Elsevier

Construction and Building Materials

Volume 215, 10 August 2019, Pages 192-206
Construction and Building Materials

Properties and material models for common construction materials at elevated temperatures

https://doi.org/10.1016/j.conbuildmat.2019.04.182Get rights and content

Highlights

  • A novel framework for deriving temperature-dependent material models is presented.

  • The framework leverages AI to understand material behavior under extreme conditions.

  • ANN and GP are used to develop unified constitutive material models.

  • The AI-derived models can modernize and standardize fire design of structures.

Abstract

Construction building materials experience physio-chemical and phase changes when subjected to elevated temperatures. These changes are often defined through temperature-dependent material models. A cross examination of adopted models reveals that such models markedly varies across open literature and fire guides (i.e. ASCE, Eurocodes etc.). This, not only complicates the process of fire analysis and design, but can also hinders ongoing standardization initiatives. In support of these initiatives, this paper leverages symbolic regression through artificial neural networks (ANN) and genetic programming (GP) to arrive at representative temperature-dependent thermal and mechanical material models for common building materials, namely: normal strength concrete, masonry, structural steel, stainless steel, cold-formed steel and wood. The proposed material models have the potential to regulate and modernize structural design under extreme loading conditions, i.e. fire. The result of this investigation demonstrates the value of utilizing artificial intelligence (AI) into comprehending the complex nature of temperature-induced effects on building materials; together with deriving associated temperature-dependent models.

Introduction

Civil constructions are to be designed to satisfy codal requirements. One such requirement is to withstand extreme events (i.e. fire/thermal loading). The ability of a structure to withstand fire and associated fire-induced forces is highly reliant on, 1) the type of material the main structural members/components are composed of, and 2) how properties of such materials are influenced by elevated temperatures [1]. This ability is often measured through experimental fire testing or, for the most part, through fire resistance evaluation. In such evaluation, thermal and mechanical characteristics of construction materials are of interest as fire resistance assessment requires carrying out a two-step analysis; thermal and structural. In the first step, rise in temperature and associated temperature propagation in a load bearing member are obtained by examining how density, thermal conductivity, and specific heat properties fluctuate with increasing temperatures. Once sectional temperatures are obtained, these are then loaded into the second step of analysis. In this step, the adverse effect of increasing temperature upon mechanical properties of construction materials; primarily comprising of strength, and modulus, is considered in evaluating the assessing behavior of a fire-exposed member [1], [2].

Thus, carrying out a proper fire resistance analysis requires thorough knowledge of thermal and mechanical properties at ambient and fire conditions. While evaluating aforementioned material properties at ambient conditions can be achieved with ease mainly due to the availability of testing standards and instrumentations, assessing properties of building materials at high temperatures is shown to be a tedious task [3]. This can be primarily ascribed to the current lack of expertise and/or standardized (i.e. agreed upon) guidance, shortage and limited access of testing equipment, and most of all complexities arising from simulating fire conditions (i.e. survivability of sensors as well as ensuring proper measurements under elevated temperatures) [4]. As a result, fewer studies managed to successfully conduct material property tests under high temperatures as compared to that at ambient conditions [5].

The outcome of these studies presented the adverse effects of fire on material properties through temperature-dependent material models which were either prepared into simplified expressions, or informative charts [6], [7]. Despite the fact that most of these research studies were carried out in 1960-90’s, the outcome of such studies continues to form the basis for currently adopted temperature-dependent material models [6], [7]. A close examination of these models exposes discrepancies arising from different testing methods and equipment, specimen configuration etc. used in high temperature tests [5], [8]. Another factor that adds further complexities is the existence of distinct variations in the make-up (composition) of construction materials tested in the 1960–90s and those available today as a result of the natural progression in materials science, differences in origin/amount/type of additives, as well as from modern production/milling procedures.

A review of literature also shows that this community has accepted two material models to be used in fire resistance assessment. For the most part, these models are adopted in North America (i.e. American Society of Civil Engineers (ASCE) design guide [6]), and Europe (Eurocodes [7], [9], [10], [11]). Despite the fact that these two models have been widely used, recent works have shown that variation in fire resistance predictions can be in the range of 25% when using temperature-dependent models adopted by ASCE or Eurocodes [5], [12]. This often complicates structural evaluation at elevated temperatures; particularly in the analysis/design for compound load effects (viz. torsion, or buckling etc.), selecting appropriate fire protection material/type/thickness, or even in carrying out design/engineering (consulting) services.

Such variation arises due to a number of reasons. For a start, these codal-adopted models imply that the micro-structure of construction materials is independent of its fabrication process, or composition/origin. Further, these models were developed using vintage devices, which are inferior to the modern and state-of-the-art equipment, and thus provided scientists with restricted testing set-ups, and possibly mediocre measurements [5], [12]. Furthermore, these material models continue to be not updated since their development; dating back to 20–30 years. Finally, while Eurocode 3 suggests the use of certain models to represent behavior of contemporary construction materials (i.e. stainless steel and cold-formed steel), ASCE design guide does not provide direction nor insights into how to account for temperature-dependent effects in materials; leaving designers with limited guidance, hence complicating the process of fire resistance evaluation.

From this work’s perspective, it is infeasible to regularly conduct temperature-dependent tests on materials – given the variety in compositions, origins and mixes. Thus, a dilemma arises highlighting the need for a uniform and modern model for construction materials at elevated temperatures. With the hope of overcoming this challenge, and in support of current inertia aimed at promoting standardization for fire resistance assessment, this study aims at utilizing Artificial Intelligence (AI) to develop modern and updated temperature-dependent models for commonly used construction materials – and this could be the first step towards realizing uniform (universal/standardized) constitutive material models. As such, this work presents a novel approach to develop temperature-dependent thermal and mechanical material models for some of the commonly used building materials, namely: normal strength concrete, masonry, structural steel, stainless steel, cold-formed steel and wood. This study starts by presenting high temperature properties of common construction materials and then showcases a proper procedure to developing an AI model capable of deriving temperature-dependent material models. To ensure precision and wide acceptance, the developed AI model integrates material models adopted by notable fire codes, standards and design guides, together with models collected from past and recent published studies/reports.

Section snippets

Temperature-dependent properties of common construction materials – an overview

Response of structures once exposed to elevated temperatures is largely a function of properties of building materials. Under such effects, thermal and mechanical1 properties fluctuate with temperature mirroring the series of phase changes that occur. As such, this section delivers a brief overview on the

A Note on high temperature material tests

A comparison between various material models plotted in Fig. 1, Fig. 2, Fig. 3, Fig. 4 shows the large variations in measured properties of common building materials. This variation is not only apparent when comparing test data points carried out in 1950s-70s with those carried out in late 1990s-2000s, but also in between data points obtained from tests within the same era. For example, the early works of Powell [21], Yawata [22], and Touloukain [23], which were carried out in 1960s-70s, did

Development of artificial intelligent model

In lieu of traditional deterministic methods, artificial intelligence (AI) simulates the reasoning process through cognitive layers arranged in a specific layout (e.g. Artificial Neural Network (ANN)). As such, ANNs are made of constitutive layers and accompanying neurons (processing units) as shown in Fig. 5. Each layer has a number of neurons that is largely contingent on the complexity level of arriving at a relation between inputs and expected outcome. On the far left side, the input layer

Deriving AI-based temperature-dependent material models

As discussed above, all data points (in terms of measured properties or codal-adopted models i.e. ASCE, Eurocodes, AS 4100, AISC, BS5950 and so on), as well as those measured through researchers as plotted in Fig. 1, Fig. 2, Fig. 3, Fig. 4.) for commonly used construction materials were input into the ANN. More specifically, temperature-dependent reduction factors of modulus property of concrete were first collected at specific temperatures i.e. 25, 100 °C etc. (as shown in Fig. 4a) [95]. It is

Practical implications and future research plans

The discussion outlined in Secs. 1–3 illustrates how the result of a given fire resistance evaluation is highly sensitive to the selected material model. As the fire response of a structural member is not known beforehand, then fire resistance predictions are often viewed with caution. In such a scenario, fire designers often face a challenge as to 1) what material model to select for fire assessment, 2) justification to select such model over others. As a result, it is common for

Conclusions

This study showcases a framework to develop temperature-dependent thermal and mechanical material properties of common building materials including normal strength concrete, masonry, structural steel, stainless steel, cold-formed steel and wood. This approach applies artificial intelligence in two forms, i.e. Artificial Neural Networks (ANNs) and Genetic Programming (GP) to achieve modern temperature-dependent material models.

These following conclusions could also be gathered from the outcome

Acknowledgment

The author would also like to thank Prof. Venkatesh K. Kodur for his continuous support.

Data availability

The raw data required to reproduce these findings are available to download from [www.mznaser.com/fireassessmenttoolsanddatabases].

Conflict of interests

The author declares no conflict of interest.

References (101)

  • Z. Xiao et al.

    Properties of partition wall blocks prepared with high percentages of recycled clay brick after exposure to elevated temperatures

    Constr. Build. Mater.

    (2013)
  • G.M. Cooke

    An introduction to the mechanical properties of structural steel at elevated temperatures

    Fire Saf. J.

    (1988)
  • T. Ranawaka et al.

    Experimental study of the mechanical properties of light gauge cold-formed steels at elevated temperatures

    Fire Saf. J.

    (2009)
  • J. Chen et al.

    Experimental investigation of cold-formed steel material at elevated temperatures

    Thin-Walled Struct.

    (2007)
  • M. Naser et al.

    Analysis of RC T-beams strengthened with CFRP plates under fire loading using artificial neural

    Networks, Constr. Build. Mater.

    (2012)
  • S. Mandal et al.

    Artificial neural network modeling to evaluate and predict the deformation behavior of stainless steel type AISI 304L during hot torsion

    Appl. Soft Comput.

    (2009)
  • M.Z. Naser

    Properties and material models for modern construction materials at elevated temperatures

    Comput. Mater. Sci.

    (2019)
  • M.Z. Naser

    Autonomous and resilient infrastructure with cognitive and self-deployable load-bearing structural components

    Autom. Constr.

    (2019)
  • M.Z. Naser et al.

    Cognitive infrastructure-a modern concept for resilient performance under extreme events

    Autom. Constr.

    (2018)
  • A.H. Buchanan et al.

    Structural design for fire safety

    (2017)
  • J.M. Franssen et al.

    Designing steel structures for fire safety

    (2009)
  • V.K. Kodur et al.

    Properties of building materials. InSFPE handbook of fire protection engineering

    (2016)
  • V.K. Kodur et al.

    Structures in fire: state-of-the-art, research and training needs

    Fire Technol.

    (2012)
  • V. Kodur et al.

    High-temperature properties of steel for fire resistance modeling of structures

    J. Mater. Civ. Eng.

    (2010)
  • Structural Fire Protection – ASCE Manuals and Reports on Engineering. Practice No. 78, New York, NY,...
  • Eurocode 3, Design of steel structures, Part1-2: General rules-structural fire design, Document CEN, European Committee...
  • Eurocode 2, Design of concrete structures, Part1- 2: General rules-structural fire design, ENV 1992-1-2, Document CEN,...
  • Eurocode 6: Design of masonry structures—Part 1-2: General rules - Structural fire design for reinforced and...
  • Eurocode 5. Design of Timber Structures. Part 1–2 General. Structural Fire Design....
  • L. Choe et al.

    Influence of material models on predicting the fire behavior of steel columns

    Fire Technol.

    (2017)
  • G.A. Khoury

    Compressive strength of concrete at high temperatures: a reassessment

    Mag. Concr. Res.

    (1992)
  • A.M. Neville

    Properties of concrete

    (1995)
  • V. Kodur

    Properties of concrete at elevated temperatures

    ISRN Civ. Eng.

    (2014)
  • T.Z. Harmathy

    Thermal properties of concrete at elevated temperatures

    J. Mater.

    (1970)
  • Society of Fire Protection Engineers (SFPE). Handbook of fire protection engineering, 4th Ed., Cleveland,...
  • B. Fredlund A model for heat and mass transfer in timber structures during fire: a theoretical, numerical and...
  • R.W. Powell et al.

    High alloy steels for use as a thermal conductivity standard

    Br. J. Appl. Phys.

    (1960)
  • Yawata Iron and Steel Co. WEL-TEN 80 material datasheet Tokyo, 22...
  • Y. Touloukian

    Thermal radiative properties for non-metallic solids

    Therm. Phys. Prop.

    (1972)
  • D. Bentz K. Prasad. Thermal performance of fire resistive materials I. Characterization with respect to thermal...
  • Knudson RM, Schniewind AP. Performance of structural wood members exposed to fire. Forest Products Journal (USA). 1975....
  • M.L. Janssens et al.

    Temperature profiles in wood members exposed to fire

    Fire Mater.

    (1994)
  • J.J. Fuller et al.

    Temperature distribution in a nailed gypsum-stud joint exposed to fire

    Fire Mater.

    (1992)
  • J Konig, L. Walleij Timber frame assemblies exposed to standard and parametric fires, Part 2: a design model for...
  • B. Fredlund A model for heat and mass transfer in timber structures during fire: a theoretical, numerical and...
  • A. Frangi Brandverhalten von Holz-Beton-Verbunddecken. ETH Zurich; 2001....
  • G.C. Thomas

    Fire resistance of light timber framed walls and floors

    (1996)
  • S. Hadvig

    Charring of wood in building fires

    (1981)
  • J.R. Mehaffey et al.

    A model for predicting heat transfer through gypsum-board/wood-stud walls exposed to fire

    Fire Mater.

    (1994 Sep)
  • BW Gammon. Reliability Analysis of Wood-Frame Wall Assemblies Exposed to Fire, Berkeley, California: University...
  • Cited by (39)

    View all citing articles on Scopus
    View full text