On Sept. 11, 2001, when President Bush ordered the grounding of all commercial aviation in the United States, the airline industry confronted a logistical nightmare. By Sept. 14, when the airlines were finally given permission to fly again, carriers faced unprecedented operations problems from a massive, system-wide schedule disruption.
Of course, the airlines had faced disruptions before. But snowstorms, mechanical breakdowns and sick pilots were insignificant when compared to the terrorist attacks. Washington D.C.’s Reagan National Airport was closed indefinitely. Thousands of airplanes had been forced to land at the nearest available airport, which meant they weren’t in the places they were supposed to be. In addition, each of the major carriers had thousands of pilots in the wrong places.
Dr. Gang Yu makes
extremely complicated industrial systems run
“Within an hour, our entire fleet was grounded at the nearest available airports,” recalls Anna White, at the time director of crew technology at Continental Airlines. “It was a monumental problem. I can honestly tell you that in the 19 years I’ve worked in crew operations, I had never seen anything like it.”
Luckily, Continental could rely on a new technology it had developed with Dr. Gang Yu, a professor of management science and information systems at The University of Texas at Austin’s McCombs School of Business. Yu and his Austin-based company, Caleb Technologies Corp., had worked for several years on CrewSolver, a piece of software that relied on complex algorithms to parse tens of thousands of variables that affect airline scheduling, including pilot and plane pairings.
“Crew pairing,” the challenge of matching pilots, planes and crews with schedules and routes, had plagued the airline industry long before the 9/11 attacks. For instance, Continental relies on four different airplanes in its fleet, 747s, 737s, MD80s and DC-10s, and they can only be flown by pilots who are trained on those specific aircraft.
After the aviation shutdown on Sept. 11, all of that information was fed into what Yu describes as his software’s “optimization engine.”
By using CrewSolver, Continental was able to pair the right pilots with the right airplanes. In fact, the airline was able to resolve 1,600 problem pairings for its fleet of 737s and MD80s in less than 20 minutes. Before CrewSolver, the airline’s scheduling department would have had to make do with spreadsheets and a monstrous pile of pencils and erasers.
In 2001, Continental estimated that Yu’s software saved the airline about $40 million, a sizable amount when you understand that the Houston-based air carrier lost $95 million that year on revenue of $8.9 billion.
In 2002, Yu’s team from Continental and Caleb Technologies won the prestigious Franz Edelman Award for Management Science Achievement. The annual award, considered the “Super Bowl” of management science, is designed to “call out, recognize and reward outstanding examples of management science and operations research practice in the world.”
The decision to give the 2002 Edelman Award to Yu and Continental was based in large part on the airline's ability to recover quickly from the 9/11 disaster thanks to CrewSolver.
“[The software] was exceptional in terms of its impact, its methodology, the management science it used in both its soundness and novelty, and also by the number of high level people at Continental who certified that it made a big difference to them,” says Donald R. Smith, an associate professor of management and marketing at Monmouth University who chaired the Edelman committee in 2002.
Yu and his company had proven they were world-class experts at modeling efficiency.
The roots of their effort stretched back to Yu’s childhood in one of the 20th century’s most notoriously inefficient societies, the People’s Republic of China under Mao Tse-tung.
The Yichang Regional Tractor Factory was set up as a way to praise Mao Tse-tung, not as a model for productivity or good workmanship. Whenever the Chinese leader produced a new teaching, or China celebrated a national holiday, workers throughout the country held parades. The parades were a way of sustaining Mao’s power and of reminding the people that they should continue supporting the Chinese Communist Party.
The workers at the Yichang tractor factory in Hubei Province in east-central China joined in the parades. But in doing so, they unintentionally demonstrated the bankruptcy of Mao’s system of state-run businesses and state-controlled manufacturing.
Yu has vivid memories of those parades. He worked in the Yichang factory as a teenager, doing various jobs, including that of handyman and electrician.
With Yu’s CrewSolver software, Continental was able to
resolve 1,600 problem pairings for its fleet of 737s and MD80s in
less than 20 minutes.
“We made our appearance by driving our tractors out,” he recalls with a smile. “If we drove out 100, we would only get 50 back because the others would break down.”
Yu also has sharp recollections of his co-workers.
“People didn’t really work,” he says. “They’d read newspapers. Nobody ever got fired. People weren’t judged by their work ethic or productivity. They were measured by their political beliefs or affiliations.”
Nearly 30 years have passed since Yu began his three-year-long stint in the tractor factory, but those three years have played a critical role in his life. The terrible inefficiencies he saw in the factory gave him clear ideas about the nature of work, productivity and how to best manage industrial operations.
“Since I’d worked in the factory, when I got to college I felt strong internal drive in trying to catch up with others and make up for time I lost. Plus, I also knew more about what I wanted to learn,” he says.
And since leaving the factory at age 17, Yu has been moving very quickly. After getting a master’s degree in physics, he set his sites on operations research and management science. Today, Yu, 45, is a renowned authority on making extremely complicated industrial systems run well, even when they are hit by massive disruptions.
“Gang is one of the best applied operations research people in North America, maybe in the world,” says Patrick Jaillet, the head of the Department of Civil & Environmental Engineering at Massachusetts Institute of Technology. Jaillet, who chaired the McCombs School’s Department of Management Science and Information Systems until 2002, says Yu’s background in physics has given him tremendous mathematical skills.
“But he’s also able to prove things,” Jaillet says. “His strength is taking on significant real world problems.”
Jaillet points out that many people had tried and failed to come up with a solution that would minimize the costs incurred by airlines due to disruptions caused by weather and other factors. It was Yu who finally found a workable solution.
Beyond Class Work
Today, the plaque awarded by the Edelman committee hangs on the east wall of Yu’s third floor office in the McCombs School. His window looks out over Speedway Avenue, and beyond that, Gregory Gym and Royal-Memorial Stadium. It’s a galaxy or two away from the world he knew as a child, growing up in Yichang.
As a teenager, Yu feared that the rest of his life might be spent working in the Yichang Regional Tractor Factory. After Mao’s death in 1976, however, reforms in China allowed Yu to compete on his merits, not his political connections, for entrance to Wuhan University and eventually a scholarship that allowed him to earn a master's degree in physics at Cornell University.
He went on to earn a doctorate in operations research at the Wharton School of the University of Pennsylvania. His dissertation, which he completed in 1989, was on vehicle routing problems in the trucking industry, a precursor to his later work with the airlines.
Dr. Yu’s team received the 2002 Edelman Award, which recognizes “outstanding
examples of management science and operations research practice in the world.”
After coming to The University of Texas at Austin in 1989, Yu began working on several different research projects, but he kept returning to the problems that were occurring in the airlines, particularly disruption management. He liked the disruption problem because the stakes were high and the need for a breakthrough solution was large and growing all the time.
In 1994, Yu and his wife Xiaomei Song, who holds a Ph.D. in physics, started Caleb (Computer Applications in Logistics Engineering and Business) in the basement of their home in order to devote more time to the problem. Within two years, the fledgling company had revenues of $1 million, and by 1997, Caleb had a contract with Continental to develop its disruption management software.
The success of Yu’s software was a function of its ability to synthesize a huge amount of information into a workable model that made use of the algorithms he had devised.
Creating a good model is one of the most difficult problems in operations research, says Smith, who continues to serve on the judging committee for the Edelman Award.
“You have to decide which combination of things capture the gist of the problem and still have the problem be soluble. You can make a model so complete that it can't be manipulated,” he says.
Once Yu had a workable model, he had to write software that could access different databases within the airline. Yu used an XML-based approach that allowed the software to pull data from Oracle, Sybase and other databases. Then he began testing the software with the airlines to assure that it was really working.
Caleb received early financial backing from Techxas Ventures, an Austin-based venture capital firm. James Schellhase, a managing director at the firm, says Techxas saw that Caleb “had a remarkable way of solving the problem that airlines had been trying to handle with spreadsheets.”
Working for Impact
Yu and Song sold Caleb in March of this year to Navitaire, a division of consulting giant Accenture. And while he’s happy that he can focus on teaching at the McCombs School, he’s also actively looking at other industries where he can apply his expertise.
“Disruption management has not been fully explored,” he says. “It worked well in airlines and now I'd like to see where else I can go with it.”
He’s been talking to various semiconductor manufacturers about how his techniques might be applied in the chip fabrication process, which, like the airline industry, is hugely capital intensive and highly vulnerable to expensive disruptions.
Whatever challenge Yu takes on next, he knows that he wants to develop something new.
“I don’t want to follow people’s steps. I don’t want to just add icing to someone else’s cake,” he says. “I want to create a big impact.”
Photos of Dr. Yu: Melissa Grimes & Mark Rutkowski