Buildings and urban infrastructure are designed to withstand earthquakes and other natural disasters, based on data from previous events. But how do you design a building to get through the extreme disaster that hasn’t happened yet?
In other words, you could design a building or infrastructure and be reasonably confident that it would withstand a moderate earthquake. But how confident would you be if the earthquake’s magnitude was 7.5 or 8 — one of those extreme events that happen rarely and for which we have limited data?
Civil & Environmental Engineering Associate Professor Floriana Petrone is working on it.
She’s developing new scientific methods — using simulations, machine learning and probability models — to better predict what happens to infrastructure in extreme disasters. Her project, Toward the Next Generation Data-Informed Probabilistic Framework for Infrastructure Seismic Risk Assessment and Uncertainty Quantification, is supported with a five-year, $575,478 National Science Foundation (NSF) CAREER award granted last summer. The Faculty Early Career Development (CAREER) Program is the NSF’s most prestigious award in support of early-career faculty.
“We are extremely excited about Dr. Petrone’s cutting-edge research and the educational opportunities it will provide to our students,” Keri Ryan, Civil & Environmental Engineering chair, said.
Closing the data gap
Advances in computer technology and simulation techniques by the early 2010s led to very detailed and accurate disaster simulations. The resulting big data (which can tell us about broad patterns) and high-fidelity models, which can show detail, produced an abundance of data about disaster scenarios.
The problem? Traditional methods of assessing structural risk rely heavily on observational data (recordings and experiments) that do not statistically cover extreme events, which leaves gaps in our ability to predict structural risk in an extreme event, according to Petrone.
“So far, we have developed our models based on the empirical data we had,” she said. “However, simulations at a large scale give, for the first time, the opportunity to address critical knowledge gaps, allowing us to verify and update our models for improved estimates of infrastructure behavior under extreme events.
“Nevertheless, while nowadays simulations can be run for any conceivable scenarios, a methodology that brings together the knowledge coming from this new simulation data is missing,” Petrone said.
The new framework she is developing will incorporate the newer disaster simulations into the traditional methods of risk assessment.
“The idea is to develop a framework that is able to pull together observational and simulation data, and leverage machine learning to obtain new, physics-informed models and predict the infrastructure response under a broad range of events, including extreme scenarios,” she said.
This would be an open framework — one that anyone can use — to leverage simulation data that already exists and incorporate new simulation data that will become available.
The endgame is to improve codes and standards around infrastructure design, so buildings and structures have a better chance of withstanding ‘the big one’ or other extreme events from natural disasters.
“The wellness of modern society is tightly tied to the functionality of urban infrastructure and facilities that support essential services,” Petrone said. “Enhancing our ability to properly assess extreme hazard loads on a local-to-site-specific scale and develop reliable probabilistic demand models is crucial for planning a broad range of surface and subsurface activities on a fully risk-informed basis, with direct implications for safety, cost and societal disruption.”
Inflecting new science and technology advancements towards engineering
Petrone earned her Ph.D. in structural engineering at Sapienza University of Rome.
Because Petrone lived in Italy, a seismically active region, she said, “I’ve always had this interest in earthquakes and how engineers can solve classical problems with new approaches.”
She moved to the United States in 2012 as a fellow of the Transatlantic Partnership for Excellence in Engineering, completing her doctoral studies while at the University of California, Davis. She worked at Lawrence Berkeley National Lab, where she contributed to the development of numerical tools to analyze how nuclear facilities would behave during earthquakes.
She joined the Â鶹´«Ã½AV as an assistant professor in 2019, working on several projects related to high-performance computing applied to civil engineering problems and advanced numerical modeling of reinforced concrete and composite structures. She has recently broadened her research interests in the field of additive construction to investigate the use of indigenous materials for the expedient deployment of in-situ infrastructure. In this field, she is leading a 3D-printed concrete project in collaboration with the U.S. Army Engineer Research and Development Center’s Construction Engineering Research Laboratory and supported by the U.S. Department of Defense.
Petrone, promoted to associate professor last summer, sees this recent CAREER award as an opportunity to advance the field of structural engineering.
“This CAREER award represents a long-term vision I want to build for advancing our approach to classical, yet not fully resolved, engineering problems,” she said. “My core objective is to advance the scientific knowledge and computational tools needed to enhance the resilience of civil infrastructure and energy systems to natural hazards. It is also a commitment to mentoring the next generation of engineers at the intersection between computer science, data analysis, and civil engineering.”
