{"id":55595,"date":"2026-05-26T12:03:58","date_gmt":"2026-05-26T16:03:58","guid":{"rendered":"https:\/\/engineering.jhu.edu\/case\/?post_type=news&p=55595"},"modified":"2026-05-26T12:03:58","modified_gmt":"2026-05-26T16:03:58","slug":"ai-based-math-model-speeds-and-improves-scientists-predictions-for-how-buildings-fare-during-natural-hazards","status":"publish","type":"news","link":"https:\/\/engineering.jhu.edu\/case\/news\/ai-based-math-model-speeds-and-improves-scientists-predictions-for-how-buildings-fare-during-natural-hazards\/","title":{"rendered":"AI-based math model speeds and improves scientists\u2019 predictions for how buildings fare during natural hazards"},"content":{"rendered":"

Quickly predicting how structures will respond to natural hazards, such as earthquakes and hurricanes, has long been one of civil engineering\u2019s toughest challenges, requiring significant computational resources and time-consuming simulations. In an effort to improve the design of buildings, bridges and other structures that can withstand such forces, a team of researchers that includes civil and systems experts Michael Shields<\/a>, Somdatta Goswami<\/a>, and Dimitris Giovanis<\/a>, is applying machine learning methods to give engineers a faster and more accurate way to understand and manage structural risk.<\/p>\n

Appearing in Engineering Structures<\/em><\/a>, the researchers say that their goal was to provide accurate simulations of numerous natural hazard scenarios with improved speed and at low computational cost.<\/p>\n

\u201cPredicting the response of structures to natural hazards is very difficult because the events are inherently random, dynamic, and have large magnitude, which can produce a complex response in the structure,\u201d says Shields.<\/p>\n

To predict the response of such highly complex systems, the team developed machine learning approaches that use neural operators\u2014math models that predict changes in response to a wide array of variables that are also changing over time. Unlike conventional machine learning models that map a fixed set of inputs to outputs, neural operators learn the underlying relationships that govern how a system responds across different conditions over time. This makes them well-suited for modeling how buildings and other structures sway, deform, and redistribute forces during events like earthquakes and high winds.<\/p>\n

The team identified two neural operators, a deep operator network, or DeepONet, and a Fourier neural operator, or FNO, to evaluate their impact on speed and computational cost of structural response prediction. They also explored approaches combining the two networks.<\/p>\n

The researchers found that using neural operators produced a significant improvement in structural modeling accuracy, while also reducing computational cost.<\/p>\n

The FNO model stood out, outperforming DeepONet and combinations of the two neural networks. Shields says the training was laborious, but once the FNO model was trained, it made consistently accurate predictions almost instantaneously. In one comparison, a traditional prediction model took 472 seconds to produce a single wind response simulation, while the trained FNO model required approximately 0.01 seconds.<\/p>\n

\u201cThere\u2019s a breakeven point at 804 simulations for the FNO model, so if the application requires more than 800 simulations, the near instantaneous prediction time pays off with substantial computational savings,\u201d Shields says.<\/p>\n

\u201cFrom a civil engineering perspective, our findings enable large-scale uncertainty quantification and faster decision-making for resilient design,\u201d says Giovanis. \u201cFrom a machine learning perspective, they demonstrate how operator learning can effectively model complex physical systems beyond traditional input-output learning.<\/p>\n

The results could potentially change the way engineers assess risk and design structures. Using the new machine learning architecture, engineers could quickly evaluate how thousands of earthquakes or high wind scenarios might affect the structural stability of a building or bridge.<\/p>\n

Goswami says that by enabling near real-time predictions of infrastructure response during extreme natural hazards, the team’s work demonstrates how AI-driven operator learning can enhance structural risk assessment.<\/p>\n

\u201cThis work is only\u00a0the beginning. It serves as a proof of\u00a0concept,\u201d Shields says. \u201cWe now need to refine these methods to reduce training data requirements, ensure they produce physically realistic predictions, and then demonstrate that they will be useful for large-scale risk and reliability studies for real infrastructure.\u201d<\/p>\n

Additional study collaborators include corresponding author Seymour M.J. Spence from the University of Michigan and Bowei Li from Texas Tech University.<\/p>\n","protected":false},"template":"","class_list":["post-55595","news","type-news","status-publish","hentry","news_categories-research","news_categories-resilient-cities","news_categories-systems"],"acf":[],"yoast_head":"\nAI-based math model speeds and improves scientists\u2019 predictions for how buildings fare during natural hazards - Department of Civil & Systems Engineering<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/engineering.jhu.edu\/case\/news\/ai-based-math-model-speeds-and-improves-scientists-predictions-for-how-buildings-fare-during-natural-hazards\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"AI-based math model speeds and improves scientists\u2019 predictions for how buildings fare during natural hazards - Department of Civil & Systems Engineering\" \/>\n<meta property=\"og:description\" content=\"Quickly predicting how structures will respond to natural hazards, such as earthquakes and hurricanes, has long been one of civil engineering\u2019s toughest challenges, requiring significant computational resources and time-consuming simulations.…\" \/>\n<meta property=\"og:url\" content=\"https:\/\/engineering.jhu.edu\/case\/news\/ai-based-math-model-speeds-and-improves-scientists-predictions-for-how-buildings-fare-during-natural-hazards\/\" \/>\n<meta property=\"og:site_name\" content=\"Department of Civil & Systems Engineering\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data1\" content=\"3 minutes\" \/>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"AI-based math model speeds and improves scientists\u2019 predictions for how buildings fare during natural hazards - Department of Civil & Systems Engineering","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/engineering.jhu.edu\/case\/news\/ai-based-math-model-speeds-and-improves-scientists-predictions-for-how-buildings-fare-during-natural-hazards\/","og_locale":"en_US","og_type":"article","og_title":"AI-based math model speeds and improves scientists\u2019 predictions for how buildings fare during natural hazards - Department of Civil & Systems Engineering","og_description":"Quickly predicting how structures will respond to natural hazards, such as earthquakes and hurricanes, has long been one of civil engineering\u2019s toughest challenges, requiring significant computational resources and time-consuming simulations.…","og_url":"https:\/\/engineering.jhu.edu\/case\/news\/ai-based-math-model-speeds-and-improves-scientists-predictions-for-how-buildings-fare-during-natural-hazards\/","og_site_name":"Department of Civil & Systems Engineering","twitter_card":"summary_large_image","twitter_misc":{"Est. reading time":"3 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/engineering.jhu.edu\/case\/news\/ai-based-math-model-speeds-and-improves-scientists-predictions-for-how-buildings-fare-during-natural-hazards\/","url":"https:\/\/engineering.jhu.edu\/case\/news\/ai-based-math-model-speeds-and-improves-scientists-predictions-for-how-buildings-fare-during-natural-hazards\/","name":"AI-based math model speeds and improves scientists\u2019 predictions for how buildings fare during natural hazards - Department of Civil & Systems Engineering","isPartOf":{"@id":"https:\/\/engineering.jhu.edu\/case\/#website"},"datePublished":"2026-05-26T16:03:58+00:00","breadcrumb":{"@id":"https:\/\/engineering.jhu.edu\/case\/news\/ai-based-math-model-speeds-and-improves-scientists-predictions-for-how-buildings-fare-during-natural-hazards\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/engineering.jhu.edu\/case\/news\/ai-based-math-model-speeds-and-improves-scientists-predictions-for-how-buildings-fare-during-natural-hazards\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/engineering.jhu.edu\/case\/news\/ai-based-math-model-speeds-and-improves-scientists-predictions-for-how-buildings-fare-during-natural-hazards\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/engineering.jhu.edu\/case\/"},{"@type":"ListItem","position":2,"name":"News","item":"https:\/\/engineering.jhu.edu\/case\/news\/"},{"@type":"ListItem","position":3,"name":"AI-based math model speeds and improves scientists\u2019 predictions for how buildings fare during natural hazards"}]},{"@type":"WebSite","@id":"https:\/\/engineering.jhu.edu\/case\/#website","url":"https:\/\/engineering.jhu.edu\/case\/","name":"Department of Civil & Systems Engineering","description":"Department of Civil & Systems Engineering","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/engineering.jhu.edu\/case\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"}]}},"distributor_meta":false,"distributor_terms":false,"distributor_media":false,"distributor_original_site_name":"Department of Civil & Systems Engineering","distributor_original_site_url":"https:\/\/engineering.jhu.edu\/case","push-errors":false,"_links":{"self":[{"href":"https:\/\/engineering.jhu.edu\/case\/wp-json\/wp\/v2\/news\/55595","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/engineering.jhu.edu\/case\/wp-json\/wp\/v2\/news"}],"about":[{"href":"https:\/\/engineering.jhu.edu\/case\/wp-json\/wp\/v2\/types\/news"}],"wp:attachment":[{"href":"https:\/\/engineering.jhu.edu\/case\/wp-json\/wp\/v2\/media?parent=55595"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}