Está en la página 1de 12
F-2181 Available online at www sciencedirect.com COMPUTERS “3° ScienceDirect AND GEOTECHNICS Computers and Geotceaics 35 2008) 86-96 ae ‘wonvevierconoeuseompge Design implications of the vertical pseudo-static coefficient in slope analysis Robert Shukha ', Rafael Baker * Facaty of Cle an Enviroment Engineering, Teion ~LLT. Teeinion Clty, Hlfa 32000, tack Received 16 Febrauy 2006; received in revise for 22 January 2007; accepted 24 January 2007 ‘Available one 9 Mae 2007 Abstract ‘Most design manuals and codes recommend a simple preudo-static (PS) approach for analysis of slope stability under seismic con- ditions. Exceptions to this recommendation include problems involving liquefaction, and exceptionally large earthquakes. These extreme situations are not considered in th present work. Most codes neglect the vertical component of the PS force. The rational basis for this practic has not been propery established inthe professional literature, andthe main purpose ofthe present work s to check the validity Of this assumption. The analysis based on the realization that the direction (up or dow), ofthe vertical PS force of a future earthquake tannot be predicted, so the designer must consider the direction Fading to a minimum safety factor, It is shown that under most con- Gitions the vertical PS foree has significant effect on slope stability and neglecting tis force leads to overestimation of safety factors. An ‘analytical solution ofthe PS stability problem in cohesionless material shows that in this case the critical eonitions are realized when the vertical PS Foree acts opposite to the direction of gravity. In such materials the errors introduced by neglecting the vertical peudo-static forees may be considered acceptable. A strength reduction numerical procedure was used to solve PS stability problems ina frictionsl- coliesive material, These solutions show that for steep slopes the critical vertical PS foree acts in the direction of gravity, but for small ‘slope inclination the eiticl direction ofthis force is upwards. The errors in safety factors introduced by the common practice of neglect ing velical PS forces are proportional tothe absolute valve ofthe ratio between vertical and horizontal pseudo-static coeficients. These errors are relatively small for intermediate slope inclinations, but they become unacceptable for very steep and very gentle slop. (© 2007 Elsevier Ltd, All rights reserved. “Keywords: Pseudo-sate slope stability, Strength reduction techniques; Vera! component of pseudo-sttic ores; Design implications 1. Introduction ‘The state of the art of geotechnical earthquake analysis of slopes is now more advanced than the simple pseudo static (PS) approach recommended for conventional pro- jects by most design guidelines and codes (e.g, USACE, [1 Burocode 8 (2), BSI [3], FHWA [4], and many other). Unfortunately however the advantage of advanced state of the art technique is to a certain extent illusory, since these technique require reliable constitutive models for Corresponding author. Tel: +972 4329232; fax +972 4 $292322, ‘Ena addeatsbakerecknion sci (R. Bake) * Currently Geotechnial engineer, GYA Soll and Foundation Engi acing Li (266.352XIS- sx front matter © 2007 Ehevier Li A rights reserved. 10.1016 compgeo. 2007.01.00 repetitive loading and large number of parameters, which ‘under usual conditions are not available to the designer. Consequently such advanced techniques are justified only in major projects or extremely large earthquakes, and the ‘vast majority of geotechnical earthquake designs are still based on the PS approach, The codes recommendations are restricted to situations for which liquefaction is not an issue, and the present work is limited to such situation, ‘The basic limitation of PS analysis of slope stability under seismic conditions, are well known, and there is no need to claborate upon this subject here. Most codes consider only the horizontal component of the PS force; however the rational for this practice is not properly documented in the professional Titerarure. The first author dealing with this subject was probably Sarma R Shutha, R Baker | Computers and Getelnles 35 (2008) 86-96 " [5] who concluded that the vertical PS force is not impor- tant enough to be considered in the design of earth dams. It should be realized however that all the derivations and umerical results in Sarma’s paper deals with ¢= 0 materi- als and planar slip surfaces. Conclusions based on these results do not mean that ina general ¢~@ material the vertical PS force can be neglected. Moreover, Sarma (5] did not used one of the well established methods (ie. lim- iting equilibrium (LE) or Strength Reduction (SR) proce dure) for evaluation of safety factors; using instead ‘Newmark’s model (6) of a block sliding on an inclined sur- face. It is noted that Newmarks model was not meant to be used for evaluation of safety factors. When this model is applied to ¢ ~ ¢ materials it ignores the geometry of eriti- cal slip surfaces, yielding therefore wrong safety factors. It appears that Sarma realized this limitation of the Newmark mode! for evaluation of safety factors, and therefore restricted consideration to ¢=0 materials when the error associated with this model is relatively small. Consequently ‘Sarma’s results are not relevant for the present work which deals with general ¢ ~ materials, We are familiar with only two papers (7,8) dealing explicitly with PS slope stability under the combined effect of vertical and horizontal PS forves in ¢~ 6 materials. Both of these papers have certain limitations which will be discussed later, but they show that under certain condi- tions the vertical Component of the PS force may have sig nificant detrimental effect on the stability of the slope. The purpose of the present work isto study the effect of vertical S forces on slope stability analysis, and check the validity ‘of common assumption that vertical PS force may be neglected, “The basic premises of the PS approach is the assumption, that seismic effects on slope stability can be approximated by considering the static body forces (dE, 4) shown in Fig, 1. In this figure p, y and # are the slope's inclina- tion, unit weight of soil, and height of the slope, respec- tively. dW/= yaa is a the body force due to gravity where da ig 2 differential area, (4B, = hiyda, dE, = kyyda} are the horizontal and vertical PS body forces representing seismic effects, respectively, and {ky, ky} are the corre- sponding non-dimensional PS coefficients. {48,0} are the sree fd ig. 1. Defisiton sketch fr the preado-tatie slope stably problem. magnitude and direction of the resultant PS body foree respectively. dP is the total body force (resultant of d& and 4M), and o is the seismic angle that characterizes the direction of dP. ‘Usually, the horizontal PS coeficient is related to the ‘maximum expected horizontal acceleration atthe site. This, acceleration is specified by various codes based on seismic information (mainly earthquake magnitude, and epicentral distance). Baker ct al 9] reviewed recommendations of dif- ferent codes, showing that: (a) Asa rule, the magnitudes of recommended &y values fare much smaller than dyga/g WET dyna 18 the maximum horizontal acceleration at the site and g is the gravitational constant. A typical recent recom. mendation (¢.g, ITTK-GSDMA {10) is fy = | (2 Where in this case dhmas is the maximum horizontal aceeleration at the level ofthe tot of the slope. A sim- ilar, but slightly more conservative, estimate based on field performance of quay walls was presented by ‘Nozu et al. (11], This recommendation is given by f (aga)! which is based on the results in Fig, 2. By its very nature the PS approach is conser~ vative, replacing a time dependent pseudo-periodic acceleration function of limited duration ay(é,x) [12} by @ constant PS body force. The design recom- mendations with respect to the relation between dines and k, presented in various codes constitute, essen- tially, “calibration” of this approach, based on field performance (Fig. 2, is a typical example of such per- formance based calibration), and numerical analysis, Consequently, in this respect, codes recommenda tions provide an adequate basis for design. {b) Most codes recommends use of f, . consider ing only the horizontal component of the PS body force. It is noted however that USACE [1] allows oT (ty a + 3 stenc onts Pt van nao fr AC pe Acros Fig. 2, Determination of based on performance of quays without Tigvelneton (after Nozu et (11. 88 Shah, R. Baker | Comper nd Geotelnics 35 (2008) 85-55 use of a vertical pseudo-static force equal to 1/2 of the horizontal one for design of retaining structutes, Gazetas et al. [13], the Indian code ITTK-GSDMA [10] and some Japanese codes require (for certain pro- jects), y= 2/3ky. Ie noted that both of these rec- ‘ommendations felate fy to ky, implying in effect that the time histories of vertical and horizontal accclerations are in phase, Inspection of acceleration records shows that this is not always the case, but adopting such an assumption lead to conservative results, which is probably justified in view of the approximate nature of PS analysis. Ii realized that ‘the recommendations of USACR [1], and Gazetas et al, [13] are based on experience with waterfront structures rather chan slopes, and it is possible that these empirical recommendations may not be relevant to slopes, (6) Numerical studies by Ling etal. (7,8, appears to indi cate that for steep slopes and small fiction angles the vertical component ofthe PS force can have a signit- icantly adverse effect on slope stability (e.g. consider- ing Fig. 2d in [7D shows that for (f= 50°, 6 = 20°), neglecting the vertical component of the PS force resulls in approximately 100% under estimate of 'N=c/(yH), ie. 100% under estimate in the magni- tude of the cohesion required fora state of LE, Such large errors can obviously not be tolerated. Therefore the common assumption k, = O requires clarification and substantiation i itis to be used as a basis for design ‘The present work does not attempt to present an add tional recommendation for the ratio ky/ky (such recom- mendations should be based on data of the type shown in Fig, 2, which at present isnot available). Rather, it aims to contribute to a better understanding of the role of vert- cal PS forces in a slope stability context, and study the engineering and design implication of existing recommen- ations for ky/ky Most codes specify PS coefficients in terms of maximum accelerations. It is reasonable therefore to assume that k/ iy is related to the ratio dymay/dhmax- Consequently it is, instructive to inspeet the typical acceleration records shown, in Fig 3 (modified from Ling etal [8), which shows accel- eration records measured in the Kobe earthquake. It is noted that the notation {Xo, Ki) used in this figure are accelerations divided by g and scaled to Kiyx = 0.3 (these variables should not be confused with the PS coefficients introduced earlier, which do not depend on time). The fol- lowing observations are relevant with respect to Fig. 3. (a) In these records the peak vertical acceleration of this earthquake is one half of the peak horizontal one, which implies @=26.6°. It is well known however that in epicentral regions the maximum vertical accel- ration can equal, or even exceed, the horizontal value (14]. Examples of this effect can be seen in the o2 ne TS ot ben . oa ( 2 40 f+ fo 0.30 Pevennnrne MBiacctecseed een MOBEEW ee : : own it lobo ~ an | ny sono f nt doodle a Me eee ta 0 mee) Fie. 3 Aeros ord ofthe Kobe eas aloe om Lig et al. (8). ‘ 7 i at 3 fi a ht a fos a3 35 - ; dt os on 1 10 1001000 HovigotlBstanoe from the Earthquake Source Fale ‘on Fig. 4. Ratios of pak vertical to horizontal acceleration a a funtion of| stance from th earthquake source (fle Yamaguchi et al (15, fecords of the Hokkaido Tohio-Oki Earthquake, the Loma Prieta Earthquake, the Northridge Earth- quake, the Hanshin Earthquake and many other records. Fig. 4 (after Yamaguchi et al. (15), shows measured values. of dmury/dmaxn for the Western Tot- tori-Prefecture Earthquake of October the 6th, 2000. R Shia, R, Baker} Computers and Grote 35 (2008) 85-96 » ‘This figure suggests that the ratio dgaxo/dgasu May be 1 function of the distance d from the site to the epi center. This function probably depends also on the earthquake magnitude Mand the fault mechanism, but this type of seismological information is not cur- rently available. The main conclusion that can be safely drawn from Fig. 4 is that there is a large range {04-1} for the ratio Ggero/Gmuxt: It is realized that Aral I HOt necessary equal to kylky, but these ‘Wo ratios are related, e.g. adopting the recommenda- tion in (10] or [11] in both the horizontal and vertical directions, means that dnuxv/mexh =| implies that also ky/h, =, ie. the geotechnical engineer must be prepared to deal with 0 values as large as 0 = 45°. (b) In Fig. 3, the maximum vertical accelerations in the up and down directions are nearly equal, and consid- ering a future earthquake it is in principle impossible to predict which one of these directions (up or down), will be associated with a greater acceleration. In prin- ciple the same difficulty exists with respect to horizon- tal accelerations which may act into the main body of the slope, or in the outward direction. However a horizontal PS force acting into the main body of the slope has a stabilizing effect, and there is no need to consider such forces in a design process. It will be shown that the situation with respect to vertical PS orees is more complicated, The observation with respect to the direction of the vertical PS forces implies that, (at best), codes may specify only the absolute value [6 rather than 0 itself, and this limita- tion has significant engineering implications. One of the main purposes of the present work is to clarify certain issues related to the effect of the direction of the vertical PS force (up or down), on results of PS stability analysis of slopes. (6) Fig. 3 shows that the maximum vertical acceleration in the “down” direction occurs almost simulta- neously with maximum horizontal acceleration. This observation justifies considering simultaneously hori- zontal and vertical PS forces as is done in Fig. 1 However it is realized that there are many records jin which the maximum accelerations in the vertical ‘and horizontal directions occur at different times. In such cases, considering ky as a function of ky (as is done by USACE (1]}, ITK-GSDMA (10]and Gazetas, ct al, (13), must be considered merely as a conserva- tive assumption. 2, Slope stability analysis In order to focus tae presentation on the main issue (effect of vertical PS forces on slope stability), a case of homogeneous slope with zero pore water pressure is con- sidered (a framework for a general “un-drained” stability analysis was presented by Baker et al. (16)). There exist two major approaches for analysis of slope stability prob- Jems (both static and PS), namely: limiting equilibrium (LE), and strength reduction (SR) procedures (a third approach; limit analysis, has been discussed in the litera- ture, but it is rarely used in practice). The details of LE. and SR techniques are well known, and they will not be elaborated upon here, except to note that both of them are based on the same notion of safety factor with respect to shear strength which is defined as e+ otan(s) - « where {r) are, respectively, the shear and normal stresses Acting at fare onthe fature plane. (c,6) ae the conven tional strength parameters cohesion and angle of internal fection respectively and F isa trial safety factor. In the LEE framework the safety factor for a given slope, Fis ob- tained by a minimization process overall potential slip su- faces considered by the LE procedure. Appliation of the SR technique involves solution ofa series of elasto-plastic oatinuam mechanies boundary value problems for va ous trial Fralues, and the magnitude of Fr is taken as the Fvalue at which the problem doesnot have a solution {e sate of global failure). Its noted that both LE and SR ynethodssufer from the sae fundamental inconsistency of Considering faire condition of a ititious material with reduced strength, rather than the ral slope stability prob- lem (see [17] fora discussion ofthis point). SR techniques, have an advantage over conventional LE procedures which introduce various static and/or kinematic assumptions. However the introduction of the variational approach by Baker and Garber (18] removed this limitation of conven: tional LE procedures, and this approach may be consid- ‘ered equivalent to SR techniques. This equivalence is supported by many observations (eg. (19-21), that for sta- ‘tic problems and linear failure criteria, LE and SR methods yield very similar results, Shukha and Baker (22) noticed that using @ non-linear failure criterion, Flac’s strength reduction method [23], shows significant mesh sensitivity, bat using an appropriate (square elements) mesh, results again with good correspondence between LE and SR re- sults, Leshchinsky and San [24] and Baker et al. [9] appliod the variational LE approach of Baker and Garber (18] to PS problems with ky=0. Baker et al. [9] demonstrated ‘good correspondence between PS results based on the vat- iational LE approach, and SR methods. ‘The main advantages of SR procedures compared with the variational LE approach is their generality and ease of implementation. However SR methods suffer from mum- ‘ber of limitations: {a) It is noted that the theoretical LE solution for slope stability problem depends on slope geometry, strength parameters and external loads; but in order to apply a SR method itis necessary to specify many additional quantities, which have only limited effect ‘on variables of engineering significance in slope sta- bility analysis (safety factors, critical slip surfaces, ” Shuto, Baker | Compucers and Goteinies 35 (2008) 86-96, and normal stress distributions along these surfaces). In many practical situations these additional input variables are not available to the practical engineer. (b) SR techniques define the critical conditions as those fat which “the corresponding continuum mechanics problem does not have a solution”. This “opera tional” definition depends on various numerical fea- tures of the program used in the application of the SR technique. In certain cases the point at which the problem does not have a solution is not well defined (calculated displacements keep increasing sgradually without a clear limiting point). (©) Shukha and Baker (22) presented # procedure for establishing critical slip surfaces based on the results of SR methods. Under certain conditions these sur- faces are very deep and wide, requiring therefore an extremely large mesh in order to analyze such prob- lems. In some limiting conditions the required mesh js so large that it is practically impossible so solve the slope stability problem using a SR technique. Despite these limitations all the numerical results pre- sented in this work are all based on Flac’s SR procedure. ‘The solution of the slope stability problem depends on the data defining this problem, which in the PS case can ‘be written as the general relation Bo = Plc, 6,7, Hy ksh) @) It is noted that the vertical component of the PS force has a dual and opposite effect on safety factors; considering a vertical PS force acting downward as shown in Fig, [ in- creases both the driving forces which contribute to failure, but also the normal stresses on potential slip surfaces, and therefore strength and stabilizing forces. These opposite ef- fects exist also when the vertical PS force is acting in the ‘upward direction. Consequently itis impossible to establish a priori the critical direction ofthis force (up or down), and the two alternative directions of this foree have to be con- sidered in the design process. The geometry of Fig. 1 implies the following relations ke Bee (ly 0-= arctan /s) (32) @ = arctan G +) 3) is noted that positive values of & correspond to situations in which the vertical PS force is acting in the direction of gravity. Expressed in the terms of {&,0), Eq. (2) can be ‘written in the equivalent forms PS= BC, $, 7H Bashy) = PG, 79H 0) “ Eq. (4) represent the conventional slope stability format which yields a safety factor Fs as a function of all input variables (including PS coeflicients) defining a particular PS stability problem. This representation requires a sepa~ rate stability chart for each slope inclination (e.g. (24). Baker et al, [9] showed that a more efficient representation ‘can be obtained by defining a critical PS coeflicient ke as the PS coefficient associated with Fs= 1 [25], and introdue- ing an alternative seismic safety factor Fy as Fe(G) = ke(O) /&(0) (5) where £(6) in the expected value of the PS coellicient at the site, in the direction 0, and ke(0) isthe critical PS coefficient in the same direction. The following comments are relevant ‘with respect to this definition: (1) The critical PS coefficient ke does not depend on information with respect to earthquakes. For inter- pretation purposes it is convenient to consider kc(6) as the resistance of the slope to earthquake loading in a certain direction 8, (2) The term k(@) represents the ellect of the earthquake at the site, and it is independent of the slope’s geom- etry and properties, (3) Fris a safety factor with respect to shear strength, while A; is a safety factor with respect to loads. In ‘general F, # Fe and these two safety factors become equal only at failure when F,= Fy= 1 In view of the common practice of neglecting the vertical PS foree itis convenient to define the ratio Fkna(9), quan- tifying the error introduced by this assumption Fe) ___e(0)/R(8) Fel) = RG =O ~ FeO = Of =O) e(O)/K(O) _ kel) ky Oy Ca) Fig. 1 shows that ky/K(0) = cos(6), and Eq, (6) can be written in the form haa) ~ kexu( 0) 608(0) = bean )/se(0) (1) vere ona) 2 7 2) ‘The end product of PS analysis is the safety factor Fk. ‘Therefore term Fkn(0) defined in Ea. (6) i @ proper crite- sion for evaluating errors associated with the common assumption 9 =0, Nevertheless, the representation in Eq, (is also convenient, separating clearly the “relative resst- ing term’ kepyx which depends on the slope geometry and material, but not on the earthquake, from the relative earthquake “loading term” K(0/fy = see(8) which depends on the eartbquake, but is independent of the slope. The irection 0 couples these two terms. The terms (kcRat ‘elk, Pigae) represent “normalized” (with respect to the conventional assumption 9 = 0), values of resistance, loads, and safety factors respectively. This terminology will prove useful in the diseussion section. ‘Accepting the basic fact that the direction (up or down), of the vertical PS force of a future earthquake eannot be predicted (2.2. Fig. 3a), @ responsible engineer has no Shiba, R Baer | Compuers and Goteclnis 35 (2008) 86-96 3 choice but to assume that both directions are equally pos- sible, and based the design on the direction resulting with the smaller safety factor. This argument leads to the follow ing dosign relations: Fill) = Min{ FE +016), FeO) = Fh 0) FH(819 = 0) 1) where oo) = Ming haa +0)P), Fa O16) (82) and we use the convention that quantities written to the Fight of the vertical line“ are considered as “given”. Eq, (8) will play a major role when discussing engineering ‘and design implications of vertical PS forces. 3. Cohesionless materials ‘Using the variational LE framework of Baker and Gar- ber (18), and a rotational transformation which simplifies solutions of certain PS problems, Shukha et a. (26] derived the following analytical solution for the safety factor sin cobesionless materials under PS loading condition Fe= tan(6)/tan( +0) ) Considering the definition of a in Ea. (3.3) it is clear that > 0 (ley is always postive, and [iy] < ky <1). Conse- quently, for cohesionless materials, the static requirement <6 is valid also under PS conditions. Setting in Eq. (9) Fe= and solving for «o yields the Following expression for the critical seismic angle cs: tan(a,) = tan(p ~ B) (10) Using Bq, (3.3), it is possible to write Eq. (10) as key Tokay 7 Mo) ay Inspection of Fig. 1 shows that key=-ecos(6) and key =kesin(0), Substituting these relations to Eq. (11) and solving for ke yields tan(d ~ 6) SRO) = sin(OjtanlT= HY (a) Eq, (12) shows that ke{f,gl0 =0) = tan( ~ f). Combin- ing this result with Eq, (12) and the definitions (6) and (7.2) yields therefore be = ko(G, 6,9) 1 ser.) = RTE i) Fina .0) = heral 8.0) 0080) = aaa (132) Fig, Sa shows the functions Fkaa(0[B,06 = 30°) based on Bq, (13.2). The points in this figure are results of Flae’s SR calculations by Shukha [27), performed for an elasto-plas- tic model, with ( 30°, 6=0, 7=18 kN/m’, ‘= 6 10° kPa, y= 0.25, H= 5m] where d is the dilation a Pou LA ae nan Thee jean, Fig. 5. Analytical solution for the safety factor ratio Aya cobesoniess pes ~ comparison with oumrical results for (a) The fonctions Rgu(lA) end (b) the functions Fyn) angle, end (£,»} are elastic constants. The SR calculations were done for various combinations of {f,0). The good correspondence between the two solutions for Fhxa(0|B,6 = 30°) shown in Fig. Sa provides a mutual substantiation of both the theoretical solution given in Eq. (13.2) and Flac’s SR numerical process. I is noted that the points in Fig. Sa, were obtained by increasing the mag- nitude of F at constant 0 until the Flac program did not converged to a statie solution, Strictly speaking such pro- cess should be classified as a load increase rather than strength reduction technique. Fig. 5b shows the functions Fkma(fl@,@ = 30°). The results in this figure are plotted for the following 3 cases (a) kyfly = 1 (or 0= 445%), which is the maximum cred- ible value of [9| based on acceleration records in epicentral regions, reported by Ebeling and Morrison {Id}, and the data in Fig. 4. (b) &y/ky=2/3 (or 0 = 433.7%), corre sponding 0 the recommendations of the Indian code (10] and Gazetas et al. [13] and (©) kJky=I/2 (or = 426.6°) which corresponds to the recommendation of the USACE (1} Fig, 536 and Eq, (13.2) have clear and significant engi- neering implications. In particular, the results in Fig. 50 show that Fld B10<0,¢= 30°) <1 while Fkpae(6|0> 0,¢-=30°)> 1 for all legitimate 0 0) for all legitimate p<, and, as a result, the critical conditions are realized at negative 8 values. Fig. 6b on the other hand shows such behavior only at small p values, while for steep slopes Plqa(Al0> 0) < Flyu(A|0 <0), resulting in the critical conditions occurring at positive values of 8, ie. for steep Irietional-cohesive slopes the critical vertical PS force acts in the direction of gravity. The dependence of the direction of the critical vertical PS force on p is due to the ellect of cohesion and slope inclination on the sizeof the critical test body. 'A significant feature in Fig. 6b isthe fact that at small values of ff the functions Flasd(#|? <0) start at values Jess than { and increase monotonically to a value larger than 1. The opposite trend is observed for the functions Fya( 0 > 0). As a result, these functions cross each other at some f value. This erossover point defines two zones; the frst zone consists of small values of f, in which FK(6I0 <0) < F(p|8 > 0). The second zone includes steep slopes satisfying Fhqa(A0 <0) > Fkna(619 > 0). The design criterion defined in Eq, (8.2) implies therefore that in ench of these two zones the function Phra nesa( fll) consists of sections of the functions Fkny(l0 <0) and Feyy(|0>0) as shows in Fig 6. For the present conditions the erossover point frm is very close to 40" forall values of 8. However, the most striking feature in Fig. 6o is the large error associ- ated with the common practie of neglecting the vertical PS force for @=45* and vertical slopes. In this case, the assumption 0=0 results in close to 70% over estimation of the safety factor Fk(f= 90", 8=-H45"), For diffrent given @ values, the errors in safety factors associated with vertical slopes are in the range of 30-70%, and such errors, are definitly not tolerable. The corresponding ertor range for gentle slopes is smaller (20-40%). The assumption 0 = 0 results with relatively small errors (less than 10%) only in the intermediate zone of slope inclinations (inthe vicinity of the transition value frm = 40°). The characteristic shapes of the functions Fens pag) in Fig. 6 (with a maxi- mum located at fra) is a consequence of the fact that the fanctions Fkga(AI9 <0) and Flay(fl0> 0) in Fig. 6 are both monotonic, but have opposite slopes. In the vicinity of B= Brom both Fkay(fl0<0) and Fkaa(fl0 > 0) are close 10 1, so lack of information with respect to the direction (up or down), of the vertical PS force has only small eect on Fen Desig (A|6). and these values remain relatively large. “The results in Fig. 6 illustrate clearly the urgent need for codes to establish some reasonable recommendations with respect to |0|. Based on the results in Fig. 4, recommended [0| values should decrease with the distance of the site from Shah, R Baker 1 Canputers ond Gooecnics 3 (2008) 86-96 os (28) msi 4p ie Pence BRU (Pet #9 Fla) gan to a Fig. 6. Numerical results for cobesive materia (ey#)~0.13,# = 25) (a) The suet factor ratio expressed athe function F(A (bth safety factor ratio expresied us the fusetion Fadl) and (c) the design criterion Farrag BOD: the epicentral region, Without recommendations with respect to the function [6(d), the engineer has no option but to adopt the limiting vaiue [@) = 45° (which in most cases is probably not realistic, resulting (for vertical slopes), with the very conservative estimate, FK(B— 90°, = 45%) =0.337K(=90",0=0). Using the stability charts presented by Baker et al, (9] or Leshchinsky and San (24) (which are valid for 00), the relation FA(B= 90°, 6 = 45°) = 0.33FK(8 = 90",0=0) makes it possible to evaluate the safety factor Fk(B =90°, 0= 459), 5, Additional case studies ‘The results in the previous section were obtained for a sin- ale set of input parameters {¢= 11.7 kPa, 6 y= I8kN/m’, E=6x10°kPa, v= 0.25, order to extend somewhat the scope of section presents some additional case studies with different values of (c,,#}. The conditions for these case studies Tablet Parameters of case studies (Fr ~ sate safety flor with respect 10 strength) Ge Cera) Patel A r 6 =a 1 0 5 2 8B oo 170 mL 0 2 ow 265 are summarized in Table | (inthis table Fs isthe static safety factor which was evaluated using the SR procedure), and their results are presented in Table 2. In these case studies we considered only the critical cases in which 0 = 45° (i.e. k,=cbky), and did not change either the deformation parameters {5=0, E=6x10°kPa, y=0.25} (it is well ‘known that in SR methods deformation parameters affect the magnitude ofthe displacements, but they have only mar= ginal effect on safety factors), or the total unit weight. Tables I and2 present che results of 3 case studies, and we will dis ‘cuss each one of them separately 3. Case Consider first the case of $=0 materials. In such materials the vertical PS body force does not affect the distribution of strength along the critical slip surface. Consequently in =0 materials for every fi the critical ‘conditions must occur when the vertical PS force acts in the direction of gravity (ie. 0> 0). Note that this conchu- sion complement the result for c=0 materials in which for every B the critical conditions are realized when the vertical PS force acts opposite to the direction of gravity (<0), Originally Case I was intended to representa situ- ation associated with g = 0. However, attempting to solve such a case with the SR method yielded an extremely deep critical slip surface restricted by the mesh boundary, and 4 Shukla, R. Baker | Compucrs and Geotenis 35 (2008) 86-96 ‘Table? Results of casestudies Cue eld 457) ead= 9) ao) Fi I 038 020 bus 037 a 03 025 040 LB i on out 078 La such a solution is obviously not legitimate. It is noted that under static conditions the ertial slip surface for {$=0, 8=90*) is rather shallow [28], however Shukha: et al [26] demonstrated (see Fig. 3 in (26) that eritical slip sur- faces for PS conditions are deeper than the corresponding surface under static condition, and the preset set of input parameters result in very deep critical slip surface even at =908. In order to avoid use of extremely large mesh wwe had to “compromise”, and consider a small but not zero 6 value (p= 5°). Ths difficulty illustrates one of the deficiencies of SR procedures. Case I represents a situation in which there is high ratio of cohesion to friction, The cro term Fla ia Table 2 is evaluated according to Eq () ic. Fea) for 9 = 45°, and the bold value Of Fhe is Fknar baie evaluated by Eq, (8.2). The result for ‘Case T show that as expected the critical conditions occur or > 0, and Fl(f = 90", = 45") = 0.57FK{ = 90°, 0=0) i.e, neglecting the verticel PS results in approximately 40% over estimation of the safety factor. 5.2, Case IT ‘The purpose of Case II was to verify that in the interme diate range of slope inclinations neglecting the vertical PS force does not introduce large errors (see Fig. 6c) when the problem's parameters are changed from {c= 11.7 kPa, O=25, H=5m) to [e=SkPa, $=32°, H= 10m), ‘The results in Table 2 verify this proposition (in Case II the error associated with neglecting the vertical PS force) are less than 10%, 5.3, Case I ‘The purpose of Case III was to verily that for small slope inclinations the critical conditions occur for vertical PS force acting opposite to the direction of gravity when the problem's parameters are changed from {e= 11.7 kPa, =25%, H=5m} to (c= 10kPa, $= 32", H= 15m). ‘The results in Table 2 verify this proposition. It may be concluded therefore that in general the results| in Tables 1 and 2 are consistent with the results showa in Fig. 6, ie, for certain conditions, neglecting vertical PS forces may results in significant under estimation of PS safety factors 5. Diseussion For interpretation purposes it is convenient to consider the term kega(9) as a normalized resistance of the slope to an earthquake loading in the direction 0, while se(®) is considered as the normalized magnitude of this loading. ‘The ratio of these terms Fhra(8) = kera(9)/sec() (Eq. (7.1) defies the nonmalized PS safety facior which quan- fifies the error associated with the common assumption 0 =0. The motivation for this terminology was introduced in Section 2, Based on Eq, (7.1) itis clear that the results presented in Fig. 6 depend toa large extent on the behavior of functions Kena(@), and it is instructive to investigate these functions Fig. a shows the functions kogy(6f) evaluated forthe same dataset used for determination of Fig. 6. Te fllow- ing comments are relevant with respect to this figure (a) All the functions kepa(8|8) have stationary minima at certain critical directions Bay, = gn(B). These min- ima are marked in Fig. 7a by the open circles. It a renal 18) Fig. 7. Ratios of cca pseodo-static coefficients for cobesive materi (et 0.13, §= 25), (a) The resistance term Kexa(Olf) and (b) Aietion of est resistance Og = R. Sha, R. Baer Comper and Gotecais 33 (2008) 86-96 os appeass therefore that there are certain directions 8, at which the slope offers a minimam resistance (0 earthquake loading. ‘The function Gea xd) is shown in Fig. 7. Inthe absence of reliable informa- tion with respect to the actual values of O (which is the usual situation), it is “tempting” to evaluate safety factors using the critical directions Oaq(f) rather than the arbitrary assumption 9 = 0. However, the performance eriteria of PS stability analysis are the safety factors (8,0), rather than only the resis- tance term ke(f,0), and Fig. 6a shows that all the functions Flqu(9[f) are monotonic, achieving their extreme values at some given [Ol. Consequently there is no justification of using Aaq 88 basis for design. (b) Inspection of Fig. Ta shows that Kena(B10 = Our) > 0.9, forall values off. Consequently, basing a desiga ‘on ‘Gog Would imply that the common assumption 0-0 is justifed, in the sense that it yields slightly conservative results, However, the safety factor functions Flna(0[A) are monotonic (Fig. 6a), achiev- ing their extreme points at some given value of (0) Combining this observation with the realization that the direction (up or down) of the vertical PS force ‘eannot be predicted, results with the very small values of Fa: peia(All) shown in Fig. 6. We speculate that the common belief that the vertical PS Force is not important for design purposes is related to the behavior of the functions Keyadfl); ignoring the fact that ke is not a measure of stability under seismic conditions. (c) Inspection of Fig. 7 shows that the function Gage = Bon(B) 38 almost linear. Bag <0 at small values of f (gentle slopes), indicating that for such slopes the extical conditions are realized when the vertical PS force acts opposite to the Jn of gravity. Conversely, for steep slopes fy > 0, and the critical conditions occur when the PS force acts in the direc- tion of gravity, Fig. 7b shows that the transition between these two cases occurs at frie = 39° which is almost identical to the transition value rma = 40° inferred in Fig. 6¢ oa the basis of considerations based on safety factors. Consequently, while dy can not be used as a basis for design, it provides an “explanation” for the results shown in Fig. 6, Stated diferently, the term cos(@), which distinguishes between Alger and epyy has only minor effect on the direction (up or down), of the critical vertical PS forces, However, the difference between kat and keyax has a very significant effect on the value of the design exiterion (82), (@) Its noted thatthe solution of Ling etal (7), which is based on the variational LE approach, utilized the & priori assumption that the vertical PS foree acts in the direetion of gravity (ue. @> 0}. They considered cohesive-frictional soils but restrict attention to only rolatively steep slopes (9 > 45°) Fig. Ge justifies their assumption that in such slopes the critical vertical PS force acts in the direction of gravity. Ling et al (8] considered both possible directions of the vertical PS force, however their LE calculations are based fon a linear (Culmann’s type) failure mechanisms: which is not appropriate for gentle slopes. In fact they characterized their methodology as being “of sign- ‘cance for classroom instruction”. Tt is noted that for certain conditions there is almost 100% difference between the solutions presented by Ling et al. [7,8 ‘This difference is not surprising since itis well known, that for relatively flat slope there is a significance di ference between the variational solution, and Cul- ‘mann’s failure mechanism. Nevertheless, even the very approximate LE approach used by Ling et al (B), predicts the correct qualitative result that in cer~ ‘ain eases the critical conditions are realized when ver~ tical PS force acts opposite to the direction of gravity (Ge. it is indeed appropriate for classroom instruc ). The present work considers the complete range of f values, and it shows that for gentle frietional- cohesive slopes, the critical conditions occur when the vertical PS force acts upwards, while for steep slopes the opposite situation is true, The limitations of the solutions presented by Ling et al. [7,8], justify the present more complete treatment of the subject. 7. Summary and conclusions Presented is a PS analysis of slopes subjected to the com- bined effect of vertical and horizontal PS forces. The main conclusions of this analysis may be summarized as follows: (a) An analytical solution for cohesionless materials sub- stantiate numerical results based on a SR method, For cohesionless materials the critical vertical PS forces always acts opposite to the direction of gravity. ‘And the error introduced by neglecting the vertical PS foroes are usually tolerable. In frictionless materi als the critical vertical PS forces actin the direction of gravity and the error introduced by neglecting those forees may under certain conditions be significant. Inall cases neglecting vertical PS forces (inthe eriical

También podría gustarte