Cement and Concrete Research 84 (2016) 1–7

Contents lists available at ScienceDirect

Cement and Concrete Research

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

Autogenous healing of sea-water exposed mortar: Quantification

through a simple and rapid permeability test
D. Palin ⁎, H.M. Jonkers, V. Wiktor
Faculty of Civil Engineering & Geosciences, Section of Materials and Environment, Delft University of Technology, Stevinweg 1, 2628 CN Delft, The Netherlands

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

Article history: Concrete has an autogenous ability to heal cracks potentially contributing to its functional water tightness and du-
Received 27 April 2015 rability. Here, we quantify the crack-healing capacity of sea-water submerged mortar specimens through a simple
Accepted 17 February 2016 and rapid permeability test. Defined crack width geometries were created in blast furnace slag cement specimens
Available online xxxx
allowing healed specimens to be quantified against unhealed specimens. Specimens with 0.2 mm wide cracks
were not permeable after 28 days submersion. Specimens with 0.4 mm cracks had decreases in permeability of
Keywords:
Microcracking [B]
66% after 28 days submersion, and 50–53% after 56 days submersion. Precipitation of aragonite and brucite in
Granulated blast-furnace slag [D] the cracks was the main cause of crack healing. Healing potential was dependent on the initial crack width,
Autogenous healing thermodynamic considerations and the amount of ions available in the crack. To our knowledge, this is the first
Permeability [C] study to quantify the functional autogenous healing capacity of cracked sea-water exposed cementitious
specimens.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction
Cracks are an unavoidable feature of reinforced concrete. Cracks
generally allow greater water permeability reducing the functionality
of, for example, water-retaining structures. If this water contains
aggressive agents, such as those found in sea-water, deterioration can
further ensue until a concrete structures' possible ruin. Concrete has
an autogenous ability to heal cracks potentially reducing its water per-
meability and extending the durability of the structures from which it
is made. In this way, accurate quantification of concrete permeability
may provide insight for better service life predictions, particularly
those in the marine environment.
Studies are available in the literature quantifying the autogenous
healing capacity of cracked cementitious materials through their
water permeability [1–3]. These studies, limited to fresh-water, are no
doubt built on research quantifying the water permeability of cracked
cementitious materials [4–7]. Quantification of crack healing through
water permeability measurements presupposes the generation of
defined (or at least explicit) crack width geometries. Wang et al., in
their permeability study [4], introduced a feedback-controlled tensile
splitting technique for generating cracks in concrete specimens. This
study and subsequent studies employing the same technique [6,7] did
not directly measure the crack width; rather they assumed them to be
equal to the lateral displacement of the specimen measured using linear
variable differential transformers (LVDTs). This may result in inaccurate
measurements due to crack branching and variability along their length.
⁎ Corresponding author.
E-mail address: d.palin@tudelft.nl (D. Palin).
Cracks generated through this technique have also been shown to relax
following unloading. Wang et al. [4] noticed that the crack openings
after unloading partially closed reducing their widths. Aldea et al. re-
corded crack reductions of 62–66% for high-strength concrete and 56%
for normal strength concrete [6] and 32–75% depending on the material
type used in a subsequent study [7]. In their study, Reinhardt and Jooss
[3] included reinforcement in their specimens and were able to form
tensile cracks within 0.01 mm of the desired crack width. Though accu-
rate, they anticipated that the crack widths, which were measured ex-
ternally, were not constant over the height of the specimens
demonstrating the difficulty of this method to create defined reproduc-
ible crack geometries. Existing permeability tests also tend to wait for a
steady state flow to be achieved, that is, for the crack inflow to equal the
crack outflow. This means that the testing periods can last ~14 [3], 20–
50 [4], 35–140 [2] and 90–100 [5] days. A long testing period may result
in healing and crack geometry changes during the test, while the
amount of water required to reach a steady state flow may not be prac-
tical for many tests. Studies quantifying autogenous healing through
permeability measurements must know the permeability of an un-
healed crack against which a healed crack can be quantified. Reinhardt
and Jooss [3], in their study, established this initial permeability by
flowing water through the cracks at the beginning of the test. This
action, however, has the danger of flushing out reactive components re-
sponsible for healing. Edvardsen [2], in her study, was able to negate this
flushing out effect by calculating the initial crack flow through
Poiseulle's law. Calculations, however, only provide an approximation
of the initial permeability. Scope is then available for a setup able to gen-
erate reference permeability values for unhealed cracks against which
healed cracks of the same geometry could be compared and quantified.

http://dx.doi.org/10.1016/j.cemconres.2016.02.011
0008-8846/© 2016 Elsevier Ltd. All rights reserved.
2 D. Palin et al. / Cement and Concrete Research 84 (2016) 1–7
The aim of the current study is to quantify the autogenous healing
capacity of cracked sea-water submerged specimens through their
water permeability. A simple and rapid permeability setup, and a
novel method for creating defined crack width geometries in cementi-
tious specimens are described. Cracked specimens were produced by
which reference unhealed permeability values could be defined and
subsequent healed specimens compared. To our knowledge, this study
is the first to quantify the functional healing capacity of cracked sea-
water exposed cementitious specimens.
2. Experimental program
A permeability test was designed and constructed to quantify the
autogenous healing capacity of cracked blast furnace slag (BFS) cement
mortar specimens submerged in sea-water. Three series were setup:
(1) to establish the reference permeability values for unhealed 0.2 and
0.4 mm wide cracks; (2) to establish permeability values for specimens
with 0.2 and 0.4 mm wide cracks after 28 days submersion; and (3) to
establish permeability values for specimens with 0.2 and 0.4 mm wide
cracks after 56 days submersion in sea-water (Fig. 1). Crack widths
and submersion times are based on the work of Palin et al. [8].
2.1. Sample preparation
Specimens were prepared following the work of Palin et al. [9]. Cyl-
inders (60 mm long and with a diameter of 33.5 mm) were cast from
BFS cement (CEM III/B 42.5N LH, ENCI) in accordance with EN 1015-
11 [10]. The applied mortar mix design is shown in Table 1. Mortar
was chosen as a representative material for concrete. The mortar cylin-
ders were designed to fit standard polyvinylchloride pipes (internal di-
ameter 35 mm) and to have two diametrically opposite grooves (2 mm
wide and 3 mm deep) running down their side. In order to cast the cyl-
inders, a master pattern was first created in polyvinylchloride. The mas-
ter pattern was used to cast silicone moulds, which were in turn used to
cast the mortar specimens. 24 h after casting, the mortar specimens
were carefully removed from their silicone moulds, tightly sealed in
polyethylene plastic bags and kept at room temperature for a total cur-
ing period of 28 days.
2.2. Crack calibration
Following curing, specimens were prepared for tensile cracking.
Each mortar cylinder was wrapped in polyethylene and steel rods
placed at their grooves before being placed between the compression
platens of an Instron 8872 servohydraulic testing machine (Instron
Corp., Canton, MA, USA). A compressive load was applied at
0.01 m.s−1 until the cylinders diametrically split from groove to groove
(Fig. 2A). The polyethylene wrap served to hold the two halves of the
specimen together (ideally so debris could not lodge in the crack
Series
0 28
Crack,
Cast Curing permeability
1(0.2)
1(0.4)
Crack,
Cast Curing submerge Healing Permeability Drying Permeability
2(0.2)
2(0.4)
Crack,
Cast Curing submerge
3(0.2)
3(0.4)
Fig. 1. Experimental scheme to quantify the autogenous healing capacity of cracked BFS cement specimens submerged in sea-water.
Table 1
Mix-design for mortar specimens.
Amount [kg/m3]
Constituent CEM III/B 42.5N LH
Cement 494
Water 247
Water cement ratio 0.5
Sand fraction [mm]
1–2 608
0.5–1 426
0.25–0.5 167
0.125–0.25 319
making it wider than the intended crack width). The specimens were
then carefully unwrapped and Perspex spacers 2.2 and 2.4 mm wide
placed between their grooves to achieve defined crack widths of 0.2
and 0.4 mm. A two-part adhesive, Plex 7742 and liquid Pleximon 801
(Evonik Röhm GmbH, Darmstadt, Germany), was mixed and applied
on either side of the spacers. Once the glue was dry, the spacers were re-
moved and the remainder of the grooves (where the spaces had been)
were glued. Specimens once cracked were silicone-glued into sections
of pipe (60 mm long), the silicone was allowed to cure for 24 h provid-
ing a watertight permeability cell. A pin was fitted at the base of the cell
to prevent the specimen shooting out from the bottom during testing.
2.3. Crack evaluation
Cracks at the specimen ends were imaged with a stereomicroscope
(Leica MZ6, Nussloch, Germany). Crack boundaries were defined by
eye, their area measured with image-analysis software (Fiji, Is just
ImageJ, http://fi ji.sc/) and the average width calculated. Images of se-
lected cracks are shown in Fig. 2.
2.4. Water permeability
Fig. 1 shows the experimental scheme devised to quantify the autog-
enous healing capacity of the cracked specimens submerged in sea-
water. The synthetic sea-water used in this study was produced from
technical grade chemicals (Sigma-Aldrich), the composition of which
is shown in Table 2. Each series consisted of two crack widths ten spec-
imens each. Specimens from series 1 were first assessed for their perme-
ability. Fig. 2B shows a schematic of the permeability setup. Five
permeability columns were set up to run in parallel. Each permeability
cell was attached via a fitting at the bottom of each permeability col-
umn. Sea-water was prepared and poured into reservoirs at the top of
each column. Taps in each reservoir were released initiating the perme-
ability test. Water flowing through the cracks was individually collected
in catchment buckets and the weight of the water recorded at 5, 10 and
30 min. The water level of each column was manually maintained
Day
56 84 112
Healing Permeability Drying Permeability
 D. Palin et al. / Cement and Concrete Research 84 (2016) 1–7
A B
33.5
Steel rod
Groove
Mortar cylinder
Crack
Polyethylene
wrap
Fig. 2. Schematic illustration of: (A) the splitting tension setup used to diametrically fracture the cylinders; and (B) the permeability setup, which has been published previously [9].
between 1 and 1.05 m, giving an almost constant water head of 0.1 bar.
Permeability cells from series 2 and 3 were submerged and vacuumed
for 2 h to remove any air bubbles trapped in their cracks, which could
hinder water flow and healing. These cells were then transferred to
plastic buckets containing 4 l of sea-water (20 ± 2 °C). The cells of series
2 were submerged for 28 days and the cells of series 3 for 56 days, in
sea-water. The buckets were sealed with lids to reduce evaporation.
After submersion, the specimens housed in their cells were assessed
for their permeability. Following the permeability test, cells from series
2 and 3 were dried in a drier at 36 °C for 28 days and again tested for
their permeability.
2.5. Precipitate characterization
Mortar specimens having the mean permeability of each series were
selected for chemical analysis. These specimens were dried, precipitates
scraped from their cracks (b5 mg) and the precipitates analysed
through Fourier-transformed infrared spectroscopy (FTIR) (Spectrum
100, Perkin-Elmer Inc., United States of America). Spectra were the re-
sult of 32 scans with a resolution of 2 cm− 1 in the range of 4000–
600 cm−1. Cross-sections of the specimens were also made, impregnat-
ed with epoxy resin and polished for both environmental scanning elec-
tron microscope (ESEM) analysis (Philips XL 30 ESEM, Eindhoven,
Netherlands) in back-scattered (BSE) mode and elemental mapping
analysis through energy-dispersive spectroscopy (EDS). In between
preparation and analysis, specimens were kept in a desiccator to avoid
continued cement hydration. EDS image maps of calcium, magnesium
and silica were merged through the apply image tool in Photoshop
(Adobe Systems, San Jose, California, USA) (Fig. 5(D, H, L and P)).
Table 2
Synthetic seawater composition based after the major constitu-
ents of seawater [11].
Compound Amount [g/l]
NaHCO3 0.19
CaCl2.2H2O 1.47
MgCl2.6H2O 10.57
Na2SO4.10H2O 9.02
KCl 0.75
NaCl 24.08
3
Resevoir
Sea-water
1.05 m
Frame
Permeability
cell
Cylinder
specimen
Catchment
3. Results
3.1. Stereomicroscope
Stereomicroscope analysis of cracks intended to be 0.2 mm wide
were on average 0.04 mm less than intended and those intended to be
0.4 mm wide were average 0.05 mm less than intended. Fig. 3 shows
randomly selected specimens from series 1 intended to have 0.2 mm
cracks, having cracks: (A) 0.16 mm, (B) 0.17 mm and (C) 0.16 mm;
and crack widths from series 1 intended to have 0.4 mm cracks, having
cracks: (D) 0.41 mm, (E) 0.38 mm and (F) 0.37 mm wide.
3.2. Water permeability
Fig. 4 shows a box plot graph depicting: the initial permeability of 0.2
and 0.4 mm cracks; the permeability of 0.2 and 0.4 mm cracks after 28
and 56 days submersion in seawater; and the permeability of the sub-
merged specimens after drying. The mean initial permeability for the
0.2 mm cracks after 5, 10 and 30 min flow was 0.6 cm3.s−1 with normal
distributions. Following 28 days submersion, the mean permeability of
the 0.2 mm cracks after 5, 10 and 30 min flow was 0 cm3.s− 1. The
same specimens after drying maintained a mean permeability of
0 cm3.s−1. Specimens featuring 0.2 mm cracks submerged for 56 days
also had a mean permeability of 0 cm3.s− 1 before and after drying.
Specimens featuring 0.4 mm cracks had initial mean permeabilities of
3.5 after 5, and 3 cm3.s−1 after 10 and 30 min flow. Following 28 days
submersion in seawater, the mean permeability after 5 min flow was
2.4 cm3.s−1, a reduction of 30%, with a wide distribution in the data, a
mean permeability of 1 cm3.s−1after 10 and 30 min flow, both having
a reduction of 66% and normal distributions. Following drying, the
specimens submerged for 28 days were less permeable, with mean
permeabilities of 0.6 after 5, 0.5 after 10 and 0.4 cm3.s−1 after 30 min
flow. Drying also resulted in box plots with smaller boxes. Specimens
following submersion for 56 days had mean permeabilities of
1.6 cm3 s− 1 after 5 min flow, a reduction of 54%, again with a wide dis-
tribution in the data, 1.5 cm3 s−1 after 10 min flow, a reduction of 50%
and 1.4 cm3 s−1 after 30 min flow, a reduction of 53%, with a normal dis-
tribution. Drying of specimens submerged for 56 days as with the dry-
ing of specimens submerged for 28 days had reduced permeabilities,
having mean permeabilities of 1.1 after 5, 0.9 after 10 and 0.5 cm3.s−1
after 30 min flow. The boxes this time were larger representing a
wider variance in the data.
4 D. Palin et al. / Cement and Concrete Research 84 (2016) 1–7
A B
0.2 mm
1 mm
D E
0.4 mm
1 mm
Fig. 3. Cracks of randomly selected specimens from series 1 intended to have 0.2 mm wide (A, B and C) and 0.4 mm wide (D, E and F) cracks.
3.3. Precipitate characterization
Fig. 5 shows both ESEM and EDS images of a selected specimen from
each series and crack size. Specimens were selected based on their hav-
ing the mean permeability of the ten tested making up each series crack
size. Fig. 5 (A, B and C) shows ESEM images of a specimen with a 0.2 mm
crack after submersion for 28 days and subsequent drying. The image is
plugged at one end with a dark grey (grey scale binary image) precipi-
tate overlaid by a white precipitate (Fig. 5A). Dark grey precipitates can
be seen in relatively smaller amounts towards the centre of the crack
(Fig. 5B). Both white and dark grey precipitates can be seen at the
crack mouth at the other end of the crack (Fig. 5C). This white precipi-
tate decreases in quantity from the crack mouth towards the centre of
the specimen. Fig. 5D shows an EDS elemental map corresponding to
Fig. 5A. This map shows these precipitates to be composed of calcium
and magnesium. Magnesium can also be seen to have intruded into
the sample. Fig. 5 (E, F and G) shows ESEM images of one of the speci-
mens with a 0.2 mm wide crack after submersion for 56 days and sub-
sequent drying. Precipitate formations after 56 days submersion are
much the same as those after 28 days submersion. ESEM images of the
specimens with 0.4 mm cracks submerged for 28 days (Fig. 5 (I, J and
K)) and 56 days (Fig.5 (M, N and O)) were not plugged, though both
can be seen to have reduced their crack mouth openings to ~ 150 μm
(I and M). Dark grey precipitates could be seen along the crack walls to-
wards the centre of both cracks (Fig. 5 (J and N)). Not bridging the crack,
both dark grey and white precipitates can be seen at the other end of the
crack (Fig. 5 (K and O)). Precipitate formations after 56 days submer-
sion, like the 0.2 mm wide cracks, are much the same as those after
28 days submersion. EDS mapping (Fig. 5 (L and P)) again shows
these precipitates to be calcium and magnesium based. It is worth
pointing out that, unhealed crack openings directly sat on the bottom
of the bucket while the plugged crack opening faced upward when sub-
merged in the sea-water. FTIR analysis of precipitates scraped from the
cracks revealed bands at 3642 and 3696 cm−1 correspond to the O—H
stretching vibration of calcium hydroxide and brucite, respectively
[12]. Calcite has characteristic absorption peaks: C—O asymmetric
stretching vibration (v3), C—O out of plane bending (v2), and C—O pla-
nar bending vibration (v4) centred at 1400, 872 and 712 cm−1, respec-
tively. While aragonite has in addition a characteristic peak stretching
vibration (v1) 1083 cm−1 and a bending vibration (v4) centred at 700
and 712 cm−1 [13].
C
F
4. Discussion
We have investigated the autogenous healing ability of cracked BFS
cement mortar specimens submerged in sea-water. To do this, a
simple and rapid permeability test was implemented. Specimens
with 0.2-mm-wide cracks were not permeable, and specimens with
0.4-mm-wide cracks had reduced permeabilities, following 28 and
56 days submersion in sea-water. Specimens also displayed reductions
in permeability between 5 and 10 min, which can be attributed to
water sorption [14,15] and resultant reductions in the crack flow vol-
umes. Further, dry specimens had lower permeabilities than wet speci-
mens; these lower permeabilities are likely due to the formation of
precipitates in the cracks as a result of drying. Specimens submerged
for 28 and 56 days showed little difference in their permeabilities
(based on the overlapping boxes of their box plots (Fig. 4)), meaning
that the majority of healing took place in the first 28 days in line with
the results of Palin et al. [8].
BFS cement is a blend of BFS and ordinary Portland cement. The
slag's latent hydraulic properties are activated by calcium hydroxide,
which is released from the cement when mixed with water. Sea-water
mainly contains magnesium chloride, magnesium sulphate and carbon-
ate ions. Cracking of marine concrete allows sea-water (pH ≈ 7.5) to
enter the cracks where its comes into contact with exposed cement
paste (pH ≈ 13). Calcium hydroxide of the cement paste being soluble
in sea-water (Ksp = 4.1 × 10−6 [15]) is leached into the crack where
the hydroxide ions react with magnesium ions in the sea-water
(Eqs. (1), (2) and (3)):
MgSO4 þ CaðOHÞ2 → MgðOHÞ2 þ CaSO4 $ 2H2 O ð1Þ
Brucite Gypsum
" #
MgSO4 þ CaðOHÞ2 þ CaO $ Al2 O3 $ CaSO4 $ 18H2 O →
Calcium monosulfoaluminate hydrate ð2Þ
MgðOHÞ2 þ 3CaO $ 3CaO $ Al2 O3 $ 3CaSO4 $ 32H2 O
Brucite Ettringite
MgCl2 þ CaðOHÞ2 → MgðOHÞ2 þ CaCl2 ð3Þ
Brucite Calcium chloride
Brucite not being soluble in the crack water (Ksp = 1.2 × 10−11 [16])
is quickly formed. At the same time, alkalis from the cement paste can
raise the pH (N8) of the crack water converting bicarbonates to
 D. Palin et al. / Cement and Concrete Research 84 (2016) 1–7 5

0.2 mm 0.4 mm
5 5
A B

Permeability after 5 min [cm3.s-1]

4 4

3 3

2 2

1 1

0 0

5 5
C D
Permeability after 10 min [cm3.s-1]

4 4

3 3

2 2

1 1

0 0

5 5
E F
Permeability after 30 min [cm3.s-1]

4 4

3 3

2 2

1 1

0 0

1(0.2) 2(0.2) 2(0.2) dry 3(0.2) 3(0.2) dry 1(0.4) 2(0.4) 2(0.4) dry 3(0.4) 3(0.4) dry

Fig. 4. Permeability data of series 1, 2 and 3 in the form of 6 box plot graphs. Graphs (A, C and E) show the permeability data for the 0.2 mm cracks after flowing for: (A) 5, (C) 10 and
(E) 30 min. Graphs (B, D and F) show the permeability data for the 0.4 mm cracks after: (B) 5, (D) 10 and (F) 30 min. The first box plot to the left of each graph shows the initial
permeability, the second to its right is the permeability after submersion for 28 days, the third is the permeability of the 28-day submerged specimens after drying, fourth is the
permeability of the specimens after 56 days submersion and the final box is the permeability of the 56-day submerged specimens after drying. Each box represents the permeability of
10 separate specimens. The top, middle and bottom line of the boxes correspond to the 75-, 50- and 25-percentile value (x.75, x.5 and x.25), respectively. The whiskers show the
minimum and maximum values.

carbonates. These carbonates are then able to react with calcium from
the seawater and mortar matrix forming calcium carbonate (Eq. (3)):
CO2 þ CaðOHÞ2 → CaCO3 þH2 O
Calcium carbonate
Aragonite (Ksp ≈ 6.65 × 10−7 [17]) is precipitated in favour of calcite
(the more thermodynamically stable of the two calcium carbonate poly-
morphs) due to magnesium ions binding to nascent calcite inhibiting
growth [18].
Carbonate and magnesium ions are quickly depleted in the crack
generating a concentration gradient between the crack and the bulk
water. This gradient causes the calcium and hydroxide ions from the
matrix to migrate along the crack until they meet an opposing front of
magnesium and carbonate ions from the bulk water. This front is con-
centrated at the crack mouth where most of the precipitate is formed
(Fig. 6). To a considerable degree, autogenous healing is influenced by
amounts of precipitate being formed between the specimens analysed
(Fig. 5), with 0.2 mm cracks having the precipitate concentrated at the
crack mouths and 0.4 mm cracks having the precipitate at the mouth
ð4Þ but also along the crack length. EDS images clearly show the healing
materials to be made up of calcium (aragonite) and magnesium
(brucite) and not silica (unhydrated cement particles) based. The lack
of silica indicates the formation of the healing materials to be autoge-
nous and not as a result of continued hydration of unreacted cement
particles. In accordance with Eqs. (1), (2) and (3), brucite formation
continues until calcium hydroxide is sufficiently depleted. Following
this, magnesium sulphate is then able to decalcify the calcium silicate
hydrate (the binding material of the cement) transforming it into
magnesium silicate hydrate, a non-cementitious material (Eq. (5)) [19]:
xCaO $ SiO2 $ kH2 O þ xMgSO4 þ mH2 O→
yMgO $ SiO2 $ nH2 O þ ðx−yÞMgðOHÞ2 þ x ðCaSO4 $ 2H2 OÞ ð5Þ
Gypsum

the crack width and thermodynamic considerations, but also by the

concentration of ions available in the crack. The concrete matrix sup- where k +m = n+ 3x − y
plies calcium and hydroxide ions and so the greater the crack surface, Conversion of calcium silicate hydrate to magnesium silicate hydrate
the more ions are available for healing. Indeed, we see relatively similar is known to result in weakening of the cement matrix [19,20]. We can
6 D. Palin et al. / Cement and Concrete Research 84 (2016) 1–7

A B C

0.2 mm 28 days
D

E F G
0.2 mm 56 days

I J K
0.4 mm 28 days

M N O
0.4 mm 56 days

Fig. 5. ESEM and EDS analysis of polished sections prepared after the permeability test. (A–C) are ESEM images of a specimen with a 0.2 mm wide crack submerged for 28 days in seawater:
(A) Shows the crack mouth which was facing upwards when submerged; (B) an area towards the centre of the crack; and (C) the crack mouth at the other end. (D) Is the corresponding
EDS analysis for image (A). (E–G) are ESEM images of a specimen with a 0.2 mm wide crack submerged for 56 days in seawater: (E) shows the crack mouth which was facing upwards;
(F) an area towards the centre; and (G) the other crack mouth. (H) Is the corresponding EDS analysis for image (E). (I–K) are ESEM images of a specimen with a 0.4 mm wide crack
submerged for 28 days in sea-water: (I) shows the crack mouth which was facing upwards; (J) an area towards the centre; and (K) the other crack mouth. (L) Is the corresponding
EDS analysis for image (I). (M–O) are ESEM images of a specimen with a 0.4 mm wide crack submerged for 56 days in sea-water: (M) shows the crack mouth facing upwards; (N) an
area towards the centre; and (O) the other crack mouth. (P) Is the corresponding EDS analysis for image (M). Yellow of the EDS images representing calcium, blue representing
magnesium and violet represennting silicate.

see possible evidence of this conversion in Fig. 5, with the incursion of
magnesium into the cement paste. This incursion is visibly more
pronounced in the 0.4 mm cracks, which is likely due to increased ion
concentrations of magnesium afforded by the larger cracks, resulting
in more leaching and replacement of calcium by magnesium ions
(Eq. (5)).
Calibration of concrete cracks with well-anticipated positions and
predictable widths is a difficult venture. Attempts have been made to
develop setups which fulfill these requirements but variability in results
between replicates has been high [4–7]. Stereomicroscope analysis of
cracks intended to be 0.2 mm wide showed them to be on average
0.04 mm less than intended, and those intended to be 0.4 mm wide
showed them to be on average 0.05 mm less than intended. Cracks
were, however, delineated by eye and so include inherent subjectivity.
Further, spacers used to define the crack widths were also machined
by hand contributing to crack width variances. Having spacers with
greater accuracies should allow for better-defined crack width geome-
tries and hence better comparison of crack permeabilities within and
 D. Palin et al. / Cement and Concrete Research 84 (2016) 1–7 7

A CO32-
Ca2+ B CaCO3

Mg
2+
Ca2+ OH- CaCO3
OH- OH- Ca 2+
Mg
2+
CO32-
OH- CO32-
CO32- CaCO3
OH - Mg(OH)2
Mg2+ Ca2+
Mg2+ CaCO3 Mg(OH)2 Mg(OH)2
Mg2+ Mg(OH)2 Mg(OH)2
Mg2+ CaCO3
CO32- Mg2+ Mg(OH)2
Mg 2+
Mg 2+ CaCO3
Mg2+
CO32- Crack CaCO3 Crack
Ca 2+
OH - Ca 2+

CO3 2-
Ca 2+ CaCO3
Ca2+
Mg2+ OH- OH- Mg2+
Ca2+ OH- CaCO3
CO32- Ca2+
OH-
Ca2+
Sea-water Mortar Sea-water CaCO3 Mortar

Fig. 6. Crack schematic with a simplified precipitation model showing: (A) the main ions in the paste, crack and bulk water; and (B) the major chemical reactions, precipitates and main
locations of the precipitates in the crack.

between tests. There were noticeable differences in flow results after 5 [2] C. Edvardsen, Water permeability and autogenous healing of cracks in concrete, ACI
Mater. J. Am. Concr. Inst. 96 (1999) 448–454.
and 10 min, with results after 5 min having considerable scatter. Little [3] H.-W. Reinhardt, M. Jooss, Permeability and self-healing of cracked concrete as a
difference was seen between crack flows after 10 and 30 min indicating function of temperature and crack width, Cem. Concr. Res. 33 (2003) 981–985.
the reaching of a steady state flow. This reaching of a steady state flow is [4] K. Wang, D.C. Jansen, S.P. Shah, A.F. Karr, Permeability study of cracked concrete,
Cem. Concr. Res. 27 (1997) 381–393.
far earlier than the many days reported in the literature [2–5]. [5] C.-M. Aldea, S.P. Shah, A. Karr, Effect of cracking on water and chloride permeability
of concrete, J. Mater. Civ. Eng. 11 (1999) 181–187.
5. Conclusion [6] C.-M. Aldea, S. Shah, A. Karr, Permeability of cracked concrete, Mater. Struct. 32
(1999) 370–376.
[7] J. Rapoport, C.-M. Aldea, S.P. Shah, B. Ankenman, A. Karr, Permeability of cracked
The functional autogenous healing capacity of marine-exposed ce- steel fiber-reinforced concrete, J. Mater. Civ. Eng. 14 (2002) 355–358.
mentitious specimens has been quantified through water permeability [8] D. Palin, V. Wiktor, H.M. Jonkers, Autogenous healing of marine exposed concrete:
Characterization and quantification through visual crack closure, Cem. Concr. Res.
measurements. A method has been described for generating defined
73 (2015) 17–24.
crack width geometries in mortar specimens. Multiple replicates were [9] D. Palin, Y. Ersan, V. Wiktor, N. De Belie, H.M. Jonkers, A rapid and Repeatable
then produced and tested giving statistical confidence from where a Method for Establishing the Water Permeability of Cracked Mortar Specimens, FIB
reference permeability for unhealed specimens could be defined and Symposium 2015, Concrete Innovation and Design, 2015.
[10] B. EN, 1015-11, Methods of Test for Mortar for Masonry‚ Part II: Determination of
subsequent unflushed-healed specimens could be compared. Cracks Flexural and Compressive Strength of Hardened Mortar, European Committee for
measuring 0.2 mm were fully healed after 28 days submersion in sea- Standardization, Brussels, 1999.
water, whereas 0.4 mm cracks healed by 66% after 28 days submersion [11] W. Stumm, J. Morgan, Aquatic chemistry, chemical equilibria and rates in natural
waters, Environ. Sci. Technol. (1996).
and by 50 and 56% after 56 days submersion. Precipitation of aragonite [12] T.L. Hughes, C.M. Methven, T.G. Jones, S.E. Pelham, P. Fletcher, C. Hall, Determining
and brucite in the cracks were principally responsible for healing. Au- cement composition by Fourier transform infrared spectroscopy, Adv. Cem. Based
togenous healing depends on the crack width, thermodynamic consid- Mater. 2 (1995) 91–104.
[13] F.A. Andersen, L. Brecevic, Infrared spectra of amorphous and crystalline calcium
erations and the amount of ions available in the crack. The general carbonate, Acta Chem. Scand. 45 (1991) 1018–1024.
setup shows great potential as a tool for quantifying the healing capacity [14] ASTM International, Significance of test and properties of concrete and concrete ag-
of autonomous self-healing concrete. gregates, ASTM, 1956.
[15] K. White, Bridge Maintenance Inspection and Evaluation, Taylor & Francis, 1992.
[16] L. Irving, The precipitation of calcium and magnesium from sea water, J. Mar. Biol.
Acknowledgements Assoc. U. K. (New Ser.) 14 (1926) 441–446.
[17] J.W. Morse, A. Mucci, F.J. Millero, The solubility of calcite and aragonite in seawater
of 35% salinity at 25 °C and atmospheric pressure, Geochim. Cosmochim. Acta 44
The authors wish to thank Arjan Thijssen for his help when
(1980) 85–94.
conducting the ESEM and EDS analysis and Gerrit Nagtegaal during [18] R. Berner, The role of magnesium in the crystal growth of calcite and aragonite from
the design and construction of the permeability setup. Research leading sea water, Geochim. Cosmochim. Acta 39 (1975) 489–504.
to these results was funded through the European Union Seventh [19] D. Bonen, Composition and appearance of magnesium silicate hydrate and its
relation to deterioration of cement-based materials, J. Am. Ceram. Soc. 75 (1992)
Framework Programme (FP7/2007-2013) under grant agreement 2904–2906.
no. 290308 — SheMat. [20] P.K. Mehta, Concrete in the Marine Environment, Taylor & Francis, 2002.

References
[1] C. Clear, The Effects of Autogenous Healing upon the Leakage of Water through
Cracks in Concrete, 1985.