Professional Documents
Culture Documents
net/publication/322537057
CITATIONS READS
13 883
5 authors, including:
Some of the authors of this publication are also working on these related projects:
Critical state behaviors and bounding surface model of rockfill material considering particle breakage View project
All content following this page was uploaded by Yang Xiao on 17 January 2018.
A R T I C L E I N F O A B S T R A C T
Keywords: Coral reefs and other calcareous deposits may experience various types of significant dynamic loading, such as
Calcareous sand those from waves and earthquakes. When submerged and subjected to earthquake loading, the potential for
Liquefaction mitigation liquefaction of calcareous deposits may cause a loss of human life and property; however, few studies have
Microbial induced calcite precipitation evaluated the liquefaction potential of calcareous sands relative to those conducted on silica sands. Accordingly,
Cyclic testing
it is critical to study the cyclic resistance of calcareous sands as well as methods to mitigate their liquefaction
Cyclic strength
potential. Microbial induced calcite precipitation (MICP) offers one such strategy that can be considered for
improving the cyclic resistance of calcareous sands, particularly for those applications below existing infra-
structure that would pose technical difficulties for traditional modes of ground improvement. This paper ex-
amines the effectiveness of MICP on the cyclic resistance of as a function of cementation solution (CS) content,
effective confining pressure, and cyclic stress ratio (CSR) through a cyclic triaxial test program. The generation
and accumulation of excess pore pressure and corresponding axial strains are compared across a range of treated
and untreated sands. This study shows that the liquefaction resistance of clean calcareous sand may be sig-
nificantly improved by the MICP treatment. Scanning electron microscope images are presented to help link the
improvement in cyclic response to the microstructural features of the microbial-induced calcite and bio-ce-
mented sand.
⁎
Corresponding author at: School of Civil Engineering, Chongqing University, Chongqing, China.
E-mail address: hhuxyanson@163.com (Y. Xiao).
https://doi.org/10.1016/j.soildyn.2018.01.008
Received 9 September 2017; Received in revised form 15 November 2017; Accepted 2 January 2018
0267-7261/ © 2018 Elsevier Ltd. All rights reserved.
P. Xiao et al. Soil Dynamics and Earthquake Engineering 107 (2018) 9–19
treating sand with MICP takes less time to improving liquefiable sand
Sand property Values
than colloidal silica grouting to achieve a similar liquefaction re-
sistance. Montoya et al. [40] studied the dynamic response of MICP- Specific Gravity (Gs) 2.79
treated siliceous sand of various strengths using centrifuge models and Minimun Void Ratio (emin) 1.22
Maximum Void Ratio (emax) 1.69
found that the pore pressure development, surface settlement, and the
D10 0.18
acceleration at the ground surface was lower than that of a comparable D30 0.28
loose sand. On the other hand, Sasaki and Kuwano [41] found that si- D50 0.36
liceous sand with a 30% clay fraction showed no increase in liquefac- D60 0.40
tion resistance following microbial precipitation of CaCO3. Coefficient of Uniformity (Cu) 2.22
Coefficient of Curvature (Cc) 1.09
There are far fewer studies that have evaluated how the magnitude
of MICP improvement and confining pressure has impacted liquefaction
resistance. In addition, the few studies that have focused on the cyclic other carbonate sands [7,42,43,46]. Based on the Unified Soil Classi-
response of marine soils like calcareous sand [42,43] have evaluated fication System [47], the sand is classified as poorly-graded, other
natural sands without MICP treatment. Calcareous sand is a biogenic properties such as the specific gravity, and maximum and minimum
material that shares many physico-chemical characteristics with cal- void ratios are shown in Table 1. The CaCO3 content in Yongxing Island
cium-based bio-cementation; accordingly, the potential for natural ef- sand was more than 90% prior to MICP treatment. Two sets of particle
ficiencies between the carbonate crystal-particle interface strength may size curves including the ranges in PSDs for most liquefiable and po-
be significant. For example, Khan et al. [44,45] conducted some initial tentially liquefiable soils [48] are compared to Yongxing Island sand in
investigations on the coral sand using different bacteria and observed Fig. 1. Based on the PSD, the potential for liquefaction of loose deposits
that unconfined compressive strengths in the range of 13–20 MPa were of this sand is high.
achievable.
The objective of this experimental laboratory investigation is to
determine the efficacy of MICP-treatment to improve the cyclic re- 2.2. Bacterial suspension and cementation solution (CS)
sistance of calcareous sand. In order to make sufficient comparison
against various treatments, the cyclic response of the untreated cal- The urease strain of Sporosarcina pasteurii was used as the active
careous sand is studied using a series of cyclic triaxial compression tests microbe for production of the calcite in the experiments conducted in
(CTCs). Then, a series of CTCs are conducted to compare the liquefac- this study. The previously frozen strain was activated in a plate
tion mitigation of different cementation solution (CS) contents, effec- medium, and then cultivated in sterilized growth media which con-
tive confining pressures, and cyclic stress ratios and to analyze the sisted of the following per-liter deionized water concentrations of:(1)
factors affecting mitigation. Scanning electron microscope (SEM) 20 g yeast extract, (2) 10 g NH4Cl, (3) 12 mg MnCl2·H20, (4) 24 mg
images are presented to help link the improvement in cyclic response to NiCl2·6H20 with a pH of 9.0, achieved using sodium hydroxide. Bacteria
the microstructural features of the microbial-induced calcite and bio- were grown in incubator shaker in 30 °C with about 36 h to reach an
cemented sand. optical density of 600 nm (OD600) of 0.8–1.0 (107 cells/mL). To remove
the abundant metabolic waste and odor, the harvested bacteria were
2. Experimental materials centrifuged at 4 °C at a speed of 4000 rpm for 15 min. Following con-
centration in the centrifuge, the resulting supernatant was removed and
2.1. Characteristics of Yongxing Sand replaced with 0.9% sodium chloride solution before the bacteria were
resuspended. The enzymatic activity of the final bacterial media was
The calcareous sands in this study were sourced from Yongxing about 1.4–2.0 mM urea/min. The cementation solution (CS) used in the
Island of the Xisha archipelago located in the South China Sea. The experiments was a mixture of 0.5 mol/L urea and 0.5 mol/L CaCl2.
particle size distribution (PSD) and a microscopic image of the
Yongxing Island sands are presented in Fig. 1. The medium to coarse,
shelly sand is characterized with widely varying grain shapes ranging
from platy to rod-like to ellipsoidal; these grain shapes can facilitate
significant interlock while exhibiting remarkably large void space like
10
P. Xiao et al. Soil Dynamics and Earthquake Engineering 107 (2018) 9–19
Fig. 2. Test setup schematics: (a) Picture of MICP treatment procedure; (b) Sketch of MICP treatment procedure. (a. Cementation solution/ Bacterial suspension; b. Peristaltic pump; c.
PVC split mold; d. Scouring pad; e. Tightening ring; f. Aspirating hole; g. Rubber membrane; h. Sand column; i. Rubber plug; j. Drainage valve; k. Effluent).
3. Experimental procedures adjusted to maintain a small head of CS at the top of the specimen.
When a given volume of CS had been pumped into the soil column, the
3.1. Sample preparation drainage valve was closed and 6 h of reaction time was allowed to
elapse, completing one CS cycle.
Several previous studies on the static triaxial response MICP-treated Additional cycles were conducted in order to complete the Ca2+
soils were conducted by precipitating the bio-cement directly in the reaction, which was determined by the measurement of pH and Ca2+
triaxial cell and apparatus [32,35]. In this study, specimens were pro- concentration in the effluent. The target volume of CS for different
duced by using a set of specially-designed specimen preparation devices treatment groups varied as shown in Table 2, and the measured change
to achieve batches of specimens. The procedure of preparing specimens in mass (Δm) of the oven-dried, treated sand was used to indicate the
is shown in Fig. 2. Fig. 2(a) depicts the MICP treatment procedure with global uniformity in treatment within a given test group.
eight soil columns being treated per pump, whereas Fig. 2(b) presents Following each treatment cycle, the Ca2+ concentration was mea-
the procedure schematically. As shown in Fig. 2(b), the test apparatus sured using the EDTA titration method described by Achal et al. [49].
consists of a polyvinyl chloride (PVC) split mold, rubber plug, rubber Upon completion of the reaction and following a rest period of ap-
membrane, scouring pad, and tightening ring. The PVC split mold has proximately 24 h, 300 mL of deionized water was circulated through
an internal diameter and height of 4 and 12.5 cm, respectively, and is the specimen to remove remaining nutrients. To facilitate comparison
transparent to allow visual inspection of the treatment process. to the treated specimens, the untreated clean sand specimens were
Specimens were prepared by air-pluviation into a split-mold lined prepared using the same general procedures excluding the treatment
with a stretched membrane under vacuum. This technique proved to be process. However, specimen U12 was prepared as a control specimen to
advantageous over other sample preparation techniques as it eliminated specifically to measure the changes in pH and Ca2+ concentration that
disturbance due to extrusion from the treatment column. The test sand could develop for numerous cycles of CS in the absence of bacterial
were went across a No. 10 (i.e., 2 mm) sieve to remove organic debris suspension, as described below.
(e.g., plant matter) and provide uniformity in the maximum grain size.
Then the sand were over-dried at 60 °C (the particle surface of calcar-
eous sand would get bark when over-dried over 100 °C). Following 3.3. Undrained cyclic triaxial tests
assembly of the treatment column and treatment (described below),
specimens of about 8 cm height and 3.9 cm diameter were produced for Undrained cyclic triaxial (UCT) tests were conducted on the pre-
cyclic triaxial testing. The air-pluviation procedures were selected to pared specimens of clean and MICP-treated calcareous sand. Prior to
produce a medium dense condition after pluviation (Table 2), and the removal of the split-mold, the top and bottom surface of each specimen
mean and coefficient of variation for the relative density of the speci- was leveled and the specimen placed on a filter paper-lined porous
mens were 47% and 5%, respectively. stone located at the base of the triaxial apparatus pedestal. Then the top
and bottom of the membrane was rolled up and down to cover the top
cap and pedestal and sealed with O-rings. A small vacuum pressure of
3.2. MICP treatment procedure
less than 20 kPa was maintained in order to preserve the integrity of the
specimen. The tests were carried out in general accordance with ASTM
Upon assembly of the sand column apparatus and specimen plu-
D5311 [50]. To aid saturation, samples were first flushed with CO2 and
viation, the specimen was flushed with deionized water and the drai-
de-aired water under an effective stress of 20 kPa, then backpressure
nage outlet was closed and raised up to the same level of the sample
saturation of 300 kPa was conducted in order to obtain a B-value of
surface to achieve equality of total head. Thereafter, the bacterial sus-
0.95 or greater. Specimens were loaded cyclically using sinusoidal de-
pension was pumped through the specimen at a rate of 80 mL/h to
viatoric stresses at a frequency of 1 Hz. Liquefaction (NL) is defined in
distribute the suspension throughout the pore volume. A detention time
this study as excess pore water pressure ratio (ru=Δu/σ′c ) greater than
of 6 h was provided to allow the bacteria to disperse and adhere to the
0.95. During these tests, axial stress, axial displacement, and pore water
soil grains.
pressure were monitored. Thirty-one UCTs were conducted, as shown in
Following detention of the bacterial suspension, several cycles of CS
Table 2, which summarizes the test conditions, including the ranges in
were pumped through the specimen as follows. Cementation solution
sand treatment, effective confining pressures, and cyclic stress ratio
was pumped into the soil sample at a rate of 40 mL/h with the effluent
(CSRs) applied during the undrained cyclic shear.
drainage valve opened slightly less than the influent to promote storage
of CS within the pore volume. The drainage valve was continuously
11
P. Xiao et al. Soil Dynamics and Earthquake Engineering 107 (2018) 9–19
Table 2
Properties of tested soil and test parameters.
Test no. Cementation solution (L) ρd (g/cm3) e0 Δm (g) Dr (%) B-value σ ′c (kPa) CSR NL
Group U
U1 0 1.132 1.463 / 48 0.98 50 0.167 312
U2 0 1.132 1.463 / 48 0.95 50 0.208 88
U3 0 1.129 1.470 / 47 0.97 50 0.250 41
U4 0 1.137 1.453 / 50 0.96 50 0.292 20
U5 0 1.134 1.460 / 49 0.97 100 0.146 460
U6 0 1.120 1.490 / 42 0.97 100 0.167 217
U7 0 1.134 1.460 / 49 0.96 100 0.188 103
U8 0 1.128 1.474 / 46 0.97 100 0.250 31
U9 0 1.120 1.490 / 42 0.97 200 0.167 125
U10 0 1.122 1.487 / 43 0.95 200 0.208 40
U11 0 1.137 1.453 / 50 0.97 200 0.250 11
U12a 0.6 1.132 1.463 / 48 / / / /
Group TA
TA1 0.2 1.126 1.477 5.64 45 0.97 100 0.188 435
TA2 0.2 1.137 1.453 5.83 50 0.97 100 0.208 137
TA3 0.2 1.132 1.463 5.64 48 0.95 100 0.250 72
TA4 0.2 1.126 1.477 5.75 45 0.98 100 0.292 21
Group TB
TB1 0.4 1.123 1.484 11.59 44 0.97 50 0.250 135
TB2 0.4 1.129 1.470 12.11 47 0.96 50 0.292 75
TB3 0.4 1.132 1.463 11.90 48 0.97 50 0.333 49
TB4 0.4 1.128 1.474 12.11 46 0.97 50 0.375 23
TB5 0.4 1.137 1.453 11.72 50 0.95 100 0.208 388
TB6 0.4 1.128 1.474 11.75 46 0.97 100 0.250 118
TB7 0.4 1.138 1.452 11.59 51 0.97 100 0.292 54
TB8 0.4 1.132 1.463 12.28 48 0.97 100 0.333 22
TB9 0.4 1.132 1.463 11.85 48 0.96 200 0.208 331
TB10 0.4 1.123 1.484 11.91 44 0.97 200 0.250 78
TB11 0.4 1.123 1.484 12.05 44 0.97 200 0.292 36
TB12 0.4 1.132 1.463 12.03 48 0.95 200 0.333 18
Group TC
TC1 0.6 1.128 1.474 17.29 46 0.97 100 0.229 250
TC2 0.6 1.137 1.453 17.83 50 0.98 100 0.250 135
TC3 0.6 1.132 1.463 17.79 48 0.97 100 0.292 60
TC4 0.6 1.128 1.474 17.35 46 0.98 100 0.333 38
Note:.
ρd =Initial dry density; e0 =Void ratio; σ ′c =Initial effective cell pressure; CSR=Cyclic stress ratio; NL =The number of cycles to liquefaction.
a
U12 conducted to illustrate changes in chemical composition from the specimen preparation techniques, only.
3.4. Scanning Electron Microscopy (SEM) imaging which was not dosed with bacterial solution prior to CS treatment.
Fig. 3(a) shows that treatment with the CS within specimens dosed with
Microscopic imaging was conducted to link the experimental cyclic the bacterial suspension produced a marked increase in pH after the
response of MICP-treated sand to the physical soil fabric produced using first cycle as a result of ammonium ion production from urea hydro-
bio-cementation. Imaging was conducted using a Zeiss Auriga-SEM; a lysis. Thereafter, the pH tended to decrease slightly. As shown in the
relatively low accelerating voltage of 5 kV was selected to ensure high Fig. 3(a), the pH of the treated specimens remained greater than the
resolution (but relatively less bright) images. Prior to conducting the control specimen U12. This observation is generally consistent with the
SEM imaging, the samples were flushed with deionized water and dried observations reported by others [52,53], and indicated that pH can
for 24 h. Thereafter, microscopic specimens were produced by adhering indicate the activity of urease during MICP treatment.
a small amount of sand to the specimen plate and then coating with Similarly, the Ca2+ concentration was measured following each
gold under vacuum conditions. Described below, the microscopic in- cycle, and was used to gage the end of the calcite precipitation reaction,
vestigation facilitates the connection of the crystal formation and indicated by the near-zero magnitude of Ca2+ (n.b., Ca2+ cannot be
bonding between the sand particles and CaCO3 crystals to the cyclic fully-consumed in the chemical reaction due to the volatilization of
response. urea; see Eq. (1)). As described earlier, the change in mass of the treated
sand (Table 2) was used as an index of the global magnitude of CaCO3
content produced as an alternate approach for the dissolution approach
4. Test results and discussion [54]. However, in the current work, the dissolution approach cannot be
used as it would not distinguish between the natural sand CaCO3 and
4.1. Physico-Chemical Characterization of the treated and untreated the MICP-induced CaCO3. The coefficient of variation in Δm in Table 2
samples ranges from 1.6% to 1.8% for each group, which indicates a sufficient
level of repeatability in the treatment regime adopted.
In the process of MICP treatment, the pH of the aqueous solution
during precipitation can be used as a key indicator of CaCO3 pre-
cipitation. The pH increases during urea hydrolysis via the enzyme 4.2. Undrained cyclic resistance
urease; the local rise in pH often causes the bacteria to serve as nu-
cleation sites for crystallization [51]. Fig. 3 presents the variation in pH The effectiveness of MICP-treatment for improving the undrained
and Ca2+ concentrations of the effluent taken from specimens following cyclic resistance was evaluated by comparing the undrained cyclic re-
each CS treatment cycle and compares the response to specimen U12, sponse of the treated and clean sand specimens under stress-controlled
12
P. Xiao et al. Soil Dynamics and Earthquake Engineering 107 (2018) 9–19
Fig. 3. Variation of the pH and Ca2+ over time during the MICP treatment: (a) pH vs. treatment cycles; (b) Ca2+ vs. treatment cycles.
Fig. 4. The typical undrained cyclic response of clean calcareous sand sheared with σ′c = 100 kPa and CSR=0.250 (Specimen U8): (a) excess pore pressure response, (b) axial strain, (c)
deviatoric stress, and (d) cyclic stress path.
conditions. Cyclic stress ratios, defined as the ratio of the applied cyclic mean excess pore pressure (Um) that is non-recoverable from cycle to
shear stress and the effective confining pressure were specified for each cycle. A red line in Figs. 4(a) and 5(a) indicates the point at which
UCT test (Table 2) to draw general trends between the number of cycles initial liquefaction is attained. For the clean untreated sand, initial li-
to liquefaction, NL, and the magnitude of shear stresses. The response of quefaction occurred after 31 cycles (i.e., NL=31). In contrast, the
typical untreated and treated specimens (U8 and TC2, respectively) treated specimen did not liquefy until NL=135.
sheared in the same condition (σ′c = 100kPa and CSR=0.250) are pre- As shown in Fig. 4, the untreated specimen exhibited a contractive
sented in Figs. 4 and 5, respectively. The two Figures depict the de- response during the first quarter cycle of loading. The initial axial
velopment of excess pore water pressure ratio and axial strain (εa ) with strains were very small for the first several cycles and experienced a
the number of loading cycles, deviatoric stress-strain curve, and cyclic slow increase in the accumulation of strain. The axial strain was −2.1%
effective stress path. Figs. 4(a) and 5(a) show that the excess pore at NL=31, and increased to 4.6% by N=37. In contrast, the deforma-
pressure generated during cyclic loading can be separated into two tion characteristics of the MICP-treated specimen shown in Fig. 5(b)
components: (1) the component associated with the loading cycle (Ucyc) depart significantly from the untreated sand. For example, at N=31,
and is generated and dissipated during each loading cycle; and (b) the the axial strain was 0.1%, and achieved liquefaction at −1.8% strain at
13
P. Xiao et al. Soil Dynamics and Earthquake Engineering 107 (2018) 9–19
Fig. 5. The undrained cyclic response of MICP-treated calcareous sand sheared with σ′c = 100kPa and CSR=0.250 (Specimen TC2): (a) excess pore pressure response, (b) axial strain, (c)
deviatoric stress, and (d) cyclic effective stress path.
14
P. Xiao et al. Soil Dynamics and Earthquake Engineering 107 (2018) 9–19
Fig. 8. Development of excess pore pressure with number of cycles for untreated and
treated calcareous sand under 100 kPa effective confining pressure and CSR=0.25.
15
P. Xiao et al. Soil Dynamics and Earthquake Engineering 107 (2018) 9–19
16
P. Xiao et al. Soil Dynamics and Earthquake Engineering 107 (2018) 9–19
17
P. Xiao et al. Soil Dynamics and Earthquake Engineering 107 (2018) 9–19
18
P. Xiao et al. Soil Dynamics and Earthquake Engineering 107 (2018) 9–19
[81] Somani RS, Patel KS, Mehta AR, Jasra RV. Examination of the polymorphs and soils. Int J Geomech 2017;17(7):04016150.
particle size of calcium carbonate precipitated using still effluent (i.e., CaCl2 + [83] Xiao Y, Sun Y, Liu H, Xiang J, Ma Q, Long L. Model predictions for behaviors of
NaCl Solution) of soda ash manufacturing process. Ind Eng Chem Res sand-nonplastic-fines mixturesusing equivalent-skeleton void-ratio state index. Sci
2006;45(15):5223–30. China Technol Sci 2017;60(6):878–92.
[82] Xiao Y, Sun Y, Yin F, Liu H, Xiang J. Constitutive modeling for transparent granular
19