Design methods for steep reinforced embankments R.A Jewell, N. Paine and R. |. Woods, Binnie and Partners emropuerioN The availability of strong polymer reinforcement materials which are resistant to corrosion provides attractive possibilities for the use of poor ‘and aggressive soils to build steep Enbanknents and retaining structures. Current knowiedge and experience from reinforced earth walls, and the findings of research into the. behaviour of reinforces “soll with reinforcement materials other than metal, “and soils other than sand, can be combined to provide desion calculations for reinforced embankments. ‘This paper presents calculations for steep reinforced embankments. Using the calculations described in the paper, & nunber of design chares for steep reinforced Slopes founded on a strong level foundation are Erogontod “in order “to faoilstate” preliainary gn studies. BEPECT OF REINFORCEMENT 1X SOIL ‘The purpdse of reinforcement used to strengthen concrete, coment composites or palymors in well, understood. the reinforcerent is positioned ta alter gdvantageously the stresses in the mstrix material to bo etrongthenod. Observations at model and field acaie on reinforced 2oil valley fand on unit cell tests in the laboratory, Fig: I, have chown that reinforcement acts t5' alter the ‘pattern of stresses in S011. to enable greater applied" loading to be supported (Schlosser et a1 1993) The pattern of stress (and strain) in reinforced soil is complox ‘and non-uniform, even "in “Isboratory unit cell testes Simplitying iaalisations are required” for calculation purposes. A direct idealisat ion thet “satisfies force “equilibrium examines potential failure surfaces through the Feinforeea "coil “and the resultant Eeinforcenent “force acting across these, as shown in'Fig 1 for laboratory tests, 70 Design calculations for steep reinforced slopes are descritied. charts for prolininary design are Dresented for the case of steep slopes over a Eompetant foundation reinforced by horizontal layers of ‘polymer grids. Design evaluations reguire a mix of calculations and parfornance criteria” to ensure satisfactory “behaviour, Descriptions of the calculations and criteria used to derive the design charts are given. A simple procedure to determine reinforcement spacings is Introduced, based ona spacing constant Q.” The relationship between the” stiffness of Eeinforcement and the soil strength deformation behaviour” influence the selection ‘of design values. A design value of soi) strength at large strain is recommended for polymer grid reinforced Soil, although dt is anticipated that equilbrivm nay "be established sooner at seller strain magnitudes. he normal component of the _tensile reinforcement “fore mohilises additional frictional resistance in the soil and the tangential componant acts directly to resist applied shear loading. (a) Axial compress: Test on (b) Direct Shear (e) Effect of reinforcement force, hy %y Py increases nornal effective stress Teduces applied shear stress Pig.1 Unit cell tests on reinforced ott (a ¢ b) and idealisations for the affect of reinfarcement (c) Both compressive and tensile strains occur when soils. are subject to. shear loading, “and Feinforcanant acts sdvantageauely when placed. in directions in which tenelie strains sccur. Tensile reinforcement forces axe mobilised by. the tensile soil strains. “the amount of tensile force which may bo generated in low modulus or extensible reincorcenent may be governed by allovable teneile strains in the Polymer iid reinforcement, Thomas Telford Limited, London, 1985soil. Care is needed in design to ensure that the agsuned reinforcement forces end Foil shear Eesistance are compatible with the oxpected Straing in the reinforced soil. This point is, Biscussed further in tho context of selecting Gocign values for" poll. and reinforeenent properties. STEEP REINFORCED SLOPES To build a slope steeper than the naturally Stable angle requires additional etabllising forces, these can provided by Eeinforeenants. placed” horizontally in the Slope. The reinforcenents carry 2xial forces Which increase the shear recistance of the seal Bnd improve stability. The case of a steep embankment built over a Gonpatent foundation is examined in this paper. The aim of design is to determine the number ‘and disposition of reinforcanents to. provide Overall equilibrium in the slope and to avoid any local overstressing. ‘The strength characteristics of reinforced soil can “be highly anisotropic, and potential fatlure surfaces pascing betveen reinforcenont layers are as inportan= a6 those intersneting Feinforcoment layers, Tig, 2a. The length of the roinforcenent. in a steep’ embankment must provide sufficient bond to ecbilise the maximum jernissible force whera the reinforcenent is intersected by eritical surfaces in the soil, The length must also be sufficient to prevent the cecurence of bodily outvard sliding of the Foinforcod sone over an vnderiying layer of reinforcements, Fig. 2a. = tial sims L surFaces , cient » fe | a8 re ee | for equilibrium earl ay " ol af | | mem envetepe of ee aie = EMBANKMENTS he design problem can be expressed as the search for’ the spacing “and length of Eeinforcaments in the slope that provides equilibrium with (1) 2 distribution of required axial Force along each reinforcement that nowhere exceeds the nazimun available force which nay be governed either By the factored reinforcenant or bond with the soil, Fig.2b 4c and (41) a reinforced soil zone of sufficient Sinersions to adequately support the bnreinforeed interior of the slope. (CHOICE OF ANALYSIS, Most studies and experience with reinforced fills ‘to date have been on compacted granular noila reinforced uniformly with metal strips and built vertically to’ form retaining walls and abutments, Sebloeser et el (1979,1983). A mix of andlyeos are weed. for design calculations, which seens appropriate given the complex and ‘indeterminate nature of reinforced soll. Limit equilibrium analysis is used to calculate a gross reinforcement quantity and diseribucion fo, provide equilibrium. A plane potential failure surtace is adopted in “the British (1373) “and Franch (1978) Rational Transportation Department codes for vertical walle. he value of reinforcement force can be locally estinated by assuming that the product of the horizontal suvass in the soil and the vertical area of soil served by the reinforcement 19 Balanced by the reinforcenent aial forces, Fig. So the “horizontal stress ‘depends on the vertical stress and the local value of the Soofticiant of sarth pressure. The combination of Lime oquilibrium analyese and Toca) checks on individual reinforcement Spacinga appone te offer a rational basis £0r design which can be adopted for reinforced ombsnienents. te y me Fig.3 Reinforcement forces calculated Fron Ioeal stresses nJEWELL ET AL ANALYSIS FOR REINFONCED. SLOPES A limit equilibeiom analysis for reinforced Slopes has been developed and is perrormed by the computer program WACGLE (2982). Four important. aspects of the analysis are Summarised below. Two-part wedge surfaces are examined. ‘This shape is ‘suitable to check both nechonisme shown in Pig. 29. Selecting @ grid cf trial wedge nodes wWetin and behind the reinforced gone, ‘the full range of potential. failure echiaions may be investigated. Current. lack Of knowledge about interciice forces in feinforced roll make a sinple mechenin attractive. At each point that a trial weage surface Grosse Feinforcenont in the geil, the: jobilised “vale of reinforcenent force ‘is Cscalated and resolved inte conponents normal Gnd tangential to the surface (as shown in Pig. i} ana included in the equilibrium equations. A conventional definition of safety factor can be"adopted. In thie case the soil shear txength available to provide equilibriue Bquals the design peak strength for the soll Sivided by the "overall. eatery factor. Similarly the reinforcement force in each layer intersected that 10. available to provide equilibrium equals the maximun design value Aiviaea by the overall safety factors (the maximum devign valve of reinforcement force is governea by either strength of bond parameters nding on where aleng its length the roinforeenent in intersected by the wedge mechanism, see Fig. 2c). Iterations ere Foquirea with the conventional definition of Safety factor in oxder to find the combination Of wedge angles giving the lowest value. The safety margin may equally voll be expresso in terns of partial factors. Typically the loadings (which inclode surcharge, soil nit weight and’ ‘pore water pressures) would be increased. by partial factors, while the resistances (which include soil shear strength, Foinforconent™ etvength ard bond) would” be reduced by partial factors. An additional Eesidual factor may be used to ellow for the consequences of failure, uncertainties in the analytics! method, etc. In this case the Eactored Joatings and resistances would be used directly in the WAGGLE computation, “and a Feingorcenent layout sought which provides the Gesired residual partial factor. Pore water pressures are included in the overall equilibrium equations, taking account Of the observatione by Turnbull «. Mvoreley (1967), and whitran &” palley (1967) on the resolution of pore water pressures. Pore water pressures. also reduce the normal effective Stress and hence the bond shear stress between the reinforcenent and the oil. The design of @ reinforcenent layout to give a desired overall safety factor’ for 2 slope proceeds by trial and efxor, like the design of bn unreinforced slope. The total number of n reinforcenent jayars and thoir spacing can he arranged to give more or leas equal safety factors on pechenisms ‘passing ‘through the reinforced zone, and on mechanisms passing most2y through the unreinforced interior of the Slope and passing out to the face betwean Feinforcenont. layers, Pig2a. Such an arrengenent usually Keeps to a minimum the total guantity of reinforcenent required. Comparison with published resulte Results from the WAGGLE prograw have been compared with published solutions for bureinforced sloper, For example, the spread Sf results found by comparing ‘the miniwon fafety factor computed by the WAGGLE proarem and values given by Bishop and Morgenstern (1860). for a range of slopes, soil. shoar strengthe and pore water pressures are summerised fn Pig. 4, “the doviacion of the ealculated safety factor from the pblished Value 1s shown plotted against the coefficient Of Intersiice Toughness; wilch 1ies in the zange zero to unity. Taking the coefficient of interslica roughnes: equal te zero for WAGLE giver results of the crder 2% to 58 on the safe side of published Solutions. The coefficient was set equal to Bere for’ the. computation of design charts described below. Se fo Ww wast © = 10% to aot 78 erat = 0 t0 0.05 T a fe 9 10 0.30 WS epren of al resutes Pig &. Comparison of safety fector coleulated by WAGSLE (Py) to Bishop § Morgenstern (1960) published resuits” (py) DBSTGN CHARTS Breliminary design investigations are greatly facilitated if solutions for standard cases can be derived from design charts, tables or equations of closed form. To this end, 2 chart design procedure has been developed ' for the preliminary design of stoop slopes reinforced By polymer grids. the derivation of the desion charts is “described below, and the bacie assumptions are listed.An interpretation of equilibrium compatability between polymer reinforcement and soi! in steep reinforced embankments is given in a. final Section, to provide a basis for selecting Gesicn values Of parameters and safety margins. ‘the charts have been devised for the following cases, Fig. 5 © Embankment slopes buile over a competent level foundation that will not be cverstressea” by “the Constructed slope. © © vniform slope with a horizontaz crest. © Uniform surcharge , along the slope crest, © Slope angles f in the range 30° to Bore © Soils vith effective stress strength parameters in the range ol 0 fra 6 = 209 to 40". © Pore water pressures in the slope exprensed by! the coefficient rin the range 0. to. (See Bishop & Morgenstern (1960) and Pig. 5). © Polymer reinforcement grids with constant length placed edjacentiy in horizontal. layers. “The design strength for the reinforcement grid allows for construction effects, Gnvironmental conditions in the soil and tine effects on reinforcenent mechanics! " behavious “during th= Gesign life of the structure: Main steps for chart deasen There are three main steps in the chart design procedure, Firstly, the naseimim horizontal force 3 roquireé to hold the slope in equilibrium when the soil and pore water preaauves are ne their design values is daternined. TF each reinforcement layer can support a meximum force Piper unit width, then the minimum number of relnforcomant layers N required for eauilibri 15 given by the ratio 1/P. Secondly, the minimun length t for the reinforcement layers is determines s0 that the Foinforces zone ie fot overatreans’ by. Precrures from the unreinforcea interior of the Slope, and to ensure adequate bond lengths Thirdly, as practical reinforcement layouts are Likely "to Be divided into zones containing layers at an equal vertical spacing, s calculation i required to derive 2 practical Spacing arrangement which Will rot lesd to local overstrossing in any ralmforeoment layer. EMBANKMENTS: one reinforced by horizontal layers Unifove sol ena, Pore water pi Tet 7 a3 Competent level foundation Pig.5. Definitions for the slove cases exonined GROSS HORIZONTAL FORCE REOUTRED FOR FOUrLIBRIUM The woximum horizontal force required to hold a Slope’ in equilibrium has been ostimated "waing the two-part wedge program WAGGLE, Tor a given Slope, a search is made hath for the wors wedge’ point leeztion and the worst. combination DE ‘wedge’ angles hich give the greatest roquired force T, Fig. 6.” This magnitude of Force just provides equilibriun on che worst or critical two-part weage when the design value Of soil chear strongth is fully mobilised, ig 6. oefinitions for tho-part wedge mechanism: The results of analyses are shown in Fig.7, ‘the foree T is plotted in a non-dimensional form against slope angle, s a ar where K is the coefficient of earth pressure. Pore wntex pressures and effective soil cohesion both affoct the magnitude of the gross horizontal force for equilibrium, A chart of the type shown in Fig. 7, applies for fixed values of the non-dimensional paranstars for Cohesion cljy q and pera water pressure u/., ‘he results shown in this case are foro! . 9 andy we 2 eo, aJEWELL EF AL D3 bat Slope: arate Pig 7. Groas horizontal force requtred for equilinriue tna steep. vareinroreed slope. A check on the results can be made by comparing the caleulated value Of the covfficient of earth pressure X for slopes with the Limit analysis results published by Chen (1975). Por the ‘case of inclined retaining walls, the Ferults cover the cases shown in Fig.’ 7 for jope angles in the range 50° to 90". the Valuer of K calculated by WAGCLE are typically 58 greater than the values given by chen for smooth walls, except for the vertical case where the coefficients are equal. This finding bears out at steep slope angles the finding for flatter siopes that. the selection of on interelice roughness coefficient equal te zero for WAGGLE gives results that are slightly conservative. Fig 8. Reinforcoment truncation ~ a) at a vertices! Line 2B) parallel to slope face REINFORCEMENT LENGTH Two simple arrangements for reinforcenant reevit fron Gither truncating reinforcement layers to a connon vertical line in the slope, or by Keeping ‘the length constant throughout the slope, Pig. 8, Both methods give the sane result for the vertical ca Comparisons were made between these tvo layouts for a number of Slope cases weing NAGGLE, with, 74 the finding that parallel truncation generally gave a more efficient use of reinforcement, & ‘Smaller quantity to provide the desired safety factor “against all potential failure nechanisns. The use of reinforcenent layers all of a” constant. length Fig.ah has been Selected for the chart designe. Three sain criteria were identified wnien govern the selection of the misimin allowable Foinforcement. length.” The three undesirable Consequences envisaged if there is insufficient reinforcement length are, 1. Roinforcenent layers near the top of he slope have insufficient length to Support the required decign forces which are” shea. to lower layers resulting in overstressingy Tig. 9a. The reinforcement length is ingursicsent to. prevent outward Sliding along the interface between the coil and a reinforcement layer, Fig. 9 be ‘The reinforced one, acting as a rigid ‘block, is not wide enough to resist the’ outward thrust of the unreintorced "soil, in the” slope interior without developing tensile vertical effective strosees anywhore along ite base, Fig. 9c. ~ critical two-part, wedge Ohan® ® max nin Pig.9. Criteria to determine the minimum Sllowable reinforcement iengch owerstressing of reinforcement layers ‘he maximum tensile force which may be Generated by frictional bond over a length of Eeinforcenent is conventionally estimated from the predict of the surface ‘sro of theroinforeoment over which bond develops, the average normal effective stress acting on this area and an angle of friction that can be generated ‘between the reinforcement and the Soil. For gri¢ reinforconont it is convenient fo treat the whole plone surface area to be available for ‘bond, ‘selecting "an appropriate angle of friction ‘to represent overall bond Shear stress, At best, the reinforcement grid may be s= effective as’ a rough sheet. in which fase the bord angle of friction would equal that of the soil, (Jewell et al, 1984). For the design chart galculations the ‘reinforcement bond angie of friction was assumed conservatively to equal half” the design feletion angle for the soil. Por a range of slope cases the profiles of required reinforcement force in the Feinforcerent layers were calculsted with variety of sasuned reinforcement length, ‘This Ehowed “that when the reingoreenant wae Shortened so that the critical two-part wedge mechanism could pass cutside the reinforced zone near the top of the elope, the Feintorconont layers lover in ‘the slope were reuized to carry proportionally greater Foreos, leading to overstrossing. tho criteria adopted to control overstressing wai that the reinforcement length | should always be Softicient to contain the critical” teo-part Wedge, ‘which is the mechanism in the siope Fequizing the largest gross horivcntal force te maintain oquilibriun, ‘she poi eion of the node and the back yedge angle 82, Fig.6, dofino the location of the Gritical' mechanisns found previously when calculating the coefficient of earth pressure values. The minigun ratio of reinforcesent Tongth’ to slope height to contain the worst mechsnisn is shown plotted against slope angle EMBANKMENTS in Pig.10 for the case of no cohesion and no pore water pressures. Outward sliding Reinforcement grids provide geod resistance to Sutward direct sliding by” soil along. the interface between the grid and the overlying oF underlying soil (Jewell et aly 1384). Resistance to direct sliding includes the contribution of soll shear resistance mobilised by the shear of soll over coil through the area oF the reinforcement grid apertures. ‘The Fosistanco to” Girect giiaing over grid reinforcenent may eften be close to the full Shear resistance of the soll, Por design chart Esloulations the Value of frictional resistance to direct sliding over grid reinforcenent vas Aswumed oqual to eighty ‘percent of the design friction angle for the s0!l. For a layout of reinforcement with layers of constant length, ‘the worst case for outward Sliding cecurs over the lowest reinforcenent layer, Fig. 9 b. The required reintorrenent Jength to provide equilibrium on the most critical direct. S1iding mechaniam “was calculated using WAGGLE, and resulta for the Ease of no cohesion and ho pore vater pressures are shown in Fig, 12. No tension on the bi A simple analysis assuming the reinforced zone to act ae a rigid block bas been carried out. The forses acting on the rigid block are self weight and. the outward thrust from ‘the unreinforced slope. interior, Pig 9c. The agnitude of outward thrust, the point of action on the reinforced ‘block ‘and the Foughness on ‘the back of the reinforced block areal] unknown. Por design chart csleulations CRITERIA 10 UEVERNINE THE MINIMUM ALLOWABLE REINCORCENRNT LENGTH FOR STEED SLOPES Slope angle Fig, 10. Minium length co. fontain the oritical two-part edge mechanism Results forthe cate Wiel no coledion or nore vater press 10 —— 2.0 = . “ wy | as 35° 20° 208 Pe 28° 25) Leg 308 302 35° 32 08 ze La ray ° ° ° so" fo” a0 a0? ao? a0? 307 60? a0 Slope angle Fig.11, Minimum length to prevent outwara sUiding Slope angle Pigs12. Minimon length to prevent tensile effective Stresses on the reinforced zone base 7JEWEL ET AL the outward thrust was assuned to equal 7 and assuned to act horizontally on the reinforced Block at one third of the slope height above Foundation 1evel, he forces on the reinforced block were resolved to give ‘vertical, horizontal “and momensnt: components acting at’ the centre of the base. Each of these forces wae accuned £0 be Balanced by a uniform disteibation of stress between the “reinforced block base and. the Foundation surface. The minimum ratio of reinforcement length to slope height to avoid effective tengile stresses securing on the base Of the reinforced block ie shown in Pig. 12, for the case of no cohesion and no pore water pressure. Composite charts for reinforcenent length Results for the three criteria ware combined on one chart to give the reqoited reinforcement Yengthy expressed as the ratio. L/it, as a function of slope angle By. for a rande of fesign soll friction angle values, “Charts aze Given in the appendix for ‘the cases of ere Sail cohesion ¢' /i=0 and pore water prevsures Wyy = 9» 0-25 and 0.50. PRACTICAL REINFORCENENT SPACTNG An aim of design should be to distribute reinforcenent Iayere within the slope so that ho one Of ‘soil ‘or layer of raintoroenent fs unduly. Stressed” in! comparison to renainder. If reinforcement were to be locally overstressed, robilising axial tension ig excece of the design value, then the risk of yupture in the reinforcement which could oad to progressive collapse of the reinforced slope would be greatly increased. he concept of locel stress equilibeson in Bincoreed soll sieges fe°Tihaeteee tee SriPiSealiy’’ a. continuous setnfortettat eps ee mere ey hold in equilibrium “the horizontal soil Tetdeses’ ota which the uses ioeat Sebilinea “Vatos’ cf oinforosnase “Povee "ty ould be: R= OS or steep slopes where the vertical stress approxinately equals the overburden pressure, the reinforcement layers. vould. carry approximately equal forces P throughout the slope if the vertical spacing was varied as the inverse of depth. or v= 2, and Py = constant value P, fron (2), wn = Syke, (2) A separate deduction based on the results of Lisif eguitibrion analysis leads to a siniler 76 The | gross horizontal force thatthe reinforcenant must provide” to maintain equilibrium in the slope can be described by equation (1). Por a slope, the increase in Tequired horizontal force fox oquilibriun can be found by difterentiating equetion (1) with Fespect to depth 2 to givey ar = Wye “ com Tf ‘the increased requirement for force is provided by reinforcemant layers each able to Support a” constant. force Py placed at a variable vertical spacing Sy, then to satisfy equilibrium the reinforcenent spacing would have to be varied, from (4), 86 (5) which reduces to equation (3). Finally, results from WAGGLE analysie of reinforced slopes with uniformly spaced reinforcement layers "typically give higher than average values of safety factor on Rechaniems in the upper parts of the slope, and lover than ‘average values ‘on more deep Seated mechanisms pagsing through the Feinforeed zone lower In the slope. With the current lack of knowledge about the effect of reinforcement spacing on Eeinforcenent forces, ard overstressing, it would be prudent te space reinforcanent. layers fon the agsunption that wach layer may locally have to support the horizontal stresses in the soll as shown in Pig. 3, and described by equation “(2). ‘The maxinun expected local force in the reinforcement would have the value, PR ‘Thus the ideal spacing arrangement to give a balanced distribution of reinforcement in the slope is when for each layer the value of naximum Corce calculated fron equation (6) equals the desien valve of reinforcement force BP This is shown diagramatically in Fig. 13. Sky (6 ey SuneZn-B ro 13 change of vertical spacing with depth to ampally Joed reintorconenta,The calculation of spacing arrangements for the reinforcenent is simplified by defining a spacing constant 0 for the slope, in terns of the winimun spacing v to be used, RB ° tw o Using 0, the relationship between the maximum allowable spacing and depth z below the crest Of ‘tho slope can be reduced toa line of Gradient unity in the non-dimensional form Shown in Pig. A. v/s, ° os 1.0 ime “e ideal spacing vy — minimum spacing Q = Spacing constant Pig.14 Variation of spacing sy with depth fo load reinforcenent layers equally. in practice, the available spacings may be selected ex rultiplee of the compaction layer thickness. Setting this thicknesr equal tov, the meximim depth 9 which reinforcament Spaced atv, 2V, 3v.,.av may be used can be calevleted fron the equation, Fy = OL (a) to use reinforcement efficiently, the largest allowable spacing should be used et any depth, fo give the step wise distribution shown in Pig. 15. A tabular procedure ean be used to calculate the numerieal valves Of reinforcement layer Gopthe below the slope crest. Start at the bottom of the slope (depth H) and place the first reinforcenent layer at foundation level. The minber of leyers in the first zone of equal spacing cap be calculated by dividing the thicknese of the zone by the required spacing af the reinforcement. The reault 35 Unlikely to be a whole number, and at is Sufficient to. round own to ‘the nearest Snteger number of layere, end add the Fensining ‘thickness to the’ overlying zone; repeating the process to the top of the slope: EMBANKMENTS Allowance for vertical surcharge Vertical surcharge ws on the slope crest ie treated at an additional thickness of filly and an increased slope height i" used for design chart calculetions, wens 3 When calculating the reinforcement spacing by the tabular procedure described above, the Gepth “of fi] representing the surcharge should he subtracted tvom the thickness of the zone Of equal spacing at the top of the slope. ° ons v/s, 1.0 4 Os 2705 92 aol depth 1 spacing &y Pig.15 Bones for reinforcement layere spaced equally at v,20,30.-+- nV. SOIL AND POLYMER GRID PROPERTIES FOR DES Gk A vertical stress in soil can give rise to an inclined shear stress and a horizontal tensile strein. If the shear stress is insufficient to maintain equilibrium in soil” slope, the introduction of horizontal reinforcement may prevent collapse, If there in no. slippage Between the foil and the reinforcenent, the soil will strein until it produces the ‘force heoded in the" veinforcenent to” attain eguilibriun. ‘he relationship between soil shear stress and horizontal train can be derived from triaxial or plane strain tests. Limit equilibrium analycie can give the horizontal out of balance Stress in tho soil. Thos the force needed in the reinforcenent can be expressed as = Hunction of etrain in the gone of maxim shear stress. The slope of the load/extension Line (the stiffness) of the reinforcement and the degree Of compaction of the sand, together with the 7JEWELL ETAL amount of reinforconent, will determine wheth equlibriun is achieved ‘before the sand ha Strained to give its peak strength or whether equilibrium vill depend upon the sand strength at large strain, When choosing values of soil and ‘reinforcenent properties and factors of safety for use in design, consideration should be given to. the relationship between the stiftnest of the reinforcement and the coil rengeh characteristics. ‘Taking the oxanple of the polymer grid Tensar 5R2 at 20%c, the work of McGown et al (1984) Indicates that at creep loads lover than 2 Kni/m the extension will be proportionally loa than 108 within 120 years." Extensions in the Fange of 4 to 78 night be expected depending on the design force BP. Howaver, the loading on the Yeinfncenant is unlikely to he constant throughout its working Life. ‘The reinforcement will be loaded during construction, first by the compaction effort and subsequently by the height of fill and any Surcharge. Equilibrium will be estabiiehed by the 904) tending to shear and thus extending the reinforcenent until the balancing force 1s provided. The initial response of ‘Tensar SR? Feinforoement can be estimated from the isochronous creep curve for a tine coreesponding “to the end of construction MeGown et al (1984). The loaded grid will extend further during the life of the structure Allowing an increase in horizontal strain in the soil. Generally this will increase tne shearing resistance of the soil and reduce the Balancing force needed in the geogrid. At worst the soll will reach its angle of shearing Fesistance with tensile strains of the order 10% and the grid would subsequently experience pure creep. At this stage in the development of reinforced Soil design, prudence dictates chat for polymer Feinforcoment’ grids, the soil strength at large” strain should be used to. check eguilibrium, ‘The design force for the polymer reinforcement Should take account of construction effects, environmental conditions in tha soil and the eefects of tine on mechanical properties during the design life of the structure. An additional overall safety factor of a conventional magaitude 1.3 to 1,5 would provide for the uncertainties of "construction alignments, unexpected external or pore water Bressure loadings and inadequacies in” the Current understanding of reinforced soil and Gee anaayst SELECTION OF DESTGN VALUES Design values for parameters should be used Girectly with the” chart design procedure. Guidance on the selection of design values is given below. 78 Geometry ‘The slope height and slope angie can usually be well defined and the expected volues used for Gesign. ‘The maximum expected value of lords which can occur simultaneously should be taken. "the soil Gait weight, pore water pressures and curchargo Joadine may’ ail attain their naximom values at ‘the sare’ tine, strength A large atrain value of soil strength would be appropriate for design with polymer arid Feinforcenent, and the critical state friction angle OP". may be used. Por. long Lived clay Fills it®yould be appropriate #0 asume ero effective cohesion. sol. Polyner reinforcement strength ‘The strength appropriste to service conditions in the ground at the end of the design life Of the structure would be suitable for design, The value chould” apply to the most severe conditions for the reinforcement that are expected. overall safety factor A conventional overall safety factor of the Order 1.3 to. 1.5 wuld be appropriate, Por Simplicity with the chart design procedure, the overall safety factor con be epplied to the Sesign value of reinforcement atrength to give a factored design force. conctustoxs Limit equilibrium analysis and local stress, calculations may be used together for the design Of reinforced soil slopes. A number of criteria Gan be eatablished to ensure the satisfactory performance of reinforced soil. A set of design Sharts for reinforced slopes have heen developed together with a procedure for their use, and these are sunmarised in the appendix. the shear Strength of the soll at large strain, and the Strength of the polymer grids in ‘the soil environment at the end of the design life are Suggestea as design values to be used with a Gonventional eatery factor. ACKNONLEDGEMENTS The two first named authors have benefited fron membership of bath the Soll Reinforcement. Design Group and the main steering Committee coordinating the “Science and Engineering Research Council's sooperative rasearch project with Netion Limited into the Civil Engineering applications of polymer zeinforcenent grids. The work on design chares was carried out as a. part of this programmes Grateful acknowleagenent is given to the menbers gf these two. committees for many stimulatingEMBANKMENTS 1 eriue worsoEss TIS 1 ovbue uoraorss 110s & 8 3g at as R Ras c 2% ' t I ef it | ea 1 g T r 8 { = a4 6 so és 6 = ao a a aoa 36 o6 f/a ausreu adore AnbueT wweunozorUTOR 2/1 Subrey edove/sabuey auemoropTe \p evbus wor.aras 1508 4 othue worsors5 110s é \ 6 3 3 6 6 6 » 4 auerorz000, amend as x aueporsz000 meena Gam 60 70 €0 Slope angle § 50 40 40 6 7 80 Slope angle # 30 79JEWEL ET AL Barth pressure coefficient om 3a 1 5. 0. 20 ae g 20 3 a6 z a Baa 4 4 faa a ' 6&3 é p 3 é 4 B20 ink Eo gue | 5 gq 3 gas ‘a daa P 2s La z Lo 304-0 3080 eo 70a” Slope angle 6 Slope angle & CHAR? DESIGN PROCEDURE 2. Select the required enbanknent dimensions and surcharge loading 2. Select design values of soil properties and pore water pressures 3. Determine earth pressure coefficient K and length of reinforcement 1 from the charts| 4, Choose "in soil" design strength properties for the reinforcement and an overall factor of safety 5. Obtain the factored reinforcenent force P 6. Choose a minimum vertical reinforcenent spacing and the spacing constant Q-P/Kyy 7. Porform tabular calculation for the number and spacing of reinforcenent layers 8, Calculate the total horizontal force required for equilibrium T=hRyK? 9. Check T/(nunber of layers

the presence of an upper foundation crust or strong layer which also may influence the mode of enbanknent Gaformation, Design mathade aze proposed to allow for those factors. Tt ie suggosted that, for consistently Fellable design, the analysis of fille over weak soils Should be baged'on the Lower Limit of the foundation, Strength available in situ, ox on tho large strain Strength. eraoo OF ANALYSTS Conmon practice in aesessing the stability GF embankinents on weak £0undations has involved the use of both effective stress analyses and total stress (¢ = 0) analyser. The advantages and disadvantages of both these Rethods have bean examined in detail (avenas et al, 1980 and Palot et al, 1982). Stability analyses baged on a total str approach aveld sone of the problems in eetective stress approach invelving ab estimation or even assumption of pore pressures fand the degree of mobilization of the shear Strength of the foundation sols, but problems Still remain concerning the mobilized strength Of the embanknent fi21 and, in particnlar, on Whether the Undrained shosring resistance of a foundation soit, as measured on intact samples or by in situ vane tests, is representative of the strength actually mobilized in sity, (Bjerrum, 1373). SHEARING RESTSTANCH OF THE FOUNDATION sort ‘ho critical period vith respect to 411 stabliity on weak foundations 1s generally immediately following construction. mm this period, tha soil is essentially undrained and in'its weakest tate. With tine, by consolidation, water is excelled from the pores of the goll and ite strength increases, Te is common to measure so!l shoaring kesistance in an undrained condition and to lise this minimum design (undrained) strength in stability analysea. Te is now recognized that any factors agfect the design undrained Strongth and that undrained strength is not, infact, a unigue property of the soil deposit! “sone of the major factors whieh affect indeained strength are as follows: © Variation with depth - As shown on Pigure 2A, ‘the undrained strength of a soft elay deposit typically varies with depeh, Ih the surface érust, 1905 95MILLIGAN AND LA ROCHELLE 96 Wich hae Joon altured by natural Weathering processes, the undrained strength 1s high but decreases with depth to a minimum balow the base of the crust. Below this depth, the undrained strength generally increases in a uniform manner, With the rate of increase in strength Sependent on che stress history. and the soll type. ‘this type of variation in strength with depth 1s common in soft clay deposits. Weak layers - secause of the nature of Seposition, it is not uncommon for soll for weak deposits to incorporate layers of weaker soils, (Pigure 2A). These Weak layers, if present, are difticult to detect and can control stabilicy (teonards, 1982). Even in cases whore discrete woak layers are not present, the gone Of minimum shear strength below the erst represents a weak sone which ig'fonnd in most deposits. Rate of shearing ~ Laboratory and field strength tests are often carried out at rates which are much more rapia than actual field loading (ive. rate of Ell Consteuction). For example, field vane fests are carted out at rates which are up to 10,000 times as fast a actual £117 constriction. Tbe undrained strength measured in common Iaboratory strength fests 2s dependent on strain rate and typleally exhibits 2 100 - 15% increase in strength for each order of magnitude dnerease in atrain rate. Anisotropy - As shown on Figure 25, the Undrained strength of soft clays depends on the direction of shearing (i.e, the undrained strength is anisotropic). The degree of anisotropy decreazes with Increasing plasticity, The complex oda of failure associated with the connon field vane teste cosults in a trend wnich does not reflect other tests in Which the boundary conditions are better defined. Strain Dependence — Many soft clays are strain-softening (Figure 3)~— In most Situations, fills ase designed euch that the actor of safety against. failure 18 dow, fe hoa), 8 thug condision, a Linge (portios of whe foundation ls deposit is overstrossed (1.2. the Smpoxed shear stress excoeds the shear strength Sf the soll). therefore, if the cou]. ta Etcdin softening, Large portions of the foundation soll will have reached at Failure a lower “Large strain” serength So that the naximan shearing resistance Soon not apply throughout she Geposit, Further, progressive fatluce will cok Song the failure surface as load ‘te Eranetarred from the over-stressed gongs to adjacent. Minder-seressed” onan gen rein geass pun sk Cre way Factors affecting undrained shear strength Fig, 2 BRITTLE, STRAIN SOFTENING g 3 Wow rT, STRAIN a Derornation, x (AFTER UA ROCHELLE ETAL TH) Fig. 2. Types of stress-strain behaviourAlthough significant research has been carried out on the strength behaviour of weak soils and, in particular, on soft clays, there are Stil unresolved problems associated with the prediction of stability of fills on soft clays. ‘This situation is well illustrated by the bredictions made by various researchers ‘since 3960, (Figure 4). Bishop and Bjerrum (1960) established the validity of the @ -o (short term-uncrained strength} approach by back analyzing a suber of failures on clays; as Shown on Pigure 4A, this approach appeareé to Provide a reasonable solution for each of the Gases studied and became a common method of analysis in geotechnical engineering. By the Garly 1970's, Le decane apparent that this appreach didnot alvays provide the correct Prediction, aa indicated on Figure 48 (Bjerrua, 1973). The magnitude of error in computed factor of safety increase with increasing plasticity when the undrained strangth was Reasured using the field vane. This trend led Bjerrum to propose an empirical correction factor to be appiled to the measured vane Strength. This nodification was chviously Conpeiling and was adopted widely. Rowever, exceptions which did not conform with Bjerrtm's correction factor were reported, (Pigure 4C) and questions arise as to the advisability of Using Byerrun's approach, based on both Fundamental and practical considerations. 1 ea a, 1 eattnonasery ou EMBANKMENTS approaches are described below and ara sinilar in one respect ~ the undrained strength pro- posed for usa in analysis is related to the Raxinum past vertical effective stress (Pc) on the 8011. Bjerrum/Mosr4 - Mesri (1975) rednter- preted Bjerrum's original data related Eo the strength at failure of eoft clay Goposits (Figure SA) and pointed out that when Bjorrum's correction factor wos applicd to sy and pe plots, there wae @ unique ratio of Su/po Of 0.22, Andependent of plasticity" add and Foot (1974) have recognized that the normalized direct simple shear undrained strength as determined on Sastructuxed clay 1s the average Strength mobilised in the foundation of ‘an enbenknent. on soft clay and have Succasstully used thig strength value to explain failures (Figure 55). As observed by Trak et al (1980), for Slightly overconeolidated cohesive soile With an cverconsolidation ratio between 1 and 2, the ratio s (DSS) /p- varies betwoen’0.18 and 0.25. La Rochelle et a1 (1973) reasoning from the premise that at failure, the strength mobilized in the soft clay Foundation would everywhere be equiy- alent to the strength at large strain, have shown that the vndrained serengt obtained at large strain (USALS) in laboratory compression tests, could kucceeefully be used in stability analysis of enbanksents on soft clays. it was later. shown by Trak et al (1980) that the ratio ¢4(USALE) /p. varios between 0.2 and U.25- Fig. 4, Undrained strength stability analyses Based on the above, if is obvious that the Selection of undrained strength to be used in Stability analysis 42 a complex issue.” What then ts the "correct" undrained strength? This question has heen addressed by a number of fosearchers using different approaches and surprisingly, there isa reasonable concensus Of opinion. ‘three of the more significant fF bb Fig. 5. Definitions of undeained shear strength at failure 7MILLIGAN AND LA ROCHELLE hus, se woulu appear that the unarained Strength at failure can be taken to Lie in the Fange of (0.2 ~ 0.25) Bo. However, the geo fechnieal profession as°a whole has not as yet adopted this or any other unigue definition of fsilure strength for defining the strength to be used in the analysis of embankment stability en soft clay. Rather, strength valuse measured ih a variety of ways are used, somotinos indiscriminately, and which 1A some cages, do nor reflect the actual field strength, PATLURE MODE In the najority of stability analyses an@ back analyses of failures, it is assuned that the failure surface is circular Walch 1s ¢ kinematically admissible failure ode (igure 6h). Possible errors in back analyses related to thrce-dimensional effects are Giscussed by Aszour et al (1981); error related to the form of expression of the factor Of safety as related to the different balance bE forces in varying methods of Limit aqull~ ibriun analyses have been exanined by Frediund etal (1981). Mowever, there are many instanoes in waich the sestmption Of a circular surface does not account for other inportant factore such a6 the presence of a woak layer, or a zone of mininum shear strength Jocated inmediately below the crust. Aetemps to fit circular failure surfaces passing through this zone can lead to unrealistic failure geonetries within the £11 and beyond the toe of the £111 slope. The assumption that failure surfaces are ciroular often appears to be congistent with surface and Subsurface evidence of failure (1.0. back ep, toe heave extent and depth of signif icant deformation in the soft clay foundation material); however, ag show on Figure 68, Other modes of failare, and in particular, predopinantly translational movement, are’ also Consistent with this physical evidence, In ‘Sone cases monitoring of lateral deformations in the soft clay clearly indicates concen- tration of deformation within the zone of ‘Ainimun strength below the crust. Sor OF Wig. €, Medes of Failure 98 it is considered that translational faiture modes are most likely to occur when the crust f2'thin and there ie'a pronounced zone of minimum shear strength below the crust.” On the other hand, cireviar feilure modes would be associated with thicker cruste and for uniform strencth-depen profiles. Analyses incorporating non-circular and translational failure modes do eiat (e.g. wedge analyses) Dut are used less frequently than clzeular arc analyses, Further, many Of the assumptions incorporated in the method Of slices, which are common in both cizcular and non-circular analysen, are not necessarily applicable to analysis of’ translational failure modes. ME EFFECT OP REINFORCEMENT 70 FILLS Based on limited prior experience, reinforce- ment at the bass of fills improves their stability on soft clay fourdations. An early exanplo of reinforcomint of #113 on soft ground is the "corduroy" road in peat bogs Where cut éreos ara laid side by side acros the intended route and Fill is placed on top of this "corduroy® arrangement.” However, Gaytlon should be exorcised intern of Snterence draw from this example, since the aode of failure associated with low road £1115 on peat is generally related to excessive Geformat ion as opposed to rupture of the foundation soils ~ the case with soft clay Foundations. The prodoninant effect of the sorduroy is to keop the [111 intact ana to Prevent distress to the road pavement asa Fosule of spreading (lateral displacement) of the fill, in association with deformation of the undoflying peat, it should also be appreciated that while corduroy roingorcenent OF low road Fills on peat may help to keep the Fill intact, it will hot reduce fand in some instances could increase) overall deformations. ‘There are a nusber of reasons why reinforce- ment of the base of fills enhances their Stability on soft clay. for example, reinforcement tends to reduce "splitting" of the Fill and therefore increases whatever Contribution the fill strength makes to Overall stability; however, it is considered that the major effect of reinforcement 18 t0 cause a beneficial change in stress distrib Gtion in soft clay foundation, The hypothesis is as follows: Consteuetion Of fills on « flat ground surface induces movement and associated shear stresses are Significant, ae indicated by the data obtained during construction of the Gloucester fest I111, (Pigure J). the effect of incorp= orating réinforcenant at the base of the fll fg to reduce tho magnitude of these shear’ Stresses at the tntevface. Thus, as the £411 tends to spread, the reinforcement 1s put into tension and, if sufficiently stiff, will reduce the taynitude of displacenent, and, therefore, result In a lower shear stress at the €Ll1/ground surface interface.20 1 DURING CONSTRUCTION z 2 9, c 2 4s _ 8 Ww DISTANCE FROM CENTRELINE m (arren 9010206 ano eDKanDS, 72) Fig. 7. shear strose at f£411/oround aurtace intereace (Gloucester Test Pili) ‘This change in external boundary stresses can Ebsult ina significant change in the stress Sistribution In the foundation material by {ntrodveing a horizontal tensile stress con ponent and by preventing rotation of the Principal stress directions at the ground Surface. Tt should be noted, however, that the vertical stress distribution will be greatly influenced by the rigidity of the Eeinforcoment and may Femain practically Unchanged {or roinforceaent of low rigidity (e.g. conventional geotextiles).. To illustrate the lefluance of reinforcerent, Figure 8 shows typical strese conditions on an element of Soil below reinforced and unveinforced filis for the ease of perfectly rigid reinforcement. Tho combination of increased total horizontal Stress and reduced boundary shear stress Causes 4 reduction in the maximum shoar stress Gn the soil element, However, the magnitude Of the reduction in shear stress associated With reinforconent decreases with depth. for Ssite whore the surface crust is thin and a typical strength profile with depth exists, the consequence of the decrease in shear Strest magnitude at shallow depth is important, Since the sone ef highest shear stress is 10 longer coincident with the zone of minimue ar strength. ‘This vesules in a deeper failure surface through more competent faterials and, as a result, a higher factor of fety (nore stability) for a given £411 height. ‘The corollary ie that on a given site, a nigher £111 can be constructed before Failure (F = 1) occurs. EMBANKMENTS, a a 5 ® BETS RANT Renee Fig. €. Tlustratlon of effect of Tenaar reinforcement on shear stress in foundation soil EXAMINATION OF EXISTING DESIGN METHODS FOR REINFORCED PILLS A variety of methods of analysis of reinforced Snbanknents on weak Foundations have been Feported in the recent Literature. The maj- Ority of these methods employ total stress (= 0) Limit equilibrium concepts to assumed circular are failure surfaces (Brakel et aly 19827 Fowler, 1982); however, sone assune tires part wedge modes of failure (Enka, 1978). the force applied by the reinforcement has been variously assumed, either to act only on the failure surface or to act tangentially to it, ‘Sewell (1982) soparates in his circular are analysis the distribution of the reinforce~ ment force required to provide equilibrium. With a apecified factor of safety (F) on tha obilized sos shear strength and the ai Eribution of sobilized force along the reinforcement at that value of F available to provide equilibrium. If the required forces fre everywhere less than the available then the embarkaent hes as a minimum, tho value of F presumed. Tt is evident that there are a variety of ascunptions Used in practice for Limit equilibriun analyses of reinforced embankment. Stability on wenk foundations. We can query ie detailed assessnents of the anount and Girection of the reinforcement force are justified 1f the assumptions made with respect fo undr. ‘ned strength and failure mode can be an error. Therefore, it 1s considered that singe the stability Of unreinforcea Fills on 99MILLIGAN AND LA ROCHELLE soft clay cennot aa yet be predicted with absolute confidence Using conventional Lint Equilibrium approaches, modification of these existing methods to indlode the effects of Feinforcement can be subject to the sama errors and uncertainties. ‘EXAMINATION OF A PROPOSED DESIGN MicTHOD Constitutive models to define stross redistrib- lution effects in the £111 and in the foundation Soll attributable to reinforcement, have not as yet been developed. At present, thare does not Zeist an agreed reilable method of analysis Which will properly account for the effect of Eeinforcesent although solutions will undoubt~ edly develop as a result of ongoing research. Re this stage of our knowledge, in using Teinforcenent to enhance embankment perform ance, we therefore have to rely on instinct ‘and growing experience. As a guide, it appears that the best available method is t9 adopt 3 nodified limit equilibrium method, despite the uncertainties discussed previously, Such a method must recognize the overwneiming inportance of a rational assessment of the foundation strength, It assuses that firstly, the total stress (#/= 0) approach is applicable for first stage construction; secondly that the lower Lint of undvained strength te adopted, as defined by an Sy/,_ value of about 0,2 to 0,257 thirdly, that the mode of failure should fit, vharevar possible, to known oF Suspected weak layers within the foundation. The factor of safety assessed by this model Should then consistently represent a lower bound and give a safe design nodel. Sinple Limit equilibrium equations suitable for expression in 2 dimensionless form can be developed as shown in Pigures 9 and 10. The equations assume 2 circular arc or transiat~ ‘onal failure through the clay foundation and & Coulomb failure surface in the fill. The assumption in the £11] allows the failure wedge to be replaced by a single active force Pa and with this configuration the contre of the erit- Seal esroular are can be shown to be on the vertical line through the mid point of the Slope. A computer search 1s necessary to Tocats the vertical distance from the ground surface to the centre of the oritical are- Figure 9 prosonts a gonoral form of tho equation sn which the clay foundation has a Srust thickness t, and average shear strength Cr. Under the crust, tha clay is assumed to be near normally congolidated with its shear strength increasing linearly with depth. The Toner bound of Sy/p_ 48 Likely to be oz. 6 Por a shallow depth of crust or of a clay foundation in relation ¢o the embankment’ Ainensions the failure mode will tend to be translational (refer to Figure 6). ‘The din~ ensionless equation for translational failure Satistying force equilibrium and assuming constant shear strength in the foundation is 100 shown in Figure 10, A comparison of the two alternative failure modes (clroular arc and frenslational) for unreinforced embankment stability, with the more rigorous method dev= eloped by’ Sarma (1973) indicated good agree~ ment for the circular are mode at'D/ti = 1.0, With the translational ode being more consere— ative at all values of D/H. Fer = 22 BE oni peat qBervaiten i Az cweernar oy 08 wees ref ae EA te Fig, 9. Circular are failure: general analysis aaa, Fig. 10. Shallow translational failure Tt can be recognized that the method proposed N{gealizes” to some extent both soil and reinforcenent behaviour and ignores possible Strain dependence for either tha soil or the Feinforcenent element) this presumption is however essential for the limit equilibricn approach and, as discussed below, for most eabes if reasonable. Strain Compatibility of the Reinforcement and Foundation Soil he strength and stiffness of Tensar geogrida, as with all other polymorics, varies with temperature and strain rate, Strength and stiffness incre:tenperature decreases. Freezing conditions therefore do not present a problem while the standard reference temperature of 20°C is un~ likely to be exceeded in temperate climates where the material is buried. woth strength end stiffness decrease with decreasing strain rate.” In addition, such material exhibits a tendency for brittle Fracture at som extension beyond the point of ductile yield, The steain, afd therefore the Stress, induced in an elencnt of reinforcenent ie depéndeat on the strains induced in the soil in the working condition. Limiting the desisn Stress to some Fraction of the yield stress does not, in itself, ensure that brittle frac~ ture will not occur; “if strains are suft- Seient, the stress level,could be high enough to cause continuing creep and eventually brittle failure, within the life of the structure. It is therefore « necessary part of stability calculations of Tensar reinforced ‘enbanknents to ensure that strains in the zone df the reinforcenent are such that brittle feacture vill not occur during the life of the structure, Currently available data suggest. that the maximum strain allowable, "the per~ formance Linit strain" for Tensar SR2, for 120 year life, is 10 per cent. tye Eactors relate to the effect of excep in the reinforcement i) As the height of the enbanknent reaches See Final value, the resulting load will produce eréep both in the Feinforcement and the foundation soil, Tf the creep rate in the reinforcement 4s larger than in the foundation soil, ‘sone oad Will be shed from the rein~ forcement into the foundation 031, and vice versa, where a higher creep Fate would be developed in the soil. The load transfer in either casa will Gease when the stress levels in the Feinforeenant and the soil will be Such that both creop rates Will have the ‘ane. value. 41) The strength ef the foundation soit and therefore its capacity to resist the strosses imposed by the embank~ ment, tends to increase with time due to consolidation. Az a consequence, the stress level and the creep rate in the foundation soil will decrease, with a resulting increase in the load transfer f¥on the reinforcement to the soll. Available data for embankments on soft clay foundations indleate that for factors of safety (6 = 0) in the range of 1.2 to 1.3, fhe maximum lateral strains in the aubsoil are several per cont ubere the masini Vertical deformation is about 10 per cant OF the embanknent height. (The average 1at~ eral deformation is 18 por cent of the maximum Settienent for sensitive soils (Pavenas, 1979) J." The performance limit strain on Toncar §n2 is Unlikely to be excoaded in these EMBANKMENTS applications and Linit equilibrium methods of Shalyeis oan thus be used safely to calculate Stability; however, whore settiomant is anticipated £0 exceed about 10 per cent of ‘the enbankuent height, the design should Include a detailed aséessnent of strain, and Linit equilibrium methods must be used with cautions 4m many applications, it is probable that Btraing will be insufficient to mobilize the fall design capacity of the reinforcensnt. ‘This situation 1s conesetent with most apo- Lications of Limit equilibrium analysis an soil mechanics (e.g. the uso of active and paccive pressures to calculate the stability Of rigid retaining walls) and the usrobilized capacity of the reinforcement can be consider~ ed as providing the required security against failure. ACKNOMEEDGEMELITE ‘The authors thank J. HM, A. Crooks and J. R. Busbridge for their critical coment on this paper ané for their significant input to the review of existing practice in the analysis of fille on soft clays. REFERENCES Agsous, A.S.; Balich, MM, and uc, Ladd (2981). “twnree-Dinensicnal stability Analyses of Four Rnbanknent Failuros™. Proc. 10th Int. Conf, ISSMPE. Vol 3, Stockholm. Bishop, A.W. and Bjerrum, L. (1960). "he Relevance of the Triexiai Test to the Solution of Stability Problems." N.G.r. No. 34, Oslo. Brakel, J., Coppens, ¥., Maagdenberg, 8.C., and isteuy, P. (1982). *Stabliity of Slopes Constructed with Polyester Reinforcing Fabric, Test Section at Almere Holland 1975.*' Proc. 2nd Int. Cont, on Geotextiles, vol 3, tas vegas, Bjerrum, 1. (1973). ‘problems of soit Mechanics and Construction on Soft Clays," State-of-the-Art Report, Session 4, Proc Gth Int. Cont. ISSMPE. Vol 3, Mancow, Bozoauk, M- and G.A. Leonards, (1972). "the Gloucester “est Fill.” Performance of Earth and Earth-Supported structures, Vol 1, Part 1, ASCE. Enka, B.V. (1978), "Design of the Suez Breakwater." Gnpublished Report. Fowler, J. (1982), "theoretical Design Con- jeerations for Fabric Reinforced Bnbank- nents." Proc. 2nd Ine. conf. on Gectextiles, Vol 3, Las Vegas. Frodiund, D.G., Krahn, J., and 0.8. Pufabl, 101MILLIGAN AND LA ROCHELLE (2981), the Reletsonahip Between Limit Equilibrium Slope Stability Methods,” Proc. 10th Ine. Conf. 188MPE, Vol 3, Stockholm. Jewell, R.A. (1982). "A Limit Bquilibriun besign Mathod for Reinforced Embankmants on Saft Foundations." Proc. 2nd Int, Conf. On Geotextiles, Vol'3, Ias Vegas. a Rochelle, P., Teak, B., Tavenae, ¥., noy, HM, (1974). "Fallure of a fest Embankment on 3 Sonsitive Champlain Clay Deposit." Can. Geot. Jour., Vol 11, Wor te Leonards, G.A, (1982). "Investigation of Pailufes."” ASCE Jour., Geot. ung. Div. vor 108, G.7-2. Mesri, G. (i975). Discussion on "Wew Design Procedure for Stabilsey of Soft Clays" by C.C. Ladd and R. Foote. ASCE Jour., Seok. fing. Dive, VoL 101, 6.7.4. Pilot, G.) Trak, B. and P. La Rochelle (1962). “uttective Stress Analyses of the Stability of Enbankments on Soft Soils." Cdn. Geot. Jour., Vol 19, No. 4 102 kaymond, G.P. (1972). "prediction of ‘Undrained Deformations and vore Pressure in Weak Clay Under two Enbanknents.” Geotechnigue, Vol 22, No. 3+ Sarma, §.K. (1973), "Stability analysis of Enbankoont and Slopes.” Gastechnigue, vol 23, No. 3. "avenas, F., Lerouetl, 8. (1980). "the Behaviour of Enbankmanes on Clay Found ations." Cdn. Geat. Jour,, vol.17, No. 2. ‘Tavenas, P., Mieuesens, Ci, Bourges, Fee (1979), "' "Lateral Displacements in Clay Foundations Under Embaakments." Cdny Goot. Jour., Yor 16, No. 3 tavenas, P., Trak, Bs and 8. teroued) (1980). "Remarks 'on the Validity of Stability analyses.” Cdn, Goot. Jour. Vol 17, No. Le ‘Trak, B., Ua wochelle, P., vavenas, Fe, Terouell, §., Roy, M. (1980). 4A New Approach to the Stability Analysis of Bnbanknents on Sensitive Clays." Can. Geot. Jour., Vol 17, No. 4.‘The use of a high tensile polymer grid mattress on the Musselburgh and Portobello bypass . Edgar, Lothian Regional Council Highways Department EwFRODUCTION ‘The Nuaselburgh and Portobello Bypass, when comploted Will Tink tho-A1 trunk road and the proposed Eesaburgt City Bypass with the eastern industrial arene of Edinburgh and the port of Lotth, eliminating the present heavily congested areas of Musselburgh and Portobello, During the design of the Bypass the problem of ensuring the stability of a 18 metre high enbanknent, over woak ground was encountered. Ivo alternatives Were considered: the excavation of the unsuitable material and ite replacement vith & suitable rock f111, or the construction of a polymer grid mattiesa over the Weak material at ground level. Costractore were invited to price both these alternatives with ihe Successful contractor electing to construct the Dolyner grid mattress which he considered tbe nore Sconomie and satisfactory solution roundabout adjacent to the Asda Superstore. ‘Ag can be geen From the sbove route description, From its commenconent at the Old Craighiit Road, the bypass rune in erly direction across easeatiaily Level arable farmland, croseing two fallway lines and one road. After crossing the Mitternii1 Marshalling Yard goods Iino whieh is carried on a 6 m high enbankment, the bypass uns north westerly ucross an arca of marshy low 1ying ground drained by several anell Streams, before crossing arable farmland to the AGOSS Neworaighall Road, After this road crossing the bypass crosses the eastern section Of the existing colliery spoil heap (iencraighall Bing) from the disused Neweratghall Colliery before passing over derelict Ian and the railway Lines from Niddrie Junction. “The bypass then continues northwards crossing the tdinburgh Suburban railway line and the valley of the Niddrie Burn, betore joining tho previously constructed and from Fig 1, the bypass crosses several roads and railways and as a result the majority of the road Will be carried on eubankmants up to & maximum heighe of approximately 18 metres. ‘The site investigation carried out revealed the Line of the road to be generally underlain by Blasial clay with a localised covering of fourse granular reised beach deposits in the east, Laboratory tests showed that the Strength of the above materials Was af Support the enbanknents proposed. However in the area of the marshy ground adjacent to the Millerhitl Marshailing Yard, where the road eubanknent reached 19 maxiqum hesest of 19 metres boreholes and subsequent Inboratory testing revealed strata of lov strength to depths of § metres below existing ground level, Calculations based upon the results of Hig 1 ~ Location Plan Polymer grid forcement. Thomas Telford Limited, London, 1985 103EDGAR laporatory tests curried out on the above low Serengeh maverials produced factors of vafety against foundation failure of the order of 1,0 2,0, This was not considered acceptable and ir was decided to Investigate alternative Solutions, to inerease (his factor of safety. SOILS INVESTIGATION: (4) Bite Deseription:~ Tho aroa of bad marshy ground over which the enbanksont is to be constructed 18, bordered to the north by the oubanknent of the Wiilerhill Marshalling Yard, and opens to the south, east and west on to gently sloping aable farmland. The marshy ground ie drsined by a ctroam running in an easterly direction which is fed by Several small streams (‘spreads’) tovards the north of the site, At the northern end of the site, an existing culvert carries the stream below the railway embankment, agen 1 Boreboeral postion |& Machitos pode pesion nao bounary fst aes Fig 2- Borehole and Trial Pit positions 104 (41) Ground Conditions:- Boreholes wore sunk by shell and auger and rotary open hole und core drilling methods fogetiier with two Trial pits excavated mechanically, The position of all these Boreholes and trial pits are shown on the site plan (Fig 2). Goneraily veneatt 0.20 m to 1.10 m of topsoil and/or clay and ash fill, the Poreholes and trial pits (35-1 to S—4 inclusive, 822, N29A, R2Gi, ROP, N23C and RasH and 85-101 and 85-1P2} encolntered glacial till, with a local covering of Alluvial deposits, resting on sedimentary Bedrock. Boreholes R22, R298 and trial pit Ss-7P2 penetrated up’ to 5.2m of alluvial materials, genoraliy comprising very soft organic clay, soft to frm, firm and Tirm fovstifr candy silty clay snd clayey sandy siit containing gravel, firm and fitm to Stiff, oftes laminated, silty clay withaay sand partings and a thin layer of clayey Sand with gravel at the base of the deposit," Trial pit 34-12 terminated within the alluvium at a depth of 3.20 m, Boreholes $5-1, RI3A, RIF, RIIG and Rat and S8-TPL peastrated between 1.05 mand 2.10 m of weathored glacial ti1i comprising soft, soft to firm, firm and firm to Seite mottled very saady silty clay, or clayey suady silt containing gravel with local layers of medium dense Silty sand. ‘The weathered glacial ti11 and the alluvium wore both underlain by unwoathered glacitl till, consisting of Stitt and very stiff sandy silty clay or locally clayey Sandy silt, containing gravel, oobviae and douldors and Secastonal thin layers of lens of sand and gravel, Boreholes ¥298, 23C and R2SD penetrated 2.90 m of alluvisl material comprising soft to firm sandy ciay and firm to ptift elayey silt with gravel, rootlets and decaying vegetable material overizing soft to firm sandy weathered glacial clay and gravel.” The weathered matarial was Again andoviain at a depth of 1-10 m by firn to atizt becoming very stil? and hard ynvoathored giacial clay with gravel Ground-water was encountered during boring and excavation in boreholes R22, ROSA, R2ID, RISC, RIK, RIVG and triad pit 85-1P2'at dopthe ringing from 0.60 £05.30 m. “Stanapipes inserted in Dorelioles R22 and RZSA indicated a standing wator level of the order of 1:50 m'and.1.00 m, respectively, below ground surtace, Boreholes S51 to 85-4, R290, Rasr, R23H and trial pit $5-TPI did not eseouster any ground water Sunaasy From inspection of the boreholes logs and aaulyses of the laboratory results obtained fron the site investigation, the underlying strata below the proposed enbanknent can be summarised 8 £01 lows:~ (a) A hard crust of topsoil /vegetation to 4 maximun thickness ef 400 mn. (b) A soft layer of alluvial materials ranging trom soft sandy silty clay to soft clayey sandy silt, with a maximum thickness of approximately G netres: (e) Stitt to hard unvesthered glactal clay with gravel recorded to a maximum thickness of approximately 1O'notres in borehole 86-2, Sedimentary bedrock:~ sandstone, midstone: «a EMBANKMENTS: [Power] sawPLe [NATURAL [NATURAL APPARENT|ANGLE OF NO. [DEPTH |MOISTURE) WET |COHESION| SHEARING (nt [CONTENT Joensity | (kNimé [aesisTaNce| (al | tigi) 19) 22 v2 [roo | a2 | | vs0 | 39 ° ea 88 a a] Jes | 20 | 208 | a7 9 Rae [ss [a [eae] 70 3 nae [04 [28 | +23 _| a7 3 Rem [vs | 17 | 207) a6 7 2s_ | iz_| are | os 4 was fie [asf is ig is si [20 | ts} 243] 95 a 3a | um | 226 | sa 6 5s | on | 225 | 208 " ‘TABLE NO 1 - Results of Shear Strength Tests [BOREWOUE] SAMPLE EEFECTNE SHEAR STHENGTH NO [DEPTH [cOneSON | ANELE OF] Tm) [INTERCEPT | SHEARING c RESISTANCE eines | "is Rae | 1s > 35 20 5 mo ry 9 250 ‘TABLE NO 2 — Shear Strongth Paranoters in tarne bf Effective seveasos ‘Test results for the alluvial and glacial clay materials aro summarised in Tables 1 and 2 above, Using these tost results, it Was found that the factor of safety agaiast Foundation failure of the alluvial Materials carrying 15 metre high enbankment ranged trom 1 to 2, Expected consolidation settlenente in the alluvial atersal and the unreathered Glacial clay vere calculated to be 275 na and 85 mm respectively, vith the majority Of the Settlement occufring within the fiesta yours. DESIGN ALTERNATIVE: (2) The excavation of the sote material aud 4Us replacement with an imported GuLtable rockf {il material, (44) The partial excavation of the soft aterial and tho displacenont of the Femainder by deposition of suitable Fockfill until finally stable, (4ii) Drainage of the soft material, together with « controlled filling of the Snbaaknent to permit disexpation of pore pressure and thue'a resultant gain in Btrengt of the weak layer (39) The construction of a high strength feagrsd materess to carry the enbankment over the weak ground, with the granular 105EDGAR £411 of the mattress acting an a dvsinage blanket. ‘The second alternative was considered unsatisfactory since it would require extremely careful control on the placing of the rockfiil with no real indication of tHe exact voluno of materials being displaced. Tho third alternative tt was felt was unworkable due to the period of timo required to construct the 15 metre high embankment. This method would aso have required strict monitoring to ensure that the rate of filling did not lead co excessive build up of pore pressure: In order to inerease the rate of settlement of the alluvial material below the embankment and prevent the build up of excess pore pressure Guring its construction, tho geogrid mattress ‘option included provision for the installstion df vertieal band drains driven through the nattress into the alluvial material below. Caleulations indicated chat’ the six metre thick layer of alluvial saterial would sortie approximately 275 mm over a period Gf four gears, but by installing a system of vertical band drains with 2 triangular equidistant Spacing of 2.5 metres, 70k of this settlement could be achieved in two years. It was therefore finally decided after coneideration to include bo: alternatives (1) and (iv) dn the contract docusents, giving the contractor the choice of construction watch he considered the ost economic and applicable to his Programe. DESIGN PHILOSOPHY FOR GEOGRID MATTHESS: (1) Introduction From a study of the site investigation results and the proposed cross-sectional Becnetry of the enbsnkment, the problon fo be analysed 1s as shown'in Fig 3. He205H0 Bez og Pe Ee2zenind Sonapsitr cay Sate ieee Fig 9 ~ Enbankmont Cros Foundation soiis From Terzaghi's expression for the nett ultinate bearing capacity of long footings, at or below the surface of any S011 conforming to general shear: section and 4 nett = oNe + 4a (rg-1) +84 MJ... CD Using the chosen maxinun and minim vallies of cohesion and angle of shearing ; the Safe bearing capacity 43 Expression divided by the 106 (ay cas) Foquired factor of safety, plug the original overburden presgure, 82. ‘Therefore for design stability FolS=q nett vhere w is the S292 Tpplled tonal intensity. «2 The calculated factors of safety ranged from approximately 1 to 2, These wore well below the target figure of 3 and the tse of « googrid foundation mattress as a means of distributing the enbankment load and thereby increasing the factor of Sarety ogainet foundation failure was considered, Drinciples of Geogrid Mattress Netlon were approached by Lothian Regional Couneil Iighvays Departaent for advice on the use and design of © Suitable mattress Sor increasing foundation otanility Aavice on the analysia of « high enbankment on a thin layer of soft foundation material was obtained by Notion from Dr iichard Bassett of King's College, London The use of a rigid geogrid mattress alters the direction of the normal slip circle faliure plane by foresng it to pase vertically through the Rattrese and a6 a result, the slip Diane is forced deeper into the stiffer layers of the underlying Stift brown sandy stity clay, Before this stiff underlying clay could shear, a plastic failure would be initiated 10 the soft alluvial layer shove, so there fore the plastic failure in the soft layer i6 the critical design condfcion. The method of analysis reties on the following characteristics of the found- ation mattress’ (2) A high tensile strength to enaure that the full shear strength of the soll 1s mobilised on the base (b) Rigidity to ensure an oven distrih- ution of the load onto the foundation materiai. This 19 provided by the high teneile strength St the googrids and the adoption of f cellular construction. (0) High friction on tho base of the Mattress, The geogrid base of the attress allows the gramiler infil material to partially penetrate through the apertures, creating « rough underside to the mattress, Design of Mattress Reference is mde to Chapter 12.5 of "Engineering Pasticity'. W Johncon and PB Mellor (VNR) which deals with the plastic failure of material between two Fough, rigid, parallel platens, when the Blaten width exeoeds the material ThickwessUsing the above reference, the pressure Gingram for Mal? the enbarkment, after Adding twice the eckesion for the offect Of passive pressure beyord the toe of the enbenknent aud neglecting the effect of upthrust within 0.48% 2h of the toe, ts shown in Fig 4. oo Et} —$— Site — he Fig 4 - Press Adopting an average, value of apparent cohesion of 53 uN/m® for the soft layer, the loud to failure for halt the enbankment. 18: ire Distribution Diagram 4.57 x 53x 4.2 = 1017.3 (43,43/244.7) x 53 x 30.6 = 18302.0 le 52x 7.5 = 1755215 26871,8 KN Imposed loading from hal? the enbaniment r 90 x 18 x 20 = 9000 KN This tives a factor of safety against foundation tailure of 2.986 Tor 4 factor of safety"Foduced to 1.0 the equivalent apparent coheston value of foundation material would be 47.73 kN/m From a Nebr Circle construction ft can be shown that the horizontal Toad to be yesisted by the geogrid mattress in: te cu sin .@ Fron Table No 2 an average value of the internal frjction gn terms of effective pressures g! is 29", ‘Therefore the hgrizontal load for Cu= 17.73 kx/m®, with a factor of safety <1 is: © a0 sia 0° = 96,57 sym. run minimum condition and applies over the centre portion of the ‘enbankment EMBANKMENTS, ‘The worst 1ikely condition ie when the factor of-safety drops to 1.5. for the eubakment under consideration, tits would require a equivalent apparent ‘conesion of 26.6 KN/n?, hich is below the minimun values dbtained' in fhe laboratory ‘The embankmet Load foundation support pressure Giagran for this condition iw shown sn Fig’ eerie) Fig 5 ‘The oxtent of the plastic zone of failure is shown by the cut-off line at the foundation Support’ value of 9000 KX. Over this plastic zone, extending to approx- Umately 14:0 metres from the edge of the mattress, the horizontal load im the googria mattress’ is:— Ty 5 = 26.6 = 56.87 KN/a.run ‘sin 29° Ascuming @ working load in the SE- 2 Googrid of 25,7 kN/m, width, (approximately 30K of the lultimte load) the Strengths of 1 metre and 0.5 metre mattress cells are:~ A metre cells 7 = 23.7 + 28.7 = 40.46 KN /m v2 0.5 metre celle T = 2 x (28.7+25.7)=80,92 N/m: 2 ‘Therofore 0.8 motre cells aro required within Lé metres of the edge Of the mattress 1 within the plastic zone and 1.0 metre colls will suffice over the reniiador of the mattress. consraucrroN The foundation mattress was constructed using a base of Tensar S3-2 googrids, with Tenear 5R-2 googride forming the sides and dispbragms. The vertical diaphragms divided the mattress into'a cellular structure consisting of 0.5 metre cells within 14 metres of both 107EDGAR enbanknent edges and 1.0 metre eclis over the Chemical Resistant to all aatural eeatre portion, as shown in Fig 6 below. Resistance occurring alkeling and A selected filisng material was then placed ie acidic soil conditions. all the celle, Biological Resistant to attack by (4) Speesrication Resistance bacteria, fungi and vermin (a) Googrids — Tensur 8-2 Sunlight Resistant to Ultra Violet Structural Resistance aitack. saare che: () Inf111 Materiat Selected filling materiel vas crushed rock a0 or sand/gravel mixture, well graded from Semxiniin sae Of dem’ dove 8 73 un Sag Oe od, id oS sieve, with not more than 10% passing the Grid WE(5/0") 920.9 238.0 78 tm'steve. Te had a uniformity Goon | maa lack Sootficiont excooting 10 Drier to Properties: ereracting./ Tensile (3S) Programme of he Works Sereagencet/a) Construction work in the ares of the See eS eS. oubanknent mattress commenced at the Liters Roll 36.2 Scglaniag of" Sepreisr 156 olin the Chazaet Gonstruction ofa new culvert” to carry the ae SOUNTEG'ARRES SARE heh" Hoee fe taversa) ouyoron~ igh ‘Tene density ei Fig 6 ~ Mattress Layout 108across the area over which the mattress Was to be constructed. While this work Was proceeding & commencenent was made bn the matrass to the south of the Stroam, Thie nocestitated a tenporary crossing of the stream ass means of access for plant and materials, This access was Also required for the Construction of the adjacent ratlvay crossing, Once the culvert was finished end the Stream diverted, Work on the mattress was now able to'proceed uninterrupted towards completion. Tae above programme of the works is shown dlagramatically in Fig 7 below. Ser Sage poewoer, | Ber | conetract New Culvert] Csite prepertion Coneiniet Mattress Divert infil Stream [—] Instell Vertical Drains[ | Fig 7 - Programe of Works (111) On Site Construction (a) Since it was intended to construct the mattress over the existing ground without removal of topsoil Or vegetation, the omy preliminary Work undertaken wae the general Tevelling of the site, and the removal of any large obstructions: Such as boulders and the 1ike, (b) The base geogrid of Tensar $8-2 in Sm wide strips was laid out over the ite commencing from the southers boundary, These strips overlapped by 150 gn and were stitched together with P.V.C. binding, Transverse Giaphragns of Tensar SR-2 were then Attached to the base grid by Similar binding and the intermediate triangular diapnrages of Tensar SR-2 were {nserted and connected to the transverse diaphragne by steel bodkins. See Figs 8 to 10. EMBANKMENTS: Fig 8 - Typical Mattress Layout Fig 9 - Triangular Col? Arrangement Fig 10 - Coupling of Diaphragme 109EDGAR 110 «wo wo In{tiai assembly of the anttress cells was Slow and rather Baphazard, until a recognised Sequence of fabrication wae adopted and the work force gained in Competence at erection, No specific Tabour skills were regiired for tho Jointing and assembly ‘work, bur 4t was found that it vas essential That the sequence of construction was adhered to and consequently @ Tair degree of supervision was required. After the initial stages where progress wae siow it was Yound that the triangular diaphragms could be more easily inserted if all the transverse diaphragms were tensioned, A simple hand winch was Used for this operation, This Greatly helped to maintain the shape Of the mattress and to develop Sound Joints between the diaphragms, thos thereasing tho rigidity of the mattress prior to Infilling- Filling of the mattress commenced fron the southern boundary once sufficient cells had been construct— ed, and. progressed northwarde with the previously filled area of mattress forming « working platform for plant and machinery, A full circle, slowing back acting crawler excavator was used to fill the mattress cells, with the material Deing shaken and not dropped into the cells from the exeavator's 0.75 culm, bucket, The cells were gradually titled with the specified material, taking care to fill the first two fous of Gells to hair height and then the Tiret row to full height. Pilling was thes continued by this mothod, Susuring thet the lending row was always half filled tefore the trating row vas fully filled No compaction of the infill material was specified, and it was found that with the'well graded granular material used, no large void were formed, The only conpaction the material did receive was fron the Binat operating over it) and to ent any damage to the vertical Giaphragns trom thie plant, and to compensate for any settlement of the material within the mattress, the cells were overtilled by. Approximately 100 ~ 180 rm. On completion of the mattress, vertical baad drains were installed over a period of five days, trials previously undertaken had chown no Gifticuity in penetrating the base weogrid of the mittress with the mandrel sed for inserting the drains, This vas borne owt by the pase of installation experienced on site by the contractor. All the Grains vere driven to refusal with tthe maximum penetration boing approximately six metres, ‘compantson oF cosrs: Hight reputable, expertencea contractors were invited to submit tenders for the estinated £5 million advance bridge and earthworks contract. Within the earthworks section of the bill ‘of quantities, contractors were asked fo prise the following two options: ) ‘The excavation of 12,000 cu metres of uneustaple material aud its Feplacenent with & suitapie imported rock{i11 @) The construction of a 1 metra thick Reogrid mattress, covering an area Of approximately’ 4120 sq metres, ingitied with suitable granular frec draining material. ‘The three lowest tenderers for the whole of the works, priced the geogrid mattress option cheapest. The average costs based upon these three lowest tenders vere £87,000 for the nattreas option and £116,000 for the sxcivation option. Howewer the averages costs based on all eight tenderers prices submitted wers 299,000 and £86,000 respectively for the mattress’ and excavation options Advantages and Disadvantages of using a Geogrid a No excavation and minimal site prepar— ation. (41) No tips required for disposal of unsuitable exeavated material, (41) No difficulty working at or below the water table, (1v) Less prone to delays due to inclement weather. (w) No difficuttios due to having to exeavate unsuitable material adjacent to existing structures, etbankments etc. (ei) Some initial training to familiarise the workforce with construction procedures is required, (vit) Labour intoneive sf nigh rate of progress to be maintained, (viii) Supervision and strict pattern of working required, Im order to maintain the rigidity of the mattress during construction and {nfilling a large degree of tensioning of the transverse vertical diaphragna is necessary From experience on site, this last point rogarding the anolnt of tensioning, appeared to have been undereatinated by the successful contrastor. Gx)The main disadvantage of the excavation option is the high cost of the imported rockfti1, which accounted for approximately 85% of the total cose of ail the tenderers prices for this option, If however a local ready supply of Pockiiit wae avellabie, or material waa being quarried or excavated oleowhere on site during the course of the works, both of which were not available ou this contract, the excavation option would obviously become more competitive. cONCLUSTONS By adopting the design of a googrid mittrose to carry the proposed embanknent over the area of weak ground, It has been possible to increase the factor of safety against foundation failure of the weak material to an acceptable: level, Careful choice of the mattress infill materiat has enabled the mattress to act ae an efrective drainage blanket for the vertical Dand drains which were installed through the mattress into the weak material below to Accelerate settlement and alleviate any excess pore pressure developed during construction of the embankment Being a new innovation, progress in construct~ ing the geogrid mattress wae slow initially stil the contractor's workforce had fami Tiar— Ged thenseives with the construction procedure Once this had been achieved and a recognised ‘soquence of construction had been adopted Progress was greatly increased. This was Pspeciaily noticeable after the first half of the mattress had been completed. With rogard to costs each possible application Sf a googrid mattress noeds to be considercd on its own merits. In many instances, whore the thickness of the weak Layer 19 Lets Lian three metres, oF where there is a ready supply Of suitable infil) material elose by, the Obvious solution from both design and cost Sousiderations will be to excavate and Subetitote the weak mtordal with Imported £111 uaterial. However, ® ready supply of infill EMBANKMENTS material is not alvays available near the Works, and excavation may prove difficult due Yo the presence of a high water table, site conditions or other restraints and itis 1p ‘such circumstances that the use of a geogrid mattress becomes cost effect AGKNOWLEDGEMENTS, ‘The author wishes to thaak Mr P J Mason, Direotor of Mighways, Lothian Regional Council for his permission to produce this paper. lle would point out that the viows and opinions expressed are his own and not necessarily Those of the Regional Council. ‘Taanka are also due to Notion Limited, Tarmac Construction, Cementaton Ground Engineering Limited and W A Fairhurst, Consulting Bogineers for thelr advice and assistance provided during the preparation of this paper, REFERENCES 1) Capper PL, Cassie WP, Geddes JD ~ Problems if Engineering Soils, London Band PN Spon Ltd (1988) 2) Johnson W, Mellor PB (¥NR) — Engineering Pasticity 8) Notion Limited (1982) — Design Suggestions for tho Use of High ‘Toneile Strength Polymer Grids in the Construction of the Musselburgh Bypass Rural Section. 4) saith oy = Elononta of Soil Mechantes for Civit and Mining Engineers London, Crosby Lockwood (1971) Second Edition 5) Taylor DW = Fundenentala of S041 vechantes (Wiley International Edition) John Wiley and Sons’ (Loudon) MmEmbankments: report on discussion R.H. Bassett, King’s College, London When the design methods described 1n paper 3.1 ere con Dared with the aldernative approach used by The Reinforced Earth Company a number of discrepancies and aitticuities ere night ighted: 4) In order to generate stresses in the reinforcement the strain necessary £0 mobilise the retnfarcenent has to be assessed, b) IF the tensile torce mobi! tsed 1s overestimated then the safety factor is aver-estinated, ) The nechanisn being strain-cantrol Ted, could ative rise to the strain corresponding to the application Of the safety factor being larger than the strain Which would cause faslure of the structure. 4) The Vnear varlation of tensile forces in the retnforeenene with depth, as suggested by the Author, did not correspond to the distribution of Gaximun tensire forces esther in the theoretical case or inextensible retnforcenent or in his practical experience. tte Paine pointed out thet his proposals were for application to slopes and he wos not avere of haw the teenniquee used by The Reinforced arth Company cou! actually be appl ted to slopes. " He suagested that ane mmajor difference was that {n slopes there was no.effect ‘bn the facing units. Application of the author's design ‘analysts to some rea} structures, showed the analysis {iteed wel, and the Dewsbury Wall vas quoted as an example The main problem in obtaining data tron real structures was one of Interpreting forces In relnforcanent. via the 2 rneasuronent of strains, particularly when the polymer: ‘orids In question were’ non-uniform material. Comparisan af the author's design methods with the work of thers was being undertaken as an twependent. tidy. Experinents conducted on sea)ed down Geoce?) mattresses at the Roya) Mil itary Col lege, kingston, Ontario were Geseribed by Prof. Jarrett, Vertical reference rods on the foundation passed un Through the gravel fIlled mattresses and were used to rneasure vertical movement at various locations of the rnattress, Displacenent of the grave} fil Ted mattress under a heam oad was conpared with displacenent of unreintorced gravels Initial ly there was 1fttTe difference tn response but hy the tive the Deas Nad oisplaced Soem a SOx adcit}onal ‘oad was required in the retnforced Geocel} case compared tothe unreinforced gravel. The data showed that the vertical movenent in the found ation mas spread over a much wider area when using the nattress indicating that the Geocel mattress has a” Gbility to spread Toad laterally, and hence must possess a measure of verding stiftness. Experience in Norway on soft softs, arshes and bogs, ‘Frequently shows shal Tew type of fet ures Te was considered that the lensar Geocel vovld prevent this type of fatlure, nat because it behaved as 9 rigtd plate or foundation ~ but because Tt took up tensile Forces In the base. The weak sof] didn't respond to nathenaties, but only to creation of weak spots. The failure in tension, rather than shear was explained by taking the example ef 2 cube under compression fram 4 Sides, and if conpressed one niilimetre, will increase tn height by 2 mil Timetres, whereas if confressed one ny 1i— Inetre vertically downward, it #11) extend omy halt & iT netre to retain the olume, Only half the eneray was therefore requires to cause failure in tension rether than In shear. Supplying. aid~ ‘tional tension at the base therefore ai lowed the Tense Geoarid todo a good Job. PARTICIPANTS: Dr duran tr Paine Professor Jarrett Hr Knutson Polymer grid Weinforcement. Thomas Telford Limited, London, 1985Embankments: written discussion contributions De. P. Symes, R. Travers Morgan & Partners ipave recently bees involved in Eeying to design sone onbanksente which are 9m Aigh on Sone very soft alluvial clay. A quick visie to the site revealed that @ meandering river zigragged its way across it and co wo decided to ase sone sort of celiuiar earth reinforcement to Span the channels. We then realised that we had Little idea ae to how to assess the benefits on the strength and at what rate we could construct the smbankment. Under normal cireumstances it mould require three-stage construction. We Mecided to contact Netlon and see what they could offer in the way of theories. The only theory currently available is based on rigid plate analysis. To investigato the problem from 8 difgerent point of view we decided to conduct Sone finite elenent analysis using an Slastoplastic formulation. This was performed for us by the Geotechnical Consulting Group. ‘the contours shown in Fig. 1 represent the Btrese level within the sabankment. ‘the Gellular mattress to be placed across the width Sf the embankment would He underlain by soft alluvial clay and by sands and gravels. Tt innpdsately Became apparent that the geogrid would fail before construction was completed — in fact after about the second Lift. At that Stage it was evident that a single layer of Cellular geogrid would not be sufficient to prevent instability. However, it was felt that Te"coula still be of benefit because wo found Sone rather interesting secondary results The Firat relates to the settlements. We ran a comparison with geogrid and without geogria to See how the settlements are affected. From Pig. 2'te'ts evident that firstly the geoyrid vas not acting as a rigid plate and secondly the geogria made virtually no difference to the settlement. So we had considerable difficulty in deciding here the benefit was originating from. That there was a benefit there was no doubt because the embanknent stood up for longer than it did without the googrid. The next thing ve locked at was the lateral Aisplacencnts. In Fig. 3 it is evident that the geogrid makos a considerable reduction in the Taterai displacements near the base of the enbanknent which is, we presume, giving the benefit in strength ana sllowing as to construct the embankment slightly wore quickly chan we could otherwise have done. bn the results of these analyses, we modified the design to reduce the stress Concentrations thet would snap the geogrid at an early stage (Fig.") by flattening the side slopes and putting a 10 m berm at the toe. The principal message iz that a cellular geogrid does not Produce rigid plate deformations and that the Theory seems to he slightly inappropriate for this case, because the benefit seams to be Felated more to lateral rather than vertical stiffness. ALLUVIAL CLAY: SANDS AND GRAVELS Fig. 1. Contours of stress lever Polymer rid reinforcement. Thomas Telford Limited, London, 1985, 13DISCUSSION summice SETTLEMENT om DISTANCE FROM CENTRELINE OF EMBANKMENT tn == WITH cea Fig. 2. Comparison of settlements with and without cellular geogrid LATERAL MOVEMENT Foxy © No azoonie Lvermicete fret DEPTH BELOW ORIGINAL GROUND SURFACE (mn) Sail rameters ; ©3641 | Coste of at Pig. 3. Comparison of lateral movement with and Fig. 4. Consideration of a potential slip without cellular geogrid surface ina slope with geogrid reinfercerent 4Me J. R. Greenwood, Department of Tran: ‘he papers by Murray (o-1) and sewell eal (3.1) present methods for analysing slope repairs and steep embankments Where googrid rolnforcenent is proposed. Those analytical hethods are somewhat complex and require Considerable computer power, To aswist the Bosigner the authors produced a series of design charts However, the practising design engineer is often uneary about the Use of design charts, particularly where layered soll systems are present. The use of complex computer programs Bay lave doubts about the validity of the base Sscuaptions and computations unless the results can be checked against a simple routine method or hand oaleulation» ‘The following simple method of analysis may be applied to check the stability of any siope or foundation and may be readily adapted to include the effects of geogrid reinforcenent. Lot we consider a potential slip surface within a slope where geogrid reinforcenent is proposed (rig. 4).. the factor of safety P may be defined most simply as suggested by Lambe & Whitman (1969) a F = shear resistance along the slip surface Shear stross along cae slip surface ‘The simple stability equation (1) is derived by consideration of vertical soll slices and the feaolution of effective stresses to determine the effective nornal force of the base of each Slice. “The analysis may be applied to both Circular and non-circular sLip surfaces (Greenwood, 1953). Fe: m 2) con “i w Ye addition of geogrid reinforcement layers may. then be considered by resolving the appropriate EMBANKMENTS permissible force (T) along the slip surface. F wie bsccat (Wut) es aan + xT ef a As for all limit analysis solutions the strain compatibility of the geoarid and soi] must be Considered together with the necessary embodiment. Toagths to prevent pull out. guation (2) is equally applicable to the eircular and non-circular undrained {# = 0) Snalysis preferred by Milligan et al. in paper 3:4 (ct becomes Su and the tan ¢' term becomes zero}. ‘The basic simple equation (1) gives satisfactory regults for all stability analyses but it aay be conservative in that it igneres the contribution of lateral soil stresses to the shear resistance available, “In a normal slope the contribution of lateral stresses nay well be insignificant Decause of the proximity of the slip surface to the edge of the slope. However, the reinforced Slope may benefit considerably from the available lateral stresses and it ie suggested That this effect may be studied by consideration of the full stability equation based on Nohr's Circle of stress (Greenwood, 1983). FruebacetW wis Heit ocean cl teten# 6) Considerable rescarch and monitoring is required before the appropriate x valves €or use in this equation canbe determined and for the present State of knowledge it is suggested that K is assumed to be 20F0. References Greenwood, J. R. (1983). A simple approach to slope Stability. Ground Engineering 16, No. ay 45-48. tame, P. W. a Whitman, R. VY. (1969), soil mechanies, pp. 363-365. Now York:Wiley. us