CONTACT: Stanford University News Service (650) 723-2558
Elisabeth Pate-Cornell: From physics to human comedy
STANFORD -- Throughout her life Marie-Elisabeth Pate- Cornell has made a practice of being an exception. Her first summer job in college was as a computer analyst in a foundry in the French Pyrenees.
"The forge was a vision of hell. As the men worked, large pieces of red-hot metal moved about them. It was very dangerous. So dangerous that women were not allowed there," she recalled. "I wanted to get the story firsthand, so I got an exception. I must have been quite a sight in my hard hat and miniskirt."
Pate-Cornell's most recent distinction is to become the first woman faculty member from Stanford University to be elected to the National Academy of Engineering (NAE), the highest honor that can be accorded to an American engineer. She is one of only 34 women and, at the age of 46, one of the younger members of this august body.
"Elisabeth is the sort of person who should be recognized early in her career. Her election has raised the caliber of the academy membership as a whole," said Jack Flipse, professor emeritus of civil and ocean engineering at Texas A&M, and a member of the NAE.
The election was a recognition of Pate-Cornell's contributions to the field of engineering risk analysis. As a professor of industrial engineering and engineering management, she analyzes the risks of failure in complex, critical engineering systems. In particular, she studies how people affect the performance of such systems in environments ranging from offshore pipelines to NASA's space shuttle to the hospital operating room. She titles her public talks on her research "From Physics to Human Comedy."
Arguably her greatest contribution to date has been to extend conventional engineering risk analysis to take organizational issues into account. Instead of concentrating exclusively on physical failure mechanisms, Pate-Cornell has developed techniques for assessing how an organization's structure and policies contribute to reducing or increasing the risk of failure of complex engineering systems.
In a study of the thermal tiles of the space shuttle, for example, she determined that the greatest risk of an accident related to the shuttle's heat shield is due to the potential failure of a relatively small number of the thousands of ceramic tiles that protect the shuttle from the blazing heat of re-entry. She proposed specific improvements that NASA has adopted to significantly reduce this source of danger.
From Senegal to France
Given her background, Pate-Cornell's choice of career is particularly unusual.
She was raised in a conservative military family. Her father was a colonel in the colonial French marines. She was born in Senegal, where she attended public school in Dakar through junior high before her family returned to France.
"After Africa, France seemed small, wet and gray," she recalled.
As a teenager, she enjoyed math problems the way other people enjoy crossword puzzles. But she was also interested in the humanities. At 17, she wrote an essay about Proust that prompted her teachers to encourage her to go into the humanities. In the French educational system, however, she had to decide between scientific and humanities tracks before going on to the university.
"If I didn't go into the sciences, I knew my knowledge there would become out of date. But I could always enjoy music, art and literature," Pate-Cornell reasoned.
This decision illustrates one of Pate-Cornell's important priorities, both personally and professionally. "An important point to me is balance of life," she said.
In the mid-1960s, computer science was a new and exciting field. Pate-Cornell found it irresistible. She enrolled in the Institut Polytechnique of Grenoble, the first school in France with a full-fledged computer science and engineering program. But, seeking balance, she also studied economics: "When you go into applied maths, it's nice to have a field to apply it to."
After graduating from Grenoble, Pate-Cornell applied to the master's program in operations research at Stanford. When she told her parents her plans, however, "they pulled long faces. It was strange to them to have a daughter do this sort of thing." She won her parents' agreement by promising to return to France after completing her degree. Having thoroughly enjoyed her time at Stanford and finding a perfect match between her interests and the Department of Engineering-Economic Systems, however, she was back at Stanford working on her doctorate a year later.
"Bill Linville was my mentor. That was the time when more women were getting into engineering at Stanford and he was very supportive. Of course, at that time my English was not very good, and I didn't really understand the way the system worked. So I imagine he watched me much like someone watching a sleepwalker on the edge of a roof. He kept me from falling off the edge," she recalled.
During this time, Pate-Cornell worked on seismic risk with civil engineering Professor Haresh Shah. Her thesis topic was a policy analysis for earthquake risk reduction: Should we reinforce all buildings to a certain level or should we invest in earthquake prediction? As part of this work, she went to an international conference on earthquake engineering in India. There she met a civil engineering professor from the Massachusetts Institute of Technology, Allin Cornell, who would play a major role in her future.
The seismic work led her to a study of optimal sensitivity of warning systems. If a warning system is too sensitive, it generates so many false alerts that people begin to ignore it. On the other hand, if it is not sensitive enough, it may either fail to warn people to a real threat or sound too late to allow people to take protective measures. Having casually observed the way that students in one of the residence halls at Stanford responded to a faulty alarm system by stuffing socks in the sensors, she generated a mathematical model to calculate the optimal sensitivity for alarm and monitoring systems that took technical system characteristics and human responses into account.
After earning her doctorate in 1978, Elisabeth Pate accepted an assistant professorship in civil engineering at MIT. She and Allin Cornell became romantically involved and were married. But the next two New England winters were particularly harsh. "It was just too cold for me. So I came back to Stanford with a husband," she said.
The move was not so easy for her new husband, however. Ten years older than his wife, Cornell had built up a comfortable professional life in Cambridge. He was well established in his field -- also in risk analysis, specializing in the failure analysis of marine structures such as offshore platforms subjected to external forces, like those from earthquakes, strong winds and waves. After a year's sabbatical at Stanford, and another year on extended leave, Cornell resigned from MIT and took a half-time position as a research professor in the civil engineering department at Stanford. He devotes the other half of his time to industry consulting.
A career at Stanford
Three months after Pate-Cornell joined the Stanford faculty, she had her first baby. "Some people were a bit surprised when three days later, I brought him into my office in a basket," she said. "Children are one of my passions in life. I had two while the tenure clock was ticking."
The way she balances the demands of work and family is to keep the two as separate as possible. "I don't keep an office at home, and I try to bring as little work home with me as possible. I try to keep my weekends free, and we try to do things together as a family. Evening and weekends, I like to cook, play the piano and the guitar, receive friends and play tennis. Recently, my daughter taught me to shoot baskets!"
This balance between work and private life is only possible, she said, because her husband is completely supportive and because they decided, early on, to hire full-time help at home.
"The other side of that coin is that we share very strong professional interests," said Cornell. "We often get into technical discussions over morning coffee. But if we really want to have a detailed discussion, we make appointments with each other."
It is a balance that she feels has worked. "It's been 15 years now and no one in the family has had a fit over it. My son and daughter were very proud when I was elected to the academy."
Pate-Cornell's research also deals with balance. "Risk analysis is all about setting priorities. We aren't infinitely rich and days have only 24 hours. So, we have to set priorities," she said.
In a 1983 paper titled "Acceptable Decision Processes and Acceptable Risks in Public Sector Regulations," she put down some of her thoughts about the trade-offs involved in applying risk analysis for regulatory purposes, an approach now espoused by the leaders of the new Republican majority in Congress.
In the paper, she pointed out that safety regulations inescapably have two effects: reducing risks for certain groups of people but at some economic cost to everyone. Without a generally accepted method for determining the trade-offs, there is a serious danger from adopting fixed-recipe risk analysis methods for regulatory purposes, she warned. "No two risks are exactly alike. Quantitative information helps to describe them, but other aspects of the problem have to be taken into account in risk management decisions, " she wrote.
The most productive approach would be to concentrate on developing a process for determining such trade-offs. "Whereas there are no universally adopted figures for 'acceptable risks' . . . there can be acceptable decision processes," she argued.
"We should use cost-benefit analysis in environment, health and safety regulations, but not for all levels of individual risks. Above a certain threshold, a risk is simply unacceptable. Below a very low threshold, we should not worry about it. But in the middle range where many of the risks we are arguing about generally fall, we should apply cost-benefit analysis techniques."
Some of Pate-Cornell's specific research projects give an idea of the potential that risk analysis carries for improving safety.
"Marie-Elisabeth has very good tastes in problems, so her solutions have considerable generality," commented Granger Morgan of Carnegie Mellon University, one of the leaders in the field.
Shortly after returning to Stanford, Pate-Cornell realized that her work could be applied in a number of different areas. In 1982-83, she and one of her doctoral students studied monitoring systems for nuclear reactors. Next, discussions with William Perry, the Stanford professor now serving as secretary of defense, led to a study of the trade-offs between the possibility of false alerts and of alarm failures in command-and-control systems like those used by the Department of Defense for nuclear missiles. She also did an analysis of fires in oil refineries that led to a recommendation that monitoring cameras be installed in specific locations to shorten the fire detection time.
In 1988, Pate-Cornell came to another important realization. There was a big gap between those like her who did engineering-based risk assessments, and those who studied organizational behavior.
The first opportunity to put this realization to work came about as the result of casual conversations with her husband and his colleagues regarding the safety of offshore oil drilling platforms. "I realized that something was missing that could make their analyses more useful in practice," she said. As a result, she worked with 30-year expert in the field Robert Bea, now at the University of California-Berkeley. The result was a paper that she published in 1990 in Science titled "Organizational Aspects of Engineering System Safety: The Case of Offshore Platforms."
The article discussed accidents in offshore oil platforms that result from the way the companies operate, and linked the frequency of certain types of accidents with specific organizational factors. Pate-Cornell concluded that the safety of offshore platforms could be improved substantially if designs were reviewed by an independent process. She estimated such a process would cost about $100,000 per platform, but could reduce the probability of failure by about 20 percent. By contrast, it would take about $9 million of steel reinforcement in each platform to achieve an equivalent degree of increased safety.
One of the turning points in Pate-Cornell's career came shortly after her oil platform study, and came from a Stanford contact, Peter Banks, who was a professor of electrical engineering and director of the NASA-supported Center for Aeronautics and Space Information Sciences at the time.
"The five faculty members involved in the center had an opportunity to get involved with a serious problem, the loss of the shuttle thermal tiles," said Banks, who is now president of the Environmental Research Institute of Michigan. "Elisabeth and I were friends and colleagues. She had the right training and skills to analyze the problem."
With Banks' encouragement, Pate-Cornell went to NASA leaders and persuaded them to support such a study. They agreed over considerable internal opposition, according to Banks. "There were some people at NASA headquarters who were concerned that her work might reflect poorly on the current management. But Elisabeth was very politically adept, as well as technically very strong. So, she got permission to do the study," Banks said.
She and a research assistant, Paul Fischbeck, now at Carnegie Mellon University, put together a risk analysis model that found 85 percent of the risk of catastrophic failure could be traced to only 15 percent of the tiles, and identified these high-risk tiles. They recommended that shuttle inspectors concentrate their final inspections in these critical areas.
"I went to Johnson Space Center in Houston to see how the tiles had been designed and what the experience had been. Then, I went to Kennedy to see how they were being maintained. I got the approval of the managers to visit the work floor, but I also spent time with the technicians over beers in the bars of Cocoa Beach to find out what was really going on," she said. Among other things, she learned that one technician had been spitting in the tile adhesive to make it set faster, despite the fact that the workers had been warned that water weakened the glue.
The shuttle study, which was published in two papers in 1993, led to her next big project, a National Research Council study of marine pipeline safety. The study panel consisted of industry experts in the pipeline business, university scientists who had studied theoretical aspects of the situation, members of the public and Pate-Cornell.
"Marie-Elisabeth was included because she had done very clever work that was unique in its approach in addressing the shuttle tile problem," said Texas A&M's Flipse, who participated in the study.
The purpose of the exercise was to improve the safety of marine oil and gas pipelines, most of which are in the Gulf of Mexico. The departments of Transportation and of the Interior had come out with two conflicting sets of regulations. So the federal government asked the council to bring the parties together -- pipeline operators, and federal and state regulators-- to address the issue and recommend some solutions.
When Pate-Cornell started, she didn't know anything about pipelines, "but she had more damn ideas than a cat has fleas," Flipse said. She made a major contribution to the final report by identifying the specific information required to use risk analysis techniques to make pipelines safer, he said.
Although the report, titled "Improving the Safety of Marine Pipelines," only came out a few months ago, pipeline operators are enthusiastic and have begun acting on the recommendations, and the departments of Transportation and of the Interior have begun using the same accident and incident report forms, Flipse said.
From the maritime and space environments, Pate-Cornell recently has moved into an entirely new area, the hospital operating room. In 1992, David Gaba, then assistant professor of anesthesia at the Stanford Medical Center, happened to read her Science article on the safety of offshore oil platforms. "Normally, I would never look at the oil literature. But I read the paper and realized that the problems were really quite similar to those that we have in anesthesia. At the time, I was helping to organize a workshop on errors in anesthesiology, so I invited her to attend the conference. Her decision to pursue funding in this area is a direct consequence of that meeting," he said.
Gaba and Pate-Cornell have collaborated on a preliminary study of patient risk during anesthesia (in press) that provides some rough estimates of which organizational changes are likely to have a significant effect on patient safety. A major push for this type of research is the Food and Drug Administration's desire to have more formal support for design decisions in medical equipment, Gaba said. "Her work opens up a number of avenues that we didn't have available previously. We would like to use some of her techniques to help manufacturers come up with better estimates of the error pathways involving their equipment that will either lead them to change their designs, or justify them to the regulators."
Another new project is work on a framework for risk analysis of global warming for the Electric Power Research Institute, the research arm of the electric power industry. In working on this project, Pate-Cornell had a number of discussions with Stephen Schneider, Stanford professor of biological sciences and a world authority on climate change.
"I've learned a lot from Elisabeth," Schneider said. "She can't change the science. That's not what she does. But she brings a fresh and objective viewpoint to controversial subjects. She can help see that subjective statements about matters such as high, medium and low risks are made with a consistent set of assumptions and that when expert opinions are weighed, it is done in an internally consistent and methodologically correct way."
A key ingredient in her success is her graduate students. Pate-Cornell currently has six students, three women and three men, who operate as a team. They meet together as a group once a week. She also has assembled a number of "friends of the group" from different disciplines who meet with them as well to discuss their specialties, and provide advice and insights. "This is the only way that I have been able to do all the things that I have done," she said.
One of the six is Linda Lakats, who came to Stanford from the University of Toronto. Originally, she was interested in production. But after taking one of Pate-Cornell's classes, she switched to risk analysis. "What attracted me about it is the fact that you can actually help people by reducing the risks that they are exposed to," she said.
According to Lakats and her fellow students, their group is tighter knit and much closer than a lot of other graduate student groups -- partly because of their adviser's relationship with them. Graduate students are the exception to Pate-Cornell's rule about keeping work and family separate. "Bringing home graduate students is different from bringing home work," her husband explained.
According to Carlo Pugnetti, Pate-Cornell is "pretty good" about letting her graduate students choose their own areas of research. "She's very open, but she's also very helpful. She is a real adviser, much more an adviser than a boss."
Last year Pate-Cornell was elected president of the Society for Risk Analysis, the professional association for risk analysts. The timing proved to be fortuitous because of the transition going on in Congress.
"The Republicans and Democrats look at risk in a very different way," Pate-Cornell said. Democrats have tended to use conservative assessments of risk. Until recently, they made little efforts to obtain the most accurate possible assessments of risk. This approach often made it difficult to determine exactly how great a given risk is, she said.
"I agree that we must take a conservative approach to risks, but the conservatism should be in the policy, not in the assessment," Pate-Cornell said.
The Republican approach, on the other hand, has been to emphasize a cost-benefit calculus. Republicans generally argue that risks should be balanced by the economic rewards that are to be gained. The party's "Contract with America" involves a number of bills designed to supersede existing regulations in the areas of environment, health and safety. For example, the original Job Creation and Wage Enhancement Act (H.R. 9) had an entire section, Title III, on risk assessment. It has since been split out as a bill (H.R. 1022) of its own. The bill states that all new federal regulations above a certain level of cost should be subjected to risk assessment and cost-benefit analysis, and only go into effect if they will provide a net economic benefit.
Because of the surge in legislative interest in risk analysis, the Society for Risk Analysis decided to sponsor a series of breakfast meetings for congressional staffers who are drafting this legislation, in order to inform them about what risk analysis can do and what it can't do. So early this year, Pate-Cornell and 11 fellow risk analysts spent several days in Washington, D.C., holding these sessions.
"This legislation represents a new and welcome departure from the previous approach because it requires a search for the best estimates of risk. But there were some serious problems with the original legislation," she said.
Pate-Cornell's general concern is that the lawmakers appeared to be expecting too much from the science. There are cases where the techniques of probabilistic risk assessment fall short or become very cumbersome to apply. A more specific concern is that they did not recognize that there are different ranges of risk that must be handled differently.
"In areas where the risk is moderate, that is, at normal levels of risk, the cost-benefit approach is justified. But there are other areas of risk that are so high that people should never be exposed to them no matter what the benefit. These are risks like that from playing Russian roulette," Pate-Cornell explained.
Will these educational efforts in Washington, D.C., pay off? "We do not know yet how successful we have been in improving the final text of these bills, but we reacted fast, gave it our best shot and are keeping our fingers crossed," she said. "It is very important that these fundamental problems are treated in a way that is scientifically sound, economically effective and ethically correct."
It's a long way from the halls of Congress to the foundry in the Pyrenees where Pate-Cornell first worked. But, looking back, she sees a connection. The subliminal image of red-hot iron and sweating workers may have influenced her to use her mathematical talent to find ways to reduce the hazards in dangerous workplaces like that distant forge. "You know, that experience may be one of the reasons why I went into the risk analysis field," she said.
This is an archived release.
This release is not available in any other form.
Images mentioned in this release are not available online.
© Stanford University. All Rights Reserved. Stanford, CA 94305. (650) 723-2300.