See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/319697836

Bio-based Self-healing Mortar: An Experimental and Numerical Study

Article  in  Journal of Advanced Concrete Technology · September 2017

DOI: 10.3151/jact.15.536

CITATIONS READS

12 1,980

4 authors, including:

Eirini Tziviloglou Zichao Pan

Delft University of Technology Tongji University
15 PUBLICATIONS   287 CITATIONS    28 PUBLICATIONS   209 CITATIONS   

SEE PROFILE SEE PROFILE

Erik Schlangen
Delft University of Technology
375 PUBLICATIONS   9,964 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

FONDECYT Initiation Project. Asphalt mixture with self-healing properties via microwave heating View project

Self Healing Asphalt for Road Pavements (SHARP) View project

All content following this page was uploaded by Zichao Pan on 26 September 2017.

The user has requested enhancement of the downloaded file.

Bio-based self-healing mortar: An experimental and numerical
study
Eirini Tziviloglou , Zichao Pan , Henk M Jonkers , Erik Schlangen
Journal of Advanced Concrete Technology, volume 15 ( 2017 ), pp. 536-543

Experimental Study on Self-Healing Capability of FRCC Using Different Types of Synthetic Fibers
Tomoya Nishiwaki , Marina Koda , Hirozo Mihashi , Takatsune Kikuta
Journal of Advanced Concrete Technology, volume 10 ( 2012 ), pp. 195-206

Development of Engineered Self-Healing and Self-Repairing Concrete−State-of-the-Art Report

Hirozo Mihashi , Tomoya Nishiwaki
Journal of Advanced Concrete Technology, volume 10 ( 2012 ), pp. 170-184

Robust Self-Healing Concrete for Sustainable Infrastructure

Victor C. Li , Emily Herbert
Journal of Advanced Concrete Technology, volume 10 ( 2012 ), pp. 207-218

Effect of Self-Healing on the Different Transport Properties of Cementitious Composites

Ahmed Alyousif, Mohamed Lachemi , Gurkan Yildirim , MustafD ŞahmaraQ
Journal of Advanced Concrete Technology, volume 13 ( 2015 ), pp. 112-123

Early-age self-healing of cementitious materials containing ground granulated blast-furnace slag

Kelly Olivier , Aveline Darquennes , Farid Benboudjema, Richard Gagné
Journal of Advanced Concrete Technology, volume 14 ( 2016 ), pp. 717-727
Journal of Advanced Concrete Technology Vol. 15, 536-543, September 2017 / Copyright © 2017 Japan Concrete Institute 536

Scientific paper

Bio-based Self-healing Mortar: An Experimental and Numerical Study

Eirini Tziviloglou1*, Zichao Pan2, Henk M. Jonkers3 and Erik Schlangen4

Received 16 February 2017, accepted 29 August 2017 doi:10.3151/jact.15.536

Abstract
Self-healing concrete can repair itself by closing micro-cracks and thus protect itself from ingress of deleterious gasses
and liquids that can affect its durability. Many self-healing concepts have been developed in the recent years which tar-
get on the recovery of water tightness after cracking. Among those systems, the bio-based healing agents have shown
promising results regarding the crack sealing performance. This paper studies the crack sealing efficiency of bio-based
healing mortar with expanded clay particles. The investigation of sealing performance is conducted through experimen-
tal and computational approaches. Image processing and crack permeability test results are compared with results ob-
tained by computer simulations. The study reveals that the experimental approaches might overestimate the crack clo-
sure percentage, while the computer simulation mostly underestimates the crack sealing. Finally, recommendations are
given to improve the results obtained by both methodologies.

1. Introduction of healing agents (Schlangen et al. 2009; de Rooij et al.

2013).
Cracking in concrete can highly influence its durability Among those systems, the bio-based healing agents
and affect the life span of a structure (Schlangen et al. are of great scientific interest, since they have two ad-
2009). Design codes indicate maximum allowable crack vantageous points over other healing agents. First, the
widths based on empirical studies (Pacheco et al. 2014). bio-based healing agents usually contain less synthetic
To minimize the risk of reinforcement corrosion, Euro- compounds. In addition, they have another significant
code 2 recommends that the maximum crack width for characteristic, namely the prolonged lifetime and the
mild environments should be 0.4 mm and 0.2 mm for possibility of repeated use. The biological healing
more aggressive exposure classes. Design calculations agents do not “expire” and can be re-activated in the
indicate that the larger cover depths will result in en- future upon further cracking. Consequently, the whole
hanced durability. Yet, self-healing concrete is an alter- concept is oriented to increase sustainability and reduce
native solution to succeed increased durability without costs.
the need to over-dimension and thus to increase the total The research of self-healing in cementitious materials
cost of a structure. often focuses on laboratory and experimental work
In self-healing concrete, the intrinsic ability of ce- (Zemskov et al. 2011). Experimental studies are very
mentitious materials to seal micro-cracks (up to 0.2 useful to test and evaluate each self-healing concept and
mm) due to prolonged hydration, carbonation of matrix its functionality. There are various ways to examine the
etc. (Hearn 1998; Edvardsen 1999; ter Heide 2005) is degree of self-healing such as: visual observation
improved by adding the healing agent. The healing (Wiktor et al. 2011; Van Tittelboom et al. 2012; Palin et
agent upon cracking is released and fills the crack vol- al. 2015; Roig-Flores et al. 2015), crack absorption tests
ume. By this means the inner part of the structure is (Wang et al. 2012;, Feiteira et al. 2016), crack perme-
protected and thus the danger of reinforcement corro- ability tests (Edvardsen 1999; Reinhardt et al. 2003;
sion is minimized. Many self-healing concepts have Jonkers 2011; Sangadji 2015; Tziviloglou et al. 2016),
been developed in the recent years using various types acoustic emission tests (Van Tittelboom et al. 2012;
Feiteira et al. 2016; Malm et al. 2016) etc. However,
there are few publications that include modelling studies
1
PhD Candidate, Faculty of Civil Engineering and regarding self-healing in cementitious materials (Joseph
Geosciences, Delft University of Technology, Delft, The 2008; Huang et al. 2012; Hilloulin et al. 2016) and even
Netherlands. *Corresponding author, fewer related to bio-based self-healing concrete
E-mail: [email protected] (Zemskov et al. 2011).
2
Assistant Professor, Department of Bridge Engineering, The current paper focuses on the study of the bio-
Tongji University, Shanghai, China. based self-healing mortar, where the healing agent is
3
Associate Professor, Faculty of Civil Engineering and embedded in lightweight aggregates (LWA). Upon crack
Geosciences, Delft University of Technology, Delft, The formation, the weak lightweight capsules break; the
Netherlands. healing agent activates and fills the open crack by pre-
4
Professor, Faculty of Civil Engineering and cipitating calcium carbonate (CaCO3). The aim of this
Geosciences, Delft University of Technology, Delft, The study is to investigate the crack closure on mortar
Netherlands.
E. Tziviloglou, Z. Pan, H. M. Jonkers and E. Schlangen / Journal of Advanced Concrete Technology Vol. 15, 536-543, 2017 537

specimens after the healing treatment via visual inspec- Table 1 Mix design of bio-based mortar.
tion and crack permeability tests. Additionally, a model Amount Density
is built to obtain a mathematical basis for further choice Compound
(kg/m3) (kg/m3)
of optimum healing agent and LWA quantity to be ap- CEM I 463 3150
plied in the self-healing mortar. Finally, the results ob- Water 231.5 1000
tained from both experimental and numerical studies are Sand 0.125/1 mm 855 1650
evaluated and recommendations for future research are LWA 1/4 mm 283 650
given. While previous research (Wiktor et al. 2011;
Tziviloglou et al. 2016) has proven (experimentally) the
enhanced healing behaviour of the bio-based mortars 2.2 Preparation and cracking of mortar prisms
compared to conventional ones, this paper addresses and 9 mortar prisms were prepared containing ordinary Port-
compares methodologies exclusively on bio-based land cement (CEM I 42.5 N, ENCI, The Netherlands)
specimens. and 0.125/4 mm sand or 0.125/1 mm sand and the im-
pregnated LWA. The mixture proportions are presented
2. Materials and methods in Table 1. The prisms (40 mm x 40 mm x 160 mm)
were reinforced with two steel wires (ø 1mm) to avoid
2.1 Preparation of the healing agent complete fracture in two parts during loading and modi-
The bacteria-based healing agent consisted of spores fied with a hole in their centre, as seen in Fig. 1. The
derived from alkaliphilic bacteria of the genus Bacillus hole was created by introducing a smooth (greased)
and organic mineral compounds. The healing agent is metal bar ø 5 mm while casting, which was pulled-out
incorporated in LWA (expanded clay particles, Liapor during demoulding. All specimens were demoulded 24 h
1/4 mm, Liapor GmbH Germany) via an impregnation after casting and kept in a room with standard tempera-
under vacuum with calcium lactate- (200 g/L), yeast ture (20 ± 2 °C) and > 95% RH for 28 days (Tziviloglou
extract- (4 g/L) and bacteria spores (108 spores/L) solu- et al. 2016).
tion. Following the impregnation, the LWA were dried Three-point-bending (with a span of 100 mm) was
for approximately 5-6 days at standard temperature (20 used for the crack introduction on 28-days-old mortar
± 2 ˚C) with 60 ± 10% RH, until a constant weight was prisms. A single crack was created in each specimen. A
achieved (Tziviloglou et al. 2016). During drying pro- vertical load was applied in the middle of the specimens
cedure, the spores can remain in dormant state, since the until the formation of a stable crack. While loading, the
pH of the environment is not adequately high for them crack opening increased constantly by 0.0005 mm/s
to activate and start germinating. It was found that after until it reached approximately 0.4 mm. The crack width
drying, the impregnated LWA increased their initial was monitored via two Linear Variable Differential
weight in average by 11.3%, due to the addition of the Transducers (LVDTs) attached on the front and the back
healing agent. The increase in weight was measured side of the specimens. When the crack width reached
experimentally in a sample of 40 LWA (20 before and 0.4 mm the specimens were slowly unloaded. After
20 after impregnation). The experiment was held as unloading, the crack width reduced to approximately
follows: 0.27 – 0.36 mm. The depth of the crack was not moni-
‐ The diameter and the weight of 20 empty LWA was tored during bending, but it was observed through mi-
measured before the impregnation procedure. croscopic images. Those images revealed that the crack
‐ After drying, 20 impregnated LWA having the same was developed along the whole height of the specimens
diameters as the empty LWA used before, were (larger on the bottom and zero on the top), therefore, the
weighted. crack depth was almost equal to the height of the speci-
‐ The difference in weight was calculated for each men, namely, 40 mm.
LWA and finally an average value was obtained.

Fig. 1 Prismatic mortar specimen modified with a hole in the centre (Tziviloglou et al. 2016).
E. Tziviloglou, Z. Pan, H. M. Jonkers and E. Schlangen / Journal of Advanced Concrete Technology Vol. 15, 536-543, 2017 538

2.3 Healing treatment voids, etc.), due to casting procedure or poor compac-
Following the crack creation 6 out of 9 specimens were tion could cause unexpected local damages after crack-
placed horizontally in a plastic container filled with tap ing. Therefore, it was decided that the estimation of the
water for crack healing, while 3 of them were immedi- crack width should be undertaken by the above de-
ately tested (crack permeability test). The 6 specimens scribed methodology, where the measuring points were
were completely immersed in water while placed on the predetermined, but they could also be adjusted by the
top of 10-mm-high spacers. The container was kept user in case that the point was coinciding with a matrix
open to the atmosphere at standard room temperature flaw. In addition, the specific measuring method has
(20 ± 2 °C) with (60 ± 10)% RH. 3 out of 6 specimens also been successfully used in the literature (Wiktor et
were left to heal for 28 days, while the other 3 for 56 al. 2011; Wang et al. 2014; Palin et al. 2015).
days. Extra water was added, to keep a constant liquid- The assessment of crack volume was performed as
to-solid ratio. follows: The images (in grey scale) were thresholded to
a grey value of 120 (Fig. 3). Afterwards, the crack area
2.4 Crack inspection (in pixels) was measured and converted to mm2. In the
Crack inspection was used to primarily evaluate the microscopic images that were taken, the grey values of
sealing efficiency of the bio-based healing agent. The the mortar matrix and of the healing products were quite
inspection was conducted in two steps; i.e. right after similar. Therefore, the thresholded grey value that was
crack creation and after healing treatment. Images of the used was sufficient to separate and detect the solids
cracks were taken by a Leica MZ6 stereomicroscope (mortar and healing product) from the crack voids. For
with a Leica DFC420 camera. Image processing soft- the estimation of the crack volume it was assumed that
ware (ImageJ) was used to estimate the width, the area the crack had a triangular shape along the height of the
and ultimately the volume of the crack on the bottom of specimen. Therefore, the crack volume was estimated
each specimen. The crack width was calculated as the by multiplying the crack surface by half of the speci-
average of four measurements (w1, w2, w3 and w4) taken men’s height. The sealing percentage αm (for every
from the bottom of the specimen as seen in Fig. 2. In specimen) was calculated as in Eq. 1. Where Vi is the
many cases, shape imperfections (rough surface, air initial crack volume and Vt is the crack volume are after
healing time t (28 or 56 days).
Vi − Vt
am = (1)
Vi

2.5 Crack permeability test

Following the crack inspection, the sealing efficiency
was studied through a crack permeability test in order to
link the functional property (crack permeability) to the
visual observations. The test was performed before and
after the healing treatment. The set-up of the test is
shown in Fig. 4.
Since the specimens contained a hole along their
length, one of the two end-sides (40 mm x 40 mm) of
the sample was sealed with a glue layer to prevent water
leakage. A connector was fixed on the other end-side,
and a plastic tube was adjusted to it to let the water pass
through the 5 mm-hole and leak out of the crack. Under
the crack, an electronic scale connected with a computer
recorded and plotted the graphs of the mass of the water
Fig. 2 Measurement of crack width on the bottom of the that leaked from the cracks as a function of time. The
specimen. test lasted approximately 10 minutes. In all cases the

Fig. 3 Crack geometry on the bottom of a specimen.

E. Tziviloglou, Z. Pan, H. M. Jonkers and E. Schlangen / Journal of Advanced Concrete Technology Vol. 15, 536-543, 2017 539

2.6 Model description

A numerical simulation was conducted at meso-scale to
estimate the sealing efficiency of the bacterial-based
healing system and to determine the optimized usage of
the healing agent. The model used in the simulation was
the same as used in the experiments (a 40 mm x 40 mm
x 160 mm prism). In the meso-scale model, the mortar
(hardened cement paste and sand) and the LWA are
taken as a homogeneous matrix and a spherical inclu-
sion, respectively.
To get the meso-scale model of the mortar, the size of
each lightweight particle was randomly generated based
on the gradation curve which was obtained by a sieve
test. The process stops when the volume fraction of all
Fig. 4 Crack permeability test set-up on prismatic sam- generated LWA reaches the requirement. Then a packing
ples. algorithm based on the “take-and-place” method (Wang
et al. 1999) was adopted to place the LWA into the
relation between the crack flow and time was linear. meso-scale model. During this procedure, a separation
Therefore, the crack flow of each specimen was calcu- check was conducted for each particle to ensure that it
lated as the slope of the resulting curve. The rate (r) of was not overlapping with any other particle which al-
the recovery of water tightness was calculated as seen in ready existed in the model. Figure 5 shows a typical
Eq. 2. Where f n − h is the average (out of three speci- example of the simulated meso-scale model.
mens) crack flow of the non-healed specimens in g/min Following the creation of the meso-scale model, an
and f h is the average (out of three specimens) crack artificial V-shaped crack was assumed in the middle of
flow of the healed specimens in g/min. Unlike the seal- the prism, according to the experimental observations.
ing percentage αm, the recovery rate r was calculated Based on the size and the position of the LWA, all the
for the total number of the tested specimens (in this case lightweight particles which were intersected by the
three). The test was performed before healing treatment crack could be identified. These LWA were considered
on a set of three and after healing on another set. The as damaged by the crack. Consequently, the healing
reason for testing different sets before and after healing agent lying inside the pores of the LWA would be re-
was to avoid any loss of healing agent during the first leased to form the chemical products which can seal the
permeability test that could possibly affect the healing crack. The sealing percentage (αs) according to the
process. This fact can influence the obtained results, in a simulation can be expressed generally as in Eq. 3,
sense that the tested cracks are not identical and might
heal differently. Yet, the results can give a good indica- Vsp Vcp ⋅ β
αs = = (3)
tion for the extend of healing that has taken place during Vcr 0.5 ⋅ d cr ⋅ wcr ⋅ l y
the healing treatment and they could be used for a solid
comparison. where Vsp , Vcp and Vcr are the volume of the sealing
product, the volume of the cracked particles and the
fn−h − fh volume of the crack, respectively; d cr is the crack depth
r= (2)
fn−h which in this case (to compare with the experimental

Fig. 5 A schematic of meso-scale model of mortar containing LWA.

E. Tziviloglou, Z. Pan, H. M. Jonkers and E. Schlangen / Journal of Advanced Concrete Technology Vol. 15, 536-543, 2017 540

results) coincides with the height of the specimen; wcr Table 2 Estimated crack width, volume and closure
is the crack width; l y is the length of the model in the y percentage.
direction shown in Fig. 5 and β is a dimensionless Crack volume
parameter which represents the amount of sealing prod- Specimen Crack width before after αm
uct that can be formed for each unit of volume of LWA. healing healing
In the simulation of a three-point bending test, the crack (mm) (mm3) (%)
is assumed to be initiated in the middle of the specimen,
i.e., x = 0.5l , where lx is the model size along the x- 1 0,289 253,98 - -
direction. It is also assumed that the crack can propagate 2 0,315 287,90 - -
through a lightweight particle, due to its lower strength 3 0,266 207,26 - -
compared to a normal aggregate. Thus, the cracked 4 0,339 334,22 114,04 65,88
LWA can be identified by checking whether they have 5 0,294 259,32 172,24 33,60
been intersected by the crack surface. Then, the total 6 0,356 312,5 52,66 83,15
volume of cracked LWA, i.e., Vcp , can be calculated 7 0,363 338,92 27,66 91,83
according to the diameter of each cracked particle. 8 0,285 243,78 77,02 68,41
The concept of the bacteria-based self-healing con- 9 0,296 274,74 3,56 98,70
crete indicates that in the presence of oxygen and water
inside the crack, the dormant bacterial spores are acti- Table 3 Estimated crack width and sealing percentage
vated. Later, the active bacteria cells convert the cal- obtained by the simulations.
cium lactate (CaC6H10O6), present in the healing agent
to calcium carbonate by using oxygen. Thus, when the Specimen Crack width αs
healing agent of the cracked LWA is released into the (mm) (%)
cement paste, the following chemical reaction will occur 4 0,339 60,14 ±3,87
(Wiktor et al. 2011): 5 0,294 69,38 ±4,27
6 0,356 57,20 ± 3,50
CaC6 H10 O6 +6O6 → CaCO3 +5CO 2 +H 2 O (4) 7 0,363 56,23 ±3,63
8 0,285 71,43 ± 4,54
Carbon dioxide (CO2) formed during the above reac- 9 0,296 68,70 ±4,24
tion can further react with portlandite Ca(OH)2 present
in the mortar matrix as a product of cement hydration: variation (γ=1.4~38.7). Thus, the simulation is per-
Ca(OH) 2 +CO 2 → CaCO3 +H 2 O (5) formed in a probabilistic approach. In addition, the
standard deviation was also considered, in order to
Based on the above chemical reactions, CaCO3 is know whether the variation of input parameters can sig-
considered as the product which seals the crack. Thus, nificantly affect the output result.
β in Equ. 3 can be calculated as:
3. Results
γ ⋅ 650 ⋅ 6 ⋅ 0,10009
β= = 0, 66 ⋅ γ (6)
0, 21822 ⋅ 2711 3.1 Crack sealing estimation
The specimens were loaded until the crack opening
In Eq. 6, 650 (kg/m3) is the packing density of the reached 0.4 mm. However, after unloading the crack
LWA used in the experiment; 0.21822 (kg/mol) is the width was varying from 0.266 to 0.356 mm, based on
molar mass of calcium lactate; 0.10009 (kg/mol) and the stereomicroscopic observations. Table 2 shows the
2711 (kg/m3) are the molar mass and density of CaCO3, measured crack width, the crack volume before and
respectively; the coefficient 6 represents the (total) mol after the healing treatment and the calculated sealing
of CaCO3 that can be produced by 1 mol of calcium percentage (αm) based on microscopic observations.
lactate (Eqs. 4 and 5); γ is the mass of the healing Specimens 1, 2 and 3 were used as reference for the
agent inside the unit mass of the LWA which is meas- crack permeability test, and so were not subjected to
ured in the experiment. Based on a test on several LWA, healing treatment. Specimens 4, 5, 6 and 7, 8, 9 were
it was found that γ varies between 1.4 and 38.7. Thus, submerged in water for 28 and 56 days respectively. It
γ was taken as a random variable following a uniform can be stated that crack closure percentage increased for
distribution in [1.41, 38.69]. the specimens immersed in water for 56 days in com-
With the value of β in Eq. 6 the sealing percentage parison with those immersed for 28 days. The crack
(αs) can be calculated according to Eq. 3. However, width exhibited a linear relation with the crack volume
since the meso-scale model is randomly generated, αs as seen in Fig. 6. In contrast with the crack
should be considered as a random variable. Hence, in width/volume which does not show a clear relation with
the following part of this paper, not only the average, the sealing percentage (Fig. 7). It could have been ex-
but also the standard deviation of the sealing percentage pected that the smaller crack volume would lead in
was considered. The reason to consider the standard more efficient the crack closure. However, this was not
deviation is that this parameter (γ) has a large degree of concluded from the microscopic observations. Table 3
E. Tziviloglou, Z. Pan, H. M. Jonkers and E. Schlangen / Journal of Advanced Concrete Technology Vol. 15, 536-543, 2017 541

Table 4 Results from crack permeability test on prismatic Table 5 Values for sealing parameters (αm, αs and r) for
mortar specimens. set of specimens immersed for 28 and 56 days in water.
Crack Water flow Sealing parameter Days of immersion
r
Sample width fn-h fh 28 56
(mm) (g/min) (%)
αm 60,88 86,31
1 0,289 81,12 -
2 0,315 134,14 - - αs 62,24 65,45
3 0,266 64,7 - r 69,41 91,75
4 0,339 - 27,46
5 0,294 - 21,66 69,41
6 0,356 - 38,04 3.2 Crack permeability results
7 0,363 - 10,9 The results from the crack permeability test and the re-
8 0,285 - 6,72 91,75 covery of water tightness for the two sets of specimens
9 0,296 - 5,88 are presented in Table 4. The results from the water
flow test on cracked non-healed specimens indicated
shows the measured crack width and the sealing per- that a larger crack area resulted in an increased flow. In
centage (αs), as derived from the simulations. Contrary addition, from the calculation of the recovery of water
to the relation between crack width and αm, the relation tightness it was derived that the longer immersion pe-
between crack width and αs is linear (Fig. 7). In other riod led in decreased water flow, therefore, a more effi-
words, the model indicates that the narrower the crack cient the sealing.
the more efficient the sealing.

Fig. 6 Estimated crack width-estimated crack volume of mortar specimens.

Fig. 7 Estimated crack width - crack closure percentage (microscopic observations) and sealing percentage (simula-
tions).
E. Tziviloglou, Z. Pan, H. M. Jonkers and E. Schlangen / Journal of Advanced Concrete Technology Vol. 15, 536-543, 2017 542

4. Discussion more realistic results. The permeability test results can

be improved if the test will be performed on the same
This study used three different parameters to evaluate set of specimens before and after healing. The danger of
the sealing efficiency of the bio-based self-healing sys- losing the healing agent during the first cycle of testing
tem. All three (αm, r and αs) were based on different can be avoided by using a liquid with higher pH (possi-
ways of investigation, namely microscopy observations, bly carbonate-bicarbonate buffer), where calcium lactate
crack permeability test and numerical simulations. is less soluble. In addition, a straight rather than a V-
However, they all specified the sealing efficiency of (a) shaped crack could cause less confusion in the interpre-
crack(s) with a certain width. The comparison of αm and tation of the results. Moreover, the image processing
αs values (Fig. 7) revealed that the sealing values ob- could be more accurate if the microscopic observations
tained by the two different methods are not in good will be replaced by 3D CT-scanning. Further experi-
agreement. This can be attributed to the assumptions mental research should also investigate whether the nu-
made for both the crack volume calculations and the merical model needs to include a “reduction factor”
numerical simulation. For example, αm is based on crack which will determine the amount from the initially ex-
volume estimation assuming that the ratio filled-to- isted healing agent that participates in crack filling
empty crack area, along the crack depth, is the same as process.
on the bottom of the specimen. This is probably overes-
timating the real volume of the CaCO3 in the crack. In 5. Conclusions
addition, (for the calculation of as) the model considers
that all the healing agent included in the LWA inter- In conclusion, this paper summarized three different
sected by the crack was released and converted into methodologies to evaluate the sealing efficiency of the
CaCO3. In reality, only a part of the healing agent is bio-based self-healing mortar; two of them were based
released, while the rest remains in the lightweight parti- on experiments (image processing and crack permeabil-
cle. Moreover, blockage (from sealing products or from ity test), while the last one was based on computer
impurities) in the crack can cause depletion of oxygen simulations. It was concluded that the experimental ap-
and therefore limited conversion of the feed into sealing proaches might overestimate the crack closure due to:
product. Furthermore, in the model, the duration of the ‐ the hypothesis that the crack filling is constant along
immersion period was not taken under consideration, the crack length, or
consequently αs was independent of the healing period. ‐ the fact that the crack permeability test was con-
Thus, αs exhibited roughly similar values for 28 or 56 ducted in different sets of specimens before and after
days of immersion (62.2% and 65.45% respectively, see healing treatment, or
Table 5). The difference in αs for 28 and 56 days could ‐ the smaller crack width at the position of the (water
be exclusively attributed to the different crack widths of passage) hole compared to the crack width on the
the specimens (4, 5, 6 and 7, 8, 9). Finally, the autoge- bottom of the specimens.
nous healing processes, which can take place during In addition, the numerical model results can slightly
water immersion, have not been considered in the simu- overestimate the volume of the filling product, due to
lation. This is an additional reason why the sealing re- the assumption that all the healing agent included in the
sults originating from the model are lower (at least for LWA intersected by the crack was released and con-
56-days-old specimens) than those from the experimen- verted to calcium carbonate. Yet, in general, the model
tal investigation. underestimates the total volume of CaCO3 produced in
Furthermore, the comparison among αs, αm and r was the crack because:
conducted for sets of specimens and not separately for ‐ the duration of the immersion period was not consid-
each specimen. Table 5 presents the average values of ered as a model variable; therefore, the sealing effi-
αm and αs, as well as r values for the sets of three speci- ciency was independent of the healing period and
mens immersed in water for 28 and 56 days. In this case, ‐ the autogenous healing processes that occur during
the results of αm and r are in good agreement for both the healing treatment in water were also not consid-
immersion periods. The αs average value for 28 days of ered.
immersion is similar to the average values of the other As a matter of fact, the reactions that take place in-
two parameters, however, for 56 days the value is sig- side the crack are quite complex and depend on several
nificantly lower. In general, r is higher than the average factors, such as the (local) crack width, the presence of
αm and αs, since the crack flow test is highly dependent oxygen, the duration of healing treatment etc. Although
on the crack opening at the position of the hole. Usually both experimental and computational methodologies
at this position (20 mm from the bottom of the speci- need some improvements to resemble a more realistic
men) the crack width is narrower than on the bottom situation, the current approaches can already provide an
and it is more likely to be fully sealed. In this case the indication of the crack sealing behaviour originating
real r value is probably overestimated. from the bio-based self-healing system.
Hence, some modifications on the experimental as
well as on the computational part are needed to obtain
 E. Tziviloglou, Z. Pan, H. M. Jonkers and E. Schlangen / Journal of Advanced Concrete Technology Vol. 15, 536-543, 2017 543

Acknowledgements Roig-Flores, M., Moscato, S., Serna, P. and Ferrara, L.,

The authors would like to acknowledge the financial (2015). “Self-healing capability of concrete with
support of European Union Seventh Framework Pro- crystalline admixtures in different environments.”
gram (FP7/2007-2013) under grant agreement no Construction and Building Materials, 86, 1-11.
309451 (HEALCON). Sangadji, S., (2015). “Porous network concrete : a bio-
inspired building component to make concrete
References structures self-healing.” Thesis (PhD). Delft
de Rooij, M., Van Tittelboom, K., De Belie, N. and University of Technology.
Schlangen, E., (2013). “Self-healing phenomena in Schlangen, E. and Joseph, C., (2009). “Self-healing
cement-based materials : State-of-the-art-report of processes in concrete.” In: S. K. Gosh Ed. Self-
RILEM Technical Committee 221-SHC”. Dordrecht, Healing Materials: Fundamentals, Design Strategies,
The Netherlands, Springer. and Applications.Wiley-VCH Verlag GmbH & Co.
Edvardsen, C., (1999). “Water permeability and KGaA : Weinheim, Germany: 141-182.
autogenous healing of cracks in concrete.” ACI ter Heide, N., (2005). “Crack healing in hydrating
materials journal, 96(4), 448-454. concrete.” Thesis (MSc). Delft University of
Feiteira, J., Gruyaert, E. and De Belie, N., (2016). “Self- Technology.
healing of moving cracks in concrete by means of Tziviloglou, E., Wiktor, V., Jonkers, H. M. and
encapsulated polymer precursors.” Construction and Schlangen, E., (2016). “Bacteria-based self-healing
Building Materials, 102, Part 1, 671-678. concrete to increase liquid tightness of cracks.”
Hearn, N., (1998). “Self-sealing, autogenous healing Construction and Building Materials, 122, 118-125.
and continued hydration: What is the difference?” Van Tittelboom, K., De Belie, N., Lehmann, F. and
Materials and Structures, 31(8), 563-567. Grosse, C. U., (2012). “Acoustic emission analysis
Hilloulin, B., Hilloulin, D., Grondin, F., Loukili, A. and for the quantification of autonomous crack healing in
De Belie, N., (2016). “Mechanical regains due to self- concrete.” Construction and Building Materials,
healing in cementitious materials: Experimental 28(1), 333-341.
measurements and micro-mechanical model.” Cement Van Tittelboom, K., Gruyaert, E., Rahier, H. and De
and Concrete Research, 80, 21-32. Belie, N., (2012). “Influence of mix composition on
Huang, H. and Ye, G., (2012). “Simulation of self- the extent of autogenous crack healing by continued
healing by further hydration in cementitious hydration or calcium carbonate formation.”
materials.” Cement and Concrete Composites, 34(4), Construction and Building Materials, 37, 349-359.
460-467. Wang, J., De Belie, N. and Verstraete, W., (2012).
Jonkers, H. M., (2011). “Bacteria-based self-healing “Diatomaceous earth as a protective vehicle for
concrete.” HERON, 56(1/2), 1-12. bacteria applied for self-healing concrete.” Journal of
Joseph, C., (2008). “Experimental and numerical study Industrial Microbiology and Biotechnology, 39, 567-
of the fracture and self-healing of cementitious 577.
materials.” Thesis, (PhD). Cardiff University, UK. Wang, J., Dewanckele, J., Cnudde, V., van Vlierberghe,
Malm, F. and Grosse, C. U., (2016). “Efficiency of self- S., Verstraete, W. and de Belie, N., (2014). “X-ray
healing agents for cementitious materials computed tomography proof of bacterial based self-
characterized by NDT.” 19th World Conference on healing in concrete.” Cement and Concrete
Non-Destructive Testing, 13-17 June 2016, Munich, Composites, 53, 289-304.
Germany. Wang, Z. M., Kwan, A. K. H. and Chan, H. C., (1999).
Pacheco, J., Šavija, B., Schlangen, E. and Polder, R. B., “Mesoscopic study of concrete I: generation of
(2014). “Assessment of cracks in reinforced concrete random aggregate structure and finite element mesh.”
by means of electrical resistance and image analysis.” Computers & Structures, 70(5), 533-544.
Construction and Building Materials, 65, 417-426. Wiktor, V. and Jonkers, H. M., (2011). “Quantification
Palin, D., Wiktor, V. and Jonkers, H. M., (2015). of crack-healing in novel bacteria-based self-healing
“Autogenous healing of marine exposed concrete: concrete.” Cement and Concrete Composites, 33(7),
Characterization and quantification through visual 763-770.
crack closure.” Cement and Concrete Research, 73, Zemskov, S. V., Jonkers, H. M. and Vermolen, F. J.,
17-24. (2011). “Two analytical models for the probability
Reinhardt, H. W. and Jooss, M., (2003). “Permeability characteristics of a crack hitting encapsulated
and self-healing of cracked concrete as a function of particles: Application to self-healing materials.”
temperature and crack width.” Cement and Concrete Computational Materials Science, 50(12), 3323-3333.
Research, 33(7), 981-985.

View publication stats