Construction and Building Materials 479 (2025) 141259

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

A multiscale model for predicting the Young’s modulus and the

thermal-expansion coefficient of concrete at high temperatures
∗
Simon Peters , Giao Vu, Günther Meschke
Institute for Structural Mechanics, Ruhr University Bochum, Universitätsstraße 150, 44801 Bochum, Germany

ARTICLE INFO ABSTRACT

Keywords: A semi-analytical micromechanical model is proposed to predict the evolution of the Young’s modulus and
Concrete at high temperatures thermal-expansion coefficient of concrete at elevated temperatures, considering three scales of observation,
Micromechanics namely cement paste, mortar and concrete. After validation with various experimental data sets, the model
Thermal-expansion coefficient
quantifies different sources of damage to concrete at elevated temperatures, indicating that the chemical
Modulus of elasticity
decomposition of cement paste has a minor influence on the evolution of the Young’s modulus, while the
Load-induced thermal strains
thermal degradation of the aggregate plays a major role. At higher temperatures, cracking is the main
mechanism driving the reduction of the Young’s modulus of the investigated concretes.
With regard to the thermal-expansion coefficient evaluated at multiple scales, load-induced thermal
strains highly effect the homogenized total strains at the mortar level. Moreover, it is demonstrated that the
dehydration degree of C-S-H increases proportionally with the measured load-induced thermal strains.

1. Introduction the correlation between the basal spacing (the distance between C-S-
H layers [20–22]) and Young’s modulus increases exponentially from
Despite extensive research in the last two decades, the exact mech- 70–115 GPa at low basal spaces (around 11–9 Å, respectively) and
anisms of explosive concrete spalling at elevated temperatures are still decreases just slightly between 11–14 Å from 68 GPa–60 GPa [23,24].
not fully understood [1,2]. One open research question is about the DeJong and Ulm conclude that partial dehydration due to heating
driving forces of thermal degradation of concrete at elevated temper- induces gradual recondensation of hydroxyl groups, structural disor-
atures. Three schools of thoughts can be identified on this matter. dering and volumetric shrinkage in the C-S-H grains. At low Ca/Si
The first school of thought assumes that microcracking induced by the molar ratios of the C-S-H these gradual recondensation effects are less
thermal mismatch of aggregates and cement paste dominates the irre- pronounced [23].
versible loss of concrete strength and stiffness at elevated temperatures. To quantify the chemo-thermo-mechanically induced concrete
Meso-scale investigations show that radial and lateral cracks at the degradation at elevated temperatures on multiple scales, several mul-
aggregate-matrix interfaces are growing during thermal loading [3–6]. tiscale continuum mechanics-based models were published [25,26].
However, this mechanical phenomenon plays a negligible role for ce- Lee et al. [27] proposed a stochiometric analysis of C-S-H and CH
ment paste degradation, where loss of elasticity [7,8] and strength [7,9] decomposition capturing the porosity increase according to a linear de-
can be measured at high temperatures. In this regard, the second hydration behavior. The influence of aggregate degradation, debonding
school of thought allocates this loss to the dehydration of the primary and interfacial cracks in the cement paste on the macroscopic concrete
hydration products (C-S-H, CH)1 in hardened cement paste [10,11] as stiffness is described phenomenologically. Zhao et al. [28] additionally
Harmathy demonstrated in his pioneering works [12,13]. considered aluminate phases in this stochiometric model and used
The third and more recent school of thought introduced by DeJong experimental dehydration kinetics of C-S-H to predict the capillary pore
and Ulm [14] attributes the thermal damage of hardened cement volume fraction of cement paste at elevated temperatures. In [29] an
paste to the thermally induced change of the packing density of C- effective medium approach was incorporated in the model to consider
S-H particles. This densification process was investigated further by separately thermo-chemically induced degradation processes and mi-
atomistic modeling simulations, see, e.g. [15–19]. It was shown, that crocracking effects on the evolution of the Young’s modulus of cement

∗ Corresponding author.
E-mail addresses: [email protected] (S. Peters), [email protected] (G. Vu), [email protected] (G. Meschke).
URL: https://www.sd.ruhr-uni-bochum.de (G. Meschke).
1
Cement chemistry abbreviations will be used in this paper: C = CaO; S = SiO; A = Al2 O3 ; F = Fe2 O3 ; S̄ = SO3 ; C̄ = CO2 ; H = H2 O.

https://doi.org/10.1016/j.conbuildmat.2025.141259
Received 19 September 2024; Received in revised form 6 April 2025; Accepted 7 April 2025
Available online 29 April 2025
0950-0618/© 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
S. Peters et al. Construction and Building Materials 479 (2025) 141259

paste. Zhang et al. [30] adapted this model to predict the macroscopic of the penny-shaped void inclusions, representing microcracks, remains
decrease of the Young’s modulus of fiber reinforced concrete at elevated unchanged, and the analysis proceeds to the next temperature step,
temperatures. In a recent analysis [31], the multiscale concepts were updating the material properties of both hardened cement paste and
extended to blended hardened cement pastes. However, load-induced aggregates accordingly. If the crack driving force exceeds the critical
thermal strains (LITS) were ignored in all the previously-mentioned energy-release rate, the microcrack length – corresponding to the radius
multiscale investigations. The main reasons for the limited focus on of the penny-shaped inclusions, increases – leading to a reduction in the
LITS are: (1) the physical mechanisms behind these quasi-instantaneous Young’s modulus of the hardened cement paste. The homogenization
strains are not fully understood [32], and (2) existing phenomeno- process is repeated at the same temperature increment by updating
logical LITS models [33–37], summarized in [38], are defined at the the Young’s modulus. The thermal-expansion coefficient of both mor-
macroscale. While Manzoni et al. [32] proposed a cement paste creep tar and concrete is altered according to the increased crack length,
model, it has not been applied in multiscale models or for temperatures subsequently reducing the mismatch stresses in the cement paste. The
above 100 ◦ C. described iterative process continues until the crack driving force no
LITS are quasi-instantaneous strains in the cement paste observed at longer exceeds the critical energy-release rate, ensuring a consistent
first heating under compressive loading. They are crucial for a reliable thermo-mechanical response of the material system.
assessment of the effect of thermal loads on concrete structures [39],
e.g. in case of fire. These strains can relax the induced stresses in a 2.1. Stress localization and homogenization
restrained concrete slab up to 100% when heated from 20 ◦ C to 100 ◦ C,
see [38,40] for a comprehensive survey of LITS. The hierarchical structure of the concrete, consisting of cement
The influence of LITS will be investigated in more detail in this paste with embedded penny-shaped cracks, fine aggregates and coarse
paper. Inspired by previous investigations [27–31], this paper pro- aggregates, is represented by a three-stage matrix-inclusion composite
poses a multiscale model based on semi-analytical Eshelby-type ho- model, each corresponding to a distinct scale of observation. Concrete is
mogenization schemes to calculate the Young’s modulus and thermal- modeled as a continuous mortar matrix with spherical coarse aggregate
expansion coefficient of concrete at elevated temperatures, quantifying inclusions, see Fig. 2(a). At the second level, mortar consists of a
the macroscopic thermal damage according to the different degradation continuous cement paste matrix containing fine aggregate (sand) inclu-
mechanisms. The following topics are mainly investigated: sions. At the smallest level, cement paste is idealized as a continuous
hydrate matrix with penny shaped void inclusions as applied in [41],
• Influence of the Young’s modulus of hardened cement paste on see [42,43] for further details.
the cracking process due to aggregate-matrix thermal-expansion The Mori–Tanaka scheme is well-suited for such matrix-inclusion
mismatch. composites, as it facilitates for the analytical homogenization of the
• Predominant temperature ranges associated with different schools thermoelastic properties of the composites. Each representative volume
of thoughts. element (RVE), occupying the domain 𝑉𝑅𝑉 𝐸 , is subdivided into a matrix
• Effect of aggregate type on Young’s modulus reduction of concrete phase and an inclusion phase, occupying the domains 𝑉𝑚 and 𝑉𝑖 ,
at elevated temperatures. respectively. Each material phase 𝑝 ∈ [𝑚, 𝑖] exhibits a specific elastic
• Impact of load-induced thermal strains on stiffness evolution and stiffness 𝐂 and an eigenstress 𝜎𝑝𝑒 with a known volume fraction, defined
thermal-expansion coefficient across multiple scales. as 𝑓𝑝 = 𝑉𝑝 ∕𝑉𝑅𝑉 𝐸 .
• Development of a micromechanical model applicable for macro- Isotropic linear elasticity is considered for all matrix and inclusion
scopic modeling. phases. Their elastic stiffness tensors C𝑝 are expressed in terms of their
bulk moduli 𝑘𝑝 and shear moduli 𝜇𝑝 . The homogenized stiffness tensor
2. Multiscale model reads as [44]

In this section, an Eshelby-based continuum micromechanics multi- Cℎ𝑜𝑚 = 𝑓𝑚 C𝑚 ∶ A𝑚 + 𝑓𝑖 A𝑖 ∶ A𝑖 , (1)

scale model for predicting the Young’s modulus of concrete at elevated and the homogenized eigenstresses as [45]
temperatures is proposed. As illustrated in Fig. 1, the multiscale model
𝑒
is divided into two parts based on the assumption of a one-sided 𝛴ℎ𝑜𝑚 = 𝑓𝑚 𝜎𝑚𝑒 ∶ A𝑚 + 𝑓𝑖 𝜎𝑖𝑒 ∶ A𝑖 , (2)
coupling of the degradation processes, i.e. assuming that chemical
where 𝑓𝑚 , 𝑓𝑖 are the volume fractions of the matrix and inclusion
dehydration is independent of cracking. In the first part, the Young’s
phase, respectively. The strain concentration tensors of the matrix
modulus evolution of cement paste at elevated temperatures is com-
A𝑚 and inclusion phase A𝑖 can be estimated using the Mori–Tanaka
puted utilizing a given heating rate and cement composition as model
scheme [46].
input. This calculation procedure is published in [31] and briefly
Assuming that the investigated composite material deforms freely
discussed in Section 3.1.
under temperature loading, the homogenized thermal expansion co-
In the second part of this study, an iterative homogenization pro-
efficient 𝛼ℎ𝑜𝑚 of the matrix inclusion composite can be computed
cedure is employed to determine the Young’s modulus and thermal-
analytically as [47]
expansion coefficient of mortar and concrete at elevated temperatures.
At each temperature increment, the homogenization of these properties 3𝑘𝑖 𝑘𝑚 (𝛼𝑚 𝑓𝑚 + 𝛼𝑖 𝑓𝑖 ) + 4𝜇𝑚 (𝛼𝑚 𝑓𝑚 𝑘𝑚 + 𝛼𝑖 𝑓𝑖 𝑘𝑖 )
𝛼ℎ𝑜𝑚 = . (3)
is performed using the Mori–Tanaka scheme (Eq. (1)), incorporating 3𝑘𝑖 𝑘𝑚 + 4𝜇𝑚 (𝑓𝑚 𝑘𝑚 + 𝑓𝑖 𝑘𝑖 )
the volume fractions and material properties of cement paste, fine The homogenization of concrete is performed through a three-step
aggregates, and coarse aggregates according to the specific concrete procedure. First, the effective properties of the cement paste, modeled
composition, see Section 2.1. Under the assumption of free thermal as a matrix-crack composite (Fig. 2c), are determined. In the second
expansion at the macroscopic concrete scale, a localization procedure step, mortar is homogenized as a matrix-inclusion composite consist-
translates the response from the concrete level to the cement paste ing of cement paste and fine aggregate inclusions (Fig. 2b). Finally,
level, enabling the computation of local stresses within each phase due the homogenized stiffness tensor and thermal expansion coefficient of
to mismatches in thermal-expansion coefficients, see Section 2.1. concrete are computed at the macroscopic scale (Fig. 2a). In the three
Subsequently, the local strain tensor in the hardened cement paste homogenization steps the properties of the representative stiffness of
is analyzed within a micro-poro-fracture mechanical framework to the matrix are incorporated, and Eqs. (1) and (3) are included as well.
calculate the crack driving force (see Sections 2.2 & 2.3). If the crack The material properties of the aggregate inclusions are provided in
driving force remains below the critical energy-release rate, the radius Section 3.3, while the stiffness of the cement paste matrix is derived

2
S. Peters et al. Construction and Building Materials 479 (2025) 141259

Fig. 1. Schematic illustration of the multiscale chemo-thermo-mechanical damage model of concrete.

Fig. 2. Multi-scale modeling of concrete: Two-dimensional sketches of three-dimensional representative volume elements: (a) Concrete composed of coarse aggregate inclusions
embedded within a mortar matrix, (b) mortar consisting of fine aggregate inclusions, embedded in a cement paste matrix, and (c) Cement paste with penny-shaped inclusions
embedded in a hydrate matrix.

based on its composition, see Section 3.1. The matrix properties at this leads to
both the mortar and concrete levels correspond to their respective
homogenized values. 𝛴𝑐𝑜𝑛 = 0 ⇒ 𝐄ℎ𝑜𝑚 𝟏 = 𝐄𝑒ℎ𝑜𝑚 = −C−1 𝑒
ℎ𝑜𝑚 ∶ 𝛴ℎ𝑜𝑚 = 𝛼ℎ𝑜𝑚 𝛥𝑇 𝟏 (8)
Since the stiffness and thermal-expansion coefficient of each con-
for the homogenized composite. Following further the derivation of
stituent are known, microscopic thermally induced stresses can be
[48], using the influence tensors in Eq. (5), the principal components
quantified by a localization procedure for a temperature change 𝛥𝑇 .
of the microscopic strain 𝜖 and stress tensors 𝜎𝑖 of the inclusions can
As suggested in [48,49], the volume averaged microscopic strains of
be calculated via
the matrix and the inclusions according to eigenstrains are computed
as 𝝐 𝑖 = 𝜖𝑖 𝟏 = (𝐴𝑖,𝑣𝑜𝑙 𝛼ℎ𝑜𝑚 + 𝐷𝑖𝑚,𝑣𝑜𝑙 𝛼𝑚 + 𝐷𝑖𝑖,𝑣𝑜𝑙 𝛼𝑖 )𝛥𝑇 𝟏,
(9)
𝜀𝑚 = A𝑚 ∶ 𝐄ℎ𝑜𝑚 + 𝐷𝑚𝑚,𝑣𝑜𝑙 𝛼𝑚 𝛥𝑇 I + 𝐷𝑚𝑖,𝑣𝑜𝑙 𝛼𝑖 𝛥𝑇 I, 𝝈 𝑖 = 𝜎𝑖 𝟏 = 3𝑘𝑖 (𝐴𝑖,𝑣𝑜𝑙 𝛼ℎ𝑜𝑚 + 𝐷𝑖𝑚,𝑣𝑜𝑙 𝛼𝑚 + 𝐷𝑖𝑖,𝑣𝑜𝑙 𝛼𝑖 − 𝛼𝑖 )𝛥𝑇 𝟏,
(4)
𝜀𝑖 = A𝑖 ∶ 𝐄ℎ𝑜𝑚 + 𝐷𝑖𝑚,𝑣𝑜𝑙 𝛼𝑚 𝛥𝑇 I + 𝐷𝑖𝑖,𝑣𝑜𝑙 𝛼𝑖 𝛥𝑇 I, and the stress components of the surrounding matrix (in the ITZ) are
where 𝐷 stands for the eigenstrain influence tensor [49] with determined by
𝐷𝑚𝑚,𝑣𝑜𝑙 = (1 − 𝑓𝑚 𝐴𝑚,𝑣𝑜𝑙 )𝑆𝑣𝑜𝑙 , 𝜎𝑚,𝑟𝑟 = 𝜎𝑖 ,
( )
𝑘 − 𝑘𝑚 −1 𝑘 4𝜇𝑚 (3𝑘𝑚 + 𝜇𝑚 )𝜖𝑖 + (3𝑘𝑚 − 2𝜇𝑚 )(2𝜇𝑚 𝜖𝑖 + 𝜎𝑖 ) − 18𝑘𝑚 𝜇𝑚 𝛼𝑚 𝛥𝑇
𝐷𝑚𝑖,𝑣𝑜𝑙 = 𝑓𝑖 𝐴𝑚,𝑣𝑜𝑙 𝑆𝑣𝑜𝑙 1 + 𝑆𝑣𝑜𝑙 𝑖 𝑆𝑣𝑜𝑙 𝑖 , 𝜎𝑚,𝜃𝜃 = 𝜎𝑚,𝜙𝜙 = ,
𝑘𝑚 𝑘𝑚 3𝑘𝑚 + 4𝜇𝑚
(5)
𝐷𝑖𝑚,𝑣𝑜𝑙 = 𝑓𝑚 𝐴𝑖,𝑣𝑜𝑙 𝑆𝑣𝑜𝑙 , 𝜎𝑚,𝑟𝜃 = 𝜎𝑚,𝑟𝜙 = 𝜎𝑚,𝜃𝜙 = 0.
( )
𝑘 − 𝑘𝑚 −1 𝑘 (10)
𝐷𝑖𝑖,𝑣𝑜𝑙 = (1 − 𝑓𝑖 𝐴𝑖,𝑣𝑜𝑙 ) 1 + 𝑆𝑣𝑜𝑙 𝑖 𝑆𝑣𝑜𝑙 𝑖 .
𝑘𝑚 𝑘𝑚
To quantify the thermal microstresses of the individual constituents, If 𝛼𝑚 < 𝛼𝑖 , the principal stress component 𝜎𝑚,𝜃𝜃 refers to tension, which
a uniform temperature change 𝛥𝑇 is considered at the concrete level may induce microcracks.
while keeping it macroscopically stress-free:

𝛴𝑐𝑜𝑛 = 0. (6) 2.2. Crack driving force

Together with the generalized Hooke’s law

A micro-poro-fracture-mechanical approach [50–53] is used to cal-
𝛴ℎ𝑜𝑚 = Cℎ𝑜𝑚 ∶ (𝐄ℎ𝑜𝑚 − 𝐄𝑒ℎ𝑜𝑚 ) = Cℎ𝑜𝑚 ∶ 𝐄ℎ𝑜𝑚 + 𝛴ℎ𝑜𝑚
𝑒
(7) culate the thermally induced crack driving force in the cement paste

3
S. Peters et al. Construction and Building Materials 479 (2025) 141259

of different concretes according to the varying thermal-expansion co- Table 1

Stoichiometric dehydration reactions and range of temperature [31].
efficients of its constituents. In this framework, initial cracks induced
by the hydration process are idealized as penny shaped inclusion Stoichiometry of reaction Temp. range [◦ C]

oriented in principal stress directions, assuming linear elastic fracture CH(𝑠) → C(𝑠) + H(𝑔) ∼ 375–500
mechanic theory at the cement paste scale. The multi step material C3 AH6(𝑠) ⇌ C3 AH1.5(𝑠) + H4.5(𝑔) ∼ 200–300
homogenization results in a three-dimensional smeared approach based 7 ⋅ C3 AH1.5(𝑠) → C12 A7(𝑠) + 9 ⋅ C(𝑠) + H10.5(𝑔) ∼ 300–400
on a micro-mechanical elastic damage formulation. C6 AS̄ 3 H26(𝑠) → C4 ASH
̄ ̄
8(𝑠) + 2 ⋅ CSH0.5(𝑠) + H17(𝑔) ∼ 70–110
In the framework concrete is modeled as a drained material (no ̄
C4 ASH ̄
12(𝑠) → C3 AH6(𝑠) + CSH0.5(𝑠) + H5.5(𝑔) ∼ 40–600
internal pressure) which results in the damage criterion for the 𝑖th crack ̄
CSH ̄
0.5(𝑠) → CS(𝑠) + 2H(𝑔) ∼ 230–250
family:
CC̄ (𝑠) → C(𝑠) + C̄ (𝑔) ∼ 480–850
𝐺𝑖 − 𝐺𝑐𝑖 ≤ 0; 𝜖̇ 𝑖 ≥ 0; (𝐺𝑖 − 𝐺𝑐𝑖 )𝜖̇ 𝑖 = 0, (11) 𝑥C ⋅ 𝑦S ⋅ 𝑧H(𝑠) → 𝑥C ⋅ 𝑦S ⋅ (𝑧 − 3)H(𝑠) + 3H(𝑔) ∼ 40–600
C3.4 S2 H3(𝑠) → C3.4 ⋅ S2(𝑠) + 3H(𝑔) ∼ 520–850
where the energy-release rate 𝐺𝑖 reads:
1 𝜕C
𝐺𝑖 = − ⟨𝐄⟩ ∶ ∶ ⟨𝐄⟩, (12)
2 𝜕𝜖𝑖
3. Material properties
with ⟨𝐄⟩ as the local strain tensor. The critical energy-release rate 𝐺𝑐𝑖
is given by:
3.1. Cement paste properties at elevated temperatures
( )1∕3
2𝜋 𝑛𝑖 2𝜋 𝑔𝑓
𝐺𝑐𝑖 = 𝑔𝑓 = , (13)
3 𝜖𝑖 3 𝑎𝑖 Based on the model [31], the Young’s modulus of cement paste at
with 𝑔𝑓 , 𝑛𝑖 as the fracture energy and the number of cracks, respec- elevated temperatures is predicted. The initial volume fractions of the
tively. 𝜖𝑖 = 𝑛𝑖 𝑎3𝑖 denotes the crack density [54] and 𝑎𝑖 describes the cement paste constituents at ambient temperature are calculated by
crack radius. using the CemGEMS web app [55] with the cement composition and
To identify the evolving crack patterns, the microscopic strain ten- w/c ratio as inputs. Afterwards, the chemical dehydration reactions in
sors 𝐄𝑐𝑟,1 , 𝐄𝑐𝑟,2 , 𝐄𝑐𝑟,3 are calculated, which lead to the propagation of the hardened cement paste as summarized in Table 1 are coupled with
each crack family when the energy-release rate is equal to its threshold the associated reaction kinetics. This approach enables the individual
value: quantification of the volume fractions of all cement paste constituents
at each level of temperature. According to the evolution of the different
1 𝜕C 2𝜋 𝑔𝑓
− I∶ ∶ I|𝐄𝑐𝑟 |2 ∶= (14) volume fractions, multiscale homogenization procedures tailored to
2 𝜕𝜖𝑖 3 𝑎𝑐𝑟
𝑖 cementitious materials, as described above, are employed to compute
For the derivation of the equations the reader may refer to [50,53]. the evolution of the homogenized stiffness during heating.
Fig. 3 shows the evolution of the cement paste stiffness for different
2.3. Damage law and initial characterization of cement paste water to cement ratios, assuming a heating rate of 3 ◦ C/min. As long
the aluminate content is low, the overall curve for OPCs follows the
Once the crack driving force, reaches the critical energy-release rate same trend.
𝐺𝑐𝑖 , the crack radii 𝑎𝑖 of the penny-shaped cracks oriented according The formulation of the thermal-expansion coefficient of cement
to an isotropic random distribution grow as long as the criterion is paste 𝛼𝑐𝑒𝑚𝑒𝑛𝑡𝑝𝑎𝑠𝑡𝑒 was obtained by fitting the experimental results pre-
fulfilled. The evolution of the crack radius is determined through an sented in [56] concerning ordinary Portland cement (OPC):
iterative approach. When the crack driving force exceeds the critical
𝛼cement paste (𝑇 ) = max(2 ⋅ 10−5 − 11 ⋅ 10−10 ⋅ (𝑇 − 20)2 , −42.8 ⋅ 10−6 ), (16)
energy-release rate, the microcracks extend incrementally by a small
value 𝛥𝑎𝑖 . Using the updated crack radius 𝑎𝑛𝑒𝑤 = 𝑎𝑖 + 𝛥𝑎𝑖 , the effective where 𝑇 is the temperature in [◦ C].
material properties of cement paste, mortar and concrete are recom-
puted via Eqs. (1) and (3). Subsequently, the stress resulting from the
3.2. Load-induced thermal strains
thermal-expansion mismatch of the composite constituents is localized
to the cement paste scale using Eq. (9). Following this, the updated
crack driving force, corresponding to the left-hand side of Eq. (14), is Experimental evidence shows that mechanically loaded concrete
evaluated. If it still exceeds the critical energy-release rate, the crack specimens undergo significant quasi-instantaneous thermal strains,
radius is further increased by 𝛥𝑎𝑖 until the failure criterion in Eq. (11) known as LITS, during first heating [38]. The physical mechanism is
is satisfied. A schematic representation of this procedure is provided in still not completely understood.
Fig. 1 for an overview of the procedure. Note that the crack driving However, Jennings proposed in [57] that creep and irreversible
force is always decreasing for a larger crack radius for the following shrinkage share a similar physical origin, namely a sliding, lateral,
reason: Larger crack radii lead to a reduced stiffness of the cement and/or rotational movement of the C-S-H chains. In this paper, we hy-
paste. thereby diminishing the stresses induced by thermal mismatch. pothesize that LITS are also driven by this mechanism [32].
The diffuse cracking, which dominates the fracture process under Wyrzykowski et al. [58] demonstrated that the C-S-H microstructure
thermal load, is described by a crack density parameter and a crack changes with increasing temperature, whatever the external loading.
aspect ratio. This concept [50] was validated at the micro and macro Hence, the changing microstructure leads to different LITS at different
scale by [41,53]. Based on these investigations, we set the initial temperature levels as a consequence of changing active sliding/rota-
microcrack parameters to tional surfaces. The temporal evolution therefore follows, qualitatively,
the same kinetic characteristics governing the dehydration process.
𝑎0 = 100 [μm], 𝑐0 = 1 [μm], 𝐺𝑖 = 10 [N/m],  = 1012 [1/m3 ], Accordingly, the evolution law for LITS is calibrated with the de-
hydration kinetics and by means of experimental results of Mindeguia
(15)
et al. [59], who measured LITS of cement paste at elevated temper-
with 𝑎0 , 𝑐0 as the initial crack length and microcrack thickness, respec- atures with 20% compressive load. Fig. 4(a) shows, that the thermal
tively.  characterizes the density of microcracks and 𝐺𝑖 the fracture evolution of the LITS with a heating rate of 2 ◦ C is very close to
energy. the dehydration kinetics of C-S-H [60], see Eq. (17), when using a

4
S. Peters et al. Construction and Building Materials 479 (2025) 141259

Fig. 3. Degradation curves of the Young’s modulus for three cement paste mix design during heating up to 650 ◦ C with a heating rate of 3 ◦ C/min as predicted by the multiscale
model [31].

Fig. 4. (a) Load-induced thermal strain of cement paste at a load of 1 MPa (b) Load induced thermal strain rate of cement paste at a load of 1 MPa.

calibrated reduced activation energy (AE) (14 [kJ/mol] compared to Table 2

Material properties of commonly-used concrete aggregates [48].
34 [kJ/mol]) and a calibration coefficient 𝑘 = 0.01095:
( Aggregate Thermal-expansion coefficient Elastic modulus Poisson’s ratio
14[𝑘𝐽 ∕𝑚𝑜𝑙] ( 1 )) type [10−6 /◦ C] [65] [GPa] [64] [-] [64]
1 1
𝜉̇ 𝐿𝐼𝑇 𝑆 = 𝑘 ⋅ 𝐻(𝑇 − 40) ⋅ (𝜉𝑒𝑞 − 𝜉𝐿𝐼𝑇 𝑆 ) 𝑒𝑥𝑝 − ,
1[ℎ] 𝑅 𝑇 40 limestone 4.75 45 0.270
dolomite 14.2 [56] 55 0.225
(17) sandstone 11.25 35 0.250
basalt 6.75 70 0.220
with 𝜉𝑒𝑞 = 1 − 𝑒𝑥𝑝(−0.01(𝑇 − 40)0.9 ) and 𝐻 denoting the Heaviside
granite 7.50 35 0.230
function. quartzite 11.75 50 0.160
Experimental studies by Khoury et al. [40] demonstrate that the
induced stresses in a heated concrete specimen scale proportionally to
′
the LITS. 𝜉𝐿𝐼𝑇 (𝑇 ) is the load-induced thermal strain rate (derivative
𝑆
of 𝜉𝐿𝐼𝑇 𝑆 at each level of temperature) for a compressive load of 1 among adjacent mineral grains inducing microcracks and chemical
MPa, see Fig. 4(b). Hence, the LITS are scaled proportionally with the conversion. The characterization of the Young’s modulus, Poisson’s
compressive stress 𝜎comp,cp in the cement paste ratio, and thermal-expansion coefficient of the aggregate types at el-
evated temperatures is derived from experimental data. At ambient
′ temperature, we adopt the average elastic parameters from [64] and
𝜖𝐿𝐼𝑇 𝑆 (𝑇 ) = 𝜎comp,cp ⋅ 𝜉𝐿𝐼𝑇 𝑆 (𝑇 ). (18)
the thermal-expansion coefficients from Emanuel et al. [65], similar
For a transient analysis Eq. (18) translates to a rate form to [48], see Table 2. The values provided represent averaged parame-
′
𝜖̇ 𝐿𝐼𝑇 𝑆 (𝑇 ) = 𝜎̇ comp,cp ⋅ 𝜉𝐿𝐼𝑇 ̇′ (19) ters, since the aggregate properties are characterized by a considerable
𝑆 (𝑇 ) + 𝜎comp,cp ⋅ 𝜉𝐿𝐼𝑇 𝑆 (𝑇 ).
scatter. They depend on the different diagenetic environments of the
quarry [66,67], which affect the mineral composition, porosity, crystal
3.3. Aggregate properties at elevated temperatures orientation and/or texture of the aggregate rock [48,65]. Starting
at ambient temperature, the normalized thermal degradation of the
To analyze different concrete compositions within the proposed aggregate properties will be discussed in the following subsections.
model, the degradation of the Young’s modulus of the aggregates has to
be characterized for the investigated temperature ranges. Irreversible
thermal degradation of the elastic parameters of aggregates has been 3.3.1. Degradation of siliceous aggregates
observed by many researchers, e.g. [61–63], in the last decades. The The thermally-induced reduction of the elastic modulus in quartz
origin of this degradation process is the differential thermal-expansion aggregate was derived by fitting the test results presented in [61],

5
S. Peters et al. Construction and Building Materials 479 (2025) 141259

Fig. 5. Young’s modulus degradation of different aggregate types at elevated temperatures.

Fig. 6. Comparison of numerically and analytically homogenized concrete models with two different sieve curves and spherical inclusions. (a) Thermal-expansion coefficient, (b)
Matrix stresses (c) Inclusion (aggregate) stresses.

see Fig. 5 and data adopted from [68] was utilized for the thermal- ples mainly composed of feldspar and quartz. The Young’s modulus
expansion coefficient evolution, see Fig. 7. Silica sand consists of up to reduction curve is included in Fig. 5.
99% quartz [69]. Therefore, in this model, the thermal and mechanical
properties of quartz are applied to represent those of siliceous sand. 3.3.2. Degradation of carbonate aggregates
Sandstones are porous sedimentary rocks which behave differently One of the most common types of sedimentary rocks is limestone
at high temperatures according to the grain size and cementing ma- containing Ca(HCO3 )2 , CaCO3 , CO2 and H2 O. The data from [73] was
terials, e.g. clayey and clayey ferruginous sandstones decrease less adopted to generate the Young’s modulus reduction curve at elevated
pronounced as compared to silica-cemented sandstones [70]. The ther- temperatures in Fig. 5. Dolomite (CaMg(CO3 )2 , [74]) is another type
mal behavior is characterized by the results available in [63], see of carbonate rock, primarily formed from CaCO3 and MgCO3 [75,76].
Fig. 5. These sandstone samples were mainly composed of quartz, The degradation curve included in Fig. 5 was generated based on the
dolomite/ankerite, feldspar and kaolinite [63]. data from [77].
Granite consists predominantly of quartz, feldspar [62] and po- In the context of the experimental investigations in [78], it has
tentially biotite [71]. Data from [72] is utilized, with granite sam- been observed, that the Young’s modulus of basalt exhibits minimal

6
S. Peters et al. Construction and Building Materials 479 (2025) 141259

Fig. 7. Comparison of the homogenized thermal strains with experimental data from [56] for (a) mortar with quartz sand and (b) concrete with quartz sand and dolomite
aggregates.

sensitivity to temperature variations. The Young’s modulus degradation as

data from [78] was used for the respective plot included in Fig. 5. {
0 if 𝑓𝐼 ≤ 35%
k ∶= 2 7 (21)
𝑓
3 𝐼
− 30
else
4. Model error and thermal-expansion coefficient homogenization
with 𝑓𝐼 as the inclusion volume fraction.
4.1. Model error
4.2. Thermal-expansion coefficient homogenization
The aggregates in concrete or mortar do not have a spherical shape
and the Mori–Tanaka homogenization scheme is not derived for high The homogenization of the thermal-expansion coefficient of con-
inclusion volume fractions above 15%. Therefore, the first step is to ver- crete with Eq. (3) has been already extensively validated in [48]
ify that the thermal-expansion coefficient and the homogenized stress for ambient temperatures. The validation at elevated temperatures is
components predicted by the micromechanical model match those of provided in the following. Cruz and Gillen [56] experimentally de-
a fully resolved FEM model with real aggregate shapes and periodic termined the thermal-expansion coefficient at elevated temperatures
boundary conditions. For this purpose, the thermal-expansion coeffi- of cement paste, mortar, and concrete based on given aggregate and
cient and homogenized stress components in a composite material with sand properties. These results are therefore well-suited to validate the
a temperature change of 1 ◦ C is determined using the computational proposed multiscale model. Utilizing the measured thermal-expansion
framework of [79] with the mesoscale generator of [80] for concrete coefficients of [56] and the elastic parameters provided in Table 2,
structures comparing two sieve curves with the largest grain size of 8 the free thermal strains of mortar and concrete are calculated for each
and 16 mm. The macroscopic results derived from the computational temperature increment using Eq. (3). These calculated strains are then
mesoscale model are compared with the micromechanical solution. As a plotted against the experimental data from [56] in Fig. 7.
reference, RVEs with spherical inclusions were homogenized addition- Fig. 7(a) shows that the homogenized thermal strains of mortar
ally to investigate the influence of the aggregate shape and the Mori do not adequately represent the measured thermal strains of mortar
Tanaka assumption. The material properties are defined as at elevated temperatures beyond 200 ◦ C, as the discrepancy between
the measured and calculated strains increases significantly beyond this
𝐸𝑀𝑎𝑡𝑟𝑖𝑥 = 30 [GPa], 𝐸𝐼𝑛𝑐𝑙𝑢𝑠𝑖𝑜𝑛 = 60 [GPa], 𝑣𝑖 = 𝑣𝑚 = 0.25 [-], point. In contrast to the homogenized thermal strains of mortar, in Fig.
(20)
𝛼𝑀𝑎𝑡𝑟𝑖𝑥 = 10 ⋅ 10 [1/ C], 𝛼𝐼𝑛𝑐𝑙𝑢𝑠𝑖𝑜𝑛 = 12 ⋅ 10−6 [1/◦ C].
−6 ◦ 7(b), the predicted thermal strains of a concrete containing dolomite
Fig. 6(a) shows that the homogenized thermal-expansion coefficient aggregates are shown to exhibit the same trend observed in the tests,
closely matches all numerically homogenized results across the inves- but the thermal strains are slightly underestimated.
tigated volume fractions, indicating that the analytical Mori–Tanaka The fact that the measured free thermal strains of mortar cannot be
homogenization scheme with spherical shapes yields sufficiently ac- sufficiently homogenized with the cement paste and sand properties has
curate results. According to Fig. 6(b), the numerically homogenized already been emphasized by [81]. They attribute the deviation in the
matrix stresses for the RVEs with a sieve curve with 8 mm maximum homogenized thermal strains at elevated temperatures to cracking. We
grain size deviate about 10% from the analytical solution for volume assume that an additional cause of the discrepancy between the homog-
fractions exceeding 40%. Homogenized RVE matrix stresses with spher- enized and measured thermal strains of mortar is due to the presence
ical inclusions are closer to the analytical solution as compared to the of LITS. This assumption is driven by the following argumentation:
analyses based on realistic shapes of the aggregates. In contrast, the Considering a mortar composite RVE with spherical inclusions rep-
numerically homogenized stresses within the aggregates are lower than resenting sand embedded in a hardened cement paste matrix which can
the analytical predictions, and the influence of the aggregate shape expand freely according to a temperature change on the composite-
is less pronounced, as shown in Fig. 6(c). In summary, the Mori– scale. With 𝛼𝑚𝑎𝑡𝑟𝑖𝑥 < 𝛼𝑖𝑛𝑐𝑙𝑢𝑠𝑖𝑜𝑛 at elevated temperatures, the homog-
Tanaka scheme is able to predict the thermal-expansion coefficient and enized volumetric stresses represent tension in the hardened cement
inclusion stresses sufficiently accurate for digital mortar and concrete paste matrix and compression in the inclusions, assuming perfect bond-
samples. To compensate for the (small) error in the homogenized ma- ing [47]. Although the homogenized volumetric stresses in the cement
trix stresses in the proposed micromechanics model at higher aggregate paste are in tension, the local principal stresses in the cement paste
volume fractions, a linear compensation factor k is multiplied with the along the interface in the radial direction to the inclusion are com-
homogenized stresses for high volume fractions of 35% to 50% defined pressive due to the continuous displacement field [82], see Eq. (10).

7
S. Peters et al. Construction and Building Materials 479 (2025) 141259

Fig. 8. Comparison of the homogenized thermal strains with experimental data from [56] for mortar with quartz sand. The dashed orange line refers to the homogenized thermal
strains without LITS considered and the red shaded area represents the model results considering the effect of LITS for a realistic range of assumptions regarding the average
matrix stresses.

The compressive stresses in the cement paste matrix in radial direction Table 3
Material and test parameters of the concrete used in [87].
induce additional strains according to the LITS phenomenon which can
be interpreted as instantaneous thermal creep. These additional strains Model input Quantity

affect the observed volumetric thermal-expansion strains [83] Cement type OPC
∑ Fine aggregates Quartzite
𝜖𝑡ℎ,𝑣 = ((𝛼11,𝑒 + 𝛼11,𝐿𝐼𝑇 𝑆 ) + 𝛼22,𝑒 + 𝛼33,𝑒 ) ⋅ 𝛥𝑇 = 𝛼𝑖𝑖 𝛥𝑇 + 𝜖𝐿𝐼𝑇 𝑆 (𝑇 ), (22) Coarse aggregates Limestone
Water/cement ratio 0.327
with 𝛼𝑣 and 𝛼𝑥𝑥,𝑒 as the volumetric thermal-expansion coefficient proportions, by weight 1:1.54:2.31
heating rate 7 ◦ C/min
and the elastic principal thermal-expansion coefficient, respectively.
𝛼11,𝐿𝐼𝑇 𝑆 describes the additional strain rate according to LITS.
To introduce the influence of LITS on the volumetric thermal-
expansion coefficient, a routine is proposed utilizing the model above In this context, the role of the interfacial transition zone (ITZ)
with the LITS master curves in Fig. 5. First, Eq. (3) is solved to homog- formed between the aggregates and the mortar matrix [85] needs to
enize the thermal-expansion coefficient of mortar at each temperature be discussed. At lower temperatures the thermal-expansion of cement
increment. Second, the stresses are localized via Eq. (10). Finally, the paste is higher then that of the aggregates, which may lead to interfacial
additional strains induced by compressive stresses in the cement paste cracking [3,4], affecting the Young’s modulus and thermal-expansion
are computed using Eq. (18). coefficient. However, it is hypothesized that the influence of interfacial
Since, the matrix stresses decay with the power of three [84] with transition zones (ITZs) on the Young’s modulus and thermal-expansion
increasing distance from the contact surface for a single inclusion coefficient at elevated temperatures is negligible as the cement paste
problem1 , instead of the maximum stress computed at the interface, contracts at higher temperatures, leading to better embedment of the
the average stress obtained from integration from 𝑟𝑖𝑛𝑐𝑙𝑢𝑠𝑖𝑜𝑛 to infinity aggregates and the closure of potential ITZs. This phenomenon may also
of the decay function is used to evaluate Eq. (18). The average value explain the strength increase observed in certain concretes exposed to
of this additional strain is obtained as 50% of the maximum stress at elevated temperatures. For further details, refer to [39,86].
the interface. In composites with high inclusion volume fraction such
as mortar, however, this average additional strain should be higher
5. Model validation and sensitivity study of the Young’s modulus
because of the short distances between the inclusions. Therefore, an
of concrete at elevated temperatures
envelope is calculated, multiplying the LITS according to Eq. (18) based
on the maximum compressive stress at the aggregate-mortar interface,
A number of test results of the Young’s modulus obtained for
with a factor ranging from 0.55 to 0.85. The resulting plot is contained
different concretes at elevated temperatures are available in the open
in Fig. 8 as a red-shaded area. The original homogenized thermal strains
literature. Initially, the proposed model is validated via well docu-
is shown as a dashed orange line. After the adequate consideration
mented experiments by [87]. Subsequently, experimental results con-
of the effect of LITS according to the proposed model, the predicted
tained in the state of the art report [86,88] are adopted in the validation
range of thermal-expansion coefficients matches the observed thermal
campaign.
strains of mortar now with a good accuracy. This result corroborates
our hypothesis, that the effect of LITS is relevant for the explanation
of the observed evolution of the thermal-expansion coefficients at high 5.1. Experiments by Castillo and Durrani [87]
temperatures.
To further validate this framework, more experimental data is nec- Castillo and Durrani [87] investigated the effect of transient high
essary. One possible procedure would be the investigation of mortars temperature on the uniaxial compressive strength and the Young’s
with various quartz sand volume fractions at elevated temperatures. modulus of high strength concrete. Type I Portland cement was used
According to the proposed framework, at low sand volume fractions of with a water to cement ratio of 0.327. The proportions of concrete
mortars the thermal-expansion should be mainly driven by the cement by weight are 1:1.54:2.31 cement, fine (siliceous sand) and coarse
matrix. aggregates (crushed limestone), respectively and the heating rate of
the specimen was 7 ◦ C/min. The input parameters are summarized in
Table 3. To enable modeling of the thermal degradation process of the
1
The decay function is 𝑥3 ∕𝑟3𝑖𝑛𝑐𝑙𝑢𝑠𝑖𝑜𝑛 . hardened cement paste the chemical composition is given in Table 4.

8
S. Peters et al. Construction and Building Materials 479 (2025) 141259

Fig. 9. Comparison of the evolution of the relative Young’s modulus 𝐸𝑇 /𝐸0 during heating as predicted by the of the model (full line) with experimental results by Castillo and
Durrani [87].

Table 4 Table 5
Chemical composition of the cement paste. Material and test parameters used in the sensitivity study investigating different
o. a. CaO SiO2 Al2 O3 Fe2 O3 SO3 MgO CaO (f) CSbH2 aggregate types.
% 62.04 20.15 4.49 2.29 2.81 2.41 0.03 0.04 Model input Quantity
Cement type OPC
Fine aggregates Quartzite & Dolomite
Coarse aggregates Limestone, Granite, Dolomite
According to Fig. 9, the ratio 𝐸𝑇 /𝐸0 of the modulus of elasticity Basalt, Sandstone, Quartzite
during heating and the initial Young’s modulus, as predicted by the Water/cement ratio 0.327
model, agrees very well with the experiments in the temperature range proportions, by weight 1:1.54:2.31
heating rate 7 ◦ C/min
from 20 to 500 ◦ C. Cracking starts at 310 ◦ C which is reflected by
the kink in the degradation curve. In the temperature range near the
quartz transition point at 574.3 ◦ C the model slightly overestimates
the Young’s modulus of concrete. One possible explanation is that the degradation curves have similar trends and the starting point of crack-
material properties of quartzitic sand cannot be accurately captured ing varies from 270 ◦ C–330 ◦ C. Calcareous sand leads to higher Young’s
due to their great variability, as documented in [89]. modulus reduction before cracking starts. Once cracking has started,
the reduction of the Young’s modulus is only slightly affected by the
5.2. Experimental results reported by Phan and Carino aggregate type. In the pre-cracking temperature range, the aggregate
type accounts for variations of the Young’s modulus of about 12%.
In the review reports [86,88], experimental results have been col- This sensitivity study suggests that the coarse aggregate stiffness has a
lected to create a data envelope of the material properties of concrete at minor influence on degradation, which agrees with many experimental
high temperatures. The data envelope of the relative Young’s modulus findings, see e.g. [90].
is taken as the basis for further validation and sensitivity studies of the
proposed multiscale model. Fig. 10 shows this data envelope, illustrat- 5.4. Influence of the initial crack density and the water-cement ratio
ing the significant scatter of the relative Young’s modulus measured
during heating. For instance, certain concrete samples experienced a Different water to cement ratios or blended cement compositions
nearly 40% loss in stiffness at 110 ◦ C, while others only exhibited lead to different initial configurations of microcracks as a consequence
smaller losses in stiffness. Reasons for these variations are differences of drying and autogenous shrinkage. The influence of the crack density
in aggregate types, moisture states, water to cement ratios or cement  , varying between 2 ⋅ 109 to 10.3 ⋅ 1012 [1/m3 ], on the Young’s
compositions. These influencing factors will be investigated in the modulus degradation of concrete at elevated temperature is shown in
following subsections. Fig. 11(a). It is observed, that the crack density significantly affects
the temperature level at the onset of cracking and the post cracking
5.3. Influence of the aggregates behavior. Low crack densities are leading to lower temperatures where
cracking starts and to a steeper decrease of the Young’s modulus of
To address the influence of the aggregate type on the thermal degra- concrete during heating. The figure also shows, that model predictions
dation, a sensitivity study within the proposed model is conducted. In for temperatures below approx. 150 ◦ C are consistently at the upper
this sensitivity study Type I Portland cement was used with a water– limit of the envelope of test results, while many experimental curves
cement ratio equals 0.327. The proportions of concrete by weight are show a more pronounced decay of the Young’s modulus. This will be
1:1.54:2.31 cement, fine (carbonate and siliceous sand) and coarse discussed in detail in Section 5.6.
aggregates (limestone, dolomite, quartz, granite, basalt and sandstone), To investigate the influence of the water-cement ratio, the input
respectively and the heating rate of the specimen was 7 ◦ C/min. All data summarized in Tables 3 and 4 is used as a reference state. In
input parameters are summarized in Tables 4 and 5. the sensitivity study, the cement stiffness is varied from a 17.2 GPa
Fig. 10 contains the predicted plots obtained for the Young’s mod- to 33.7 GPa which relates to a range of the water-cement ratio from
ulus of concrete at elevated temperatures, assuming the different con- 0.6 to 0.25, respectively [91]. Fig. 11(b) shows that the water-cement
figurations of the concrete constituents as listed in Table 5. For com- ratio has only a minor effect on the Young’s modulus reduction. The
parison, the data envelope from the review papers [86,88] is included temperature where cracking starts decreases by approx. 30 ◦ C with in-
in the figure as a gray-shaded area. The figure shows that all computed creasing cement paste stiffness because the induced thermal mismatch

9
S. Peters et al. Construction and Building Materials 479 (2025) 141259

Fig. 10. Influence of different aggregate types on the thermally induced reduction of the Young’s modulus as predicted by the multiscale model for (a) calcareous sand mortar
(thick blue lines) and (b) siliceous sand mortar (thick red lines). For comparison, the range of experimental plots obtained from different concrete mixtures reported in [86,88] is
included (gray-shaded area).

Fig. 11. Influence of (a) initial crack densities and (b) initial cement paste stiffnesses on the thermally induced degradation of the Young’s modulus of concrete. The range of
model predictions is shown as contour plots in both figures. For comparison, the range of experimental plots obtained from different concrete mixtures reported in [86,88] is
included (gray-shaded area).

stresses are lower and the thermal-expansion coefficient depends on the Concrete mixtures containing stiffer aggregates typically demon-
stiffness of the constituents and the volume fraction, see Eq. (3). strate an earlier initiation of localized microcracking due to the ther-
Concretes containing larger aggregates generally exhibit higher mally induced mismatch stresses in comparison to a reference concrete
crack densities, resulting in a delayed onset of local microcracking containing less stiff aggregates. Additionally, concrete made with ther-
at elevated temperature levels and a more gradual reduction of the mally stable aggregates also exhibits a lower temperature for the initial
Young’s modulus compared to a reference concrete with smaller aggre- microcracking, as the aggregates retain their stiffness while the cement
gates. Conversely, the use of admixtures and additives, such as calcined paste undergoes decomposition.
clay, silica fume, or blast furnace slag, reduces the initial crack density,
leading to earlier cracking and to a steeper decline in the Young’s 5.6. Influence of load-induced thermal strains in Young’s modulus measure-
modulus during heating. ments

5.5. Evaluation of damage mechanisms of concrete at elevated temperatures Fig. 11 shows, that in the temperature range below 150 ◦ C, none
of the model predictions evaluated for different compositions is able to
One advantage of computational modeling is that it makes the predict the strong decay of the concrete stiffness observed for some of
decoupling of various intertwined phenomena possible, something that the concrete mixes documented in e.g. [92,93]. Obviously, this rapid
generally cannot be achieved from experiments. By decoupling the phe- loss of stiffness cannot be explained only by the thermally induced loss
nomena, each contribution can be investigated by itself. This approach of stiffness of aggregates or dehydration of the cement paste. We hy-
is employed in the present study, utilizing the input data from Tables pothesize that one reason for the rapid loss of stiffness at temperatures
3 and 4. below 150 ◦ C for hot state tested concrete specimens is partly due to
The results from the analysis of the individual contributions in LITS, based on the following reasoning.
Fig. 12 show that the influence of the chemical cement paste degra- In the collection of experimental data reported in [86,88], the mois-
dation has only a minor effect. This is reasonable, since aggregates ture states and testing procedures have not been documented. The tests
and sand occupy the largest volume fraction in concrete as compared performed by Castillo and Durrani [87], whose results crowd the upper
to the cement paste. Consequently, the intrinsic thermal degradation part of the envelope, are based on dried samples. The samples providing
of aggregates has a comparably larger influence before the onset of stiffness reduction plots at the lower bound have not been dried and
microcracking. After that, the evolution of microcracking is the most therefore tested in hot-state. In a hot-state compression test, LITS
important mechanism leading to the stiffness reduction of concrete cannot be differentiated from elastic strains due to their instantaneous
during heating. nature. When the Young’s modulus is calculated from a compression

10
S. Peters et al. Construction and Building Materials 479 (2025) 141259

Fig. 12. Analysis of different damage sources for the degradation of the Young’s modulus of concrete at high temperatures: Cement paste degradation, aggregate degradation and
microcracking.

test conducted at 20% or 40% of the ultimate compressive strength, a significantly greater reduction in Young’s modulus at 100 ◦ C than the
as is common in civil engineering, the elastic strains are overestimated estimated 20%. This may be attributed to the higher volume fraction
because the additional LITS are included as elastic strains. Similarly, of cement paste in high-performance concrete, which can induce more
recent investigations [94] conclude, that under the assumption, that LITS, as well as to the presence of additives like silica fume, which in-
the strains measured during loading at 45 ◦ C are elastic and that creep crease the volume fraction of C-S-H in the cement paste, where LITS are
strains develop exclusively during a load plateau, leads to a 7.6% likely generated. Further tests, however, are required to confirm these
overestimation of the elastic strains. This is in line with the proposed or possible other effects. Concretes showing substantial reductions in
hypothesis. The overestimation of the elastic strains of a compression Young’s modulus at 100 ◦ C in hot state testing will be investigated in
test of concrete at 100 ◦ C is quantified within the presented framework future work.
and compared with the experimental envelope by [86,88]. Ultrasonic measurements and acoustic tests are employed to further
Proposing a unit load of 1 MPa on a concrete RVE in uniaxial justify the hypothesis on the reduction of concrete elastic modulus up
compression and the material parameters are adopted from Table 3, to 150 ◦ C (in the hot state), that is primarily attributable to LITS. If
which represent a commonly used concrete composition. The elastic the reduction in the elastic modulus of concrete heated up to 150 ◦ C
stress of the concrete RVE is localized to the cement scale according is primarily attributed to LITS, it should not be detectable in ultrasonic
to [85,94] measurements or acoustic tests. This is because the propagation ve-
locity of elastic waves remains unaffected by creep-like mechanisms
𝜎𝑝 = (B𝑣𝑜𝑙
𝑝 I
𝑣𝑜𝑙
+ B𝑑𝑒𝑣
𝑝 I
𝑑𝑒𝑣
) ∶ 𝜎𝑐𝑜𝑚𝑝𝑜𝑠𝑖𝑡𝑒 , 𝑝 = [ Matrix,Inclusion ] (23)
such as LITS. Such experiments are provided in, e.g. [96]. Fig. 3(c)
leading to in [96] shows, that no significant acoustic events can be measured
before 150 ◦ C and Fig. 3(d) indeed confirms, that no ultrasonic speed
𝜎𝑐𝑝,𝑥𝑥 = 0.8599 ⋅ 𝜎𝑐𝑜𝑛,𝑥𝑥 , (24)
change occurs below 150 ◦ C.
where 𝜎𝑐𝑝,𝑥𝑥 and 𝜎𝑐𝑜𝑛,𝑥𝑥 are the compressive stresses in loading direction
in the cement paste and in the concrete, respectively. The elastic strains 6. Conclusions
in the concrete are calculated as
1 [MPa] 1
𝜖𝑐𝑜𝑛,𝑥𝑥 = = = 2.523 ⋅ 10−5 , (25) A semi-analytical multiscale model has been proposed for predicting
𝐸𝑐𝑜𝑛,𝑇 =100◦ 𝐶 39632 the Young’s modulus of concrete at elevated temperatures. Based on
and the additional strain induced in the cement paste by the LITS at the sensitivity studies carried out in this research project, the following
100 ◦ C are obtained as conclusions can be drawn:

𝜖𝑐𝑝,𝑥𝑥,𝐿𝐼𝑇 𝑆 = 0.8599 ⋅ 𝑓𝐿𝐼𝑇 𝑆,𝑇 =100◦ 𝐶 = 0.8599 ⋅ 1.43292 ⋅ 10−5 = 1.232 ⋅ 10−5 . • The homogenized elastic moduli and thermal-expansion coeffi-
cients proposed in this study for high temperatures are in very
(26)
good agreement with the experimental tests. This corroborates
Since only 32% of the total volume fraction is occupied by cement the possibility of micro-to-macro upscaling for the elastic mod-
paste, the total strains can be calculated: ulus and thermal-expansion coefficient via semi-analytical ho-
mogenization, as confirmed by the comparison of analytical and
𝜖𝑡𝑜𝑡,𝑥𝑥 = 𝜖𝑐𝑜𝑛,𝑥𝑥 + 0.32 ⋅ 𝜖𝑐𝑝,𝑥𝑥,𝐿𝐼𝑇 𝑆 = 2.917 ⋅ 10−5 . (27)
numerical results.
This calculation leads to the conclusion that around 15% of the exper- • The thermo-chemical processes induced in the cement paste by
imentally determined degradation of the Young’s modulus is related high temperatures have a minor impact on the overall reduction
to LITS. Adding about 4% effect associated with chemo-thermo degra- in concrete stiffness over the whole temperature range. This is
dation in the cement paste and the aggregates, the predicted share primarily due to the markedly higher volume fractions of the
on the total stiffness degradation at 100 ◦ C is 19%. This calculated aggregates relative to the hardened cement paste.
and the experimentally determined value by [95] is included in Fig. • At temperatures below 200 ◦ C, the thermally-induced aggregate
13 by an asterisk. Fig. 13 reveals a significant difference between degradation mainly influences the stiffness reduction of concrete.
the Young’s modulus reduction considering LITS in comparison to the Hence, any concrete containing thermally-stable aggregates, such
analysis that accounts only for the degradation of cement paste and as basalt, remains stable at high temperatures until the occurrence
aggregates. Moreover, Fig. 13 illustrates that certain concretes exhibit of local microcracking, whose activation is controlled by the

11
S. Peters et al. Construction and Building Materials 479 (2025) 141259

Fig. 13. Comparison of analysis results with and without considering LITS in the experimental envelope by [86,88]. (a) whole temperature range (b) close up.

thermal incompatibility between the aggregate and the cement cases, where ordinary Portland cement is substituted by CO2 reduced
paste. cement types, such as CEM III/B or CEM II/B-Q.
• The initial crack density in the hardened cement paste signifi- The multiscale model provides a basis for further analyses to predict
cantly influences the temperature at which localized microcrack- thermal decomposition and related material properties, such as intrinsic
ing starts. It also affects the subsequent post-cracking behavior. permeability, thermal conductivity or diffusivity of different mortars
Lower initial crack densities lead to an earlier initiation of micro- and concretes at elevated temperatures.
cracking and a more pronounced decline in the Young’s modulus
of concrete at elevated temperatures. Reproducibility and code
• The apparent reduction in concrete stiffness observed during
loading at moderate temperatures has been found to be mainly
The code of the paper is written in Matlab and will be made
attributable to nonrecoverable thermally load induced strains
available on request.
(LITS) and not caused by the reduction of the Young’s modulus
of the concrete associated with microcracking. This conclusion
has been corroborated by analyses of thermal loading tests heat CRediT authorship contribution statement
up to 105 ◦ C with and without considering LITS, which show an
error of almost 20% w.r.t. the experimental value, when LITS are Simon Peters: Writing – original draft, Software, Methodology,
disregarded, while the error decreases to less than 3% when LITS Conceptualization. Giao Vu: Software, Formal analysis. Günther
are considered in the model. Meschke: Writing – review & editing, Methodology, Funding
• For an accurate homogenization of the free thermal-expansion acquisition.
coefficient of mortar it is essential to consider LITS in the homog-
enization procedure. Declaration of Generative AI and AI-assisted technologies in the
• The evolution of the degree of C-S-H dehydration has been found writing process
to closely agree with the experimentally-determined evolution
of LITS, which are assumed to be associated to creep-like C- During the preparation of this work the authors used ChatGPT in
S-H sliding under compressive loading. Hence, this observation order to improve language and readability. After using this tool/service,
justifies the conclusion, that the degree of C-S-H dehydration the authors reviewed and edited the content as needed and take full
characterizes the sensitivity to chain sliding and rotation at the responsibility for the content of the publication.
C-S-H scale.
Declaration of competing interest

Commonly-used design tools for assessing concrete behavior at elevated The authors declare that they have no known competing finan-
temperatures primarily rely on phenomenological models, which often cial interests or personal relationships that could have appeared to
lack detailed information on microscale processes [97]. The proposed influence the work reported in this paper.
model addresses this limitation by explicitly accounting for mecha-
nisms occurring at smaller scales. Due to the semi-analytical nature Acknowledgment
of the homogenization approach, the model offers a closed-form so-
lution, facilitating its integration into existing design tools for robust
Financial support was provided by the German Research Founda-
damage prediction in high-temperature environments. In contrast to
tion (DFG) in the framework of the project ‘‘Thermohydraulic spalling
phenomenological models, the proposed multiscale framework is able
mechanisms in concretes with different binders with and without PP-
to attribute damage mechanisms to individual design parameters in
fibers under fire exposure: An experimental and numerical analysis’’
concrete compositions, such as aggregate type, cement class, w/c ratio.
(Grant ME 1848/11-1). This support is gratefully acknowledged.
This allows to systematically investigate practical scenarios for the
improvement of fire resistance of concrete compositions. One example
is the choice of the aggregate types according to their stiffness and Data availability
thermal expansion coefficient, which has been shown to be the main
influencing factors for the initiation of microcracking and damage Data will be made available on request.
evolution. Another example is the analysis of the damage behavior in

12
S. Peters et al. Construction and Building Materials 479 (2025) 141259

References [26] Y. Zhang, J.W. Ju, Q. Chen, Z. Yan, H. Zhu, Z. Jiang, Characterizing and
analyzing the residual interfacial behavior of steel fibers embedded into cement-
[1] R. McNamee, Fire Spalling of Concrete: Theoretical and Experimental Studies based matrices after exposure to high temperatures, Compos. Part B: Eng.
(Ph.D. thesis), KTH Architecture and Built Enviroment, 2013. 191 (2020) 107933, URL https://www.sciencedirect.com/science/article/pii/
[2] C. Jiang, J. Fang, J.-Y. Chen, X.-L. Gu, Modeling the instantaneous phase S1359836819365266.
composition of cement pastes under elevated temperatures, Cem. Concr. Res. [27] J. Lee, Y. Xi, K. Willam, Y. Jung, A multiscale model for modulus of elasticity
130 (2020) 105987, URL https://www.sciencedirect.com/science/article/pii/ of concrete at high temperatures, Cem. Concr. Res. 39 (9) (2009) 754–762.
S0008884619303035. [28] J. Zhao, J. Zheng, G. Peng, K. Breugel, Prediction of thermal decomposition of
[3] Y. Fu, Y. Wong, C. Tang, C. Poon, Thermal induced stress and associated cracking hardened cement paste, J. Mater. Civ. Eng. 24 (2012) 592–598.
in cement-based composite at elevated temperatures - Part I: Thermal cracking [29] J. Zhao, J. Zheng, G. Peng, A numerical method for predicting Youngs modulus
around single inclusion, Cem. Concr. Compos. 26 (2004) 99–111. of heated cement paste, Constr. Build. Mater. 54 (2014) 197–201.
[30] Y. Zhang, J.-W.W. Ju, H. Zhu, A novel multi-scale model for predicting the
[4] Y. Fu, Y. Wong, C. Tang, C. Poon, Thermal induced stress and associated
thermal damage of hybrid fiber-reinforced concrete, Int. J. Damage Mech. 29
cracking in cement-based composite at elevated temperatures–Part II: thermal
(3) (2019) 1–26.
cracking around multiple inclusions, in: Fire Resistance, Cem. Concr. Compos.
[31] S. Peters, G. Meschke, Computational multiscale modelling of thermally induced
26 (2) (2004) 113–126, URL https://www.sciencedirect.com/science/article/pii/
dehydration of blended hardened cement paste, Cem. Concr. Res. 180 (2024)
S0958946503000878.
107485, http://dx.doi.org/10.1016/j.cemconres.2024.107485.
[5] A. Khennane, G. Baker, Plasticity models for the biaxial behaviour of concrete
[32] F. Manzoni, T. Vidal, A. Sellier, X. Bourbon, G. Camps, On the origins of transient
at elevated temperatures, Part II: Implementation and simulation tests, Comput.
thermal deformation of concrete, Cem. Concr. Compos. 107 (2020) 103508, URL
Methods Appl. Mech. Engrg. 100 (2) (1992) 225–248, URL https://www.
https://www.sciencedirect.com/science/article/pii/S0958946519313514.
sciencedirect.com/science/article/pii/004578259290184L.
[33] Y. Anderberg, S. Thelandersson, Stress and deformation characteristics of con-
[6] K. Willam, I. Rhee, Y. Xi, Thermal degradation of heterogeneous concrete
crete at high temperature: 2. Experimental investigation and material behaviour
materials, J. Mater. Civ. Eng. 17 (2005).
model, Bull. 46 (1976) 86.
[7] W. Dias, G. Khoury, P. Sullivan, Mechanical properties of hardened cement paste
[34] R. de Borst, P.P. Peeters, Analysis of concrete structures under thermal loading,
exposed to temperatures up to 700 C (1292 F), ACI Mater. J. 87 (1990) 160–166.
Comput. Methods Appl. Mech. Engrg. 77 (3) (1989) 293–310, URL https://www.
[8] S. Masse, G. Vetter, C. Haehnel, F. Boch, Elastic modulus changes in cementitious
sciencedirect.com/science/article/pii/0045782589900790.
materials submitted to thermal treatments up to 1000C, Adv. Cem. Res. 14
[35] M. Terro, Numerical modeling of the behavior of concrete structures
(2002) 169–177.
in fire, ACI Struct. J. 95 (2) (1998) 183–193, URL https://www.
[9] J. Komonen, V. Penttala, Effects of high temperature on the pore structure and
scopus.com/inward/record.uri?eid=2-s2.0-0032028656&partnerID=40&md5=
strength of plain and polypropylene fiber reinforced cement pastes, Fire Technol.
417c55907f1ae01975856192122f82a1.
39 (2003) 23–34.
[36] C.V. Nielsen, C.J. Pearce, N. Bićanić, Theoretical model of high temperature
[10] Z. Bažant, M. Kaplan, Concrete at High Temperatures: Material Properties and
effects on uniaxial concrete member under elastic restraint, Mag. Concr. Res. 54
Mathematical Models, Longman, 1996.
(4) (2002) 239–249, http://dx.doi.org/10.1680/macr.2002.54.4.239.
[11] F.-J. Ulm, O. Coussy, Z. Bažant, The ‘‘chunnel’’ fire. I: Chemoplastic softening in
[37] C.J. Pearce, C.V. Nielsen, N. Bićanić, Gradient enhanced thermo-mechanical
rapidly heated concrete, J. Eng. Mech. 125 (3) (1999) 272–282.
damage model for concrete at high temperatures including transient thermal
[12] T.Z. Harmathy, Thermal properties of concrete at elevated temperatures, in: creep, Int. J. Numer. Anal. Methods Geomech. 28 (7–8) (2004) 715–735.
Research Paper (National Research Council of Canada. Division of Building
[38] G. Torelli, P. Mandal, M. Gillie, V.-X. Tran, Concrete strains under transient
Research); No. DBR-RP-426, ASTM J. Mater. 5 (1) (1970) 47–74.
thermal conditions: A state-of-the-art review, Eng. Struct. 127 (2016) 172–188,
[13] T. Harmathy, Fire Safety Design and Concrete, Longman, London, England, 1993. URL https://www.sciencedirect.com/science/article/pii/S0141029616304229.
[14] M.J. DeJong, F.-J. Ulm, The nanogranular behavior of C-S-H at elevated [39] U. Schneider, Concrete at high temperatures – A general review, Fire Saf. J.
temperatures (up to 700 ◦ C), Cem. Concr. Res. 37 (1) (2007) 1–12, URL 13 (1) (1988) 55–68, URL https://www.sciencedirect.com/science/article/pii/
https://www.sciencedirect.com/science/article/pii/S0008884606002134. 0379711288900331.
[15] M.A. Qomi, L. Brochard, T. Honorio, I. Maruyama, M. Vandamme, Advances [40] G.A. Khoury, B.N. Grainger, P.J.E. Sullivan, Transient thermal strain of concrete:
in atomistic modeling and understanding of drying shrinkage in cementitious literature review, conditions within specimen and behaviour of individual
materials, Cem. Concr. Res. 148 (2021) 106536. constituents, Mag. Concr. Res. 37 (132) (1985) 131–144, http://dx.doi.org/10.
[16] G. Constantinides, F.-J. Ulm, The nanogranular nature of C-S-H, J. Mech. Phys. 1680/macr.1985.37.132.131.
Solids 55 (1) (2007) 64–90, URL https://www.sciencedirect.com/science/article/ [41] T. Iskhakov, J. Timothy, G. Meschke, Expansion and deterioration of concrete due
pii/S0022509606001062. to ASR: Micromechanical modeling and analysis, Cem. Concr. Res. 115 (2019)
[17] X. Wang, T. Li, P. Wei, D. Li, N. Han, F. Xing, Y. Gan, Z. Chen, Computational 507–518, http://dx.doi.org/10.1016/j.cemconres.2018.08.001.
study of the nanoscale mechanical properties of C-S-H composites under dif- [42] C. Pichler, R. Lackner, H.A. Mang, A multiscale micromechanics model for
ferent temperatures, Comput. Mater. Sci. 146 (2018) 42–53, URL https://www. the autogenous-shrinkage deformation of early-age cement-based materials, Eng.
sciencedirect.com/science/article/pii/S0927025618300302. Fract. Mech. 74 (2007) 34–58, Fracture of Concrete Materials and Struc-
[18] P.A. Bonnaud, H. Manzano, R. Miura, A. Suzuki, N. Miyamoto, N. Hatakeyama, tures http://dx.doi.org/10.1016/j.engfracmech.2006.01.034, URL http://www.
A. Miyamoto, Temperature dependence of nanoconfined water properties: Appli- sciencedirect.com/science/article/pii/S0013794406000464.
cation to cementitious materials, J. Phys. Chem. C 120 (21) (2016) 11465–11480, [43] A. Zaoui, Continuum micromechanics: Survey, J. Eng. Mech. 128 (8) (2002)
http://dx.doi.org/10.1021/acs.jpcc.6b00944. 808–816.
[19] Y. Zhang, S. Zhang, Q. Chen, Y. Shen, J.W. Ju, M. Bauchy, Insights into the [44] R. Hill, Elastic properties of reinforced solids: Some theoretical principles, J.
effect of high temperature on the shear behavior of the calcium silicate hydrate Mech. Phys. Solids 11 (5) (1963) 357–372.
by reactive molecular dynamics simulations, Int. J. Damage Mech. 31 (7) (2022) [45] V. Levin, Thermal expansion coefficients of heterogeneous materials, Mekh.
1096–1112, http://dx.doi.org/10.1177/10567895221093395. Tverd. Tela 2 (1) (1967) 83–94.
[20] R. Shahsavari, M. Buehler, R. Pellenq, F.-J. Ulm, First-principles study of elastic [46] Y. Benveniste, A new approach to the application of Mori-Tanaka’s theory in
constants and interlayer interactions of complex hydrated oxides: Case study of composite materials, Mech. Mater. 6 (1987) 147–157.
tobermorite and jennite, J. Am. Ceram. Soc. 92 (2009) 2323–2330. [47] D. Gross, T. Seelig, Bruchmechanik. Mit einer Einführung in die Mikromechanik,
[21] J.E. Oh, S.M. Clark, H.-R. Wenk, P.J. Monteiro, Experimental determination of Springer, 2007.
bulk modulus of 14Å tobermorite using high pressure synchrotron X-ray diffrac- [48] H. Wang, H. Mang, Y. Yuan, B. Pichler, Multiscale thermoelastic analysis of
tion, Cem. Concr. Res. 42 (2) (2012) 397–403, URL https://www.sciencedirect. the thermal expansion coefficient and of microscopic thermal stresses of mature
com/science/article/pii/S0008884611002894. concrete, Materials 12(17):2689 (2019).
[22] J.E. Oh, S.M. Clark, P.J. Monteiro, Does the Al substitution in C-S-H(I) change [49] B. Pichler, C. Hellmich, Estimation of influence tensors for eigenstressed mul-
its mechanical property? Cem. Concr. Res. 41 (1) (2011) 102–106, URL https: tiphase elastic media with nonaligned inclusion phases of arbitrary ellipsoidal
//www.sciencedirect.com/science/article/pii/S0008884610002140. shape, J. Eng. Mech. 136 (8) (2010) 1043–1053.
[23] G. Geng, R.J. Myers, M.J.A. Qomi, P.J.M. Monteiro, Densification of the inter- [50] L. Dormieux, D. Kondo, F. Ulm, Microporomechanics, Wiley & Sons, 2006.
layer spacing governs the nanomechanical properties of calcium-silicate-hydrate, [51] E. Lemarchand, L. Dormieux, F.-J. Ulm, Micromechanics investigation of expan-
Sci. Rep. 7 (1) (2017) 10986. sive reactions in chemoelastic concrete, Philos. Trans.: Math. Phys. Eng. Sci. 363
[24] Y. Zhang, Q. Zhou, J.W. Ju, M. Bauchy, New insights into the mechanism (2005) 2581–2602.
governing the elasticity of calcium silicate hydrate gels exposed to high tem- [52] L. Charpin, A. Ehrlacher, A computational linear elastic fracture mechanics-based
perature: A molecular dynamics study, Cem. Concr. Res. 141 (2021) 106333, model for alkali–silica reaction, Cem. Concr. Res. 42 (4) (2012) 613–625.
URL https://www.sciencedirect.com/science/article/pii/S0008884620316136. [53] R. Esposito, The Deteriorating Impact of Alkali-Silica Reaction on Concrete:
[25] A. Caggiano, D. Said Schicchi, G. Etse, M. Ripani, Meso-scale response of concrete Expansion and Mechanical Properties (Ph.D. thesis), 2016.
under high temperature based on coupled thermo-mechanical and pore-pressure [54] B. Budiansky, R. O’Connell, Elastic moduli of a cracked solid, Int. J. Solids Struct.
interface modeling, Eng. Fail. Anal. 85 (2017) 167–188. 12 (1976) 81–97.

13
S. Peters et al. Construction and Building Materials 479 (2025) 141259

[55] D.A. Kulik, F. Winnefeld, A. Kulik, G.D. Miron, B. Lothenbach, Cemgems - [76] Subagjo, W. Wulandari, P.M. Adinata, A. Fajrin, Thermal decomposition of
an easy-to-use web application for thermodynamic modelling of cementitious dolomite under CO2-air atmosphere, AIP Conf. Proc. 1805 (1) (2017) 040006,
materials, RILEM Tech Lett. 6 (2021) 36–52, URL https://letters.rilem.net/index. http://dx.doi.org/10.1063/1.4974427, URL https://aip.scitation.org/doi/abs/10.
php/rilem/article/view/140. 1063/1.4974427.
[56] C. Cruz, M. Gillen, Thermal expansion of portland cement paste, mortar and [77] N. Sengun, Influence of thermal damage on the physical and mechanical
concrte at high temperatures, Fire Mater. 4 (1980) 66–70. properties of carbonate rocks, Arab. J. Geosci. 7 (12) (2014) 5543–5551.
[57] H.M. Jennings, Refinements to colloid model of C-S-H in cement: CM-II, Cem. [78] T. Nahhas, X. Py, N. Sadiki, Experimental investigation of basalt rocks as storage
Concr. Res. 38 (3) (2008) 275–289, URL https://www.sciencedirect.com/science/ material for high-temperature concentrated solar power plants, Renew. Sustain.
article/pii/S0008884607002761. Energy Rev. 110 (2019) 226–235, URL https://www.sciencedirect.com/science/
[58] M. Wyrzykowski, P.J. McDonald, K.L. Scrivener, P. Lura, Water redistribution article/pii/S1364032119302813.
within the microstructure of cementitious materials due to temperature changes [79] B. Anglin, R. Lebensohn, A. Rollett, Validation of a numerical method based on
studied with 1H NMR, J. Phys. Chem. C 121 (50) (2017) 27950–27962, http: Fast Fourier Transforms for heterogeneous thermoelastic materials by comparison
//dx.doi.org/10.1021/acs.jpcc.7b08141. with analytical solutions, Comput. Mater. Sci. 87 (2014) 209–217, URL https:
[59] J.-C. Mindeguia, I. Hager, P. Pimienta, H. Carré, C. La Borderie, Parametrical //www.sciencedirect.com/science/article/pii/S0927025614001219.
study of transient thermal strain of ordinary and high performance concrete, [80] V. Holla, G. Vu, J.J. Timothy, F. Diewald, C. Gehlen, G. Meschke, Computa-
Cem. Concr. Res. 48 (2013) 40–52, URL https://www.sciencedirect.com/science/ tional generation of virtual concrete mesostructures, Materials 14 (14) (2021)
article/pii/S0008884613000422. 3782, http://dx.doi.org/10.3390/ma14143782, URL http://dx.doi.org/10.3390/
[60] J. Wang, L. Lacarriere, A. Sellier, Multicomponent modelling of cement paste ma14143782.
dehydration under different heating rates, Mater. Struct. 52 (1) (2019) 6. [81] M. Rodríguez, C. López, I. Carol, J. Murcia, High temperature effects in mortar
[61] I. Ohno, K. Harada, C. Yoshitomi, Temperature variation of elastic constants of and concrete specimens using a meso-mechanical model with fracture based zero-
quartz across the alpha-beta transition, Phys. Chem. Miner. 33 (2006) 1–9. thickness interface elements, Comput. Plast. XI - Fundam. Appl. COMPLAS XI
[62] F. Homand-Etienne, R. Houpert, Thermally induced microcracking in granites: (2011) 1596–1605.
characterization and analysis, Int. J. Rock Mech. Min. Sci. Geomech. Abstr. [82] S. Li, G. Wang, Introduction to Micromechanics and Nanomechanics, World
26 (2) (1989) 125–134, URL https://www.sciencedirect.com/science/article/pii/ Scientific, 2008.
0148906289900016. [83] R. Schapery, Thermal expansion coefficients of composite materials based on
[63] H. Sun, Q. Sun, W. Deng, W. Zhang, C. Lü, Temperature effect on mi- energy principles, J. Compos. Mater. 2 (1968) 380–404.
crostructure and P-wave propagation in Linyi sandstone, Appl. Therm. Eng. [84] S. Timoshenko, J. Goodier, Theory of Elasticity, second ed., McGraw-Hill Book
115 (2017) 913–922, URL https://www.sciencedirect.com/science/article/pii/ Company, 1951.
S1359431117301552. [85] M. Königsberger, B. Pichler, C. Hellmich, Micromechanics of ITZ–aggregate
[64] A. Gudmundsson (Ed.), Rock Fractures in Geological Processes, Cambridge interaction in concrete Part I: Stress concentration, J. Am. Ceram. Soc. 97 (2014)
University Press, 2011. 535–542, http://dx.doi.org/10.1111/jace.12591.
[65] J. Emanuel, J. Hulsey, Prediction of the thermal coefficient of expansion of [86] L. Phan, N. Carino, Fire performance of high strength concrete: Research needs,
concrete, J. Am. Concr. Inst. 74 (1977) 149–155. 2000.
[66] R. Mao, X. Mao, L. Zhang, R. Liu, Effect of loading rates on the characteristics [87] C. Castillo, A. Durrani, Effect of transient high temperature on high-strength
of thermal damage for mudstone under different temperatures, Int. J. Min. Sci. concrete, ACI Mater. J. 87 (1990) 47–53.
Technol. 25 (5) (2015) 797–801, URL https://www.sciencedirect.com/science/ [88] L. Phan, Fire Performance of High-Strength Concrete: A Report of the
article/pii/S2095268615001299. State-of-the-Art, Tech. Rep., National Institute of Standards and Technology,
[67] A. Török, A. Török, The effect of temperature on the strength of two different 1996.
granites, Central Eur. Geol. 58 (2015) 356–369. [89] T. Yin, Y. Chen, X. Li, Q. Li, Effect of high temperature and strain rate on the
[68] A. Jay, The thermal expansion of quartz by X-ray measurements, Proc. R. Soc. elastic modulus of rocks: a review, Int. J. Earth Sci. 110 (8) (2021) 2639–2660.
Lond. A 142 (846) (1933) 237–247, http://dx.doi.org/10.1098/rspa.1933.0165. [90] V. Kodur, Properties of concrete at elevated temperatures, in: I.G. Raftoyiannis,
[69] J. Svidró, A. Diószegi, J.T. Svidró, The origin of thermal expansion differences in R. Kacianauskas, J. Wang (Eds.), ISRN Civ. Eng. 2014 (2014) 468510.
various size fractions of silica sand, Int. J. Cast Met. Res. 33 (6) (2020) 242–249, [91] C.-J. Haecker, E. Garboczi, J. Bullard, R. Bohn, Z. Sun, S. Shah, T. Voigt,
http://dx.doi.org/10.1080/13640461.2020.1838078. Modeling the linear elastic properties of Portland cement paste, Cem. Concr.
[70] A. Török, M. Hajpál, Effect of temperature changes on the mineralogy and Res. 35 (10) (2005) 1948–1960.
physical proper ties of sandstones. A laboratory study, Restor. Build. Monum. [92] F. Furumura, T. Abe, Y. Shinohara, Mechanical properties of high strength
11 (2005). concrete at high temperatures, in: Proceedings of the Fourth Weimar Workshop
[71] S.-Q. Yang, P. Ranjith, H.-W. Jing, W.-L. Tian, Y. Ju, An experimental on Performance Concrete: Material Properties and Design, HAB, 1995, pp.
investigation on thermal damage and failure mechanical behavior of gran- 237–254.
ite after exposure to different high temperature treatments, Geothermics [93] S. Huismann, Materialverhalten von hochfestem Beton unter thermomechanischer
65 (2017) 180–197, URL https://www.sciencedirect.com/science/article/pii/ Beanspruchung (Ph.D. thesis), Deutscher Ausschuss für Stahlbeton e.V. (DAfStb),
S0375650516301110. Berlin, 2010, (in german).
[72] C. Saiang, K. Miskovsky, Effect of heat on the mechanical properties of selected [94] E. Binder, M. Königsberger, R. Díaz Flores, H.A. Mang, C. Hellmich, B.L.
rock types - A laboratory study, in: Harmonising Rock Engineering andthe Pichler, Thermally activated viscoelasticity of cement paste: Minute-long creep
Environment - Proceedings of the 12th ISRM International Congresson Rock tests and micromechanical link to molecular properties, Cem. Concr. Res.
Mechanics, 2012, pp. 815–820. 163 (2023) 107014, URL https://www.sciencedirect.com/science/article/pii/
[73] X.-b. Mao, L.-y. Zhang, T.-z. Li, H.-s. Liu, Properties of failure mode and S0008884622003064.
thermal damage for limestone at high temperature, Min. Sci. Technol. (China) [95] U. Diederichs, U. Jumppanen, V. Penttala, Material properties of high strength
19 (3) (2009) 290–294, URL https://www.sciencedirect.com/science/article/pii/ concrete at elevated temperatures, in: IABSE 13th Congress, Helsinki, 1988.
S1674526409600545. [96] S. Huismann, F. Weise, B. Meng, U. Schneider, Transient strain of high strength
[74] J.R. Goldsmith, Thermal stability of dolomite at high temperatures and pressures, concrete at elevated temperatures and the impact of polypropylene fibers, Mater.
J. Geophys. Res. 85 (B12) (1980) 6949–6954. Struct. 45 (5) (2012) 793–801.
[75] R. McIntosh, J. Sharp, F. Wilburn, The thermal decomposition of dolomite, [97] C. Davie, C. Pearce, K. Kukla, N. Bićanić, Modelling of transport processes
Thermochim. Acta 165 (2) (1990) 281–296, URL https://www.sciencedirect. in concrete exposed to elevated temperatures - An alternative formulation for
com/science/article/pii/004060319080228Q. sorption isotherms, Cem. Concr. Res. 106 (2018) 144–154, URL https://www.
sciencedirect.com/science/article/pii/S0008884616307001.

14