B C F T G - R S W: Technical Paper by S.C.R. Lo, J. Bosler and M. Gopalan

Technical Paper by S.C.R. Lo, J. Bosler and M.
Gopalan
BEARING CAPACITY FAILURE OF A TRAPEZOIDAL, GEOSYNTHETIC-REINFORCED SOIL WALL

ABSTRACT: Bearing capacity failure of geosynthetic-reinforced soil walls with a trapezoidal reinforced zone was studied analytically by a series of finite difference stress analyses. Non-associated flow rules and modelling of both construction and the path to collapse were adopted in the analyses. A range of factors, including wall height, soil parameters, and path to failure were studied. For each trapezoidal reinforced soil wall, its equivalent rectangular reinforced wall was also analysed in order to provide a benchmark for evaluation of simplified design models. The results of the analysis were used to calibrate three possible simplified design methods. The performances of these design models were evaluated by the margin of conservatism relative to their equivalent rectangular wall. KEYWORDS: Bearing capacity, Geosynthetics, Reinforced soil. AUTHORS: S.C.R. Lo, Associate Professor, School of Civil Engineering, University College, University of New South Wales, Canberra, ACT, 2600, Australia, Telephone: 61/2-62688349, Telefax: 61/2-2688337; E-mail: [email protected]; J. Bosler, Principal Geotechnical Engineer, Snowy Mountain Engineering Corporation, Cooma, Australia, Telephone: 61/2-64520203, Telefax: 61/2-64520400, E-mail: [email protected]; and M. Gopalan, formerly Research Officer, School of Civil Engineering, University College, University of New South Wales, Canberra, ACT, 2600, Australia, Telephone: 61/2-62688349, Telefax: 61/2-2688337, E-mail: [email protected]. PUBLICATION: Geosynthetics International is published by the Industrial Fabrics Association International, 1801 County Road B West, Roseville, Minnesota 55113-4061, USA, Telephone: 1/651-222-2508, Telefax: 1/651-631-9334. Geosynthetics International is registered under ISSN 1072-6349. DATES: Original manuscript received 25 May 1999, revised version received 28 September 1999 and accepted 1 October 1999. Discussion open until 1 July 2000. REFERENCE: Lo, S.C.R., Bosler, J. and Gopalan, M., 1999, Bearing Capacity Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall, Geosynthetics International, Vol. 6, No. 5, pp. 383-416.
GEOSYNTHETICS INTERNATIONAL
S 1999, VOL. 6, NO. 5
383
LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall
INTRODUCTION
A reinforced soil (RS) wall usually has a rectangular reinforced zone, and the majority of research on RS walls is based on this wall configuration. If the bearing pressure on the foundation soil requires reduction, the length of the reinforcement near the base is occasionally increased to form a stepped-back wall. For RS walls built on competent foundations, one may argue that the reinforcement length near the base may be reduced to form a trapezoidal reinforced soil (TRS) wall (Figure 1). A TRS wall may lead to savings in both reinforcement and select fill quantities. TRS walls are also desirable for stabilising existing embankments with steep-sided slopes as it eliminates or minimises excavation into an existing slope during construction. However, BS 8006 (1995) states that the use of TRS walls should only be considered where foundations are formed
Trapezoidal reinforced zone H
Sloping virtual back idealisation
Equivalent rectangular wall idealisation
Foundation level LB BER
Figure 1.
Trapezoidal reinforced soil (TRS) wall and the simplified dimensions.
384
S 1999, VOL. 6, NO. 5
by excavation into rock or when good foundation exist. This clause may be interpreted as there being a concern for the bearing capability of foundations because BS 8006 (1995) has already prescribed the design equations against bearing capacity failure. R57 (1996), the RS wall design specification of the Roads and Traffic Authority, NSW, Australia, limits the use of TRS walls to a rock-type foundation. Federal Highway Administration (FHWA) design guidelines (Christopher et al. 1989) require TRS walls to be designed with simplified design rules based on the equivalent rectangular reinforced soil wall and global stability analysis . performed using a reinforced soil program. The FHWA design guidelines (Christopher et al. 1989) define the global stability design procedure for a rectangular wall; it is not clear whether conducting a global stability analysis on the equivalent rectangular wall in lieu of the actual trapezoidal wall is considered acceptable. If the global stability analysis procedures need to be applied to an actual trapezoidal wall, a number of problems will be encountered. The (global) factor of safety against bearing capacity failure, Fb , is conventionally calculated as follows: Fb =
F B
z
qu
(1a)
where: qu = ultimate bearing capacity; Fz = summation of all vertical forces acting on the reinforced block; and Bi = effective bearing width of the foundation during bearing capacity collapse, which is given by: B = B 2e (1b)
where: B = width of the foundation; and e = eccentricity of the foundation reaction. The eccentricity is assumed to be positive (i.e. the wall is leaning outwards) or zero. The ultimate bearing capacity can be rationally computed from soil properties and foundation geometry as follows: q u = c Nc i c + p o N q iq + 1 B N i 2 (2)
where: Nc , Nq , and N = bearing capacity factors; ic , iq , and i = load inclination correction factors; c = soil cohesion; po = equivalent uniform surcharge pressure; and = unit weight of the soil. The expressions for these factors are given by Vesic (1975) and reproduced in the Appendix. Equation 2 neglects both shape and embedment correction factors, which are well justified simplifications for RS walls. The inclination correction factors, however, were retained in Equation 2 because, for RS walls, these factors can be significantly less than unity. Equations 1a and 1b follow the same principle as BS 8006 (1985) and R57 (1996), which express the design equations in the following limitstate format: q* u
F
B
*
* z
(3)
where * denotes factored values or values to be calculated with factored parameters: equality applies when the design requirement is just met. Equation 3 will reduce to
S 1999, VOL. 6, NO. 5
385
Equation 1a if all partial factors and the global factor of safety are set to unity. Equation 1a will be used in the current paper as it provides a single value for comparing analysis results. For a rectangular or stepped-back wall, B is unambiguously defined as equal to the reinforcement length at the base of the wall, LB . For a TRS wall, the reinforcement width at the wall base is shorter than the average wall width, and both B and Bi are not clearly defined. Furthermore, the active force acting on a trapezoidal reinforced zone and, hence, base eccentricity are not defined by conventional calculation procedures. Therefore, the application of Equations 1 to 3 for a TRS wall, unlike a rectangular wall, is problematic. The behaviour of a stable reinforced soil wall has been studied using finite element analysis by a number of researchers (Rowe and Ho 1997; Kapurapu and Bathurst 1995). Finite element analysis was also used by researchers, for example, Jaber et al. (1992) and Kapurapu and Bathurst (1992), to predict the internal failure of the Denver, Colorado, USA trial wall (Wu 1992). Internal instability of RS walls has also been studied using centrifuge testing (Bolton and Pang 1982; Jaber and Mitchell 1991). However, there is a paucity of information on the bearing capacity failure of TRS walls as evident from cases compiled by Zornberg et al. (1997). The current paper investigates, numerically, the predicted collapse of a TRS wall involving a bearing capacity mode of failure, with the primary objective of developing simplified design model(s). It is recognised that the construction logistics for a TRS wall may be more complicated if the reinforcing elements have to be prefabricated. But, such complications are not applicable to geosynthetic reinforcements that can be cut to required nonuniform lengths on site. Hence, the current study is restricted to trapezoidal, geosynthetic-reinforced soil walls. Although the use of finite element analysis in identifying collapse of structures has been reported (de Borst and Vermeer 1984; Frydman and Burd 1997), numerical problems for structures near collapse, especially for non-associated flow materials, still remain a major challenge. Alternatively, a finite different analysis based on an explicit iterative solution scheme may be used (Frydman and Burd 1997). This type of analysis is numerically more stable during collapse and, hence, was used in the current study. Three different simplified design methods for bearing capacity are proposed. It is most important to continue the analysis to wall collapse in a bearing capacity mode because these design methods are based on limit equilibrium and assume a collapse limit state. (For a highly nonlinear system, the behaviour under working conditions may not be the same behaviour at the incipient collapse state.) These methods were then evaluated using a finite difference stress analysis of a series of TRS walls.
METHODOLOGY
To perform a bearing capacity check using Equations 1a, 1b, and 2, a trapezoidal wall must be idealised into a simple wall configuration using one of the following methods.
S Method A. The TRS wall is idealised as an equivalent rectangular (ER) wall as defined in the FHWA design guidelines based on an equivalent reinforced area (Figure 1). The width of the ER wall, BER , is defined by BER H = area of the trapezoidal reinforced zone. This method will often yield B > LB , where B is the effective ER ER bearing width of the ER wall foundation and B = BER - e. ER
386
S 1999, VOL. 6, NO. 5
S Method B. The TRS wall is again idealised as an ER wall, but the effective width
to be used in Equations 1a and 2 is calculated by Bi = Min{ B , LB }. ER
S Method C. The TRS wall is idealised as a wall with a sloping back (Figure 1). This
means B is unambiguously given by LB and the active force acting on the idealised wall is reduced. This idealisation may yield a negative eccentricity (i.e. wall leaning toward retained fill); in such a case, Bi will be set to LB . The calculated Fb values will generally be different because the above three methods simplify a TRS wall geometry using different procedures. It is important to emphasise that these idealisations are only for the purpose of calculating the factor of safety, Fb , against bearing capacity failure using the simple Equations 1a, 1b, and 2. To calibrate the three design methods, a number of TRS walls were analysed with a finite difference stress analysis program called FLAC: Fast Lagrangian Analysis of Continua (FLAC 1996). In FLAC, the equations of equilibrium and stress-strain behaviour are expressed in finite difference form and are solved by an explicit iterative scheme. The analysis is inherently incremental and can be taken to collapse along a path prescribed by the user. The numerical accuracy of the solution can be controlled by either specifying the number of iterations (referred to as steps in FLAC) per increment, or by continuing the iterations until the specified tolerance is achieved. In the current paper, such an analysis will be referred to as a FLAC analysis. The program includes looping control commands and an internal programming language for implementing user-defined procedures. The numerical principles of a FLAC analysis are detailed in Cundall and Board (1988). Each FLAC analysis in the current study modelled wall construction, working condition behaviour, near-collapse behaviour, and collapse behaviour. FLAC analyses have been successfully used in studying the behaviour of innovative soil retaining systems (Lorig 1991), dynamic response of geosynthetic RS walls (Bathurst and Hatami 1998), and N -bearing capacity factors (Frydman and Burd 1997). It is also important to evaluate the behaviour predicted by the FLAC analysis relative to that of conventional rectangular RS walls. Therefore, for every TRS wall studied in the current paper, its equivalent rectangular wall as defined by the FHWA design guidelines (Christopher et al. 1989) was also analysed.
3 3.1
ANALYSES Wall Configuration
For the analysis to be representative of a scenario where bearing capacity is a design concern, the trapezoidal walls have to satisfy the following requirements:
S The foundation soil is competent (otherwise a TRS wall will not be used). S The factor of safety against bearing capacity failure is in the range of two to three,
depending on which method is used in calculating Fb . This range of values corresponds to the minimum value required or inferred by most design codes. S Premature internal instability is avoided. These requirements limit the range of wall configurations. A benchmark wall configuration as shown in Figure 2 was adopted. The benchmark wall is 8 m high, which repre-
S 1999, VOL. 6, NO. 5
387
10 reinforcement layers @ 0.8 m vertical spacing Select fill Wall facing
General fill
8.0 m
Ground level
0.4 m 3.2 m
Foundation level
0.8 m 0.8 m 0.8 m
7.0 m Foundation soil
Figure 2.
Schematic of the benchmark TRS wall.
sents a large-sized geosynthetic RS wall. Most of the analyses were based on this wall. The value of LB was set to the minimum value of 0.4H as recommended by BS 8006 (1995). The reinforcement length at and near the top of the wall was 0.7H. The width of the equivalent rectangular (ER) wall as defined in the FHWA design guidelines, BER , was 4.64 m. The bulk unit weight of the backfill was assumed to be 20 kN/m3. The strength parameters of the fill materials and foundation soil were: General fill: Select fill: Foundation soil: = 30, c = 0 = 40, c = 0 f = 30, c f = 10 kPa
where: = internal friction angle of the soil; c = soil cohesion; and the subscript f denotes foundation soil. The foundation soil parameters were chosen to provide the relevant range of factors of safety. The chosen soil parameters are reasonable for a large-
388
S 1999, VOL. 6, NO. 5
sized geosynthetic RS wall built on relatively competent soil. The water table under working conditions was assumed to be at a depth exceeding 7 m. The factors of safety against bearing capacity failure, calculated based on the three simplified methods, are denoted as Fb (x), where: F b(A) = 3.28 F b(B) = 2.50 F b(C) = 3.30 Note that a sloping-back idealisation gave a negative eccentricity of -0.197 m; and, in such a case, Fb (C) was calculated by setting Bi= LB . The full method of calculation is given in the Appendix. 3.2 Numerical Analyses
It is important to re-iterate that the actual wall configuration was used in the FLAC analyses, not the simplified one assumed by the design model. The grid used in the FLAC analyses is shown in Figure 3. Wall construction was modelled in a layer-by-layer manner as illustrated in Figure 4. Both the left and right vertical boundaries were allowed to move in the vertical direction, i.e. as vertical rollers. The bottom boundary was modelled as fixed. The ground level in front of the wall was 0.4 m above the foundation level. This extra 0.4 m of soil was modelled as a uniform surcharge of 8 kPa. Incremental construction without using any temporary lateral props was assumed in the analyses. All of the soil layers were modelled as elastic-plastic, two-dimensional (2-D) elements (referred to as zones in FLAC) with the Mohr-Coulomb failure criterion as the yield surface. The Youngs moduli and Poissons ratio of the 2-D elements are listed in Table 1. These values are comparable to those assumed by Rowe and Ho (1997). A non-associated flow rule with a zero dilatancy angle was used. It is recognised that the dilatancy angle is related to the difference in friction angles between the peak and critical void ratio states (Bolton 1986); thus, a zero dilatancy angle is a slightly conservative assumption for the select fill. The facing panels were modelled as beam elements with low flexural stiffness in order to give a flexible facing condition. These beam elements were also assigned a low axial stiffness to model the presence of compressible strips between the panels. Moment releases were inserted at every panel height; this implied the use of shear keys or similar interlocking details between the facing panels, which is considered to be appropriate for most permanent walls. Interface elements were inserted between the wall panels and 2-D soil elements. The interface friction angle was assumed to be the same as that of the select fill because the back of the wall panels usually has a rough finish. The soil reinforcement was modelled as elastic bar elements (referred to as cables in FLAC) with an axial stiffness of 1,450 kN/m. Such a stiffness value is considered to be representative of the long-term stiffness of high-strength geosynthetics and similar in magnitude to values used by Rowe and Ho (1997). The bar elements were given a high yield strength to avoid premature failures by reinforcement rupture. The bond strengths between the reinforcement and the select fill were assigned based on: (i) the design equations prescribed in R57 (1996) for polyester straps; and (ii) a strap width of 90 mm with attachment points at a horizontal spacing of 0.8 m. These bond
S 1999, VOL. 6, NO. 5
389
15.00
10.00
Origin (0,0)
5.00
0.00
---5.00
---10.00
---15.00
---12.50
---7.50
---2.50
2.50
7.50
12.50
17.50
Horizontal distance, x (m)

Figure 3. Finite difference grid used in FLAC for the benchmark wall analyses.
strengths were based on large-scale pullout testing (Lo 1998) and, thus, were average mobilised values during reinforcement pullout. In theory, element values are more appropriate for a FLAC analysis. Furthermore, polyester straps generally have a more severe pullout situation compared to geogrids because of the small interface area and concentration of reinforcement force. Thus, the bond strengths assigned in the FLAC analyses were reasonable lower bound estimates. However, the factors of safety against pullout, calculated as per the FHWA design guidelines (Christopher et al. 1989), were in the range of 3.75 to 4.75. Hence, there was no concern of premature failure due to reinforcement pullout.
390
S 1999, VOL. 6, NO. 5
Vertical distance, y (m)
(a)
Wall facing alignment Foundation level
(b)
8 kPa surcharge
(c)
8 kPa surcharge
Vertical roller
0.25 0.25 0.25 0.20 0.20
7m
15 m
Fixed bottom boundary 18 m
Figure 4. FLAC modelling of the wall construction sequence: (a) initial; (b) the first reinforcement layer; (c) the second reinforcement layer.
Table 1.
Elastic soil properties.

Material E (MPa) 30 30 30 50 100
General fill Select fill Foundation soil: a) 0.0 to 1.0 m depth b) 1.0 to 2.0 m depth c) 2.0 to 7.0 m depth Note: E = Youngs modulus, = Poissons ratio.
S 1999, VOL. 6, NO. 5
391
A typical FLAC analysis consisted of the following three stages:
S Stage I (working condition): The wall was constructed in a layer-by-layer manner

(Figure 4). The gravity load of each layer was turned on in 10 increments. S Stage II (near-collapse condition): The foundation parameters were reduced such that Fb (B) = 1.1 S Stage III (collapse condition): The foundation soil parameters were further reduced until a collapse condition was detected in the FLAC analysis. Completion of Stage I was referred to as the working condition because it was a conservative representation of the service condition. Upon completion of Stage I, the analysis was then switched (from small deformation) to a large deformation mode. Completion of Stage II was referred to as the near-collapse condition and was not defined by actual wall behaviour. It is a reference state with a small margin of safety against bearing capacity failure and was used to track wall behaviour as collapse was approached. Simplified design Method B, which gave the lowest Fb value, was used as the criterion to ensure that Stage II occurred before collapse was detected by FLAC. It is recognised that the combination of foundation parameters to achieve Fb (B) = 1.1 is not unique. It was achieved by reducing the foundation parameters to tan * and c * f f on a pro rata basis and applying a water table to the foundation level. Because a number of cases were analysed, these reduced foundation parameters were simply a convenient and tenable combination that served as a standardised point to track the behaviour change as the wall approached bearing capacity collapse. Completion of Stage III (i.e. collapse) was achieved by further reducing the foundation parameters in small increments of 0.5_ for * or 0.5 kPa for c * . Collapse was def f tected when proper convergence could not be achieved. The numerical strategy was to solve the equations iteratively to a specified tolerance (in terms of the maximum out-ofbalance force). Improper convergence was identified by run-off increase in either the number of iterations or wall displacement. The latter indicator always gave the same run-off state irrespective of whether the vector length of the maximum wall displacement, max , or maximum lateral wall displacement, x , or settlement near the toe of the wall, y , was used as a measure of wall displacement. If increases in run-off for both the number of iterations and wall displacement were detected, they occurred at the same state. Although the analysis could still proceed after an increase in run-off for either the number of iterations or wall displacement, a further reduction of 0.5 to 1_ for * (or f 1.0 kPa for c * ) would lead to a condition of false convergence. This could be checked f by prescribing additional iterations without any change in soil or reinforcement parameters. If the convergence was false, the unbalanced force would not reduce with additional iterations (under a constant state condition), but would oscillate to higher values. The run-off increase for wall displacement could also continue with these additional iterations until a bad geometry error occurred. The analysis could not proceed once false convergence occurred; thus, the methodology for detecting improper convergence is appropriate and has an overall uncertainty of less than 1.5_ for * or 1.5 kPa for c * . f f For each TRS wall analysis, a corresponding analysis (following the same stage and loading path) for its equivalent rectangular (ER) wall, as defined in the FHWA design guidelines (Christopher et al. 1989), was also conducted. Therefore, a reference for evaluating the performance of the TRS wall, as discussed in Section 3.3.3, was also established.
392
S 1999, VOL. 6, NO. 5
3.3
Results of the Benchmark Wall Analyses
3.3.1 Working Condition The displacement pattern of the TRS wall is presented in Figure 5. The plotted displacement vectors are total values; therefore, the actual post-construction displacements can be smaller because some of the displacements can be taken out during construction. The kinematics is a combination of overall translation, tilting near the foundation level, and bulging along the wall facing. This deformation mode can also be seen in the horizontal displacement profile of the wall facing (Figure 6). The horizontal displacement profile of the ER wall is also plotted in Figure 6. Both displacement profiles show a pattern of wall tilting. This implies that the distribution of reinforcement length of a TRS wall is more effective in limiting wall tilt; therefore, it would be conser-
Working condition displacement vectors 0 2 10-1 Maximum vector = 1.150 10-1

Figure 5. FLAC-generated displacement vectors for the benchmark wall during working conditions.
S 1999, VOL. 6, NO. 5
393
Elevation (m)
TRS wall: working condition TRS wall: near collapse ER wall: working condition
Horizontal displacement at the wall facing, x (mm)

Figure 6. Horizontal displacement of the 8 m-high wall facing.
vative to assess the maximum lateral displacement of a TRS wall by applying existing empirical rules to the corresponding ER wall. The settlement profile at the foundation level is presented in Figure 7, which shows the foundation tilt over the length LB . The settlement profile of the ER wall is also plotted in Figure 7; foundation tilt, similar to that of the TRS wall, is clearly evident. The settlement profiles of both the TRS and ER walls are different from the concave shape resulting from an elastic analysis (Figure 7). This is due to the occurrence of plastic (yield) zones, despite the fact that factors of safety during the working stage are adequate. It was not possible to test the validity of the three proposed methods for bearing capacity design based on the FLAC analysis results for a working condition because the
394
S 1999, VOL. 6, NO. 5
Distance from wall facing (m)
Elastic analysis
Settlement (mm)
Elastic-plastic analysis
LB BER
Figure 7.
Foundation settlement of the 8 m-high wall.
results predicted using the bearing capacity equation, i.e. Equation 2, correspond to a collapse state. However, the distribution of foundation pressure, yy , along the wall foundation (Figure 8) also indicates a forward-tilting foundation. The foundation pressure along LB is approximately trapezoidal and has an eccentricity of approximately 0.2 m. A jump in yy occurs at a distance LB and minor oscillations in yy also occur between LB and BER . Both features are indicative of a forward-tilting reinforced zone. However, Method C, as detailed in the Appendix, implies a negative eccentricity of -0.197 m (i.e. leaning back toward the retained fill), which is inconsistent with the kinematics implied by the FLAC analysis. The foundation pressure for the ER wall is also shown in Figure 8 and is similar to that of the TRS wall, except the trapezoidal distribution extends over a longer length BER . 3.3.2 Near-Collapse Condition A near-collapse condition is defined by Fb (B) = 1.1 in the current study, where Fb is calculated using Equations 1a, 1b, and 2 based on Method B. This condition was achieved at a state defined by:
S 1999, VOL. 6, NO. 5
395
yy (kPa)
LB
BER

Figure 8. Foundation pressure for the 8 m-high wall.
S * = 26_ and c * = 8.45 kPa, where the superscript * denotes a reduced value; f f S the water table at the foundation level (referred to as flooding); and S other parameters identical to the normal working condition.
The calculations demonstrating that Fb (B) = 1.1 during such a state are described in the Appendix. The effect of the water table at the foundation level can be modelled by reducing the bulk density to buoyant (effective) density because the analyses did not consider transient seepage. Progressive reduction of the unit weight to the buoyant unit weight is mathematically equivalent to specifying a progressive increase in (hydrostatic) pore water pressure as there is no unbalanced pore pressure in the horizontal direction. Proper convergence was always achieved and, therefore, the wall must have been stable at Fb (B) = 1.1, which implies that Method B is unlikely to have any significant unsafe error.
396
S 1999, VOL. 6, NO. 5
The displacement pattern of the TRS wall during the near-collapse condition is presented in Figure 9. Once again, the kinematics is a combination of overall translation, tilting at the foundation level, and lateral displacement (i.e. bulging). The lateral displacement profile of the wall facing (Figure 6) appears to be the result of a forward translation of that manifested during the working condition. The tilting at the foundation level is higher than that of the working condition as evident by comparing Figures 7 and 9; however, the distribution of foundation pressure (Figure 8), is close to that of the working condition. This can be explained by the onset of significant plasticity in the foundation zone. The resultant of the foundation reaction on LB again has a positive eccentricity, which is contrary to that implied by Method C. 3.3.3 Collapse Condition A FLAC analysis can proceed until collapse, unless the analysis results unambiguously shows that proper convergence to equilibrium has not been achieved. Hence, after a near-collapse condition (Stage II) was achieved, the analysis was continued by pro-
Near-collapse condition
Displacement vectors Maximum vector = 1.30810 -1
2 10 -1

Figure 9. FLAC-generated displacement vectors for the benchmark wall during the near-collapse condition.
S 1999, VOL. 6, NO. 5
397
gressively reducing the friction angle of the foundation material, until collapse was detected. Proper convergence is achieved for * 24_, thus, indicating stability (Figure f 10). However, when * is stepped from 23.5 to 23_, the maximum wall displacement, f max , and the number of iterations required to achieve the specified tolerance suddenly and significantly increases. This is clearly indicative of imminent collapse. The FLAC analysis could not proceed when * was reduced to less than 22.5_; therefore, the colf lapse state was assumed to exist at * = 23.5_. The values of the other parameters were f identical to those for the near-collapse condition as given in Section 3.3.2. This procedure of identifying a collapse state in a FLAC analysis may be marginally conservative, but it enables an examination of behaviour during the collapse condition. The wall displacement pattern during this condition indicates that the significant increase in wall displacement was largely due to foundation movements, thus, causing the reinforced zone to move as a coherent mass. The distribution of foundation pressure during col-
1.00 Iterations Wall displacement
80,000
Wall displacement (m)
0.50
60,000
Collapse 0.00 22 23 24 * (_) f

Figure 10. Detection of imminent wall collapse for the 8 m-high TRS wall.
50,000 25 26
398
S 1999, VOL. 6, NO. 5
Number of iterations
70,000
lapse is presented in Figure 11. The foundation pressure may be approximated as uniformly distributed over the length of the bottom layer of reinforcement, LB . A design approach can be applied to the collapse state determined by a FLAC analysis. The calculated factor of safety of unity can be obtained only if the design approach is in exact agreement with the FLAC analysis results. Three methods are proposed for calculating the factor of safety against bearing capacity failure of a TRS wall. Therefore, the collapse state of a benchmark TRS wall has the following three method-dependent Fb values for bearing capacity failure: F b(A) = 1.085 F b(B) = 0.850 F b(C) = 1.120 Both Fb (A) and Fb (C) were slightly in excess of unity despite the wall being at a collapse state; thus, these two methods are slightly unsafe relative to the results of the FLAC analysis. Fb (B) was significantly less than unity and, therefore, Method B has a fair mar-
LT
Average foundation pressure
yy (kPa)
LB
Reduced f value ( * ) f Reduced cf value ( c * ) f

Figure 11. Distribution of foundation pressure for the benchmark TRS wall at collapse.
S 1999, VOL. 6, NO. 5
399
gin of conservatism. Initially, Method A may appear to be superior to Method B because Fb (A) is closer to unity; however, it is important to assess the margin of conservatism relative to that of a conventional rectangular RS wall. Hence, the collapse state of the ER wall was also determined using a FLAC analysis. The analysis proceeded according to the following stages: the working condition with f = 30_ and cf = 0; the near-collapse condition with the water table at the foundation level, and * = 24_ and c * = 7.7 kPa f f as defined by a factor of safety of 1.1 for the ER wall; and the progressive reduction of * to collapse. Following the same algorithm as the TRS wall, collapse was detected f at * = 21.5_, c * = 7.7 kPa, and the water table at the foundation level. This state had f f Fb (ER) = 0.845, where ER denotes an equivalent rectangular wall. Thus, the application of Method B to a TRS wall gave a margin of conservatism similar to a conventional ER wall. The choice of design method will be further discussed in Section 7. 3.4 Alternative Paths to Collapse
The collapse state may not be unique, i.e. it may be load path dependent, because a non-associated flow rule was used in the FLAC analysis. Therefore, different alternative load paths to collapse were also studied. The first alternative path was achieved by progressively reducing the value of cf , with other parameter values identical to the Stage II values (Figure 12). This algorithm leads to a run-off increase in the number of iterations at c * = 4.5 kPa . A similar run-off f increase in wall displacement, max , was also observed at this point. The analysis could not proceed when c * was reduced to 3.0 kPa; hence, collapse was assumed to occur at f c * = 4.5 kPa, with the other parameter values similar to the Stage II values. During this f state, the distribution of foundation pressures over LB was essentially uniform, i.e. similar to the collapse state achieved by a reduction of * as evident from the comparison f shown in Figure 11. The application of the bearing capacity design equations to this collapse state (as defined by a FLAC analysis) gave the following factors of safety against bearing capacity failure: F b(A) = 1.144 F b(B) = 0.884 F b(C) = 1.170 The ER wall was also taken to collapse along this load path by progressively reducing the value of c * . During the collapse determined by the FLAC analysis, which had a f different set of reduced strength parameters of * = 24_ and c * = 3.5 kPa, the Fb (ER) f f was 0.849. The second alternative path was achieved by reducing the foundation stiffness during flooding. It is reasonable and tenable to assume that foundation stiffness will have a minimal effect on collapse in a bearing capacity mode; therefore, in the presented analyses, the foundation stiffness during flooding was not reduced. However, in a nonlinear system, a change in stiffness will lead to a change in stress path; thus, an additional FLAC analysis was conducted whereby the foundation stiffness was reduced during flooding in accordance with the following equation:
400
S 1999, VOL. 6, NO. 5
c * (kPa) f 4 100,000 5 6 7 8 9 10
90,000 Collapse by reducing f ( * ) f Number of iterations Benchmark wall 80,000 Foundation stiffness reduced with flood
70,000
f = 23.5_ 60,000 Collapse by reducing cf ( c * ) f 50,000 22 23 24 * (_) f

Figure 12. Influence of the path on collapse for the 8 m-high TRS wall.
25
26
New stiffness = Original stiffness
Buoyant unit weight Bulk unit weight
(4)
This analysis gave a run-off increase in the number of iterations when * is stepped f from 23.5 to 23_ (Figure 12), i.e. at exactly the same values as the benchmark wall. A similar run-off increase in wall displacement was also observed. The analysis could not proceed when * was reduced to 22.5_; hence, the collapse state detected is identical f
S 1999, VOL. 6, NO. 5
401
to the benchmark wall analysis, conforming to expectation. The ER wall was also taken to collapse along this load path, i.e. with a reduction of foundation stiffness during flooding. The collapse state, so determined, was also identical to that of the ER benchmark wall.
4 4.1
INFLUENCE OF REINFORCEMENT PARAMETERS Reinforcement Stiffness
The possible influence of reinforcement stiffness on bearing capacity failure was studied by repeating the FLAC analysis on the benchmark wall geometry, but with the following two different reinforcement stiffness patterns:
S a reduced axial stiffness value of 650 kN/m for all reinforcement layers; and S an axial stiffness of 3,000 kN/m for the top reinforcement layer and the axial stiffness
increasing linearly with depth to 6,000 kN/m for the bottom reinforcement layer. The reinforcement stiffness of the latter case was considerably higher than what is commonly used in the construction of geosynthetic-reinforced soil walls. The lower reinforcement stiffness led to an increase in horizontal displacement and more significant lateral wall displacement along the face of the wall, whereas the higher reinforcement stiffness led to a reduction in lateral wall displacement (Figure 13). However, the changes in reinforcement stiffness only has a slight affect on foundation pressure as evident from the comparison presented in Figures 14a and 14b. The analysis with higher reinforcement stiffness has a run-off in the number of iterations occurring between 24 to 23_ (Figure 15a), but the wall displacement, max , unambiguously increases at 23.5_ (Figure 15b). Therefore, the collapse state was considered to occur at * =23.5_, which f is identical to that of the benchmark wall with a reinforcement stiffness of 1,450 kN/m. The case of reduced reinforcement stiffness gives a run-off increase in the number of iterations and displacement when * = 23_ (Figures 15a and 15b), which occurs after f increasing the * value by only 0.5_. This increase is within the numerical uncertainf ties in detecting collapse. The analysis of both reinforcement patterns, i.e. high and low stiffness reinforcements, could not proceed when * was reduced to 22_; hence, the f three design methods are not affected by considerable variations in reinforcement stiffness. It is hypothesised that this occurs because, for the range of reinforcement stiffness studied and in conjunction with the bond strength assigned, the reinforced zone was effectively bound into a coherent mass in the context of bearing capacity failure. 4.2 Interface Strength of Reinforcement
The bond strengths assigned in the analysis were lower estimates relative to a range of representative values; therefore, an additional FLAC analysis was conducted with the values of the bond strengths doubled. All other input parameters were identical to that of the benchmark wall. The increase in bond strength has no detectable affect on lateral wall displacement during the working condition (Figure 13). A run-off increase in wall displacement occurs when * is stepped from 23 to 22.5_, which is 0.5_ less f than the benchmark wall; however, a similar run-off in the number of iterations is not
402
S 1999, VOL. 6, NO. 5
Working condition
Elevation (m)
Benchmark wall Low reinforcement stiffness High reinforcement stiffness High reinforcement bond strength Soil stiffness variation
Horizontal displacement at the wall facing (mm)

Figure 13. Influence of soil and reinforcement parameters on lateral displacement of the 8 m-high TRS wall.
as obvious (Figure 15b). The analysis could not proceed when * was reduced to 22_. f This means a slight increase in bearing capacity may result from a considerable increase in bond strength, and the findings based on the benchmark wall analysis (using lower bound estimates of bond strength) may contain a small conservative error. Therefore, no further study was carried out.
INFLUENCE OF THE FILL SOIL PARAMETERS
The soil parameters used in the analyses are considered to be reasonably representative of the average value for the elastic-plastic model and are consistent with values reported in previous work (Lo et al. 1996; Rowe and Ho 1997). Furthermore, the current paper is a bearing capacity study, and it is highly unlikely that fill stiffness can have a significant effect on foundation collapse. To verify that fill stiffness (and its variation with depth) has a negligible influence on collapse in a bearing capacity mode, an addi-
S 1999, VOL. 6, NO. 5
403
Near-collapse condition
Average foundation pressure yy (kPa)
LB
LB
Benchmark wall Low reinforcement stiffness High reinforcement stiffness

Figure 14. TRS wall. Influence of reinforcement stiffness on the foundation pressure for the 8 m-high
tional analysis was conducted by linearly increasing the Youngs modulus of the fill from 10 MPa at the surface to a maximum value of 50 MPa at the maximum depth of 8 m. A Youngs modulus of 10 MPa at the surface was adopted to avoid possible numerical problems due to an excessively low Youngs modulus. This stiffness pattern had an average Youngs modulus of 30 MPa, which is identical to the benchmark wall analysis. Lateral wall displacements during the working condition were essentially identical to the benchmark wall analysis (Figure 13). The detected collapse state was identical to the benchmark wall as indicated by a run-off increase in the number of iterations and wall displacement when * was stepped from 23.5 to 23_ (Figures 15a and 15b). The f analysis could not proceed when * was reduced to 22.5_; thus, the detected collapse f state was identical to the benchmark wall analysis.
404
S 1999, VOL. 6, NO. 5
(a) 100,000
* (_) f 22 23 24 25 26
90,000 Number of iterations (b)
80,000
70,000
60,000
50,000
Benchmark wall Soil stiffness increases with depth High reinforcement stiffness Low reinforcement stiffness High reinforcement bond strength
Wall displacement, max (m)
0.50
0.00
22
23
24 * (_) f
25
26
Figure 15. Influence of soil and reinforcement parameters on the collapse of the 8 m-high wall: (a) number of iterations; (b) maximum wall displacement.
S 1999, VOL. 6, NO. 5
405
INFLUENCE OF WALL CONFIGURATION
The possible influence of wall configuration on the findings in Section 3 was examined by repeating some of the analyses using a 5 m-high TRS wall. This wall configuration (Figure 16) has dimensions similar to the minimum requirements of BS 8006 (1995). The equivalent rectangular (ER) wall has a width to height ratio of 0.646, i.e. BER = 3.235 m). The boundary conditions and distances to boundaries, fill properties, reinforcement properties, and facing panel construction were identical to those of the benchmark wall. However, the strength parameters of the foundation soil (at working condition) were changed to f = 25_ and cf = 10 kPa, such that calculated Fb values would be similar to that of the benchmark wall. The Youngs modulus values for the FLAC analysis were reduced in proportion to tanf . The analysis of the wall followed the same procedure as that of the benchmark wall. After simulating the construction of the wall to the working condition (Stage I), the analysis proceeded to a near-collapse condition (Stage II) by reducing the foundation parameters to * = 22_, c * = 6.4 kPa and applying a water table at the foundation level. f f
6 reinforcement layers @ 0.8 m vertical spacing Select fill Wall facing 5.0 m General fill Ground level
0.4 m 2.4 m
Foundation level
0.8 m 0.8 m
7.0 m Foundation soil
15 m
18 m
Figure 16.
Schematic of the 5 m-high TRS wall.
406
S 1999, VOL. 6, NO. 5
The analysis reached collapse by progressively reducing the value of * and reducing f c * to a minimum value 6.4 kPa. Imminent collapse (as indicated by run-off increase f in both wall displacement, max , and the number of iterations) was detected when * f = 18.5_. During this state, the factors of safety against bearing capacity failure were: F b(A) = 0.925 F b(B) = 0.780 F b(C) = 0.993 A FLAC analysis was also conducted on a 5 m-high ER wall. Following the same algorithm for the TRS wall, imminent collapse was detected at * = 16.5_, c * = 6.4 f f kPa, and with the water table at the foundation level. This resulted in Fb (ER) = 0.745, which was close in value to Fb (B). Therefore, Method B for a 5 m-high TRS wall gave a margin of conservatism close to the conventional design of an ER wall. The kinematics of the 5 m-high wall during Stage I (working condition) and Stage II (near-collapse condition defined by Fb (B) = 1.1) are also similar to that of the benchmark wall. During the working condition, both the lateral displacement of the wall facing, x , (Figure 17a) and the foundation settlement (Figure 17b) indicate a displacement pattern consisting of tilting, translation, and wall bulging. The foundation pressure distributions for all three stages are presented in Figure 18. The eccentricities for all three distributions were positive, i.e. toward the wall facing. These observations do not support Method C, which implies a zero or negative eccentricity (toward the retained soil). The influence of an alternative path to collapse was also investigated using the 5 mhigh wall. Upon completion of Stage II, the cohesion of the foundation material was progressively reduced until collapse was detected at c * = 3 kPa. The corresponding calf culated factors of safety against bearing capacity were: F b(A) = 1.003 F b(B) = 0.826 F b(C) = 1.061 The 5 m-high ER wall was also taken to collapse along this alternative path in a FLAC analysis; the corresponding Fb value was 0.810. Therefore, the influence of the alternative path to collapse is minimal. A FLAC analysis was also conducted on a 5 m-high wall identical to Figure 16, but with stronger foundation parameters of f = 30_ and cf = 10 kPa (during the working condition). This wall had a high margin of safety against bearing capacity failure and generated significantly less displacement. However, the FLAC analysis results were the same: during the collapse detected by the FLAC analysis, Method A and Method C resulted in Fb values slightly in excess of unity, whereas Method B resulted in an Fb value less than unity and close to that of an ER wall.
S 1999, VOL. 6, NO. 5
407
(a)
Elevation (m) Horizontal displacement at the wall facing, x (mm) (b) Distance from wall facing (m)
LT
Settlement (mm)
LB
Figure 17. Displacement of the 5 m-high TRS wall during the working condition: (a) horizontal displacement at the wall facing; (b) settlement at the foundation level.
408
S 1999, VOL. 6, NO. 5
150
LT
100 yy (kPa)
Average foundation pressure
50 LB 0 0 5 Distance from wall facing (m)

Figure 18. Distribution of foundation pressure for the 5 m-high TRS wall. Working condition Near-collapse condition Collapse condition
10
DISCUSSION
The analyses presented in Section 6 assume bearing capacity failure would not be triggered by premature reinforcement rupture. This assumption must be considered by the design procedure to prevent this failure from occurring. Therefore, the design procedure for calculating the maximum reinforcement tension at the ith level, Tmax (i), also needs to be investigated. The distribution of the Tmax (i) values for the benchmark TRS wall are compared with those of the ER wall in Figure 19. The working condition (Stage I), near-collapse condition (Stage II), and collapse condition (Stage III) are included in the comparison. The difference in Tmax between the TRS wall and the ER wall is slight; however, a slight increase in Tmax at and near the wall base is observed as the wall state changes from working to collapse. This observation is common to both TRS and ER walls and was likely caused by a loss of foundation restraint. The distributions of Tmax for the 5 m-high wall were also compared and a similar result was obtained, these results imply that, for the purpose of design against reinforcement rupture, the maximum reinforcement for a TRS wall can be calculated based on its corresponding ER wall. For the cases discussed in the current paper, the Fb values corresponding to the collapse states, as determined by FLAC analyses, are summarised in Table 2. Cases corresponding to variations in reinforcement and fill parameters are not recorded in Table 2 because these parametric variations only resulted in small, if any, changes in the collapse state. It is evident from Table 2 that, during the collapse state as detected by the FLAC analyses, with the exception of one case, Method A and Method C consistently gave Fb > 1. Hence, these two design methods may be interpreted as unsafe, although
S 1999, VOL. 6, NO. 5
409
6 Elevation (m)
ER wall, working condition ER wall, near-collapse condition ER wall, collapse condition TRS wall, working condition TRS wall, near-collapse condition TRS wall, collapse condition
0 0 10 20 30 40 Maximum reinforcement tension, Tmax (kN/m)

Figure 19. Comparison of reinforcement tension during different conditions for the 8 m-high wall.
only by a small margin. Furthermore, Method C results suggest that the foundation reaction along LB had a negative eccentricity, i.e. leaning backward toward the retained soil, which is contradictory to the findings of the FLAC analysis. Method B consistently gave Fb values between 0.850 to 0.907 and, hence, this method had a reasonable margin of conservatism relative to the FLAC analysis results. The margin of conservatism inferred by the FLAC analyses of TRS walls should be similar to that of a conventional rectangular wall because the FLAC analysis has not yet been calibrated against experimental results. Therefore, the performance of the simplified Method B design can be measured using Fb (B) - Fb (ER), whereby a small value indicates good performance. For all of the analysed cases, Fb (B) - Fb (ER) was less than 0.04; thus, one may argue that the preferred design method should be Method B.
410
S 1999, VOL. 6, NO. 5
Table 2.
Factors of safety against bearing capacity failure based on FLAC analyses.

Factor of safety against bearing capacity failure for the TRS wall Method A Method B Method C
Load path to collapse
ER wall
8m high wall: a) Reduction of f b) Reduction of cf c) Reduction of f and foundation stiffness reduced during flooding 5m high wall: a) Reduction of f b) Reduction of cf
1.085 1.144 1.085
0.850 0.844 0.850
1.120 1.170 1.120
0.845 0.849 0.845
1.935 1.003
0.780 0.826
0.993 1.061
0.745 0.810
CONCLUSIONS
FLAC analyses were successfully used to analyse bearing capacity collapse of trapezoidal, geosynthetic-reinforced soil walls. According to the FLAC analyses, the possible influence of reinforcement stiffness, load path to collapse, reinforcement parameters, fill parameters, and wall height on the collapse state, if any, was found to be very small. Three simplified design methods against bearing capacity failure were examined using FLAC analyses. The margin of conservatism of Method B as applied to TRS walls appears to be consistent with that of conventional RS walls. As such, Method B is considered to be a more suitable simplified design method.
ACKNOWLEDGEMENT Part of the research reported in this paper was conducted when the first author was on sabbatical leave at the Hong Kong University of Science and Technology.
REFERENCES BS 8006, 1995, Strengthened/reinforced soils and other fills, British Standards Institution, 161 p. Bathurst, R.J. and Hatami, K., 1998, Seismic Response Analysis of a GeosyntheticReinforced Soil Retaining Wall, Geosynthetics International, Vol. 5, No. 1, pp. 127-166. Bolton, M.D., 1986, The Strength and Dilatancy of Sands, Geotechnique, Vol. 36, No. 1, pp. 65-78. Bolton, M.D. and Lau, C.K. ,1993, Vertical Bearing Capacity Factors for Circular and Strip Footings on Mohr Coulomb Soil, Canadian Geotechnical Journal, Vol. 30, No. 4, pp. 1024-1033.
S 1999, VOL. 6, NO. 5
411
Bolton, M.D. and Pang, P.L.R., 1982, Collapse Limit States of Reinforced Earth Retaining Walls, Geotechnique, Vol. 32, No. 4, pp. 349-367. Christopher, B.R., Gill, S.A., Giroud, J.P., Juran, I., Schlosser, F., Mitchell, J.K. and Dunnicliff, J., 1989, Reinforced Soil Structures: Volume I. Design and Construction Guidelines, Report No. FHWA-RD-89-043, Washington, DC, USA, November 1989, 287 p. Cundall, P. and Board, M., 1988, A Microcomputer Program for Modelling StrainStrain Plasticity Problems, Proceedings of the Sixth International Conference in Numerical Methods in Geomechanics, Vol. 3, Balkema, Innsbruck, Austria, April 1988, pp. 2101-2108. de Borst, R. and Vermeer, R.A., 1984, Possibilities and Limitations of Finite Element Analysis, Geotechnique, Vol. 34, No. 2, pp. 199-210. FLAC, 1996, Fast Lagrangian Analysis of Continua, Version 3.3, User manual, Vols. 1-3, Itasca. Frydman, S. and Burd, H.J., 1997, Numerical Studies of Bearing Capacity Factor N, Journal of Geotechnical and Environmental Engineering, Vol. 123, No. 1, pp. 20-29. Jaber, M. and Mitchell, J.K., 1991, Behaviour of Reinforced Soil Walls at Limit States, Performance of Reinforced Soil Structures, McGown, A., Yeo, K., and Andrawes, K.Z., Editors, Thomas Telford, 1991, Proceedings of the International Reinforced Soil Conference held in Glasgow, Scotland, September 1990, pp. 53-58. Jaber, M., Schmertmann, G.R. and Collin, J.G., 1992, Prediction of Geosynthetic Reinforced Soil Wall Performance Using the Finite Element Analysis, GeosyntheticReinforced Soil Retaining Walls, Wu, J.T.H., Editor, Balkema, 1992, Proceedings of the International Symposium on Geosynthetic-Reinforced Soil Retaining Walls, Denver, Colorado, USA, August 1991, pp. 305-311. Kapurapu, R. and Bathurst, R.J., 1995, Behaviour of Geosynthetic Reinforced Soil Retaining Walls Using the Finite Element Method, Computer and Geotechnics, Vol. 17, pp. 279-299. Kapurapu, R. and Bathurst, R.J., 1992, Prediction of Geosynthetic Reinforced Soil Wall Performance Using the Finite Element Analysis Geosynthetic-Reinforced Soil Retaining Walls, Wu, J.T.H., Editor, Balkema, 1992, Proceedings of the International Symposium on Geosynthetic-Reinforced Soil Retaining Walls, Denver, Colorado, USA, August 1991, pp. 329-342. Lo, S.C.R., 1998, Pull-Out Resistance of Polyester Straps at Low Overburden Stress, Geosynthetics International, Vol. 5, No. 4, pp. 361-382. Lo, S.C.R., and Gopalan, M., 1996, Determination of Delta Value for Reinforced Soil Wall, Geomechanics in a Changing World, Jaksa, M.B., Kaggwa, W.S. and Cameron, D.A., Cameron, Editors, Institution of Engineers, Australia, Vol. 1, Proceedings of the Seventh Australi -New Zealand Conference on Geomechanics, Adelaide, Australia, July 1996, pp. 151-156. Lo, S.C.R., Li, S.Q., Gopalan, M. and Gao, Z., 1996, Analysis and Design of a Tied Back-to-Back Geosynthetic Reinforced Soil Wall, Geotechnical Engineering, SE Asian Geotechnical Society, Vol. 27, No. 1, pp. 37-50.
412
S 1999, VOL. 6, NO. 5
Lorig, L.J., 1991, Analysis of Novel Retaining Structures Using Explicit Finite Difference Codes, Computer Methods and Advances in Geomechanics, Beer, G., Booker, J.R. and Carter, J.P., Editors, Balkema, Vol. 1, Proceedings of the International Conference in Computer Methods and Advances in Geomechanics, Cairns, Australia, May 1991, pp. 157-164. R57, 1996, Design of Reinforced Soil Walls, Quality Assurance Specification R57, Roads and Traffic Authority of New South Wales, Australia, 67 p. Rowe, R.K. and Ho, S.K., 1997, Continuous Panel Reinforced Soil Wall on a Rigid Foundation, Journal of Geotechnical and Environmental Engineering, Vol. 123, No. 10, pp. 912-920. Vesic, A.S., 1975, Bearing Capacity of Shallow Foundations, Foundation Engineering Handbook, Winterkorn, H.F. and Fang, H.Y., Editors, Van Nostrand Reinhold Co., New York, New York, USA, pp. 121-147. Wu, T.H.J., 1992, Geosynthetic-Reinforced Soil Retaining Walls, Wu, J.T.H., Editor, Balkema, 1992, Proceedings of the International Symposium on Geosynthetic-Reinforced Soil Retaining Walls, Denver, Colorado, USA, August 1991, 375 p. Zornberg, J.G., Mitchell, J.K. and Sitar, N., 1997, Testing of a Reinforced Slope in a Geotechnical Centrifuge, Geotechnical Testing Journal, Vol. 24, No. 4, pp. 470-480.
NOTATIONS Basic SI units are given in parentheses. B BER Bi B* B ER c cf c* f E e Fb Fx Fz F* z H ic , iq , i Kax = = = = = = = = = = foundation width (m) foundation width of equivalent rectangular wall (m) effective foundation width for bearing capacity calculation (m) factored foundation width (m) effective bearing width of equivalent rectangular wall foundation (m) soil cohesion (Pa) cohesion of foundation soil (Pa) reduced/factored soil cohesion (Pa) Youngs modulus of soil (Pa) eccentricity of foundation reaction (m) = global factor of safety against bearing capacity failure (dimensionless) = horizontal force acting on foundation (N) = vertical force acting on reinforced block (N) = factored vertical force acting on reinforced block (N) = height of wall (m) = load inclination factors for bearing capacity equation (dimensionless) = active lateral earth pressure coefficient (dimensionless)
S 1999, VOL. 6, NO. 5
413
Lavg LB LT Nc , Nq , N po qu q* u Tmax x y max x y f * f yy
= = = =
= = = = = =
= = = = = = = = = =
average width of reinforced zone (m) length of reinforcement at base of wall (m) length of reinforcement at top of wall (m) bearing capacity factors (dimensionless) equivalent uniform overburden pressure at foundation level (Pa) ultimate bearing capacity (Pa) factored ultimate bearing capacity (Pa) maximum reinforcement tension (N/m) horizontal distance (m) vertical distance (m) maximum wall facing displacement (m) maximum lateral/horizontal wall facing displacement along (m) settlement of wall near toe (m) internal friction angle of soil (_) internal friction angle of foundation soil (_) reduced/factored internal friction angle of soil (_) unit weight of soil (N/m3) interface friction angle between vertical back of idealised wall and retained fill (_) Poissons ratio of soil (dimensionless) foundation pressure (Pa)
ABBREVIATIONS ER: RS: TRS: equivalent rectangular (reinforced soil wall) reinforced soil (wall) trapezoidal reinforced soil (wall)
414
S 1999, VOL. 6, NO. 5
APPENDIX
This section outlines the procedure to calculate the bearing capacity.
A-1
CALCULATION OF FACTORS OF SAFETY FOR THE BENCHMARK WALL DURING THE WORKING CONDITION
Bearing capacity factors are calculated based on the following expressions from Vesic (1975): N q = exp( tan ) tan 2 N c = (Nq 1) cot N = 2 (Nq + 1) tan
+ 2 4
(A-1) (A-2) (A-3)
At the working condition where = 30_ and c = 10 kPa, Equations A-1 to A-3 give Nq = 18.4, Nc = 30.14, and N = 22.4. It is recognised that N is semi-empirical; however, the above expression for N gives a value close to the more rigorous solution given by Bolton and Lau (1993) for a rough strip footing. The inclination factors given by Vesic (1975) are: i q = (1 + K) 2 i c = iq 1 iq Nc tan (A-4) (A-5) (A-6)
i = (1 + K) 3 where:
K=

x z
F c cot 1 + F B F
z
(A-7)
and Fx is the horizontal force acting on the foundation. Thus, both forces acting on the reinforced block are required to calculate the inclination factors. The frictional angle between the virtual back of the idealised wall and the retained general fill is calculated using the empirical equation given in the FHWA design guide (Christopher et al. 1989) for inextensible reinforcements:
S 1999, VOL. 6, NO. 5
415
= 1
L avg H 0.2
(A-8)
where Lavg is the average width of the reinforced zone. As discussed by Lo and Gopalan (1996), Equation A-8 can also be used for high strength geosynthetics. For Methods A and B, an ER wall (Figure 1) is used to calculate the forces acting on the reinforced zone by setting Lavg = BER = 4.64 m, H = 8 m, and = 18.6_. This, in conjunction with = 30_, gives the active lateral earth pressure coefficient Kax = 0.2826 and, thus, the horizontal activating force, Fx = 180 kN. A statics analysis results in Fz = 803 kN, which gives the following inclination factors: iq = 0.627, ic = 0.6056, and i = 0.4965. Moment equilibrium calculations give an eccentricity value, e = 0.4245, thus, B = 3.791. In Method A, Bi = B , which gives qu = 696 kPa and FB (A) = 3.28. In ER ER Method B, because B > LB , Bi is set to the lower value of LB = 3.2, which gives qu ER = 696 kPa and FB (B) = 2.50. For Method C, the force calculation is based on an idealised, sloping-back wall (Figure 1). The virtual back of the wall is inclined at 17.7_ to the vertical. Because the average width of the reinforced zone remains unchanged, = 18.6_. This, combined with = 30_, gives Kax = 0.1939, therefore, Fx = 124 kN. A statics analysis results in Fz = 708 kN, which gives the following inclination factors: iq = 0.701, ic = 0.684, and i = 0.587. Moment equilibrium calculations give an eccentricity value, e = -0.197 m, thus, the base width of the idealised, sloping-back wall, Bi = 3.2 m. This results in qu = 730 kPa and FB (C) = 3.31. A-2 CALCULATION OF Fb (B) FOR THE BENCHMARK WALL DURING THE NEAR-COLLAPSE CONDITION
The forces acting on the idealised ER wall (Figure 1b) are identical to the working condition; however, the foundation strength parameters are reduced to * = 26_ and f c * = 8.45 kPa, where * indicates reduced values. Also, the unit weight of the foundation f soil is set equal to the buoyant unit weight to simulate the water table at the foundation level. This gives the following bearing capacity factor values: Nq = 11.85, Nc = 22.25, and N = 12.53. The inclination factors are: iq = 0.627, ic = 0.5926, and i = 0.4965, thus, qu = 278 kPa and FB (B) = 1.10.
416
S 1999, VOL. 6, NO. 5

B C F T G - R S W: Technical Paper by S.C.R. Lo, J. Bosler and M. Gopalan

Uploaded by

Copyright:

Available Formats

B C F T G - R S W: Technical Paper by S.C.R. Lo, J. Bosler and M. Gopalan

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

B C F T G - R S W: Technical Paper by S.C.R. Lo, J. Bosler and M. Gopalan

Uploaded by

Copyright:

Available Formats

Technical Paper by S.C.R. Lo, J. Bosler and M.

BEARING CAPACITY FAILURE OF A TRAPEZOIDAL, GEOSYNTHETIC-REINFORCED SOIL WALL

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

Trapezoidal reinforced zone H

Sloping virtual back idealisation

Equivalent rectangular wall idealisation

Foundation level LB BER

Trapezoidal reinforced soil (TRS) wall and the simplified dimensions.

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

ANALYSES Wall Configuration

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

10 reinforcement layers @ 0.8 m vertical spacing Select fill Wall facing

7.0 m Foundation soil

Schematic of the benchmark TRS wall.

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

Horizontal distance, x (m)

S 1999, VOL. 6, NO. 5

Vertical distance, y (m)

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

Wall facing alignment Foundation level

Fixed bottom boundary 18 m

Elastic soil properties.

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

A typical FLAC analysis consisted of the following three stages:

S Stage I (working condition): The wall was constructed in a layer-by-layer manner

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

Results of the Benchmark Wall Analyses

Working condition displacement vectors 0 2 10-1 Maximum vector = 1.150 10-1

Vertical distance, y (m)

Horizontal distance, x (m)

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

Horizontal displacement at the wall facing, x (mm)

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

Distance from wall facing (m)

Foundation settlement of the 8 m-high wall.

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

Distance from wall facing (m)

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall

Vertical distance, y (m)

Displacement vectors Maximum vector = 1.30810 -1

Horizontal distance, x (m)

S 1999, VOL. 6, NO. 5

LO, BOSLE AND GOPALAN D Failure of a Trapezoidal, Geosynthetic-Reinforced Soil Wall