Application Constitutive Model Swelling Rock
Application Constitutive Model Swelling Rock
Application Constitutive Model Swelling Rock
ABSTRACT: Swelling due to chemical processes is a well-known problem in tunnelling in anhydritic rocks and certain types of claystone.
If the swelling rock mass is exposed to water due to tunnel excavation or natural water influx, in anhydritic rocks large deformations of more
than 1 m can be observed, which are typically concentrated at the tunnel invert. Estimating swelling deformations and swelling pressures is
of paramount importance for the design of durable underground structures in such materials. This paper presents the results of a numerical
back analysis of measured swelling deformations with a simple constitutive model, using swelling parameters derived from laboratory
swelling tests.
2. SWELLING PROCESSES
2.1 Clay swelling
Swelling of clays comprises two different mechanisms:
Innercrystalline swelling and osmotic swelling.
Innercrystalline swelling is related to the incorporation of water
within the clay particles and requires a very large pressure (up to
400 MPa in the case of Montmorillonite) to be suppressed. As such
high pressures are usually not provided by the natural environment,
innercrystalline swelling has already taken place in most natural
clays and does not contribute to swelling deformations after tunnel Figure 2 Swelling test results of bentonite-claystone mixtures
excavation. (Wang et al. 2012)
The more relevant swelling mechanism from a practical point of
view is osmotic swelling, which is caused by differences in cation 2.2 Anhydrite swelling
concentration in the clay matrix and in the free pore water (Figure
1). After innercrystalline swelling has taken place, sodium cations Swelling of rock containing anhydrite like the gypsum keuper
(surrounded by water molecules) tend to align at the surface of the formation in south-western Germany is the result of the chemical
clay particles. A electric double layer is formed, which consists of reaction of unleached anhydrite (CaSO4) with water to gypsum
the negatively charged aluminosilicate layers at the centre and a
Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.4 December 2012 ISSN 0046-5828
(CaSO4·H2O), Eq. 1. The specific volume of gypsum is about 60% The swelling rate is determined by the parameter q, which gives the
higher than that of the unleached anhydrite. time to reach the final swelling strain based on the initial inclination
of the time swelling curve (Figure 4).
Ca 2 SO 24 2H 2 O CaSO 4 2H 2 O (1)
iq t iq t 1 e t /q t
(3)
The gypsum is initially dissolved in the pore water and only starts to
precipitate once the gypsum saturation concentration is exceeded.
The growth of gypsum crystals in cracks and macropores further iq t
i
q t
iq t (4)
contributes to the volume increase resulting from the chemical q t
reaction itself. Due to the high solubility of gypsum, swelling can
occur in parts of the rock mass which are different from the area In incremental form, the swelling strain increment in the current
where the anhydrite was initially dissolved. time step is calculated from the information provided at the
The applicability of Grob’s swelling law to anhydrite is highly beginning of the step:
debated. Due to the slow evolution of swelling deformations in
anhydritic rock, there are no experimental results available for final
swelling deformation and swelling pressures (Anagnostou et al. iq t t iq t
i
q t
iq t
t (5)
2010). Extrapolation of laboratory test results seems to indicate, that q t
swelling deformations remain high and almost constant up to a
stress level of ~4.0 MPa with a sharp subsequent drop to almost zero
(Pimentel 2007). This is in contrast to field observations in the
Freudenstein test gallery, where invert heave was significantly
reduced by increasing the support pressure with prestressed anchors,
even though no final swelling state has been reached yet. Most
constitutive models for swelling rock employ Grob’s swelling law
also for anhydritic rock (Wittke & Wittke 2005).
kqi is the (axial) swelling parameter, i is the axial stress and q0i is Positive volumetric strains (loosening of the material) result in faster
the maximum swelling stress in that direction. The swelling curve is approach of the final swelling strain, while negative volumetric
limited at c = -10 kPa to avoid excessive swelling strains at low or strains delay or may even stop the evolution of the swelling strains.
tensile stresses. Swelling strains are calculated in the coordinate This approach accounts for the dependency of the swelling rate on
system of principle stresses without any interaction of swelling in the penetration rate of water, which changes with the permeability
the different directions. of the rock mass and the thickness of the swelling rock layer. Due to
their dependency on the thickness of the swelling rock layer, A0, Ael
and Apl usually cannot be derived from laboratory swelling tests.
ε ε el ε pl ε q (7)
Figure 3 Semi-logarithmic swelling law (Grob 1972)
Within a finite element calculation scheme, an iterative procedure
3.2 Evolution of swelling with time on stress point level is required to find the stress state which satisfies
The model is based on exponential decay of the difference between the constitutive equations for the given total strain increment. The
the current swelling strain and the final swelling strain (Kiehl 1990). model employs an approach proposed by Heidkamp & Katz (2002),
Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.4 December 2012 ISSN 0046-5828
which is based on an implicit backward-Euler-scheme with deformation (swell pressure test) or no swelling pressure (free
integrated line search. swelling test).
Czurda & Ginther (1983) distinguished between samples of
4. PFAENDERTUNNEL CASE STUDY undisturbed molasse marl (series A, Figure 6), retrieved outside the
fault zones, and the fault zone material (series B, Figure 7). Series A
4.1 Project description and geological overview samples showed higher maximum swelling potential, but lower
The 6.7 km long first tube of the Pfaendertunnel near Bregenz maximum swelling pressures than the samples of series B. This
(Austria) was constructed in 1976-1980 according to the principles difference was attributed to relaxation and swelling of the series B
of the New Austrian Tunnelling method (NATM). While top samples before the samples could be tested. Due to fissuring and
heading and bench excavation were carried out without major higher water content, evolution of swelling strains may have taken
difficulties, significant invert heave of up to 30 cm was observed place faster in the series B samples, which resulted in part of the
after about 75% of the tunnel length was excavated. These swelling to occur before testing. Consequently, the in situ swelling
observations lead to detailed laboratory investigations of the potential of the series B material may be higher than tested and in
swelling characteristics of the Pfaenderstock material, an extensive the range of the undisturbed series A samples.
monitoring program and to the installation of additional anchors in It is also worth noting, that Czurda & Ginther (1983) could only
the tunnel invert. apply axial stresses of up to 300 kPa, which were not sufficient to
suppress swelling completely. The maximum swelling pressures
were therefore extrapolated from tests at lower stress levels,
assuming the validity of Grob’s swelling law. As this procedure lead
to very high maximum swelling pressures for the series B samples
(>10 000 kPa), Czurda & Ginther (1983) speculated that the q – yy
- curve may show a kink at higher stresses, resulting in maximum
swelling stresses similar to the series A samples.
Figure 5 Pfaendertunnel cross section 1st tube (after John & Pilser
2011)
The Pfaenderstock consists of various sedimentary molasse rocks Figure 6 Swelling test results, series A after Czurda & Ginther
(sandstone, conglomerate, claystone, marl), which were deposited in (1983)
the area north of the Alps between the early Oligocene and the older
Pliocene. The rock mass is characterised by significant
interstratification, with layers varying in thickness and in general
dipping into the longitudinal tunnel direction. The maximum
overburden is 350 m. Minor water inflow was observed in the areas
close to the tunnel portals, while the central part of the tunnel was
essentially dry during excavation.
and outside the fault zone area. The upper boundary of their test Table 2 Swelling parameters
results roughly coincides with the mean of the series A samples
reported by Czurda & Ginther (1983). The samples with significant Parameter Set 1a Set 1b Set 2a Set 2b
swelling also exhibited approximately isotropic behaviour, i.e. both Swelling potential
3.0 3.0 0.75 0.75
the maximum swelling stresses and the swelling potentials were kq [%]
almost independent of sample orientation. Results shown in Figure 8 Max. swelling
1000 1500 4000 4000
are mean values, averaged over the three axial directions. Weiss et stress q0 [kPa]
al. (1980) also analysed the mineral content of the samples and A0 5.0e-3 2.5e-3 3.0e-3 0.0
found a strong correlation of maximum swelling stress and Ael 0.0 0.0 0.0 9.0
Montmorillonite content. Apl 0.0 0.0 0.0 0.0
For the back analysis of the field measurements two swelling
parameter sets are considered, which represent the upper and lower
boundary of the test results of Czurda & Ginther (1983). The time
swelling parameters A0, Ael and Apl were calibrated to match the in
situ time-swelling curve.
with swelling parameter set 1b. The parameters of the two layers at
the surface were kept constant.
Increasing the rock stiffness resulted in a slight reduction of final
swelling deformation and a marked increase of swelling pressure at
the tunnel invert (Figure 15). In the present case study, the increase
of support stiffness at the tunnel crown obviously has more
influence than the higher swelling pressure at the tunnel invert. The
overall influence of rock stiffness on swelling deformations is small.
6. REFERENCES
Anagnostou, G., Pimentel, E., and Serafeimidis, K. (2010)
“Swelling of sulphatic claystones – some fundamental
questions and their practical relevance”, Geomechanics and
Tunnelling, 3 (5), pp. 567-572.
Czurda, K. A., and Ginther, G. (1983) “Quellverhalten der
Molassemergel im Pfänderstock bei Bregenz, Österreich“,
Mitt. österr. geolog. Ges., 76, pp. 141-160.
Grob, H. (1972) “Schwelldruck im Belchentunnel“, Proc. Int. Symp.
für Untertagebau, Luzern, pp. 99-119.
Heidkamp, H., and Katz, C. (2002). „Soils with swelling potential -
Proposal of a final state formulation within an implicit
integration scheme and illustrative FE-calculations”, Proc. of
the 5th World Congress on Comp. Mec., Vienna, Austria.
Hering, S., and Walliser, T. (2011) “Criteria for selecting a
tunnelling method using the first and the second tube of the
Pfänder tunnel as example”, Geomechanics and Tunnelling,
4(11), pp. 527-533.
John, M. (1982) “Anwendung der neuen österreichischen
Tunnelbauweise bei quellendem Gebirge im
Pfändertunnel“ Proc. of the 31st Geomechanik Kolloquium,
Salzburg, Austria.
John, M., Marcher, T., Pilser, G., and Alber, O. (2009)
“Considerations of swelling for the 2nd bore of the
Pfändertunnel”, Proc. of the World Tunnel Congress 2009,
Budapest, Hungary, pp. 50-61.
John, M., and Pilser, G. (2011) “Criteria for selecting a tunnelling
method using the first and the second tube of the Pfänder
tunnel as example”, Geomechanics and Tunnelling, 4(11), pp.
527-533.
Kiehl, J. R., (1990) „Ein dreidimensionales Quellgesetz und seine
Anwendung auf den Felshohlraumbau“, Proc. 9. Nat. Rock
Mech. Symp., Aachen 1990, pp. 185 - 207.