LABORATORYTESTPREDICTIONSOFTHECYCLICAXIAL RESISTANCEOFAPILEDRIVENINNORTHSEASOILS MJ Rattley Fugro GB Marine Limited Wallingford UK L Costa Scottish Power Renewables London UK RJ Jardine Imperial College London UK WCleverly Offshore Wind Consultants Limited London UK Abstract including generalised interaction diagrams but they may not be able to address site-specific aspects of soil behaviour layering pile geometries or cyclic loads. Applying ‘one size fits all' cyclic loading factors is likely to be either inappropriate or uneconomic in wind farm applications where turbines may be spaced kilometres apart with geology and water depth varying between positions. This paper describes how carefully controlled farm site in the North Sea. Interpretation of cyclic laboratory tests on sand and clay within an effective stress framework allowed the potential cyclic degradation of pile shaft friction to be assessed. The practical ap- proach outlined is shown to provide a feasible route to assessing turbine-specific cyclic factors for offshore pile design. 1.Introduction the degradation of pile resistance. A range of meth- 1.1Background ods for estimating this effect is available in the liter- Assessing how cyclic loading affects offshore jacket ature but no standardised application method exists foundation piles may be more important than previ- in practice. One widely adopted approach applied ously appreciated particularly for structures where for screening purposes involves the use of simple in- the cyclic axial pile load ponent is significantly teraction diagrams which relate the cyclic capacity higher than the static (average) load (Jardine et al. of the pile to the static capacity. Figure 1 shows one 2012). International standards such as API (2011)such example as presented by Jardine and Standing remend allowing for cyclic effects but this is of- (2012) for steel tubular piles driven in dense sand at ten assumed to be implicit in the static methodolo- Dunkirk N France. EA-Pfihle (2014) present a gies and factors of safety applied under storm load- similar cyclic design approach based on the interac ing conditions. The recent drive for increased tion diagram proposed by Kirsch et al. (2011) which foundation design efficiency in large offshore wind is applied routinely in German OWF projects. farm (OWF) developments has prompted a rise in practitioners’ use of modern design methods such as Interaction diagrams afford simple screening tools the Imperial College Pile (ICP) method (Jardine et that rely on patibility between the soil condi- al. 2005). Such methods provide a more robust ti tions for which the charts were prepared and those at physically reasonable approach to static axial ca- the design location. Precise matches are unlikely at pacity. However cyclic loading needs also to be most offshore sites. Key site-specific characteris- considered explicitly by the designer. Jardine et al. tics of cyclic soil behaviour layering pile geome- (2005) suggest how this may be approached; this tries or cyclic loads are therefore likely to be over- paper addresses the details of practical application looked. Applying single ‘one size fits all’ interaction for a major OWF project. charts is likely to be either inappropriate or uneco- nomic when initial screening indicates a need for de- 1.2 Design Approaches tailed cyclic design especially with offshore wind In ultimate limit state (ULS) axial pile design the eturbine jackets that may be spaced kilometres apart
tions. Approaches for more advanced local soil-pile mation were sampled over the expected foundation analysis are outlined by Jardine et al. (2012) and ad- depths at multiple locations. vocated for pile design in carbonate soils by Erbrich et al. (2010). Such analysis is most pertinent when 1.4 Soil Properties concerns arise with regard to i) potentially brittle Tables 1 and 2 summarise some key geotechnical soil response and progressive pile failure and i) parameters for the soil units considered. sites with highly variable or strongly contrasting. soil layering. Table 1: Summary of key soil properties for sand soils Formation [%]a Dso [mm] ] qe [MPa] Yarmouth Roads 90-110 0.17-0.20* 32 20-60 Smith’s Knoll 50-70 0.12 28 10-30 0.8 Westkapelle 80-90 0.14 31 40-70 ata *Range includes formation sublayers Table 2: Summary of key soil properties for clay soils Formation PI[-] YSR dl°] qe Westkapelle [MPa] 15-20 2-4 2-4 20 4-6 Westkapelle 25-30 13 3-5 0.2- 2.Cyclic Laboratory Testing 2.1 TestingProgramme Cyclic degradation laws can be formulated using cy- clic triaxial (CTXL) (Aghakouchak et al. 2015) cy- Oaverago/max static 0.4 0.6 clic constant normal stiffness (CNS) tests (Erbrich et al. 2011) or cyclic direct simple shear (CSS) labora- Figure I: Interaction diagram from cyclic testing of piles driv- tory tests.While simple shear tests offer less en in dense sand at Dunkirk (Jardine and Standing 2012) prehensive measurements than triaxial tests of the developed stress and strain conditions they better One route to developing local shaft degradation laws match the pile kinematic boundary conditions. is to conduct cyclic laboratory testing on site- specific soils. This approach provides data on the el- Constant volume CSS tests match the constant vol- ement level response of the soils which can be used ume conditions expected during undrained cycling to analyse either global pile degradation or local in clays and provide notionally infinite normal stiff- soil-pile response under cyclic loading. The bounda- ness for drained sand cases.However the constant ry conditions applied in cyclic laboratory testing af- volume condition is applied irrespective of in situ fect the results and this must be considered carefully. drainage conditions load frequency and variations The laboratory testing approach adopted recently for in cyclic load level. Arguably CNS testing should the East Anglia One (EAONE) OWF is reported in provide more realistic measurements of changes in this paper. normal effective stress during drained shearing al- lowing also for soil response outside of the direct 1.3 Site Surmary shear zone along the pile. However such tests re- EAONE which is sited in the southern North Sea quire an estimation of the mass horizontal stiffness involves 102 7MW wind turbine generators (WTG) of the soil a parameter that is pressure-dependent whose substructures consist of 3-legged jackets varies non-linearly with distance from the pile and founded on driven open-ended piles; water depths depends on the pile diameter. CNS values are not range from 38 m to 50 m. The site geology is typical easily obtained or estimated but are likely tobevery of the region consisting of Holocene sediments high in dense North Sea sands. overlying Pleistocene strata that reached their cur- rent state though a series of glaciations and prise Hollow Cylinder Apparatus offers the best equip- predominantly dense sands and high strength clays. ment for simple shear testing as it provides better The key formations considered are the Yarmouth boundary conditions and full descriptions of the Roads Smith’s Knoll and Westkapelle formations specimens’ stress and strain states. But routine CSS binations of which are present across the whole tests provide useful practical measurements in which site. Sand and clay units of the Westkapelle for- the constant volume condition is obtained by dynam-
ic control. With most mercial CSS apparatus od for equalised effective radial stresses at the pile the controls lead to very high but finite normal wall but with the parameter h which represents the stiffness. tip depth of the pile relative to the soil depth fixed to a value of 0.5 m. This approach resulted in equiv- Over 40 CSS tests were performed across 5 soil alent overconsolidation ratio (OCR) values between units as part of a prehensive cyclic laboratory 2 and 5 depending on specimen depth. The mini- testing programme for EAONE. A small number of CTXL tests were performed to provide a benchmark above the pile tip. To model conditions higher up the for the measured CSS response. Intact clay CSS shaft specimens were unloaded to a normal stress specimens were cut at 90° from core samples during equal to the effective radial stress acting on a pile trimming. For reconstituted sand it was not practi- wall after equalisation predicted according to the cable to create test specimens with grain alignments ICP method. Following Aghakouchak et al. (2015) representative of the soil state around the pile after driving. Instead the sand CSS tests were prepared stress (drained) cyclic pre-shearing at their final by moist paction to a representative 60% < D 1500^ 10 - >1500 Smith's Knoll Sand 12 60 81 - 152 0.10 - 0.20 Westkapelle Sand 14 85 324-1004 0.15 - 0.84 0.10 - 0.20 10->1500 Clay (LP) 3 558 0.32 - 0.80 0.32 >1500 Clay (HP) 4 702 0.45 - 1.10 0.23 67 - >1500^ LP = lower plasticiy HP = higher plasticity = cyclic loading stages
For each specimen tested the application of cy led methodology. The OWF considered adopted the ICP to changes in o. Slight increases were observed un- (Jardine et al. 2005) in which shaft resistance is der the lowest cyclic stress ratios and over a limited controlled by the local radial effective stress during number of initial cycles at higher cyclic stress ratios loading and the interface friction angle at failure. in very dense specimens. However cycling above a limiting Tcy value led to marked o reductions that Cyclic loading does not influence 8 (as defined in grew progressively as cyclic loading continued. Fig- Equation 1) which was determined by ring-shear in- ure 2 presents an example of the evolution of o for terface tests. The key was then to predict from the a high level cyclic test performed on a dense sand cyclic laboratory element tests the reductions in pile specimen that led to failure within 25 cycles. shaft radial effective stress expected during storm events. Merritt et al. (2012) and Jardine and Stand- 1.0 ing (2012) describe one such approach. 0.8 The proposed cyclic degradation model aims to first g06 capture the changes in o measured in the CSS tests. In the interpretation the o values on pletion of each constant amplitude load cycle are referred to D =95% the initial effective stress ono at the start of cyclic 0.2 /t=0.5 loading. Subtracting one from the other gives o 0.0 and then dividing by ono allows the proportional loss 10 20 30 40 Number of eydes 50 of effective stress to be quantified. The effective stress reductions grow with the number of applied Figure 2: Ilustration of normalised reduction in normal effec- cycles N for cyclic stress amplitudes above a given tive stress with number of cycles in α CSS test on dense sand 'zero degradation’ threshold. As noted earlier bene- ficial increases in normal effective stress may be ex- pected below this threshold although these are not 100 30 considered further here. CTXL Jardine and Standing (2012) propose that for cycling josno ssm Tav/rz the proportional loss of effective stress is sseo plied. Losses can then be expressed in terms of Tey/Trzf and N in the form of Equation 2: D =9% t4/t=9.6 (2) where the material coefficients A B and C define the Figure 3: Comparison of variarions in two-dimensional norma? rate of effective stress reduction under cyclic load- effective stress from CSS and CTXL tests on very dense sand ing. Jardine and Standing (2012) note that an equiva- lent log (N) expression may be more appropriate in some soils than the power law given in Equation 2. CTXL tests were performed to verify the effective In addition although it was adequate for the exam- stress changes measured in the larger number of ple case presented it may not be appropriate in some CSS tests. Comparison of the data from each test cases to neglect the average load ponent and a type was generally favourable in the terms plotted in more plex form of Equation 2 may be necessary. Figure 3 which pares a pair of CTXL and CSS tests performed under similar cyclic stress condi- 3.2 Pile-Soil Interface Stability tions. In the simple shear test arrangement the samples were prevented from shear at the top and base load- 3.Interpretation of Cyclic Test Data ing platen interfaces by constraining rings located 3.1 Interpretation Method along the loading platen. Noting that interface shear The cyclic design approach should be framed to be failure limits the shear stresses that can be sustained theoretically consistent with the static pile design by soil positioned at the pile-soil interface the cyclic
degradation model focuses primarily on the pre-V which is accumulated under Metastable conditions failure CSS test response. Pile shaft failure was de- over which significant numbers of cycles may be fined as the point at which the shear stress ratio (T/o'n) mobilised during the CSS test reached the pile capacity losses. expected limiting pile-soil interface value (tan) for each soil unit. 3.3Derivation of ModelParameters Cyclic model parameters were derived from the CSS Higher t/o' ratios cannot be mobilised close to the tests listed inTable 3 to quantify the effective stress pile-soil interface as soil-pile slip initiates when changes expected under a range of cyclic conditions. T/o' = tan8. Figure 4 illustrates this principle The following degradation parameter is helpful to where the state line corresponding to interface slip the discussion below: given by Equation 1 with the o for the pile inter- face being equated to in the simple shear test is S= 0'no (3) plotted in grey. 3.3.1Sand cases D =9% Figure 5 provides an illustration of the model pa- t/ty=05 rameter derivation for Westkapelle sand. Data ex- tracted at the end of load cycles 5 and 200 are plot- 0.2 tedagainst Tcy/Trzf、 covering tests that failed 00 (giving losses of effective stress exceeding 95% of ） and those where no failure was observed within the maximum number (1500) of cycles imposed. 02 Despite some scatter a reasonable linear correlation is observed for the two load cycle number cases that 0.0 2O 04 0.6 80 1.0 12 support the expression presented in Equation 4: Figure 4: llustration of test dato exclusion outside the limiting SN= (4) stress state where a and B are fitting parameters describing the The stress states lying to the left of the state line trend gradient and intercept respectively.Thefunc shown in Figure 4 could not be mobilised by piles in tion is subject to limits so that it predicts no change the field as local shaft failure would occur with each in effective stresses or shaft capacity under cyclic further cycle leading to shear stress transfer to great- loading with Tey/Trzf less than the threshold B. er depth on the shaft and hence extra pile displace ment. 1.0 0 Loa cydei=5 Jardine et al. (2012) employed cyclic stability dia 0.8 Loal edei-200 grams to illustrate the interactive effects on pile ca- Trmdi=s pacity of cyclic and mean loads and the number of 0.6 Tredi-200 cycles applied. Their diagrams indicate three distinct phases of pile-soil interface response: a cyclically 0.4 Stable region where there is no reduction of load capacity after a large number (say 1000) of cycles a 0.2 Metastable phase where some reduction of load ca- pacity occurs after a lower number N of cycles and 0.0 880 an Unstable zone where cyclic failure develops with- 0.1 0.2 0.3 0.5 0.8 50 in a relatively small number (say 100) of cycles. In the shaft degradation study reported in this paper all Stable cycles could be eliminated as causing no neg- Figure 5: Proportional loss of normal effective stress after 5 and 200 cycles for CSS tests on Westkapelle sand ative effect. At the other extreme globally Unstable cyclic conditions were avoided in design as re- Overall Equation 4 fits each set of data plotted well. mended by Jardine et al. (2012). The study therefore The experimental deviations fall within the range focused on modelling the reduction in pile capacity that could be related simply to the dynamic control
北海泥沙中驱动桩的循环轴向阻力的实验室试验预测.pdf
