Long-term resilient and permanent deformation behaviour of Controlled Low-Strength Materials for pavement applications

Transportation Geotechnics 2 (2015) 108–118

Contents lists available at ScienceDirect

Transportation Geotechnics journal homepage: www.elsevier.com/locate/trgeo

Long-term resilient and permanent deformation behaviour of Controlled Low-Strength Materials for pavement applications Marco Bassani a,⇑, Sadaf Khosravifar b,1, Dimitrios G. Goulias c,2, Charles W. Schwartz d,3 a

Politecnico di Torino, Department of Environment, Land and Infrastructures Engineering (DIATI), 24, Corso Duca degli Abruzzi, Torino 10129, Italy University of Maryland, Department of Civil and Environmental Engineering, 1173 Glenn L. Martin Hall, College Park, MD 20742, United States c University of Maryland, Department of Civil and Environmental Engineering, 0147A Glenn L. Martin Hall, College Park, MD 20742, United States d University of Maryland, Department of Civil and Environmental Engineering, 1173G Glenn L. Martin Hall, College Park, MD 20742, United States b

a r t i c l e

i n f o

Article history: Received 12 September 2014 Revised 13 December 2014 Accepted 14 December 2014 Available online 19 December 2014 Keywords: Controlled Low-Strength Materials Flowable fill materials Sulpho-aluminate cement Resilient modulus Repeated load triaxial tests Constitutive models

a b s t r a c t The paper deals with the long-term stiffness characterisation of Controlled Low-Strength Materials (CLSMs) for pavement applications in substitution of granular fill materials. Three alternative CLSM mixtures, two with ordinary Portland cement and a third one with an ultra-rapid sulpho-aluminate cement, were examined. Two different sample aspect ratios were considered and the samples were subjected to different testing conditions in terms of saturation, loading time and repetition. The investigated CLSMs are insensitive to variations of loading frequency and to water saturation, and sensitive to sample aspect ratio. They exhibit a significant increase in stiffness under repeated load triaxial testing and a low permanent strain accumulation. Finally, they exhibit an increase in resilient modulus when the deviatoric stress increases. Ó 2014 Elsevier Ltd. All rights reserved.

Introduction Controlled Low-Strength Materials (CLSMs) are flowable mortars consisting mostly of cement, fine aggregate and water (Folliard et al., 2008; Alizadeh et al., 2014; Etxeberria et al., 2013). They often include admixtures to enhance both fluid and hardened properties, as well as secondary materials and by-products (e.g. fly ash, recycled materials, and alternative binders) to improve mixture properties and reduce the quantities of natural materials for lower production costs. Also known as flowable fill materials (FFM), CLSMs are employed in a variety of appli⇑ Corresponding author. Tel.: +39 011 564 5635. E-mail addresses: [email protected] (M. Bassani), sadafkh@umd. edu (S. Khosravifar), [email protected] (D.G. Goulias), schwartz@umd. edu (C.W. Schwartz). 1 Tel.: +1 530 531 5030. 2 Tel.: +1 301 405 2624. 3 Tel.: +1 301 405 1962. http://dx.doi.org/10.1016/j.trgeo.2014.12.001 2214-3912/Ó 2014 Elsevier Ltd. All rights reserved.

cations where the use of granular fill materials, or the excavated soils, do not provide the required performance and/ or cause longer construction time and thus increased cost. CLSMs can easily flow and fill irregular voids and trenches and they do not require compaction or vibration (i.e., they are self-levelling). They harden in a reasonably short time, reach mechanical properties similar or superior to those of soils, and maintain such properties over a long period of time while resisting adverse environmental effects and loadings. Such properties are greatly appreciated in road construction, maintenance works and pavement repairs. A great variety of CLSMs with different short and long-term performance may be obtained by modifying the composition, proportion of components and mixing operations. To meet the required performance, the mechanical properties of CLSMs must be known and controlled during the design and construction stage. More specifically for pavement applications, CLSM should be

109

M. Bassani et al. / Transportation Geotechnics 2 (2015) 108–118

excavatable when hardened and at the same time be resistant and stiff enough to bear and distribute the loads. In case of road applications (e.g. trench backfilling, bridge approaches, pavement repairs), the mechanical properties of CLSMs also need to be comparable to those of granular layers in order to assure a balanced distribution of stress and strains under the repeated traffic load. Unlike soils and unbound granular material (UGM), CLSMs may show a gain in strength and stiffness due to curing that can have a significant implication for pavement applications (Folliard et al., 2008). For example, the gain in stiffness and strength during curing may gradually make the excavation (e.g. trench reopening) more difficult, resulting in an additional cost and labour. It is thus important that CLSMs have stable properties during their service life that are close to those of surrounding soils and UGM. Strength and stiffness gain depends on the hydration phenomena and/or pozzolanic activity in those mixtures containing OPC or fly ash (Folliard et al., 2008; Bertola et al., 2013). In most CLSMs, fly ash is employed to partially substitute the cement in the mixture (Du et al., 2002), reducing thus mixture cost. It is also used to increase the flowability of mixtures thanks to its very fine dimension and spheroidal shape (Folliard et al., 2008; Du et al., 2002; Turkel, 2007). Similar effects and properties may be reached using functional admixtures with different proportions of the basic constituents. From the physicalvolumetric point of view, CLSMs can be divided into two main families based on mix composition (Folliard et al., 2008): (a) mixtures containing fly ash and (b) mixtures without fly ash. The first group provides a highly flowable fresh mixture with a very low void content (lower than 3%). Group b, include admixtures (i.e., air-entraining agents) to produce a soft and flowable mixture with a high void content (15–30%) and low density. CLSMs that do not contain fly ash seem to be more suitable to support road pavements in utility beddings and bridge approaches. The reason is that the use of the air-entraining agent leads to a lower density and higher air voids, improving insulation properties and frost resistance. Meanwhile, it contributes to a lower water/cement ratio and therefore decreases the segregation, bleeding phenomena, and related costs. Furthermore, a higher air void content hampers the long-term strength gain, thus assuring easy of future excavatability (Folliard et al., 2008). For these specific applications, stiffness over time needs to be investigated in order to achieve comparable behaviour to those exhibited by soils and UGMs. In the case of repeated loading conditions, the most accepted stiffness parameter used to characterise base, subbase and subgrade materials is the resilient modulus (MR). Referring to CLSMs, Folliard et al. (2008) advocate that future research on MR is necessary to draw meaningful conclusions on the effects of mixture parameters on such fundamental material property. This paper presents the results of an extensive laboratory investigation focused on the evaluation of long-term stiffness properties of CLSMs at various moisture conditions and stress levels. In particular, three different CLSMs suitable for pavement applications were investigated by means of the resilient modulus test according to AASHTO

T307 (American Association of State Highway and Transportation Officials, 2007). The resilient modulus of CLSM can directly be used in structural analysis models to calculate the pavement response to wheel loads and to design pavement structures (American Association of State Highway and Transportation Officials, 2008).

Experimental study This investigation focused on the characterisation of three CLSMs having various composition and properties. The ingredients used for these mixtures, as well as mixture compositions are presented next. The mix formulations followed recommendations based on past studies (Bertola et al., 2013). A primary focus in the present study was the assessment of the use of sulpho-aluminate cement (SAC) to replace the ordinary Portland cement (OPC) traditionally employed in producing flowable fill materials. CLSMs made with SAC have been shown to have mechanical properties very similar to reference unbound granular materials normally employed in subbase and pavement repairs (i.e., trench backfilling, pavement rehabilitation). On the basis of such experience, the investigation focused primarily on two commercial formulations of CLSM with OPC and an alternative mix with SAC. Resilient modulus tests were carried out to characterise these CLSM mixtures, and to evaluate the effects of testing parameters and conditions on such materials. These mixtures do not contain fly ash or secondary recycled materials in order to control curing time effects. Tests were carried out after 90 days of curing in order to achieve the required mixture stiffness and to minimise stiffness variation at early stages due to the ongoing hydration process.

Materials The materials used in the investigation included two cements: an ordinary Portland Type I/II cement (OPC) and a sulpho-aluminate cement (SAC), a natural sand with a gradation shown in Table 1, and an air-entraining agent in powder form with 0.85 g/cm3 density. The main characteristics of the cements are presented in Table 2. The OPC represents a typical cement used in CLSMs, while the SAC represents the alternative cement that has been shown to effectively reduce setting times and control strength and stiffness gain over time (Bertola et al., 2013).

Table 1 Sand sieve analysis. Sieve

# mm

Passing %

#8 #16 #30 #50 #100 #200

2.360 1.180 0.600 0.300 0.150 0.074

100.0 99.9 1.2 0.2 0.0 0.0

110

M. Bassani et al. / Transportation Geotechnics 2 (2015) 108–118

Table 2 Cement characteristics. Characteristic

Standard

Units

Type I/II

SAC

Blaine fineness #325 Sieve Retained Gilmore time of set Initial Final Vicat time of set Initial Final Pack set index Air content Autoclave expansion Compressive strength 1½ h 3h 1 day 3 days 7 days 28 days

ASTM C204 ASTM C430

cm2/g %

3756 5.4

7529 4.0

ASTM C266

min min

170 300

n/a n/a

ASTM C191 ASTM C1565 ASTM C185 ASTM C151

min min – % %

129 min 231 min 3 6.7 0.011

n/a 13 min n/a n/a 0.021

ASTM C109 on mortar cubes

MPa

n/a n/a 12.97 21.84 28.52 37.09

23.54 24.13 32.65 n/a 45.54 50.61

(35.0%) (58.9%) (76.9%) (100%)

(46.5%) (47.7%) (64.5%) (90.0%) (100%)

Table 3 Mixture composition. Mixture

Units

Mix 1

Mix 2

Mix 3

Cement type Cement Sand 0–2 Air-entraining agent Effective total water Effective water/cement ratio

– kg/m3 kg/m3 kg/m3 dm3/m3 –

OPC Type I/II 180 1300 0.5 202 1.12

OPC Type I/II 160 1350 0.5 202 1.26

SAC 100 1460 0.5 230 2.30

From the hydration point of view, the hardening effect of OPC is associated with the hydration of the calcium silicate, while for the SAC, ye’leemite is combined with calcium sulphate to produce aluminium hydroxide and nonexpansive ettringite (Alizadeh et al., 2014). In addition, SAC has a faster hydration rate due to its greater specific surface area, around twice that of OPC (Table 2). The setting time of SAC according to the Vicat test results was only 13 min, much shorter than that of OPC (300 min). When used in CLSM, both cements exhibit greater setting times than what is measured in the Vicat test. The setting time for SAC is 475–645 min but is still lower than that of OPC, providing necessary and sufficient time for field placement. Compressive strength test results on cubes of reference mortars (according to ASTM C109) demonstrate that SAC develops a very high strength in a relatively shorter time than OPC and that the strength remains relatively constant, while OPC continues to gain strength for a longer period of time. Specifically, SAC develops 64.5% of its 28-day compressive strength in only 1 day, and about 10% from day 7 to 28, while OPC achieves only 35% of its 28-day strength in the first day and around 23% from day 7 to 28. Consequently, the use of SAC instead of OPC may eventually lead to a potential reduction in cement content. Sample preparation and testing Three CLSM mixtures were prepared according to the job mix formulas presented in Table 3. Mix 1 and Mix 2

contain OPC in quantities equal to 180 and 160 kg/m3 respectively, while Mix 3 contains SAC in a lower quantity of 100 kg/m3 due to its higher compressive strength (Table 2). Mixes were prepared using a laboratory portable tilting drum mixer. The dry materials including sand, cement, and the powdered air-entraining agent were first mixed for 30 s. Then, water was slowly poured in within 1 min, and finally the whole mixture was blended for an additional 2 min. During the preparation, the quantity of added water was controlled and modified with respect to the planned mix composition (245 dm3/m3 for Mix 1 and Mix 2 and 230 dm3/m3 for Mix 3), in order to achieve the same fluidity that was observed during the slump test. Thus, Mix 1 and Mix 2 had a lower water content (Table 3). Each mixture was poured into single-use plastic cylinder moulds to form samples of two different dimensions: (a) small sample with a diameter of 100 mm and a height of 200 mm, and (b) big samples with a diameter of 150 mm, and a height of 300 mm which were later sawed to produce a height of 200 mm. The moulds were tightly closed with a plastic lid during the 90 days of curing at 20 °C to avoid any humidity loss and moisture exchange with the environment. After curing, samples were extracted and tested in a UTM-100 triaxial apparatus to measure the resilient modulus MR according to the AASHTO T307 protocol (American Association of State Highway and Transportation Officials, 2007). MR is determined by applying a compressive pulse load on specimens subjected to a constant confining

111

M. Bassani et al. / Transportation Geotechnics 2 (2015) 108–118 Table 4 Mixture volumetrics and densities. Sample

Small D = 100 mm H = 200 mm AR = 2

Large D = 150 mm H = 200 mm AR = 1.33

Mixture

Dry density Cured density Saturated density Moisture content @ cured Moisture content @ saturation Content of water permeable voids Dry density Saturated density Moisture content @ saturation Content of water permeable voids

Units

g/cm3 g/cm3 g/cm3 % % % g/cm3 g/cm3 % %

Mix 1

Mix 2

Mix 3

l

r

l

r

l

r

1.505 1.625 1.630 7.9 8.3 12.5 1.524 1.640 7.6 11.6

– 0.016 0.020 1.1 1.3 2.0 – 0.010 0.7 1.0

1.512 1.613 1.623 6.7 7.3 11.1 1.479 1.628 8.9 13.1

– 0.011 0.007 0.7 0.5 0.7 – 0.029 0.3 0.5

1.311 1.423 1.460 8.5 11.3 14.9 1.294 1.483 14.6 18.9

– 0.021 0.021 1.6 1.6 2.1 – 0.015 1.2 1.5

Fig. 1. Microscopic view of three sections of Mix 1 (A), Mix 2 (B) and Mix 3 (C) samples.

pressure. The standard haversine load pulse lasts 0.1 s, and reaches the maximum cyclic load after 0.05 s. After each pulse, the constant contact load is maintained at a magnitude equal to 10% of the maximum applied load for 0.9 s, forming a total cycle duration of 1 s. Two additional loading rates were also considered in the investigation: (a) haversine pulse with 0.056 s loading time and a cycle duration of 0.504 s, and (b) haversine pulse with 0.5 s loading time and a cycle duration of 4.5 s. To avoid any uneven transmission of the applied loads that may lead to local shear stresses on the top and bottom of the sample, end friction reducing elements consisting of two latex sheets separated by silicon grease were used between the loading platens and the ends of the specimen. MR is calculated as the ratio of the peak axial repeated deviator stress (rd) to the recoverable (resilient) axial strain of the sample. The base/subbase materials testing sequence of AASHTO T307 (American Association of State Highway and Transportation Officials, 2007) were used to evaluate the resilient properties of the investigated CLSMs. According to the standard, 500–1000 of conditioning cycles at 103.4 kPa maximum axial and confining stresses (r3) were applied before the fifteen test sequences of different axial and confining stress conditions consisting of 100 cycles per sequence. The same sample was tested in three different moisture conditions in the following order: (a) ‘‘cured’’ condition referred to the test performed immediately after the extraction of sample from the sealed plastic mold after

90 days of curing; (b) ‘‘dry’’ condition after 5 days of airdrying; and (c) ‘‘saturated’’ condition after immersing the sample for 24 h in water. The larger samples (diameter = 150 mm, height = 300 mm) were sawed to a final height of 200 mm to achieve an aspect ratio (AR = height/diameter) of 1.33. A diamond saw was used for this purpose when the samples were still in the plastic moulds to preserve their integrity. The water used in the sawing process modified the water content of the samples, thus, no data were available for the cured conditions both for density and for resilient modulus tests. Sample volumetrics The volumetric properties of the three mixtures are summarised in Table 4 for the two sample sizes considered in this study. Microscopy analysis was also carried out to explore the internal structure of the three CLSM mixtures. As can be seen from Fig. 1, the air-entraining agent has worked similarly in the three mixtures creating a honeycomb structure in the cement matrix, in which the pores are isolated from each other. The air-entrained voids have diameters between 70 and 150 lm, which correspond to an average cross section of 3 * 104 lm2. The cement matrix is perfectly bonded to the aggregate grains, and the aggregate surface is partially visible in all three images of Fig. 1. From the image analysis, it can be observed that some randomly distributed larger and irregularly shaped

112

M. Bassani et al. / Transportation Geotechnics 2 (2015) 108–118

entrapped air voids are present in all three mixtures. Such voids are located between the aggregate grains and do not appear to be part of an interconnected void network. Consequently, the absorption capacity of the samples is limited to those superficial voids. This conclusion is also supported by the volumetric analysis (Table 4) where it is shown that the volume of voids permeable to water is very small thus leading to very low densities when compared to CLSMs mixtures containing fly ash (Folliard et al., 2008). Compressive strength The term controlled low-strength is usually referred to cement mortars whose maximum unconfined cubical compressive strength (UCCS) is below 8.27 MPa (1200 psi) (Deng and Tikalsky, 2008). Compressive strengths from 0.35 to 0.69 MPa are typical of materials with a bearing capacity of high quality compacted aggregate (American Concrete Institute, 2008). According to ACI 229R-99 (American Concrete Institute, 2008), the maximum compressive strength that ensures future excavatability is 1.38 MPa (200 psi), although the literature includes the case of a CLSM with a UCCS of 2.1 MPa (300 psi) containing fine sand and fly ash that was excavated with a backhoe (Krell, 1989). To classify CLSMs, the Removability Modulus (RE) (Folliard et al., 2008) is also used to estimate the longterm excavatability according to the formula:

RE ¼ 0:619  106  c1:5  UCCS0:5

ð1Þ

3

where c is the density in kg/m , and UCCS is in kPa. When RE is lower than 1, it indicates that the CLSM will be easily removable using excavators. Table 5 reports the results of the compressive strength carried out on cylindrical specimens (UCS) for the three mixtures and the two AR. The UCCS which is necessary to evaluate RE (Eq. (1)), have been estimated according to Lamond et al. (2006):

UCCS ¼ f AR  f cy=cu  UCS

ð2Þ

where fAR is the factor that converts strength data of a cylindrical sample with AR < 2 into the values of a slender cylinder (AR P 2) and fcy/cu is the factor that converts the strength of a slender cylinder into cubic sample strength. Data in Table 5 were derived assuming a fAR value of 1.06 Table 5 Summary of compressive strength and removability modulus results after 90 days of curing time. Removability modulus (RE)

MPa

Estimated unconfined cubical compressive strength (UCCS) MPa

2.194 1.543

2.802 1.859

1.949 1.559

1.33 1.55

1.933 1.215

2.470 1.464

1.750 1.393

1.25 1.48

0.169 0.242

0.216 0.291

0.423 0.502

Mix

AR

Strain at failure

(%) 1

1.33 2

1.67 1.82

2

1.33 2

3

1.33 2

Unconfined cylindrical strength (UCS)

–

for the case of samples with AR equal to 1.33, and a value of 1.205 for fcy/cu. Mix 1 and Mix 2 have compressive strengths greater than the maximum suggested value for excavatability, as confirmed by the RE modulus which is greater than 1. Mix 3 has a very low compressive strength with an RE modulus ranging from 0.479 to 0.502. With the exception of Mix 2, the AR does not seem to have the same effect on strength results for all the mixtures. Similar values of strain at failure have been recorded in the range 1.14– 1.85% for all the mixtures.

Results and discussion A summary of resilient modulus testing results is presented in Table 6. The average (l) and standard deviation (r) of MR was calculated based on the testing results at three moisture conditions (air dry, cured and saturated), two AR (1.33, 2), and three different levels of rd: (a) low level, equal to 70 ± 10 kPa, (b) medium level, equal to 140 ± 10 kPa, and (c) high level, equal to 280 ± 10 kPa. The results show that samples with AR equal to 2 exhibit higher MR values than those with an AR equal to 1.33. No large differences were observed between the air dry and saturated conditions. Surprisingly, samples of Mix 1 in the cured conditions had an MR significantly lower than the other two conditions. Lower values were also observed in the case of Mix 3 for low and medium rd. Furthermore, the test results on the samples with 150 mm diameter (AR = 1.33) exhibit a lower standard deviation (r) in comparison to the results from the 100 mm samples (AR = 2). It appears that for larger samples the testing device is able to provide a less variable load, limiting dispersion around the average values. The effects of stress dependency, aspect ratio, saturation conditions, loading time, and loading repetitions on resilient modulus were examined next.

Stress parameters dependency The influence of the stress parameters on MR was examined and example results are presented in Fig. 2 (large sample of Mix 1 in dry condition). The figure shows that a relationship exists between MR and rd, r3, and bulk stress (h) which is equal to rd + 3r3. The results show that MR of CLSM is stress dependent, with a strong influence of rd, a moderate influence of h, and a weaker influence of r3, according to the corresponding coefficient of correlation (R2). Similar results were observed for other samples, mixtures, and moisture conditions. The MR of CLSM is a week function of r3 reflecting the rigidity of the cement matrix (Puppala et al., 2011). The MR has a stress hardening dependence on rd, and h, with the dependency being stronger with rd (with R2 = 0.99). This is dissimilar to that of UGM whose stress hardening MR is mainly governed by h Lekarp et al. (2000). Since h is per se a function of rd and r3, the low dependence on r3 justifies the MR of CLSM being mainly governed by rd as opposed to h.

113

M. Bassani et al. / Transportation Geotechnics 2 (2015) 108–118 Table 6 Summary of MR results in MPa. Conditions

Air dry

Mix

AR

1

1.33 2

2

1.33 2

3

1.33 2

l r l r l r l r l r l r

Cured

Saturated

L

M

H

L

M

H

L

M

H

224 34.5 323 57.0 209 25.8 249 60.4 189 15.4 239 72.9

298 30.7 408 25.3 283 23.5 330 71.3 238 15.9 303 96.6

374 19.6 519 31.1 365 7.5 447 49.4 309 5.8 373 81.9

– – 184 49.1 – – 305 44.7 – – 173 26.2

– – 225 34.1 – – 373 40.0 – – 230 29.9

– – 340 36.2 – – 479 44.8 – – 367 37.0

214 21.1 313 31.5 174 16.1 284 40.6 194 19.2 218 73.3

288 21.4 394 23.8 233 12.0 381 41.5 252 10.2 281 88.9

369 8.7 518 11.2 308 5.5 498 42.4 327 7.3 382 68.1

Note: L = low rd (70 ± 10 kPa), M = medium rd (140 ± 10 kPa), H = high rd (280 ± 10 kPa).

Sample aspect ratio effects The effect of aspect ratio on resilient modulus was examined for the three mixtures at various moisture conditions. Overall, higher modulus and stress dependency was observed for samples with 100 mm diameter and 200 mm height (AR = 2). The results for Mix 2 and for airdry and saturated conditions are shown in Fig. 3. A similar tendency was observed for other mixtures and moisture conditions.

Saturation condition effects

Fig. 2. Resilient modulus as a function of stress parameters.

Fig. 4 shows the MR test results obtained on a sample with AR = 2 for Mix 2. The reported MR values represent the mean value of resilient modulus during the last five pulses of each sequence. The standard deviation of the last

Fig. 3. Resilient modulus evolution for Mix 2 as a function of different aspect ratios.

114

M. Bassani et al. / Transportation Geotechnics 2 (2015) 108–118 700

Resilient modulus (average) [MPa]

200

Resilient modulus (standard deviation) [MPa]

600

180

R² = 0.9633

160

500

140 R² = 0.8119 120

400

R² = 0.8266

Average air-dry Average cured Average saturated St.dev. Air-dry St.dev. cured St.dev. saturated Air-dry Cured Saturated

300

200

100 80 60 40

100 20 0

0 0

50

100

150 Deviatoric stress [kPa]

200

250

300

Fig. 4. Resilient modulus as a function of degree of saturation.

five pulses is shown as well in Fig. 4 in the second y-axis. A linear interpolation is used to fit the average values of the three different saturation conditions: air-dry, cured, and saturated. No specific trends are observed between the different water content. The standard deviation results indicate that the variation decreases when the deviatoric stress increases, a conclusion that supports the concept that at higher loads the testing device produces more stable loading conditions. According to the volumetric and image analyses, voids are primarily concentrated in the cement matrix and are isolated from larger entrapped voids in the sample. Water permeation is limited to the entrapped voids, thus different saturation conditions should not significantly affect the resilient properties of the investigated CLSMs. Any variation in the results may be attributed to random deviations. In contrast, soils and UGMs are more sensitive to water content and saturation level (Lekarp et al., 2000). MR tends to decrease with a growing saturation level due to the increase in pore-water pressure and decrease in effective stress, as well as the lubricating effect of water. The same effect can be inferred in relation to the pore suction, which decreases with a higher water content. The effect of water content on MR of UGMs is certainly affected by the size distribution of particles, which also governs the dimension of internal pores, and consequently the suction phenomena.

Loading time duration effects Four different loading times were selected to evaluate any dependency of resilient properties on loading effects. Load durations ranged from 0.056 to 0.5 s, and included 0.1 s and 0.25 s. Fig. 5 shows the results of two tests performed on a small scale sample (AR = 2) of Mix 1 in cured

condition, and on a larger sample (AR = 1.33) of Mix 3 in dry condition. All the recorded test results (i.e., five per stress level) have been plotted to highlight dispersion of data for each stress level as indicated by AASTHO T307 (American Association of State Highway and Transportation Officials, 2007) testing protocol for subbase materials. The results show that these CLSMs are not sensitive to changes in loading frequency or duration. According to Lekarp et al. (2000) similar conclusions were drawn for granular materials at almost all moisture conditions. Granular materials exhibit a stiffness reduction with an increase of load frequency only at moisture contents approaching saturation due to the development of excess pore water pressures and the consequent reduction of effective stress (Hicks and Monismith, 1971).

Loading cycles effects Fig. 6 shows the average permanent strain and the resilient modulus obtained on dry small samples (AR = 2) of Mix 1, 2 and 3 in the range from 1 to 106 pulses under repeated load triaxial test at a frequency of 5 Hz. Testing conditions also considered a loading time of 0.1 s, a deviatoric stress of 280 kPa, and a confining pressure of 140 kPa. During the test, after 10–15 cycles the testing parameters stabilized at these predefined values. It can be observed that at the start of the testing phase all materials accumulate a significant amount of total permanent strain, but no sample failure occurred during the tests. The results also indicate that CLSM is able to reach stable conditions after few loading applications. Mix 1 and Mix 2 accumulate a permanent deformation at a significantly lower rate than UGM under identical stress conditions. Werkmeister et al. (2004) investigated UGM of granodiorite at a dry density of 2.26 g/cm3 and a moisture content of 4.0% in the same repeated load triaxial test. The UGM may reach compaction during this initial phase,

115

M. Bassani et al. / Transportation Geotechnics 2 (2015) 108–118 600

Resilient Modulus [MPa]

500

400

300

Mix 1, sample 6, LT = 0.056 s Mix 1, sample 6, LT = 0.1 s

200

Mix 1, sample 6, LT = 0.25 s Mix 3, sample 1, LT = 0.056 s Mix 3, sample 1, LT = 0.1 s

100

Mix 3, sample 1, LT = 0.5 s Deviatoric Stress [kPa] 0 0

50

100

150

200

250

300

350

Fig. 5. Resilient modulus as a function of loading time (LT).

100

600

500

Resilient modulus [MPa]

Average permanent strain [103•m/m]

10

1

300

200 CLSM Mix 1

CLSM Mix 1 CLSM Mix 2 CLSM Mix 3 Granodiorite

0.1

400

CLSM Mix 2 CLSM Mix 3

100

Cycle #

Cycle # 0.01 1.E+00

1.E+02

1.E+04

1.E+06

0 1.E+00

1.E+02

1.E+04

1.E+06

Fig. 6. Average permanent strain and resilient modulus as a function of number of pulsing loads repetitions.

while for the strongest CLSM mixtures (Mix 1 and Mix 2) the permanent strain increment is more gradual reflecting the low compressibility of the honeycomb structure (Fig. 1). These results confirm field observations where CLSMs are widely appreciated in trench backfilling and pavement repairs since they avoid any troublesome settlement and failure of the pavement under load. The strain rate of Mix 1 and Mix 2 remained small at a million of loading cycles (Table 7). In the case of Mix 3, which is much softer than the other two CLSM mixtures as a result of differences in composition (Table 3), it showed higher densification and higher permanent strain accumulation in the preliminary cycles but the rate of strain accumulation decreased and eventually became

constant after 102 cycles. In the case of UGM, the rate of permanent strain increased after 2105 cycles and the sample observed shear failure albeit showing a lower densification in the preliminary stage as compared to Mix 3. MR increased for all the CLSM samples as a consequence of a decrease in resilient vertical strain during the repeated loading.

Modelling of resilient modulus A number of computational models for the resilient behaviour of unbound granular materials have been proposed over the years. Lamond et al. (2006) and Andrei

116

M. Bassani et al. / Transportation Geotechnics 2 (2015) 108–118

Table 7 Permanent strain and average permanent strain rate at a deviatoric stress of 280 kPa and a confinement pressure of 140 kPa for CLSMs and Granodiorite.

*

Cycle interval

Average permanent strain rate [103 m/m/cycle] Mix 1

Mix 2

Mix 3

UGM*

–

1550 501104 11042.5104 2.51045104 51041105 11052105 21055105 5105106

– 1.1429 0.0181 0.0024 0.0036 0.0012 0.0006 0.0001 0.0001

– 501.9326 0.2161 0.0158 0.0231 0.0063 0.0024 0.0004 0.0006

– 2.8571 0.0704 0.0100 0.0160 0.0040 0.0010 0.0017 Failure

5.1956 0.0540 0.0037 0.0013 0.0062 0.0004 0.0012 0.0000

Values estimated from graph reported in Lamond et al. (2006).

provides the more conservative and repeatable results. Results from air-dry and water-saturated conditions have been included and combined since the materials have proved to be insensitive to water content. Mix 3 exhibited a lower stiffness as compared to Mix 1 and Mix 2 due to differences in composition (Table 3). Table 9 includes the mean (l) and standard deviation (r) of ki values and the coefficient of determination (R2) of the fitted model. The results in Table 9 demonstrate that the best fitting for the investigated CLSM mixtures is provided by the M-EPDG model included in the NCHRP 1-37A (Andrei et al., 2004). The Uzan model (Uzan, 1985) is able to fit the data as well since it requires only three parameters, while the model of Barksdale and Itani (1989) is less accurate despite the use of four parameters. Similarly, Andrei et al. (2004) observed that in the case of granular materials, the predictive model with the best goodness-of fit employs the highest number of parameters. The M-EPDG model considers the most significant stress variables that were highlighted in the Section ‘‘Stress parameters dependency’’. A comparison of resilient properties of CLSMs with unbound granular materials (Kim and Kim, 2007), stabilized clay with fly ash and cement kiln dust (Solanki et al., 2010) is included in Table 10, and contains the ranges of values for the Uzan (Uzan, 1985) and M-EPDG model

Table 8 Synthesis of models. References to authors

Variables

Restrictions on ki

Hicks and Monismith (1971) Uzan (1985) M-EPDG model Design Guide: Design of New and Pavement Structures (2004) Thompson and Robnett (1979) Barksdale and Itani (1989)

h h, rd h, rd (or soct)

k3 = k4 = k5 = k6 = 0 k4 = k5 = k6 = 0 k4 = k6 = 0, k5 = 1

rd rd, r3, p

k2 = k4 = k5 = k6 = 0 k2 = k5 = 0

et al. (2004) have presented an extensive comparison of various resilient modulus models. In this work, the most popular models were considered for the analysis as shown in Table 8. A generalised formula that considers the combination of h, r3, rd, and average atmospheric pressure (Pa) of 101 kPa is represented as following:

log M R ¼ logðk1 Pa Þ þ k2 logðh=P a Þ þ k3 logðrd =Pa þ k5 Þ þ k4 logðr3 =Pa Þ þ k6 logðP=Pa Þ

ð3Þ

where MR, h, Pa, rd and r3 are in kPa. Pa is used to generate dimensionless calibration factors (ki) in Eq. (3). Table 9 reports the calibration results from the tests carried out on the large samples (AR = 1.33) which

Table 9 Synthesis of model calibration for AR = 1.33. Model

Hicks and Monismith (1971) Uzan (1985)

M-EPDG model Design Guide: Design of New and Pavement Structures (2004)

Thompson and Robnett (1979) Barksdale and Itani (1989)

Mix 1

k1 k2 k1 k2 k3 k1 k2 k3 k5 k1 k3 k1 k3 k4 k6

Mix 2

Mix 3

l

r

R2

l

r

R2

l

r

R2

1313.11 0.4990 2102.79 0.2660 0.1872 1398.38 0.3009 0.4965 1.0000 2608.99 0.3410 2170.21 0.0894 0.2742 0.6381

123.4 0.0477 327.7 0.0869 0.0457 137.5 0.0911 0.1380 – 127.7 0.0193 240.6 0.1206 0.3261 0.4098

0.870

1168.67 0.5392 1876.98 0.2978 0.1943 1243.73 0.3403 0.4927 1.0000 2433.02 0.3663 2211.36 0.1824 0.0401 0.3495

147.7 0.0588 143.7 0.0327 0.0229 149.3 0.0419 0.0375 – 145.0 0.0375 115.3 0.1180 0.2579 0.3249

0.774

1145.59 0.4628 1972.76 0.2016 0.2101 1232.32 0.2266 0.5922 1.0000 2168.38 0.3271 1715.17 0.0446 0.4594 0.8266

59.6 0.0210 561.3 0.1144 0.0908 77.1 0.0918 0.2156 – 57.1 0.0289 334.9 0.2080 0.3676 0.5860

0.835

0.938

0.942

0.867 0.937

0.844

0.848

0.755 0.824

0.939

0.955

0.871 0.931

117

M. Bassani et al. / Transportation Geotechnics 2 (2015) 108–118 Table 10 Synthesis of model calibration for AR = 1.33. Model

CLSM

Uzan (1985)

M-EPDG model Design Guide: Design of New and Pavement Structures (2004)

k1 k2 k3 k1 k2 k3

UGM (1)

Stabilized clay (2)

Min

Max

Min

Max

Min

Max

1877 0.202 0.187 1232 0.227 0.493

2103 0.298 0.194 1398 0.340 0.592

100 0.250 1.100

1400 1.200 0.100

803 0.046 0.635 1425 0.362 2.424

22399 0.403 0.100 26139 0.275 0.364

Data from (1) Kim and Kim (2007), and (2) Solanki et al. (2010).

parameters (Design Guide: Design of New and Pavement Structures, 2004). A change in sign is evident for the k3 parameter when comparing CLSM with other materials. k3, which is the factor of MR dependency on rd is positive in the case of CLSM and negative in the case of stabilized clay. The MR is rd hardening in the case of CLSM (see Fig. 2) while it is rd softening for stabilized clay. In the case of UGM, the deviator or shear stress is considered much less influential on material stiffness than other stress variables (Lekarp et al., 2000; Nazarian et al., 1996). Some observers have detected a slight softening dependency for low stress level, and a slight hardening at high stress levels (Hicks and Monismith, 1971); others have observed a hardening effect of rd (Uzan, 1999; Heydinger et al., 1996).

the values measured on specimens with 150 mm diameter and 200 mm height (AR = 1.3). The examined CLSMs were insensitive to water saturation due to their low void content and limited interconnected voids. A large amount of impermeable pores is located in the cement matrix between solid particles, as monitored by microscopic analysis, but these form a honeycomb structure that does not easily enter the moisture. This gives the material the ability to provide consistent behaviour at different water contents. Overall, the stiffness of CLSMs is not sensitive to variation in water content and loading frequency. The MR is rd hardening, meaning it increases by an increase in rd. The resilient properties tend to improve with increasing load repetitions.

Conclusions

Acknowledgements

This study examined the resilient characterisation of alternative CLSM formulations that do not include fly ash. Of the three mixtures examined, two were prepared with ordinary Portland cement while the third one with an ultra-rapid sulpho-aluminate cement. Different mix compositions were adopted in the sample preparation. Samples with different aspect ratios of the three CLSM mixtures were prepared and subjected to different testing conditions in terms of stress states, loading frequencies, loading repetition and saturation level. Similar to granular materials, CLSMs seem to be insensitive to variations of loading frequency. They both have nonlinear stress dependent resilient moduli. In case of CLSM, the resilient modulus significantly increases by the increase in deviatoric stress and is to lesser extent affected by the confining pressure dissimilar to UGM. CLSM mixtures exhibit a significant increase in stiffness under repeated load triaxial testing. The comparison with granular materials indicated that the rate of permanent strain accumulation is also very low, which implies that the use of these CLSM formations can be successful in applications where there is a need to limit the settlement (i.e., pavement repairs, trench backfilling, bridge approach). The effects of various stress conditions on mixture behaviour were further examined. The results indicated a high variation of resilient modulus as a function of specimen AR. In particular, small and slim samples with diameter of 100 mm and 200 mm height (AR = 2) exhibited a resilient modulus higher than

The research work included in this paper was possible thanks to the funding support of the Compagnia di San Paolo – Italy – and the Politecnico di Torino with the call titled: ‘‘Bando per il Finanziamento di Progetti di Internazionalizzazione della Ricerca’’, approved with the Rectoral Decree n. 208 of the 24th of May, 2013. Testing was carried out in the Laboratory of the Department of Civil and Environmental Engineering of the University of Maryland (College Park, US), which hosted Dr. Bassani as a visiting professor during the investigation. Buzzi Unicem USA Inc. is greatly acknowledged for the support in supplying the materials.

References Alizadeh V, Helwany S, Ghorbanpoor A, Sobolev K. Design and application of controlled low strength materials as a structural fill. Constr Build Mater 2014;53:425–31. American Association of State Highway and Transportation Officials. Standard method of test for determining the resilient modulus of soils and aggregate materials. Washington, DC: AASHTO; 2007. AASHTO designation: T 307-99. American Association of State Highway and Transportation Officials. Mechanistic-empirical pavement design guide. A manual of practice. Interim ed. Washington, DC: AASHTO; 2008. American Concrete Institute. Controlled Low-Strength Materials. ACI manual of concrete practice, ACI 229R–99, 2008. Andrei D, Witczak MW, Schwartz CW, Uzan J. Harmonized resilient modulus test method for unbound pavement materials. Transp Res Rec 2004;1874:29–37. Barksdale RD, Itani SY. Influence of aggregate shape on base behavior. Transp Res Rec 1989;1227:173–82.

118

M. Bassani et al. / Transportation Geotechnics 2 (2015) 108–118

Bertola F, Bassani M, Canonico F, Bianchi M. Use of rapid-hardening cement for Controlled Low-Strength Materials for pavement applications. Transp Res Rec 2013;2363:77–87. Deng A, Tikalsky PJ. Geotechnical and leaching properties of flowable fill incorporating waste foundry sand. Waste Manage 2008;28:2161–70. 2002 Design Guide: Design of New and Rehabilitated Pavement Structures, Draft Final Report, National cooperative highway research program, NCHRP Study 1-37 A, Washington, DC; 2004. Du L, Folliard KJ, Trejo D. Effects of constituent materials and quantities on water demand and compressive strength of controlled lowstrength material. J Mater Civ Eng 2002;14(6):485–95. Etxeberria M, Ainchil J, Pérez Ma E, González Alain. Use of recycled fine aggregates for Control Low Strength Materials (CLSMs) production. Constr Build Mater 2013;44:142–8. Folliard KJ, Du L, Trejo D, Halmen C, Sabol S, Leshchinsky D. Development of a recommended practice for use of controlled low-strength material in highway construction. NCHRP report 597. Washington, DC: Transportation Research Board; 2008. Heydinger AG, Xie Q, Randolph BW, Gupta JD. Analysis of resilient modulus of dense- and open-graded aggregates. Transp Res Rec 1996;1547:1–6. Hicks RG, Monismith CL. Factors influencing the resilient properties of granular materials. Highway Res Rec 1971;345:15–31. Kim D, Kim JR. Resilient behavior of compacted subgrade soils under the repeated triaxial test. Constr Build Mater 2007;21:1470–9.

Krell WC. Flowable fly ash. Concr Int 1989;11(11):54–8. Lamond J, Peilert J. Significance of tests and properties of concrete & concrete making materials. ASTM STP 169D, West Conshohocken, PA; 2006. Lekarp F, Isacsson U, Dawson A. State of the art I: resilient response of unbound aggregates. J Transp Eng 2000;126(1):66–75. Nazarian S, Pezo R, Picomell P. Testing methodology for resilient modulus of base materials. Texas Department of Transportation; 1996. Report TX-94 1336-1. Puppala AJ, Hoyos LR, Potturi AK. Resilient moduli response of moderately cement-treated reclaimed asphalt pavement aggregates. J Mater Civ Eng 2011;23(7):990–8. Solanki P, Zaman MM, Dean J. Resilient modulus of clay subgrades stabilized with lime, class C fly ash, and cement kiln dust for pavement design. Transp Res Rec 2010;2186:101–10. Thompson MR, Robnett QL. Resilient properties of subgrade soil. J Transp Eng 1979;105(1):71–89. Turkel S. Strength properties of fly ash based controlled low strength materials. J Hazard Mater 2007;147:1015–9. Uzan J. Characterization of granular material. Transp Res Rec 1985;1022:52–9. Uzan J. Granular material characterization for mechanistic pavement design. J Transp Eng 1999;125(2):108–13. Werkmeister S, Dawson AR, Wellner F. Pavement design model for unbound granular materials. J Transp Eng 2004;130(5):665–74.