Mohamed Rusthi, Poologanathan Keerthan, Mahen Mahendran and Anthony Ariyanayagam
This research was aimed at investigating the fire performance of LSF wall systems by using 3-D heat transfer FE models of existing LSF wall system configurations.
Abstract
Purpose
This research was aimed at investigating the fire performance of LSF wall systems by using 3-D heat transfer FE models of existing LSF wall system configurations.
Design/methodology/approach
This research was focused on investigating the fire performance of LSF wall systems by using 3-D heat transfer finite element models of existing LSF wall system configurations. The analysis results were validated by using the available fire test results of five different LSF wall configurations.
Findings
The validated finite element models were used to conduct a parametric study on a range of non-load bearing and load bearing LSF wall configurations to predict their fire resistance levels (FRLs) for varying load ratios.
Originality/value
Fire performance of LSF wall systems with different configurations can be understood by performing full-scale fire tests. However, these full-scale fire tests are time consuming, labour intensive and expensive. On the other hand, finite element analysis (FEA) provides a simple method of investigating the fire performance of LSF wall systems to understand their thermal-mechanical behaviour. Recent numerical research studies have focused on investigating the fire performances of LSF wall systems by using finite element (FE) models. Most of these FE models were developed based on 2-D FE platform capable of performing either heat transfer or structural analysis separately. Therefore, this paper presents the details of a 3-D FEA methodology to develop the capabilities to perform fully-coupled thermal-mechanical analyses of LSF walls exposed to fire in future.
Details
Keywords
Irindu Upasiri, Chaminda Konthesingha, Anura Nanayakkara and Keerthan Poologanathan
Elevated temperature material properties are essential in predicting structural member's behavior in high-temperature exposures such as fire. Even though experimental…
Abstract
Purpose
Elevated temperature material properties are essential in predicting structural member's behavior in high-temperature exposures such as fire. Even though experimental methodologies are available to determine these properties, advanced equipment with high costs is required to perform those tests. Therefore, performing those experiments frequently is not feasible, and the development of numerical techniques is beneficial. A numerical technique is proposed in this study to determine the temperature-dependent thermal properties of the material using the fire test results based on the Artificial Neural Network (ANN)-based Finite Element (FE) model.
Design/methodology/approach
An ANN-based FE model was developed in the Matlab program to determine the elevated temperature thermal diffusivity, thermal conductivity and the product of specific heat and density of a material. The temperature distribution obtained from fire tests is fed to the ANN-based FE model and material properties are predicted to match the temperature distribution.
Findings
Elevated temperature thermal properties of normal-weight concrete (NWC), gypsum plasterboard and lightweight concrete were predicted using the developed model, and good agreement was observed with the actual material properties measured experimentally. The developed method could be utilized to determine any materials' elevated temperature material properties numerically with the adequate temperature distribution data obtained during a fire or heat transfer test.
Originality/value
Temperature-dependent material properties are important in predicting the behavior of structural elements exposed to fire. This research study developed a numerical technique utilizing ANN theories to determine elevated temperature thermal diffusivity, thermal conductivity and product of specific heat and density. Experimental methods are available to evaluate the material properties at high temperatures. However, these testing equipment are expensive and sophisticated; therefore, these equipment are not popular in laboratories causing a lack of high-temperature material properties for novel materials. However conducting a fire test to evaluate fire performance of any novel material is the common practice in the industry. ANN-based FE model developed in this study could utilize those fire testing results of the structural member (temperature distribution of the member throughout the fire tests) to predict the material's thermal properties.