fi ovastide, M, de Buban,P. & Dorian, L (195) Gdotcbnae 48, No 1, 25-24 Bearing capacity of a foundation resting on a soil reinforced by a group of columns M. BOUASSIDA*, P. DE BUHANt and L, DORMIEUXt ‘A now design method for a foundation on a soil reinforced by columns is described. A lower bound of the bearing capacity is determined within the framework of the yield design theory. Te takes into account the threedimensional nature of the problem and is applicablo to a wide range of ‘geometries. A parametric study on the improve ment of the beating capacity as 2 function of the proportion of reinforcement, and on the strength characteristics, is presented. A complete analyticad solution is given for the strength of 2 composite cell subjected to a triaxial loading, which provides an insight into the reinforcement mechanism. KEYWORDS: bearing capacity, design, fallue, foot ingsfocodatons, ground improvement, reinforced sols Le préseat article propose une méthode de dimen- sionnement dune foundation sur un sol renforcé par colonnes, basée sur la théorie du caleul a la rapture. Une borne inférieore de la capacité por ‘ante est obtenae en tenant compte de la nature tridimensionnclle de Pouvrage. Elle est applicable pour une grande variété de géométries. On présente tune Gude paramétrique sur Pamélioration de la ceapacité portante en fonction de la proportion de enforcement et des caractérstiques dx matériau renforgant ot du matériau intial. On sintéresse en outre 4 la résistance en compression d'une éprou- vette composite soumise a un chargement triaxial classique, pour laquelle une solution consplite est fournie sous forme analytique. Elle apporte un Gelairage interessant sur le mécanisme du renforce- ment par colonaes. INTRODUCTION As a soil improvement method, the columns ‘inforsoment technique has tinee purposes: to Feduce te setilement of highly compres Sich os Soft clays, 10 astelerate the stage of primary consolidation, and to. enhance” the Elanng cipscity. This Poper is concerned with thelast ofthese, ‘Common material for columns are well graded cican sands, gravels or stones, hence the term Stone columns. (AboshiTehinoto, Enoki & Harads, 1979; ‘Soyen, 1985) However, the ‘eduction of settlements rather than aaah Ti iloret Toundations EP porally the pemary contieraion- when ‘Geiigning SOON wceTunee Walnarn & Booker, 1985; Schweiger & Pands, 1986). As far_as bearing capacity is concerned, a remarkable per ‘enue ab be Obtaad by mig the nate f° hime (he Hime oil with a Tow” percentay Manuscript reeeved 15 February 1993; revised mano- seript accepted 9 December 1993 Discussion on ths Paper closes i June 1995; for further details see pi, "'Formenly Laborstoite de Mécanique des Soles, Palaisau, France, now Ecole Nationale des ingénieurs de Tunis 4 Boole Polytechnique, Palaiea, 25 columns technique, Broms & Boman, 1939), The Strength ofthe Column smaterial can be deeibed by a Mohs-Coulomb eniterion (with appropriate flion ange and coteion) while the native sol Ismodeled ass purely cobesive materi ‘Design calelations concerning. the Gearing capacity of such einford soils have mainly bee preseatd for a tench-like reinforment under {he plane strain assumption (Madhav & Vitker, 1908; Prochnic & Shahrour, 1981) and for an foolted column in an axisymmetric. condion (Hughes Withers 1974; Braun, 1978). Few Sttompts Have been made to deal with the ease of asoil reinforced by a group of columns. nthe fat part of te Paper, the steagth of composite ylindial ell subject to taxa tt ‘CORAUONS-BTAVESTESTEN- WTR THe_rigorous ‘eSreea-Trmewone oF we Vere aes REST ie, 7; Salengon, 19837 ie solution ofthis anvilary problem provides an interesting insight into the way in which the native sol actually reinforced by” the inwoducton of Columns. The second pat is devoted to te case of a group of columns. A lover-bound solution for the bearing capacity ofa foundation resting on teinforead soll derived on the bas of the previously determined solution, and a quantia- the analyis of the gain of strength due tothe tenforosment i developed6 BOUASSIDA, DE BUHAN AND DORMIEUX COMPOSITE CELL SUBIECTED TO THE ‘TRIAXIAL TEST Consider a cylindrical composite cell @ of height h made up of « circular reinforcement ‘column sutrounded by the native soil. The shape of the cross-section of the cell by any plane z = 25 is not specified. To illustrate this, a square-shaped cells shown in Fig. 1, The reinforcement column ‘occupies the domain r a) is ‘occupied by the native soll, which is a purely ‘cohesive material. Its strength is described by a ‘Tresea criterion fie) =0,-0,-2C<0 o Both criteria are expressed in terms of total stresses and ate therefore relevant for a short- term analysis, The ratio of the cross-sectional area na? of the reinforcing column to the total cross-sectional area 5 of the cell is denoted by n, the proportion of the reinforcement. ‘This composite cell is subjected to classical tri- axial test conditions. The cell is in contact with ‘wo smooth and rigid plates located in the planes 2=0 and z=h, The lower plate is kept fixed ‘hile the upper one is given a vertical downward translation motion of velocity V. The force exerted by the upper plate on the cel is denoted by Q = —Ge, (Q.> 0 when the vector Q is orient- ed downwards), A uniform confining pressure p is acting on the lateral surface, which is referred to as 4,. Thus, the loading of the cel is character- Fig, 1. Composit cell under triaxial loading ized by the two parameters p and Q. A statically admissible (SA) stress field o and a kinematically admissible (KA) velocity field U are respectively ‘subjected to the following conditions (a) oSA with pend Q Aeyyox, «0 Ga) yn) = pm on 8, (4 = me, inward oriented unit vector normal to 28) (36) @) an [ eas os ‘The convention of summation over repeated sub- setipts is adopted in equations (3a) and (30) with 1 orj= He ©) UKA with ¥ U,=0forz=0;U,=—Vforz=h (4) Generally, a given load (9, p) is said to be safe in the sense of limit analysis or yield design) iit can be equilibrated by a siress field that complies with the strength criterion everywhere in the cell. The tim of this section is to determine the domain K of safe loads (Q, p) restricted to the region defined by Q20 and p20. Practically, this involves determining, for a given pressure p, the maximum value of Q such that (Q, p) is @ safe load. This value is referred to as Q°(p) O°) = max{O > 0.(Q, 9) € KI ‘The set of points (Q*(p), p) for p > 0 defines the boundary of K (Fig, 2). 0"(p) can be determined by a combination of the so-called static and kine matic approaches, which are now described. Lawer-bound estimate of K: static approach ‘The static approach simply involves. imple- menting this definition of K. Thus, in order to ° Pom . Fig. 2. Domaie K of sae loads (p, 2)BEARING CAPACITY OF A FOUNDATION a prove that a given load (Q, p) belongs to K, itis necessary to exhibit a stress field o that is SA with (Q, p) (ie. that satisfies equation (3)) and complies withthe strength criteria (equation (1) in the domain r Et n+ Kym) +2CL—+2C%K] Improvement of the statie approach. Let b denate the minimum distance from the axis ofthe reinforcement column in the eell {0 the lateral boundary surface #8, . For any radius ¢ ¢ [, b) 2 class of aaisymmetiical stress files is defined eS ed ig 3, Threecone axaymmetrcal stress Held in the three intervals (0, a3, (a, ¢] and r > by ig.3) n= 00) = Bi tag D+ 2C (208) acr a respectively. The vertical force equilibrated by such a stress field is f 6:88 = SMI = 1 + nK,) + 209K, 4+ 2C(1 + nllK,— Din) — 404291 (a2)8 BOUASSIDA, DE BUHAN AND DORMIEUX where x denotes the ratio c/a. Thus a lower bound estimate of O° (7) obltned for any value chee a bhie.foreny ©e Cl, ba] LO) 1-1 + nk) +2C WK, + 2C(1 + nliK,— Din(s) -i0+ 2) a) ‘The maximum value, of the ‘right-hand side of equation (13) is achieved for x= /(K,— 1) ‘Thus, if 2 2, the difference A= nCg (9) between the lowerbound estimates of Q°() given in equations (9) and (14) is a postive number. In ther words the second static approach improves the frst one as soon as 9 2 19°3", which is always satisfied in the ease of a column made up of & purely retional material, where the friction angle fs usually ofthe order of 35° stone columns). The second condition, K, <1 + (@/a), will also be achieved for the ustal valucs of 9, a and 6 encountered in practice. Its assumed below that this condition is always satis. IK, <2 (p< 19"), which could correspond to the Gate of lime columns, the best lower-bound estimate of 0°) provided by equation (13) is obtained for x= 1, For this value of x, the present static approach reduces to equation (9 Tt is now established by means of a Kinematic ‘approach that equation (14) when 2 a (native soil) wyaifes2e-of%) a9 This velocity field is kinematically admissible with ¥ and remains continuous across the inter= face The density x(d) can be computed from equation (20) @ itra HK, <2: @) =201V/8) #K,> 2:7 2) to equation (36, The validity of equation (39) is subject tothe condition K, <1 + (D/2a) % iC denice Fig. 7. Zoning of the mass of sil below the foundation2 BOUASSIDA, DE BUHAN AND DORMIEUX Interpretation based on the concept of equivalent material ‘The previous result may be interpreted based on the concept of equivalent material and related notions of equivalent cohesion C* and angle 9, as introduced with reference to the composite cel, Indeed, adopting a heuristic point of view, one could be tempted to substitute the equivalent material thus defined for the heter- geneous reinforced soil, so that the solution of the initial bearing capacity problem (Fig. 8(@)) would reduce to that of the equivalent problem shown in Fig. 8(). Proceeding further with this kind of reasoning, 1 simple static approach can be implemented on the latter problem by considering the following stress field (@) within the zone composed of the equivalent material oan = 8p = C5 2CK, + 20° JK," } or this zone beg y= 2C5 0, @ Such a stress field obviously complies with the Strength criteria of the original soil and of the equivalent material in each of the two zones. It cequilibrates the force Ox" = 2(CK," + CHK) 2) ‘hich could be interpreted as a lower-bound esti- rate for the equivalent bearing capacity problem. Now, referring to equations (29) and (30), it appears immediately that Q,,"” is identical with the lower-bound estimate Q,, given for the intial problem by equations (36) and (39) for K, <2 and K, > 2 respectively However, the concept of equivalent material should be handled very cautiously as it bears no significance apart from the particular situation where the reinforeed soil is submitted to a triexial loading (axisymmetrical about the direction of reinforcement). It does not apply, for example, to the description of the overall strength properties ‘of the reinforced soil when subjected to off-axis forces. To altempt to implement a static approach on the equivalent problem, using stress fields with inclined principal directions, would lead to completely mislending conclusions. The idea of homogenization, which is only ly conveyed in the notion of equivalent 31 can be rigorously formulated (de Buhan & Salenson, 1990}. It relies on the assumption that the reinforcing inclusions are regularly dis- tributed throughout the soil mass, and that the spacing between two such successive inclusions can be considered small enough when compared with a characteristic length of the problem, such as the width of the footing. Such a theoretical ‘approach has already been successfully applied in the geotechnical feld to the design of reinforced earth structures {de Buhan, Mangiavacchi, Nova, Pellegrini & Salengon, 1989). Contrary to what might be suggested by the preceding notions of equivalent cobesion and friction angle, it provides ‘lear evidence of the anisotropic strength proper- ties of the reinforced soll as a homogeneous material Quantitarice assessment of the bearing capacity improvement ‘Although the exact value of the bearing capac- ity semains unknown, and only lower-bound est- ates Q,,4n) given by equations (36) and (9) are available, it seems reasonable to assess the bearing ‘capacity improvement due to the reinforcement by means of the ratio R= Qedrh/QedO), where Ors) represents the value of the lower bound obtained for the unrcinforced structure (Q,d0) = 4aC) (@) for K, <2 =14%«,-94F nated aS ve] (O) for Ky >2 (43) a4, 94S HY Ral tf, aS Ym] + tate where g(g) is as given in equation (15). As is apparent from this equation, the reinforcement Fig. & Bearing capacity problem: (0) initials 6) equivaletBEARING CAPACITY OF A FOUNDATION 3 effect can be attributed to both the cohesive (rough C'/C) and the frictional properties (hrough g or K,) of the column material, The Increase of (the “lower-bound cstimate of) the bearing capacity is directly proportional to the volume fraction of the columns. ‘All these results are summarized in Fig. 9 where the contour lines of R (called the bearing capacity enbancement factor) a8 a function of the non-dimensional strength parameters of the rein- forcing material (C*/C, 9) are represented for three typical values of its proportion (y = 0-1, 02 and 0-3), Assuming, for example, that the eolurnns are placed into the foundation soil following a periodic square-shaped mesh, these values corres- pond approximately to 2a/D = 0:36, 0:50 and 0662, where 2a is the diameter of the columns and D their horizontal spacing. ‘Two situations commonly encountered in prac- tive are now examined. Reinforcement by stone columns. The reinfor- reinforced “THeTein sn barely SST. theoretical result confrms.the idea that. sing the Bearing eapacty is sot the prime ‘Sppeciive oN MEKMTTeMTorosment, Reinforcement by lime colums. The constituent material is obtained by mixing a small proportion of lime with the in situ clay, thus increasing the Strength of the latter by a factor of up to 20. Although significantly lower than in the case of the stone column material, a friction angle can also be measured. As can be seen from Fig. (b) (lor =02), the bearing capacity can easily be Increased threfold even ia zero ftition angle is assumed CONCLUSIONS “The improvement of the bearing capacity of a foundation soil expected from its reinforcement by columns has been investigated in two stages. Fist, a triaxial loading test on a representative volume comprising a single column of reinfores- reat was simulated, The strength properties of Sich ¢ composite cll expressed in terms of com- pressive forse and confining pressure have been Setermined exactly through the combined imple- mentation of the yield design static and kincmatic approaches. Such a study paves the way for a possible experimental validation of the theoretical Predictions. “Making use of the optimal stress field exhibited in the static solution of the composite cell model, 2 general lower-bound estimate for the bearing Fig. 9, Contour fines of the Bearing capacity erbance- ment factor R: (a) 9 = 01, (6) n= 025 (6) = 0334 ROUASSIDA, DE BUHAN AND DORMIEUX capacity of a foundation-raft lying on a soil rein- forced by a group of columns was thea deter- ‘mined in the form of readily usable analytical expressions. The main interest ofthis estimate lies in the fact that it provides a reliable assessment for the three-dimensional bearing capacity problem, regardless of the shape of the founda- tion and the geometrical arrangement of the columns, It is & convenient engineering design tool, allowing the efficiency of different technical solutions to be compared in terms of bearing capacity improvement. Moreover, it can be shown that the range of applicability of the fower-bound ‘olution developed in this Paper (ccquations (36) and (39) can be extended to take gravitational forces into account, and to the case Where the strength propertics of the column material are expressed in terms of effective stresses, However, it remains to be seen whether this lower-bound ‘solution is sufficiently close to the exact value of the bearing capacity. This ‘could be determined by implementing upper bound kinematic solutions for the problem, pro- vided that the shape of the foundation and the exact distribution of the columns were specified, in which case a homogenization method should probably be used NOTATION radius of the column Surface ofthe foundation ‘cohesion of the native sit ‘cobesion ofthe enforcing material ‘cobesion ofthe equivalent material minimal dance between the axes of two columns K domain of safe loads (Q, ) applied to the com- posite cll Fellclent of passive sites of the reinforcing satel KS coulficient of passive sess of the equivalent ‘material Oy lowec-bound estimate of the foundation bearing capacity Q°() maxinaum force sustained by the ell for @ con- fing presure p R bearing capacity ed0Q,40) proportion ofthe reinforcement enhancement factor ction angle ofthe reinforcing material ‘g* letion angle ofthe equivalent material REFERENCES [Abothi, Hy, Ichimoto, E. Enoki, M. & Harada, K. (4979). The composer: a method to improve charac. Aeraties of soft clays by inclusion of large diameter sand columns, Proc. Int Collog. Sol Reinforcement, Pert, vol. pp. 211-216. Balsam, N.P. Booker, S. R. (1989). Bifect of stone column yield on sstlomeat of rig foundations in Sabiized clay. dne. J Numer. Analyt. Methods Geomech, 9, NO. 4 Brauns J. (1978), Die Anfangstragast von Schottersiu- Ten im bindingen Untergrond. Bautechnik 8, 263- ™. mis, B. B, & Borntn, P. (1979). Lime columns—a "oe foundation method, J. Geotech Engng Dio. Am. ‘See. Cie, Engrs 108, GT, 539-556 e Bahan, P, Mangiavacch, R, Nova, R, Pellegrini, C. & Salengon, 3. (1989). Yield design of reinforoed fearth walls ‘by a homogenization method. Geotecoigve 39, No.2, 189-201. 4 Bohan, P. & Salengon, J. (1990) Yield strength of reinforced solls as anisotvopic mele. tn. Yielding, damage and Jatlre of anisotropic. soils (ed. J.P. Boehlen, EGFS, pp. 791-803, London: Mechanical Engineering Publications Chea, We F. (1975). Lint analysis and sot plasticity. Fhsevier Hughes, JM, & Withers, N. J. (197). Reinforcing of Sot cohesive soils with stone columns. Greund Brora 42. Madhav, M. R. & Vitkar, P. P. (1978. Sup footing on ‘weak clay siabilzed with granular trench or pil. Gar. Gectoch. J.18, 605-609. eushnick, E.& Shabrour, 1. (1999). Application dela Ahéorie de. Thomoginéisation aux. eolonnes al- lastées, Ann st, Tech Batim, No. 496, Série sos et fondations 213, 117-127 Salengon, 1. (1983). Cale la rupture et analyse tite ‘Pars: Press de L'Eeole Nationale des Ponts et ‘Chaussées, Salengon, J. (1990), An introduction to the ytd design theory and ts application in soil mechanics. Bur. J. Mech. AS, No.5, 477-500. Schweiger, H. F. & Pande, G. N. (1986), Numerical ‘analyse of sione column supported. foundations. ‘Comput. Geoech. 2, 347-372. Soyan, B. (1985, Métodes de dimensionnement des ‘olonnes ballasts, Bull. Lins. Labs Ponts Chauss, No. 135, 35-51