February 1998
Practice Lesson:
OR Helps DOE Diffuse Explosive Issue
By Nancy Bistritz
The subject of nuclear weapons has long been associated with criticism, concern and politics. Such was the case in 1994 when the implementation of the START I and START II Treaty required the U.S. Department of Energy to reduce the nuclear weapons stockpile and to replace the oversized and inefficient production facilities.
Each of these treaties were the basis for the development of the nuclear-weapons-stockpile memorandum that the president approves annually. This memorandum specifies the amount of tritium the DOE is required to produce [Interfaces 28:1].
In March 1994, President Bill Clinton approved a nuclear-weapons-stockpile plan which stated that the United States needed a tritium supply facility that would be online by 2011.
How it works
One of the key "ingredients" in a nuclear weapon is tritium. According to Eric Schweitzer, former Deputy Director of the Office of Reconfigurations in the DOE, a nuclear weapon goes through two explosive stages. The first stage, often referred to as the "plutonium pit," contains a hollow ball of plutonium that is wrapped in high explosives. When the weapon is detonated, the high explosives explode, causing the plutonium to become critical.
The explosive power of the plutonium is boosted with the inclusion of tritium. This first stage acts as a "trigger," which initiates the explosive power of the second stage. In this stage, highly enriched uranium explodes as a result of the first stage, making for a much stronger explosion overall.
Plutonium and uranium have extremely long half lives of thousands of years, tritium has a half life of approximately 12 years. Therefore, the tritium needs to be periodically replenished. While the DOE has sufficient supplies of plutonium and uranium, a production facility for tritium is needed.
Since 1960, the DOE has operated 14 reactors which produced tritium. However, in 1994, none of the reactors were operational, and tritium had not been produced in the country since 1988. What the United States had been doing was dismantling surplus weapons, recovering the tritium and reusing it in existing weapons. The 1994 Nuclear Weapons Stockpile Plan established the need for a new facility by 2011, but left implementation up to the DOE which needed to determine the type, size and location of a facility that would meet the requirements for tritium by 2011.
Defining the Task
Detlof von Winterfeldt, a professor of planning at the University of Southern California, had been working with the DOE on applying decision analysis methods for project management in the Department's Office of Reconfiguration and Defense Program. When he began working on this project, there were approximately 70 people working on various aspects — mainly environmental issues, engineering concerns and cost.
"We didn't emphasize so much fancy work and modeling and statistics in the beginning, but we did say, 'Okay, how can we get our arms around all this information?' " von Winterfeldt said.
Several alternatives were considered: building one of several types of new reactors; building a linear accelerator; or relying on production in existing commercial light-water reactors. Each had its own set of advantages and disadvantages. Environmental groups opposed building a new nuclear power plant. Tritium production in a commercial reactor was relatively inexpensive, but faced several institutional issues. Accelerators were more expensive and had not been previously used for large-scale production [Interfaces 28.1].
According to von Winterfeldt, the analysis took place in two rounds: the first in 1994, the second in 1995. During the first round, von Winterfeldt was primarily concerned with data collection to assess the various alternatives against a set of criteria including cost, the degree to which the 2011 schedule could be met, the degree to which the capacity was being met, and health/safety risks.
"What we laid out was a matrix where we had the alternatives and approximately 20 criteria," von Winterfeldt said. "Then, decision makers could just sift through the facts and say, 'This one looks good here, this one looks good there.'
"But because there was so much information, we built a model that could help the decision maker evaluate the trade- offs among the criteria."
That work, concluded in December 1994, narrowed the alternatives to two candidates: the accelerator and the small light water reactor. Another strong contender — supported by Congress, the DOE and other agencies — was the use of a commercial reactor.
This alternative had been considered a viable option because it was so inexpensive; however, with it came several political and institutional issues. First, the DOE wasn't sure a utility company would be willing to provide tritium production services. Second, the DOE wasn't sure using a commercial reactor was compatible with the Atomic Energy Act. Third, there was a question of international policy to consider.
"How do we look in the world if we discourage everyone in sight not to use their commercial reactors for nuclear weapons production, but then we do something like that ourselves?" von Winterfeldt said. "We don't exactly come across as the great moral leaders."
Second Stage
In 1995, von Winterfeldt and the DOE began the second stage of the analysis. Previously, von Winterfeldt had used a set of approximately 20 different criteria to compare the alternatives. However, when he rescoped the analysis — in early 1995 — von Winterfeldt determined that many of these criteria were unnecessary.
"We had 20 different criteria, and it turned out that most of these were irrelevant like air pollution — these facilities just don't pollute a lot," he said. "We focused on what mattered which was cost, production assurance and environmental concerns."
After limiting the number of criteria, von Winterfeldt put OR techniques to work. "We started with production assurance. With both production assurance and cost, you would like to get a number for it,' " von Winterfeldt said, adding that there was a tremendous amount of uncertainty involved, so they conducted a risk/uncertainty analysis.
"The purpose of the uncertainty analysis was to appropriately express the range of uncertainty in the numbers that we came up with," von Winterfeldt said. "The Secretary of Energy was the decision maker, and we felt it was inappropriate to tell the secretary, 'Here are the precise numbers. Option A costs $3 billion; Option B costs $4 billion. Option A can be online in 2009; Option B can be online in 2010.' "
Using formal probability elicitations, event tree analyses and Monte Carlo simulations, a production simulation was developed that combined the results of the schedule and production risk analysis to predict production of Tritium for each alternative 40-year period.
In the case of production assurance, all of the alternatives met the requirements in terms of how much they could produce, but two types of reactors — the heavy water reactor and gas cooled reactor — had extremely large scheduling problems, meaning there wasn't a high assurance that those facilities could be online by 2011.
While the production risk/uncertainty analysis proved it could eliminate a contender, it also proved that alternatives, not previously considered viable, could stand a chance. "There was concern about the accelerator having production/capacity problems — that maybe it wouldn't be able to produce as much as we needed because we'd never built a facility that big," said von Winterfeldt.
"That analysis proved, though, that it had a pretty good production assurance with the accelerator."
The Results
In September 1995, von Winterfeldt took his analysis to the Secretary and her staff, who decided to take a dual-track approach to solving the problem: Theywould pursue the commercial reactor to determine whether the institutional issues could be resolved and would also simultaneously pursue the accelerator to see if it could meet both production goals and schedule requirements.
The process of studying both of these options was scheduled to take approximately three years. A decision is expected in 1998.
"It was a compromise, but I think it was also wise. There are still things that need to be learned about both options," said von Winterfeldt, adding that after briefing the Secretary, both he and Schweitzer worked on drafting the record of decision that justified the Secretary's decision in the form of a legal document. This effort was extremely rewarding for von Winterfeldt, who said, as a decision analyst, compiling this official record of decision was not in his job description.
"Normally what we as decision analysts do is put the results on somebody's table and say, 'Here are the recommendations.' Then you go away and they do with it what they want. In this case, we developed the analysis and the report and we pursued it to a set of recommendations and then we tied it together in a formal legal document," he said.
For many at the DOE, including Schweitzer, this was the first time they had worked with someone using OR methods. When reflecting on the project, Schweitzer said he thought it helped with the decision. "It came at an extremely complex problem and broke it down into its logical components so that you could understand them, and then it allowed you to 'roll them up' in a logical way so that you could understand the basis for the recommendations you were making," Schweitzer said.
In addition to using new problem-solving methods, there were several other sensitive issues both von Winterfeldt and Schweitzer faced including how to deal with decision makers who weren't accustomed to such techniques. According to Schweitzer and von Winterfeldt, there was a constant concern among those involved that the final analysis would, in essence, be a biased one.
"Even though we tried very hard to explain to them that the analysis was not going to make the decision for them, but rather would inform their decision, there was still a nagging feeling in many decision makers' heads of, 'Will my hands be tied once this analysis is done?' " said von Winterfeldt. "That's a real serious obstacle if you're talking about OR methods that are directly designed to support decision making."
von Winterfeldt said that while he has no "pet answers" on how to overcome that particular obstacle, he did say experience and working with an "educated and supportive" decision maker helps.
In addition to learning that experience pays off, von Winterfeldt said he learned that a good OR analyst is flexible, has ups and downs, and hangs in there. "I kept reminding myself, 'This is a very important national decision. It's a problem to which our OR tools are immensely useful. If I can't hang in there and make these tools work on a problem like this, then I'm not a good analyst — I'm not a good member of the OR community'. I did it for myself and for the field. And it worked in the end."
Reader Service Form
OR/MS Today copyright © 1998 by the Institute for Operations Research and the Management Sciences. All rights reserved.
Lionheart Publishing, Inc.
506 Roswell Street, Suite 220, Marietta, GA 30060, USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com
Web Site © Copyright 1998 by Lionheart Publishing, Inc. All rights reserved.