Montrsio. L. & Nova, R. (1997), Géotechnique 47. No. 1, 49-60 Settlements of shallow foundations on sand: geometrical effects L. MONTRASIO* ‘The paper is an extension of a previous work concerning the behaviour of model strip shallow foundations subjected to eccentric or inclined loading. The effects on the overall behaviour of the shape and the embedment of the founda- tion are considered here. The structure of the mathematical model employed to describe the behaviour of the strip foundations is recalled. The results of an extensive series of tests on model foundations of rectangular shape and various degress of embedment are then pre- sented, together with the numerical results obtained by means of the mathematical model. It is shown that experimental evidence is generally well matched by the theory. The values of the constitutive parameters of the mathematical model in the various cases con- sidered are compared. It is shown that only two ‘out of nine vary significantly with the shape of the foundation. Embedment has a small influ- ence on the value of an additional two para- meters; the other five remain constant. The results of a back-analysis on a full-scale embedded foundation under eccentric and in- clined load are discussed. KEYWORDS: bearing capacity; footings/oundations; ‘model tests; plasticity; sands; settlement and R. NOVA* Larticle est un’extension d'un travail précedent dont on analyze le comportement d'une semelle filante, soumise l'action combinée des charge- ments inclings et excentrés. Ici on considére les éffets de la forme et du niveau de base de la fondation sur son comportement. D’abord, on rappelle briefment la structure du modéle mathématique utilisé pour decrire le comporte- ‘ment des semelles filantes. On présente en suite le résultats expérimentaux obtenus dans une serie d’essais sur fondations en modéle reduit et les correspondents resuitats calculés 4 aide du modéle mathématique. On montre que la thé- orie donne des bonnes prédictions. Le valeurs des paramitres constitutifs dans les différents cas condiderés sont comparés. On montre que, entre les neufs paramétres qui characterisent le modéle, deux seulement sont influencés par la forme de la fondation, tandis que sept restent constants. Le niveau de fondation a influence sur quatre paramétres. On discute finalement le esultats d’un analyse a posteriori d'una fonda- tion réelle soumise a un chargement incliné et excentré. INTRODUCTION A simple method to evaluate settlements and rotation of rigid shallow foundations on sand under the combined action of inclined and eccentric loads has been presented by the authors (Nova & Montrasio, 1991a). The theory allows the full load~ displacement curves to be predicted with reasonable accuracy for strip foundations founded directly on the free surface of the soil. In particular, the theory correctly models the non-linearity of the load~ displacement and moment-rotation curves, the irreversibility on unloading, and the coupling of vertical and horizontal displacements and rotation. ee script received 24 January 1994; revised manuscript accepted 17 March 1995. Discussion on this paper closes 2 June 1997; for further details see p. i * Politecnico di Milano, Italy. 49 It also treats in a unifying framework both dis- placements under working loads and failure condi- tions. However, in order to use such a theory in engineering applications a step forward is neces- sary. Foundation bases are embedded a few metres below the free surface of the soil. Moreover, quite often the foundation is rectangular and plane-strain conditions cannot be assumed. In addition, some of the constitutive parameters of the mathematical model are different from those used in traditional geotechnical analysis of foundations, such as the elastic modulus of the soil or the bearing capa- city of the foundation. Little experience existed to characterize such new parameters. For these Teasons it was decided to pursue a new series of tests with embedded foundations of different rectangular shape. The -density and type of the sand used were different from those in the previous Paper. The results of such tests are presented50 MONTRASIO AND NOVA below, together with the numerical results calcu- lated by means of the mathematical model, The possible range of the values of the parameters is discussed. The theory is applied to bback-predict the experimental results of a full-scale foundation. THE MATHEMATICAL MODEL. A short description of the mathematical model is now given. See Nova & Montrasio (1991a) for a detailed presentation. The model was formulated on the basis of two basic hypotheses (a) the foundation and the soil can be considered as a macro-element for which the loadings act as generalized stress variables while the displacements and rotation of the foundation (only rigid strip foundations were considered) are the corresponding generalized strain vari- ables. (b) the constitutive law of the macro element, i.e the relationship between generalized stress and strain rates, is rigid-plastic_strain-hardening with a non-associated flow rule. Formally this can be expressed as q= Cag 0) where @ is the vector of the generalized non- dimensional stress variables, defined as f) ifr Oa) hear) Hyp (2) m MU M/pe and q is the vector of the generalized strain variables ” 2 sfe My a ) g Be where V and H are the vertical and horizontal loads respectively; M is the overturning moment; the variables v and u are the corresponding displace- ments; 0 is the foundation rotation; B is the foundation width; My, 4, are constitutive Parameters; Vy is the failure load for a foundation under centred vertical load and the parameters 1 and y govern the shape of the failure locus. The matrix C is the compliance matrix, which depends on the state of generalized stress Q, on the ‘direction’ of the stress increment dQ, and on the history of loading. It is assumed That a load- ing function f delineates between virgin loading associated with permanent plastic deformation afd unloading-reloading which is considered to occur without strain. The ‘direction’ of the strain increment dq when plastic deformation occurs is given by a plastic potential g, also dependent on the state of stress ag dg = AE a 0g The value of the plastic multiplier A depends on the intensity of the load increment and on the ‘history’ of the macro element via a coefficient known as hardening modulus K () where the tilde indicates a transposed vector. In the Plastic range the compliance matrix is therefore given by =~ 27 6 (= 25050 6 Loading function and plastic potential each depend on two constitutive parameters (f on 1 and y, g on Au and yy). The loading function depends on the history through a parameter po which in turn depends on plastic generalized strains and on four additional parameters (Ro, Vy a, 7). ‘The hardening parameter pc gives the intercept of the loading function with the V/Vjq axis. It can be interpreted as a sort of pre-consolidation stress as in the strain-hardening plasticity models such as the Cam clay model. When pc attains its limit value, pc = I, the loading function coincides with the failure locus. In this case, the intercept of the locus with the V axis is Vg The law that governs the evolution of pe with the displacements plays the role of a hardening rule of the ‘material’ and allows K to be deter- mined Ff Ope | Ox Ac Ala] | 3Q It can be shown that Rp is the initial stiffness of the foundation under centred vertical load, while Vy, is the failure load in the same conditions. The Parameters 4 and w govern the shape of the loading. function, while Aw and yy rule that of the plastic potential. Another shape parameter f is common to both, and can be taken as a constant (B = 0:95), since it seems to be the same for different types of foundations and sands. The parameters a and y control the evolution of the hardening rule. In principle, only three tests are necessary to determine the entire set of parameters: centred vertical loading, one inclined loading and one eccentric loading. The aim of the paper is to give suggestions for the possible range of values of the parameters without the need to perform such tests, It is apparent that, although very simple, such tests cannot be performed on a prototype foundation, fo)SETTLEMENTS OF SHALLOW FOUNDATIONS ON SAND) sl TEST PROGRAMME, MATERIALS AND APPARATUS. In order to study the influence of the geometric effects on the foundation, various series of tests were performed. The first was devoted to investi- gating conditions different from plane strain; the second to the study of the effects of embedment. A more sophisticated set-up than the original apparatus was used for these test series (Montrasi 1994). The upgrading of the testing facilities. was made possible by the sponsorship of ISMES, Bergamo, who put at the authors’ disposal the container used for the calibration chamber (Bellotti et al. (1985)) together with all the necessary in- strumentation for measuring displacements. The container is a metallic cylinder 1-2m in diameter and 081m in height, made of two smaller cylinders out of the three constituting the original container for the calibration chambet. The main advantage of using such a container was that the procedure for uniform sample preparation by pluviating sand was already well established (Bellotti er al., 1985), Moreover, the container width allowed several foundation tests to be per- formed with the same sample, with no apparent influence either of boundary’ conditions or of previous tests conducted in other parts of the sample surface. Theoretically identical tests were Performed; the foundation was put at different Fig. 1. Specimen and loading frame positions on the sand surface with negligible differences in results. Figure | shows a specimen after deposition and a loading frame: Fig. 2 shows the foundation and the loading ram together with the system of trans- ducers used to measure displacements and rotation, ‘The foundation has several grooves fitting with the loading blade. The load can therefore be applied with different eccentricities and inclinations, with negligible spurious overturning moments. The loading ram is a servo-controlled oleo-dynamic32 MONTRASIO AND NOVA piston by means of which a constant load rate can bbe imposed on the foundation. In order to perform tests in which the vertical load is constant but the horizontal load and/or overturning moment increase, a different loading system is used; a horizontal steel wire is connected to a vertical stick, 85mm high (not shown in Figs 1 and 2). Horizontal load and overturning moment were given by applying a dead weight to the wire which pulls the foundation via a pulley. A load cell was used to quantify the frictional loss of the pulley. Since loads are given in finite steps, the corresponding load-displacement curves have steps. ‘The sand used for this series of tests is a dense (Dz = 0:96) Ticino sand whose grain size distri- bution curve is shown in Fig. 3, after Bellotti et al. (1985). Grain size data are given in Tables 1 and 2. The difference in type of sand and density with respect to the previous series of tests will allow the general applicability of the theory to be checked. It is worth noting that the high density and the consequent high stiffness make unnecessary the Sra Sang or Demet eo H20"s | 2°06" 02 008 06s b008 0-002 109, coogi | INU L TMNIT LAME AMUNELL | HU TULLE MULE MET MITE mE LAE AME EE AMT 46 HL MME PHO MEE AMIE UN AMEE THQ NT MILT OHM MULE LTE A AML AMIEL S00" 0 Aso ° ° Grain size: mm Fig. 3. Grain size curve of Ticino sand (modified from Bellotti er al. (1985)) Table 1. Grain size distribution of Ticino sand Grain Percentage diameter: mm passing 1 100 084 997 042 129 025 14 0-149 o1 0-106 00 0074 oo Table 2. Grain size data of Ticino sand _ Du Dao: | Yein® | Ymue® | Gi mm mmm _| kW/m? | kW/m? | kNim? 029 [oar | ost | 1365 | 1668 | 2649 * Minimum and maximum dry weight were determined by the ASTM method, use of the levelling by hovering (James, 1965) widely used by Nova & Montrasio (1991a) in order to avoid the spurious effect of the lateral surcharge caused by the foundation settlement. In this case the limit load is attained for displacements of the order of $%-7% of the foundation width, values for which the lateral surcharge influence is negli- gible, as shown by Nova & Montrasio (1991a). Three rectangular aluminium foundations 8 em wide and with length-width ratio of 1, 2 and 3 were used to investigate the influence of the foundation shape. The square foundation was used also to investigate the effect of the foundation roughness, and of the embedment, For all but one series of tests, sand was glued to the foundation base to guarantee perfect roughness of the base; in fone series the smooth metallic surface of the base was clean. The load programmes followed were (a) inclined loading (with no overturning moment) from zero to failure (6) eccentric loading (with no horizontal load) from zero to failure (©) application of a horizontal load and over- turning moment at constant centred vertical load, from zero to failure (4 application of a horizontal load at constant eccentric vertical load, A total of 90 tests were performed. Some of these tests were to check the repeatability of results. In Particular, the effect of the position of the foundation, the influence of the constraint imposed on the loading ram to maintain fixed eccentricity, the influence of the type of control of the load- ing piston (in steps or servo-controlled), and the horizontal loading loss due to the pulley friction were investigated. All spurious effects had a negligible influence on test results, which were ‘consequently highly repeatable. TESTS RESULTS AND MODELLING The effects of foundation shape are considered first. Fig. 4 shows the results of central vertical loading “on surficial foundations with different length-width ratios, referred to below as aspect ratio a, together with the corresponding numerical simulations. Constitutive parameters were chosen in order to get the best fit for the square foundation data. A better fit could be easily obtained even for the two rectangular foundations. However, for the sake of limiting the number of constitutive parameters, only the failure value Vy was taken as an independent parameter, while the non-dimensional ratio Vw/BRo was’ taken as, constant, all the other parameters being the same for the three foundations considered. Fig. 5 shows that the three loaddisplacement curves of Fig. 4SETTLEMENTS OF SHALLOW FOUNDATIONS ON SAND — Experimental a + Calculated 4 Exporiment Calculated ° a 6 9 12 vem Fig. 4. Load-displacement curves for centred vertical loading: aspect ratio a= L/B 12 09 03 Fig. 5. Normalized load-displacement curves of Fig. 4 collapse approximately into one when vertical loading is normalized with respect to the calcu- lated Py. Figure 6 shows experimental data and calculated failure loci for inclined centred loading and eccentric vertical loading for the three foundations considered. The only parameter depending on foundation shape is Viw, which is the same as that used to fit data shown in Fig. 4. Figure 7 shows a comparison between calculated results and experimental data of an eccentric loading (e/B=0-125) on the three foundations. Fig. 8 shows the results of a typical inclined test (HIV = 0-25), The aspect ratio of the foundations affects the experimental results only quantitatively, the qualitative trend being independent of it. The numerical calculations fit the experimental data with the same set of parameters fairly well, with the obvious exception of Vy, for the reasons discussed. MB:KN o Fig. 6. Comparison of experimental data and caleu- lated failure loci for different foundation shapes: (a) inclined centred loading; (b) eccentric vertical loading The effect of more complex load conditions is examined next, for the square foundation. Fig. 9 shows the failure loci for eccentric horizontal load, for two different values of the vertical load. ‘Two types of tests were performed. In the first series the vertical load was increased from zero to the value shown in Fig. 9 with fixed eccentricity. The horizontal load was then applied at the foundation base until failure was attained. The failure points of this test series are fitted quite well by the assumed expression of the failure locus. In the second series of tests the vertical load was increased up to the given value. A horizontal loading was then applied at a different position of the stick in order to give horizontal loading as well as overturning moment at constant vertical load. The failure points of this loading are not well fitted by the theoretical curve, which overestimates the bearing capacity of the foundation. The reason for the discrepancy is possibly that hardening is not isotropic, as assumed, but depends on the load path followed. Fig. 10 shows, however, that the overall agreement between calculated and experi- mental curves is reasonably good. ~Another series of tests was devoted, as men- tioned above, to the study of the effect of embedment. Only the square foundation was used. Three different depth ratios, d= D/B, of the34 MONTRASIO AND NOVA — Exporimental Calculate foundation base were considered: d= 0-1, 0:25, 0-5. Inclined and eccentric loading were treated separately in this case. Figure 11 shows the experimental and calculated failure loci in a purely inclined test. For compar- ison the data discussed above relative to the foundation on the free surface are also shown. In order to fit the data with the theoretical model in this case it is necessary to vary not only Va but also the shape of the failure locus, changing the parameter 1, which increases linearly with (Fig 12), An additional parameter (the slope of the y-d line) is therefore necessary to describe the effect of the embedment. Figure 13 shows a comparison for a typical inclined loading (H/V = 0-25). It is worth mentioning that Ry varies linearly with" so that ‘BRo/Vy is constant, as are all the other parameters except of Vy and ju. * Conceptually similar results were obtained for cecentric loading. In this case it is the parameter W that varies linearly with the embedment. Figs 14-16 show the failure loci, the variation of y with d and the results of a typical eccentric loading (M/BV = 0-125) respectively Finally, the square foundation was used to investigate the foundation roughness. The failure loci of a rough and a smooth foundation in inclined and eccentric loading tests are compared in Figs 17 and 18, and the computed results are shown. As expected, the only parameter that is highly influenced by foundation roughness is s., which is 25% less in the case of the ‘smooth’ foundation. Figs 19 and 20 show the experimental data and calculated results for a typical inclined loading test and a typical eccentric loading test respectively. In all the cases considered the agree- ment between experimental evidence and theore- tical results is satisfactory. DISCUSSION The original constitutive model was character- ized by nine constitutive parameters: Vi, RoB/Va, 1H, Ys % Ay a, 7, B. It has been shown here that theSETTLEMENTS OF SHALLOW FOUNDATIONS ON SAND) 55 xa} SKN 015, Hk Fig.9. Comparison of experimental data and fai loci for eccentric horizontal load. o foundation at different vertical toi (moment increases with vertical I load is applied); (+, #) second series (moment is given by ‘the application or horizontal eccentric load. at constant vertical load) Fig. 10. Comparison of calculated and experimental Fesults for one test of the first series (e/B = -0-125 ‘and one test of the second series (MIHB = 0-275) a} Experimental Es Calculates 08 Fig. 11. Comparison of experimental data and com- uted failure loci for inclined loading paths and ‘different embedments © Experimental = Fitting sy + xo 1 Ho=048 x= 072 a Fig, 12, Variation of with the embedment foundation shape affects only Vy, the embedment affects only Vy, 4 and y, and foundation rough- ness influences only 4. Such a result is convenient for practical pur- poses, since Vy can be estimated from the friction angle of the soil via a convenient bearing capa- city formula, which naturally takes account of the shape of the foundation and the effect of the embedment. On the other hand, the observed linear variation of 4 and yi makes the choice of a convenient value relatively simple. In order to hint at the possible values of the parameters, all the cases examined so. far for different types of foundations and sands are now considered. In Montrasio & Nova (1988a,b) and Nova & Montrasio (1991a), the sand employed was a uniform silica sand from the river Adda with a mean grain size Dso = 0:2 mm. Two densities were considered, Dp = 0:2 and 0:6. The foundation was a model 100mm wide loaded in plane-strain conditions. In Nova & Montrasio (1991b) the results of a56 MONTRASIO AND NOVA — Experimental . Experimental Calculates — Fiting y= y+ 0d W203 @=03 08 06 o-. 02 ° 02 04 os a Fig. 1S. Variation of y with the embedment — rperimental Calculated Fig. 13. Comparison of experimental data and calcu- lated results for a typical inclined test (H/V = 0-25) Ast Excrimenal H Caloulated oy os 1 tS 2 858 Vek ° 3 6 a 2 °8:mm Fig. 14. Comparison of experimental data and com- uted fai ire_loci for eccentric loading path and Fig. 16. Comparison of experimental data and caleu- different embedments Inted results for a typical eccentric test (elSETTLEMENTS OF SHALLOW FOUNDATIONS ON SAND. 7 Sen a oa z zo i oot f ° Vin Fig. 17. Comparison of experimental data and com- puted failure loci for inclined loading paths and rough and smooth foundations = Rough Experimental + Smooth - Caleuiates ° os 1 15. 2 veKN Fig. 18. Comparison of experimental data and com- uted failure loci for eccentric loading paths and Fough and smooth foundations circular foundation on Hostun RF sand (Dy 01636) (Meimon, Corté & Flavigny, 1991) eccen- trically loaded in the Nantes centrifuge at 90g was analysed. The constitutive parameters were deter- mined a posteriori from the results of model foundation tests in this case and in all the aforementioned ones. The influence of stress level on intrinsic sand parameters such as friction and dilatancy angles, which are different, in general, in centrifuge and Ig tests, is therefore automatically taken into account. Table 3 presents the values of the parameters used for the numerical simulations. It is interesting to note that the parameter RoB/Viy — Experimental es = Calculated Fig. 19. Comparison of experimental data and calcu- lated results for a typical inclined test (HIV = 0-25) and rough and smooth foundations = Experimental Caleulated 12. 19h 09. 03 Fig. 20. Comparison of experimental data and caleu- lated results for a typical eccentric test (e/B = 0-125) and rough and smooth foundations seems to vary linearly with Dg. Fig. 21 shows the variation of such a parameter in all the cases considered. The experimental values given by Passalacqua, Berardi & Cabella (1991), who considered model tests of a circular foundation on Ticino sand, fit well with the line derived from the values given in the table. ‘As mentioned above, most of the parameters remain constant for Ticino sand whatever the shape or the embedment of the foundation. The para- meter 8 is constant for all the sands| considered. The type and density of sand has little influence on the other parameters. BACK-PREDICTION OF A FULL-SCALE TEST In 1990 the Institut fir Bodenmechanik, Fels- mechanik und Grundbau of the University of Karlsruhe organized a symposium in which a58 MONTRASIO AND NOVA Table 3. Values of the constitutive parameters in several cases Sand Pm [wmm) a Toa [We [Robi] « | w | A |x )@ py De kN ‘Adda 02s | oo | o | ps* ii [9 [oae [oss | 25 | 3 | 28) in| oss ‘Adda 06 | too) oO | Ps | 17 | 10 | 048 }o33 | 25 | 3 | 283 | 171 | o9s Addat ois | 100 | 0 ps | o9 | 66 | — | — | | -|- ‘Addat O18 | 10 fozs | ps] ras} sf — | — | — | —) 2} S| o ‘Addat O18 | oo fos | ps | is} s | —} —} —} —} DPT] st Hosunt | 06 | 566 | 0 1 535] isa} — | — | —) — |] Ll] Ticino 096 | 8 | 0 1 16 | 30° | o48 |o3s | 25 | 35 | 283 | 71 | 095 ino 096 | 8/0 | 2 | 4 | 30 | o4s foas | 25 | 35 | 283 | 171 | o9s Ticino 096 | 80 | 0 3 | st] 30 | o48 Joss | 25 | 35 | 283 | 171 | o9s Ticino 096 | 30 | 0 it] 16 | 30 | oas|o3s | 25 | 35 | 283 | 171 | 09s Ticino 096 | 80 | or | 1 18 | 30 | 054 ]o38 | 25 | 35 | 283 | 1-71 | 09s Ticino 096 | so | o2s| 1 225) 30 | 066 |o425] 25 | 35 | 283 | 171 | o9s Ticino 096 | 80 | os | 1 28 | 30 | os }os | 25 | 35 | 283 | 171 | 093 Ticinot | 045 | 250 | 0 os ee ee Ticinot | O45 | 500 | 0 fae fis} — | — |} —} —] 2} 2] Ticinos | 0-7 | 250 | 0 i yi27 | a6) — | —} —} — | 2} 2] = Ticinot | 07 | Soo | 0 18 [so | 3s | —}| —}|—-}—}]—-]}—]— * Plane-strain condition, {Only vertical loading test, Smooth foundation, § Cireular foundation. = Experimental = Interpolation R,8/Vy = 300, 2, Fig. 21. Variation of RoB/Vy with relative density number of participants were asked to predict the behaviour of a foundation loaded eccentrically with an assigned inclination (Fig. 22). Predictions in general were not very good, including those of the authors. It is possible, however. to try to interpret the experimental data in the light of the findings presented in this paper. ‘The most relevant parameter to be estimated is Vis as shown above, From the experimental data available (triaxial compression tests, cone pressure- meter tests and dynamic penetration tests) it was possible to estimate ' = 44°, From Vesic’s (1975) formula for bearing capacity, the limit vertical load is 2200 kN, 080m ls le— n= 1:00m Fig.22. Karlsruhe foundation and loading scheme (after von Wolffersdorff, 1991) Ro was chosen to correspond to the appropriate relative density (0-66) in Fig. 21. All the other Parameters were determined on the basis of theSETTLEMENTS OF SHALLOW FOUNDATIONS ON SAND) 39 1200. 200 200 z 2 coo & 100 300- = °! Thboow Fa Ee ‘mm 36: 500) 120 400 ” = ovo. g = 200; so +00 » ° ° 0% 4 & 6 9 2 40 6 o 120) :” Eco 20. -0"0 10 30 wo 96: mm Fig. 23. Comparison of experimental dat lated results for the foundation of Fig. 22 and calcu- data in Table I. The back-calculated results are compared with experimental data in Fig. 23. The assumed values of constitutive parameters are given in Table 4. CONCLUSIONS The aim of the research was to determine which modifications should be made to the method proposed by the authors for the settlement predic- tion of shallow foundations on sand in order to take account of geometrical effects such as the shape of the foundation or the embedment of the base. A number of experimental tests were per- formed on model foundations, varying the aspect ratio and the depth-width ratio. The effect of foundation roughness was also investigated. The basic conclusions of this study are as follows. (a) By using the method proposed by the authors it is possible to back-calculate with sufficient accuracy the observed behaviour of founda- tions under eccentric and inclined loading, intespective of foundation shape and the depth Table 4. Constitutive parameters for Karlsruhe test back-analysis «|vlaqvTxTelr[e [oss [25 [3s [283 [a7 | oss of the embedment, provided some dependence of the model parameters on geometrical effects is taken into account. (6) The only parameter that is macroscopically affected is the failure value under centred loading ¥y4 Such a parameter can be estimated by means of the usual geotechnical formulae for bearing capacity. The initial stiffness Ry is affected as well, but the ratio RoB/Viq is approximately constant and appears to. be influenced mainly by relative density. (6) The parameters 4 and y, which govern the shape of the yield locus, are not affected by foundation shape while varying linearly with the embedment-foundation width ratio. (4) The parameter jis also affected by foundation Toughness. A ‘smooth’ steel foundation is characterized by a value of = 0:35, which is as much as 73% of the value appropriate for a rough foundation with sand glued to the base of the foundation. (e) The parameters y, 4, a, y are affected by neither aspect ratio “nor embedment ratio. Moreover, in all the cases considered so far they have been taken as constant irrespective of the type and density of sand. (/) The parameter B= 0-95 seems to be appro priate for any type of silica sand at any density. An example of a back-calculation of the behaviour of actual foundation under inclined and eccentric loading has been presented. Agreement with measured data appears to be reasonably satisfactory. ‘The application of the method to the prediction of actual foundations is therefore possible, in principle at least. However, in order to apply with confidence the method proposed to actual design problems, a further step is necessary. A series of back-analyses of instrumented foundations should be performed and correlations obtained in ordinary tests in situ and in the laboratory should be established between traditional soil constants and the parameters used in this theory. This task is very difficult, since itis rare to have such complete information on the soil and the foundation movements. One special case was studied (Nova & Montrasio, 1995). As more case histories become available, the comparison of pre dicted and observed behaviour will allow further re- finement of parameter evaluation and increasingly reliable predictions. ACKNOWLEDGEMENTS The research was partly funded by the Ministry of University, Research and Technology of the Re- Public of Italy. The sponsorship of ISMES, which ut at the authors” disposal many of its laboratory facilities, is also gratefully acknowledged, together60 MONTRASIO AND NOVA with the assistance of ISMES technical personnel. Special acknowledgement is due to Miss Laura Norbis for her valuable collaboration in performing some of the tests NOTATION, aspect ratio L/B foundation width embedment D/B depth of foundation base ‘mean diameter eccentricity of vertical load yield function Plastic potential normalized horizontal load horizontal load hardening modulus foundation length normalized overturning moment overturning moment vector of generalized strain variables vector of generalized stress variables initial stiffness of the load-displacement curve under centred vertical loading horizontal displacement vertical displacement vertical load. failure value of V under centred vertical loading hardening parameter shape factor of f and g hardening parameter normalized horizontal displacement ‘normalized rotation ‘normalized vertical displacement foundation rotation wing, plastic multiplier shape parameter for shape parameter for g normalized vertical load intercept of f on & axis friction angle vee shape parameter for f shape parameter for g P se aneP ones Pos tans foundation roughness Senedd s Preosmo us wee REFERENCES Bellotti. R.. Crippa, V.. Morabito, P, Pedroni, S., Baldi, G., Fret. C., ‘Ostricat, D, Ghionna, V,,Jamiol- kowski, M. & Pasqualini, "E. (1985). Laboratory validation of in situ tests. AGI Jubilee Volume, Proa 11th Int. Conf. Soil Mech.. San Francisco, 251-265, James, R. G. (1965) Siress and strain fields in sand. PhD thesis, University of Cambridge. Meimon, ¥., Corté, J. F & Flavigny, E, (1991). Validation ‘de modéles pour le calcul de fondations superficelles, Proc. 10th Eur, Conf. Soil Mech. Florence 1, 143 146, Montrasio. L. (1994). Un metodo per il calcolo degli Spostamenti di fondazioni su sabbia soggette a carichi ‘eccentrici e inclinati. Tesi di Dottorato, Politecnico di Milano, Montrasio. L. & Nova, R. (1988). Assestamenti di una fondazione modello sotto carico inelinato: risultati sperimentali e modellazione matematica, Riv. [tal Geotec. 22, 1, 35-49, Montrasio. L. & Nova, R. (19886). Un metodo di calcolo degli assestamenti di fondazioni superfcili sottoposte alla azione di carichi eccentrici ed inclinati. Proceed. ings of Monselice CNR National Meeting, Vol. 2, pp. 141-152, Nova, R. & Montrasio, L. (1991a), Settlements of shallow foundations on sand. Géotechnique 42, No. 2, 243- 256. Nova, R. & Montrasio, L. (1991). Analysis of settle ‘ments of shallow foundations on sand. Proc. 10th Eur. Conf. Soil Mech. Florence, 505-509, Nova, R. d& Montrasio, L. (1995). Un'analisi di stablita del Campanile di Pisa’. Rix. Ital. Geotee., 29, 2, 39-93, Passalacqua. R.. Berardi, R. & Cabella, E. (1991). Experimental modelling of soil foundation interaction, compared with numerical analysis. Proc. Sth Int Conf, Numer. Methods Exp. Meas., CMEM/91, 675— 686, Vesic. A. S. (1975). Bearing capacity of shallow foundations. Foundations engineering Handbook (ed, 4H. F Winterkom and H. Y. Fang), pp. 121-147. New York: Van Nostrand Reinhold Von Wolffersdorff, PA. (1991), Probebelastung zur Baugrundtagung 1990: Versuchsergebnisse und Aus- Wertung des Prognosewettbewerbes, Geotechnik 14. No. 1, 16-21