Research Projects

The lack of a robust framework for evaluating and approving building plans and layouts in the context of fire safety poses significant risks to occupants and first responders. Inadequate evaluation processes result in buildings with critical design flaws that compromise their ability to facilitate safe evacuation during a fire emergency. A lack of thorough evaluation can lead to non-compliance with established fire safety standards, allowing substandard buildings to be built and putting occupants at risk in the event of a fire. This problem can be addressed by implementing a comprehensive building plan evaluation and approval framework that identifies fire safety risks, ensures enforcement of building codes and regulations, and incorporates standardized processes for fire safety assessments. In this context, this study will develop an Artificial Intelligence (AI)-based building plan evaluation and approval framework that can identify fire safety loopholes in design. Knowledge is scarce in the literature about how building plans can impact the evacuation time and effectiveness of emergency response. No study so far has developed a predictive model to estimate the available safe egress time (ASET) and required safe egress time (RSET) in a residential/commercial/hospital building based on the architectural layout and other relevant design parameters. This study will aim to fill this research gap by developing an AI-based multi-task learning framework that can accurately predict the ASET and RSET in buildings with various occupancy types, design plans, and architectural layouts. It may be noted in this context that ASET is the time available from the ignition of fire till the point when the condition becomes untenable for evacuation due to smoke, temperature, or toxic gases. On the other hand, RSET refers to the time required for all occupants in a building to safely evacuate after the detection of a fire. A building is considered fire-safe if the ASET is larger than γ times the RSET, where γ is a pre-determined safety factor that typically lies between 1.5 and 2. The training data required for ASET and RSET predictions will be generated synthetically using a fire dynamics simulator and an egress modeling software, respectively. Additionally, a very simple user interface will be developed which can be conveniently used to quickly check the ASET and RSET for a given building plan and thereby assess fire safety. It will also recommend necessary adjustments to the plan if it fails to meet the fire safety requirements. This tool can be utilized by statutory approving authorities to sanction design plans and verify compliance. It will also assist the structural engineers and architects in the fire-safe planning and design of buildings. Overall, it will minimize fire-related risks in buildings and enhance the safety of occupants and first responders.