Effect of Fly Ash on Self-healing of Cracks in [609290]

Effect of Fly Ash on Self-healing of Cracks in
Concrete
K.A.S.D. Ratnayake and S.M.A. Nanayakkara
Department of Civil Engineering
University of Moratuwa
Sri Lanka
[anonimizat], [anonimizat]

Abstract —The design of water retaining structures is mainly
based on the serviceability limit state crack control. The allo wable
crack width depends on the self-healing ability of concrete and the
use of supplementary cementitious material like fly ash in conc rete
mixes might affect it. Therefore, an experimental investigation was
carried out to find the influence of fly ash on self-healing pr ocess.
Water is sent through artificially induced cracks in a specimen for
autogenous healing to take place at a constant pressure gradien t
across the specimen. To determine the level of self-healing, th e flow
through the crack was measured with time to obtain the sealing
time. Fly ash percentages of 20%, 30% and 40% were tested along
with a 0% fly ash mix. Insignificant variation was shown for in itial
d r o p i n f l o w r a t e a c r o s s f l y a s h p e r c e n t a g e s u s e d i n t h i s s t u d y .
Sign ific a n t r ed uc tion in se a lin g tim e wa s o b se rv ed fo r 20% an d
30% fly ash mixes as compared to 0% fly ash whereas higher fly ash percentages (40%) showed insignificant reduction.
Keywords— water retaining stru ctures; self-healing; crack
width; concrete; fly ash; sealing time
I. INTRODUCTION
A. Introduction to self-healing of cracks in concrete
The phenomenon of self-healing of cracks in concrete has
been observed and studied for a long time. The process of self-healing can be attributed to three major processes [1]:
• Further hydration of unhydrated cement at the cracked
surface
• Recrystallization of portlandite leached from the bulk
paste
• Formation of calcite (CaCO
3)
Design of water retaining structures (WRS) is based on the
crack width limitation criteria. Previous studies have shown th at
supplementary cementitious material (SCM) used in the concrete mix have a large impact on the self-healing performance. SCM includes pulverized fuel ash (fly ash) (FA), silica fume and ground granulate d blast-furnace slag (ggbfs)
B. Previous studies on self-healing
Various parameters affecting this mechanism, such as crack
width, water seepage rate through cracks and temperature have been studied in earlier researches [2]. Studies have been done on
strength, ultrasonic pulse velocity (UPV), rapid chloride
permeability (RCP) and sorptivity of concrete, considering high
percentages of FA in the mix [3]. Most of these are implicit methods of quantifying the ability to self-heal and do not
simulate the real situation of water retaining structures where
water is flowing through a crack. The effect of pressure gradie nt
across a crack in concrete with ordinary Portland cement (OPC)
has also been studied and an optimum pressure gradient for sealing crack widths of different sizes was found [4]. Further,
smart cementitious materials which can self-heal with minimal external support have also been explored [5].
C. Test methods used to study se lf-healing in previous studies
• Tensile splitting test used to split cylindrical specimen and
water sent through the interc onnected crack across the
diameter and time taken to seal the crack was used to quantify performance [4]
• Cylindrical specimen was loaded to 70% and 90% of
compressive strength to induce interconnected cracks.
UPV, RCP test and sorptivity were measured to quantify how much the specimen had sealed [3]
• Crushing of specimen to 75 μm fine particles and testing
for hydration degree to see if healing can be sufficiently done by the unhydrated cement [6]
D. Trends shown with other parameters
Higher FA content showed positive trends in unhydrated
OPC, reduction of porosity [6], RCP test values, and an optimal
FA content was recognized for each. FA mixes have shown better performance in self-healing than ggbfs [5]
Comparison of Portland limestone cement and OPC have
shown no appreciable difference in self-healing [7]. For a give n
hydraulic gradient across a crack, optimal crack width was identified to seal crack in minimum time. Experiments conducted in the range 20 °C to 80 °C concluded that the rate o f
self-healing increased from 20 °C to 80 °C.
E. Importance and identifica tion of knowledge gap
However, the study of how supplementary cementitious
material such as fly ash, silica fume and ggbfs affect the proc ess
of self-healing has not been looked into in depth. The most commonly used SCM in the construction industry, especially in the construction of WRS is FA. With the introduction of FA, the
percentage of OPC in the mix reduces. Thereby, the amount of Ca(OH)
2 produced from the hydration of OPC varies and
consequently may affect the self-healing process. Hence the
978-1-5386-4417-1/18/$31.00 ©2018 IEEE
2642018 Moratuwa En gineerin g Research Conference (MERCon)

effect of fly ash on the self-healing process of cracks in WRS
was focused in this study.
II. OBJECTIVES
The main objective was to study how sealing time of a crack by
self-healing process is affected by using Class F FA as a SCM.
III. METHODOLOGY
A. Selection of methodology
The purpose of the research is to find out how cracks in WRS
respond to the flow of water through them and get information on how fast the crack seals by itself. Implicit quantification of
certain changes of properties in material will not resemble thi s
situation properly. Therefore, to simulate this exact situation ,
cracks needed to be induced artificially and water passed
through the cracks should be monitored with time.
Previous work done in this regard showed that splitting of a
cylindrical specimen as in the tensile splitting test was a via ble
option since it ensured an interconnected crack across its diameter in a cylindrical specimen [4].
B. Specimen details
A cylindrical specimen of 150 mm diameter specimen and
200 mm length was chosen. To avoid the separation of the two halves when subjected to tensile splitting test, 3 of 6 mm mild
steel bars were inserted in the direction perpendicular to the expected crack propagation direction at approximately 75 mm
axial spacing. The bars were held in place by connecting the 3
R6 bars onto 2 R6 bars running axially through the sample much like a ladder formation as shown in Fig 1. See Fig 2 for schematic arrangement.
BS 8007:1987 recommends 0.5 water/cement ratio for mixes
having FA and a maximum cement content of 450 kgm
-3. Hence
a Grade 35 mix having the mix proportions given in Table I was chosen as the base proportions (i.e. OPC only).
Fig 1. Reinforcement arrangement and sizing of specimen (dimens ions in mm)
TABLE I. BASE MIX PROPORTIONS FOR 1 M3
Content Amount (kg)
OPC 400
Water 200
Fine aggregate (River sand) 810
Coarse aggregate 990 The objective of the research is to find the performance of
FA over OPC in the self-healing process. Hence, the total
cement content of the mix was kept the same and the OPC in the
mix was replaced with FA on the basis of mass percentage.
Initial survey of several mix designs for 35A (used for water
retaining structures) showed that approximately 25% of cement in it consists of FA and approximately 0.35 ratio of FA/OPC. Therefore, the replacement of OPC was done in 20%, 30% and 40% of total cement content.
Due to density the difference in FA (2150 kgm
-3) and OPC
(3150 kgm-3), the other components also needed adjustment for
1m3 of concrete. The selected mixes are given in Table II.
Out of the 3 cylindrical specimens made for each mix, only
2 usable specimens could be obtained for 0% FA.
Along with the cylindrical specimens, three 150 mm cubes
were cast for each mix for obtaining strength properties as wel l.
TABLE II. MIX PROPORTIONS USED IN EXPERIMENT
% FA FA (kg) OPC
Cement
(kg) Water
(kg) Fine
aggregate(R/
Sand) (kg) Coarse
aggregate (20
mm) (kg)
0 0 400 200 810 990
20 79.0 316.0 197.5 799.8 977.6
30 117.8 274.8 196.3 794.9 971.5
40 156.0 234.1 195.0 789.9 965.5
(a) Front view

(b) Side view
Fig 2. Schematic diagram of arrangement (dimensions in mm)
265

C. Crack width and pressure gradient
For 0.1 mm crack width, pressure gradients ranging from 4
to 10 m/m are acceptable since reasonable interpolation from previous research shows that approximately 5 m/m will seal a 0.1 mm crack in the minimum time [4]. For 200 mm specimen, the head required for this pressure gradient ranges from 0.8 m to
2 m. The arrangement of the test specimen is shown in Fig 2.
Crack width was measured by an optical microscope (20X
magnification) with a scale (Fig 3). Steel straps on either end of
the specimen were used to control the crack width to the requir ed
size. The crack width was measured at several places to determine an average measured crack width
D. Curing
The self-healing phenomenon is required when the WRS is
filled with water, which may happen a long time after
construction of the structure. Hence, the self-healing ability should be tested when a majority of FA and OPC has been hydrated.
Normally, FA mixes have a lower early strength gain and a
better late strength gain. The study by Y.M. Zhang et al [8] fo und
that at 28 days, the fraction of FA reacted is more than 75% as
that at 90 days (S4 represents 40% FA mix in Fig 4, S4(A) has 40% FA with 3% Na
2SO 4, S6 has 60% FA in mix etc.).
Since a major portion of the fly ash in the mix is expected to
react within 28 days, it is acceptable to start conducting the testing self-healing of the specimens at 28 days.
The cylindrical specimen was cured for 28 days before the
tensile splitting test was done.
Fig 3. 0.1 mm crack through microscope
Fig 4. Fraction of reacted FA with time [8] Cubes were also cured for 28 days and checked for
compressive strength.
The final setup of the apparatus was as shown in Fig 5.
E. Data collection
Buckets were used to collect the water and the data point was
taken to be at the average of the start and end time.
Since literature review showed a large initial drop in the
discharge for all samples tested in a similar manner to this, t he
time between consecutive measurement of discharge rates was kept low for the initial flow measurements to accurately plot t he
flow rate decline vs time.
݁ݐܽݎݓ݋݈ܨ ሺܳሻൌሺܹ
௧௢௧௔௟െܹ௕௨௖௞௘௧ሻ
οݐെሺͳሻ
Δt = t 2 – t 1
t1 – time at start of water collection period
t2 – time at end of water collection period
Wtotal – total weight of bucket and water inside
Wbucket – weight of bucket
Q – flow rate at time (t 1+t2)/2
After the initial drop in flow rate, interval between readings
were lengthened such that sufficient quantity of water was collected to take a reading (approximately 500 ml – 3000ml). The measurements were taken until the cracks were sealed fully.
The average flow rate in the time interval between time t
1 and t 2
was obtained from equation (1). For reasonable approximation, the flow rate obtained was assumed to be the flow rate at time (t
1+t2)/2.
IV. RESULTS
The compressive strengths of cubes are given in Table III.
Average measured crack widths of cylindrical specimens
w e r e i n t h e r a n g e o f 0 . 1 m m t o 0 . 1 5 m m i n m o s t o f t h e specimens. Crack widths of some samples could not be lowered beyond 0.2 mm even after maximum tightening of steel strap.
Since the two surface crack sizes measured on either end of
Fig 5. Final arrangement of specimens and apparatus
266

TABLE III. FA% AND COMPRESSIVE STRENGTH
FA % Compressive strength (MPa)
0 39.4
20 36.8
30 36.7
40 33.3

the specimen will not give a proper value of the crack size wit hin
the sample, the initial flow rate which was calculated from the
initial flow measurement was used to find the average crack width of the specimen, assuming equation of laminar flow through 2 parallel plates as given by equation (2).

ݍൌοݓ݀݌
ଷ
ͳʹ݈ߟെሺʹሻ
where
q – initial water flow rate (m3/s)
Δp – differential water pressure between inlet and outlet of
the crack / N/m2
d – surface crack length /m w – crack width /m l – flow path length of a crack /m η – absolute viscosity of water /Ns/m
2
The calculated crack width from this method and
corresponding measured crack width for each sample is given in Table IV.
The number of the sample name represents the percentage of
FA in the mix and the letter gives a unique ID to the particula r
sample (eg: 2A – 20% FA of sample A).
Since the calculated crack width gives a more realistic
equivalent crack width representing the crack width variation along the specimen, this will be used in the analysis of result s.
All of the samples showed a rapid drop in flow rate at the
start (See Fig 7) and then the flow rate reduction was gradual (See Fig 8).
From the flow rate data, the initial reduction of flow rate to
0.01 ml/s was calculated by linear interpolation of the two closest values. The time taken for that reduction for different
samples were plotted against the different crack widths as show n
in Fig 9.
TABLE IV. MEASURED AND CALCULATED CRACK WIDTH OF
SPECIMENS
Sample name Measured average
crack width (mm) Calculated crack width
(mm)
0A 0.12 0.094
2A 0.20 0.199
2B 0.13 0.138
2C 0.11 0.125
3A 0.10 0.119
3B 0.13 0.145
3C 0.13 0.132
4A 0.098 0.154
4B 0.11 0.116
4C 0.26 0.237 Fig 6. Graphical representation of crack size distribution
Fig 7. Flow rate with time (initial drop)
Fig 8. Flow rate with time (gradual drop)
From Fig 9, it can be seen that between the crack sizes of
0.094 to 0.15 mm, the time taken for flow rate to drop to 0.01 ml/s was approximately the same for all specimens considered.
Considering 20% FA specimens only, it can be seen that the time
has not changed significantly from one specimen to the other.
The same can be observed for specimen with 30% fly ash
(3B and 3C). 3A, however seems to be an outlier in this case. The non-uniform crack width across the sample may have affected the initial drop.
The same can be observed for specimen with 30% fly ash
(3B and 3C). 3A, however seems to be an outlier in this case. 0.0000.0010.0100.1001.000
0 240 480 720 960 12001440Log flow rate (ml/s)
Time (h)00A 0.094
20B 0.138
20C 0.125
30A 0.119
40B 0.1160.0100.1001.00010.000
04 4 8 4 8 0Log flow rate (ml/s)
Log time (h)0A 0.094 2A 0.199 2B 0.138
2C 0.125 3A 0.119 3B 0.145
3C 0.132 4A 0.154 4B 0.1160A4B3A2C3C2B3B4A2A4C
0.000.050.100.150.200.250.30crack width (mm)
SampleCalculated
CW
Measured
CW
267

Fig 9. Time taken for specimens to reach 0.01 ml/s flow for dif ferent specimens
The non-uniform crack width across the sample may have
affected the initial drop.
However, considering a much larger crack size of 0.20 mm,
the time for this initial drop has been understandably delayed due to the large crack size. Fig 10 shows the time taken to sea l
the cracks completely.
Some specimens were not used in this comparison since the
flow rate of those specimens showed a sudden increase after an accidental discontinuity in water supply. The sudden increase i n
flowrate could be assumed to be due to the dislocation of the particles inside the crack.
For comparison purposes, the sealing time for specimens
were normalized to that of a 0.10 mm crack width specimen using the sealing time variation with crack width for pressure
gradient of 10 given in the reference [4].
The control sample 0A with a crack width of 0.094 mm has
only a deviation of 13.3% from the value obtained for the same crack size from Fig 11 at a pressure gradient of 10.
The curve corresponding to pressure gradient 10 shown in
Fig 11 can be used to normalize sealing times for crack widths
in the range of 0.025 mm to 0.15 mm. Since the same pressure
gradient was used in this experiment, it is acceptable to use t his
curve. The accuracy of the relevant curve in Fig 11 has an R
2
value of 0.9285, which is acceptable.
From this curve, the time taken for sealing a 0.1 mm crack is
Fig 10. Graph of time taken for specimens to seal 698 hours. Hence, all values will be normalized for 0.1 mm
crack width (CW) using a normalization factor in equation (3).
The normalized sealing time (ST) is calculated by equation (4).
ݎ݋ݐ݂ܿܽ݊݋݅ݐܽݖ݈݅ܽ݉ݎ݋ܰ ൌܶܵሺܹܥͲǤͳ݉݉ሻ
ܶܵሺ݉݉ݔܹܥሻെሺ͵ሻ
ܶܵሺܹܥͲǤͳ݉݉ሻ
ܶܵሺ݉݉ݔܹܥሻൌ͸ͻͺ
͵Ͷʹͺ͸ʹǤͷͶݔଶെͷͷͺͲǤͺͺݔ ൅ʹͺʹ͹Ǥʹͳ
ܶܵ ൌ ݎ݋ݐ݂ܿܽ݊݋݅ݐܽݖ݈݅ܽ݉ݎ݋ܰ ൈ݈ܶܵ݁݌݉ܽݏ݈ܽݑݐܿܣെሺͶሻ
Based on the normalized sealing times given in Table V and
shown in Fig 12, the following observations can be made:
• FA mixes having 20% and 30% have shown
approximately 25% reduced sealing time as compared to
the 0% FA mix
• FA mixes having 40% have shown only 6% reduction in
sealing time compared to 0% FA mix
V. CONCLUSION
The initial rapid drop in flow rate was common to all the
mixes and insignificant variation was shown with respect to FA percentages.
A s f o r t h e p e r f o r m a n c e o f f u l l c l o s u r e o f t h e c r a c k b y s e l f –
healing, there was a significant difference in the time taken w ith
respect to FA percentages.
Since comparison between FA percentages and sealing time
could not be done with samples having different crack widths,
TABLE V. NORMALIZED SEALING TIME
Fig 11. Sealing time vs crack width [4] Sample Crack
width
(mm) Time
for
crack
to seal
(h) Normalization
factor Actual
sample
sealing
time (h) Normalized
ST (h)
0A 0.094 632 1.1038 716 790
2B 0.138 1687 0.4137 1365 565
2C 0.125 1237 0.5641 1075 606
3A 0.119 698 0.6531 861 562
4B 0.116 993 0.7024 1052 739 0A4B3A
2C3C2B3B4A2A
0100200300400500600700800900
0.09 0.11 0.13 0.15 0.17 0.19 0.21time taken to reduce flow to 0.01
ml/s (h)
crack width (mm)
0A4B
3A2C2B
05001000150020002500
0.09 0.10 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.20Total time taken to seal 100% (h)
crack width (mm)
268

Fig 12. Normalized sealing time variation for different FA mixe s
could not be done with samples having different crack widths,
the sealing time of the specimens were normalized to 0.1 mm.
• FA mixes having 20% and 30% have shown
approximately 25% reduced sealing time as compared to the 0% FA mix
• FA mixes having 40% have shown only 6% reduction in
sealing time compared to 0% FA mix
From these results, it can be seen that 20% and 30% FA
mixes perform better than 0% FA in sealing cracks in the range
of 0.1 mm to 0.15 mm, which also matches with the literature.
However, very high percentage of FA such as 40% does not
show significant improvement in performance as much as the
20% and 30%.
It is not recommended to use high volume FA mixes in the
concrete mix for WRS (more than 40% FA) A
CKNOWLEDGMENT
The authors would like to thank Siam City Cement (Lanka)
Ltd for providing the fly ash used in the experiment. The assistance given by the technical staff of the Structural Testi ng
and Material Testing Laboratories in the Department of Civil Engineering is also acknowledged.
R
EFERENCES
[1] H. Huang, G. Ye, C. Qian and E. Schlangen, “Self-healing in cem entitious
materials: Materials, methods and service conditions,” Materials and
Design, vol. 92, p. 499–511, 2016.
[2] H.-W. Reinhardt and M. Jooss, “Permeability and self-healing of cracked
concrete as a function of temperature and crack width,” Cement and
Concrete Research, vol. 33, p. 981 – 985, 2003.
[3] M. Sahmaran, S. B. Keskin, G. Ozerkan and I. O. Yaman, “Self-he aling
of mechanically -loaded self consolidating concretes with high volumes of
fly ash,” Cement & Concrete Composites, vol. 30, p. 872–879, 2008.
[4] S. Nanayakkara, “Self-healing of cracks in Concrete under water
pressure,” in New Technologies for Urb an Safety of Mega Cities in Asia,
Tokyo, 2003.
[5] M. Sahmaran, G. Yildirim and T. K. Erdem, “Self-healing capabil ity of
cementitious composites incorporating different supplementary
cementitious materials,” Cement & Concrete Composites, vol. 35, p. 89–
101, 2013.
[6] P . T e r mkha jo r nk it , T . N aw a, Y . Y a mash ir o a nd T . S a ito , “ S e l f-he aling
ability of fly ash –cement systems,” Cement & Concrete Composites, vol.
31, p. 195–203, 2009.
[7] S. M. A. Nanayakkara and T. Elakneshwaran, “Self-healing of cra cks in
concrete with Portland Limestone Cement,” in 6th International Congress
in Concrete Technology, Dundee, 2005.
[8] Y. M. Zhang, W. Sun and H. D. Yan, “Hydration of high-volume fl y ash
cement pastes,” Cement & Concrete Composites, vol. 22, pp. 445 -452,
2000.
0A
2B2C3A4B
0100200300400500600700800900Sealing time (h)
Sample
269