Researchers at the University of East London (UEL) have developed a new framework for assessing earthquake risks, using AI-assisted technology to examine how soil conditions, ground shaking and structural strength influence a building's ability to withstand seismic activity.
Led by Dr Jawed Qureshi, Senior Lecturer in Structural Engineering and Design, the study tested a computer model of a five-storey reinforced concrete building using recorded ground-motion data from the devastating Northridge and Kobe earthquakes in 1994 and 1995, drawing on four specialist engineering technologies to assess everything from soil conditions to structural strength in a single coordinated workflow. The research is published in Buildings journal.
Understanding what happens when the ground shakes
Designing buildings to withstand earthquakes is a complex challenge. Engineers must consider not only the strength of an earthquake but also the ground beneath a building, how the structure moves and whether its individual components can withstand the forces involved. These factors are often assessed using separate engineering tools, making it difficult to understand how they interact and influence a building's overall performance.
The UEL study explored a framework described as Proto-Artificial General Intelligence (Proto-AGI), which brings together four specialist technologies under human supervision to provide a more complete assessment of structural risks.
Dr Qureshi said:
“Structural engineers have always known that buildings do not fail in isolation. What happens beneath the ground affects how a structure moves, and how a structure moves determines whether it survives. Now, for the first time, we have a framework that joins all those pieces together.”
Lessons from two major earthquakes
To demonstrate the approach, the researchers used recorded ground motions from the 1994 Northridge earthquake in California and the 1995 Kobe earthquake in Japan, two of the most destructive earthquakes of the twentieth century.
They applied the earthquake records to a computer model of a five-storey reinforced concrete building to examine how it would respond.
The results revealed that although the Northridge record produced approximately 3.4 times the median seismic demand of the Kobe record at the building's natural period, the difference in peak structural movement was only 1.32 times.
The comparison shows that the relationship between ground shaking and structural movement is not straightforward. A larger difference in the shaking a building experiences does not necessarily translate into an equally large difference in its movement.
The characteristics of the earthquake, the soil beneath the building and the structure’s design all influence the outcome, which is why stronger shaking does not necessarily produce a proportionately greater amount of building movement.
Examining these factors together could help engineers understand how vulnerabilities develop across a structure, rather than considering each part of the assessment in isolation.
Could AI change the future of earthquake engineering?
The research explores how AI-assisted technologies could improve the way engineers assess buildings and infrastructure.
Rather than replacing engineers, the proposed approach brings digital tools together, with human specialists overseeing the process and interpreting the results.
This coordinated approach could provide a foundation for more advanced engineering systems, combining information from different disciplines to build a clearer understanding of structural performance.
The findings also highlight how digital technologies could help the construction industry develop infrastructure better able to withstand natural hazards.
Further testing is needed before the approach could be used routinely in engineering practice.
The initial study used one soil profile and two earthquake records, with engineers manually coordinating information between the technologies. Future research could explore larger earthquake datasets, greater automation and information from building inspections and structural monitoring.
Dr Qureshi added:
“The goal was never to replace engineering judgement with AI. It was to give engineers a clearer, more complete picture before they make their decisions. That is what this research delivers.”
The research, led by Dr Jawed Qureshi and co-authored with Bala Karthika Balakrishnan, is published in Buildings journal and freely available at https://doi.org/10.3390/buildings16163324
Journal
Buildings