You are on page 1of 40
SLOPE STABILITY COMPUTATIONS NILMAR JANBU* CONTENTS Notation 47 1. Introduction 48 Scope, 48 Factor of Safety, 48 i Shear Strength Expressions, 49 Limit Equilibrium, 49 2. Elementary Mechanical Principles 50 Purpose, 50 Stress Conditions, in Soil Elements, 51 Failure Conditions, in Soil Elements, 52 Stress Vector Curves, 55 ‘Stress Conditions along Classical Rupture Surfaces, 58 NOTATION ey cohesion, in general he c’, cohesion intercept, on the basis of effective stress i, Ces F = limit equilibrium parameter L E, total horizontal interslice force, with Al, boundary values Ey and Ey 6, subscript for limit equilibrium M, f subscript for failure Nes FP, factor of safety Np H, height N, AH, — horizontal component of force on the base AN of astice P, ‘Professor Technical University of Norway, institte ofGeo AP, technics and Foundation Engineering, Trondheim, Norway. P. Plastic versus Brittle Failure, 60 Stresses at Limit Equilibrium, 60 Limit Equilibrium as a Unifying Principle, 62 The Generalized Procedure of Slices 63 Purpose, 63, Basic Principles, 64 Working Formulas for Slope Stability Analysis, 65 Iteration Procedure, 67 Examples, 73 Routine Procedures, 77 Earth Pressure and Bearing Capacity Determinations, 82 Derivation of Equations of Equilibrium, 84 References 86 dis thru: slice number length length of shear surface along the base of the slice ance between shear surface and line of moment bearing capacity factor tan? (45+ 4) = flow value 4(V,—1) = modified flow value, or number total normal force on base of slice vertical line load, in general vertical line load on a slice yet q+ AP/Ax = total overburden a 43 SLOPE STABILITY COMPUTATIONS. Q, horizontal line load, in general AQ, horizontal line load ona stice surcharge 4 R,— 4(o,~0) = Mohr radius R, Mohr radius at failure s._ total shear strength (6 = 0) Si, undrained shear strength AS, shear force along base of sli T, vertical interslice force, with boundary values T, and 7, 1 AT/Ax= slope of T—x curve ut, pore pressure U, resultant water force AV, vertical component of force on the base of a slice W, weight AW, total weight of each slice, including sur- charge and line load Ax, slice width x, horizontal distance (abscissa) y. vertical distance (ordinate) 2 depth in soil a, angle of shear surface, in general 1. INTRODUCTION Scope ‘The main purposes of this paper are: 1. To review the basic concepts and the elemen tary mechanical principles of the limit equili- brium procedure for stability analysis. 2. To present, with examples, a limit equilibrium ‘method for computing the internal stresses and the average factor of safety of slopes, using shear surfaces of any shape. 3. To illustrate the general applicability of the proposed method for solving slope stability, earth pressure, and bearing eapacity problems. Only plane strain or axisymmettical stress condi tions are considered. The axisymmetric case is used only for interpretation of the results of triaxial tests Factor of Safety Engineering design must include a margin of safety for several reasons: to guard against ultimate fail: ure: to avoid intolerable deformations: and to ¢ ancertainties associated with the measurement of soil properties and with the analysis, ay 45+46 = angle of shear failure plane a, 45-+44,= angle of critical shear plane at limit equilibrium B, —slopeangle y ——unitweight «strain a ratio, regarding location of line of @ total normal stress o, a —u= effective normal stress ow jor principals on mediate principal stress a. minor principal stress Om Moytos) = Mohr center stress Gore uniaxial compressive strength (unconfined) o!, — c'jtand’ = reference siress in Mohr-Coul- comb diagram 1. shear stress th shear strength 7, intercept in veetor curve & angle of internal friction, in general tand', friction coefficient on effective stress basis tan,, tan 6'/F = limit equilibrium parameter The degree of safety against ultimate failure is usually expressed numerically by a factor of safety, which may be defined as follows: factor of safety = Senet a ‘The structural engineer is normally interested in the ratio of tensile (or compressive) strength to tensile almost exclusively interested in the ratio of shear Tn mathematical terms Eq. 1 reads F= @ where F = factor of safety 1/= shear strength along some shear surface += equilibrium shear stress along the same shear surface. In general, the purpose of slope stability analysis is to obtain the minimum value of F. In this chapter the im the soil mass is also corresponding state of stres: explored, Equation 2 shows that a complete slope stability ls requires two very different investiga strength: and (2) analysis « tions: (1) measurement of the she: INTRODUCTION 49 computation of the shear stress. A calculated value of the factor of safety therefore reflects the inaceur acies of both the strength measurements and the stress analysis. At present the uncertainties associ ated with the measurement of strength are likely to the computation of the introduce larger errors than shear stress. Shear Strength Expressions e value of the shear stress at which a soil fails in shear is called the shear strength. The shear strength of soils is usually expressed by Coulomb's equation, aya etotand (3) where ¢ and tan 6 are shear strength parameters and, ris the normal stress on the failure surface. in practice, one has to distinguish between total and effective stress analyses, When total stresses are used, ponding parameters are often de- noted c, and dy; when effective stresses are used the parameters are denoted c’ and $'. Hence y= eutotandy (4a) (4b) a= eto" tan where @ is total normal stress and 0" is effective normal stress on the failure surface. A stability an alysis based on Eq. 4a or 4b is often called a c,¢- analysis. Theoretically, when @=0 the shear strength expression becomes yaers 6) This equation implies that the shear strength is in dependent of total normal stress. In practice Eq. 5 frequently is used when unconsolidated, undrained tests without pore pressure measurements are per formed on saturated clays: the corresponding strength is termed the undrained shear strength sy If the soil is isotropic and if the undrained shear strength is independent of total normal stress, then for a given soil $= constant = 5, 6) ‘The undrained shear strength of clays may vary with depth. A stability and Eq. 6 will be called an s,-analysis, For short-term problems, involving layered soi sand and clay, one may have a combination of a c,¢+ analysis and an s,-analysis along shear surfaces alysis based on total stress through the various layers. From a practical point of view it should be of varameters s, ¢, and strongly emphasized that 1 & are not “material constants” for soils, since these parameters depend on a large number of factors such as void ratio and structure of the soil, effective stress level, method of loading, definition of failure, and sample disturbance. This is why the most diff cult part of a stability analysis is to obtain reliable shear strength data that simulate properly all per tinent factors involved in practice. In the equilibrium theories of elasticity the soil pro perties are expressed in terms of a stress-strain rela tionship. In the corresponding equilibrium theories of plasticity the stress-strain relationships do not enter, and the material properties are most con veniently expressed in terms of an equation for the state of limiting equilibrium. ‘A state of limiting equilibrium will be said to exist when the shear stress 7 is expressed as a @ shear strength. one obtains where F = factor of safety and Now by combining Eqs. 4b and tang" 73 (8a) = e,-to" tan dy (8b) in which the equilibrium parameters ¢, and tan are defined as follows: tan d= ne” (9) F is a factor of safety with respect to shear strength and 1/F is the degree of mobilization of the shear strength. Equations 4b and 8b define a strength line and mobilized stress line in a 7-0" plot. Both lines inter: sect the stress axis at the same point 0, corres ponding to a negative reference stress (Fig. 1): o (10) tang’ tan de and 7 have a stress axis only It should be noticed, however, that common point of intersection on t if the degree of mobilization is the same for c* and tan 6, as is assumed in this chapter Use of the factor of safety with respect to shear strength (Eq. 7) leads to simple ways of illustrating the differences between a failure condition and a 50 state of limiting equilibrium in small elements as well as in extensive soil masses. Figure | shows that when 6’ >0 and F > 1 the inclination a, of the plane on which F = minimum (ie., the eritical plane) is different from a for the failure plane. The corres ponding difference between the failure surface for F=1 and the critical surface for F > 1is also illus: trated in Fig. 1 for the case of bearing capacity of shallow footings on a weightless soil with ’ > 0. Equation 7 plays the same role in limit equilibr theories as Hooke’s law plays in the theory of elas- ticity, as may be seen by comparingthe two equations term by term: Limit equilibrium: ay Elasticity. ork (2) ‘The analogy is: The shear stress r corresponds to the tensile (or compressive) stress 0; the degree of shear mobilization 1/F corresponds to mobilized strain ¢; the shear strength 7) corresponds to Young's modu lus E. The ultimate values of 1/F and € are both reached at failure. In limit equilibrium theories the state of failure is readily included as Fi» = 1, where as in the theory of elasticity the state of failure, or flow, could similarly be defined as ¢ = ¢, The limit equilibrium procedures for stability an- alysis are usually based on the assumption that F constant along the entire shear surface, which means p= a5 + be Fig. 1. a ape a bee SLOPE STABILITY COMPUTATIONS that a weighted average value of F is obtained. The consequences of shear stress concentrations in critical zones must therefore be studied separately, say by assuming reasonable distributions of F along the shear surface, or by using a stress-strain rela- tionship instead of the limit equilibrium equation (Eq. 2. ELEMENTARY MECHANICAL PRINCIPLES Purpose ‘The main purposes of this section of the chapter are: 1, To review, for easy reference, classical know edge about equilibrium stresses in soil ele- ments, and to summarize classical principles and formulas used in describing two-dimen: sional failure conditions for ideal soils. To review the principles of stress-vector curves, and to illustrate their application to the interpretation of shear tests on soil samples. . To derive the shape and location of classical failure surfaces and the corresponding state of stress, and to demonstrate, by examples, the difference between ideal plastic and ideal brittle failures, To illustrate the application of limit equil brium principles in obtaining information about the stress conditions at limit equilibrium and the shape of the shear surface corresponding to minimum safety factor. Shear stress and critical shear surfaces at limit equilibrium and at failure ELEMENTARY MECHANICAL PRINCIPLES st 5. To illustrate that earth pressure, bearing cap. acity, and slope stability belong to one and the same family of problems and that limit equili brium theory provides a unified treatment of such problems. A review of this kind is a prerequisite for going into more complicated analyses, such as those that are dealt with in Section 3. Stress Conditions in Soil Elements In soil mechanics, total normal stress o is positive when compressive, and shear stress 7 is positive for counter clockwise rotation (Fig. 2). In two dimensions the equations for the stresses, on a plane inclined at an angle a relative to the direction of the minor principal stress 0; are a= 07 cos? ator sin? a (as) Tq = sin 60s alo; —o5) as) where 7 is the major principal stress and o; is the minor principal stress. These analytical expressions are most conveniently displayed graphically by Mohr stress circles as follows. Using the symbols (see Fig. 2), R=i(o,—0,) = radius of Mohr cirele (15) normal stress corresponding tocenter of Mohr circle (16) Om = Ho, +05) Eqs. 13 and 14 become cos 2a az sin2a as) Fig.2, Stresses onthe a ‘Squaring and adding these two equations yields (g~ Om)? 742 = RE (a9) ‘This is the equation for the Mohr stress circle, with radius R and with its center at a on the axis of normal stress (Fig. 2). Dividing Eq. 18 by Eq. 17, one obtains tan2a= (20) ‘The meaning of 2a and ais illustrated in Fig. 2. ‘The effective normal stress a’ and the pore pres: sure w are related to the total normal stress in fully saturated soils by Terzaghi’s principle of effective stress: o'=o- (2) Free pore water can take no static shear stresses, that is, pore pressure acts equally in all directions. Therefore the Mohr circle for effective stresses has the same diameter as the Mohr circle for total stres ses but is shifted to the left on the normal stress axis by an amount u, as shown in Fig. 3. Thus the ex pressions for the shear and normal stresses on plane 4, in terms of effective stresses, are the same as =Rsin2a (23) and R=4(o,—03) = Hoj—o}) 7. le fe 52 SLOPE STABILITY COMPUTATIONS. Fig. 3. The Mohr stress circle for effective and total stress. From the Mohr diagram, as well as from Eqs. 18 and 13, it ean be seen that the maximum shear stress facts on the planes inclined at a=+45°, and is given by te = R= @n rae Mo,-e,) ==Hoi—05) and the corresponding normal stresses on these planes are (25) 05 = Om= Mort) Os = 4 = Moi +05) (26) Equations 24, 25, and 26 define the top points on the Mohr circles. Thus one can represent the entire Mohr circle by plotting this one point, and this is often done in the interpretation of the results of various shear tests on soils. In slope stability analysis, as well as for the inter- pretation of test results, it may sometimes be neces ary to determine the principal stress direction from known shear and normal stresses on two mutually perpendicular planes (Fig. 4). The expressions for the principal stresses and the maximum shear stress Tet Qn 01 =Ho,+02) + |Fmaxl (28) (ost 4) — lr (29) If the two planes on which the stresses are known are not perpendicular, the expressions become more complicated. The lower part of Fig. 4 shows two examples of soil elements subjected to given systems of no and shear stresses, say in a direct shear box. main purpose of the examples is to illustrate that the directions and magnitudes of the maximum shear stresses may be radically different from the applied shear stresses. For instance, 7pax = 3.9 while 3.0 in Example 1, and tnax = 3.2 against r= 2.0 for Example 2. The directions of the maximum shear stress are drawn directly on the elements for illustra- tion purposes. In the interpretation of direct shear tests these elementary conditions appear to be easily overlooked. Failure Conditions in Soil Elements Stress conditions at failure are most conveniently studied with the aid of a Mohr-Coulomb diagram Fig. 5). Figure 5a shows the effective stress con: dition at failure when c' and o" are constant. Points P and P’ define the inclination, * ay, of the critical shear planes. [tis readily seen that y= = 1S +48") (30) At failure the major effective principal stress can ELEMENTARY MECHANICAL PRINCIPLES 53 Fig.4. Determination of the principal stresses in plane strain be written as follows (Terzaghi, 1943): Neos + 2c'VNy (31) in which N; (= tan® ay) is often ealled the flow value, Figure 5b shows the total stress condition at failure for the ease of 6 = 0. The major total principal stress is then equal to oy = ost 2s (32) ‘The failure planes are inclined at a= 45°. ‘The key sketches above the Mohr diagrams in Fig. 5 show one small shaded element within each of the two samples, These elements are hounded by failure planes on all four sides. In many analyses it may be of considerable importance to know the stress conditions in such elements, and therefore they are drawn to an enlarged scale in Fig. 6. The effective normal stress o’ on the failure planes AB, BC, CD, and DA is equal to Sit Noa (33, ‘The shear stress on these failure planes, that is, the shear strength, is The angles between the theoretical failure planes are (906). For the case of ¢ = 0 the theoretical failure planes are perpendicular to each other (Fig. 68). The total normal stress and the shear stress (i.e., the shear strength) on these planes are Moy+os) (35) (oye) (30) Equation 31 has often been used in interpreting triaxial tests on nonmetallic solid materials (e.g, see Terzaghi, 1945). For dry materials, like concrete and rock, the equation is preferably written as follows: oy=Nyortou (37) where a»; = unconfined compressive strength N,= flow value Equation 37 is illustrated in Fig, 7a. Ifthe results of triaxial tests plot as a straight line on this figure the values of ¢ and 6 can be computed from the formulas tang = ACI (38) 54 ‘SLOPE STABILITY COMPUTATIONS |_— Fale panes aes 'tng 7 Gi ats ste o Faire planes for 6 = 0 Fig.5. Mohr-Coulomb diagrams. fon mar. Corresponding to these large extension strains in the Figure 7b shows the results of triaxial te ble (von Karman, 1911). Using Eqs. 38 and 39, one y-direction, it has been observed that failure by obtains tan 6 = 0.72 and ¢ = 2400 t/m splitting in the oy-direetion often occurs in triaxial If no volume change takes place during plane tests at low confining pressure on brittle materials. then Figure 8 contains photos of a number of different failure patterns for rock and ice specimens (Hawkes 541 (plane strain (40a) and Mellor, 1970), The most ideal failure patterns, 1 , consisting of numerous failure elements, are shown 1 (triaxial) (400) in the last photograph in Fig. 8. 0) or triaxial test strain tests ELEMENTARY MECHANICAL PRINCIPLES 55 os >0 wero Fig.6. Stresses on infinitesimal elements bounded by failure planes. Figure 9 illustrates large-scale, in situ failure pat ey were visible on the day of the Verdalen way, on May 19, 1893. For details of the slide see Holmsen (1953). The sharp, angular ridges pointing up from lakes of liquid mud may be the remains of large-scale failure elements, Research on failure patterns in situ and in soil elements leads the author to believe that the shape ‘and location of rupture figures in soil is primarily governed by the stress system in the soil. Except for soils subject to progressive failure (brittle rup- tures), even large variations in induced strain appear to be of less significance, at least for the geometry of the failure terns as lide in ® ature Fig.7._ Linear failure envelopes in principal Stress Vector Curves The stress changes applied to a sample in the labor atory should model as closely as possible the existing stress and the stress changes that take place in situ. ‘The purpose of this section is to describe useful ways of depicting graphically the stress and strain changes that occur in situ or during the laboratory strength test. ‘The path along which a significant stress point moves during stress changes from a starting condi tion toward and along a failure condition is called a stress vector curve. It is recommended that the strain be plotted at several points along the vector curve to 20000 noon tung! = 072 toncin? ° ° 300 soni 3s diagrams. 58 SLOPE STABILITY COMPUTATIONS Fig. 8. Failure patterns in test samples. (After Hawkes and Mellor, 1970.) make it more informative. There are four ways of plotting stress vector curves: (I) principal stress vee tors, a versus 0; (2) modified principal stress vee tors, Ko! —o5) versus a: (8) stress paths, to} —0:) versus (oj + 0%); and (4) limit equilibrium veetors, 18" on critical plane. Principal Stress Vector. If ois plotted against 0, failure is defined by the classical equation (31) or Eq. 37, which is illustrated in Fig. Ta (see also Terzaghi, 1945). However, if one uses the same scale along both stress axes, the classical failure line (Eq. 31) is often so steep (V,= 3 to 10) that accuracy may be lost. This shortcoming can be overcome by using modified principal stress vectors. Modified Principal Stress Vectors. The coordinates of the modified principal stress vector are obtained by rewriting Eq. 31 Moi—0}) =4N—DoyteVN, (A) By using a modified flow value N, defined by the relationship Np=14+2N (42) Eq. 41 may be reduced to Ry= Noj+7 (43) where R; is the radius of the Mohr failure circle and tq (=c’VN,=c'V1+42N) is the intercept on the Reaxis (Fig. 10a). The modified flow value Vis hence the slope of the failure line in an R-o! plot. ‘A modified principal stress vector is therefore obtained by plotting half the deviator stress R versus minor effective principal stress «7, If the envelope of this vector is a straight line (constant e” and '), the slope NV and the intercept 79 can be used to cal: culate the shear strength parameters (Fig. 10a) N VFN (wy tan 6" (4s) Vi+IN The use of modified principal stress vectors is illus trated in Fig. 106. An isotropically consolidated, un- drained triaxial test on loose sand, with porosity of 46%, gave values of N= 1.28 and 7) 0 for strains between 4 and 10%. Hence Eqs. 44 and 45 yield tan 6! = 0.68 and c' = 0, ‘Aline corresponding to F = 1.4 in Fig. 10b inter: sects the vector at a point where €, ~ 1%, which means that for this test a major principal strain of ‘1% was needed to create a state of limit equilibrium: corresponding to a factor of safety of 1.4. Figure 11 shows the results of three consolidated- undrained tests on a medium stiff sity clay. The common failure line for these tests has a slope N 1.17 and an intercept 7y= 1.4 #/m®. Equations 44 and 45 yield tan 6! = 0.64 and o’ = 0.76 tin’, For more detailed information about the applicability of the modified principal stress vector curves, see Janbu (70) Stress Path. In an effective stress path plot, S(a}—0) versus 4(o} +03), a Mohr-Coulomb fail tre envelope with constant c’ and q is represented by a straight line. The intercept and slope of this line are readily derived from Fig. 12 R,= (o,-+07,) sing! (46) Since a} = c' cot 6’, one obtains R, = asin g! +c! cos 6! (a7) This is the velope on t squation of a Mohr-Coulomb failure en: e stress path plot. Hence the intercept of Rediei-ei) 20 20 ton? Fig. 10. Modified principal st 58 SLOPE STABILITY COMPUTATIONS. 3 Fig. 11. Results of CU tests on silty clay interpreted by vector curves, the envelope of the stress paths is c’ cos 6! and the slope is equal to sin ¢' (Fig. 126). For detailed in. formation about the application of stress paths, see Lambe (1967). Limit Equilibrium Vectors. The path along which the critical stress point moves as the stresses are changed is called a limit equilibrium vector curve. Fora given stress circle in a Mohr-Coulomb diagram, and for a known 7-line, the critical stress point e is obtained simply by drawing a tangent to the stress circle from the reference point 0 (Fig. 13a). At this point r= 7/F pine Figure 136 shows the results of an anisotropically consolidated-undrained triaxial test on saturated “undisturbed” silt. The shear strength parameters fare found to be c’=0 and 6’ = 33°, and a major principal strain of approximately 3% was needed to reach a theoretical state of failure, defined as follows: 4) tt (48) ‘This state is approximately maintained between rh ow Foe strains from 3 to 15% (because of the decrease in ore pressure for increasing strain). Figure 136 shows that a state of limit equilibrium corresponding 5 was reached in this test at ¢, = 0.7%. Stress Conditions along Classical Rupture Surfaces The elementary stress conditions derived earlier are often helpful in the study of the stress conditions along failure surfaces in general and along zone boundaries in particular. For some simple cases one ray also use these elementary principles to deduce the shape and location ofthe failure surface. Figure 14a shows a bearing capacity problem for weightless soil (y= 0), plane strain (¢,=0), and a smooth base. For the condition @ = O one may argue as follows. The surface 00’ is assumed smooth, hence the load q is a principal stress. Therefore, the inclination of the failure planes under oo must be 45°, Consequently, the failure planes from the edges of the load q intersect at point m at right angles. Surface 00" is also assumed smooth and therefore the failure planes are inclined at 45°, and they intersect at point n. In the vicinity of points mand n the infinitesimal elements (bounded by four failure planes) are squares of the type shown in Fig. 6b. The failure elements’ along curve mn are also squares, and therefore curve mn must be a circle. ‘Thus the failure zone consists of four triangular (Rankine) zones and a circular seetor, the Prandtl zone. The direction of the major principal stress along mn changes from vertical at m to 45° at m’ to horizontal at. The stress conditions within the triangular zones are as follows for y= 0. On ae, 0; =4 and on 27, 3 =q~2s (Eq. 36); hence the normal stress on am=4—s. On oe", oy =p and on fz’, a hence the normal stress on fio = p-+s. For the sector = p+2s; 7 Yi Hesse) Fig. 12. Principle of stress paths ELEMENTARY MECHANICAL PRINCIPLES 59 Fig. 13. ‘omn moment equilibrium about point o yields H(q—s)RE= dasR? +4 pts) RE The resultant of the normal stress o along 7m passes through point 0 and does not enter the moment equation. Solving for q, one obtains (49) which is the well-known formula for the ultimate bearing capacity for smooth, shallow footings on soils with constant shear strength, as first shown by Prandtl (1920). For the ease of c= 0 and @ > 0 (Fig. 145), the arguments for deriving the failure surface for y= 0 are very much the same as above, except that the infinitesimal failure elements are now rhomboid (Fig. 6a). Therefore angle o'mo = 90°—¢, angle ono" =90°+4, and the smooth curve from points a= (at2)s+p i Limit equilibrium vector curve, m to nis a logarithmic spiral with the equation R=Re, Fee (50) ‘Thus the failure zone consists of four Rankine zones and a logarithmic spiral sector. Along the spir- al the major principal stress changes from vertical at point m to horizontal at point n. The stress con- ditions in the triangular zones are as follows. On 6, 0 = 4, while os = q/N,on cm; hence the normal stress on ami is obtained from Eq. 33: —4 a «1 On ae", 0; = p. while o = pN,on ne’; hence e 2) nar Tell berpeereeceon ssical rupture figures. 60 SLOPE STABILITY COMPUTATIONS Along the spiral jm the resultant stress passes through point o. Hence moment equilibrium about point o yields Banh = Wonk? Since the angle between Ry and Ry is o = finds from Eq. 50 Ryo = Ruger? By introducing this relationship, og, and into the moment equation, one obtains Nop (N= Nyer™n) (53) Here N,= bearing capacity factor for load p for shallow footings, in accordance with the Prandtl theory. Plastic versus Brittle Failure To illustrate the differences between ideal plastic and ideal brittle failures, two simple cases will be studied. The examples are limited to materials of constant shear strength. ‘The first example is a flexible, uniformly distribu ted strip load q on the smooth surface of a material of constant shear strength s. When plastic failure is developed along the entire length of the eritical shear surface, the bearing capacity qy is given by the well- known formula (Prandtl, 1920) an (w+ 2)8 (54) For an elastic material a flexible, surface strip load q leads to a maximum shear stress max along the load edges equal to 4 55 e (55) In an ideal brittle material, failure must occur as soon as Zax becomes equal to s, because the stress taken at the failed point must be transferred to adjacent points, which immediately fail themselves. Hence the surface-bearing capacity q,, for a brittle failure is (56) Accordingly, the ratio between the bearing capac- ities for plastic and brittle failure is aa 7+2_ 1 6 QF The second example is a vertical, unsupported cut of height H below the horizontal surface of a soil with unit weight y and constant shear strength s, If a state of complete plastie failure is developed along the entire length of the critical shear surface (45° plane through the toe), the critical height is given by the classical formula 4s Ha= > 57) Fi ( However, the vertical stress at an element near the toe is 0) =H and the horizontal stress = 0. Therefore the maximum shear stress at the toe, taken as (0; —05), equals yl (58) In an ideal brittle material failure will occur wh Tmax = 8 Hence the critical height in a brittle material is (59) and the ratio of critical heights for plastic and brittle behavior becomes Hp _ Hy? c Limit equilibrium analysis is based on the assump: tion of ideal plastic behavior. The foregoing exam- ples show that for a given strength, ideal brittle materials fail at a lower average shear stress on the failure surface than ideal plastic materials. How. ever, most soils that would be termed “brittle” or “plastic” lie somewhere between ideal brittle and ideal plastic materials. Therefore the difference between “plastic” and “brittle” behavior for real soils is smaller than for ideal soils. Stresses at Limit Equilibrium For any point ¢ along a linear r-line (Fig. 15) defined by constant limit equilibrium parameters ce and tangy, the relationship between the principal stresses is given by Of = Noh 2GVN (el) in which (45 +4.) (62) N,= tan? ag = The normal stress o” and the shear stress r along the critical shear plane a, at equilibrium are ait Naot (63) (61) ELEMENTARY MECHANICAL PRINCIPLES 61 — auiteum o Fig, 15. Definition of limit equilibrium and comparison with the failure condition. ‘The limit equilibrium values tan a, and N, are most conveniently obtained in terms of tan, by the expressions VN. = tana, = tan bet VIF tant d, (65) The consequence of using the factor of safety to define the limit equilibrium condition is that one can readily discriminate between failure shear surfaces (Fin = 1) and critical shear surfaces at equilibrium (Fain > 1). Since the difference is solely due to the difference between @, and 6’, these surfaces are different from one another only when 6 > 0, as illustrated in Fig. 15. Figure 150 shows the condition of active earth pressure in sand fora completely smooth wall. In this case the limit equilibrium shear surface (dashed) is more deep-seated than the failure surface. Corres: pondingly, the active earth pressure at limit equili brium is larger than the failure value Figure 15c shows the conditions for passive earth pressure in sand for a completely smooth wall. The limit equilibrium surface is more shallow than the failure surface, and the passive pressure at failure is greater than the limit equilibrium pressure. Similar differences would occur for rough walls, although the critical shear surfaces would then be curved. Assume that by means of a general slope stability analysis one has determined the shape and location of the critical shear surface at limit equilibrium Fig. 16a). Then 7 and a” are known along the surfac Fig. 18. Boundary 62 ‘SLOPE STABILITY COMPUTATIONS. For such a case one can readily obtain the principal stresses along the critical shear surface by solvin Eqs. 63 and 64 with respect to o and a. The result of=o + VNer (66) (67) The critical shear surface at point @ intersects the ground surface (a plane of principal stress) at an angle ag, where a = 45° +4. (68) ‘The stress condition at point a (an expansion zone) is (09) The critical shear surface at point b intersects the ground surface (a plane of principal stress) at an angle ay, where 45°40, (70) The stress condition at point b (a compression zone) 20. mm llustrated in Fig. 168. ritical shear surface the d the These stress conditions ar For any point along t angle a; between the major principal stress shear surface equals ay = 45°44, (72) Knowledge of boundary and stress conditions can often lead to a reduction of the number of trial sur faces required for determination of the minimum factor of safety in stability analysis. In slope stability analysis the zone around point a in Fig, 16a deserves special attention, because of the tendency to develop tensile stress 0. s when o* At limit equilibrium, tension may exist down to a theoretical depth z,, corresponding to o = 0 in Eq, 61, while o = 7z. Solving for z = z, one obtains 2e.V = ee 73) F (i fective unit weight In the short term, tension may be a reality. In the long term, however, it is probably unrealistic to count on tension being maintained at its theoretical value. For this reason, stability analyses are often made by assuming that tension eracks exist to a depth x Limit Equilibrium as a Unifying Principle In Fig. 17 a half-space of soil is loaded over half its surface with a uniform stress q. Through the soil are drawn two shear surfaces, both of which are sym- metrical about the vertical ob. The average shear stress 7 that is required for equilibrium along either of these surfaces can be expressed by the following formula: (74) where W is a dimensionless parameter called the stability number If the soil is isotropic with respect to shear strength, and if the shear strength s is constant, the average factor of safety with respect to shear failure along each surface can be computed as follows: (75) The maif object of stability analysis is to find the location and shape of the shear su the minimum factor of safety: ian Fe. rant setce Fig. 17. Illustration of the common principle involved in earth pressure, bearing capacity, and slope stability analysis ‘THE GENERALIZED PROCEDURE OF SLICES 63 where Neis the critical stability number, that is, the minimum value of N, Along any surface except the critical surface the computed stability number N will be greater than N, unless some incorrect simplifying, assumptions are made in the analysis. For example, stability number for the critical circle in Fig. 17 is . while the statically correct Prandi! surface is to; = 5-14, which is about Thus the computation of the safety is analogous to the problem of minimizing (or maximizing a function of two variables, the loca- tion and shape of the shear surface. It is not sufi cient, in general, to select a shape and minimize on the basis of location only. The importance of con sidering both of these interrelated variables increases considerably if the soils involved have highly non- uniform shear strengths, ‘The stability problem in Fig. 17 ean also be con- sidered as a bearing capacity problem. Solving Eq. 7A for q, one obtains “lower. nimum factor of q For constant limiting shear stress, 7 = s/F, the mini: mui value of Vis equal to the bearing capacity num ber, NV. = 5.14. The corresponding equilibrium bear- ing pressure q, then becomes Nee (78) Here qe represents the “allowable hearing pressure" for a surface load and for a specified factor of safety with respect to shear strength, say F = 2. ‘One may also study the stability problem of Fig. 17 by consideration of the horizontal limit equilib rium earth pressures at depth z on both sides of the vertical ob. For constant limiting shear stress, 7= S/F, and for all shear surfaces that are symmetrical about ob one obtains the following formulas for the equilibrium horizontal pressures at depth z: Py=qtys—WNr (79) Pp = ytiNe (80) In a perfect state of limit equilibrium, the shear ical ob is equal to the shear surface, abc. In this case the 5.14) leads to maximum Py stress 7 along the ve stress along the crite: minimum N (ie, Ne and minimum P», which are the active and passive pressures at limit equilibrium (for a completely rough face ob). The bearing capacity equation (77) can be derived by setting Eq. 79 equal to Eq. 80. of the problem illustrated in Fig. 17 shows that active and passive earth pressure prob- lems, bearing capacity problems, and slope stability problems can be solved by one and the same set of basic principles, In the next section of this chapter the main em phasis is on slope stability problems, but the prin ciples discussed can be applied to any problems in volving limit equilibrium of soil masses. 3. THE GENERALIZED PROCEDURE OF SLICES Purpose ‘The generalized procedure of slices can be used for any of the following purposes: 1. To solve stability problems for irregular topo- graphy and layered soils of different shear strengths, using shear surfaces of any arbitrary shape. 2. To explore a statically reasonable state of stress both along the shear surface itself and n the body of soil located above that surface 3. To unify the treatment of such categories of soil stability problems as earth pressure, bearing capacity, and slope stability For the sake of brevity the procedure is referred to here as the GPS procedure. ‘The basie principles of the GPS procedure were laid down for the first time by the author in a dis- cussion at the conference on Stability of Earth Slopes held in Stockholm in 1954. An abbreviated routine procedure for computing the factor of safety for composite shear surfaces was published by Janbu, Bjerrum, and Kjaernsli (1956). Re: cognizing that earth pressure and bearing capacity problems could be solved by the same principles, Janbu (1957) extended the GPS procedure in a paper ‘entitled “Earth Pressures and Bearing Capacity Cal: culations by Generalized Procedure of Slices. In more recent years a number of authors have dealt with the problem of stability analysis for shear surfaces of any shape. The papers by Morgenstern and Price (1965) and Nonveiller (1965) are now well known, Both these papers are based on the same basic principles as the GPS procedure. The differ: ences are to be found in the assumptions regarding interslice forces and in the details of the computa: tio The GPS procedure was so briefly covered in 195; 64 SLOPE STABILITY COMPUTATIONS that its general character and scope of application appear to have remained almost undetected. In recent years the author has had several requests for a more complete presentation of the theoretical and practical aspects of this procedure, particularly for slope stability analysis, and this book offers a most appropriate opportunity to do so, Basic Principles Figure 18 shows a cross section of a slope for the ral case in which the terrain, the external loads, the boundary conditions, the soil profile, and the shear surface may be irregular. The problem to be solved may be an earth pressure, bearing capacity, or slope stability problem, The Shear Surface. The shear surface ab in Fig. 18 is only one out of an infinite number of possible shapes and locations of such surfaces. The shear sur face that gives the minimum factor safety, the crit- ical load, or the critical earth pressure is termed the critical shear surface. When the factor of safety is unity the corresponding surface is called a failure surface (or a potential slip or sliding surface). Theoretical and practical evidence is now so ex. tensive that the required number of trials for locat ing the critical shear surface, even for complicated composite surfaces, may be kept to a fairly small number. The Slices. The soil mass located between the ground surface and the assumed shear surface is divided into a number of slices by vertical lines. A cross section of such a slice is indicated by the dotted area in Fig. 18. This same slice is drawn to an enlarged scale and the system of notation for Fig. 18. Definitions and notations used for the generalized procedure of slices the forces acting on all four boundaries of the slice is shown on Fig. 19, ‘The resultants of the total interslice forces are E and 7, in the horizontal and vertical directions, respectively; AS and AN are the resultants of the shear stress 7 and total normal stress «acting over the length Al along the shear surface. Hence Aj Al and AS =rAl. In the general case, E, T, AN, and AS are unknowns. Basie Assumptions. The following basic assump. tions are used for the general case: 1. Plane strain conditions apply 2. The equilibrium shear stress along the shear surface is given by the equation F where ty is the shear strength and F is the factor of safety. The shear strength will be ex pressed as a function of normal stress, as out. lined earlier. (Insomecases, simplifyingassump- tions regarding the factor of safety are intro- duced after the equilibrium equations are established.) 3. The total resultant AW is assumed to act where AW = AW, + q\x+ AP intersects the base, (al) Fig. 19. Forces acting on the boundaries ofa single slice. ‘THE GENERALIZED PROCEDURE OF SLICES 65 4, The position of the line of thrust for the total side force E is assumed to be known, Numer ieal investigations have shown that the conse quences of changing the position of the line of thrust within wide margins have insignificant effects on the computed factor of safety. More: over, our knowledge about stress distribution from earth pressure theories is now sufficient to make it a fairly simple matter to select a proper line of thrust. Known Quantities. The analysis is begun by selecting a trial surface and subdividing the soil ‘mass into an appropriate number of slices (which is about 5 to 10n most practical cases). From the shape and location of this surface (Fig. 19) one obtains for each slice (dimensionless) (ength) tan a= slope of shear surface Ax=slice width (82) The average total vertical stress at the base of cach slice is computed from a yetat ge > (foreejlength®) (83) ‘ax For stability analyses based on limit equilibrium, the shear strength parameters must be known, and their values may be different from slice to slice. Horizontal forces located either at the ground surface (inclined loads) or in the interior of the soil mass (earthquake force) can be taken into account. ‘The magnitude of the horizontal force is designated AQ and za is the distance that AQ acts above the assumed shear surface. When the position of the line of thrust is chosen, one measures hy = vertical distance between shear surface and Tine of thrust (length) ope ofline of thrust’ (dimensionless) (84) tana Both quantities are measured at each intersection, and not in the middle of the slice. (The same is true forzq and AQ/Sx. For the case of c’ = 0 the line of thrust should be selected at or very near the lower third point, fy ~ Jz. For c' > 0 the line of thrust should be located above this point in a compression zone (passive condition) and somewhat below it in an expansion In tension zones, it is re (as defined on zone (active conditio commended that eracks of dep page 62) be assumed, or that a theoretical tensile force E, (negative) be introduced above =, (Fig. 18). In both eases the shear surface starts at depth 2, Equilibrium Conditions. The requirement for static equilibrium must be satisfied both for indi vidual slices and for the whole body. The equations for equilibrium are derived later in this section, The complete set of basic equations, which must be satisfied simultaneously, is summarized below for reference purposes ectlo—u) tan de (85) ptimrtana (86) SE= AQ +(p+ Sx tan a—rAx(1+tan*e) 67) ae, a0 ede (88) T=-Etan a, where t= dT/dx. Equation 85 is the defining equa- tion for the state of limit equilibrium, Eq. 86 is the equation for vertical equilibrium of each slice, Eq 87 is the equation for a combination of horizontal and vertical equilibrium for each slice, and Eq. 88 is the equation for moment equilibrium for a slice of infinitesimal width. The requirement for overall horizontal equilib- rium leads to the following equation (Figs. 18 and 19): (89) Sar= Overall vertical and overall moment equilibrium are automatically satisfied by the foregoing equations when the boundary forces on the end slices are properly handled. In slope stability analy safety and four unknowns for each slice, and r, must be determined. For bearing capacity and earth pressure problems the unknowns that cor- respond to the average factor of safety for the slope stability problem are the average bearing capacity and the resultant earth pressure (Por Pp), respec the average factor of tively. Working Formulas for Slope Stability Analysis ‘The required formulas for calculating the average he shear surface, and the average factor of safety, the interslice forces and stresse factor of safety along interfaces between slices are readily derived from Eqs. 85 to 89. fuerage Safety Factor. Inserting Eq. 87 into 89, 66 SLOPE STABILITY COMPUTATIONS one obtains SAO Hip +H Astana] Ey 3 ran(+tanta) IP into the expression above, and constant (equal to the overall aver- d shear surface), Introducing r= considering F age safety factor along the selec and solving the equation for F, one finds © 7/Ax( + tanta) s (90) E,-Bx+ 3 [30+(p+ 0 tan ads] where p= AW/Ax and ¢= AT/Ax. For d= 0 the value of the shear strength 7) = ¢ = sy is directly known in force per unit area, Ci quently, Eq. 90 is explicit with respect to F and the only unknown is ¢, which can be obtained from Eq. £88 by iteration, However, in the general case when 6’ > 0 a for mula for the shear strength 1, must be derived. When introducing Eq. 86 into the general shear strength expression, one obtains y= e'+(p+t—u—rtan a) tan d! Introducing + = 7)/F and solving for 7,, one finds (p+ t—u)tan é T+U/F)tan 6 tana ey ‘The formula for the average safety factor for the general case is obtained by incerting Eq. 91 into Eq. 90. For simplicity its advantageous to introduce abbreviated terms in the formula for F, particularly for longhand calculations (see next seetion). Abbreviations. For each slice the following abre- viations will be ws AQ+(p+0Mxtana (92) A= 1)Ax(1+ tanta) (98) Inserting Eqs. 91 and 93 into Eq. 90, the formula for the average factor of safety is reduced to F=—__ oy a Eyt DB For ! = 0(z/= s,), the factor of safety occurs only ‘on the left of Eq. 94 and can be solved for directly For ' > 0, F appears on both sides of Eq. 94 and a trial-and-error solution is required. By introducing Eq. 91 into 93, the A term for each slice can be calculated in three steps, as follows A’ =[c'+(p+t—u)tang'}Ax (95) L+0/F) tan 4! tana I+ tanta (96) “ on ‘The intermediate term ng is plotted versus tan a for different values of tan ¢/F in Fig. 20. In longhand cal culations one can save time by obtaining n, from Fig, 20(as discussed in the next section), Stresses 7 and o along the Shear Surface. Once the A term for each slice is computed, the shear strength ris obtained from Eq. 93: (98) _ 1 HH eh Fig.20. Values of n, as function of tan a and tan 4, tang. THE GENERALIZED PROCEDURE OF SLICES 67 ‘The total normal stress on the shear surface is now calculated as follows: ptt—rtana (9) according to Eq. 86. Incerslice Forces. Introducing Eqs. 92 and 93 into Eq. 87, and observing that 7 = 7/F, one obtains for each slice =p-4 AE=B-5 mming the AE values for each slice, starting at point a in Fig, 18, one obtains the total side force E on each interface from the formula E=E,+3AE (ony (100) Along this same interface the vertical shear force Tis equal to (Eq. 88) dE. do T=—Etanathg. 20g, (102) After obtaining the T' values for all interfaces, one computes AT for each slice, and the corresponding values = AT/Ax. All the preceding equations must be satisfied simultaneously by iteration procedures, as described later. Interslice Stresses. The average horizontal total stress @ is by definition = (103) ane If, for simplicity, one assumes a Iinear distribution of o% along interfaces, one can determine the stresses Gy and oy, at the ground level and at the shear surface, respectively. Since the computed E value corresponds to a selected value of he = 7: (Fig 21), horizontal equilibrium and moment equilibrium Fig.21. Notations for stresses on vertical sides offices, about the base are satisfied when On = (A-6n)on (04) On = (6n~2on (205) The average vertical shear stress is by definition r * (106) Let the factor of safety against shear along a verti- cal interface be designated by F,. If one assumes F,= constant over the entire height of the interface, fone can compute the vertical shear stresses at ground level and at the shear surface: = fe gyms aor to= f+ 04 ins (ao7) = toi, tan! (108) The average value of F, for each interface is calcul ated as shown below. Factor of Safety along Interfaces. On the basis of the known, or assumed, pore pressure distribution ‘one can compute the resultant horizontal water fore Uy and the horizontal effective resultant E’ = E— Uy. This resultant produces an average horizontal effective stress, = E"/z, from which the shear strength 7.yon the interface ean be computed: '+o%, tang! (109) One can now determine the average factor of safety F, with respect to shear failure along cach interface from the formula F,= ty!7e, where 75 = Hence Jian 6 Fr, (ato) T In a theoretically correct solution F, =F. For layered soils the computation of average F, values becomes much more complicated. Iteration Procedure ‘The working formulas given earlier can be solved ¢ iteration procedure used in slide rule calculations is explained with the aid of Fig. 22 and Table 1 by iteration. TI Data from Profile. First a trial shear surface is drawn and the soil mass is subdivided into a number of slices. To save space only four slices are used in Fig. 22. From the profile one obtains the following quantities for each slice: the average slope of the 68 10m et ten fmt as ‘SLOPE STABILITY COMPUTATIONS 10 (torsim®) ° i 5 tons) Fig. 22. Safety factor, interslice forces, and stresses along a noncircular shear surface. red below the center of shear surface, tan a, me gravity of the slice: the width of the slice, Ax; the average total overbustlen, p (which in this example is the average pore pressure u shear simply equal to along the base of each slice: the average strength parameters c' and tan d', both of which may differ from slice to slice; and finally, an horizontal force AQ acting on any slice. In this example AQ = 0. These quantities are recorded for each slice in the first seven columns in Table 1. Note that for the sign convention used in Fig, 22, tan a= dy/dxis negative for slice 4. Calculation of Fa. Experience has shown that it is simplest to start the iteration letting ¢ = 0, because the formulas can then be solved with the di value of the safety factor in col tumns [to 7 of Table 1. Th ‘THE GENERALIZED PROCEDURE OF SLICES corresponding to fq = 0 is called Fy, It is obtained as follows. For each slice calculate Bg for to = 0(Eq. 92): By = AQ+ptanadx au) d Record the values in column 8 in Table 1, 8 obtain 3By at the bottom of the column. Next eal alate Aj for t= 0 (Ea. 95) Ay= (e+ (p—w)tand'Jax (112) In order to obtain the no values needed in col: umn 10, one should already know the Fy value. In principle, Fy would have to be obtained by iteration Tn practice, time may be saved by summing the A; values, thus obtaining an approximate value of Fo for taj = 1. In Table 1 Fy= 131.3/98.3 = 1.34. By assuming Fy= 14, somewhat above the estimated Fi, one can obtain ny from Fig. 20 or from Eq. 9. Then Aya (i) (For 6 = 0 one can obtain Ap directly from Eq. 93 for 1/=5y.) Compute ¥4y, at the bottom of column 11, and obtain F for ty = 0 (Eq. 94), = Fe +38, =1385 (114) If the computed Fy is considerably more than 5% different from the assumed Fy (in go), then a re. calculation is recommended. Here, the assumed, F y= 1.4 is s0 close to the calculated Fy = 1.385 that a recalculation is unnecessary. Calculation of Ey. The initial values of the forces E, are calculated in two steps. First, Eq. 100 yields the change of E over each slice (for fa = 0}: Ay 5 Bo Fe (11s) For instance, for slice 2 in Table 1, one AEs nds =15.1 (tons per meter) ‘The calculated ABy values are shown in column 12, where YAE, = O because E, = Ey = 0. Starting with the known boundary value Ey at point a, one obtains the By force for each interface by adding the AZ, values, one by one, ending with the given boundary force Ey at point b. The E forces act between slices, and column 13 is subdivided accordingly, by displaced horizontal lines. 69 Computation of T,. On the basis of the initial values of Ey and an assumed line of thrust, one ob: tains the first approximation T, of the T forces (Eq. 102): Eytan ayt bt 2, 2 ea dx dx D This calculation is contained in columns 14, 15, 17, and 18 in Table 1. In general the derivative of E with respect to x at each interface (col and 28) can be obtained either by measurements on E~x plots (Fig. 22) or by computations, as of a chord value between slices i and i+ 1 as follows (d8) =a dx)ison ARAM a?) For instance, the slope of Ey between slices 1 and 2is (dEe) 7415.1 _, 9, (22) = ESB 1.74 (tons per meter) A line of thrust is now drawn on the profile and its slope tana, and distance fy from the shear surface are measured for each interface (columns 16 and 17). In Fig. 22 the line of thrust is assumed to be located at the lower third point, except near the toe where it is taken somewhat higher, and near point a where it is somewhat lower. (When AQ > 0 the table must also contain columns for zy and dQ/dx.) Between slices 1 and 2 one finds from Table 1 Tag =~ 11.7 X0.63+ 1.2% 1.74 =~ 5.29 (tons per meter) At the boundaries, points a and b, Ty = T,=0. Known 7 and 7, values must he recorded in column 18, at top and bottom, respectively, in order to ob- tain correct AT values for the end slices (column 19). The quantities in columns 14 to 18 refer to inte slices, and columns 19 to 25 refer ice, For instance, the faces hetw to average values for each change of Tver slice i equal: AT, = Tear T as) Applying this general equation to slice 3 in Table 1. one obtains for the frst iteration AT, =~ 3.38— (-8.96) 4.5.58 tons/meter) ATIAx For each slice one can then obtain ¢ Calculation of Fy. The values of T, (or t)) are used to find an improved value of the factor of safety F When comparing the general formulas for B and 4 70 SLOPE STABILITY COMPUTATIONS Table 1. The teration Procedure = initial step Data from profile I Calculation of F, AE, and E, Se OF 2 8 @ 8 6 M/S @ Go an aa no tana kop ou tang’ ad | 8, A AEs 1113 5 os7_o | 264 138 068 20311 2——050-110-10.1 40410067 —0-| 556555 089560 is 301g 110” 86-348 10067 0] 17.0490 1.05456 =165, ‘#008 6023-116 10067 0 ]-07 130098133103} a "3383 WS Ta eZ SOT iteration Fi 1.385 sculatonof7, Calculation of FE. and Slice (14) (15) ay [as oy [2 (22) (23) (2m) 28) | (28) 20 a, & 7 Ql 509 120 102 08s isa 100) =0 367 —035| 37 531 ase 583172] 100 v5sa r0si| 180 527105504 —159} 2 05 S158 016 11 —338)-333-+056[-08 153098186 —113 =o 3-913 (313) Fis) 1387 = BA 1357 _ Weration Fy= pA = OZ 1.495 Calculation of. “ Caleulation ot Fe, AE,andé, | Stresses Sie 2) 28 @ Go an} oa |e GH Go Gn Ge [Ga] a aD 0 a, 6 tana, nan fa ms A Mb] &] on ow 417 -095|216 109 066 165 104 f=! 465 —043 [532525 098 536 17.0 204 272006033 18 885/75 25 F048] 180 526 1.05502 —i59}274 Hig =159 016 11 —3571-357 +060[-08 153098157 —115 FSO 8=920 (53 1H) 1360 _F=0 “Forces in metric tons, lengths in meters, hence stresses in tons per sa (Eqs. 92 and 95) with the formulas for By and 4 (Eqs. IL and 112) and observing that (Ax= AT, one arrives at the following general relationships: B r By+ AT tana us) 4+ AT tang! (120) The: columns 21 and 2 e equation are used to calculate B, and 4’, in in Table 1. 15, ‘umn 23) and computin, Assuming F, id thus obtaining n,, (col 4, = Aili, (column 24), one are meter. obtains ~ Ey + XB, 185 Since the computed F is very close to the assumed value of 1.5 used in computing n,,, a recalculation of ng, and F, is not necessary. Iteration. In the preceding example the difference between Fy and F, is only 7% and the analysis could hence be terminated at this point. However, for irregular geometries and nonuniform soil properties ‘THE GENERALIZED PROCEDURE OF SLICES n one may encounter larger differences, in which case one should continue the iteration procedure as fol Tows. First one obtains AE, = B\—A,/F;, and then Ey by summation (columns 25 and 26). The second iteration is based on 7, which is calculated on the basis of E, (columns 27 to 31). After obtaining AT one computes B. and 4; from Eqs. 119 and 120, and continues to obtain y., As, and F,, Here F:= 1.48 = F, = 1.485, and the convergence is there- fore very rapid. This is usually the case for homo: geneous conditions, and particularly when the shear surface is selected in a statically reasonable way that is, the boundary conditions and the curvature of the shear surface are compatible. Computing Shear Surface Stresses 7 and o. The stresses along the shear surface are obtained by means of Eqs. 98 and 99 as average values for each slice. The calculation can be done for each itera n, but in Table I the result is shown only after the second iteration. For example, for slice 2, where F( +tanta)A: 1.48(1 + 0.50911.0 = 20.4, one obtains for the second iteration 53.6 2.63 10.1 + (— 0.43) ~ 2.63 x 0.50 = 8.36 When these stresses are computed for all slices (columns 40 and 41) the analysis for this surface is completed, since one has now obtained the follow ing: 1. The average factor of safety along the shear surface 2. The interslice forces, E and T, on all interfaces. 3. The avera ‘rand @, along the base of each slice. Checking F,, The average factor of safety F, along the interfaces ean be computed by Eq. 110 for race between homogeneous conditions. For the slice 2 and 3: 5.2m, E=27.2t/m 2x04 10.8 8.851/m, c= 1.01/m*? tan 6! = 0.67 hence which is larger than F. A more detailed analysis of P, for this shear surface, using seven slices, has disclosed that F, ~ F near the middle of slice 2 in Fig. 22; otherwise F, > F. (See F, diagram in Fig, 23. Plotting of Results. Figure 22 contains diagrams of imterslice forces and shear surface stresses plotted versus horizontal distance. The values of E and T are plotted for the interfaces, while o and 7 are plotted at the middle of the slices. In the £ diagram the horizontal water force Us is also included so that £’ ean be obtained. In the 7 diagram the pore pressure w along the shear sur- face is included so that a" can be obtained. For the o diagram the boundary values at points @ and 6 are calculated by means of Eqs. 69 and 71 as follows. Since tan, = 0.67/1.48 = 0.452, one obtains N= 2.41 (Eq. 65); hence with cy = 1.01.48 = 0.675, ng = BETS X15 pT 0.61 (tons per square meter) and In a zone around point a there is tension down to a theoretical depth of z= 1.7 meters (Eq. 73) For the 7 diagram one can obtain the boundary values at points a and b from Eqs. 69 and 71: 2x 0.67 T+2.41 ty = Nera = 0.95 ‘These boundary values are plotted in Fig. 22. The initial values Eo, 9, 7», and 7, are shown by crosses, and the values computed from the second iteration are shown by the solid points. The convergence is seen to be very rapid. For comparison the values p= yz are shown by a dash-dotted curve in the o-diagram in Fig, 22. Along the upper part of the shear surface & < y2, whereas toward the toe o is equal to yz, or slightly greater, as expected, 40 (tons per square meter) (ons per square meter) Sliderte and Computer Galeataton. Asie One additonal hour may be needed Issn 1 Tamar wi the dea of the GPS procedure ers. It is also advantageous to ‘on to the use of comp n SLOPE STABILITY COMPUTATIONS 7 = 2 toneim® tung’ = 067 as* Fig.23. Distribution of stresses on shear surface and on sides of slices plot the values of £ and T as they are obtained in each iteration. In this way one will soon get firsthand, experience with the procedure that will he particular- ly valuable in using computer programs, The GPS procedure has been programed in ALGOL on the UNIVAC 1107 computer by Grande (1969). ‘The program is in principle built up in the same way as the sliderule calculation described above. Tabl le of the input and the final output data for a noneircular shear surface ig. 22. Seven close to that shown in used. The input data are tana, Ax, p, uy’, tand', and AQ as midslice quantities, and tan ay and hy for cach interface. (For the most general case =, 0dr, Ey, Ey, To, and Ty would also be included here.) The final output data are the overall factors of safety Fy and F, the side forces T and E on each slice, and 7 and. atthe base of each ‘The final output data are printed after « number of nding to a specified precision. For iterations corres the example shown in 1 the change in F between two iterations should not ble 2 it was required that ‘THE GENERALIZED exceed 0.001. Even with this very stringent require ment the computer stopped after three iterations, The execution time was about 1.4 seconds. ‘The stress distribution diagrams of Fig. 23 are ob- tained from the output data in Table 2 and Eqs. 103 to 110. The total stress distribution along the shear surface and along the interfaces for slice 3 are shown separately in Fig. 24 Examples ‘The generalized procedure of slices is applied here to five different profiles. Homogeneous Soil Profile. The profile shown in 22 and 23 has been analyzed on the computer to investigate the influence of (1) the T forces, (2) the location and shape of the shear surface, (3) the posi tion of the line of thrust, and (4) the number of slices. Some 30 trial surfaces were used. Five of the trial surfaces passing through points D and Tare shown Fig. 25. Here Fy and F are plotted against the survature rat Fy and F have a minimum Table 2. Computer Input and Output Data ~~ Noneircular Tan Alfa Delta x P 1480 2.00 3.000 0.950 3.00 7.100 0.610 5.00 91800 0.370 6.00 10.600 0.260 6.00 9.500 0.050 6.00 6.400 = 0.090 4.00 2.100 FO=1.380 F= 1.474 Tan alfa. HT 0.920 0.80 0.00 0.630 1.40 0.00 0.440 1.80 0.00 0.320 1.80 0.00 0.200 1.60 0.00 0.120 0.80 0.00 Executiontime 1.422 Seconds - Libr: PROCEDURE OF SLICES B value (although not sharply defined) for a specific d/l ratio, but the minimum Fy occurs for a deeper sur. face than minimum F. This circumstance is logical since Fy corresponds to T= 0, which means no shear restraint in the vertical direction, In this example the T forces increase the mini. mum factor of safety from Fy= 1.36 for T= 0, to F=147, that is, by 8%, and the dL-ratio for the critical shear surface is reduced from 0.17 to 0.13, The influence of the T forces on F, 7, 0, and E is, further illustrated in Fig. 26 for shear surface 18 of Fig. 25. Separate investigations were made for sur faces passing to the right and left of point T in Fig 25. The results indicate that F increases slowly in both directions. Moreover, when changing the loca- tion of D the minimum factor of safety increased slightly in both direetions Figure 27 shows the final result of the analysis re- garding the shape and location of the shear surface. ‘The figure illustrates the following: 1. The critical shear surface is not well defined for surface. Example u c TanFl Delta 1200 1.000 0670 0.00 2840 1.000 0.670 0.00 3920 1000 0670 0.00 4240 1.000 0670 0.00 3800 1000-0670 0.00 2560 1000 0670 0.00 0840 = 1.000 0670 0.00 T E Tau 0.000 0.000 0.801 0.268 -0691 2.753 1.503 1.821 -3724 11.512 2.336 3.647 -7768 22.915 3012 5.135 8.429 25.656 3.040 5.196 6712 21.453 2673 4,389 -2616 7.499 1614 2.059 0.000 0.000 ary of 1 MAR 67

You might also like