Construction and Building Materials 162 (2018) 253–264

Contents lists available at ScienceDirect

Construction and Building Materials

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

Characterization of concrete affected by delayed ettringite formation

using the stiffness damage test
Eric R. Giannini a,⇑, Leandro F.M. Sanchez b, Atolo Tuinukuafe c, Kevin J. Folliard d
a
RJ Lee Group, 350 Hochberg Road, Monroeville, PA 15146, USA
b
Department of Civil Engineering, University of Ottawa, 161 Louis-Pasteur, K1N 6N5 Ottawa, ON, Canada
c
Department of Civil, Construction and Environmental Engineering, The University of Alabama, Box 870205, Tuscaloosa, AL 35487-0205, USA
d
Department of Civil, Architectural and Environmental Engineering, The University of Texas at Austin, 301 E Dean Keeton St, Stop C1700, Austin, TX 78712-0273, USA

h i g h l i g h t s

The Stiffness Damage Test (SDT) can reliably assess DEF-damaged concrete.

SDI and PDI indices have a near-linear correlation to DEF expansions up to 0.40%.

SDT may characterize DEF damage more effectively than ASR damage.

The statistical significance of SDT analyses by SDI and PDI is confirmed by ANOVA.

Future work should consider combined ASR and DEF damage, and correlating SDT to DRI.

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

Article history: The stiffness damage test (SDT) may provide more information than conventional mechanical tests used
Received 28 March 2017 to assess concrete affected by expansive reactions. A new approach to analyzing SDT data involving the
Received in revised form 14 October 2017 calculation of stiffness damage and plastic damage indices (SDI and PDI) has successfully characterized
Accepted 2 December 2017
the expansion of concrete affected by alkali-silica reaction (ASR). This study is the first to implement
Available online 15 December 2017
SDI and PDI to characterize concrete affected by delayed ettringite formation (DEF). The SDI and PDI
parameters have a nearly linear relationship to expansions under 0.40%. ANOVA confirms the statistical
Keywords:
significance of the SDI and PDI parameters and that they provide information not available from standard
Concrete
Stiffness damage
elastic modulus tests.
DEF Ó 2017 Elsevier Ltd. All rights reserved.
Delayed ettringite formation
ASR
Damage characterization

1. Introduction cient load-carrying capacity. Compressive strength, though an

Extraction and testing of core samples from concrete structures
is extremely useful in the diagnosis and evaluation of structures
affected by durability-related problems, such as alkali-aggregate
reactions (AAR), cyclic freezing and thawing damage, and delayed
ettringite formation (DEF). Core sampling is akin to taking a biopsy
in medicine – an invasive procedure that generally does little harm
to the structure. Cores are typically subjected to mechanical tests
and petrographic examination to determine the type and severity
of the distress, and to determine whether the structure has suffi-
⇑ Corresponding author.
E-mail addresses: EGiannini@rjleegroup.com (E.R. Giannini), Leandro.Sanchez@
uottawa.ca (L.F.M. Sanchez), atuinukuafe@crimson.ua.edu (A. Tuinukuafe), folliard@
mail.utexas.edu (K.J. Folliard).
important indicator of concrete quality and a critical input for
structural analyses, is not affected as much as stiffness, tensile
strength, or flexural strength in concrete suffering from internal
swelling reaction mechanisms (ISR) such as AAR and DEF [1–3].
Compressive strength testing is also destructive to the specimen
and precludes also performing petrographic examination of the
same specimen. The stiffness damage test (SDT), however, was
found to provide useful information on the degree of damage in
concrete, such as reduction in stiffness and extent of microcracking
damage; it is thus considered a more complete technique than a
simple elastic modulus test [4,5]. Moreover, it has been found that
when loads of 40% or less of the compressive strength of the con-
crete area used, the SDT remains a non-destructive test of the con-
crete, even at very high levels of deterioration from ISRs [5]. Hence,
the SDT can be coupled with further chemical or microscopic

https://doi.org/10.1016/j.conbuildmat.2017.12.012
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
254 E.R. Giannini et al. / Construction and Building Materials 162 (2018) 253–264
evaluations performed on the same specimens, to obtain maxi-
mum information about the condition of the structure [4–11]. It
is worth noting that the non-destructive character of the SDT is
an asset, although the most important aspect of this mechanical
testing procedure is its ability to quantify the amount of inner dis-
tress of damaged concrete, which requires the application of loads
of at least 30% (ideally 40%) of the compressive strength of the
concrete.
SDT attempts to quantify damage to concrete from AAR and
other mechanisms by analyzing stress vs. strain data obtained
when cyclic loads are applied to core samples. Several versions of
the test and associated data analyses have been proposed, but
the basic principle remains the same; compared to sound concrete,
damaged concrete will have a reduced stiffness and will accumu-
late more plastic strain and thus dissipate more energy during
the test. While the use of SDT as a tool for characterizing the effects
of AAR on concrete structures has been widely-described and
reported in the literature [5,12–17] and most extensively for
alkali-silica reaction (ASR), there are only limited reports of its
use for characterizing DEF-affected concrete [4,18]. Recent
research on the use of SDT for ASR-damaged concrete has resulted
in new methods of data analysis [5,8,10,11], which are now applied
to prior SDT data generated for DEF-affected concrete in this paper.
DEF is a form of internal sulfate attack in concrete, driven by
curing temperatures in excess of 65–70 °C and unfavorable cement
chemistry [19–21]. Cement hydration and formation of C-S-H is
greatly accelerated with increased curing temperatures. With sus-
tained temperatures above 65 to 70 °C, ettringite becomes thermo-
dynamically unstable; hydration reactions are unable to form
ettringite, and previously-formed ettringite decomposes and
returns to solution [22]. The rapidly-growing ‘‘inner” C-S-H traps
dissolved sulfates and alumina before they can react to form ettrin-
gite [20,21,23]. After temperatures decrease to levels more com-
monly experienced by concrete in service, thermodynamics again
favor the formation of ettringite. Trapped sulfates and alumina
may be released from the C-S-H and react with water and mono-
sulfate to form ettringite; this can lead to deleterious expansion
and cracking of the concrete [20]. Similar to AAR, the microscale
characteristics of DEF include extensive microcracking. In contrast
to AAR (especially ASR), where microcracking extends through
both the paste and aggregate particles, microcracking from DEF is
primarily limited to the cement paste, which expands away from
the aggregate particles. The resulting gaps and microcracks are
often filled with recrystallized ettringite deposits instead of ASR
gel. More importantly, because DEF damage development is not
intrinsically linked to the mineralogy of the aggregates or the dis-
tribution of reactive phases, as is the case in ASR [24], it is possible
that SDT could be more generally useful as a tool to diagnose and
characterize the extent of damage from DEF.
2. Background on SDT
The stiffness damage test (SDT) was originally developed by
Crouch and Wood [12] and involved applying five cycles of 5.5
MPa compressive load to a core specimen and measuring its
stress–strain response. In ASR-damaged concrete, the elastic mod-
ulus decreases, while the stress–strain hysteresis loops increase in
size, and increasing amounts of plastic strain accumulate during
the course of the test [12,13]. Chrisp et al. [13] placed emphasis
on the elastic modulus, plastic strain and size of the hysteresis
loops of the second through fifth cycles and largely discarded the
data from the first load cycle, primarily because they believed that
the first cycle induced new damage in the concrete. They also pro-
posed calculating a non-linearity index (NLI), defined as the secant
modulus at half the maximum load divided by the secant modulus
at the maximum load; this was thought to account for the orienta-
tion of cracking in damaged concrete.
Further development of the test method by Smaoui et al. [14]
resulted in a recommended loading level of 10 MPa and identified
the area of the first hysteresis loop and the accumulated plastic
strain over all five cycles as the most important parameters. They
proposed that a linear relationship between these parameters
and ASR expansion could be established using laboratory speci-
mens or core samples extracted from larger specimens of known
expansion levels [14,15]. They also noted that concrete made with
different reactive aggregates will exhibit varying responses in the
stiffness damage test; that is, linear relationships must be estab-
lished for multiple reactive aggregate types in order to estimate
the expansion of a variety of field structures. Fig. 1 shows typical
stress–strain data obtained using this version of the test for an
undamaged sample and one damaged by ASR.
The 10 MPa version of the test has been adopted by the US Fed-
eral Highway Administration [25]. Several papers describe the
application of this version of the SDT to field structures damaged
by ASR [26–29], but no studies have been published regarding
the use of SDT to characterize damage from DEF or a combination
of ASR and DEF using the updated procedure and analysis devel-
oped by Smaoui et al.
More recent works by Sanchez et al. [5,8,10,11,30] suggest that
the maximum applied stress in the test should not be a fixed value
because useful data can only be obtained if it is at least 30% of the
compressive strength. If the load is less than 30% of the compres-
sive strength, which would be greater than 10 MPa for concrete
stronger than 33.3 MPa, and greater than 5.5 MPa for all but the
weakest concretes, then it is very difficult to quantify ASR damage
with SDT. As a result, Sanchez et al. [10,11] recommended conduct-
ing the test at 40% of the current strength, which would be the
same load specified for the determination of the static modulus
of elasticity by ASTM C469 [31].
3. Scope and research significance
This paper presents a study of the use of SDT to characterize
two sets of concrete cylinders, made with different aggregates
and conditioned to increasing levels of expansion and damage
from DEF. The results of standard secant modulus of elasticity tests
and compressive strength tests on the same specimens are also
presented and analyzed for their utility in characterize the pro-
gress of damage. The notable improvement over earlier work to
characterize DEF-affected concrete is the use of analytical tools
developed by Sanchez and coworkers that seek to correlate SDT
damage indices to petrographic damage features quantified
through the Damage Rating Index (DRI) [5–11,32]. The SDT data
are also analyzed using tools previously developed by Smaoui
et al. [14]. Finally, ANOVA analysis of the data is presented to iden-
tify the most significant variables influencing the test results.
4. Materials and methods
Two sets of 100  200 mm cylinder specimens were fabricated for this study.
The specimens made use of the same concrete mixtures used by Giannini and Fol-
liard [16] in a parallel study on the mechanical properties of ASR-affected concrete;
however, the conditioning regime (described in Section 4.2) was designed to limit
the development of ASR and promote the development of DEF.
4.1. Materials and mixture proportions
Table 1 presents the aggregates used in this study, and Table 2 shows the pro-
portions for the two concrete mixtures. Mixture 1 contained aggregates C1 and F1,
and Mixture 2 contained aggregates C2 and F2. Aggregates C1 and F2 are considered
non-reactive with regards to ASR; the 1-year expansion in ASTM C1293 testing for
these two aggregates tested together is 0.01% [20]. When tested per ASTM C1293,
reactive aggregates C2 and F1 have been documented to exhibit 1-year expansions
 E.R. Giannini et al. / Construction and Building Materials 162 (2018) 253–264 255

12 12

10
Expansion 10
Expansion
0.009% 0.272%
8 8

Stress (MPa)
Stress (MPa)
6 6

4 4

2 2

0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
Axial Strain ( m/m) Axial Strain ( m/m)
Fig. 1. Typical stiffness damage test data for undamaged (left) and ASR-damaged (right) concrete [4]. Test was conducted per the procedure recommended by Smaoui et al.
[14].

Table 1
Details of aggregates used in the study.

Aggregate Source Mineralogy and Type Oven-dry SG Absorption (%)

C1 (coarse) San Antonio, Texas Crushed limestone 2.50 2.72
C2 (coarse) Bernalillo, New Mexico Partially crushed gravel with rhyolite and mixed volcanics 2.56 1.47
F1 (fine) El Paso, Texas Natural sand with quartz, feldspars, siliceous volcanics, and chert 2.59 0.69
F2 (fine) San Antonio, Texas Manufactured limestone sand 2.42 3.77

Table 2
Mixture proportions. Mixture 1 used aggregates C1 and F1 and Mixture 2 used
aggregates C2 and F2. Aggregate quantities are in the saturated surface-dry condition.

Quantity, kg/m3
Component Mixture 1 Mixture 2
Coarse aggregate 1107 1161
Fine aggregate 626 561
Cement 420 420
Water (w/c = 0.42) 176 176
NaOH 1.90 1.90

of 0.16% and 0.59%, respectively [20]. An ASTM C150 Type I Portland cement with an
equivalent alkali content of 0.90% Na2Oeq was used for both mixtures, and the alkali
content of the mixture was boosted to 1.25% by mass of cement through the addi-
tion of 50% w/w NaOH solution to the mixing water. This enabled the fabrication of
ASR- and DEF-affected concrete from a single mixture.
4.2. Specimen preparation
Concrete was mixed per ASTM C192 and a total of 45 cylinders were cast (24 for
Mixture 1 and 21 for Mixture 2). Three cylinders from each mixture were tested for
28-day compressive strength, and three cylinders were instrumented as reference
specimens for expansion measurements. The remaining 18 cylinders from Mixture
1 and 15 cylinders from Mixture 2 were not instrumented; their expansions were
inferred based on the average expansions of the reference cylinders for each mix-
ture. Including the three reference cylinders for each mixture, three cylinders each
were tested at seven levels of expansion for Mixture 1 and six levels of expansion
for Mixture 2.
To promote development of DEF, the cylinder molds were sealed and the spec-
imens subjected to a high-temperature curing regime developed by Kelham [19]
and illustrated in Fig. 2. Specimens were placed in a ramping oven programed to
increase the air temperature from 23 °C to 90 °C over a period of 4 h, hold at 90
°C for 12 h and decrease to 23 °C over a period of 4 h. The oven was not moisturized;
the seal on the molds was considered to be adequate to minimize drying during
high-temperature curing. Although the internal temperature of the specimens
was not monitored during this process, upon removal from the oven, the exterior
surface temperature of the cylinder molds was measured to be 41 °C, which sug-
gests that the specimen temperatures lagged the air temperature profile somewhat
during the ramping periods. Given that the air temperature was held at 90 °C for 12
h, it is assumed that the specimens reached and maintained that temperature for
much of that time period. Specimens were demolded and moist-cured at 23 °C
for an additional 24 h and then stored in saturated limewater at 23 °C. The limewa-
ter promotes leaching of alkalis from the cylinders, thereby suppressing the devel-
opment of ASR [20,33]. Therefore, despite using a highly-alkaline mixture and
alkali-silica reactive aggregates (C2 and F1), the measured expansion should be
attributable primarily, if not almost entirely to DEF, and not ASR.
Fig. 2. Curing temperature profile to promote development of DEF.
The three cylinders from each mixture designated for expansion measurements
were instrumented immediately following demolding, after completing the 24-h
high-temperature curing cycle. A drill press was used to drill out a small opening
on each end, and 6.4-mm- diameter stainless steel gauge studs were inserted and
bonded with epoxy; while the epoxy hardened, the specimens were wrapped in
plastic to avoid moisture loss.
The change in length of the instrumented cylinders was measured in reference
to a 173-mm- long Invar bar using a digital comparator with 0.002 mm resolution.
Expansions were measured periodically and the average expansions for each set of
instrumented specimens were assumed to be representative of the average expan-
sions of the non-instrumented specimens. When the expansions of the reference
specimens reached a target level (based on intervals given below), a set of three
non-instrumented cylinders was subjected to the SDT (followed by a standard
secant modulus and compressive strength test). An initial set of specimens for each
mixture was tested at seven days of age to provide baseline (undamaged) data; sub-
sequent sets would be tested as close as possible to expansion intervals of 0.10% for
Mixture 1 (until approximately 0.50%) and 0.08% for Mixture 2 (until approximately
0.32%), and the final, instrumented set was tested at the end of the experimental
program.
In preparation for the mechanical tests (including SDT), the ends of the speci-
mens were ground plane and parallel using a cylinder end-grinder on the day the
tests were conducted. After grinding the ends and between each test, the specimens
were stored in water to maintain them in a moisture condition in accordance with
ASTM C469 [31] and ASTM C39 [34].
4.3. Stiffness damage test
At each expansion level, three specimens were subjected to the SDT, following
the procedure first outlined by Smaoui et al. [14]. Specimens were placed in a com-
pressometer as described by ASTM C469 [31] and subjected to five load cycles using
256 E.R. Giannini et al. / Construction and Building Materials 162 (2018) 253–264

a closed-loop hydraulic testing machine with a 100 kN capacity as shown in Fig. 3.

Because the platens installed on the load frame were fixed, a spherically-seated pla-
ten was inserted on top of the specimen. A loading and unloading rate of 0.1 MPa/s
was used, with a peak compressive stress of 10 MPa. An LVDT mounted in the com-
pressometer provided specimen deformation data, while the MTS controller
recorded the applied load. All data were sampled four times per second. The
load-deformation data was used to calculate and plot the stress–strain curve for
each specimen.
Two general methods of SDT data analysis were employed, hereafter referred to
as the Smaoui Method and Sanchez Method. The original parameters recommended
by Chrisp et al. [13] were developed when the test was conducted using a lower
maximum load and ultimately unsupported by any correlation to the degree of
damage, and so were not used in this study.

4.3.1. Smaoui Method analysis

The Smaoui Method analysis focused on two parameters: the area of the hys-
teresis loop of the first loading cycle (A1) and total accumulated plastic strain over
all five loading cycles (S5). These parameters were found by Smaoui et al. to corre-
late nearly linearly to expansion from ASR [14], although more recent works are less
supportive that this is always the case [5,8,10,11,16,32]. These parameters are
extracted by numerical analysis of the stress and strain data generated by the
SDT, and are illustrated in Fig. 4. The average A1 and S5 at each level of expansion
are calculated and plotted against expansion.
4.3.2. Sanchez Method analysis
Sanchez et al. [5,8,10,11,30,32], following the above mentioned works of
Smaoui et al. [14,15,26], conducted extensive studies of SDT on AAR-affected con-
crete. Using the cyclic loading technique, they evaluated concrete specimens fabri-
cated in the laboratory presenting different compressive strengths (25, 35 and 45
MPa) and incorporating a wide range of reactive aggregates (both coarse and fine),
as well as concrete cores extracted from AAR-affected structures. The authors
showed that SDT starts being diagnostic and still non-destructive (i.e. tested sample
may be used to other tests after SDT cycles) when loads equal or greater to 30% of
the compressive strength are applied, although the procedure should ideally be car-
ried out at 40% of the compressive strength in order to be considered a diagnostic
test procedure. However, in many practical situations, the information regarding
mix design and 28-day compressive strength of in-service infrastructure is quite
limited, and even if quality control test records are available, in-situ strengths
Fig. 4. Illustration of the two parameters extracted for the Smaoui Method SDT data
analysis.
may vary considerably throughout the structure. Sanchez et al. [5,8,10,11,30,32]
suggested that the most suitable yet practical approach would be a two-step proce-
dure presented hereafter:

Step 1: Determine the compressive strength on cores extracted from undam-
aged or less-damaged locations in the structural member under investigation,
and;

Step 2: Use 40% of the value obtained in Step 1 to perform the SDT.
In addition, some practical parameters were found to influence SDT outcomes,
such as the moisture content and conditioning prior to conducting the test, location
and direction of the core (i.e. longitudinal vs. transverse vs. vertical), environmental
conditions and depth from the member surface (especially in cores drilled from dis-
tressed structures) [10].
The Sanchez Method was found very promising for assessing AAR damage and
progress in concrete, especially with the use of indices as output parameters of the
test, namely Stiffness Damage Index (SDI) and Plastic Deformation Index (PDI),
which represent respectively the ratio of dissipated energy or plastic deformation
(SI or DI, respectively) to the total energy or deformation induced in the concrete
(SI + SII or DI + DII, respectively) over the five cycles of testing (Fig. 5). Finally, fol-
lowing the works by Chrisp and co-workers [13], Sanchez et al. [7–9,11] confirmed
that the Non-Linearity Index (NLI), calculated by the slope of the stress–strain curve
from the origin to the half of the maximum load applied, divided by the slope curve
taken from the origin to the maximum load in the first cycle, would be an interest-
ing complementary index to consider, particularly for the assessment of cores
extracted from field structures, because this index accounts for preferred crack ori-
entation often found in AAR-affected reinforced concrete structures.

4.4. Other mechanical tests

Following the stiffness damage test, the static secant modulus of elasticity and
compressive strength of the specimens were determined in accordance with C469
[31] and ASTM C39 [34]. An estimated compressive strength (based upon an
assumed slight decline in strength from tests at the previous expansion level)
was used to determine the applied load for the first elastic modulus test, and sub-
sequent tests were based on the compressive strength of the previously tested spec-
imens in the set. The goal was to apply a load of 40% of the compressive strength;
however, the actual applied load varied from 29 to 46% of the compressive strength.

Fig. 5. Determination of the Stiffness Damage Index (SDI) and Plastic Deformation
Fig. 3. Test setup for stiffness damage test [4]. Index (PDI) based on the output from the Stiffness Damage Test (SDT).
 E.R. Giannini et al. / Construction and Building Materials 162 (2018) 253–264
Secant moduli should be nearly identical for peak loads of 30 to 45% of the compres-
sive strength [35]; the loads used in this study are generally within that range. This
sequence of tests (SDT ? E ? fc) typically involves the application of increasing
loads from one test to the next, so each test is not expected to influence the one
(s) which follow. In fact, ASTM C469 specifies that no data be recorded in the first
loading cycle, which supports the idea that there should be no influence on the
measured elastic modulus when the SDT is conducted at up to 40% of the compres-
sive strength. However, when the compressive strength is less than 25 MPa, the
loads used in for SDT will exceed those used to determine the elastic modulus when
the maximum SDT load is fixed at 10 MPa. To test this, the elastic moduli obtained
from both the SDT (average of 2nd and 3rd cycles, per suggestions in [10] and ASTM
C469 testing were compared and analyzed.
5. Results
5.1. Specimen expansions
The average expansions of the reference specimens are shown
in Fig. 6. The specimens from Mixture 1 exhibited much greater
expansion than those from Mixture 2 after approximately 50 days
after casting. However, both Mixture 1 and Mixture 2 specimens
expanded more quickly than the cylinders cast from those mix-
tures that were conditioned to expand only from ASR using storage
over water in sealed container at 38 °C. Expansion data for the ASR-
affected cylinders were presented alongside the data in Fig. 6 in
earlier works by Giannini and co-workers [16,18]. The difference
in expansion behavior between the two mixtures is not particu-
larly surprising; prior research has noted that the aggregates have
influence on expansion in DEF-affected mortars, and that this
appears to be most strongly correlated to aggregate shape and tex-
ture, and the resulting quality of the paste-aggregate bond, rather
than to alkali-reactivity [20,21,36,37]. Thus, it follows that the
manufactured limestone sand in Mixture 2 may be expected to
suppress expansion from DEF compared to Mixture 1, which con-
tained a natural sand.
5.2. Stiffness damage test
Figs. 7 and 8 present representative examples of the stress–
strain data obtained at each expansion level for Mixtures 1 and
2, respectively. Qualitatively speaking, as the expansion level
increases, increased signs of damage can be seen in the stress–
strain plots for both mixtures in the form of larger hysteresis areas
(indicating dissipated energy), greater accumulated plastic strain,
and reduced stiffness (measured by slope of the stress–strain curve
in the loading phases of each cycle). Furthermore, at higher expan-
sion levels, the loading portion of the stress–strain curve for the
first cycle is increasingly concave in shape, while the subsequent
loading cycles become increasingly convex in shape. These charac-
teristics also found in AAR-damaged materials, and are linked to
Fig. 6. Average expansions of instrumented cylinders from the two mixtures. Error
bars represent the range of expansions measured.
257
slip across crack surfaces while the cracks close under compression
loading [38].
5.3. Elastic modulus and compressive strength
Tables 3 and 4 present the elastic modulus and compressive
strength for Mixture 1 and Mixture 2, respectively. The tables show
the moduli and strength of individual specimens, as well as the
average value for all three tested at each expansion value. The
tables also show the average elastic modulus of each set of three
specimens as a percentage of the value that would be expected
based on the average compressive strength of the specimens (E,%
of Predicted). This value is calculated as:
Emean
E; % of Predicted ¼ pffiffiffiffi ð1Þ
4730 f c
where the denominator of the equation is the SI-unit formula used
in design equations to estimate elastic modulus in the ACI 318
building code for normalweight concrete [39].
6. Analyses and discussion
6.1. SDT – Smaoui Method analysis
Smaoui et al. [14,15] proposed two primary parameters to
quantify SDT results for ASR: (1) the area of the first loading cycle,
which will be referred to as A1 and; (2) the accumulated plastic
strain after five loading cycles, which will be referred to as S5.
Fig. 9 presents A1 plotted against expansion for each mixture and
Fig. 10 presents S5 plotted against expansion for each mixture.
Error bars in these figures represent the range of A1 and S5 values
measured for individual specimens. A linear regression is also
shown for the two mixtures in both Figs. 9 and 10. Each data point
represents the average of three cylinders tested at that expansion
level.
The linear regression of A1 vs. expansion has an R2 of 0.86 for
Mixture 1, and 0.79 for Mixture 2. In the case of S5, the linear
regression yields an R2 of 0.87 for Mixture 1 and 0.72 for Mixture
2. A better linear fit is evident for both A1 and S5 for Mixture 1 if
the final expansion level is excluded, with R2 values of 0.97 and
0.96, respectively. Comparing Figs. 9 and 10, it is clear that both
parameters show a very similar relationship to expansion and it
appears that the expansive phenomenon has two distinct phases:
a) an initial linear trend up to expansions of about 0.40% and; b)
a second phase where the rate of increase in the damage parame-
ters is somewhat slower compared to the initial phase.
6.2. SDT – Sanchez Method
According to Sanchez et al. [8,10,11], the SDT is more successful
in characterizing damage in concrete (i.e. more accurate and less
influenced by a number of test parameters such as test load)
through the use of indices, SDI and PDI (stiffness damage and plas-
tic deformation indices, respectively), as described in 4.3.2.
Fig. 11 shows a plot of the above indices against expansion for
Mixture 1. Linear fits are shown in Fig. 11a and logarithmic fit
curves in Fig. 11b. Because the first set of specimens were tested
at 0.000% expansion, this data was excluded when developing
the logarithmic fit. The SDI values range from 0.14 to 0.46 and
PDI values range from 0.13 to 0.42, for expansion levels varying
from 0% to 1.01%. Moreover, the variation of both indices as a func-
tion of expansion presented a ‘‘concave” (i.e. close to a logarithmic
function) trend as well as a somehow two-phase behavior: a) a
roughly linear trend up to 0.4% expansion and; b) a trend for level-
ing off at expansions greater than 0.4%. Therefore, when all the
258 E.R. Giannini et al. / Construction and Building Materials 162 (2018) 253–264

Fig. 7. Example stress–strain plots for Mixture 1 specimens (1 specimen per expansion level).

expansion levels are considered together, a linear regression
(Fig. 11a) does not indicate a very good fit for both indices (R2 of
0.72 and 0.69 for SDI and PDI, respectively). However, when the
last expansion level is removed, as shown in Fig. 12, the linear
trend is more clearly seen, and the linear regression has an R2 of
0.97 for both indices. The logarithmic curves in Fig. 11b provide
a much improved fit for both indices (R2 of 0.93 and 0.88 for SDI
and PDI, respectively) through the highest expansion level. How-
ever, the steep slope of the curve at 0.1% expansion suggests this
logarithmic fit may be problematic for very small expansions and
predict negative SDI and PDI values. A practical engineering
approach may be to use a two-part linear fit, with the slope
decreasing significantly at 0.4% expansion.
Fig. 13 shows a plot of the SDI and PDI against expansion for
Mixture 2. Linear fits are shown in Fig. 13a and logarithmic fit
curves in Fig. 13b. SDI values range from 0.18 to 0.32 and PDI val-
ues range from 0.16 to 0.33, for expansions up to 0.45%. In this
case, for both indices, a linear trend is also apparent, and linear
regression yields an R2 of 0.90 for SDI and 0.70 for PDI. The
improved linear trend for Mixture 2 compared to Mixture 1 (pri-
marily with regards to SDI) is likely related to the fact that the
maximum expansions were less for Mixture 2. The logarithmic fit
curves yield an R2 of 0.92 for SDI and 0.57 for PDI. In the case of
Mixture 2, there is little benefit derived from the use of a logarith-
mic fit over this range of expansions; this approach may be best
suited only when dealing with expansions well in excess of 0.40%.
 E.R. Giannini et al. / Construction and Building Materials 162 (2018) 253–264 259

Fig. 8. Example stress–strain plots for Mixture 2 specimens (1 specimen per expansion level).

Table 3
Elastic modulus and compressive strength results for Mixture 1.

Exp. Elastic Modulus (GPa) Compressive Strength (MPa) E,% of SDT Load
(%) Individual Specimens Mean Individual Specimens Mean Predicted (% of fc)
0.00 22.6 24.6 22.2 23.1 27.6 27.6 26.9 27.3 93.5 36.6
0.10 24.4 18.8 16.9 20.0 33.2 26.1 25.9 28.4 79.5 35.2
0.21 20.0 10.8 13.5 14.8 30.7 23.1 24.8 26.2 61.0 38.1
0.34 8.5 9.1 6.9 8.2 20.2 20.4 18.4 19.7 39.1 50.9
0.42 8.4 5.7 8.5 7.5 20.5 16.5 19.3 18.7 36.6 53.4
0.51 6.3 7.2 6.6 6.7 19.1 18.2 17.7 18.3 33.1 54.5
1.01 6.3 4.6 6.6 5.8 16.9 14.3 16.6 16.0 30.7 62.7

Table 4
Elastic modulus and compressive strength results for Mixture 2.

Exp. Elastic Modulus (GPa) Compressive Strength (MPa) E,% of SDT Load
(%) Individual Specimens Mean Individual Specimens Mean Predicted (% of fc)
0.01 23.1 23.8 23.8 23.6 32.9 33.9 32.2 33.0 86.8 30.3
0.08 20.4 21.6 18.4 20.1 31.9 33.7 33.1 32.9 74.2 30.4
0.18 16.2 17.5 20.1 17.9 33.4 30.9 33.0 32.5 66.6 30.8
0.27 13.4 15.5 19.7 16.2 28.9 30.4 28.7 29.3 63.2 34.1
0.33 19.6 20.7 15.5 18.6 32.7 35.6 30.4 32.9 68.5 30.4
0.45 8.7 9.5 10.6 9.6 27.3 25.0 23.1 25.2 40.3 39.7
260 E.R. Giannini et al. / Construction and Building Materials 162 (2018) 253–264

Fig. 9. Average area of the first loading cycle (A1) vs. expansion for Mixtures 1 and
2. Area is expressed as dissipated energy per unit volume (1 J/m3 = 1 Pa). Error bars
indicate the range of values calculated for individual specimens.

Fig. 11. Indices proposed by Sanchez vs. expansion level for Mixture 1 presenting
all the expansion levels with (a) linear fit, and (b) logarithmic fit. Error bars (black =
SDI, grey = PDI) indicate the range of values calculated for individual specimens.

Fig. 10. Average accumulated plastic strain over all five loading cycles (S5) vs.
expansion for Mixtures 1 and 2. Error bars represent the range of values calculated
for individual specimens.

An examination of the scatter in the data presented in Figs. 9–

13 also reveals that the SDI and PDI parameters exhibit signifi-
cantly less scatter than A1 and S5. As a percentage of the average
values at each expansion level, the range of individual SDI values
is an average of 24%, while the range of A1 values is on average
56% of the average value at a particular expansion level. The same
calculations for PDI and S5 yields 42% and 65%, respectively. While
this is a simplistic way to analyze the scatter of SDT data and the
calculated parameters, this further supports the use of the Sanchez
parameters of SDI and PDI in the analysis of SDT data for DEF-
damaged concrete.
Fig. 12. Indices proposed by Sanchez vs. expansion level for Mixture 1 disregarding
the last expansion level. Error bars (black = SDI, grey = PDI) indicate the range of
6.3. The use of SDT as a quantitative tool for assessing damage in values calculated for individual specimens.
concrete

Although the results discussed in the previous sections seem to
be interesting and promising, the use of the SDT in assessing con-
crete damage should be discussed, especially regarding how the
method detects, measures, and most importantly ‘‘considers” dis-
tress in damaged concrete affected by different mechanisms. The
output parameters proposed by Sanchez et al. [8,10,11] are able
to measure the actual ‘‘damage” state of a concrete based on the
hysteresis area and the plastic deformation under the stress vs.
strain curve of a tested material. In previous works, the reliability
of this approach was shown to be unrelated to the characteristics
of the material tested (e.g. compressive strength), and interesting
and comparable results were found for a wide range of AAR-
affected mixtures [5,8,10,11]. Under compression, the cracks and
flaws found in the damaged medium (i.e. cracked concrete) tend
to close. If the load applied is insufficient, the cracks remain
opened after load application, whereas if the load is too high, it cre-
ates new damage in the tested material. The latter seems to pre-
vent the use of fixed loading procedures proposed by Chrisp
et al. [13] and Smaoui at al. [14,15] for quantitatively assessing real
structures because in most practical cases, the characteristics of
 E.R. Giannini et al. / Construction and Building Materials 162 (2018) 253–264
Fig. 13. Indices proposed by Sanchez vs. expansion level for Mixture 2 with (a)
linear fit, and (b) logarithmic fit. Error bars (black = SDI, grey = PDI) indicate the
range of values calculated for individual specimens.
field concrete are unknown, especially in relatively old
infrastructure.
Sanchez et al. [8,10,11] verified that in order to be considered a
quantitative tool, the SDT procedure should somehow ‘‘challenge”
the sample analyzed, and thus a load around 40% of the sample
strength should be used. Moreover, the prior authors determined
that for AAR damage up to very high expansion levels (expansion
levels of around 0.3–0.4%), the procedure proposed would still
remain non-destructive.
In the case of AAR damage up to 0.3–0.4% expansion, Sanchez
et al. [8,10,11] verified the presence of a similar concave trend of
SDT/PDI vs. expansion levels, as is seen for Mixture 1 in this study.
The authors explained this trend according to microscopic analysis
made through the use of the Damage Rating Index (DRI) [7,9]. At
the onset of ASR expansion, a number of cracks are formed within
the aggregates particles. As expansion increases, new cracks are
again formed in the aggregate particles but more importantly,
the cracks already present increase in length and width, eventually
reaching the cement paste. Finally, the previously-formed cracks
link together forming a network of cracks. The ‘‘concave” or level-
ing off trend of the SDI and PDI was found to coincide primarily
with the last step of crack development, when the cracks are
growing and linking, rather than forming, because the indices
measured by the SDT are obtained by the number of cracks (which
are not necessarily greater at the last steps of the chemical reac-
tion) vs. the width of the cracks (which keeps increasing as a func-
tion of ASR development). Furthermore, the behavior of ASR gel
within the cracks was also questioned by the authors, and probably
they play a non-negligible role on the leveling off trend obtained.
Martin et al. [40] investigated the use of SDT according to the
261
Sanchez procedure for assessing concrete distress caused by com-
bined ASR and DEF and the results were promising. A fairly linear
trend was found in this case, which after some microscopic analy-
sis using the DRI, may be attributed to two main reasons: 1) the
different nature of DEF distress development, which occurs almost
exclusively in the cement paste and; 2) the lack of reaction prod-
ucts completely filling cracks caused by DEF.
In this work (Section 6.2), a linear trend of SDT parameters vs.
expansion was found for Mixture 2 concrete, whereas a two-step
behavior was found for Mixture 1 (a linear first step and a second
step with a concave or logarithmic trend). In the case of Mixture 1,
at very high expansions (>0.60%), the SDT parameters leveled off
even as expansion continued to increase. Without petrography of
the specimens to support a microstructural explanation, it is
unclear why this happens; however, prior works by Sanchez
et al. [8,10] suggest that sample conditioning parameters (e.g.
moisture content) may have some contribution to this observed
effect on SDT results. In this research, the samples remained satu-
rated throughout the conditioning (limewater soak) and testing
processes, effectively eliminating moisture as a variable. Therefore,
in order to fully understand the observed trends in SDT results, a
comprehensive petrographic evaluation is still needed and should
be addressed in future works. Otherwise, regarding practical appli-
cation of the tools, the largest expansion values attained by the
Mixture 1 specimens (>1%) are unlikely to be observed in the
majority of DEF-damaged civil structures. This is particularly true
for more heavily-reinforced concrete structures, where the steel
provides a confining force that significantly restrains expansion.
Reinforcement yielding, for example, happens at a strain of approx-
imately 0.20%; if expansions are restrained such that yielding does
not occur, the linear ‘‘first-step” relationship between SDT param-
eters and expansion may be valid for damaged concrete that would
be found in service. The analysis of the data obtained in the previ-
ous sections seems to once more support the capacity of the SDT in
characterizing damage due to DEF, as well as AAR.
6.4. Compressive strength and elastic modulus
Considering that it is quite simple to also test for the elastic
modulus and compressive strength of a specimen following SDT,
some discussion of the value of those parameters in characterizing
damage from DEF is warranted. Fig. 14a presents the static elastic
modulus (E) results (ASTM C469 test) for both mixtures vs. expan-
sion (although with the final expansion level removed for Mixture
1), and Fig. 14b compares this data against average moduli calcu-
lated from the 2nd and 3rd SDT loading cycles. Fig. 15 presents
the average compressive strength results vs. expansion. Error bars
in Figs. 14a and 15 represent the range of moduli and compressive
strengths measured for all specimens at each expansion value. Full
data for individual samples are given in Tables 3 and 4. The elastic
moduli of the specimens were greatly affected by damage from
DEF, decreasing as much as 75% for Mixture 1 and 59% for Mixture
2, compared to baseline (zero expansion) values. Compressive
strength decreased 41% for Mixture 1 and 24% for Mixture 2, com-
pared to the baseline values.
Although compressive strength is a significant parameter in
assessing the load-bearing capacity of a structure, it is ultimately
the least sensitive parameter investigated with respect to charac-
terizing DEF progress in concrete. Furthermore, as the data in
Tables 3 and 4 show, the elastic modulus is significantly under-
pffiffiffiffi
predicted by the equation E ¼ 4730 f c given in the ACI 318 Build-
ing Code. Measured elastic moduli were as much as 60–70% below
values calculated based on compressive strength. Therefore, not
only is the elastic modulus a more suitable parameter than
262 E.R. Giannini et al. / Construction and Building Materials 162 (2018) 253–264

compressive strength to characterize the degree of damage from

DEF, it must be tested for directly.
Fig. 14b shows there is a close agreement between the elastic
moduli measured using the ASTM C469 method and the 2nd and
3rd SDT loading cycles. This further supports the argument that
the SDT is nondestructive in nature, provided that the applied load
is approximately 40% or less of the compressive strength. The SDT
moduli were generally slightly higher than the ASTM C469 moduli;
for SDT loadings less than 40% of the compressive strength, the SDT
moduli were, on average, 4% higher than the ASTM C469 moduli.
The deviation between these moduli values increased for higher
expansions of Mixture 1 (0.344%), where the SDT loading
exceeded 50% of the compressive strength. A reexamination of
Figs. 7 and 8 show that the stress–strain curves of the 2nd through
5th SDT loading cycles become increasingly concave at higher
expansion levels; this means that as SDT loads exceed those
applied in the ASTM C469 test, the secant moduli calculated from
the SDT data would be expected to exceed those determined by
ASTM C469. As cracks are closed under load, the stress–strain
response of DEF-damaged concrete would be expected to stiffen
slightly; this behavior becomes more pronounced as damage
increases, and is reflected in the increasingly concave shape of
the stress–strain curve, and the disparity in moduli determined
by these two methods. However, it should be noted that either
method would be preferable to estimating the elastic modulus
from compressive strength data.

6.5. Statistical analysis

Fig. 14. (a) Average elastic modulus (ASTM C469) vs. expansion for Mixtures 1 and
2. Final expansion level excluded for Mixture 1. (b) Average elastic moduli from
ASTM C469 and SDT 2nd and 3rd cycles compared.
Fig. 15. Average compressive strength vs. expansion for Mixtures 1 and 2.
In order to assess the statistical validity and significance of the
database generated as part of this study, analyses of variance (two-
variable ANOVA) were performed, taking into account both the
individual specimens and the average results obtained at each
expansion level.
First, two-variable ANOVA was performed on the individual test
results (i.e. accounting for each specimen evaluated) obtained from
Sanchez Method in order to verify the significance of the individual
outcomes obtained at each expansion level studied (with a confi-
dence level of 95%) for all the parameters analyzed in this research
(i.e. PDI, SDI, and E, respectively). The results, presented in Table 5,
indicate that all the different expansion levels provided significant
results (i.e. the result of each set of specimens, composed of three
different cylinders, showed equal results for the level of signifi-
cance specified), as all the ‘‘F values” obtained were lower than
the ‘‘Fcritic” and the ‘‘p values” were greater than 0.05 for each case.
These results thus enabled the use of the average values obtained
at each expansion level to appraise the significance of the out-
comes obtained through the use of the Sanchez Method.
Second, following the previous statistical results, two-variable
ANOVA was performed on the average values obtained for each
set of specimens at each expansion level studied (Table 6). Analyz-
ing the data obtained, one may notice that all the results obtained
through the use of Sanchez Method (i.e. PDI, SDI and E, respec-

Table 5
Two-variable ANOVA performed on the individual test results obtained through the use of Sanchez Method.

Mixture Variable F Fcritic F > Fcritic a p-value p<a

1 SDI 1.30 3.74 No 0.05 .30 No
PDI 1.93 3.74 No 0.05 .18 No
E 1.93 3.74 No 0.05 .18 No
2 SDI 0.91 3.89 No 0.05 .43 No
PDI 0.36 3.88 No 0.05 .70 No
E 0.36 3.88 No 0.05 .70 No
 E.R. Giannini et al. / Construction and Building Materials 162 (2018) 253–264
Table 6
Two-variable ANOVA performed on the average test results obtained for the different parameters used in Sanchez Method.
Mixture Variable F Fcritic
1 SDI 8.88 2.76
PDI 9.25 2.76
E 30.59 2.76
2 SDI 8.95 3.00
PDI 8.85 3.00
E 45.20 3.00
tively) could be considered statistically significant, as all the
‘‘F values” obtained were largely greater than the ‘‘Fcritic” and the
‘‘p values” were much less than 0.05, confirming thus the analyses
and data interpretation made throughout this research.
7. Conclusions
In this work, the SDT was used to assess concrete mixtures
affected by DEF and presenting a wide range of deterioration levels.
From this study, the following conclusions can be made:

The SDT was previously found to provide useful information on
the degree of damage of ASR-affected concrete such as its stiff-
ness and inner damage. Furthermore, it has been found that
when loads equal or lower than 40% of the material’s compres-
sive strength are used, the SDT remains a non-destructive
method up to very high levels of deterioration of the material
tested.

SDT is suitable as a mechanical testing tool to quantitatively
characterize DEF evolution and damage in laboratory concrete.
Moreover, it is believed that the SDT would even be more suit-
able to appraise damage from DEF than from ASR, especially
because DEF microscopic damage features are almost entirely
found in the cement paste and/or ITZ. Thus, conceptually, the
SDT parameters would be less influenced by aggregate charac-
teristics (rock type, mineralogy, etc.) than when testing ASR-
damaged concrete. Further analyses are needed to validate this
hypothesis.

The SDI and PDI parameters developed by Sanchez et al. and
found to be suitable for assessing ASR damage can be directly
employed to characterize DEF-affected concrete even if the
loads applied in SDT are slightly lower than 40% of the compres-
sive strength.

The SDI and PDI parameters are more effective ways of charac-
terizing DEF-damaged concrete than previous parameters iden-
tified by Smaoui et al.

Both the SDI and PDI parameters obtained from the SDT and the
static elastic modulus provide statistically significant results
and could be considered unique ways of assessing damage from
DEF.

Future research to advance the use of SDT to characterize DEF
damage in concrete should seek to use petrographic examina-
tion to correlate the SDT and expansion data to the Damage Rat-
ing Index. It is also recommended that consideration be given to
combined deterioration from ASR and DEF, because DEF is more
commonly found in concert with ASR in field structures than as
the sole damage mechanism.
Acknowledgements
This research was supported in part by the Texas Department of
Transportation as part of research project 0-6491. The authors
263
F > Fcritic a p-value p<a
Yes 0.05 .0003 Yes
Yes 0.05 .0002 Yes
Yes 0.05 .00 Yes
Yes 0.05 .0007 Yes
Yes 0.05 .0008 Yes
Yes 0.05 .00 Yes
would also like to acknowledge the contributions of Kerry Kreit-
man, Zach Webb, Brian Hanson, Eric Schell, and Dennis Fillip.
References
[1] ISE, Structural Effects of Alkali-Silica Reaction, Institution of Structural
Engineers, London, 1992, 45.
[2] T. Ahmed, E. Burley, S. Rigden, A.I. Abu-Tair, The effect of alkali reactivity on
the mechanical properties of concrete, Constr. Build. Mater. 17 (2003) 123–
144, https://doi.org/10.1016/S0950-0618(02)00009-0.
[3] R.N. Swamy, Assessment and rehabilitation of AAR-affected Structures, Cem.
Concr. Compos. 19 (5-6) (1997) 427–440, https://doi.org/10.1016/S0958-9465
(97)00035-8.
[4] E.R. Giannini, Evaluation of Concrete Structures Affected by Alkali-Silica
Reaction and Delayed Ettringite Formation PhD Dissertation, The University
of Texas at Austin, Austin, Texas, 2012.
[5] L.F.M. Sanchez, B. Fournier, M. Jolin, J. Bastien, Evaluation of the Stiffness
Damage Test (SDT) as a tool for assessing damage in concrete due to ASR: test
loading and output responses for concretes incorporating fine or coarse
reactive aggregates, Cem. Concr. Res. 56 (2014) 213–229, https://doi.org/
10.1016/j.cemconres.2013.11.003.
[6] L.F.M. Sanchez, S. Multon, A. Sellier, M. Cyr, B. Fournier, M. Jolin, Comparative
study of a chemo–mechanical modeling for alkali silica reaction with
experimental evidences, Constr. Build. Mater. 72 (2014) (2014) 301–315,
https://doi.org/10.1016/j.conbuildmat.2014.09.007.
[7] L.F.M. Sanchez, B. Fournier, M. Jolin, J. Duchesne, Reliable quantification of AAR
damage through assessment of the damage rating index (DRI), Cem. Concr. Res.
67 (2015) 74–92, https://doi.org/10.1016/j.cemconres.2014.08.002.
[8] L.F.M. Sanchez, B. Fournier, M. Jolin, J. Bastien, Evaluation of the Stiffness
Damage Test (SDT) as a tool for assessing damage in concrete due to ASR: input
parameters and variability of the test responses, Constr. Build. Mater. 77
(2015) 20–32, https://doi.org/10.1016/j.conbuildmat.2014.11.071.
[9] L.F.M. Sanchez, B. Fournier, M. Jolin, M.A.B. Bedoya, J. Bastien, J. Duchesne, Use
of the Damage Rating Index (DRI) to quantify damage due to alkali-silica
reaction in concrete incorporating reactive fine and coarse aggregates, ACI
Mater. J. 113 (2016) 395–407, https://doi.org/10.14359/51688983.
[10] L.F.M. Sanchez, B. Fournier, M. Jolin, J. Bastien, D. Mitchell, Practical use of the
Stiffness Damage Test (SDT) for assessing damage in concrete infrastructure
affected by alkali-silica reaction, Constr. Build. Mater. 125 (2016) 1178–1188,
https://doi.org/10.1016/j.conbuildmat.2016.08.101.
[11] L.F.M. Sanchez, B. Fournier, M. Jolin, D. Mitchell, J. Bastien, Overall assessment
of alkali-aggregate reaction (AAR) in concretes presenting different strengths
and incorporating a wide range of reactive aggregate types and natures, Cem.
Concr. Res. 93 (2017) 17–31, https://doi.org/10.1016/j.
cemconres.2016.12.001.
[12] R.S. Crouch, J.G.M. Wood, Damage Evolution in AAR Affected Structures ,
International Symposium on Fracture Damage of Concrete and Rock, Vienna,
July 1988.
[13] T.M. Chrisp, P. Waldron, J.G.M. Wood, Development of a non-destructive test
to quantify damage in deteriorated concrete, Mag. Concr. Res. 45 (165) (1993)
247–256, https://doi.org/10.1680/macr.1993.45.165.247.
[14] N. Smaoui, M.-A. Bérubé, B. Fournier, B. Bissonette, B. Durand, Evaluation of the
expansion attained to date by concrete affected by alkali silica reaction. Part I:
experimental study, Canadian, J. Civ. Eng. 31 (5) (2004) 826–845, https://doi.
org/10.1139/l04-051.
[15] N. Smaoui, B. Fournier, M.-A. Bérubé, B. Bissonette, B. Durand, Evaluation of the
expansion attained to date by concrete affected by alkali-silica reaction. Part
II: application to nonreinforced concrete specimens exposed outside,
Canadian, J. Civ. Eng. 31 (6) (2004) 997–1011, https://doi.org/10.1139/l04-074.
[16] E.R. Giannini, K.J. Folliard, Stiffness Damage and Mechanical Testing of Core
Specimens for the Evaluation of Structures Affected by ASR, in Eds. T. Drimalas,
J.H. Ideker, B. Fournier, in: 14th Int. Conf. on Alkali-Aggregate Reactions in
Concrete, Austin, TX, USA, 2012, 10.
[17] A.F. Bentivegna, J.H. Ideker, Application of prognosis and diagnosis techniques
of ASR for a historic structure, in Eds. H.M. Bernardes, N.P. Hasparyk, in: 15th
Int. Conf. on Alkali-Aggregate Reaction in Concrete, São Paulo, Brazil, 2016.
[18] E.R. Giannini, K.J. Folliard, J. Zhu, O. Bayrak, K. Kreitman, Z. Webb, B. Hanson.
Non-Destructive Evaluation of In-Service Concrete Structures Affected by
Alkali-Silica Reaction (ASR) or Delayed Ettringite Formation (DEF) -Final
264 E.R. Giannini et al. / Construction and Building Materials 162 (2018) 253–264
Report, Part I, No. FHWA/TX-13/0-6491-1, Center for Transportation Research,
Austin, Texas, 2013.
[19] S. Kelham, The Effect of cement composition and fineness on expansion
associated with delayed ettringite formation, Cem. Concr. Compos. 18 (1996)
171–179, https://doi.org/10.1016/0958-9465(95)00013-5.
[20] K.J. Folliard, R. Barborak, T. Drimalas, L. Du, S. Garber, J. Ideker, T. Ley, S.
Williams, M. Juenger, B. Fournier, M.D.A. Thomas, Preventing ASR/DEF In New
Concrete: Final Report, Center for Transportation Research, Austin, Texas,
2006.
[21] H.F.W. Taylor, C. Famy, K.L. Scrivener, Delayed Ettringite Formation, Cem.
Concr. Res. 31 (5) (2001) 683–693, https://doi.org/10.1016/S0008-8846(01)
00466-5.
[22] T. Ramlochan, The Effects of Pozzolans and Slag on the Expansion of Mortars
and Concrete Cured at Elevated Temperature PhD Thesis, University of
Toronto, Toronto, 2003.
[23] K.L. Scrivener, M.C. Lewis, Effect of Heat Curing on Expansion of Mortars and
Composition of Calcium Silicate Hydrate Gel, Ed. B. Erlin. Ettringite - The
Sometimes Host of Destruction, ACI SP-177, American Concrete Institute,
Farmington Hills, Michigan, 1999. 93-104.
[24] R.P. Martin, C. Bazin, F. Toutlemonde, Alkali aggregate reaction and delayed
ettringite formation: common features and differences, in Eds. T. Drimalas, J.H.
Ideker, B. Fournier, in: 14th Int. Conf. on Alkali-Aggregate Reactions in
Concrete, Austin, TX, USA, 2012, 10.
[25] B. Fournier, M.-A. Bérubé, K.J. Folliard, M. Thomas, Report on the Diagnosis,
Prognosis, and Mitigation of Alkali-Silica Reaction (ASR) in Transportation
Structures, Report No. FHWA-HIF-09-004, Federal Highway Administration, U.
S. Department of Transportation, Washington, DC, 2010.
[26] M.-A. Bérubé, B. Durand, B. Bissonette, B. Fournier, N. Smaoui, Evaluation of the
expansion attained to date by concrete affected by alkali silica reaction. Part
III: application to existing structures, Canadian, J. Civ. Eng. 32 (3) (2005) 463–
479, https://doi.org/10.1139/l04-104.
[27] L. Sanchez, P. Salva, B. Fournier, M. Jolin, N. Pouliot, A. Hovington, Evaluation of
Damage in the Concrete Elements of the Viaduct ‘‘Robert-Bourassa-Charest”
After Nearly 50 Years in Service, in Eds. T. Drimalas, J.H. Ideker, B. Fournier, in:
14th Int. Conf. on Alkali-Aggregate Reactions in Concrete, Austin, TX, USA,
2012, 10.
[28] M.D.A. Thomas, K. J. Folliard, B. Fournier, T. Drimalas, P. Rivard, Study of
Remedial Actions on Highway Structures Affected by ASR, in Eds. T. Drimalas, J.
H. Ideker, B. Fournier, in: 14th Int. Conf. on Alkali-Aggregate Reactions in
Concrete, Austin, TX, USA, 2012, 10.
[29] S. Tremblay, B. Fournier, M. Thomas, T. Drimalas, K. Folliard, The Lomas
Boulevard Road Test Site, Albuquerque (New Mexico) - A Case Study on the
Use of Preventative Measures Against ASR in New Concrete, in Eds. T.
Drimalas, J.H. Ideker, B. Fournier, in: 14th Int. Conf. on Alkali-Aggregate
Reactions in Concrete, Austin, TX, USA, 2012, 10.
[30] L. Sanchez, B. Fournier, M. Jolin, Critical Parameters of the Stiffness Damage
Test for Assessing Concrete Damage due to Alkali-Silica Reaction, in Eds. T.
Drimalas, J.H. Ideker, B. Fournier, in: 14th Int. Conf. on Alkali-Aggregate
Reactions in Concrete, Austin, TX, USA, 2012, 10.
[31] ASTM C469/C469M-14, Standard Test Method for Static Modulus of Elasticity
and Poisson’s Ratio of Concrete in Compression, ASTM International, West
Conshohocken, Pennsylvania, 2014. doi: 10.1520/C0469_C0469M.
[32] L.F.M. Sanchez, Contribution to the Assessment of Damage in Aging Concrete
Infrastructures Affected by Alkali-Aggregate Reaction Ph.D. Dissertation,
Université Laval, Québec, 2014.
[33] M. Thomas, K. Folliard, T. Drimalas, T. Ramlochan, Diagnosing Delayed
Ettringite Formation in Concrete Structures, Cem. Concr. Res. 38 (6) (2008)
841–847, https://doi.org/10.1016/j.cemconres.2008.01.003.
[34] ASTM C39/C39M-17, Standard Test Method for Compressive Strength of
Cylindrical Concrete Specimens, ASTM International, West Conshohocken,
Pennsylvania, 2017. doi: 10.1520/C0039_C0039M-17.
[35] S. Popovics, Strength and Related Properties of Concrete: A Quantitative
Approach, John Wiley & Sons, New York, 1998.
[36] P.E. Grattan-Bellew, J.J. Beaudoin, V.-G. Vallée, Effect of Aggregate Particle Size
and Composition on Expansion of Mortar Bars due to Delayed Ettringite
Formation, Cem. Concr. Res. 28 (8) (1998) 1147–1156, https://doi.org/10.1016/
S0008-8846(98)00084-2.
[37] R. Yang, C.D. Lawrence, J.H. Sharp, Effect of type of aggregate on delayed
ettringite formation, Adv. Cem. Res. 11 (3) (1999) 119–132, https://doi.org/
10.1680/adcr.1999.11.3.119.
[38] S. Mindess, J.F. Young, D. Darwin, Concrete, second ed., Prentice Hall, Upper
Saddle River, New Jersey, 2003.
[39] ACI Committee 318, ACI 318-14, Building Code Requirements for Structural
Concrete, Farmington Hills, Michigan: American Concrete Institute, 2014.
[40] R.-P. Martin, L. Sanchez, B. Fournier, F. Toutlemonde, Diagnosis of AAR and
DEF: Comparison of Residual Expansion, Stiffness Damage Test and Damage
Rating Index, in Eds. H.M. Bernardes, N.P. Hasparyk, 15th Int. Conf. on Alkali-
Aggregate Reaction in Concrete, São Paulo, Brazil, 2016.