The objective of this paper is to present a comprehensive study of the effects of ball burnishing on the low cycle fatigue endurance of cylindrical specimens subjected to alternative bending stress, by constructing fatigue S-N curves to evaluate the impact of the process on the general fatigue behavior of AISI 1038. The specimens were burnished through five different force-number of passes couples, displacing the as-machined-material S-N curve towards higher lifespans as the degree of cold work is increased, either by effect of a higher force and more burnishing passes. The derived surface hardening effect also proved to be correlated analytically with the fatigue lifespan increase at all tested maximum-stress levels, as prove the correlation of the parameters that represent the potential equation associated to the Wöhler model with the hardening effect. It is concluded that the process performed with 120 N and 7 passes results in the most favorable results, represented by a 41% increase of hardness and a 77% increase in the lifespan in the best of cases.

Ball burnishing is a surface finishing process performed by a sphere that rolls over the target surface while applying a controlled force. By effect of the exerted pressure, the peaks that define the surface texture are deformed, thus reducing the scale of its topological features. The process is known for achieving a triple enhancing effect on the surface integrity of different materials: a decrease in surface roughness, an increase of the surface and subsurface layers hardness, and a reinforcement of the compressive residual stress to a depth of nearly 1 mm [1].

Due to the fact that ball burnishing includes such a high amount of processing parameters (force, feed, lateral pass width, number of passes, ball diameter, rotation speed, original surface roughness and residual stress, lubrication, ball material, etc.), its study is often guided by applying experimental design techniques. In this context, different materials have been extensively studied during the last years. For instance, El-Tayeb et al. [2] proved the positive results of the process on an Aluminum 6061 alloy, finding that 220 N was a good burnishing force to reduce roughness and increase hardness of cylindrical specimens by 30%. Gomez-Gras et al. [3] found positive results on AISI 1038 workpieces after ball burnishing, and defined the optimal parameters to obtain a proper surface finishing.

Generally speaking, the aforementioned studies are based on evaluating the results of ball burnishing in terms of one or more of the descriptive surface integrity parameters, i.e. roughness, hardness or residual stress. However, an alternative perspective to consider its results, is relating them to the enhancement of functional responses of the target workpiece. Such is the case of the mechanical behavior of parts under fatigue regimes. Indeed, it has been proved that, since fatigue cracks are triggered by surface defects, and their nucleation and propagation depend on the surface roughness and residual stress state of the surface and subsurface layers of the material, the application of surface finishing processes is necessary enhance the fatigue life [4].

The bibliography shows numerous references dealing with the experimental assessment of the effect of laser shock peening, shot peening or ball burnishing, as finishing operations, on the increase of fatigue life [5]. Ball burnishing has proved to have superior effects over the results of other finishing processes. Wagner [6] focused his research on high cycle fatigue, and concluded that the residual stress induced by ball burnishing were compressive to a depth of 500 μm, unlike those obtained by shot peening, which were present only to 200 μm. The fatigue life experienced by burnished specimens also proved to be longer, which was related to the fact that higher residual stresses prevent to a higher extent crack growth.

On the other hand, ball burnishing has proved to introduce more homogenous plastic deformation mechanisms and more stable residual stresses than the shot peening or laser shock peening, according to McClung [7]. Drechsler et al. [8] compared the fatigue life of Ti-10V-2Fe-3Al specimens treated by shot peening and the burnishing process, finding that burnished specimens were able to support higher alternative loads during longer periods. Furthermore, a direct relation between the burnishing force and cycles to failure was found, being 1200 N the boundary value from which no enhancement was detected. The superior performance of burnishing on shot peening was observed in other materials such as Inconel®718 [9] or Ti-6Al-4V [10]. Although most references compare shot peening with ball burnishing, others evidence the superior results of burnishing on laser shock peening as well, as proved Golden et al. [11].

Ball burnishing has also shown a higher limiting effect on crack growth of all materials tested in bibliographical references, compared to shot peened ones in notched specimens, as a much greater depth is affected by compressive residual stresses. This effect is further explained at Prevey et al. [12], where a sequential burnishing study is performed. Inconel®718 specimens subjected to a four bending point fatigue test showed crack initiation after 4000 cycles. Then, they were burnished, causing the crack growth to be arrested until 30,000 cycles. The load had to be increased up to 30% so that the specimen could be lead to fatigue crack failure. Avilés et al. [13] observed similar results on AISI 1045 rotating bending stress specimens, with a change in crack propagation mechanisms according to burnishing parameters. Fatigue limit was increased by 21%.

The positive effects of ball burnishing have also been presented as a technique to counterbalance anticipated fatigue failure in corrosive or high temperature environments. Prevey et al. [14] proved that burnishing 7075-T6 aluminum specimens increased their expected life in saline ambient by a tenfold, in comparison to machined specimens. Nalla et al. [15] tested fatigue life of Ti-6Al-4V specimens in a controlled atmosphere at 450 °C, finding satisfactory lifespan increase. The authors explain this fact as a consequence of grain refinement caused by ball burnishing. Nikitin & Altenberger [16] arrived to similar conclusions on AISI 304 stainless steel parts at 600 °C.

The works described in this paper have been executed with the aim of showing extensive experimental data to find a correlation between hardness after ball burnishing, as surface integrity descriptor, and fatigue life of AISI 1038 specimens. Due to the fact that both hardness and residual stress represent the metallographic state of the material, the first one is much more accessible to be measured and it is easier to be industrially implemented. For that reason, we discarded the second one in this analysis. Specifically, the effect of different levels of force and number of passes are studied, since they define the degree of plastic strain caused on the material’s surface. For each parameter combination, a Wöhler curve is constructed through extensive bending fatigue experimental tests. This is an important aspect of this paper as fatigue behavior linked to ball burnishing are treated in the bibliography focusing on reduced set of data. Furthermore, few Wöhler curves can be found. For instance, Sadeler et al. [17] constructed reduced S-N curves to analyze the effect of three different fluid pressure levels with a hydrostatic tool, but no consideration about the number of passes was done. Other authors, such as Avilés et al. [18] or Blason et al. [19], constructed S-N curves, but only included one ball burnishing condition, so that the difference between parameter levels was not assessed. Seemikeri et al. [20] studied the effect of different burnishing parameters through DOE on surface integrity and fatigue, but the authors did not correlate directly burnishing results with fatigue performance.

At sight of the previous references, this work focuses on providing a direct link between a functional response (fatigue) and surface integrity descriptors obtained after numerous parameter combinations for ball burnishing, to provide a phenomenological approach to fatigue enhancement. Indeed, as five different ball burnishing conditions are considered, and specimens are tested six times at six different maximum stress levels, the information derived from these tests allows the identification of a correlation between the effects of plastic strain arising from ball burnishing, and the fatigue behavior of the steel alloy under different stress levels. Results are also processed statistically to assess the influence of the burnishing force and the number of passes on the fatigue life, considering the highest and lowest maximum stress level to which specimens were subjected during the fatigues tests. These results shall support the observations made through the Wöhler curves, and establish the best parameter combination to maximize fatigue life of this kind of specimens.

2Materials and methods2.1Ball burnishing toolThe ball burnishing tool used in this experimental campaign was designed and characterized by Gomez-Gras et al. [21]. This kind of tool is based on burnishing force regulation through a calibrated spring lodged inside a cylindrical part (Fig. 1A). Consequently, the burnishing force depends on the compression length experimented by the spring, and varies linearly according to Hooke’s Law. The tool is equipped with a ball of hardened chromium steel (100Cr6) with a hardness value of approximately 57–66 HRC and a diameter of 6 mm.

Although this tool is also able to perform the vibration-assisted ball burnishing process, this work only includes the testing through conventional ball burnishing. This decision was taken due to the fact that only the influence of the force and number of passes on the fatigue life wanted to be assessed. To attach the tool to the lathe, a special adapting system was designed (Fig. 1B and C).

2.2Fatigue specimenFatigue specimens where manufactured from cold drawn AISI 1038 bars of 12-mm diameter. A stress-strain curve and the material properties obtained from tensile tests are shown in Fig. 2.

The specimens were manufactured in a CNC lathe, to obtain the geometry shown in Fig. 3. The manufacturing parameters used were fitted so that the surface average roughness (Ra) was around 1 µm. The specimen’s length and diameter were defined according to the characteristics of the fatigue testing machine, whereas the diameter reduction ensured the stress concentration in a specific section of the specimen. The transition surface between the initial section and the final one was solved through a radius of 4 mm. This value would allow the burnishing operation on the resulting fillet in the diameter transition area with the 6-mm ball installed in the burnishing tool.

2.3Rotating bending fatigue machineThe tests were executed with a GUNT WP 140 rotating bending machine. The force was controlled through a dynamometer installed on a manual actuator, through which the 8-mm diameter edge of the specimen was vertically loaded. The specimen rotated during the test at a constant speed of 2800 min−1, thus generating the alternative bending moment that accounts for the fatigue load. Due to this low rotation speed, the tested number cycles was limited to obtain a reasonable timespan. Therefore, the low cycle fatigue regime was chosen as the object of this study. Hence the selected stress amplitudes.

The relationship between the applied force and the maximum bending stress can be calculated through the Euler-Bernoulli expression particularized for a cantilever beam, point-loaded at the tip, with round section (Eq. 1)

where Sa is the maximum stress due to bending moment, F is the applied force, L is the cantilever length, and d is the beam minimum diameter.For the elaboration of the Wöhler curves, different forces were applied to generate the respective maximum bending stresses (Table 1). As the material provider indicated that the fatigue limit of the cold drawn raw material was 318 MPa, the testing forces were selected to obtain higher maximum stresses, in an attempt to search the fatigue strength of the ball-burnished specimens.

2.4Experimental designA two-level factorial design with one central point was planned, including two factors: the burnishing force and the number of passes (Fig. 4). The levels of these parameters were decided based on results obtained by the research team in previous experiments, ensuring a significant variation of the response variable, that is, the expected fatigue lifespan for different fatigue stresses. For that reason, 80-N and 120-N forces were considered as extreme points, coupled with 1 and 7 passes. Consequently, the central point of the experimental was set at 100 N and 4 passes, deriving in five different testing conditions (Table 2).

For each factor combination included in Table 2, a Wöhler curve was represented. To construct those curves, the maximum bending stresses included in Table 1 were considered. For each stress level, six repetitions were performed. This value was suggested by Lee et al. [22] as an admissible number of points in the Wöhler curve for preliminary fatigue tests. Therefore, 36 tests were performed for each ball burnishing condition (including non-burnished specimens as control values). In overall, 216 fatigue tests were performed to generate six Wöhler curves, that are analyzed in the next section.

3Result discussion3.1Hardness resultsSurface hardness represents the degree of plastic deformation caused on the specimen. It was measured through Vickers microindentation tests by applying a 5-N load. As the specimen surfaces are cylindrical, these measurements were corrected through numerical factors included in the ASTM E92-82 standard. For each specimen, 20 indentations were taken around the circumference next to the chamfer where the stress was concentrated, guaranteeing that the space between them was enough so that the results were not compromised. It was considered the most representative value for the hardness in the fracture section. Subsequently, a confidence interval was calculated for each dataset to obtain a representative hardness value associated to all burnishing conditions.

Fig. 5A represents these intervals as a function of the force and number of passes applied on the specimen. It can be observed that hardness increases as the degree of plastic deformation increases, which confirms the impact of the variation of these parameters in the cold work experienced by surface layers. An alternative representation is shown in Fig. 5B, where the mean surface hardness calculated through an analysis of variance (ANOVA) presents a clear increasing trend of the Vickers hardness as the force and number of passes increase.

3.2Fatigue behavior through the Wöhler curvesFig. 6 represents the S-N curves obtained for all tested specimens. A six-point cloud represents the lifespan results for each burnishing condition tested at a specific maximum stress. Therefore, it represents the expected fatigue life associated to that burnishing condition tested at that maximum alternating stress.

The non-burnished specimens show the lowest expected life for all stress levels. Therefore, it can be stated that ball burnishing improves the fatigue life of AISI 1038 parts. This result agrees with the conclusions presented by other researchers [7,13], but it is worth highlighting the relevance of Fig. 6 in the fact that it evidences that the extent to which fatigue life is improved is dependent on the force and/or number of passes applied through ball burnishing. Indeed, the resulting fatigue curves are visibly displaced towards a higher number of expected cycles to failure as a result of the increase of the force and number of passes.

By analyzing the relative position of all curves, it can also be concluded that a decrease of the burnishing force can be compensated by an increase in the number of passes, since both factors have a similar impact on the surface hardening, and, therefore, the improvement in terms of fatigue. In fact, being the hardening effect dependent on the level of force and the number of passes, these results show that the increase of cold work caused by ball burnishing on the surface layers of the material has a positive impact on the improvement of the expected fatigue life in rotating bending tests. Therefore, by combining the level of force and the number of passes, the user is able to impact directly on the expected fatigue life of the material.

The point clouds that represent the fatigue results for each burnishing parameter set disperse as the maximum stress decreases, being more scattered at the minimum stress level of 360 MPa. This should not be forgotten, even when considering the curves derived from the fitting. For instance, although the curves corresponding to 80N-7passes and 100N-4passes are overlaid in the Wöhler graph, they come from different point clouds, meaning that the effect is not exactly the same. In fact, they showed a difference of around 10 HV. They are however the parameter combination leading to the most similar results. On the contrary, results for 460 MPa are clearly grouped for each burnishing condition. This could be caused by a fastest stress relaxation derived from a higher stress level, which leads to the homogenization of results in a shorter time. This can be also seen numerically in Table 3.

Mean values and standard deviation of lifespan for all burnishing conditions at all maximum stresses tested.

Maximum stress (MPa) | Specimen type | |||||
---|---|---|---|---|---|---|

As machined | 80 N 1 pass | 80 N 7 passes | 100 N 4 passes | 120 N 1 pass | 120 N 7 passes | |

360 | 18567 ± 101 | 23966 ± 64 | 24991 ± 174 | 25491 ± 333 | 26736 ± 305 | 31422 ± 209 |

380 | 12226 ± 131 | 18877 ± 142 | 19895 ± 82 | 19895 ± 135 | 21434 ± 74 | 25114 ± 94 |

400 | 8533 ± 61 | 12427 ± 85 | 16441 ± 183 | 15791 ± 116 | 16735 ± 66 | 19196 ± 78 |

420 | 6139 ± 91 | 8360 ± 76 | 11608 ± 124 | 11608 ± 133 | 13912 ± 53 | 15309 ± 106 |

440 | 5274 ± 73 | 6267 ± 59 | 7834 ± 42 | 7834 ± 45 | 10627 ± 54 | 11147 ± 83 |

460 | 4927 ± 156 | 5265 ± 49 | 6099 ± 69 | 6099 ± 13 | 8014 ± 35 | 9349 ± 77 |

An alternative way to observe the dependence of surface hardening and fatigue life, is by analytically correlating the HV representative values with their correspondent power law curve parameters. Indeed, the points that compose the clouds in Fig. 6 can be represented by the Wöhler model, that is, a mathematical description of the S-N curve through a power law curve, as represented by Eq. (2). This equation typically defines the fatigue behavior of materials inside the elastic domain. The coefficients corresponding to the Wöhler models of each curve are represented in Table 4.

where Na are the cycles to failure, Sf is a constant typical of each type of specimen, and Sa is the maximum stress to which they are subjected during the fatigue test.Coefficients of every Wöhler curve (Eq. (2)) for each ball burnishing condition.

Specimen type | Sf (MPa) | b |
---|---|---|

As machined | 1414 ± 23 | −0.140 ± 0.002 |

80 N – 1 pass | 1653 ± 44 | −0.151 ± 0.003 |

80 N – 7 passes | 1909 ± 71 | −0.163 ± 0.004 |

100 N – 4 passes | 1917 ± 58 | −0.163 ± 0.003 |

120 N – 1 pass | 2866 ± 91 | −0.203 ± 0.003 |

120 N – 7 passes | 2723 ± 66 | −0.195 ± 0.002 |

Fig. 7 shows that both parameters of the Wöhler curve present a linear relation with the hardness exhibited by each burnishing condition. The lowest hardness level is associated to the as-machined specimens, which also show the extreme values for both the constant and exponent of the Wöhler curve. As a higher amount of cold work is achieved on the surface through ball burnishing, these parameters increase and decrease, respectively. Indeed, this observation confirms that a higher degree of plastic strain due to ball burnishing is represented graphically by an increase in the offset between the curves as a higher burnishing force and number of passes are applied, and analytically by an increase of the Sf parameter. It is also worth noting that the point corresponding to the 100 N with 4 passes is overlapped with the point 80 N with 7 passes, as was observed in the ANOVA calculated for the hardness results, which leads to conclude that both sets of parameters deliver the most similar results in terms of hardness and fatigue life.

3.4Influence of force and number of passes on the cycles to failureAt sight of the results represented in the Wöhler graphs in the previous subsection, the cycles to failure presented by the burnished specimens corresponding to the maximum and minimum stress levels have been statistically analyzed. An ANOVA analysis, with a confidence level of 5%, has been performed on the bending fatigue test results to assess the factorial effects of force and the number of passes. As an example, Fig. 8 represents the graphical results of the mean effect of each parameter level on fatigue lifespan for the maximum stresses of 360 and 460 MPa. This representation confirms the direct relation between the force with the expected cycles to failure at all maximum stresses. However, no statistically significant response modification between 1 pass and 4 passes is detected by the ANOVA test (the p-value associated to this variable is higher than 5%). Therefore, declaring that 120 N and 7 passes are the best process parameters leading to fatigue life enhancement at all stress levels is a valid statement, and confirms the graphical information shown by all Wöhler curves in Fig. 6, and the previous discussion about Fig. 7.

3.5Visual inspection of crack growthIn order to analyze the sections of the specimens after fatigue failure, different optical images were acquired through a MOTIC SMC microscope equipped with a MOTICAM 3 digital camera. First of all, it was confirmed that all specimens broke at the sections where stress concentration occurred due to the fillet radius along which the section changed. That area had been externally subjected to ball burnishing, meaning that the process cannot prevent stress concentration.

In most of the cases, the subcritical fatigue crack propagation area of the specimen was distributed as a ring, around the final fractured area at the center of the section. This surface corresponds to the material that supported higher stress during the test. Fig. 9A represents the described effect at the fracture section of a non-burnished specimen. As the fatigue test advanced, the crack nucleated at the surface, and propagated until the subcritical area was not able to bear the stress, thus harming the part integrity. It is at that point that the stress is concentrated near the cylinder axis, causing final failure. Consequently, the different aspect of the area inside the subcritical ring corresponds to the material that caused a catastrophic failure.

4ConclusionsAISI 1038 specimens have been burnished under different combinations of loading force and number of passes. From the observed results, the following conclusions can be drawn:

- 1
Ball burnishing increases the expected fatigue lifespan for AISI 1038 specimens, according to rotating bending tests. This enhancement is favored by the cold work induced by the process on the specimen, correlated to the self-hardened surface on the stress concentration area.

- 2
The burnishing force and number of passes are mutually substitutive in terms of fatigue resistance enhancement, as only the specimens treated with both the highest force and number of passes, and both their lowest values, are significantly different from the other three burnishing conditions (central point and combination of lowest and highest values).

- 3
Performing the ball burnishing operation with a 6-mm ball and 120 N and 7 passes the AISI 1038 material is a suitable combination for the application of the process in terms of fatigue life enhancement.

The authors declare no conflicts of interest.

FundingFinancial support for this study was provided by the Ministry of Science, Innovation and Universities of Spain, through grant RTI2018-101653-B-I00, which is greatly appreciated.