MIT Department of Nuclear Science and Engineering
George E. ApostolakisProfessor of Nuclear Engineering; Professor of Engineering Systems
Diploma '69 (mechanical & electrical engineering), National Technical University of Athens; MS '70 (engineering science), PhD '73 (engineering science & applied mathematics), California Institute of Technology.
Risk assessment and operations management for complex engineering systems; nuclear and toxic waste risk assessment and management; nuclear reactor safety; applications of decision analysis to engineering problems; risk-informed regulation; management of infrastructure risks from terrorism.
We are developing a methodology to apply Probabilistic Risk Assessment (PRA) and traditional "deterministic" methods to the design process, using the new Gas Fast Reactor (GFR) as case study. The use of the integrated top-down PRA methodology will result in more effective and economical design solutions than can be achieved via purely deterministic methods. At the same time, we are investigating the applicability of certain "deterministic" regulatory requirements to a risk-informed licensing process.
Safety margins are incorporated into the design and operation of nuclear power plants. However, their effect on PRA -- that is, the impact on core damage frequency and the large early release frequency -- is not known. Currently, we are investigating the uncertainties in reliability assessments of passive safety systems, and potential reductions in margin as fuel burnup is increased in light-water reactors.
In 1998, the Nuclear Regulatory Commission (NRC) issued a White Paper in support of Risk-Informed Decision-Making. "[NRC] regulations can be strengthened and resources allocated to ensure that they are focused on the most risk-significant equipment and activities, and to ensure a consistent and coherent framework for regulatory decision-making."
To facilitate this effort, we are developing a catalogue of NRC decision-making processes. This includes reviewing documents and interviewing key NRC decision makers. This catalogue will help us identify common elements among the various decisions and will serve as the basis for selecting the decision-making processes that will serve as case studies. For the selected case studies, we will evaluate the quality of the available risk information and assess how defense-in-depth and safety margins could compensate for the weaknesses in risk information so that robust decisions could be made. We will also place these decision-making processes in the context of formal theories of decision-making, e.g., multi-attribute decision theory, in order to test the coherence of the existing processes.
Protecting critical infrastructures from terrorist threats presents an enormous challenge. Given that society cannot afford absolute protection, it is necessary to identify and prioritize vulnerabilities. To do so, we are developing a methodology to model infrastructures as interconnected digraphs and employ graph theory to identify scenarios of vulnerability. These are assessed for susceptibility to terrorist attack using multi-attribute utility theory. The methodology is illustrated by application to the MIT campus.
Michael W. GolayProfessor of Nuclear Engineering
BME '64 University of Florida; PhD '69 (nuclear engineering) Cornell University.
Risk and reliability; decision support and system evaluation methods, energy policy and management.
Work has focused upon improving methods for framing and resolving safety regulatory questions on a risk-informed basis. Important topics include treatment of uncertainty, measures of probabilistic risk assessment quality, software reliability, and treatment of individual high-consequence events.
Decision support and system evaluation methods: Important decisions often involve substantial elements of uncertainty. Our work in this area utilizes a probabilistic formulation of problem structure and decision-maker beliefs; this approach is complemented by deterministic simulation of systematic feedbacks and non-linear amplifications of system perturbation. Important recent applications have included nuclear technology performance assessment, including proliferation resistance, new technology R&D management, nuclear power plant organizational performance analysis, and large construction project organizational performance.
