Benchmark 2 – Springback of a Jaguar Land Rover Aluminium

The aim of this benchmark is the numerical prediction of the springback of an aluminium panel used in the production of a Jaguar car. The numerical simulation of springback has been very important for the reduction of die try outs through the design of the tools with die compensation, thereby allowing for the production of dimensionally accurate complex parts at a reduced cost. The forming stage of this benchmark includes one single forming operation followed by a trimming operation. Cross-sectional profiles should be reported at specific (provided) sections in the part before and after springback. Problem description, tool geometries, material properties, and the required simulation reports are summarized in this benchmark briefing.


INTRODUCTION
Springback is one of the most important problems for the sheet metal forming industry due to the strong geometrical deviations which occurs through elastic recovery after forming. These deviations can lead to many manufacturing difficulties such as joining parts together into a more complex assembly. Springback is influenced by the forming operations and the degree of constraints imposed by the geometry of the part but it is also strongly dependent on the material properties of the blank sheet. For aluminum, springback behaviour is more complex because of its strong plastic anisotropy and low Young's modulus. Consequently, inaccurate material models can lead to major or unexpected deviations in the prediction of springback.
The main objective of this benchmark is to predict the springback of a single stage formed panel, assess the influence of material models and quantify the influence of different numerical modelling techniques that affect springback prediction. Numerical techniques includes the finite elements used, integration rules, implicit or explicit code analysis, contact and friction models and the use of emerging techniques such as isogeometric analysis and meshless methods.
The kinematic hardening effect of bending and unbending deformation through the different die radius and curvatures of the tools can significantly influence the nature and prediction of panel's springback. The springback prediction of different loading/unloading forming operations requires the use of appropriate kinematic and/or combined kinematic/isotropic hardening models, together with sophisticated flow rules and yield functions. Cyclical shear tests for different levels of pre-strains were therefore performed for the material characterisation of the kinematic/isotropic hardening (the Bauschinger effect) for this benchmark study and the measured shear strain-stress curves are summarised in the attached excel file "Cyclical_Shear.xls".

BLANK MATERIAL
The blank material to be used in this benchmark is the aluminium alloy (AA6451-T4) with thickness t = 3.0 mm. The elastic mechanical properties are given in Table 1. The uniaxial tensile yield stress and r-values are given in Table 2. The equal biaxial tensile yield stress and the biaxial r-value are given in Table 3. The material constants for the hardening curve at 0 degrees from the rolling direction (RD) are described in Table 4 for the Voce hardening law. The Voce hardening curve gives a better fitting to the experimental results at 0 degrees from RD. The material constants for Barlat's Yld2000-2d yield function are provided in Table 5 with the eight  Table 6. Cyclical shear mechanical tests were conducted (with the specimen at 0 degrees from RD) for different pre-strains so that a full characterization of the kinematic and/or combined kinematic/isotropic hardening can be conducted effectively for the numerical simulation of the springback of the aluminium panel. The plots for the shear stress vs shear strain for the different pre-strain levels are shown in Figure  2. The excel file "Cyclical_Shear.xls" with the full data for the cyclical shear tests is available on the website of the conference. The rolling direction is specified schematically in Figure 3, with the rolling direction making an angle of 87 with the global x-axis.

SIMULATING THE FORMING OPERATION
The simulation involves three operations: forming, trimming and springback. The drawing occurs continuously in a single action process during which the die moves at 100 mm.s -1 . The CAD geometries for the blank, the lower punch, the upper die and the binder, as well as a mesh for the punch, die and blank holder are provided. The parts/tools are provided in their corresponding orientation and position in the global axis and the forming direction is aligned to the global z-axis, whilst no symmetry plane exists as shown in Figure 4. Participants should not move the tool position in the x-y plane.
The indicative values for the coefficient of friction to be used in the forming operations are: i) 0.08 for Pam-Stamp and LS-DYNA; ii) 0.14 for AutoForm. The lower punch, binder and upper die are illustrated in Figure 4. Only one blank material (3 mm thick) is investigated in this benchmark, properties of which are given in the previous section. The required simulation boundary conditions are given in Table 7.    Table 7  Binder: loading (z-direction), see Table 7 4.1.3 Blank holding force The blank holding force is defined in table 7. It should be applied after the binder has been moved into position.

Trimming
The trim line is illustrated in Figure 5 (the red line/edge) and it is provided in the attached IGES file.

Springback Analysis
The locations of the boundary conditions (BCs) to be defined for springback analysis simulation are depicted in Figure 6. A 3-2-1 locating configuration will be used for part measurement. Points 1 and Point 2 correspond to the centre of the holes shown in Figures 5 and 6.

Point 1 -Pin BC (all dimensions in mm)
The blank is restrained in all global translation directions, X, Y, Z at Point 1 with the coordinates, (-749.3, 75.5, 206.2).

Point 2 -Slot (all dimensions in mm)
A local coordinate system is to be defined and restrained in translation directions, y', z'. The coordinates of the origin (Point 2) of the local coordinate system is (711.0, 83.8, 220.0) and the vector defining the free x' local axis is (30.0, 10.0, 0.1).

Point 3 -Simply Supported (all dimensions in mm)
The blank is restrained in global translation direction, Z at Point 3 with the coordinates, (-68.7, -46.5, 193.4).

Simulation Files
CAD geometry (IGES) files are provided for the die face, binder, blank, punch and the trim line. The trim lines are indicated by lines in the IGES file.

BENCHMARK REPORT
The due date for benchmark submission is listed on the website. All results are to be reported using the benchmark report template which can be downloaded from the conference website.

General Description
 Benchmark participant: name, affiliation, address, email and phone number.


Simulation software: name of the FEM code, general aspects of the code, basic formulations, element/mesh technology, type of elements, number of elements, contact property model and friction formulation.


Simulation hardware: CPU type, CPU clock speed, number of cores per CPU, main memory, operating system, a breakdown of CPU time for the three stages and analysis methods adopted (e.g. explicit or implicit) for each operation.  Delegate's remarks on the results template.

Simulation Results Required
The following information are requested from your simulation:  Die stroke (mm) vs. total punch force (kN) from the simulation during forming, reported for at least every 5 mm of die movement.
 Blank thickness after forming at Sections I, II and III (as shown in Figure 7). The sections are provided as IGES files and the normal vectors of these sections are provided in Table 8, whilst the origin points coincide with the points defined in sections 4.3.1.1, 4.3.1.2 and 4.3.1.3, respectively. Local in-plane axes are defined for each section as described in figures 8, 9 and 10 and Table 9 and, together with the normal vectors from Table 8, they form a right-handed local coordinate system that should be used for the report of the blank thickness after forming.


Profiles of the formed sheet at Sections I, II and III, taken of the punch-side surface for two different instants: (i) end of the forming operation and (ii) after trimming and springback. The profiles should be plotted in graphs with local coordinate system defined by local axes described schematically in figures 8, 9 and 10 and Table 9 and the normal vectors from  As an option, the part after springback can be reported in the form of a geometric (*.stl) file. The committee will report the springback results from correlation with the real part after springback. This will be carried out by aligning the springback result to the measured data by using the same three BC points from section 4.3 -Springback Analysis.       Combined Swift -Hockett-Sherby formulation based on raw tensile test data provided by the organizing committee upon request Main goal was to investigate influence of friction on springback prediction of the part. Pressure dependent friction was described by a power law, i.e. Reference Pressure -4MPa; Pressure Exponent -0.85; Reference friction coefficient -0.12. In this submission, springback analysis was performed with boundary conditions requested in the benchmark briefing.        Figure 6.7. Profile after springback for Section II: BM2_01, BM2_02, BM2_03.                 Figure 6.25. Thickness for Section II: BM2_01, BM2_02, BM2_03.       Figure 6.31. Thickness for Section III: BM2_01, BM2_02, BM2_03.       Figure 6.37. Punch Force: BM2_01, BM2_02, BM2_03.