TUNNEL INFLOW ASSESSMENT IN DISCONTINUOUS ROCK MASSES:

FROM NUMERICAL MODELING TO EMPIRICAL EQUATIONS

Paola Gattinoni, L. Scesi, S. Terrana

Politecnico di Milano (Dept. of Environmental, Hydraulic, Infrastructures and Surveying

Engineering), Piazza Leonardo da Vinci 32, I-20133 Milano, Italy

Keywords: Tunnel inflow, fractured rocks, numerical modeling, empirical formula

INTRODUCTION
Frequently, the excavation of medium-deep tunnels without waterproofing brings about an
hydrogeological risk, not only because it can interfere with springs as well as surface waters, but
also because it can cause trouble for the workers safety and for the efficiency of the tunnel drainage
systems.
The paper deals with the analysis of drainage process involved in tunnel construction. In particular,
the work compares the traditional analytic methods for tunnel inflows forecasting to the modeling
approach, with specific reference to the case of anisotropic rock masses. The study was carried out
through a parametrical modeling, so that groundwater flow was simulated with the DEM Model
UDEC 2D. The influence of geo-structural setting (as dip and dip direction of discontinuities, with
reference to their permeability) on tunnel drainage process was quantified. Results were compared
with the ones obtained through the well-known analytic formulas (i.e. Goodman and El Tani
equations) pointing out the following aspects:
the geological- structural setting critical for hydrogeological risk in tunnelling;
the influence of rock mass anisotropy;
the reliability of these analytic formulas for discontinuous rock masses.
Finally, using modeling results, the previous cited analytic formulas were corrected to point out an
empirical equation that gives the tunnel inflow as a function of the geological-structural setting.
DEFINITION OF THE PROBLEM AND ACTUALLY KNOWLEDGE
In the last few years many studies about tunnel drainage effect were carried out (Anagnostou, 1995;
Barla, 2000; Gisotti et al., 2002; Civita et al., 2002; Pazzagli et al., 2001; Picarelli et al., 2002;
Reuter et al., 2000), therefore it was possible to define the contribution that hydrogeology can offer
to the different tunnels planning stages, with particular reference to:
the forecast of groundwater tunnel inflow (Goodman et al., 1965; Knutsson et al., 1996;
Ribacchi et al., 2002; Park et al., 2008; Perrochet et al., 2007; Cesano et al., 2003; Hwang et
al., 2007);
the hydrogeological environmental impact: springs regime, groundwater level drawdown
and surface water impoverishment (Dematteis et al., 2001; Molinero et al., 2002; Gattinoni
et al., 2006; Loew, 2002).
For both problems it is necessary to start from a complete knowledge of the rock mass, because the
water flow is controlled by joints orientation, by their hydraulic characteristics and by rock
fracturing conditions (Gattinoni et al., 2007; Lee et al., 1993; Min et al., 2004; Snow, 1969; Louis,
1974). For example, the permeability can greatly increase along shear zones and the direction of
these zones determines the shape and the size of the drainage potential areas.
1

For the evaluation of the tunnel inflow some Authors suggest analytic formulations (Table 1), valid
for infinite, homogeneous and isotropic aquifers. Otherwise, some Authors suggest analytic
formulas for finite thickness aquifer (Custodio, 2005) and for anisotropic aquifer (Kawecki, 2000).
All these analytic formulas, both for isotropic or anisotropic aquifers, allow to evaluate the tunnel
inflow at whole saturation conditions and in homogeneous medium. On the contrary, discontinuous
rock masses are a typical anisotropic and non homogeneous medium. In this case it is necessary to
integrate the analytic methods with numerical modeling (Dunning et al., 2004; Molinero et al.,
2002), or to adapt the analytic formulas to the specific hydrogeological structure.
Table 1 - Analytic formulas for the tunnel inflow assessment. All the above cited formulas are based on the
hypothesis of homogeneous and isotropic aquifer, horizontal water table and r << H.
K = hydraulic conductivity; L = length of the tunnel; H = depth of the tunnel centre from the water table; h
= hydraulic head into the tunnel; S = specific storage coefficient ; r = tunnel radius (with lining: re is the
external radius and ri is the internal radius); R = tunnel radius of influence; Kl =tunnel lining hydraulic
conductivity; D = hydraulic load above land surface; t = time; x = spatial coordinate along the tunnel axis
with the origin at the entry of the permeable zone; v = drilling speed; (L-x) = Heaviside step function (also
named unit step function; when (Lx) < 0, (L-x) = 0; when (L-x) > 0, (L-x) = 1).
Steady state

2 KL ( H h )
Q =
2H 2h
ln

Goodman (1965)

2 KL ( H h )

Q =

H Dh
ln
+

H Dh

1
r

2KL ( H h) ln(re / ri ) K
1 +
Q=

ln( R / re ) ln( R / re ) K l
2 KL ( A + D )

Q =

(H D )
ln
+

where A = ( H D)

Lei (1999)

Ribacchi et al.
(2002)

Water table below land surface.

Hydrostatic load constant along the
tunnel border.
Water table above land surface.
Hydrostatic load constant along the
tunnel border.

Tunnel lining. Hydrostatic load

constant along the tunnel border.
Water table above land surface.
Hydrostatic load along the tunnel
border depending on the stage.

(H D)2
1
2

Park et al.
(2008)

(1 2 )
(1 + 2 )

1
r

= ( H D ( H D) 2 r 2 )

Q = 2KL

2 1 ( H h)

2 + 1 ln

where
El Tani (2003)

( H h)
( H h) 2
=

1
r
r2

Transient state
Jacob & Lohman
(1952)

4 KL [ H ( t ) h ]
Q (t ) =
ln( 2 . 25 KLt / Sr 2 )
vt

Q (t ) = 2
0

K [ H (t ) h )] ( L x)

x
K
t
ln 1 +
(

)
v
Sr 2

dx

Perrochet et al.
(2005)

Water table below land surface.

Hydrostatic load along the tunnel
border depending on the stage.
Extension for non horizontal water
table.

Hydrostatic load constant along the

tunnel border
Hydrostatic load constant along the
tunnel border
Extension for heterogeneous aquifer
by Perrochet et al. (2007)

NUMERICAL MODELING OF THE TUNNEL INFLOWS

To overcome the previously described limits in the use of analytic formulas for tunnel inflow in
rock masses, a modeling approach was used. The simulations were carried out using the Universal
Distinct Element Code (UDEC; Itasca, 2001), which is a numerical program able to perform 2-D
analysis of fluid flow through a fractures network. The modeling approach allows to determine both
the flow and the drawdown induced by the tunnel excavation in different structural and
hydrogeological conditions. The aim was to create a sufficient data set of tunnel inflows, in several
geological-structural setting, to enable a quantitative comparison between numerical and analytic
estimations.
Model implementation
For the model implementation an horizontal tunnel having N-S direction was considered; the
parametrical simulations were carried out for the following geological structural setting:
two discontinuity families having the same strike of the tunnel axis (N-S) and the dip
direction toward E (the first family) and W (the second family);
variable dips;
variable spacing and aperture.
The numerical modeling interested a rectangular domain (100x200 m2) considering the tunnel
excavation on domain border (Figure 1), that is to assume a symmetric geostructural and
hydrogeological setting. The conceptual model implementation includes following steps:
definition of joint characteristics: dip (ranging from 0 to 90, spacing (ranging from 3 m to
12 m) and aperture (ranging from 1.03E-04 m to 1.63E-04 m);
definition of the hydrogeological setting of the area: water table 10 m below land surface
and no recharge;
tunnel project parameters: depth equal to 40 m from water table, radius equal to 5 m and no
waterproof lining.

Figure 1 - Examples of modeling domain and their tensor ellipses (light blue): K1 and K2 are the tensor
components (respectively horizontal and vertical) belonging to the same plane of the cross section whereas
the other tensor component (K3) is perpendicular to the model cross section. Discontinuity dip: A)
Vertical/Horizontal; B) E/45 and W/45; C) E/20 and W/20 ; D) E/70 and W/70. In red are indicated
the impermeable boundary conditions and in violet are indicated the constant head boundary conditions.

The following boundary conditions were applied (Figure 1):

impermeable boundary along the bottom;
impermeable boundary along the same border of the tunnel location;
hydrostatic load on the opposite side as regards the tunnel location (groundwater supply
boundary condition).
Then, the numerical simulations were carried out only solving the hydraulic problem, whereas the
mechanical aspects were at first neglected, according to the following procedure:
simulation in pre-excavation conditions;
simulation in post-excavation conditions, till the steady state was reached.
Results and discussion
For each simulation the following features were pointed out in post-excavation steady state:
tunnel inflow,
water table drawdown along the tunnel axis as regards the pre-excavation conditions.
The results show that the tunnel inflow decreases when joint dip increases. This effect is amplified
for low values of spacing. Actually, in saturated medium the water flow towards the tunnel is
mainly horizontal; consequently joints with low dip (close to 0) favour the drainage processes and
increase the inflow (Figure 2). Otherwise, joints with orientation orthogonal to the water flow
direction cause the decrease of discharge and, therefore, the water table drawdown results higher
with the increase of joints dip. This is shown by the water table profiles of the drainage process
induced by tunnel with reference to the variation of joints dip (Figure 3).
Logically the model also shows that with the increase of the joints opening, the hydraulic
conductivity increases as well according to an exponential low and that the higher fracturing degree
is (high frequency), the higher drawdown and tunnel inflow will be.

Figure 2 - Trend of the tunnel inflow versus the discontinuities frequency (discontinuities aperture kept
constant) for several geostructural setting.

Figure 3 - Trend of the water table drawdown along tunnel axis versus the discontinuities frequency
(discontinuities aperture kept constant) for several geostructural setting.

It is important to notice that the model implementation does not provide for the recharge,
consequently the tunnel radius of influence cannot be defined because it coincides with the
dimension of the domain.

COMPARISON BETWEEN NUMERICAL MODELING RESULTS AND ANALYTIC

FORMULAS
The inflow values obtained from numerical modeling were compared with the results obtained by
the analytic formulas (Table 1) valid for steady state and water table below land surface (Goodman
and El Tani equations). The comparison pointed out that the analytic formulas overestimate the
tunnel inflow. This overestimation is bigger for geostructural setting having discontinuities with
higher dips (Figure 4). In particular, El Tani formula gives a little better estimation than the more
simple Goodman formula, but the trend is totally similar. So that the use of the most complex El
Tani equation doesnt seam to fit in a better way the tunnel inflow in discontinuous rock masses.
Based on the comparison between the numerical results and the tunnel inflows calculated with the
Goodman equation, the following empirical relation was pointed out:
Q = aQGb

(1)

where Q (m3/s) is the tunnel inflow in discontinuous rock mass, QG (m3/s) is the tunnel inflow
obtained from the Goodman equation, a and b are empirical dimensionless coefficients depending
on the geostructural setting. In particular, these empirical coefficients depend on:
the horizontal component of discontinuities,
the hydraulic conductivity anisotropy ratio,
the orientation of the hydraulic conductivity tensor.

Figure 4 - The points represent the simulated tunnel inflows versus the ones calculated with Goodman
equation for different geostructural setting. The continuous lines arise from the power regression of the
simulated values. The corresponding regression coefficients are shown in the graphic.

According to these considerations, a dimensionless coefficient F was defined (Figure 5):

n

F=

cos

i =1

K min

K max

0.5

(2)

where n is the number of discontinuity sets, i is the dip of ith discontinuity set, Kmin and Kmax are
respectively the minimum and maximum components of the hydraulic conductivity tensor, whereas
is defined by the following equation:
1
1

if
if

min > 45
min 45

(3)

assuming that min is the angle between the Kmin direction and the horizontal plane.
Consequently, the empirical dimensionless coefficients a and b were defined as a function of F,
obtaining:
b = ln 3.463F 0.0342
3.448F 0.8834
a=
0.6805
3.2411F

( 4)
for

F <1

for

F 1

(5)

Figure 5 - Trend of the coefficient a versus the coefficient F. The series points simulations1 and
simulations2 represent the interpolated values obtained from the simulations of Figure 4, respectively for
F<1 and F1. The continuous line is the corresponding regression for F<1, whereas the dotted line is the
regression for F1. The series verification1 and verification2 represent the verification points respectively
for symmetric and asymmetric geological structural setting.

The equation (1) is valid only if the ratio Q/QG is lesser than the unit, that corresponds to values of
the coefficient F decreasing with the increase of the value of QG.
The above equation was verified considering the following cases (Figure 5):
symmetric conditions, in presence of two or four discontinuity sets;
asymmetric conditions in presence of two, three or four discontinuity sets.
As far as the asymmetric geometry is concerned, the modeling was carried out in two phases;
actually, the left and right sides of the domain (always considering the tunnel excavation on a
domain border) were simulated separately and the tunnel inflow was calculated adding the two
contributions. The results obtained from these simulations showed a good agreement with the
empirical formula, also for asymmetric geostructural settings (Figure 5), with an average
overestimation lesser than 10%.

CONCLUSIONS
In the last few years, the tunnel inflows forecasting has assumed a great applicative interests, both
for the environmental impact assessment and for the tunnel drainage system design. At this aim, the
analytic equations found in the technical literature could be very useful, but often they dont reflect
the real phenomena of the tunnel inflow in rock masses. Actually, these equations are based on the
hypothesis of homogeneous aquifer, and then they dont give good agreement for heterogeneous
fractured medium.
In this latter case, the numerical modeling could provide the best results, but only with a detailed
conceptual model of the water circulation, high costs and long simulation times.
As in practice the data available during the design are generally limited and not sufficient to
describe the real situation, a simple equation, able to take into account the fractures network, was
defined. This empirical equation was pointed out on the basis of parametrical modeling and allows
7

to forecast the tunnel inflow as a function of joint characteristics. This first evaluation of the tunnel
inflow can be very useful to identify the areas where in-depth studies are required.
In the future, the research will be focused on the scale effect, in term of different ratio between the
tunnel radius and the joints spacing, and on the stress field, that can affect both the joint aperture
and the water pressure.

REFERENCES
Anagnostou, G. (1995), The influence of tunnelling on the Hydraulic head, Int J. Numer. Anal. Meth. Geomech., 19,
pp. 725-746.
Barla, G. (2000), Lessons learnt from the excavation of a large diameter TBM tunnel in complex hydrogeological
conditions, International Conference on Geotechnical and Geological Engineering, pp. 938-995.
Cesano, D., Bagtzoglou, A.C. and Olofsson, B. (2003), Quantifying fractured rock hydraulic heterogeneity and
groundwater inflow prediction in underground excavations: the heterogeneity index, Tunneling and Underground
Space Technology, 18, pp. 19-34.
Civita, M., De Maio, M., Fiorucci, A., Pizzo, S. and Vigna. B. (2002), Le opere in sotterraneo e il rapporto con
lambiente: problematiche idrogeologiche, Meccanica e Ingegneria delle rocce, pp. 73-106.
Custodio, E. and Llamas, M.G. (2005), Idrologia sotterranea, Dario Flaccovio Editore.
Dematteis, A., Kalamaras, G. and Eusebio, A. (2001), A systems approach for evaluating springs drawdown due to
tunneling, World Tunnel Congress AITES-ITA 2001, Vol.1, pp. 257-264.
Dunning, C.P., Feinstein, D.T., Hunt, R.J. and Krohelski, J.T. (2004), Simulation of Ground-Water Flow, SurfaceEater Flow, and a Deep Sewer Tunnel System in the Menomonee Valley, Milwaukee, Wisconsin, USGS Scientific
Investigations Report, 2004-5031.
El Tani, M. (2003), Circular tunnel in a semi-infinite aquifer, Tunnelling and Groundwater Space Technology, 18,
pp.49-55.
Gattinoni, P. and Scesi, L. (2007), Roughness control on hydraulic conductivity in fractures, Hydrogeology Journal,
15, pp. 201-211.
Gattinoni, P. and Scesi, L. (2006), Analisi del rischio idrogeologico nelle gallerie in roccia a media profondit,
Gallerie e grandi opere sotterranee, 79, pp. 69-79, Patron Editore (Bologna).
Gisotti, G. and Pazzagli, G. (2002), Linterazione tra opere in sotterraneo e falde idriche. Un recente caso di studio,
World Tunnel Congress AITES-ITA 2001, Vol. 1, pp. 327-334.
Goodman, R.E., Moye, D.G., Van Schalkwyk, A. and Javandel, I. (1965), Ground water inflow during tunnel driving,
Eng. Geol., 2, pp. 39-56.
Hwang, J.H. and Lu, C.C. (2007), A semi-analytical method for analyzing the tunnel water inflow, Tunneling and
Underground Space Technology, 22, pp. 39-46.
Itasca (2001), UDEC, Users guide, Itasca Consultino Group Inc., Minneapolis, Minnesota.
Jacob, C.E. and Lohman, S.W. (1952), Nonsteady flow to a well of constant drawdown in an extensive aquifer,
Transaction of the American Geophysical Union, 33(4), pp. 559569.
Kawecki, M.W. (2000), Transient flow to a horizontal water well, Ground Water, 38(6), pp. 842-850.
Knutsson, G., Olofsson, B. and Cesano, D. (1996), Prognosis of groundwater inflows and drawdown due to the
construction of rock tunnels in heterogeneous media, Res. Proj. Rep. Kungl Tekniska, Stokholm.
Lei, S. (1999), An analytical solution for steady flow into a tunnel, Ground Water, 37(1), pp. 23-26.
Loew, S. (2002), Groundwater hydraulics and environmental impacts of tunnels in crystalline rocks, Meccanica e
Ingegneria delle rocce, pp. 201-217.
Louis C., 1974. Introduction lhydraulique des roches. Bur. Rech. Gel. Min. 4 (III), 283-356.
Min, K.B., Jing, L. and Stephansson, O. (2004), Determining the equivalent permeability tensor for fractured rock
masses using a stochastic REV approach: Method and application to the field data from Sellafield, UK, Hydrogeology
Journal, 12, pp. 497-510.

Molinero, J., Samper, J. and Juanes, R. (2002), Numerical modelling of the transient hydrogeological response
produced by tunnel construction in fractured bedrocks, Engineering Geology, 64, pp. 369-386.
Park, K.H., Owatsiriwong, A. and Lee, G.G. (2008), Analytical solution for steady-state groundwater inflow into a
drained circular tunnel in a semi-infinite aquifer: a revisit, Tunnelling and Underground Space Technology, 23, pp.
206-209.
Pazzagli, G. (2001), Linterazione tra opere in sotterraneo e falde idriche. Un recente caso di studio, AITES-ITA 2001,
Progress in tunnelling after 2000, Ed. Teuscher P., Colombo A., Patron Vol. 2, pp. 327-334.
Perrochet, P. (2005), A simple solution to tunnel or well discharge under constant drawdown, Hydrogeology Journal,
13, pp. 886-888.
Perrochet, P. and Dematteis, A. (2007), Modelling Transient Discharge into a Tunnel Drilled in Heterogeneous
Formation, Ground Water, 45(6), pp. 786-790.
Picarelli, L., Petrazzuoli, S.M. and Warren, C.D. (2002), Interazione tra gallerie e versanti, Meccanica e Ingegneria
delle Rocce, pp. 219-248.
Reuter, E., Kopp, B. and Lemke, S. (2000), Hallandsas Tunnel. Waterproofing system with a 4 mm thick plastic
membrane, Tunnel, 6, pp. 39-45.
Ribacchi, R., Graziani, A. and Boldini, D. (2002), Previsione degli afflussi dacqua in galleria e influenza
sullambiente, Meccanica e Ingegneria delle rocce, pp. 143-199.
Snow, D.T. (1969), Anisotropic permeability of fractured media, Water Resources Res, 5.