Physical modelling of large diameter piles in coarse-grained soil

Transcription

1 Physical modelling of large diameter piles in coarse-grained soil K. T. Brødbæk 1, A. H. Augustesen and M. Møller COWI A/S, Aalborg, Denmark S. P. Hyldal Sørensen Aalborg University, Denmark ABSTRACT Monopiles are an often-used foundation concept for offshore wind turbine converters. These piles are highly subjected to lateral loads and overturning bending moments due to wind and wave forces. To ensure enough stiffness of the foundation and an acceptable pile-head deflection, monopiles with diameters of 4 to 6 m are typically employed. In current practice these piles are traditionally designed by means of the p-y curve method although the method is developed and verified for slender piles in sand with diameters up to approximately 2 m. One of the limitations of the p-y curves used in current design (e.g. [1] and [2]) is the effects of diameter on the initial part of the p-y curves. This part is especially important in connection with the serviceability limit state and fatigue. The effects of diameter on the p-y curves can be investigated by means of either numerical analyses or physical modelling (large- or small-scale). This paper investigates the effects of diameter on the initial part of the p-y curves by small-scale testing. A new and innovative test setup is presented. In order to minimize scale effects the tests are successfully carried out in a pressure tank enabling the possibility of increasing the effective stresses. The test setup is thoroughly described in the paper. Two non-slender aluminium pipe piles subjected to lateral loads have been tested in the laboratory. The piles are heavily instrumented with strain gauges in order to obtain p-y curves, displacement and bending moment distributions along the pile. Keywords: large-diameter piles, lateral load, laboratory test, pressure tank, p-y curves, cohesionless soil 1 INTRODUCTION Several concepts for offshore wind turbine foundations exist. The choice of foundation primarily depends on site conditions and the dominant type of loading. The most common used foundation concept is monopiles, which are single steel pipe piles driven open-ended. Monopiles typically have diameters around 4 to 6 m and a pile slenderness ratio (L/D) around 5 where L is the embedded length and D is the outer diameter. In current design of monopiles, the p-y curve method is normally employed as described by [3]. A p-y curve describe the non-linear relationship between the soil resistance acting against the pile wall, p, and the lateral deflection of the pile, y. Several formulations exist depending on the type of soil. The p-y curves for coarse-grained material employed in the offshore design regulations, e.g. [1] and [2] are given as a hyperbolic function. The expression is developed based on the testing of two identical, instrumented piles with diameters of 0.61 m and embedment lengths of 21 m installed at Mustang Island, Texas as described by [4]. The tests at Mustang Island included a total of seven load cases. Furthermore, the tests were conducted for only 1 COWI A/S, Thulebakken 34, DK-9000 Aalborg, Denmark. ktbr@cowi.dk

2 one pile diameter, one type of sand, and for circular pipe piles. Due to the very limited number of tests to validate the method, the influence of a broad spectrum of parameters on the p-y curves remains still to be clarified. According to [5] the p-y curve method is valid for diameters up to 2 m. For modern wind turbine foundations only small pile head rotations are acceptable. Furthermore, the strict demands to the total stiffness of the system due to resonance in the serviceability mode increase the significance of the p-y curves initial slope and hereby the initial stiffness of the soil-pile system. It seems questionable that the initial stiffness of the p-y curves are independent of the pile properties among these the pile diameter. The research within the field of diameter effects gives contradictory conclusions as described by [6]. Figure 1. Pressure tank - test setup. This paper evaluates the influence of pile diameter on the initial part of the p-y curves through small-scale tests. Such tests are however attached to significant scale effects. An often introduced source of error is the low stress levels relating the soil parameters and in specific the angle of internal friction to vary strongly with the effective stresses. Therefore, it is an advantage to increase the effective stresses to a level where the angle of internal friction is independent of a possible stress variation during the tests. Furthermore, an increase in effective stress level minimizes fluctuations in the measured signals. The low stress level problem can be overcome by testing piles in a centrifuge or in a pressure tank where overburden pressure can be applied by regulating the air pressure. This paper presents the results obtained by testing two instrumented closed-ended aluminum pipe piles in a pressure tank. The outer diameters of the piles are 60 and 80 mm, respectively. The piles are subjected to a quasi-static horizontal load at a constant vertical eccentricity. 2 TEST SETUP 2.1 Pressure tank The tests have been carried out in a pressure tank at Aalborg University, Denmark. A cross sectional view of the test setup is illustrated in Figure 1. The diameter and the height are approximately 2.1 m and 2.5 m, respectively. The test tank is installed in a load frame resting on a reinforced foundation. The pressure tank is furnished with trap doors for test preparation. Furthermore, it contains portholes where the measurement devices etc. are led out. At the top hatch a hydraulic piston is mounted for pile installation. The tank contains up to 0.69 m of fully saturated sand, cf. Figure 1. A highly permeable layer of gravel is located underneath the sand for drainage. The increase in effective stress level, compared to traditional 1G tests, is created by increasing the air pressure in the upper part of the tank, which is separated from the lower part by a rubber membrane. Hence, a homogeneous increase in the effective stresses is introduced at the soil surface via the elastic membrane. At large overburden pressures the effective stress level along the pile is almost constant. In that case the tests reflect the behaviour of a slice or part of a pile located in sand with an almost constant angle of internal friction at a given depth. In contrast, at low overburden pressures the effective vertical stresses varies from zero at the soil surface to approximately 4 kpa at the pile toe. Hence, the tests reflect the behaviour of a pile installed in sand with varying and very large angles of internal friction due to the low stress levels applied. To ensure limited excessive pore pressures, the soil is connected to an ascension pipe. The pile is led through a sealing in the membrane

3 allowing the pile to be extended above the soil surface. A rubber band is attached to the outer perimeter of the membrane. On the outside of this vertical rubber band two mouldings are attached, cf. Figure 2. The purpose of these is to minimize gaps due to imperfections between the tank wall and the membrane. The rubber band and mouldings are pushed towards the tank wall, and kept in place, by a coil. The coil is made up from a fire hose and has the exact same diameter as the pressure tank. When inflated, the coil provides an equal pressure at the rubber band through the whole perimeter of the membrane. Approximately 16 cm of water is poured in the upper part of the tank to ensure that the soil is fully saturated if leaks in the interface between the membrane and walls of the tank are present. Furthermore, the dynamic viscosity of water is about 55 times higher that in air, which minimize the flow through any potential gaps. It has been observed that a water volume of approximately liter/hour (corresponding to 2% to 3% of the total water volume) passes through gaps between the membrane and the walls of the pressure tank. order to protect the strain gauges and cords against water. 10 strain gauges (HBM, type K-LY43-3/120) are mounted at 5 levels as shown in Figure 3. At each level two foil strain gauges are mounted in grooves milled in the pile with a mutual angle of 180 oriented in the cross-sectional plane. The grooves are sealed to protect the strain gauges. The depth, width, and length of the mill outs are approximately 2, 6, and 10 mm, respectively. The corresponding reduction in bending stiffness is negligible compared to the bending stiffness of the remaining profile. The cables from the strain gauges are led into the pile and through a hermetic packed hole in the pile-head. A hydraulic piston, cf. Figure 1, is employed to actuate the test piles horizontally. The line of attack of the pistons is 0.37 m above the soil surface. The pile and the hydraulic piston are connected via a steel wire. The force acting on the pile is measured by a force transducer (HBM, type U2B 10 KN), which is placed in series in between the hydraulic piston and the wire. Lateral deflections are measured at three levels as shown in Figure 4, above the soil surface by means of wire transducers (ASM GMBH, type WS R1K-L10). Figure 2. Cross-section of the joint between the membrane and the walls of the pressure tank. 2.2 Measuring system Quasi-static tests on two instrumented aluminum pipe piles with outer diameters of 60 and 80 mm, respectively, have been conducted. Both piles have a slenderness ratio of L/D = 5 corresponding to an embedded length of 300 and 400 mm, respectively. The wall thickness s of the piles are 5 mm and they are closed-ended in Figure 3. Strain gauge levels. Measures are in mm. 2.3 Soil conditions The soil employed is Baskarp Sand no.15, which is graded sand from Sweden with the characteristics given in Table 1. The hydraulic conductivity is m/s which ensures drained conditions during the tests.

4 conducted. The employed CPT is a prototype with a diameter of 15 mm. A typical profile of the cone tip resistance, q c, is shown in Figure 5. The tests denoted "references" present the results of CPT tests prior to pile installation. As the profiles of q c are without particular variation the soil is considered homogeneously compacted. After the pile is installed and the soil is vibrated the membrane is very carefully placed on the soil and more water is poured into the tank. Figure 4. Measurement of horizontal displacement at three different levels. Measures are in mm. Table 1. Material properties Baskarp Sand No. 15, after [7]. Specific grain density, d s 2.64 Maximum void ratio, e max Minimum void ratio, e min d 50 = 50 % - quantile 0.14 mm U = d 60 / d Test preparation and test execution Before installing the test pile the sand is mechanically vibrated to ensure a homogeneous soil. The piles are installed in one continuous motion by means of a hydraulic piston mounted in the top of the pressure tank. The test pile is installed in the centre of the tank, and the strain gauges are aligned in the vertical plane of the horizontal load. During installation an upward gradient (pore water pressure) of 0.9 has been applied in order to ease the installation of the test pile. After the installation of the pile the sand is mechanically vibrated minimizing the disturbances from the pile installation, i.e. a homogeneous compaction of the sand is ensured. Pile displacements are prevented during vibration by fixation of the pile to the hydraulic piston mounted on the top hatch. After installation the compaction and homogeneity of the soil have been controlled by cone penetration tests (CPT). Four CPT's with a distance of 0.5 m from the centre of the pile and two 0.16 m from the neutral sides of the pile, i.e. the sides perpendicular to the load direction, are 2.5 Interpretation of strain gauge measurements The p-y curves are derived based on the bending moment distribution along the pile, M(x), and the pile bending stiffness, E p I p : M( x) yx ( ) = dxdx (1) EI p p 2 d M( x) px ( ) = (2) 2 dx The double integration of the discrete data points with respect to depth does not result in significant errors. However, double differentiation of the discrete signals gives an amplification of measurement uncertainties. In order to minimize these errors the piecewise polynomial curve-fitting method described by [8] is employed. When using this method the moment distribution is estimated by fitting five successive moment data points to three order polynomials. In order to ensure a proper relation between the strain gauge output and bending moment distributions, calibrations of the piles have been conducted by loading the piles transversely while supported as simply supported beams. In order to eliminate effects of stress concentrations, four load series with varying locations of the point load have been conducted for each pile. At each location load series consisting of seven load steps of 20 kg, from 0 to 120 kg have been applied.

5 x [mm] q c [N] References vertical distance between the two displacement transducers and the measured horizontal deflections the rotation of the pile at the height of the hydraulic piston is obtained Figure 5. Output from CPT's. 3 RESULTS Six tests have been conducted at overburden pressures (applied as air pressures in the tank), p 0 = 0 kpa, p 0 = 50 kpa and p 0 = 100 kpa. During the tests, the soil is brought to failure, unloaded, and reloaded in order to estimate the ultimate soil resistance and the elastic behavior of the soil. Figure 7. Relationships between normalised soil resistance and normalised displacement (p 0 = 0 kpa). Figure 6. Lateral pile displacement at different stress levels for D = 80 mm. Figure 6 presents the lateral pile displacements with depth for the pile with D = 80 mm. A prescribed displacement of 10 mm at the level of the hydraulic piston is applied. The lateral displacement can be separated into two components; deformation of the pile due to bending and rotation of the pile as a rigid object. The pile deformation due to bending is calculated according to equation (1). The pile rotation is obtained by the displacement transducers at the top of the pile and at the height of the hydraulic piston, cf. Figure 5. Based on the Figure 8. Relationships between normalised soil resistance and normalised displacement (p 0 = 100 kpa). As shown in Figure 6 the pile behaves almost as a rigid object when p 0 = 0 kpa. When applying an overburden pressure the pile deformation caused by bending becomes more significant, but the pile rotation is still dominant. Due to the rigid behaviour of the pile, the deflection at the pile toe must be negative, which is not the case in all tests. This could be due to uncertainties in the vertical distance between the displacement

6 transducers, c.f. Figure 5, and the few numbers of strain gauges, which may lead to uncertainties when determining the rotation. Similar results is obtained for the pile with D = 60 mm. Figure 7 and 8 present normalised p-y curves at two depths. Generally, for a given deflection, y, and pile diameter, D, the pressure, p [kn/m], acting against the pile wall increases with increasing depth, x. Moreover, for a given deflection and depth, p increases with D. From Figure 7 and 8 it is also shown that the initial stiffness of the p-y curve is highly dependent on the pile diameter with the highest initial stiffness relating to the largest pile diameter. When p 0 = 0 kpa the initial stiffness of the pile with an outer diameter of 80 mm is approximately 3 to 4 times higher than the stiffness for the pile with an outer diameter of 60 mm. Higher initial stiffness relating to the highest diameter is in agreement with the results presented by [9] and [10]. In contrast, [11], [12] and [13] conclude that the effects of the diameter on the initial stiffness of the p-y curves are insignificant. It must be emphasized that the results presented in this paper are indicative and show general tendencies. Much more research within the area is needed. 4 CONCLUSIONS A new and innovative test setup used for testing horizontally loaded piles in the laboratory is presented. The test setup is an alternative to centrifuge testing. Focus has been paid to the physical description of the setup, the measuring system, test preparation and execution. The paper presents the results of six quasistatic tests on two non-slender monopiles. The tests are carried out at varying overburden pressures from 0 to 100 kpa. The two aluminum pipe piles both have a slenderness ratio, L/D, of 5. The piles are each instrumented with a total number of 10 strain gauges located at five levels along the piles. The conclusions that can be drawn are: - The two non-slender piles deflect as almost rigid objects given only one point of zero deflection. Hereby, negative deflections at the pile toe are observed. - The initial stiffness of the p-y curve is highly dependent on the pile diameter; the larger pile diameter the larger stiffness. This observation conflicts with the recommendations in the design regulations. It must be emphasized that the results presented in this paper are indicative and show general tendencies. REFERENCES [1] API, Recommended practice for planning, designing, and constructing fixed offshore platforms - Working stress design, API RP2A-WSD, American Petroleum Institute, Washington D.C, 21. Edition, [2] DNV, Foundations - Classification Notes No. 30.4, Det Norske Veritas, Det Norske Classification A/S, [3] L.C. Reese, W.F. Wan Impe, Single Piles and Pile Groups Under Lateral Loading, Taylor & Francis Group plc, London, [4] W.R. Cox, L.C. Reese, B.R. Grubbs, Field Testing Of Laterally Loaded Piles in Sand, Proceedings of the 15th Annual Offshore Technology Conference, Houston, Texas, 2079, [5] J.M. Murchison, M.W. O Neill, Evaluation of p-y relationships in cohesionless soils, Analysis and Design of Pile Foundations, Proceedings of a Symposium in conjunction with the ASCE national Convention, , [6] K.T. Brødbæk, M. Møller, S.P.H. Sørensen, Review of p-y relationships in Cohesionless Soil, DCE Technical Report no. 57, Department of Civil Engineering, Aalborg University, Denmark, [7] K. A. Larsen, Static Behaviour of Bucket Foundations, Department of Civil Engineering, Aalborg University, Denmark, [8] K. Yang,R.Liang, Methods for Deriving p-y Curves from Instrumented Lateral Load Tests, Geotechnical Testing Journal, 30(1), Paper ID GTJ100317, [9] D.P. Carter, A Non-Linear Soil Model for Predicting Lateral Pile Response, Rep. no. 359, Civil Engineering Dept., Univ. of Auckland, New Zealand, [10] L.F. Ling, Back Analysis of Lateral Load Test on Piles, Rep. no. 460, Civil Engineering Dept., Univ. of Auckland, New Zealand, [11] K. Terzaghi, Evaluation of coefficients of subgrade reaction, Geotechnique, 5(4), , [12] S.A. Ashford, T. Juirnarongrit, Evaluation of Pile Diameter effect on Initial Modulus of Subgrade Reaction, Journal of Geotechnical and Geoenvironmental Engineering, 129(3), , [13] C.C. Fan, J.H. Long, Assessment of existing methods for predicting soil response of laterally loaded piles in sand, Computers and Geotechnics, 32, , 2005.