RATZ Version 4-2 LOAD TRANSFER ANALYSIS OF AXIALLY LOADED PILES M. F. Randolph February 2003 RATZ Version 4-2 LOAD TRANSFER ANALYSIS OF AXIALLY LOADED PILES M. F. Randolph February 2003 The accuracy of this program has been checked, and it is believed that, within the limitations of the analytical model, results obtained with the program are correct. However, the author accepts no responsibility for the relevance of the results to a particular engineering problem. Technical support in relation to operation of the program, or in respect of engineering assistance, may be obtained from the author, who reserves the right to make a charge for such assistance. Address: Centre for Offshore Foundation Systems The University of Western Australia Perth Western Australia 6009 Telephone: +61 8 9380 3075 Email: mark.randolph@uwa.edu.au February 2003 RATZ Manual Version 4-2 M.F. Randolph CONTENTS Page No. PART A: GENERAL DESCRIPTION 1 INTRODUCTION ............................................................................................1 2 LOAD TRANSFER CURVES .........................................................................2 2.1 Monotonic Loading...............................................................................2 2.1.1 2.2 Hyperbolic Pre-peak Load Transfer Curve ...............................3 Cyclic Loading ......................................................................................4 2.2.1 Standard Yield Criterion ...........................................................4 2.2.2 Alternative Yield Criterion A ...................................................5 2.2.3 Alternative Yield Criterion B ...................................................6 2.3 Cyclic Residual Shaft Friction ..............................................................6 2.4 Simulating Creep Effects ......................................................................6 2.5 Group Effects ........................................................................................7 2.6 Pile Base Response ...............................................................................8 2.6.1 Hyperbolic Base Response .......................................................8 3 EXPLICIT TIME INTEGRATION ..................................................................9 4 EXTENSIONS TO PROGRAM.......................................................................10 5 4.1 Residual Stresses ...................................................................................10 4.2 External Soil Movement .......................................................................10 4.3 Thermal Strains .....................................................................................11 4.4 Measured Test Data ..............................................................................11 EXAMPLE ANALYSES..................................................................................11 5.1 Bored Pile in Interbedded Clays and Sands ..........................................11 5.2 Driven Precast Concrete Pile ................................................................11 5.3 Rock-socketed Pile................................................................................12 5.4 Offshore Drilled and Grouted Pile ........................................................12 5.5 Monotonic Loading...............................................................................13 5.6 Cyclic Loading ......................................................................................13 PART B: PROGRAM DOCUMENTATION 6 STRUCTURE OF PROGRAM ........................................................................15 6.1 7 Overview of Program Operation ...........................................................15 PROGRAM INPUT ..........................................................................................16 7.1 Data Input..............................................................................................16 7.2 Data Items .............................................................................................16 7.2.1 Problem Title ............................................................................16 February 2003 7.3 8 RATZ Manual Version 4-2 M.F. Randolph 7.2.2 Pile Segement Properties ..........................................................16 7.2.3 Pile Details and Base Response ................................................16 7.2.4 Soil Details................................................................................17 7.2.5 Miscellaneous Parameters.........................................................17 7.2.6 Soil Profile ................................................................................17 7.2.7 Loading Details .........................................................................18 7.2.8 Residual Loads ..........................................................................18 7.2.9 Downdrag..................................................................................19 7.2.10 Thermal Strains .......................................................................19 Measured Load Test Data .....................................................................20 PROGRAM OUTPUT ......................................................................................20 8.1 Printed Output .......................................................................................20 8.2 Graphical Output ...................................................................................20 9 REFERENCES .................................................................................................21 10 FIGURE TITLES..............................................................................................23 FIGURES .....................................................................................................................24 February 2003 RATZ Manual Version 4-2 M.F. Randolph PART A: GENERAL DESCRIPTION 1 INTRODUCTION The overall load deformation response of an axially loaded pile depends on the axial compressibility of the pile as well as on the shape of the load transfer curves for the soil around the pile. These curves may be thought of as the integrated (in a radial direction) effects of shear strains in the soil, leading to relationships between the shear stress applied at the pile wall and the resulting local displacement of the soil. Unless the pile is very compressible, changes in vertical stress induced in the soil will be small compared with the induced shear stresses. This justifies the load transfer approach, where the soil continuum is idealised as a number of separate horizontal layers, each with its own load transfer curve, (see Figure 1). The compression or extension of the pile may be calculated from consideration of the variation of axial load, P, down the length of the pile. For a pile of radius, ro, and cross-sectional stiffness, (EA)p, the variation of axial load is given by dP 2ro o dz (1) where o is the shear stress at the pile wall. The axial strain in the pile is dw P z EA p dz (2) Combining these two equations gives the governing equation for axial compression of d2w dz 2 2ro EA p o (3) Traditionally, this equation has been solved, in conjunction with appropriate load transfer curves relating o and w, by dividing the pile into elements, and adopting a finite difference approximation for the second derivative of w. For non-linear load transfer curves, this approach requires iteration at each increment of axial load. For an adequate discretisation of the pile, the computing effort required becomes considerable, particularly if the effects of cyclic loading are to be investigated. An alternative approach, which avoids the assembly and solution of any system of equations, is the so-called explicit time approach described, for example, by Cundall and Strack (1979). The technique involves the introduction of time as a variable (artificially in the case of straightforward static loading) and following the load applied to the pile head as it propagates down the pile, causing local accelerations, velocities and thus 1 February 2003 RATZ Manual Version 4-2 M.F. Randolph displacements of the pile elements. By using a sufficiently large number of load increments, and time steps which are small in comparison with the time taken for a shock wave to travel across each pile element, any loading path for the pile may be followed in a stable fashion. The approach is described in more detail later. 2 LOAD TRANSFER CURVES In choosing appropriate load transfer curves for the soil, it is convenient to divide the response of the soil into two distinct phases (for example, Kraft et al (1981)). Initially, the soil will deform as a continuum, with shear strains induced according to the applied shear stress at the pile wall. For a given shear stress, these shear strains will give rise to displacements of the soil around the pile that are proportional to the pile radius - even for non-linear stress-strain response of the soil. For this stage of loading, it is therefore helpful to consider relationships between o and w/ro (that is the normalised pile displacement). As the shear stress level reaches the maximum value that the interface between pile and soil can withstand, a rupture surface will develop close to the pile. Thereafter, the shear stress level may be expected to vary with the absolute displacement of the pile relative to the soil, and it becomes more logical to consider relationships between o and the absolute displacement, w. Figure 2 shows this two stage approach. 2.1 Monotonic Loading In the program, a general form of load transfer curve has been adopted, although specific curves could be incorporated with minor modifications. The form of curve has no precise theoretical basis, but allows sufficient flexibility to model most aspects of soil response, for example non-linear, work hardening behaviour prior to peak, followed by strain-softening after peak. There are three main stages to the curve (see Figure 2): (1) A linear stage where = kw/ro, which extends from zero shear stress up to a fraction (xi) of the peak shear stress, p. (2) A parabolic stage, with initial gradient, k, and final gradient of zero when o = p. (3) A strain-softening stage, where the current value of shaft friction is related to the absolute pile displacement by o p 1.1 p r 1 exp 2.4(w / w res ) (4) where r is the residual value of shaft friction, reached after an additional displacement of wres, with w being the post-peak displacement. The parameter (eta) controls the shape of the strain softening curve. The value of the initial gradient, k, may be related to the shear modulus, G, for the soil 2 February 2003 RATZ Manual Version 4-2 M.F. Randolph (Randolph and Wroth, 1978; Baguelin and Frank, 1979). The ratio k/G will generally lie in the range 0.2 - 0.3, with the smaller value corresponding to more slender piles. Following Randolph and Wroth (1978), the relationship between k and G may be expressed as k G (5) The parameter (zeta) is given by n2.51 / ro n rm / ro (6) where is the ratio of the average shear modulus over the depth of penetration of the pile, to that at the bottom of the pile shaft ( = G /G in the notation of Randolph and Wroth). The other parameters in equation (6) are Poisson’s ratio for the soil, , and the pile length, l. For typical pile geometries, the value of is about 4, giving k/G = 0.25. Figure 3(a) shows the pre-peak load transfer curve for the case of = 4, G/s = 400 and = 0 (parabolic from origin). The peak shaft friction is mobilised at a displacement of 1 % of the pile diameter, which agrees with recommendations for load transfer curves in the API RP2A guidelines. The parameter, , which defines where the load transfer curve becomes non-linear, may be taken anywhere in the range 0 to 1. A value of zero corresponds to a parabolic load transfer curve right from the origin, while a value of unity corresponds to a linear load transfer curve right up to failure. For the strain softening part of the load transfer curve, typical values for the parameter are about unity, with lower values giving more abrupt, and higher values more gradual, strain softening. Figure 3(b) shows examples of load transfer curves for values of = 0.8, 1 and 1.2, for the case of r/p = 0.6 and wres taken as 4 % of the pile diameter. 2.1.1 Hyperbolic Pre-peak Load Transfer Curve An alternative form of pre-peak load transfer curve has been implemented for monotonic loading only, based on a hyperbolic stress-strain model for the soil. The basic hyperbolic model gives a secant shear modulus which decreases linearly with stress ratio, as G sec G i 1 R f ( / f ) (7) where f is the failure shear stress, Gi is the initial tangent shear modulus and Rf is a parameter which controls the degree of non-linearity, and is generally taken in the range 0.8 - 1 for most soil types. This relationship may be integrated in order to derive an equivalent load transfer curve for a pile (Randolph, 1977; Kraft et al, 1981). The final form of load transfer curve may be expressed as 3 February 2003 RATZ Manual Version 4-2 w * M.F. Randolph o ro Gi (8) where the parameter * now varies with the shear stress level according to r / r * n m o where R f o p 1 (9) Figure 3(a) shows a comparison of the pre-peak load transfer curve obtained using the hyperbolic stress-strain response, compared with the parabolic curve. It may be seen that a value of Rf = 0.95 gives a response which matches the parabolic curve very closely. Lower values of Rf give curves which are more linear, and may be matched using the linear-parabolic model, with increasing values. 2.2 Cyclic Loading In order to simulate degradation of load transfer under the action of cyclic loading, particular attention has been paid to the manner in which unloading and reloading is modelled. Consider a monotonic loading path ABCD in Figure 2, where the current failure stress at point C is f. On unloading and reverse shearing (path CE), the failure shear stress is assumed to be the same magnitude, that is -f. Initially, unloading will be elastic. However, at some point yield will occur. If there has been no previous history of unloading, then yield is taken to occur at the same point as for loading, that is at a shear stress of -p. The yield point for reloading (and also for unloading where previous unloading has occurred) is calculated from the detailed path followed. 2.2.1 Standard Yield Criterion In the standard yield criterion, the yield point is assumed to move down the unloading path at a rate of 0.5(1 - ) times the rate of the current stress-displacement point. Thus if the pile reverses direction such that the mobilised shear stress decreases by , then the yield stress reduces by 0.5(1 - ). In practice, this is achieved by defining a yield point in relation to the minimum shear stress, min, and the peak shaft friction, p, as y min 0.51 p min (10) The form of this expression automatically ensures that the yield stress is always greater than p (the initial yield point). A check is made to ensure that the yield stress never increases during unloading. Once yield occurs, the plastic (non-recoverable) component of displacement is treated as equivalent to post-peak monotonic displacement. This leads to gradual degradation of the shaft friction from peak to residual, even though there may be no net accumulation of 4 February 2003 RATZ Manual Version 4-2 M.F. Randolph displacement. The particular algorithm for calculating the yield point implies that, under one-way cyclic loading between zero and a maximum value of max, degradation to residual shaft friction will ultimately occur if max exceeds the elastic limit given by max 0.51 p (11) Under symmetric two way loading between limits of ±max, the corresponding elastic limit is 1 max p 3 (12) For cyclic loading that exceeds the limits given by equations (7) and (8), the rate of degradation is a function of the shape of the strain softening part of the monotonic loading curve, together with the extent by which the cyclic shear stress exceeds the above limits. The expressions (11) and (12) allow the safe cyclic range to be varied by means of the parameter . For = 1, the cyclic range extends right up to the peak shaft friction, while for = 0, the safe cyclic amplitude varies linearly with the mean shear stress, from ± p/3 for symmetric two-way cyclic loading (mean shear stress of zero), through ± p/4 for one-way cyclic loading (mean shear stress equals cyclic amplitude) down towards zero as the mean shear stress approaches the peak shaft friction. This form of variation of the safe cyclic range is similar to that proposed for metals by Goodman, and is shown in Figure 4. In the more conventional cyclic stability diagram shown in Figure 5, the yield criterion becomes a straight line (see examples for = 0 and = 0.333). 2.2.2 Alternative Yield Criterion A The safe cyclic zone corresponding to = 0 (between the upper dashed and lower solid lines in Figure 4, or beneath the lower straight line in Figure 5) was considered to be potentially optimistic for the calcareous sediments of the North West Shelf, and possibly for other compressible or loose soils which are susceptible to volume decrease under cyclic loading. As such, the algorithm in RATZ has been modified to allow the safe cyclic zone to be reduced (in the extreme case, to nothing). In place of equation (10), the yield stress is taken as y min p min (13) with an overriding criterion that the yield stress should not be less than p. Clearly, if is taken as zero, yield occurs immediately, and the safe cyclic range disappears. The cyclic loading algorithm in RATZ may be used to establish theoretical fatigue curves, showing the number of cycles to failure for cyclic loading of varying amplitude and different mean shear stress. Figure 6 shows such fatigue curves for the original RATZ 5 February 2003 RATZ Manual Version 4-2 M.F. Randolph algorithm, for two-way (zero mean stress) cyclic loading and one-way (zero minimum stress) cyclic loading. Similar curves are shown in Figure 7 for the modified algorithm, taking = 0.1. The shape of the fatigue curves depend on the various load transfer parameters for the soil, and also on the radius of the pile. These parameters are summarised in Table 1. The fatigue curves shown in Figures 6 and 7 are compared with data from laboratory tests on intact samples of calcarenite. The data are reasonably consistent with the theoretical fatigue curves. 2.2.3 Alternative Yield Criterion B Both the above yield give linear threshold contours in the cyclic stability diagram shown in Figure 5. In practice, however, threshold contours from actual pile tests have tended to be convex, as shown by the two parabolic curves (Poulos, 1988). Although such an approach is less conservative than the linear thresholds discussed above, an alternative yield criterion has been implemented in order to match measured data. This is referred to as the ‘parabolic’ yield criterion in the main menu of the program. 2.3 Cyclic Residual Shaft Friction In granular soils, particularly lightly cemented sands, it is often found that the shear transfer during two-way displacement limited cyclic loading is very low. This may be modelled within the program by nominating a ‘cyclic residual shaft friction’, which limits the shear transfer within a limited zone. The manner in which this operates is shown in Figure 8. Initially the two markers that control the limit of the cyclic residual zone are positioned together at the origin. Before failure, the two markers move together, staying at the point where an elastic unloading response would intersect the displacement axis. After failure, the upper limit marker (the circle) continues to track the current position of the shear stress:displacement point, while the lower limit stays fixed. On unloading, the path BCDE is followed, with the shear resistance staying at the residual value within the ‘failure’ zone between the two markers. If the element is reloaded, from point F, the path FGHIJ is followed. The shear transfer remains at the cyclic residual until the upper marker is reached, after which the shear transfer reverts back to the main curve. Note, however, that the shaft friction at point I will be lower than at B, due to the plastic displacement (and accompanying degradation) that occurs during the cycle HCFGH. 2.4 Simulating Creep Effects In most geotechnical constructions, the effects of creep are small, amounting to perhaps 10 to 20 % additional movement compared with consolidation effects. The case of an axially loaded pile can be rather different, especially for long slender piles, since the shear stress level adjacent to the pile may be close to failure over the upper part of the pile. The creep can then become more significant, and can lead to creep rupture, whereby small 6 February 2003 RATZ Manual Version 4-2 M.F. Randolph movements accumulate and lead to a rupture surface at the pile wall, with peak shaft friction values lower than in a short term load test. Figure 9(a) shows a possible change in the form of the load transfer curve due to the effects of creep. At high shear stress levels, the displacements are increased in a localised band of soil adjacent to the pile, and the original peak value of shaft friction is never achieved. The explicit time approach used in the analysis is essentially a strain controlled approach. At each load increment, the current displacement of the pile determines the shear stress mobilised at the pile wall. Implementation of creep effects is thus most easily accomplished by a stress relaxation process, whereby the load transfer curve is shifted by a small amount over each time increment, as shown in Figure 9(b). In the program, the degree of movement is based on a creep rate law of the type proposed by Singh and Mitchell (1968) m dw t A o exp o dt t f (14) There are three constants in this equation, A, and m. In addition, the time to is the time associated with a standard rate of testing (at which the values of A and are determined). In order to express the creep in a non-dimensional form, equation (14) has been re-cast in the form m t w w * exp o t f (15) where w* is the displacement to mobilise peak shaft friction (point B in Figure 2). Typical values for the constants and m are 6 - 8 and 0.75 - 1.2 respectively (Singh and Mitchell (1968)). A parametric study indicates that values of should generally lie in the range 0 (no creep) to 0.01. The effects of different rates of loading are best investigated by varying the value of , thus avoiding the need for analyses with an excessive number of loading increments. In the program, creep is assumed to occur only between the yield point (o ≥ p) and failure. 2.5 Group Effects Pile group effects may be assessed by analysing a single pile with an appropriately modified load transfer curve. In the program, this is accomplished following the procedure suggested by Focht and Koch (1973) for laterally loaded piles. The user specifies a ‘group settlement ratio’, Rs, by which the elastic part of the load transfer curve is factored, to allow for interaction between piles in a group. The settlement ratio may be estimated following Randolph (1979), whereby the parameter is replaced by 7 February 2003 RATZ Manual Version 4-2 r * n m ro M.F. Randolph r n m si (16) The summation is taken over all remaining piles in the group, spaced at si from the pile in question, with the parameter rm being the maximum radius of influence of the pile, defined in equation (6). The settlement ratio, Rs, is then taken as the ratio */. It should be emphasised that only the elastic component of the load transfer curve is factored in the manner described above. The plastic component of the load transfer curve for single pile response is then added on to the factored elastic component in order to obtain the ‘group’ load transfer curve. Figure 10 illustrates this process. 2.6 Pile Base Response The load-displacement response of the pile base has been taken as parabolic in shape, with the vertex of the parabola at the ultimate base capacity. Unloading and reloading has been assumed elastic, subject to the condition of no tension being developed at the pile base. The response under monotonic loading is as shown by curve OACD in Figure 11. The base capacity is assumed to remain constant, independent of the magnitude of base displacement. For displacements outside the range of the initial parabola, the parabola is ‘dragged’ along to the right (increasing displacement) or left (decreasing displacement) without changing shape. The final position of the parabolic curve is shown dashed in Figure 11. Small unload-reload loops are elastic, with a gradient parallel to the initial gradient of the parabola (see AB in Figure 11). Larger loops are governed by the zero tension condition and the current position of the parabola (see DEFGD in Figure 11). 2.6.1 Hyperbolic Base Response A hyperbolic base response has been implemented, assuming a secant stiffness that decreases linearly with increasing load level. To be consistent with the parabolic base model, where a limiting end-bearing pressure of qbf is mobilised at a base displacement of wbf, the initial gradient is taken as k bi 2 q bf w bf (17) The secant stiffness at any pressure, qb, is then given by k b sec k bi 1 8 qb q bf q 2 bf w bf q 1 b q bf (18) February 2003 3 RATZ Manual Version 4-2 M.F. Randolph EXPLICIT TIME INTEGRATION As discussed earlier, pile-soil interaction is calculated using an explicit time integration procedure to solve the non-linear problem. The effect of pile compression or extension is incorporated by first dividing the pile into a number of elements, each of mass m. Each element is connected by a spring of stiffness, K, which represents the pile stiffness. For a pile of embedded length, l, divided into n segments, the value of the spring stiffness may be calculated as K EA p / n (19) Each element of the pile is connected to the soil by a non-linear spring, representing the load transfer curve at that particular depth (see Figure 1). At any stage during the analysis, a net force will act on any given pile element, which will cause the element to accelerate, leading to changes in its velocity, v, and displacement, w. For an element of mass, m, with out of balance force, F, the changes in velocity and displacement over a time period, t, may be calculated as t m (20) w t t w t v t t / 2 t (21) v t t / 2 v t t / 2 Ft and The form of these equations assumes that the force, Ft, is the average value of the force over the time period t-t/2 to t+t/2. Similarly, vt+t/2 is the average velocity over the period t to t+t. Each increment of load may effectively be seen as involving two scans of all the pile elements. In the first scan, the forces acting on each pile element are evaluated, and the net force calculated. In the second scan, the element velocities and displacements are updated according to equations (20) and (21). There are three components to the force acting on each pile element. The internal force on the ith element in the pile is given by Fa i K w i1 w i K w i1 w i (22) while the force from the static component of the soil displacement is Fb i 2ro / n o i (23) In addition, stability of the solution is maintained by incorporating a damping force given 9 February 2003 RATZ Manual Version 4-2 M.F. Randolph by Fc i 2ro / n k ' v i (24) The quantity k' is taken from solutions for the dynamic response of soil around a pile, and is related to the shear modulus, G, of the soil and the soil density, , by (Novak et al, 1978); Simons and Randolph, 1985) k ' D f G (25) where the quantity Df is a factor specified by the user, ranging between zero and unity. The net force on the pile element is the algebraic sum of the three forces given above. For a pile loaded at a constant strain rate, the inclusion of a damping force such as that in equation (25) will give rise to a spurious excess force at the head of the pile, generally of a relatively small magnitude. This is allowed for within the program by subtracting the additional force prior to outputting the load displacement relationship for the pile (see later for discussion of program output). 4 EXTENSIONS TO PROGRAM In addition to straightforward monotonic and cyclic loading of the pile, a number of additional situations may be analysed. These are recent extensions, and include (a) external soil movement, causing downdrag, (b) thermal strains in the pile, (c) residual loads developed during pile installation, and (d) matching of a measured loaddisplacement response. These extensions are discussed briefly below. 4.1 Residual Stresses A more common cause of residual stresses within the pile are those set up during pile installation, generally in the form of a compressive load locked in at the base of the pile, and negative values of shear stress between pile and soil, particularly towards the lower end of the pile. The compressive base load may arise either from pile driving (especially for full-displacement piles) or from a base grouting operation on a cast-in-situ pile. A set of residual loads along the pile shaft is input prior to starting the analysis. 4.2 External Soil Movement At any stage during the loading history of the pile, the effect of external soil movement (either upward, causing heave, or downward, causing downdrag and negative friction) may be analysed. Essentially, the zero points (and other key markers) on the load transfer curves are shifted to reflect the soil movement at that depth. The shear stress at the pile shaft is then a function of the relative movement between soil and pile, rather than the absolute pile movement. 10 February 2003 4.3 RATZ Manual Version 4-2 M.F. Randolph Thermal Strains In some circumstances, curing of the concrete or grout in a large diameter pile can lead to very high temperatures within the pile. As the concrete sets, the pile starts to cool down and significant shrinkage may occur. This will give rise to residual stresses between the pile and the soil. The effect of such thermal strains may be modelled by specifying, at the initial loading stage, a distribution of thermal strains down the length of the pile. The effect of these strains is then computed prior to starting the main analysis. It should be noted that, for long piles, thermal strains may lead to actual slip between pile and soil. 4.4 Measured Test Data In the interpretation of pile load tests, it is useful to be able to compare the predicted and measured responses at the pile head. Test data giving the measure load-settlement response may be input and compared with the computed response, allowing iteration of the soil and pile parameters until a close match is achieved. This process is comparable to that of matching stress-wave data from dynamic tests. 5 EXAMPLE ANALYSES Several example problems are presented here in order to illustrate use of the program. The first example concerns a bored pile, for which a measured load-settlement response is also available. Other examples include a relatively stubby rock-socketed pile with a strain-softening shaft response, a driven precast pile with significant residual stresses, a downdrag example and a slender offshore pile subjected to monotonic and cyclic loading. 5.1 Bored Pile in Interbedded Clays and Sands Figure 12 shows a fit to the measured load test on a bored pile, cast under bentonite. the pile was 800 mm in diameter, by 20 m long. The response at small displacements is shown in the upper diagram for clarity. The ultimate shaft capacity has been modelled as 4.3 MN (using a hyperbolic soil model), while the base capacity has been taken as 3.5 MN (7 MPa), again with a hyperbolic response. The form of the hyperbolic base model is such that full mobilisation of the base capacity will require a base movement of 200 300 mm. As such, the total ‘deduced’ capacity of 7.8 MN exceeds the practical capacity of about 7 MN, mobilised at a settlement of 80 mm. The average tangent shear modulus for the shaft response is Gi = 200 MPa (G/p = 2350) and for the base is 68 MPa (Gib/qbf = 9.7). 5.2 Driven Precast Concrete Pile The second example is a precast concrete pile, 300 mm square, that was driven 34 m through layered alluvial deposits to found in a dense sandy gravel layer. Figure 13 shows 11 February 2003 RATZ Manual Version 4-2 M.F. Randolph a comparison of the computed response with and without allowance for residual stresses. In this instance, the residual stresses have been obtained from the back-analysis of a dynamic test on the pile. The pile response under working load would be significantly stiffer due to the presence of residual stresses in the pile. 5.3 Rock-socketed Pile The third example is a hypothetical example of a 10 m long by 1.4 m diameter rock-socketed pile, embedded in siltstone with an unconfined compression strength of about 5 MPa. Peak and residual shaft friction of 1000 kPa and 600 kPa have been adopted, with a displacement of 30 mm required to degrade from peak to residual (wres = 30 mm, = 1). The shear modulus of the siltstone was taken as G = 700 MPa, with the yield parameter as 0.5. The base capacity was taken as 25 MPa, mobilised at a displacement of 140 mm (10 % of the pile diameter). The ‘ideal’ capacity of the pile may be calculated from the peak shaft friction and base capacity as 44.0 MN (shaft) plus 38.5 MN (base), giving a total of 82.5 MN. In practice, strain-softening of the shaft response will occur before the base resistance has built up, and the actual capacity should be based on the residual shaft friction of 600 kPa. This leads to a lower capacity of 26.4 MN plus 38.5 MN, which equals 64.9 MN. The actual response of the rock-socket is shown in Figure 14. The load-displacement and shear stress-displacement responses are plotted at depths of 0, 4.8 and 9.8 m. The bottom curve effectively gives the response at the pile base. It may be seen that a peak load of 44.2 MN is reached at a displacement of about 8 mm, after which the load decreases, due to strain-softening of the shaft response. Subsequently, the load increases again towards the final capacity of 64.9 MN. The form of the response emphasises the need to choose working loads carefully. Clearly, there is potential for large differential settlements between piles if the working load is taken in the region of 40 MN. 5.4 Offshore Drilled and Grouted Pile The final example is an offshore drilled and grouted pile embedded in calcarenite which shows significant strain-softening and degradation under cyclic loading. The pile dimensions are: Embedded length = 70 m; Outer radius = 1.0 m; Wall thickness = 60 mm; Young’s modulus of steel = 210 GPa. The soil properties for the shaft resistance are tabulated below. The base resistance of the pile has been ignored in this example. 12 February 2003 RATZ Manual Version 4-2 M.F. Randolph Shear modulus of soil = 500 MPa; Parameter = 4; Yield point is at = 0; Peak shaft friction = 400 kPa; Residual shaft friction = 40 kPa; Displacement to residual = 1000 mm; Strain-softening parameter = 0.7 Cyclic residual shaft friction = 4 kPa. 5.5 Monotonic Loading Figure 15 shows the pile response to monotonic loading, with different amounts of creep. Although the theoretical capacity (assuming a rigid pile) is 176 MN, the actual capacity under static loading (with no creep) is only 151 MN, due to effects of progressive failure. As the value of the creep parameter, , is increased (which may be viewed as equivalent to decreasing the loading rate for a constant degree of creep), the peak load decreases. Thus peak loads of 150 MN, 132 MN and 87 MN are achieved for values of 0.001, 0.01 and 0.1 respectively. Within the framework of assumptions in the program, the theoretical lower limit to the pile capacity corresponds to the residual shaft friction of 40 kPa, which would give a capacity of 18 MN. The phenomenon of progressive failure of the pile may be seen clearly in Figure 16, which shows the axial pile load and shear stress mobilisation at 3 depths down the pile (for zero creep). At peak load, the shaft friction in the upper pile elements has degraded to about 70 % of the maximum value. 5.6 Cyclic Loading The pile response under a typical design storm sequence has been analysed with both the original and the modified yield algorithm. The yield parameter has been reset to 0.1 for the modified yield algorithm, since a value of zero would imply continuous degradation under cyclic loading of any amplitude, which is considered unrealistic. The results of the analyses are shown in Figures 17 and 18. The storm starts with large numbers of load cycles of low amplitude, builds up to the maximum design loading with a single cycle of 62.5 MN ± 23.5 MN, and then decays symmetrically. In the example, about 1000 load cycles have been applied in 25 stages, although in a real design it may be necessary to analyse a somewhat greater number. As an indication of the speed of analysis, the example took about 2 minutes to compute on a 1 GHz (Pentium 3) microcomputer. Some interesting effects may be observed from Figures 17 and 18: • during the early cyclic loading, the mean shear stress in the upper part of the pile 13 February 2003 RATZ Manual Version 4-2 M.F. Randolph reduces, tending towards symmetric two-way loading, even though the overall cyclic loading is one-way; • the element shown (at a depth of 9 m) reaches failure on the first cycle and then again under the peak design loading; the effect of the very low cyclic residual shear stress may be noted, from the exaggerated ‘S’ shape to the response; • little further displacement of the pile occurs during the second half of the storm sequence, although the upper soil elements continue to degrade; • the final ‘post-cyclic’ capacity of the pile is reduced to 136 MN using the original yield algorithm, and 132 MN using the modified algorithm, compared with the original monotonic capacity of 151 MN; • comparing Figures 17 and 18, it is clear that significantly greater displacements accumulate with the modified yield algorithm, although the decrease in post-cyclic capacity is only small. The effect of cyclic loading on pile capacity will vary with the relative stiffness of the pile. Piles that are less compressible will show less concentration of load transfer at the head of the pile. As such, cyclic loading at typical design levels will tend to cause less degradation. This has been discussed by Randolph (1983). In the example shown here, for a very compressible pile, the upper five elements all degrade to residual shaft friction. However, the final capacity shows only moderate reduction, since the effects of progressive failure during the final monotonic loading are reduced. The program outputs the status of each of the soil elements at the end of each loading stage. This information may be used to follow the pattern of degradation through a cyclic loading sequence. Figure 19 shows results from the analysis using the modified yield algorithm. The upper part of the Figure shows the distribution of current shaft friction down the pile at two stages during the storm sequence, while the lower part shows the variation of pile capacity with each stage of loading. The ‘current’ capacity after each stage has been estimated by linear interpolation, from the monotonic capacities before and after cyclic loading, and the integrated shaft friction (the ideal capacity) at the end of each loading stage. 14 February 2003 RATZ Manual Version 4-2 M.F. Randolph PART B - PROGRAM DOCUMENTATION 6 6.1 STRUCTURE OF PROGRAM Overview of Program Operation The program comprises an Excel workbook, which provides the framework for data input and output of results. The workbook calls a dynamic link library (DLL) compiled from Fortran code, which performs the main numerical computations. In order for Excel to find the DLL file, the directory must be set to that containing it. The simplest way of achieving this is to keep a copy of this file in the same directory as the RATZ workbook, and to open the workbook file by going through the full ‘File: Open’ procedure (not just double-clicking on the file name). This resets the directory in which Excel will look for the DLL to the current one. The program performs an analysis of the axial response of a pile under monotonic or cyclic loading. The pile, which may be made up in a number of segments of varying properties, is treated as an elastic bar, able to deform one-dimensionally along its length. The pile is partly (or wholly) embedded in soil which may consist of a number of different layers, with homogeneous or heterogeneous properties in each layer. The loading may be specified in a number of different stages, with each stage being defined in terms of monotonic or cyclic variation of load or displacement. The main operations involved in running the program, with optional items shown in square brackets, are as follows: (1) Input or edit data (see ‘Data’ worksheet) (2) Run analysis (press ‘Run’ button, or type Ctrl-r). (3) View graphical output The main output from the program is graphical. However, tabulated output is also provided, reflecting the input data and summarizing results at intervals defined by the user. In addition, output is provided giving the status of every soil element at the end of each loading stage (see‘Degradation’ worksheet). This latter feature is primarily used for detailed analysis of degradation under cyclic loading. In addition to the main input worksheet (‘Data’), two additional worksheets for data input are provided. The first of these (‘Xtradata’) gives information on (a) residual loads in the pile; (b) downdrag soil movement; and (c) thermal strains. The second additional data worksheet (‘Testdata’) allows tabulation of the results of a load test, so that measured and computed response can be compared. 15 February 2003 7 7.1 RATZ Manual Version 4-2 M.F. Randolph PROGRAM INPUT Data Input Data input is on the worksheets ‘Data’, ‘Xtradata’ and ‘Testdata’ must be confined to the areas with yellow background. Standard Excel techniques, including formulae may be used. It is important within the array areas for pile or soil properties, or for the loading stages, not to leave any gaps between successive entries. The system of units assumed by the program comprises forces in kN and lengths in m (and hence modulus or shear stress in kPa), although in principle other consistent sets of units could be used. 7.2 Data Items 7.2.1 Problem Title The title may comprise any sequence of alphanumeric characters or punctuation marks. 7.2.2 Pile Segement Properties The pile is considered to be made up of a number of segments (maximum: 5). The segment properties are the segment length, the outer and inner pile diameters and the Young’s modulus of the pile material. For non-circular, or H-section piles, the ‘diameters’ should be chosen so as to match both the gross cross-sectional area of the pile (for example, the encompassing rectangle for an H-section pile) and also the actual cross-sectional area of the pile section. 7.2.3 Pile Details and Base Response The pile penetration into the soil must be specified, and also the number of elements that the embedded part of the pile is to be divided into. Typically, pile elements of one or two diameters in length should be sufficient for an accurate analysis. The pile base may be assigned a diameter, db, that is different from the shaft diameter. The base response of the soil is taken as parabolic. The user specifies the ultimate base pressure, qbf and the pile displacement, wbf, at which this is mobilised. The properties of a parabola are such that the initial gradient of the pile response will be 2qbf/wbf, which implies a shear modulus for the soil in the region of the pile base of G b 0.25(1 )d b q bf w bf (27) The user should check whether this represents a reasonable modulus, consistent with values expected at the base of the pile. The base response may be switched to the hyperbolic model by specifying the pile 16 February 2003 RATZ Manual Version 4-2 M.F. Randolph displacement for full mobilisation of end-bearing pressure in ‘mm’ rather than ‘m’. (There are few occasions where this could lead to any confusion!) Essentially, the initial tangent shear modulus may be taken from Equation 27, and the secant shear modulus is then assumed to decrease linearly to zero as qb approaches qbf. 7.2.4 Soil Details The global load transfer parameters and creep parameters have been discussed already, in Section 2 of this manual. 7.2.5 Miscellaneous Parameters The damping factor should be set in the range 0.1 - 0.3 in order to ensure numerical stability. The strain-softening switch is set to 0 for degradation from peak to residual shaft friction that is linear with displacement, or 1 for the exponential function given in Equation 4. 7.2.6 Soil Profile The soil properties are specified in terms of a number of different layers, with the load transfer parameters for each soil element down the pile obtained by interpolation within each layer. A fully homogeneous deposit is indicated by specifying only the first line of data in this input area. Up to seven soil properties must be specified for each layer. The key parameters that determine the shape of the load transfer curve are: the shear modulus, G, the yield threshold, (xi), the peak shaft friction, p, the residual shaft friction ratio, r/p, and the displacement to residual, wres. In addition, for non-linear strain-softening, a shape factor, (eta), is specified that determines the exponential shape. Finally, for cyclic loading only, the cyclic residual shaft friction ratio,cr/p must be specified (see section 2). The absolute value of the yield ratio, , may be taken anywhere in the range 0 to 1 (inclusive), depending on what degree of non-linearity is required before peak shaft friction is reached. This parameter also controls the cyclic response of the soil, through the yield algorithms described in section 2.2. In order to switch to the hyperbolic shaft model, the parameter should be set to a number between 1 and 2 (any number higher than 2 will be reduced to 2). The program then assumes a value for Rf of Rf = - 1 (26) Note that the hyperbolic model may only be used for monotonic loading. Values of peak and residual shaft friction should be assessed from normal design 17 February 2003 RATZ Manual Version 4-2 M.F. Randolph procedures, while the two parameters that control the strain softening part of the curve, wres and should be adjusted to give the required shape (see Equation 4). Generally, a value of wres in the range 30 - 100 mm appears to be appropriate for monotonic loading, together with a value for of around unity. For cyclic loading, where the residual shaft friction can be very low, it may be appropriate to adopt a much larger value of wres (see example in section 5.4). 7.2.7 Loading Details Multiple stage loading may be specified (up to 100 stages), with each stage either displacement or load controlled, and either monotonic or cyclic. In practice, displacement control is more useful for monotonic loading, while load control is generally more appropriate for cyclic loading. The tabulated output from each load stage may be either slim-line (pile head and ground surface response only) or full (tabulated load, displacement and shear stress down the embedded section of the pile). When choosing the number of output stages, it should be remembered that graphical output occurs at ten times the frequency of tabular output. Thus, for monotonic loading, 10 - 20 output stages should be sufficient to give good definition to the plotted load-displacement curve. For cyclic loading it will generally be sufficient to plot 10 points per cycle (so only one tabular output per cycle). The size of load or displacement increment should generally be chosen so as to give displacement increments that are no greater than 0.01 % of the pile diameter. For very stiff piles, or where rapid strain softening occurs, increments as low as 0.001 % of the diameter may be required in order to retain stability. Under monotonic loading, the overall number of increments is determined by the required total pile displacement (or final load). Under cyclic loading, the maximum load (or displacement) and the minimum load (or displacement) are specified, with the number of increments per cycle being chosen according to the incremental criterion given above. In general, 200 - 500 load increments will be needed per cycle. 7.2.8 Residual Loads A profile of residual loads may be specified in the ‘Xtradata’ worksheet. The loads are specified as a series of ordered pairs (one pair per line) giving (a) the depth (in m), and (b) the magnitude of the residual force (in kN). The program interpolates between each given depth in order to establish the residual force at any intermediate depth. Thus, a simple triangular distribution of residual force in a pile of embedded length 20 m, could be specified as: 18 February 2003 0 0 20 1500 RATZ Manual Version 4-2 M.F. Randolph where 1500 (kN) is the magnitude of the residual force at the pile base. The program checks that there is sufficient shaft capacity to equilibrate the requested residual force, and if not adjusts the residual force accordingly. 7.2.9 Downdrag In addition to specifying ‘monotonic’ or ‘cyclic’ control, a third option is provided which is ‘downdrag’. If this option is selected for a particular loading stage (but only once in any analysis), the profile of soil movements specified on the ‘Xtradata’ worksheet is applied over the specified number of increments. The form of the soil movement profile is a series of ordered pairs (one pair per line) giving (a) the depth (in m), and (b) the magnitude of the soil movement (in m, downwards positive). The program interpolates between each given depth in order to establish the soil movement at any intermediate depth. Thus, a simple triangular distribution of soil movement from 0.1 m at the soil surface to zero at a depth of 25 m, could be specified by: 0 0.1 25 0 7.2.10 Thermal Strains Analysis of the effects of thermal strain may be achieved by specifying a profile of thermal strains in the ‘Xtradata’ worksheet. Ordered pairs of (a) depth (in m) and (b) thermal strain (contractive strain taken as positive) are specified, with the depth measured from the top of the pile (independent of the penetration of the pile). The program then interpolates between the given depths in order to evaluate the strain at each pile element, which are then listed in the output file (where the stated strain corresponds to the average strain in the section of pile immediately above each node). Note that only the embedded section of the pile is divided into elements. Hence, for a free-standing length of pile of length X, embedded length of , the thermal strain for element 1 is between the top of the pile and a distance X + 0.5/N from the top. The thermal strain for the ith element is between X + (i-1.5)/N and X + (i-0.5)/N from the top of the pile. A thermal strain loading stage is activated automatically if zero loads or displacement (under the monotonic section if loading switch set to monotonic, or cyclic section if loading switch set to cyclic) are specified. The program then picks up the thermal strains from the ‘Xtradata’ worksheet and applies them over the specified number of increments. During a thermal strain loading stage, specifying ‘Load’ control allows free movement 19 February 2003 RATZ Manual Version 4-2 M.F. Randolph of the pile head, under constant load, while specifying ‘Displacement’ control fixes the top of the pile (zero change in displacement) and lead to a build up of pile head load due to thermal strains. 7.3 Measured Load Test Data Data from a pile load test may be entered on the ‘Testdata’ worksheet, and plotted to allow comparison with the computed pile response, facilitating the back-analysis of pile load tests. The form of the data is a series of ordered pairs giving (a) the displacement (in m) and (b) the load (in kN). 8 8.1 PROGRAM OUTPUT Printed Output Tabulated output is provided on the ‘Output’ worksheet, and is reasonably self explanatory, with the exception of the two profiles of load given in the full output. The profile (P) is calculated from the internal pile strains, with the component of force due to damping in the soil subtracted. The profile (S) is calculated from the cumulative shear stress acting on the pile. Agreement between the two profiles of load is an indication of the accuracy and relative stability of the numerical solution. In general, the two profiles should not differ by more than 5 % near the pile head, and up to 10 % near the pile base (where the loads will be smaller). In addition to the main output, the status of each soil element at the end of each loading stage is provided in the ‘Degradation’ worksheet. This information is extremely useful in following the pattern of soil degradation during a cyclic loading sequence (see section 5.4, and Figure 19). The tabulated data may be presented graphically by modifying one of the plots in the ‘Userarea’ worksheet. 8.2 Graphical Output The main output from the program is graphical, and the user may adjust the form of the graphs to suit their preferences. All available data are tabulated in the ‘Plotoutput’ worksheet, with pile loads, displacements and shaft friction specified at the pile head and up to 5 levels down the pile. 20 February 2003 9 RATZ Manual Version 4-2 M.F. Randolph REFERENCES 1. Baguelin F. and Frank R.A. (1979), Theoretical studies of piles using the finite element method, Proc. Int. Conf. on Num. Methods in Offshore Piling, ICE, London, 83-92. 2. Cundall P.A. and Strack O.D.L. (1979), A discrete numerical model for granular assemblies, Geotechnique, 29(1):47-66. 3. Focht J.A and Koch K.J. (1973), Rational analysis of the lateral performance of offshore pile groups, Proc. 5th Offshore Technology Conf., Houston, 2, Paper OTC 1896, 701-708. 4. Kraft L.M., Ray R.P. and Kagawa T. (1981), Theoretical t-z curves, J. of Geot. Engng Div., ASCE, 107(11):1543-1562. 5. Poulos H.G. (1988), Cyclic stability diagram for axially loaded piles, J. Geot. Eng. Div., ASCE, 114(8):877-895. 6. Randolph M.F. (1977), A Theoretical Study of the Performance of Piles, PhD Thesis, University of Cambridge. 7. Randolph M.F. and Wroth C.P. (1978), Analysis of deformation of vertically loaded piles, J. of Geot. Eng. Div., ASCE, 104(GT12):1465-1488. 8. Randolph M.F. (1979), Discussion in Conf. on Numerical Methods in Offshore Piling, ICE, London, 197. 9. Randolph M. F. (1983), Design considerations for offshore piles, Proc. Conf. on Geotechnical Practice in Offshore Engineering, Austin, 422-439. 10. Randolph M.F. (1994), Design methods for pile groups and piled rafts, State-ofthe-art Lecture, Proc. 13th Int. Conf. on Soil Mech. and Found. Eng., New Delhi, 5:61-82. 11. Singh A. and Mitchell J.K. (1968), General stress-strain-time function for soils, J. Soil Mech. and Found. Eng. Div., ASCE, 94(SM1):21-46. 21 February 2003 RATZ Manual Version 4-2 M.F. Randolph TABLE 1 LOAD TRANSFER PARAMETERS ADOPTED FOR FATIGUE CURVES Original Algorithm Modified Algorithm Shear modulus, G (MPa) 200 or 1000 200 or 1000 Load transfer parameter, 4 4 Yield parameter, 0 0.10 Residual shaft friction (%) 18 10 1000 1500 0.7 0.65 Cyclic residual friction (%) 1 1 Fatigue threshold (%) 33 10 Radius of pile, ro (m) 1.15 1.15 Disp. to residual (mm) Shape parameter, 22 February 2003 RATZ Manual Version 4-2 M.F. Randolph 10 FIGURE TITLES Figure 1 Idealisation of pile in load transfer analysis Figure 2 Details of load transfer curve Figure 3 Example load transfer curves Figure 4 Goodman diagram showing linear projection to allow for stress bias Figure 5 Cyclic stability diagram Figure 6 Fatigue curves for original yield algorithm (a) Shear modulus of G = 200 MPa (b) Shear modulus of G = 1000 MPa Figure 7 Fatigue curves for modified yield algorithm (a) Shear modulus of G = 200 MPa (b) Shear modulus of G = 1000 MPa Figure 8 Modelling low cyclic residual shaft friction Figure 9 Modelling effects of creep (a) Effects of soil creep (b) Displacement of load transfer curve due to creep Figure 10 Modelling group effects by factoring load transfer curve Figure 11 Response of pile base Figure 12 Measured and simulated response of bored pile (a) Small displacements (b) Overall response Figure 13 Effect of residual stresses on pile head and base response Figure 14 Rock-socket example: 10 m long by 1.4 m diameter Figure 15 Load-displacement response of offshore pile with varying amounts of creep Figure 16 Offshore drilled and grouted pile: 70 m long by 2 m diameter Figure 17 Offshore drilled and grouted pile: cyclic loading, original algorithm Figure 18 Offshore drilled and grouted pile: cyclic loading, modified algorithm Figure 19 Shaft friction and pile capacity during cyclic loading (a) Profiles of failure shaft friction at different stages (b) Progress of degradation through storm 23 February 2003 RATZ Manual Version 4-2 M.F. Randolph FIGURES P P Lumped mass, m Spring stiffness, k w Load transfer curves along pile shaft Load transfer at pile base (a) Actual pile (b) Idealisation of pile Figure 1 Idealisation of pile in load transfer analysis 24 February 2003 RATZ Manual Version 4-2 M.F. Randolph Shear stress, o wres B p C r parabolic D f p linear A Normalised displacement w/ro wo yield f E Figure 2 Details of load transfer curve 25 Displacement, w Proportion of peak shaft friction / p February 2003 RATZ Manual Version 4-2 M.F. Randolph 1 0.9 0.8 0.7 0.6 0.5 Hyperbolic: Rf = 0.8 0.4 0.3 Hyperbolic: Rf = 0.95 0.2 RATZ Parabola: xi = 0 0.1 API Guidelines 0 0 0.2 0.4 0.6 0.8 Displacement/pile diameter (%) 1 Proportion of peak shaft friction / p (a) Pre-peak response 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 RATZ: xi = 0; eta = 1.3 0.2 RATZ: xi = 0; eta = 1 0.1 RATZ: xi = 0; eta = 0.7 0 0 1 2 3 4 5 Displacement/pile diameter (%) (b) Post-peak response Figure 3 Example load transfer curves 26 6 February 2003 RATZ Manual Version 4-2 M.F. Randolph max /p and min/p 1 Pr n tio ec oj ine l 1-way loading 0.33 2-way loading max min 0.5 cyclic 0.25 mean 0 1 min/p -0.33 Figure 4 Goodman diagram showing linear projection to allow for stress bias 27 February 2003 RATZ Manual Version 4-2 Limiting condition Normal algorithm: xi = 0 Normal algorithm: xi = 0.333 Revised algorithm: xi = 0 Revised algorithm: xi = 0.333 1 0.9 0.8 0.7 cyclic / p M.F. Randolph 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 / me an p Figure 5 Cyclic stability diagram 28 0.8 1 February 2003 RATZ Manual Version 4-2 M.F. Randolph 1 1-way Triaxial 1-way CNS 2-way CNS Resonant Column Samples not failed RATZ: 2-way cyclic f 0.8 0.6 0.4 0.2 RATZ: 1-way 0 0 1 2 3 4 5 6 7 8 7 8 Log(No. of cycles) (a) Shear modulus of G = 200 MPa 1 1-way Triaxial 1-way CNS 2-way CNS Resonant Column Samples not failed RATZ: 2-way cyclic / f 0.8 0.6 0.4 0.2 RATZ: 1-way 0 0 1 2 3 4 5 6 Log(No. of cycles) (b) Shear modulus of G = 1000 MPa Figure 6 Fatigue curves for original yield algorithm 29 February 2003 RATZ Manual Version 4-2 M.F. Randolph 1 1-way Triaxial 1-way CNS 2-way CNS Resonant Column Samples not failed RATZ: 2-way cyclic / f 0.8 0.6 0.4 0.2 RATZ: 1-way 0 0 1 2 3 4 5 6 7 8 7 8 Log(No. of cycles) (a) Shear modulus of G = 200 MPa 1 1-way Triaxial 1-way CNS 2-way CNS Resonant Column Samples not failed RATZ: 2-way cyclic / f 0.8 0.6 0.4 0.2 RATZ: 1-way 0 0 1 2 3 4 5 6 Log(No. of cycles) (b) Shear modulus of G = 1000 MPa Figure 7 Fatigue curves for modified yield algorithm 30 February 2003 RATZ Manual Version 4-2 Figure 8 Modelling low cyclic residual shaft friction 31 M.F. Randolph February 2003 RATZ Manual Version 4-2 M.F. Randolph Shear stress, o without creep with creep Displacement, w (a) Effect of soil creep Shear stress, o displaced curve 1 2 w Displacement, w (b) Displacement of load transfer curve due to creep Figure 9 Modelling effects of creep 32 February 2003 RATZ Manual Version 4-2 A Shear stress O B C M.F. Randolph D Group pile Single pile OC = R s OA CD = AB Displacement Figure 10 Modelling group effects by factoring load transfer curve Base pressure, qb C qbf D G A 2 B E F O wbf Base displacement, wb Figure 11 Response of pile base 33 February 2003 RATZ Manual Version 4-2 M.F. Randolph 6 Load (MN) 5 4 3 2 Measured RAT Z 1 0 0 2 4 6 8 10 80 100 Displacement (mm) (a) Small displacements 8 7 Load (MN) 6 5 4 3 Measured 2 RAT Z 1 0 0 20 40 60 Displacement (mm) (b) Overall response Figure 12 Measured and simulated response of bored pile 34 February 2003 RATZ Manual Version 4-2 M.F. Randolph 4000 Pile head Residual stresses Load (kN) 3000 No residual stresses Residual stresses 2000 1000 Pile base No residual stresses 0 0 20 40 60 80 100 120 Displacement (mm) Figure 13 Effect of residual stresses on pile head and base response 35 February 2003 RATZ Manual Version 4-2 70000 M.F. Randolph Pile head z = 4.8 m 60000 Pile head load (kN) z = 9.8 m 50000 40000 30000 20000 10000 0 0 0.02 0.04 0.06 0.08 0.1 Pile head displacement (m) (a) Load-displacement response Mobilised shaft friction (kPa) 1200 1000 z = 4.8 m z = 9.8 m 800 600 400 200 0 0 0.02 0.04 0.06 0.08 0.1 Pile head displacement (m) (b) Load transfer response Figure 14 Rock-socket example: 10 m long by 1.4 m diameter 36 February 2003 RATZ Manual Version 4-2 M.F. Randolph 160 140 Pile head load (MN) 120 100 80 60 Beta = 0 Beta = 0.001 40 Beta = 0.01 Beta = 0.1 20 0 0 20 40 60 80 Pile head displacement (mm) Figure 15 Load-displacement response of offshore pile with varying amounts of creep 37 100 February 2003 RATZ Manual Version 4-2 Pile head z = 9.5 m 160000 140000 Pile head load (kN) M.F. Randolph z = 29.5 m z = 49.5 m 120000 100000 80000 60000 40000 20000 0 0 0.02 0.04 0.06 0.08 0.1 Pile head displacement (m) (a) Load-displacement response Mobilised shaft friction (kPa) 450 400 350 300 250 z = 9.5 m 200 z = 29.5 m z = 49.5 m 150 100 50 0 0 0.02 0.04 0.06 0.08 0.1 Pile head displacement (m) (b) Load transfer response Figure 16 Offshore drilled and grouted pile: 70 m long by 2 m diameter 38 February 2003 RATZ Manual Version 4-2 Pile head z = 9.5 m 160000 140000 Pile head load (kN) M.F. Randolph z = 29.5 m z = 49.5 m 120000 100000 80000 60000 40000 20000 0 0 0.02 0.04 0.06 0.08 0.1 Pile head displacement (m) (a) Load-displacement response Mobilised shaft friction (kPa) 500 400 300 200 100 0 -100 0 0.02 0.04 0.06 -200 0.08 0.1 z = 9.5 m z = 29.5 m z = 49.5 m -300 Pile head displacement (m) (b) Load transfer response Figure 17 Offshore drilled and grouted pile: cyclic loading, original algorithm 39 February 2003 RATZ Manual Version 4-2 140000 Pile head z = 9.5 m 120000 Pile head load (kN) M.F. Randolph z = 29.5 m z = 49.5 m 100000 80000 60000 40000 20000 0 0 0.02 0.04 0.06 0.08 0.1 Pile head displacement (m) (a) Load-displacement response Mobilised shaft friction (kPa) 500 400 300 200 100 0 -100 0 0.02 0.04 0.06 -200 0.08 0.1 z = 9.5 m z = 29.5 m z = 49.5 m -300 Pile head displacement (m) (b) Load transfer response Figure 18 Offshore drilled and grouted pile: cyclic loading, modified algorithm 40 February 2003 RATZ Manual Version 4-2 M.F. Randolph Shaft friction (kPa) 0 100 200 300 400 500 0 10 Depth (m) 20 30 40 50 Peak Max. storm 60 End storm Residual 70 (a) Profiles of failure shaft friction at different stages Ideal and actual pile capacity (MN) 180 170 160 Ideal capacity (integrated failure shaft friction) 150 140 130 Actual capacity (allowing for strain softening) 120 0 5 10 15 20 Loading stage 25 30 (b) Progress of degradation through storm Figure 19 Shaft friction and pile capacity during cyclic loading 41