390
International RILEM Conference on Materials, Systems and Structures in Civil Engineering
Conference segment on Service Life of Cement-Based Materials and Structures
22-24 August 2016, Technical University of Denmark, Lyngby, Denmark
of air at the air-concrete interface whose moisture content depends on the distance to the
surface of concrete. This is important at the initiation of drying, because the high moisture
content initially observed at the concrete surface gives rise to a boundary layer of air with
high humidity at the air-concrete interface, which initially slows the drying kinetics. The
thickness of this boundary layer strongly depends on the ambient conditions (air flow,
temperature) and the material surface condition (surface roughness and moisture) [8]. Such
boundary conditions can be more accurately represented by equation 13, in which J
BC
represents the flow of water vapor at the air-concrete external interface, hr
s
is the relative
humidity at the interface and hr
e
is the external air relative humidity. The value of h
BC
is not
well defined and can reach values from 1. 10
-2
[8] to 8. 10
-7
[9] in natural conditions.
(13)
4. Results and discussion
The ongoing tests related to the evolution of the humidity gradient measured by relative
humidity sensors will not be presented in this section, which focuses on the mass loss. The
results presented in Figure 3a indicate that when increasing the drying radius, the mass loss as
a function of time is decreased. The curves representing the mass loss as a function of the
square root of time (Figure 3b) present three successive stages. First, during the first hours
after the drying starts, the mass loss increases with an accelerating rate. This stage, which is
particularly marked in G4, lasts until approximately one day. Then, during the second stage,
the mass loss increases as a linear function of the square root of time (which corresponds to
the exact analytical simulation for semi-infinite media). This stage lasts longer as the drying
radius increases. After more than one month, G1, which has the larger drying radius, still
exhibits a linear evolution of the mass loss as a function of the square root of time. On the
other hand, the second stage ends approximately after 3 days. Finally during the third stage,
the mass loss increases with a decreasing rate. This stage occurs until the equilibrium between
the concrete sample and the ambient air is reached, at which point no mass variations will be
observed. These stages are described more thoroughly in following sections.
391
International RILEM Conference on Materials, Systems and Structures in Civil Engineering
Conference segment on Service Life of Cement-Based Materials and Structures
22-24 August 2016, Technical University of Denmark, Lyngby, Denmark
Figure 3: Various representations of the mass loss as a function of time and drying radius
4.1. Stage 1: initial drying and boundary layer
During the first hours and up to one day after the drying starts, the mass loss can be expressed
as a non-linear function of the square root of time. This observation can be explained by the
presence of a boundary layer of air at the concrete-air interface. Initially, all water extracted
from the sample is present in the form of water vapour close to the concrete surface. This
tends to increase the relative humidity of the ambient air, and therefore decrease the drying
kinetics. Progressively, the water contained in the porosity in direct contact with the concrete
surface evaporates. Due to the high diffusivity coefficient of water vapour in air in
comparison with the concrete(equivalent) permeability and diffusivity, the relative humidity
of the boundary layer decreases, until eventually reaching a value close to the actual ambient
conditions.
In order to calibrate the value of parameter h
CL
, preliminary tests are performed by measuring
the relative humidity of the air against the drying surface through time. These measurements
are shown in Figure 4a. Right after drying is initiated, the relative humidity at the surface is
close to 80%. Then, progressively, it decreases until reaching a value of 55% at 5 days (120h).
Initially, in order to correspond to these measurements, the best fit value of h
BC
is situated
392
International RILEM Conference on Materials, Systems and Structures in Civil Engineering
Conference segment on Service Life of Cement-Based Materials and Structures
22-24 August 2016, Technical University of Denmark, Lyngby, Denmark
between 2.10
-6
and 10
-5
. However, progressively, the experimental dots tend to reach values
closer to h
BC
of 5.10
-5
. One reason for this is that the exchange coefficient might change in
time. This experiment indicates that as the saturation degree close to the surface decreases, the
rate of exchange of water from the surface to the ambient air also decreases. Figure 4b shows
the effect of changing h
BC
on the mass loss during the first hours after drying. In the following
simulations, a value h
BC
=10
-5
is chosen.
Figure 4: Effect of the exchange coefficient h
BC
on the surface relative humidity, and
comparison with experimental values (markers)
The results illustrated in Figure 5a show that the simulation strategy developed in section 3
allows representing the initial mass loss kinetics. The experiments are shown by dots, while
the numerical simulations are represented by dashed lines. In these simulations, all parameters
are kept constant, except for the geometry of the sample. The values of the parameters are all
extracted from [1], except for the intrinsic permeability, which is fixed at 1.15x10
-21
. For G4,
the value of the parameter
h
BC
is different than for other geometries. This is due to the specific
drying surfaces of G4, one of which is the “top” side of the sample which was not in contact
with the mould during setting and hardening.
Figure 5: Initial mass loss during stage 1 (dots = experiments, dashed lines = simulations)
The surface roughness and porosity is therefore significantly different from all other surfaces,
justifying a change in the exchange coefficient h
BC
. In the same way, Figure 5b indicates that