ACI MATERIALS JOURNAL TECHNICAL PAPER

Title no. 105-M71

Performance of Slag Concrete in Marine Environment

by Michael D. A. Thomas, Allan Scott, Ted Bremner, Alain Bilodeau, and Donna Day

This paper reports the findings from an evaluation of concrete revealed some interesting trends in terms of the relative
containing ground pelletized blast-furnace slag after 25 years of performance of the concrete mixtures after 10 to 15 years of
exposure in a marine tidal zone. Twelve large concrete blocks (305 x exposure and these have been reported elsewhere.1
305 x 915 mm [1 x 1 x 3 ft]) were retrieved from the exposure site In 2003, a decision was made to retrieve one of the two
and were sampled for evaluation in the laboratory. The specimens
blocks from each of the mixtures as they reached an age of
were from three series of mixtures with a water-cementitious material
ratio (w/cm) of 0.40, 0.50, and 0.60; within each series, the 25 years for testing in the laboratory. The testing will include
concrete mixtures contained 0, 25, 45, or 65% slag by mass of total the determination of mechanical properties, depth of chloride
cementitious materials. penetration, and the measurement of various mass transport
Severe surface erosion was observed for concretes containing properties. This paper reports the results from testing the
higher levels of slag (45 or 65%) with a w/cm of 0.50 or 0.60, the blocks from Phase I, which were placed in 1978, that
extent of damage increasing with increasing slag levels. The contained up to 65% of a pelletized slag.
surface condition, however, was satisfactory for all concretes with
a w/cm of 0.40. Laboratory testing included the determination of RESEARCH SIGNIFICANCE
the depth of chloride ion penetration, compressive strength, The Treat Island exposure site provides a unique opportunity
modulus of elasticity, splitting strength, chloride permeability
to study the performance of concrete under very severe
(ASTM C1202), chloride diffusion, and hardened air-void parameters.
exposure conditions. This paper presents long-term data,
The depth of chloride penetration after 25 years of exposure was
greater than 100 mm (4 in.) for all the control concretes (that is, few of which exist already, for concrete with various levels
without slag) regardless of the w/cm. The slag concretes showed of ground pelletized blast-furnace slag. Such data are essential
significantly greater resistance to chloride ion penetration (generally for determining the real effect of supplementary cementing
<50 mm [2 in.] for 45 and 65% slag), the depth of penetration materials, such as slag, on the ability of concrete to resist the
decreasing with increasing slag content and decreasing w/cm. penetration of chloride ions. The long-term diffusion data
The beneficial effect of slag in terms of reducing mass transport from this study will serve as valuable input for emerging
was confirmed by laboratory tests with the slag concretes mathematical models for predicting the service life of
showing more than tenfold reductions in chloride permeability concrete exposed to chloride ions and freezing-and-thawing
and diffusion coefficients. conditions in service.
These data indicate that the use of slag at relatively high levels
of replacement by North American standards (for example, 45 to
65%) results in a significant increase in the performance of EXPERIMENTAL INVESTIGATION
concrete in a very aggressive marine environment provided the Twelve concrete blocks (305 x 305 x 915 mm [1 x 1 x 3 ft])
w/cm is kept low (that is, w/cm ≤ 0.40). from Phase I of the investigation were collected from the
marine exposure site at Treat Island, Maine, on August 19,
INTRODUCTION 2003. The blocks had been exposed to tidal conditions in the
In 1978, the Canadian Centre for Mineral and Energy Bay of Fundy for 25 years representing approximately
Technology (CANMET) initiated a study on the marine 18,250 cycles of wetting and drying and 2500 cycles of
performance of concrete containing supplementary freezing and thawing. The concrete specimens were produced
cementing materials (SCM), which involved placing large from three series of mixtures with different w/cm values of
concrete blocks (305 x 305 x 915 mm [1 x 1 x 3 ft]) in the 0.40, 0.50, and 0.60. At each w/cm, there were four blocks
tidal zone of the marine exposure site at Treat Island, with slag replacement levels of 0, 25, 45, and 65%. Mixture
Maine.1 Between 1978 and 1994, 16 different series of proportions for the concrete mixtures are given in a previous
concrete mixtures were placed at Treat Island; these paper.1 The slag used was a ground pelletized blast-furnace
mixtures had water-cementitious material ratios (w/cm) in slag from Hamilton, ON, Canada. The portland cement used
the range of 0.40 to 0.60 and various levels (up to 80% by met the requirements of ASTM C150 Type I. The chemical
mass of cementing material in some cases) of fly ash, slag, analyses of the cement and slag are given in Table 1. The
and silica fume. Other parameters investigated included the portland cement had a C3A content of 11.4% (Bogue
use of lightweight aggregate, fibers, epoxy-coated steel, and calculation). The specific surface area of the portland
alkali-silica reactive aggregates.1 cement and slag were 339 and 466 m2/kg, respectively. The
28-day compressive strength of standard-cured cylinders cast
The exposure conditions at Treat Island, which is situated
from the 12 mixtures are given in Table 2. The fine aggregate
at the entrance of the Bay of Fundy, are extremely aggressive
with among the highest tides in the world and an average of
100 freezing-and-thawing cycles per annum. Testing to date
has included visual examination and rating, pulse velocity, ACI Materials Journal, V. 105, No. 6, November-December 2008.
MS No. M-2008-064 received February 20, 2008, and reviewed under Institute publication
and resonance frequency of the exposed blocks and policies. Copyright © 2008, American Concrete Institute. All rights reserved, including
compressive strength and elastic modulus determinations of the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including authors’ closure, if any, will be published in the September-
cylinders under laboratory conditions. The testing has October 2009 ACI Materials Journal if the discussion is received by June 1, 2009.

628 ACI Materials Journal/November-December 2008

 were cut to produce specimens for testing rapid chloride
Michael D. A. Thomas, FACI, is a Professor of civil engineering at the University of
New Brunswick, Fredericton, NB, Canada. He is a member of ACI Committees 201, permeability, bulk diffusion, and chloride profiles. These
Durability of Concrete; 221, Aggregates; 232, Fly Ash and Natural Pozzolans in Concrete; cores were drilled along the center line of the long face of the
233, Ground Slag in Concrete; 234, Silica Fume in Concrete; 236, Material Science block such that the cut edge of the core was at least 100 mm
of Concrete; 308, Curing Concrete; 365, Service Life Prediction; and E803, Faculty
Network Coordinating Committee. (4 in.) from the edge of the block. Specimens for bulk diffusion
and RCPT were cut from the center of these cores as shown
Allan Scott is employed at the University of New Brunswick. in Fig. 2. This was done to minimize the chloride content of
ACI member Ted Bremner is employed at the University of New Brunswick. He is a the specimens used prior to testing.
member of ACI Committees 122, Thermal Properties and Performance of Concrete
and Masonry Systems; 213, Lightweight Aggregate and Concrete; 214, Evaluation of
Results of Tests Used to Determine the Strength of Concrete; 357, Offshore and Mechanical properties
Marine Concrete Structures; and 365, Service Life Prediction. Compressive strength, splitting tensile strength, and static
modulus of elasticity tests were performed on 114 mm
Alain Bilodeau is Senior Research Engineer, Recycling and Stabilization Unit, CANMET-
MMSL, Ottawa, ON, Canada His research interests include the use of supplementary
(4-1/2 in.) diameter cores. A length of approximately 38 mm
cementing materials and durability of concrete. He is a member of ACI Committee (1-1/2 in.) was cut from each end of the core to provide a
232, Fly Ash and Natural Pozzolans in Concrete.

Donna Day is employed with the U.S. Army Corp. of Engineers.

used was a natural sand and the coarse aggregate was a river
gravel with a nominal maximum size of 37.5 mm (1-1/2 in.).
Some of the blocks exhibited severe surface scaling, the
extent of damage generally increased with an increase in w/cm
or slag content. The concretes with a w/cm of 0.40 generally
performed well and visual differences between concretes
with different levels of slag were small. At a w/cm of 0.50 or
0.60, however, the slag concretes suffered more surface
scaling, especially at the highest level of replacement (65%
slag). Block B11 was split into two pieces, and this was
attributed to the action of freezing-and-thawing cycles that
opened a fine crack that was observed on the specimen at the
time of placing on the exposure site. The fine crack is
believed to have resulted from stresses during handling and
placing the specimen on the exposure. Figure 1 shows Fig. 1—Photographs of blocks at 25 years of marine exposure;
photographs of the blocks with a w/cm of 0.60. w/cm = 0.60.
Eight 114 mm (4-1/2 in.) diameter cores and two 102 mm
(4 in.) diameter cores were drilled from each block; the cores
were drilled through the block and were approximately
305 mm (12 in.) long. The larger cores were used to determine
mechanical properties, whereas the 102 mm (4 in.) cores

Table 1—Chemical analysis of portland cement

and slag
SiO2 Al2O3 Fe2O3 CaO SO3 MgO Na2Oe LOI
Type I 20.56 5.59 1.99 63.07 4.63 1.44 0.92 1.27
Slag 36.84 9.53 1.11 36.92 0.10 11.07 0.52 — Fig. 2—Sectioning of 102 mm (4.02 in.) cores for analysis.

Table 2—Specimen numbering and mechanical properties of concrete

Compressive strength, MPa (psi)
Cylinder splitting strength at Modulus of elasticity at
w/cm Slag, % Specimen no. 28-day 25-year 25 years, MPa (psi) 25 years, GPa (ksi)
0 A5 29.0 (4205) 47.4 (6873) 3.67 (532) 28.7 (4162)
25 A3 29.3 (4249) 42.3 (6134) 3.46 (502) 23.4 (3393)
0.40
45 A7 30.5 (4423) 43.4 (6293) 3.64 (528) 23.6 (3422)
65 A11 27.2 (3944) 37.2 (5394) 3.40 (493) 20.3 (2944)
0 B5 24.4 (3538) 45.7 (6627) 3.39 (492) 25.9 (3756)
25 B3 27.9 (4046) 38.5 (5583) 3.88 (563) 26.3 (3814)
0.50
45 B7 28.1 (4075) 40.1 (5815) 3.75 (544) 22.2 (3219)
65 B11 23.5 (3408) 38.6 (5597) 3.95(573) 19.8 (2871)
0 C5 26.5 (3843) 37.3 (5409) 2.76 (400) 22.4 (3248)
25 C3 24.8 (3596) 38.6 (5597) 3.46 (502) 21.6 (3132)
0.60
45 C7 25.5 (3698) 32.3 (4684) 2.45 (355) 21.5 (3118)
65 C11 19.8 (2871) 32.9 (4771) 2.96 (429) 18.3 (2654)

ACI Materials Journal/November-December 2008 629

specimen with an aspect ratio of 2. Three cores were used for The cores used for evaluating mechanical properties were
each of the compressive strength and modulus of elasticity cut close to the exposed edges of the block such that in some
tests, and two cores for the splitting tensile strength tests. places, the curved surface of the core was within 12 mm (1/2 in.)
The test results are shown in Table 2. of the exposed surface of the block. If deterioration due to
The compressive strength of the 25-year-old cores generally either freezing-and-thawing cycling or chemical attack by
decreased with increasing w/cm or slag content of the concrete the constituents of the seawater, or both, occurred during
mixture. On average, the strength of the cores at 25 years was exposure, it is likely that such deterioration would not occur
51% higher than the strength of 28-day-old cylinders from uniformly across the cross-section of the core. Such hetero-
the same mixture. There was no consistent trend between the geneity is likely to influence the measured mechanical
ratio of 25-year to 28-day strengths and the slag content of properties. The lower modulus of elasticity exhibited by the
the mixture, however. These data are not generally consistent concrete containing slag and the lower-than-expected
with strength data for concrete containing slag. At a given w/cm, strength gain observed for these concretes indicates that
slag would be expected to contribute significantly to the there is likely greater microstructural damage in the concrete
long-term strength of the concrete and both the 25-year containing slag; this is consistent with the visual condition of
strengths and the ratio of 25-year to 28-day strength would the blocks.
be expected to increase with slag content.2
The modulus of elasticity of the 25-year-old cores also Air-void analysis
showed a significant decrease with increasing slag content. Harden air-void analysis was determined by examination
The measured modulus of the cores was, on average, 23% of polished specimens in accordance with ASTM C457. The
lower than the modulus predicted from the compressive original target for the air content was 6% ± 1%. The results
strength of the cores using the following relationship from of the air-void analysis on the fresh concrete showed that all
the ACI Building Code3 the specimens were within this range.1 The results from the
hardened air-void analysis reveal that all but two concrete
mixtures achieved the desired level of air. The concrete with
E c = 4731 f c ′ MPa (1a) w/cm = 0.40 and 45% slag had an air content of 3.4% and the
concrete with w/cm = 0.50 and 65% slag had an air content
of 4.7%. These concretes did, however, show satisfactory
E c = 57, 000 f c ′ psi (1b) spacing factors of less than 0.11 mm (0.004 in.) and none of
the concretes had a spacing factor greater than 0.23 mm
In the authors’ experience, this is indicative of microstructural (0.009 in.) which is the maximum average spacing factor
damage, the stiffness being more sensitive to microcracking permitted by Canadian specifications for this exposure
than the compressive strength. The splitting tensile strengths condition (CSA A23.1-04).
of cores show a decreasing strength with an increasing w/cm,
but there is no discernable trend between strength and slag Rapid chloride permeability testing
content for a given w/cm. The rapid chloride permeability test (RCPT) (ASTM
C1202) can be used to give a general indication of the ability
of the concrete to resist the movement of chlorides, although
the test really measures the electrical conductivity of the
concrete. The results of the RCPT are provided in Table 3.
The impact of both w/cm and slag content are evident in
these data, although clearly the level of slag has a much more
dominant effect.
The concretes without slag all had high chloride ion
penetrability according to the classification system in
ASTM C1202. Increasing the w/cm to 0.60 results in a
substantial increase in the penetrability of the concrete
without slag. All of the slag concretes had RCPT values
below those of the concrete without slag regardless of w/cm.
Very significant reductions in the RCPT value are observed
with increasing slag level and, at the higher levels of slag, the
impact of w/cm is small within the range studied. On average,
concretes with 25, 45, and 65% slag had RCPT values that
Fig. 3—Chloride profile for concrete at w/cm = 0.40 after were 1/3, 1/5, and 1/10 of the values for concrete without slag.
25 years of marine exposure.
Chloride profiling and bulk diffusion
Table 3—Results of rapid chloride permeability The chloride profiles of the specimens after 25 years of
tests, coulombs exposure to marine conditions were determined by slicing a
Slag level, % centrally located 102 mm (4 in.) core at 10 mm (0.4 in.)
w/cm 0 25 45 65
intervals. The disks were then ground to a powder and
analyzed to determine the total chloride concentration. The
0.40 4096 2065 825 621
chloride profile expressed as a percentage of total chlorides
0.50 4203 2112 1190 513
by mass of concrete is shown in Fig. 3 through 5.
0.60 >12400* 3615 1562 770 It should be noted that some of the blocks showed severe
*
Estimated from 5-hour data due to excessive temperature rise during test. scaling, resulting in a loss of the mortar at the surface. In such

630 ACI Materials Journal/November-December 2008

cases, zero depth was established as the outermost surface of time in seconds; and Da is the apparent diffusion coefficient
the remaining coarse aggregate. This may produce an error in m2/second.
of 1 to 2 mm (0.04 to 0.08 in.) for the severely scaled samples. Best-fit values of Da and Cs were found by curve-fitting
It can be seen from Fig. 3 to 5 that the inclusion of slag, in Eq. (2) to the chloride profiles using the least-squares
increasing proportions, generally results in a substantial approach and assuming the initial chloride content of the
reduction in the chloride concentration at any given depth. concrete prior to exposure is zero, that is, Cb = 0 in Eq. (2).
Comparing concrete with the same levels of slag, it can be The calculated values of Da and Cs are given in Table 4 and
seen that the chloride concentration at the surface is greater diffusion coefficients are plotted in Fig. 6. Decreasing the w/cm
for concrete with a low w/cm, but the chloride concentration and increasing the slag content results in significant reductions
at depth is lower for concrete with a low w/cm. The higher in the calculated diffusivity. Averaging the data for all mixtures
chloride concentration at the surface of the concrete with at the same w/cm, it can be seen that reducing the w/cm from
a lower w/cm is likely the result of increased binding in the 0.60 to 0.40 reduces the diffusivity by almost four times. For
concrete with the higher cement contents and also the concrete with w/cm = 0.40, the diffusivity is reduced by
increased loss of cement paste due to scaling in the concrete
approximately 6, 14, and 18 times compared with the control
with a higher w/cm.
when the cement is replaced by 25, 45, and 65% slag, respectively.
If it is assumed that the predominant mass transport
process is ionic diffusion and that the mass transport properties Calculating diffusion coefficients from chloride profiles
remain constant with time, the following numerical solution obtained from structures exposed to marine conditions or
to Fick’s second law can be used to calculate diffusion deicing salts is a commonly used approach. Such calculations,
coefficients from the chloride profiles in Fig. 3 through 5 however, are based on the assumption that the diffusion
coefficient remains unchanged with time and this is not the case
especially for concrete containing slag or fly ash.4,5 Thus, the
Cx – Cb
------------------ = 1 – erf ⎛ ----------------⎞
x values calculated from the 25-year-old chloride profiles
(2)
Cs – Cb ⎝ 2 D t⎠ should not be seen as a true representation of the diffusion
a
coefficient of these concretes and the values are presented
where Cx is the chloride concentration at depth x and time t,
%; Cs is the chloride concentration at the surface, %; Cb is
the background chloride content, %; x is the depth, m; t is the

Fig. 6—Chloride diffusion coefficients calculated from chloride

profiles after 25 years of exposure.

Table 4—Diffusion coefficients and surface

concentrations after 25 years of marine exposure
and after 98 days of ponding (bulk diffusion test)
Fig. 4—Chloride profile for concrete at w/cm = 0.50 after Values from 25-year Values from 98-day
25 years of marine exposure. chloride profiles bulk diffusion test

Slag, Cs, Da, 10–12 m2/s C s, Da, 10–12 m2/s

w/cm 2
% % concrete (in. /year) % concrete (in.2/year)
0 0.63 4.34 (0.212) 1.12 11.3 (0.552)
25 1.03 0.69 (0.034) 1.26 3.37 (0.165)
0.40
45 1.00 0.32 (0.016) 1.03 1.57 (0.077)
65 1.02 0.24 (0.012) 1.58 0.61 (0.030)
0 0.48* 20.4* (0.997) 1.12 39.0 (1.906)
25 0.63 1.46 (0.071) 0.98 5.76 (0.282)
0.50
45 0.73 0.75 (0.037) 1.19 2.75 (0.134)
65 0.68 0.46 (0.022) 1.06 1.47 (0.072)
0 * * No fit
0.58 15.6 (0.763)

0.60
25 0.50 2.42 (0.118) 0.98* 71.1* (3.475)
45 0.66 1.34 (0.065) 1.29 4.12 (0.201)
Fig. 5—Chloride profile for concrete at w/cm = 0.60 after 65 0.52 0.89 (0.044) 1.52 2.10 (0.103)
*Low (r2
25 years of marine exposure. regression coefficient < 0.90).

ACI Materials Journal/November-December 2008 631

merely to allow a comparison to be made between the Samples for bulk diffusion tests were immersed in a solution
different concretes. of sodium chloride (2.8 Molar) for a period of 98 days prior
Because it is the chloride content at the depth of the steel to profiling. Profile grinding of the specimens was conducted
that is the main concern, it is interesting to compare the chloride using 1 mm (0.04 in.) depth increments. The samples were
concentration in the different concretes at depths where steel then analyzed for total chloride concentration. In this case,
reinforcement is likely to be encountered. For convenience, the background chloride concentration of the concrete
the chloride content in the depth increment from 60 to 70 mm specimens prior to immersion in the chloride solution could
not be assumed to be zero due to previous exposure of the
(2.4 to 2.8 in.) was used for this purpose as it is consistent
concrete to the marine environment. Best-fit values of Da,
with the minimum depth of clear cover of 60 mm (2.4 in.) Cs, and Cb were found by curve-fitting Eq. (2) to the chloride
specified by the Canadian Standards Association, CSA profiles using the least-squares approach. The bulk diffusion
A23.1-04, “Concrete Materials and Methods of Concrete coefficients are presented in Table 4.
Construction/Methods of Test and Standard Practices for Figure 8 shows how the diffusion coefficient changes with
Concrete,” for reinforced concrete exposed to chlorides. both w/cm and slag content. It was not possible to fit Eq. (2)
These data are shown in Fig. 7. The chloride concentration at to the bulk-diffusion data for the mixtures with w/cm = 0.60
this depth in the plain PC concrete (without slag) is above that contained zero or 25% slag as the data did not produce
0.25% regardless of the w/cm and thus significantly greater profiles consistent with the diffusion process. This is probably
than the commonly used chloride threshold value, Ct = due to high chloride concentrations present in the samples
0.05%, above which it is assumed that steel corrosion is prior to test. For the control mixtures without slag, reducing
initiated. Consequently, these concretes cannot be the w/cm from 0.50 to 0.40 results in a three-times reduction
assumed to provide adequate protection to embedded steel in the diffusion coefficient. Although the magnitude of these
(if it were present). At w/cm = 0.40, the threshold value has changes is significant, the impact of slag is even greater. For
not been reached at a depth of 60 to 70 mm (2.4 to 2.8 in.) in concrete with w/cm = 0.50, adding 25, 45, or 65% slag
any of the slag concretes. At w/cm = 0.50, concrete with 45 reduces the diffusion coefficient by approximately 7, 14, or
or 65% still meets this condition and at w/cm = 0.60, the 26 times, respectively. For concrete with w/cm = 0.40
concrete with 65% slag would still provide protection to (maximum w/cm specified by ACI 318-08 and CSA A23.1-04
for reinforced concrete exposed to chlorides), the diffusion
steel with this cover after 25 years of marine exposure.
coefficient at 25 years was reduced by approximately 3, 7, or
19 times by the addition of 25, 45, or 65% slag.

Comparison of data with service-life model

The chloride profile results obtained from the study of
25-year-old concrete continuously exposed to a marine
environment provides a useful opportunity for the validation
of existing service-life models. The data for the concrete
mixtures with w/cm = 0.40 were compared with predicted
output from a service-life model.6 This model accounts for
the time-dependent reduction in the diffusion coefficient of
concrete by using the following relationship

28 m 28 m
D t = D 28 ⎛ ------⎞ + D ult ⎛ 1 – ⎛ ------⎞ ⎞ (3)
⎝ t⎠ ⎝ ⎝ t⎠ ⎠

Fig. 7—Chloride concentration at 60 to 70 mm (2.4 to 2.8 in.)

after 25 years (also shown is chloride threshold value, where Dt and D28 are the diffusion coefficients at time t and
Ct = 0.050%). 28 days, respectively, and Dult is the ultimate diffusion
coefficient when the degree of hydrations reaches 100%.
Values for D28 and m are estimated by the model based on
the mixture proportions (w/cm and quantity of pozzolans or
slag). The values selected by the model for these concrete
mixtures are provided in Table 5. Dult is assumed to be the
diffusion coefficient at 100 years and is estimated from the
following equation (the Dult values for the different mixtures
are shown in Table 5)

Table 5—Values selected by service life model for

concrete with w/cm = 0.40
Slag, % D28, 10–12 m2/s (in.2/y) Dult, 10–12 m2/s (in.2/y) m
0 9.10 (0.445) 1.41 (0.0689) 0.260
25 9.10 (0.445) 0.506 (0.0247) 0.403
Fig. 8—Chloride diffusion coefficients calculated from bulk 45 9.10 (0.445) 0.223 (0.0109) 0.517
diffusion tests conducted on 25-year-old concrete samples. 65 9.10 (0.445) 0.123 (0.00601) 0.600

632 ACI Materials Journal/November-December 2008

 m properties, the depth of cover, the exposure (surface concen-
D ult = D 28 ⎛ ------------------⎞
28
(4) tration), and the chloride threshold level required to initiate
⎝ 36,500 ⎠
corrosion Ct. The service-life model6 assumes a value of Ct =
0.05% (by mass of concrete), however, significantly higher
The early-age diffusion coefficient D28 is determined from values have been reported in the literature.7 Figure 10 shows
the w/cm and is unaffected by the level of slag in the the predicted time to corrosion, for reinforced concrete
concrete. The slag influences the value of m, which controls produced with these different concrete mixtures and with
the rate of decrease of the diffusion coefficient with time. either 50 or 75 mm (2 or 3 in.) of cover to the steel. Also
The value of m is given by the equation shown is the influence of the chloride threshold value Ct
assumed in the analysis. The predictions were made on the
m = 0.26 + 0.40 (FA/50 + SG/70) (5) basis of Eq. (3) to (5), using the values given in Table 5 for
D28, Dult, and m, and the average monthly temperatures in
where FA and SG are the levels of fly ash (zero in this case) the service-life model6 for the location closest to Treat Island
and slag used in the concrete expressed as a percentage of the (Portland, Maine). The use of slag clearly leads to significant
total cementitious material. The service-life model6 does not increases in the time to corrosion and, hence, the service life
allow the value of m to exceed 0.60. of reinforced concrete; and the difference attributed to the
The output from the service-life model6 is compared with use of slag is more apparent where increased depths of cover
the data collected from the blocks in Fig. 9. Generally, the are used or where an increased corrosion threshold is
predicted profile from the model is reasonably close to the assumed. For example, assuming a depth of cover of 75 mm
actual profile determined on cores cut from the blocks. This (3 in.) and a chloride threshold of Ct = 0.10%, the predicted
is quite remarkable considering that the only inputs required time to corrosion for concrete with 0, 25, 45, and 65% slag is
for the model were w/cm, slag percentage, and selection of approximately 15, 30, 75, and 150 years, respectively.
the location (Eastport) and exposure (marine tidal). The
model assumes a constant surface concentration of Cs = SUMMARY
0.80% for a tidal exposure. The results from this study clearly show the importance of
Service life models can also be used to predict future chloride achieving a low w/cm for concrete exposed to freezing and
profiles for the blocks or to determine what the time to corrosion thawing in a marine environment, especially when slag is
would be for reinforced concrete produced with the same used. Indeed, the influence of w/cm increases as the slag
materials and exposed to the same conditions. The time to content increases and for concrete with 45 and 65% slag,
corrosion, or initiation period, ti is a function of the concrete severe scaling occurred at w/cm = 0.50 and 0.60. Concrete

Fig. 9—Chloride profiles predicted by service life model6 Fig. 10—Effect of slag, depth of cover, and chloride threshold
versus experimental data for 25-year-old concrete blocks value Ct on predicted times to corrosion for concrete (w/cm =
with w/cm = 0.40. 0.40) in marine tidal zone.

ACI Materials Journal/November-December 2008 633

with w/cm = 0.40 generally performed well with regard to paid to the w/cm, however, if the slag concrete is likely to be
scaling even at the higher levels of slag. exposed to freezing and thawing.
Concretes with a lower w/cm and higher levels of slag
showed reduced chloride penetration after 25 years of exposure REFERENCES
to the tidal conditions of the Treat Island marine exposure 1. Malhotra, V. M., and Bremner, T. W., “Performance of Concrete at
site. Although the impact of the w/cm was significant, the Treat Island, U.S.A.: CANMET Investigations,” Proceedings of the Third
level of slag had a much greater influence on the chloride ion CANMET/ACI International Conference on Performance of Concrete in
Marine Environment, SP-163, V. M. Malhotra, ed., American Concrete
penetration. For concrete with w/cm = 0.40, the level of chloride Institute, Farmington Hills, MI, 1996, pp. 1-52.
at a depth of 60 to 70 mm (2.4 to 2.8 in.) (typical specified 2. ACI Committee 233, “Slag Cement in Concrete and Mortar (ACI
cover for marine structures) in the plain concrete without 233R-03),” American Concrete Institute, Farmington Hills, MI, 2003, 19 pp.
slag was more than five times the 0.05% value often used as 3. ACI Committee 318, “Building Code Requirements for Structural
the threshold required to initiate corrosion. The chloride Concrete (ACI 318-02) and Commentary (318R-02),” American Concrete
concentrations at this depth in the slag concretes was only 1/10 Institute, Farmington Hills, MI, 2002, 443 pp.
or less of the threshold value and only slightly above the 4. Thomas, M. D. A., and Bamforth, P. B., “Modelling Chloride Diffusion
in Concrete: Effect of Fly Ash and Slag,” Cement and Concrete Research,
background (initial) chloride concentration. V. 29, 1999, pp. 487-495.
The bulk diffusion test data confirmed the impact of slag 5. Thomas, M. D. A., and Matthews, J. D., “Performance of PFA
on the chloride resistance. The diffusion coefficients were Concrete in a Marine Environment—10 Year Results,” Cement and
between 3 and 19 times lower due to the presence of slag in Concrete Composites, V. 26, No. 1, 2004, pp. 5-20.
concretes with w/cm = 0.40. 6. Riding, K., “Early Age Concrete Thermal Stress Measurement and
Modeling,” PhD thesis, University of Texas at Austin, Austin, TX, 2007.
The data indicate that considerable extensions in the 7. Glass, G. K., and Buenfeld, N. R., “Chloride Threshold Levels for
service life of marine-exposed concrete can be achieved by Corrosion Induced Deterioration of Steel in Concrete,” Chloride Penetration
incorporating slag in the concrete. Due attention must be into Concrete, L.-O. Nilsson and J. Ollivier, eds., 1995, pp. 429-440.

634 ACI Materials Journal/November-December 2008