Calculating Risk, Increasing Reliability
When you take a plane trip, drive across a bridge or ride the commuter train to work, you trust that those structures and systems are safe. Likewise, pilots flying combat missions depend on their planes and astronauts hurtling into space depend on the rockets propelling them.
Sankaran Mahadevan, professor of civil and environmental engineering, works on ways to increase the reliability and decrease the risks of those and other complex structures and systems. His research regarding railroad wheels, spacecraft, dams, bridges and even nuclear waste dumps has the potential to save human lives and millions of dollars.
In the past, engineers had to test multiple samples of something to see how many failed. They would then add more material and components than needed to increase the reliability of their designs. That approach doesn’t work with today’s large structures and complex systems.
“Skyscrapers and bridges can’t be put through full-scale testing as can small mechanical and electrical devices,” Mahadevan says. “You can’t test the reliability of large systems like space shuttles and warplanes by waiting to see what fails. No matter what the system, we have to be concerned about how reliable it is.”
Mahadevan and his colleagues in the Structural Reliability Research Group are developing computer models that can predict with a high degree of confidence whether a system will fail, when failure is likely to occur and how to prevent such failure.
Computer Programs Take the Risk
Dr. Maha, as he is known to his students, also directs the Vanderbilt Risk and Reliability Engineering and Management doctoral program, the largest and most prestigious of its kind in the world. “It began in 2001 as an Integrative Graduate Education and Research Traineeship (IGERT) grant from the National Science Foundation,” he explains. “That focused on studying and developing multidisciplinary computational and experimental methods for assessing and managing risk and reliability.”
Today, the program is self-sustaining, with governmental and private partners that include the Transportation Technology Center, Sandia National Laboratories, Federal Aviation Administration, NASA, U.S. Air Force, U.S. Department of Energy, Boeing Co., Bell Helicopter Textron and Union Pacific. Those organizations and others provide about $1 million in funding each year for research projects on structural reliability and durability, optimization and decision making under uncertainty, structural health monitoring, and reliability and risk engineering and management.
“You can’t test the reliability of large systems like space shuttles by waiting to see what fails.”
~ Sankaran Mahadevan
Mahadevan is currently applying his expertise to NASA spacecraft. His team is working on calculating risk and uncertainty in such large systems by incorporating multiple disciplines like structures, aerodynamics, propulsion, mass and geometry into the computer programs. An acceptable risk for spacecraft is typically about one in 10,000.
“The question then becomes how good are our models?” Mahadevan says, noting that there are many assumptions and very little data on which to base such predictions. “It has been shown that different models will yield very different predictions. So we’re developing rigorous methods to verify and validate our computer models.”
“We are also monitoring the health of large systems by placing sensors on the vehicle that will detect real-time damage, diagnose the problem, and offer a prognosis as to how long the vehicle can be used before repair or grounding,” he continues. “One research question is to determine how many sensors are needed and where they should be installed.” Another concerns the reliability of the sensor itself.
Planes, Trains and Nuclear Waste
Mahadevan’s reliability methods can be used in the design, manufacture, operation and maintenance of equipment and systems in many fields. Engineers and scientists call that life-cycle risk management. His research for the Federal Highway Administration, for example, identified which 2,000 bridges throughout the country should carry advanced structural health monitoring instrumentation. The team also developed a cost-effective way to inspect train wheels that demonstrated a 400 percent return on investment for partner Union Pacific.
Current research includes a U.S. Air Force Research Laboratory project to develop rapid diagnosis and prognosis methods for warplanes. The group is also working on applying risk and reliability management to large complex systems like homeland security and transportation. Through a consortium of universities known as the CRESP project, the Department of Energy funds an effort to model the durability and uncertainty of concrete storage facilities for low-level nuclear waste. The team also has received a five-year, $1.2 million award from the FAA to develop advanced methods to predict fatigue and fracture, and the related uncertainty, in helicopter rotor components.
Mahadevan joined the Vanderbilt engineering faculty in 1988, after earning his Ph.D. from Georgia Tech. His work has been recognized with numerous awards, including the Distinguished Probabilistic Methods Educator Award from the Society of Automotive Engineers and the Outstanding Professional Service Award from the American Society of Civil Engineers. In 2006, he received Vanderbilt’s Joe B. Wyatt Distinguished University Professor Award for his achievements in developing significant new knowledge from multidisciplinary research.