Das QUESTION VII FELLENIUS, SWEDEN SECOND CONGRESS ON LARGE DAMS WASHINGTON, D. C., 1936 CALCULATION OF THE STABILITY OF EARTH DAMS* Wotmar Fe.uenrus Professor in Hydraulic Sirwetures af ihe Royal Technical University, Stockholim; Lieutenant Colonel in the Royal Corps of Engineers Sweden To the first Congress on Large Dams, held in Stockholm in 1933, the writer contributed a paper on question IIa, viz: Research Methods for Ascertaining Whether a Given Material is Suitable for Use in the Construction of an Earth Dam {Re ort No. 23). As the question was not formulated to include calculations on the stability of dams, the writer confined himself to the observation that such calculations were often founded upon the basis of curved sliding surfaces, and a reference to a published work of his (1926-27) concerning such calculations." As some other reports submitted to the same Congress dealt more fully with calculations, the ensuing discussion on question Ila was diverted from the researc: ds of the materials to the subject of calculations, and in the first of the resolutions on this question the requested that questions concerning the calculation of the Congr ty of H darth dams should be accorded special attention in the discussions at a forthcoming session. Notwithstanding that some discussion took place at the first Con- gress concerning calculations, the writer considers it advisable to refer here to some of the fundamental prineiples presented in the aforementioned work." ~*Caleuls de slabiild des barrages en terre. Berechnung der Stabiliiat von Erddammen. Céleulo de feed aes de ies sprenss de tierra, ins a, stati berdkningar ar, med friktion ach eharion fl cick: or ws = ‘atten rena vel eee ry latische Berechnungen mit 445All calculations concerning stability are founded on the assumption that the sliding surfaces are circular-cylindrical, The forces working in these sliding surfaces are assumed to be cohesion alone—the tial force independent of the normal pressure—or combined cohesion and friction. In the latter case we have the Coulomb equation T=k+jNn; (wD where T=the tangential sliding force per unit of the sliding surface; ‘=the cohesion per unit of the slidin; gurinee; {=the friction coefficient (=tge; ¢ the angle of internal friction); N=the normal pressure per unit of the surface. ef first case—pure cohesion—is a special instance of the latter, with /=0. Calculations generally have for their object the determination of the properties of earth necessary for equilibrium in different given sliding surfaces, That sliding surface which needs the greatest cohe- sion plus the ind angle of friction for equilibrium is the most dangerous of id the degree of security varies as to the ratio of the existing to the needed strength of the earth, I, Pure Conesion Eanta In the case where only cohesion exists in the earth the calculations are very much simplified, uation of equilibrium for the turn- ine moyemeate round the axis Thee 6 cylindrical sliding surface ADC is (see fig. Pa=kar @) and P a t= @) where (see fig, 1): =the needed cohesion per unit of the sliding surface; P=the weight of the mass of earth over the sliding surface per unit of length (of the axis); a=the horizontal distance between the axis (1/4) of the sliding surface and the centre of gravity (Z) of the mass of over the sliding surface; scans ree or ike hi siding arte hi ADO, a=the of the urface (t vette (See Fellenius, W.: Kaj- och j (Quay and landslides in Gothenburg), Teknisk "Tidskrift ate . U. m2, p 17-19, Stockholm 1918.) For 4 plane slope thee calculation can be worked out numerically and, only with respec to sliding wycd aces passing through the foot it of the slope, the + ded esion in a homogeneous earth may be expressed by: rath; 4) where v=the unit weight of the earth; 446h=the o height of the slope (=the difference in height between horizontal limiting the slope). e=a a of cohesion that is a somewhat complicated trigono- expression ( (9, «, w) on fig. 2), the a of which lifferent angles of slope, 6 . metrical is is arrived at by for dif Fiavze 1.—Stability of slope for co- hesion alone for curved ing sur= face through the toe of the slope. = Stabilité du talus, mda simple, surface cylindrique de glissement ‘traversant la base du talus. — Sta- bilitat einer Boschung, nur Kohdsion, kromme Gigltiche Gurl den Fuse: der — Estabilidad talud, cohesion Cizaple, superficie eilindrica de dealizamiento que atra- viesa la base del talud. The issible cohesion, 4, being given, the possible height, 4, of the slope at different angles ¢ of slope is obtained from (4): 4k 1 hates (5) On figure 2 the factor of cohesion, c, is shown by a full-drawn line for different angles of slope. The eft part of this line (for @<10°) —_—s Figeae 2.—Diagram showing the factor of cohesion at different an; f slope: Pure ‘sliding passing through the tow of the seen tees gramme indiquant le facteur jon & des angles différenta de pente; eohésion sim) surface de nt traversant Ia base du talus, — Dia- gramm des Koh: nur Kohasion, ‘Gieitfsche « dureh den Fusspunkt der Boeehung. indieando el factor de cohesién en diferentes dngulos de Ja ini paca sidn simple, superficie de deslizamiento que atraviesa la base del tal has here been adjusted in accordance with an observation recently made to the writer by Mr. Ohde, Berlin. On the same figure the corresponding factor at cohesion, ‘based upon calculations assuming plane sliding surfaces omte3)i is shown by a dotted Ins. 447The difference between the results of the two methods of caleula- tion is clearly shown in this figure. One sees that the factor of cohesion is always greater for curved than for plane sliding surfaces, The increase for a vertical wall (@=90°) is 4.4 ent, but this rapidly increases at diminishing angle of slope, and is for slope 1 : 1 (@=45°) about 65 percent, and for slope 1 : 2 (§=26°34') about te fos eam ishes to know the possible height of » slo , for example, one wishes to know possi tof a in @=30° and the earth has a given unit weight y=1.6 t/m® and & cohesion k=1.5 t/m", one obtains as cp=0.62 (from fig. 2) the fol- lowing height: Figure 3.—Most dangerous curved sliding surfaces through the toe of the slope at different angles of slope; pure cohesion. — surfaces ques de glissement traversant lm base du talus A des angles de pente afte nies solaior. simple, —— Gattherliohats feel eee ‘durch: 1 Fusspunl Bésehui ei verschiedenen ungswinkeln; nur Kohasion. — Las superficies sine Btras de deslizamiento mds gue piraviesan la base del talud en diferentes dngulos de inclinacién; in a On figure 3 the most dangerous circular-cylindrical sliding surfaces, drawn through the toe of the slope, are shown for the following slopes: 8:1 (60°); 1: 1 (45°); 1: 1.5 (98°41’); 1:2 (26°34’); 1:3 (18°26’); and 1; (11°19’), It will be found, in nearly all these cases, that the sliding surfaces Project below the horizontal plane through the toe of slope. aturally, entirely homogeneous material has here been assumed, and the results are therefore not valid in those most common cases where the ground on which the dam rests differs in character from the dam fill. In such cases one must determine (1) the most d. us sliding surface that does not touch the underlying ground, and (2) the most dangerous of the sliding surfaces that pass through both the dam fill and the underlying ground, and then try to ascertain which of these two is the more dangerous. 448Still more di us sliding surfaces may—if the underlying ground is of similar or of looser material than the dam fill—be drawn through points beyond the toe of the slope. (See fig. 4.) It has been shown that in a homogeneous, pure cohesion material, the most dangerous sliding surface at angles of slope less than Fiacne 4.—Stability of slope; pure cohesion, curved sliding surfaces ing outside the toe of the slope. — Stabilité du talus eohésion simple, surfaces cylindriques de Gimemant traversant des Poe en dehors la base du talus. — bilitét einer Bosshung, nur Konision krumme Gleit- ausserhalb des ktes der hesion superficies cilindricas de que atraviesan puntos fuera de Is base del talud. 6=53° theoretically is situated in infinity, that is, as deep as possible, and has a center angles ~ 133) "= = ~1 right angle. The cohesion corresponding to this is: om Estabilidad del talud; co- kath sin 133 34’ —tho.723 (6) From this, for & given cohesion, t, a limit value is obtained of the possible height of the slope: hay (7p Il. Frierton ann Conzsion Compinep In this case the calculations for curved sliding surfaces are very com- plicated and can scarcely be carried out otherwise than graphically, One may write approximately—especially at small angles of slope— (see fig. 1 and. a and (2)): Pazx(ke+fP)r (8) but often this approximation is not sufficiently close to be permissible for other than preliminary calculations. ‘ ihe graphic calculation is shown on figure 5. The method is as ‘ollows: The mass of earth over the supposed sliding surface is divided into vertical elements, having weights P,, P:, Ps, ete. The forces of co- hesion acting at ‘the sliding surface Ay, Ky, Ky, ete., are calculated upon a certain assumption of the amount of the working or permissible cohesion. With the P and K forces a continuous force polygon is drawn, Assuming that the othe for forces op acting at the sliding surface have directions at all points making le of friction with the normal to the sliding surface at corresponding places and, therefore, are tangents to a circle haying the same center as the sliding surface and a radius=sing><[the radi of the sliding surface], a plan of forces and a corresponding line polygon is then drawn. The loci and the directions of the forces working in the surfaces between the different elements are not previously known, but in 449drawing the plan of forces and the line polygon one may, within given limits, make certain assumptions conderneg them. if the calcula- tion is based upon reasonably similar assumptions it will not influence the result to any great extent. . In the case of equilibrium, the line polygon must be reasonably situated and the force polygon must cl itself. Should the foree polygon not close itself, new drawings must be made and the origi- 0 Hf assumed values of » or k, or both, must be changed until equilibrium is obtained, In order to determine the ises for eatuilibrinm of a given slope, one can either assume that the cohesion is known and the angle hues 5.— Stability of slope; friction and cohesion, curved sliding, surface surfaces cyl ues de gliseement la base du tal einer Béschung; Reibung und Kohision, krumme Gleitfidche durch den Fuss- punkt der Boschung. — sbilidad del talud, rozamiento y cohesién, superficies eilfndricas de deslizamiento que atraviesan la base del talud. of friction necessary for equilibrium, or that the of friction is known and seck the amount of cohesion necessary for equilibrium. te generally gives an infinite number of combinations and possi- ilities. With the abject of subjecting the problem to a systematic investi- ation and of obtaining « general result, the writer has chosen certain lopes and has treated them according to the following method. This is founded on the fact that the constructions at a certain angle of slope in slopes of different heights are exactly uniform, if one assumes 450values of cohesion proportional to the corresponding heights of the slopes. In all the cases considered herein it is assumed that the slope is just in equilibrium. ‘or a certain angle of slope, one must first assume that the earth material is a purely cohesive, and calculate the corresponding value of the cohesion k,, necessary for equilibrium. &, is calculated as in the preceding chapter (fig. 2 and (4)): k= 0) valid for a circular-cylindrical sliding surface through the toe of the The next step is to try to find, for the same slope, the most dan- gerous sliding surfaces for the cohesion=0.75 k,, 0.50 k, and 0.25 k, and to determine at the same time the corresponding values of the angle of friction, ¢v.75, ¢u.s0 Ad gp.25 just Necessary to assure the equi- 6 WON Ser WS ete, So = Sal SS ge st ES. Ficune 6.—Diagram showing the connection between angle of alo of friction and the relative vi of cohesion; curved sliding surface throug the toe of the slope. — Diagramme démontrant le ray entre l’angle de pente, Vangle de frottement et la valeur correspondante de cohésion; surfaces eylin- ‘isct Kohision; krumme Gleitfiche dureh den Fusspunkt der Boschung. — Dia- grama demostrando la relacidn entre el dngulo de inclinacidn, el dngulo de rosamiento y el valor de cohesién correspondiente; superficies cilindricas de deslizamiento que atraviesan la base del talud. librium of the slope. How this is done is shown in figure 5. Fora certain value of &, that sliding surface is the most dangerous which needs the greatest angle of friction. If one we a curve for the greatest values of friction obtained, # 1.0=0, #075, Pom Od y oo5 8 abscissas, @ correspondi Telative values of cohesion, 1.00, 0.75, 0.50, and 0.25, as ordinates, this curve will be typical for the angle of slope in question. Owing to the uniformity already mentioned, this curve is evidently guite independent of the height of the slope and of the gravity of the earth 1 and is valid for any slope of the chosen angle which 1s just in uilibrium, igure 6 illustrates some curves obtained in this way for different angles of slope. 451Figure 7 shows the loci of the centers of some of the different sliding surfaces, varying according to changes of angle of slope and of rela- tive value of cohesion. In this figure the toe of the slopes, A, is assumed to remain fixed. The center loci radiate from the centers of the most dangerous sliding surfaces in case of pure cohesion. These sliding surfaces are drawn in the figure. These curves of center loci are asymptotical to the normals to the slopes at their middle points. ‘For k=0, that is, pure friction earth, the most dangerous slid- ing surface evidently is plane and lies in the slope itself, that is p=@ Ficune 7.—Centers of the sliding surfaces at different angles of slope and different relative values of ion; sliding surfaces ing through the toe of the slope. — Les centres des surfaces de glissement 4 des angles de pente différents et des valeurs de cohésion différentes ites; surfaces do traversant la base du talus. — Die Mit jak der Gleit- liche bei verschiedenen Bosch keln und nen tivwerten der Kobision; Gleitfidchen durch den Fusspunkt der Béschung. — Centros de wuperficies de deslizamiento en diferentes dnguloa de inclinacién y de los laa diferentes valores de cohesién correspondientes; superficies de deslizamiento que atraviesan la base del talud. . Figure § shows for the slope 1:3 the loci of the different sliding surfaces intersecting the toe of the slope for the relative values of cohesion: 1, 0.75, 0.50, 0.25, and 0, and the coordinates of the corre- sponding centers of the sliding surfaces. We now return to figure 6, in order to show the use of the drawn curves for the calculations of different cases. If in a given case, the angle of friction, ¢, and the height, h, up to which the earth is just standing in the angle of slope, #, are known, and the value of the co- hesion of that earth must be ascertained, the proportion between the desired value of cohesion and the value of k,, valid for the case in question, may be obtained from the corresponding @ curve in figure 6 and from the known value of ¢. 452Ex. ¢=10°; h=10 m; @=33°41'; y=1.6 t/m’, From (9) and figure 2 one obtains: i= Leu 0.636 =2.54 and from figure 6: E:k,=0.54 7 k= 0.54-2.54= 1.37 tm? If, however, ¢ is not known, but it is known that the earth is stand- ing at the angles of slope=6,, and 6, up to the respective heights A, Fravr® §&.—Displacement of the sliding surface by changing the relative cohesion (slope 1:3); ing passing through the toe of the al Le déplacement de ia surface de nt par changement de la vi de eohésion correspondante (pente 1:3); surfaces de glissement traversant la base du talus. — Die Verschiebung der Gleitflache bei Verinderung des Relativ- wertes der Kohision (Bésebung 1:3); Gleitfiichen durch den kt der Béschung. — Desplazamiento la superficie de deslizamiento al cambiar el valor de cohesién ndiente {inclinacién 1:3); superficies de desliza- miento que atraviesan la del talud. and i the maximum height fA; at the angle of slope 4, may be ascer- tained in the following manner: tk From figure 2 are obtained for (,, #,, and 4 the values of 7 C1, Cy, and oy; ‘kya oj ky TH oy and eqs e, (10) In figure 6 the @,, , and # curves may now be sought or drawn by interpolation. If, now, the unknown value of ¢ the still unknown ordinates of the 6, &, and 6, curves are named b,, b; and by, the eohesion of the earth is evidently: k=bky,=bokoy= bole (11) from which, as a beginning, with the aid of (10) one obtains: b * If now, in figure 6, in the vicinity of the place where ¢ can be supposed to be located, an aid curve is drawn the ordinates of which 453,at every point have the relation to the corresponding ordinates of the 8 curve=/ycy : Aye; this curve will evidently cross the curve at the desired ¢ value, and the ordinate of the crossing int will be=6,. By drawing this sxdinate line, by and by are obt as the ordinates for the intersection of that line and the # and 4 curves. From (11) and (10) are obtained: noe and from this the desired height, hf, (by, bay Sy and ¢ and ¢ now being known). it is certainly possible to sane the greatest angle of a that can be reached at a given height, As, in the same way, but it i Frovre §.—Diagram showing the relation between ‘misslble heights of the for different angles of slope, #, and for a slope; friction and co- m combined; circular-cylindrical sliding surface intersecting the toe of slope. — Diagramme démontrant le rapport entre les hauteurs de talus tolérées pour de divers angles de pente, » Bt pour talus com! de cohésion et de frottement, surface eylindrique eirculaire de glissement traversant Ia base du talus. — im des Zusammenhanges swischen mog- lichen Béschungshéhen bei nen Béschungswinkeln, 4, und bei senkrechter Wand unter Annahme susammenwirkender Reibung und Kohision nebst kreiszylindrischer Gleitfiiche durch den Fusspunkt der Boschung. — Diagrama mostrando la relacién entre las alturas de Jos taludes itidas para diversos dngulos de jnelinacion, @ dn, 8 ¥ para el talud vertical; combinacién de co- y de rozami Hentoy, superficie cilindrica circular de deslizamiento que atraviesa la base del talud. require many tests involving a great expenditure of time. For this reason, in figure 9 a diagram has been plotted over the proportion AB : hyo for di erent angles of slope, 8, 4 this repartee has been caleu- lated upon the assumption of cireular-c cal sliding surfaces intersecting the toe of the slope The ordinates in this diagram have been calculated by the aid of figure 2 and figure 6 in the following manner: 454