Analysis During Design Reduces Cost and Time
Finite Element Analysis has been around since the 1950s but still has not acquired the complete trust of the design community. By understanding and using FEA properly, however, it can accomplish a great deal - solve for stress in a part, reduce prototyping, shave weight and much more.
By Suchit Jain
Vice President, Analysis Software, SolidWorks Corp.
Despite its growing popularity, analysis software remains a bit of a mystery in much of the design and engineering communities. Increases in desktop processing power and steady improvements in speed, ease of use, and the meshing of complex assemblies have made analysis software accessible to virtually anyone in the design process. Yet it faces a lingering skepticism, in part because most designers and even a few engineers aren’t exactly sure what analysis applications do with design data, or how they arrive at their conclusions. One minute their design is on the screen as a three-dimensional rendering, and in the next it’s festooned with colors that purport to reveal its strengths and weaknesses. Why should they trust it?
Today’s analysis applications are mature, reliable products built on 50-year-old engineering principles and 30 years of software development experience that stretches back to the FORTRAN and mainframe days. Used properly, they yield trustworthy results that are already driving efficiencies and cost savings in industries ranging from consumer goods to aerospace. Companies that use analysis software throughout the design process typically have fewer costly late-stage design errors and need fewer physical prototypes. For the price of learning what’s in the “black box” of analysis software, companies stand to reap considerable gains.
A reliable legacy
Most analysis software on the market today employs the finite element analysis (FEA) method. The FEA process consists of subdividing all systems into individual components or “elements” whose behavior is easily understood and then reconstructing the original system from these components. This is a natural way performing analyses in engineering and even in other analytical fields, such as economics. The approach of using discrete components to solve full systems has been used in structural mechanics since early 1940s, and the term FEA was first used in 1950s. Aerospace engineers adopted FEA in the 1960s to analyze aircraft designs.
COSMOSWorks 2003 allows users to query or probe results on any
section of a model, allowing them to study a variation of results along
the section plane. (Click on image to enlarge)
FEA moved from manual calculations in the ‘50s to FORTRAN applications running on mainframes and Cray supercomputers during the 1960s and ‘70s. Analysts ran design data through the applications, which yielded numeric results the analysts would interpret for the designers. By the ‘80s, analysis software came out of the back room to run on high-powered workstations that displayed 3D images rather than numeric data. Analysis was, however, still mainly the domain of specialized analysts who did nothing else except analyze designs.
Picture of the Mars Rover in which
COSMOSWorks was used for
detailed stress analysis and
mass minimization of the rover's
Though it was an improvement over the mainframe days, it was still an awkward system. Analysts would suggest modifications to the design, but designers would not always know how to apply those modifications.
The picture improved dramatically in the mid ‘90s. As desktop CAD became a design industry staple, software vendors responded by integrating analysis with CAD software so users could model and test designs in the same environment. That eliminated re-creating a design in an analysis application, which was time consuming and often caused errors.
Analysis software at work
Analysis software is finding roles at every juncture of the design process from concept to finished product, in industries ranging from consumer products like kids’ scooters to aerospace components.
Under the Hood of Analysis Software
From the mainframe era to today, FEA's heart has been the engineering equation based on Hooke's law F=Ku, or force equals stiffness multiplied by distortion. Force is the variable that determines distortion, so applying different forces shows users how their designs will respond under various conditions. Over the years, vendors have developed software that simulates real-world forces, such as torque and centrifugal force, to yield more accurate results.
FEA applications build matrices of the Hooke's Law equations according to a design’s parameters. As the name suggests, FEA applications select a finite number of elements in a design to analyze and make calculated assumptions about the rest of the design based on established knowledge about matrices. For example, a control arm on a car suspension is one continuous shape. An analysis application will test the control arm by dividing the geometry into ‘elements,’ analyzing them, then simulating what happens between the elements. The application solves the matrices of Hooke's equations and displays the results as color-coded 3D images, red usually denoting an area of failure, and blue denoting areas that maintain their integrity under the load applied.
Regardless of its cost or complexity, that’s what every FEA application does. Interfaces and algorithms have improved over the years, but under the hood most FEA applications differ only in their method for solving F=Ku . Mathematicians and engineers have written hundreds of papers on efficiently solving the matrix equations: F=Ku, from “Gauss elimination” to “Conjugate gradient” methods. Today most FEA software employs a combination of these techniques to solve various FEA problems.
The other factor that determines how well analysis software works has little to do with the software itself, but rather who’s using it. Analysis software is prey to the same “garbage in, garbage out” problem as any application. Users who do not know how to apply the proper force won’t get good results. Beyond that, if they do not understand fundamental concepts like yield they will not be able to interpret results accurately. In those cases, it doesn’t matter how well the software works. However, vendors are steadily improving their products’ ease of use to help young engineers and designers without engineering educations to better use and understand analysis applications.
Companies use it early in the design process to reduce prototyping costs, and later to optimize designs by eliminating unnecessary materials while maintaining structural integrity.
Pasadena, Calif.-based Alliance Spacesystems Inc. (ASI) used SolidWorks’ COSMOSWorks analysis software in its design of a robotic arm for the NASA probe now on its way to Mars. The company used COSMOSWorks to optimize its design to meet NASA’s strict weight-rigidity guidelines.
The robotic arms weigh only 7.7 pounds and are four feet long with five degrees of freedom. Meeting weight guidelines was paramount because the probe is designed to bounce on inflated pads when it lands on the hard surface of Mars.
If it’s too heavy and lands with too much velocity, the probe could be damaged or destroyed. In addition to being light, however, the robotic arm also had to meet stiffness requirements so it can operate scientific instruments, such as scrapers to take rock samples, with precision. ASI’s engineers use optimization software for detailed stress analysis and mass minimization of the arm’s parts. The engineers vary the parts’ geometry to reduce mass while controlling the maximum allowable stress or deflection.
“Using some of the optimization features, we can typically take 15 to 20 percent additional mass out,” said ASI Vice President of Engineering Jim Staats. “That doesn’t sound like much, but if you wanted to reduce your car 20 percent, you’d have to take the whole engine and transmission out, so it’s a big chunk.”
Like ASI, Sub-Atlantic, an Aberdeen, Scotland-based maker of remote controlled submersible vehicles, uses analysis software for design optimization. Sub-Atlantic also uses it for new product design, as in the recent production of a lightweight, low-cost submersible to supplement its high-end products, which are used in undersea oil exploration and marine research. Sub-Atlantic wanted its new “Navajo” to weigh no more than 70 pounds so it could be launched by hand from a small boat with no special equipment. Sub-Atlantic used analysis software aggressively to cut down on the “Navajo” submersible’s weight and test the durability of its control housing under various depths of ocean water.
“Once engineers produce designs, the analysis software tells them how they will behave in real-world conditions,” said Colin Millum, Sub-Atlantic’s director of mechanical engineering and design. "The subs are designed to be weight-neutral in water so they can descend or rise according to the surface operator’s instructions. Analysis software lets us anticipate where high stresses occur and optimize the design by adding or removing buoyancy elements so that, above water, the vessels are as light as possible.”
The company is prototyping the Navajo because it is the company’s first product built from plastic instead of aluminum, but Millum said that is the exception at Sub-Atlantic, where analysis software has replaced most physical prototyping.
“We prototyped our latest project because we were working with a completely new material, but typically we’re confident enough of the results we get from the analysis software to go straight from design to production,” Millum said.
Out of the black box, into the bank
Today’s analysis products range from high-end workstation-based software aimed strictly at engineers to desktop software accessible by anyone in the design cycle. Under the covers, however, they draw on the same 40 years of software development experience and perform the same basic procedure, albeit with variations in methods. Properly applied, they can help every phase of the design process become faster and more cost effective. Analysis software has a direct, tangible benefit on companies’ bottom line performance. No longer confined to the rarefied realms of specially trained analysts, it can now sit on desktops anywhere in the organization, delivering reliable results that can shave time off design cycles and reduce physical prototyping costs.
About the Author
Suchit Jain started his technology career in 1994, supporting then-new COSMOS analysis software as an entry-level engineer with Structural Research and Analysis Corp. (SRAC). Jain rose through the ranks to vice president of marketing, helping COSMOS grow into one of the most widely used analysis products in the computer-aided design (CAD) industry. Today, he is responsible for marketing communications, technical sales and support, analyst relations, and overall product management for COSMOS, which is now owned by SolidWorks.
Jain holds a bachelor’s degree in civil engineering from the Indian Institute of Technology in Bombay, and a master’s degree in structural mechanics from the University of Southern California. He has presented at the National Design Show, AUTOFACT, and several SolidWorks user conferences. His articles and columns have been published in CAD industry journals such as Machine Design and Design News.