Organizing and evaluating uncertainty in geotechnical engineering

Probabilistic methods are potentially useful in four stages of a typical project: site characterization and evaluation, evaluation of designs, decision making, and construction control. In evaluation of projects, it can be useful to express risk numerically. When uncertainties can be quantified and model errors are understood, reliability theory may be used. Event-tree analysis can be a framework for effectively applying judgment concerning uncertainty. The use of quantified risk in decision making is limited by standards for acceptable risk; good communication with a client is essential. Unless clients or regulators are interested in quantifying risks as part of decision making, engineers will continue to rely on traditional methods. When risks are large and the costs of absolute safety are large, clients are interested in discussing risks. Issues concerning the adequacy of existing structures such as earth darns are stimulating interest in risk assessment, and there will be spin-off from developments in earthquake engineering. More and better examples of applications of probabilistic methods are needed.