You are on page 1of 5

96

H. FARHADIAN
AND
OTHERS
JOURNAL
GEOLOGICAL
SOCIETY
OF INDIA
Vol.80, July 2012, pp.96-100

Optimization of Analytical Equations of Groundwater Seepage


into Tunnels: A Case Study of Amirkabir Tunnel
H. FARHADIAN, A. AALIANVARI and H. KATIBEH1
Mining, Metallurgy and Petroleum Dept., Amirkabir University, Tehran, Iran
Dept. of Mining and Metallurgy Engg., Amirkabir University of Technology, Hafez Ave., Tehran, Iran
Email: katibeh@aut.ac.ir

Abstract: In this study, with investigation of validity limits of analytical equations in computing groundwater seepage
discharge into tunnels in different values of r/h (tunnel radius/water table height above tunnel), using optimization by
regression analysis, we have tried to converge the results of analytical equations for any value of r/h. There are various
experimental, analytical and numerical methods by which it is possible to calculate the amount of groundwater seepage
into tunnels. Due to their simplicity and practical base theory, analytical methods have been applied more frequently.
Studies imply that amount of real seepage into tunnel is significantly less than what is anticipated from analytical
equations; on the other hand, the results of seepage calculations using these equations are depended on tunnel geometry
and medium conditions. Previous investigations confirmed that in a tunnel for which, r/h is more than 0.4; these results
are highly different both from one another and from the real seepage inflow. In this paper, optimization has been successfully
performed on analytical equations so that the results of these methods are converged to each other for any value of r/h.
Keywords: Tunnel seepage discharge, Groundwater, Optimization, Analytical equation.

INTRODUCTION

One of the most important problems in tunneling


operation is groundwater seepage into tunnels. The most
significant difficulties, resulted from water seepage into
tunnels, include reduction of rock mass stability around the
tunnel, imposition of extra pressure on temporary and
permanent supporting systems, reduction of operational
velocity (rate), and creating financial problems.
Due to impossibility of recognition and exact
determination of all effective factors on groundwater flow
into tunnels, especially during drilling operation in rocks,
exact prediction of groundwater seepage into drilled tunnels
is difficult. So, analytical methods and equations, because
of their simplifications and practical theories, have many
uses in calculation of groundwater infiltration into
tunnels.
The most important researches about calculation of the
rate of groundwater seepage into tunnels are studies of
Goodman et al. (1965), Freeze and Cherry (1979), Heuer
(1991), Lei (1999), Karlsrud (2001), Lombardi (2002) and
El-Tani (1999, 2003).
In addition to analytical methods which imply basic
estimation of infiltration rate, regarding to the basic
equations in seepage flow, using numerical methods such

as finite element method, finite difference method, distinct


element method, or finite volume method, it is possible to
simulate groundwater seepage flow into tunnels and
calculating seepage rates in different boundary conditions
and material properties. On the contrary of the analytical
methods, numerical methods do not have simplicity in
computation, and need overall information about boundary
conditions and material properties. As a result, numerical
methods are more complicated and their application needs
more time and equipments, but in case there is accurate
input data, they provide more exact results compared to
analytical methods.
The results of the above mentioned analytical methods
in estimating the rate of groundwater seepage into tunnels
diverge in great values of r/h (ratio of tunnel radius, r, and
water table height above tunnel axis, h). In this study,
investigating the validity limits of these analytical equations
for estimating groundwater seepage rate into Qomroud
tunnel for different values of r/h, using single regression
analysis, it is tried to optimize the analytical equations so
that the results of all these equations converge for any value
of r/h (including r/h>0.4). Then, the results of optimized
equations and their convergence have been validated with
observed seepage flow into Amirkabir tunnel.

0016-7622/2012-80-1-96/$ 1.00 GEOL. SOC. INDIA


JOUR.GEOL.SOC.INDIA, VOL.80, JULY 2012

OPTIMIZATION OF ANALYTICAL EQUATIONS OF GROUNDWATER SEEPAGE INTO TUNNELS

97

ANALYTICAL RELATIONS OF GROUNDWATER


SEEPAGE FLOW INTO TUNNEL AND THEIR
VALIDITY LIMITS

Analytical methods considering the parameters such as


rock mass permeability, water table height above tunnel axis,
tunnel radius etc., estimate the groundwater seepage rate
into tunnels. Table 1 shows the analytical equations used to
estimate seepage flow to tunnels.
In these equations H0 is the distance from tunnel axis to
groundwater table, Z, overburden thickness, r, tunnel radius,
k, equivalent permeability of rock mass in the direction of
seepage flow, and QL, the groundwater seepage rate in unit
length of tunnel. Figure 1 shows the applied parameters in
the equations presented in Table 1.
The presented analytical equations are not valid under
these conditions:
1 The vertical seepage flow towards tunnel
2 Bedding in rock around the tunnel
3 Inexact determination of the permeability of rock mass.
El-Tani (2003), applied Mobius transformation method
and Fourier series and presented a new analytical solution
for seepage flow calculation into tunnels. He also reviewed
different equations in this regard and discussed various

Fig.1. Tunnel condition in analytical equations and definition


of its parameters. The tunnel is located in saturated zone.
1 - Ground surface, 2 - groundwater table which is possible
to be upper than the burden, 3 - Water motion zone; the
equivalent permeability parameter k, is calculated in this
zone, 4 - Inner zone of tunnel.

conditions of their application. He showed that when the


ratio of tunnel radius to water table height above tunnel is
less than 0.4 (tunnel in large depth below water table) the
results of these equations are close to each other. But in
larger ratios the results of the mentioned methods start to
diverge more and more.

Table 1. Analytical equations of groundwater seepage flow into circular tunnels

Goodman 1965)

This equation has three basic following defaults, radius


flow, no bedding in rock, and accurate exertion of
equivalent permeability.

h
2z
)
r

Q = 2K

ln(

Freeze and Cherry


(1979)

Q=

Heuer (1991)

Q=

2 KH o
2H
ln( o )
r

These researchers, have revised equation 1, by substituting


z instead of H.

2 KH o 1

2z
ln( ) 8
r

Heuer reduction coefficient (1/8) and some changes in


denominator has been exerted to revise equation 2.

Q = 2K

Lei (1999)

El-Tani (1999)

Karlsrud (2001)

Q = 2K

h
h
h
ln( + ( ) 2 1)
r
r

r 2
)
2h
r 2
2h
r
[1 ( ) ]ln( ( )2
2h
r
2h
1 3(

Q = 2K

El-Tani (2003)

JOUR.GEOL.SOC.INDIA, VOL.80, JULY 2012

ln

2h
r
1 + 0.4
r
h

Q = 2k

EL-Tani has defined equation 5 as an optimum relation by


considering above mentioned relations.
A compound of equations 1 and 3, according to field
observations, is edited for reducing error in deep and
shallow tunnels (under water table).

h
2h
ln( 1)
r
h

Q = 2k

Lombardi (2002)

In this relation, Goodman method has been corrected with


more exact exertion of real conditions

2 1 h
2 + 1 ln

In this equation, Karlsrud method has corrected by more


accurate conditions.
In this equation, El-Tani, has applied Mobius
transformation method and Fourier series and presented a
new analytical solution for flow calculation, in which
= (h / r) ((h2 / r2) 1)1/2

98

H. FARHADIAN AND OTHERS


0.08

METHOD

0.07

Optimization of Different Analytical Method for Tunnels with


r/h>0.4

One of the most important applications of regression


analysis is the prediction of future anticipating value for y
parameter in terms of a defined amount of regressive variable
x. Empirical conversions are used for confirming these
different methods of calculating seepage discharge, so that
the result presumptions of different methods are improved.
Regression models are often used for data analysis of

Q Karlsrud

0.06
Q Lei

Q (m^3/day)

Anvari and Katibeh (2004), considering measured


groundwater seepage rates into Qomroud tunnel and
comparing them with the obtained results of the presented
analytical equations, showed if the ratio of tunnel radius to
water table height above tunnel (r/h) is less than 0.4, the
results of analytical equations no. 1, 4, 6, 8 in Table 1, will
have significant correlation with each other, while for larger
values, the results start to diverge gradually and show more
error.
Figure 2 shows the estimated groundwater seepage rates
into tunnel, obtained from different analytical methods and
various r/h values in sections of Qomroud tunnel. As it is
observed, for a tunnel with r/h0.4 or h4r, the analytical
solutions show very close results, but for a tunnel with
r/h0.4 divergence in the results is obvious.

0.05
Q El

0.04
Q Goodman

0.03

Q Lombardi

0.02
0.01
0
0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

r/h

Fig.2. Estimated groundwater seepage rates into Qomroud tunnel


against r/h values, using different analytical methods.

unscheduled tests, such as data registered during historical


observations.
In many problems, there are two or more variables that
are naturally related and it is required that the relation be
defined. Regression analysis is a statistical technique for
modeling and investigating the relations between two or
several variables. This regression model is fitted to a set of
data. In some cases, the real relation mode between y and
x1, x2,, xk, follows the known function y = (x1,x2,,xk),
but in many cases the functional relation is unknown and
the analyzer should choose a suitable empirical function
for . Usually, a multidimensional model is used to find
this relation. However, in many cases it has not been used
properly. The attention must be paid to selecting the model
variables and in determining the approximate function. It is
possible to find a statistical relation among some variables,

Table 2. Analytical equations and their optimized form for circular tunnels with r/h>0.4, using single variable regression
Reference

Goodman (1965)

Equation

Optimised equation for circular tunnels with r/h>0.4 using


single variable regression

QGo = 2K

Lei (1999)

QL = 2K

Karlsrud (2001)

Q Ka = 2k

QLo = 2k
Lombardi (2002)

El-Tani (2003)

h
2h
ln( )
r

h
h
ln( + ( ) 2 1)
r
r

h
2h
ln( 1)
r

Q Q Go * = 572 . 45 Q Go

67 . 138 Q Go + 2 . 0271

Q Q L * = 39 .429 Q L + 6 .3384 Q L 0 .18 .07


2

Q Q Ka* = 7.6392Qka 1.5399Q Ka + 0.1323


2

h
2

1 + 0.4 r ln 2h

h r

Q El = 2k

2 1 h
2 + 1 ln

Q Q Lo* = 186 .77Q Lo 19.382Q Lo + 0.5601


2

Q QEl* = 174.39QEl 18.1QEl 0.527


2

JOUR.GEOL.SOC.INDIA, VOL.80, JULY 2012

OPTIMIZATION OF ANALYTICAL EQUATIONS OF GROUNDWATER SEEPAGE INTO TUNNELS

Q* Goodman

0.03

Q* Lombardi

0.02
0.01
0
0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

r/h

Fig.3. Estimated groundwater seepage rates into Qomroud tunnel


against r/h values, using optimized analytical methods.

Observed
inflow

2600-2900

2400-2600

2300-2400

2200-2300

2100-2200

2020-2100

2000- 2020

0.01

distance (m)

Fig.4. Estimated groundwater seepage rates into Amirkabir tunnel


using analytical methods, along 2 to 3 km.

it can be seen, along 2.020 to 2.200 km, r/h is greater than


0.4, so in this portion of tunnel, the results of estimated
seepage using different methods show considerable
divergence and differ from the observed seepage. But from
2.200 to 3.000 km of tunnel, since r/h is less than 0.4, the
results of different methods are significantly same and are
in agreement with observed seepage. Applying the optimized
equations presented in Table 2 reduces the above mentioned
difference for 2.020 to 2.200 km, so that the results of
different methods coincide to each other and with the

CASE STUDY

10
Q* Karlsrud
Q* Lei

Log Q (cubic meter/day)

Q* El

Q* Goodman
Q* Lombardi
0.1

Observed
inflow

0.01
2000- 2020

Amirkabir tunnel, located in the northwest of Tehran,


Iran, has been designed and performed to transfer water from
Amirkabir dam to Tehran. One of the problems occurred in
performing this project is possibility of groundwater inflow
to the tunnel during drilling operation. This case has been
chosen to evaluate the validity of the presented analytical
and introduced optimized equations in this paper.

Q Lombardi
0.1

2600-2900

Q* El

0.04

Q Goodman

2400-2600

Q (m^3/day)

Q* Lei

0.05

Q El

2300-2400

0.06

Q Lei

2200-2300

Q* Karlsrud

Q Karlsrud

2100-2200

0.07

10

2020-2100

0.08

tunnel was divided into 14 different geological units


according to their sedimentology that generally encompasses
various sedimentary-volcanic sets of Karaj Formation. Its
petrology contains layers of tuff, sandstone, fine grained
conglomerate and siltstones, as well as lava and agglomerate.
In this study, the possibility of groundwater seepage inflow
from 2 to 3 km of the tunnel has been considered. This length
of tunnel is divided into 2 sections from geological point of
view: GAT1 (monzodiorite, gabbro) and GTA2 (sandstone
and tuff layers).
Rates of seepage discharge into the tunnel have been
estimated for 2 to 3 km of tunnel using 7 computational
sections. In this length of tunnel the ratio of r/h changes
from smaller to larger values than 0.4. The estimated seepage
rates from analytical methods have been shown in Fig.4. As

Log Q (cubic meter/day)

while, they are not really related and scientifically they are
totally apart from each other.
As mentioned above, for a circular tunnel with
r/h>0.4, the calculated groundwater seepage rates into
tunnels by analytical methods are different. In this research,
considering the real values of groundwater seepage rate
into Qomroud tunnel and effective parameters on seepage
flow, and using single variable regression method and
value engineering, it has been tried to optimize the analytical
methods so that they result in more exact and converged
answers for any value of r/h (including the ratios larger
than 0.4).
Table 2 shows the deformed and optimized equations
which have been obtained in this study. Because of these
empirical conversions, results of these various methods
converge to a balanced condition. These amounts of
QEl*,QLo*,QKa*,QL*,QGo*, are the optimized seepage rates
which have been obtained from presented equations in table
2 and converge to their average, Q . Therefore, having one
of the optimized form of analytical equations, optimal
amount of Q can be estimated.
Figure 3 shows the diagram of estimated values of
groundwater seepage rate into Qomroud tunnel, by
optimized equations in Table 2. As can be seen, the estimated
seepage rates into tunnel using optimized equations for r/h
more than 0.4, are converged and very close to the mean
value of different methods.

99

distance (m)

Geology of the Area

In geological studies which have been performed, the


JOUR.GEOL.SOC.INDIA, VOL.80, JULY 2012

Fig.5. Estimated groundwater seepage rates into Amirkabir tunnel


using optimized analytical methods, along 2 to 3 km.

100

H. FARHADIAN AND OTHERS


Q Karlsrud

10

Q (m^3/day/m)

Q Goodman
Q Lombardi
Observed
inflow

0.1

Log Q (cubic meter/day)

Q* Lei

Q El

Q* Karlsrud

10

Q Lei

Q* El

Q* Goodman
Q* Lombardi
Observed
inflow

0.1

0.01
0

0.01

0.2

0.4

r/h

0.6

0.8

0.2

0.4

r/h

0.6

0.8

Fig.6. Estimated groundwater seepage rates into Amirkabir tunnel


against r/h values using analytical methods, along 2 to
3 km.

Fig.7. Estimated groundwater seepage rates into Amirkabir tunnel


against r/h values using optimized analytical methods, along
2 to 3 km.

observed seepage discharge into the tunnel. The results of


this operation have been shown in Fig.5.
In Fig.6, r/h values for 7 computational sections have
been plotted against groundwater seepage rates into
Amirkabir tunnel. As the figure shows two sections has r/h
value larger than 0.4. So in these sections, the estimated
seepage discharges due to different methods (before
optimization) are showing considerable difference. Applying
optimized equations from Table 2, the results of all the
methods converge in these 2 points also and coincide with
the observed seepage discharge. The output of this operation
has been displayed in Fig.7.

using analytical equations into tunnel was compared with


real value and for r/h larger than 0.4, there was considerable
difference between calculated and real values. Using singlevariable regression method and value engineering, optimized
equations with lower error were found for r/h greater than
0.4.
QGo*, QL*, QKa*, QLo*, QEl* are optimized values that
obtained from various analytical equations of groundwater
seepage flow to tunnels, and are equal to Q that is the
average of single-variable regression for r/h higher than 0.4
and lead to a balanced convergent state.
According to the performed optimization in this study,
all the presented methods can be used suitably to calculate
the seepage discharge into tunnels, even for large values of
r/h (r/h>0.4), so that the results show good accordance with
each other and with the real seepage discharge. From 2.02
to 2.2 km along Amirkabir tunnel, r/h was larger than 0.4,
hence, optimized equations were used and a good
approximation to the real seepage in the r/h larger than 0.4
was achieved.

CONCLUSION
Studies on Qomroud and Amirkabir tunnels showed that
the ratio of tunnel radius to the height of water table above
tunnel axis (r/h) can be an effective factor on the accuracy
of different analytical methods to calculate groundwater
seepage discharge into tunnels. Calculated water seepage

References
AALIANVARI, A., KATIBEH, H. and MAHMUDABADI, H. (2009) Estimation of
equivalent permeability in Amirkabir tunnel alignment with neural
network. 21st Internat. Mining Congress and Exhibition of Turkey,
May 6-8, Antalya.
CHENG-HAW LEE., BOR-WEI DENG and JUI-LIN CHANG, (1995) A continuum
approach for estimating permeability in naturally fractured rocks.
Engg. Geol., v.39, pp.71-85.
DOUGLAS, C.M. (2001) Design and analysis of experiment. 5th edn., John
Wiley & sons.
ELTANI , M. (2003) Circular tunnel in a semi-infinite aquifer. Jour.
Tunneling and Underground Space Tech., v.18, pp.49-55.
FREEZE, R.A. and CHERRY, J.A. (1979) Groundwater. Englewood, New
Jersey, Prentice-Hall Inc., TIC 217571.
GOODMAN , R., M OYE , D., S CHALKWYK , A. and JAVENDEL , I. (1965)
Groundwater inflow during tunnel driving. Engg. Geol., v.1 , pp.150162.
HWANG, J.H. (2007) A semi-analytical method for analyzing the tunnel
water inflow. Jour. Tunnelling and Underground Space Tech., v.22,
pp.39-46.
HEUER, R.E. (1995) Estimating rock-tunnel water inflow. Proc. Rapid

Excavation and Tunneling Conf., June 18-21.


KATIBEH, H. and AALIANVARI, A. (2009) Development of a new method for
tunnel site rating from groundwater hazard point of view. Jour. Appld.
Sci., v.9, pp.1496-1502.
LEE, C-H., DENG, B-W. and CHANG, J-L. (1995) A Continuum Approach
for Estimating Permeability in Naturally Fractured Rocks. Engg. Geol.,
v.69, pp.71-85.
LEI, S. (1999) An analytical solution for steady flow into a tunnel. Ground
Water, v.37, pp.23-26.
MERI, DAVID (1985) Unconfined groundwater flow calculation into a tunnel.
Jour. Hydrology, v.82, pp.69-75.
PARK, K.H., OWATSIRIWONG, A. and LEE, J.G. (2008) Analytical solution
for steady-state groundwater inflow into a drained circular tunnel in
a semi-infinite aquifer: A revisit. Jour. Tunnelling and Underground
Space Tech., v.23, pp.206-209.
PRIEST, S.D. and HUDSON, J. (1976) Discontinuity Spacing in Rock. Int.
Jour. Rock Mech. Min. Sci. & Geomech. Abstr. v. 6, pp.165-148.
RAYMER, J.H. (2003) Predicting groundwater inflow into hard-rock tunnels:
Estimating the high-end of the permeability distribution. RETC,
pp.201-217.

(Received: 19 December 2009; Revised form accepted: 18 July 2011)


JOUR.GEOL.SOC.INDIA, VOL.80, JULY 2012

You might also like