【作者】 齐剑峰；

【导师】 栾茂田；

【作者基本信息】 大连理工大学 ， 岩土工程， 2007， 博士

【摘要】 近年来，海洋资源与海洋空间的开发和利用日益引起国内外的高度重视。然而由于海洋环境极端恶劣，而且海洋建筑物地基往往由软黏土、粉质黏土与可液化砂土等不良工程土壤组成，在对海域土质的循环特性没有深入了解的情况下进行工程建设与设计，将会造成海床与地基失稳，以致带来巨大的生命和财产损失。因此，为配合海洋工程设计与建设，必须进行黏土的循环特性试验研究。在深海和极浅滩海域，吸力式沉箱基础受到了海洋石油界的广泛关注，但是目前对于吸力式沉箱等新型海洋基础的工作机理及其稳定性分析尚未得到统一认识，因此为完善吸力式沉箱基础的设计理论体系与计算方法，亟需开展系统地试验研究、理论分析与数值计算等方面的综合研究。为此，论文针对饱和黏土试样，利用土工静力—动力液压—三轴扭转多功能剪切仪等土工试验设备，对UU试验条件下黏土的单调剪切试验特性、动剪切模量与阻尼比等动力特性与循环软化变形特性进行了较为深入而系统地对比分析与研究。在大量的单调与循环剪切试验基础上，建议了能够考虑应变速率效应的双曲线静力本构模型以及考虑静应力与循环应力联合作用的循环软化总变形拟静力计算模型，并在大型有限元软件ADINA平台上对建议的模型进行了数值实施与二次开发，考虑吸力式沉箱基础单调加载阶段地基变形速率的影响，对沉箱地基的竖向与水平向单调承载力以及循环软化变形进行了数值分析。论文的主要研究内容及所取得的研究成果包括如下几个方面：（1）饱和黏土制样技术的改进。通常获取原状土样花费较大，而且原状土样的饱和程度往往较低，土工试验时需要对试样进行真空处理使其饱和，经过如此处理的试样很难反映实际应力状态，而对于实际的非饱和土样目前无论在实验技术还是计算理论方面尚处于探索阶段。因此，采用实验室土样制备技术制备黏土试样，对制备的土样进行试验研究，是目前探索饱和土的变形与强度特性一种途径。通过对比泥浆加压固结制样法和真空预压制样法存在的优缺点，在吸取两种制样方法优点及其使用经验的基础上，对饱和黏土的制样设备进行了改进。改进黏土制样方法具有操作简便、制样时间短等优点，所制备试样具有成份均一、饱和程度高，且易于切削制备圆柱或空心圆柱试样等特点。（2）饱和黏土的单调剪切特性与非线性静力本构模型。在CU和UU条件下进行了单调三轴和单调扭剪试验，通过对试验结果的综合比较分析，着重探讨了应变速率和约束压力对黏土应力—应变关系特性、强度特性与孔隙水压力发展特性的影响。结果表明，当应变速率较低时，应力—应变关系一般经历弹性、屈服、应变强化等阶段，而当应变速率较高时，应力—应变关系在经历弹性、屈服、应变强化达到峰值之后，在较高应变条件下还会发生应变软化，当应变超过一定值之后，随着应变增加，不同应变速率下的偏应力差别减小；约束压力对应力—应变关系模式的影响较小，但在CU试验中，屈服极限和强度随约束压力的增加而增大，而在UU试验中，应力—应变关系与强度不依赖于约束压力；不排水强度随应变速率的增加而增大，通过采用参考应变速率及其相应的不排水强度进行归一化处理，在双对数坐标下，不排水强度与应变速率之间的经验关系近似地符合线性关系。考虑不固结不排水条件和应力—应变关系的应变速率效应，建议了饱和黏土的非线性本构模型，在达到破坏时的偏应力q_f之前，应力—应变关系基本上符合双曲线模式，而在超过q_f之后的应力—应变关系按理想塑性处理，所建议的模型含有4个待定常数，由UU三轴试验确定。（3）循环剪切变形特性与应变破坏标准。通过单向与双向耦合循环剪切试验对饱和黏土的不排水剪切变形特性与应变破坏标准及其孔隙水压力发展特性进行了研究，试验结果表明：初始剪应力对于黏土的循环应力—应变关系模式具有显著的影响，当初始剪应力较小或循环应力分量较大时，试样的变形以循环效应为主，而当初始剪应力较大或循环应力分量较小时，此时试样的变形呈现出明显的累积效应特征；在循环扭剪试验中，尽管没有施加任何轴向应力，但由于试样的循环扭动而产生轴向应变，并随循环应力增大，轴向偏差应变增长速率加大；在竖向—扭转耦合剪切试验中，尽管在耦合双向都施加对称的循环应力，但剪切向应变分量的循环效应比较明显，而轴向应变的累积效应比较显著，这与普通循环三轴或扭剪试验的结果不同。在单向循环剪切试验中，黏土试样的单向总应变同时包含不断累积的残余应变和循环应变两部分，通过对各种循环应力路径模式下的试验数据比较分析，建议了一个能够综合考虑轴向应变和剪切向应变，同时反映残余应变与循环应变特征的综合应变算式，该算式与单向总应变的表达式是一致的，试验数据表明这个综合应变具有普遍适用性和较好的稳定性。在不固结不排水条件下，无论对于循环扭剪试验还是竖向—扭转耦合试验，孔隙水压力总是在所施加的约束压力大小附近波动，而且随循环应力的增加孔隙水压力的波动范围有所加大，但并不具有固结不排水试验中孔隙水压力累积上升的特征。（4）大应变(＞10^-3)条件下动剪切模量与阻尼比等动力特性。单级与分级加载循环剪切试验表明：采用多个试样单级加载与采用一个试样分级加载所得的应力—应变滞回圈、初始骨干曲线、动剪切模量与阻尼比都较为一致，采用分级加载试验测定动剪切模量与阻尼比等动力特性是可行的；初始剪应力对分级加载的应力—应变滞回圈的发展模式和动剪切模量与阻尼比都有较为显著的影响，初始剪应力对动剪切模量与阻尼比的影响主要体现在动力参数的逐次变化过程上；对于某一给定的剪应变幅值，耦合循环应力中的轴向偏差应力对动剪切模量具有增大作用，动剪切模量发展与轴向偏差应力的大小关系密切，尤其是在扭转向循环剪应力较小的情况下，轴向偏差应力越大，动剪切模量的增大越显著，而当扭转向循环剪应力较大时，轴向偏差循环应力对动剪切模量的增大效用降低。（5）循环剪切强度特性及其简便确定方法。循环剪切强度与循环次数、初始剪应力之间关系曲线的确定，对于海洋平台设计与稳定性分析具有重要意义。循环剪切试验表明，对于给定的初始剪应力，循环剪切强度的大小是由循环软化效应和应变速率效应共同作用的结果。结合单调强度与应变速率之间的对应关系，提出了近似消除速率效应的循环强度归一化方法，通过归一化表明，不同破坏次数的循环强度与初始剪应力之间关系的数据点落在一个相当狭窄的区域内，初始剪应力与归一化循环剪切强度之间关系可用一个二次方程拟合表示。以此为基础，对循环强度简便确定方法进行了初步探讨，通过少量单调与循环剪切试验对待定参数确定之后，就可以估算不同循环次数的循环剪切强度，对于解决海洋地基稳定性分析中的试验工作量巨大等问题具有一定意义。（6）循环蠕变特性与循环软化总变形计算模型。对循环剪切试验结果的分析表明：循环软化总应变由初始静应力引起的应变和循环蠕变两部分组成。其中，循环蠕变主要受循环应力幅值、初始静应力大小及其加载速率的影响。根据循环蠕变随循环次数的变化特点，将循环次数与时间等效，仿照Singh-Mitchell蠕变模型参数确定方法，建议了循环蠕变的经验计算模式。进而对循环软化总应变的两个组成部分，即初始静应变与循环蠕变，分别结合非线性弹性理论与屈服面流变理论，建立了循环软化总变形计算模型。考虑到不排水条件，所建议计算模型仅含有7个试验参数，由单调与循环剪切试验可以确定。尽管该模型属于拟静力模型的范畴，但是模型不仅能够区分施加荷载的性质（单调荷载或循环荷载），而且能够分别计算单调加载与循环加载两个加载过程所产生的变形，并且考虑了单调荷载及其加载速率对总变形的影响。（7）循环软化总变形计算模型的有限元实施。循环软化总应变计算模型是基于单调与循环剪切试验建立的拟静力模型，较为准确地反映了静应力与循环应力共同作用所产生的循环蠕变等循环剪切特性。尽管大型有限元软件ADINA不仅具有强大的求解非线性问题的功能，而且具有优秀的前、后处理界面，但不具有反映这种循环特性的本构模型，需要对其进行二次开发。为此，论文对非线性弹性材料模型与循环蠕变材料模型的二次开发进行了探索，进而对循环软化总应变计算模型在大型有限元软件ADINA平台上进行了数值实施，并详细叙述了模型的二次开发过程及其计算方法与步骤。（8）吸力式沉箱地基的单调承载力与循环软化总变形的有限元数值分析。考虑单调加载阶段吸力式沉箱地基变形速率的影响，对沉箱地基的单调承载力与循环软化总变形进行了数值分析。计算结果表明：对于单个沉箱基础，黏土地基的竖向与水平向单调承载力随着地基变形速率的增加而增大；对于给定的单调荷载与循环荷载组合，竖向与水平向循环软化总变形随循环次数的增加而增长，循环软化总变形的逐次增长速率逐渐减小。在竖向单调与竖向循环荷载联合作用下，基础附近泥面处土体由于基础的下沉挤压而向上移动，地基变形模式主要为沉箱附近土体的整体沉陷；在竖向单调荷载与水平向循环荷载联合作用下，沉箱基础与土体出现开裂，沉箱基础顶部的水平向转动位移随循环次数的增加而增大。更多还原

【Abstract】 In recent years, the exploitation and utilization of marine source and space are highlynoticed by international society. However, due to the extremely bad marine environment andsoft and weak ground foundation composed of disadvangtageous engineering soils, such assoft clay, silty clay and fluid sand, the instability of the seabed and ground foundation beneathharbor construction will occur and induce the expense of huge wealth and life if oceanengineering is designed before the cyclic behavior of marine soil is not understood. Therefore,in order to cooperate with ocean engineering design, it is necessary to study cyclic behavior ofsoils by performing the geotechnical tests. Suction caisson foundation is paid attention to bymarine petroleum kingdom in the shallow maritime space and deep sea area, but because theworking mechanics and computational method for stability analysis of suction caissonfoundation are not clarified yet, so it urgently needs to carry on the research for theoryanalysis and computational method in order to perfect the design theory system.In this dissertation, the geotechnical testing equipments including the advanced apparatusfor static and dynamic universal triaxial and torsional shear soil testing are employed toperform montonic triaxial tests and cyclic shear tests under unconsolidated-undrainedcondition. Through a series of tests on saturated clay, the montonic shear behavior, dynmicbehavior including dynamic shear modulus and damping ratio and cyclic softeningdeformation behavior of clay are synthetically examined and studied. Based on analyzingthese abundant monotonic and cyclic shear testing results, the static hyperbolic constitutivemodel for considering the effect of strain rate and cyclic softening general strain calculationalmodel for considering the effects of the combination of static and cyclic stresses arerespectively recommended. Numerical implementation of the finite element method for thesetwo models and secondary development are completed on the platform of the universal finiteelement software ADINA. Furthermore, emphasizing the effect of deformation rate of groundfoundation, the numerical analysis for montonic bearing capacity and cyclic softening generaldeformation of the ground foundation beneath suction caisson are carried out. The maininvestigative contents and achievements consist of the following parts.（1） The improved preparing technique of saturated clay samples. Due to huge expense inobtaining original clay samples and many disadvantages in analyzing the geotechnical testingresults for original clay samples, such as low saturation degree, moreover because it is verydiffcult to reflect the actual stress state when the original clay samples are dealt with throughvacuum suction pressure in geotechnical testing in order to analyze the testing results byutilizing saturated soil theory due to undeveloped unsaturated soil theory, so it is an approachto investigate the strength and deformation behavior of saturated clays by utilizing saturatedclay samples prepared in the laboratory. Through comparing with one-dimensional slurry consolidation method and slurry vacuum suction consolidation method, the advantages ofthese two methods are assimilated and the equipment for preparing clay samples has beennewly designed. The improved method possesses the advantages of convenient operation andshorter time of preparing samples. The prepared clay samples have the homogeneous propertyand highly saturated degree, and were easy to cut for preparing the cylinder orhollow-cylinder testing samples.（2） Monotonic shear behavior and static non-linear constitutive model of saturated clay.The monotonic triaxial and torsional shear tests for saturated clay samples are performedunder consolidated-undrained （CU） and unconsolidated-undrained （UU） conditions. Throughcomparing with these testing results, the studies are emphasized on the effects of strain rateand confined pressure on stress-strain relations and strength behavior. Experimental dataindicate that strain hardening appears at low strain rate while strain softening occurs afterpeak strength at high strain rate. Nevertheless, the difference of deviatoric stress at differentstrain rates will decrease after the strain increases to a certain value. Undrained strength ofthis clay increases with increase of strain rate under the condition of UU tests. In addition,they are revealed that strain-stress relation and strength behavior are mainly affected by strainrate. However they are almost independent on confining pressure under the condition of UUtests. An empirical linear relation which correlates shear strength and strain rate under doublelogarithm coordinates is established through normaiization which makes use of referencestrain srate and corresponding reference strength. A non-linear constitutive model isrecommended through considering the UU experimental condition and the effects of strainrate on stress-strain relations. Before the failure deviatoric stress q_f is attained, stress-strainrelations accord with hyperbolic pattern. After the failure deviatoric stress q_f is attained,stress-strain relations are disposed according to ideal plasticity. This model only includes 4experimental constants which are obtained through monotonic triaxial tests under UUcondition.（3） Cyclic shear deformation behavior and cyclic strain failure criterion: Through a seriesof cyclic torsional shear tests and coupled vertical and torsional shear tests on saturated clayunder unconsolidated-undrained condition, undrained shear deformation behavior, cyclicstrain failure Criterion and pore water pressure development behavior are examined. Thetesting data show that the patterns of cyclic stress-strain relations are distinctly influenced byinitial shear stress. The predominant deformation behavior is characterized by a cyclic effectunder the condition of low initial shear stress or high cyclic stress. Whereas, the predominantbehavior is characterized by an accumulative effect when initial shear stress is high and cyclicstress is low. Although no axial deviatoric stresses are exerted in cyclic torsional shear tests,axial deviatoric strain occurs due to cyclic torsional action. With the increase of cyclic stress,the magnitude of increment of axial deviatoric strain for each cycle increases. But thepredominant strain is shear strain. Although symmetric cyclic stresses are exerted intwo-direction coupled cyclic shear tests, the component of cyclic torsional shear strain hasdistinct cyclic effect, but the component of axial deviatoric strain is characterized by an accumulative effect. These specialities are different from cyclic triaxial or cyclic torsionaltests. For various combinations of the static and cyclic stresses, the cyclic single-directiongeneral strain is composed of the strain caused by initial shear stress and the strain induced bythe combination of static and cyclic stresses. Through comparing with experimental results forall cyclic stress patterns, a general strain failure criterion for clay is recommended forconsidering the integrative effects of torsional shear strain and axial deviatoric strain as wellas cyclic strain and accumulative strain. This failure criterion is considered to be stable innumber for all cyclic stress patterns by comparing to strain criterion currently used. The porewater pressure always fluctuates around the confined pressure with the increase of cyclenumber. It seemed that the fluctuant amplitude increases with the increase of cyclic stress. Butthe buildup of pore water pressure does not occur in cyclic single-direction shear tests andtwo-direction coupled shear tests.（4） Dynamic behavior including dynamic shear modulus and damping ratio under thecondition of large strain (＞10^-3). The testing data in single-graded and graded loading cyclicshear tests indicate that the graded loading method is doable to determine dynamic behaviorsince initial backbone curves, dynamic shear modulus and damping ratio obtained from singlegrade loading by many soil samples are consistent with the ones obtained from graded loadingby one soil sample. Initial shear stress influences development pattern of stress-strainhysteresis loops, dynamic shear modulus and damping ratio measured by graded loading.Effects of initial shear stress on dynamic shear modulus and damping ratio are mainlyreflected by variation pattern step by step during graded loading. The deviatoric stress incoupled cyclic stress raises dynamic shear modulus for a given shear strain amplitude.Especially for the case of small amplitude of cyclic shear stress, along with increase of axialdeviatoric stress, dynamic shear modulus increases obviously. Whereas, for the case of largeamplitude of cyclic shear stress, the increscent effect of axial deviatoric cyclic stress ondynamic shear modulus decreases.（5） Cyclic shear strength behavior and simple and convenient determination approach forcyclic shear strength. The relations of cyclic strength, cycle numbers and initial shear stresshave the significance for design and stability analysis of offshore platform. Testing data incyclic shear tests indicate that cyclic shear strength is determined by the effects of cyclicsoftening and strain rate for given initial shear stress. Based on the relation between montonicstrength and strain rate, the normalization approach of eliminating the effect of strain rate isrecommended. Through utilizing this normalization approach, the data points between cyclicstrength for different cycle number and initial shear stress locate in a narrow strip. Therelation between initial shear stress and normalization cyclic shear strength may be expressedby a quadratic equation. The simple and convenient determination approach of cyclic shearstrength is discussed according to this relation. The cyclic strength of different cycle numbermay be estimated after five experimental constants are determined through a few monotonicand cyclic shear tests. It is significant to reduce experimental workload in the stabilityanalysis for offshore ground foundation. （6） Cyclic creep behavior and cyclic softening general strain computational model. Thetesting data in cyclic shear tests show that cyclic softening general strain is made up of thestrain caused by initial shear stress and cyclic creep strain induced by the combination of thestatic and cyclic stress. Cyclic creep strain is mainly influenced by peak value of cyclic stress,initial shear stress and strain rate used for exerting initial shear stress. According to thespecialities of cyclic creep strain and based on the equivalent between time and cycle number,an experiential cyclic creep strain calculational expression is recommended by using themeasure of equivalent stress. Furthermore, cyclic softening general strain model is set up,whose two compositive parts, namely initial static strain and cyclic ceep strain, arerespectively employing non-linear elasticity theory and yielding surface rheology theory. Thismodel only includes 7 experimental constants, which can be determined in monotonic andcyclic shear tests. Although this.model belongs to the pseudo-static model, it can differentiatethe loading properties, such as monotonic loading or cyclic loading. This model not only canconsider the effect of strain rate used for exerting initial shear stress but also can respectivelycalculate the deformations in the monotonic loading and cyclic loading phase.（7） Implementation of finite element method for cyclic softening general straincomputational model. On the base of a great many monotonic and cyclic shear tests, cyclicsoftening general strain computational model is established, and it exactly reflects the cycliccreep behavior for various combination of static and cyclic stress. Although universal finiteelement software ADINA not only possesses the function of solving nonlinear problem butalso has the outstanding interface of forward and backward disposal, it does not include theconstitutive model which reflects this cycIic shear behavior. So it is necessary to developnumerical implementation on the software ADINA platform. In the dissertation, the secondarydevelopments for non-linear material model and cyclic creep material model on the softwareADINA platform are firstly carried out. Then numerical implementation of cyclic softeninggeneral strain computational model is carried out. Moreover, its development process andcalculational methods are elaborately explained.（8） Numerical analysis of finite element for monotonic bearing capacity and cyclicsoftening deformation of ground foundation beneath suction caisson. Emphasizing on theeffect of clay ground deformation rate during monotonic loading, the numerical analysis formontonic bearing capacity and cyclic softening general deformation of the clay ground hasbeen carried out. The calculational results show that vertical and horizontal montonoicbeating capacity of single suction caisson increases with the increase of clay grounddeformation rate. For the given combination of static and cyclic stresses, vertical andhorizontal cyclic softening deformations increase along with the increase of cycle number.The increscent magnitude of cyclic softening deformation for each cycle decreases duringcyclic loading. For. the combination of vertical montonic and cyclic loads, the soil bodylocating on the surface near the suction caisson will move upward due to extrusion of thedescending foundation, and the main deformation pattern of the foundation exhibits the dentof the whole soil body near suction caisson. Whereas, for the combination of vertical monotonic loads and horizontal cyclic loads, the cranny between soil body and f suctioncaisson appears, and the horizontal and rotational displacement of suction caisson increasesalong with the increase of cycle number.更多还原