Cement and Concrete Research 106 (2018) 1–11

Contents lists available at ScienceDirect

Cement and Concrete Research

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

Determination of mortar setting times using shear wave velocity evolution T

curves measured by the bender element technique
⁎
J. Zhua,b, J.N. Caob, B. Batec, , K.H. Khayatd
a
Jones Lang LaSalle, 1 Zhongxinsi Road, Futian District, Shenzhen 518048, China
b
Department of Civil, Architectural and Environmental Engineering, Missouri University of Science and Technology, Rolla, MO 65409, United States
c
Institute of Geotechnical Engineering, College of Civil Engineering and Architecture, Zhejiang University, Hangzhou, China
d
University Transportation Center and the Center for Infrastructure Engineering Studies, Department of Civil, Architectural and Environmental Engineering, Missouri
University of Science and Technology, Rolla, MO 65409, United States

A R T I C L E I N F O A B S T R A C T

Keywords: Although identified as a good indicator to characterize freshly cast cementitious materials, shear wave velocity
Mortar (Vs) alone has not been used successfully to determine the initial and final setting times (ti and tf). The challenge
Early age originates from the large Vs range that can vary from < 50 m/s to > 2000 m/s for cementitious materials at early
Shear wave velocity age (typically < 24 h). To overcome these challenges, modifications to traditional bender element and to spe-
Initial setting time
cimen geometry were made to obtain Vs versus time (Vs(t)) curves of six early age mortars at different water-to-
Final setting time
Bender element
cement ratios, some with chemical admixtures. Derivative methods were then proposed to obtain ti and tf. The
peak time (tpeak′) in the first-order derivative of Vs(t) curves correlate well to the final setting time (R2 = 0.979),
while the peak time (tpeak″) of the second-order derivative of Vs(t) curves correspond well to the initial setting
time (R2 = 0.950).

1. Introduction (approximately 1490 m/s), which is on the same order of magnitude as

that of the fresh cementitious material where Vp can vary from ap-
Monitoring freshly cast cementitious materials (paste, mortar, and proximately 100 m/s in the fresh state to over 4000 m/s in the hardened
concrete) at early age (approximately the first 24 h) is desired in quality state. Subsequently, P-wave velocity originating from the solid portion
assurance and quality control (QA/QC), and for long-term performance of the cement-based material during curing is shielded by water, which
prediction. Initial and final setting times are the two key parameters to makes it a poor indicator of the curing process when used alone. Carette
characterize cementitious material properties of early age. Initial set- and Staquet [11] combined Vp and Vs results to determine setting time
ting time denotes the time when a cementitious material is sufficiently of mortars, and concluded that the P-wave is less sensitive to the setting
rigid to withstand a certain pressure and the material starts losing its process than S-wave. The hydraulic pressure method monitors the set-
plasticity. Final setting time denotes the time when the developments of ting/hardening process using wall hydraulic pressure that are in contact
strength and stiffness start, and the plasticity is completely lost. Both with the concrete through the formwork (from hydrostatic to zero as
setting times are useful parameters in the transportation, casting, and the specimen cures from slump to fully hardening) [7,12]. Hydraulic
consolidation of cementitious materials and are key parameters for pressure method can be implemented in the field. However, this
strength development at early age and for formwork removal [1,2]. method necessitates the use of high accuracy pressure sensors. Pore
The standard methods of measuring setting times are based on the water pressure sensors in contact with concrete can then be used to
penetration resistance test (ASTM C403) for mortar or concrete and the evaluate the rate of setting.
Vicat needle test (ASTM C191) for paste. Both tests are destructive la- Previous studies suggest that shear wave velocity (Vs) is a good
boratory tests. The isothermal calorimetry method was employed to indicator of the curing process of cementitious materials, such as
determine setting times from heat evolution curves [3–8]. The above mortars [11] and cement-soil mixtures [13,14]. This is because shear
methods are at the specimen scale. The ultrasonic pulse velocity mea- waves propagate primarily through the solid skeleton of a material and
surement [9,10] is widely used in the field due to its non-destructive is not significantly influenced by the presence of water or air. In spite of
nature and its sensitivity to the presence of air pockets, abnormalities, its use in evaluating characteristics of in-situ soils and cemented soil in
or defects. However, water in the concrete leads to a high Vp value the laboratory, shear wave velocity alone is not commonly used to

⁎
Corresponding author at: Institute of Geotechnical Engineering, College of Civil Engineering and Architecture, Zhejiang University, Hangzhou, China.
E-mail address: [email protected] (B. Bate).

https://doi.org/10.1016/j.cemconres.2018.01.013
Received 14 February 2017; Received in revised form 4 November 2017; Accepted 11 January 2018
Available online 02 February 2018
0008-8846/ © 2018 Elsevier Ltd. All rights reserved.
J. Zhu et al. Cement and Concrete Research 106 (2018) 1–11

monitor the early age properties of cementitious materials. This is be- Table 1
cause early-age cementitious materials have a larger range of stiffness Mixture proportioning of tested mortars.
variations during the curing process where Vs can vary from around
Mixture Mix 1a Mix 2 Mix 3 Mix 4 Mix 5 Mix 6
50 m/s in the fresh state to over 2000 m/s in the hardened stage. This is
considerably greater than soils that normally vary from 30 m/s (kaoli- w/c 0.50 0.43 0.37 0.43 0.43 0.43
nite [15]) to 350 m/s (iron oxide-coated sand, [16]) under normal Cement (kg/m3) 673 713 751 713 713 713
Sand (kg/m3) 1137 1203 1267 1203 1203 1203
loading conditions (< 400 kPa). The quick stiffness (Vs) increment in-
Water (kg/m3) 337 313 277 313 313 313
creases the resonant frequency of the specimen, which subsequently Set retarder (ml/100kgc) – – – 195 – 220
mandates the increment of the exciting frequency to ensure a good Set accelerator (ml/100kgc) – – – – 1500 –
signal-to-noise ratio [17,18], reduces wavelength, and poses challenges Unit weight (kN/m3) 21.34 22.28 22.63 22.08 21.78 22.08
in maintaining the wavelength ratio (Rd) requirement (Rd > 2) needed a
Repeated three times to verify the repeatability of material properties.
to minimize the near-field effect in < 24 h [19,20]. This is the major
reason that existing studies using traditional piezoceramic material-
based tools, such as the traditional bender element (BE) [21] and pie- close to the lower limit of the specification for aggregate used to make
zoelectric ring actuator [7], provide only partial Vs evolution data masonry mortar (ASTM C144). Two chemical admixtures were in-
(< 15 h) for cementitious materials generally with longer curing pro- troduced to alter the range of setting times: a hydration controlling
cess (normally > 24 h). However, in view of its advantages, such as in- admixture that retards setting times by controlling the hydration of the
situ and laboratory experimental capabilities and well-established ap- cement, and a non-chloride accelerating admixture that accelerates
proaches for soils, the bender element technique could be suitable to cement hydration.
evaluate the stiffness of cement-based materials at early age.
This study aims at modifying the traditional bender element to 3. Mixing design
obtain Vs evolution curves of mortar at early-age spanning from 0
to > 24 h. The study also seeks to determine the initial and final setting Six mortar mixtures were used, as shown in Table 1. Mortar mix-
times based on Vs evolution results. The following objectives are pro- tures with w/c of 0.50, 0.43, and 0.37 are referred to as Mix 1, Mix 2,
posed accordingly: (1) modify the geometry of mortar specimen and and Mix 3, respectively. The dosage rates of the admixtures (i.e., set
change the geometry, alignment and coating of a traditional bender accelerator and set retarder) were selected based on the criteria in
element used for soils to cover Vs variations at early-age of cement- ASTM C 494: Normal variation of delaying in the initial setting time is
based materials; and (2) measure the evolution of shear wave velocity between 1 and 3.5 h when using a set retarder, or between 1 and 3.5 h
of mortar specimens at early age (up to 96 h) with the modified bender earlier when using an accelerator. Mix 4 and Mix 6, modified from Mix
element testing system. Six mortar mixtures of different water-to-ce- 2 (w/c of 0.43), contained 3 fl oz./cwt (fluid ounce/cement hundred-
ment ratios (w/c), including one with a set accelerator and two with set weight) (195 ml/100kgc) and 3.4 fl oz./cwt (220 ml/100kgc) hydration
retarders were prepared to embrace a broad range of setting times. The controlling admixture, respectively. Mix 5 was modified from Mix 2
evolution of Vs with time was analyzed to evaluate setting times, which with w/c of 0.43, and incorporated 23 fl oz./cwt (1500 ml/100 kg) ac-
were compared to values measured by ASTM standard penetration re- celerating admixture. The procedure used for mixing mortar is in
sistance test. In addition, a calorimetry test-based method was also compliance with ASTM C 305. A mechanical mixer was used. In each
evaluated and used to correlate to setting times determined from pe- test, a batch of 25 l of mortar was prepared, and 21 l was placed in the
netration resistance test. formwork.

4. Bender element testing system

2. Materials
A bender element (BE) testing system for measuring Vs of ce-
A Type I portland cement was used. Missouri River sand. A well mentitious materials was used. The BE system consisted of three pairs of
graded river-bed sand was used. The sand was sieved through No. 4 bender elements, a signal generation and acquisition system, and a
sieve with D50 of 0.7 mm and Cu (coefficient of uniformity, D60/D10) of wooden formwork measuring 0.61 × 0.305 × 0.14 m3
2.74. The grain-size distribution of the sand is shown in Fig. 1 and is (length × width × height, inner geometry) (Fig. 2). The system was
developed with modifications on a traditional bender element system.
100 The details of each component and the rationale behind the corre-
Missouri River Sand sponding modifications are discussed below.
90
ASTM C144 lower limit
80 5. Bender element test setup
ASTM C144 upper limit
Cumulative Passing (%)

70
Two-layered brass-reinforced piezo actuators were cut into bender
60 element plates with dimension of 23 × 11.5 × 2 mm3
(length × width × thickness). This size is larger than typical sizes
50 ranging from 12 × 5 × 0.5 mm3 to 20 × 12.7 × 2 mm3 [22] used in
40 soil testing. The selected larger size is expected to enhance signal
strength given the long travel distance in a large specimen and the
30 initially paste-type materials (possibly weak signal due to the few
20
contacting points for Vs propagation). A parallel-type connection was
also adopted over series-type connection for stronger received signals
10 [23].
From inside out, coatings of a bender element that is typically used
0
0.01 0.1 1 10
for geotechnical applications follow the order of polyurethane, silver
Sieve Size (mm) conductive paint, and epoxy coatings (Fig. 2b) [15,17,23,24]. Mod-
ifications were made to these coatings to accommodate testing of ce-
Fig. 1. Grain-size distribution of sand used in this study.
mentitious materials, which are corrosive with high pH. For the

2
J. Zhu et al. Cement and Concrete Research 106 (2018) 1–11

purple
primer
PVC Installed on a
3 days later cement socket
traditional coating modified coating

Fig. 2. Bender element testing system for monitoring of early age stiffening of cementitious materials: (a) bender element coatings, (b) coating comparison, (c) signal generation and
receiving system, and (d) mortar specimen and formwork.

Fig. 3. Penetration resistance evolution with time for the

7000 Mix-1 (w/c=0.50) six investigated mortars.
Mix-2 (w/c=0.43)
Mix-3 (w/c=0.37)
6000
Mix-4 (w/c=0.43+1.95 mL/kg retarder)
Mix-5 (w/c=0.43+ accelerator)
Penetration Resistance (psi)

5000 Mix-6 (w/c=0.43+2.20 mL/kg retarder)

4000

3000

2000

1000

0
100 200 300 400 500 600 700 800
Elapsed Time (min)

modified BE system, polyurethane was still used in direct contact with grounding (drain wire embedded into the testing material) [23,24].
the piezoceramic plate as the insulation and waterproofing coating. (2) PVC cement was used instead of epoxy due to its higher moisture
Three to five polyurethane layers were painted onto the piezoceramic resistance, good flexibility, chemical resistance and durability. A
plate to strengthen the coating and to avoid possible electricity shorting layer of Oatey purple primer was used to roughen up the surface of
path. Outside the polyurethane coating, however, the following mod- the polyurethane-coated piezoceramic plate, and to provide better
ifications to the traditional coatings were made: mechanical bonds between the polyurethane and PVC cement [19]
(Fig. 2b).
(1) Silver paint coating was not used in this study, despite its electrical (3) Outside of the previously mentioned coatings, a plastic sheet was
shielding ability that prevents cross-talking [23]. This is because: (i) tightly wrapped around bender element unit to further protect the
no obvious improvement in the quality of signals was observed unit in a corrosive environment.
using silver painting in a side-by-side comparison test on dry sands
using bender elements with and without silver conductivity coating; To sum up, the coatings on a BE unit follow the order of (from inside
(ii) the BEs are more prone to electrical short-circuiting during the to outside): polyurethane; purple primer; PVC cement; and plastic wrap
silver coating procedure, which leads to low success rate of BE (Figs. 2a–b).
fabrication; and (iii) a sufficient electrical shield seemed to be Several bender element tests were performed on three individual
provided by parallel BE made with twisted coaxial cable with specimens with Mix 1 to evaluate repeatability.

3
J. Zhu et al. Cement and Concrete Research 106 (2018) 1–11

Fig. 4. Heat evolution versus time for the six tested mortars.

12 Fig. 5. Example input and output signals of (a–b) square

8 (a) function and (c–d) sine function for a mortar specimen with
0.5 w/c (Mix I) determined after 8 h of age.
Input (V)

4
0
-4
-8
-12
0.10
Output (mV)

0.05
(b)

0.00

-0.05 travel time = 0.40 ms

-0.10
12
8 (c)
4
Input (V)

0
-4
-8
-12
0.10
(d)
Output (mV)

0.05
0.00
-0.05 travel time = 0.43 ms
-0.10
-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
Time (ms)

6. Formwork required to ensure the travel time sufficiently larger than the system lag.
(3) On the other hand, in order to receive clear signals, travel distance
There are a few considerations in choosing the dimensions of the should not be too long. (4) To avoid near-field effect, the travel distance
formwork. (1) There is inherent system time lag due to the bender ele- to wavelength ratio (Rd ratio) should be no less than two [25]. In this
ment plate, coatings, wiring, and the electrical equipment [19,20]. Kang study, the tip-to-tip travel distance was chosen to be 0.27 m, which gave
et al. [20] determined the system lag of the same bender element testing a travel time at least 8 times greater than the system lag while satisfying
system with a smaller BE (12.7 × 8.0 × 0.6 mm, the wavelength ratio requirement for the majority of the time (resonant
length × width × thickness) to be 6–11 μs. (2) The stiffness of hardened frequency and wavelength evolve over time, see also the last paragraph
cementitious materials, such as mortar in this study, is usually higher of this section). Subsequently, the inner dimensions of the formwork
than that of common soils. Therefore, long wave travel distance is were 0.61 × 0.305 × 0.14 m3 (length × width × height) (Fig. 2d).

4
J. Zhu et al. Cement and Concrete Research 106 (2018) 1–11

10 MHz signal generator, a linear amplifier, a 4 pole LP/HP filter, and a

100 MHz oscilloscope. The stiffness and attenuation (energy dissipa-
tion) of cement paste, mortar, and concrete at early age (up to 72 h)
evolve rapidly as they change from slurry state to a semi-solid state.
Consequently, their resonant frequencies increase drastically (estimated
to be from 100 Hz to 14,000 Hz), while the attenuation of the received
electrical signals likely decrease over time. To receive strong signal and
weak noise, the exciting frequency of the input sine wave was also
adjusted to be close to the resonant frequencies. Square wave, con-
taining a wide frequency range that covers the evolving natural fre-
quency of the mortar, was also used. Cutoff frequencies were 1 Hz (high
pass) and 50 kHz (low pass), respectively. The amplitude of the wave-
form generator was 10 V.

7. Penetration resistance test

To determine the initial and final setting times of mortar, penetra-

tion resistance test was performed in accordance to ASTM C403 and
Fig. 6. Shear wave velocity versus elapsed time curves from bender element tests on three
AASHTO T197. Loading apparatus, penetration needles with bearing
individual mortar specimens with Mix 1. areas of 645, 323, 161, 65, 32, and 16 mm2, and tamping rod were
used. The penetration resistance was calculated by dividing the re-
corded force by the needle bearing area. Six to nine undisturbed pe-
Thick (0.038 m in thickness) wooden board was chosen as the
netration readings of penetration resistance were recorded at different
formwork material for the following two reasons: (1) it is rigid enough
elapsed times for each test. The time taken to penetrate 25.4 mm depth
to resist any lateral movement due to the lateral pressure exerted by the
was about 10 ± 2 s. Elapsed time was calculated from the time when
fresh mortar; (2) its thickness can ensure tight fixation of the BEs on the
water was added to cement. Initial and final setting times are de-
wall and provided good contacts between the bender and mortar.
termined to be the elapsed times at penetration resistance of 3.5 MPa
Wooden boards were assembled together by screws, which is easily
and 27.6 MPa, respectively.
removed after each test. Oil lubrication along the inside wall of the
formwork, together with screw connections, enabled the reuse of
wooden formwork. 8. Calorimetry test
Three pairs of BEs were installed in pre-drilled holes (diameter,
0.022 m), aligning perpendicularly to the bottom of the formwork to A calorimetry test was carried out, in compliance with ASTM C1679
prevent possible voids immediately underneath the benders during to evaluate the heat flow generated by the hydration reaction of cement
mortar placement. The vertical alignment of the benders also avoided over time. I-Cal 8000 Isothermal Calorimeter and the accompanying
interference by compressive waves reflected from sidewalls because of CalCommander software were used. Fifty to 150 g of mortar sample was
their elongated travel path [20]. The distance between the center of a placed in a clean reusable plastic cup. The lid was closed until testing to
BE and the mortar surface was 0.038 m. minimize heat exchange with the surrounding air. TheArrhenius' law
The signal generation and acquisition system consisted of a 1 mHz- (Eq. 1) was used to describe the temperature dependency of the hy-
dration rate of cement:

Fig. 7. Shear wave velocity versus elapsed time curves.

5
J. Zhu et al. Cement and Concrete Research 106 (2018) 1–11

Fig. 8. Vs evolution over time of mortars with different w/c.

Legends beginning with “Carette”, “Liu” and “Soliman”
refer to Carette and Staquet [11], Liu et al. [21], and So-
liman et al. [7], respectively.

Table 2
Initial and final setting times determined from different methods.

Mortar Initial setting time (h) Final setting time (h)

Penetration resistance BE (tpeak″) BE (a) Calorimetry Penetration resistance BE (tpeak′) Calorimetry

Mix-1 5.00 4.59 5.16 4.48 6.42 6.50 6.56

Mix-2 4.33 5.32 4.20 4.14 5.67 6.20 5.78
Mix-3 3.28 3.03 3.85 3.46 4.83 5.10 5.07
Mix-4 6.53 6.78 6.36 7.75 8.08 8.50 9.30
Mix-5 3.28 3.13 2.88 3.46 4.38 3.80 3.83
Mix-6 9.58 9.17 9.56 8.71 11.50 11.50 10.34
R2 – 0.950 0.981 0.910 – 0.979 0.908

Fig. 9. The rate of heat evolution curve and its first-order de-
rivative curve for Mix 1.

6
J. Zhu et al. Cement and Concrete Research 106 (2018) 1–11

constant temperature (20.0 °C) was maintained during all calorimetry

tests to avoid inconsistent results induced by temperature variation.

9. Results

The penetration resistance versus elapsed time curves for six mix-
tures is shown in Fig. 3.
The rate of heat evolution versus elapsed time relationship for the six
fresh mortars measured by calorimetry test is shown in Fig. 4. The
process of cement hydration can be broken down into five stages. In the
first stage, a large amount of heat is rapidly generated after cement
contacts with water. Hydration activity slows down in Stage 2, which is
also known as the dormant period. Mortars with set retarder (Mix 4 and
Mix 6) have longer dormant periods. In Stage 3, heat release accelerates
followed by a rapid reaction between calcium hydrate (CH) and cal-
cium silicate hydrate (CSH) [7]; both the initial setting (beginning of
solidification) and final setting (complete solidification and beginning
of hardening) occur in Stage 3. Hydration products are formed at slower
rates during Stage 4 (deceleration) and Stage 5 (diffusion limited).
The received BE signals from both the sine and square wave forms
are plotted in Fig. 5. The time of arrival of the received shear wave was
determined using the zero-crossing point with the x-axis of the half
peak before the first major peak (Fig. 5b and d). Then, the travel time
was determined from time zero to the time of arrival, and the shear
wave velocity was calculated by dividing the tip-to-tip distance by the
Fig. 10. Comparison between final setting times obtained from the penetration resistance travel time. The travel times determined from square and sine wave
and those from shear wave velocity derivative method. Solid symbols are experimental
forms should be the same [17,26]. In practice, however, minor differ-
data from this study; open symbols are from Carette and Staquet [11].
ences (< 10% in this study) in the travel time by different wave forms
exist. For instance, in the case of the Mix 1 mortar made with 0.5 w/c, a
7% difference in travel time was observed between square and sine
wave forms at 8 h of age. In such a case, an average shear wave velocity
was reported.
Three repeated Vs versus time relationships for Mix 1 agrees well
with each other (Fig. 6), which suggests good repeatability of the BE
system. The variations of shear wave velocity with elapsed time for the
six mortar mixtures are illustrated in Figs. 7 and 8. The results show
that: (1) shear wave velocity increased monotonically with time during
the first 24 h. Three phases of the Vs evolution curve can be observed: a
gentle slope before approximately 5 to 8 h, a steep increment from
about 5 to 15 h, and a gentle slope approximating to a plateau value
after about 10 to 15 h. (2) Shear wave velocity of hardened (> 20 h)
mortar ranged from 1700 to 2100 m/s. (3) At the same elapsed time in
mortar made without a set accelerator or retarder, Vs of fresh mortar
with low w/c was higher. The use of a set retarder significantly delayed
hydration reaction of cement by approximately 3 h and 5 h when in-
corporated of dosage rates of 195 ml/100 kg and 220 ml/100 kg, re-
spectively (Table 2). (5) The use of a set accelerator increased Vs in the
initial 9 h, and then registered a lower Vs value than the mortar made
without any set accelerator did (Fig. 7), which suggests that the use of
set accelerator decreased the Vs of the hardened mortar.

10. Discussion

10.1. Comparison of Vs evolution at early age to previous studies

Fig. 11. Comparison between initial setting times obtained from penetration resistance
and those either from Vs evolution methods (tpeak″ and parameter a) or from calorimetry
Compared to the shear wave velocity (Vs) measurements of mortar
method. Solid symbols are experimental data from this study; open symbols are from mixtures previously reported by Soliman et al. [7], Liu et al. [21], and
Carette and Staquet [11]. Carette and Staquet [11], it can be observed that for the same curing
time, the range of the magnitude of Vs in this study (0–2100 m/s) is
similar to that in previous studies, and that the variations of Vs with
k = Ze−Ea RT (1) elapsed time curves in this study share the same S-shaped feature as
those in the literature (Fig. 8). It can also be noted that Vs of the mortar
where k is the hydration rate constant, Z is a proportionality constant, made with w/c of 0.50 (1697 m/s) after 24 h of curing in this study is
Ea is the activation energy for the reaction, R is the ideal gas constant in lower than that of a hardened mortar at the same w/c (2260 m/s) re-
J/(mol.K), and T is the temperature in Kelvin. Eq. 1 suggests that the ported by Vipulanandan and Garas [27]. In addition to the longer
warmer the mortar, the faster the hydration reaction is. Therefore, a curing time in the latter study, a higher Vs value could also originate

7
J. Zhu et al. Cement and Concrete Research 106 (2018) 1–11

Fig. 12. Fitting curves using Weibull, the lognormal, and

Gamma cumulative distributions, and modified Fredlund and
Xing equation. Data labeled “measured” refers to the measured
Vs data in this study.

Table 3 subsequent time steps can be taken directly from the calorimetry test
Parameters of Weibull, lognormal, and Gamma cumulative distributions, and Fredlund results to achieve sufficient accuracy since approximately 500 time
and Xing equation. steps were recorded between the valley of Stage 2 and the peak of Stage
Equation Parameters Mix 1 Mix 2 Mix 3 Mix 4 Mix 5 Mix 6
3. The rate of heat evolution versus elapsed time curve and its first order
derivative curve for Mix 1 is shown in Fig. 9 as an example. The initial
Weibull α 2.178 2.651 2.303 3.018 1.773 3.384 and final setting times of all six mortar specimens obtained using the Ge
β 9.953 8.933 8.124 11.286 6.955 14.083 et al. [3] method are presented in Table 2, and were correlated to those
R2 0.996 0.995 0.994 0.998 0.992 0.999
Lognormal μ 2.097 2.031 1.904 2.285 1.695 2.511
measured from the penetration resistance test, as shown in Figs. 10 and
σ 0.516 0.416 0.486 0.411 0.637 0.331 11, respectively. For the calorimetry measurement in this study, both tcp
R2 0.998 0.996 0.998 0.994 0.997 0.996 and tcp′ values correlate to the final and initial setting times obtained
Gamma α 3.934 5.795 4.422 6.447 2.707 9.227 from standard penetration tests reasonably well (with R2 of 0.910 and
β 2.295 1.405 1.665 1.617 2.341 1.396
0.908, respectively). These good correlations seem to suggest that the
R2 0.998 0.996 0.997 0.996 0.995 0.998
Modified a 8.068 6.696 6.190 9.778 4.817 14.346 calorimetry method, albeit yielding slightly lower R2 values
Fredlund n 3.015 5.019 3.464 3.683 2.677 4.047 (0.908–0.910) than those (0.950–0.981) from the proposed Vs-based
and Xing m 2.266 1.656 1.921 2.162 1.825 3.707 method in this study, is viable. For the calorimetry measurement in
R2 0.999 0.995 0.999 0.997 0.998 0.999 Carette and Staquet [11], however, both tcp and tcp′ values correlate
poorly to tf and ti values obtained from standard penetration tests
(Figs. 10–11). It is not clear whether the poor correlations origin from
from the larger maximum size of the aggregate, better quality cement
the narrow range of the initial and final setting times for the mortars in
and mixing, as well as the systematic shift in the measurement method
Carette and Staquet [11] or the calorimetry method proposed by Ge
and first arrival time determination [20].
et al. [3]. Further calorimetry testing and analysis are warranted before
In the following sections, the initial and final setting times are de-
conclusions can be drawn regarding the validity and limitations of the
termined from the penetration resistance, calorimetry test method, and
calorimetry method.
shear wave velocity (Vs) test. The ti and tf values deducted from the
calorimetry and Vs tests are compared to those obtained from the pe-
10.4. ti and tf determined from Vs evolution curves
netration resistance test, which was used as a benchmark.

The evolution of Vs versus elapsed time could be used to determine

10.2. ti and tf determined from penetration resistance test the initial and final setting times [7]. Soliman et al. [7] plotted the first
derivative of Vs (dVs/dt) versus time curve, and defined the times cor-
The initial and final setting times determined from penetration re- responding to the “lower concave point” and the “highest convex point”
sistance test (ASTM C403) ranged from 197 to 575 min and from 263 to in the major Vs increment portion of the dVs/dt versus time curve as the
690 min, respectively as shown in Table 2 and Fig. 3. initial and final setting times, respectively. However, it is not always
possible to obtain the ‘lower concave point’ and the ‘highest convex
10.3. ti and tf determined from calorimetry test point’ because dVs/dt versus time curve is often monotonic in the major
Vs increment portion (see also Fig. 13). This is because the dVs/dt versus
Ge et al. [3] proposed that the initial setting time correlates to the time curves in this study and that reported by Carette and Staquet [11]
peak time (tcp′) of the first-order derivative curve of Stage 3 of the rate do not have such local peak or valley points (Fig. 13) and noise (or
of heat evolution versus elapsed time curve, while the final setting time fluctuation due to the variation of measured Vs values) could be mis-
correlates to the peak (tcp) of Stage 3 of the rate of heat evolution versus interpreted as such local points due to the discrete nature of the dVs/dt
elapsed time curve. Instead of curve fitting, differences between two versus time curves.

8
J. Zhu et al. Cement and Concrete Research 106 (2018) 1–11

Fig. 13. Original, first, and second derivatives of Vs versus

elapsed time curves.

Table 4
Fitting parameters of lognormal and modified Fredlund and Xing equations for experimental data of mortars with clinkers (REF), municipal solid waste incineration electrostatic
precipitator fly ash at different ratios (EFA-1 and EFA-2), thermal power station fly ash at different ratios (TFA-1 and TFA-2), reduced water/binder ratio of 0.4 (R-0.4), and air entraining
agent (R-AEA) in Carette and Staquet [11].

Equation Parameters REF TFA-1 TFA-2 EFA-1 EFA-2 R-0.4 R-AEA

Lognormal Vs,max 1782 1784 1666 1803 1538 2075 1682

Vs,min 66 67 85 64 83 66 44
Μ 1.996 2.173 2.257 2.072 2.043 1.966 2.094
Б 0.613 0.598 0.533 0.596 0.456 0.654 0.597
R2 0.9999 0.9998 0.9994 0.9999 0.9997 0.9992 0.9997
Modified Fredlund and Xing Vs,max 3907 2817 1539 2968 1741 95,670 8140
Vs,min 57 52 62 50 70 72 40
a 4.575 6.062 10.959 5.388 6.777 3.799 4.756
n 3.079 2.947 2.855 2.993 3.583 3.271 3.328
m 0.383 0.686 4.002 0.629 1.548 0.012 0.140
R2 0.9999 0.9999 0.9998 0.9999 0.9999 0.9996 0.9998

To solve the above challenges, new methods have been proposed to above-mentioned equations give satisfactory accuracy with high R2
determine the initial and final setting times, as elaborated below. The values of 0.992 to 0.999 (Table 3).
fitted Vs versus elapsed time curves with commonly seen S-shaped cu- Furthermore, it is recommended to calculate the first and second
mulative distribution equations, such as the Weibull cumulative eq. derivatives of the fitted cumulative equations. Take lognormal cumu-
[28] (Eq. 2), the lognormal cumulative equation [29] (Eq. 3), or the lative equations for example (Fig. 13), the time at the peak (tpeak′) of the
Gamma cumulative equation [30] (Eq. 4). The modified Fredlund and first derivative curve (Vs′), i.e. the inflection point of the original Eq. 3,
Xing [31] equation (Eq. 5) is also used because it's a parameter char- was used to correlate to the final setting time (tf) measured from the
acterizes the onset of the major increment of Vs from the initial gentle standard penetration resistance test:
slope (dormant period), and therefore can be used to estimate the initial
′ + 0.17
t f = 0.96tpeak (6)
setting time (see below for details).
α with R2 = 0.979 (Fig. 10). A similar procedure for modified Fredlund
−⎛ x ⎞
y=1−e ⎝β⎠ (2) and Xing equation gives:

1 (ln x − μ)2 ′ − 0.09

t f = 0.96tpeak (7)
−
y= e 2σ 2
xσ 2π (3) 2
with R = 0.982 (Fig. 10). Applying the same methods to the Vs evo-
1 lution curves reported by Carette and Staquet [11] yielded reasonable
y= γ (α, βx ) good fitting results, with R2 = 0.861. The fitting results shown here
Γ (α ) (4)
suggest that the proposed inflection point method is valid to estimate
m
1 the final setting time.
y = θs − θs ⎡ ⎤
⎢ ln(e + (x/ a)n ⎥ (5) The time (tpeak″) corresponding to the peak of second derivative (Vs″)
⎣ ⎦
of lognormal distribution correlates to the initial setting time (ti) de-
where α, β, μ, σ, a, n and m are fitting coefficients for Eqs. 2–5. termined from standard penetration resistance test as follows (Fig. 11):
The fitted curves are shown in Fig. 12, and the fitting parameters
and the coefficient of determination (R2) are listed in Table 3. All of the t i = 1.04t′′ peak + 1.05 (8)

9
J. Zhu et al. Cement and Concrete Research 106 (2018) 1–11

Table 5
Fitting coefficients and R2 values of the proposed correlation between initial and final setting times to and the fitted parameters in the Vs evolution methods.

Fitting method Correlation to ti/tf Data in this study Data from Carette and Staquet [11]

k b R2 k b R2

Cumulative lognormal tf = k ⋅ tpeak′ + b 0.96 0.17 0.979 1.37 −0.11 0.861

Modified Fredlund and Xing 0.96 −0.09 0.982 1.13 1.09 0.836
″
Cumulative lognormal ti = k ⋅ tpeak + b 1.04 1.05 0.950 1.39 1.92 0.742
Modified Fredlund and Xing 1.03 1.09 0.937 1.45 1.25 0.648
Modified Fredlund and Xing ti = k ⋅ a + b 0.70 −0.50 0.981 0.75 1.59 0.697

with R2 = 0.950. Similarly, for modified Fredlund and Xing equation setting time than the experimental method used in Carette and Staquet
we have [11]. With the non-destructive nature and reliable results, the proposed
bender element is a promising tool to determine initial and final setting
t i = 1.03t′′ peak + 1.09 (9)
times and to monitor early age characteristics of mortars as well as
2
with R = 0.937. Besides, the parameter a in modified Fredlund and other cementitious materials.
Xing equation correlates to ti as
Acknowledgement
t i = 0.70a − 0.50 (10)

with R2 = 0.981. The authors would like to thank the United States Department of
Applying the same approach to the measured shear wave velocity Transportation and the Center for Infrastructures Engineering Studies
versus elapsed time curves in Carette and Staquet [11] resulted in cor- for the financial support to Mr. Jianfeng Zhu for his master studies. The
relations shown in Fig. 11 with R2 values of 0.742 and 0.697 for the authors would also like to thank the One-Thousand-Young-Talents
lognormal and modified Fredlund and Xing equation, respectively Program of the Organization Department of the CPC Central Committee
(Table 4). It should be noted that one of the mixtures in the Carette and as well as the 100-Talents Program of Zhejiang University for their fi-
Staquet [11] evaluation was removed from the fitting of modified nancial support. This work is also partially sponsored by the National
Fredlund and Xing equation because the initial fluctuating segment of Natural Science Foundation of China (Award No.: 51779219). In ad-
the measured Vs evolution curve resulted in fitting disagreeable with dition, the Key Laboratory of Soft Soils and Geoenvironmental
the visual judgment (e.g., the two tangent line method) and gave an Engineering of the Ministry of Education is acknowledged.
unreasonably high a value. Two reasons accounting for the low R2
values are postulated. Firstly, the initial fluctuating segment of the References
measured Vs evolution curves by Carette and Staquet [11] caused a
large variation in initial setting time estimation (e.g. the R-0.4 sample); [1] V. Garnier, G. Corneloup, J.M. Sprauel, J.C. Perfumo, Setting time study of roller
compacted concrete by spectral analysis of transmitted ultrasonic signals, NDT&E
Secondly, the range of initial setting times was narrower than that Int. 28 (1995) 15–22.
measured in this study, causing a larger variation. That said, both tpeak″ [2] Z. Li, L. Xiao, X. Wei, Determination of concrete setting time using electrical re-
and parameter a from the shear wave velocity evolution monitoring sistivity measurement, J. Mater. Civ. Eng. 19 (2007) 423–427.
[3] Z. Ge, K. Wang, P.J. Sandberg, J.M. Ruiz, Characterization and performance pre-
method can give reasonable estimation of the initial setting time
diction of cement-based materials using a simple isothermal calorimeter, J. Adv.
(Table 5). The above results and analysis also suggest Vs evolution curve Concr. Technol. 7 (2009) 355–366.
measured by the BE method in this study can offer more reliable pre- [4] M.P. Hofmann, S.N. Nazhat, U. Gbureck, Real-time monitoring of the setting re-
dictions of both the initial and final setting times than that measured by action of Brushite-forming cement using isothermal differential scanning calori-
metry, J. Biomed. Mater. Res. 79B (2006) 360–364.
the method used in Carette and Staquet [11] does. [5] V. Rahhal, R. Talero, Calorimetry of Portland cement with silica fume, diatomite
and quartz additions, Constr. Build. Mater. 23 (2009) 3367–3374.
[6] J.P. Sandberg, S. Liberman, Monitoring and Evaluation of Cement Hydration by
11. Summary Semi-Adiabatic Field Calorimetry, Concrete Heat Development: Monitoring,
Prediction, and Management, vol. 241, Curran Associates. Inc., NY, 2007, pp.
In this study, a method is proposed for the use of the shear wave 13–24.
[7] N.A. Soliman, K.H. Khayat, M. Karray, A.F. Omran, Piezoelectric ring actuator
velocity (Vs) evolution curves of freshly cast mortar to estimate the technique to monitor early-age properties of cement-based materials, Cem. Concr.
initial and final setting times using a specially designed bender element Compos. 63 (2015) 84–95.
system. The traditional bender element testing system used in geo- [8] G. Zhang, J. Zhao, P. Wang, L. Xu, Effect of HEMC on the early hydration of
Portland cement highlighted by isothermal calorimetry, J. Therm. Anal. Calorim.
technical engineering was modified to obtain Vs versus elapsed time
119 (2015) 1833–1843.
relationship of the cement-based materials, whose stiffness increased [9] C.-W. Chung, P. Suraneni, J.S. Popovics, L.J. Struble, Setting time measurement
quickly at early age (from 0 to > 24 h). Details of the modifications of using ultrasonic wave reflection, ACI Mater. J. 101 (2012) 109–118.
[10] G. Trtnik, G. Turk, F. Kavčič, V.B. Bosiljkov, Possibilities of using the ultrasonic
the bender element system and their rationales were elaborated.
wave transmission method to estimate initial setting time of cement paste, Cem.
Cumulative distribution equations, such as the lognormal equation Concr. Res. 38 (2008) 1336–1342.
and modified Fredlund and Xing equation were used to fit the measured [11] J. Carette, S. Staquet, Monitoring the setting process of mortars by ultrasonic P and
Vs evolution curves determined in this study as well as data reported by S-wave transmission velocity measurement, Constr. Build. Mater. 94 (2015)
196–208.
Carette and Staquet [11] for mortars made with w/c of 0.4 to 0.5. The [12] S. Amziane, Setting time determination of cementitious materials based on mea-
test results were compared to the initial and final setting times obtained surements of the hydraulic pressure variations, Cem. Concr. Res. 36 (2006)
from the standard penetration resistance test. Test results indicated that 295–304.
[13] J. Silva, M. Azenha, A.G. Correia, C. Ferreira, Continuous stiffness assessment of
the time corresponding to the first inflection point of the Vs evolution cement-stabilised soils from early age, Geotechnique 63 (2013) 1419–1432.
curve can correlate well to the final setting time (R2 = 0.979). The [14] J. Silva, M. Azenha, A.G. Correia, J. Granja, Continuous monitoring of sand–cement
initial setting time was found to correlate well either to the peak time of stiffness starting from layer compaction with a resonant frequency-based method:
issues on mould geometry and sampling, Soils Found. 54 (2014) 56–66.
the second derivative of the Vs evolution curve or to the parameter a in [15] X. Kang, G.C. Kang, B. Bate, Measurement of stiffness anisotropy in kaolinite using
the modified Fredlund and Xing equation (R2 = 0.950 and 0.981, re- bender element tests in a floating wall consolidometer, Geotech. Test. J. 37 (2014)
spectively). Correlation results also suggested that the modified bender 1–16.
[16] J.M. Larrahondo, H. Choo, S.E. Burns, Laboratory-prepared iron oxide coatings on
element used in this study can yield a more reliable estimation of the

10
J. Zhu et al. Cement and Concrete Research 106 (2018) 1–11

sands: submicron-scale small-strain stiffness, Eng. Geol. 121 (2011) 7–17. bender elements, Soils Found. 49 (2009) 631–650.
[17] X. Kang, B. Bate, Shear wave velocity and its anisotropy of polymer modified high [23] J.-S. Lee, J.C. Santamarina, Bender elements: performance and signal interpreta-
volume class F fly ash-kaolinite mixtures, J. Geotech. Geoenviron. 142 (12) (2016), tion, J. Geotech. Geoenviron. 131 (2005) 1063–1070.
http://dx.doi.org/10.1061/(ASCE)GT.1943-5606.0001562, 04016068 ASCE [24] B. Bate, H. Choo, S.E. Burns, Dynamic properties of fine-grained soils engineered
(IF=2.464). with a controlled organic phase, Soil Dyn. Earthq. Eng. 53 (2013) 176–186.
[18] J. Zhu, Determination of Setting Times by Shear Wave Velocity Evolution in Fresh [25] J. Marjanovic, J. Germaine, Experimental study investigating the effects of setup
Mortar Using Bender Element Method, Department of Civil, Architectural, and conditions on bender element velocity results, Geotech. Test. J. 36 (2013) 187–197.
Environmental Engineering, Missouri University of Science and Technology, Rolla, [26] J.C. Santamarina, K.A. Klein, M.A. Fam, Soils and Waves: Particulate Materials
MO, USA, 2015. Behavior, Characterization and Process Monitoring, John Wiley & Sons Ltd, New
[19] B.M. Montoya, R. Gerhard, J.T. DeJong, D.W. Wilson, M.H. Weil, B.C. Martinez, York, NY, USA, 2001.
L. Pederson, Fabrication, operation, and health monitoring of bender elements for [27] C. Vipulanandan, V. Garas, Electrical resistivity, pulse velocity, and compressive
aggressive environments, Geotech. Test. J. 35 (2012). properties of carbon fiber-reinforced cement mortar, J. Mater. Civ. Eng. 20 (2008)
[20] X. Kang, G.-C. Kang, B. Bate, Closure to "discussion of "measurement of stiffness 93–101.
anisotropy in kaolinite using bender element tests in a floating wall [28] W. Weibull, A statistical distribution function of wide applicability, J. Appl. Mech.
Consolidometer" by X. Kang, G.-C. Kang, and B. Bate" by Coffman, Salazar and Trans. 18 (1951) 293–297.
Zhao, Geotech. Test. J. 39 (2016) 1079–1083. [29] L.H. Ahrens, The lognormal distribution of the elements (a fundamental law of
[21] S. Liu, J. Zhu, S. Seraj, R. Cano, M. Juenger, Monitoring setting and hardening geochemistry and its subsidiary), Geochim. Cosmochim. Acta 5 (1954) 49–73.
process of mortar and concrete using ultrasonic shear waves, Constr. Build. Mater. [30] P.G. Moschopoulos, The distribution of the sum of independent gamma random
72 (2014) 248–255. variables, Ann. Inst. Stat. Math. 37 (1984) 541–544.
[22] S. Yamashita, T. Kawaguchi, Y. Nakata, T. Mikami, T. Fujiwara, S. Shibuya, [31] D.G. Fredlund, A. Xing, Equations for the soil-water characteristic curve, Can.
Interpretation of international parallel test on the measurement of G max using Geotech. J. 31 (1994) 521.

11