Vous êtes sur la page 1sur 11

Computers and Geotechnics 37 (2010) 504514

Contents lists available at ScienceDirect

Computers and Geotechnics


journal homepage: www.elsevier.com/locate/compgeo

The ultimate uplift capacity of multi-plate strip anchors in undrained clay


R.S. Merield a,*, C.C. Smith b,1
a b

Centre for Geotechnical & Materials Modelling, School of Engineering, The University of Newcastle, NSW 2308, Australia Department of Civil and Structural Engineering, University of Shefeld, Sir Frederick Mappin Building, Mappin Street, Shefeld S1 3JD, UK

a r t i c l e

i n f o

a b s t r a c t
Soil anchors are commonly used as foundation systems for structures requiring uplift resistance such as transmission towers, or for structures requiring lateral resistance, such as sheet pile walls. Anchors commonly have more than one plate or bearing element and therefore there is a complex interaction between adjacent plates due to overlapping stress zones. This interaction will affect the failure mode and ultimate capacity. However, no thorough numerical analyses have been performed to determine the ultimate pullout loads of multi-plate anchors. By far the majority of the research has been directed toward the tensile uplift behaviour of single anchors (only one plate). The primary aim of this research paper is to use numerical modelling techniques to better understand plane strain multi-plate anchor foundation behaviour in clay soils. A practical design framework for multi-plate anchor foundations will be established to replace existing semi-empirical design methods that are inadequate and have been found to be excessively under or over conservative. This framework can then be used by design engineers to more condently estimate the pullout capacity of multi-plate anchors under tension loading. 2010 Elsevier Ltd. All rights reserved.

Article history: Received 4 August 2009 Received in revised form 5 October 2009 Accepted 8 December 2009 Available online 12 March 2010 Keywords: Anchor Pullout capacity Finite elements Limit analysis Upper bound and lower bound Clay

1. Introduction 1.1. Background and objectives Multi-plate anchors are geotechnical foundations that can be used as either tension or compression members and are composed of a number of plates welded along a central steel shaft, as illustrated in Fig. 1. The plates are typically constructed as a helix with a carefully controlled pitch. The anchors can have more than one plate located at appropriate spacing on the shaft. The central shaft is used to transfer axial loads to the anchor plates. Unfortunately, current understanding regarding the behaviour of buried foundations, and multi-plate anchors in particular, is somewhat unsatisfactory and remains essentially unchanged for some 20 years. With the advent of complex numerical modelling software and laboratory centrifuge facilities, there is an opportunity to thoroughly investigate the multi-plate anchor problem with the aim of improving our knowledge of such foundations and achieving economies in their design and application. As a starting point, this paper will concentrate solely on the plane strain case of multi-plate anchors with the aim of better understanding the fundamental mechanics of the problem. Further research is planned, where more complex 3D nite element analyses will be employed to better represent the true problem geometry.
* Corresponding author. Tel.: +61 49 215735; fax: +61 49 216991. E-mail address: richard.merield@newcastle.edu.au (R.S. Merield). 1 Tel.: +44 114 2225717; fax: +44 114 2225700. 0266-352X/$ - see front matter 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.compgeo.2010.02.004

The complex nature of buried foundation behaviour has meant there are many conicting theories reported in the literature. Accurately modelling anchor behaviour is a difcult problem that presents many challenges, as pointed out by Das [1] and more recently by Merield [2]. In general, anchor capacity is a function of (1) soil type and density; (2) the capacity of each individual bearing element (i.e. plate or helix); (3) the adhesion between the plate/shaft and surrounding soil; (4) the interaction between each bearing element; (5) the orientation of the anchor; and (6) the embedment depth. Any combination of these variables will significantly affect the observed mode of failure and thus the ultimate capacity of a buried anchor in tension. Yet, the authors believe few of these critical variables are well understood and they have received inadequate attention in the literature. The objectives of the present paper are: (1) a more fundamental understanding of multi-plate anchor behaviour and failure mechanisms, (2) a rigorous design methodology and framework in the form of equations and design charts that can be used by practicing engineers.

1.2. Previous studies There have been numerous theoretical studies that address the uplift of single horizontal anchors in clay as summarised by Merield et al. [3,4]. In contrast, there are very few publications that deal with the theoretical problem of multi-plate anchor foundations.

R.S. Merield, C.C. Smith / Computers and Geotechnics 37 (2010) 504514

505

soil between the blades and bearing capacity of the top blade, given by

Q u 2n 1Scu Bcu Nc rv

where n is the number of anchor plates along the shaft, S is the spacing between each plate, cu is undrained soil cohesion, B is anchor plate width, rv = cH is the total vertical soil pressure at the top plate, and Nc = (2 + p) is the bearing capacity factor for general shear [6]. This theory is dependent on a specic mode of failure which in turn will be a function of the anchor spacings, as discussed later. Anchor manufacturers strongly believe that for deeply buried anchors that the theoretical capacity is best calculated using an individual bearing method based on Terzaghis [6] general bearing capacity equation which is used without any modications. The assumed failure mechanism in the individual bearing method consists of each anchor plate displacing the overburden soil in a logarithmic spiral mode. Thus, the capacity of each plate can be estimated using the well known Terzaghi bearing capacity equation. The total ultimate pullout capacity is the sum over n blades in clay, as given by:

Qu

1 X n

Bn cu N c rv n

Fig. 1. Problem denition.

Kulhawy [5] presents what appears to be one of the only methods of analysis for the uplift capacity of multi-plate anchors. In his model, circular anchors were assumed, where the upper plate develops a cylindrical shear surface that controls its behaviour. The soil between the plates becomes an effective cylinder if the plates are sufciently close together. The shearing resistance along the interface is said to be controlled by the friction angle and state of stress in the disturbed cylinder of soil above the anchor. Key terms in the general model are the modied bearing capacity factors, the in situ stresses and how they are modied by installation effects, and the soil strength along the shear surfaces. Kulhawy [5] gives guidance on how to estimate these factors based entirely on past observations and therefore the overall design procedure is again semi-empirical in nature. 1.3. Multi-plate anchor capacity theories currently in use Anchor manufacturers have chosen to use one of three methods for predicting pullout capacity, namely, cylindrical shear, individual bearing, and an empirical method based on installation torque. In general, manufacturers suggest that the cylindrical shear and individual bearing methods be used to determine capacity and the installation torque method be used as a eld verication tool only. In the cylindrical shear method, the entire volume of soil between the anchor plates is assumed to be mobilized. The term cylindrical was used as the method was originally applied to anchor plates that were circular in shape. This can be easily modied to address continuous strip anchors. The ultimate pullout capacity of a multi-plate anchor was dened as a combination of (1) shear along the slip planes extending vertically between the outside edge of the anchors and (2) individual bearing capacity of the top anchor. Terzaghis [6] bearing capacity theory was generally adopted for estimating the individual bearing capacity of the top anchor. For strip plate anchors in clay (uu = 0), the ultimate pullout capacity/per unit length of a multi-blade anchor using the cylindrical shear analogy is a combination of shear along the strip of

where Bn is the width of plate n, and rvn = c(H + (n 1)S). All other parameters have been dened previously. Both the cylindrical shear and individual bearing methods have similarities to the semi-empirical methods presented by Mitsch and Clemence [7] and Mooney et al. [8] with one significant exception, namely, they do not use adjusted bearing capacity factors or earth pressure coefcients to reect the difference between soil uplift and footing bearing capacity. The authors believe that the concept of applying existing bearing capacity solutions derived initially for surface footings, to the problem of anchor uplift, is not appropriate and is a poor representation of the real problem. Firstly it should be stated that the bearing capacity equation assumes the principle of superposition is valid in relation to soil strength, soil surcharge and soil weight. The principle of superposition remains largely unproven when applied to the uplift of multi-plate anchors in soil. In addition, the standard bearing capacity equations and factors were derived from specic log-spiral failure surfaces and there is no evidence to suggest this is the shape of the failure surface observed when an anchor is subject to tension loads. A more compelling argument could be made to use Terzaghis bearing capacity theory if the anchor was loaded in compression like most foundations, as opposed to tension. One of the main aims of this research is to assess the existing semi-empirical design methods given in Eqs. (1) and (2) and to provide a more appropriate design framework for multi-plate anchor foundations. 2. Problem of anchor capacity 2.1. Problem denition A general layout of the problem to be analysed is shown in Fig. 1. The anchor has a total of n individual plates of width B spaced evenly along the shaft at a distance of S. The anchor spacing ratio is dened as S/B and the anchor embedment depth is H/B. Referring to Fig. 1, the total pullout capacity for the anchor system can be expressed as a summation of the individual plate capacities, namely as;

Qu

n X i1

Q ui

506

R.S. Merield, C.C. Smith / Computers and Geotechnics 37 (2010) 504514

(a)

(b) (c) (d)

Fig. 2. Deep anchor behaviour.

where Qui is the ultimate pullout capacity of anchor i, and n is the total number of anchor plates. The average ultimate bearing pressure for each anchor is

qui

Q ui B

After Rowe et al. [9], the analysis of anchor behaviour can be divided into two distinct categories, namely those of immediate breakaway and no breakaway. In the immediate breakaway case it is assumed that the soil/anchor interface cannot sustain tension so that, upon loading, the vertical stress immediately below the anchor reduces to zero and the anchor is no longer in contact with the underlying soil. This represents the case, where there is no adhesion or suction between the soil and anchor. In the no breakaway case the opposite is assumed, with the soil/anchor interface sustaining adequate tension to ensure the anchor remains in contact with the soil at all times. This models the case, where an adhesion or suction exists between the anchor and the soil. In reality it is likely that the true breakaway state will fall somewhere between the extremities of the immediate breakaway and no breakaway cases. The anchor analyses presented in this paper are performed for the immediate breakaway case only. This will result in conservative estimates of the actual pullout resistance. After allowing for immediate and no breakaway behaviour, anchors can be further classied as shallow or deep, depending on their mode of failure. This point is illustrated in Fig. 2. An anchor is classied as shallow if, at ultimate collapse, the observed failure mechanism reaches the surface (Fig. 2a and b). In contrast, a deep anchor is one whose failure mode is characterised by localised shear around the anchor(s) and is not affected by the location of the soil surface (Fig. 2c and d). In this case, even though the model permits immediate breakaway, a no breakaway result is what the model produces, because the localised failure mode requires zero volume change within the undrained plastic analysis.

For an individual anchor plate of width, B, and soil type, (c, cu), there exists a critical embedment depth, Hcr (Fig. 2), at which the failure mechanism no longer extends to the soil surface and becomes fully localised around the anchor(s). This is true for both single and multi-plate anchors. When this type of behaviour occurs, the total ultimate capacity of the anchor will have reached a maximum limiting value Q (the asterisk on Qu denotes deep failure). u Physically, this transition arises because the undrained shear strength is assumed to be independent of the mean normal stress. From a practical point of view, this result is important as embedding the anchor beyond Hcr will not lead to an appreciable increase in total anchor capacity Qu (see Fig. 2). In the context of a multi-plate anchor arrangement (Fig. 1), the type of deep failure mechanism will be a function of the anchor spacing ratio S/B. If the anchor spacing ratio S/B is large enough then each plate anchor will act independently from each other and produce a collapse mechanism like that shown in Fig. 2d. This transition will occur at a critical value of S/B = (S/B)cr. However, if the anchor spacing ratio S/B is small enough, such that each plate anchor is inuenced by the location of adjacent anchors, then the collapse mechanism is likely to be global like that shown in Fig. 2c. From a design perspective the maximum efciency of a multi-plate anchor Q will be achieved if each of the anchor plates u acts independently of each other.

2.2. Upper bound mechanisms for deeply buried multi-plate anchors By deriving several analytical solutions for the ultimate capacity of anchors, a useful check can be made on the numerical nite element scheme. The upper bound theorem states that if a set of external loads can be found acting on a compatible failure mechanism such that the work done by these loads in an increment of displacement is equal to the work done by the internal stresses, these external loads are not lower than the true collapse load. In

R.S. Merield, C.C. Smith / Computers and Geotechnics 37 (2010) 504514

507

(a)

(b)

(c)

(d)

Fig. 3. Upper Bound anchor mechanisms for deep failure modes (a) Individual anchors (Rowe, 1978), (b) Interacting dual plate anchor Mechanism 1, (c) Interacting dual plate anchor Mechanism 2, and (d) Interacting dual plate anchor Mechanism 3.

its simplest form the failure mechanism may be assumed to be comprised of rigid blocks, and by examining different block arrangements, the best (least) upper bound value can be found. For a fully bonded or deep anchor plate i, Rowe derived an upper bound solution for ultimate pullout capacity Q using the ui mechanism shown in Fig. 3a as 11.42 or

Q S u 6 2p 4n 1 B Bcu

As mentioned previously, it is important to realise that the theoretical capacity of any given multi-plate anchor does not continue to increase indenitely. There will be a limiting value placed on Eq.

Q 2 3pBcu ui

where the asterisk is used here to signify a deep failure mode. Eq. (5) can be used to calculate the limiting capacity for a single deep anchor plate as discussed previously. For a multi-plate anchor arrangement the situation is more complex due to the possible interaction between plates. For a deep global failure mode that encompasses all the anchor plates (i.e. S/B is small), three possible mechanisms are shown in Fig. 3bd. The mechanisms consist of several rigid block and fan zones and it can be shown that:

 p  S Q u 2 3p 2 2p Mechanism 1 B Bcu  S p Qu 2 3p 2 2 Mechanism 2 B Bcu Qu S 6 2p 4 Mechanism 3 B Bcu

6 7 8

It is clear from the above three equations that Mechanism 3 provides a better (lower) estimate of the collapse load for values of S/B > $1 and will therefore be adopted. This mechanism can easily be extended to anchors with a total of n individual plates, and the capacity is simply given by

Fig. 4. Analytical upper bound solution for the limiting ultimate capacity of a multiplate anchor system.

508

R.S. Merield, C.C. Smith / Computers and Geotechnics 37 (2010) 504514

(9). That is, the maximum total capacity of a multi-plate anchor system Q cannot exceed the sum of the individual anchor plate u capacities (Eq. (5)) assuming they act independently of each other. Therefore it can be stated that

Q 6 u

n X i1

Q ui

10

If all the anchor plates are deep and act independently of each other then

Q n2 3pBcu u

11

By equating Eq. (9)(11) we can determine the critical anchor spacing ratio S=Bcr , where the transition occurs between global (Fig. 2c) and individual (Fig. 2d) deep anchor failure. This indicates that

  2 3p 4p S n1 4 B cr

12

Eq. (9) in conjunction with Eqs. (11) and (12) have been used to create Fig. 4. For this upper bound mechanism S=Bcr 2:85 for large n which is similar to the critical value perceived by many manufacturers which is around three S=Bcr % 3 for axi-symmetric anchors. Fig. 4 can be used to estimate the limiting ultimate capacity of a multi-plate anchor system. Although the mechanism in Fig. 3d is a valid upper bound, it may not be the best upper bound and therefore the accuracy of the results in Fig. 4 are largely unknown and will be investigated in the following sections. 2.3. Pull-out capacity of individual anchor plates in undrained clay In contrast to formulating a methodology for multi-plate anchors based on bearing capacity theory (i.e. Eqs. (1) and (2)), it is proposed to extend the plate anchor methodology of Merield et al. [3]. Therefore, the ultimate anchor bearing capacity in undrained clay for each anchor plate qui will be expressed as a function of the undrained shear strength in the following form:

Fig. 5. Typical nite element mesh.

In addition, by equating Eqs. (14)(16), the critical overburden ratio cH=cu cr , where the transition from shallow to deep failure will occur can be estimated as

cH
cu


cr

N N coi c

17

It is important to note here that the inequality Q u 6 Q must u also be satised globally using Eqs. (9) and (10). 3. Finite element analysis The displacement nite element software ABAQUS was used for solving this problem. The ABAQUS model consisted of two parts: the anchor and the soil. A typical mesh for this problem, along with the applied displacement boundary conditions, is shown in Fig. 5. The actual distribution and concentration of elements varied as a function of the anchor embedment and geometry. The optimum distribution of elements was obtained using the mesh adaptivity process option available in ABAQUS. The unstructured mesh primarily comprised 4-noded quadrilateral plane strain elements which were found to provide the best solution convergence. The overall mesh dimensions were selected to ensure that the zones of plastic shearing and the observed displacement elds were contained within the model boundaries at all times. The soil was modelled as an isotropic elasto-perfectly plastic continuum with failure described by the MohrCoulomb yield criterion. The elastic behaviour was dened by a Poissons ratio m 0:49, and a ratio of Youngs modulus to shear strength of E/ cu = 400. Because the anchor is assumed to be much stiffer than the soil it comes into contact with, the anchor was modelled as a discrete rigid body. This simplication provides signicant computational savings. Although it is likely that shaft friction contributes to the capacity, the term is generally ignored in anchor design because of the uncertainties involved. This assumption is considered conservative. Therefore the shaft was not considered in the ABAQUS analyses. For the problem presented, the anchorsoil interface was modelled using the surface to surface small-sliding master/slave approach as formulated in ABAQUS. The master was taken to be the anchor and the slave was the soil. A small-sliding approach was

qui

Q ui cu Nci B   qui cH Ncoi cu cu c0   qui cu c0

13

where for a homogeneous soil prole

Nci

14

and the term Ncoi is an anchor break-out factor dened as

Ncoi

15

Implicit in (13) is the assumption that the effects of soil unit weight and cohesion are independent of each other and may be superimposed. It was shown by Merield et al. [3] that this assumption generally provides a good approximation to the behaviour of anchor plates in purely cohesive undrained clay. It is exact for the no-breakout condition. The above equations reect the complex nature of the pullout factor N cn , as observed by Rowe [10] and Merield et al. [3], which is a function of both the embedment ratio and overburden pressure. The latter dependence is expressed in terms of the dimensionless quantity cH=cu and implies that if the ratio of cH=cu is large enough, then the anchor will behave as a deep anchor even at shallow embedment depths. When using Eq. (5) the inequality Q ui 6 Q must also be satisu ed for each individual anchor plate. Therefore the value of N ci has a limiting value of N that is easily obtained from Eq. (5) as c

N 2 3p c

16

R.S. Merield, C.C. Smith / Computers and Geotechnics 37 (2010) 504514

509

seen as appropriate given that geometrically non-linear effects were neglected. The strength of the anchor-soil interface was modelled using an interface roughness, a, where the maximum shear stress at the interface, smax = acu. For all analyses a smooth interface strength has been modelled, corresponding to a 0. A no-tension condition allowing separation (or breakaway) of the anchor from the soil has been imposed at the anchorsoil interface. When the soil selfweight is set to zero, separation occurs immediately there is any normal motion of the anchor away from the soil. Small strain analyses were performed on the pre-embedded anchor therefore neglecting the installation and large strain aspects of the problem. Such a wished in place analysis is of course a simplication of what is actually a very complex problem. A truly rigorous analysis of this problem would need to include large deformations, large strains, material remoulding and contact conditions that include self-contact behaviour. To determine the collapse load of the anchor, displacement-dened analyses were performed in which the anchor was displaced vertically. A prescribed displacement was applied to the rigid anchor, resulting in uniform vertical motion of the entire anchor. The total displacement was applied over a number of sub-steps and the total nodal contact force along the anchor/soil interfaces in the vertical (V) direction was used to calculate the total ultimate load Q u . By observing the loaddisplacement response, a check can be made to ensure that the limiting soil resistance has been reached and that overall collapse has occurred. In addition, the load carried by each individual anchor can be observed.

discontinuities for a specied arrangement of nodes. While the proximity of the numerical solution to the true solution in general remains unknown, performing a nodal density study provides a good indication of the improvement in accuracy which can be expected as the nodal spacing tends towards zero. Particularly when cohesive (Tresca) materials are involved the accuracy of the solutions will typically be found to be extremely good, even when relatively low nodal densities are used. In this study 1000 nodes were typically adopted in each analysis. 5. Results and discussion Finite element analyses were performed to obtain estimates of the anchor break-out factor Ncoi (15) for a range of embedment depths and anchor geometries. Analyses were conducted on two and three plate anchor geometries n 2; 3 at anchor spacing ratios of S=B 14 and embedment depths of H=B 110. A range of DLO numerical upper bound analyses were also performed for the above cases with additional analyses performed on anchors with four and ve plates n 4; 5. These results, along with the effects of overburden pressure, are discussed in the following sections, where possible, past experimental and numerical results are compared to results obtained from the current study. 5.1. Single plate anchors in weightless soil An upper bound value of the break-out factor Ncoi for single plate anchors was proposed by Merield et al. [3] as

4. Discontinuity layout optimisation analysis In addition to the ABAQUS nite element modelling, a new numerical limit analysis procedure, discontinuity layout optimisation (DLO) [11] was used to solve the multi-plate anchor problem. The implementation of DLO in LimitState: GEO was utilised in this paper. The results are compared with the ABAQUS nite element solutions. Only a brief summary of the method is provided here. The reader is referred to Smith and Gilbert [11] for a full description of the method. The procedure presented by Smith and Gilbert [11] is able to determine the layout of discontinuities that produce the critical sliding block mechanism and associated upper bound limit load for any plane plasticity problem. In the procedure, a grid of nodes is set out over the problem domain. Linear Programming is used to select the critical layout of discontinuities corresponding to the optimal collapse mechanism from a large set of potential discontinuities connecting every node to every other node. The potential discontinuities, are permitted to crossover one another giving a much wider search space than when such discontinuities are located only at the edges of nite elements of xed topology. The key distinctiveness of the DLO procedure lies in the expression of the limit analysis problem formulation entirely in terms of lines of discontinuity, rather than in terms of elements (as is usually the case). Using DLO a large number of potential discontinuities are set up and these take up many different orientations and are free to cross each other arbitrarily. In contrast, with element-based formulations discontinuities are typically restricted to lie only at the edges of elements positioned in a xed mesh, and, since overlapping elements are not normally allowable, discontinuities are not permitted to cross. This means that when a critical mechanism is being sought the search space is very much more restricted. The DLO procedure is also inherently able to handle singularities which can be challenging for FE based approaches. The proposed procedure has been found to be very robust in use and can be guaranteed to determine the most critical layout of

  H Ncoi 2:76 loge 2 B

18

The value of the break-out factor N coi obtained from the nite element analyses of single plate anchors carried out in this study can be approximated by the following equation:

Ncoi 2:08 2:47 ln

  H B

19

For a fully bonded or deep anchor plate, the ABAQUS nite element and DLO solution for the ultimate pullout capacity Q were ui found to be 11.57 and 11.42 respectively which compares favourably with the exact solution of Rowe [10] as given in Eq. (5). 5.2. Multi-plate anchors in weightless soil The computed FE and DLO estimates of the anchor break-out factor N coi are shown graphically in Fig. 6 for an anchor with two plates n 2. Fig. 6a plots the break-out factor for the anchor closest to the ground surface (Anchor 1, i 1) as dened by Eq. (15), i.e. N co1 . Results for anchor spacings of S=B 14 are also shown in this gure. The results in Fig. 6a indicate that there is only a small reduction in the capacity of the shallowest anchor plate (Anchor 1) due to the anchor plate below. For practical design purposes, however, it can be concluded that the break-out factor for the uppermost anchor is largely unaffected by the location of the lower anchors and appears almost independent of S=B. Fig. 6b plots the FE break-out factor for the anchor immediately below the shallowest anchor (Anchor 2, i 2) as dened by Eq. (15), i.e. Nco2 . The results show that the capacity of the anchor below the shallowest anchor is largely independent of the overall embedment ratio H=B, particularly for values of S=B 6 2. To conrm whether this observation also holds true for anchors with more than two plates n > 2, an extensive set of numerical upper bound analyses were performed using the DLO method. This included analysing anchors with up to ve plates spaced evenly along the shaft. The average value of the break-out factors Ncoi

510

R.S. Merield, C.C. Smith / Computers and Geotechnics 37 (2010) 504514

(a)

Fig. 7. Average DLO upper bound break-out factors for multi-plate anchors in weightless soil, n = 25. For each value of n, values of i = 25 were modelled with essentially no change in the result.

(a)

(b)

(c)

(b)
Fig. 6. Break-out factors for dual (n = 2) plate anchors in weightless soil (a) anchor 1 and (b) anchor 2.

(d)

(e)

(f)

Fig. 8. FE Displacement contours for anchors in weightless clay.

are shown in Fig. 7. The results conrm that, regardless of the number of plates along the shaft, the break-out factor for all anchor plates below the upper anchor can be conservatively estimated using Eq. (19) assuming they are embedded at an equivalent depth ratio of H=B S=B. Fig. 7 also plots the values of Ncoi using Eq. (19), conrming that the estimates are generally conservative. For ratios of S=B > 3, Ncoi increases more noticeably with H=B, and therefore Eq. (19) may be overly conservative in these cases. However, as typical plate spacing ratios are three or less, using Eq. (19) should be adequate for design purposes. The displacement contours at collapse for anchors in weightless soil are illustrated for various embedments and anchors in Fig. 8. For anchors at small embedment ratios (H/B = 1) the failure mechanism of the shallowest anchor consists of the upward movement of a rigid soil block immediately above the anchor. As the anchor embedment depth increases, the zone of plastic shearing extends outward from the anchor edges and causes an increase in the area over which deformations occur at the surface.

The FE displacement pattern for the shallowest anchor (Anchor 1) remains largely unchanged regardless of the proximity of the underlying anchors. In addition, the mechanism for the lower anchors is independent of the overall embedment ratio H=B and is primarily related to the local equivalent embedment ratio of S=B. This is highlighted by comparing Fig. 8a with Fig. 8f and d with Fig. 8e, and is consistent with the ndings shown in Figs. 6 and 7. 5.3. Multi-plate anchors in soil with weight The numerical results discussed above have been limited to soil with no unit weight, and therefore the effect of soil weight (overburden) needs to be investigated. If the assumption of superposition is valid then it would be expected that the ultimate anchor capacity would increase linearly with the dimensionless overburden pressure cH=cu . The results from further FE analyses which include cohesion and soil weight, shown in Fig. 9 conrm that this is

R.S. Merield, C.C. Smith / Computers and Geotechnics 37 (2010) 504514

511

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 11. Displacement contours illustrating transition from deep global to deep individual plate failure mode (n = 3). Fig. 9. Effect of soil weight on ultimate pullout capacity FE results (n = 2, S/B = 1).

indeed the case. This conclusion is in agreement with the observations of Rowe [10] and Merield et al. [3]. Fig. 9 shows that the ultimate anchor capacity increases linearly with overburden pressure up to a limiting value. This limiting value reects the transition from shallow to deep anchor behaviour, where the mode of failure becomes fully contained around the anchor. At a given embedment depth, an anchor may behave as either shallow or deep, depending on the dimensionless overburden ratio cH=cu . For this particular problem (n 2, S=B 1) the failure mode at the limiting capacity is of the form shown in Fig. 2c, also shown in Fig. 10a and b. As mentioned previously, the type of deep failure mechanism will be a function of the anchor spacing ratio S=B which, if large enough, will be of the form shown in Fig. 2d. The FE displacement patterns for a number of deep anchor problems are shown in Figs. 10 and 11. These gures illustrate the transition between the two types of deep anchor failure mechanism previously shown in Fig. 2c and d. A global deep failure mechanism is visible in Figs. 10a and b and 11ac. A transitional mechanism is shown in Figs. 10c and 11d and e while a truly local deep failure can be observed in Fig. 11f. The failure mechanisms for shallow anchors with unit weight will be the same as that shown previously for the weightless soil case (Fig. 8).

A summary of the ABAQUS and DLO numerical upper bound results for the limiting ultimate capacity of a multi-plate anchor system can be seen in Fig. 12. Results for two and three anchor plates n 2; 3 are shown in this gure. A review of the results shown in Fig. 12 suggests that Eq. (12) does not adequately predict the transition from a global mechanism encompassing all anchor plates, to an individual deep failure mechanism, where a local failure mechanism exists around each anchor plate. It has been found that a better estimate of the ultimate limiting capacity is provide by the following expression:

Q S u 2 3p n 1p B Bcu

20

Which differs from Eq. (9) and provides a better t to the FE results. This indicates the assumed upper bound mechanism Fig. 3d is not the critical or optimum one.

(a)

(b)

(c)
Fig. 12. Comparison of analytical and FE solutions for the limiting ultimate capacity of a multi-plate anchor system cu = 50 kPa, c = 15 kN/m3, 0.4 m, 0.2 m, and 3 m.

Fig. 10. Displacement contours illustrating transition from deep global to deep individual plate failure mode (n = 2).

512

R.S. Merield, C.C. Smith / Computers and Geotechnics 37 (2010) 504514

When Eq. (20) is combined with Eq. (11), the critical embedment ratio is given as

(ii) If

  S 2 3p B cr p

21

S0 2 3p > B p
each anchor plate will behave independently, deep local (Fig. 2d), and the total ultimate capacity Q u can be calculated using Eq. (11) as

which predicts the transition between failure modes very well as shown in Fig. 12. 5.4. Suggested procedure and practical example Listed below are the suggested steps for estimating the ultimate capacity of a plane strain multi-plate anchor system based on the analyses discussed above. The anchor spacing S is assumed constant along the anchor shaft. 1. Determine representative values of the material and geometric parameters c; cu , H/B and S/B 2. For the anchor plate closest to the surface (Anchor 1) calculate cH=cu cr using Eqs. (16) and (17).

Qu Q u n2 3p Bcu Bcu
END 6. If SHALLOW (4(ii)) (i) Estimate the capacity of the upper anchor Q u1 (Anchor 1) from Eq. (13) as

cH
cu


cr

N coi

where i 1;   H ; and N 2 3p 2:08 2:47 ln c B

N Ncoi c

Q ui cu Nci ; i 1; B   qui cH Ncoi N ci cu cu c0


(ii) If

cS
cu

 P

cS
cu


cr

3. For the remaining anchor plates below Anchor 1, calculate cS=cu cr by substituting H = S and using Eqs. (16) and (17).

cS
cu


cr

All remaining anchors will have individual deep failure modes (Fig. 2d) and their capacity Q ui can be calculated from Eq. (5) as

N coi

where i 1;   S ; and N 2 3p 2:08 2:47 ln c B

N Ncoi c

Q 2 3pBcu ; ui
n X 12

i 2n

The total ultimate anchor capacity can now be calculated

Q u Q u1
4. For the anchor plate closest to the surface (Anchor 1), if (i) If END (iii) If

Q ui

cH
cu

 P

cH
cu


cr

) DEEP

cS0
cu

 <

cH
cu


cr

The anchor plate will fail in a deep failure mode. (ii) If

cH
cu

 <

cH
cu


cr

) SHALLOW

Then all remaining anchor plates will fail in a shallow failure mode and the capacity of each remaining plate can be calculated from Eq. (13) substituting S for H. The total ultimate anchor capacity can now be calculated

The anchor plate will fail in a shallow failure mode. It is important to remember that the value of H in the above equations is xed by the depth of the rst plate and so will cancel out in these comparisons. 5. If DEEP (4(i)) (i) If

Q u Q u1

n X i2

Q ui

END It is anticipated that a similar procedure will be applicable to the design of axi-symmetric anchors with minor change to the coefcients. 5.5. Example of application

S0 2 3p 6 B p
where S0 n 1S. The collapse mechanism will be deep global (Fig. 2c, and the total ultimate capacity Q u can be calculated using Eq. (20) as

Qu Q S u 2 3p n 1p B Bcu Bcu
END

We now illustrate how to use the results presented to determine the ultimate pullout capacity of an anchor in clay. Problem: An anchor with ve plates of width 0.2 m is to be embedded 3 m (H = 3 m) in a homogeneous clay. The anchor spacing is 0.4 m. Determine the ultimate pullout capacity given the clay has a shear strength cu = 50 kPa and unit weight c = 15 kN/m3 The problem geometry is shown in Fig. 13a. The systematic procedures given above will now be used to determine the ultimate anchor capacity.

R.S. Merield, C.C. Smith / Computers and Geotechnics 37 (2010) 504514

513

Nc1 Nco1

cH
cu

8:77 0:9 9:67

Q u1 Bcu Nc1 0:2509:67 96:7 kN


(ii) S0 = (n 1)S = (5 1)0.4 = 1.6 m

cS0
cu

  <

    151:6 cS 7:63: 0:48 and from above 50 cu cr 

Therefore

cS0
cu

cS
cu


cr

Then all remaining anchor plates will fail in a shallow failure mode The capacity of each remaining plate can be calculated from Eq. (13) remembering that

(a)

(b)

H S 2; B B Q ui cu Nci B N ci Ncoi

and H 0:4 where i 25

Fig. 13. Example problem (a) problem layout and (b) displacement contours at ultimate load (ABAQUS FE).

1. Given cu = 50 kPa, c = 15 kN/m3, S = 0.4 m, H = 3 m determine

from Eq: 19   H 2:08 2:47Ln2 3:79 Ncoi 2:08 2:47Ln B cH 150:4 0:12 cu 50 cu 3:79 0:12 3:91 cu Q u4 Q u5 Bcu Nci 0:2503:91 39:1 kN
5 X i2

cH

H=B 3=0:2 15 and S=B 0:4=0:2 2


2. For the anchor plate closest to the surface (Anchor 1) calculate cH=cu cr using Eqs. (16) and (17).

Therefore

Nci Ncoi Q u2 Q u3

cH

Ncoi
and

  H 2:08 2:47Ln 2:08 2:47Ln15 8:77; B


N c 2 3p 11:42 therefore

The total ultimate anchor capacity can now be calculated

Q u Q u1

Q ui 253:1 kN

cH
cu

cr

N c

Ncoi 11:42 8:77 2:65

3. For the remaining anchor plates below Anchor 1 calculate cS=cu cr using Eqs. (16) and (17).

The ABAQUS nite element analysis of this problem predicted a total ultimate capacity of 250 kN. The displacement contours at collapse are shown in Fig. 13b which conrm the shallow failure mode predicted in the calculations steps above. Using cylindrical shear theory (Eq. (1)) the capacity is calculated as 220 kN (13%) while using individual bearing theory estimates the capacity as 314 kN (+24%). 6. Conclusions 1. A practical and straightforward design framework for plane strain multi-plate anchor foundations buried in undrained soils has been presented to replace existing semi-empirical design methods that are inadequate and have been found to be excessively under or over conservative. 2. The proposed method avoids the need to apply bearing capacity theory, which was derived initially for surface footings, to the problem of anchor uplift. This approach is considered a signicant approximation. 3. The new design framework is derived from a series of rigorous FE and DLO upper bound analyses on a wide range of problems and has been veried through a practical example. It has been derived for the case, where no suction forces exist between the anchor and soil, and the capacity of the shaft has been ignored. 4. The results indicate that the capacity of individual anchor plates within a multi-plate anchor (where all blades are of equal width) are largely independent of the other anchors and dependent only on the blade width, the soil strength and distance to the next blade (or surface in the case of the top most blade).

  S Ncoi 2:08 2:47Ln 2:08 2:47 ln2 3:79; B


and

cS
cu

N c

2 3p 11:42 therefore

cr

N c

Ncoi 11:42 3:79 7:63

4. For the anchor plate closest to the surface (Anchor 1),

cH
cu

    153 cH 0:9 and 2:65 therefore 50 cu cr

0:9 < 2:65 ) SHALLOW


5. For the anchor plate closest to the surface (Anchor 1), (i) Estimate the capacity of the upper anchor Q u1 (Anchor 1) from Eq. (13) as

Q u1 cu N c1 B

514

R.S. Merield, C.C. Smith / Computers and Geotechnics 37 (2010) 504514 [4] Merield RS, Lyamin AV, Sloan SW, Yu HS. Three-dimensional lower bound solutions for stability of plate anchors in clay. J Geotech Geoenviron 2003;129(3):24353. [5] Kulhawy FH. Uplift behavior of shallow soil anchors an overview. ASCE; 1985. p. 125. [6] Terzaghi K. Theoretical soil mechanics. New York: John Wiley and Sons; 1943. p. 510. [7] Mitsch MP, Clemence SP. Uplift capacity of helix anchors in sand. Detroit (MI, Engl): ASCE; 1985. [8] Mooney JS, Adamczak SJ, Clemence SP. Uplift capacity of helix anchors in clay and silt. ASCE; 1985. p. 2647. [9] Rowe RK, Davis EH. Behaviour of anchor plates in clay. Geotechnique 1982;32(1):923. [10] Rowe RK. Soil structure interaction analysis and its application to the prediction of anchor behaviour. Australia: University or Sydney; 1978. [11] Smith C, Gilbert M. Application of discontinuity layout optimization to plane plasticity problems. Proc Roy Soc AMath Phys Eng Sci 2007;463(2086):246184.

5. The exception to this rule is, where a global deep mechanism forms involving ow around from the top most plate to the base plate. This normally occurs when the intermediate anchor spacing is small. 6. Based on the above, the framework is able to encompasses the complex relationship between the various modes of failure, the anchor geometry, and the soil properties with a straightforward procedure and also provides a generic basis for which other anchor plate congurations (e.g. axi-symmetric) might be assessed. References
[1] Das BM. Earth anchors. Amsterdam: Elsevier; 1990. [2] Merield RS. Numerical modelling of soil anchors. Australia: University of Newcastle; 2002. [3] Merield RS, Sloan SW, Yu HS. Stability of plate anchors in undrained clay. Geotechnique 2001;51(2):14153.

Vous aimerez peut-être aussi