What All Engineers Need to Know Before Using Finite Element Analysis

Bob Williams
Development Manager
Algor, Inc.
Pittsburgh, PA

Increasingly, engineers in every industry are choosing to integrate finite element analysis (FEA) into their design cycle in order to ensure that their designs are safe, efficient and cost-effective and to more quickly get products to market. However, analysis is not as simple as putting a CAD model into any FEA package. For each scenario, the engineer needs to determine the design goals and criteria, ensure that the chosen FEA package contains the necessary features and capabilities and learn techniques to properly interpret results.

Today, engineers have more types of analysis software to choose from than ever before. For many years, engineers were limited to using linear static stress analysis to predict the displacements and stresses that would result when a model was subjected to the given loads and boundary conditions. More recently, finite element packages have been extended to include nonlinear static stress, dynamic stress (vibration), fluid flow, heat transfer, electrostatic and FEA-based stress and motion analysis capabilities. Frequently, these capabilities are combined to perform analyses that consider multiple physical phenomena, and are tightly integrated within a CAD environment through an intuitive, easy-to-use interface.

This article will briefly discuss some FEA basics and then outline the analysis, modeling and results interpretation considerations that engineers need to think through when deciding to use FEA.

FEA Basics

FEA is a numerical approximation that is used to solve engineering problems, based on the engineering handbook formulae and hand calculations that engineers have traditionally used to validate their work. A finite element model is a discrete representation of the continuous, physical part that is being analyzed. This discrete representation is created using nodes and elements. Nodes are connected together to form elements. The nodes are the discrete points on the physical part where the analysis predicts the response of the part due to applied loading. This response is defined in terms of nodal degrees of freedom (DOF). For stress analysis, up to six degrees of freedom are possible at each node (three components of translation and three components of rotation). Depending on the element type selected (e.g., beam, plate, 2-D and 3-D elements, etc.), the number of required degrees of freedom at each node is determined. 

The grid of connecting elements at common nodes comprises the mesh. When adjacent elements share nodes, the displacement field is continuous across the shared element boundary and loads can be transferred between the elements. In order for forces or other loads (such as temperatures) to transfer between parts in an assembly, the nodes on the different parts have to be shared where the parts meet.

Analysis Considerations

In any analysis scenario, an engineer first needs to decide the significant physical phenomena to which the design will be exposed in order to decide what type of analysis should be performed. By far, the most common concern for engineers involves maximizing the part’s durability, which is its ability to withstand the mechanical stresses created during use in the real world. 

Determining whether the design will be subject to static or dynamic conditions is the first step in determining how to go about any type of stress analysis. For years, engineers have used linear and nonlinear static FEA software to calculate stresses at a single instance in time. This method can provide acceptable results if the problem is purely static and the design in question does not experience impact, motion or changes in the applied loads over time. However, if the problem is not purely static, even an FEA expert may not be able to make valid assumptions about the loads and boundary conditions necessary to use linear or nonlinear static FEA to properly validate their design. The best alternative is to simulate the scenario in which the design will be used as accurately as possible in order to avoid the inaccuracies due to assuming forces inherent in traditional linear or nonlinear static FEA.

There are several common methods for simulating events. A growing trend in FEA is the simulation of large-scale motion using finite element models. Simulating large-scale motion is often critical to be able to accurately replicate the real-world behavior of many mechanisms and to determine how components will perform under conditions of impact, contact or other loads of concern. Engineers benefit from simulating the large-scale motion of finite element models because these models can accurately determine the resultant deformation and stresses without the need to make assumptions about the forces at work in the mechanical event. Ideally, the chosen method should provide the flexibility to consider either linear or nonlinear material behavior and offer automatic contact features. The available methods can vary widely in terms of ease-of-use, accuracy and functionality. Following are the three common methods available:

• Motion Load Transfer requires engineers to use a separate kinematics package in order to obtain approximate loads for FEA. Therefore, it does not actually simulate the motion of the finite element model, but instead produces loading from rigid-body motion with assumed stiffnesses. This loading is then used to perform a static stress analysis. 
• Explicit Timestep Method determines a solution by “marching” along in time, extrapolating from the solution of the prior timestep. This method is fast for a given timestep, but requires many small timesteps to complete large-scale motion problems, often making the processing time impractical. In addition, detailed knowledge of the finite element model may be required in order to achieve reliable results. 
• FEA-based stress and motion analysis uses an automatic timestepping scheme that incorporates an implicit timestep method to yield a highly efficient and accurate solution. This type of software simultaneously produces motion, deformation and stresses in a single “What-You-See-Is-What-You-Get” process. This method is very stable and only requires that analysts specify the duration of the event and a solution capture rate in order to obtain an accurate solution. In addition, the automatic timestepping scheme enables the analysis to run at larger timesteps for periods of relative inactivity, such as a constant acceleration or deceleration, and then automatically reduces the timestep size to capture periods of critical activity, like surface-to-surface contact, local buckling or impact. 

In addition to considering a part’s ability to withstand the mechanical stresses created during use in the real world, some FEA software enables engineers to predict the effects of extreme temperatures or temperature changes (heat transfer analysis), the flow of fluids through and around objects (fluid flow analysis) and voltage distributions over the surface or throughout the volume of an object (electrostatic analysis). The net effect of more than one of these cases may need to be considered as well. 

It should be said that engineers who want to perform FEA within a CAD environment should make sure that the chosen CAD/FEA solution actually offers all of the necessary FEA and motion simulation capabilities within CAD, using the same interface. Many FEA vendors have a full range of FEA capabilities but offer only a subset of them within CAD, thus requiring the use of a different interface for some of the FEA capabilities. 

Modeling Considerations

After determining the type of analysis that is required, the engineer must produce a finite element model with appropriate analysis parameters, such as loads, constraints and a suitable mesh size. In many cases, an engineer has a CAD model already available. The three available methods of CAD/FEA interoperability can vary widely in terms of ease-of-use, accuracy and functionality:

• The CAD universal file format method requires the engineer to export the CAD solid model to a neutral file format, such as IGES, ACIS or Parasolid, and then import the neutral file into the FEA system for setup and analysis. Although this method usually enables engineers to take full advantage of an FEA vendor’s capabilities, most engineers find other, more seamless methods easier to use. In addition, translation to a universal file can result in the loss of CAD geometry data within the FEA package.
• With the “one window” CAD/FEA method, an FEA vendor offers seamless and accurate capture of CAD data to the FEA system and builds common and often times limited analysis capabilities into a particular CAD solid modeler. This method typically limits things like element types available and meshing and analysis options in the name of ease-of-use. This CAD/FEA interoperability method usually requires transitional mesh enhancement tools in order to obtain a mesh that will produce accurate FEA results. This method usually requires the engineer to purchase other software and learn another interface to handle analysis beyond the basic tools provided and to interface with other CAD solid modelers as may be necessary in a multi-CAD environment.
• The “one window away” CAD/FEA method offers the same seamless and accurate capture of CAD data to the FEA system whether the CAD and FEA packages are on the same or different computers, and utilizes the same interface for working within multiple CAD packages typical in a multidivisional or consulting environment. The autonomy that the FEA package retains with this method also enables the FEA vendor to provide more element types and meshing and analysis options without requiring the need for other software or sacrificing accuracy and ease of use. Engineers learn just one interface for all analysis needs. Engineers can look for a “one window away” CAD/FEA solution to provide options including but not limited to:
  • the flexibility to define 2-D, truss, beam, plate/shell and other element types that can offer significant processing speed benefits and the ability to combine them with solid elements; 
  • automatic solid meshing with a wide variety of solid element types such as brick, tetrahedra and hybrid, which consist of bricks on the model surface with tetrahedra inside (a choice many engineers prefer because hybrid models combine the speed of tetrahedra with the accuracy of bricks);
  • controls for the element quality of the automatic solid mesh and built-in, default aspect ratio checking to control the accuracy and precision of the solid elements; 
  • control over CAD feature suppression during the meshing process; 
  • a robust automatic surface meshing scheme, such as “surface-inward meshing” technology that puts the best-shaped elements on the surface where stresses tend to be highest and does not require transitional mesh enhancement tools for repairing low-quality meshes that may result from an imposed-grid meshing scheme; 
  • automatic midplane meshing with the capability to heal the gaps inherent with the midplane extraction of a thin, solid part to plate/shell elements to take advantage of the significant processing speed benefits available with plate/shell elements;
  • automatic features for simulating different types of contact among several separate objects, including surface-to-surface contact, static and sliding friction and impact among several independently moving objects or between moving objects and a stationary object; 
  • kinematic elements to replace flexible elements in parts of solid models where stress concerns are low in order to reduce analysis run times;
  • special engineering elements such as actuator elements for simulating the axial expansion and retraction of hydraulic cylinders, hydrodynamic elements for fluid-structure analyses, coupling elements to limit the extension of mechanical linkages and dashpot elements to damp vibrations;
  • Windows-style, right-click functionality for all analysis types for applying, modifying and deleting loads, constraints and FEA properties; and 
  • the capability to set up load and constraint sets that can be activated in different combinations over a series of analyses with different design scenarios for an efficient work flow.

Results Interpretation Considerations

Once the results from an analysis are obtained, the engineer asks the question, “How do I know my answers are correct?” Therefore, it is important that the FEA software provides tools to aid in the verification or validation of the analysis of any design. Ideally, these tools would not only include displacement and stress contours but also precision contours that provide qualitative and quantitative indications of the degree to which a model complies with the assumptions of the finite element theory. 

Often, the analysis and modeling choices that an engineer makes will determine how easy it is to interpret results. For example, if a linear static stress analysis was performed, only contours at a single instant in time will be available. The stresses displayed will need to be interpreted in some way, such as comparing the values to the yield stress of the material used in the analysis. In addition, the engineer has assumed that the one instant analyzed in the linear static stress analysis represents the worst case scenario. If, on the other hand, a FEA-based stress and motion analysis was performed on a solid assembly, interpreting the results will be much easier because the model can be seen to move, flex, bend and even break over time. These time-dependent results can be recorded in an animation file, such as the Windows .avi format. In addition to animation capabilities, engineers should expect the FEA software to have a fast, easy-to-use visualization tool for reviewing and presenting results for all analysis types with other integrated presentation options, such as automatic options to generate image files of result contours and plots, VRML files and customizable HTML reports. 

These tools can be used to prepare a presentation of results to other engineers and even non-engineers, like managers and clients. Carefully considering the analysis, modeling and results interpretation issues discussed here will help engineers to present their designs with more confidence in the validity of their FEA results.