Cement and Concrete Research 56 (2014) 203–212

Contents lists available at ScienceDirect

Cement and Concrete Research

journal homepage: http://ees.elsevier.com/CEMCON/default.asp

Compressive strength and hydration processes of concrete with

recycled aggregates
Eduardus A.B. Koenders a,b, Marco Pepe c, Enzo Martinelli c,⁎
a
COPPE-UFRJ, Rio de Janeiro, Brazil
b
Microlab, Delft University of Technology, The Netherlands
c
Dept. of Civil Engineering, University of Salerno, Italy

a r t i c l e i n f o a b s t r a c t

Article history: This paper deals with the correlation between the time evolution of the degree of hydration and the compressive
Received 28 February 2013 strength of Recycled Aggregate Concrete (RAC) for different water to cement ratios and initial moisture condi-
Accepted 28 November 2013 tions of the Recycled Concrete Aggregates (RCAs). Particularly, the influence of such moisture conditions is inves-
tigated by monitoring the hydration process and determining the compressive strength development of fully dry
Keywords:
or fully saturated recycled aggregates in four RAC mixtures. Hydration processes are monitored via temperature
Concrete (E)
Recycled aggregates
measurements in hardening concrete samples and the time evolution of the degree of hydration is determined
Hydration (A) through a 1D hydration and heat flow model. The effect of the initial moisture condition of RCAs employed in
Temperature (A) the considered concrete mixtures clearly emerges from this study. In fact, a novel conceptual method is proposed
to predict the compressive strength of RAC-systems, from the initial mixture parameters and the hardening
conditions.
© 2013 Elsevier Ltd. All rights reserved.

1. Introduction
It is commonly known that the construction sector is characterised
by a significant demand for both energy and raw materials [1]. More-
over, recent research highlighted that the concrete industries' share to
green-house gas emissions is about 10% of its global production [2]
and almost half of it can be directly or indirectly related to the produc-
tion of cement [3]. Therefore, initiatives that aim at reducing the sectors'
enormous demand for energy and raw materials [4] should be devel-
oped and implemented massively, in order to enhance the sectors'
sustainability perspective. Developments such as the possibility to
pre-design structures for “deconstructing” or to reuse Construction
and Demolition Waste (CDW) [5] are nowadays pursued with the
same objective. As a matter of fact, due to the significant volumes of
concrete produced each year, developments that aim at reusing
recycled constituents, such as aggregates, fibres or binders, are rapidly
emerging. Particularly, partial replacement of the “natural” raw mate-
rials or substitution of regular Portland cement with lower CO2 impact
supplementary cementitious materials are gaining interest to reduce
of the environmental footprint of the building industry [6].
Moving forward from these initial considerations and keeping in
mind the guidelines about future recycling objectives recently issued
⁎ Corresponding author. Tel.: +39 089 964098; fax: +39 089 964045.
E-mail addresses: e.a.b.koenders@coc.ufrj.br, e.a.b.koenders@tudelft.nl
(E.A.B. Koenders), mapepe@unisa.it (M. Pepe), e.martinelli@unisa.it (E. Martinelli).
by the European Union [7], a wide experimental activity was carried
out at the Laboratory of Materials testing and Structures (LMS) of the
University of Salerno (Italy) [8]. Part of this experimental activity is
intended at observing the hydration process [9] that takes place during
the hardening phase of four Recycled Aggregate Concrete (RAC) mix-
tures characterised by a different degree of saturation of the Recycled
Concrete Aggregates (RCAs). In fact, it is widely recognised that one of
the key issues of using recycled aggregates in concrete is basically relat-
ed to the higher water absorption capacity of such aggregates with re-
spect to the corresponding natural materials generally used for
concrete production [10]. Thus, a higher amount of water possibly
absorbed by RCAs during mixing is supposed to affect the actual values
of water/cement ratio and, then, to have its consequences on both the
cement hydration process and the final mechanical properties of RAC
systems.
Beyond the mainly empirical investigations generally carried out to
observe the possible influence of recycled constituents on the mechan-
ical behaviour of concretes [11] and mortars [12], this paper proposes a
more fundamental approach by investigating the evolution of the hy-
dration processes and its potential correlation with the compressive
strength for the four concrete mixtures under consideration.
To this end, the four different RAC mixtures that are tested in com-
pression at different isothermal curing ages are described in Section 2.
Moreover, one sample of each mixture was monitored during the first
week of curing to measure the time evolution of temperature develop-
ing inside the concrete as a result of the exothermic cement hydration
reaction. Particularly, temperatures were measured on a 15-cm-edge

0008-8846/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.cemconres.2013.11.012
204 E.A.B. Koenders et al. / Cement and Concrete Research 56 (2014) 203–212

cubic sample of concrete poured in an insulating plastic box which – Saturated condition (SAT): the coarse aggregates are saturated for
realised semi-adiabatic conditions for the cement hydration reactions. 24 h in water and, before mixing, their surface was dried with a
Section 3 reports the main experimental results in terms of both cloth.
time evolution of temperatures and average compressive strength de-
As a matter of fact, natural aggregates used in ordinary concretes are
veloped in cubic specimens tested at different curing ages ranging
generally characterised by a low water absorption capacity and their
from 2 to 28 days. The temperature measurements obtained under
corresponding portion of “absorbed water” can easily be accounted for
the aforementioned semi-adiabatic conditions were intended to be
in the concrete mix design. On the contrary, a higher water absorption
employed for identifying a theoretical model, able to simulate the heat
capacity of RCAs clearly depends on their production process [17]. Par-
production and the time evolution of the degree of hydration in the
ticularly, internal damage and cracks, due to demolition and crushing,
cured concrete samples [14].
result in a non-negligible influence of RCAs water absorption capacity
Section 4 outlines the theoretical basis of such a model and gives full
on the concrete mix performance, in both the fresh state (in terms of
attention to the theoretical assumptions on which it is based. It shows
actual workability and rheological properties) and the hardened one
the relationship between the heat production of the hydration reactions
(in terms of mechanical properties) [10,13]. The processing of the
and defines the “degree of reaction” and the “degree of hydration” [15]
RCAs as considered in this study has led to the following ranges of the
concepts.
grain sizes:
It may be worth to mention that this research is finally aiming at
obtaining a quantitative relationship between the evolution of the – N3, nominal size 20–31.5 mm;
degree of hydration and the compressive strength for the concrete – N2, nominal size 10–20 mm;
mixes under consideration. As a matter of fact, similar relationships – N1, nominal size 4–10 mm;
were already found for ordinary concrete mixes, with ordinary natural – Sand, nominal size small than 4 mm.
aggregates [16,23–25]. This study aims at generalising such relation-
Table 1 describes the actual composition of the four concrete
ships by taking into account the influence of both water/cement ratio
mixtures characterised by the two aforementioned values of the nomi-
and the initial moisture conditions of recycled aggregates. In this regard,
nal water-to-cement ratios and the two initial moisture conditions. The
Section 5 describes the linear correlations emerged between the degree
amount of RCAs is kept constant to 30% of the total amount of
of hydration and the average compressive strength for the four concrete
aggregates, with a distribution over the different fractions, i.e. a total
mixtures investigated. The regular variation of such correlations, which
replacement of the coarse fraction, a partial replacement of the finer
are clearly affected by the nominal water/cement ratio and the initial
fraction, and no replacement of sand.
moisture conditions of RCAs, is the main finding of this paper and out-
The volume of water absorbed by the saturated aggregates is not
lines the conceptual approach for predicting the compressive strength
included in the calculation of the w/c ratio. The absorbed volumes are
of RAC for different mix parameters and moisture conditions.
estimated by considering the amount of the various aggregate fractions
and their respective water absorption capacity, which was determined
2. Materials and methods
on both the natural and recycled aggregates [8]. Table 1 reports the
amounts of water absorbed by recycled aggregate in saturated condi-
The two following subsections describe the materials employed for
tions, apart from the regular mixing water.
producing the test specimens and the experimental methods carried
Finally, a common Portland cement, type CEM I 42.5 R [19], was used
out this study.
as a binder in all concrete mixes.
2.1. Materials
2.2. Experimental methods
Four different concrete mixtures with a 30% recycled aggregate
For all four tested concrete mixtures, the hydration process was
replacement are considered in the present study with emphasis on
monitored by measuring the temperature evolution in the centre of a
the relationship between the compressive strength development and
concrete cube during the first seven days after casting. To this end, a
the corresponding time evolution of the degree of hydration. Two key
cubic sample of each concrete mix was cured in semi-adiabatic condi-
parameters, i.e. the water to cement ratio and the initial moisture condi-
tions with the aim of measuring the temperature evolution and to use
tion of the recycled aggregates, are considered in this study and their
this data as input for the proposed simulation model to calculate the
impact on the final concrete quality is investigated. Particularly, their in-
hydration process. For this purpose, such a concrete sample was cast
fluence on both the hydration reaction and the evolution of the com-
within an insulated mould with a rib size of 150 mm (Fig. 1). Four out
pressive strength of concretes made with recycled aggregates is
of the six side planes of the cubic sample bordered with a thick layer
investigated. Therefore, two mixtures are designed with a different
(about 100 mm) of insulation material, whereas two planes were insu-
nominal water/cement ratio, i.e. 0.45 and 0.60. Moreover, for the
lated with a significantly thinner layer (of about 40 mm). Therefore, the
recycled aggregates two different initial moisture conditions were
heat produced by the hydration reaction was supposed to be mainly
realised, according to the two following definitions:
dissipated through the two faces bordering with the thinner layers of in-
– Dry condition (DRY): the coarse aggregates are dried for 24 h in an sulation material. Since these two faces are placed opposite from each
oven with a constant temperature of 100 °C; other (namely, the top and bottom of the system depicted in Fig. 1) a

Table 1
Key parameters about concrete mixes.

Mix CEM I Mixing water Absorped water w/c Recycled concrete Natural aggregates [kg/m3]
[kg/m3] [l/m3] [l/m3] aggregates [kg/m3]

N3 N2 N2 N1 Sand

0.45DRY 410 185 0 0.45 400 100 300 130 760

0.45SAT 410 185 21 0.45 400 100 300 130 760
0.60DRY 310 185 0 0.60 400 100 300 130 850
0.60SAT 310 185 22 0.60 400 100 300 130 850
 E.A.B. Koenders et al. / Cement and Concrete Research 56 (2014) 203–212
Fig. 1. Insulated mould inducing semi-adiabatic boundary conditions on the curing con-
crete sample.
1D heat flow was actually supposed to occur. The evolving exothermic
reaction under semi-adiabatic conditions leads to a temperature en-
hancement in the cube, which also affects the kinetics of the cement hy-
dration reaction. The evolution of the hydration temperature inside the
cube was monitored with respect to the ambient temperature using two
thermocouple wires (Fig. 2).
Moreover, apart from the one cured within the insulated mould, ten
other concrete samples of each mixture were cast and cured in a water
bath under isothermal conditions at a temperature TR of 22°C. Couples
of these samples were tested in compression (Fig. 3) after 2, 3, 7, 14
and 28 days of hardening, with the aim of determining the average
strength at the aforementioned curing ages.
3. Results
This section reports the results obtained from the experimental pro-
gramme described in Section 2. To this end, the two following subsec-
tions outline the results of the temperature and compressive strength
measurements, respectively.
3.1. Temperature development
The time evolution of temperature is measured for the four concrete
mixes described in Table 1. The different results obtained for such mix-
tures reveal the actual influence of the mix ingredients and moisture
conditions of the recycled aggregates on the cement hydration
reaction. Fig. 4 shows the temperature development measured in the
two concrete samples with a w/c = 0.45. It can be observed that the
sample with dry aggregates reaches the highest peak value for the
Fig. 2. Temperature measurement inside insulated mould.
205
Fig. 3. Compression test.
temperature as well as a slightly higher rate of the hydration reaction
for the ascending branch. This could likely be attributed to the relatively
lower amount of water in this mix with dry aggregates. As a matter of
fact, the role of the w/c ratio on the hydration process is well known
and this observation is basically in line with this well-established
knowledge [20]. On the contrary, the sample with saturated recycled
aggregates exhibited a slightly longer reaction period, which could like-
ly be attributed to the higher amount of available water, and a slower
temperature decay in the post-peak branch.
Similar considerations can be drawn from Fig. 5 for the two concrete
mixtures with a w/c of 0.60. Higher peak values for the temperature
were achieved in the concrete made with dry aggregates, but in this
case the post-peak temperature decay observed in the two samples
showed similar rates.
3.2. Compressive strength development
The influence of the two parameters under consideration, i.e. the w/c
factor of the mix and the moisture condition of the recycled aggregates,
on the hydration reaction of the four mixtures (Table 1) is expected to
have a clear effect on the time evolution of the compressive strength.
Fig. 6 shows the results of the compressive strength development in
terms of average compressive cube strength Rcm, according to the pro-
cedure described in Section 2.2. The results confirm that the initial mois-
ture content of the recycled aggregates plays an important role in the
development of the compressive strength of concrete (Fig. 6). In this
regard, dry aggregates lead to higher values of the Rcm over the investi-
gated time span of 28 days, for both w/c ratios considered in this study.
Moreover, the compressive strengths measured from the samples with
recycled aggregates in dry condition and a w/c = 0.6 (denoted as
0.60DRY) lost ≈60% of their 28-day strength when applying saturated
recycled aggregates. For the corresponding tests done for the w/c ratio
30
Temperature increase ΔT=T-TR [°C]
0.45 DRY
25
0.45 SAT
20
15
10
0
0 12 24 36 48 60 72
Time t [h]
Fig. 4. Time evolution of temperature on curing concrete samples (w/c = 0.45).
206 E.A.B. Koenders et al. / Cement and Concrete Research 56 (2014) 203–212

30
total amount of heat Q(t) produced up to the time t, the degree of hy-
Temperature increase ΔT=T-TR [°C]

dration can be effectively determined through the following analytical

25
definition:

20 Q ðt Þ
0.60 DRY α h ðt Þ ¼ ; ð1Þ
Q max
15
0.60 SAT
where Qmax is the heat potentially produced after reaction of the total
10 amount of cement present in the concrete mix.

5 4.1. Hydration processes occurring under ideal adiabatic conditions

0 As a matter of fact, the rate of heat production is significantly influ-

0 12 24 36 48 60 72
enced by the temperature developed inside a concrete which, in turn,
Time t [h]
is affected by the produced reaction heat itself. Particularly, a simple an-
Fig. 5. Time evolution of temperature on curing concrete samples (w/c = 0.60). alytical relationship can be settled under adiabatic conditions between
the temperature variation ΔTa(t) and the reaction heat Qa(t) produced
of 0.45 with recycled aggregates under dry and saturated conditions this at time t [9]:
reduction was ≈ 50%. As a matter of fact, both mixtures made with
CQ a ðt Þ CQ max
saturated aggregates were characterised by a very low compressive ΔT a ðt Þ ¼ ¼ α a;h ðt Þ; ð2Þ
ρc cc ρc cc
strength, apparently as a result of the higher total amount of water
(mixing water + absorbed water) which potentially led to a higher
where ρc is the mass per unit volume of concrete, C the amount of
value of the actual (local) w/c ratio, depending on the possible release
cement per unit volume of concrete mix, cc the specific heat of curing
of the absorbed water from the aggregates into the mix. However, this
concrete and αa,h(t) = Qa(t)/Qmax is the corresponding degree of
ability has not been explicitly considered in this study.
hydration (the subscript “a” recalls the occurrence of “adiabatic” condi-
The correlation between the hydration processes and the resulting
tions). Thus, the actual temperature Ta(t) developed inside a concrete
compressive strength can be figured out by calculating the evolution
sample cured under adiabatic conditions can be easily written as
of the degree of hydration from the temperature curves as presented
follows:
in Figs. 4 and 5 and by comparing the results with the strength measure-
ments reported in Fig. 6. A deeper investigation towards the quantita- C
tive relationship between these hydration-related measurements and T a ðt Þ ¼ T R þ ΔT a ðt Þ ¼ T R þ Q ðt Þ; ð3Þ
ρc cc a
the compressive strength is proposed in the following sections.
where TR is the room temperature, which is supposed to be also the
4. Theoretical model for simulating the hydration process temperature of the concrete constituents during the initial mixing
of concrete phase. Based on the definition given in Eq. (1) and the specific relation-
ship (2) between temperature variation and reaction heat under adia-
The chemical reaction process that develops inside concrete during batic conditions, a relationship can be developed to determine the
setting and hardening is due to the hydration of cement and the forma- degree of hydration under such conditions, through temperature mea-
tion of hydration products. This hydration reaction is strongly exother- surements:
mic in nature and results in the production of a significant amount of
heat that is proportional to the development of the chemical reaction Q a ðt Þ ΔT a ðt Þ
α a;h ðt Þ ¼ ¼ α ; ð4Þ
process. The actual state of such a reaction at time t can be described Q max ΔT a; max a;h; max
by the so-called “degree of hydration” αh(t) which is a dimensionless
quantity defined as the ratio between the amount of cement that has αa,h,max being the asymptotic value (for t → ∞) of the degree of hydra-
reacted at the time t and the total amount of cement in a concrete tion αa,h developed under adiabatic conditions for the concrete samples
mix [9]. Since the amount of hydrated cement is proportional to the under consideration.

70

60

50
Rcm [MPa]

40

30

20

10

0
0.45DRY 0.60DRY 0.45SAT 0.60SAT
2 days 32.87 25.44 13.77 6.47
3 days 35.77 28.94 15.47 8.19
7 days 41.62 32.07 18.55 9.85
14 days 43.61 35.75 21.17 14.41
28 days 44.56 34.99 22.75 13.99

Fig. 6. Time evolution of the average compressive strength on concrete samples.

 E.A.B. Koenders et al. / Cement and Concrete Research 56 (2014) 203–212 207

Experimental observations carried out on the concrete samples
cured under adiabatic conditions led to the following analytical rela-
tionship to possibly describe the time evolution of the produced reac-
tion heat Qa(t) [9]:
Q a ðt Þ ¼ Q a; max e ð t Þ ;
β
 − τ
where the Qa,max⁎ ≤ Qmax is the actual amount of heat produced at the
end of the reaction (realised under adiabatic conditions), whereas τ
and β control the shape of the function which describes the time evolu-
tion of Qa(t).
Finally, as a matter of principle, the following relationship can be
written to relate the two quantities Qmax⁎ and Qmax:


 quantity V(T) for (at least) two different temperatures. However, the
Q a; max ¼ Q max  lim α a;h ðt Þ ¼ Q max α a;h; max :
t→∞
4.2. Hydration processes developing under “semi-adiabatic” conditions
Since concrete curing generally occurs under non-adiabatic (or
semi-adiabatic) conditions, the development of the hydration reaction
will be actually affected also by the temperature field, in both time
and space, and depends on the heat flow inside the concrete and the ex-
change of heat with the environment. As a matter of principle, the non-
stationary problem of heat transfer that takes place within a concrete
sample during the hydration process can be described by a Partial Dif-
ferential Equation (PDE). In this study, a PDE with a one dimensional
space domain is considered that represents the situation as presented
in Fig. 7, and can be written as follows [14]:
2
∂T ∂ T
ρc cc ¼ λc  2 þ qc ðx; t Þ;
∂t ∂x
dissipated) by the same hydration reaction. Particularly, the effect of the
(absolute) temperature T on the kinetics V(T) of a wide class of chemical
properties (among which the cement hydration reaction) can be
expressed through the well-known “Time–Temperature equivalence”
principle as proposed by Arrhenius [9]:
ð5Þ
E
−RT
A
V ðT Þ ¼ AV e ; ð9Þ
where AV is a reference value of the reaction rate, EA is the so-called ap-
parent activation energy (in J/mol) of the chemical reaction under con-
sideration and R is the universal gas constant (R ≈ 8.3145 J/molK).
As a matter of principle, the values of AV and EA which characterise a
certain chemical process could be determined by measuring the rate
ð6Þ
applications proposed in Section 5 of this paper will be based on assum-
ing EA = 33,000 J/mol, according to results available in the scientific lit-
erature [9]. Moreover, Eq. (9) can be employed to express the
relationship between the actual rate of heat production qc(T) and the
corresponding heat rate qa(Ta), which would have been developed by
the reaction under adiabatic conditions.
Therefore, if a hydration process develops under semi-adiabatic con-
ditions and has reached a degree of hydration αh(t) = Q(t)/Qmax, an
equivalent time teq can be determined from the corresponding adiabatic
process for the same value of αh(t) and, then, for an equal amount of
heat production Qa(teq):
 
Q a t eq ¼ α h ðt ÞQ max : ð10Þ
and from the following analytical expression results teq by having
assumed the relationship (5) to describe the ideal adiabatic hydration
ð7Þ
process (Fig. 8):

where λc is the thermal conductivity coefficient of concrete and qc(x,t)  β

is the rate of heat production, which can be expressed as follows: − τ
α h ðt Þ τ
¼ ⇒t eq ¼ n io 1 : ð11Þ
t eq
e h
α h; max
− ln ααh ðt Þ
β

∂Q c h; max
qc ðx; t Þ ¼ C ; ð8Þ
∂t
Although, by definition, the amount of heat produced in a real semi-
where, in principle, Qc = Qc(x,t) is the function which describes the
adiabatic process Qc(t) and the corresponding ideal adiabatic one
total heat produced in a point x of the 1D domain at time t, under gen-
Qa(teq) are the same, two different temperatures, denoted as T(t) and
eral boundary conditions.
Ta(teq), develop in the two aforementioned processes as a result of
As a matter of fact, the rate of heat production qc is strongly affected
their different boundary conditions. However, the Arrhenius principle,
by the temperature T developed inside the concrete sample. Thus, qc
analytically interpreted by Eq. (9), can be applied to settle a relationship
and T fields are strictly related to each other, as the latter controls the re-
between the two rates of heat production (namely, qc and qa) and two
action kinetics and results from the actual heat produced (and possibly
different temperatures (namely, T(t) and Ta(teq)) developed in the
two corresponding reactions, for the same degree of hydration
αh(t) = αa,h(teq):

EA
− E T ðt Þ−T ðt Þ
q ½T ðt Þ e RT ðtÞ − A  a eq
h c  i ¼ EA
¼ e R T a ðteq ÞT ðtÞ : ð12Þ
qa T a t eq −
e RT a ðteq Þ

The assumption of the Arrhenius principle can be accepted in the

case of chemical reactions that are clearly characterised by a unique
and dominating reaction phase [18], as is the case of concrete mixtures
with Portland cement (CEM I or CEM IV according to EN-197-1 [19]). In
such cases, the following set of differential equations can be derived by
solving Eq. (12) with respect to qc[T(t)] and introducing the corre-
sponding expression in the heat transfer PDE (8):

∂T ∂2 T h  i −EA T a ðteq Þ−T ðtÞ

ρc cc ¼ λc 2 þ qa T a t eq e R T a ðteq ÞT ðtÞ ; ð13Þ
Fig. 7. Schematic of the one-dimension domain considered for the hydration process. ∂t ∂x
208 E.A.B. Koenders et al. / Cement and Concrete Research 56 (2014) 203–212

where the rate of heat production qa can be defined as follows: 5. Analysis of results

h  i dQ a
qa T a t eq ¼ C j : ð14Þ The experimental results reported in Section 3 are employed for
dt t¼t eq identifying the theoretical model outlined in Section 4 for the four
RAC mixtures under consideration and, then, simulating their hydration
and the function Qa(t) is described by the analytical expression (5). process and determining the time evolution of the degree of hydration
The initial starting condition of the temperature field T(x,t = 0) rep- αh(t) under isothermal conditions. The two following subsections de-
resents the initial mix temperature inside the sample. Particularly, in scribe the inverse identification procedure and explain the methodolo-
the mixing phase, the temperature of the concrete constituents is gy to correlate αh(t) and Rcm(t). Finally, a further subsection is devoted
equal to the room temperature TR: to examine the obtained correlations and the way in which they are in-
fluenced by the two key parameters considered in the present study.
T ðx; t ¼ 0Þ ¼ T R : ð15Þ
5.1. Indirect identification of hydration reaction processes
Moreover, the two insulation layers at x = ±L/2 (see Fig. 7) induce
semi-adiabatic conditions, and the following relationships apply to the The model formulated in Section 4 and, particularly, the solution
whole reaction process: of the integral-differential Eqs. (13)–(14), with the applied initial
and boundary conditions (15) and (16)–(18), respectively, can be
T ðx ¼ −L=2; t Þ ¼ T left ðt Þ employed to simulate the time evolution of the temperature field
ð16Þ
T ðx ¼ L=2; t Þ ¼ T right ðt Þ: which develops inside the hardening concrete [14]. The resulting
theoretical simulation of the temperature field can be symbolically
where Tleft(t) and Tright(t) are the two temperatures developed at the denoted as follows:
inner surface of the insulation layer and can be related to the room tem-
 
perature TR by imposing continuity of the heat flow through the same T th ¼ T th x; t; qr ; q f ð19Þ
faces:

T left ðt Þ−T R where the vectors qr and qf contain the two sets of parameters listed
∂T
λp  ¼ λc  j ; ð17Þ below:
tp ∂x x¼−L=2
h i
q f ¼ τ; β; λp ð20Þ
T right ðt Þ−T R ∂T
λp  ¼ −λc  j : ð18Þ
tp ∂x x¼L=2 h i
qr ¼ T R ; Q max ; α h; max : ð21Þ
where tp and λp are the thickness and the heat conductivity of the insu-
lation layer. In the following, the values for the parameters of the vector qr are as-
It is worth highlighting that the above solutions neglect the convec- sumed “a priori”. Particularly,
tion effect possibly developing in the neighbourhood of the outer sur-
faces of the insulation layer. The temperature measured on those – TR is the room temperature which has been measured for all the four
surfaces is, in fact, rather close to TR during the whole reaction time. mixes;
However, in principle, a more correct application of Eqs. (17) and (18) – Qmax = 450 kJ/kg, as generally assumed for CEM I [16];
could be achieved by replacing TR with the actual temperature mea- – αh,max is determined as a function of the w/c ratio according to the
sured on both aforementioned surfaces. following analytical relationship [20,21]:
Finally, it should be emphasised that Eqs. (13)–(14) are integral-
1:031  w=c
differential in nature, due to the presence of the auxiliary variable teq de- α h;max ¼ : ð22Þ
0:194 þ w=c
fined as a function of the degree of hydration αh(t) via Eq. (11). The lat-
ter depends of the total amount of produced heat Qc(x,t) which is, in
fact, an integral quantity. Therefore, the solution of Eqs. (13)–(14) can Since the time evolution of the temperature T developed in the centre
only be carried out numerically. The analyses reported in the following of the concrete sample at x = 0 in the equivalent 1D domain (see Fig. 7)
are based on a finite-difference scheme, whose detailed formulation and was monitored for all four mixtures cast in the insulated mould under
validation is described in [14]. semi-adiabatic conditions (Figs. 1 and 2), the following objective
error parameter can be defined for the difference between the experi-
ΔQa[Ta(teq)] qa[Ta(teq)]Δt mental measurements Texp and the outcome of the numerical simulation
Hydration Heat - Q

Tth(x = 0, tk; qr, qf):

ΔQc[T(t)] qc[T(t)]Δt
  Xn h  i2
ðkÞ
Δ qr ; q f ¼ Texp −T th x ¼ 0; tk ; qr ; q f ð23Þ
k¼1

teq t Time - t where T(k)

exp is the temperature measured from the experiments at the
time tk in the reference position (x = 0).
Therefore, the indirect identification procedure of the model pre-
Temperature - T

αh T
sented in Section 4 can be written as the following unrestrained optimi-
T(t) Adiabatic sation problem:
Semi-Adiabatic h  i
q f ¼ argmin Δ qr ; q f ; ð24Þ
Ta(teq) qf

which leadsh to the values

i for the parameters of the adiabatic hydration
Fig. 8. Hydration temperature and heat in semi-adiabatic and ideally adiabatic conditions. curve q f ¼ τ; β; λp and results in the best approximation between the
 E.A.B. Koenders et al. / Cement and Concrete Research 56 (2014) 203–212 209

corresponding numerical simulation and the measured temperature 353

curve. Particularly, the first two parameters (τ,β) were independently 343
calibrated for each one of the four mixtures under consideration, as
they determine the hydration heat produced under adiabatic conditions 333

according to Eq. (5). On the contrary, a common value of λp was cali-
brated for all the specimens under consideration, as it is directly related
to the insulation material and sets the effect of the semi-adiabatic con-
ditions on the shape of the simulated temperature curve.
Table 2 reports the results of the parameters identified by solving
the optimisation problem described by Eq. (24). As τ is related to the
contra flexure point of the adiabatic hydration curve described by
Eq. (5), this parameter can also be indicative for the setting period
of the concrete (dormant stage). The results in Table 2 show a consis-
tent trend for the values obtained from the identified parameters.
Both mixtures with a w/c = 0.60 show a higher value for τ in comparison
with the w/c = 0.45 mixtures. Moreover, the values for the parameter β,
which controls the curvature of Eq. (6), shows only slight variations and
indicates that the shape of the curves of all four mixtures is rather similar.
The comparison between the temperature measured in semi-
adiabatic conditions and the corresponding simulations carried out by
means of the identified models are represented in Figs. 9 to 12. The re-
sults show that the theoretical model is very much capable of simulating
the time–temperature curves for all four mixtures considered. In some
cases the theoretical curve even completely coincides with the experi-
mental measurements. However, it can also be observed that the simu-
lated curves are underestimating the temperature development during
first few hours of hydration (dormant stage). This result is a conse-
quence of the applied empirical formulation for the adiabatic hydration
curve expressed by Eq. (5), which does not explicitly describe this stage
of the reaction process.
5.2. Degree of hydration and compressive strength
Adopting the identified parameters τ and β, that represent the adia-
batic hydration curves for each of the four concrete mixtures, in prin-
ciple, other (non-adiabatic) hydration processes can be simulated by
solving Eqs. (13)–(14) and by adjusting the relevant boundary condi-
tions. Particularly, the concrete specimens that have been tested in
compression were cured in a water bath under isothermal conditions,
and their hydration process can now be simulated by solving
Eqs. (13)–(14) with the identified parameters and applying the bound-
ary conditions that correspond to the isothermal curing process:
T ðx; t ¼ 0Þ ¼ T R ;
T ðx ¼ −L=2; t Þ ¼ T R
T ðx ¼ L=2; t Þ ¼ T R :
Temperature [K]
323
313
303
293
0.45DRY Room temperature
283 Adiabatic (Simulation) Semi-Adiabatic (Simulation)
Semi-Adiabatic (Experimental)
273
0 12 24 36 48 60 72
Time [h]
Fig. 9. Experimental–theoretical comparison in terms of temperature evolution during the
hydration process (0.45DRY).
the curve obtained for adiabatic conditions is proportional to the corre-
sponding temperature curve presented in Fig. 9. Moreover, the time
evolution of αh(t) under semi-adiabatic conditions is initially close to
the ideal adiabatic process and, then, diverts from it as a result of the re-
duction in temperature due to the dissipation of heat through the insu-
lation layer. Finally, the αh(t) curve simulated for the hydration reaction
under isothermal conditions is beneath the two aforementioned curves
and shows the most moderate development of the hydration process: a
lower heat production associated with a lower temperature develop-
ment under isothermal conditions is the main reason for this. The con-
siderations addressed so far are also applicable for the other three
concrete mixtures, whose results are presented in Figs. 14 to 16.
With the time evolution of the degree of hydration being simulated
for the four batches of specimens tested in compression, a relationship
could be established between the simulated values of αh(t) and the cor-
responding experimental compressive strength results Rcm(t), for each
day these concrete mixtures have been tested.
Fig. 17 shows the results of the correlation that emerged between
the two aforementioned quantities. Although such a linear correlation
between αh(t) and Rcm(t) was already observed and documented in
the scientific literature for ordinary concrete [16,22–25], the results ob-
tained from this study point out that this approach can also be applied
for concretes with recycled aggregate replacements, and implicitly ac-
counts for the water absorption and moisture condition issues of such
ð25Þ materials. However, the potential effect of absorption and moisture
contribution on the prevailing degree of hydration of RAC systems is
not explicitly considered in this study. This relevant issue is part of a cur-
ð26Þ
rently running research activity.

Eqs. (13)–(14) and (25)–(26) were actually solved for all the four
353
mixes by considering TR = 22 °C and the corresponding time evolution
of the degree of hydration αh(t) under isothermal conditions was ob- 343
tained via Eq. (1).
333
Fig. 13 shows the results of the simulated time evolution for the de-
Temperature [K]

gree of hydration for mixture 0.45DRY. In order to show the effect of the 323
temperature development on the hydration process, results are pre-
sented for adiabatic, semi-adiabatic and isothermal conditions. Due to 313
the absence of heat dissipation and according to Eq. (4), the shape of 303

Table 2 293
0.45SAT Room temperature
Parameters calibrated through indirect identification of the hydration model.
283 Adiabatic (Simulation) Semi-Adiabatic (Simulation)
System parameters MIX w/c αh,max Qmax τ β Semi-Adiabatic (Experimental)
[kJ/kg] [h] 273
0 12 24 36 48 60 72
3
ρc cc = 2500 kJ/m K 0.45DRY 0.45 0.70 450 5.0 1.50
Time [h]
λc = 2.5 W/mK 0.45SAT 0.45 0.74 450 6.8 1.10
tp = 0.04 m 0.60DRY 0.60 0.76 450 9.5 1.60
Fig. 10. Experimental–theoretical comparison in terms of temperature evolution during
λp = 0.072 W/mK 0.60SAT 0.60 0.80 450 8.0 1.60
the hydration process (0.45SAT).
210 E.A.B. Koenders et al. / Cement and Concrete Research 56 (2014) 203–212

353 1.0
MIX0.45DRY
0.9
343

Degree of hydtarion- αh
0.8
333 Adiabatic Semi-Adiabatic [TR=22°C]
Temperature [K]

0.7
323 0.6

313 0.5
Isothermal [TR=22°C]
0.4
303
0.3
293
0.60DRY Room temperature 0.2
283 Adiabatic (Simulation) Semi-Adiabatic (Simulation) 0.1
Adiabatic (Experimental) Semi-Adiabatic (Experimental) 0.0
273 0 12 24 36 48 60 72
0 12 24 36 48 60 72
Time-t [h]
Time [h]
Fig. 13. Time evolution of the degree of hydration—0.45DRY.
Fig. 11. Experimental–theoretical comparison in terms of temperature evolution during
the hydration process (0.60DRY).

When looking again to the results presented in Fig. 17 two explicit

Nevertheless, the results of the present study in terms of the
emerged linear correlations between the compressive strength devel-
opment and the time evolution of the degree of hydration (Fig. 17),
can be simulated by the family of simple relationships formulated as
follows:
α‐α 0
Rcm ¼ Rcm;max : ð27Þ
1‐α 0
where the values for the best fit (Rcm,max, α0) are determined through a
least-square regression method for the four mixtures and are reported
in Table 3. This formula can be used to correlate the compressive
strength with the degree of hydration relationship and implicitly ac-
counts for the effect of both w/c ratio and moisture conditions of the
recycled aggregates on the hydration reaction of concrete mixtures
considered.
5.3. Remarks about the physical significance of the emerged
linear correlations
A more detailed analysis makes clear that the results presented in
Fig. 17 actually show the compressive strength development of RAC ver-
sus the degree of hydration of the paste inside the RAC system. This ob-
servation makes the paste strength the critical component of RAC
systems, and not the strength of the recycled aggregate. Therefore,
what is governing in this respect is the effect of RCAs on the strength
of the paste. This controls the final strength of a RAC system.
353
343
333
phenomena can be observed that potentially affect the strength of the
paste, which are:
1. the relative weakness of recycled aggregate particles relative to nat-
ural aggregate particles and its consequence for the internal stress
distribution in the paste;
2. the potential interaction of the absorbed water of recycled aggre-
gates and its consequence for the (local) water to cement ratio of
the paste.
The first phenomenon is caused by the fact that recycled aggregate
particles, in general, still contain portions of old (damaged) cement ma-
trix that consists of anhydrous cement and hydration products that
form a porous microstructure. Along with the embedded natural parti-
cle fractions, recycled aggregates show a much lower stiffness than
the solid natural ones. This reduction in stiffness might introduce larger
internal deformations in the RAC paste leading to higher stress concen-
trations, which can reduce the overall strength of RAC systems. Howev-
er, in this study the total aggregate volume and the recycled aggregate
replacement ratio were similar, i.e. 30%, and, therefore, this is not the
reason for the observed differences in compressive strength capacity.
The second phenomenon is related to the water absorption of
the recycled aggregate particles and its potential effect on the nominal
w/c ratio. Due to a possible release of absorbed water from the recycled
aggregates into the paste during mixing, an increase of the (local) w/c
ratio might occur that weakens the paste in RAC systems, causing a re-
duction of the paste strength, and with this, an observed reduction of
the compressive strength of the RAC. From Fig. 17, it can be seen that
this effect may be significant. Based on this, mixing concrete with satu-
rated recycled aggregates (SAT) will lead to a higher actual value of the
1.0
MIX0.45SAT
0.9
Degree of hydtarion- αh

0.8
Temperature [K]

Adiabatic Semi-Adiabatic [TR=22°C]

323 0.7
0.6
313
0.5
303 Isothermal [TR=22°C]
0.4
293 0.3
0.60SAT Room temperature
283 Adiabatic (Simulation) Semi-Adiabatic (Simulation) 0.2
Semi-Adiabatic (Experimental) 0.1
273
0 12 24 36 48 60 72 0.0
Time [h] 0 12 24 36 48 60 72
Time-t [h]
Fig. 12. Experimental–theoretical comparison in terms of temperature evolution during
the hydration process (0.60SAT). Fig. 14. Time evolution of the degree of hydration—0.45SAT.
 E.A.B. Koenders et al. / Cement and Concrete Research 56 (2014) 203–212 211

1.0 60

Compressive strength - Rcm [MPa]

MIX0.60DRY 0.45DRY
0.9
50
Degree of hydtarion-αh

0.8
Adiabatic Semi-Adiabatic [TR=22°C]
0.7 40
0.6 0.60DRY
0.5 30
Isothermal [TR=22°C]
0.4 0.45SAT
20
0.3
0.2 10
0.60SAT
0.1
0
0.0 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
0 12 24 36 48 60 72
Time-t [h] Degree of hydration - αh

Fig. 15. Time evolution of the degree of hydration—0.60DRY. Fig. 17. αh − Rcm correlations.

water to cement ratio of the RAC paste, than when mixing concrete with
dry recycled aggregates (DRY). Besides this, dry recycled aggregates
may also absorb part of the mixing water and even reduce the (local) ac-
tual w/c ratio in RAC systems. These two opposite effects induced by
the moisture condition of the recycled aggregates will also be reflected
by the development of the compressive strength. Therefore, for the
compressive strength results presented in Fig. 17, the following should
be considered:
– a higher slope obtained in the correlations refer to the mixes with
dry aggregates (whose actual values of w/c can be significantly
lower than the nominal one), and
– a lower slope for the mixes with saturated aggregates, whose actual
value of w/c is certainly higher than the nominal one.
Finally, the proposed curves can be intended as an operational meth-
od to determine the actual values of the compressive strength for differ-
ent w/c ratios and different moisture conditions of the recycled
aggregates inside RACs. Further data, also covering the quality of
recycled aggregates and their amount within concrete mixtures, are
needed to determine a more comprehensive correlation between the
relevant parameters analysed in this paper and a final compressive
strength model for RACs.
6. Conclusions
In this paper, a novel model is proposed to simulate the develop-
ment of the compressive strength to the progress of the hydration reac-
tion in Recycled Aggregate Concrete (RAC). Particularly, the possible
influence of two key parameters, i.e. the w/c ratio and the moisture con-
dition of recycled concrete aggregates (RCAs), on the hydration reaction
1.0
MIX0.60SAT
0.9
Semi-Adiabatic [TR=22°C]
0.8
Degree of hydtarion-αh
of the paste inside a RAC system and on the compressive strength devel-
opment for four different mixtures with RCAs have led to the following
conclusions:
– the proposed model to relate the compressive strength development
to the evolution of the hydration processes provides researchers
with a deeper insight and a more fundamental understanding of
the development of mechanical properties of the concrete at differ-
ent curing temperatures: the proposed method allows for a direct
interpretation of phenomena which are often studied according to
a merely empirical approach;
– the time monitoring of the temperature developed inside a harden-
ing concrete made it possible to indirectly identify the hydration
process of the four mixtures considered in this study;
– the simulated time evolution of the degree of hydration of the
paste inside a concrete with recycled aggregates and the mea-
sured compressive strength of the RAC system showed a clear
linear correlation;
– these correlations turned out to be strongly affected by both the
w/c ratio and moisture conditions of recycled aggregates;
– the proposed analytical model to describe the correlation be-
tween the degree of hydration of the paste inside a RAC system
and the compressive strength of RAC considers the paste to be
the governing component. The stiffness and potential moisture
interaction of the recycled aggregates with the RAC paste deter-
mines the final strength of RAC-systems.
– finally, the linear correlation emerged between degree of hydra-
tion and compressive strength indicates a novel conceptual de-
sign approach for RAC systems.
Future research will focus on collecting more experimental data to
enrich the knowledge on the proposed hydration approach to predict
the compressive strength of mixtures based on the correlation between
αh(t) and Rcm(t).

Adiabatic Acknowledgements
0.7
0.6 The present study is part of the activities carried out by the Authors
0.5 within the “EnCoRe” Project (FP7-PEOPLE-2011-IRSES n° 295283;
Isothermal [TR=22°C]
0.4 www.encore-fp7.unisa.it) funded by the European Union within the
0.3
Table 3
0.2 Parameters describing the linear correlation αh − Rcm.
0.1
Mix Rcm,max α0
0.0
0 12 24 36 48 60 72 0.45DRY 76.3 0.26
Time-t [h] 0.45SAT 33.3 0.18
0.60DRY 47.9 0.11
0.60SAT 22.9 0.49
Fig. 16. Time evolution of the degree of hydration—0.60SAT.
212 E.A.B. Koenders et al. / Cement and Concrete Research 56 (2014) 203–212
Seventh Framework Programme. It was developed during the visits of
both the first author at the University of Salerno (Italy) and the other
two co-authors at the Universidade Federal do Rio de Janeiro (Brazil),
occurred in the period May-December 2012.
References
[1] S. Moll, S. Bringezu, H. Schütz, Resource Use in European Countries—Material Flows
and Resource Management, Wuppertal Institute for Climate, Environment and, En-
ergy, Wuppertal (DE), 2005.
[2] J. Fiksel, Design for Environment, Creating Eco-efficient Products and Processes,
McGraw-Hill, 1996.
[3] J.A. Moya, N. Pardo, A. Mercier, Energy Efficiency and CO2 Emissions: Prospective
Scenarios for the Cement Industry, JRC Scientific and Technical Report, EUR 24592
EN, 2010.
[4] BRE, Developing a strategic approach to construction waste—20 year strategy,
Proceedings of the CIB Task Group 39—Deconstruction Meeting, Garston, Watford
WD25 9XX (UK), 2006, (available online at: http://www.bre.co.uk/filelibrary/pdf/
rpts/waste/ConstructionWasteReport240906.pdf; accessed on 20 February 2013).
[5] P. Crowther, Design for buildability and the deconstruction consequences, Proceed-
ings of the CIB Task Group 39—Deconstruction Meeting, Karlsruhe, Germany, April 9
2002.
[6] C. Edvardsen, K. Tølløse, Environmentally “Green” Concrete Structures, Proceedings
FIB-symposium “Concrete and environment” in Berlin, October 2001.
[7] European Parliament and Council, Directive 2008/98/EC on waste and repealing cer-
tain Directives, November 19 2008.
[8] C. Lima, A. Caggiano, C. Faella, E. Martinelli, M. Pepe, R. Realfonzo, Mechanical behav-
iour and durability of sustainable concrete made with recycled aggregates and fly
ash, Constr. Build. Mater. 47 (2013) 547–559.
[9] K. van Breugel, Simulation of hydration and formation of structure in hardening
cement-based materials, (Ph.D. Thesis) Delft University of Technology, Delft
(Netherlands), 1991. (295 pp.).
[10] K.W. Anderson, J.S. Uhlmeyer, M. Russell, Use of Recycled Concrete Aggregate in
PCCP: Literature Search, Washington State Department of Transportation, Report
WA-RD 726.1, June 2009.
[11] M. Malešev, V. Radonjanin, S. Marinković, Recycled concrete as aggregate for struc-
tural concrete production, Sustainability 2 (2010) 1204–1225.
[12] V. Corinaldesi, G. Moriconi, Behaviour of cementitious mortars containing different
kinds of recycled aggregate, Constr. Build. Mater. 23 (1) (2009) 289–294.
[13] A. Caggiano, C. Faella, C. Lima, E. Martinelli, M. Mele, A. Pasqualini, R. Realfonzo, M.
Valente, Mechanical behavior of concrete with recycled aggregates, Atti del 2°
Workshop on “The new boundaries of structural concrete”, Università Politecnica
delle Marche—ACI Italy Chapter, Ancona (Italy), Settembre 18–16 2011, pp. 55–62.
[14] E. Martinelli, E.A.B. Koenders, A. Caggiano, A numerical recipe for modelling hydra-
tion and heat flow in hardening concrete, Cem. Concr. Compos. 40 (2013) 48–58.
[15] E.A.B. Koenders, Simulation of volume changes in hardening cement-based materials,
(Ph.D. Thesis) Delft University of Technology, Delft (Netherland), 1997. (171 pp.).
[16] S.J. Lokhorst, Deformation behavior of concrete influenced by hydration related
changes of the microstructure, Internal Report Nr. 5-99-05, Delft University of Tech-
nology, Delft (Netherland), 1999, (178 pp.).
[17] R.D. Toledo Filho, E.A.B. Koenders, M. Pepe, G.C. Cordeiro, E.M.R. Fairbairn, E. Martinelli,
Rio 2016 sustainable construction commitments lead to new developments in recycled
aggregate concrete, Proceedings of the ICE-Civil Engineering 166 (6) (2013) 28–35.
[18] G. de Schutter, L. Tearwe, Degree of hydration-based description of mechanical
properties of early age concrete, Mater. Struct. 29 (6) (1996) 335–344.
[19] EN 197-1, Cement—Composition, specification and conformity criteria—Part I:
Common cements, 2001.
[20] R.H. Mills, Factors Influencing Cessation of Hydration in Water-Cured Cement
Pastes, Special report n.90, Proceedings of the symposium on the Structure of
Portland Cement Paste and Concrete, Highway Research Board, Washington, D.C.,
1966, pp. 406–424.
[21] T.C. Hansen, Physical structure of hardened cement paste. A classical approach,
Mater. Struct. 19 (114) (Nov–Dec 1986) 423–436.
[22] I. Pane, W. Hansen, Investigation of blended cement hydration by isothermal calo-
rimetry and thermal analysis, Cem. Concr. Res. 35 (2005) (2005) 1155–1164.
[23] A.W. Gutsh, Properties of early age concrete—experiments and modelling, Proceed-
ings Int. RILEM Conference on Early Age Cracking in Cementitious Systems, 2003,
pp. 11–18.
[24] M. Laube, Werkshoffmodell jungen Betons, (Ph.D. Thesis) TU Braunschweig, 1990.
[25] M. Krauss, M. Hariri, F. Rostásy, Investigation of fracture processes in young concrete
using electronic speckle pattern interferometry (ESPY), Proceedings Int. RILEM Con-
ference on Early Age Cracking in Cementitious Systems, 2003, pp. 199–208.