Continuous Real-Time Hybrid Simulation Method for Structures Subject to Fire

Traditional prescriptive methods for structural fire design, focusing on individual member failure criteria and ignoring system-level performance, are being replaced by performance-based design. Performance-based methods consider system-level performance and actual fire hazards, assessed using computer simulations like Hazard I, FIRECAM, and FDS. While these tools effectively assess fire hazards, predicting structural fire performance and providing design guidelines remains challenging. Numerical models can analyze structural fire performance but struggle to accurately model material degradation at elevated temperatures and changing boundary conditions. Full-scale tests offer system-level validation but are expensive and limited. Hybrid fire simulation, combining physical testing of critical structural components with numerical modeling of the remainder, reduces uncertainty and allows for more comprehensive system-level investigation. Previous hybrid fire tests focused on individual components or smaller-scale structures. This research addresses the need for a robust method capable of simulating large-scale structures under fire conditions.

The paper reviews existing literature on structural fire design methods, highlighting the limitations of prescriptive methods and the advantages of performance-based approaches. It discusses the use of numerical models and the challenges in accurately modeling material behavior at elevated temperatures. The limitations of full-scale fire testing due to cost and resource constraints are noted. Existing literature on hybrid fire simulation methods is also reviewed, citing previous studies that used this approach for various structural components but lacked the robustness and scale of the proposed method. The authors mention previous work using a ramp-and-hold method and its drawbacks as motivation for the current research.

The proposed continuous hybrid fire-simulation method uses the University of Toronto Simulation Framework (UT-SIM) and a LabView-based network interface (NICON) for actuator control. A continuous real-time algorithm with error compensation accounts for testing setup deformation. The ISO 834 fire curve is used, but the method is adaptable to parametric fire curves. A 3D numerical model considers convection, conduction, and radiation for directly exposed members, while a simplified 1D model considers only conduction for indirectly exposed members. Nonlinear structural analysis uses a modified Newton-Raphson method without iterations within a time step to maintain synchronization with the furnace temperature. The continuous method uses polynomial extrapolation and interpolation (third-order Lagrange polynomial) to generate displacement commands at 100 Hz, overcoming limitations of the ramp-and-hold method (force drop during hold, unbalanced force from continuous thermal expansion, accuracy dependence on increment size, and limitations on numerical model complexity). An error-compensation scheme continuously compensates for loading-frame deformation using measured force and predefined frame stiffness. The method's stability is analyzed using a single-degree-of-freedom (SDOF) system, showing that stability depends on the ratio of specimen stiffness to actual loading frame stiffness (βs) and the ratio of estimated to actual loading frame stiffness (βf). Stability is ensured when βf is slightly greater than 1. A multiresolution numerical model with detailed 3D modeling of the RBS connection and simplified frame elements for the rest of the structure is employed. The experimental setup involves heating a physical column specimen in a furnace, recording temperature history, and measuring actuator displacement and force. Preliminary tests determine thermal boundary conditions, loading-frame stiffness, and validate the error-compensation scheme. The main test involves a hybrid fire simulation with gravity and thermal loads, comparing results with numerical simulation and a previous ramp-and-hold test.

The continuous hybrid fire simulation method demonstrated improved accuracy and robustness compared to the previous ramp-and-hold method. The continuous method allowed for longer time steps and more complex FE models in the numerical substructure. Accurate compensation for loading system deformation requires careful evaluation of the loading frame stiffness. The ratio between the estimated and actual loading frame stiffness significantly influences the system stability. Force fluctuations observed in the ramp-and-hold method were eliminated. The use of a slightly higher initial stiffness in the error-compensation scheme ensured stability but could introduce a small displacement error. Ideal behavior requires a linearly elastic loading frame and accurate stiffness estimation. The multiresolution numerical model, incorporating a detailed RBS connection model, effectively captured the structural behavior. Results from the continuous hybrid simulation showed a closer match to the experimental data than the numerical simulation, particularly in the plastic deformation regime. The continuous method proved effective at simulating the complex behavior of a steel frame under fire loading, accurately predicting displacement and force histories throughout the heating and cooling phases.

The study successfully addressed the research question by developing and validating a robust continuous hybrid fire simulation method. The findings demonstrate the method's superiority over previous approaches, offering enhanced accuracy and efficiency. The method's ability to handle complex numerical models and longer time steps is significant, expanding the applicability of hybrid simulation to larger and more realistic structures. The detailed analysis of stability and error compensation provides practical guidance for conducting such simulations. The successful validation using a large-scale experiment strengthens the method's credibility. The results contribute to the advancement of performance-based fire design by providing a more accurate and reliable tool for predicting structural behavior under fire conditions. The findings have implications for improving fire safety design codes and practices.

This study successfully improved upon the existing ramp-and-hold hybrid fire simulation method by introducing a continuous real-time approach. The continuous method offers greater accuracy and efficiency, enabling the use of more complex numerical models and longer time steps. Careful evaluation of the loading frame stiffness is crucial for stability and accurate error compensation. The use of a multiresolution numerical model successfully captured the complex behavior of the steel frame under fire conditions. This improved method enhances the accuracy and reliability of performance-based fire design, leading to safer structures.

The study's focus on an idealized structure with a simplified fire scenario limits the generalizability of the findings. The assumption of heat conduction as the primary heat transfer mechanism may not be universally applicable. The limitation of the testing apparatus to a single degree of freedom restricts the simulation of more complex structural behaviors and fire scenarios. The nonlinear elastic behavior of the loading frame introduces some uncertainty in the error-compensation scheme. Future research could address these limitations by expanding the scope to more complex structures, fire scenarios, and multiple degrees of freedom.