weak side, and to overcome this, additional post-processes are essential. Slide diamond burnishing, known for its enhancement of surface roughness, residual stress, microhardness, and other properties, was combined with grinding in this research after 3D printing of MetcoAdd 17-4PH-A to mitigate the mentioned shortcomings. This study aimed to analyze the effects of each process on surface roughness, residual stress (both on the surface and in-depth), and microhardness. Workpieces were ground with the same parameters and burnished with four levels of force, feed, and number of passes. The L16 Taguchi experimental design was used to optimize the process parameters and to study their effects. For surface roughness, the optimum parameters were found to be 60 N force, 0.02 m/min feed rate, and three passes. The longitudinal surface residual stress has optimal values at 80 N force, 0.02 m/min feed rate, and four passes. In the case of transverse surface residual stress, the optimal values were 60 N force, 0.17 m/min feed rate, and three passes. Microhardness was maximized with 60 N force, 0.02 m/min feed rate, and one pass. Additionally, the in-depth residual stress for selected surfaces was investigated, and 100 N force showed a deep burnishing effect. Further multi-objective optimization using desirability function analysis found that the optimal parameters for all responses were achieved at the fourth burnishing force level (100 N), the first tool feed level (0.02 m/min), and the fourth number of passes level (four passes). Ultimately, both grinding and burnishing processes exhibited significant enhancements in the measured parameters.

1. Introduction

The importance of surface finish cannot be overlooked, as it impacts various product qualities, such as critical character fatigue life, friction, wear resistance, and lubrication retention, which can be extended to operational noise level increase. Proper surface finish can prevent the premature failure of a component. Despite its high surface roughness, dimensional accuracy, isotropic character, long printing time, and other shortcomings that attract many researchers, additive manufacturing is gaining crucial popularity in the automotive, aerospace, and biomedical industries. Complicated parts that are difficult to produce with the conventional method are much easier to produce with AM as they appear in the CAD model. Customers’ demand for customized and high-precision products can be easily answered by this technology, with no time elapsed during tool change and machine setup. In the aerospace industry, lightweight and good-durability products are preferred for fuel efficiency and higher performance, which are promising features of 3D metal printing [1]. Strength-to-weight ratio enhancement of 3D-printed metals is another important ongoing area of research in the automotive industries for better fuel efficiency [2,3]. From the biomedical perspective, patient-specific implant and prosthetics manufacturing is easier to achieve with additive manufacturing compared to conventional manufacturing processes [4,5].

While 3D printing has revolutionized the manufacturing industry, it has challenges, like poor surface finish and mechanical properties, caused by porosity, distortion due to thermal effects, scanning methods, shrinkage defects while cooling, and irregularities in layer deposition. There is therefore a need for a special post-processing technique that can largely eliminate these challenges [6]. In most of the literature, post-processing is categorized into three main groups: mechanical, chemical, and thermal. Removing support structures, finishing processes to modify surface integrity, and dimensional accuracy are examples of mechanical post-process requirements. Post-processing techniques, such as annealing [7], shot peening [8], and burnishing [9], can significantly improve the strength and durability of 3D-printed parts. Additionally, post-processing can also help to reduce the effects of anisotropic behavior, improve dimensional accuracy, and eliminate voids or porosity [10]. Overall, to meet the stringent dimensional and surface finish requirements of many engineering components, it is crucial to optimize the process variables and implement appropriate post-processing techniques [11].

The challenges posed by current AM processes, as outlined in the introductory paragraph, necessitate effective post-processing techniques to overcome issues such as poor surface finish and compromised mechanical properties. The burnishing process emerges as a strategic solution, offering multiple advantages. Firstly, it provides a superior surface finish, addressing one of the primary concerns in AM [12]. The research conducted by Varga G. et al. on titanium alloy printed by the selective laser melting method using slide diamond burnishing showed improved 2D and 3D surface roughness parameters [13]. An additional advantage of the burnishing process is its ability to induce compressive residual stress on the burnished material, thereby enhancing the material’s service life by modifying its fatigue life [14,15,16]. Furthermore, it contributes to a significant increase in both surface and subsurface microhardness [16], addressing mechanical property enhancement challenges. Furthermore, the burnishing process stands out due to its ease of machine setup, requiring a short period of time for completion. Moreover, it proves to be economically advantageous compared to conventional post-processing methods, as the authors of [17] mentioned in their conclusion. Finally, its chip-less and coolant-free characteristics make it an environmentally friendly process, which is demanding to achieve for machining process solutions currently. This multifaceted approach positions burnishing as a valuable technique in mitigating the challenges associated with 3D-printed metals, offering not only enhanced material properties but also practical benefits in terms of efficiency and cost-effectiveness.

Ultrasonic vibration-assisted burnishing (UVAB) is a well-known burnishing process applied in diverse industrial applications to maximize the benefits of the burnishing process by introducing an ultrasonic vibration accessory [18]. Teramachi et al. [19] used this mechanism in their study to improve the surface and subsurface integrity of an additive-manufactured AlSi10Mg workpiece. They observed an improvement in the surface roughness, hardness, and filling ratio. Additionally, the influence of post-process conditions and their sequence on the surface integrity of 316L steel made using selective laser melting was studied by [15]. Sanding, polishing, and burnishing sequences to finish the product were experimentally applied. The authors found that these post-processes are key aspects of smart manufacturing to improve the surface properties of the component. When dealing with biomedical equipment, wear and corrosion resistance are the most important material property characteristics. Ma C. et al. [20] treated nickel–titanium (NiTi) using simultaneous ultrasonic striking and burnishing to address the potential for toxic Ni ion release due to poor surface finish and other concerning properties associated with surface integrity. Another study, by Salmi M. and his team [18], was conducted on 3D-printed cobalt–chrome and 316L stainless steel by laser sintering EOSINT M 270 and EOSINT M 280, with an objective of characterizing the effective burnishing process parameters (feed and spring compression). A feed of 0.05 mm/rev and spring compression of 1.5 mm were the optimal parameters for reduced surface roughness and an increase in hardness.

Another challenge for additive manufacturing is the availability of printable materials in powder or wire form for the continuously changing system technology and processes [6]. The different steel families, titanium, aluminum on a limited scale, nickel, magnesium, and their alloys, are the most commonly reported in the literature. Numerous burnishing process experiments were conducted on Co-Cr and 316L stainless steel [18], nickel–t [21], Inconel 718 [21], and aluminum alloy AlSi10Mg (3.2382) [9] to study parameters effect on different responses. To the authors’ knowledge, there is no burnishing process parameter study on printed MetcoAdd 17-4PH-A.

Charfeddine Y. et al. [22] simulated combined and separate grinding and burnishing processes for 100Cr6 steel. They found that combined grinding and burnishing gave better surface quality in terms of residual stress than only surface burnishing. Karthick Raaj R. et al. [16] studied the effect of grinding and low-plasticity burnishing on nickel-based superalloy 718 printed by electron beam additive manufacturing. In their results, the grinding-plus-burnishing process showed improved surface finish and higher hardness and compressive residual stress. Furthermore, 316L stainless steel printed by selective laser melting was shot-peened before being burnished, and an improved surface was achieved [23]. G. Brock et al. and Attibi S. et al. also studied burnishing’s effect on the surface roughness and other properties of 316L stainless steel, using turning as the initial machining process [24,25]. Other combined post-processes, like milling and burnishing [9], magnetic-field-assisted polishing and burnishing [15], as well as heat treatment and burnishing [26], are also common industrial practices.

Since continuous printing technologies change, and newly developed materials require thorough research to mitigate the existing challenges of the manufacturing process, this study investigates the advantages of removing excess materials and irregularities by the grinding process and smoothing the surface, as well as mechanical and metallurgical enhancements by the burnishing process. It analyzes the effect of burnishing parameters on the surface roughness, residual stress, and microhardness of 3D-printed MetcoAdd 17-4PH by using a modified burnishing tool. This study focuses on variations in burnishing force, feed, and the number of passes, while maintaining a consistent speed. By systematically altering these parameters, the experiment aimed to comprehensively examine and understand the impact of burnishing on surface characteristics.

2. Materials and Methods

2.1. Experimental Setup

The experiment begins by metal-3D-printing the workpieces, ensuring consistency in the printing process by keeping all the parameters constant for all workpieces. Subsequently, the printed workpieces undergo a grinding process using identical parameters, focusing on uniformity to maintain controlled conditions, as the main aim is to study the impact of burnishing parameters by using grinding as a pre-finishing method. Following this, burnishing is conducted on the ground workpieces, employing an L16 Taguchi design (orthogonal array) using Minitab 18 software for systematic optimization. To comprehensively study the effects of both grinding and then burnishing of the 3D-printed product, measurements of surface integrity components, including surface roughness, residual stress, and microhardness, were then taken (Figure 1). This iterative approach allows for a thorough assessment of the influence of each process on the final properties of the 3D-printed metal components.

2.2. Materials

MetcoAdd 17-4PH-A powder was selected as a material to prepare the workpieces used in the experiment. Its exceptional characteristics give significant promise in various applications. Classified as an iron-based alloy, with a composition of FeCrNiCu, MetcoAdd 17-4PH-A shares similarities in its properties with 17-4PH, EN 1.4542, and UNS S17400 [27]. Its production through gas atomization results in a spheroidal morphology, making it an ideal candidate for additive manufacturing purposes, specifically in powder-bed-type processes. This martensitic, precipitation hardening stainless steel demonstrates versatility in applications ranging from aerospace and nuclear technologies to chemical processing and oil refining equipment.

Furthermore, its utilization extends to food processing equipment and the production of surgical parts, proving the alloy’s adaptability and strength across diverse sectors. The printing process performed using ORLAS CREATOR 3D Metal Printer ©, by OR LASER, based in Dieburg, Germany, has two stages with different scanning parameters. The first scanning stage operated at 50% power with 107 W, a speed of 1200 mm/s, and a single repetition. The second scanning stage utilized 85% power with 182 W, a speed of 600 mm/s, and also a single repetition. For both stages, the beam diameter was set at 40 µm, layer thickness at 50 µm, and hatch spacing at 40 µm, with an 80% overlap. The scanning rotation angle was 120°, and the scanning style employed was continuous zig-zag. The selected parameters reflect careful consideration of the trade-offs between speed, quality, and mechanical properties in the metal 3D printing process. The two-stage approach allows for a more controlled and effective printing process that optimizes material use and enhances the final part’s performance. A 60 mm × 26 mm × 10 mm size workpiece, as in Figure 2, was printed to accommodate three burnishing areas measuring 17 mm × 20 mm each, with enough of a margin between them, and from the two sides. Six workpieces were prepared with specified dimensions for the L16 Taguchi design of the experiment.

2.3. Grinding and Burnishing Processes

Perfect Jet MCV-M8 CNC milling machine (Ping Jeng Machinery Industry Co., Ltd., Taichung City, Taiwan) was utilized for the burnishing process. The ground surface underwent burnishing following the parameters presented in

Table 1. Uncoded L16 orthogonal array response results and their S/N ratios.

				Surface Roughness		Transverse Stresses		Longitudinal Stresses		Microhardness
Run	F(N)	f (m/min)	Pass	Ra (µm)	S/N ratio	σ-Trans (Mpa)	S/N Ratio	σ-Long (Mpa)	S/N Ratio	HV 0.2	S/N Ratio
1	40	0.02	1	0.32	9.90	−784	57.88	−581	55.29	444.3	52.95
2	40	0.07	2	0.28	11.06	−842	58.51	−581	55.28	433.3	52.74
3	40	0.12	3	0.30	10.46	−890	58.99	−567	55.08	370.7	51.37
4	40	0.17	4	0.26	11.70	−750	57.51	−578	55.23	423.0	52.56
5	60	0.02	2	0.19	14.42	−847	58.56	−551	54.83	438.0	52.83
6	60	0.07	1	0.35	9.12	−994	59.95	−609	55.69	451.3	53.09
7	60	0.12	4	0.27	11.37	−680	56.65	−527	54.44	442.5	52.76
8	60	0.17	3	0.28	11.06	−649	56.25	−560	54.97	457.3	53.19
9	80	0.02	3	0.15	16.48	−802	58.08	−436	52.79	440.0	52.86
10	80	0.07	4	0.37	8.64	−881	58.90	−500	53.97	425.0	52.56
11	80	0.12	1	0.55	5.19	−933	59.40	−494	53.88	411.0	52.27
12	80	0.17	2	0.53	5.51	−805	58.12	−532	54.52	420.0	52.61
13	100	0.02	4	0.32	9.90	−984	59.86	−502	54.02	442.0	52.90
14	100	0.07	3	0.39	8.18	−906	59.15	−561	54.98	424.3	52.54
15	100	0.12	2	0.42	7.54	−896	59.05	−566	55.06	420.3	52.47
16	100	0.17	1	0.56	5.04	−918	59.26	−498	53.94	445.0	52.97

. Grinding process was conducted by a Metallkraft FSM 4080 series surface grinding machine (Stürmer Maschinen, Hallstadt, Germany) with a 6A60M8V 38 Granit wheel running at a speed of 1450 RPM. From our preliminary test, applying burnishing process to the as-built surface gives poor surface finish, which suggests the need for an additional process. To address this issue, grinding was used before burnishing process. To investigate the surface response based on the other parameters, the burnishing speed remained constant at 3000 mm/min, achieved by the milling table as the spindle was stationary to hold the tool.

The burnishing tool used in this experiment, depicted in Figure 3, was applied in our previous works [28], and the results were good enough to reuse it in this study. A slide diamond burnishing tool with a 3 mm radius burnishing head, initially designed to be used with lathe, adapted with a tool holder feature for compatibility with CNC milling machine, was used. This modification helps the technician to use the same tool interchangeably between lathe and milling machine. The machining plan involved a stationary spindle head holding the burnishing tool while moving the milling table, following G and M code instructions to guide the workpiece in an elliptical path at a constant speed. Burnishing parameters were adjusted based on a Taguchi design, with particular attention to ensuring the lowest possible coefficient of friction through lubrication with SAE 15W-40-grade traditional oil. Controlling burnishing force was crucial, and it was monitored using a Kistler 9257A force sensor, a Kistler 5011A signal processing charge amplifier (both from Kistler Instrumente AG, Winterthur, Switzerland), and an NI Compact DAQ 9171 four-channel signal acquisition unit (NI Hungary Kft. Debrecen, Hungary), all integrated with LabView for real-time data display and control.

2.4. Surface Roughness, Residual Stress, and Microhardness Measurements

To follow the properties’ changes due to the grinding and burnishing process, surface roughness, residual stress, and microhardness measurements were taken after printing, after grinding, and after burnishing process. To avoid errors, surfaces were measured at least three times, and their average was taken for analysis.

The as-built, ground, and burnished surfaces underwent three-dimensional topography analysis using an AltiSurf 520 surface measuring instrument (Altimet SAS, Thonon-les-Bains, France) by applying a CL2 confocal chromatic sensor and MG140 magnifier. Altimap 6.2 software from Digital Surf was used for surface roughness evaluation based on the Ra parameter. ISO 21921-2:2021 [29] and ISO 21920-3:2021 [30] standards were used with a 4 mm sampling length and 0.8 mm cut-off. Arithmetic average roughness Ra was selected to study the peak and valley height changes during grinding and burnishing to follow surface roughness enhancement [31].

The analysis of residual stress, both surface and in-depth, using the Xstress G3 diffractometer (Stresstech GmbH, Rennerod, Germany), was crucial for comprehending the stress distribution within the workpiece after all three processes (printing, grinding, and burnishing). The measurement of surface residual stresses and their distribution was performed using the X-ray diffraction technique, focusing on both transverse and longitudinal directions. This was facilitated by a Stresstech G3R-type centerless diffractometer (Stresstech GmbH, Rennerod, Germany), which was equipped with a chromium X-ray source and a 2 mm diameter collimator. Stress values were calculated employing the ${\sin }^2\psi$ method. The experiment was designed with five tilts in the positive and negative directions, incorporating oscillations of 2 degrees and reaching maximum tilting angles of ±45°. For every tilt position, an exposure time of 4 s was applied. For the purpose of measuring stress distribution at depth, the QETCH 100 M electrolytic etcher from QATM (ATM Qness GmbH, Mammelzen, Germany) was utilized for the removal of steel layers, with the thickness of these etched layers being precisely measured by the Mitutoyo ABSOLUTE depth gauge (Mitutoyo Europe GmbH, Neuss, Germany).

Concerning the microhardness analysis, a Tukon 2100B device, by Wilson Instruments, Norwood, MA (USA) was used to assess the variations in hardness across the burnished surfaces. Surfaces were prepared for the indentation process by cleaning them to avoid any possible errors. An indenter with a 0.2 kg load was applied for 10 s. Three measurements were achieved in three different places on each surface, and an average value was then taken for the purpose of the analysis.

2.5. Experimental Design

2.5.1. Taguchi Orthogonal Design of Experiment

The effects of burnishing force, feed, and number of passes on surface roughness, residual stress, and microhardness in the burnishing process were investigated using Taguchi experimental design with an L16 orthogonal array. Each burnishing parameter was studied at four levels while keeping the burnishing speed constant. Taguchi’s design advantage lies in its ability to efficiently explore multiple factors and their interactions with a minimal number of experimental runs. By utilizing the orthogonal array, the study systematically varied the burnishing parameters across a limited number of experiments, allowing for a comprehensive analysis of their effects on the responses.

The results from the Taguchi experimental design provide valuable insights into the optimal combination of burnishing parameters for improving surface roughness, residual stress, and microhardness. By analyzing signal-to-noise ratios (S/N), this study can identify the most influential factors and their optimal levels. Intending to minimize surface roughness and maximize residual stress and microhardness, the “smaller-the-better“ for surface roughness and “larger-the-better“ for residual stress and microhardness principles are employed, and they are given in Equations (1) and (2).

$$\mathrm{Smaller-the-better\; (STB):} \mathrm{S}/\mathrm{N}=-10\times log10 \left({\displaystyle \frac{1}{n}}{\sum }_{i=1}^n\left({\displaystyle \frac{yi}{\overline{y}}}\right)\right)$$

(1)

where $yi$ represents individual observations, $\overline{y}$ is the average, and $n$ is the number of observations.

$$\mathrm{Larger-the-better\; (LTB):} \mathrm{S}/\mathrm{N}=10\times log10 \left({\displaystyle \frac{1}{n}}{\sum }_{i=1}^n {\left({\displaystyle \frac{yi}{\overline{y}}}\right)}^2\right)$$

(2)

where $yi$ represents individual observations, $\overline{y}$ is the average, and $n$ is the number of observations.

2.5.2. Multi-Objective Optimization Method

Taguchi’s optimization technique is suitable for single-response optimization application. Using another optimization method, which deals with more responses, is compulsory. The Desirability Function (DF) a multi-optimization technique proposed by Derringer and Suich [32], with application potential for different designs of experimental methods [33,34], and its simplicity has made it popular. Surface roughness minimization (Equation (3)), residual stress, and microhardness maximization (Equation (4)) optimization targets were applied in this paper.

$$di(max)=\left\{\begin{array}{c}0 if Yi<Y″\\ {\left({\displaystyle \frac{Yi-Y\prime }{Y\prime -Y″}}\right)}^{\mathrm{r}} if Y″\le Yi\le Y\prime \\ 1 if Yi>Y\prime \end{array}\right.$$

(3)

where $di\left(max\right)$ is the desirability function of each individual response for the maximization optimization target, $Y\prime$ is the maximum response, and $Y″$ is the minimum response.

$$di(min)=\left\{\begin{array}{c}0 if Yi<Y″\\ {\left({\displaystyle \frac{Yi-Y″}{Y″-Y\prime }}\right)}^{\mathrm{r}} if Y″\le Yi\le Y\prime \\ 1 if Yi>Y\prime \end{array}\right.$$

(4)

where $di\left(min\right)$ is the desirability function of each individual response for the minimization optimization target, $Y\prime$ is the maximum response, and $Y″$ is the minimum response. r defines the distribution of the desirability function between the minimum and maximum range. For both Equations (4) and (5), we considered a linear relation equal to 1.

$$D\left(Tot\right)={\prod }_{i=1}^n{di}^{wi}$$

(5)

where $D\left(Tot\right)$ is the overall desirability function, and wi (0 < wi < 1) is the weight of each response. Principal component analysis method was used to calculate the relative weight used in this multi-objective optimization.

The results from the uncoded L16 orthogonal array Taguchi experimental design applied in this experiment, including the measured surface responses and level of variables, are given in Table 1 and

Table 2. Input parameter levels.

Levels	Force (N)	Feed (mm)	No. Passes
1	40	0.02	1
2	60	0.07	2
3	80	0.12	3
4	100	0.17	4

. All responses analyzed in this experiment were repeated at least three times, and their average was taken to avoid measurement error.

Percentage changes realized due to the introduced surface enhancement machining methods are calculated using Equation (6) to help in understanding the increased or decreased degree of surfaces’ responses. The terms initial and final in the equation represent the starting measured value and the value after a specified machining process is applied. For an overall observation, changes achieved by burnishing process for the 16 surfaces are averaged. The calculated and measured results are presented in

Table 3. Measured surface responses.

Responses	3D-Printed	Ground	Burnished (Average)
Ra (µm)	11.6	1.17	0.346
σ-Trans (Mpa)	129.9	−189.9	−847.7
σ-Long (Mpa)	43.5	98.5	−540.3
HV 0.2	241	350.6	430.5

and

Table 4. Calculated % changes.

Percentage of Increase	3D-Printed to Ground	Ground to Burnished	3D-Printed to Burnished
Ra (µm)	−89.9	−70.4	−97.0
σ-Trans (Mpa)	−246.2	346.4	−752.6
σ-Long (Mpa)	126.4	−648.5	−1342.1
HV 0.2	45.5	22.8	78.6

.

$$\% change=\left({\displaystyle \frac{final-initial}{final}}\right)\ast 100$$

(6)

3. Results and Discussion

Surface roughness, residual stress, and microhardness are the three responses analyzed to study the effects of grinding and burnishing processes.

3.1. Surface Roughness

Analyzing the effect of the burnishing parameters on each burnished surface and comparing it to that of the surface before and after the grinding process gives a clear picture of the favorable combinations. During the grinding process, the higher peaks created during printing process are removed by the hard grains of the grinder to give a smoother surface. In the burnishing process, a hard smooth spherical tool is pressed against the surface to plastically deform the remaining peaks to the valley for the same purpose. According to Figure 4, surface 9 has the smallest Ra value, followed by surfaces 5, 4, and 7, respectively, with Ra values of 0.15 µm, 0.19 µm, 0.26 µm, and 0.27 µm. If we divide all the experimental runs into two halves, one with eight surfaces, as a low-roughness profile, and another with eight surfaces, with a high-roughness profile, we can obtain a clear comparative image of the parameters’ effects. For the low-profile-value half, the burnishing force is represented by 40 N four times, 60 N three times, 80 N one time, and 100 N zero times. These numbers help us to conclude that low burnishing forces of 40 N and 60 N give a low-roughness profile. From the feed point of view, the values are as follows: 0.02 m/min three times, 0.07 m/min one time, 0.12 m/min two times, and 0.17 m/min two times. From these numbers, there is no clear trend in the number of passes affecting the low-surface-roughness profile. The third parameter, the number of passes, is represented one time by one, two times by two, three times by three, and two times by four. Two, three, and four passes each gave lower-profile surfaces than only one pass.

For the second higher roughness values profile part, 100 N of burnishing force is predominantly reflected on four surfaces (13, 14, 15, 16), which are all the surfaces for 100 N force in the experimental design. The contribution of 80 N force to high roughness was also significant, as three of the possible four are categorized in the high region. This leads us to the conclusion that applying a high burnishing force can create excessive plastic deformation and, as a result, a rough surface. In the case of the feed, as in the low-roughness surfaces, it showed no clear trend. Four surfaces were burnished by 0.07 and 0.02 m/min of feed, and the other four surfaces by 0.12 and 0.17 m/min. However high feed can lead to a high-roughness profile. A collective generalization on the last four surfaces suggests that two of the last four feeds are 0.12 m/min and two are 0.17 m/min. Concerning the number of passes, the results show that a smaller number of passes gives high roughness. This is evident, as surfaces 6, 11, and 16 were burnished using one pass, surfaces 12 and 15 were burnished two times, and surfaces 14 and 13 were burnished three and four times each.

The reason for the roughness profile height increase when the burnishing force increases and the number of passes decreases is that the effect of applying a high burnishing force can excessively deform the material plastically, which leads to material overflow to the two sides of the burnishing tool. This also happened with low numbers of passes compared to the low-roughness profile. From numerous studies, it is known that high burnishing force and an increased number of passes are causes of rough surfaces [35,36]. Figure 5 and Figure 6 show the changes after each machining process. From these figures, we can observe the change in the trend of the surface roughness.

Figure 6 and Figure 7 show the physical change of the surface with tool trace because of the plastic deformation in the process. After the material removal process using the grinding process and the surface plastic and elastic deformation using the burnishing process, the surface roughness of the workpieces was enhanced significantly, as shown in Table 4 and Figure 4.

3.2. Residual Stress

The burnishing process is essential for enhancing the performance and longevity of mechanical components by inducing compressive residual stress at the surface. This shift from tensile to compressive stress significantly improves fatigue resistance, as compressive residual stresses counteract tensile stresses during operational loads, thereby inhibiting crack initiation and propagation. This improvement in surface integrity and smoothness contributes to the overall strength and durability of materials, making burnishing critical for producing high-performance engineering components.

The type of residual stress—tensile or compressive—induced in a material depends on the machining process and the applied force. For example, pressure applied by burnishing tools causes the plastic deformation of surface layers, compressing the underlying layers and creating compressive residual stress as the deformed surface tries to revert to its original shape. In stainless steel, the initial tensile residual stress in the transverse and longitudinal directions (as observed in Figure 8 and Figure 9) becomes compressive after grinding and is maintained through burnishing. Residual stresses were measured longitudinally, along the machining direction, and transversely, at 90 degrees to the machining direction. For as-built materials, the longitudinal direction corresponds to the printing direction, and the transverse direction is perpendicular to it; for ground and burnished materials, the measurement directions align similarly.

3.2.1. Transverse Residual Stress

The as-built stainless steel workpieces had a tensile residual stress of 130 MPa, which then changed to a compressive residual stress of −190 MPa, as shown in Figure 8. The same figure shows the increase in compressive residual stress of the burnished surfaces. When compared in percentage form, the average burnishing process showed a 346% increase over the ground surface. This indicates a huge amount of enhancement by the burnishing process.

Like the other surface integrity properties, the burnishing parameters have different effects on residual stress. Surfaces 8, 7, and 4, with minimum increases, share low burnishing forces of 60 N, 60 N, and 40 N, high numbers of passes, of three, four, and four, and high feeds, of 0.17 m/min, 0.12 m/min, and 0.17 m/min, respectively. This indicates that the applied forces are not strong enough to induce maximum compressive residual stress. Even though the number of passes is not small, their effect is reduced as the applied force is low. A high feed also has a negative effect, as it indicates higher spacing between two parallel tool paths. In the maximum increase category, surfaces 6, 13, 11, and 16 have similarities in terms of the applied force and number of passes, as shown in Table 1. Higher force and a smaller number of passes, for except surface 13, with four passes, imply favorable parameters for maximum increase. This illustrates that a high burnishing force is dominant in inducing maximum compressive residual stress and causes the material to be compressed.

3.2.2. Longitudinal Residual Stress

The longitudinal compressive residual stress, like the transverse, showed tendencies of increase and decrease, depending on the burnishing parameters depicted in Figure 9. Since all the surfaces showed an increasing trend, the 16 surfaces were categorized into two halves, named minimum and maximum increase, based on the degree of increase. A clear relationship between the longitudinal residual stress and the burnishing parameters is shown only with the burnishing force when the surfaces experience minimum increase. Surface 9, burnished with 80 N of force, a 0.02 m/min feed, and three passes, possesses the lowest value, of −436 MPa. To have a better understanding, we can check the additional force parameters of surfaces 11, 16, and 10, which have higher values than surface 9, but still have a low value. Unlike the transverse residual stress, the longitudinal residual stress experienced less of an increase when higher burnishing force values were used. From these observations, we can say that the longitudinal residual stress has an inverse relation to the burnishing force, but an unclear or weak relation with the feed and the number of passes with low increase. Surfaces 6, 1, and 2 experienced the highest increases, with values −609 MPa, −581 MPa, and −5801 MPa. For all these surfaces, the common parameter behavior of minimum forces of 60 N, 40 N, and 40 N, feeds 0.07 m/min, 0.02 m/min, and 0.07 m/min, and numbers of passes of one, one, and two, respectively, were observed. These indicate that lower forces, feeds, and numbers of passes give better longitudinal residual stress. Generally, the burnishing force is the dominant factor, showing a good relationship to the longitudinal residual stress. This is due to the ability of force to affect the plastic deformation phenomenon.

3.2.3. In-Depth Residual Stress

The surface and subsurface study of a material gives an immense perspective on the influence of different machining processes on properties like fatigue, strength, and wear resistance. Depth residual stress measurements of some selected surfaces were taken into account to study the grinding and burnishing processes’ effects on the subsurface. The measurements taken before and after grinding, as well as after the burnishing of surfaces 3, 8, 9, and 14 in both directions (longitudinal and transverse), are shown in Figure 10 and Figure 11. The four burnished surfaces were selected based on the same number of passes (three) and representative forces 40 N, 60 N, 80 N, and 100 N, with varying feeds, as they were directly selected from the orthogonal experimental array. The study of the effect of the burnishing force was the aim when investigating the depth residual stress, as its dominance is reported in many studies of regular materials processes.

Due to the reason discussed in the surface residual stress section, different machining methods induce different stress types. At the surface, the as-built workpiece in each direction has tensile stress. However, when the depth increases, first it increases in the positive stress direction abruptly, in the case of the transverse direction, and gradually in the longitudinal direction, and then it decreases gradually until it attains a stable condition. This indicates the stress variation at different depths of the as-built material due to the printing process [37]. The grinding process did not have much of an effect on the initial state of the stress, except in the transverse direction up to a depth of 4 µm, which changed the stress polarity to a magnitude of −232 MPa. In the transverse direction, the as-built and ground surfaces show the same trend of sharp increase and then gradual stabilization. In both directions, the burnished surfaces show a similar effect of burnishing force on surfaces 3, with 8, and 9, with 14. The changing effect with the increase in depth could have been due to the varying feed rate, otherwise, a clear increasing effect of burnishing with an increase in force magnitude is demonstrated. Another important observation is the distinct group-like similarity shown in both longitudinal and transverse directions. If we compare the feed rates of the 40 N and 60 N forces (0.12 m/min and 0.17 m/min) with those of the 80 N and 100 N forces (0.02 m/min and 0.07 m/min feeds), the former are in the higher group if we categorize the feed into two groups, of higher and lower. In both directions, the printing, grinding, and burnishing processes induced stress apart from their magnitude and direction. The burnishing process effect is clearly shown in both pictures, with higher depth and stress magnitude and burnishing force increases inducing higher compressive residual stress. Generally, the burnishing process shows significant compressive residual stress enhancement at the subsurface level of the as-built material. Even though it is in small amounts, the grinding process also shows a pronounced increase.

3.3. Microhardness

Understanding the microhardness provides insights into the material’s resistance to plastic deformation and wear, which is essential for evaluating the durability and performance of a workpiece in practical applications. The HV 0.2 microhardness results measured for the as-built, ground, and burnished surfaces with different burnishing parameters are depicted in Figure 12. From the measured microhardness results, different material properties, like mechanical properties, the performance of the material, and its behavior at different conditions, can be inferred. As shown on the bar graph, the microhardness values were found to have increased after the grinding process and burnishing process. The as-built material’s initial hardness was 241 HV, and it was increased to 35 HV after grinding, and to 371 HV and 457 HV if we take surfaces 3 and 8 as representative minimum and maximum values of the burnished surfaces. As presented in Table 4, a 45.5% increase in grinding effect, an average increase of 22.8% after burnishing compared to grinding, and a general 78.6% increase with reference to the printed surface hardness were measured. This shows that both post-processes of the 3D-printed stainless steel have a positive effect on increasing the microhardness of the material. Heat generation due to the abrasive nature and the induction of different stress types are the main characteristics of the grinding process, which can lead to work hardening of the material. This work hardening process, either by grinding or burnishing, is the main reason for dislocation movement, which affects the hardness of the material. Teimouri et al. [38] discussed this issue with uniform microstructure refinement due to severe plastic deformation, and because of this microstructure refinement, the hardness increases.

Diving into the burnishing results of each surface to study the effects of the parameters changes, we found a weakly associated general trend in the increasing or decreasing direction of the parameters in terms of hardness change, despite the increased hardness results, as discussed previously. Surface number 3, with 371 HV, and number 11, with 411 HV, are surfaces with a minimum increase. If we check their burnishing parameters, surface 3 is burnished with 40 N of force, a 0.12 m/min feed, and three passes, while surface 11 is burnished with 80 N of force, a 0.12 m/min feed, and one pass. For both surfaces, the second highest feed is the only common parameter, which is also shared by surfaces 12, 15, and 4, with feeds of 0.17 m/min, 0.12 m/min, and 0.17 m/min, respectively. In this case, we can say that high feed has a minimum hardness increase. On the other end, surfaces 8 and 6 have the maximum hardness increase, with values of 457 HV and 451 HV, respectively. Their parameters are 60 N of force, a 0.17 m/min feed, and three passes for surface 8, 60 N of force, a 0.07 m/min feed, and one pass for surface 6. Here, 60 N of force was used in both experiments, but the other parameters were different. If we check surfaces 16, 1, and 7, which have an increased hardness on the maximum side, to obtain general information, low forces of 40 and 60 N were used, except in surface 16, and few passes were shown, except in surface 7. From these, we can say that low force, few passes, and slow feed are the characteristics of the maximum increase in microhardness.

3.4. Statistical Analysis

The designed experiment helps to conduct valid and reliable statistical analyses to infer different conclusive information from the experimental results. The burnishing force, feed, and number of passes are the independent variables, and the mean roughness Ra, residual stress, and hardness are the dependent variables. The signal-to-noise ratio is a commonly used tool to study the influence of input variation on response parameters. Independent variables influence categorization and the distinguishing of the optimum parameter levels is carried out by the S/N ratio by applying smaller-the-better for roughness and larger-the-better for the other three response measures.

Minitab software was used to calculate the S/N ratio and draw the plot of their main effects. The delta value, which is the difference between the maximum and minimum mean of the factor level of the parameter, indicates their dominance in the response. The feed, pass, and force influence are first, second, and third. Another important graph is Figure 13, which represents the main effects plot for the S/N ratio, which shows the best levels for each response factor. The second level of the burnishing force (60 N), the first level of the feed (0.02 m/min), and the third level of the number of passes (three) give the optimum Ra value. Surface numbers 5 and 9 are the surfaces burnished near these parameters, which have minimum roughness compared to the other surfaces, and this agrees with the result.

3.4.1. Taguchi Analysis for σ Trans

In the case of residual stresses (transverse and longitudinal), the calculated delta value shows that the burnishing force is the most dominant factor, followed by the feed and number of passes, consecutively. These indicate that the burnishing force plays an important role in creating residual stresses in both directions. This agrees with the negative effect of increasing force on surface roughness deterioration by the excessive movement of materials and residual stress enhancement. Despite the fact that force has a dominant effect in both the residual stress types, Figure 14 and Figure 15 show that the effects of force and the number of pass levels are not equivalent. For example, the optimal level of forces in the transversal residual stress is 100 N, while in longitudinal residual stress, it is the minimum level of 40 N. Further explanation is also possible based on the direction of the line and its slope, which indicates an increase with a higher rate in the transverse and a decrease with a slower rate in the longitudinal direction. The main effects plots depict that 100 N of force, a 0.07 m/min feed, and one pass give optimal transverse residual stress. For longitudinal residual stress, 40 N of force, a 0.07 m/min feed, and two passes give an optimum result.

3.4.2. Taguchi Analysis of Microhardness

Based on the larger is better signal-to-noise ratio analysis quality characteristics presented in Figure 16, the feed makes the most significant contribution to the microhardness response, followed by the force and pass. The exact figures indicating the second level of force (60 N), the first level of feed (0.02 m/min), and the first level of the number of passes (one) are the optimum levels for the burnishing parameters. In addition to these, the figure shows a continuously changing direction and slopes, which can be interpreted as unstable effects of the burnishing parameters.

3.4.3. Multi-Objective Optimization

The optimized results of each response using the Taguchi method gave different optimal burnishing settings, which are not representative of the overall optimal burnishing setting. While calculating the overall desirability function using Equation (5), it is essential to calculate the weights of the responses based on their contribution instead of treating them equally. The principal component analysis method is very convenient for solving this kind of problem.

Table 5. Principal component analysis.

Eigenanalysis of the Correlation Matrix
Eigenvalue	1.5716	1.0287	0.8087	0.591
Proportion	0.393	0.257	0.202	0.148
Cumulative	0.393	0.65	0.852	1
Eigenvectors
Variable	PC1	PC2	PC3	PC4
Ra	0.616	0.125	−0.371	0.683
σ Trans	−0.632	−0.005	0.279	0.723
σ Long	0.107	0.912	0.392	−0.051
HV 0.2	0.457	−0.39	0.794	0.091

, calculated using Minitab software, shows that principal component 1 (PC1) has a higher proportion (39.3%). The corresponding weight of each response is squared before being applied, as their sum must be one to generate the overall desirability values from the normalized

Table 6. Normalized responses and overall desirability, with their rank.

Surface	Ra	σ Trans	σ Long	HV 0.2	D(Tot)	Rank
1	0.5854	0.3889	0.8397	0.8500	0.5399	8
2	0.6829	0.5579	0.8370	0.7231	0.6392	4
3	0.6341	0.6976	0.7600	0.0000	0.0000	15
4	0.7317	0.2928	0.8183	0.6039	0.4883	10
5	0.9024	0.5726	0.6672	0.7769	0.7269	3
6	0.5122	1.0000	1.0000	0.9308	0.7642	2
7	0.7073	0.0887	0.5291	0.8289	0.3181	11
8	0.6829	0.0000	0.7194	1.0000	0.0035	14
9	1.0000	0.4414	0.0000	0.8000	0.6008	5
10	0.4634	0.6721	0.3691	0.6269	0.5715	7
11	0.0244	0.8228	0.3362	0.4654	0.1903	13
12	0.0732	0.4521	0.5560	0.5692	0.2384	12
13	0.5854	0.9703	0.3846	0.8231	0.7658	1
14	0.4146	0.7448	0.7222	0.6189	0.5737	6
15	0.3415	0.7147	0.7523	0.5727	0.5160	9
16	0.0000	0.7785	0.3581	0.8577	0.0000	15

. Each experimental run’s total desirability function was ranked to know the best burnishing setting, with higher desirability. Experiment number 13, with a burnishing setting of 100 N of burnishing force, a 0.02 mm feed and four passes, is the best for reduced roughness and higher residual stress and microhardness.

For the orthogonal experimental design, the effects of parameters can be distinguished based on the values of overall desirability [39]. The mean and delta values of the desirability were calculated for the burnishing parameters and are presented in

Table 7. Mean response table DFA.

Level	Force	Feed	Pass
1	0.42	0.66 *	0.37
2	0.45	0.64	0.53
3	0.4	0.26	0.29
4	0.46 *	0.18	0.54 *
Delta	0.06	0.48	0.24
Rank	3	1	2

. The fourth burnishing force level, first feed level, and fourth number of passes level are the optimal values, with an asterisk sign in the table, and this agrees with experiment 13 as the optimal setting.

4. Summary and Conclusions

In the present research work, successive grinding and slide diamond burnishing experiments were carried out on 3D-printed MetcoAdd 17-4PH-A stainless steel samples with the aim of studying the burnishing parameters’ effect on the surface roughness, residual stress, and microhardness. Workpieces were printed and ground with the same setup, and measurements of the aforementioned responses were taken before and after grinding. The burnishing force, feed, and number of passes were the changed burnishing factors. The combined grinding and burnishing process showed significant improvement in terms of surface integrity. Our findings are summarized as follows:

• An increased residual stress in transverse directions with an average increase of 346% was observed with a high burnishing force and a lower number of passes. The 130 MPa initial as-built surface was modified by grinding to −190 MPa and significantly increased in all the burnished surfaces.
• A low burnishing force with a higher number of passes showed a surface roughness decrease. Forces of 40 N, 60 N, and 80 N with repetitive burnishing resulted in a smoother surface. The effect of the feed in this case was not clear. All the burnished surfaces demonstrated better surface roughness when compared with the as-built and ground surfaces.
• Unlike the transverse residual stress, the longitudinal surfaces experienced an increasing trend when lower forces (40 N and 60 N) and fewer passes (one and two) were used. Another important observation was that the ground surface’s stress was unchanged with direction. Burnishing processes, like in the transverse direction, significantly increased the residual stress.
• Both grinding and burnishing processes enhanced the microhardness of the printed surfaces. A 45.5% increase in grinding, an average increase of 22.3% after burnishing compared to grinding, and 78.6% compared to the initial state, were accomplished.
• The in-depth residual stress study showed almost no change due to the grinding process, but after burnishing, the compressive stress distribution was up to 0.11 mm by 40 N and 60 N, and up to 0.22 mm by 80 N and 100 N. The feed variation in the burnishing process also showed a clear distinction when higher and lower force ranges were used.
• This study demonstrates that multi-objective optimization using desirability function analysis effectively identified the optimal parameters for all the burnishing process responses. By selecting a burnishing force of 100 N, a tool feed rate of 0.02 m/min, and four passes, the process achieved desirable outcomes for multiple performance objectives.