1059

Proceedings of the XVI ECSMGEGeotechnical Engineering for Infrastructure and DevelopmentISBN 978-0-7277-6067-8

© The authors and ICE Publishing: All rights reserved, 2015doi:10.1680/ecsmge.60678

des pieux sous sollicitations cycliques, pour lesquels

on considère généralement uniquement une dégrada-

tion du frottement sous chargement cyclique. La

phase d’amélioration observée irait alors plutôt dans

le sens d’une stabilisation du comportement du pieu,

après, bien sûr, une phase initiale de dégradation qui

doit quand même être prise en compte.

Il sera bien évidemment nécessaire de confirmer ces

résultats par des essais complémentaires, en particu-

lier in situ si possible afin de confirmer les résultats

obtenus en laboratoire.

7 REMERCIEMENTS

Les auteurs de cette communication tiennent à re-

mercier le projet SOLCYP (composé d’un projet

ANR couplé à un Projet National), l’Agence natio-

nale de la recherche française (ANR) ainsi que

l’IREX (organisme gestionnaire des deux projets)

pour leur support financier et l’environnement scien-

tifique et technique dans le cadre duquel ces travaux

ont pu être réalisés.

8 REFERENCES BIBLIOGRAPHIQUES

Al-Douri, R.H. & Poulos, H.G. 1995. Predicted and observed cy-

clic performance of piles in calcareous sand. J. Geot. Eng. Div.,

ASCE 121(1), 1-16.

Andersen, K.H., Kleven, A. & Heien, D. 1988. Cyclic soil data for

designof gravity structures. J. Geot. Eng. Div., ASCE 114 (5)

517-539.

Boulon, M. & Foray, P. 1986. Physical and numerical simulation

of lateral shaft friction along offshore piles in sand. Proceed-

ings 3rd

Int. Conf. on Numerical Methods in Offshore Piling,

Nantes, France, 127-147.

Chan, S.F. & Hanna, T.H. 1980. Repeated loading on single piles

in sand. J. Geot. Eng. Div., ASCE 106 (2) 171-188.

Chin, J.T. & Poulos, H.G. 1996. Tests on model jacked piles in

calcareous sand. Geot; Testing J. 19(2) 164-180.

Foray, P., Tsuha C.H.C., Silva M., Jardine R.J. & Yang, Z.X.

2010. Stress paths measured around a cyclically loaded pile in

a calibration chamber. Proceedings Int. Conf. on Physical

Modelling in Geotechnics, London, 933-939.

Karlsrud, K., NAdim, F. & Haugen, T. 1987. Piles in clay under

cyclic axial loading – Field tests and computational modelling.

Publication n°169, Norwegian Geotechnical Institute.

Le Kouby, A., Canou, J. & Dupla, J.C. 2004. Behavior of model

piles subjected to cyclic loading. Proceedings Int. Conf. on Cy-

clic Behavior of Soils and Liquefaction Phenomena, Bochum,

Germany, 159-166.

Lehane, B.M., Jardine, R.J., Bond, A.J. & Frank, R. 1993. Mecha-

nisms of shaft friction in sand from instrumented pile tests.

J. Geot.Eng. Div., ASCE 119 (1), 19-35.

Lehane, B.M. & White, D.J. 2005. Lateral stress changes and shaft

friction for model displacement piles in sand. Canadian Geot.

J. 42, 1039-1052.

Schlosser, F. & Guilloux, A. 1980. Le frottement dans le renfor-

cement des sols. Revue française de Géotechnique 16, 65-77.

Tali, B. 2011. Comportement de l’interface sol-structure sous sol-

licitations cycliques. Application au calcul des fondations pro-

fondes. Thèse de doctorat de l’Université Paris-Est / Ecole des

Ponts ParisTech, Champs-sur-Marne, France, 225 p.

Uesugi, M., Kishida, H. & Tsubakihara, Y. 1989. Friction between

sand and steel under repeated loading. Soils and Foundations

29 (3), 127-137.

White, D.J. & Lehane, B.M. 2004. Friction fatigue on displace-

ment piles in sand. Geotechnique 54 (10), 645-658.

Performance-based assessment of rammed aggregate piers

Une evaluation sur le comportement des colonnes ballastees pilonnees

K.O. Cetin*1, E. Kurt Bal2 and L. Oner2 1 Department of Civil Engineering, METU, Ankara, Turkey

2 Sentez Insaat, Istanbul, Turkey * Corresponding Author

ABSTRACT Within the confines of this paper, a normalized displacement-based capacity mobilization scheme is presented for rammed aggregate piers (RAP). For this purpose, field load tests, performed on 63 RAPs, in Turkey, were assessed. Site investigations at these sites revealed that generalized soil profiles are mostly composed of normally consolidated clay layers extending to a depth of 18 m. Below this depth, usually medium dense to dense sand / hard greywacke / very stiff to hard clay layers are present. A weighted mean SPT N60 assess-ment procedure was utilized to estimate representative soil strength and stiffness parameter in cohesive soils. The results of RAP load tests were summarized in the form of normalized mobilized capacity versus settlement curves as a function of representative SPT N60 values. The normalized field load test database revealed that: i) the shaft resistance is observed to be fully mobilized at normalized displacements of 40 % of RAP diameter for very soft clays and at 10 % of RAP diameter for firm clays, ii) up to normalized displacements of 2-5% of RAP diameter, 30-50% of the shaft resistance capacity is mobilized in a rather linear elastic manner, iii) normalized capacity mobilization response of RAPs is more flexible than the ones of bored concrete piles, iv) under compressive loads, RAPs exhibit a strain hardening re-sponse, as a result of which, the design-basis capacity is dominated by allowable settlement criterion. The proposed normalized capacity and displacement response curves, presented herein enable displacement (performance)-based assessment and design of RAPs.

RÉSUMÉ Cet article a pour objet l'étude de la capacité de mobilisation de déplacement d'une colonne ballastée pilonées (RAP). A cet effet il a été effectué en Turquie 63 essais de chargement sur les RAPs. Les études de terrain sur ces sites ont révélé que les profils de sols géné-ralisés sont principalement composés de couches d'argile normalement consolidées s'étendant à une profondeur de 18 m. En dessous de cette profondeur, habituellement un sable moyennement dense allant à un sable dense / disque grauwacke / très rigide des couches d'ar-gile durs sont présents. Un SPT N60 procédure d'évaluation moyenne pondérée a été utilisée pour estimer la résistance du sol représentatif et le paramètre de rigidité dans les sols cohérents. Les résultats de tests de chargement de (RAP) ont été résumées sous forme de courbe de la capacité mobilisée normalisée en fonction de valeurs de tassement SPT N60 représentatives. La base de données de essais de chargement sur le terrain normalisé a révélé que: i) la résistance cylindrique est observée pour être pleinement mobilisé pour des déplacements norma-lisées de 40% du diamètre de RAP pour les argiles très doux et à 10% du diamètre de RAP pour les argiles fermes, ii) jusqu'à déplacements normalisées de 2-5% du diamètre de RAP, 30-50% de la capacité de résistance du cylindre est mobilisée d'une manière élastique plutôt li-néaire, iii) la réponse de la mobilisation des capacités normalisée des RAPs est plus flexible que celles de pieux forés en béton , iv) sous charges de compression, RAPs présentent une réponse de durcissement, à la suite de laquelle, la capacité de dimensionnement est dominée par le critère de tassement admissible. Les courbes normalisées proposées permettent de définir la capacité de déplacement évalué selon la conception des RAPs.

1 INTRODUCTION

Ground improvement engineering solutions in the form of rigid column intrusions are frequently used to eliminate i) static bearing capacity and excessive settlement problems, ii) seismic soil liquefaction-

induced failures and deformations. Stiff rammed ag-gregate pier (RAP) elements serve as an alternative to existing conventional solutions (e.g.: deep founda-tions or over excavation and replacement of com-pressible soils). Within the confines of this manu-script, the deformation performance of 50 cm

Geotechnical Engineering for Infrastructure and Development

1060

diameter RAP elements, constructed by bottom-fed dry method (Geopier-Impact System), is evaluated. The resulting response is summarized as normalized load-settlement curves obtained from full scale load tests. For this purpose, 63 RAP field load tests were used, which were constructed at thirteen different soil sites in Turkey. The field load test results and the proposed capacity mobilization curves will be pre-sented after a brief review of the existing literature.

2 AN OVERVIEW OF LITERATURE

Barksdale and Bachus (1983) defined three distinct failure modes for stone columns subjected to vertical loading: bulging, shearing, and punching failures. A schematic illustration of these modes is presented in Figure 1. For the assessment of the failure load trig-gering uniquely bulging-induced failures Datye and Nagaraju (1975), Hughes and Withers (1974) and Madhav and Vitkar (1978) presented analytical and /or numerical solutions, as presented in Equations 1-4, respectively. On the other hand, Wong (1975), Barksdale and Bachus (1983) solutions are widely used for the assessment of shearing-induced failures. For stone columns contructed in very soft clayey deposits, punching-induced failure mechanism can be assessed by Aboshi et al. (1979).

Figure 1. Failure mechanisms of a single stone column in a homo-geneous soft layer (Barksdale and Bachus, 1983).

qult= γczkpc + 2co kpc 1+sinφs

1-sinφs (1)

qult= Fc

1Co +Fq1 Qo

1+sinφs

1-sinφs (2)

qult= σro + 4Co

1+sinφs

1-sinφs (3)

qult= Co Nc + 12 γc BNγ +γc Df Nq (4)

where; γc ∶ unit weight of soil z : total depth of the limit of bulge of the column kpc : coefficient of passive earth pressure of soil co : cohesion ∅s : angle of internal friction of stone column Fc', Fq' : cavity expansion factors Qo : mean stress within the zone of failure σro : initial radial effective stress B : foundation width Nc, Nq, N: dimensionless bearing capacity factors Df : depth of foundation On the basis of these existing studies, it can be concluded that the ultimate bearing capacity of a stone column is a function of the column diameter, strength and stiffness responses of stone column and native soil materials. Usually, the column length is judged to have a negligible effect on the "long" col-umn ultimate bearing capacity. This conclusion is al-so supported by the results of model tests. It was shown that the load transfer mechanism is simply due to skin friction or adhesion along the shaft (Hughes and Withers, 1974). Test results also indicated that the ultimate capacity of stone column is governed primarily by the maximum radial reaction (confine-ment) of the soil which is limited by bulging failure, and extend of vertical movement in the stone column was limited to about 4 times the column diameter. In the literature there exist a number of alternative ap-proaches to assess the bearing capacity of a single column and group of columns (e.g. Etezad et al., 2006). Effect of column diameter on bearing capacity has also been investigated on the basis of laboratory tests, which were performed on 40, 50 and 70 mm diameter stone columns with constant length to di-ameter ratio of six (Ali et al., 2010). Results of these studies suggested that relatively small diameter stone columns mobilize larger capacities at the same level of deformations. In simpler terms, smaller diameter stone columns mobilize their capacity much faster with increasing vertical displacements. Bae et al., 2002, studied the factors affecting the failure mecha-nism of stone columns with laboratory model tests, and compared their findings with finite element mod-el solutions. They concluded that bulging failure for a single stone column is usually observed at a depth of 1.6 to 2.8 columns diameter.

3 CONSTRUCTION OF COLUMNS

As discussed in previous section, one of the main pa-rameters affecting both capacity and deformation be-havior of stone columns is the construction process. In the field 63 rammed aggregate piers were installed by Geopier-Impact construction procedures. Impact elements are constructed by following steps: (1) a closed ended mandrel with a diameter of 36

cm is pushed into the design depth by applying static driving forces assisted with vertical dy-namic vibration (Figure 2a).

(2) the mandrel and hopper are continuously fed with aggregate (Figure 2b).

(3) the ramming action is applied with 100 cm up / 67 cm down compaction effort, during which vertical vibration is also introduced (Figure 2c). The vertical ramming actions expand the diame-ter from 36 cm to 50 cm, if 100 cm up and 67 cm down compaction procedure is selected.

Figure 2. The construction of Impact RAPs.

Considering the significant influence of construc-

tion method, findings presented herein are only ap-plicable to stone columns installed by following the Geopier-Impact construction procedure. Use of them for columns produces via other techniques will be misleading.

4 SITE INVESTIGATION

As part of the site investigation program, series of conventional boreholes were drilled extending to 23m – 40m depths. At various depths, standard pene-tration tests were performed. As a part of investiga-tion program, both disturbed and undisturbed soil samples were retrieved. Figure 3 presents representa-

tive soil profile documented after the site investiga-tion program performed at one of the select sites (Yalova site). It mostly consist of normally consoli-dated, low to high plasticity, soft to stiff clay (CL-CH), where scattered silty and sandy layers extend-ing to depths of 4m - 18m from existing ground level. Below this layer, medium dense to dense gravelly, clayey, silty sand / hard greywacke / very stiff to hard clay are located. Groundwater table is reported to be at approximately 0.0m – 5.0m depth range.

LLWPL

0

30

0 25 50

S1S2S3S4S5S6S7S8

SPT N60 (blows/30cm)

0

30

0 25 50 75 100

LL, PI & (%)

0

10

20

30

0

Silty CLAY

Silty CLAY

Soil Layer

LLPI

Silty SAND

Surface Soil

Figure 3. A representative soil profile at Yalova site

5 FIELD LOAD TEST

This test is widely referred to as "quick" tests due to relatively rapid application of the loading scheme. The test procedure is very similar to pile load tests defined by ASTM D 1143. As part of the test, test load is directly applied on the pier, as opposed to al-ternative distributed application of the load on both the site soil and pier which is widely referred to cell loading. Field load tests were performed by closely following the loading scheme summarized in Table 1. Staged loading starting with 5% of the service load has been continued until the pier is tested under 150% of its service load. Then an unloading proce-dure was followed.

The modulus load tests of Geopier-Impact ele-ments often incorporate tell-tales at different eleva-tions within the pier. The tell-tale consists of a hori-zontal steel plate that is attached to two sleeved vertical bars extending to the top of the pier. During the load test, displacements at top of the pier and at

1061

diameter RAP elements, constructed by bottom-fed dry method (Geopier-Impact System), is evaluated. The resulting response is summarized as normalized load-settlement curves obtained from full scale load tests. For this purpose, 63 RAP field load tests were used, which were constructed at thirteen different soil sites in Turkey. The field load test results and the proposed capacity mobilization curves will be pre-sented after a brief review of the existing literature.

2 AN OVERVIEW OF LITERATURE

Barksdale and Bachus (1983) defined three distinct failure modes for stone columns subjected to vertical loading: bulging, shearing, and punching failures. A schematic illustration of these modes is presented in Figure 1. For the assessment of the failure load trig-gering uniquely bulging-induced failures Datye and Nagaraju (1975), Hughes and Withers (1974) and Madhav and Vitkar (1978) presented analytical and /or numerical solutions, as presented in Equations 1-4, respectively. On the other hand, Wong (1975), Barksdale and Bachus (1983) solutions are widely used for the assessment of shearing-induced failures. For stone columns contructed in very soft clayey deposits, punching-induced failure mechanism can be assessed by Aboshi et al. (1979).

Figure 1. Failure mechanisms of a single stone column in a homo-geneous soft layer (Barksdale and Bachus, 1983).

qult= γczkpc + 2co kpc 1+sinφs

1-sinφs (1)

qult= Fc

1Co +Fq1 Qo

1+sinφs

1-sinφs (2)

qult= σro + 4Co

1+sinφs

1-sinφs (3)

qult= Co Nc + 12 γc BNγ +γc Df Nq (4)

where; γc ∶ unit weight of soil z : total depth of the limit of bulge of the column kpc : coefficient of passive earth pressure of soil co : cohesion ∅s : angle of internal friction of stone column Fc', Fq' : cavity expansion factors Qo : mean stress within the zone of failure σro : initial radial effective stress B : foundation width Nc, Nq, N: dimensionless bearing capacity factors Df : depth of foundation On the basis of these existing studies, it can be concluded that the ultimate bearing capacity of a stone column is a function of the column diameter, strength and stiffness responses of stone column and native soil materials. Usually, the column length is judged to have a negligible effect on the "long" col-umn ultimate bearing capacity. This conclusion is al-so supported by the results of model tests. It was shown that the load transfer mechanism is simply due to skin friction or adhesion along the shaft (Hughes and Withers, 1974). Test results also indicated that the ultimate capacity of stone column is governed primarily by the maximum radial reaction (confine-ment) of the soil which is limited by bulging failure, and extend of vertical movement in the stone column was limited to about 4 times the column diameter. In the literature there exist a number of alternative ap-proaches to assess the bearing capacity of a single column and group of columns (e.g. Etezad et al., 2006). Effect of column diameter on bearing capacity has also been investigated on the basis of laboratory tests, which were performed on 40, 50 and 70 mm diameter stone columns with constant length to di-ameter ratio of six (Ali et al., 2010). Results of these studies suggested that relatively small diameter stone columns mobilize larger capacities at the same level of deformations. In simpler terms, smaller diameter stone columns mobilize their capacity much faster with increasing vertical displacements. Bae et al., 2002, studied the factors affecting the failure mecha-nism of stone columns with laboratory model tests, and compared their findings with finite element mod-el solutions. They concluded that bulging failure for a single stone column is usually observed at a depth of 1.6 to 2.8 columns diameter.

3 CONSTRUCTION OF COLUMNS

As discussed in previous section, one of the main pa-rameters affecting both capacity and deformation be-havior of stone columns is the construction process. In the field 63 rammed aggregate piers were installed by Geopier-Impact construction procedures. Impact elements are constructed by following steps: (1) a closed ended mandrel with a diameter of 36

cm is pushed into the design depth by applying static driving forces assisted with vertical dy-namic vibration (Figure 2a).

(2) the mandrel and hopper are continuously fed with aggregate (Figure 2b).

(3) the ramming action is applied with 100 cm up / 67 cm down compaction effort, during which vertical vibration is also introduced (Figure 2c). The vertical ramming actions expand the diame-ter from 36 cm to 50 cm, if 100 cm up and 67 cm down compaction procedure is selected.

Figure 2. The construction of Impact RAPs.

Considering the significant influence of construc-

tion method, findings presented herein are only ap-plicable to stone columns installed by following the Geopier-Impact construction procedure. Use of them for columns produces via other techniques will be misleading.

4 SITE INVESTIGATION

As part of the site investigation program, series of conventional boreholes were drilled extending to 23m – 40m depths. At various depths, standard pene-tration tests were performed. As a part of investiga-tion program, both disturbed and undisturbed soil samples were retrieved. Figure 3 presents representa-

tive soil profile documented after the site investiga-tion program performed at one of the select sites (Yalova site). It mostly consist of normally consoli-dated, low to high plasticity, soft to stiff clay (CL-CH), where scattered silty and sandy layers extend-ing to depths of 4m - 18m from existing ground level. Below this layer, medium dense to dense gravelly, clayey, silty sand / hard greywacke / very stiff to hard clay are located. Groundwater table is reported to be at approximately 0.0m – 5.0m depth range.

LLWPL

0

30

0 25 50

S1S2S3S4S5S6S7S8

SPT N60 (blows/30cm)

0

30

0 25 50 75 100

LL, PI & (%)

0

10

20

30

0

Silty CLAY

Silty CLAY

Soil Layer

LLPI

Silty SAND

Surface Soil

Figure 3. A representative soil profile at Yalova site

5 FIELD LOAD TEST

This test is widely referred to as "quick" tests due to relatively rapid application of the loading scheme. The test procedure is very similar to pile load tests defined by ASTM D 1143. As part of the test, test load is directly applied on the pier, as opposed to al-ternative distributed application of the load on both the site soil and pier which is widely referred to cell loading. Field load tests were performed by closely following the loading scheme summarized in Table 1. Staged loading starting with 5% of the service load has been continued until the pier is tested under 150% of its service load. Then an unloading proce-dure was followed.

The modulus load tests of Geopier-Impact ele-ments often incorporate tell-tales at different eleva-tions within the pier. The tell-tale consists of a hori-zontal steel plate that is attached to two sleeved vertical bars extending to the top of the pier. During the load test, displacements at top of the pier and at

Cetin, Kurt Bal and Oner

Geotechnical Engineering for Infrastructure and Development

1062

the tell-tale plate were recorded which enable relative displacement (straining) of the pier element.

Table 1. Typical test procedure.

No Time (min.) (min./max.)

Load (%)

No Time (min.) (min./max.)

Load (%)

0 15 / 60 5 8 15 / 60 133 1 15 / 60 16 9 15 / 60 150

2 15 / 60 33 10 N/A 100

3 15 / 60 50 11 N/A 66

4 15 / 60 66 12 N/A 33

5 15 / 60 83 13 N/A 0

6 15 / 60 100 14 N/A 100

7 60 / 240 116 * 15 N/A 0

* The load increment that represents approximately 115% of the design maximum stress on the Rammed Aggre-gate Pier shall be held for a minimum of 60 minutes and until the rate of deflection is less than 0.254mm per hour or less, or for a maximum duration of 4 hours.

100

80

60

40

20

0

Settl

emen

t(m

m)

0 5 10 15 20 25

Applied Load (ton)

Tell-TaleTop

Figure 4. A representative load test results from Yalova site.

6 FIELD LOAD TEST RESULTS

As previously discussed, 63 loading tests were per-formed on rammed aggregate piers (RAP) installed in soft to stiff clay soils to assess the bearing capacity and stiffness response of individual piers. The rammed aggregate piers were constructed at thirteen different soil sites, and all piers were constructed to a final diameter of 50 cm with varying lengths of 8.0 m to 17.0 m. Within the scope of this paper, the ulti-mate bearing capacity was assessed using load vs. displacement responses. The hyperbola fitting ap-

proach was used for the load tests, at which ultimate capacity could not be reached during loading. Similar to Reese and O’Neill (1988), the field load test re-sults were presented by normalized responses (i.e.: graphs of load normalized by ultimate bearing capac-ity, versus settlement normalized by pier diameter. Then these normalized responses were grouped as functions of representative N60 values, which repre-sent confining soils strength and stiffness characteris-tics. Table 2. A summary of input parameters of the compiled database.

Project Sites N60,rep. RAP Length

(m) No

Afyon-1 3-5 8/11/16 1-5 Afyon-2 9 14/17 6-9 Aydın 7-8 13/18 10-13 Bursa 11-14 16/17 14-26

Gaziantep-1 12 7/8 27-28 Gaziantep-2 13 9 29-30 İstanbul-1 13 10 31 İstanbul-2 2-3 8/14 32-33 Kayseri 22 17 34-35 Sivas 5-15 7/9/10/12 36-43

Yalova 6-12 12/14/16 44-51 Yozgat-1 8-12 8/10/12/15/17 52-57 Yozgat-2 4-10 9/10/12/15 58-63

The corrected SPT N60 values from hammer energy efficiency were used to calculate the representative N60 values. In the estimation of representative N60 values, a linear weighting scheme, linearly decreas-ing from 1 at the ground surface to 0 at the tip of the pier, was used to overweight the shallower soils shear contribution as compared to deeper ones. This weighting is preferred due to the fact that mobilized pier capacity is mostly due to skin friction between the piers and the soil and it mobilizes first at shal-lower depths first. The representative SPT N60 values (SPT N60,rep.) obtained from weighted arithmetic were summarized in Table 2. A representative load-settlement curve is shown in Figure 4 for illustration purposes.

6.1 Mobilization of the skin friction

Reese and O’Neill (1988) assessed a number of com-pression pile load test data obtained from full-size drilled piers constructed in cohesive and cohesionless

soils. On the basis of test results, they developed normalized load-transfer curves for isolated drilled piles. These curves express the mobilized capacity of piles as a function of normalized settlement. Inspired by this, 63 load test results were similarly evaluated in order to assess capacity mobilization responses of Impact piers constructed in the cohesive soils. The ultimate bearing capacity (Qult) was obtained from load vs. displacement response. For the tests where Qult is not achieved during the test, hyperbola fitting method was used to assess the ultimate capacity. The average values of the displacements at the top of the pier and at the tell-tale plate were calculated in order to estimate the average deformations exerted on the Impact pier elements. Field load tests were grouped on the basis of representative SPT N60 values, which vary in the range of 2 to 22 blows/30 cm.

500

400

300

200

100

0

Settl

emen

t(m

m)

0 5 10 15 20 25 30 35 40 45

Applied Load (ton)

side resistanceside resistance (estimated)average of side and tip resistanceaverage of side and tip resistance (estimated)

Figure 5. Estimation of Qult using load-settlement responses Yalova site.

For illustration purposes, the bearing capacity mo-bilization response is shown in Figure 5 for a cohe-sive soil layer with SPT N60 = 6 blows/30 cm. On the same figure, the maximum capacity estimation by us-ing a hyperbolic curve is also shown. Monitored val-ues are shown by solid lines, while extrapolated re-sponse by using hyperbolic expression is shown by dash lines. The curves of the load normalized by estimated Qult versus settlement normalized by diameter (D=50cm) was plotted for each field load test. The resulting responses obtained from Yalova site, which was represented by an SPTN60 = 6 blows/30 cm was shown in Figure 6. Figure 7a presents the normalized curves for 63 field load tests. The soil sites have a representation SPT N60 value ranging from a mini-mum of 2 to a maximum of 22 blows /30 cm.

0 50 100 150/D (%)

0

0.2

0.4

0.6

0.8

1

1.2

Q/Q

ult

side resistanceside resistance (estimated)average of side and tip resistanceaverage of side and tip resistance (estimated)

Figure 6. The graphs of normalized load-settlement (SPT N60 =6).

To illustrate the variation of normalized responses with representative SPT N60 values, responses corre-sponding to minimum and maximum representative N60 values are shown in Figure 7b with a larger scale.

0 10 20 30 40 50/D (%)

0

0.2

0.4

0.6

0.8

1

1.2

Q/Q

ult

0 10 20 30 40/D (%)

0

0.2

0.4

0.6

0.8

1

1.2

Q/Q

ult

SPT N60=2, side resistanceSPT N60=2, average of side and tip resistanceSPT N60=22, side resistanceSPT N60=22, average of side and tip resistance

Figure 7. The graphs of normalized load-settlement.

For comparison purposes, the normalized pile ca-

pacity curves for skin friction in cohesive soils pro-posed by Reese and O’Neill (1988), is presented to-gether with the estimated normalized responses for Impact pier elements, as shown in Figure 8a. The normalized load-settlement curves assessed using by the average values of the displacements at top the pier and at the tell-tale plate for SPT N60 values of between 2 to 22 are shown in Figure 8b.

Qult

Detail-A

%10 %40

SPTN60 = 22

SPTN60 = 2

(a)

(b)

1063

the tell-tale plate were recorded which enable relative displacement (straining) of the pier element.

Table 1. Typical test procedure.

No Time (min.) (min./max.)

Load (%)

No Time (min.) (min./max.)

Load (%)

0 15 / 60 5 8 15 / 60 133 1 15 / 60 16 9 15 / 60 150

2 15 / 60 33 10 N/A 100

3 15 / 60 50 11 N/A 66

4 15 / 60 66 12 N/A 33

5 15 / 60 83 13 N/A 0

6 15 / 60 100 14 N/A 100

7 60 / 240 116 * 15 N/A 0

* The load increment that represents approximately 115% of the design maximum stress on the Rammed Aggre-gate Pier shall be held for a minimum of 60 minutes and until the rate of deflection is less than 0.254mm per hour or less, or for a maximum duration of 4 hours.

100

80

60

40

20

0

Settl

emen

t(m

m)

0 5 10 15 20 25

Applied Load (ton)

Tell-TaleTop

Figure 4. A representative load test results from Yalova site.

6 FIELD LOAD TEST RESULTS

As previously discussed, 63 loading tests were per-formed on rammed aggregate piers (RAP) installed in soft to stiff clay soils to assess the bearing capacity and stiffness response of individual piers. The rammed aggregate piers were constructed at thirteen different soil sites, and all piers were constructed to a final diameter of 50 cm with varying lengths of 8.0 m to 17.0 m. Within the scope of this paper, the ulti-mate bearing capacity was assessed using load vs. displacement responses. The hyperbola fitting ap-

proach was used for the load tests, at which ultimate capacity could not be reached during loading. Similar to Reese and O’Neill (1988), the field load test re-sults were presented by normalized responses (i.e.: graphs of load normalized by ultimate bearing capac-ity, versus settlement normalized by pier diameter. Then these normalized responses were grouped as functions of representative N60 values, which repre-sent confining soils strength and stiffness characteris-tics. Table 2. A summary of input parameters of the compiled database.

Project Sites N60,rep. RAP Length

(m) No

Afyon-1 3-5 8/11/16 1-5 Afyon-2 9 14/17 6-9 Aydın 7-8 13/18 10-13 Bursa 11-14 16/17 14-26

Gaziantep-1 12 7/8 27-28 Gaziantep-2 13 9 29-30 İstanbul-1 13 10 31 İstanbul-2 2-3 8/14 32-33 Kayseri 22 17 34-35 Sivas 5-15 7/9/10/12 36-43

Yalova 6-12 12/14/16 44-51 Yozgat-1 8-12 8/10/12/15/17 52-57 Yozgat-2 4-10 9/10/12/15 58-63

The corrected SPT N60 values from hammer energy efficiency were used to calculate the representative N60 values. In the estimation of representative N60 values, a linear weighting scheme, linearly decreas-ing from 1 at the ground surface to 0 at the tip of the pier, was used to overweight the shallower soils shear contribution as compared to deeper ones. This weighting is preferred due to the fact that mobilized pier capacity is mostly due to skin friction between the piers and the soil and it mobilizes first at shal-lower depths first. The representative SPT N60 values (SPT N60,rep.) obtained from weighted arithmetic were summarized in Table 2. A representative load-settlement curve is shown in Figure 4 for illustration purposes.

6.1 Mobilization of the skin friction

Reese and O’Neill (1988) assessed a number of com-pression pile load test data obtained from full-size drilled piers constructed in cohesive and cohesionless

soils. On the basis of test results, they developed normalized load-transfer curves for isolated drilled piles. These curves express the mobilized capacity of piles as a function of normalized settlement. Inspired by this, 63 load test results were similarly evaluated in order to assess capacity mobilization responses of Impact piers constructed in the cohesive soils. The ultimate bearing capacity (Qult) was obtained from load vs. displacement response. For the tests where Qult is not achieved during the test, hyperbola fitting method was used to assess the ultimate capacity. The average values of the displacements at the top of the pier and at the tell-tale plate were calculated in order to estimate the average deformations exerted on the Impact pier elements. Field load tests were grouped on the basis of representative SPT N60 values, which vary in the range of 2 to 22 blows/30 cm.

500

400

300

200

100

0

Settl

emen

t(m

m)

0 5 10 15 20 25 30 35 40 45

Applied Load (ton)

side resistanceside resistance (estimated)average of side and tip resistanceaverage of side and tip resistance (estimated)

Figure 5. Estimation of Qult using load-settlement responses Yalova site.

For illustration purposes, the bearing capacity mo-bilization response is shown in Figure 5 for a cohe-sive soil layer with SPT N60 = 6 blows/30 cm. On the same figure, the maximum capacity estimation by us-ing a hyperbolic curve is also shown. Monitored val-ues are shown by solid lines, while extrapolated re-sponse by using hyperbolic expression is shown by dash lines. The curves of the load normalized by estimated Qult versus settlement normalized by diameter (D=50cm) was plotted for each field load test. The resulting responses obtained from Yalova site, which was represented by an SPTN60 = 6 blows/30 cm was shown in Figure 6. Figure 7a presents the normalized curves for 63 field load tests. The soil sites have a representation SPT N60 value ranging from a mini-mum of 2 to a maximum of 22 blows /30 cm.

0 50 100 150/D (%)

0

0.2

0.4

0.6

0.8

1

1.2

Q/Q

ult

side resistanceside resistance (estimated)average of side and tip resistanceaverage of side and tip resistance (estimated)

Figure 6. The graphs of normalized load-settlement (SPT N60 =6).

To illustrate the variation of normalized responses with representative SPT N60 values, responses corre-sponding to minimum and maximum representative N60 values are shown in Figure 7b with a larger scale.

0 10 20 30 40 50/D (%)

0

0.2

0.4

0.6

0.8

1

1.2Q

/Qul

t

0 10 20 30 40/D (%)

0

0.2

0.4

0.6

0.8

1

1.2

Q/Q

ult

SPT N60=2, side resistanceSPT N60=2, average of side and tip resistanceSPT N60=22, side resistanceSPT N60=22, average of side and tip resistance

Figure 7. The graphs of normalized load-settlement.

For comparison purposes, the normalized pile ca-

pacity curves for skin friction in cohesive soils pro-posed by Reese and O’Neill (1988), is presented to-gether with the estimated normalized responses for Impact pier elements, as shown in Figure 8a. The normalized load-settlement curves assessed using by the average values of the displacements at top the pier and at the tell-tale plate for SPT N60 values of between 2 to 22 are shown in Figure 8b.

Qult

Detail-A

%10 %40

SPTN60 = 22

SPTN60 = 2

(a)

(b)

Cetin, Kurt Bal and Oner

Geotechnical Engineering for Infrastructure and Development

1064

On the basis of these normalized field responses, it is concluded that: i) the shaft resistance is observed to be fully mobilized at normalized displacements of 40 % of RAP diameter for very soft clays to 10 % of RAP diameter for firm clays, ii) up to normalized displacements of 2-5% of RAP diameter, 30-50% of the shaft resistance capacity is mobilized in a rather linear elastic manner, iii) normalized capacity mobi-lization response of RAPs is more flexible than the ones of bored piles, iv) under compressive loads, RAPs exhibit a strain hardening response, as a result of which the design-basis capacity is dominated by allowable settlements. The proposed normalized ca-pacity and displacement response curves enable per-formance-based assessment and design of RAPs.

0 25 50 75 100 125 150RAP/DRAP (%)

0

0.2

0.4

0.6

0.8

1

1.2

Q/Q

ult

0 0.5 1 1.5 2 2.5PILE/DPILE (%)

range of results, Reese & O'Neill, 1988trend line, Reese & O'Neill, 1988SPT N60=2, side resistanceSPT N60=2, average of side and tip resistanceSPT N60=22, side resistanceSPT N60=22, average of side and tip resistance

0 10 20 30 40RAP/DRAP (%)

0

0.2

0.4

0.6

0.8

1

1.2

Q/Q

ult

SPT N60=2, average side and tip resistanceSPT N60=22, average of side and tip resistane

Figure 8. a) Comparision of normalized load-settlement curves for RAP elements and Reese & O’Neill (1988) Method, b) normalized load-settlement for SPT N60 = 2 and 22. 7 SUMMARY AND CONCLUSIONS

Within the confines of this paper, a normalized de-formation (performance) based capacity mobilization assessment scheme is presented for rammed aggre-gate piers (RAP). For this purpose, field load tests performed on 63 RAPs in Turkey were assessed. The

loading scheme was chosen to be very similar to pile load tests defined by ASTM D 1143. As part of the test, load is directly applied on the pier. The impact pier elements are loaded to 150% of the maximum top-of-pier stress. Relative deformation response of the pier was monitored through tell-tales installed at different elevations within the pier. The results of RAP load tests were summarized in the form of nor-malized mobilized capacity versus settlement curves as functions of representative SPT N60 values. The normalized field load test database revealed that:

i) the shaft resistance is observed to be fully mobi-lized at normalized displacements of 40 % of RAP diameter for very soft clays to 10% for firm clays,

ii) up to normalized displacements of 2-5% of RAP diameter, 30-50% of the shaft resistance capaci-ty is mobilized in a rather linear elastic manner,

iii) normalized capacity mobilization response of RAPs is more flexible than the ones of bored con-crete piles, iv) under compressive loads, RAPs exhib-it a strain hardening response, as a result of which the design-basis capacity is dominated by allowable set-tlement criterion. The proposed normalized capacity and displacement response curves enable perfor-mance-based assessment and design of RAPs.

REFERENCES

Ali, K., Shahu, J.T. & Sharma, K.G. 2010. Behaviour of Rein-forced Stone Columns in Soft Soils: An Experimental Study, Indi-an Geotechnical Conference. ASTM D1143 – 81. Reapproved 1994. Standard Test Methods for Deep Foundations Under Static Axial Compressive Load. Bae, W.S., Shin, B.W. & An, B.C. 2002. Behaviors of Foundation System Improved with Stone Columns, Kitakyushu, Japan. Barksdale, R.D. & Bachus, R.C. 1983. Design and Construction of Stone Column. Report No.FHWA/RD-83/026, Springfield, VA. Huges, J.M.O. & Withers, N.J. 1974. Reinforcing of Soft Cohesive Soils with Stone Columns. Ground Engineering, 7(3): 42-49. Datye, K.R. & Nagaraju, S.S. 1975. Installation and Testing of Rammed Stone Columns. Proc. of IGS Specialty Session, Ban-galor, India: 101-104. Madhav, M.R. & P.P. Vitkar (1978) Strip footing on weak clay stabilized with a granular trench or pile. Canadian Geotechnical Journal, 15(4), 605 – 609. Aboshi, H., Ichimoto, E., Harada, K., Emoki, M. 1979. The Com-poser- A Method to Improve the Characteristics of Soft Clays by Inclusion of Large Diameter Sand Columns. Proc. of Int. Conf. on Soil Reinforcement, E.N.P.C, 1, Paris: 211–216. Reese, L. C. & O’Neill, M. W. 1988. Drilled Shafts: Construction Procedures and Design Methods, Pub. No. FHWA-HI-88-042, U.S., Washington, D.C., 564–564.

(a)

(b)

Pile response to lateral spreads: analysis of ultimate soil pressures

Réponse d’un pieu à l’écoulement lateral: Analyse des pressions de sol ultimes

Y.K. Chaloulos*1, G.D. Bouckovalas1 and D.K. Karamitros2

1 National Technical University of Athens, Athens, Greece 2University of Bristol, Bristol, UK

* Corresponding Author

ABSTRACT Recent experimental results suggest that available methods for the design of piles under liquefaction-induced lateral spreading may underestimate soil pressures and lead to un-conservative solutions. Due to this evidence, the kinematic interaction between single piles and liquefied ground has been thoroughly examined by means of 3-D numerical analyses and new empirical relations are established which take into account the dilative soil response at the upper part of the pile and the associated increase of pile reaction. Example application of the new relations shows that the aforementioned effect becomes important for liquefiable soils with relatively low permeability (e.g. fine, silty sands).

RÉSUMÉ Des résultats expérimentaux récents suggèrent que les méthodes disponibles pour le dimensionnement des pieux à l’écoulement latéral induit par la liquéfaction pourraient sous-estimer les pressions du sol et conduire à des solutions non-conservatives. C’est pourquoi l’interaction cinématique entre des pieux isolés et un sol liquéfié a été examinée en profondeur par des analyses numériques 3D ; de nouvelles relations empiriques ont par ailleurs été établies prenant en compte la dilatation du sol à la partie supérieure du pieu et l’augmentation induite de la pression du sol. Une illustration de ces nouvelles relations montre que l’effet ci-devant mentionné devient important pour des sols liquéfiables à faible perméabilité (sables fins et limoneux).

1 INTRODUCTION

Pile damage due to soil liquefaction and lateral spreading first drew the attention of the Geotechnical Community in the 1964 Niigata earthquake. Current-ly, pile design against such phenomena is performed with the “Beam on Nonlinear Winkler Foundation” method, alternatively known as the p-y method. Namely, the pile is simulated with beam elements, while soil-pile interaction is modeled through hori-zontal Winkler springs which follow a certain force-displacement law (p-y curve). Under these condi-tions, lateral ground displacements are imposed at the fixed end of the springs. It is realized that the accuracy of the simulation mainly depends on the nonlinear force-displacement relationship adopted for the Winkler springs. Thus, it is no surprise that a large number of studies, mostly

experimental, have been dedicated to the investigation of the parameters that affect the shape of p-y curves, as well as to the development of empirical correlations for their evaluation. The majority of the existing correlations are based on relations for nonliquefied soil [e.g. (API 2002)], while the effects of liquefaction are incorporated either through appropriate reduction factors [e.g. Brandenberg et al. (2005)] or through empirical correlations for the residual strength of the liquefied soil [e.g. Cubrinovski and Ishihara (2007)]. In either case, the pursued reduction is solely related to the relative density (Dr) of the sand. Nevertheless, there are recent experimental data [e.g. Tokimatsu and Suzuki (2009)] which imply the effect of additional parameters, such as the properties of the pile itself (bending stiffness, installation method, head constraints), as well as, the insitu