Simplified Approach To The Design Of Concrete Block Retaining Walls

J. Du Plessis, C. Martella, T. GreenIn a global stability analysis more than one slip failure surface is analysed, with the criticalfailure surface being the one with the lowest value calculated for the factor of safety. For manyCBR walls the governing design criteria is global stability. When only considering overturningand sliding, generally, shorter lengths of geosynthetic reinforcing is required. However, a deepseated slip failure may occur, which would only be identified using FEA software.An example of a FEA result indicating a deep-seated slip surface can be seen in Figure 8 below.Figure 8. Example of a global stability analysis.Unfortunately, it appears not to be standard practice to check the global stability of CBR walls.It is essential to remember that even if the design checks for overturning and sliding aresatisfied, a global stability analysis is always required, especially for walls in excess of 6.0mhigh. This is to check the possible slip circle failure that passes around the reinforced zone andthat the reinforcement length is adequate. The failure mechanism may also be affected byexternal factors such as increased surcharge and slopes on the toe and crest.In addition to determining the global stability, FEA software has the ability to take multiplefactors into account. The effect of cohesion, groundwater seepage, complex surcharge, seismicloading etc, can be accurately modelled and the effect on stability of the CBR wall determined.When no geotechnical information is available for the site, representative strength parametershave to be estimated for the design, but they need to be checked during construction.In addition, the shear strain distribution found in a global stability check can be used todetermine the reinforcing type needed to ensure the wall stability. This is an important aspectfor optimising the design and can to be taken into consideration to mitigate the cost of the wall.3 Case studiesIn recent years, there has been an increase in the number of failures of concrete block retainingwalls to the extent that ECSA’s (Engineering Council of South Africa) Investigating Committeehas identified these walls as problem structures. The failure of these walls is usually attributedto construction or design deficiencies. Two cases are discussed below, with different reasonsfor failure.19

9th SAYGE Conference 20173.1 Case 1 – Concrete block retaining wall: CollapseIn this case, a portion of a CBR wall collapsed (Figure 9) because of both inadequate designand incorrect construction. This wall was between 4.0m and 8.0m high, and was constructed at70 using TB490 blocks in an "open face" configuration.Figure 9. Collapsed wall.According to the original design, 2.5m long strips of Polyfelt Rock GX100 had to be placedevery fifth row of blocks. After the collapse, when the blocks where removed, it was observedthat no geosynthetic strips were actually placed behind the wall. Regardless, the design lengthwouldn’t have been sufficient to guarantee the wall stability as original design allowed for 2.5mlong strips (i.e. only 31.25% of the retained height).Additionally, placing the geosynthetic strips every fifth row of blocks, which is quite a widespacing, reduced the effectiveness of the reinforcing. In this case, the slip failure surface couldpass through the reinforced zone without being intersected by the strips.The mistakes made on this project can be summarised as follows:Vertical spacing of geosynthetic reinforcing too large.Length of geosynthetic reinforcing too short.Inadequate supervision during construction, hence the missing geosynthetic strips.···The only solution in this case was to demolish and rebuild the CBR wall correctly, withadequate reinforcing at a suitable vertical spacing.3.2 Case 2 – Concrete block retaining wall: Excessive displacementIn this case, the CBR wall was 3.5m high, constructed at 80 , reinforced with RockGridPC100/100 geosynthetic strips placed every 3rd row of blocks and subject to a minimumreinforcing length of 2.5m behind the blocks. In addition to this, the top 1.5m of backfillmaterial was stabilised up to a storm water pipe approximately 3.0m behind the crest of theCBR wall.A portion of a concrete surface bed directly above the concrete block retaining (CBR) wall wasshowing signs of distress, and movements occurred along the entire wall length. The concretesurface bed moved approximately 30mm horizontally towards the CBR wall, and verticallyupwards by approximately 10mm. Following this distress, a line of concrete surface bed panelsbehind the wall, 35-40m in length, were broken out. Within this exposed zone, there appearedto be a tension crack running parallel to the CBR wall, approximately 3m behind the top of thewall, with this tension crack roughly above a sub-surface storm water pipe (Figure 10).20

J. Du Plessis, C. Martella, T. GreenFigure 10. Tension crack along the entire wall length, above a sub-surface storm water pipe.The original design was reviewed using finite element analysis software. The CBR wall waschecked for three different load conditions: 20-ton vehicle load, 50-ton vehicle load and anominal surcharge of 10kPa, using an ru 0.1 to reflect little to no ground water in the fill.In all cases the finite element analysis yielded a factor of safety considerably higher than 1.5.Given that the design of the CBR wall appeared adequate, external factors were considered toexplain the movement in the retaining wall. The most pertinent of these was the storm waterpipe that ran parallel to the affected CBR wall. In this case the presence of the stabilised filltowards the top of the wall prevented water from draining freely out the CBR wall, developinghydrostatic pressure against the stabilised fill and encouraging water to flow along the backfillaround the storm water pipe.Therefore, to investigate this scenario another finite element analysis was conducted. In thiscase, a nominal surcharge of 10kPa was applied throughout with the ru increased to 0.5 to reflectsaturation of the backfill. Hydrostatic pressure was also added behind the stabilised fill at thetop of the wall.The finite element analysis for this scenario yielded a factor of safety of 1.15. This suggeststhat while the wall in relatively dry conditions is perfectly adequate, saturation behind thestabilised fill reduces the factor of safety to 1.15, which is less than the temporary minimumrequirement of 1.3. In addition, the expected movements are reasonably high, and correlatewell with the movements observed on site. The FEA result is shown in Figure 11.Figure 11. Wall displacements.21

9th SAYGE Conference 2017In this case, the wall failed in the serviceability state, with an increase in horizontaldisplacements between dry and saturated case. This was calculated at approximately 20mm to25mm. It is interesting to note in this case that the addition of stabilised fill, however wellintentioned, actually contributed to the failure of this wall.4 ConclusionAfter some case studies of failed CBR walls were reviewed, a few common trends and aspectsthat typically cause problems have been identified.Many of the failures of CBR walls are the result of water ingress from external and internalsources, leading to the formation of a slip plane behind, through or beneath the walls. In somecases, the walls were extended beyond their original design height or subjected to surchargeloading for which they were not designed. Internal instability problems were common due toinadequate reinforcement design and installation.The quality and compaction of the backfill is another recurring problem. All too often, materialavailable on site is used as backfill without due regard to its drainage characteristics, strength,compaction requirements and sensitivity to water ingress.Typical design related issues include incorrectly assumed soil properties, inadequate provisionfor surface and subsoil drainage, and incompatibility of the design with the actual conditionson site. Other issues attributable to the designer include inadequate construction monitoringand poor standard of the construction drawings.Design parameters for in situ soils and the backfill are often assumed without any testing andwithout the necessary inspections and site control during construction. In some cases, it mayhappen that designers lacked a basic understanding of soil mechanics principles and the mannerin which walls act, and some designers make use of empirical methods without understandingtheir limitations.Furthermore, it is highly recommended that finite element analysis software is utilised toconfirm hand calculations – specifically for larger walls where global stability is very often thegoverning stability criteria.ReferencesCoulomb, C. A. 1776. Essai sur une application des regles des maximis et minimis a quelquesproblemes de statique relatifs a l'architecture. Memoires de Mathematique de l’AcademieRoyale de Science 7, Paris.Jewell, R. 1996. Soil reinforcement with geotextiles. London: Thomas Telford.Lambe, T.W. and Whitman, R.V. 1969. Soil Mechanics. New York: Wiley.SANS 207:2011. 2011. The design and construction of reinforced soils and fills.Shukla, S. and Yin, J. 2006. Fundamentals of geosynthetic engineering. London: Taylor &Francis/Balkema.22

Proceedings of the 9th South African Young Geotechnical Engineers Conference,13, 14 & 15 September 2017 – Salt Rock Hotel, Dolphin Coast, Durban, KwaZulu-NatalThe Influence of Material Grading on Strength andStiffnessL. Geldenhuys11University of Pretoria, Pretoria, Gauteng, louis.geldenhuys@up.ac.zaAbstractFull scale testing of road pavements can be costly and time consuming. There are alsodifficulties observing and understanding the failure mechanisms and damage accumulation ofroad pavements under numerous load cycles. Centrifuge modelling of scaled down roadpavements solves some of these problems. It has been proposed that centrifuge testing of ultrathin continuously reinforced concrete pavements (UTCRCP) is done. Research has been doneon scaling down the concrete, but it is also necessary to scale down the base material. Triaxialtests were done on three soil samples for which the grading was truncated to different maximumparticle sizes (6.35 mm, 2.0 mm and 1.18 mm). It was found that the shear strength parametersas well as the Young’s modulus for the tested grading range did not change significantly dueto the change in grading for the three samples.Keywords: grading analysis, triaxial test.1 Introduction1.1 BackgroundRoad pavements typically consist of compacted layerworks with a rigid or flexible surfacing.Full scale testing is common practice for road pavement analysis. The benefits of doing scaleddown testing of road pavements have become apparent. This is partly due to the savings in costsby building smaller models. Parametric studies can be done on smaller models since theturnaround time between testing and sample preparation is shorter, allowing more tests to bedone in the available time. Variables, such as soil compaction and moisture content, can bebetter controlled in smaller models. Environmental effects can be better controlled accordingto the needs of the experiment during small scale testing.It has been proposed that centrifuge tests of scaled-down ultra-thin continuously reinforcedconcrete pavements (UTCRCP) are done. This would result in the concrete layer in thecentrifuge model being much thinner than the actual size used for the full-scale pavement. Theconcrete mix design would have to be altered so that smaller aggregate and reinforcing steelcan be used. Research has been done and concrete slabs have been developed that are scaleddown and yet represent the behaviour and strength of the full-scale concrete adequately23

9th SAYGE Conference 2017(Kearsley et al., 2014). The soil layerworks for UTCRCP are typically 150 mm thick. Thelayers in a 1:10 centrifuge model would be 15 mm thick. The maximum aggregate diameter oncompacted road layerworks is 0.25 of the layer thickness. This rule would have to be appliedto the scaled-down model too. It is therefore necessary to investigate whether, and by howmuch, the properties of the material used in the layerworks changes if the maximum aggregatediameter is restricted to a smaller value. This research stemmed from that problem.1.2 Centrifuge TestingSelf-weight is an important factor to consider when modelling geotechnical problems(Schofield, 1980). This is because the behaviour of geotechnical materials is heavily dependenton the stress state imposed on the soil. This is an intrinsic property of model size. The size ofthe problems that need to be modelled in the field of geotechnical engineering makes itnecessary to scale them down. If the model is scaled down, the stresses in the soil are notproportional. It is therefore necessary to keep the stress state in the soil the same by increasingthe self-weight of the model. This can be done in a centrifuge.1.3 Triaxial TestLaboratory testing of soils is an important part of the material characterisation of a geotechnicalproblem. The triaxial test is a common method of testing soils in a controlled environment. Thesample’s consolidation parameters, permeability, strength and stiffness can be acquired fromthe triaxial test. A stress path can also be imposed on a sample and the response thereof can beobserved. The triaxial test also allows control over drainage. The pore pressures, volumetricstrains and axial strains can be easily measured. There is also control over the applied principalstresses. This, combined with its versatility, makes it a popular test method for soils (Bishop &Henkel, 1962).1.4 Particle Size Distribution of modelled materialsThe particle size distribution (PSD) of a soil influences the shear resistance and deformabilityof a material. This is because it determines the number of inter-granular contact points (Lineroet al., 2007). It also determines the density to which a soil sample can be compacted to.The particle size distribution of the coarser particles can be determined by a sieve analysis.Various methods are used to determine the grading of the finer components.The maximum particle size has to sometimes be limited. This could be due to a limitation inapparatus size. It could also be because a scaled-down model is being tested. There are twobasic methods which can be used to change the grading of a material in such a way as to limitthe maximum particle size. The “truncated” scaling method involves removing all the materialover a certain particle size. The “parallel” scaling method results in a PSD curve that is parallelto the PSD of the original field sample. The sequence of grain sizes is maintained and the resultthereof is a true scaling of the material (Lee, 1986; Verdugo & Gesche, 2003).When scaling the grading of a material down, it is important to realise that the various portionsof the material in the grading curve do not all have equal contribution on the behaviour of thematerial. This can be illustrated with clay. A small change in the amount of clay in a materialcan have a significant change in the behaviour of that material. When using the parallel gradingtechnique, the fine material is changed from silt to clay (according to particle size). This is adisadvantage of using this method for scaling down materials.24

L. Geldenhuys2 Experimental Procedure2.1 MaterialThe material for the samples was acquired from a road construction site on the N1 highwaynear Trompsburg. It was classified on site as a G7 base material. This means that it is a naturalmaterial (soil, sand or gravel) with a plasticity index of less than 12 and a grading modulusbetween 0.75 and 2.7 (COLTO, 1998). The MOD AASHTO density was 2385 kg/m3 and theoptimum moisture content was 5.3%. The grading curve of this material is shown in Figure 1.Figure 1. Grading curves of the original sample.2.2 ExperimentsThree samples were made from the sample obtained from site. To ensure that each samplecreated from the original sample was representative of the original sample, a sample splitterwas used. The original batch was split into two halves. These were then also split into two. Thisresulted in four samples being created, of which three were used in the experimental procedure.The truncated grading method was used rather than the parallel grading method. The firstsample (Sample A) was truncated to 6.35 mm, the second (Sample B) was truncated to 2 mmand the third to 1.18 mm (Sample C). The specimens tested in the triaxials were created fromthese three samples. The grading curves of the three samples are shown in Figure 2.Figure 2. Grading curves for the three truncated samples.25

9th SAYGE Conference 20172.3 Experimental SetupConsolidated undrained triaxial tests were performed on samples with a diameter of 50 mm anda height of 100 mm. Samples were prepared to a target dry density of 2000 kg/m3. This targetdensity was determined by compacting samples with varying moisture contents, usingconventional triaxial preparation equipment and using maximum compaction effort.The samples were first saturated until a B value of at least 0.96 was achieved. Thereafter,samples were consolidated at effective stresses of 100, 300 and 500 kPa. Drainage duringconsolidation was provided for through a porous disk at the top and bottom of the sample.Samples were then sheared at a rate, determined by the time taken to consolidate, of0.1 mm/min. The confining pressure for each sample was the same as the effective stress atwhich it had been consolidated to. The naming of each sample indicates this (e.g. A 300 istruncated to 6.35 mm and consolidated at 300 kPa). The load, axial displacement and porepressures were recorded. Tests were ended once a displacement of 15 mm had been reached. Asummary of the samples tested is given in Table 1 below.Table 1. Properties of the prepared specimens.SampleA 300A 500B 100B 300B 500C 300C 500Dry density ofpreparedsample (kg/m3)2009201519801964197519971997Moisturecontent (%)4.84.87.47.47.46.66.33 Results3.1 Shear Strength ParametersThe shear strength of the soil was determined by plotting the Mohr-Coulomb line for eachsample and the friction angle (ϕ) was calculated. This required at least two samples tested atdifferent effective stresses. The shear strength of the soil determines the bearing capacity ofthe soil as well as the failure mechanism that will develop. Only one specimen was tested at aconfining stress of 100 kPa.The following figures show stress paths of the triaxial tests. Sample A (Figure 3) was slightlydenser than the other two samples, but it is interesting that Samples B and C (Figures 4 and 5respectively) show a stronger dilatant behaviour. The Mohr-Coulomb line was fitted throughthe estimated failure points. This allows the friction angle (ϕ) to be calculated. Although thefriction angle is sensitive to the estimated best fit line through the failure points, it remainedwithin a range of 38.5 and 39.5 .26

L. GeldenhuysFigure 3. Stress path and Mohr-Coulomb failure envelope for Sample A.Figure 4. Stress path and Mohr-Coulomb failure envelope for S

example of geosynthetically reinforced concrete block retaining wall is shown in Figure 1. Figure 1. Example of CBR wall 1.1 Uses of concrete block retaining walls Concrete block retaining walls are reasonably versatile structures, and generally provide a more cost effective solution when compared to reinforced concrete walls or gabion structures.