1 INTRODUCTION
The notch sensitivity (q) is a very important design parameter, since its estimation directly influences the design against fatigue failure of mechanical parts. Also notch sensitivity is associated with many other mechanical design concepts such as: stress concentration factor (SCF) or K _{t,} fatigue SCF or K _{f} and short cracks propagation. Indeed, the SCF is a wellknown parameter used in mechanics of solids to estimate the stress amplification caused by a mechanical part discontinuity, like a notch. Its estimation depends entirely on the loading type and the geometry of the discontinuity. It is also wellknown that for variable load the stress amplification, for ductile materials, is usually overestimated by K _{t}. To reduce the conservatism in the design of mechanical parts that are subjected to variable loading, the fatigue SCF is used. This variable is also a function of notch sensitivity factor (q) that takes into account, for notches, both: notch radius (r) and material ultimate strength (S_{ut}), recognizing that material influences on the stress amplification near notch discontinuity.
Research has been done in this area by ^{Castro and Meggiolaro (2013}), which associated the notch sensitivity with the relatively fast generation of tiny nonpropagating cracks at notch root. According to the authors it indicates that the crack initiation is controlled by the stresses acting at notch tip and the fatigue crack growth behavior of short cracks emanating from notch root could be affected by the generated stress gradients. So, a crack could start at a notch tip and then stop after a short growth, due to the decreasing of the driving force in spite of its growing size. Thus, q should depend not only on the notch tip radius but also on the notch depth. ^{Castro et al. (2014}) addresses semiempirical q factors that consider the notch tip stress gradient influencing on the fatigue behavior of short cracks. According to the authors, properly calculation of q values can be done from its fatigue limit (Se), and crack propagation threshold (ΔK_{th}), considering all the notch geometry and loading characteristics, without the need of any adjustable parameter.
^{Hu et al. (2012}) made a review of various concepts that are related to q as fracture process zone, damage process zone and theory of critical distance (TCD), suggesting an exponent m to estimate q. The results indicate that there is an optimal m value for each material. Also, ^{Arola and Williams (2002}) evaluated the apparent K _{t} of surface texture on the fatigue life of highstrength lowalloy steels (HSLAS). It was found that the fatigue life of the used HSLAS was surfacetexturedependent, with the fatigue strength decreasing with the increase of the surface roughness.
^{Chapetti and Gerrero (2013}) proposed the integration of fracture mechanics concepts to account for both: the fatigue notch sensitivity and the notch size effect. In fact, for components tested under cyclic loading containing notches or holes of a given K _{t}, with their fatigue strengths measured, the smaller are the notch or hole sizes, smaller is the decrement of K _{f} value.
In this article, an integrated approach was implemented to estimate the notch sensitivity of high strength wires (HSW). To accomplish it, K _{f} was obtained from experimental results of notched and notched free specimens. The estimation of K _{t} was implemented through the use of finite element commercial software, doublechecked by an analytic approach as in ^{Young (1989}). Thus, the estimation of q was obtained through the utilization of classical relationships between K _{t} and K _{f}.
2 EXPERIMENTAL APPROACHES
The experimental approaches were divided in: material characterization, fatigue apparatus and fatigue procedure. The material characterization was divided in chemical analysis, tensile tests, optical microscopy, hardness and roughness tests. The fatigue apparatus describes the specimen dimensions, servohydraulic machine setup and test accessories. The fatigue procedure explains the staircase method used to determine the fatigue limits. The material characterization is addressed in sequence.
2.1 Material Characterization
The material characterization was done through the implementation of chemical analysis, tensile tests, optical microscopy, hardness and roughness tests of the HSW specimens. Table 1 shows the results of chemical analysis done by the use of optical emission spectroscopy technique. Table 2 shows the tensile tests results performed to determine the material main monotonic mechanical properties, such as: Yield Strength (S_{y}); Ultimate Strength (S_{ut}) and Elongation (EL). For tensile tests, three specimens of 200 mm length were used with 50 mm gage length for elongation measurements. It can be seen from Tables 1 and 2 that high values of S _{y} and S _{ut} are consistent with high strength steels (HSS) and compatible with a SAE 1065 steel material.
Chemical composition in weight  

element  C  Si  Mn  P  S  Cr  Mo  Ni  Al 
(%)  0.64  0.22  0.74  0.019  0.007  0.037  0.009  0.023  0.028 
Specimen  S_{ y } (MPa)  S_{ ut } (MPa)  EL (%) 

1  1150  1260  15.3 
2  1180  1270  15.0 
3  1150  1250  14.6 
average  1160  1260  14.9 
The Young modulus and Poison ratio were estimated, respectively, as E = 200 GPa and ν = 0.29. Also, it is important to observe that a high S _{ut} is directly related to high fatigue resistance (S_{f}).
To implement the hardness tests an Instron Wolpert machine was used with a Rockwell C indenter. To obtain the surface roughness it was used a TesaRugosurf 10 equipment. Tables 3 and 4 show the hardness and the roughness results, respectively.
Measurement  Ra longitudinal (μm)  Ra transversal (μm) 

1  0.22  0.58 
2  0.14  0.65 
3  0.22  0.45 
average  0.19  0.56 
Note that the average hardness of the HSS, of Table 3, resulted in a quite high value for steel without the use of any thermomechanical process, as quenching. Also, note that the average roughness measurement, in Table 4, revealed a quite different roughness depending on the test direction. Optical microscopy was used to assess the microstructure wires aspect. The observed transversal and longitudinal (reference to the rolling direction) cross sections are shown in Fig. 1.
It can be observed that the microstructure is composed of elongated grains, in the rolling direction, of ferrite and pearlite. The volumetric fraction of pearlite is consistent with an eutectoid carbon steel composition. It can be also observed fine elongated white areas between pearlite grains which correspond to the primary ferrite. Such microstructural aspect suggests a severe cold working, compatible with the last rolling pass that could be applied under the critical temperature. Such procedure is in line with materials aiming to attain high yield strength and ultimate tensile strength levels. The fatigue apparatus is addressed in sequence.
2.2 Fatigue Apparatus
To experimentally assess K _{f} value, it was necessary to obtain the notched and notched free specimens fatigue limits. To accomplish this objective notches and notched free specimens were tested in a servohydraulic test machine, equipped with a four points bending fixture.
The specimens were submitted to different loading levels, maintaining the same loading ratio R = σ_{min} / σ_{max}. All specimens were manufactured in house by cutting 200 mm length bars. Concerning the specimen’s surface finishing, they were far from highly polished; indeed all surface finishing was maintained as received, with arithmetic average of absolute values of roughness (R_{a}) shown in Table 4. In this way, it was considered that the fatigue curves were related to real parts rather than the usual high polished specimens. Fig. 2 shows both specimens types.
The dimensions were identical for both specimens’ types, notched and notched free, unless by a 0.4 mm shallow notch for the notched specimens generated (at transversal direction) by a broaching process in half its longitudinal length. Figure 3 shows the servohydraulic test machine used to implement the experimental fatigue tests, where Fig. 3.a shows the servohydraulic test machine general view and Fig. 3.b shows a specimen positioned on the inferior part of the four points bending fixture.
Fig. 4 shows the four points bending fixture schematic drawing that is used to show some important geometric and loading characteristics.
The distances between rollers were defined in function of the specimen length. The bending moment is considered uniform between the upper rollers. The experimental tests were conducted using pulsating loads, to guarantee the upper rollers contact with specimen during the tests. The fatigue loading levels were defined as a ratio between the specimen maximum stress (σ_{max}) and the S _{ut}, always maintaining the same loading ratio R.
The maximum force (F_{max}) applied by the test machine was calculated as in ^{Crandall et al. (1978}):
Where, I is the specimen second moment of area, a is the distance between lower and upper rollers (see Fig. 4) and c is the specimen semiheight (note that for notched specimen, the notch depth is deducted).
The four points bending apparatus was used, as shown schematically in Fig. 4, to apply the pulsating load. The ratio was maintained at R = 0.1. As the pulsating sine load has nonzero mean stress it was necessary to transform both, mean stress (σ_{m}) and alternating stress (σ_{a}), on equivalent alternating stress (σ _{a_eq}), as in ^{Budynas and Nisbett (2014}) and ^{Castro and Meggiolaro (2016}):
It is also possible to rewrite equation 2.b to be a function of σ _{a_eq}:
The fatigue procedure is addressed in sequence.
2.3 Fatigue Procedure
One of the main objectives of this article is focused in obtaining, with maximum possible accuracy, fatigue limits of both notched and notched free HSS specimens. To accomplish this objective the staircase method was used, as in ^{Dixon (1965}) and Dixon (1948). This method uses a simple idea: a specimen is submitted to a certain stress level, after a certain number of cycles if this specimen breaks, the next specimen has the stress level decreased by a predefined stress decrement. However, if the specimen doesn’t break, then the stress level is increased by a predefined stress increment. This process continues until all specimens are used or reaches a significant number of specimens that are statistically representative according to ^{Dixon and Mood (1948}). The staircase method flowchart is presented in Fig. 5. This simple method can produce a much better estimative of the notched and notched free specimen fatigue limits than the traditional Wohler approach.
Where C
_{1} is a positive constant smaller than one that multiplies S
_{ut} to estimate the first maximum stress σ_{max1}, C
_{2} is a positive constant bigger than one that divides the constant C
_{1} to generate the fraction of S
_{ut} called stress increment. s is the increment or the decrement of stress and n is the number of specimens utilized. i is a counter that varies from 1 to n+1.
The numerical approach is addressed at the next section.
3 NUMERICAL APPROACH
It is well known that the SCF produced by a given notch depends on the mode of loading and on the notch geometry. There are many ways to estimate it: by theory of elasticity, as in ^{Timoshenko and Goodier (1951}); by experimental measurements as in ^{Pilkey and Pilkey (2008}); using, for example, photoelasticity method, as in ^{Dally and Riley (1991}), or by using a numerical method like finite element method, as in ^{Fish (2007}). In this section a numerical approach, using finite element method, is used to estimate the stress concentration factor of the shallow notch.
3.1 Numerical Model
In this article it was used the commercial software ANSYS WORKBENCH v14.0 to evaluate SCF produced by a shallow notch broached on HSS specimen. For numerical model purposes the specimen material was considered isotropic and linear elastic. The mesh was generated with hexahedrons elements PLANE 186 with plane stress option, with one node at each vertex, as shown in Fig. 6.a. The finite element model was created in 2D dimension (in the XY plane) to reduce the computational effort, as shown in Fig. 6.b.
The 2D model with plane stress option was considered based on ^{E399 ASTM standard (2009}). In item 7.1.3 of this standard, the minimum specimen width to develop plane strain condition is related to the material S_{y}/E ratio. Using the HSS material properties, this ratio was calculated as 0.0058, which corresponds in this standard to a minimum recommended width of 64 mm to develop plain strain condition. However, as the HSW specimen has 9 mm of width, it is quite reasonable to suppose that the state of stress is much closer, even at half specimen width, to plane stress than to plane strain condition.
Due to the notch depth is much smaller than the specimen height, in fact it is a shallow notch, and to ensure good elements aspect ratio, mainly around the notch, a refined region around it was provided as shown in Fig. 7.a. Figure 7.b shows a photo of the real notch machined by a manual broaching machine.
To guarantee the finite element model implementation consistency, a mesh convergence analysis was implemented, refining the element size, (δ) around the notch radius. Different δ were evaluated: 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, and 0.075 mm. The mesh convergence was studied in function of the maximum normal stresses in the xdirection (see Fig. 6.b). Fig. 8 shows a graphical representation of the mesh convergence study.
The mesh refinement of δ = 0.075 mm was selected for this analysis because the notch boundary can be described with a reasonable accuracy. Both, experimental and numerical results are presented at the next section.
4 RESULTS
The results are divided in two parts: experimental and numerical. At experimental part, the fatigue limit results obtained from the experimental tests are shown and, at numerical part, the SCF result obtained from the finite element method is shown.
4.1 Experimental Results
Fig. 9 shows a representation of the test sequence, for both notched and notched free specimens, applying the staircase method, as in ^{Pollack et al. (2006}), which algorithm can be seen in Fig. 5.
Note that all stresses are divided by S _{ut}, generating a ratio (σ _{a_eq} / S _{ut})·100, which varies from 0 to 100%. Fig. 10 presents all fatigue results, with common points that were obtained by usual fatigue approach and upanddown points by using staircase method. The common points are represented by black squares and the upanddown points are represented by blue circles. Note that an arrow represent that specimen didn’t fail, for all figures. There is one arrow per specimen, unless in Fig. 10.b, where the not broken specimens were so near each other, that just one arrow was used.
The fatigue results which are graphically shown in Fig. 10, correspond to actual fatigue results values that are shown at Appendix, in Tables 6 and 7. Also in Fig. 14, at the same Appendix, the staircase method flowchart explains, in a graphical way, the two S _{e} estimation, which are represented by blue lines in Fig. 10. In Figs. 11.a and 11.b show a detailed view of upand down points.
The fatigue limit results shown on Table 5 were obtained through the utilization of staircase method flowchart  estimation of S _{e} range, shown in Fig. 14 at Appendix. The experimental fatigue results are available at Tables 6 and 7, also at Appendix, or graphically in Fig. 10.
Specimen  μ  σ  Fatigue Limit (%)  

Common points  notched free  35.5 

33.9 ≤ (S _{ e } / S _{ ut } ) · 100 ≤ 37.1 
notched  15.5 

14.8 ≤(S _{ e } / S _{ ut } ) · 100 ≤ 16.2  
Upanddown points  notched free  32.7 

32.3 ≤(S _{ e } / S _{ ut } ) · 100 ≤ 33.0 
notched  12.7 

12.6 ≤(S _{ e } / S _{ ut } ) · 100 ≤ 12.8 
Note: μ and σ were maintained from the notation adopted in ^{Pollack et al. (2006}).
The mean and the standard deviation of fatigue limit, respectively, μ and σ, of Table 5 were assessed for two cases: if the failures were the majority event, the lower stress level of survival occurrence was estimated or if the survivals were the majority event, the lower stress level of failure occurrence was estimated. Note in Table 5 that fatigue limit for both notched and notched free specimens were estimated with less dispersion when it was used the upanddown approach, based in the staircase method when compared with the common points, obtained with usual approach.
The numerical results were assessed in sequence.
4.2 Numerical Results
To calculate the SCF, the specimen model was loaded in pure bending, to simulate the four points bending test. The bending moment value was imposed to generate a unity normal stress on longitudinal fiber at the same height of the notch root but far from it. Using this approach, the maximum stress value at the notch root turns out to be exactly the value of the SCF. Fig. 12.a shows the finite element results in a usual representation and in Fig. 12.b it was used a path approach, at the same height of the notch root, to explicitly show the stress concentration distribution.
As expected, the stresses form a high gradient curve in the neighborhood of the shallow notch root. The ratio between maximum stresses at notch root and nominal stresses at a region far from notch, was used to estimate K _{t}:
Note that using the analytical approach shown at Appendix the K _{t} was estimated as 2.79, which is quite close to the finite element estimative of 2.84. So, the K _{ t } value was estimated inside the following range: 2.79 ≤ K _{ t } ≤ 2.84.
At next section, the material notch sensitivity factor is assessed.
5 DISCUSION
In this section the results obtained by experimental and numerical approaches are used to estimate the HSW notch sensitivity factor.
The fatigue limit range for each specimen condition can be obtained from Table 5: 32.3 ≤ (S _{ e } _/S _{ ut } )_{ notched free } ∙ 100 ≤ 33.0 and 12.6 ≤(S _{ e } /S _{ ut } )_{ notched } ∙ 100 ≤ 12.8. The fatigue stress concentration factor can be stated, as ^{Budynas and Nisbett (2014}):
So the fatigue stress concentration factor K _{f} can be estimated, within the following range: 2.56 ≤ K _{ f } ≤ 2.58. In sequence, the notch sensitivity can be estimated through the utilization of a classic Peterson’s equation, as ^{Budynas and Nisbett (2014}):
Finally, applying the ranges calculated, respectively SCF and fatigue SCF ranges in equation (06), the range within the notch sensitivity can be estimated as: 0.86 ≤ q ≤ 0.87.
Analyzing the range for notch sensitivity factor, it can be said that the notch sensitivity narrow range obtained by the application of the proposed integrated approach revealed to be quite effective, manly due the use of the staircase method in the assessment of fatigue limits of the notched and the notched free specimens. On the other hand, the notch sensitivity range revealed to be high, indicating a probable brittle behavior, which is compatible to high strength materials. In fact, although the high ultimate strength is a desirable mechanical property to increase the fatigue resistance, the high notch sensitivity can induce deleterious effects to mechanical structures submitted to fatigue loading.
6 CONCLUSIONS
An integrated approach was used to assess the notch sensitivity of HSW. Material characterization was obtained through the implementation of chemical analysis, tension tests, hardness tests, roughness tests and optical microscopy of the HSW. Fatigue tests were accomplished with two types of wire specimens, the notched and the notched free, resulting in very distinct fatigue limits, which were used to estimate the fatigue SCF. Finite element method and analytical approaches were used to estimate the SCF with good matches between them. Finally, the notch sensitivity factor was obtained through the utilization of a well know Peterson’s equation.
In fact the utilization of the experimental/numerical integrated approach for this research was revealed quite promising in generating an estimative of HSW notch sensitivity value. In particular, the utilization of staircase method at experimental approach, really improves the fatigue limit estimation, which taking into account the natural tendency of fatigue results dispersion, contributed, in a very significant way, to obtain such a reasonable estimative of notch sensitivity.