Building and Environment 89 (2015) 59e71

Contents lists available at ScienceDirect

Building and Environment

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

Hygrothermal behavior of a massive wall with interior insulation

during wetting
M. Guizzardi a, b, *, D. Derome b, R. Vonbank b, J. Carmeliet a, b
a
b

ssische Technische Hochschule, Zrich, Switzerland

ETH Zrich, Eidgeno
Empa, Swiss Federal Institute for Materials, Science & Technology, Dbendorf, Switzerland

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 10 November 2014
Received in revised form
24 January 2015
Accepted 24 January 2015
Available online 20 February 2015

To improve the energy efciency of historical buildings, the introduction of an interior insulation is often
the only possible solution in order to preserve their valuable external facades. However, this intervention
changes the wall hygrothermal conditions and can negatively affect its hygrothermal performance.
Furthermore, interior insulation is often difcult to apply due to practical problems of irregularities in
geometry and heterogeneity of materials. A newly developed high insulating render is used as interior
insulation on a masonry test wall and is exposed to controlled but severe wetting conditions, while the
hygrothermal parameters in its different components have been recorded. In addition to provide a welldocumented dataset for model validation, this investigation highlights the hygrothermal behavior of a
masonry wall internally insulated with a new developed highly insulating render. The experimental
work described in the present paper is designed in order to record data to be used for the validation of a
numerical model for the parametric study of the hygrothermal behavior of internally insulated masonry
walls.
2015 Elsevier Ltd. All rights reserved.

Keywords:
Interior insulation
Hygrothermal conditions
Masonry wall
Moisture content

1. Introduction
To improve the energy efciency of historical buildings, the
introduction of an interior insulation is often the only possible
solution in order to preserve their valuable external facades. One of
the main issues to be considered in insulating solid walls consists in
nding equilibrium between the reduction of heat loss and the
preservation of the wall integrity through the careful choice of
materials, installation techniques and assembly design. The long
term performance of energy efcient renovation strategies must be
assessed as moisture accumulation and interstitial condensation
are difcult to detect. Moisture related damages can be costly to
solve and can compromise the structural stability as in the case of
some bricks sensitive to freeze-thaw cycles [27,28,45], although
sound test procedures have been developed to assess the freezethaw resistance of brick and brick/mortar systems [17]. Thus, the
choice of the insulation material to be applied is crucial for a safe
long term performance of the retrotted wall.

ssische Technische Hochschule,

* Corresponding author. ETH Zrich, Eidgeno
Zrich, Switzerland. Tel.: 41 76 7373873.
E-mail address: michelaguizzardi@empa.ch (M. Guizzardi).
http://dx.doi.org/10.1016/j.buildenv.2015.01.034
0360-1323/ 2015 Elsevier Ltd. All rights reserved.

Furthermore, in historical buildings, interior insulation is often

difcult to apply due to irregularities in the geometry of the wall,
heterogeneity of materials, different interventions, damages and
repairs during the building life, etc. For example, insulation under
the form of rigid panels requires at surfaces to attach the panels,
while old walls can present surface irregularities. For this reason,
the use of an insulating interior render is often a more suitable
solution. The application of a wet layer has the advantages of
requiring a lower number of construction steps, thus making the
construction method simpler, of providing exibility in the presence of unevenness of the wall, and of allowing gap lling in order
to get a continuous contact between the insulation layer and the
substrate. However the components of the render must be inert
and thus chemically suited to the old materials of the structure. In
addition, as often the exterior nishing cannot be modied, it is
important to avoid the use of vapor tight insulation solutions in
order to allow moisture drying from the wall to the indoor environment if needed. For these specic interventions, an insulating
render has been developed [37].
Moisture is known to be one of the main causes of damage in
building facades and wind-driven rain a main external moisture
source for exterior walls [4,23,24]. Different experimental measurements have been performed with different types of structures

60

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

and loadings in order to evaluate the hygrothermal performance of

building envelopes exposed to wind-driven rain. Pioneer work on
rain penetration [12,13,40] has opened to way for more advanced
hygrothermal evaluation methods of envelope systems exposed to
environmental loading, both in North America [5,6,14] or in Europe
[18,21]. Several eld measurements recorded hygrothermal data
from real buildings, such as Knzel [26], Knzel [26] Abuku et al.
[1], Sass et al. [36] Straube et al. [39]. Other experiments have been
performed using wall samples and large-scale testing facilities that
allow the application of controlled environmental loads. Many of
these studies refer to wood framed walls as in Derome et al. [10],
Maref et al. [29] Maref et al. [30], Teasdale-St-Hilaire et al. [41]
while a more limited number deals with large-scale experiments
of solid walls [2,34,19]. In the work of Teasdale-St-Hilaire et al. [42],
wood-framed walls are studied and the percentage of rain water
inltration through an intentional defect is determined. Piaia et al.
[34] analyzed water penetration through a wall of concrete block.
Other studies have been performed to investigate retrotted masonry walls with an additional internal insulation layer [32,38,46]
and the presence or not of vapor barriers [11,7,8]. Johansson et al.
[19] performed an experimental campaign similar to that exposed
in the present paper, for the analysis of the hygrothermal performance of masonry envelopes with an interior insulation layer of
vacuum insulation panels. In this study, the conditions of wooden
beams embedded in the wall are recorded. Wood members, in fact,
are seen to undergo risky conditions after retrot, as assessed in
Morelli et al. [33], with a risk for corrosion of fasteners due to
changes in hygrothermal conditions [47]. In many cases these
measurements have been used to validate numerical models that
are currently used in building physics calculations and design
[20,35]. Despite these numerous studies, very little has been done
to document rain water as the source of water movement massive
masonry walls. Such studies under realistic conditions would
require very long monitoring efforts. As a result, the behavior of
reinsulated massive masonry walls during wetting conditions as
yet to be assessed.
In this work, the behavior of a massive masonry wall, insulated
on its interior surface with the insulating aerogel render and
exposed to controlled environmental loading, is documented. In
order to maximize the wetting rate and the moisture penetration
through the different layer, freezing temperatures are avoided and
a weathering is used to expose the wall sample to temperatures in
the range between 20  C and 50  C. The large-scale wall assembly
consists in a load-bearing two-width brick masonry without any
cavity or air space, nished with an external lime-based render and
the internal insulating layer, and is tted with two wooden beams.
The monitoring protocol focuses on documenting with a high
number of sensors the moisture transport in the wall, with focus on

liquid moisture using specially developed sensors, in order to give a

complete description of the wetting process in all layers. As for
many, other large-scale experiments, as described in Maref [15],
Geving et al. [31] or Teasdale-St-Hilaire et al. [43] the results of the
experimental campaign that is described in this paper have been
used for the validation of a numerical model for the calculation of
the hygrothermal performance of building facades [16].
2. Experimental set up
2.1. Material description
The materials used for the construction of the wall sample
(Fig. 1) have been selected to represent an average assembly as
found in Swiss residential buildings dating from the 1850e1920
period. The choice of materials for the test wall has been made to
match material properties measured on samples taken from
different historical reference buildings from the Zurich area.
However, for the external layer, an exterior render with a high
capillary absorption coefcient has been selected in order to reduce
the experimental duration. For the same reason, no paint has been
applied at the exterior surface of the wall. The selected render is a
mixture of hydraulic lime, white slaked lime and cement, with
crushed and round limestone or sand grains with a dimension of
0e4 mm (mortar F 662, available on the market and produced by
Fixit). Some water and air-retaining additives are present in the
mixture. As mentioned, the value of the capillary absorption coefcient, Acap, of the render is higher than the average of the historical renders while other hygrothermal properties, that have been
measured in laboratory and are presented in Table 1, are consistent
with those of the material samples from the reference buildings.
For the choice of the clay bricks to build the masonry, full bricks
have been preferred because holes can affect moisture transport.
The full clay brick chosen has properties similar to the ones of site
samples. For the mortar to be used in the masonry joints, as no
samples from reference buildings were available, the IBP database
for historical materials, as available in the software Wu, has been
taken as reference. For mortars, the Acap values lie in the range
between 0.001 kg/m2s0.5 and 0.127 kg/m2s0.5 and the retained
mortar value lies in the middle of this range. To get a good contact
between the render and the masonry, and between the masonry
and the internal insulation, a thin rough layer of mortar has been
applied on the interior and exterior surfaces of the masonry, after
the wall had dried.
On the internal side of the wall, a layer of 6 cm of the high
insulation render, mentioned above, has been applied. The material
is a lightweight mortar (200 kg/m3) that can be applied both as
external and internal render. It consists of hydrophobized granular

Table 1
Table of properties of the materials used for the test wall construction.
External render

Clay brick

Cement mortar

Aerogel render

Wood

Density
Thermal conductivity
m-value
Acap

1668 kg/m3
0.464 W/mK
11 0.077 kg/m2s0.5

1553 kg/m3
0.684 W/mK
14e15 0.115 kg/m2s0.5

1623 kg/m3
0.633 W/mK
17e21 0.023 kg/m2s0.5

200 kg/m3
0.027 W/mK
40.032 kg/m2s0.5

Sorption isotherm

RH
w [kg/m3]
0
0
0.3
4.2
0.5
17.9
0.8
50.5
0.95
59.7
286.8 [kg/m3]

RH
w [kg/m3]
0
0
0.3
1.4
0.5
1.9
0.8
10.5
0.95
17.1
303.8 [kg/m3]

RH
w [kg/m3]
0
0
0.3
2.44
0.5
5.56
0.8
20
0.95
113.2
3
148 [kg/m ]

RH
w [kg/m3]
0
0
0.3
1.1
0.5
2.55
0.8
10
0.95
43.4
3
400 [kg/m ]

455 kg/m3
0.23 W/mK
3.8 (longit.) 0.007 kg/
m2s0.5
(trans.) 0.032 kg/
m2s0.5
RH
w [kg/m3]
0
0
0.3
45
0.5
53
0.8
73
0.95
107
3
534 [kg/m ]

wsat

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

silica aerogel (60e90 vol.%) and lime-based binder. This type of

binder ensures compatibility with construction materials usually
found in historical building walls. The use of the aerogel aggregates
confers high thermal resistance and low density to the material,
making it suitable for thermal insulation applications. The thermal
conductivity of the aerogel render is 25 2 mW/mK, at 23  C and
50% RH, which is lower than other rendering values (65 mW/mK).
This property of the aerogel render is inuenced by the pumping
pressure applied to the wet mixture during the automatic laying
phase. Further additives enhance the workability of the wet
mixture [37]. The material has a capillary absorption coefcient
Acap of 0.06 kg/m2s0.5.
At the internal side of the test wall, two pieces of wood of 50 cm
length with a cross section of 11.5 x 14.5 cm2 have been embedded
in the wall to a depth of 18 cm to simulate the intersection of the
beams of the oor with the external wall in real buildings. Spruce
wood has been used for these components as it is a species of wood
commonly used in historical buildings and its material properties
are well known and tested in numerical simulation studies.
2.2. Geometry and construction of the wall specimen
The assembly is designed to be representative of historical
building envelopes of the Swiss building stock. The overall dimensions of the wall, 180 cm high by 150 cm wide, match the size of
the weathering machine used for generating the climatic loading
on the external surface of the wall. The thickness of the masonry
wall is 26 cm, corresponding to two bricks layers plus the mortar
joint. In plan, the wall has a T shape. This is for stability reasons as
the wall needs to be moved for drying and then placement in the
weathering machine. Full clay bricks, i.e. without holes, of dimensions of 6  12  25 cm3, are used. On the inside of the wall,
two wood beams are inserted in the brick wall, to a depth of one
brick width, positioned at 140 cm from the bottom and in the
middle of each half of the wall.
In building practice, mortar joints are not perfectly uniform in
thickness and may contain voids, leading to an imperfect brick/
mortar contact. Nevertheless, for this wall, all joints are fully lled
with mortar to ensure very good contact at the brick/mortar
interface and thus to minimize the effect of interface resistance to
moisture transport. To ensure an efcient curing of the mortar, as
water uptake by a dry brick reduces the amount of water available
in the mixture for hydration [9], all bricks are wetted in water for
50e60 s before laying (during which they absorb approximately
100 g of water, equivalent to an average moisture content of
approximately 55 kg/m3).
Once the masonry structure of the wall is built, it is dried for two
weeks at laboratory ambient conditions (23  C, 50% RH). Then it is
placed for ve weeks in an oven at 40  C. After an additional two
weeks of conditioning in the laboratory, the thin rough coat of
mortar and the remaining layers are added to the interior and
exterior sides. On the exterior side, a 3 cm layer of lime-cement
render is placed. It is nished with an additional 0.5 cm render of
lime-cement mortar with bigger aggregates to prevent damages of
the surface due to the sprayed water. At the interior side, a 6 cm
layer of aerogel render is laid. The choice of this newly developed,
high performing insulation render is connected to the goal of the
main doctoral project this experimental campaign belongs to,
consisting in the study of the effect of interior insulation of existing
masonry walls. This is a vapor open and hydrophobic material expected to allow drying of eventual moisture entering the wall
during the wetting phases.
The two spruce beams are placed in the test wall during the
masonry construction in symmetric positions. Then, they are
removed to prevent moisture uptake from the masonry during

61

curing and to be equipped with the grids of moisture content pins.

The beams are repositioned in the dry masonry wall, before the
application of the layer of interior insulation aerogel render. Before
adding the insulating render, any space between the wood beams
and their pocket is carefully sealed with the cement mortar in order
to get a good contact with the surrounding masonry.
2.3. Monitoring system
Sensors are inserted in the wall in order to constantly monitor
the hygrothermal conditions at different positions. The overall
schematic representation of the monitoring is shown in Fig. 2. Four
types of sensors have been used: probes for temperature and
relative humidity, electrical conductivity sensors for brick and for
mortar and wood moisture content pins.
Sensors measuring both temperature and relative humidity
(RH) (models SHT 15 and SHT75 from Sensirion) are placed in the
following positions: at the masonry/exterior render interface, at the
masonry/internal insulation interface, in mortar joints between
bricks and at three different thicknesses in the aerogel render layer
(Fig. 3). In addition, a T-RH sensor is placed in the wood beam end
pocket, in front of the beam head. As the sensors can be damaged
by high pH values in the drying phase of the mortar and as it is
impossible to insert them after curing, the sensors are protected
with an envelope of gore-tex fabric that allows reaching hygrothermal equilibrium with the surrounding material, while preventing liquid, specially the alcalin water of wet mortar paste, to
contact the sensor. The T-RH sensors are initially calibrated by the
manufacturer. The calibration is further veried in the climatic
chamber in 16 T-RH conditions: at 20  C for 15, 25, 35, 45, 55, 65, 75,
85 and 95% RH and at 0  C-95%, 5  C-85%, 10  C-75%, 15  C-65%,
25  C-45% and 30  C-35% RH.
Sensors measuring the electrical conductivity changes of the
materials have been used for monitoring liquid presence in the
bricks and the mortar. These sensors apply an electrical voltage
across a layer of material and the electrical conductivity is
measured. For full description of this approach, see Ref. [16]. Six
special bricks have been drilled to be equipped with two series of
ve pairs of sensors, one series is close to the brick edge in contact
with the mortar joint and the other is placed at the center of the
brick, see Fig. 4a. Of the special bricks, four are placed in the
external line of the masonry, two in a lower row 50 cm from the
bottom and two higher at 140 cm from the wall base, and two are
placed in the inner layer of the masonry, 140 cm from the bottom.

Fig. 1. Schematic 3-dimensional views of the T shaped test wall a) from the outside,
and b) from the inside.

62

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

Fig. 4. Photos of the moisture content sensors used to monitor the test wall, a) for
bricks, seen in a sectioned brick for visualization, b) for the mortar, mounted before
application of the render.

connections of each array, see the arrangement of measurement

connections in Fig. 5a. From the recorded electrical conductivity
values, the moisture content in mass % in the monitored positions is
calculated using the relation for spruce given by the pins manufacturer Delmhorst. The correlation values and the function are
given in Eq. (1):

MC% 21:737$R0:122

Fig. 2. 3D schematic view of the positions of all sensors in the different layers of the
test wall: temperature and relative humidity sensors (blue), brick electrical conductivity sensors (red) and electrical conductivity sensors for electrical conductivity in
renders and mortar joints (green). (For interpretation of the references to colour in this
gure legend, the reader is referred to the web version of this article.)

The sensors for mortar measurements, where the electrical contact

pairs are mounted on a support as shown in Fig. 4b, are embedded
in the external render, in the aerogel render internal layer and in
the mortar joints in the masonry, both horizontally and vertically.
For moisture content measurements in the two wood beam
heads, wood moisture probes are applied. Eight pins of 25 mm from
Delmhorst 26-ES/100  1.500 , with teon coating along their length
except for the tips for measurement at the tip positions (Fig. 5), are
arranged in a orthogonal array. Cylindrical holes of 15 mm depth
are drilled, then the pins are pushed into the wood for a length of
20 mm, controlling for the precise depth with the drilling machine
control tool. Once all the moisture pins have been precisely positioned at the right distance and depth, and connected through
electrical wires to the circuit, the holes are lled with electrical
insulating glue resin in order to ensure that the electrical resistance
is measured solely between the tips. Pins are arranged in three
arrays of 8 pins and electrical conductivity is measured along the 10

[1]

2.4. Experimental protocol

A weathering chamber is used to expose the exterior side of the
test wall to controlled hygrothermal loadings (Fig. 6). The facility
allows wetting of the surface through 6 sprinklers, spraying water
on the sample rate at the controlled rate of 2 l/m2$min water. The
wall can be irradiated by 20 Ultra-Vitalux lamps of 300 W that
allow a radiation mix similar to that of natural sunlight of
approximately 600 W/m2 on the sample surface. The temperature
is controlled through a 3 kW heater and a 5 kW cooler in the range
between 20  C and 50  C. A dehumidier removes moisture and
allows keeping the RH in the desired range when the spurts are
turned off. The interior side of the wall is exposed to ambient
laboratory conditions that are kept constant at 23  C and 50% RH.
First the experimental set-up and the functioning of the monitoring system is veried through one continuous and heavy wetting load. The experiment consists in 48 h of wetting. Then, after a
full drying, the test wall is exposed to cyclic excitations that alternate wetting, drying and radiation cycles. Although recent studies
in predicting rain deposition have been made [3,22,24], we apply a
more conservative approach with applying liquid uptake conditions. One cycle consists on one hour of wetting followed by one
hour without wetting nor radiation, and then four hours of radiation. These six-hour cycles have been repeatedly applied for four
months. To start the test, the rst of these cycles was prolonged to
15 h. Before and between the two experiments, the wall sample is
dried rst in oven and then with radiation lamps.
2.5. Verication of the experimental set up

Fig. 3. Positions of the T-RH sensors at different thicknesses in the cross section of the
test wall.

The results of the 48 h of continuous wetting of the wall are

presented to verify and demonstrate the capacity of the monitoring
system. During this shorter wetting, the sensors in the exterior
render and in the external layer of the masonry are solicited.
Data from the sensors in the exterior render show the wetting
process of the external wall layer in Fig. 7. As no paint is applied to
the exterior surface and the render used has a high capillary absorption coefcient (Acap 0.138 kg/m2s0.5), the wetting of the

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

63

Fig. 5. Monitoring of the wood beam, a) schematic representation of the arrangement of the measurement array, b) photos of wood moisture probes.

whole render layer is completed quite fast, within four hours. From
the ten contact pairs situated positioned across the layer, the rst
two outer contacts display an almost immediate reaction right after
the start of spraying of water on the wall surface. These contacts are
in fact 1 and 3 mm deep in the render layer. The wetting process is
markedly faster in the lower sensor G2 positioned at 50 cm from
the bottom of the wall, compared to G1 that is placed in a higher
position, 140 cm from the wall base. In G2, the water front takes
approximately 4 h to reach the deepest measurement position
close to the interface between the render and the masonry, while in
G1 it takes only 1.5 h. In both sensors, the deepest position, i.e. the
dotted line number 10, wets before position 9 in G1 and positions 7,
8 and 9 in G2, while, for all other positions, the arrival of the water
front is consistent with the depth of the contacts. This effect is
probably due to water that bypassed the masonry, e.g. through
eventual cracks in the render and masonry, to reach the interface
between the render and the masonry, where it is taken up from the
interior side of the render.
After passing through the external render, water reaches the
masonry wall. The electrical conductivity measurements in the
exterior brick width indicate the water front proceeding deeper and
reaching the two most exterior sensors of the bricks, as shown in
Fig. 8b. The times of moisture arrival at corresponding positions in
the different bricks are consistent, although the values measured

are different. The moisture front reaches the rst position after
approximately 20 h for all exterior bricks, except in the central line
of brick 2 where it takes more than 50 h. The positions deeper in the
brick are not reached by water during this test.
The sensors in the mortar joints in the exterior layer of the
masonry show a behavior that is consistent with that of the bricks.
The mortar sensors cover the rst three centimeters, corresponding
to the position of the rst two brick contacts, as shown in Fig. 8a. As
seen in Fig. 8b, the reaction in the contacts of sensors in the horizontal joints G3 and G4 closer to the exterior wall surface (green
lines) occurs at 15e20 h from the start of the wetting while in the
sensors in the vertical joints, i.e. G5 and G6, the increase starts after
5e10 h and the increase of electrical conductivity is higher. It thus
seems that the liquid water penetration is faster through the vertical joints than the horizontal ones. This is probably due to the fact
that the mortar in the vertical joints is less compact than in the
horizontal one which is subjected to the mass of above masonry.
This results in differences in the pore structure and in the hygrothermal properties of the material. Also, this loading guarantees
better contact at the brick/horizontal joint mortar interfaces, while
in the vertical joints, micro cracking due to mortar shrinkage is
more likely. This causes the presence of pathways for moisture
transport that speeds up the liquid uptake in the vertical joints. The
concordance of the brick and mortar data gives a consistent

Fig. 6. Laboratory photos of a) weathering chamber without the wall, where the 20 lamps and the 6 sprinklers are visible, (b) the test wall installed in the chamber with connections
of the sensors to the data logger. Note the presence of wood beams protected by plastic sheet until the start of the test.

64

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

Table 2
Table of water front arrival times in sensors B1, B2, B3 and B4.

Side line

Central line

Sensor position and

depth [mm]

Water front arrival times [hours]

B1

B2

B3

B4

1e15 mm
2e35 mm
3e62.5 mm
4e90 mm
5e110 mm
6e15 mm
7e35 mm
8e62.5 mm
9e90 mm
10e110 mm

20
49
e
e
e
20
50
e
e
e

18
44
e
e
e
27
e
e
e
e

21
48
e
e
e
26
44
e
e
e

17
42
e
e
e
20
49
e
e
e

3. Experimental results
During the four months of cyclic environmental loading, the
wall sample is monitored with all the installed sensors. The
experimental protocol consists in a cycle loading repeated for four
months, with a rst longer cycle (17 h). During the experiment, the
moisture front crosses all the layers of the test wall and reactions
are observed in all the sensors.

3.1. Results from sensors in the external render

Fig. 7. Measurement results from moisture content sensor in the external render in the
rst 15 h of wetting (a) in sensor G1 on the right side of the wall and (b) in sensor G2
on the left side of the wall.

description of the wetting process in the external layer of the

masonry.
This short experiment demonstrates that the monitoring system
designed for this investigation can capture precisely the water
uptake process in the wall. We can then proceed to a more complex
and extensive wetting experiment, whose results are presented in
the next section Table 2.

The wetting of the render is monitored by two electrical conductivity sensors and two T-RH sensors (respectively G1 and G2 and
T-RH 1 and 2). In Fig. 9, the electrical conductivity values recorded
by one of the sensors placed in the depth of the render (G1) are
plotted for the rst 10 days of the experiment. Each curve corresponds to one couple of connections of the sensor, the proles are
plotted with different reference lines and values are normalized to
the initial the value recorded before the beginning of the
experiment.
The data recorded by the electrical conductivity sensor show
that the full render layer is fully wetted by the end of the rst cycle,
as shown by the electrical conductivity curves that show an increase corresponding to the initial wetting time of each cycle. The
positions 1, 2 and 3, located on the exterior side, present smoother
proles compared to other contacts, while the cycling loading is
more visible in the remaining of the render layer. Nevertheless,
after 8 or 9 days, all positions display a smooth reading, indicating
that the liquid content is uniform over the render and over the
cycles.

Fig. 8. a) Schematic representation of the positions of the sensors in the mortar joint compared to the connections of the brick sensor. b) Graph of water front arrival times detected
by the mortar sensors in the joints (green lines) and the brick sensors (red dots) versus position in the layer. (For interpretation of the references to colour in this gure legend, the
reader is referred to the web version of this article.)

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

Fig. 9. Measurement results from moisture content sensor in the exterior render
(sensors G1) in the rst 10 days of cyclic wetting. The 10 lines correspond at the
different measurement positions in the thickness of the render.

Temperature and relative humidity data recorded by the sensor

at the interface between the render and the masonry are presented
in Fig. 10. These data also show that the effect of the wetting front is
detected at the interface already during the rst cycle. After the
initial at line, the RH curve increases over the rst cycle. Once
wetted, the render/masonry interface remains wet, as seen with the
results of the electrical conductivity sensors. Thus across the
render, the effects of the drying period, i.e. the radiation loading,
are limited to changes of temperature.
3.2. Results of the sensors in the masonry
The masonry is monitored with electrical conductivity sensors
in the bricks and the mortar joints, and by temperature and RH
probes. The electrical conductivity sensors in the two layers of the
masonry give an interesting insight on the wetting process. Such
proles from one brick placed in the external layer of the masonry

65

are plotted in Fig. 11. From the shape of the proles, two stages are
discerned: the initial moisture arrival reaction (highlighted with
red circles), corresponding to the rst sensible increase in the
values measured by the sensor, and the secondary moisture arrival
time, that corresponds to the steepest conductivity increase when
the wet steady state is reached (highlighted with blue circles). In
these proles, the effect of the cyclic loading on the electrical
conductivity is visible and uctuations remain constant after the
wet steady state is reached. The initial and secondary moisture
front arrival times are plotted in Fig. 12 for all four bricks placed in
the external layer. The external layer of the masonry wets almost
linearly in function of time from the outer positions towards the
inner ones.
The moisture front arrival times are determined also for the
bricks in the internal layer of the masonry. In this case, the two
steps observed in the external layer are not clearly discernable as
the curves are smoother and the process does not reach a steady
state during the experimental time. Thus, in this case, only one
arrival time is dened for each curve and the determined values are
plotted in Fig. 13. Sensor positions show a quasi-simultaneous reaction time, independent of contact position with almost simultaneous arrival time on both sides of the bricks, i.e. positions 1 and 5,
in contact with mortar. The electrical conductivity increases a bit
later in the contacts at the center of the brick, i.e. positions 2, 3 and
4.
After the brick, we consider the liquid transport in the mortar.
Data from the four electrical conductivity sensors placed in the
mortar joints between bricks close to the external surface (G3-6)
show curves similar to those of the bricks placed at the same depth
in the masonry. An example of the proles is given in Fig. 14. The
values plotted are normalized with respect to the average value in
the nal wet state. For this dataset also, the initial and secondary
arrival times are identied, corresponding to the red line and blue
circles in Fig. 14. The difference between the two reaction times
increases with the depth of the measurement position. The secondary arrival times, corresponding to the actual liquid front, are
plotted for all the four sensors in Fig. 15. The positions closer to the
outside surface (position 1) wet in a few days from the beginning of
the experiment, while all sensors positions react within 15 days.
The data recorded by the temperature and relative humidity
sensors placed in the mortar joint between the two layers of bricks
and in front of the wood beam are plotted in Figs. 16 and 17. In these

Fig. 10. Temperature and relative humidity data from one of sensor (T-RH 1) at the interface between the external render and the masonry.

66

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

Fig. 13. Arrival time of the moisture front, determined by electrical conductivity
measurements, across two bricks in the interior layer of the masonry (B5 and B6).
Fig. 11. Electrical conductivity versus time measured across brick B1, located in the
exterior layer of the masonry.

Fig. 12. aeb. Graph of arrival times (in days) of the moisture front for the different
depths of four external layer bricks (B1 to B4) as detected by electrical conductivity
measurement, a) initial arrival, b) secondary arrival.

positions, the effect of cyclic loading is visible. As shown in Fig. 16,

the relative humidity increases at each cycle and 100%RH is reached
after 18, 20 and 38 days respectively for the three different sensors
placed in the center of the masonry wall. The two sensors in the air
space in front of the wood beams (Fig. 17) show larger uctuations
in terms of RH and reach 100% RH after 38 and 40 days respectively.
In Fig. 18, the values recorded by the T-RH sensors at the interface between the masonry and the interior insulation are plotted.
These proles exhibit small increases in the average RH from
approximately 20 to 30 days that corresponds to the arrival times
recorded by the contacts in the bricks closer to the interior surface.
This moisture is due a water leakage that is observed from the
lower part of the wall. Most probably the formation of cracks between the wall and its wooden base during the experiment causes
water leakages out of the weathering chamber. We observed that
the water coming from the inner chamber of the machine moved
through the lower joint of the wall and was taken up by the interior
insulation layer to a height of 30e40 cm, corresponding to the
position of the brick and T-RH sensors (B5-6 and T-RH 9e10).

Fig. 14. Measured electrical conductivity at 10 measurement positions in the mortar

joint between the brick, at location G3, during the rst 50 days of the experiment.

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

Fig. 15. Secondary arrival times detected by electrical conductivity for the different
measurement positions in the mortar of the outer layer of the masonry, from G1 to G4.

3.3. Results from sensors in the wood beam

The wooden elements embedded in the masonry structure
simulate the connection of the exterior envelope with oor beams.
The data recorded by the three grids of moisture pins in the
wooden beam are used to generate the iso-contour moisture content maps at 35, 58, 85 and 110 days from the beginning of the
experiment, interpolating the MC values of the measurements
positions to cover the area on the beam equipped with the sensors,
see Fig. 19. Results show that the moisture is absorbed by the wood
from the head surface, which is exposed towards the outer brick of
the masonry. The effect of moisture absorbed from the side edges of
the beam is also detectable.
The moisture content measured by different probes at the same
depth is averaged as shown in Fig.20a and plotted in Fig. 20b. After
a certain delay into the experiment, the moisture content is seen to
increase rst rapidly and then more slowly. For the sake of
describing the moisture movement in the beam, the results of all
moisture probes are analyzed by two parameters as graphically
dened in Fig. 20c, namely two curve slopes are retained, the

67

primary and secondary moisture content increasing rates. For the

four grids closer to the wood beam head, the primary rate corresponds to the average slope of the curve between the 40th and the
50th day from the initial wetting. The secondary rate corresponds
to the average slope of the curve calculated between the 58th and
the 84th day from the beginning of the experiment.
In Fig. 21, the primary and the secondary increasing rates obtained from the rst two nets of probes of each beam are compared.
The diagram bars are arranged according to the distance of the
corresponding position from the head surface, as shown in Fig. 20a.
Considering the rst grids W1 and W2, we can observe that positions 2, 5 and 8, that are closer to the surface (15 mm), show a
higher MC increasing rate than the ones in positions 3, 6 and 9, at
40 mm from the surface. Positions 1, 4, 7 and 10 at 27.5 mm present
intermediate values. This distribution is clearer for the right beam
especially considering the secondary rate. From the analysis, we
can see that the left beam reacts faster than the right one because it
presents higher initial increasing rates in all positions. Grid W3 also
reacts faster than W4 and these two react a little bit later compared
to W1 and W2, after approximately 40 days.
The moisture content increasing rates of the measurement positions at the same depth are averaged and plotted in function of
the distance from the head surface in Fig. 22. The rate is higher
closer to the head of the beam, decreases quickly in the rst 4 cm
and further decreased in deepest positions.
4. Analysis of results
The experiment of this test wall exposed to controlled environmental conditions gives information about the wetting process
and the moisture movements through the different layers of the
wall. Liquid water travels fast in the most external layer, i.e. the
render, which is monitored with electrical conductivity sensors and
by two T-RH sensors at the interface between the render and the
masonry. The high rate of wetting in the most external layers is
observable both during the short and long wetting experiments.
Graphs in Fig. 7 show that all 10 measurement points in the render
depth react to water front within a few hours. It is expected that the
presence of a layer of paint would signicantly reduce the absorption rate of the sprayed water in the exterior nishing.
The water front needs longer time to penetrate the masonry
wall as the interface between the rst two layers, the render and

Fig. 16. Temperature and relative humidity versus time, measured in the mortar joints between the external and internal bricks of the masonry by T-RH sensors 0, 3 and 4.

68

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

Fig. 17. Temperature and relative humidity versus time, measured in the air space between the external masonry and the timber beam, by T-RH sensors 15 and 16.

the masonry itself, introduces a hydraulic resistance to the transport of liquid water, which is the main moisture source for the
exterior building envelopes and its transport is driven by capillary
forces. Considering the electrical conductivity measurements in the
depth of the external layer of brick the water front takes 1.5 to 3.5
days to reach the rst sensor position (1.5 cm from the brick edge).
Then the liquid front travels through the brick depth almost linearly
with time and the wetting times recorded by the different sensor
positions are consistent with their distance from the external surface (Fig. 11). The water front takes 21e30 days to cross the exterior
brick.
Considering the values recorded by the electrical conductivity
sensors G3-G6, inserted in the mortar joints among the external
layer of bricks and covering the 3 cm of material close to the
interface with the exterior render, we can observe that reaction
times to the water front arrival are consistent with those measured
by the brick sensors at the corresponding depths, although the
transport process in the brick is faster than in the mortar, as shown
in Fig. 8. After 20 h of wetting almost all the sensors positioned at a
depth of 15 mm inside the brick already register the passage of the
water front, while at the measurement positions at the corresponding depth in the mortar joints the water front comes with a
delay from 5 to 10 h. The same observation can be made for the last
measurement positions in the mortar sensor, that record the water
arrival at the same time as the second connection of the brick

sensor, that is 12 mm deeper. The conclusion is consistent with the

porous structure, and consequently with the capillarity properties
of the two materials: clay bricks have a more homogeneous and
ner pore structure, that results in a higher capillary suction than
the mortar. Furthermore we can observe the effect of the different
pore structure between the vertical and horizontal mortar joints:
the transport process is faster in the horizontal joints (corresponding to the position of sensors G5 and G6) than in the vertical
ones (sensors G3 and G4) as can be observed in Fig. 8. The mortar in
the vertical joint is expected to be more compact, with minor voids
among the aggregates and the cement binder, due to the vertical
load acting on it during the curing phase and to the pressure
manually applied on the layer during the construction process.
Deeper in the masonry, the moisture front travels through the
internal brick layer between the 30th and the 45th days of the
experiment and the wetting process is not linear as for the bricks in
the external layer. The difference between the wetting times for the
external and internal layers of bricks indicates that the intermediate mortar joint behaves as a second capillary barrier that slows
down the penetration of liquid water though the different layers.
T-RH measurements in the intermediate mortar joint show
moisture content increases that are consistent with data recorded
by the other sensors: a value of 100% RH is reached approximately
after 20 days from the beginning of the rst wetting cycle (Fig. 12),
while the sensors in the space in front of the wooden beam heads,

Fig. 18. aeb. Temperature and RH proles versus time measured at the interface between the masonry and the interior aerogel render: sensors T-RH 9 in (a) and T-RH10 in (b).

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

69

Fig. 19. Isocontours of moisture content measured by the moisture content pins at four times during the experiment, a) 35, b) 58, c) 85 and d) 110 days.

which are placed at the same depths but in contact with different
materials, reach the same value after 35 and 40 days (Fig. 13). This
time discrepancy is probably due to a small air space between the
masonry and the wood beam surface, that was detected during the
demolition of the setup. Such air space would locally slow down the
wetting process compared to the movements in the other part of
the wall where continuity among the different layers is provided.
On the inside of the test wall we do no detect signicant
moisture content increases. Sensors reactions to hygroscopic condition changes are smoother as the effect of vapor and liquid phase

is more difcult to distinguish. Electrical conductivity sensors

placed in the rst 3 cm of the interior insulating render do not
provide interesting information. On the other hand, RH increases
are recorded at the interface between the masonry and the insulation layer (Fig. 15). From the proles, it is not possible to identify a
clear increasing step but the average values increase after 40e60
days of measurements and the recorded RH values stabilize after
100 days around 30% at the end of the experiment. Finally, the
wood beams inserted in the masonry for a depth of 19e20 cm and
equipped with moisture content pins, show a reaction time to

Fig. 20. a) Schematic representation of the measurement position in the wood beams, b) Averaged moisture content versus time for the left beam, c) Schematic representation of
the two rates retained for analysis.

70

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

Figs. 21. aeb. Comparison of moisture increasing rates calculated from measurement results of grids W1-W3 (a) and W2-W4 (b). For each measurement position, the primary (left
column of each pair) and secondary (right column) moisture increasing rates are compared. Positions are in the graphs are arranged according to their distance from the head
surface (see Fig. 19b).

moisture content that is similar for all positions in a range of 35e40

days after the experiment start.
5. Outlook and further perspectives
The dataset of the wetting of a massive masonry wall provides a
complete picture of the transport of moisture, mostly in liquid state,

through a masonry assembly. Such dataset is readily useable for

modeling validation. Based on these ndings and on the hygrothermal characterization of all the material used in the construction
of the wall sample, a numerical model has been developed and
validated. The model has been then used for a comparative analysis
between different interior insulation solutions and for a parametric
investigation on the inuence of the external render properties on
the global wetting and drying process of the building faade.
Furthermore the temperature and relative humidity values recorded during the experimental campaign described in the present
paper are applied as input materials in the model for rot damage
developed by Viitanen et al. [44]. A description of the material
characterization, of the numerical model and the results of the
comparative analysis is given in Guizzardi [16].
As a note, a further experiment on the drying behavior of the
wall would have been interesting, although not possible in the time
frame of this project.
6. Conclusions

Fig. 22. Plot of the moisture increasing rate in function of the distance from the head
surface of the left beam.

The experiment on this masonry wall specimen exposed to

controlled but heavy weathering conditions gives insightful information on the wall wetting behavior. The monitoring system gives
reliable results and data from the different types of sensors placed
in the different materials and in the different layers allow a precise
description of moisture movement, especially of liquid water front
penetration. The effect of interfaces between material layers is
nicely observed at several planes through the assembly. The
monitoring system and the newly developed electrical conductivity
sensors performed well in capturing the global moisture behavior
and may provide an interesting opportunity for the evaluation of
real building applications, where the effects due to energy retrot
interventions could be monitored for long time.

M. Guizzardi et al. / Building and Environment 89 (2015) 59e71

Acknowledgements
The support of the Competence Center Energy and Mobility
under the SuRHiB (Sustainable Renovation of Historical Buildings)
project is acknowledged.
References
[1] Abuku M, Blocken B, Roels S. Moisture response of building facades to winddriven rain: comparison of eld measurements and numerical simulations.
J Wind Eng Ind Aerodynam 2009;97(5e6):197e207.
[2] Bitsuamlak GT, Arindam Gan Chowdhury AG, Sambare D. Application of a fullscale testing facility for assessing wind-driven-rain intrusion. Build Environ
2009;44:2430e41.
[3] Blocken B, Defraeye T, Derome D, Carmeliet J. High-resolution CFD simulations
of forced convective heat transfer coefcients at the facade of a low-rise
building. Build Environ 2009;44:2396e412.
[4] Blocken B, Derome D, Carmeliet J. Rainwater runoff from building facades: a
review. Build Environ 2013;60:339e61.
[5] Burch DM, Thomas WC, Mathena LR. Transient moisture and heat transfer in
multilayer non-isothermal walls e comparison of predicted and measured
results. In: Proceedings of the ASHRAE/DOE/BTECC/CIBSE Conference, thermal
performance of the exterior envelopes of building IV, Orlando, Florida; 1989.
p. 513e31.
[6] Burch DM, Thomas WC. An analysis of moisture accumulation in a woodframe wall subjected to winter climate. In: Proceedings of the ASHRAE/DOE/
BTECC conference. Thermal performance of the exterior envelopes of building
V, Clearwater Beach, Florida; 1992. p. 467e79.
[7] Carmeliet J, Derome D. Temperature driven inward vapor diffusion under
constant and cyclic loading in small-scale wall assemblies: part 1 experimental investigation. Build Environ 2012a;48:48e56.
[8] Carmeliet J, Derome D. Temperature driven inward vapor diffusion under
constant and cyclic loading in small-scale wall assemblies: part 2 heatmoisture transport simulations. Build Environ 2012b;47:161e9.
[9] Derluyn H, Janssen H, Carmeliet J. Inuence of the nature of interfaces on the
capillary transport in layered materials. Constr Build Mater 2011;25(9):
3685e93.
[10] Derome D, Saneinejad S, Carmeliet J, Karagiozis A. Comparison of small- and
large-scale wall assembly specimens exposed to similar experimental conditions. In: Proceeding of nordic building physics symposium, Copenhagen, June
2008; 2008. p. 1015e22.
[11] Derome D, Sananeijad S. Inward vapor diffusion due to high temperature
gradients in experimentally tested large-scale wall assemblies. Build Environ
2010;45:2790e7.
[12] Doshi H. Rain water on building cladding: two case studies. In: Proc. 10th Canadian Conf. Building science and techn., Ottawa, May 12e13, 1; 2005. p. 116e26.
[13] Duff JE. Moisture distribution in wood-frame walls in winter. For Prod J For
Prod Res Soc 1968;18(1):60e4.
[14] Dumont R. Wood frame wall moisture accumulation study. Saskatchewan
Research Council; 1993 [SRC Publ. No. 1-4800-25-C-93, 31].
[15] Geving S, Karagiozis A. Field measurements and computer simulations of
hygrothermal performance of wood frame walls. 1997. In: GEving S, editor.
Moisture design of building construction, hygrothermal analysis using simulation models, part II: collection of papers and reports; 1996 [Ph.D. thesis, June
1997, NTNU Trondheim, Norway].
[16] Guizzardi M. Hygrothermal performance assessment of novel interior insulation solutions [Ph.D. thesis]. ETH Zrich: Chair of Building Physics; 2014. J
Building Physics.
[17] Hendry EAW. Masonry walls: materials and construction. Constr Build Mater
2001;15:323e30.
[18] Janssens A, Hens H, Silberstein A, Boulant J. The inuence of underroof
systems on the hygrothermal behavior of sloped insulated roofs. In: Proceedings of the ASHRAE/DOE/BTECC conference, thermal performance of
the exterior envelopes of building V. Clearwater Beach, Florida: ASHRAE;
1992. p. 368e78.
[19] Johansson P, Time B, Geving S, Jelle BP, Kalagasidis AS, Hagentoft CE, et al.
Interior insulation retrot of a brick wall using vacuum insulation panels:
design of a laboratory study to determine the hygrothermal effect on existing
structure and wooden beam ends. In: Thermal performance of the exterior
envelopes of whole buildings XII international conference, Clearwater Beach,
FL, USA; 2013 [December 5e9].
[20] Kehl D, Ruisinger U, Plagge R, Grunewald J. Wooden beam ends in masonry
with interior insulation e a literature review and simulation on causes and
assessment. In: 2nd central European symposium of building physics, Vienna,
September 9e11; 2013.
[21] Kohonen R. Thermal effects of airows and moisture on exterior wall structures. In: Proceedings of the ASHRAE/DOE/BTECC conference, thermal performance of the exterior envelopes of buildings III, American society of

[22]

[23]

[24]

[26]
[27]
[28]
[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]
[39]

[40]

[41]

[42]

[43]

[44]

[45]

[46]

[47]

71

heating, refrigerating and air-conditioning Engineers, Clearwater Beach,

Florida; 1985. p. 583e605.
Kubilay A, Blocken B, Derome D, Carmeliet J. CFD simulation and validation of
wind-driven rain on a building facade with an Eulerian multiphase model.
Build Environ 2013;6:69e81.
Kubilay A, Derome, Blocken B, Carmeliet J. High-resolution led measurements of wind-driven rain on an array of low-rise cubic building. Build Environ 2014a;78:1e13.
Kubilay A, Derome D, Blocken B, Carmeliet J. Numerical simulations of winddriven rain on an array of low-rise cubic buildings and validation by eld
measurements. Build Environ 2014b;8:283e95.
Knzel H. Effect of interior and exterior insulation on the hygrothermal
behavior of exposed walls. Mater Struct 1998;31(2):99e103.
Laycock EA. Ten years of frost testing at Shefeld Hallam University. Constr
Build Mater 2002;16:195e205.
riaux et
Maage M. Frost resistance and pore size distribution in bricksMate
Construction 1984;17(5):345e50.
Maref W, Lacasse MA, Kourglicof N. A precision weighing system for helping
assess the hygrothermal response of full-scale wall assemblies. In: Performance of exterior envelopes of whole building VIII: Integration of building
envelopes (Clearwater Beach, FL,12/2/2001); 2001. p. 1e7 [December 12
(NRCC-45202)].
Maref W, Booth DG, Lacasse MA, Nicholls M. Drying experiment of woodframe wall assemblies performed in the climatic chamber EEEF, specication of equipment used in EEEF-environmental exposure envelope facility. IRC
Research Report. 2002. p. 42 (RR-105).
Maref W, Lacasse MA, Booth D. Large-scale laboratory measurements and
benchmarking of an advanced hygrothermal model. In: Building for the
future: the 16th CIB World building congress, Rotterdam, Netherlands; 2004.
Mensinga P, Straube J, Schumacher C. Assessing the freeze-thawn resistance
of clay brick for interior insulation retrot projects. In: Proceedings of thermal
performance of the exterior envelopes of whole buildings XI, Clearwater
Beach, FL, USA, December 5e9; 2010.
Morelli M, Schefer GA, Nielsen TR, Svendsen S. Internal insulation of masonry
walls with wooden oor beams in northern humid climate. In: Proceedings of
the thermal performance of the exterior envelopes of whole buildings XI international conference, Clearwater Beach, FL, USA, December 5e9; 2010.
Piaia JCZ, Cheriaf M, Rocha JC, Mustelier NL. Measurements of water penetration and leakage in masonry wall: experimental results and numerical
simulation. Build Environ 2013;61:18e26.
Ruisinger U. Long-term measurements and simulations of ve internal insulation systems and their impact on wooden beam heads. In: 2nd Central
European symposium of building physics, Vienna, September 9e11; 2013.
Sass O, Viles HA. Wetting and drying of masonry walls: 2D-resistivity monitoring of driving rain experiments on historic stonework in Oxford, UK. J Appl
Geophys 2010;70(1):72e83.
Stahl T, Brunner S, Zimmermann M, Ghazi Wakili K. Thermo-hygric propreties
of a newly developed aerogel based insulation rendering for both exterior and
interior applications. Energy Build 2012;44:114e7.
Straube JF, Schumacher CJ. Interior insulation retrot of load-bearing masonry
walls in cold climates. J Green Build 2007;2(2):42e50.
Straube J, Ueno K, Schumacher C. Internal insulation of masonry Walls: nal
measure guideline, building america report e 1105. In: Svendsen S, editor.
Driving rain: experimental research on the resistance of external walls against
rain penetration. 1955. Oslo: Norwegian Building Research Institute; 2011
[Report 20, 19].
Svendsen SD. Driving rain: experimental research on the resistance of
external walls against rain penetration. Rapport 20. Oslo: Norges byggforskningsinstitutt; 1955.
Teasdale-St-Hilaire A, Derome D. State-of-the-art review of simulated rain
inltration and environmental loading for large-scale building envelope
testing. ASHRAE Trans 2005;111(2):389e401.
Teasdale-St-Hilaire A, Derome D. Methodology and application of the simulated wind-driven rain inltration in building envelope experimental testing.
ASHRAE Trans 2006;112(2):656e70.
Teasdale-St-Hilaire A, Derome D. Comparison of experimental and numerical
results of wood-frame wall assemblies wetted by simulated wind-driven rain
inltration. Energy Build 2007;39:1131e9.
Viitanen H, Toratti T, Makkonen L, Peuhkuri R, Ojanen T, Ruokolainen L, et al.
Towards modeling of decay risk of wooden materials. Eur J Wood Prod
2010;68:303e13.
Whiteside D, Newman AJ, Kloss PB, Willis W. Full-scale testing of the resistance of water penetration of seven cavity lls. Build Environ 1980;15:
109e18.
Wilkinson J, DeRose D, Straube J, Sullivan B. Measuring the impact of interior
insulation on masonry walls in a cold climate. In: Proceedings of Canadian
building science and technology conference, Montreal; 2009. p. 97e110.
Zelinka S, Glass S, Derome D. The effect of moisture content on the corrosion
of fasteners embedded in wood subjected to alkaline copper quaternary
treatment. Corros Sci 2014;83:67e74.