Computer Models Have Increasing Value in Assessing Proliferation Risks, Nuclear Expert Says

Computer models can be useful risk-assessment tools in the arena of nuclear proliferation, though even the most advanced cannot “predict” whether a given candidate country will try to develop nuclear weapons or not, nuclear engineer Man-Sung Yim told a Capitol Hill luncheon gathering organized by AAAS.

Ongoing work on quantitative models has progressed to the point where they can, for example, help nuclear watchdogs decide where to focus limited resources among countries with the potential to go nuclear and they have, among other things, identified key variables that seem to discourage a country from going nuclear. Yim spoke 8 October at an event sponsored by the AAAS Center for Science, Technology and Security Policy (CSTSP) as part of the Congressional Science and Security Seminar Series.

Man-Sung Yim

Man-Sung Yim

Improving the early detection of potential threats is important to the future of the commercial nuclear power industry, as it attracts a growing number of developing nations, Yim said. A key question for the industry’s supporters, he added, is: “Can the peaceful use of nuclear power be expanded without affecting the world’s nuclear proliferation?”

Detection of proliferation threats has depended primarily on expert judgments developed from the laborious processing of massive amounts of information, noted Yim, a professor of nuclear engineering at North Carolina State University who is also affiliated with Oak Ridge National Laboratory.

His talk at the Cannon House Office Building focused on researchers’ efforts over recent decades to augment this somewhat subjective risk-assessment process by developing statistical modeling that could provide more rigorous or systematic signposts pointing to future proliferators. He said the effort has had mixed results, is sometimes “unreliable,” and “needs more work.”

There has been some skepticism among nuclear experts about the predictive capabilities of computer simulations, said Pierce Corden, a visiting scholar with the AAAS Center for Science, Technology and Security Policy, although such modeling “helps order your thinking” and therefore has problem-solving value.

“We’re not predicting the future,” Yim had told the group, “but we are trying to project what is expected to happen” based on a set of settled assumptions, “and trying to get some useful information out of it. That’s all. It may be a useful aid for warning. If that objective can be met, then imprecisions, maybe some errors, can be forgiven.”

In past decades, Yim said, the relationship between peaceful nuclear power and nuclear weapons had been something of a taboo topic among the nuclear engineering community. “But the world has changed,” he said. “Time has passed. And people in the civilian nuclear sector are concerned about the connection.”

The worst fears and predictions of proliferation experts have failed so far to materialize. Despite the spread of commercial nuclear power facilities, there has been no nuclear domino-effect leading to the rapid multiplication of nuclear weapon states—and there are numerous theories about why this is so.

As the statistics make clear, “it’s not easy to develop infrastructure for civilian nuclear power,” Yim said. “And today it is very difficult actually to develop nuclear weapons out of this industry.” Among other things, it requires vast commitments of money and other resources as well as political will.

Since the 1950s, beginning with the Atoms for Peace program and continuing under the International Atomic Energy Agency regime, almost 60 countries have expressed an interest in developing nuclear power for peaceful purposes, Yim said. “Many failed.”

There are currently some 441 reactor units in operation in 29 countries around the world. The United States leads with 104, followed by France with 58, Japan (54), Russia (32), South Korea (21), and India and the United Kingdom, with 19 each.

Almost 60 additional nuclear power plants are under development in more than a dozen countries, including four in the United Arab Emirates. China has 19 plants under construction; Turkey, Indonesia, Vietnam, Egypt and Israel are planning new commercial reactors and new nuclear energy programs are beginning in Jordan, Libya, and Poland.

On the military side, out of a total of 23 countries that have explored joining the “club,” Yim observed, nine possess nuclear weapons. (These include the five official nuclear weapons states recognized in the Non-Proliferation Treaty—the United States, Russia, United Kingdom, France, and China—and four countries not party to the treaty—India, Pakistan, and North Korea, plus Israel, which declines to confirm its nuclear status.)

One of the important arguments for using quantitative computer models, Yim said, is that “the process helps you understand and identify key factors that affect the nuclear decisions” of various countries. “Many countries had an interest” in developing nuclear weapons “and then stopped,” he said. “Why did they stop? Modeling efforts may help us to understand.”

The models show that several variables consistently “inhibited” proliferation, he said, including a country’s possession of a successful commercial nuclear power industry to generate electricity, and its establishment of democracy.

Yim said the modeling uses a data base (all unclassified material) covering the nuclear status of 114 countries over a many decades. The researchers used two basic approaches—regression and event history analytical models—that seek to indicate, for example, how various outcomes are affected by the use of various inputs, and to see what approaches best “predict” the outcomes that actually happened in given cases.

They plowed in quantifiable factors such as a country’s economic status (scholars say it takes a Gross National Product of roughly $100 billion to be able to afford a viable nuclear weapons program), and its nuclear technical capabilities, its political motivation and security concerns. The challenge in the research is how to consider non-quantifiables such as leaders’ psychology and desire for prestige.

One purpose of his talk, Yim said, was to encourage more collaboration on the computer modeling effort among members of the nuclear community. “We are expanding our database and looking for help.”


See biographical information on nuclear engineer Man-Sung Yim.