You are on page 1of 11

2010 Structures Congress 2010 ASCE

110

p-y Based Approach for Buckling Analysis of Axially Loaded Piles under Scoured Conditions Cheng Lin1, Caroline Bennett2, Jie Han3, and Robert L. Parsons4 ABSTRACT Buckling of piles may become a severe issue for bridges when scour increasingly erodes away soils around the pile foundation, exposing unsupported lengths above the ground surface. A p-y based approach for buckling analysis was developed to consider buckling behavior of piles due to scour. The procedure utilized multilinear spring stiffnesses to approximately represent p-y curves at different depths. The approximated spring stiffnesses were incorporated into the structural analysis model, and buckling analyses were conducted. The p-y based approach for buckling analysis was verified by laboratory tests presented in the literature. Buckling of axially loaded piles due to scour was investigated based on the p-y based approach. The results showed that multilinear soil stiffness may be successfully used to represent the p-y curve as a result of considerable agreement of deflections of piles calculated based on the multilinear springs and p-y curves. The calculated buckling load based on p-y approach generally compared well with the measured results. Scour significantly reduced the buckling capacity of piles. Effects of sand density and fixity of pile head were also considered for buckling analysis of piles at different scour depth. INTRODUCTION Scour removes soil in the vicinity of pile foundations of bridge or offshore platforms, which can significantly reduce the capacity of the pile foundation and consequently may threaten the safety of the structure. The possible failure modes of scour-susceptible pile foundations can be lateral and/or vertical. Lateral failure includes kick-out of pile tip and pushover failure, while vertical failure may correspond to plunging and buckling failure (Ramey et.al., 2007). Buckling of fullyembedded pile foundations is usually not considered as an issue unless very soft soil is encountered or liquefiable soil is considered for potential seismic damage. However, buckling may become a severe issue when scour increases the unsupported pile length, resulting in slender piles. A case study reported by Avent and Alawady

Department of Civil, Environmental, & Architectural Engineering (CEAE), The University of Kansas, 2150 Learned Hall, 1530 W. 15th Street, Lawrence, KS 66045
1

Cheng Lin, Graduate Research Assistant, cheng@ku.edu Caroline Bennett, Assistant Professor, crb@ku.edu Jie Han, Associate Professor, jiehan@ku.edu Robert Parsons, Associate Professor, rparsons@ku.edu

2010 Structures Congress 2010 ASCE

111

(2005) showed that severe scour deteriorated the piles of bridges and subjected the substructure of the bridges to potential buckling failures. Several approaches (Davisson and Robinson, 1965; Gabr et al., 1997; Heelis et al., 2004) were developed to analyze the buckling of axially loaded piles. A procedure developed by Davisson and Robinson (1965), which used the governing bending differential equation for buckling deflection to determine the buckling loads of piles, has been adopted in the AASHTO (2007) LRFD Bridge Design Specifications for preliminary design. In this procedure, a partially-embedded pile is treated as a pile having a fixed base of an equivalent depth below the ground, while the boundary condition of pile head is the same as that of the actual pile. The equivalent depth to fixity, Le, is determined for clays and sands respectively using Eqs. (1) and (2). The total length of a pile as a result is the summation of the equivalent length, Le, and unembedded length, Lu. The buckling load of piles is then calculated using the Euler equation as shown in Eq. (3). For clays:
L e = 1.4( E p I w / E s ) 0.25

(1)

For sands:
L e = 1.8(E p I w / n h ) 0.2

(2)

Pcr =

2 E p I w [K(L u + Le )]2

(3)

where: Le = equivalent embedded length of piles; Ep = modulus of elasticity of the pile; Iw = weak axis moment of inertia for the pile; Es= elastic modulus of clay; nh = rate of increase of soil modulus with depth; Pcr = buckling load of piles; K = effective length factor; and Lu = unembedded length of pile. This method is inherently limited due to the assumption of constant or linearly increasing horizontal subgrade modulus with depth. It is more likely that the subgrade modulus at each depth will vary with deflection rather than being constant. Therefore, the potential for buckling of unsupported pile lengths and the determination of stability under lateral loading should be evaluated using p-y curves (AASHTO, 2007). The method is not capable of considering the buckling of fullyembedded piles. Moreover, Davisson and Robinsons method tends to encounter difficulty when dealing with nonhomogeneous soils, i.e. layered soils. The present study uses a p-y based approach for analyzing the buckling of axially loaded piles. A laboratory test conducted by Lee (1968) was used to compare Davisson and Robinsons method and the p-y based method for buckling analysis of axially loaded piles. The p-y based approach for buckling analysis was then used to evaluate the buckling behavior of axially loaded piles in sands under different scour depths. Effects of sand density (loose, medium, and dense) and fixity of pile head on the buckling of piles were also considered in the present study.

2010 Structures Congress 2010 ASCE

112

THE P-Y BASED APPROACH FOR BUCKLING The p-y based approach for analyzing the buckling of piles was conducted in the structural analysis software, STAAD Pro. 2007 in conjunction with a laterally loaded pile foundation analysis conducted in the software, LPILE 5.0. Widely disseminated and generally available p-y curves (Reese and Van Impe, 2001) were used to describe lateral soil resistance versus pile deflection at different depths using a series of nonlinear springs as shown in Figure 1. The p-y curves at various depths were generated based on properties of a pile and soil when the pile under lateral loading was analyzed using the software package LPILE 5.0. Multilinear soil springs were determined using piecewise slopes to approximate the p-y curve, as shown in Figure 2. The approximated multilinear springs representing soil resistance to piles were then provided as input to the structural analysis model in STAAD Pro. Multilinear soil springs were also required to account for negative deflections in the structural analysis in STAAD Pro. 2007. Therefore, the above procedure was to integrate p-y curves into the structural analysis model using approximated mulitilinear springs; accordingly the buckling analysis could be fully accomplished in the structural analysis.

Px Py Mt Py

Px

Mt

p x=x1 p x=x2 y p x=x3 y x y

Figure 1. Illustration of p-y curves used in a pile analysis (Reese and Van Impe, 2001)

2010 Structures Congress 2010 ASCE

113

k3 k2 k1 k0 k7 k6 k5 k4 y

Figure 2. Approximation of a p-y curve

COMPARISON WITH EXPERIMENTAL RESULTS A series of laboratory tests on pile buckling reported by Lee (1968) were used to verify the p-y based approach used herein for buckling analysis. The objective of the experimental tests performed by Lee (1968) was to verify the accuracy of the method developed by Davisson and Robinson (1965). From the experimental testing, Lee (1968) concluded that Davisson and Robinsons method successfully predicted the measured results. Two of four tests in the Lee (1968) study were selected for use within this current study because only two tests provided buckling capacity. An aluminum (Al) tube and a steel rod were used as piles for the two experimental tests. Dimensions and properties of the piles are shown in Table 1. The aluminum pile was embedded 33 in. into sand with relative density of 100%; the steel rod pile was placed 33 in. into dense sand having relative density of 70%. The soil properties, i.e. friction angle, effective unit weight, and coefficient of subgrade reaction, Kpy , were estimated in the present study based on the limited given parameters, such as relative density, maximum ratio, and minimum void ratio in Table 2. Kpy in the p-y method was taken as 225 lb/in3 for dense sand, 90 lb/in3 for medium sand, and 25 lb/in3 for loose sand (Reese and Van Impe, 2001). Sand at a relative density of 70% was treated as medium loose sand by Lee (1968), even though it is commonly classified as dense sand (Kulhawy and Mayne, 1990). This treatment could likely be because the coefficient of subgrade reaction, nh, was determined to be approximately 9.0, which ranged from 24 for a medium sand to 8

2010 Structures Congress 2010 ASCE

114

for a loose sand, as recommended by Terzaghi (1955). The deviation of values of nh and Kpy according to Reese and Van Impe (2001) was because Terzaghi (1955) used the slope of secant to p-y curves to determine nh rather than the initial soil reaction However, AASHTO (2007) modulus which was used to determine Kpy. recommended the selection of coefficient of subgrade reaction to be close to Kpy. Accordingly, the coefficient of subgrade reaction, Kpy, was selected as a value approaching that for a loose sand. In Lee (1968) experimental tests, all piles were pinned at the top and fixed at the bottom.
Table 1. Dimensions and properties of model piles (Lee, 1968) Total length (in.) 64 62.25 Length in sand (in.) 33 33 Outside diameter (in.) 0.316 0.355 Inside diameter (in.) 0.251 -Section area (in2) 0.029 0.099 Moment of inertia (in4) 2.86 10-4 7.810-4 Youngs modulus (ksi) 1.0104 3.0104

Pile type Al Tube Steel Rod

Table 2. Properties of sand Relative density (%) 100 70 Friction angle (degrees) 43.5 41.5 Effective unit weight (lb/in3) 0.038 0.035 coefficient of subgrade reaction Kpy (lb/in3) 225 30

Pile type in Al Tube Steel Rod

Maximum void ratio 1 1

Minimum void ratio 0.618 0.618

To determine whether multilinear spring stiffnesses can be used to approximately describe the p-y curve, pile responses under lateral loading were determined in STAAD Pro based on the multilinear spring stiffness approximation; these results were compared with results obtained using LPILE which has p-y curves already incorporated into the software. Figure 3 shows that the pile lateral deflection calculated based on the two approaches (multilinear spring stiffnesses approximation and p-y curves) matched very well. This result verified that the multilinear spring approximation used in STAAD Pro was able to satisfactorily represent the true p-y behavior of the soil effects on the piles. As a result, it was determined that adequate modeling of structures and soils could be integrated in a structural analysis performed in STAAD Pro. The calculated buckling loads for the aluminum tube and steel rod using the p-y based approach for buckling analysis and the Davisson and Robinson (1965) approach have been summarized in Table 3. The buckling loads obtained using both approaches agreed well. Calculated values for the aluminum tube matched the measured values quite well; however, there was a small deviation (10%-20% offset) from the measured values for the steel rod. The reason for this difference could be that the selection of Kpy for the p-y based approach was not based on test data, and therefore, the Kpy value may be different from the actual value. In general, it was found that the computations based on both approaches were reasonable for the buckling analysis of axially loaded piles.

2010 Structures Congress 2010 ASCE

115

Deflection of the top of Al tube (in) -15 5 25 45 0

Deflection of the top of steel rod (in) -15 5 25 45 65 0


Depth measured from the top of steel rod (in)

Depth measured from top of the Al tube (in)

10 20 30 40 50 60 70 Staad Lpile

10 20 30 40 50 60 70 Staad Lpile

(a) Lateral load = 5 lb

(b) Lateral load = 40 lb

Figure 3. Comparison of lateral responses of (a) aluminum tube and (b) steel rod as analyzed using Staad Pro and LPile. Table 3. Comparison of buckling loads Pile type Calculated values (lbs) P-y curve-based Davisson and Robinsons approach approach 48 49 334 329 Measured values (lbs)

in Al Tube Steel Rod

50-55 280-300

BUCKLING OF PILES UNDER SCOURED CONDITIONS Pile buckling analyses using the p-y curve approach were performed on an 80 ft long HP10X24 pile in loose, medium, and dense, submerged sands. The soil properties are given in Table 4. Scour was applied to the pile foundation which was initially fully embedded with the pile head level at the ground surface. Water surface was assumed to be at the same elevation as the ground surface. Four scour depths, 10 ft, 20 ft, 30 ft, and 40 ft were investigated analytically (Figure 4). The base of each pile was fixed due to the sufficient length of the pile embedded in the soil to provide the fixity for the pile tip. However, different boundaries for pile heads were considered to account for the fact that different foundation systems have various boundary conditions, as shown in Figure 5. Therefore, the effects of sand density and boundary conditions on the buckling of piles under scour were evaluated.

2010 Structures Congress 2010 ASCE

116

P Initial ground surface 10 ft 20 ft 30 ft 40 ft Ground surface after scour

Sands

Figure 4. Design of scour depths for the buckling analysis of HP10x42 pile

Free

Fixed head with sway

Pinned

Fixed

Figure 5. boundaries of HP10x42 pile

Buckling of the HP10x42 pile with the free head condition was evaluated in different sand conditions, and the results are presented in Figure 6. It can be seen

2010 Structures Congress 2010 ASCE

117

that as the scour depth increased, the buckling capacity of the HP pile decreased significantly. Small difference was observed regarding the effect of sand density on the buckling load; only minor deviation was found at shallow scour depths, exhibiting the decrease in the buckling load with a change of sand density from dense to loose. Moreover, the variation of buckling capacity of the HP pile with scour depth was nonlinear with a decreasing tangent slope. The current study only considered elastic buckling. An increasingly sophisticated curve would show the combination of inelastic and elastic buckling. This is because at a relatively shallow scour, the unsupported pile length is relative short, and the pile would be expected to be relatively short or intermediate in length. However, as scour progressed, the pile becomes more slender and consequently should buckle in an increasingly elastic manner.
Table 4. Properties of submerged sands Sand type Loose sand Medium sand Dense sand Frictional angle (degrees) 29 33 39 Effective unit weight (lb/in3) 0.019 0.028 0.039 Coefficient of subgrade reaction, Kpy (lb/in3) 20 60 125

160 140 loose sand medium sand dense sand

Buckling load (kips)

120 100 80 60 40 20 0 0 10

20 30 Scour depth (ft)

40

50

Figure 6. Buckling load versus scour depth in different sands (free pile head)

Effects of different boundary conditions on the buckling of piles due to scour were also investigated. Four boundary conditions of the pile head, i.e. free of support, fixed head with sway (no rotation), pinned, and fixed, were considered. The results shown in Figure 7 illustrate the buckling loads supported by piles with various head boundary conditions in loose sand at different scour depths. The trends in the figure show that scour reduced the buckling capacity no matter what boundary of pile head

2010 Structures Congress 2010 ASCE

118

was. Fixity of the pile significantly affected the buckling loads supported by the piles. Differences in buckling capacities for different boundary conditions were found to be more obvious at small scour depths. However, as scour depth increased, the variation of buckling load due to different boundary conditions was overtaken by the overall pile slenderness. It should be mentioned that the calculated buckling loads exceeding the pile yielding capacity of 620 kips (equal to 50 ksi times 12.4 in2) should not be taken into account, as shown in Figure 7. This result was attributable to the sole consideration of elastic buckling within the buckling analyses.
2000 1800 1600 1400 1200 1000 800 600 400 200 0 0 10

free head swayed head pinned head fixed head


Yielding load (yield stress 50 ksi)

Buckling load (kips)

20 30 Scour depth (ft)

40

50

Figure 7. Buckling load versus scour depth with different pile head boundary conditions (loose sands)

CONCLUSIONS A p-y based approach for buckling analysis was developed for the buckling of piles due to scour. This approach was verified through the comparison of the experimental data (Lee, 1968) and the calculated results using the method of Davisson and Robinson (1965). The p-y based approach has advantages over other methods for buckling analysis. It enables meaningful inclusion of soil effects on pile behavior in the structural analysis model, including lateral, vertical, and torsional soil resistance; it can provide a more reasonable solution when considering nonlinear soil behavior; it enables consideration of nonhomogeneous soils, such as layered soils; and it can be applied to the analysis of fully or partially embedded piles. Several conclusions were drawn in this study as follows: (1) Use of multilinear springs as a close approximation of soil p-y behavior was successful; lateral deflections of piles using both multilinear soil spring stiffness approximation and p-y curves matched very well.

2010 Structures Congress 2010 ASCE

119

(2) The p-y based approach was found to have the results consistent with those from the method of Davisson and Robinson (1965) and measured from the experimental tests performed by Lee (1968). (3) Scour reduced the buckling load of the HP10X42 pile; the results in this study considered elastic buckling; however a more sophisticated solution should also include inelastic buckling. (4) The effect of sand density on the buckling capacity of HP10X42 piles was found to be insignificant. (5) The boundary conditions of the pile affected the buckling distinctly at small scour depths but indistinctly at greater scour depths. ACKNOWLEDGEMENTS This research project is funded by the Kansas Department of Transportation (KDOT) through the KTRANS program. Mr. John Jones at KDOT is the project monitor. The authors would like to express their appreciation to KDOT for this support. The results and opinions presented in this paper are the authors and do not reflect the policy and recommendation of KDOT. REFERENCES AASHTO (2007). "AASHTO LRFD Bridge Design Specifications." American Association of State Highway and Transportation Officials, Washington, DC. Avent, R. R., and Alawady, M. (2005). "Bridge scour and substructure deterioration: Case study." J Bridge Eng, 10(3), 247-254. Davisson, M. T., and Robinson, K. E. (1965). "Bending and buckling of partially embedded piles." Proceedings of 6th International Conference on Soil Mechanics and Foundation Engineering, , Canada, 243-246. Gabr, M. A., Wang, J. J., and Zhao, M. (1997). "Buckling of piles with general power distribution of lateral subgrade reaction." Journal of Geotechnical and Geoenvironmental Engineering, 123(2), 123-130. Heelis, M. E., Pavlovic, M. N., and West, R. P. (2004). "The analytical prediction of the buckling loads of fully and partially embedded piles." Geotechnique, 54(6), 363-373. Kulhawy, F. H., and Mayne, P. W. (1990). "Manual on Estimating Soil Properties for Foundation Design." Electric Power Research Institute, Palo Alto. Lee, K. L. (1968). "Buckling of Partially Embedded Piles in Sands." Journal of the Soil Mechanics and Foundations Division, ASCE, 94(1), 255-270. Ramey, G. E., Brown, D. A., Hughes, M. L., Hughes, D., and Daniels, J. (2007). "Screening tool to assess adequacy of bridge pile bents during extreme flood/scour events." Practice Periodical on Structural Design and Construction, 12(2), 109-121.

2010 Structures Congress 2010 ASCE

120

Reese, L.C. and Van Impe, W.F. (2001). Single piles and pile groups under lateral loading, Rotterdam: A.A.Balkema. Terzaghi, K. (1955). "Evaluation of Coefficients of Subgrade Reaction." Geotechnique, 5, 297-326.

You might also like