The material used in this study was Al6061-T0 alloy with formability improved through annealing. A tensile test was conducted in accordance with the KS B 0802:2003 standards to determine the mechanical properties of the material. The gauge length and diameter of the specimen were 25 mm and 5 mm, respectively. Table 1 shows the mechanical properties determined from the tensile test.

The Cockcroft-Latham ductile fracture criteria were used to evaluate the shear and fracture behavior of the material during trimming based on results in previous studies [6][10][11]. Many theories were proposed to predict ductile fracture. The Cockcroft-Latham model is known to be easily applicable and produce accurate results for the bolt trimming process.

where ε¯ is the effective strain, σ_max is the maximum principal stress, ε¯f is the effective strain at fracture, and C₁ is the critical damage value. The critical damage value of Al6061-T0 was found to be 74.4 from the tensile test results.

The trimming process was analyzed using DEFORM-2D, a commercial FEA software program. The analysis model was selected based on the cross section A-A’, where the maximum scrap of the material occurs during trimming, as shown in Figure 1. During analysis of the trimming process, axisymmetric models can be used to reduce the analysis time compared to 3D models and they are useful in quantitatively evaluating formation of non-uniform burr along the shear surface [3][5]. Figure 2 shows the axisymmetric FEA model used for the trimming analysis. Smaller meshes were used in the shear zone to increase the accuracy of the description of shear and fracture. As a result, 8,494 nodes and 8,230 elements were generated. Moreover, the deformation history accumulated during forging was imposed on the material by performing the analysis of the multi-stage cold forging process prior to trimming analysis [12]. The friction constant (m) was determined to be 0.1 based on results from a previous study [11], and the trimming punch speed was set to 100 mm/sec based on the specifications of cold formers for prototype production.

Figure 1: 
Description of the cross-section for FE-analysis

Figure 2: 
FE-model of the trimming process

The trimming process for the Al6061 bolt head was optimized using the Taguchi method. The blade radius (BR) of the punch, land width (LW) of the bottom die, and the stop distance (SD) between the punch and the bottom die were selected as design parameters, as shown in Figure 3. Unlike sheet trimming, which is completed in a single shearing process, bolt trimming is divided into the two shearing processes as shown in Figure 4 [6]. In the first sequence, the punch performs shearing up to the determined stop distance while pressing the material. In the second sequence, the final shearing process is performed while the knockout pin installed inside the bottom die pushes on the connection between the bolt and scrap (opposite the punch movement). If the end of the trimming punch align with the end of the bottom die land section (LW = 0) as shown in Figure 5, a relatively large crack zone occurs in the middle of the shear surface, thereby degrading the shear surface quality. According to results from a previous study, applying a compressive hydrostatic stress during trimming delays cracking occurrence and increases the effective shear surface [1]. Therefore, cracks must be suppressed during shearing to secure a sound shear surface until final shearing by applying a reaction force against the trimming load to the bottom die. Accordingly, the land width of the bottom die was offset by 0.5, 1.0, and 1.5 mm from the shear surface along the bolt head. The application ranges of the punch blade radius and the stop distance were determined from the results in a previous study [5]. To generate the L₉ (3³) orthogonal array table, all parameter ranges were divided into three levels. Tables 2 and 3 show the design parameters with three levels and L₉ (3³) the orthogonal array table created for using the Taguchi method, respectively. FEA was conducted based on the orthogonal array table.

Figure 3: 
Design parameters required in the trimming process

Figure 4: 
Schematic illustration of the trimming process: (a) initial state, (b) 1^st sequence, and (c) 2^nd sequence.

Figure 5: 
Crack propagation during trimming: (a) 16.6% stroke, (b) 39.1% stroke, (c) 66.2% stroke, and (d) 98.0% stroke.

The shape defects (DF_shape) and peak trimming load (Load_peak) were set as objective functions to examine the effects of the design parameters (selected above) on the shear surface quality and shear load. The shape defects are shown in Figure 6 and defined in Equation (2) as follows:

Figure 6: 
Description of shape defects

where A_R, A_U, and A_B represent the roll-over, under-filling, and burr areas, respectively.

Figure 7 shows the trimming load curve as a function of the punch stroke. As shown in the figure below, multiple shearing and fractures take place during trimming. The peak trimming load occurred during the initial shearing because it had the maximum shear area. Later, the load gradually decreased with the periodic increase and decrease according to the punch stroke. As it is favorable when both the shape defects and peak trimming load are smaller, the signal-to-noise (SN) ratio was calculated by applying the loss function of the-smaller-the-better characteristics in Equation (3):

Figure 7: 
Punch stroke-trimming load curve

where y is a measured value and n is the number of measurements.

Table 4 and Table 5 show the results of FEA and the Taguchi method to the shape defects and peak trimming load, respectively. Figure 8 shows the effect of each parameter on shape defects and peak trimming load. The process parameter that had the greatest impact on shape defects was the blade radius, followed by the stop distance and land width. The process conditions that created the highest SN ratio were 0.2 mm blade radius, 1.0 mm land width, and 0.25 mm stop distance.

Figure 8: 
Main effect plots for objective functions: (a) shape defects and (b) peak trimming load

The peak trimming load showed high sensitivity only to the shape of the punch (blade radius), and the influence of land width and stop distance were insignificant. The SN ratio of the trimming load tended to be inversely proportional to the blade radius. This is because the contact area between the punch and the material increases as the blade radius increases [5]. Therefore, the optimum process conditions for the trimming load were 0.2 mm blade radius, 1.5 mm land width, and 0.75 mm stop distance.

The optimum process conditions of the land width and stop distance were selected based on the shape defects that exhibited relatively high sensitivity. Therefore, the optimum conditions in the trimming process derived from the FEA results with the Taguchi method were 0.2 mm blade radius, 1.0 mm land width, and 0.25 mm stop distance.

Based on the above results, FEA was conducted. Figures 9 and 10 show the damage distribution and hydrostatic stress distribution of shear surface at various punch stroke values, respectively. Although locally high damage values appeared near the middle of the shear surface, a good shear surface without crack defects formed during preliminary analysis was obtained. This can be attributed to the compressive hydrostatic pressure that acted on the shear zone between the punch and the material.

Figure 9: 
Damage distribution in the optimum process condition: (a) 19.9% stroke, (b) 40.4% stroke, (c) 60.9% stroke, (d) 88.7% stroke, (e) shear by KO_pin, and (f) final state.

Figure 10: 
Mean stress distribution in the optimum process condition: (a) 19.9% stroke, (b) 40.4% stroke, (c) 60.9% stroke, (d) 88.7% stroke, (e) shear by KO_pin and (f) final state.

Moreover, as with previous studies, final shearing by the knockout pin was performed while cracks occurred on the shear surface A-A’ with a certain angle. However, the ‘dragged through’ phenomenon mentioned in a previous study was not observed [6]. This appears to be because analysis was conducted in the previous study without considering the accurate critical damage value of the material through experiments. As the shape of the shear surface varies depending on the critical damage value during trimming, it is very important to acquire accurate mechanical properties and a critical damage value for the material through experiments.

Figure 11 and Table 6 show the analysis results for the shear surface shape and trimming load. Satisfactory shape defects and peak trimming load results were obtained.

Figure 11: 
Sheared plane profile in FEA

However, as shown in Figure 12, roll-over occurred on the top of the bolt head due to the indentation of the punch. In general, during sheet trimming, roll-over occurs along the punch load direction due to a bending moment. On the other hand, in the case of a bolt trimming process, the material around the punch blade section rises as much as the volume of the punch’s penetration due to the reaction force from the bottom die.As roll-over is proportional to the blade radius, it seems desirable to design the blade radius of the punch to be as small as possible.

Figure 12: 
Velocity distribution on the upper part of the bolt head: (a) 0.025% stroke (b) 0.15% stroke (C) 0.3% stroke (d) 0.45% stroke.