Arab J Sci Eng

DOI 10.1007/s13369-017-2655-3

RESEARCH ARTICLE - CIVIL ENGINEERING

Effect of Glass Powder Added Grout for Deep Mixing of Marginal

Sand with Clay
Hanifi Canakci1 · Hamza Güllü1 · Mohanad Isam Kwana Dwle1

Received: 14 January 2017 / Accepted: 8 June 2017

© King Fahd University of Petroleum & Minerals 2017

Abstract The deep mixing has recently become a pop- of 3% glass powder similar to the response trend of UCS,
ular technique for improvement of weak or problematic while their performances are higher due to 4% clay content
soils in order to increase bearing capacity and reduce settle- in contrast to the trend of UCS. As a consequence, this study
ment. Despite various successful applications in construction demonstrates the use of glass powder added grout to be ben-
projects, specifically improvement of marginal sand (i.e., eficial for deep mixing of marginal sand, in which 20% clay
loose poorly graded sand) using new materials in the view- inclusion contributes more.
point of strength development is still open for the attempt
of deep mixing regarding the binder of grout, curing time Keywords Deep mixing · Grout · Glass powder · Cement ·
and clay content. Thus, in this paper, effect of glass powder Sand · Clay
added to cement-based slurry grout in different proportions
(0, 3, 6, 9% by dry weight of binder) has been experimentally
investigated for deep mixing of marginal sand with various 1 Introduction
clay contents (4, 8, 20%). An experimental program has been
developed for this purpose conducting the Vicat, unconfined The deep mixing is one of the ground improvement tech-
compressive strength (UCS) and ultrasonic pulse velocity niques that have been successfully applied for obtaining
(UPV) tests for the performances of soilcrete samples. From soil–cement columns for weak or problematic soils (i.e.,
the testing results, it is found that: (i) The glass powder is loose poorly graded sands, soft or expanded clays, organic
not able to accelerate the setting times of grout, (ii) the bulk soils) in the quality of increase in bearing capacity and
density does not significantly change with the glass pow- decrease in settlement [1]. In the deep mixing technique,
der, clay content and curing time, (iii) the most performed the weak soil is mixed with cementitious materials to obtain
UCS value is obtained due to the addition of 3% glass pow- a soil–binder column to enhance the strength and compress-
der at the 28-day curing for the soilcrete samples of 20% ibility characteristics [2]. The deep mixing columns due to
clay content of sand, while all UCS magnitudes obtained this technique are formed due to penetrating of rotary augers
for the soilcrete samples are considered acceptable for the into the soil and then mixing slurry or grout with in situ
construction of soil–cement column, (iv) the elastic modulus soil [3]. Throughout the mixing process, the deep mixing
correlates well with the UCS values (R ≥ 0.83) at the major- technique uses rotating mixer shafts, paddles or jets that pen-
ity of soilcrete samples and (v) the UPV values of soilcrete etrate into the soil while injecting and mixing the cement
samples at 28-day curing perform better due to the addition or other additives [4]. Generally, the mixing in this tech-
nique has been conducted by wet and dry methods dependent
upon water content of natural soil. It can be said that the dry
method is more suitable when the water content of natural
B Hamza Güllü soil is high. In the wet method, cementitious slurry is injected
[email protected]
through large diameter to a specified depth to produce the
1 Department of Civil Engineering, University of Gaziantep, soil–binder column. On the other hand, the mixing tool is
27310 Gaziantep, Turkey rotated into the natural soil applying the dry cementitious

123
 Arab J Sci Eng

materials pneumatically injected and blended with the soil to (ii) binder type and proportions, (iii) binder-to-water ratio,
obtain the soil–binder column in the dry method. Through- (iv) mixing the soil–binder and preparation technique of
out the reactions of in situ soil with the soil–cement columns, specimen and (v) curing conditions influence the laboratory
the strength development is obtained along time by these two behavior of deep mixing treated soils. However, it should
methods to improve the weak soil. While the dry mix method be noted that significance of the variables could be varied
is generally considered less expensive, the strength of its final depending upon the site conditions and project requirements.
product is also considered less than the one from the wet mix If strength of weak ground is the main issue to enhance the
[3,5]. ground characteristics, the deep mixing is mostly performed
Through the laboratory and field tests, it is reported [6] that by use of cement [2,17]. It is well known that the cement
a review of significant studies indicates effectiveness of the is composed of calcium silicates and calcium aluminates,
deep mixing method for the soils that require a treatment. The and when it is combined with water, hydration occurs for
performances on the physical integrity, strength and leach- obtaining the cementing compounds (i.e., calcium silicate
ability criteria for obtaining the soil–cement columns due hydrate, calcium aluminate hydrate and calcium hydroxide).
to a laboratory-scaled drilling auger were investigated for Due to these cementing compounds, both the granular and
sandy soil by the past study [7]. Effect of the soil mixing fine-grained soils are successfully stabilized with the use
technique to the soil–cement mixes was examined for the of cement. However, it is important to emphasize that par-
stratified sand by the past work [3] through the investigation ticularly for enhancement of fine-grained soils, pozzolanic
of the influences due to the thickness of soil layers, natu- reaction occurred between the calcium hydroxide released
ral moisture content of soil on the unconfined compressive during hydration and soil alumina, and soil silica becomes
strength, leachability, grout type, mixing mode, installation prominent rather than the hydration [18,19]. Although the
procedure, auger design, curing period, etc. The laboratory cement is the most frequently adopted binder in the deep
investigations of deep mixing conducted by several studies mixing process, the cement also including some admixtures
[8,9] show that there is strong dependency of engineering (or additives) (i.e., fly ash, lime, bottom ash, blast furnace
behavior of the treated soil on the organic content in the slag) in the binder has become popular in recent for ground
soil. During the examination of the deep mixing process improvement [20].
of lime–cement columns at a test site, it was found that The glass powder is a very recent material of admixture
strength of formed soil–binder columns increased at the var- that could be used together with cement as a part of the
ious installation processes [10]. A novel investigation [11] binder for modification of the engineering characteristics of
was carried for marine clay on the strength (unconfined soil based on the findings obtained in the previous efforts
compressive strength) performances obtained between the [21–23]. The past studies [21–23] show that powder form
cement deep mixing and dry jet mixing. Effect of deep soil of waste glass is suitable for stabilization in the viewpoint
mixing treatment method for reducing heave movements of economy when replaced with the cement and provides
of underlying expansive soils was investigated by the past friendly environment due to the waste disposal. It is reported
work [12] in the laboratory conditions through the consider- [22] that the glass powder addition is able to increase the
ation of several binder types for selection of the appropriate maximum dry density and to decrease the optimum moisture
binders for the field use of deep mixing. It is reported that content and Atterberg limits of soft soil. Similar findings have
the soil–cement technique has been used successfully as a been obtained for the soft soil studied by Nuruzzaman and
base layer for shallow foundations, slope protection for earth Hossain [21]. Canakci et al. [23] investigated modification
dams and prevention for sand liquefaction [13,14]. The bear- of clay due to the addition of glass powder (0, 3, 6, 9 and
ing capacity and failure mode of the group of deep mixing 12% by dry weight of the clay) and found that the addition of
columns were assessed for soft soil by means of instrumen- glass powder into clay had a significant effect on the strength
tation through a produced physical model in the past study and consistency properties of the clay. The glass powder has
[15]. The literature review does not only present the fac- also been encountered in the applications of concrete technol-
tors that affect the performance of treated soils, but also ogy within the binder (i.e., cement + glass powder) together
encourages the engineers for a comprehensive understand- with cement [24,25]. It is reported that the glass powder is
ing of the deep mixing parameters by further investigations. found successful during the cement hydration by modeling
It is understood from the literature surveying that majority of the degree of hydration of cement paste with the glass pow-
deep mixing works has been performed for enhancement of der [24]. Despite the potential ability of glass powder for the
soft soil, while the studies are few specifically for marginal ground improvement reported in past works, the literature
sands (i.e., loose poorly graded sand) improved by this tech- surveying indicates that there is a lack of effort that uses the
nique. glass powder added to the cement slurry grout for the pur-
Based on the findings in previous works, it is reported pose of deep mixing of marginal sand (loose poorly graded
[16] that several factors that primarily include (i) soil type, sand).

123
Arab J Sci Eng

This article aims to experimentally investigate the effect preparation and the employed testing methods for measur-
of glass powder added to the cement-based slurry grout for ing the performances of samples could be beneficial for the
improvement of the marginal sand (i.e., loose poorly graded experimenters in practice for deep mixing.
sand) through the purpose of deep mixing technique. The
sand was assumed to have different clay contents in situ
that can be identified by the classifications: (i) poorly graded 2 Experimental Study
sand (SP) (4% clay), (ii) poorly graded sand with clay (SP-
SC) (8% clay content) and (iii) clayey sand (SC) (20% clay The experimental study investigates performances of the
content). It is the authors’ knowledge from the literature glass powder added grout for deep mixing of loose sand (SP,
surveying that the influences of the varying clay contents SP-SC, SC) with different clay contents. The Vicat, uncon-
included in the sand attempted in this paper have been stud- fined compressive strength and ultrasonic pulse velocity tests
ied very few for the deep mixing purpose as well as the grout have primarily been conducted for the performance measure-
performances [26]. The clay content plays a major impact on ments. The bulk density and elastic modulus of specimens
the mechanical and physical properties of soilcrete samples. have also been estimated during the testings. Except the Vicat
Usually, the clay and peat soils provide the lowest strengths test, all specimens have been performed by three replicate
(2–8 MPa), while the best responses are obtained with sandy testing, and their average values have been taken into account.
and gravelly soils (12–19 MPa) [27,28]. Although the low
soil particle sizes considerably limit the mechanical prop-
erties of the soilcrete samples, it is found that 25% clay 2.1 Experimental Program
content in sandy soil results in the best strength for the cement
dosages lower or equal to 210 kg/m3 of soil [29]. On the other The experimental program of the study is presented in
hand, the strength increases with the clay percentage between Table 1. The water/binder ratio, percentage of binders,
5 and 15%, resulting in the best strength due to the 15% clay quantity of binder and water in grout, USCS (unified soil clas-
in another study [26]. By the effort given in this study on sification system) identification of sand to be treated with clay
the investigation of effect of glass powder added to the grout contents and quantities of sand and clay have been included
for the enhancement of marginal sand via deep mixing, it in the experimental program (Table 1).
is believed that the experimental program developed in this The sand to be treated in situ was assumed to have the
present study with the inclusion (i.e., glass powder, cement, different clay contents of 20, 8 and 4%. In accordance with
sand, clay) proportions, the applied procedure for sample the USCS, the sand deposit including clay to be treated is,
respectively, classified as: (i) SC (clayey sand), (ii) SP-SC

Table 1 Experimental testing program employed in the study

Mix ID Grout Soil
w/b Binder (%) Water Binder quantity, gr USCS classification Clay con- Sand, gr Clay, gr
quantity, tent (%)
gr

PC GP w PC GP S C

M1 1.2 100 0 156 130 0 SC (clayey sand) 20 571.2 142.8

M2 1.2 97 3 156 126.1 3.9 SC (clayey sand) 20 571.2 142.8
M3 1.2 94 6 156 122.2 7.8 SC (clayey sand) 20 571.2 142.8
M4 1.2 91 9 156 118.3 11.7 SC (clayey sand) 20 571.2 142.8
M5 1.2 100 0 156 130 0 SP-SC (poorly graded sand with clay) 8 656.88 57.12
M6 1.2 97 3 156 126.1 3.9 SP-SC (poorly graded sand with clay) 8 656.88 57.12
M7 1.2 94 6 156 122.2 7.8 SP-SC (poorly graded sand with clay) 8 656.88 57.12
M8 1.2 91 9 156 118.3 11.7 SP-SC (poorly graded sand with clay) 8 656.88 57.12
M9 1.2 100 0 156 130 0 SP (poorly graded sand) 4 685.44 28.56
M10 1.2 97 3 156 126.1 3.9 SP (poorly graded sand) 4 685.44 28.56
M11 1.2 94 6 156 122.2 7.8 SP (poorly graded sand) 4 685.44 28.56
M12 1.2 91 9 156 118.3 11.7 SP (poorly graded sand) 4 685.44 28.56
w/b = water/binder, binder = cement + glass powder, PC = cement, GP = glass powder, S = sand, C = clay, USCS unified soil classification
system

123
 Arab J Sci Eng

decisions of these proportions were strongly benefited from

the previous studies [23,30]. It is found in the previous study
[30] that 10% addition of glass powder contributes higher
strength than the remaining proportions (20, 30 and 40%).
The weight of binder to the soil in this present study for
all testing was taken as 18% in accordance with the previ-
ous suggestion [31], in which it is reported that it is better to
use the binder less than 20% of the soil in the soil–cement
columns for deep mixing. This could be considered as an eco-
nomical proportion for the cement usage due to the ground
improvement as well as the strength enhancement for deep
mixing.

2.2 Materials

The cement (PC) for deep mixing due to the binder com-
position of grout used in this study is an ordinary Portland
cement with the type of CEM 1-42.5R of industrial origin in
accordance with ASTM [32]. Some chemical and physical
properties of cement are given in Table 2. As for the glass
powder in the binder (i.e., cement + glass powder) added to
the grout, it was obtained from the waste soda green glass
bottles collected from the environment (stores, industry) in
Gaziantep (Turkey). For use in the grout, the green glass bot-
tle was washed to remove unwanted materials like corks and
labels and then dried to produce powder form. The glass pow-
der was produced by crushing the green glass bottles using
the manual treadmill until obtaining the required sizes. The
particle sizes of glass powder used in this study were pass-
ing the sieve size of 150 µm in accordance with the previous
study [33]. Some proposed values of chemical and physi-
Fig. 1 Trial samples prepared at water/binder = 1 and water/binder = cal properties of the glass powder [34] used in the study are
1.2
presented in Table 3.
As regards the soil used to be treated, the sand (S) included
in the soil deposit is the poorly graded sand and the clay
(poorly graded sand with clay) and (iii) SP (poorly graded
sand).
As for the grout, the water/binder ratio was employed as Table 2 Some chemical and physical properties of cement used in the
1.2. This is in accordance with the previous study [6], in study
which it is reported that among three w/b ratios (0.8, 1 and Constituent Value
1.3) used for deep mixing, the best one is obtained as w/b
= 1 regarding the performance of unconfined compressive CaO (%) 62.12
strength. However, trial of the w/b = 1 for preparing the SiO2 (%) 19.69
grout mixtures in this present study resulted in the voids in Al2 O3 (%) 5.16
the samples that can be visually seen in the soilcrete specimen Fe2 O3 (%) 2.88
(Fig. 1). On the other hand, the water-to-binder ratio of 1.2 MgO (%) 1.17
was found to produce the grout to be more homogeneous SO3 (%) 2.63
and workable providing with less voids. Thus, the water- K2 O (%) 0.88
to-binder ratio of 1.2 was decided to be appropriate for the Na2 O (%) 0.17
study. In concern with the glass powder addition to the grout, Loss on ignition (%) 2.99
the glass powder as an admixture in the binder (i.e., cement Specific gravity 3.15
+ glass powder) has been tested for the proportions 0, 3, 6 Blaine Fineness (m2 /kg) 394
and 9% by dry weight of binder, as shown in Table 1. The

123
Arab J Sci Eng

Table 3 Some proposed chemical and physical values for the glass 100
powder

Percentage passing (%)

Constituent Value 80

Silica (SiO2 ) 74%

60
Sodium oxide (Na2 O) 13%
Lime (CaO) 10.50%
Alumina (Al2 O3 ) 1.30% 40

Another components accumulate 1.20%

Clay
Density at 20◦ (g/cm3 ) 2.52 20

Young’s modulus at 20◦ (GPa) 72

Sand
0
0.01 0.1 1 10
Table 4 Some index and physical properties of soil (sand, clay) used Particle size ( mm )
in the study
Fig. 2 Particle size distribution of sand and clay
Property Sand Clay

Soil classification (USCS) SP CL

Specific gravity (GS ) 2.667 2.77 Finally, in order to obtain the testing specimens, the mix-
D10 (mm) 0.24 – tures were put into the splitting molds (with the height of
D30 (mm) 0.59 – 110 mm and the height to diameter ratio of 2:1) with the
D60 (mm) 1.8 – plastic collars around and sealed thick plastic bottom. Fol-
Cu 7.5 – lowing the recommendations of Bhadriraju et al. [16], the
Cc 0.806 –
mixtures in the molds were compacted in three equal layers
using 5 mm diameter of rod by 30-blow count that covers
Liquid limit (LL) – 41
entire surface of the specimen to remove the entrapped air
Plastic limit (PL) – 25
bubbles, and then, the specimens were allowed to leave 24
Plasticity index (PI) – 16
hr in the molds (Fig. 3c). After 24 hrs, the specimens were
Swelling (%) – 3.58
extracted from the molds and kept for the curing times of 7,
14 and 28 days in the water bath in the control room of labora-
tory (by the temperature of 20◦ ± 3◦ and relative humidity of
(C) is the low plasticity clay in accordance with the USCS. 55–70%), prior to the testing performances (i.e., unconfined
Some index and physical properties of soil (sand and clay) are compressive strength, ultrasonic pulse velocity, bulk density,
given in Table 4. The particle size distributions of sand and elastic modulus) (Fig. 3d). The curing times were selected in
clay are illustrated in Fig. 2. The sand particles between the accordance with the previous work [37] suggested for deep
sieves 0.15 and 2 mm were used for all testings. As mentioned mixing. The sample preparation procedure of mixtures men-
earlier, the clay particles (passing the sieve 0.425 mm) were tioned above was also same for the Vicat test. However, it
added to the sand at the proportions of 20, 8 and 4% by dry should be noted that immediately after preparation of the
weight of soil (sand + clay) in order to artificially obtain mixtures, the Vicat tests were carried out for measuring the
the soil deposit (sand + clay) that was assumed to represent setting times.
the field, respectively, as SC (i.e., 20% clay addition), SP-SC As for the testing methods, the Vicat tests for finding the
(i.e., 8% clay addition) and SP (4% clay addition). initial and final setting times of the mixture pastes have been
performed only for the binder of grout mixtures (i.e., cement
2.3 Sample Preparation and Testing Methods + glass powder) in accordance with ASTM [38]. Following
this procedure, it is measured that the grout reaches its initial
The procedure given in the previous guides [16,35,36] was setting time when the needle becomes 5mm above the bottom
followed for preparing the sample mixtures in this research of the plate. The final setting time is estimated when no any
(Fig. 3). First, the cement, glass powder and water were indent left due to the cylindrical needle.
mixed together for 1 min to obtain the grout. Then, the grout In regard to the unconfined compressive strength (UCS)
was added to the sand and clay. After that, the grout and tests of the specimens (for the curing times of 7, 14 and
soil (sand and clay) were mixed together using a laboratory 28 days), they have carried out following the procedure given
mixer with 5 L capacity for 10 min (Fig. 3a, b). In order by ASTM [39]. The compression machine used for the UCS
to obtain the sample mixtures to be homogeneous, the mix- tests was the one with the mechanical control type having
ing velocity was allowed to be 280 rpm for the mixtures. the capacity of 50 kN. All the UCS testings were performed

123
 Arab J Sci Eng

Fig. 3 Sample preparation: a materials (glass powder, cement, sand, clay), b mixing the grout (cement and glass powder) and soil, c samples in
mold, d soilcrete samples for curing

under the constant strain rate of 1%/min. The UCS response UCS tests become the main effort of experimental study in
is either the maximum axial stress or the stress at 5% axial order to understand the performances of the grout with glass
strain, whichever occurs first during the performance of test- powder for deep mixing of sand with clay.
ing. From the stress–strain curves of the specimens due to the Finally, the ultrasonic pulse velocity (UPV) tests have
UCS tests, the elastic modulus (Es) of the specimens has been been carried out for the specimens at the 7, 14 and 28
estimated regarding the secant modulus. It is reported [40] curing days, in accordance with ASTM [42]. A PROCEQ
that the UCS values could be considered as the key parame- type apparatus measuring in a wide range of interval from
ter for the quality measurement of deep mixing. Some UCS 24 to 500 kHz was used for the UPV testing. Following
ranges of treated soils (improved soils) dependent upon the the procedure [42], the longitudinal pulse wave velocity of
binder quantity and different soil types have been proposed the cylindrical specimens is determined by the longitudinal
in literature [5,41] (Fig. 4). In this present investigation, the vibration pulses applied in contact with the specimen surface.

123
Arab J Sci Eng

Fig. 4 Range of compressive strength of improved soils dependent upon the binder dosage (modified from Abbey et al. [5])

Table 5 Classification of UPV [43] 300

initial final
Class UPV (m/s) Definition 250
Setting Time (min)

1 <2500 Very low velocity 200

2 2500–3500 Low velocity
150
3 3500–4000 Middle velocity
4 4000–5000 High velocity 100

5 >5000 Very high velocity 50

0
M1-0%GP

M2-3%GP

M3-6%GP

M4-9%GP
After traveling through the specimen, the pulses are received
in measurements of the pulse velocity electronically. The
UPV test can be considered the non-destructive test method
to understand the material homogeneity. The magnitude of Fig. 5 The initial and final setting times of grout paste (i.e., cement +
glass powder)
measurements can be interpreted by some categorizations
given in past work [43] (Table 5).
addition of glass powder may be attributed to the particle
size of glass powder that is larger than the particle size of
3 Results and Discussion cement. It can be said that the larger particle sizes of glass
powder result in the setting times longer. Furthermore, the
All the results obtained for the setting times (Fig. 5), bulk den- workability of paste due to the glass powder and cement
sity (Fig. 6), unconfined compressive strength (Fig. 7), elastic could be considered on the influence of the setting times in
modulus (Fig. 8), elastic modulus versus strength (Fig. 9) and the view of concrete technology. It is reported in the past
ultrasonic pulse velocity (Fig. 10) are illustrated in Figs. 5, study [44] that the workability of mix with glass powder is
6, 7, 8, 9 and 10. found better than the one without the glass powder.

3.1 Setting Times 3.2 Bulk Density

For the setting times, due to the Vicat test (Fig. 5), it is seen For the bulk density of soilcrete samples obtained for the
that the setting times (the initial and final setting times) of average values of 7-, 14- and 28-day curing times (Fig. 6),
grout paste proportionally increase with the increased glass it is appeared that there is no significant change in the bulk
powder addition. The increase in setting times due to the density due to the glass powder addition as well as the clay

123
 Arab J Sci Eng

25 4%Clay (SP)
20%Clay (SC) 8%Clay (SP-SC)
Bulk density (kN/m3)

20

15

10

0
M1-0%GP

M2-3%GP

M3-6%GP

M4-9%GP

M5-0%GP

M6-3%GP

M7-6%GP

M8-9%GP

M9-0%GP

M10-3%GP

M11-6%GP

M12-9%GP
Fig. 6 The bulk density of soilcrete samples

12000
8%Clay (SP-SC) 4%Clay (SP)
10000
UCS (kPa)

8000

6000
4000 20%Clay (SC)

2000
7-day 14-day 28-day
0
Mix1-0%GP

Mix2-3%GP

Mix3-6%GP

Mix4-9%GP

Mix5-0%GP

Mix6-3%GP

Mix7-6%GP

Mix8-9%GP

Mix9-0%GP

Mix10-3%GP

Mix11-6%GP

Mix12-9%GP

Fig. 7 The UCS performances of soilcrete samples

800
20%Clay (SC) 8%Clay (SP-SC) 4%Clay (SP)
Elastic modulus (MPa))

600

400

200 Fig. 9 The correlations of elastic modulus versus unconfined compres-

7-day 14-day 28-day sive strength for soilcrete samples (Es: elastic modulus)
0
Mix10-3%GP

Mix11-6%GP

Mix12-9%GP
Mix1-0%GP

Mix2-3%GP

Mix3-6%GP

Mix4-9%GP

Mix5-0%GP

Mix6-3%GP

Mix7-6%GP

Mix8-9%GP

Mix9-0%GP

compared to the native cement (i.e., 0% glass powder) in this

study, they could be assessed as favorable for the strength
considerations.
Fig. 8 The elastic modulus of soilcrete samples
3.3 Unconfined Compressive Strength

content of sand. On the other hand, it is found in the past Regarding the performances of unconfined compressive
study [45] that the density significantly decreases with the strength (UCS) of soilcrete samples (Fig. 7), it is found that
increased clay content. As replacement of cement in concrete the UCS values present the performances in the range nearly
applications, it is found [46] that increase in the addition of from 4 to 12 MPa. The obtained UCS ranges mostly appear
glass powder leads to decrease in the density of samples. The to be consistent with ones obtained from the past studies
differences in the bulk density compared to the past works [5,26,41] (see Fig. 4). It can be said from the results that the
could be attributed to the mix proportions employed in this tested soilcrete samples having the strengths within the range
present study. However, it should be emphasized that since 4–12 MPa obtained here for construction of the soil–cement
the magnitudes of bulk densities do not significantly change columns could be acceptable for enhancement of loose sand

123
Arab J Sci Eng

3400 sion (i.e., SC) appears to be more favorable for attempting

20%Clay (SC) via deep mixing. Here, it is important to note that beyond the
3200
20% clay in the sand particularly for the developed experi-
UPV (m/s)

3000 mental program in this study, the strength could decrease and
4%Clay (SP)
be problematic for deep mixing. The increase in the strength
2800 8%Clay (SP-SC) due to the rise of clay content (but up to the limited rate
7-day 14-day 28-day
of 20% in this study) may be attributed to the contributions
2600 of cohesion and friction both developed between the solid
Mix1-0%GP

Mix2-3%GP

Mix3-6%GP

Mix4-9%GP

Mix5-0%GP

Mix6-3%GP

Mix7-6%GP

Mix8-9%GP

Mix9-0%GP

Mix10-3%GP

Mix11-6%GP

Mix12-9%GP
particles of soilcrete samples [4,48,49]. This increase could
also be explained by the physicochemical viewpoint. When
the clay content increases in the sand + clay mixtures for
the given water content, the pore sizes between the parti-
Fig. 10 The UPV performances of soilcrete samples
cles decrease that results in increase in the ratio of bound
water (i.e., the water associated with the molecular and elec-
trical forces surrounding the clay particles) to unbound water
with clay in the bearing capacity and settlement in accordance (i.e., the water between the particles and located outside the
with the UCS performances [4,47]. It is observed from the limits of bound water). Since the strength is related to the
results (Fig. 7) that the UCS values proportionally increase forces that bind the water to the clay particles, it proportion-
with the curing times. For the 7-day curing time, the UCS ally increases with the increased relative bound water [50].
decreases with the increased glass powder for all samples of As reviewed from the previous studies [26,29,45], the sand
the clay content. For the curing time of 14 days, the UCS deposits including up to the limited amount of clay content
decreases with the increased glass powder only for the clay (i.e., 10% by Carasca [45]; 15% by Helson et al. [26]; and
contents of 4 and 20%, while the addition of 3% glass powder 25% by Szymkiewicz [29]) present better strength improved
for 8% clay content of sand produces the UCS better than the by different cement dosages. From the revision, it can be said
native cement (i.e., 0% glass powder) and other inclusions that the clay content of sand that results in better strength
(i.e., 6, 9%). Regarding the 28-day curing time, the addition for deep mixing in this study (i.e., 20%) reasonably appears
of 3% glass powder only is found to increase the UCS per- within the ranges proposed by the past studies.
formances, while the glass powder addition greater than 3%
decreases the UCS values, compared to the native cement for 3.4 Elastic Modulus
all soilcrete samples of clay contents of sand. In conclusion,
the best UCS responses are obtained by the 3% glass pow- As concerned with the elastic modulus (Es) of soilcrete sam-
der additions after 28-day curing for all clay contents. These ples (Fig. 8), it is estimated that the Es of the tested samples
all findings summarized above clearly indicate the impact is able to perform the responses in the range nearly from
of curing time and clay content on the UCS performances 300 to 700 MPa. The obtained range for deep mixing in this
of soilcrete samples due to the glass powder added to the study appears to be in agreement with the past studies [6,31],
grout. There is no past study attempting the glass powder for in which the elastic modulus is nearly estimated between
deep mixing of sand to compare the results obtained here, 200 and 700 MPa for silty sand soils with different cement
as emphasized earlier. However, it can be said that generally dosages [31], and between 200 and 1000 MPa for loose sandy
increase in the addition of glass powder leads to decrease in soils with different water/cement ratios [6]. From the range
the strength in different applications including the concrete obtained in this study, it can be interpreted that the estimated
technology [30] and soil stabilizations [23]. This is in agree- Es of soilcrete samples indicates a relatively good stiffness
ment with the trend of strength responses obtained in this for the soil–cement columns to improve the loose sand. It is
present study due to the glass powder addition. observed from the results (Fig. 8) that the Es at the majority of
On the effect of clay content to the UCS performances the samples presents similar response trends of UCS (Fig. 7)
of soilcrete samples (Fig. 7), it can be generally said that regarding the curing time, glass powder addition and clay
the UCS performances increase with the increased clay con- content. From the results, it can be generally concluded that
tent up to the limited inclusion rate (i.e., 20% in this study). the best Es values are obtained due to the 3% glass powder
It is found from the results that the 20% clay content (i.e., additions after 28-day curing for all clay contents. In regard
the soil classification of SC) produces the best UCS per- to the effect of clay content, it is found that the best Es value
formances for deep mixing. On the other hand, the UCS of soilcrete sample for deep mixing is obtained due to the
performances of soilcrete samples decrease when the clay 20% clay content of sand (i.e., the poorly graded soil with
content is reduced to 8% (i.e., SP-SC) and 4% (i.e., SP). In clay inclusion, SC), similar to the response trend of UCS
conclusion, the poorly graded sand with the 20% clay inclu- presented earlier. As for the correlations of Es versus UCS

123
 Arab J Sci Eng

for the soilcrete samples (Fig. 9), it is observed that the Es The obtained pulse velocities in the categorization of very
values increase with the increased UCS values resulting in low velocity could be reasonably considered in the viewpoint
varied levels of correlations. It is illustrated in Fig. 9 that of strength quality, since the soil treated here is the loose sand
majority of the Es versus UCS responses (i.e., 20% clay con- with clay. From the results (Fig. 10), it is generally observed
tent of 28 days, 8% clay content of 7, 14 and 28 days, 4% that the response trends of UPV are found somewhat dif-
clay content of 7 and 14 days) yield well correlation per- ferent from the ones of UCS specifically regarding the clay
formances in the strong level of correlation coefficients (R content. In this view, it is appeared that the 4% clay added to
≥ 0.83). In the remaining ones (i.e., 20% clay content of 7 sand mostly produces better UPV performances. This differ-
and 14 days, 4% clay content of 28 days), it can be said that ence response between UCS and UPV may be attributed to
there still exist some correlations between the Es and UCS the homogeneity of samples [51]; however, it needs an exten-
(0.29 ≤ R ≤ 0.72), where a low performance is obtained sive study for verification. Relevant to this, it is reported [51]
by the 20% clay content of sand at the curing time of 14 that less clay content could lead to the particles being closer
days. The low correlation performance could be attributed producing less voids that result in higher velocity. Due to the
to the stress–strain response estimated for Es. However, it reasons criticized above, a lack of correlation tested between
needs a further study recommended for the confirmation due the UPV and UCS is observed in this study. On the effect of
to the effect of curing time. From the correlations of Es ver- glass powder, it is found that 3% addition of glass powder
sus UCS (Fig. 9), in conclusion, it can be said that the Es contributes better UPV performances at the curing time of 28
values mostly correlate with the UCS values. This could pro- days for all clay contents, while the performances decrease
vide the engineer in practice to more understand the factors beyond the 3% glass powder. As a general trend of glass pow-
(i.e., curing time, clay content, glass powder addition) that der, it is reported from the concrete works [30] that increase
affect the strength and stiffness responses of the soil–cement in the addition of glass powder leads to decrease in the UPV.
columns via deep mixing. This finding presents similar response trend of glass powder
obtained for the topic of deep mixing performed in this study.
In connection with this, some microstructural illustra-
3.5 Ultrasonic Pulse Velocity tions that show intensity of ingredients or arrangement of
solids and voids due to the glass-powder-added grout as
Finally, for the performances of ultrasonic pulse velocity typically given in Fig. 11 could be beneficial for the rea-
(UPV) (Fig. 10), they are found to measure the responses sons behind the experimental results obtained here. The two
nearly varied from 2600 to 3400 m/s. This range indicates the typical SEM pictures indicate cementitious compounds that
degree of low velocity proposed in the past work (Table 5).

Fig. 11 Typical microstructural

views by SEM images and XRD
results for some favorable
soilcrete specimens (SC, 28
days): a without the addition of
glass powder (i.e., native
cement, 0% glass powder), b
3% addition of glass powder

123
Arab J Sci Eng

suggest the binding phase in the treated grout (i.e., with 3% tudes due to the clay content of sand are obtained better
glass powder) (Fig. 11b) to be denser, compared to native by the inclusion of 20% clay. It can be said that all the
cement (Fig. 11a). Crystals can be occurred in the porosity magnitudes of UCS performances obtained from the soil-
and sand-paste interface. Cement hydration reaction tends to crete samples could be proposed for construction of the
fill the pores providing favorable strength. From the XRD soil–cement columns to enhance the loose sand with clay
pictures with chemical elements identified, their presences in the bearing capacity and settlement.
strongly support the formation of cementitious compounds 4. The Es values at the majority of the soilcrete samples
in the treated grout. The dominating peaks (C, O, Ca, Al, are mostly estimated to present similar response trends
Si) in the XRD pictures show considerable amount of ele- of UCS regarding the curing time, glass powder addition
ments which might come from cement, glass powder, sand and clay content. It is found that the Es proportionally
and clay. The considerable amount of silica together with increases with the UCS resulting in the performances of
aluminum and calcium in the treated grout could be con- strong correlations (R ≥ 0.83) at the majority of soilcrete
sidered as source of strength development alongside cement samples (i.e., 20% clay content of 28 days, 8% clay con-
hydration reaction. The bonding of silica and alumina could tent of 7, 14 and 28 days, 4% clay content of 7 and 14
give major distribution effects among the elements during days). On the other hand, there still exist some correla-
the treatment of sand using grout with glass powder. Here, tions between the Es and UCS (0.29 ≤ R ≤ 0.72) in the
an emphasis could be given that influence of clay on the remaining samples (i.e., 20% clay content of 7 and 14
strength development becomes more complex requiring an days, 4% clay content of 28 days).
additional microstructural and chemical study, as reported in 5. The UPV results generally indicate that 3% inclusion
past study [26]. Nonetheless, it can be said that in early ages, of glass powder at 28-day curing provides better per-
clay particles contribute to an increase of apparent strength formances similar to the response trends of UCS. On
with a compression of solid phase during hydration process the other hand, 4% clay content of sand due to the
that provides a supplementary cohesion to the material [26]. UPV produces better performances on the contrary of the
Despite the criticisms of grout dependent upon microstruc- UCS results. However, this strongly requires an extensive
ture given above, for being convinced robustly an extensive study for the verification to be robust.
study on the microstructural changes is required as a separate
effort proposed in future work. Acknowledgements The experimental study of this present paper has
been conducted at the Civil Engineering Laboratory of University of
Gaziantep, through the M.Sc. thesis of third author. The authors would
like to present their gratitude to the anonymous reviewers for carefully
4 Conclusion reviewing the manuscript and providing valuable comments.

Effect of glass powder inclusions (0, 3, 6 and 9%) added to

the cement-based slurry grout has been experimentally inves-
tigated in this paper on the enhancement of poorly graded References
loose sand with different amounts of clay (20, 8 and 4%) for
1. Massarsch, K.R.: Deformation properties of stabilized soil
the purpose of deep mixing technique. Based on the findings columns. In: International Conference on Deep Mixing Best Prac-
obtained from the experimental study, the following conclu- tice and Recent Advances, Stockholm, Sweden. pp. 129–144
sions could be drawn for the deep mixing of sand: (2005)
2. Bruce, D.: An introduction to the deep mixing methods as used
in geotechnical applications. Vol. III, The Verification and prop-
1. As the glass powder inclusion in the grout increases, the erties of treated ground, Report No. FHWA-RD-99– 167, U.S.
setting times of grout pastes increase. This indicates that Department of Transportation, Federal Highway Administration
high inclusions of glass powder are not able to accelerate (2001a)
3. Al-Tabba, A.; Ayotamuno, M.J.; Martin, R.J.: Soil mixing of
the setting times of grout paste. stratified contaminated sands. J. Hazard. Mater. 72(1), 53–75
2. The bulk density of soilcrete samples does not signifi- (2000)
cantly change with the glass powder additions regarding 4. Coduto, D.P.: Geotechnical Engineering: Principles and Practices.
the curing time and clay content. Prentice-Hall, Englewood Cliffs (1999)
5. Abbey, S.J.; Ngambi, S.; Ngekpe, B.E.: Understanding the per-
3. The UCS values are obtained better due to the inclusion formance of deep mixing columns improved soils-a review. Int. J.
of 3% glass powder at the 28-day curing time for all Civil Eng. Technol. 6(3), 97–117 (2015)
clay contents. At the remaining curing times (7 and 14 6. Esmaeili, M.; Gharouni-Nik, M.; Khajehei, H.: Evaluation of deep
days), the UCS values mostly decrease with the increased soil mixing efficiency in stabilizing loose sandy soils using labo-
ratory tests. Geotech. Test. J. 37(5), 1–11 (2014)
inclusions of glass powder. As the clay content of sand 7. Al-Tabbaa, A.; Lander, S.A.; Evans, C.W.: The performance of
increases (but to the limited amount, i.e., 20% in this model auger in the in situ stabilization/solidification of a contami-
study), the UCS responses increase. The UCS magni- nated sand. Environ. Technol. 18(9), 913–920 (1997)

123
 Arab J Sci Eng

8. Haley and Aldrich Inc.: Data report on soil mix design and testing. I- 28. Leoni, F.M.; Bertero, A.: Soil mixing in highly organic materials:
95/Route 1 Interchange, Alexandria, VA, VDOT, Project No. 0095– the experience of LPV111. New Orleans, Louisiana (USA) (2012)
96A-106, PE-101 (2001) 29. Szymkiewicz, F.: Evaluation des propriétés mécaniques du matéri-
9. Francisco, G.M.: Implementation of wet soil mixing method in aux soil-mixing. Université Paris-Est (2011)
organic soils. Research Report, Department of Engineering, Uni- 30. Khatib, J.M.; Negim, E.M.; Sohl, H.S.; Chileshe, N.: Glass powder
versity of Cambridge, UK, p. 94 (2003) utilisation in concrete production. Eur. J. Appl. Sci. 4(4), 173–176
10. Larsson, S.; Dahlstrom, M.; Nilsson, B.: Uniformity of lime– (2012)
cement columns for deep mixing: a field study. Ground Improv. 31. Farouk, A.; Shahien, M.M.: Ground improvement using soil–
9(1), 1–15 (2005) cement columns: experimental investigation. Alex. Eng. J. 52(4),
11. Liu, S.Y.; Zhang, D.W.; Liu, Z.B.; Deng, Y.F.: Assessment of 733–740 (2013)
unconfined compressive strength of cement stabilized marine clay. 32. ASTM C150/C150M: Standard specification for Portland cement.
Mar. Georesour. Geotechnol. 26(1), 19–35 (2008) West Conshohocken, PA (2015)
12. Puppala, J.A.; Madhyannapu, R.S.; Nazarian, S.; Yuan, D.; Hoyos, 33. Rahman, M.; Nateriya, R.: Study of waste glass powder as poz-
L.: Deep soil mixing technology for mitigation of pavement zolanic material in concrete. Int. Res. J. Eng. Technol. (IRJET) 2,
roughness. FHWA/TX-08/, Report 0-5179-1, Texas Department of 1847–1850 (2015)
Transportation, Austin, TX (2008) 34. Seward, T.P.; Vascott, T.: High temperature glass melt property
13. Porbaha, A.; Tanaka, H.; Kobayashi, M.: State of the art in deep database for process modeling. Wiley-American Ceramic Society,
mixing technology: part II-applications. Ground Improv. J. ISS- ISBN: 978-1-57498-225-1, p. 291 (2005)
MGE 2(2), 125–139 (1998) 35. JGS 0821-2000: Practice for marking and curing stabilized soil
14. Da Fonseca, A.V.; Cruz, R.C.; Consoli, N.C.: Strength properties of specimens without compaction (in Japanese). Japanese Geotech-
sandy soil–cement admixtures. Geotech. Geol. Eng. 27, 681–686 nical Society Standard.
(2009) 36. ASTM D4320/D4320M-09: Standard practice for laboratory
15. Yin, J.H.; Fang, Z.: Physical modeling of a footing on soft soil preparation of chemically grouted soil specimens for obtaining
ground with deep cement mixed soil columns under vertical load- engineering parameters. West Conshohocken, PA
ing. Mar. Georesour. Geotechnol. 28(2), 173–188 (2010) 37. Hansson, T.; Parry, L.; Graham, M.; Troughton, V.; Eriksson, H.:
16. Bhadriraju, V.; Puppala, A.J.; Madhyannapu, R.S.; Williammee, Limix: a deep dry mixing system used at channel tunnel rail contract
R.: Laboratory procedure to obtain well-mixed soil binder samples 440. In: Proceedings of Underground Construction 2001 Sympo-
of medium stiff to stiff expansive clayey soil for deep soil mixing sium and Exhibition, London, Institute of Materials, Minerals and
simulation. Geotech. Test. J. 31(3), 1–14 (2007) Mining, London (2001)
17. Bruce, D.A.: Practitioner’s guide to the deep mixing method. 38. ASTM C 191-04b: American society for testing and materials.
Ground Improv. 5(3), 95–100 (2001b) Standard test methods for time of setting of hydraulic cement by
18. Mindess, S.; Young, F.J.; Darwin, D.: Concrete, 2nd edn. Prentice Vicat needle. West Conshohocken, PA
Hall, Upper Saddle River (2003) 39. ASTM D5102-09: Standard test methods for unconfined com-
19. Little, D.N.; Males, E.H.; Prusinski, J.R.; Stewart, B.: Cementi- pressive strength of compacted soil-lime mixtures. West Con-
tious stabilization. Transportation Research Board, Transportation shohocken, PA
in the New Millennium, A2J01: Committee on Cementitious Sta- 40. Terashi, M.; Kitazume, M.: QA/QC for deep-mixed ground: current
bilization, Chairman: R.K.Seals (2016). http://onlinepubs.trb.org/ practice and future research needs. Proc. Inst. Civ. Eng. Ground
onlinepubs/millennium/00016.pdf. Last access on 22 Dec 2016 Improv. 164(3), 161–177 (2011)
20. Van Impe, W.F.; Verástegui Flores, R.D.; Mengé, P.; Van den 41. Kitazume, M.; Terashi, M.: Deep Mixing Method. CRC
Broeck, M.: Considerations on laboratory test results of cement Press/Balkema Publishers, Netherlands (2013)
stabilised sludge. In: Proceedings of Deep Mixing 05, International 42. ASTM C597-16: Standard test method for pulse velocity through
Conference on Deep Mixing Best Practice and Advances, 23–25 concrete. West Conshohocken, PA
May, Stockholm, Sweden, pp. 163–168 (2005) 43. Anon, O.H.: Classification of rocks and soils for engineering geo-
21. Nuruzzaman, D.; Hossain, M.A.: Effect of soda lime glass dust on logical mapping, part 1-rock and soil materials. Rep. Comm. Eng.
the properties of clayey soil. Global J. Res. Eng. 14(5-E), 16–22 Geol. Mapp. Bull. Int. Assoc. Eng. Geol. 19, 364–371 (1979)
(2014) 44. Borosnyói, A.; Kara, P.; Mlinárik, L.; Kase, K.: Performance of
22. Fauzi, A.; Djauhari, Z.; Fauzi, U.J.: Soil engineering properties waste glass powder (WGP) supplementary cementitious mate-
improvement by utilization of cut waste plastic and crushed waste rial (SCM)-workability and compressive strength. Recycl. Waste
glass as additive. Int. J. Eng. Technol. 8(1), 15–18 (2016) Recovery 65(3), 90–94 (2013). doi:10.14382/epitoanyag-jsbcm.
23. Canakci, H.; Aram, A.L.; Celik, F.: Stabilization of clay with waste 2013.17
soda lime glass powder. Procedia Eng. 161, 600–605 (2016) 45. Carasca, O.: Soil improvement by mixing: techniques and perfor-
24. Schwarz, N.; Neithalath, N.: Influence of a fine glass powder on mances. Energy Procedia 85, 85–92 (2016)
cement hydration: comparison to fly ash and modeling the degree 46. Vasudevan, G.; Pillay, S.G.K.: Performance of using waste glass
of hydration. Cem. Concr. Res. 38(4), 429–436 (2008) powder in concrete as replacement of cement. Am. J. Eng. Res.
25. Nassar, R.U.; Soroushian, P.: Green and durable mortar produced 2(12), 175–181 (2013)
with milled waste glass. Mag. Concr. Res. 64(7), 605–615 (2012) 47. Bowles, J.E.: Foundation Analysis and Design, 5th edn. McGraw-
26. Helson, O.; Beaucour, A.L.; Eslami, J.; Noumowe, A.; Gotteland, Hill, New York (1996)
P.: Physical and mechanical properties of soilcrete mixtures: soil 48. Holtz, R.D.; Kovacs, W.D.; Sheaahan, T.C.: An Introduction to
clay content and formulation parameters. Constr. Build. Mater. 131, Geotechnical Engineering, 2nd edn. Prentice Hall, Englewood
775–783 (2017) Cliffs (2011)
27. Lunardi, P.: Ground improvement by means of jet-grouting. In: 49. Dafalla, M.A.: Effects of clay and moisture content on direct shear
Proceeding ICE Ground Improvement, pp. 65–85 (1997) tests for clay-sand mixtures. Adv. Mater. Sci. Eng. 2013(Article ID
562726), 1–8 (2013). doi:10.1155/2013/562726

123
Arab J Sci Eng

50. Trask, P.D.; Close, J.E.H.: Effect of clay content on strength of 51. Lorenzi, A.; Tisbierek, F.T.; Da Silva Filho, L.C.P.: Ultrasonic
soils. In: Proceedings of 6th Conference on Coastal Engineering, pulse velocity analysis in concrete specimens. In: IV Conferencia
Gainesville, Florida, Paper No: 50, pp. 827–843 (1957). doi:10. Panamericana de END Buenos Aires, Octubre, pp. 1-13 (2007)
9753/icce.v6.50.

123