Using FEA in your modeling design
If you are considering having your engineering problem computer modeled by the finite element method, it is good to have a general understanding of the method.
Many of the laws of physics may be represented by partial differential equations (PDE). Some can be solved analytically for simple shapes, such as parallelepipeds, whereas others cannot be solved in the closed form, but must be solved numerically. Regardless, there are almost no useful shapes representative of engineered parts for which the PDEs can be solved.
Finite element analysis (FEA) takes the PDEs and breaks the part into many small pieces (finite elements), for which the PDEs may be solved, at least numerically. The PDEs may represent the stress, strain, heat and mass transfer, radiation or similar, for a part or assembly of parts. Basic concepts of continuity, energy and mass conservation, applied at the boundaries of the array of elements, allow for an accurate model of the actual system. These can also be solved simultaneously in what is sometimes referred to as a “multi-physics” model.
One can appreciate that a very large system of equations results from these basic equations across thousands of “finite elements” representing the work piece. So, the engineering problem becomes a massive system of equations simultaneously solved in the software. The unsolvable PDEs are reduced to matrix equations. Steady state or transient solutions can then be obtained.
Transient solutions can also be obtained, albeit with much more lengthy computations. At Glew Engineering, we have run transient radiation and heat transfer models that have run for weeks, to simulate a one hour event. Also, mechanical event simulation (MES), event horizons—or situations that change over time—can be modeled. These can also be solved in what is sometimes referred to as an “event simulation.”
Finite Element Analysis is an example of world class engineering methodology. Some people and organizations are firm believers, while some are yet to be convinced. Success depends on the software, the modeler, and knowledge of the material parameters.
At Glew Engineering, we believe that it is best to start by modeling a prototype or known system that is similar to the one being engineered. Then, the FEA model can be tested against physical systems and its veracity measured. Afterward the model can be modified to represent a new design with greater accuracy and confidence.
There are a number of advantages in using the FEA method to model your design. In some cases, the computer can model experiments that are very expensive to run in more economical modes. Often times, it is faster to work through models than wait for parts and prototypes. One advantage that comes into play with these design iterations is the ability to makes slight changes to the model to test the advantage of one design over another. Another advantage is the ability that it allows your organization to keep the laboratory and test facilities available for other work until the model results indicate that the design could work.
Often times, when a design does not perform as desired, the results often times are disappointed customers and repair costs abound in unpleasant quantity. Sometimes, the multitude of field conditions are difficult to replicate in the engineering laboratory, but can be simulated by the FEA method. Engaging a third party to perform an FEA on your situation can help to go a long way to diffusing customer dissatisfaction and show that you’re are serious about solving the problem. Smaller companies can assuage concerns of much larger customers who routinely use FEA, by utilizing consulting groups to perform the FEA for them.
Practical questions concerning the overall project determine how the results will “stick.”
Has the actual hardware design project just started, or is it down the road some weeks or months? Is the design project encountering difficulties? Has a field failure occurred? Is the finite element model a suggestion of a customer, of management, or your own best practices methodology? What level of accuracy is required from the model? Answering some of these basic questions is important to understanding the required output and, the timeframe—and to locating, and funding the work.
The most efficient, highest quality output will result from an experienced modeler working with a software package that he knows thoroughly. Even then, the FEA software is not for the faint of heart, due to its advanced and idiosyncratic nature. Before the actual model can be constructed, CAD files of the design must be shared with the modeler, and a description of the design materials planned for the design must be shared. While the modeler can access databases to find relevant material constants, some specialized materials may have unique properties that are key to the outcome of the model.
Frequently, the output of modeling exercises is to determine sensitivity of output parameters to material properties of key components. This can lead to manufacturing constraints on these components, or to a re-design which reduces sensitivity to the components.
Modern Design Strategy
For a complex mechanical-thermal system design, there is no substitute for Finite Element Analysis. Within the design cycle, the sooner it is employed, the greater the benefits. Many top tier companies require that computer modeling by FEA be performed on all new parts. Some regulatory agencies now prefer FEA over traditional engineering calculations based on cook book equations.