Structural design models
for tunnels
11, Duddeck
}. Erdmann
‘Universitat Braunschweig, Germany
‘synopsis
the ITA working group on structural design models for
‘unnelling will publish in 1982 the replies to a
‘questionnaire in synopsis form. In continuation of that
work the results of an investigation into the most common
design models for soft ground tunnels are given by way of a
‘comparative review of progress to date in this field. The
iain differences in the assumptions with regard to the
ifferent models are stated. Diagrams for the hoop forces,
bending moments and radial displacements illustrate the
differences in the design values evaluated for three different
models ~ the continuum model, the Muir Wood design
‘model and the bedded beam model without bedding at the
crown region. A comparison with free parameters
necessitates analytical solutions, so only citcular cross:
sections are investigated. Nevertheless, the results of the
investigation may also be valid for non-circular cross-
sections and more refined numerical analyses. There is a
‘end towards agreement on the basic assumptions and on,
the design model in regard to both shallow and deep
cunnels,
In the period of planning and tendering for a tunnel
structure the engineer has to rely — apart from his
experience ~ on a structural model, from which he may
derive criteria to establish whether the design is suitable,
safe and economical. One of the working groups of the
International Tunnelling Association (ITA) chose the task.
investigation
geology line and orientation
‘ground characteristics ]
prim. stress, strenath ,
fissures, anisotropy.etc. | |
probing and
[ rock mechanies
experience ‘excavation method i
estimation structural elements
mechanical __,|_[ -statical system
model 31 analysis
a
safety concept __Bi_[~ design criteria
failure hypotheses” I|_~_ prognosis.
“verification” of insitu monitoring
tunnel design _deformations stop ?
(es
~
for the actual state only, Sgfe~
[Shunsue Sttehinacge’ — “Bate
translation of reality into design model marked by]
Fig 1 Design procedure for tunneling (structural design model
includes sections marked with vecial bar)
Youk (ya
cof gathering information on design models for tunnelling
that are currently applied. A questionnaire sought data on
site investigations, assumed loads, static systems, safety
concepts, design criteria and in-situ monitoring for four
different types of tunnel structure, The replies to the
questionnaire will be published in 1982.1
From this first step information is disseminated on an
international level as to the kinds of structural models that
are available and applied. As would be expected, the replies
reflect national developments in the art of tunnelling,
traditions and experience.
‘The present paper is part of a second step intended to
compare the design models, analysing why and in what
‘ways they differ from one another. The scope of the
‘comparison is at present restricted as it deals only with
tunnelling models for soft ground and that part of the full
structural model (Fig. 1) from which criteria are drawn for
the design of the tunnel lining. The reason for the first
restriction is, simply, that for soft ground internationally
applied models are fairly well defined and more developed
than those for tunnels in hard rock. With regard to the
second restriction, the authors are well aware that safe
tunnelling involves much more than the selection of a
suitable mechanical model, but all the other parts of the
complete structural model defy translation into objective
analysis and mathematical language.
Papers that applied a much broader approach to all the
complex aspects of tunnelling, néluding different design
‘models, have been presented.*?
In many instances the strest~strain—deformation
problem of a tunnelling procedure can be solved only by
the application of a numerical analysis, ¢. the finite
element method. The geometry of the opening and the
stratigraphic layers of the ground are in most cases not
simple, Non-linear material behaviour is involved. The
consecutive phases of the process of driving and support are
‘important for the final stresses in the linings. Comparison of,
the results on the basis of numerical solutions is, however,
very difficult and does not yield any insight into
dependence on soil and design parameters.
‘The analysis given here is, therefore, additionally
restricted to solutions for circular cross-sections. Since the
main objective is directed towards the basic assumptions of
the models the conclusions are also more generally valid for
non-circular cross-sections and for numerical analyses.
‘The following questions are to be answered for the
different design models for tunnels in soft ground: Which
basic assumptions are applied to derive a model? Which are
generally agreed? Which are different? To what extent are
the design criteria (thrust, bending moments, displacements)
affected by the different assumptions?
Basic assumptions for structural design models for tunnels in
soft ground
Soft ground requires immediate support as, for example, in
driving a shield-excavated tunnel or by applying shoterete
with the shore-time closure of the full ring, Therefore,
general agreement exists on the following basic assumptions.
(1)__ For the design model of the linings it may be
sufficient to consider only a cross-section on the assumption
of plane strain conditions for the lining and the ground,
(2) The active soil pressures on the lining are taken as
‘equal to the primary stresses in the undisturbed ground
because the ground is soft. Hence, itis assumed that for the
final stage (years after construction) the ground will
eventually return to the same condition as before the
‘tunnelling, except for the passive stresses due to the
deflection of the lining. Changing groundwater levels, traffic
83Fig. 2 Plane strain continuum model
-O-%
6 =yH
Oh = Ky
K, = const.
Fig. 3 Bedded ring model: erown without bedding!”
vibration, ete,, may be the cause of this, For future
monitoring results may offer the opportunity to ascertain
‘which eases (eg, type of soil and depth of tunnel) this
B.ssxmption is too conservative. Different intermediate
situations influenced by the driving procedure and the
placing of the supporting elements are neglected,
(3) Between the lining and the ground there exists 2
Ove
On
Y= volume weight
bond either for radial and tangential deformations or for
radial deformations only. With this assumption the model
complies with the equilibrium conditions as well as with
the compatibility conditions at the boundary between the
lining and the ground. This is quite different from the
‘concept of introducing a lining resistance, expressed in
terms of a force, proposed by the Austrian school. Here
only equilibrium conditions are considered.
(4) Because of the lining—ground relationship
deformation of the lining results in reaction stresses in the
ground. A continuum model includes this effect
‘aucomatically. For a beam model bedding springs with
appropriate bedding moduli have to be applied. The bond at
every place around the lining gives rise to a reduction in the
‘Joading’ ground pressure where the lining deflects inwards,
If such a ‘load reduction’ were not allowed, bedding (and
the equivalent of this for che continuum model) has to be
neglected in those parts of the cross-section where inward
deformation occurs, ie. principally at the crown (see Figs. 3
and 5).
(5) The material behaviour of ground and lining is
‘generally assumed as being elastic. More refined theories may
also include non-linear and plastic material laws, which,
however, in most cases require the application of numerical
methods (e.g. FEM),
For more detailed discussions of the assumptions the
papers given in the list of references should be studied.
Contributions to development of design models for tunnels
‘with circular cross-sections
‘The following brief report on progress to date is inevitably
incomplete, a number of important contributions having
been omitted,
‘Until 1960 most tunnels were designed, if by analysis and
not merely by experience alone, by methods that arbitrarily
assumed ‘loads’ acting on the lining. Even allowance for
bedding forces was taken as given forees. Hence,
compatibility conditions between ground and lining were not
‘considered,
‘Schmid’ in 1926 seems to have first analysed an elastic
continuum to take into account the interaction between
ground and lining, but his work was not recognized by
practical engineers, possibly because he considered a rather
thick lining.
In Voellmy’s (1937) doctoral thesis® a thinner lining is
assumed, but the omission of che tangential components of
the earth’s pressure and the tangential displacements means
that the results obtained do not comply with the
: a1 uo N -
Fig. 4 Plane strain continoum model and characteristic distribution of radial displacement, u, hoop
forces, N, and bending moments, M
a4y=volume weight
Fig. 5
Bedi ting mode! for shallow tunnels without tension bedding a crown’® (distribution of
radial displacement, u, ground reaction pressure, o, hoop forces, N, and bending moments 4)
equilibrium and compatibility conditions of an elastic
continuum.
in 1944 Bull” introduced the assumption, which is valid
for shallow tunnels with limited overburden, that tension
bedding in the crown region should be neglected. The
bedded beam calculation i very tedious because of the
proposed evaluation and superposition of influence
functions for 16 single forces along the ring. Moreover,
‘overall equilibrium in the vertical direction is gained only by
2 downward rigi-body displacement.
Engelbreth* in Norway had in 1987 already derived a
closed formula for stress and deformation of the lining
deduced from a continuum model (Fig. 2). In 1961 Morgan?
assumed a priori thatthe circular lining deforms into an
elliptical mode. Thus, angential stresses and the
corresponding deformations are omitted. From this
inuitive approach some parts ofthe full solution forthe
continuum model are lost. The resulting bending moments
are theoretically too small
In 1964 Schulze and Duddeck"® published complete and
closed solutions for the model (Fig. 3) intended for
application for shallow cunnels with Limited overburden.
Hence, bedding atthe crown is omitted where tension
springs may cause load reductions. Results were given for
the direct design in diagrams for bending moments, hoop
forces and radial displacements for those three points of the
lining ae which values of a relative maximum occur. The
radial bedding modulus, K,, i still a free parameter, and the
tangential stresses may be included in or omitted from the
load parameters
In 1966 Windels" extended the approach of Schulze and
Duddeck to take into account second-order effects
(geometrical nonlinearity) and an approximation for pre-
deformation of the lining owing to the erection procedure,
eg, of tubbing rings. Thus, buckling instability is also
covered for the model shown in Fig. 3. Further
investigations of geometrical and physical nonlinear effects
(eg. plastic hinges) have been published by German workers
(o0t refered to here)
In 1967 Windels™ published an extensive and complete
treatise on the circular tunnel in an elastic continuum
ig, 2), geometrical non-linearity and pre-deformation also
being covered. Bending moments, hoop forces and radial
deflections were given in diagrams for the cos 2 stress
‘mode either with fll bonding or excluding tangential
stresses. The ratio of relative stffnesses of ground and lining
is taken as a free parameter.
Peck!® in 1969 proposed a semi-empirical approach that,
in exsence, is not very different from the model given in
Fig. 3 and the solutions presented in 1964," together with
the effects of compressibility stiffness and flexibility
stiffness. Tangential ground stresses were not taken into
consideration
‘Muir Wood" in 1975 gave a corrected version of
“Morgan's? more intuitive approach, again assuming an
elliptical deformation mode. The tangential ground stresses
axe included, but that part of the radial deformation which
is due co the tangential stresses is omitted (reference ro
previous work! would have enabled the correct model
(Fig. 2) to have been reached). Making allowances for some
predecompression of the ground around the opening before
the lining is placed, Muir Wood proposed to take only 50%
(or two-thirds, as Panet had suggested in 1973) ofthe inital
ground stresses into consideration. The moments may be
Feduced even more by reducing the lining stiffness by an
amount equivalent to the effect of less rigid hinges (results
from this method are compared with other design methods
later ~ see Figs. 12 and 15).
In 1976 Curtis! supplemented Muir Wood’s formulae’
by the inclusion of radial deformation due to tangential
stress. In addition, some valuable extensions in regard to
time-dependent behaviour (ereep) were given in simple
expressions for the change of stiffness ratio with time, The
todel proposed is that of Fig. 2.
Einstein and Schwartz!® proposed, after extensive study.
of the available methods for the design of deep tunnels, a
simplified analysis that is derived from the continuum
model (Fig. 2) and yields practically the same results as
those of Engelbreth® and Windels'* (see later).
‘Ahrens and co-workers!” investigated both models of
Figs. 2 and 3, Appropriate bedding moduli were derived for
the beam model (Fig. 3, but with full bedding also atthe
cxown) that yield the same solutions a those for the
continuum model of Fig. 2. Tangential bedding has to be
introduced, Differences either fr full bond or for tangential
slip are given, The results are incorporated into the German
recommendations.*
To summarize progress it may be stated thatthe methods
available are simple enough (even with design graphs) for
practical application, that the design engineer may choose
cither the model given in Fig. 2 (For moderately deep
tunnels) or that given in Fig. 3 (for shallow tunnels without
reduction of ground pressure at the crown), thatthe free
parameters also allow consideration of a reduced initial
{ground stress (say, 50% or even 30%) for deep tunnels in
stiffer ground, aking into account some decompression at
85