Computational Approaches to Fire-structure Interaction and Real-time Fire Monitoring

Abstract

Structural fires in buildings are a persistent hazard in modern infrastructure and a potential threat to occupants and firefighters alike. In the pre-construction phases of building design, there has been a growing interest in the use of simulated natural fires as a direct input for modeling the fire-structure interaction problem. One common method to simulate the natural fire is through the use of computational fluid mechanics (CFD), where the fire development is treated as a fluid-flow problem. The structure is typically modeled using finite element analysis (FEA), as in other applications of structural engineering. However, the link between the fire and solid domains has not been standardized for structural fire engineering purposes in research or practice. In the post-construction phases of the building lifecycle, for example, fire safety engineering is used to reduce the potential for fire and minimize threats to building occupants. While there have been many improvements in materials and design to attempt to limit the ignition of new fires in buildings over the last several decades, the fact remains that fire events still pose significant risks to firefighters. With this in mind, methods must be developed to provide new technology and solutions for the modern firefighter by using advances in computation and visualization. In this dissertation, modeling the fire-structure interaction problem and providing a real-time system for fire monitoring are the two main focuses. Both contributions serve to improve the two different ends of the structural fire spectrum: research, analysis, and design on one end and sensor-assisted firefighting on the other. First, to address the problem of modeling the fire-structure interaction, an approach is provided which used CFD-based boundary conditions from a fire simulation as input for the FEA-based model of the structure. The main components of this research are (i) the development of fire-to-structure coupling methods linking the fluid and solid domains, (ii) the extension of a layered thermal shell element to include mechanical degrees of freedom (DOF) and provide a coupled thermo-mechanical shell element, and (iii) the use of these two contributions together to analyze structures exposed to local fires. The trapezoidal rule for numerical integration was used to represent spatially non-uniform heat fluxes computed in the CFD fire simulation as equivalent nodal fluxes in the FEA model of the structure. The thermo-mechanical shell element was coupled from its individual formulations using virtual work methods and ensuring consistency in the use of the layered representation of the governing equations for conduction heat transfer and structural deformation in the element. Second, a proposed system for real-time fire monitoring in the post-ignition fire state of a building was developed to improve the technology of sensor-assisted firefighting from a computing perspective. The software system was designed to coordinate various data streams from a simulated wireless sensor network (WSN) to sub-models responsible for performing real-time calculations using the data measured by sensors. An event detection model for assessing smoke toxicity, burn threats, and fire spread was implemented as the first sub-model and its real-time performance was analyzed. Results from the event detection model were used to present an example of visualization using a building information model (BIM) as the platform for communicating hazard warnings to the incident commander and firefighters in this study.