Professional Documents
Culture Documents
Introduction
Although gravity retaining wall systems have been adapted to retaining fills, excavation retaining walls are the most frequently used type of retaining wall and have experienced continued development throughout the 20th century. The development and diversification of the technique relating to retaining walls, sheet piles, diaphragm walls, Lutecian type interpile sheeting, parisian interpile sheeting and cut-off walls, as well as the technique of using supports such as struts with pretensioned anchors have made it possible to build more and more subsurface walls: transport infrastructure, various networks, building bases, parking areas, etc. [Delattre, 2000]. This development and application of technologies resulted in a major development of computation methods. In fact, this form of structure requires the diversification of the computation schemas of earth pressure because their kinematic characteristics are different from those of gravity walls, the most popular kind of retaining walls to date.
* This is the follow-up article of an article dedicated to traditional methods and the coefficient method of reaction [DELATTRE, 2002]. It will be completed by an article about the application of the finite elements method on retaining walls since the 1970s.
B ULLETIN
DES LABORATOIRES DES
P ONTS
ET
C HAUSSES - 244-245 -
2003 -
REF .
4457 -
PP .
31-51
31
Gravity walls are rigid and their movements combine lateral displacement and warping when the supported soil experiences pressure. As a result the soil is decompressed laterally and is brought to a state of limited equilibrium of the thrust [Delattre, 2002]. The earth pressure computation methods assocated with this kinematic characteristic of the wall were proposed by Coulomb [1776] and Boussinesq [1882] and have undergone many experimental validations [e.g. Darwin, 1883; Feld, 1923]. Contrary to this kinematic characteristic of warping in rigid gravity walls that are used to retain fills, flexible retaining walls have more complex kinematic characteristics. Two main factors are responsible for this diversification in the walls [Delattre, 2002]: Their relative flexibility leads to a variable deflection of the structure, as well as a redistribution of the thrust between the areas where supports exist and the the areas between supports; The ways in which the retaining walls are implemented provoke a kinematic reaction of the entire wall that moves from a top rotation (warping) into a base rotation. Original tools needed to be developed for ascertaining these new forms of interaction between the wall and the supported soil. The problem/task nevertheless proved to be difficult. In practice, it was necessary to wait for the appearance of the finite element method so that the different forms of soilstructure interaction concerned could be included into the details. In the mean time, empirical and semi-empirical solutions were proposed by engineers. These solutions, which were initially developed for calculating the stress in structural element, were otherwise to be used to assess the deformation of the structure and the supported soil block. The development of these empirical and semiempirical methods as well as that which concerns stress assessment and distortion of structures is the focus of this article.
32
B ULLETIN
P ONTS
ET
C HAUSSES - 244-245 -
2003 -
REF .
4457 -
PP .
31-51
The study of earth pressure on flexible walls The experimental validation of the empirical schema adopted by Christiani and his successors of the Danish school was continued by Stroyer [1935]. He started a series of tests on reduced models using a device made specially for studying earth pressure distributions on flexible walls. In this device, the rigid wall pivoting at the base is replaced by a wall that is jointed at the top and at the base and that is also sufficiently flexible to undergo a deflection in its middle portion. This device allowed Stroyer to observe that the stress in the middle part of the wall that was submitted to deflection, reduced as the deflection increased. This meant that the wall was flexible. He otherwise observed that this stress reduction in the middle part of the wall was accompanied by a redistribution of stress towards the fixed points at the top and base of the wall and that this did not lead to reduction of the resultant stress on the wall. Tschebotarioff and Brown [1948], then Rowe [1952, 1961] and Masrouri [1986] continued this first experimental study by equally conducting studies on reduced flexible models. The works of Tschebotarioff helped to underline the fact that the redistribution of stress by arching appeared only on walls sheeted from top to bottom, anchored rigidly at the top (to a platform, according to the circumstances), then dredging (Fig. 1) and that such redistributions of stress should not be considered as walls filled or anchored with tension rods that are relatively flexible. The experiments of Rowe [1961] and Masrouri [Masrouri, 1986; Masrouri and Kastner, 1991] involved walls sheeted from top to bottom, then excavated with struts installed in the advancement. Similar to the experiments of Stroyer, Rowes experiments (Fig. 2), carried out on a flexible wall and rigid struts, clearly showed a redistribution of stress on the wall, consisting of a reduction of stress between the supports and a concentration of stress on the supports. For her part, Masrouri concentrated on studying the influence of strut rigidity and the intensity of initial prestressing on stress distributions on a semi-flexible wall. She showed that, for a flexible support that was not prestressed, the stress distribution behind the wall corresponded to the distribution calculated using the theories of thrust while the displacement in the base of the wall remained limited, but that this was no longer true the moment the prestressing and rigidity of the strut rose or when displacement in the base of the wall increased. She actually showed that both the increase in stiffness of the support and the increase of initial prestressing caused the pressures applied to the wall in the support area to increase while the areas between the supports could be relieved. In the same way she showed that if excavation is pursued in order to generate strong movements in the base, stress concentrations will increase to the right of the support. Similar work carried out this time on structures anchored with tension rods, enabled Masrouri to obtain analogue results.
Soft clay
After filling
Filling
Dragage
Fig. 1 - Stress distribution stemming from different construction procedures [according to Tschebotarioff and Brown, 1948].
B ULLETIN
P ONTS
ET
C HAUSSES - 244-245 -
2003 -
REF .
4457 -
PP .
31-51
33
Fig. 2 - Stress distribution taken on a reduced flexible wall model with struts rigidly installed in the advancement of the excavation [Rowe, 1961].
Aft
er
exc
ava
tion
P ONTS
ET
C HAUSSES - 244-245 -
2003 -
REF .
4457 -
PP .
31-51
The structure is dimensioned using the single direction thrust bearing method, with the thrust being calculated using an angle of wall-soil friction equal to /2. Firstly, the thrust on the structure is calculated using the Coulomb method by stipulating a wall-soil friction equal to zero. The diagram of the thrust that is obtained is then modified by constructing a parabola that reduces the pressure of the coefficient q at the centre of the two supports anchored from above and the application point of the bearing resultant and increasing from 1.5 q to the top anchorage level. To find q:
q
1,5q A A
1/2 L
10H + 2L -P q = k ---------------------10H + 3L m where pm is the average value of stress distribution calculated with the Coulomb method. H is the charge applied above the anchorage level, equal to the height of the soil above the anchorage level, established while taking into account the submerged weight of the soil. L is the distance between the two support points that support the wall and k is a coefficient approximately equal to 0.8. The pressure distribution that is calculated helps to ascertain the bending moment of the wall and the anchorage force. The depth given to the wall can be obtained by multiplying the depth necessary for limiting equilibrium by 2 .
1/2 L
2/3 D d 2 1/3 D
Fig. 3 - Danish computation method for walls anchored at the top [in Brinch-Hansen, 1953].
experienced by the retaining wall at each level, crossed with the number of excavation phases and then the depth, before its blocked by a support. The measurements taken on these structures have revealed pressures on retaining walls that do not obey the triangular law of distribution predicted in the theories of Rankine or Boussinesq. Contrary to what was expected, the earth pressures were stronger in the middle part of the wall than in the bottom part.
P ONTS
ET
C HAUSSES - 244-245 -
2003 -
REF .
4457 -
PP .
31-51
35
Base rotation
Translation
Top rotation
Fig. 4 - Dependence of the diagram of the thrusts appplied by the soil to the kinematics of the wall [according to Terzaghi, in Ohde, 1938].
Ever since, many authors have reconsidered, using experimental study results, the influence of the kinematic characteristic of a structure on thrust distribution. The obtained results mainly concern thrust distribution applied to a wall for the three wall kinematics characteristics: translation, top rotation and base rotation. Concerning the transaltion movement, Sherif et al. [1982], as well as Fang et al. [1997], have demonstrated that thrust distribution during movement of the wall stays noticeably triangular. This result is qualified by Fang and Ishibashi [1986], for whom the thrust distribution deviates slightly from the triangular distribution, the thrusts being slightly stronger than expected with this distribution in the upper section of the wall and slightly lower in the bottom part of the wall. The case of walls pivoting at the top have been studied by James and Lord [1972]. The study revisited the measuring devices used in the case of wing wall abutments [James and Bransby, 1970]. Although the device did not enable them to measure the thrust applied by the soil on the wall, owing to the fact that the thrust cells used were dimensioned to measure stress on the wing wall abutments, it enabled them to gain access to the soil block deformations. They also observed that the deformations were located following a band for which the curve resembled the arc of a circle. This band of deformation started at the base of the wall and progressed, throughout the displacement of the wall, towards the surface of the soil block. James and Lord otherwise observed that the volume of the soil, while slipping along this band of deformation is all the weaker bacause the sand is dense: the deformation band develops all the more behind the wall as the sand is loose. The thrust distribution applied by the soil to the wall for this structural kinematic characteristic was studied, always on reduced models, by Fang and Ishibashi [1986]. They observed that while the wall is rotating, moving from the "at rest" state, the stresses applied at the top of the wall have a tendency to decrease in the lower part. They also highlighted the stress distributions for which, in the upper quadrant of the wall, the stress is significantly higher at the earth pressures at rest state, while in the middle section, this stress approaches the theoretical thrust and that in the bottom quadrant, this pressure decreases until it reaches values close to zero at the foot of the wall. They otherwise showed that the initial density state of the sand noticeably influenced the thrust distribution. In this way, the thrust distribution for loose sand, while being similar to the distribution described above, is not far removed from a triangular distribution. For greater initial densities, the thrust distribution is clearly more removed from a triangular distribution with the stress being concentrated at the top increasing the density state of the material. Fang and Ishibashi explain this result by using the notion of anchorage efects created at the edges of the fixed point at the top of the wall. For structures pivoting at the base, they finally observed that during rotation of the wall, the thrusts decreased less rapidly in the bottom section of the wall than at the top. In this way, they showed that the wall movement led to the develoment of an equilibrium state limit in the top part of the soil block (approx. the upper half), while in the bottom part, the soil remained at an intermediate state between the thrust state and the rest state.
36
B ULLETIN
P ONTS
ET
C HAUSSES - 244-245 -
2003 -
REF .
4457 -
PP .
31-51
distribution on the wall. In fact, when the movement of the upper part of the wall is insufficient at this level to authorize the lateral decompression of the soil mass involved in the mobilisation of the thrust, an anchorage effect appears resulting in the increase in thrusts in the top part of the wall and a reduction at the bottom part.
B ULLETIN
P ONTS
ET
C HAUSSES - 244-245 -
2003 -
REF .
4457 -
PP .
31-51
37
Fig. 5 - Stress computation for the kinematic characteristic of rotaton for a wall at the top [Ohde, 1938
1 h e' G Eaw Ea Q : Soil reaction along the fracture plane Ea : Thrust on the wall G : Soil block weight Qw Q
a) Analysis diagram.
Ql
The German school Laboratory tests carried out by Press [1942] or Lehmann [1942] have demonstrated, like the tests of Terzaghi [1934], that earth pressures largely depend on the wall kinematics. Lehmann focused mainly on the issue of stress distribution at the moment when the soil decompresses in the lower part of the soil block, which is the case for top rotation. For these tests carried out on sand, he used a closed box with sheets of glass on the sides in order to observe the behaviour of the soil block and the wall. The wall, which was 100cm high and 98cm wide, was divided into four rigid parts attached with hinges. In doing so, he could simulate top rotation as well as a succession of earth movements by turning one joint after the other. By measuring the stress in the springs used as struts, he was always able to trace an average line of stress distribution according to wall displacement, the centre of rotation, wall friction and parasite effects (if they were identified, like soil-glass friction) (Fig. 6a). From these tests, Lehmann formulated a semi-empirical method to calculate thrust strength using Coulombs method (his measurement results never exceeded this value) and to redistribute it using a line that encompasses all the results (Fig. 6c). In a second series of articles, Ohde [1948, 1949, 1950, 1951, 1952] revisited his ideas of the end of the 1930s [Ohde, 1938], and qualitative tests that demonstrate the different fracture planes according to the kinematic characteristic of the wall. He presented strength distribution figures in agreement with Lehmanns results indicated above.
38
B ULLETIN
P ONTS
ET
C HAUSSES - 244-245 -
2003 -
REF .
4457 -
PP .
31-51
Ei : Stress in the struts ei : Distributed pressure e : Relative pressure e = E / ( . b . ha) ha : Height of zone of influence for a strut b : Width of zone of Influence (b = 1 m)
: Soil unit weight h : Free height of wall F : Earth pressures according to Coulomb eL : Maximum trapezoidal pressure according to Lehmann
a) Strut stress E0 to E4 for a displacement of 14 mm. b) Test results with a rotation of level sections of the wall (max. = 12 mm). c) Proposals of thrust redistribution. Fig. 6 - Lehmanns tests on reduced models [1942].
The recent computation methods used in Germany were based on these first studies and additional studies done between 1960 and 1970. Briske [1958] proposed an initial synthesis of strength redistribution according to the type of retaining wall and the number of struts used for non-coherent soil (rectangular or trapezoidal redistribution). Subsequently, Mller-Haude and Scheibner [1965], Heeb et al. [1966], Briske and Pirlet [1968], Breth and Wanoschek [1969] and Petersen and Schmidt [1971] describe the work and measurements taken during the construction of underground railways in Berlin, Stuttgart, Cologne, Frankfurt am Main and Hamburg. For strutted interpile sheetings built on the coherent soil of Stuttgart, Heeb et al. [1966] propose a triangular distribution with most of the struts mid-way up the wall. This idea was later revisited by Breth and Wanoschek [1969] for the more rigid walls of bored piles in the clay of Frankfurt. Briske [1971] stressed the importance of where the first strut was situated, the effect of time, as well as the depth attained before the following struts are put in place. In Stuttgart (2 layers of struts) as in Frankfurt am Main (5 layers), the excavation below the second to last strut was relatively deep compared to the spacing between the layers. Consequently, after installation, there was a large amount of stress on the bottom struts and this even increased further with time because of creep. Since 1970, tension rods have been used more frequently and this led to a new series of laboratory tests. Schmitt and Breth [1975] first performed tests with one layer of tension rods. One year later, a second publication announced the results of tests using three layers of tension rods [Breth and Wolff, 1976]. Some time afterwards, Briske [1980] analysed the observation results in situ of tests carried out on anchored walls. Tests carried out on reduced models demonstrated that the amount and distribution of tension rods did not have a large influence on the earth pressures on the wall although the pressures decreased when the tension rods got longer. Upon comparing strutted and anchored walls, Breth and Wolff [1976] discovered that the strength concentration around the supports was less strong for anchored walls. In their opinion, the strength distribution behind strutted walls depended mainly on the way in which the excavation was dug while the advancing effect of excavation was less visible in the case of tension rods. They explained this observation to be the result of elasticity and the more symmetric vertical spacing of the anchorage system.
B ULLETIN
P ONTS
ET
C HAUSSES - 244-245 -
2003 -
REF .
4457 -
PP .
31-51
39
The EAB, the recommendations of the work group on excavation retaining walls, is the result of work carried out by several German researchers under the direction of J. Schmidbauer and A. Weissenbach. The publication of these technical rules started in 1970 and continued for a few years. They were published in the Die Bautechnik journal and gathered since 1980 in a work published by the editor Ernst & Sohn. They are well established in present day Germany and even considered to have legislative value. The recommendations address the following issues: