‘AI predicts the risk level of hazardous gases underground’: KICT develops management system

[Financial News] The Korea Institute of Civil Engineering and Building Technology said on the 2nd that it had developed an “Underground Environmental Risk Management System” that measures hazardous gases and oxygen levels in real time at underground construction sites and uses artificial intelligence (AI) to predict changes in risk.
According to KICT, carbon monoxide, hydrogen sulfide, volatile organic compounds and dust can be generated at underground construction sites during excavation and blasting, the operation of diesel equipment, welding, and waterproofing and painting work. In underground spaces where ventilation is limited, hazardous gases can spread quickly and oxygen levels can fall, making it important to continuously monitor working conditions and respond to risks proactively.
To address this, a research team from KICT’s Geotechnical Research Division combined Internet of Things (IoT) sensors with AI prediction technology, integrating real-time data collection and analysis, risk prediction, estimation of gas-generation locations, and alerts into a single system.
The system collects information in real time on hazardous-gas concentrations measured at the site, temperature and humidity, and airflow, and analyzes it using AI. By predicting the movement paths of hazardous gases and changes in their concentrations, it has expanded safety management—previously centered on measurement and alarms—to focus on prediction and response.
The system can also estimate gas concentrations in areas without sensors and predict the direction in which hazardous gases will spread and where they originate. The research team tested the system’s performance in indoor experiments replicating various environments, as well as in actual large-diameter conduits (large conduits) and culvert-type tunnels (underground passage-type tunnels). The tests confirmed that measurements from just two sensors could be used to estimate the distribution of gas concentrations at 14 locations, and that the section where gas was generated could be identified based on information collected for approximately 30 seconds after gas generation. The system was also designed to use data from different types of sensors together, establishing the potential to link it with existing field measurement equipment.
KICT President Park Sun-kyu said, “This technology is significant because it identifies risks in underground working environments that are invisible to the eye in advance, giving workers time to respond,” and added, “Starting with tunnels, we will apply it to a range of underground spaces and contribute to preventing serious accidents at construction sites.”

[email protected] Yeon Ji-an Reporter