Professional Documents
Culture Documents
Consoli Et Al. 2020 - 3 Areias - Vidro - CC PDF
Consoli Et Al. 2020 - 3 Areias - Vidro - CC PDF
099]
Waste glass is a solid residue usually available near urban centres, where it is discarded after being
used as a container for one of several products. Carbide lime (CL) is a by-product of the
manufacture of acetylene gas. The present research evaluates the potential of combining these
two wastes as a possible hydraulic cement to enhance the behaviour of sands. Pozzolanic reactions occur
between silica in its amorphous phases (found in GG) and Ca++ (present in CL) in an alkaline
environment. The effect of the amount of GG, the CL content and the porosity of the blend on
the strength, stiffness and durability of compacted sand–GG–CL mixes is quantified for three
distinct sands. An original parameter, termed the porosity/binder index (η/Biv), can be used
to normalise the behaviour of the unconfined compressive strength (qu), the shear modulus at small
strains (G0) and the accumulated loss of mass (ALM) of the compacted mixes, considering GG + CL as
the binder. Results have shown unique normalised tendencies for qu, G0 and ALM with η/Biv for the
three studied compacted sand–GG–CL blends, even though each sand has a distinct grain size
distribution.
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
2 CONSOLI, DA SILVA CARRETTA, FESTUGATO, LEON, TOMASI AND HEINECK
Several common stabilisers (coal fly ash and bottom ash, short time periods? Are qu, G0 and ALM related to η/Biv in
blast furnace slag and cement kiln dust, among others) compacted sand–GG–lime mixes? Is G0 related to qu in
have been studied as additives to cement (e.g. Schwarz & compacted sand–GG–lime mixes? If so, is this relation
Neithalath, 2008). Glass powder (particles smaller than dependent on η/Biv? In poorly graded (uniform) silica
0·075 mm) has seen increasing use in various applications as sands, does the average particle size (D50) have any effect
an alternative to these stabilisers. Rangaraju et al. (2016) on strength, stiffness and loss of mass values and in their
studied the use of milled glass fibre waste partly substituting relations with η/Biv?
Portland cement and found superior mechanical and dur-
ability behaviour in mortar and concrete mixtures when
compared to control mixtures. Islam et al. (2017) found out EXPERIMENTAL PROGRAMME
that 20% replacement of Portland cement with waste glass The materials and methods used for this research are
powder in concrete was optimal from both economic and discussed below.
environmental viewpoints. Nassar & Soroushian (2012) and
Soroushian (2012) used GG as a partial replacement for
cement, as they deemed it an environmentally sustainable
material, which is largely available at a low cost. Güllü et al. Materials
(2017) investigated the use of cement-based grout combined Three poorly graded sands (SP) with mean particle
with glass powder to enhance clay soil by way of a deep diameters (D50) of 0·16 mm, 0·25 mm and 0·35 mm were
mixing technique. The authors found that a 3% replacement used throughout this investigation, respectively Osório sand,
of Portland cement by glass powder increased unconfined Rio Pardo (RP) sand and Porto Alegre (POA) sand. Fig. 1
compressive strength by 15%. shows the grain size distribution curve of the three sands
Despite all the previous work on the use of waste glass studied. Fig. 2 shows optical imaging of the three sands
in concrete and as a partial replacement for cement, a specific studied and of the GG particles. Osório sand is an eolic sand
research study concerning compacted sands–GG–CL (see Fig. 2(a)), while RP and POA sands are fluvial sands
blends is yet to be developed. The possibility of using (see Figs 2(b) and 2(c)). Their physical properties are shown
an environmentally sustainable geomaterial prepared with in Table 1. All of the sands studied are basically quartzitic.
finely ground waste glass and CL blended with sands and Particle forms vary from 1 : 1 to 1 : 2 from the shortest to
compacted as the beds of pipelines and spread footings, the longest axes; roundness varies from sub-rounded to
bases/sub-bases of pavements and other possible earth- angular; and grain surface texture varies from polished
works might be of interest for geotechnical and geoenviron- to small shaped markings.
mental engineers who wish to develop sustainable Ground glass grains are shown to be mainly angular
geo-solutions. with distinct shapes (see Fig. 2(d)). The physical properties
In order to assess the potential of using sand–GG–CL and grain size distribution curve of GG are displayed
blends, characterisation of the GG from physical, chemical in Table 1 and Fig. 1. The GG utilised in the testing origi-
and mineralogical viewpoints is necessary. Moreover, nated from domestic waste colourless glass bottles
strength tests are usually employed as a way to examine that were ground in a ball mill, being classified (ASTM,
the influence of variables such as the porosity of the blend 2006) as silt (ML). By employing the degree of crystallinity
and the amount of GG and CL. A logical dosage procedure method (Kern et al., 2012), it was determined that the
for each distinct amount of pozzolan for soil–coal fly ash amount of amorphous GG was more than 99%. Chemical
(pozzolan)–lime mixtures was created by Consoli et al. analysis has shown silicon dioxide to be the main constituent
(2011a) when they took into consideration the porosity/lime of GG.
index (η/Liv – considering lime as a binder) as a proper para- Carbide lime was used throughout this investigation. Such
meter to assess specific relations regarding strength for each lime is a by-product of the manufacture of acetylene
distinct amount of pozzolan for soil–coal fly ash (pozzolan)– gas. Testing revealed that a #200 sieve retains 23·2% of
lime mixtures. Nevertheless, no study has examined the the CL used in this study, thus meeting the 25% requirement
applicability of the porosity/binder index (η/Biv) for different of the ASTM C997 standard (ASTM, 2003). It possesses a
compacted sand–GG–CL mixes (considering GG plus CL specific surface area of 22·6 m2/g and a volumetric surface
as the binder) in terms of unconfined compressive strength area of 49·5 m2/cm3. Stoichiometrically, it consists of 81·0%
(qu), shear modulus at small strains (G0) and durability
(accumulated loss of mass (ALM) after wetting–drying
cycles). This seeks to determine the relationship between 100
η/Biv and G0, qu, and ALM for three different compacted 90 Ground glass
silica sand–GG–CL mixes. Osório sand
Glass bottles were not a usual waste until a few years ago, 80 RP sand
Percent finer by weight: %
because returnable bottles were reused and not discarded. 70 POA sand
In recent years, with changes in consumers’ habits, bottling
60
companies have stopped reusing them. Since this change,
glass bottles have started to be discarded, exponentially 50
increasing the amount of glass as waste. Few research studies 40
have focused on the use of glass in Portland cement-
30
based concrete as a substitute for aggregates (in large
pieces) or even partial substitution of Portland cement 20
(ground). The present research will focus on answering 10
the following key issues about the potential employment
of GG–CL blends in poorly graded (uniform) silica sands 0
0·0001 0·001 0·01 0·1 1 10
of different average grain sizes (D50). Is it possible to stabilise Grain size: mm
sand with lime and GG? If so, which quantities of GG,
CL and compaction energy are ordinarily employed to Fig. 1. Grain size distribution of Osório sand, RP sand, POA sand
produce strength, stiffness and durability improvements in and GG
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
GROUND WASTE GLASS–CARBIDE LIME AS A SUSTAINABLE BINDER 3
500 µm 500 µm
(a) (b)
500 µm
(c) (d)
Fig. 2. Optical imaging of (a) Osório sand; (b) RP sand; (c) POA sand; and (d) scanning electron microscope image of GG grains
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
4 CONSOLI, DA SILVA CARRETTA, FESTUGATO, LEON, TOMASI AND HEINECK
Table 2. Specimens’ target moulding characteristics for unconfined compression, initial stiffness and durability tests
study varied between 20 and 30% and were chosen after undercompaction levels were 3·5% for the first compacted
preliminary evaluation in which less than 20% of the GG was layers, 1·5% for the second compacted layers and 0·0% for
ineffective and more than 30% was considered excessive the third compacted layers. The sand–GG–CL blends were
by the authors, due to possible high cost. CL contents varied statically compacted in three layers into cylindrical split
from 3 to 11%, following international (Mitchell, 1981) moulds, to a target dry unit weight. The static compaction
experience. Modified compaction effort Proctor tests was performed using a hydraulic loading machine. Each layer
(ASTM, 2012) were carried out for each one of the three was compacted to a predetermined height. The specimens
different silica sands (Osório, RP and POA sands), blended typically failed over a defined shearing plane, which extended
with two distinct amounts of GG (20 and 30%); and 5% through the three layers of the specimen, indicating satisfac-
of CL. Results for all studied sands have shown that tory uniformity. However, not every specimen’s uniformity
increasing the amount of GG increased the dry unit weight was checked and as such cannot be guaranteed due to the
of the compacted specimens. The maximum dry unit weights inherent limitations of the method used.
of each sand, obtained with 30% GG, were 18·75 kN/m3 As exhibited in equation (1) (Consoli et al., 2011a),
(Osório sand), 18·70 kN/m3 (RP sand) and 18·70 kN/m3 porosity (η) is a function of the dry unit weight (γd) of the
(POA sand) with optimum water contents (ωopt) of 11, 10·2 sand (S), GG and CL content. Each substance (specific sand,
and 9·5%, respectively. All specimens were moulded with a GG and CL) has a unit weight of solids (γsS, γsGG, and γsCL),
water content of about 11% and were moist cured for 7 days. which also must be considered for computing porosity.
γd S=100 GG=100 CL=100
η ð%Þ ¼ 100 100 þ þ ð1Þ
ðS=100Þ þ ðGG=100Þ þ ðCL=100Þ γsS γsGG γsCL
Consoli et al. (2011b) have shown that the effect of diverse The volumetric binder content (Biv) allows the unification
soil structures formed during compaction of soil–binder of the pozzolan and CL contents into a single index that
blends at distinct water contents, together with porosities of represents the influence of the cementitious material’s
the specimens and the amounts of binder inserted in them content created by the combination of these two components.
play a role in the strength of such materials. Studies by For the calculation of this parameter, the sum of the
Stracke et al. (2012) specifically on the influence of moulding volumetric contents of both materials is considered, as
water content on compressive strength of early strength presented in equation (2). In equation (2), VGG is the
Portland cement Osório sand have shown that a 2% decrease volume of GG, VCL is the volume of CL, V is the total
in the water content (from 12 to 10%) ended up resulting in a volume, mGG is the weight of GG, mCL is the weight of CL,
change of 4·1% in strength. So, the effect of the moulding γsGG is the GG unit weight of solids and γsCL is the CL unit
water content (11%) used for the three studied compacted weight of solids.
sand–binder blends is minor, but exists. A target dry unit
VGG þ VCL
weight for a particular specimen was then established to be Biv ð%Þ ¼ 100
the dry weight of the compacted sand–GG–CL mix divided V
by the total volume of the specimen (ASTM, 2009). ðmGG =γsGG Þ þ ðmCL =γsCL Þ
Following the undercompaction method proposed by Ladd ¼ 100 ð2Þ
V
(1978), the sand–GG–CL blends were statically compacted
in three layers into cylindrical split moulds, to the target A dosage parameter, the porosity/binder index (η/Biv), that
dry unit weight. Rotta et al. (2003) and Consoli et al. (2000, is proposed in this study can be calculated by dividing
2006, 2007, 2016, 2017) have systematically used the equation (1) by equation (2), and allows for unification of the
undercompaction method proposed by Ladd (1978) for influence of the porosity and the binder content (pozzolanic
preparation of stabilised specimens. On average, the material (GG) and CL).
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
GROUND WASTE GLASS–CARBIDE LIME AS A SUSTAINABLE BINDER 5
Details of the theoretical and actual values of the specimens qu = 3·08 Biv0·28 (R2 = 0·99)
tested for strength and stiffness are presented in online qu = 2·52 Biv0·28 (R2 = 0·98)
supplementary Appendix 1, while specimens tested for loss 1600 qu = 2·02 Biv0·28 (R2 = 0·97)
of mass are presented in online supplementary Appendix 2. η = 31%
η = 30%
ASTM D559 (ASTM, 2015). Test procedures determine 350 η = 33%
mass losses produced by recurrent (12) wet–dry series. Every η = 36%
cycle begins by placing specimens under water for 5 h at
300
23° ± 2°C. Then, the samples are oven dried for 42 h at
71° ± 2°C. Finally, specimens are brushed 18 to 20 times
using a force of approximately 15 N. Some specimens, 250
representative of those moulded for unconfined compression
and ultrasound tests, were chosen to be moulded in order to 200
evaluate their durability. The chosen dosages maintained the
same amounts of GG, quantities of CL and dry unit weights
as for the strength and stiffness tests. The variations of GG 150
and CL contents, as well as the range of dry unit weights
studied for each one of the studied sands are presented in 100
Table 2. All blends were cured for 7 days. 2·1 2·2 2·3 2·4 2·5 2·6
Biv0·28
(c)
RESULTS AND ANALYSIS
Influence of porosity, binder content and porosity/binder Fig. 3. Variation of unconfined compressive strength (qu)
index on qu with adjusted volumetric binder content Biv0·28 considering
Figure 3 displays the variation of unconfined compressive distinct porosities (η) for (a) Osório sand; (b) RP sand and (c) POA
strength (qu) with adjusted volumetric binder content Biv0·28 sand
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
6 CONSOLI, DA SILVA CARRETTA, FESTUGATO, LEON, TOMASI AND HEINECK
(expressed in the present study as a percentage of cementi- qu = 257 751·62 η –3·60 (R2 = 0·87)
tious binder (GG plus CL) volume to the total volume of the qu = 212 851·81 η –3·60 (R2 = 0·95)
1600
specimen) for the compacted Osório sand–GG–CL qu = 119 546·67 η –3·60 (R2 = 0·71)
(Fig. 2(a)), compacted RP sand–GG–CL (Fig. 2(b)) and Biv0·28 = 2·49
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
GROUND WASTE GLASS–CARBIDE LIME AS A SUSTAINABLE BINDER 7
Colour qu × η/Biv0·28
Osório sand: qu = 8·30 × 106 (η/Biv0·28) –3·60 R2 = 0·86
RP sand: qu = 3·31 × 106 (η/Biv0·28) –3·60 R2 = 0·64
POA sand: qu = 2·44 × 106 (η/Biv0·28) –3·60 R2 = 0·77
Empty 20% GG
1600 Filled 30% GG
Pattern
1400 03% CL
Unconfined compressive strength, qu: kPa
05% CL
07% CL
1200 08% CL
11% CL
1000 γd = 18·2 kN/m3
γd = 17·5 kN/m3
γd = 17·2 kN/m3
800
γd = 16·5 kN/m3
γd = 16·2 kN/m3
600
γd = 15·5 kN/m3
400
200
0
10 11 12 13 14 15 16 17 18 19 20
η/Biv0·28
Fig. 5. Variation of unconfined compressive strength (qu) with adjusted porosity/binder index for compacted Osório sand–GG–CL, compacted RP
sand–GG–CL and compacted POA sand–GG–CL blends aimed at 7 days of curing. A full-colour version of this figure can be found on the ICE
Virtual Library (www.icevirtuallibrary.com)
Table 3. Correlation characteristics between qu and η/Biv (parameter F ), G0 and η/Biv (parameter T ), G0 and qu (parameter S) and ALM (after
12 wet–dry cycles) and η/Biv (parameter L) of the sand–GG–CL mixtures studied
Osório sand 8·30 106 0·86 17·9 106 0·73 2169·5 0·94 6·42 106 0·97
Rio Pardo (RP) sand 3·31 106 0·64 5·99 106 0·41 1785·3 0·85 5·07 106 0·88
Porto Alegre (POA) sand 2·44 106 0·77 2·39 106 0·31 985·3 0·82 6·00 106 0·62
The use of the porosity/binder index to normalise the failure of both the cemented and soil matrix phases; and
strength of Portland cement (seen as the binder) treated soils strain compatibility between the composite and its two
has been shown by Consoli et al. (2007, 2016). They showed phases applies. Through a volumetric averaging approach
that rates of change of strength with porosity (η) and the (Diambra et al., 2011, 2013; Diambra & Ibraim, 2015), the
inverse of the volumetric binder content (1/Biv) are not stress state of the cemented composite is derived from the
generally the same. Thus, the application of a power (as a rule stresses of its constituents at failure. To account for the soil
0·28; Consoli et al. (2007, 2016)) to Biv is required for the matrix strength, the model requires the calibration of the
rates of η and 1/Biv to be compatible. parameter a, among others, which can be estimated through
According to Consoli et al. (2016), the unconfined com- the results of triaxial tests. The parameter a links the peak
pressive strength of Portland cement (seen as the binder) strength of the soil matrix (represented using the density-
treated soils follow equations such as equation (4). dependent deviatoric stress and mean stress ratio qm/pm) to a
" #E state parameter (defined as the ratio between the porosity at
η the critical state (ηcs) and the current porosity (η)), as shown
qu ¼ A ð4Þ in equation (5), where M represents the critical state strength.
ðBiv ÞD a
qm η
in which A, D and E are scalars. ¼ M cs ð5Þ
pm η
Based on the concept of superposition of the failure
strength contributions of the soil and cement phases, The model developments yield equation (6), which
Diambra et al. (2017) provided a theoretical derivation for describes the unconfined compressive strength of cemented
the unconfined compression strength of artificially cemented granular soils.
granular soils. The theoretical model assumes that the com- " #a
posite cemented material is isotropic. The material granular η
qu ¼ K ð6Þ
matrix follows concepts of critical state soil mechanics ðBiv Þ1=a
and the strength of the cemented phase is described by
the Drucker–Prager failure criterion. The behaviour of the Diambra et al. (2017) also pointed out the similarity between
cemented soil at failure is determined by superposing equations (4) and (6) and suggested the exponents D and E
the strength contributions of the two phases; failure of the in equation (4) are dependent on the soil matrix related
composite cemented soil occurs as a result of a simultaneous parameter a: D = 1/a and E = a. The multiplying parameter
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
8 CONSOLI, DA SILVA CARRETTA, FESTUGATO, LEON, TOMASI AND HEINECK
A in equation (4) is the result of the combined properties evaluation of small-strain stiffness, as shown by equation (7).
of the sand matrix and the cement phase. According to According to the authors, this power relationship is a
the authors, the key governing parameters seem to be the simplification of the well-established Hardin’s type relation-
frictional strength of the matrix and the strength of the ship for the initial small-strain shear stiffness of cemented
cemented phase. The factor A is also affected by the exponent soils under unconfined loading conditions. Hardin’s type
a, which is controlled by the soil matrix properties. For the relationships usually present materials’ stiffness as a function
cemented soil matrixes studied, the observed value of E is of void ratio (or porosity), structure-related characteristic and
3·60 (Fig. 4), leading to an a of 0·28, which is in accordance mean effective stress. The power of the porosity (η) term in
with the values found by the authors mentioned. the empirical relationship accounts for the shape of the
normal compression line (NCL) of the uncemented soil in
the η–p′ plane, while the volumetric cement content (Civ)
Influence of the porosity/binder index on G0. Figure 6 term accounts for the shift of the NCL of the cemented soil
exhibits the distinction for initial shear modulus (G0) with with respect to its uncemented NCL state. Diambra et al.
an adjusted porosity/binder index η/Biv0·28 for compacted (2019) showed the exponential term depending on the η may
Osório sand–GG–CL, compacted RP sand–GG–CL and be seen as an approximation of the current Hvorslev pressure
compacted POA sand–GG–CL blends cured for 7 days. for uncemented soil and it is a term which accounts for the
Fig. 6 also denotes that the porosity/binder index is useful overconsolidation state or f (e) function of the material. The
in normalising initial shear stiffness results for the three exponential term linked to the Civ may represent the shift of
compacted sand–GG–CL mixtures. Reasonable correlations, the NCL for cemented soil and is a measure of the current
described as power functions, exist between G0 and η/Biv of state of cementation.
the three distinct silica sand–GG–CL mixtures studied
(equation (7) and Table 3).
!360
η Influence of the porosity/binder index on G0/qu. Increasing
G0 ðMPaÞ ¼ T ð7Þ the amounts of the GG and CL inclusions strengthens and
B028
iv stiffens the three sand matrixes. G0 plotted against qu is
in which T is a scalar. presented in Fig. 7 for: compacted Osório sand–GG–CL
Equation (7), relating G0 and η/Biv, was shown to be blends considering 20% and 30% of GG, 3%, 5% and 7% of
identical to equation (3), which relates qu and η/Biv. Once CL; compacted RP sand–GG–CL blends considering 20%
again, the scalar (T ) decreases when D50 increases (see and 30% of GG, 5%, 8% and 11% of CL; and compacted
Table 3). As previously explained, G0 increases when the D50 POA sand–GG–CL blends considering 20% and 30% of GG,
of the soil particles decreases because increasing the number 5%, 8% and 11% of CL – all considering 7 days of curing.
of the soil particles in a given volume, and consequently Good linear correlations between G0 and qu for the three
increasing the number of contacts among particles, results in different silica sand–GG–CL mixtures studied (equation (8)
a better redistribution of stresses among the particle contacts, and Table 3).
which in turn increases the stiffness of the assembly. G0
Diambra et al. (2019) provided a theoretical justification ¼S ð8Þ
qu
for the existence of empirical power relationships governed
by porosity/volumetric cement content index for the in which S is a scalar.
Colour G0 × η/Biv0·28
Osório sand: G0 = 1·79 × 107 (η/Biv0·28) –3·60 R2 = 0·73
RP sand: G0 = 5·99 × 106 (η/Biv0·28) –3·60 R2 = 0·41
POA sand: G0 = 2·39 × 106 (η/Biv0·28) –3·60 R2 = 0·31
4000 Empty 20% GG
Filled 30% GG
3500 Pattern
03% CL
05% CL
Initial shear modulus, G0: MPa
3000 07% CL
08% CL
2500 11% CL
γd = 18·2 kN/m3
γd = 17·5 kN/m3
2000
γd = 17·2 kN/m3
γd = 16·5 kN/m3
1500
γd = 16·2 kN/m3
γd = 15·5 kN/m3
1000
500
0
10 11 12 13 14 15 16 17 18 19 20
η/Biv0·28
Fig. 6. Variation of initial shear modulus (G0) with adjusted porosity/binder index for compacted Osório sand–GG–CL, compacted RP sand–
GG–CL and compacted POA sand–GG–CL blends aimed at 7 days of curing. A full-colour version of this figure can be found on the ICE Virtual
Library (www.icevirtuallibrary.com)
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
GROUND WASTE GLASS–CARBIDE LIME AS A SUSTAINABLE BINDER 9
Colour G0 × qu
Osório sand: G0 = 2169·49qu R2 = 0·94
RP sand: G0 = 1785·34qu R2 = 0·85
4000 POA sand: G0 = 985·34qu R2 = 0·82
Empty 20% GG
Filled 30% GG
3500
Pattern
03% CL
Initial shear modulus, G0: MPa
3000 05% CL
07% CL
08% CL
2500
11% CL
γd = 18·2 kN/m3
2000 γd = 17·5 kN/m3
γd = 17·2 kN/m3
1500 γd = 16·5 kN/m3
γd = 16·2 kN/m3
1000 γd = 15·5 kN/m3
500
0
0 0·2 0·4 0·6 0·8 1·0 1·2 1·4 1·6
Unconfined compressive strength, qu: MPa
Fig. 7. qu plotted against G0 for: Osório sand–GG–CL compacted blends considering 20% and 30% of GG, 3%, 5% and 7% of CL; RP sand–
GG–CL compacted blends considering 20% and 30% of GG, 5%, 8% and 11% of CL; and POA sand–GG–CL compacted blends considering
20% and 30% of GG, 5%, 8% and 11% of CL, all considering 7 days of curing. A full-colour version of this figure can be found on the ICE Virtual
Library (www.icevirtuallibrary.com)
Such relations suggest that for the three compacted sand– Figures 9(a) and 9(b), 10(a) and 10(b) and 11(a) and 11(b)
GG–CL blends studied, G0/qu is a scalar (and consequently present the normalised variation of unconfined compressive
independent of the porosity/binder index); G0/qu is basically strength ½ðqu Þ=ðqu at η=Biv028 ¼14 Þ, the normalised variation of
affected by D50, the smaller D50 is, the larger is G0/qu. initial shear stiffness ½ðG0 Þ=ðG0 at η=Biv028 ¼14 Þ and the normal-
ised variation of ALM after 12 wet–dry cycles
½ðALMÞ=ðALMat η=Biv028 ¼14 Þ with adjusted porosity/binder
Influence of the porosity/binder index on durability (wetting index η/Biv0·28 for compacted Osório sand–GG–CL, com-
and drying cycles). The ALM in each cycle is more pacted RP sand–GG–CL and compacted POA sand–GG–
pronounced for specimens with higher porosities and lower CL blends considering a normal plot (Figs 9(a), 10(a) and 11
cement contents, as shown in Figs 8(a)–8(c). According to (a)) and a logarithm plot (Figs 9(b), 10(b) and 11(b)),
the durability requirements of USACE (1994), the maximum respectively. In the logarithm plot, considering each one of
allowable weight loss after 12 wetting–drying cycles is 11% the variables studied (qu, G0 and ALM), unique straight lines
for granular material with a reduced plasticity index. For this are shown. In total, 115, 101 and 23 specimens’ results were
study, the wetting–drying durability requirements were only used for assessing qu, G0 and ALM through equations (10),
reached for the three sand–GG–CL blends with an adjusted (11) and (12) in predicting the mechanical performance of
porosity/binder index (η/Biv0·28) of less than 16, which corres- three distinct granular soils stabilised with GG + CL, leading
ponds to 20 of the 23 mixtures studied. Therefore, most to a coefficient of determination (R 2) equal to 0·75 (qu), 0·70
specimens studied showed an ALM of less than 11%. The (G0) and 0·90 (ALM), established by the minimum mean
specific characteristics of each specimen analysed (dry unit square error. It is important to point out that equations (7)–
weight (γd) of the specimens, GG content and CL content) (9) consider three poorly graded sands with distinct mean
can be verified in Figs 8(a)–8(c), where one can see which of particle diameter (D50 = 0·16 mm for Osório sand–GG–CL
them meet the criteria. blends, D50 = 0·25 mm for RP sand–GG–CL mixes and
Results have shown similar power correlations of ALM D50 = 0·35 mm for POA–GG–CL mixtures).
after the 12 cycles and η/Biv for the three silica sand–GG–CL !360
mixtures studied (equation (9) and Table 3). qu η
" #525 ¼ 134 10 4
ð10Þ
qu ðfor ðη=B028 Þ¼14Þ B028
η iv iv
ALMð%Þ ¼ L 028 ð9Þ
Biv !360
G0 η
in which L is a scalar. ¼ 134 10 4
ð11Þ
G0 ðforðη=B028 Þ¼14Þ
iv
B028
iv
!525
Normalisation of strength, stiffness and ALM results. In ALM η
6
order to normalise the results obtained for the different ¼ 120 10 ð12Þ
ALMðforðη=B028 Þ¼14Þ B028
iv
sands, the method proposed by Consoli et al. (2017) was iv
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
10 CONSOLI, DA SILVA CARRETTA, FESTUGATO, LEON, TOMASI AND HEINECK
Colour ALM × η/Biv0·28
Osório sand: 12 cycles ALM = 6·42 × 10–6 (η/Biv0·28)5·25 R2 = 0·97
Filled 30% GG
30 Empty 20% GG
Pattern
Accumulated loss of mass, ALM: % 03% CL
05% CL
07% CL
08% CL
20 11% CL
γd = 18·5 kN/m3
γd = 18·2 kN/m3
γd = 17·5 kN/m3
γd = 17·2 kN/m3
γd = 16·5 kN/m3
10 γd = 16·2 kN/m3
γd = 15·5 kN/m3
0
10 11 12 13 14 15 16 17 18 19 20
η/Biv0·28
(a)
Colour ALM × η/Biv0·28
RP sand: 12 cycles ALM = 5·07 × 10–6 (η/Biv0·28)5·25 R2 = 0·88
Filled 30% GG
10 Empty 20% GG
Pattern
03% CL
Accumulated loss of mass, ALM: %
8 05% CL
07% CL
08% CL
11% CL
γd = 18·5 kN/m3
6 γd = 18·2 kN/m3
γd = 17·5 kN/m3
γd = 17·2 kN/m3
4 γd = 16·5 kN/m3
γd = 16·2 kN/m3
γd = 15·5 kN/m3
2
0
10 11 12 13 14 15 16 17 18
η/Biv0·28
(b)
Colour ALM × η/Biv0·28
POA sand: 12 cycles ALM = 6·00 × 10–6 (η/Biv0·28)5·25 R2 = 0·62
Filled 30% GG
Empty 20% GG
10
Pattern
03% CL
Accumulated loss of mass, ALM: %
05% CL
8 07% CL
08% CL
11% CL
γd = 18·5 kN/m3
6 γd = 18·2 kN/m3
γd = 17·5 kN/m3
γd = 17·2 kN/m3
γd = 16·5 kN/m3
4
γd = 16·2 kN/m3
γd = 15·5 kN/m3
0
10 11 12 13 14 15 16 17 18
η/Biv0·28
(c)
Fig. 8. Variation of ALM after 12 cycles with porosity/binder index for (a) compacted Osório sand–GG–CL, (b) compacted RP sand–GG–CL
and (c) compacted POA sand–GG–CL blends aimed at 7 days of curing. A full-colour version of this figure can be found on the ICE Virtual
Library (www.icevirtuallibrary.com)
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
GROUND WASTE GLASS–CARBIDE LIME AS A SUSTAINABLE BINDER 11
10 G0 /G0 (at η/Biv = 14) × η/Biv0·28
qu/qu (at η/Biv = 14) × η/Biv0·28
7 Osório sand
9 Osório sand RP sand
RP sand POA sand
8 POA sand ± 1σ (standard deviation)
6
± 1σ (standard deviation) ± 2σ (standard deviation)
7 ± 2σ (standard deviation) ± 3σ (standard deviation)
qu/qu (at η/Biv0·28 = 14)
4
5
4 3
3
2
2
1
1
0 0
10 11 12 13 14 15 16 17 18 19 20 10 11 12 13 14 15 16 17 18 19 20
η/Biv0·28 η/Biv0·28
(a) (a)
qu/qu (at η/Biv = 14) × η/Biv0·28 G0 /G0 (at η/Biv = 14) × η/Biv0·28
10 Osório sand 8 Osório sand
8 RP sand
6 RP sand
6 POA sand
POA sand
± 1σ (standard deviation)
4 ± 1σ (standard deviation)
4 ± 2σ (standard deviation)
± 2σ (standard deviation)
± 3σ (standard deviation)
qu/qu (at η/Biv0·28 = 14)
2 2
1·0 1·0
0·8 0·8
0·6 0·6
0·4
0·4
0·2
0·2
0·1
8 9 10 20 30 0·1
η/Biv0·28 8 9 10 20 30
η/Biv0·28
(b)
(b)
Fig. 9. Normalised variation of unconfined compressive strength
½ðqu Þ=ðqu at η=Biv028 ¼14 Þ with adjusted porosity/binder index for com- Fig. 10. Normalised variation of initial shear modulus
pacted Osório sand–GG–CL, compacted RP sand–GG–CL and ½ðG 0Þ =ðG 0 at η=Biv028 ¼14 Þ with adjusted porosity/binder index for com-
compacted POA sand–GG–CL blends aimed at 7 days of curing pacted Osório sand–GG–CL, compacted RP sand–GG–CL and
considering (a) normal plot and (b) logarithm plot compacted POA sand–GG–CL blends aimed at 7 days of curing
considering (a) normal plot and (b) logarithm plot
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
12 CONSOLI, DA SILVA CARRETTA, FESTUGATO, LEON, TOMASI AND HEINECK
10 ALM/ALM (at η/Biv = 14) × η/Biv0·28 3
Osório sand
9
RP sand
POA sand 2
8
± 1σ (standard deviation)
ALM/ALM (at η/Biv0·28 = 14)
7 ± 2σ (standard deviation)
Standardised residuals
1
6
5 0
3 –1
2
–2
1
0
–3
10 11 12 13 14 15 16 17 18 19 20
0 0·5 1·0 1·5 2·0 2·5
η/Biv0·28 Fitted value (qu/qu (at η/Biv0·28 = 14))
(a) (a)
10 3
8
6
2
4
Standardised residuals
2 1
qu/qu (at η/Biv0·28 = 14)
1·0
0
0·8
0·6
ALM/ALM (at η/Biv = 14) × η/Biv0·28 –1
0·4
Osório sand
RP sand
0·2 POA sand
± 1σ (standard deviation) –2
± 2σ (standard deviation)
0·1
–3
8 9 10 20 30
η/Biv0·28 0 0·5 1·0 1·5 2·0 2·5
Fitted value (G0/G0 (at η/Biv0·28 = 14))
(b)
(b)
Fig. 11. Normalised variation of ALM after 12 cycles
½ðALMÞ=ðALMat η=Biv028 ¼14 Þ with adjusted porosity/binder index for 3
compacted Osório sand–GG–CL, compacted RP sand–GG–CL and
compacted POA sand–GG–CL blends aimed at 7 days of curing
considering (a) normal plot and (b) logarithm plot 2
Standardised residuals
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
GROUND WASTE GLASS–CARBIDE LIME AS A SUSTAINABLE BINDER 13
cement as a binder of compacted sands. The GG L lime content (expressed as a percentage)
particles interact with calcium in a hydrated alkaline Liv volumetric lime content (expressed as a
environment, yielding binder products such as calcium percentage)
silicate hydrate gel, which is the same product formed m mass
from the hydration of Portland cement. This represents qu unconfined compressive strength
R2 coefficient of determination
a major advance in the techniques of sustainable soil V volume of specimen
improvement. γd dry unit weight
(b) It can be observed that the smaller the mean sand γs unit weight of solids
particle diameter is, the greater the effect of the η porosity
stabilisation results with finely ground waste glass plus η/Biv porosity/binder index
CL by means of the observed gains in terms of strength, η/Liv porosity/lime index
stiffness and durability. The coordination number of
particles of uniform soils with different grain sizes is the
same. Considering a constant porosity value, the number
of particles in a given volume will increase with the REFERENCES
reduction of particle diameter. Increasing the number of ABNT (Brazilian Standards Association) (2000). NM 23: Portland
particles will consequently increase the total number of cement and other powdered materials: determination of
contacts between particles. The increasing qu and G0 and unit weight of the grains. Rio de Janeiro, Brazil: ABNT
(in Portuguese).
reducing ALM when D50 is reduced for specimens with ASTM (2003). C977: Standard specification for quicklime and
the same porosity and the same quantity of cementing hydrated lime for soil stabilization. West Conshohocken, PA,
agent is explained by the fact that increasing the number USA: ASTM International.
of soil particles, and consequently increasing the number ASTM (2006). D2487: Standard classification of soils for
of contacts among particles, results in a better engineering purposes. West Conshohocken, PA, USA: ASTM
redistribution of stresses among particle contacts, which International.
in turn increases not only the strength, but also the ASTM (2008). D2845: Standard test method for laboratory
stiffness and durability, of the assembly; determination of pulse velocities and ultrasonic elastic constants
(c) The adjusted porosity/binder index (η/Biv0·28) for of rock. West Conshohocken, PA, USA: ASTM International.
ASTM (2009). D7263: Standard test methods for laboratory
compacted soil (three distinct sands in the present
determination of density (unit weight) of soil specimens. West
study)–GG–CL mixes is an appropriate parameter to Conshohocken, PA, USA: ASTM International.
evaluate unique relationships in terms of: (a) ASTM (2010). C39: Standard test method for compressive strength
unconfined compressive strength (qu); (b) shear of cylindrical concrete specimens. West Conshohocken, PA,
modulus at small strains (G0); and (c) durability USA: ASTM International.
(ALM after wetting–drying cycles). The GG plus ASTM (2011). C25: Standard test methods for chemical analysis of
CL are considered to be a binder. Different amounts of limestone, quicklime, and hydrated lime. West Conshohocken,
GG, different amounts of CL and distinct dry unit PA, USA: ASTM International.
weights were considered. ASTM (2012). D1557: Standard test methods for laboratory
(d ) It was also found that the strength, stiffness and ALM compaction characteristics of soil using modified effort
(2700 kN-m/m3). West Conshohocken, PA, USA: ASTM
after 12 wet–dry cycles were normalised for different International.
sand–GG–lime mixes, providing unique equations for ASTM (2013). C511: Standard specification for mixing rooms,
each mechanical property studied. moist cabinets, moist rooms, and water storage tanks used in the
(e) The method proposed by Consoli et al. (2017) for testing of hydraulic cements and concretes. West Conshohocken,
strength was adjusted and expanded upon for the blends PA, USA: ASTM International.
studied. This provided satisfactory results, which allowed ASTM (2015). D559: Standard test method for wetting and drying
the determination of the curve that controls not only compacted soil–cement mixtures. West Conshohocken, PA,
strength, but also stiffness and ALM for an entire range USA: ASTM International.
of porosities and binder contents for the three distinct Basu, D. & Puppala, A. (2015). Principles of sustainability and
their applications in geotechnical engineering. In Geotechnical
sands. This also contributes to minimising the number of
synergy in Buenos Aires 2015 (eds A. O. Sfriso, D. Manzanal and
specimens that need to be moulded and tested, thus also R. J. Rocca), pp. 162–183. Amsterdam, the Netherlands: IOS
reducing project development cost and time. Press.
(f ) Further studies will be carried out in the future to check Bicca Neto, V. (2015). Commitment business for recycling – review.
whether current findings can be applied to other sands São Paulo, Brazil: CEMPRE Editor (in Portuguese).
considering different gradations (e.g. well-graded BSI (2015). BS 5930: Code of practice for ground investigations.
sand), mineralogy, and so on. London, UK: BSI.
Chang, T. S. & Woods, R. D. (1992). Effect of particle contact bond
on shear modulus. J. Geotech. Engng 118, No. 8, 1216–1233.
Consoli, N. C., Rotta, G. V. & Prietto, P. D. M. (2000). Influence of
ACKNOWLEDGEMENTS curing under stress on the triaxial response of cemented soils.
The authors wish to express their appreciation to Edital Géotechnique 50, No. 1, 99–105, https://doi.org/10.1680/geot.
12/2014 FAPERGS/CNPq – PRONEX (project # 16/2551- 2000.50.1.99.
0000469-2) and CNPq (INCT, Produtividade em Pesquisa Consoli, N. C., Rotta, G. V. & Prietto, P. D. M. (2006). Yielding–
and Universal) for funding the research group. compressibility–strength relationship for an artificially cemen-
ted soil cured under stress. Géotechnique 56, No. 1, 69–72,
https://doi.org/10.1680/geot.2006.56.1.69.
Consoli, N. C., Foppa, D., Festugato, L. & Heineck, K. S. (2007).
NOTATION Key parameters for strength control of artificially cemented
a, A, D, E and K scalars soils. J. Geotech. Geoenviron. Engng 133, No. 2, 197–205.
B binder content (expressed as a percentage) Consoli, N. C., Dalla Rosa, A. & Saldanha, R. B. (2011a). Variables
Biv volumetric binder content (expressed as a governing strength of compacted soil–fly ash–lime mixtures. J.
percentage) Mater. Civ. Engng 23, No. 4, 432–440.
D50 mean particle diameter Consoli, N. C., Rosa, D. A., Cruz, R. C. & Dalla Rosa, A. (2011b).
G0 shear modulus at small strains Water content, porosity and cement content as parameters
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.
14 CONSOLI, DA SILVA CARRETTA, FESTUGATO, LEON, TOMASI AND HEINECK
controlling strength of artificially cemented silty soil. Engng mechanics and foundation engineering, pp. 509–565.
Geol. 122, No. 3–4, 328–333. Rotterdam, the Netherlands: Balkema.
Consoli, N. C., Ferreira, P. M. V., Tang, C. S., Marques, S. F. V., Mohajerani, A., Vajna, J., Cheung, T. H. H., Kurmus, H.,
Festugato, L. & Corte, M. B. (2016). A unique relationship Arulrajah, A. & Horpibulsuk, S. (2017). Practical recycling
determining strength of silty/clayey soils – Portland cement applications of crushed waste glass in construction materials: a
mixes. Soils Found. 56, No. 6, 1082–1088. review. Constr. Build. Mater. 156, 443–467.
Consoli, N. C., Marques, S. F. V., Floss, M. F. & Festugato, L. Nassar, R. U. & Soroushian, P. (2012). Green and durable mortar
(2017). Broad-spectrum empirical correlation determining produced with milled waste glass. Mag. Concrete Res. 64, No. 7,
tensile and compressive strength of cement-bonded clean 605–615, https://doi.org/10.1680/macr.11.00082.
granular soils. J. Mater. Civ. Engng 29, No. 6, 06017004. Rangaraju, P. R., Rashidian-Dezfouli, H., Nameni, G. &
Diambra, A. & Ibraim, E. (2015). Fibre-reinforced sand: interaction Amekuedi, G. Q. (2016). Properties and performance of
at the fibre and grain scale. Géotechnique 65, No. 4, 296–308, ground glass fiber as a pozzolan in Portland cement concrete.
https://doi.org/10.1680/geot.14.P.206. In Proceedings of the 11th international concrete sustainability
Diambra, A., Ibraim, E., Russell, A. R. & Muir Wood, D. (2011). conference. Silver Springs, MD, USA: National Ready Mixed
Modelling the undrained response of fibre reinforced sands. Concrete Association.
Soils Found. 51, No. 4, 625–636. Rotta, G. V., Consoli, N. C., Prietto, P. D. M., Coop, M. R. &
Diambra, A., Ibraim, E., Russell, A. R. & Muir Wood, D. (2013). Graham, J. (2003). Isotropic yielding in an artificially cemented
Fibre reinforced sands: from experiments to modelling and soil cured under stress. Géotechnique 53, No. 5, 493–501,
beyond. Int. J. Numer. Analyt. Methods Geomech. 37, No. 15, https://doi.org/10.1680/geot.2003.53.5.493.
2427–2455. Saldanha, R. B., Scheuermann Filho, H. C., Mallmann, J. E. C.,
Diambra, A., Ibraim, E., Peccin, A., Consoli, N. C. & Festugato, L. Consoli, N. C. & Reddy, K. R. (2018). Physical-
(2017). Theoretical derivation of artificially cemented granular mineralogical-chemical characterization of carbide lime: an
soil strength. J. Geotech. Geoenviron. Engng 143, No. 5, 04017003, environment-friendly chemical additive for soil stabilization. J.
https://doi.org/10.1061/(ASCE)GT.1943-5606.0001646. Mater. Civ. Engng 30, No. 6, 06018004.
Diambra, A. Ibraim, E., Festugato, L. & Corte, M. B. (2019). Schwarz, N. & Neithalath, N. (2008). Influence of a fine glass
Stiffness of artificially cemented sands: insight on characteriz- powder on cement hydration: comparison to fly ash and
ation through empirical power relations. Road Mater. Pavement modeling the degree of hydration. Cem. Concr. Res. 38, No. 4,
Des., https://doi.org/10.1080/14680629.2019.1705379. 429–436.
Encyclopaedia Britannica (2019). See https://www.britannica. Soroushian, P. (2012). Strength and durability of recycled aggregate
com/technology/glass (accessed 11/06/2019). concrete containing milled glass as partial replacement for
German, R. M. (2014). Coordination number changes during cement. Constr. Build. Mater. 29, 368–377, https://doi.org/
powder densification. Powder Technol. 253, 368–376. 10.1016/j.conbuildmat.2011.10.061.
Güllü, H., Canakci, H. & Al Zangana, I. F. (2017). Use of cement Stracke, F., Jung, J. G., Korf, E. P. & Consoli, N. C. (2012). The
based grout with glass powder for deep mixing. Constr. Build. influence of moisture content on tensile and compressive
Mater. 137, 12–20. strength of artificially cemented sand. Soils and Rocks 35,
Islam, G. M. S., Rahman, M. H. & Kazi, N. (2017). Waste glass No. 3, 303–308.
powder as partial replacement of cement for sustainable USACE (U.S. Army Corps of Engineers) (1994). Flexible pavement
concrete practice. Int. J. Sustain. Built Environ. 6, No. 1, design for airfields, Technical Manual No. TM5-822-13.
37–44, https://doi.org/10.1016/j.ijsbe.2016.10.005. Washington, DC, USA: USACE.
Kern, A., Madsen, I. C. & Scarlett, N. V. Y. (2012). Quantifying Wiacek, J. & Molenda, M. (2014). Effect of particle size distrib-
amorphous phases. New York, NY, USA: Springer. ution on micro- and macromechanical response of granular
Ladd, R. (1978). Preparing test specimens using undercompaction. packings under compression. Int. J. Solids Structs 51,
Geotech. Test. J. 1, No. 1, 16–23. No. 23–24, 4189–4195.
Massazza, F. (1998). Pozzolana and pozzolanic cements. In Lea’s Yang, L. & Salvati, L. (2010). Small strain properties of
chemistry of cement and concrete, 4th edn (ed. P. C. Hewlett), sands with different cement types. Proceedings of the 5th
pp. 471–635. London, UK: Arnold. international conference on recent advances in geotechnical
Mitchell, J. K. (1981). Soil improvement – state-of-the-art report. In earthquake engineering and soil dynamics, San Diego, CA,
Proceedings of the 10th international conference on soil USA.
Downloaded by [ Universidade Federal do Rio Grande do Sul] on [05/08/20]. Copyright © ICE Publishing, all rights reserved.