catia v5 sheet metal design

MODELING, FORMING, AND MODAL ANALYSIS OF SHEET METAL PARTS USING CAE TOOLS

Andy O. Fox, Raghu Echempati, Ph.D., P.E. Kettering University Mechanical Engineering Flint, Michigan, USA

ABSTRACT CAE tools can be used to study the characteristics and reduce the cost of sheet metal parts that are used in products. Using an instrument panel that is used in a car as an example which is made up of sheet metal components the basic process of analyzing the components and assembly to optimize its design is discussed. The paper is mostly educational in the sense that the integrated procedures and analysis presented here can be adapted in a senior level course and at a university that has state-of-the-art CAE tools as discussed in this paper. Several tutorials have been developed that are user-friendly and show how the subsequent analysis can be conducted. To the best of the knowledge of the authors, no such tutorials exist, or are available to students at a university. To start out, solid modeling of the individual sheet metal components using different CAD programs is discussed. Then a discussion on how these solid models can be imported to different CAE programs to be meshed and then subsequently exported to high end solvers like LS-Dyna or MSC Nastran is presented. The integrated analysis that was conducted for this paper was forming analysis of the individual components, followed by modal analysis and gauge optimization of the entire instrument panel assembly. Also, a design of experiments based on Taguchi method is discussed which was done to determine the effects that the input factors have on the results of the forming simulations that were conducted. DOE studies were performed on a couple different components and the results are briefly discussed. It is believed that the contents of this paper serve as an educational tool to the students and the instructors involved in understanding and/or teaching sheet metal forming simulation. Sample tutorials will be presented at the conference meeting. INTRODUCTION CAE tools can be used for many different purposes in engineering to make better products. A situation where CAE tools would become useful is for the analysis of an IP (Instrument Panel) assembly. The solid models of the components of the assembly can easily be created using a software program like Unigraphics or Catia. Sheet metal forming analysis can be performed on each component of the assembly to see if each part is formable. Then modal analysis can be performed on the assembly to determine the frequency of the first few modes. During this process the strain energy density of the components can be determined to find areas in the assembly that can be improved to stiffen the structure. After this a gauge optimization can be performed to minimize the mass of the structure while using the results from the modal analysis of the entire assembly (e.g. frequency of modes) as constraints to maintain its original performance characteristics by not allowing the frequency of the first few modes to drop below the original frequencies of the assembly or predetermined values can also be used for the constraints. Solid modeling of the components can easily be done using software programs like Unigraphics, Catia, or SolidWorks. They make the creation of components easy with their solid modeling capabilities. To create the models basic solid shapes can be made and then special operations like extruding faces or adding radii to corners are used to create the components. CAE tools can be used to study the characteristics of sheet metal parts during the manufacturing process. This analysis saves time and money in the initial development of new products by making it easier to test new designs. Instead of doing actual physical tests on the product to see if the design will work, the product can be analyzed on a computer to produce the same results that would be obtained from the actual physical tests. Time is saved because

many simulations will not take nearly as long as the actual testing of many stamped parts and also money is saved because no money will need to be spent on material (e.g. steel, dies). A DOE (Design of Experiments) can also be performed on a sheet metal stamping process using the forming results from a CAE program like LS-Dyna or HyperForm which can then be input into a data analysis program like the regression analysis tool in Excel. These software programs when used for a DOE are useful when a wide variety of variables need to be tested and the influence of the variables and their interactions need to be determined. One purpose of CAE tools is to analyze parts or assemblies while in service. The main purpose of the IP of a car is to support the steering wheel and it needs to be stiff enough so that the driver does not notice any significant amount of vibration in it. One of the analyses that can be performed using CAE software is a modal analysis. For an IP the analysis can be used to determine the frequency of vibration of the major modes of the steering wheel (e.g. vertical and horizontal). One of the outputs that can be specified for this analysis is the strain energy density. By looking at this for each of the modes the areas with high strain energy density can be located easily to determine where the components should be redesigned to be made stiffer or include more welds.

SOLID MODELING At the beginning of the design process solid models of the components can be made using a CAD software programs like Unigraphics1,2, Catia, or SolidWorks. You can start the components with basic geometrical shapes like rectangles, cones, and cylinders. Boolean operations can be performed on these to add and subtract the geometry of the parts. Faces or outlines can be extruded to create more shapes. Many other operations like creating radii and hollowing out solid geometry to create shells can be performed on the solid geometry to easily create the solid models of the components. Figure 1 below shows a solid model of a sheet metal component that can be used in an IP assembly that was made using Unigraphics. The main reason to use a CAD program to create the solid models is that they are specifically made to create solid models. They usually have specific features that are designed to make it easy to create sheet metal components. They also have the ability to create parameterized models which makes it easy to make modifications to your designs.

FIGURE 1 – SOLID MODEL OF A BRACKET CREATED USING UNIGRAPHICS

FORMING ANALYSIS HyperMesh is one example of a CAE software program that can be easily used to mesh geometry for a stamping simulation and was used in this case. First the solid model geometry is imported into HyperMesh, although HyperMesh can also be used to create its own geometry. Once this is done the midsurface of the sheet metal components are extracted. The mesh, which contains the finite elements used in the simulation, is then created from the midsurfaces. The mesh can be used in different solvers. LS-Dyna and HyperForm One_Step are two examples of solvers that can be used. Material properties and thicknesses of the actual components are assigned to the mesh. Loads,

constraints, prescribed motion and other information are assigned to the FEM (Finite Element Model) and the simulation is run. One of the assumptions for the forming analysis is that the power law of plasticity is valid for the forming of the 3D parts. Two software program used to do stamping simulations were LS-Dyna3 and HyperForm One_Step4. The analysis in HyperForm One_Step is much easier to setup the FEM and the simulation time is much faster as well. This is because in HyperForm One_Step all that is needed is the final shape of the stamped part. This part is then meshed, material properties and thicknesses are assigned to it, the stamping direction is defined, and then the simulation is ran and solved in one step. For LS-Dyna the simulation is solved in increments and takes longer. For LS-Dyna you need to start with the original blank shape and then deform it into the final shape by creating all of the components that play a part in the stamping process (e.g. die, blank, binder, punch). Figure 2 shows the setup that was used in the LS-Dyna analysis.

FIGURE 2 – Die, Blank, Binder, and Punch used in the LS-Dyna simulation

One of the outputs that is generated from HyperForm One_Step is a mesh of the original flat blank shape that would be used to make the final shape of the part. This can be used in the FEM for the LS-Dyna simulation. One advantage of LS-Dyna though is that the actual shape of the original blank can be used in the analysis. Therefore different forming results can be obtained by using different blank shapes or adding things like beads to the die to create more friction to see if there is a better way to form the final part. One of the main results in a forming analysis is the percent thinning of a component after it has been formed. Usually if a part exceeds 20% thinning it can start to split resulting in failure of the component. This is one of the main problems in the forming of sheet metal components. A simulation can be used to determine the percent thinning of components. The lower the percent thinning is the less likely it is that splits will form in the bracket. Figure 3 shows the results of the percent thinning of a bracket that was analyzed using LS-Dyna and HyperForm One_Step.

(a)

(b)

FIGURE 3 - PERCENT THINNING - (a) HyperForm One_Step, (b) LS-Dyna

Since Fig. 3 shows that both software programs predict a maximum thinning of around 17% the bracket should not develop any splits during the stamping process. Wrinkling can also become a problem if it becomes excessive which is shown in dark blue in Fig. 3. Since many simulations can be performed on a stamping process in order to obtain several designs, a DOE (design of experiments) is used in this paper to determine which factors and interactions play the most important role in the forming process and hence effect the thinning. There were four factors of interest in the process, the yield strength of the sheet metal, its thickness, the friction coefficient, and the tonnage of the press. A 2k factorial can be performed easily by setting up a spreadsheet that uses factor values and the thinning (response) of each complete trial as inputs and then the results (effects and interactions) are determined by using contrasts 5. This can also easily be done using the regression data analysis tool in Excel. The results of the DOE can be used to find the optimum settings of the four factors to minimize the percent thinning (the response) of one of the corners of the part which was shown in red in Fig. 3. Table 1 shows the factor values that were used in the analysis. Table 1 – Factor level used in the DOE Studies Factor Level

Yield Strength (Mpa)

Thickness (mm)

Friction

Tonnage (Tons)

-1 +1

186 500

0.8 3.2

0.05 0.70

20 80

The results of the DOE experiments showed that the yield strength was insignificant and could be disregarded. It was also found that the thickness and friction should be kept as low as possible to reduce the chance of splits occurring. An additional study is being done on another component to determine if the factors have the same influence on the formability of the components.

MODAL ANALYSIS Modal analysis6 on parts and assemblies can be performed with CAE software. An instrument panel is used to support the steering wheel in a car and needs to be stiff enough to keep its frequency of vibration high to reduce the driver’s awareness of the vibration in the steering wheel during operation. The first two modes of an IP are shown in Fig. 5 below along with their corresponding frequencies.

Mode 2 42.2 Hz

Mode 1 35.3 Hz

FIGURE 5 – INSTRUMENT PANEL ASSEMBLY

The bracket shown in Fig. 3 is a part of the IP assembly and is used to connect the steering column to the IP assembly. The forming analysis that was shown in Fig. 3 on that bracket is a separate analysis from the modal analysis that is shown in Fig. 5. The CAD data shown in Fig. 1 and Fig. 5 was obtained from a finite element model archive on the internet7. For the instrument panel all of the components are made from sheet metal except for the steering wheel. Therefore the entire model is made up of shell elements except for a few solid elements in the steering wheel. Once the mesh is created the components can be connected using 1D rigid beams to simulate the welds. Finally the constraints are applied to the attachment points which hold the IP in place. The FEM is then exported to some solver to perform the modal analysis. HyperMesh has a built in solver that can be used called OptiStruct that was also used. The IP shown had frequencies of the steering wheel in the range of 35-45 Hz. If the frequency of vibration of the steering wheel falls below this range it is likely that the driver of the car will notice and it will become uncomfortable or distracting after a while so the IP needs to be stiff enough to keep its frequency high. One output that can be exported during the modal analysis simulation is the strain energy density. This is an output that identifies the components and attachment locations that are deforming the most for each mode. The components strain energy density can be viewed for each mode shape and corresponding frequency. Components that have a high strain energy density can be made thicker, welded together more, or redesigned to stiffen the assembly. Viewing the plot of the strain energy density is useful in finding areas that can easily be improved that otherwise are not easy to find to stiffen a structure. Figure 6 shows the strain energy density for one of the areas around the steering column on the IP.

FIGURE 6 – STRAIN ENERGY DENSITY

The red areas shown in Fig. 6 have high strain energy density for the mode shown. This can be reduced by either making the parts thicker, redesigning the components, or adding more weld around components interfaces. A longer bead of weld was added to the bracket interface that had a high strain energy density shown in Fig. 7 to increase the stiffness of the assembly.

ADDED WELD FIGURE 7 – EXTRA WELD ADDED TO THE COMPONENT INTERFACE

By making the weld longer the frequency of mode 1 shown in Fig. 5 increased from 34.9 Hz to 35.3 Hz and mode 2 increased from 41.8 Hz to 42.2 Hz. REFERENCES [1] UG NX 5.0 Cast Library [2] http://training.ugs.com [3] LS-DYNA 971 User’s manual, 2007. [4] HyperWorks User’s manual, 2007. [5] Montgomery, Douglas C. (2005) “Design and Analysis of Experiments” – Sixth Edition, John Wiley and Sons, INC., Hoboken, NJ. pgs. 204-254 [6] Palm III, William J. (2007) “Mechanical Vibration” John Wiley and Sons, INC., Hoboken, NJ [7] http://www.ncac.gwu.edu (2009)