Beruflich Dokumente
Kultur Dokumente
Topics
DOI: 10.1002/geot.200800033
Helmut F. Schweiger
336 © 2008 Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · Geomechanik und Tunnelbau 1 (2008), Heft 5
336_344.qxd 24.09.2008 14:23 Uhr Seite 337
3.1 Short summary on classes of constitutive models Linear (nonlinear) elastic-perfectly plastic models
These models are an improvement on elastic models be-
Linear or non-linear elastic models cause they introduce a criterion limiting the strength of
Although easy to use with a small number of input param- the material but the deficiencies in representing the stress-
eters models based on elasticity do not play a significant strain behaviour for stress levels well below failure remain.
role in the analysis of geotechnical problems because soils The most widely used model in this category is the Mohr-
in general to not behave as an elastic material, probably Coulomb model (Figure 3) for drained analyses and the
with the exception of very stiff clays under low stress lev- Tresca criterion for undrained analyses (in terms of total
els. For monotonic primary loading non-linear elastic stresses) respectively. These models are available in all
models have been applied successfully in the past [2] but codes designed for geotechnical applications. Approxi-
for more complex stress paths these models are not well mations in form of Drucker-Prager and Von Mises crite-
suited. Elastic models cannot account for stress path de- ria have been used in the past for computational conve-
a)
b)
Figure 7. Graphical representation of a 3 surface kinematic
hardening model after [7]
Bild 7. Graphische Repräsentation eines Modells mit kine-
matischer Verfestigung, nach [7]
Figure 6. Graphical representation of a doubel hardening Figure 8. Graphical representation of anistropic yield sur-
model: a) p-q-space, b) principal stress space faces for structured soil after [10]
Bild 6. Graphische Repräsentation eines Modells mit dop- Bild 8. Graphische Repräsentation von anisotropen
pelter Verfestigung a) p-q-Diagramm, b) Hauptspan- Fließflächen für Boden mit Struktur, nach [10]
nungsraum
when strains are very small. Figure 9 illustrates schematical- with a small strain stiffness model which was not available
ly the decay of the shear modulus with shear strains as ob- at the time when the benchmark exercise was performed.
served in experiments [11]. In order to take this behaviour Figure 10 shows the geometry of the problem. The excava-
into account enhancements of elastic-plastic models have tion is supported by a diaphragm wall with three rows of
been presented in the literature whereas different strategies ground anchors. Excavation and groundwater lowering in-
are possible [12] [13]. Some models are enhanced by intro- side the pit has been modelled step by step. All analyses
ducing a function for a strain dependent stiffness inside the have been performed with the FE-code Plaxis V6 [15]. Re-
yield surface, others add another surface inside the yield sur- sults for the final excavation stage obtained from using the
face (e.g. so called bubble models). The importance of tak- Mohr Coulomb model (MC) and the so-called Hardening
ing into account small strain stiffness effects when solving Soil model (HS), which belongs to the class of isotropic
practical problems has long been recognized but only more double hardening models as described above, are com-
recently commercial codes provide these models. It can be pared. In addition the Hardening Soil Small model (HSS)
anticipated that the use of small strain stiffness models will [12], which is the extension to account for small strain
be common practice in near future. stiffness effects, and a hypoplastic model (HP) [22] have
been used. Figure 11 compares lateral displacements of
4 Effect of constitutive model on results of boundary value
problems
this cannot be answered in a general way because it de- portance because it will govern internal forces in wall and
pends on the particular situation. Eventually the question struts. Numerical analyses have certainly the advantage of
has to be seen in connection with the purpose of a numer- taking into account soil-structure interaction in a more ra-
ical analysis for a particular problem. In the following an tional way than in conventional analysis, e.g. the effect of
attempt is made to give a rough guidance on what is need- the relative stiffness differences between soil and support
ed. system is automatically considered. However, one has to
acknowledge that experience is not (yet) as comprehen-
Ultimate limit state analysis sive as with conventional calculation methods, an excep-
Leaving aside the discussion whether or not numerical tion being possibly shallow tunnelling.
methods based on continuum mechanics should be used
for evaluating failure mechanisms and factors of safety at Serviceability analysis
all, it can be suggested that application of failure criteria, Assessing displacements, stresses and internal forces un-
such as the Mohr Coulomb criterion, is sufficient for this der working loads for a complex geotechnical problem is a
purpose [19]. Thus ULS-analysis is not demanding from a task which requires advanced skills in soil mechanics and
constitutive modelling point of view but numerically it is, computational methods. The accuracy which can be
because convergence criteria, element type and discretisa- achieved depends to a large extend on the constitutive
tion will play an important role in this type of analysis model, which has been shown, but also on a good knowl-
[20]. If failure modes in the post peak range have to be in- edge of ground conditions and material parameters for the
vestigated conventional finite element formulations can- relevant soil layers. In general this is the case for near sur-
not be used and special techniques or other methods have face structures, in particular in urban areas where ground
to be employed [9]. conditions are usually well defined and reliable material
Whether numerical analyses should be used for de- parameters are available. In these cases numerical meth-
sign is currently under wide discussion, in particular with ods with appropriate constitutive models will provide a
reference to Eurocode7. Comparison of numerical and good estimate of the behaviour of the structure and thus
conventional calculations for deep excavations have are a very useful tool for defining support and other auxil-
shown that significant differences in bending moments iary measures. The often heard argument that parameters
may be obtained and thus the method of analysis will have for advanced models are not available form geotechnical
a strong influence on the design of the support system reports and are therefore not applicable is not relevant. It
[21]. In these cases the constitutive model is again of im- is true that advanced constitutive models require more in-
put parameters than simple failure criteria but some of [7] Baudet B.A. and Stallebrass S.E.: A constitutive model for
them can be estimated with reasonable accuracy for differ- structured clays. Geotechnique 54 (2004), pp. 269-278.
ent types of soil, provided the user has sufficient theoreti- [8] Wheeler, S.J., Näätänen A., Karstunen M. and Lojander M.:
cal background. An anisotropic elasto-plastic model for natural soft clays.
Canadian Geotechnical Journal 40 (2003), pp. 403–418.
On the contrary, if ground conditions are not well
[9] Galavi, V.: A multilaminate model for structured clay incor-
known, significant inhomogeneity in soil layers is expect- porating inherent anisotropy and strain softening. Mitteilung-
ed and the mechanical behaviour is not well investigated, shefte Gruppe Geotechnik Graz, Heft 32. TU Graz, 2007.
numerical methods can only provide qualitative results. [10] Leroueil, S. and Vaughan, P.R.: The general and congru-
Nevertheless they are useful in assessing advantages and ent effects of structure in natural soils and weak rocks. Geo-
disadvantages of different construction methods or exca- technique 40 (1990), pp. 467–488.
vation sequences. [11] Atkinson, J.H. and Sallfors, G.: Experimental determina-
tion of stress-strain-time characteristics in laboratory and in
6 Summary situ tests. In Associazione Geotechnica Italiana (ed.): Proc.
10th European Conference on Soil Mechanics and Founda-
A brief summary of classes of constitutive models avail- tion Engineering, Vol. 3, pp. 915–956. Florence, Italy, 1991.
[12] Benz, T.: Small-Strain Stiffness of Soil and its Numerical
able for modelling soil behaviour has been presented and
Consequences, PhD Thesis, University of Stuttgart, 2007.
their features briefly described. It has been made clear that [13] Jardine R.J., Potts D.M., Fourie A.B. and Burland J.B.:
it is highly problem dependent whether a simple model is Studies of the influence of non-linear stress-strain character-
sufficient or a more advanced model is required. As a gen- istics in soil-structure interaction. Geotechnique 36 (1986),
eral rule elastic-perfectly plastic models suffice for evalu- pp. 377–396.
ating failure mechanisms but for calculating deformations [14] Schweiger, H.F.: Musterlösung und Parameterstudie für
at stress levels well below failure these models are not suit- dreifach verankerte Baugrube – Anhang zu Empfehlungen
able, at least not for problems with complex stress paths Nr. 3 „Baugruben“ des Arbeitskreises AK 1.6 „Numerik in der
such as excavation problems. If appropriate models and Geotechnik“. Geotechnik 25 (2002), pp. 101–109.
parameters are used, numerical methods are an estab- [15] Brinkgreve, R.B.J., Broere, W. and Waterman, D.: Plaxis,
Finite element code for soil and rock analyses, users manual.
lished, reliable tool and in many cases the only way to as-
Delft, 2006.
sess deformations and stresses under working load condi-
[16] Scharinger, F.: A multilaminate model for soil incorporat-
tions. With respect to determining failure mechanisms nu- ing small strain stiffness. Mitteilungshefte Gruppe Geotech-
merical methods also play their role, with the significant nik Graz, Heft 31. TU Graz, 2007
advantage that the shape of the failure surface is not a pri- [17] Schweiger, H.F., Kofler, M. and Schuller, H.: Some recent
ori assumed as in most conventional analyses, but is a re- developments in the finite element analysis of shallow tun-
sult of the calculation. nels. Felsbau 17 (1999), pp. 426–431.
The frequently asked question whether numerical [18] Addenbrooke, T.I., Potts, D.M. and Puzrin, A.M.: The in-
analysis with advanced constitutive models leads to a fluence of pre-failure soil stiffness on the numerical analysis
more economic and/or safer design is difficult to answer of tunnel construction. Geotechnique 47 (1997), pp. 693–712.
because commonly adopted factors of safety are implicitly [19] Schubert, P. and Schweiger, H.F.: Zur Standsicherheit der
Ortsbrust in Lockerböden. In W. Schubert (ed.): Proc. ISRM
linked to traditional calculation models and the true fac-
Regional Symposium EUROCK 2004 & 53rd Geomechanics
tor of safety is generally not known. However, it is gener-
Colloquy, pp. 99–104. Essen: Glückauf, 2004.
ally accepted that numerical methods with advanced ma- [20] Schweiger, H.F.: Standsicherheitsnachweise für Böschun-
terial models (when properly used) lead to a much better gen und Baugruben mittels FE-Methode durch Abminderung
understanding of the mechanical behaviour of complex der Scherfestigkeit. In Schanz (ed.): Workshop: Nachweise
geotechnical structures and as such facilitate the design für Böschungen und Baugruben mit numerischen Methoden.
procedure and risk assessment significantly. pp. 19–36. Schriftenreihe Geotechnik Heft 11. Bauhaus-Uni-
versität Weimar, 2003.
[21] Schweiger, H.F.: FE-Berechnung von Baugruben mit den
References Nachweisverfahren des EC7. In Grabe (ed): Tagungsband
zum Workshop Bemessen mit Finite-Elemente-Methoden.
[1] Kolymbas, D.: Eine konstitutive Theorie für Böden und an- pp. 17–25. TU-Hamburg Harburg, 2007.
dere körnige Stoffe. Publ. Series of Institut für Boden- [22] von Wolffersdorff, P.-A.: Ausgewählte Probleme bei der
mechanik und Felsmechanik der Universität Fridericiana in Berechnung von Stützkonstruktionen mit der Methode der
Karlsruhe, Vol. 109, 1988. Finiten Elemente. In Brinkgreve, Schad, Schweiger & Wil-
[2] Duncan, J.M. and Chang, C.-Y.: Nonlinear analysis of stress land (eds.): Proc. Symp. Geotechnical Innovations. pp.
and strain in soils. J. Soil Mechanics and Foundations Divi- 679–699. Essen: Glückauf, 2004.
sion, 1970, pp. 1629–1653.
[3] Griffiths, D.V.: Failure criteria interpretation based on
Mohr-Coulomb friction. J. Geotech. Enging. 116 (1990), pp.
986–999. Ao. Univ.-Prof. Helmut F. Schweiger
[4] Roscoe, K.H. and Burland, J.B.: On the generalized stress- Technische Universität Graz
strain behaviour of “wet” clay. IN: Engineering Plasticity, pp. Institut für Bodenmechanik und Grundbau
535–609. Cambridge: Univ. Press, 1968. Arbeitsgruppe Numerische Geotechnik
[5] Yu, H.S.: CASM: A unified state parameter model for clay and Rechbauerstraße 12
sand. Int. J. Num. Analyt. Meth. Geom. 22 (1998), pp. 621–653. A-8010 Graz
[6] Vermeer, P.A.: A double hardening model for sand. Geo- Austria
technique 28 (1978), 413–433. helmut.schweiger@tugraz.at