Journal of Non-Newtonian Fluid Mechanics 214 (2014) 18–27

Contents lists available at ScienceDirect

Journal of Non-Newtonian Fluid Mechanics

journal homepage: ht tp : / /www.elsevier .com/locate / jnnfm

Measurement of yield stress of cement pastes using the direct shear test

http://dx.doi.org/10.1016/j.jnnfm.2014.10.0090377-0257/� 2014 Elsevier B.V. All rights reserved.

⇑ Corresponding author.E-mail addresses: jassaad@ndu.edu.lb (J.J. Assaad), jharb@ndu.edu.lb (J. Harb),

ymaalouf@ndu.edu.lb (Y. Maalouf).

Joseph J. Assaad a,⇑, Jacques Harb b, Yara Maalouf c

a Holderchem Building Chemicals, PO Box 40206, Lebanonb Civil Engineering Department, Notre Dame University, PO Box 72, Lebanonc Notre Dame University, PO Box 72, Lebanon

a r t i c l e i n f o a b s t r a c t

Article history:Received 8 April 2014Received in revised form 19 October 2014Accepted 21 October 2014Available online 30 October 2014

Keywords:Cement pasteCohesionDirect shearVaneYield stress

The direct shear test is widely used in soil mechanics to determine the cohesion (C) and angle of internalfriction (/). This paper aims to assess the suitability of this test to evaluate yield stress (s0) of cementpastes having different flowability levels. Special emphasis was taken to eliminate friction between shearboxes, thus allowing the measurement of C ranging from several kPa to just a few Pa. Tests have shownthat the maximum shearing stress prior to failure is not a material constant, but rather varies with thenormal stress as per the Mohr–Coulomb law. Good correlations between C and s0 determined usingthe vane method were established. Nevertheless, the vane method was found to over-estimate s0 whenthe blades are positioned inside the specimen, particularly for cohesive materials.

� 2014 Elsevier B.V. All rights reserved.

1. Overview on yield stress measurements

The yield stress (s0) is of interest for various industries; it isregarded as the transition stress between elastic solid-like behav-ior and viscous liquid-like behavior [1]. The measurement of s0 isgenerally performed using direct rheometric techniques that con-sist of slowly shearing the material and recording the peak shearstress required to initiate flow. However, such measurements arenot easy to implement, given the need to accurately monitor vari-ations of shear stresses at low shear rates [2–4]. Most importantly,the wall-sample interactions can result in slip effects associatedwith displacement of the dispersed phase(s) away from the bound-aries, resulting in a low-viscosity particle-depleted layer near thewall and under-estimation of s0 [5,6]. The probability of wall slipincreases when dealing with smooth walls, relatively small gaps,low flow rates, and concentrated suspensions of large and floccu-lated particles. A common way to reduce the extent of slip is toroughen the wall’s surface in order to increase friction with thesuspension [7,8]. It is to be noted that s0 can also be determinedusing indirect techniques that consist of extrapolating to zeroshear rate a series of shear stress vs. shear rate rheological data.Nevertheless, such measurements are very sensitive to theassumed constitutive model as well as the accuracy and range ofexperimental flow data especially at low shear rates [2,4].

Over the last decades, various techniques have been more orless successfully developed to overcome the complications relatedto wall slip and enable reliable measurement of s0 [9–13]. Themost popular techniques were those realized under quasi-staticconditions and whose basic principle requires that shearing takesplace within the material itself, i.e. not between the material andan object. Hence, ideally, this requires that a virtual plane of mate-rial should move inside the suspension, and the material–materialshearing stresses recorded at low shear rates [9,10,14]. The peakshear stress needed to initiate flow can thus be considered as the‘‘true’’ s0 of tested material.

The vane method is probably the most popular for measurings0, since slip is physically impossible and shearing completelyoccurs within the material [6,9,15]. Its concept originated from soilmechanics, where vanes are used to determine shear strength ofsoils as described in ASTM D2573 [16]. Hence, a four, six, oreight-bladed vane of diameter D and height H, connected to astress-controlled rheometer is fully immersed in the material androtated at sufficiently low shear rate to determine the maximumtorque required to initiate flow. Nguyen and Boger [9] suggesteda series of criteria for satisfactory s0 measurements of various con-centrated suspensions including H/D < 3.5, DT/D > 2, Z1/D > 1, andZ2/D > 0.5 (see Fig. 1 for notations). Elsewhere, Nguyen and Boger[3] reported that Z1 + H + Z2 > 2H. Alderman et al. [17] utilized aset-up where Z2 = H, Z1 = ½ H, and DT = 3D.

To calculate s0, the maximum torque (Tm) is taken as the alge-braic sum of shear stress exercised by the lateral area (Ts) andthe vane’s upper and lower areas (Te), such that:

J.J. Assaad et al. / Journal of Non-Newtonian Fluid Mechanics 214 (2014) 18–27 19

Tm ¼ Ts þ 2Te:

In terms of shear stress, the torque can be written in Eq. (1) as:

Tm ¼p2

D2ss

� �H þ 4p

Z D=2

0ser2dr ð1Þ

where r is a radial coordinate, se is the shear stress on the upper andlower circular ends of the cylinder, and ss is the stress on the curvedcylindrical surface. Assuming that yielding occurs at the cylindricalsurface defined by the tips of the blade, and that se and ss are uni-form and equal to s0 at maximum torque, Tm becomes equal to:

Tm ¼pD3

2

!HDþ 1

3

� �s0 ð2Þ

During s0 evaluation of various emulsions, Yoshimura et al. [18]considered positioning the top edges of the blades vane alignedwith the upper material’s surface, so as to eliminate the stress con-tribution from emulsion located above the blades on torque mea-surements. Consequently, Tm becomes equal to Ts + Te and thefactor (1/3) in Eq. (2) is replaced by (1/6), as follows:

Tm ¼pD3

2

!HDþ 1

6

� �s0 ð3Þ

To measure s0 at very low shear rates, Zhu et al. [10] and Zhanget al. [14] developed the plate technique in which a slotted platesubmerged in a test material is pulled out slowly while measuringthe required load that comes from the material’s resistance to thismotion. The plate was hung to a balance through very thin stainlesssteel wires and its velocity precisely controlled from 0.003 to60 mm/min [10,14]. The height of slots was at least 100 times largerthan the maximum particle size in the suspension. This strengthensthe assumptions that the suspension remains static in the slots withno secondary flow, and that shearing occurs only at the slot edges

Fig. 1. Notations used for vane configuration.

(i.e., material in slot shearing against material in bulk). Zhanget al. [14] considered that, unlike the vane method, the plate tech-nique does not rely on an assumed yield surface area. The s0 wascalculated as Fnet = (F � Fi) divided by the slotted plate area; whereF refers to the force recorded by the balance, and Fi is the initialforce reading calculated as the gravitational force due to plate andwire mass minus the buoyant force in suspension.

The plate technique was found adequate to determine s0 of var-ious non-Newtonian fluids such as bentonite and TiO2 suspen-sions; however, several difficulties were encountered whentesting cement pastes [14]. For example, s0 could be over-esti-mated if the plate is not in a fully vertical position during testing,given that the force measured would be higher than that for a ver-tical plate. Another difficulty in determining s0 occurs in caseswhen it is important to use a correction factor for edge effects[14]. The determination of this factor is time-consuming, as itrequires the use of various plate sizes and batching of cementpastes with different water-to-cement ratios (w/c).

Assaad and Harb [19] proposed using the triaxial and uncon-fined compression tests to overcome the complications related toslip effects, secondary flow, or confinement conditions encoun-tered in rheometric techniques. These tests are widely used in geo-technical applications to analyze the soil’s shear strengthproperties, including cohesion (C) and angle of internal friction(/), and are standardized by ASTM D2166, D2850, and D4767[20–22]. Two main drawbacks were however attributed to thesetests, including a considerable time needed for specimen prepara-tion (i.e., around 15–20 min) and inadequacy of testing flowablemixtures having a flow exceeding around 140 mm, as per ASTMC1437 [23]. A cohesion threshold of around 4 kPa was determinedon tested mortars, below which the specimens are no longer capa-ble to self-stand in a vertical position for testing [19]. Tests realizedunder drained conditions displayed higher C values than those per-formed under undrained ones, given the resulting increase in fric-tion generated between solid particles within the matrix.

2. Use of direct shear to measure s0

The direct shear test is the oldest and simplest method used insoil mechanics to determine the C and / parameters, and analyzefailure mechanisms occurring along interfaces [24–26]. The proce-dure for specimen preparation is quite simple, and drawbackrelated to verticality encountered in triaxial and unconfined com-pression tests is not present [19]. In this test, two portions of a spec-imen are made to slide along each other by the action of steadilyincreasing horizontal shearing force while a constant load is appliednormal to the plane of relative movement. The direct shear test isrealized under quasi-static conditions, and shearing takes placewithin the material itself along a pre-defined interface representedby the horizontal surface area of the shearing box. This physicallyenables the determination of ‘‘true’’ s0, since all problems relatedto wall slip and secondary flow are eliminated. The direct shear isstandardized equipment documented in ASTM D3080 [27] andavailable in most research centers.

Besides its use in geotechnical applications, the direct shear testhas been popular when studying rheology of extrudable materialslike plastics, fiber composites, rubbers, clays, and asbestos [28,29].In fact, conventional shear-driven rheometers such as parallel plates,rotors, and concentric cylinders are not adapted to measure rheologyof highly cohesive pastes due to the difficulty of sample preparation,wall slip, and plug flow. Alfani and Guerrini [29] reported that thedirect shear is among the most promising and suitable methods forrheological characterization of cement-based extrudable materialsincluding the interfacial flow behavior between the bulk materialsand equipment forming wall systems. For adequate extrusion, Toutouet al. [30] found that s0 has to be high enough to allow the material to

20 J.J. Assaad et al. / Journal of Non-Newtonian Fluid Mechanics 214 (2014) 18–27

retain its shape when leaving the extruder; the common minimumusable s0 is evaluated at around 20 kPa.

Limited studies exist in literature pertaining to the suitability ofdirect shear to assess s0 of cementitious materials possessing flow-able nature. L’Hermite and Tournon [31] were among the fewresearchers who considered this test to study shear resistance ofnormal-consistency fresh concrete. Keeping the displacement rateconstant, the authors found that shear stress increases linearlywith the degree of distortion up to a maximum value, thendecreases towards a steady state region. A linear relationshipbetween normal and shear stresses was established following theMohr–Coulomb law:

s ¼ C þ r0 tan / ð4Þ

The s and r0 in Eq. (4) refer to shear resistance and normal effec-tive stress resulting from the solid grains, respectively.

Hendrickx et al. [32] reported that s0 of relatively stiff mortarsintended for masonry and brick-laying applications are difficult tomeasure using traditional rheometric techniques; geotechnicalmethods such as the vane and direct shear should instead be usedto quantify shear strength. The authors reported that the maxi-mum stress in an undrained mortar is not a material constant,but rather depends on the rate of stress increase and normal loadapplied. Tested mortars exhibited C values ranging from 1 to3 kPa, whereas the / values varied from 25� to 47� [32]. Girishand Santhosh [33] developed a 150-mm cubic direct shear appara-tus to evaluate the Bingham rheological properties of fresh con-crete. Excessively high s0 values varying from 13 to 80 kPa were,however, reported depending on w/c and coarse aggregate concen-tration. The authors attributed the high s0 values to the quasi-sta-tic nature of direct shear along with the high aggregate frictionencountered during shearing at low rates [33].

Lu and Wang [34,35] considered the direct shear to measure‘‘true’’ s0 and validate a constitutive model developed for predict-ing the yield behavior of cementitious materials. The model is prin-cipally based on the concept of excess cement pastes and shearstresses developed between two cement particles that arise fromthe combined effect of Van der Waal’s and electrostatic forces.The s0 determined at zero normal stress was found to decreasefrom 1673 to 474, 202, and 142 Pa for cement pastes having w/cof 0.3, 0.35, 0.4, and 0.48, respectively [34]. The corresponding /values decreased from 6�, to 5�, 2.2�, and 2�, respectively. Noattempts were made, however, to test cement pastes and mortarswith higher w/c or to quantify the effect of friction between shearboxes on test results.

This paper seeks to evaluate the suitability of direct shear testfor assessing s0 of cement pastes possessing cohesive to flowablenature. Special care was placed to reduce friction emanating fromthe shear boxes and allow measurements of C values in the order offew Pa. Also, various emulsions displaying no time-dependentbehavior were tested, and C responses are validated using s0 deter-mined by the vane method. The paper also emphasis relevant phe-nomena related to the friction angles that can be deduced fromdirect shear testing as well as effect of vane positioning on s0 mea-surement and its computation using either Eq. (2) or Eq. (3). Datapresented in this paper can be of interest to researchers in variousindustries to facilitate inter-laboratory comparison and unifyquantification of ‘‘true’’ s0 using standardized testing protocols.

3. Experimental program

3.1. Materials and mix proportions

Portland cement conforming to ASTM C150 Type I was used inthis project. The Blaine surface area [36] and specific gravity of the

cement were 4050 cm2/g and 3.14, respectively. The cement hadC3S, C3A, and Na2Oeq. values of 63.5%, 6.1%, and 0.71%, respectively.A naphthalene sulphonate based high-range water reducer(HRWR), complying with ASTM C494 Type F, having a specificgravity of 1.18 and solid content of 35% was used. A liquid cellu-losic-type viscosity-modifying admixture (VMA) was alsoemployed. Its specific gravity and solid content were 1.11 and25%, respectively. A sodium gluconate based set-retarder wasincorporated in all pastes at a rate of 0.2% of cement mass, to avoidhydration effects on test results.

Ten cement pastes exhibiting different flowability levels weretested. The flow was evaluated by determining the material’s aver-age diameter after spreading on a horizontal surface. An ASTMC1437 [23] mini-slump cone having top diameter, bottom diame-ter, and height equal to 70, 100, and 50 mm, respectively, was used.As summarized in Table 1, the flowability varied from highly cohe-sive (i.e., flow of 100 mm) to highly flowable (i.e., flow of 250 mm).The required flowability was secured by selecting the appropriatew/c and, if necessary, adjusting the HRWR. Depending on w/cand targeted flowability, the VMA was added at different rates tominimize bleeding and sedimentation. Such phenomena maydirectly affect stability of cement pastes with direct influence onC and s0 measurements [37]. All cement pastes were batched witha laboratory mixer using water cooled to constant temperature of20 ± 3 �C. Water was first introduced in the mixer followed gradu-ally by the cement and HRWR over 1 min. After a rest period of30 s, the VMA and set-retarder were added and mixing resumedfor 2 additional minutes. The bulk fresh unit weight of cementpastes was determined using a calibrated container having a vol-ume of 0.5–l (Table 1).

Five grades of poly-vinyl acetate (PVA) emulsions commonlyused as adhesives for wood, furniture, paint, and paper industrieswere tested [38]. These smooth white colored emulsions are pro-duced by polymerisation of vinyl acetate and subsequent alcohol-ysis of the polyvinyl acetate that is formed. Such materials displayexceptional stability with no time-dependent behavior, thus mak-ing them ideal systems for rheological characterization and valida-tion purposes [18]. The PVAs were selected to exhibit differentyield stresses; their physical properties including flowability deter-mined using the mini-slump cone are listed in Table 2. The solidcontent and maximum particle size were determined using mois-ture analyzer and laser particle-size instruments, respectively.The viscosity was measured using Anton Paar coaxial cylinder rhe-ometer (RheolabQC) using a sandblasted internal rotating cylinderto minimize slippage. The testing procedure consisted of pre-shearing the PVA at 100 s�1 over 1 min to break down the suspen-sion’s structure and ensure a reference state. Subsequently, thematerial is subjected to a hysteresis loop varying from 0 to100 s�1 and back to 0 s�1. The elapsed time for each ascendingand descending part is 1 min. The download curve of the datarecorded using a computer was used to determine the viscosity.

3.2. Determination of C using the direct shear test

An ELE Direct Shear apparatus complying with ASTM D3080[27] requirements was used in this study. It comprises a metalshearing box divided into two halves horizontally, and measuring100 mm diameter and 58 mm height. The lower section of thebox can move forward at different constant velocities varying from0.001 to 9 mm/min, while the upper section remains stationary(Fig. 2). The apparatus is constructed to ensure minimum framedistortion during testing. The normal load applied on the uppersection of the shearing box is provided by a dead weight connectedto a horizontal beam, which in its turn, is connected to a loadingyoke that applies the normal load on the loading platen of speci-men container [24]. The horizontal beam produces a loading ratio

Table 1Cement paste composition for different flowability levels.

w/c HRWR, % of cement VMA, % of cement Flow, mm Fresh unit weight, kg/m3

Paste #1 0.29 0 0 100 2060Paste #2 0.31 0.1 0 110 2005Paste #3 0.36 0 0 135 1960Paste #4 0.4 0 0 160 1910Paste #5 0.43 0 0 170 1870Paste #6 0.45 0 0 195 1830Paste #7 0.48 0 0.15 200 1815Paste #8 0.5 0.1 0.25 225 1810Paste #9 0.52 0.15 0.4 235 1785Paste #10 0.55 0.2 0.6 250 1760

Notes: All pastes contained a fixed amount of set-retarder, i.e. 0.2% of cement.The flow and unit weight are given within ±10 mm and ±25 kg/m3, respectively.

Table 2Physical properties of tested PVAs.

Flow, mm Solid content, % Viscosity, Pa s Maximum particle size, lm Specific gravity pH

PVA #1 110 55 1.65 2.5 1.09 4.5PVA #2 140 53.5 1.13 2.4 1.08 4.5PVA #3 155 40 0.84 2.75 1.04 4.3PVA #4 170 34 0.61 2.4 1.05 4.7PVA #5 200 28 0.37 2.35 1.03 4.6

J.J. Assaad et al. / Journal of Non-Newtonian Fluid Mechanics 214 (2014) 18–27 21

of 10:1 between the weight hanger and loading platen. The systemis fully automated through high precision load cell and LVDT trans-ducers to provide real-time control of all loadings and strains tak-ing place during shearing. The maximum load cell capacity is 50 Nhaving a resolution of 0.01 N. The resulting shear and normal stres-ses were calculated by dividing the horizontal and vertical loads bythe specimen’s cross-sectional area, i.e. 7850 mm2.

Several experiments were performed to determine the ‘‘net’’forces required to overcome friction between the upper and bot-tom plates during shearing. A low-viscosity lubricant oil wasapplied between the plates, and tests were run without any mate-rial being filled in the shear boxes. Shearing forces were relativelyhigh and fluctuating within 1.4 ± 0.8 N (i.e., shear stresses of178.3 ± 101.9 Pa), which may lead to erroneous responses espe-cially when testing materials displaying low yield stresses in theorder of a few Pa. The problem of friction between plates is wellknown when large-size direct shear boxes are used in geotechnicalinvestigations, and improvements have been suggested with theuse of Teflon rods, rollers, or needle linear bearings [39,40].

In order to minimize friction and adapt the shearing box to test-ing flowable materials with relatively low C values, four perfectly

Fig. 2. Set-up for the direct shear test.

aligned 10-mm long channels were laser-grooved in the bottompart of the shear box (Fig. 3a). A steel ball having 2.5-mm diameterwas then placed in each channel, thus allowing the lower plate ofthe shear box to behave like a roller with respect to the upperplate. Additionally, the upper plate of the shear box was veryslightly laser-grooved (Fig. 3b), in order to avoid eventual twistingduring movement. The resulting gap between the upper and lowerplates was 10 ± 1 lm. Using this set-up, the ‘‘net’’ shearing forcerequired to overcome friction was reduced to 180 mN (i.e.,22.9 Pa), and most importantly, its fluctuation reduced to ±40 mN(i.e., 5.1 Pa). The value of 180 mN was systematically removed toobtain the ‘‘net’’ shearing stresses when calculating C. During test-ing, the 10-lm clear opening between the upper and bottom plateswas greased, so as to avoid leakage of cement pastes or PVAsthrough the gap.

The C and / parameters are considered being determined underdrained conditions, given the non-waterproof design of the shearbox container and poor control of expelled interstitial liquid[24,19,41]. Two bronze porous stones were used to encase radiallythe specimen (Fig. 2). Special fiberglass filters capable of retainingall particles greater than 1 lm were provided between the cementpaste or PVA and contact faces of the porous disks. It is to be notedthat the upper porous stone was not placed when tests are realizedwithout normal load; rather all materials that protruded above thetop surface of the upper shear box plate were cut off to a planeusing a straight edge trowel.

Given that the shear box is divided into two horizontal halves,the specimen was placed in three approximately equal layers, thusreducing the risks of void formation within the interfacial regionwhere failure is expected to occur. The compaction of each layerwas realized by tamping the material depending on its flowability.Hence, 10 gentle tamps were performed for PVA #1, PVA #2, andcement pastes possessing flowability lower than around 175 mm.The shear box was slightly tamped on its exterior surfaces for otherPVAs and cement pastes having more than 175 mm flow. Thecement paste specimens were allowed to rest for 30 s in the shear-ing box prior to beginning of testing at given displacement rate. Atend of test, the total horizontal displacement of the bottom shearbox varied from 1 to around 3 mm, depending on flowability oftested materials.

Fig. 3. Photos for (left) direct shear box, and (right) grooving details in the upper and bottom plates.

Fig. 4. Typical torque variations with time for Paste #3 and PVA #2 for differentvane positioning (Z1 = 40 mm or Z1 = 0 mm).

Fig. 5. Typical shear stress variations with horizontal displacement.

22 J.J. Assaad et al. / Journal of Non-Newtonian Fluid Mechanics 214 (2014) 18–27

3.3. Determination of s0 using the vane method

Anton Paar (RheolabQC) rheometer was used to evaluate s0 ofcement pastes and PVAs. The rheometer’s torque maximum capac-ity and resolution are 75 and 0.01 mN m, respectively; while itsrotational speed can vary from 0.01 to 1200 rpm. The vane usedconsisted of four blades arranged at equal angles around the mainshaft; its height (H) and diameter (D) were 24 and 12 mm, respec-tively. A cylindrical recipient having 120 mm height and 100 mmdiameter was used for s0 measurements of cement paste andPVA materials.

Two types of s0 measurements were realized, depending on theposition of the four-bladed vane with respect to tested material. Inthe first measurement, the vane was inserted and centered in therecipient, thus making Z1 = 40 mm (see Fig. 1). In case of cohesivematerials possessing flow less than around 175 mm, special carewas placed to distribute and level uniformly the specimen abovethe vane inserted inside the recipient. In the second measurement,the top edges of the vane were aligned with the upper material’ssurface, thus making Z1 = 0 mm. The cement pastes or PVAs werestirred manually between both measurements in order to mitigatethe formation of preferential shear planes due to particle orienta-tion. The cement pastes were allowed to rest for 30 s prior to eachtest, given their thixotropic nature that may affect s0 responses[37]. The total time duration for s0 measurements in both positionsdid not exceed 10 min after the initial mixing of cement withwater.

Shearing was applied at low rotational speed of 0.1 rpm, a valuetypically recommended in literature [3,4,6,18] and also found ade-quate by experimentation for direct s0 measurements. The changesin rheometer’s torque for various cement pastes and PVAs arerecorded as a function of time. As can be seen in Fig. 4, all profilesexhibited an elastic linear region whereby the material resistsshearing, until reaching a maximum torque (Tm) indicating break-age of majority of bonds and yielding of structure. The magnitudeof Tm decreased when the vane is positioned on the upper mate-rial’s surface (Z1 = 0 mm), due to reduced stress contributionresulting from the specimen self-weight. The s0 was calculatedusing Tm following Eq. (2) or Eq. (3), depending on the vane’s posi-tion. Detailed discussion on s0 determination using the vanemethod can be seen in other references [such as 1–3,15,17], andis beyond the scope of this article.

4. Test results and discussion

4.1. Shear stress profiles determined from direct shear test

Typical shear stress vs. horizontal displacement profiles deter-mined at 0.5 mm/min without normal load for different cementpastes and PVA #2 are illustrated in Fig. 5. For both types of

materials, the profiles resemble to a large extent those obtainedwhen determining s0 using the vane method. Initially, the shearstress varies almost linearly until reaching a maximum peak valueindicating failure of bonds. Further horizontal displacement causesthe stresses to decrease towards a steady state region. As men-tioned earlier, the calculation of C was realized using the net shearstress obtained by subtracting the peak value from 180 mN (due tofriction), and dividing by 7850 mm2.

Two photos taken right after completion of tests for Paste #1and Paste #2 having flow of 100 and 110 mm, respectively, aregiven in Fig. 6. As can be seen, the yielding plane is clearly demar-cated horizontally within the suspension, thus reflecting the ‘‘true’’nature of the stress that is being measured. It is to be noted that itwas difficult to take photos for other tested materials, given that

Fig. 7. Effect of displacement rate on C values for tested cement pastes.

Fig. 8. Effect of displacement rate on C values for tested PVAs.

J.J. Assaad et al. / Journal of Non-Newtonian Fluid Mechanics 214 (2014) 18–27 23

the specimen flows along the sides right after removal of the shear-ing box.

4.2. Determination of appropriate displacement rate

4.2.1. Data obtained from cement pastesThe logarithmic variations of C vs. displacement rate for differ-

ent cement pastes tested without normal load are plotted in Fig. 7.Regardless of consistency, the data obtained show a minimum in Cresponses at displacement rates varying in the range from 0.3 to0.7 mm/min. For example, when the displacement rate decreasesdown to 0.05 mm/min, the C value increases up to 38.5 and397 Pa for Paste #8 and #3, respectively. The material’s capabilityto withstand higher shear stresses at low displacement rates canbe related to the fact that the cement paste has enough time torebuild its structure and re-orient its particles within the shearingplane. Hence, for example, a considerable elapsed time of 20–40 min is required to displace the shear box by 1–2 mm, respec-tively, at a rate of 0.05 mm/min, whereby cement hydration cou-pled with significant loss in consistency may occur. Conversely,speeds higher than around 2 mm/min would rip apart the struc-tural network bonds and lead to an increase in C values. Suchtrends are in complete agreement with those determined usingthe vane method by other researchers [6,15]. Thus, at the mini-mum of C vs. displacement rate curves (Fig. 7), the stress appliedby the direct shear box is just great enough to overcome the restor-ing forces due to reorientation of particles, thixotropic recovery ofbroken bonds, and structural development due to cement hydra-tion reactions.

4.2.2. Data obtained from PVAsThe C vs. displacement rate curves obtained for various PVAs

are illustrated in Fig. 8. Unlike cement pastes whose propertiesare time-dependent, the C values of PVAs remained almost con-stant for displacement rates less than around 1 mm/min and vary-ing within the accuracy of testing (see next paragraph). However,higher speeds led to gradual increases in C as a result of the mate-rial’s viscous resistance to shearing.

Based on Figs. 7 and 8, displacement rates varying from 0.3 to0.7 mm/min shall be considered appropriate for determining C ofcement pastes and PVAs. The entire duration that extends fromthe filling and compacting the material in the shear box till per-forming the test would not exceed around 10 min.

4.2.3. Determination of PVA viscosity by direct shearIt is interesting to note that the PVA viscosity values determined

by coaxial cylinders could be reproduced, to a certain extent, by thedirect shear test. As can be seen in Fig. 9, linear regressions follow-ing the Bingham behavior are obtained between the various C val-ues and corresponding shear rates (c). The c was determined by

Fig. 6. Photos of Paste #1 (left) and Paste #3 (right) taken after t

assuming that flow takes place between two parallel planes, onemoving at constant speed and the other remaining stationary.Hence, c (in s�1) becomes equal to the displacement rate (inmm/s) divided by the distance between parallel planes (assumedto be 0.01 mm). Nevertheless, the over-estimation of viscositydetermined by direct shearing could mostly be related to the lowrange of shear rates applied that varied up to 14 s�1 (Fig. 9). Thus,the use of higher shear rates as well as adequate control of the gapopening between the upper and bottom plates would be needed tomake the direct shear test a reliable viscosity measurement tool.

4.3. Evaluation of repeatability of C and s0 responses

In order to evaluate repeatability of testing, the C and s0 mea-surements were realized 3–5 times (a new cement paste was bat-ched for each test). The coefficient of variation (COV) is calculatedas the ratio between standard deviation of responses and their

he materials have been taken out from the direct shear box.

PVA #2:y = 1.49x + 112.48R² = 0.89

PVA #1:y = 3.76x + 192.92R² = 0.95

PVA #4:y = 0.78x + 22.82R² = 0.88

0

50

100

150

200

250

0 2 4 6 8 10 12 14 16

Shearstress,Pa

Shear rate, sec-1

Viscosity, Pa.sec Coaxial cylinder Direct shear PVA #1 1.65 3.76 PVA #2 1.13 1.49 PVA #4 0.61 0.78

Fig. 9. Relationships between shear stress vs. shear rate determined on PVAs usingthe direct shear test.

24 J.J. Assaad et al. / Journal of Non-Newtonian Fluid Mechanics 214 (2014) 18–27

mean values, multiplied by 100. Table 3 summarizes all C and s0

along with their COV values (the / values are also reported inTable 3, but will be discussed in Section 4). The relationshipbetween C and flow is given as: C, Pa = 6277.1� exp[�0.026 � (flow, mm)]

with coefficient of correlation (R2) equal to 0.73.

4.3.1. Data obtained from direct shear testIn general, materials having relatively cohesive nature exhibited

adequate repeatability of C responses. Hence, the COV remainedless than around 6% for cement pastes having flow less than170 mm and for PVA #1 and #2. The COV increased up to 9% and14.3% for cement pastes having 225 and 250 mm flow, respec-tively. The PVA #4 and #5 resulted in COVs of 11.5% and 15.8%,respectively. The adequate repeatability resulting from cohesivematerials could be related to a more uniform distribution of shearstresses along the failure plane, as can typically be seen in Fig. 6 forPaste #1 and #2. Conversely, the C responses of relatively flowablematerials may become quite sensitive to the distribution of stres-ses as well as various steps implemented during testing includingthe eventual leaking of materials between the upper and lowerplates.

4.3.2. Data obtained from vane testUnlike repeatability of C responses, materials possessing flow-

able nature resulted in the lowest COV for s0 that remained less

Table 3Direct shear and vane results – repeatability evaluation.

Flow, mm Direct shear test

C, Pa COV,a % /, degree

Paste #1 100 559.5 5.1 7.4Paste #2 110 467.2 4.3 7.2Paste #3 135 326.8 4.4 6.4Paste #4 160 197 6.2 6.5Paste #5 170 168.3 5.8 4.9Paste #6 195 91.2 8.3 4.3Paste #7 210 43.5 7.7 4Paste #8 225 26.6 9 3.1Paste #9 235 10.1 11.8 2.5Paste #10 250 9.3 14.3 2.4PVA #1 110 195.5 4.6 5.5PVA #2 140 111.4 6.2 3.8PVA #3 155 41.7 8 3.3PVA #4 170 24.2 11.5 2PVA #5 200 6.7 15.8 1.4

The reported C and s0 values are averages of 3–5 measurements.n/a refers to ‘‘not available’’.

a COV of direct shear is determined on materials tested without normal load at 0.5 m

than around 6% for cement pastes having flow higher than200 mm and PVA #4 and #5. The COV tended to increase graduallyfor cohesive materials; the maximum COV of 13.3% correspondedto Paste #1 having a flow of 100 mm. In fact, homogeneity of cohe-sive materials adjacent to the vane’s blades was remarkably dis-turbed when the vane was inserted inside the recipient prior toshearing. This was visually noticed and believed to create someair pockets around and above the vane that can alter shear stressesrequired to breakdown the structure. The improvement in repeat-ability of s0 when the vane is positioned at Z1 = 0 confirms the ear-lier statement, as this action reduces disturbance of material priorto shearing. Hence, for example, the COV decreased from 13.3% atZ1 = 40 mm to 10.4% when the vane was positioned at the uppermaterial’s surface (Z1 = 0 mm).

4.4. Determination of angle of internal friction, /

A minimum of three tests realized at different normal loads areperformed to obtain the / values of tested materials (a new batchof cement paste was prepared and used for each test). Specialemphasis was placed to avoid rotation of cover plate that can influ-ence the distribution of normal load on top of specimen. In fact, thecontact point of this plate acts like a ball-and-socket around whichit can freely rotates, particularly when testing flowable materials.This problem is well known in soil mechanics and improvementshave been proposed in literature [42]. In the case of this study,the loading yoke assembly was detached from the system to allowbetter control of the squeezing pressure. Light dead weights (i.e., inthe order of 200 and 400 g) were used to apply concentrically thenormal stress on top of specimen. The use of light weights to createthe normal loads was found essential to maintain good homogene-ity and reduce bleeding of cement pastes, particularly those pos-sessing flow larger than around 200 mm.

Typical shear stress vs. horizontal displacement profiles deter-mined at 0.5 mm/min at different normal loads for various materi-als are illustrated in Fig. 10. For all loadings, the shear stresses varyalmost linearly until reaching a maximum peak value indicatingfailure of bonds. The increase in peak values with increasing nor-mal load is caused by higher friction and interlocking of solid par-ticles during shearing. An approximately linear relationshipfollowing the Mohr–Coulomb theory exists between the maximumshear resistance and normal stress, as shown in Fig. 11. The C and /

Vane method

Vane position: Z1 = 40 mm Vane position: Z1 = 0 mm

s0 from Eq. (2), Pa COV, % s0 from Eq. (3), Pa COV, %

762 13.3 657.2 10.4611.3 n/a 522 n/a466.2 10.7 388.3 n/a242.7 8 211 6.3183.5 n/a 149.6 n/a133.2 8.1 88.6 755 n/a 47.2 n/a31.9 5.3 23.5 n/a18 3.8 13.9 5.410.2 5.7 8.6 5.9341.7 9.7 264 7.5172 6 122.1 n/a62.6 n/a 43 n/a40.2 6.3 22.6 611.4 5.5 7 5.8

m/min displacement rate.

J.J. Assaad et al. / Journal of Non-Newtonian Fluid Mechanics 214 (2014) 18–27 25

values are taken as the intercept on shear stress axis and slope oflinear fitting line, respectively.

The relationship between / and flow for all tested materials isplotted in Fig. 12 (the correlation with C is also plotted). As canbe seen, the amplitude of / increases with decreased flowability;the maximum value of 7.4� corresponded to the flow of 100 mm.As concluded by other researchers [32,34], this indicates that themaximum shear stress is not a material constant, but rather variesdepending on the normal stress applied. It is to be noted that therange of / values that varied from 2� to 7.4� for cement pastes(Table 3) is in complete agreement with that found by Lu andWang [34]. This reflects the standardized nature of the direct sheartest that could be used to unify characterization of cementitiousmaterials.

Fig. 11. Mohr–Coulomb failure curves for fresh cement pastes.

Fig. 12. Relationships between / with respect to flow and C values for all tested

4.5. Analysis and interpretation of absolute C and s0 values

Given that repeatability of C and s0 responses is differentlyaffected by flowability, the cement pastes and PVAs were groupedfor analysis purposes into three categories including cohesive,flowable, and highly flowable (Table 4). For each category, the lin-ear regression model linking s0 with C values along with resultingR2 is given. The developed model was forced to intercept the originof axis, thus taking the form s0 = A C, where A is the coefficient ofproportionality. Also, the root-mean square error (RMSE) that isan indicator of the residual size or spread around the regressionmodel is given in Table 4. The normalized RMSE (NRMSE), in per-cent, calculated as the ratio between RMSE and range of s0 values(i.e., maximum minus minimum value) can be used as an index toreflect the accuracy of the relationship. A NRMSE value of 0% indi-cates a perfect model, whereas a value of 100% indicates that themodel is inaccurate.

materials.

4.5.1. Effect of vane positioning on s0 measurementsHigh R2 values greater than 0.95 are obtained for all categories;

albeit the A-coefficients are affected by the vane positioning andflow level. For example, when the vane was positioned atZ1 = 40 mm, the s0 determined on materials possessing cohesivenature was considerably higher than C (i.e., the A-coefficient is1.38). The A decreased to 1.21 and 1.27 for materials possessingflowable and highly flowable nature, respectively, at Z1 = 40 mm.Conversely, the effect of positioning the top edges of the bladesvane aligned with the upper material’s surface (Z1 = 0 mm)resulted in a net decrease in A-coefficients (Table 4). For example,such decrease was equal to 0.99 and 1.04 for flowable and highlyflowable materials, respectively.

Fig. 10. Effect of normal stress on variations of shear stresses determined using thedirect shear test.

To better interpret such variations, a series of s0 measurementsdetermined on cement pastes and PVAs were realized while posi-tioning the vane at Z1 = 20 mm and then Z1 = 60 mm (Fig. 13).The s0 computed using Eq. (2) clearly shows that yield stressincreases with increasing Z1. This concludes that positioning thevane at Z1 = 0 mm has the advantage of substantially reducingthe contribution of normal stresses on torque measurements, thusresulting in closer absolute values between s0 and C. In otherwords, this indicates that the common approach that consists oninserting the vane inside the material tends to over-estimate s0,particularly for cohesive cement pastes possessing the highest bulkunit weights (Table 1). The variations of / values with flow plottedin Fig. 12 corroborate this statement; the lower the flow, the higheris the influence of normal stresses on the shear stresses. Suchresults are in complete agreement with Fall et al. [43] whoreported that yield stress can be ascribed to the frictional behaviorof the granular matrix under normal stresses due to gravity, anddepends on the difference between densities of solid particlesand surrounding liquid. Lu and Wang [34] reported that the stres-ses potentially affecting measurements conducted by the vanemethod may be due to the specimen self-weight located abovethe shearing zone or any confining stress created from the vaneboundaries in the rheometer device.

The relationships between s0 and normal stress measured fromtop surface of specimen to mid-point of the 24-mm high vane forPaste #3, #5, #8, and PVA #2 are shown in Fig. 14. The normalstress was calculated as the ratio between the load (i.e., volumeof cylindrical-shaped material located from top surface to mid-height of vane � density � gravity acceleration) divided by the cir-cular surface area defined by the tips of the vane blade. Clearly, s0

Table 4Comparison between C and s0 absolute values.

Materials included in comparison Vane position: Z1 = 40 mm Vane position: Z1 = 0 mm

Cohesive materials: flow of 100–150 mm Paste #1 A-coefficient 1.38 1.17Paste #2 R2 0.97 0.98Paste #3 RMSE, Pa 31.92 23.76PVA #1 Range of s0, Pa 590 535.1PVA #2 NRMSE, % 5.4 4.4

Flowable materials: flow of 150–200 mm Paste #4 A-coefficient 1.21 0.99Paste #5 R2 0.95 0.98Paste #6 RMSE, Pa 15.13 13.46PVA #3 Range of s0, Pa 202.5 188.4PVA #4 NRMSE, % 7.47 7.14

Highly flowable materials: flow of 200–250 mm Paste #7 A-coefficient 1.27 1.04Paste #8 R2 0.97 0.97Paste #9 RMSE, Pa 3.15 3.36Paste #10 Range of s0, Pa 43.6 40.2PVA #5 NRMSE, % 7.23 8.35

Fig. 13. Effect of vane positioning, or Z1, on s0 calculated using Eq. (2).

26 J.J. Assaad et al. / Journal of Non-Newtonian Fluid Mechanics 214 (2014) 18–27

increases linearly with normal stress following the Mohr–Coulombempirical equation; y = ax + b where ‘‘a’’ represents the slope ofrelationship and ‘‘b’’ the yield stress that intersects the y-axis atzero normal stress. It is interesting to note that the slope of rela-tionships roughly coincides with the / values determined fromdirect shear test. For example, the slopes varied from 10.8� to 4�and 0.8� for Paste #3, #5, and #8, respectively; while the corre-sponding / was found equal to 6.4�, 4.9�, and 3.1�, respectively.The wider range of values obtained from the vane test may reflectincreased sensitivity to the effective hydrostatic stress in thematrix of solid particles that depends on the difference betweendensities of particles and surrounding liquid [43]. The correlation

Fig. 14. Relationships between s0 and normal stress calculated from top ofspecimen to mid-height of vane.

between the rate of change of s0 with normal stress with respectto / for the various cement pastes is given in Eq. (5).

ds0

dNormal stress¼ 2:99/� 9:15 R2 ¼ 0:93 ð5Þ

4.5.2. Evaluation of accuracy of relationshipsThe effect of vane positioning (Z1 = 40 mm vs. 0 mm) did not

considerably influence the RMSE values calculated for flowableand highly flowable materials (Table 4), reflecting limited spreadof responses around the regression model. Nevertheless, the RMSEvaried remarkably from 31.92 to 23.76 Pa for cohesive materialstested when the vane was positioned at Z1 = 40 or 0 mm, respec-tively. Physically, this indicates that the vane insertion inside cohe-sive material leads to higher spread of data.

The NRMSE of all tested categories was less than around 8%,reflecting high accuracy of relationships established between s0

and C. However, in general, the NRMSE tended to increase whenevaluating responses of materials exhibiting higher flowability lev-els. For example, the NRMSE increased from 4.4% to 7.14% and8.35% for cohesive, flowable, and highly flowable materials, respec-tively, when Z1 = 0 mm. This can numerically be explained by therelatively increased range of s0 values of cohesive materials,thereby reducing the NRMSE ratio.

5. Summary and conclusions

This paper does not aim at substituting the vane method bydirect shear test, especially knowing that the vane method iswidely used, simple, and versatile. Rather, the paper seeks to eval-uate the suitability of direct shear to determine s0 which couldpossibly makes it a referee test to, for instance, unify quantificationof yield stress or validate constitutive models intended for yieldbehavior of materials. Other potential applications of the directshear may include the evaluation/validation of the vane shapes(such as four, six, or eight blades, slotted or not) and dimensions(H, D, and ratio H/D) on s0 responses. Special emphasis was placedthroughout the paper to eliminate friction between shear boxes,thus allowing the measurement of the full flowability spectrumwith C values ranging from kPa to just a few Pa.

Test results showed that the shear stress vs. horizontal displace-ment profiles required for determining C resemble to a large extentthose obtained when determining s0 using the vane method. Dis-placement rates varying from 0.3 to 0.7 mm/min were found ade-quate during testing. The maximum shear stress, however, is not amaterial constant but rather varies depending on the rate of nor-mal stress applied. Adequate repeatability of C responses was

J.J. Assaad et al. / Journal of Non-Newtonian Fluid Mechanics 214 (2014) 18–27 27

obtained for cohesive materials, but then started to reduce gradu-ally with increased flowability.

The direct shear was efficiently used to validate the effect ofvane positioning on yield stress responses. Hence, over-estimationof s0 occurs when the blades are positioned inside the specimen,particularly for cohesive materials. Conversely, positioning thevane at Z1 = 0 mm eliminates the contribution of material’s selfweight on torque measurements and results in almost identical Cand s0 absolute values. This shows that the rheological tests con-ducted using the vane method to measure s0 without consideringthe effect of normal stress are not accurate.

Acknowledgment

The authors wish to thank the financial support provided by theLebanese Council for Scientific Research (CNRS) – Lebanon.

References

[1] H.A. Barnes, The yield stress—a review—everything flows?, J Non-NewtonianFluid Mech. 81 (1999) 133–178.

[2] C.F. Ferraris, L.E. Brower, Comparison of Concrete Rheometers, InternationalTests at LCPC (Nantes, France), NISTIR 6819, National Institute of Standardsand Technology, 2001. 25 p.

[3] Q.D. Nguyen, D.V. Boger, Yield stress measurement for concentratedsuspensions, J. Rheol. 27 (4) (1983) 321–349.

[4] P.F.G. Banfill, D.C. Saunders, On the viscometric examination of cement pastes,Cem. Concr. Res. 11 (1981) 363–370.

[5] H.A. Barnes, A review of the slip (wall depletion) of polymer solutions,emulsions and particle suspensions in viscometers; its cause, character, andcure, J. Non-Newtonian Fluid Mech. 56 (1995) 221–251.

[6] A.W. Saak, H.M. Jennings, S.P. Shah, The influence of wall slip on yield stressand viscoelastic measurements of cement paste, Cem. Concr. Res. 31 (2001)205–212.

[7] S.V. Kao, L.E. Nielsen, C.T. Hill, Rheology of concentrated suspensions ofspheres I. Effect of the liquid–solid interface, J. Colloid Interface Sci. 53 (1975)358–366.

[8] G.V. Vinogradov, G.B. Froishteter, K.K. Trilisky, E.L. Smorodinsky, The flow ofplastic disperse systems in the presence of the wall effect, Rheol. Acta 14(1975) 765–775.

[9] Q.D. Nguyen, D.V. Boger, Direct yield stress measurement with the vanemethod, J. Rheol. 29 (1985) 335–347.

[10] L. Zhu, N. Sun, K. Papadopoulos, D. De Kee, A slotted plate device for measuringstatic yield stress, J. Rheol. 45 (2001) 1105–1122.

[11] N.A. Park, T.F. Irvine, F. Gui, Yield stress measurements with the falling needleviscometer, in: Proc. of the Xth Int. Congress on Rheology, Sydney, Australia,vol. 2, 1988, pp. 160–162.

[12] P.H.T. Uhlherr, K.H. Park, C. Tiu, J.R.G. Andrews, Yield stress from fluidbehaviour on an inclined plane, in: B. Mena, A. Garzia-Rejon, C. Rangel-Nafaile(Eds.), Proc. IXth Int. Cong. on Rheology, Advances in Rheology, Mexico, vol. 2,1984, pp. 183–190.

[13] P.H.T. Uhlherr, J. Guo, T.N. Fang, C. Tiu, Static measurement of yield stressusing a cylindrical penetrometer, Korea–Aust. Rheol. J. 14 (1) (2002) 17–23.

[14] M.H. Zhang, C.F. Ferraris, H. Zhu, V. Picandet, M.A. Peltz, P. Stutzman, D. De Kee,Measurement of yield stress for concentrated suspensions using a plate device,Mater. Struct. 62 (2010) 43–47.

[15] P.V. Liddel, D.V. Boger, Yield stress measurements with the vane, J. Non-Newtonian Fluid Mech. 63 (2–3) (1996) 235–261.

[16] ASTM D2573, Standard Test Method for Field Vane Shear Test in Cohesive Soil,American Society for Testing and Materials, West Conshohocken, PA, USA,2008.

[17] N.J. Alderman, G.H. Meeten, J.D. Sherwood, Vane rheometry of bentonite gels, J.Non-Newtonian Fluid Mech. 39 (3) (1991) 291–310.

[18] A.S. Yoshimura, R.K. Prud’homme, H.M. Princen, A.D. Kiss, A comparison oftechniques for measuring yield stresses, J. Rheol. 31 (1987) 699–710.

[19] J.J. Assaad, J. Harb, Assessment of thixotropy of fresh mortars by triaxial andunconfined compression testing, Adv. Civ. Eng. Mater. 9 (5) (2012) 1–19.

[20] ASTM D2166-06, Standard Test Method for Unconfined Compressive Strengthof Cohesive Soil, American Society for Testing and Materials, WestConshohocken, PA, USA, 2006.

[21] ASTM D2850-03a, Standard Test Method for Unconsolidated-UndrainedTriaxial Compression Test on Cohesive Soils, American Society for Testingand Materials, West Conshohocken, PA, USA, 2007.

[22] ASTM D4767-11, Standard Test Method for Consolidated Undrained TriaxialCompression Test for Cohesive Soils, American Society for Testing andMaterials, West Conshohocken, PA, USA, 2011.

[23] ASTM C1437-07, Standard Test Method for Flow of Hydraulic Cement Mortar,American Society for Testing and Materials, West Conshohocken, PA, USA,2007.

[24] A. Aysen, Soil Mechanics; Basic Concepts and Engineering Applications, Taylor& Francis, London, 2006. 450 p.

[25] G.A. Miller, T.B. Hamid, Testing and modeling of soil-structure interface, J.Geotech. Geoenviron. Eng. 130 (8) (2004) 851–860.

[26] C.N. Liu, Y.H. Ho, J.W. Huang, Large scale direct shear tests of soil/PET-yarngeogrid interfaces, Geotextiles Geomembr. 27 (2009) 19–30.

[27] ASTM D3080/D3080M-11, Standard Test Method for Direct Shear Test of SoilsUnder Consolidated Drained Conditions, American Society for Testing andMaterials, West Conshohocken, PA, USA, 2011.

[28] F. Micaelli, Extrudable Cement-Based Pastes: Some Rheological Aspects,Thesis, University of Pisa, Italy, 2003, 150 p.

[29] R. Alfani, G.L. Guerrini, Rheological test methods for the characterisation ofextrudable cement-based materials – a review, Mater. Struct. 38 (2005) 239–247.

[30] Z. Toutou, N. Roussel, C. Lanos, The squeezing test: a tool to identify firmcement-based material’s rheological behaviour and evaluate their extrusionability, Cem. Concr. Res. 35 (2005) 1891–1899.

[31] R. L’Hermite, G. Tournon, Vibration of Fresh Concrete, Annales, InstitutTechnique du Batiment et des Travaux Publics, Paris, 1948. 25 p.

[32] R. Hendrickx, K. Van Balen, D. Van Gemert, Yield stress measurement of mortarusing geotechnical techniques, IN: RILEM Int. Symp. on Rheology of CementSuspensions Such as Fresh Concrete, 2009, 10 p.

[33] S. Girish, B.S. Santhosh, Determination of Bingham parameters of freshPortland cement concrete using concrete shear box, Bonfring Int. J. Ind. Eng.Manag. Sci. 2 (4) (2012) 84–90.

[34] G. Lu, K. Wang, Investigation into yield behavior of fresh cement paste: modeland experiment, ACI Mater. J. 107 (1) (2010) 12–19.

[35] G. Lu, K. Wang, Theoretical and experimental study on shear behavior of freshmortar, Cem. Concr. Compos. 33 (2011) 319–327.

[36] ASTM C204, Standard Test Method for Fineness of Hydraulic Cement by Air-Permeability Apparatus, American Society for Testing and Materials, WestConshohocken, PA, USA, 2011.

[37] J.J. Assaad, K.H. Khayat, H. Mesbah, Assessment of thixotropy of highlyflowable and self-consolidating concrete, ACI Mater. J. 100 (2) (2003) 99–107.

[38] C.S. Chern, Principles and Applications of Emulsion Polymerization, Wiley Pb.,2008. 252 p. ISBN: 978-0-470-12431-4.

[39] S.H. Liu, D. Sun, H. Matsuoka, On the interface friction in direct shear test,Comput. Geotech. 32 (2005) 317–325.

[40] G. Barla, F. Robotti, L. Vai, Revisiting large size direct shear testing of rock massfoundations, IN: C. Pina, E. Portela, J. Gomes (EdS.), 6th Int Conf on DamEngineering, Lisbon, 2011, p. 10.

[41] J. Assaad, J. Harb, K. Khayat, Use of triaxial compression test on mortars toevaluate formwork pressure of SCC, ACI Mater. J. 106 (5) (2009) 439–447.

[42] M.L. Lings, M.S. Dietz, An improved direct shear apparatus for sand,Geotechnique 54 (4) (2004) 245–256.

[43] A. Fall, F. Bertrand, G. Ovarlez, D. Bonn, Yield stress and shear banding ingranular suspensions, Phys. Rev. Lett. 103 (178301) (2009) 4.