Learning from Experience - Characterization of Dam Incidents

Learning from the in-service performance of dams is an important element of dam engineering and safety. In many respects, dam engineering is an empirical, trial-and-error undertaking. While present day methods of analysis in geotechnical, structural, and geologic engineering are sophisticated, we continue to rely on observational evidence to understand a dam is responding to the constant hydraulic and environmental loading cycles that it experiences. This makes a great deal of sense since dam engineers are keen to point out, each dam site, dam design and structure are unique – one-of-a-kind systems. Therefore, modern dam engineering and dam safety rely on detailed, structure-specific monitoring to track the performance of individual structures. In addition, the profession also requires data on dam incidents across the population of dams to be gathered, archived, and evaluated. While individual dams are unique, documenting and evaluating the performance of the broader population provides a foundational basis for understanding dam performance and identifying experiences and lessons the industry can benefit from. An important element to develop the level of understanding required to learn from the performance of dams is to establish a data system that captures information that supports the advancement of dam engineering and safety. In this paper we describe an approach to characterizing dam incidents. The approach includes a graphical method to represent dam incidents. The method, known as an influence diagram, represents events in an incident, along with factors that contributed to the events in the timeline. A challenge with influence diagrams is the image cannot be easily search if retained only as an image. To address this, we have developed a scheme that defines the influence diagram in a text string, referred to as an ‘influence string’. Additionally, we have written a code that makes it possible to read an influence string and generate an influence diagram on demand. In this paper, we lay out considerations for developing this data system and illustrate, through an example, an approach to capture information on dam incidents (in a database) and display the influence diagram for the user.