Skip navigation
Electronics Industry Search
Advertisement
Email
Print
Reprints/License
RSS

A Measure of Confidence

New method brings consistency to evaluating fatigue life of critical welds on cars, ships, and other structures

Lawrence D. Maloney -- Design News, April 3, 2005

Until recently, engineers responsible for specifying welds needed for such applications as auto frames and chassis typically included one common ingredient in their calculations: a "fudge factor."

The reason? They based their calculations primarily on finite element analysis of stress points or "hot spots" in areas to be welded--a method that can yield widely different results, depending on the computer model and the size of the mesh. To be on the safe side, engineers tended to "overengineer" the welds or the thickness of materials to be welded in order to compensate for the uncertainties of their fatigue modeling methods. In addition, they ordered time-consuming and expensive tests of welded structures to back up their calculations.

Now, however, industries ranging from autos to offshore exploration to shipbuilding can at last gain a healthy dose of confidence in their fatigue predictions, thanks to a new modeling method developed by Pingsha Dong and his colleagues at Battelle's Center for Welded Structures Research in Columbus, OH.

"For the past 25 years, experts in the field have been trying to address the inadequacies in stress analysis for fatigue design of welded structures so that companies would not have to compensate for poor data," says Dong, who holds a Ph.D. in computational mechanics from the University of Michigan. "Eventually, the industry and academia gave up. We did not."

A "Mesh-Insensitive" Method

Among past approaches for predicting fatigue life, engineers would typically use computer-generated meshes to analyze stresses outside a welded area—then extrapolate the results to the weld toe. Dong's approach also employs finite element analysis, but it focuses instead on nodal forces in the structure and weld. Using these nodal forces, the Battelle researchers then developed a mathematical formula that determines the stresses. What's more, the Battelle model achieves accurate, consistent results, regardless of the size of the mesh used.

To the general public, Dong's new methodology may seem arcane. Yet manufacturing experts believe that it can revolutionize industry, which spends billions each year on testing, modeling, and predicting the life of welded structures. Tarsem Jutla, a respected expert on welding design at Caterpillar, calls the Battelle breakthrough "incredibly significant" and compares its impact on engineering to that of mapping the human genome in the scientific world.

Adds Hari Agrawal, a senior technical specialist at Ford: "This is a robust, consistent methodology. No matter who uses it, you get the same answer."

Ford and Caterpillar are among an initial group of user companies that have been testing Dong's patent-pending methodology, which Battelle calls Verity™. Agrawal was so impressed with Dong's work that he asked his assistance in 2001 to integrate this "mesh-insensitive" method into a new proprietary fatigue analysis tool called FLOW (Fatigue Life of Welds).

To validate the Battelle methodology, Ford did extensive cyclic-loading tests on both aluminum and steel samples. Data from such tests helped Ford formulate a master S-N curve, which engineers use to predict fatigue life. Says Agrawal: "No matter what load you apply and no matter what geometry you have, the mesh-insensitive model gives you a very accurate way of determining the fatigue life of a weld."

Battelle's Pingsha Dong, center, developed a ground-breaking methodology for predicting fatigue life. His Battelle colleagues include J.K. Hong (left) and Zhenning Cao.

Agrawal notes that Ford's FLOW tool not only helps determine the optimum size of a weld, but where to start and stop the weld. Ford operations in North America, Europe, and Brazil are already using the tool, which is said to be saving the company millions of dollars on materials and testing.

"We all know the competitive nature of the auto industry, and the great pressures to reduce the design cycle," Dong says. "The repetitive tests needed with previous fatigue modeling methods add to costs and lengthen development time." He believes that the greater accuracy resulting from Battelle technique can reduce the number of tests by least 50 percent and may also allow engineers to use lighter materials, a key goal in many auto and aerospace applications.

Spreading the Word

Despite support by Ford, Caterpillar, Chevron Texaco, and other blue chip companies, the Battelle methodology has had to overcome strong skepticism by industry experts. A big turning point came, however, in 2003 when the Society of Automotive Engineers (SAE) held a competition to predict the fatigue life of a rectangular hollow section joint. This "Weld Challenge," sponsored by SAE's Fatigue Design & Evaluation Committee, attracted eight of the top minds in structural analysis from industry and academia. Each gave predictions for the fatigue life of an aluminum part, which consisted of a 4 × 4 inches section and a 2 × 6 inches section, joined together by a 5/16-inch weld.

Comparing these predictions with actual lab tests on eight samples under a 4,000-lb load, the SAE found Battelle's mesh-insensitive approach to be the only accurate predictor of fatigue life—regardless of the mesh size employed. The Battelle method not only estimated failure at about 75,000 cycles but also pinpointed the location of the crack in the samples under test.

Stress analysis of fatigue life of a tubular component, typically used in auto applications, shows how Dong's "mesh insensitive" approach (Verity) delivers uniform results, despite the mesh size. In contrast, conventional methods yield highly variable results.

Dong's work got even more recognition in 2004 when he received the Henry Ford II Distinguished Award for Excellence in Automotive Engineering at the SAE World Congress in Detroit.

Meanwhile, interest from industry has been growing. Some 15 major companies are now testing the methodology in applications that include: automotive, agriculture and construction equipment, heavy trucks, oil rigs and refineries, pressure vessels, and shipbuilding. The newest Joint Industry Partnership (JIP) also includes two regulatory organizations.

"We are getting more and more inquiries about Verity, and the JIP approach is a great way for industry to get exposure to our methodology," says David Stroud, Battelle's director of Technology Development.


Web Resources
For more on new welding methods:

»Battelle Labs Verity method for fatigue analysis
http://rbi.ims.ca/4407-541
»SAE Fatigue Design & Evaluation Committee
http://rbi.ims.ca/4407-542
»Paper on "mesh insensitive" structural stress method
http://rbi.ims.ca/4407-543
Advertisement

Sponsored Content

Technology Marketplace

Email
Print
Reprints/License
RSS
Find a supplier on oemsuppliersearch.com

Talkback
Advertisement
Reed Business Information Resource Center

Featured Company


Advertisement

Design News Partner Zones

Light Matters: The Unsung Heroes of Modern Health Care
First, let's define "no-compromise." In an ideal configuration, this lamp would use a high-brightness LED (HBLED) that is built into a small, integrated package and is able to produce a large quantity of focused light, operate with a high level of reliability and generate no audible noise. Is this difficult? Yes, but it is possible.
Read More


Design Engineers' Portal for Sensing and Machine Safety
Whatever industry you're in, or whatever product you manufacture, the right sensors to automate your plant, and to improve your overall efficiency, quality and safety are a must. You'll find Banner Engineering to be an amazing resource of products, training and people with expertise.


Test & Measurement World Machine Vision & Inspection Report
Topics include machine-vision software, Power over Camera Link, thermal imaging and frame grabbers. Read More

Design News Partner Zone Directory »

Please visit these other Reed Business sites