Disaster response aided by technology

In the wake of the devastating Pueblo-Morelos earthquake in September 2017, a team of SOM engineers traveled to Mexico City to contribute to post-disaster recovery efforts. Shortly after the 7.1-magnitude earthquake struck, structural engineers based in our San Francisco and Los Angeles offices set out to document building damage and provide technical support to local structural reconnaissance efforts. 

One of the first international teams of engineers on the ground in Mexico, the group documented building damage patterns before the crucial work of cleanup and rebuilding began. They also assisted local officials in assessing critical and significantly damaged structures. As part of the documentation, the group took hundreds of photos that were then shared with the Earthquake Engineering Research Institute (EERI) database in support of the institute’s mission to mitigate earthquake risk around the world. Each photo included the notation of specific findings including damage observed, the severity of the damage, and whether it was structural or non-structural. Transforming what is traditionally a time consuming, manual process, the group trained a machine learning model to identify areas of damage.

SOM had organized reconnaissance​ teams in the aftermath of several major earthquakes before, including the 1985 Mexico City earthquake. On each mission, the information gleaned from documentation, combined with the research conducted by numerous organizations and professionals involved in post-disaster recovery efforts, is intended to help cities become more resilient against seismic events in the future. Transforming what is traditionally a time consuming, manual process, the group trained a machine learning model to identify areas of damage.

Beyond earthquakes: Everyday applications for machine learning

Building on the successful machine learning effort in Mexico, SOM has since explored other uses for the technology for the assessment of buildings in all stages of construction, and even long after.

The engineering group has trained AI models to recognize signs of change over time so that clients can identify a potential slow-developing deterioration. Through this process, an aging building can be protected, documented, and then analyzed for any possible problems. Algorithms can take an image and transform it into a construction drawing, allowing contractors to make any necessary adjustments. Inspections can take place remotely for building sites across the world. Drone photography or laser scanning can assist with the mapping of an entire structure, not only to make any necessary changes but to give clients a better idea of structural components that are not otherwise visible. Shop or design drawings can then be carefully compared to what’s on site, and then monitored. 

Through the experience in Mexico City, SOM has now applied machine learning tools for buildings in need of construction verification as well as buildings long since completed.