Materials and Structures

DOI 10.1617/s11527-012-0004-2

ORIGINAL ARTICLE

Mechanisms of strength loss in underwater concrete

Joseph J. Assaad • Camille A. Issa

Received: 5 September 2012 / Accepted: 17 December 2012

Ó RILEM 2012

Abstract In-situ hardened properties of underwater Keywords Underwater concrete  Washout loss 
concrete (UWC) are affected by washout loss and Permeability  Strength loss  Falling-head method
water infiltration occurring during the casting and
post-casting stages. This paper evaluates the suitabil-
ity of the falling-head method determined using a 1 Introduction
permeameter cell to assess water infiltration or,
permeability (k), in freshly mixed UWC. Correlations Underwater concrete (UWC) develops lower in situ
between k and washout loss determined using the CRD performance than concrete cast and consolidated
C61 test method are established. The paper also seeks above water. Typical UWC residual compressive
to quantify the coupled effect of washout loss and strengths reported in literature were in the order of
water infiltration on the drop in UWC compressive 80–90 % for concrete cast using the tremie/hydroval-
strength. Test results showed that Darcy’s law is valid ve technique [1], 50–70 % for self-consolidated UWC
to evaluate water permeability in UWC. The k and depending on turbulence of water and location of
washout loss values are found to be well correlated extracted cores for strength testing [2], and as low as
within each other. UWC mixtures subjected to wash- 40 % for concrete having a slump of 230 mm made
out or water infiltration exhibited lower compressive without anti-washout admixture (AWA) [3]. The
strengths, as compared to reference concrete sampled ultimate bond stress and slip behavior of steel bars
in dry conditions. The drop in UWC strength was embedded in UWC were found to be directly affected
attributed to a combination of factors including by mixture composition and casting conditions.
washout loss of cementitious phase together with Residual stresses varying from 55 to 95 % were
relative increases in aggregate concentration and recorded, depending on the level of washout loss and
specified water-to-cement ratio. degree of aggregate segregation [4, 5].
It is reasonable to attribute the drop in UWC
strength to different mechanisms occurring during the
J. J. Assaad (&)  C. A. Issa casting and post-casting stages. During the casting
Lebanese American University, Byblos, Lebanon stage, the decrease in UWC strength may be related
e-mail: [email protected]; [email protected] due to washout loss and aggregate segregation which
C. A. Issa arise from a combination of factors such as agitation of
e-mail: [email protected] concrete during placement and consolidation, turbu-
J. J. Assaad lence of water, interfacial concrete/water velocity, and
Holderchem Building Chemicals, Amchit, Lebanon depth of the casting location [1, 4]. From the other
 Materials and Structures

hand, the drop in strength following the end of casting, measurement of washout, and was later adopted by the
i.e. during the period extending from the end of casting US Corps of Engineers by its standard version CRD
until onset of hardening, can be attributed to water C61 [10]. The test consists on placing approximately
infiltration inside the newly placed concrete. Water 2 kg of fresh concrete in a wire-mesh basket and
infiltration causes leaching of freshly mixed cemen- allowing it to free fall vertically in a column of water.
titious hydrating compounds and increase in the The basket is then retrieved at constant speed and
specified water-to-cement ratio (w/c), thus reducing weighed to determine washout loss. The CRD C61 test
strength and service life of the underwater structure. was reported to provide good comparison basis of
Limited relationships or design provisions have been various UWC mixtures in the laboratory [4]. However,
proposed in literature to quantify the effect of washout limited information can be deduced regarding infiltra-
loss and/or water infiltration on the drop in UWC tion of water inside the fresh concrete after being
strength. In fact, most of the reported strength data were placed and left at rest exposed to surrounding water
determined using samples different than those used for until the onset of hardening.
washout evaluation, thus making difficult the establish- The notion of water infiltration, or simply perme-
ment of a link between residual strength and washout loss. ability (k), has been mostly developed in soil mechan-
For example, the common approach to evaluate com- ics to evaluate the ease with which fluid moves
pressive strength of UWC consists on dropping concrete through the tortuous path of a solid skeleton with
into moulds placed in water tanks. This approach does not interconnected voids [11, 12]. The value of k (cm/s) is
simulate practical conditions, neither accurately estimate deduced from the measured percolating flow rate (Q,
the degree of washout loss that the fresh UWC sample has cm3/s) through a specimen section (S, cm2) from
undergone during the dropping process in water [6, 7]. Darcy’s law as follows: Q/S = k i, where i is the
Also, it is not accurate to correlate washout loss to vertical hydraulic gradient (dimensionless). There-
compressive strength determined on cores extracted from fore, the objective of the first phase of this paper is to
existing underwater structures, given the direct effect of determine the suitability of the permeameter test to
the casting method on washout loss [2, 8]. assess the rate of water infiltration in fresh concrete
In previous studies, Assaad et al. [9] proposed and mortar mixtures intended for underwater applica-
evaluating the residual compressive and bond strength tions. Universally available in most research centers,
of UWC during the casting stage using the same concrete the permeameter test is standardized by ASTM D2434
samples that were used for washout measurements. Such and D5084 [13, 14], and quite simple to be realized
samples were found to adequately reflect actual UWC under both laboratory and field conditions. Special
properties characterized by lower content of cementi- emphasis is placed on evaluating the repeatability of
tious phase due to washout loss and relative increase in the permeameter test as compared to the CRD C61.
coarse aggregate concentration due to segregation taking The second phase of this paper seeks to quantify the
place during casting and self-consolidation under water. drop in UWC strength due to coupled effects of
The testing method to evaluate the drop in strength washout loss and water infiltration. The loss in
consisted on subjecting a series of UWC samples to strength was determined using the same concrete
washout as per the CRD C61 test method, and then re- samples that were used for washout and permeability
mixing vigorously the washed samples in a clean measurements. Such data can be of special interest to
container for subsequent use in strength determination. researchers, concrete engineers, and contractors to
The authors concluded that the residual compressive and properly estimate the UWC residual strengths during
bond strengths can be correlated to washout loss with real underwater placements.
coefficients of correlation (R2) greater than 0.84 [4, 5].

3 Experimental program
2 Testing of UWC and objectives of this project
3.1 Materials
Several testing methods are available in literature to
evaluate washout loss of UWC [1, 3]. The plunge test Portland cement and silica fume conforming to ASTM
has been mostly popular as it provides a quantitative C150 Type I and C1240, respectively, were used in
Materials and Structures

this project. The surface area of the cement (Blaine) mixing, the set-retarder was used at dosages of 0.6 and
and silica fume (B.E.T.) were 340 and 20,120 m2/kg, 0.4 % of the cement mass in UWC and CEM mixtures,
respectively; and their specific gravities were 3.14 and respectively.
2.22, respectively. The cement had a C3S, C3A, and
Na2Oeq. values of 60.4, 6.6, and 0.73, respectively. 3.3 Batching of mixtures
Continuously graded ASTM C33 crushed limestone
aggregate with 20-mm nominal size and well-graded The concrete mixing procedure consisted of homog-
siliceous sand with 4.75-mm nominal size were enizing the sand and coarse aggregate with half of the
employed. The coarse aggregate and sand had fineness mixing water, then introducing the cementitious
moduli of 6.4 and 2.5, respectively, and bulk specific materials gradually over 30 s. The remaining part of
gravities of 2.71 and 2.69, respectively. water along with the HRWR, AWA, and set-retarder
A polycarboxylate-based high-range water reducer were then added and mixed for 1 min. After a rest
(HRWR) conforming to ASTM C494 Type F was period of 30 s, the concrete was remixed for two
used; its solid content and specific gravity were equal additional minutes. The same mixing procedure was
to 40 % and 1.11, respectively. A powder welan gum adopted for the mortar, except that the coarse aggre-
based AWA was used; it was diluted in 5 % solution of gates were removed from the batch. Testing and
the mixing water prior to addition to mortar or sampling of all concrete and mortar mixtures were
concrete. Finally, a sodium gluconate-based set- made at room temperature of 23 ± 2 °C and
retarder was used to reduce workability loss during 50 ± 5 % relative humidity.
testing.
3.4 Characterization of UWC and CEM properties

3.2 Mixture proportions 3.4.1 Determination of washout loss

As summarized in Table 1, three UWC series made Right after preparation of samples, washout of UWC
with 350, 400, and 450 kg/m3 of cement were was determined using the CRD C61 test which
investigated in this project. The sand-to-total aggre- consists on placing a fresh concrete sample in a
gate ratio remained fixed at 0.46. The mortars were wire-mesh basket and allowing it to free fall vertically
proportioned using the concrete-equivalent-mortar in a 1.7 m high column of water [10]. The sample is
(CEM) approach, i.e. same concrete composition then retrieved slowly to the surface and weighed to
except that the aggregate particles coarser than determine washout loss. Cumulative washout mass
4.75 mm were replaced by an equivalent quantity of loss after 3 drops in water is reported as W3. Around
sand to take into consideration the amount of water 2.5 kg are normally used when testing washout loss as
that can be absorbed onto their surfaces during mixing. per the CRD C61 test method.
More details on the CEM approach can be seen in An adapted version of the CRD C61 test was used to
Schwartzentruber and Catherine [15] and Assaad et al. determine washout of CEM containing no coarse
[16]. aggregates [17]. The testing method consists on
In each series, different combinations of w/ placing around 1 kg of fresh mortar in a wire-mesh
c (0.35–0.65), silica fume (0–10(% of cement)), and perforated basket and subjecting it to free fall in a 1 m
AWA (0–0.08(% of cement)) were tested. Unless high water column. The basket was similar to the one
otherwise specified, the dosage of HRWR was used in the original CRD C61 test, except that the
adjusted to secure a concrete slump of 200 ± 10 nominal diameter of the circular perforations was
mm as per ASTM C143; or an equivalent mortar free reduced to 1.5 mm (instead of 3 mm) in order to limit
flow of 205 ± 10 mm on the flow table as per ASTM the extent of material loss during filling of the basket.
C1437. The relationship between concrete slump and The distance between centers of adjacent perforations
CEM free flow is given by: Concrete slump = 2.1 9 remained the same as in the original basket (i.e.,
CEM free flow - 230 (R2 of 0.94) [16]. 5 mm). After 15 s at the bottom of the test column, the
To minimize the effect of workability changes on basket is retrieved at a speed of 0.5 m/s. Only one drop
washout loss and permeability measurements after in water is used to determine washout loss of the
 Materials and Structures

Table 1 Mixture composition of evaluated UWC and CEM mixtures

Series # 1 Series # 2 Series # 3

Type I cement (kg/m3) 350 400 450

Silica fume (% of cement) Varying from 0 to 10 %
w/c Varying from 0.35 to 0.65 %
AWA (% of cement) Varying from 0 to 0.08 %
HRWR (% of cement) Various dosages to achieve different consistencies
UWC mixture
Sand (0–4.75) (kg/m3) 930 890 840
Coarse aggregate (4.75–20) (kg/m3) 1,080 1,040 980
Volume of sand/volume of cement paste 1.36 1.14 0.96
Volume of coarse aggregate (L/m3) 401 386 364
Sand/(sand ? coarse aggregate) 0.46 0.46 0.46
Corresponding CEM mixture
Sand (0–4.75) (kg/m3) 1,060 1,005 940
Volume of sand/volume of cement paste 1.56 1.29 1.07

Table 2 Differences between original and adapted CRD C61 test methods
CRD C61 for UWC testing Adapted CRD C61 for mortar testing

Height of water tube (m) 1.7 1

Diameter of water tube (mm) 200 200
Diameter of the basket (mm) 130 130
Nominal diameter of the basket’s perforations (mm) 3 1.5
Distance between adjacent perforations (mm) 5 5
Sample used for testing (kg) 2–3 Around 1
Number of drops in water Three drops One drop
Washout referred to as: W3 W1

mortar, referred to as W1 [17]. The differences filter papers were placed between the upper and lower
between the original and adapted CRD C61 test perforated plates and tested material to retain around
methods are described in Table 2. 96 % of all particles greater than 1 lm. De-aired
distilled water at room temperature was used during
3.4.2 Determination of permeability testing to minimize the amount of air dissolved in
water which may affect permeability measurements.
A commercially available soil permeameter apparatus More details on permeability testing can be seen in
was employed for testing water permeability [18]. various geotechnical books [11].
Steel-made cells having diameter/height dimensions The permeameter stand consisted of a metal frame
of 102/105 and 152/115 mm/mm were used for testing with water tank adjustable in height between 1,500
CEM and UWC mixtures, respectively. Their inner and 5,000 mm. This allows monitoring the extent of
surfaces were coated with a thin layer of specially hydraulic gradient (i) applied on top of the tested
grease to reduce eventual water leakage along the specimen to values ranging from 10 to around 55. The
sides of the cell. To avoid entrapped air, the tested i is being calculated as the ratio of total head of water
material was well compacted using a tamping device under motion to the height of tested specimen. After
in three layers of approximately similar heights. Two opening the inlet water valve on top of the cell, outflow
Materials and Structures

is observed to ensure a continuous flow regime (i.e., 15 mm and 1 %, respectively. Average strength
indicating complete saturation of the specimen) where values determined from two compression tests were
water constantly trickles out from the outflow valve considered for each mix design.
(Fig. 1) [11, 18]. The time needed to reach such As previously, the drop in compressive strength due
regime varied from 4 to 10 min, depending on mixture to water infiltration was determined using the same
composition. After ensuring continuous flow, the concrete sample that was used for testing permeability.
value of k was determined using the falling-head Right after measuring k, the top cover of the perme-
equation given by: ameter cell is opened and water that was accumulated
  on top of the specimen during testing is carefully
a L h1 decanted. The fresh concrete was then moved to a
kðcm/sÞ ¼   ln
A Dt h2 clean container, homogenized manually, and filled in
100 9 200 mm steel cylinder. The homogenization
where a (cm2) is the cross-sectional area of the inlet
process was realized in consistent and similar manners
water valve (equal to 0.384 cm2), A (cm2) cross-
for all mixtures, so as to minimize the effect of such
sectional area of specimen, L (cm) height of specimen,
parameter on strength results. It is to be noted that one
and Dt (s) time needed for the total head to drop from
permeability test realized using the 152/115 mm/mm
clearly marked graduations h1 to h2 (Fig. 1) [11].
diameter/height cell was enough to secure the needed
Generally, Dt in the range of 2–10 min was obtained
material for filling the 100 9 200 mm cylinder. The
throughout testing to drop the level of water from h1 to
hardened concrete was then demoulded after 24 h,
h2, again depending on the mixture composition. The
cured in water, and tested for compressive strength
maximum total time that was needed to prepare the
after 28 days following ASTM C39 test method.
mix (i.e., concrete or mortar), place in the cell, ensure a
continuous flow regime, and testing permeability was
less than about 25 min.
4 Test results and discussion

3.4.3 Sampling of UWC for strength determination 4.1 Phase I: assessment of water infiltration:
comparison with CRD C61 test method
The sampling procedures for determining the drop in
UWC compressive strength due to washout loss or 4.1.1 Validity of Darcy’s law to evaluate k of UWC
water infiltration are described below (note that and CEM
reference strength determined on concrete sampled
in dry conditions, i.e. without washout or permeability Permeability computed on the basis of Darcy’s law is
testing, was also measured). limited to the condition of complete saturation of
Right after measuring W3 using the CRD C61 test, specimen and laminar flow characterized by straight
the fresh concrete sample was moved to a clean and parallel flow lines [11]. Hence, for example, for
container and covered by a wet burlap. Another 2.5 kg highly plastic clays of low permeability, Darcy’s law
of fresh UWC was then taken from the same batch, may not hold as the flow rate is so small and the soil
subjected to similar washout loss testing, and vigor- considered being impervious. From the other hand, at
ously mixed with the first sample to account to around high i values, the flow becomes turbulent with
4 kg of fresh concrete (taking into consideration the fluctuations in fluid velocity both in parallel and
material lost due to washout). The so-obtained transverse directions. This can result in significant
concrete was then filled in a 100 9 200 mm steel changes in the specimen structure due to washing out
cylinder for strength determination [9]. The methods of the finest particles and dissolution/precipitation of
for compacting the concrete in the cylinder, demold- ions towards the bottom of the sample (leaching
ing after 24 h, curing in water, capping, and testing at effect).
28 days were according to ASTM C39 test method. It To evaluate validity of Darcy’s law, permeability
is important to note that the time needed to complete measurements were conducted at different i values
the two testing cycles did not exceed 12 min, whereby ranging from 10 to 55 on various UWC and CEM
variations in slump loss and W3 values were limited to mixtures. As can be seen in Fig. 2, the variations in
 Materials and Structures

Fig. 1 Permeameter test

and sketch for the falling-
head method Water standpipe,
Area a

Inlet water h1
valve
h2

Specimen,
L Area A

Datum

Outflow

Cylinder

k due to different hydraulic gradients did not exceed cementitious materials, the void volume representing
5 % from the mean value, indicating that permeability the degree of porosity through which water percolates
is independent from i. This result is in agreement with can be considered equal to the water volume, and thus
other research studies [18, 19], indicating that k of expressed as a function of w/c [19]. Therefore,
fresh mortars and concrete determined using the mixtures prepared with lower w/c are expected to
falling-head method can be regarded as valid when possess lower porosity which can slow down water
computed with respect to Darcy’s law. percolation in the tested specimen.
The use of higher cement content resulted in
4.1.2 Effect of mixture composition decreased permeability, for given w/c. A k(CEM) value
on k measurements of 0.9 9 10-5 cm/s was obtained for the mortar made
with 450 kg/m3 cement and 0.35 w/c (Fig. 3). This can
Table 3 summarizes k(UWC) and k(CEM) values deter- be attributed to increased packing density which
mined on UWC and CEM mixtures using the 152 and hinders water movement. Furthermore, mixtures con-
102 mm diameter cells, respectively, at hydraulic taining higher cement contents require lower HRWR
gradients (i) of 20 ± 2. Also shown in Table 3 are the demand to achieve the needed flow ability (Table 3),
corresponding washout loss (W3 and W1) values. due to the relative increase in mixing water and
Effect of w/c and cement content Typical variations decrease in internal friction resulting from lower
of k(CEM) determined on mortar mixtures made with aggregate contents. The reduced HRWR dosages can
various w/c and cement contents are illustrated in thus increase cohesiveness of CEM and UWC mix-
Fig. 3. Mixtures made with reduced w/c are shown to tures and result in lower permeability.
yield lower permeability values. For example, a Effect of AWA and silica fume All UWC and CEM
decrease in k(CEM) from 6.2 to 3.2 9 10-5 cm/s was mixtures incorporating AWA exhibited a decrease in
noted for the CEM prepared with 350 kg/m3 cement permeability, particularly when such agent is added at
and w/c of 0.63 or 0.38, respectively. In newly mixed high concentration (Table 3). For example, k(UWC)
Materials and Structures

Fig. 2 Effect of hydraulic 10

gradient on k values

Mixtures with 350

kg/m³ cement content

k x 10 , cm/s
6

-5
4

2 Mortar: w/c = 0.55 Concrete: w/c = 0.55

Concrete: w/c = 0.4 Mortar: w/c = 0.4

0
0 10 20 30 40 50 60
Hydraulic gradient (i) value

Table 3 Results of k and washout loss determined on UWC and CEM mixtures
Mix Cement Silica Water w/c AWA CEM properties UWC properties
# content fume (% (kg/ (% of
(kg/m3) of cement) m3) cement) HRWR k(CEM) 9 10-5 W1 HRWR k(UWC) 9 10-5 W3
(% of (cm/s) (%) (% of (cm/s) (%)
cement) cement)

1 350 – 133 0.38 – 0.83 3.2 16.2 1.02 5.1 7.1

2 – 140 0.4 – 0.64 3.8 19.5 0.81 5.4 7.6
3 – 140 0.4 0.015 0.66 2.7 15.1 0.83 4.2 7.2
4 – 192.5 0.55 – 0.46 5 24.1 0.6 8.2 9.1
5 – 192.5 0.55 0.06 0.5 1.9 17.6 0.66 2.6 6.4
6 7 192.5 0.55 – 0.48 3.6 21.8 0.62 3.9 8.5
7 – 220.5 0.63 – 0.4 6.2 27.6 0.48 7.1 12.2
8 400 – 152 0.38 – 0.7 2.8 13.5 0.86 4.2 4.6
9 – 168 0.42 0.01 0.55 2.1 14.6 0.65 3.6 4.1
10 – 184 0.46 – 0.46 3.3 15.8 0.59 6.1 7.6
11 4 184 0.46 – 0.47 2.8 11.7 0.6 5.4 6.3
12 – 184 0.46 0.05 0.5 1.7 10 0.63 3.7 5.2
13 – 204 0.51 – 0.38 4.2 18.4 0.47 4.3 8.3
14 10 204 0.51 – 0.42 1 16.8 0.5 1.8 5.1
15 – 260 0.65 – 0.28 5.3 25.2 0.36 7.3 9
16 450 – 157.5 0.35 – 0.52 0.9 9.2 0.68 2 2.5
17 – 157.5 0.35 0.02 0.55 0.13 5.6 0.7 0.9 2
18 2 157.5 0.35 – 0.52 0.74 8.4 0.69 1.8 2.4
19 – 225 0.5 – 0.3 1.6 14.6 0.37 3.1 6.1
20 – 225 0.5 0.08 0.35 0.28 8.2 0.42 0.63 3.6
21 8 225 0.5 – 0.34 0.85 12.4 0.4 1.6 4.2
22 – 279 0.62 – 0.22 3.1 17 0.27 3.9 6.7
HRWR adjusted to achieve concrete slump of 200 ± 10 mm and CEM free flow of 205 ± 10 mm
 Materials and Structures

Fig. 3 Effect of cement 7

content and w/c on k(CEM) 6.2
determined on CEM 6

k (CEM) x 10 , cm/s
5
5

-5
4
3.2 3.1
3

2 1.6
0.9
1

0
w/c = 0.38 0.55 0.63 0.35 0.5 0.62

Cement content = 350 kg/m3 450 kg/m3

Fig. 4 Effect of AWA and 10

silica fume on k(UWC)
8.2
determined on UWC
k (UWC) x 10 , cm/s

8
-5

3.9
4 3.1
2.6
2 1.6
0.63
0
Control 0.06% 7% Silica Control 0.08% 8% Silica
AWA fume AWA fume
w/c = 0.55 0.5

Cement content = 350 kg/m3 450 kg/m3

decreased from 8.2 to 2.6 9 10-5 cm/s when the The fine silica fume particles are known by their
AWA was added at 0.06 % in the concrete made with ability to improve compaction and reduce bleeding of
350 kg/m3 cement and 0.55 w/c (Fig. 4). This is due to freshly mixed mortars and concrete [17, 18], thus
the mode of function of the AWA polymers that leading to reduced permeability (Fig. 4). For example,
adhere to the periphery of water molecules, thus the addition of 8 % silica fume reduced k(UWC) from
adsorbing and fixing part of the mixing water [3]. With 3.1 to 1.6 9 10-5 cm/s for the concrete made with
hydration of the polymer, the AWA continues to 450 kg/m3 cement and 0.5 w/c.
aggregate and begins to expand as it imbibes water. Effect of coarse aggregates Good relationship is
Molecules in adjacent polymer chains can intertwine obtained between k values determined on CEM and
and develop attractive forces through hydrogen bond those determined on UWC (Fig. 5). With the excep-
and polymer entanglement which can further increase tion of two data points (Table 3), k(UWC) was higher by
overall thixotropy and resistance of the mixture 1.2–2.2 the value of k(CEM). Just like in soils, this
towards water permeability [3]. indicates that permeability in cementitious-based
Materials and Structures

Fig. 5 Relationship 10
between k(UWC) and k(CEM)
values
y = 1.418 x
8 R² = 0.804

k (UWC) x 10 , cm/s
6

-5
4

0
0 1 2 3 4 5 6 7
-5
k (CEM) x 10 , cm/s

materials increases with the increase in particle size relatively moderate R2 of 0.69 determined on concrete
[11]. can be related to the presence of coarse aggregates
which reduces repeatability of testing, as will be
4.1.3 Relationships between washout loss discussed below.
and permeability
4.1.4 Repeatability of permeameter test versus CRD
The effect of mixture composition on W3 and W1 C61 test
determined using the original or adapted CRD C61 test
procedures is quite similar to the variations in Table 4 summarizes the average values, standard
k evaluated using the permeameter test (Table 3). deviation, and coefficient of variation (c.o.v.) deter-
For example, washout loss decreases with the increase mined following repeating five times the permeability
in cement content, decrease in w/c, and addition of and washout loss measurements on UWC and CEM.
higher concentrations of AWA or silica fume. More The mixture used for testing was made of 350 kg/m3
discussion on the effect of mixture composition on cement, 0.5 w/c, and various dosages of HRWR to
washout loss can be seen in Ref. [17]. It is to be noted yield three different consistency levels.
that washout loss determined on CEM is greater by For both UWC and CEM, the increase in consis-
approximately 2–3 times the value determined on the tency led to increased k and washout loss (Fig. 7). For
corresponding concrete (W3 = 0.396 9 W1, with R2 example, k(CEM) increased from 4.2 to 10.5 9 10-5
of 0.84). This is mainly related to the absence of coarse cm/s when the flow increased from 205 to 235 mm,
aggregates in the CEM, which increases the tendency respectively. The corresponding W1 increased from
towards washout upon contact with water [17]. The 18.5 to 25.3 %, respectively. This can be related to a
relatively lower mass of tested CEM compared to that decrease in the mixture’s cohesiveness and stability
of UWC (1 vs. 2.5 kg) may also contribute to such which allow water to permeate easily in the matrix
differences in washout loss [17]. (during the permeameter test) or lead to higher washed
Acceptable R2 values are obtained between out fine particles (during the CRD C61 test).
k (910-5 cm/s) and washout loss (%) determined The variations of c.o.v. determined on various
using the original or adapted CRD C61 on UWC and UWC and CEM are plotted in Fig. 8. Generally
CEM mixtures (Fig. 6). This suggests that the CRD speaking, the repeatability of testing decreases when
C61 and permeameter tests are both appropriate and the mixture’s consistency increases. For example, the
complement each other to assess UWC properties c.o.v. determined from k(CEM) values increased from
during the casting and post-casting stages. The 1.3 to 2.2 and 3.7 % when the CEM flow increased
 Materials and Structures

Fig. 6 Relationships 30
between k and washout loss
values determined on UWC k(CEM) vs. W1
and CEM 25
k(UWC) vs. W3 y = 3.08 x + 7.57
R² = 0.81

Washout loss, %
20

15

10

y = x + 2.18
5
R² = 0.69

0
0 1 2 3 4 5 6 7 8 9
-5
k x 10 , cm/s

Table 4 Repeatability of permeameter versus CRD C61 tests

CEM properties UWC properties
-5
k(CEM) 9 10 (cm/s) W1 (%) k(UWC) 9 10-5 (cm/s) W3 (%)

CEM flow = 205 ± 10 mm 4.24 17.56 7.1 8.4

4.16 19.50 6.8 8.15
4.21 18.10 7.19 8.5
UWC slump = 200 ± 10 mm 4.11 17.82 7.04 8.76
4.13 19.34 6.96 8.06
Average 4.17 18.46 7.02 8.37
SD 0.05 0.90 0.15 0.28
c.o.v. (%) 1.30 4.85 2.11 3.35
CEM flow = 220 ± 15 mm 7.10 19.56 9.7 10.3
6.92 22.50 10.4 9.4
7.02 20.10 9.84 11.25
UWC slump flow = 400 9 400 ± 15 mm 7.25 21.90 11.2 10.7
7.30 23.15 10.1 12.1
Average 7.12 21.44 10.25 10.75
SD 0.16 1.55 0.60 1.01
c.o.v. (%) 2.22 7.22 5.81 9.42
CEM flow = 235 ± 15 mm 9.95 21.10 11.1 12
11.00 25.80 10.92 10.6
10.60 24.15 12.9 13.8
UWC slump flow = 550 9 550 ± 20 mm 10.41 27.20 10.86 14.2
10.30 28.00 11.5 15.4
Average 10.45 25.25 11.46 13.20
SD 0.39 2.74 0.85 1.90
c.o.v. (%) 3.70 10.86 7.38 14.37
Materials and Structures

Fig. 7 Effect of 28
consistency on k and 25.3
washout of UWC and CEM 24
k x 10-5, cm/s

UWC and CEM properties

Washout, %
20 18.5

16
13.2
11.5
12 10.5
8.4
8 7.0
4.2
4

0
205 mm 235 mm Slump = 200 Slump flow =
mm 550 mm

CEM free flow UWC workability

Fig. 8 Variations in c.o.v. 16

determined from 14.4
permeability and washout Permeability (k) values
measurements
12 Washout loss values 10.9
9.4
c.o.v., %

8 7.2 7.4
5.8
4.9
3.7 3.4
4
2.2 2.1
1.3

0
205 mm 220 mm 235 mm Slump = S. flow = S. flow =
200 mm 400 mm 550 mm

CEM free flow UWC workability

from 205 to 220 and 235 mm, respectively. The For all tested UWC and CEM, the c.o.v. determined
corresponding c.o.v. determined from W1 varied from from k values was remarkably lower than the one
4.9 to 7.2 and 10.9 %, respectively. From the other obtained using the original or adapted CRD C61 test
hand, better repeatability reflected by lower c.o.v. (Fig. 8), suggesting that the repeatability of permea-
resulted from the CEM, as compared to the corre- mater test is better than the one resulting from the
sponding concrete. The presence of coarse aggregates CRD C61. For example, for the CEM having a flow of
can act as obstacles, thus leading to variations in the 205 mm, the c.o.v. increased from 1.3 to 4.9 % when
rates of water percolations throughout the specimen determined from the k(CEM) and W1 values, respec-
during permeability testing. During washout testing, tively. Such increase was from 2.1 to 3.4 % for the
the aggregates may block the holes of the CRD C61 UWC having a slump of 200 mm. The improvement in
basket and result in reduced repeatability [20]. repeatability determined from permeability may be
 Materials and Structures

Table 5 Loss in UWC compressive strength due to washout and water infiltration
Mix f0 c(dry) Loss in strength due to Loss in strength due to water Total strength Residual
# (MPa) washout infiltration loss (%) strength (%)
W3 f0 c(W3) Df0 c(W3) k(UWC) 9 10-5 f0 c(k) Df0 c(k)
(%) (MPa) (%) (cm/s) (MPa) (%)

1 42 7.1 37 11.9 5.1 38.6 8.1 20 80

3 40.8 7.2 34.1 16.42 4.2 38.4 5.88 22.3 77.7
4 34.5 9.1 25.8 25.22 8.2 30.5 11.59 36.81 63.19
6 36.9 8.5 28.9 21.68 3.9 34 7.86 29.54 70.46
7 30.2 12.2 20.8 31.13 7.1 26.1 13.58 44.7 55.3
9 43.7 4.1 40.1 8.24 3.6 41.2 5.72 13.96 86.04
11 41.2 6.3 36.5 11.41 5.4 38.5 6.55 17.96 82.04
12 38.5 5.2 35 9.09 3.7 35.6 7.53 16.62 83.38
14 39.6 5.1 35.8 9.6 1.8 36.1 8.84 18.43 81.57
15 31.4 9 26 17.2 7.3 28.5 9.24 26.43 73.57
17 59.4 2 56.7 4.55 0.9 58.3 1.85 6.4 93.6
18 60.1 2.4 59.4 1.16 1.8 58.4 2.83 3.99 96.01
19 44.6 6.1 38.8 13 3.1 41.2 7.62 20.63 79.37
20 44 3.6 41.5 5.68 0.63 42.2 4.09 9.77 90.23
22 34.8 6.7 27.1 22.13 3.9 33 5.17 27.3 72.7
Df0 c(W3) (%) = [f0 c(dry) - f0 c(W3)/f0 c(dry)] 9 100; Df0 c(k) (%) = [f0 c(dry) - f0 c(k)/f0 c(dry)] 9 100; total strength loss (%) = Df0 c(W3) ? Df0 c(k);
residual strength (%) = 100 - total strength loss

due to the fact that the samples are well confined calculated as [(strength determined in dry condi-
within the cell and not subjected to any kind of tions) - (strength measured after washout or perme-
agitation or turbulence during testing. ability measurement)/(strength determined in dry
It is interesting to note that the difference in c.o.v. conditions)] 9 100. The total strength loss is taken
between k and washout loss values increases when the as the summation of strength losses following washout
mixture possesses higher level of consistency (Fig. 8). (Df0 c(W3)) and permeability (Df0 c(k)) measurements,
For example, such difference was limited to 1.3 % (i.e., whereas the residual strength is equal to 100 - total
3.4 - 2.1 %) for the UWC having a slump of 200 mm; strength loss.
whereas it increased to 7 % (i.e., 14.4 - 7.4 %) for the
same UWC having a slump flow of 550 mm. The 4.2.1 Effect of washout and water infiltration
significant increase in c.o.v. determined from washout
loss values can be related to the fact that such mixtures Typical variations of UWC strength loss due to
inherently flow out through the holes of the CRD C61 washout or water infiltration are plotted in Fig. 9.
basket even before any immersion in water, thus Regardless of the mixture composition, the loss in
reducing repeatability [20]. In other words, the strength for concrete sampled following washout is
permeameter responses can be considered as reason- found greater than the one determined from a mixture
ably repeatable, despite the increase in consistency. sampled following water infiltration. This indicates
that UWC strength loss is more pronounced during the
4.2 Phase II: mechanisms of UWC strength loss casting stage due to physical losses of cementitious
phase, relative increase in coarse aggregate concen-
The compressive strengths after 28 days of tested tration, together with potential infiltration of water
UWC mixtures sampled in dry conditions or following within the freshly mixed concrete during agitation.
washout and permeability measurements are summa- The reduction in strength determined on mixtures
rized in Table 5. The loss in strength (Df0 c) is sampled following permeability testing is mainly
Materials and Structures

Fig. 9 Variations in
strength loss for UWC Slump of UWC = 200 +/-10mm
sampled after washout and 30
permeability measurements

Loss in strength, %
Loss due to water infiltration
Loss due to washout
20

10

0
Cement 350 350 350 400 400 450 450
w/c 0.38 0.55 0.63 0.42 0.65 0.35 0.5
AWA - - - 0.01 - 0.02 0.08

attributed to water infiltration which may result in an example, Df0 c(W3) increased from around 5 to 20 %
increase in the specified w/c. Nevertheless, both when W3 increased from 3 to 6 %, respectively. The
mechanisms occurring during the casting and post- increase in Df0 c(k) values was from around 5 to 14 %
casting stages should be accounted for to adequately when k(UWC) increased from 2.5 to 8 9 10-5 cm/s,
quantify the drop in UWC strength. respectively. As already explained, the decrease in
As can be seen from Fig. 9, the loss in strength due strength for mixtures exhibiting increased W3 and
to washout is highly affected by the mixture compo- k(UWC) values can be attributed to a combination of
sition. For example, Df0 c(W3) increased by around factors such as washout loss and relative increase in
three-folds from 11.9 to 31.1 % for the concrete made specified w/c. The maximum strength loss determined
w/c of 0.38 and 0.63, respectively, containing 350 kg/ in this study was of 31 and 14 % for UWC sampled
m3 cement. Such variation in strength loss was limited after washout and permeability measurements, respec-
to 8.1 and 13.6 %, respectively, when the concrete is tively. It is to be noted that Df0 c(W3) can be correlated to
sampled following permeability testing. Practically Df0 c(k) with acceptable R2 of 0.62 (Fig. 11).
speaking, this indicates that the mixture composition is
a crucial factor that severely affects hardened UWC 4.2.2 Effect of hydration time on the rate of water
properties during the casting stage. Thus, mixtures infiltration
properly proportioned to exhibit high cohesiveness
levels (i.e., with relatively low w/c and/or incorporat- In order to define the time period during which water
ing AWA or silica fume) are expected to reduce UWC infiltration may adversely affect UWC strength, the
vulnerability to casting conditions. Conversely, strength variations in k values were monitored for various
loss is affected quite moderately by the mixture UWC and CEM mixtures at different elapsed times
composition following the end of casting until onset of following initial mixing [18] The specimen remained
setting. in the permeameter cell during testing, and the inlet
The relationships between W3 versus Df0 c(W3) and water valve opened to realize the measurement at
k(UWC) versus Df0 c(k) for tested mixtures are plotted in specific time intervals, and then closed back again. In
Fig. 10. The increase in W3 and k(UWC) values is parallel, the setting time was determined on a separate
shown to be well correlated to Df0 c measured follow- specimen by penetration resistance according to
ing washout and permeability testing, suggesting that ASTM C403 test method.
the magnitudes of the CRD C61 and permeameter tests Typical variations of k(CEM) and penetration resis-
are suited to reflect variations in strength loss. For tance obtained on CEM made with 350 kg/m3 cement
 Materials and Structures

Fig. 10 Relationships 15 10
between W3 versus Df0 c(W3)
y = 0.58 x -0.096
and k(UWC) versus Df0 c(k)
R² = 0.63
12 8

Washout of UWC (W3), %

k (UWC) x 10 , cm/s
9 6

-5
y = 0.307 x + 2.04
R² = 0.87
6 4

W3 vs. Loss due to washout

3 2
k vs. Loss due to permeability

0 0
0 5 10 15 20 25 30 35
Loss in compressive strength, %

Fig. 11 Relationship 35
between Df0 c(W3) and Df0 c(k)
Loss in strength due to washout, %

R² = 0.62
30

25

20

15

10

0
0 3 6 9 12 15
Loss in strength due to water infiltration, %

and 0.5 w/c are plotted in Fig. 12 (hydraulic gradient between solid particles and voids within the newly
was set at 25). As expected, the k(CEM) gradually hardened cement paste. Therefore, this indicates that
decreased as a function of time during the initial water infiltration may adversely affect UWC strength
200 min after mixing due to the loss in mixture’s as long as the mixture did not reach initial setting.
consistency and continuous change in the pore
network structure as the cement reacts with water. 4.2.3 Prediction of total strength loss due to washout
Beyond around 250 min which roughly coincides with and water infiltration
the mortar’s initial setting time, the k(CEM) dropped
sharply towards zero. Garcia et al. [21] reported that Theoretically, the total strength loss of UWC which is
initial setting time can be used to reflect a certain the summation of Df0 c(W3) and Df0 c(k) is expected to
percolation threshold which affects connectivity depend on the casting conditions and volume of
Materials and Structures

Fig. 12 Effect of hydration 5 25

time on k and penetration
resistance

Penetration resistance, MPa

4 20
Beginning of
setting time
Permeability

k x 10 , cm/s
3 15
Setting time

-5
2 10
Mortar with 350 kg/m³
cement and 0.5 w/c

1 5

Initial set at
3.5 MPa
0 0
0 50 100 150 200 250 300 350 400
Time, minutes

material to be placed. For example, the tremie/ water infiltration during the post-casting stage.
hydrovalve technique is often used for casting mass A considerable drop in total strength exceeding
concrete. The steel tremie pipes are mounted vertically 25 % is expected to occur within this category of
in water with a hopper at the top to act as a reservoir for mixtures.
the supply of fresh concrete [8]. The pipes are partially – For W3 and k(UWC) varying between 3.5 and 7 %
filled with UWC and inserted within freshly placed and 2–4 9 10-5 cm/s, respectively, the UWC
concrete to reduce turbulence and minimize washout mixtures become less vulnerable to washout loss
loss during the casting process [2]. In such situations, and water infiltration. The total strength loss
Df0 c(W3) can be neglected as the concrete does not get resulting during the casting and post-casting stages
in direct contact with surrounding water during the ranges from 10 to 25 %.
casting process, and the drop in strength can mostly be – Mixtures exhibiting W3 and k(UWC) less than 3.5 %
related to water infiltration (i.e., Df0 c(k)). From the and 2 9 10-5 cm/s, respectively, are obviously
other hand, the skip and toggle bag methods are most most recommended for underwater castings as the
suited for small pours where fresh concrete is filled mechanisms leading to strength loss are mini-
into special buckets [8]. For discharging, the bottom mized. The total drop in strength is expected to be
bucket’s door is opened to allow a free release of less than 10 %.
concrete into the surrounding water. The total strength
loss is thus expected to occur from a combination of
4.2.4 Total strength loss during casting of thin
Df0 c(W3) and Df0 c(k).
sections
Depending on the intrinsic W3 and k(UWC) deter-
mined using the CRD C61 and permeameter tests,
The drop in UWC strength is particularly important
respectively, three total strength loss zones obtained
when undertaking repair works involving the casting of
from the summation of Df0 c(W3) and Df0 c(k) are
thin underwater surfaces [22]. In such situations, the
distinguished in Fig. 13, as follows:
same UWC subjected to washout loss during placement
– UWC mixtures possessing W3 and k(UWC) greater is expected to be affected by water infiltration following
than 7 % and 4 9 10-5 cm/s, respectively, are the end of casting. To simulate such conditions, several
characterized by reduced cohesiveness which may UWC mixtures were tested using the CRD C61 test
increase tendency towards washout loss during the method, remixed in a clean container, and then used
casting stage together with increased potential for themselves for permeability testing. The so-recuperated
 Materials and Structures

Fig. 13 Effect of W3 and 14

k(UWC) on total strength loss
12
R² = 0.73

Washout loss of UWC (W3), %

For k > 4 x 10-5 cm/s and W3> 7%:
10 Total strength loss > 25%

6
For 2 < k < 4 x 10-5 cm/s and
4 3.5% < W3< 7%:
10% < Total strength loss < 25%

2
For k < 2 x 10-5 cm/s and W3< 3.5%:
Total strength loss < 10%
0
0 2 4 6 8 10
-5
k (UWC) x 10 , cm/s

Fig. 14 Total strength loss

for UWC sampled under 50
different conditions Δf'c(W3) + Δf'c(k)
Total loss in strength, %

40 Δf'c(W3+k)

30

20

10

0
Cement 350 350 400 400 450 450
w/c 0.38 0.63 0.42 0.65 0.35 0.5
AWA - - 0.01 - 0.02 0.08

fresh samples were then employed for filling the determined using both methods was less than 6 %
100 9 200 cylinder and determining the total strength (Fig. 14). The UWC made with 350 kg/m3 cement and
loss, referred to as Df0 c(W3?k). 0.63 w/c exhibited a Df0 c(W3?k) of 53.1 %, compared to
As shown in Fig. 14, the Df0 c(W3?k) tends to be 44.7 % when Df0 c(W3) and Df0 c(k) were summed
higher than the summation of Df0 c(W3) and Df0 c(k). Such together.
increase in Df0 c(W3?k) is mainly related to the fact that
the mixture’s cohesiveness used for permeability
testing has already been weakened due to washout 5 Summary and conclusions
loss, thus resulting in increased k(UWC) value, and in its
turn, increased strength loss. With the exception of one This paper demonstrated the suitability of the falling-
test, the variation between total strength losses head method realized using a soil permeameter cell to
Materials and Structures

assess water infiltration in freshly mixed UWC and 4. Assaad JJ, Issa CA (2013) Effect of washout loss on bond
mortar mixtures. This method is standardized by behavior of steel embedded in underwater concrete. ACI
Struct J (in press)
ASTM D2434 and D5084, and quite simple to be 5. Assaad JJ, Issa CA (2012) Bond strength of epoxy-coated
realized under both laboratory and field conditions. bars in underwater concrete. Constr Build Mater 12:667–674
Permeability measurements can be computed using 6. Perry SH, Holmyard JM (1992) Scaling of underwater
Darcy’s law, and are found to gradually decrease as a concrete repair materials. Technical report OTH 89-298, vol
62. Health and Safety Executive, Liverpool
function of time after mixing until the mixture reaches 7. Moon HY, Shin KJ (2006) Evaluation on steel bar corrosion
initial setting. Similarly to washout determined using embedded in antiwashout underwater concrete containing
the CRD C61 test, permeability is directly affected by mineral admixtures. Cem Concr Res 36:521–529
the mixture composition. The increase in cement 8. McLeish A (1994) Underwater concreting and repairing. In:
McLeish A (ed) ISBN 470234032, p 155. Edward Arnold,
content, decrease in w/c, and addition of silica fume London
and AWA are found to reduce k values. The perme- 9. Assaad JJ, Daou Y, Salman H (2011) Correlating washout to
ameter test offers better repeatability in testing as strength loss of underwater concrete, construction materials.
compared with the CRD C61 test, particularly for In: Proceedings of the institution of civil engineers, vol 164
(CM3). ICE, London, pp 153–162
mixtures possessing increased workability levels. 10. CRD C61-89A (1989) Test method for determining the
The drop in strength for concrete sampled follow- resistance of freshly-mixed concrete to washing out in
ing washout loss was found greater than the one water. In: Handbook for concrete. US Army Experiment
determined following water infiltration, indicating that Station, Mississippi, pp 1–3
11. Das BM (2011) Principles of geotechnical engineering, 7th
the mechanism of UWC strength loss is more edn. Cengage Learning, Toebben Drive Independence, p 651
pronounced during the casting stage. The reduction 12. Boynton SS, Daniel DE (1985) Hydraulic conductivity tests
in strength determined on mixtures sampled following on compacted clay. J Geotech Eng 111(4):465–478
permeability testing is mostly attributed to water 13. ASTM D2434 (2006) Standard test method for permeability
of granular soils (constant head). ASTM, Philadelphia
infiltration which can increase the specified w/c. The 14. ASTM D5084 (2010) Standard test methods for measure-
total strength loss of UWC depends on the casting ment of hydraulic conductivity of saturated porous materials
conditions and volume of material to be placed. Three using a flexible wall permeameter. ASTM, Philadelphia
strength zones varying from less than 10 %, between 15. Schwartzentruber A, Catherine C (2000) Method of the
concrete equivalent mortar (CEM): a new tool to design
10 and 25 %, and larger than 25 % were established concrete containing admixture. Mater Struct 33:475–482
depending on the intrinsic washout loss (W3) and 16. Assaad JJ, Harb J, Chakar E (2009) Relationships between
k(UWC) values. key astm test methods determined on concrete and concrete-
equivalent-mortar mixtures. ASTM Int J 6(3):14
Acknowledgments The authors wish to acknowledge the 17. Assaad JJ, Daou Y, Harb J (2011) Use of CEM approach to
financial support of the National Council for Scientific Research develop and optimize high-performance underwater con-
(CNRS), Lebanon and the University Research Council of the crete. J Mater Civ Eng ASCE 23(7):1094–1102
Lebanese American University (LAU), Byblos, Lebanon. 18. Assaad JJ, Harb J (2013) Use of the falling-head method to
assess permeability of freshly mixed cementitious-based
materials. J Mater Civ Eng (in press)
19. Picandet V, Rangeard D, Perrot A, Lecompte T (2011)
Permeability measurement of fresh cement paste. Cem
References Concr Res 41(3):330–338
20. Sonebi M, Bartos PJM, Khayat KH (1999) Assessment of
1. Hughes BP (1961) Development of an apparatus to deter- washout resistance of underwater concrete: a comparison
mine the resistance to separation of fresh concrete. Civil between CRD C61 and new MC-1 tests. Mater Struct 32:
Eng Public Rev 56(658):633–634 273–281
2. Sonebi M, Khayat KH (1999) Effect of water velocity on the 21. Garcia A, Castro-Fresno D, Polanco JA (2008) Evolution of
performance of underwater self-consolidating concrete. penetration resistance in fresh concrete. Cem Concr Res
ACI Mater J 96(5):519–528 38:649–659
3. Khayat KH (1998) Viscosity-enhancing admixtures for 22. Assaad JJ, Daou Y, Khayat KH (2009) Simulation of water
cement-based materials: an overview. Cem Concr Compos pressure on washout of underwater concrete repair. ACI
20:171–188 Mater J 106(6):529–536