HE

1 8.5
. A3 7
no
DOT- iRT NO. UMTA-MA-06-01 00-79-1
TSC-
UMTA-
79-49
IMPROVED DESIGN OF
TUNNEL SUPPORTS:
EXECUTIVE SUMMARY

Herbert H. Einstein, Amr S. Azzouz,

Charles W. Schwartz, Walter Steiner

MASSACHUSETTS INSTITUTE OF TECHNOLOGY

Department of Civil Engineering
Cambridge, MA 02139

A of

DECEMBEr/1979
FINAL REPORT

DOCUMENT IS AVAILABLE TO THE PUBLIC

THROUGH THE NATIONAL TECHNICAL
INFORMATION SERVICE, SPRINGFIELD,
VIRGINIA 22161

PREPARED FOR

U.S, DEPARTMENT OF TRANSPORTATION

I'.
Urban Mass Transportation Administration
Office of Technology Development and Deployment
Office of Rail and Construction Technology
Washington, DC 20590
 NOTICE

This document is disseminated under the sponsorship

of the Department of Transportation in the interest
of information exchange. The United States Govern-
ment assumes no liability for its contents or use
thereof.

NOTICE

The United States Government does not endorse pro-

ducts or manufacturers. Trade or manufacturers'
names appear herein solely because they are con-
sidered essential to the object of this report.
 Tvctinical Report Docunt«ii1utton Po{^

1. Report No. 2. Covornmont Accotsion No. 3. Rocipiont'* Catalog No.

UMTA-MA-06-0100-79-15
4. Title artd Subtitle
m 5. Report Date
TBANSPOfiTATIGN December 1979
IMPROVED DESIGN OF TUNNEL SUPPORTS: 6. Performing Orgoniietion Code
EXECUTIVE SUMMARY iR ^ 1 1S80
8. Performing Orgoniiotion Report No.
7. Author' i) Einstein, A. S. Azzpuz, LIBRARY DOT-TS C-UMTA- 79-49
C. W. Schwartz, and W. Steiner
9. Performing Organization Name and Addrei* ]0. Work Un.t No. (TRAIS)

Massachusetts Institute of Technology* MA-06-0100(UM948/R9745)

Department of Civil Engineering 1]. Controct or Grant No.

Cambridge, Massachusetts 02139 DOT-TSC-1489

13. Type of Report and Period Covered
12. Sponsoring Agency Nome ond Address
Final Report
U.S. Department of Transportation
Jan. 19 7 8- Aug. 1979
Urban Mass Transportation Administration
400 Seventh Street, S.W. 14. Sponsoring Agency Code

Washington, DC 20590 UTD-30

15. U.S. Department of Transportation
Suppiementory Notes
Research and Special Programs Administration
*under contract to: Transportation Systems Center
Cambridge, Massachusetts 02142
16. Abstract This report focuses on improvement of design methodologies related to the
ground-structure interaction in tunneling. The design methods range from simple ana-
lytical and empirical methods to sophisticated finite element techniques as well as an
evaluation of tunneling practices in Austria and Germany. The purpose of this report
is to provide the tunneling profession with improved practical tools in the technical
or design area. These design tools provide more accurate representations of the
ground-structure interaction in tunneling.
The Executive Summary is the first of six publications to be published on the Improved
Design of Tunnel Support. Volumes 1 through 5 will be published in March, 1980. This
Executive Summary summarizes improvements in the methodology available to tunnel de-
signers with the objective of reducing the cost of tunnel construction in the United
States. This report summarizes each of the five volumes. Volume 1: Simplified Ana-
lysis for Ground- Structure Interaction , describes a simplified analysis method, which
incorporates the effects of relative stiffness, face-support distance, and ground
yielding on ground-structure behavior; Volume 2: Aspects of Yielding in Ground-Struc-
ture Interaction addresses the effects ground yielding has on ground-structure inter-
,

action; Volume 3: Finite Element Analysis of the Peachtree Center Station in Atlanta ,

reports the application of a three-dimensional finite element analysis of an Atlanta,

Georgia station; Volume 4: Tunneling Practices in Austria and Germany ; and Volume 5:
—
Empirical Methods for Rock Tunneling Review and Recommendations , examines empirical
methods in tunneling and provides a guide to the advantages, limitations, and applica-
bility for the best known empirical methods.

17. Key Word* Empirical Methods in Tunnel 18. Distribution Statement'

Design; European Tunneling Practice;
Finite Element Analysis; Ground-Structure Available to the public through the
Interaction; Tunnel Construction; Tun- National Technical Information Service,
nel Design; Tunneling; Tunnels and Tun- Springfield, Virginia 22161
neling; Underground Construction
19. Security Claxif. (of this report) 20. Security Cletsif. (of this page) 21. No. of Pages 22. Pr.e#

Unclassified Unclassified
55 AQ4
Form DOT F 1700.7 (8-72) Reproduction of completed pogc outhoriced
 S S r I ~cV E 3 "
s & ^ &c ^

III i I .it
II
s s
s| 1
t i E im Hi lllill

e
1
s

Jf I 6 e € S S 3 E -EE
65
©
O
ct tt It |or 61 81 H 91 91 »1 Ct Cl 11 01 6 9 1 9 S 9 C t

,1 In In,
e 1 ,

lllllllll lllllllll tiiiiiitiiiiiiiitit lllllllll lllllllll itniiiii lllllllll lllllllll lllllllll lllllllll lllllllll lllllllll lllllllll lllllllll lllllllll lllllllll lllllllll lllililtl lllllllll lllllllll lllllllll

>
© T|T T|T T|T T|T T|T T|T T|T T|T T|T T|T T|T T|T T|T T|T T|T T|T
© '•’P'

9 8 7 6 6 4 3 3 t

g S E i '^~e~e1 3 E66 EE

M • 3
6 E
k i
s If • i • 1 1
I i : =
1
il •i 4
if
•
1st 3 3 h. i 2

X a « ^ if
A8 0
0 o o
«o «
o
•
V
• ••
0 o o
s
* 00 000*^00 2
(9 ~a ® - 2 ^
X
IM << e < Ji 3 »
> K
IM
s o.
2
2 • 5 •* “
Isll •ft
“ • g 5 s -s

III assi^
fill lllli Hi Ss=’ hlill

1
M -c-vVl ? fs o E » i'c'k-

11
 PREFACE

The uncertainties associated with the design and con-

struction of transit tunnels have resulted in inflated con-

struction costs which include sizable margins for error to

reduce the risk of tunnel construction for contractors. The

Urban Mass Transportation Administration (UMTA) through a

series of contracts to universities and private engineering

firms has attempted to improve existing design methodologies,

construction techniques, and contractual arrangements in order

to reduce the associated risks and produce more cost-effective

tunneling for transit and other tunneling applications.

This contract with the Massachusetts Institute of

Technology (MIT) focuses on improvement of design methodologies

related to the ground-structure interaction in tunneling.

The design methods range from simple analytical and empirical

methods to sophisticated finite element techniques as well as

an evaluation of tunneling practices in Austria and Germany.

The information contained in these six publications will be

invaluable to tunnel designers and constitutes a major step

toward more cost-effective tunneling.

We would like to acknowledge contributions made by

members of the project review board: Messrs. R. Beale,

H. Gasp, G. Fox, T. McCusker, H. Parker, and H. Sutcliffe.

In addition to the authors , several undergraduate students

worked on the project notably B.K. Low and E. Whitbeck.

iii
Finally, the information retrieved in Europe was possible

only with the generous help provided by our colleagues.

They are individually mentioned in Volume 4.

IV
 TABLE OF CONTENTS

Section Page

1. INTRODUCTION 1

2. SIMPLIFIED ANALYSIS FOR GROUND-STRUCTURE

INTERACTION IN TUNNELING 7

3. ASPECTS OF YIELDING IN GROUND-STRUCTURE

INTERACTION 17

4. FINITE ELEMENT ANALYSIS OF THE PEACHTREE

CENTER STATION IN ATLANTA 21

5. TUNNELING PRACTICES IN AUSTRIA AND GERMANY. 27

6. EMPIRICAL METHODS FOR ROCK TUNNELING REVIEW

AND RECOMMENDATIONS 35

7. REFERENCES 47

APPENDIX REPORT OF NEW TECHNOLOGY 49

v/vi
 1. INTRODUCTION

This six volume series of reports presents the results

of a one and one-half year effort sponsored by the Urban Mass

Transportation Administration (UMTA) , U.S. Department of Trans-

portation (DOT), under Contract No. DOT-TSC-148 9 . The work re-

ported here is a part of ongoing systematic research to improve

tunnel design and construction that began with the tunnel Cost

Model and with studies under grant OS-60136, Office of University

Research, DOT. Improvement of tunnel design and construction

through this research is achieved by advancing specific technical,

operational, and contractual aspects of tunneling and, in par-

ticular, by employing these advances to integrate the entire

tunnel design-construction process. As indicated in previous

work and as documented in these reports, major improvements

leading to more economic tunneling must come from integration of

the entire tunneling process. Nevertheless, in order to provide

the tunneling profession with some practical tools as early as pos-

sible, the research is structured such that the investigations

of the individual components not only contribute to the overall

integration of the tunneling process, but also produce results

that can be applied directly and independently. The nature of

these individual components were previously identified (Einstein,

et al,, 1977). Specifically, they are: (a) improvements in

1
analytical and empirical methods to be used during design and

construction, (b) procedures for better decision making and cost

assessments in all phases of the tunneling process (planning,

design, construction planning, and construction), and (c) contrac-

tual details that will lead to more economical tunnel construc-

tion. The decision-making procedure will, in addition to its

direct use, serve as the framework for improving the tunneling

process and fully incorporating other new advances into this

process. During the research, strong but not exclusive considera-

tion is given to observational methods, since some of these meth-

ods have already achieved considerable integration of the tunneling

process and have incorporated some innovative individual compo-

nents. Many of the technical and economic advances in European

and particularly in Austrian and German tunneling practices can

be traced to the use of observational methods as an approach to-

ward cost effective tunnel design and construction.

The goal of this research is far-reaching and encompassing.

For this reason, and because the results not only benefit

transportation tunneling, it is a long-term effort supported by

two government agencies: UMTA and the National Science Foundation

(NSF) . In order to have identifiable research tasks under each

sponsorship, the technical or design oriented aspects are supported

by UMTA under the aforementioned contract (period January 15,

1978 - August 1979), while the operational and contractual aspects.

2
as well as the overall effort to improve the tunneling process,

are funded by NSF under grant DAR-7709116 (period March 15, 1978 -

March, 1979) . Since the UMTA-sponsored research started first,

it included an information collection trip to Austria and Germany.

This trip produced information on the technical aspects of

European tunneling practice, and on the overall tunneling process

for transportation tunnels in both urban and mountainous environ-

ments; thus, this trip served as a basis for the NSF-funded project.

The purpose of this report, which documents the results of

the UMTA-supported research, is to provide the tunneling profes-

sion with improved practical tools in the technical or design

area. These design tools provide more accurate representations

of the ground-structure interaction in tunneling. They range

from simple analytical and empirical methods to sophisticated

finite element techniques, thus reflecting the practical dif-

ferences in input information and output requirements of the

various phases in the design-construction process. Also, some

findings on integration of the tunneling process are reported,

and practitioners are given detailed descriptions and evaluations

of tunneling practice in Austria and Germany. Throughout the

report, contributions to a better understanding of ground-

structure interaction, including several specific new

developments, are made.

3
 Volume 1 describes a simplified analysis method, geared
toward hand calculations, that incorporates the effects of

three of the most significant factors influencing the ground-

structure behavior: (1) the relative stiffnesses of the ground

and support, (2) the distance between the face and the point

of support installation, and (3) yielding of the ground mass.

The method is intended for use in preliminary design, for

parametric studies in final design, and for updating the

design during construction.

Volum.e 2 addresses a specific aspect of ground-structure

interaction, the ill-understood and controversial topics of

ground yielding and loosening, and reports new findings in

this area. It is based on previous research, on the knowledge

gained during the developmcent of the simplified analysis and

the application of the more sophisticated finite element

techniques, and on specific studies of strain softening ground

behavior. It should be emphasized, however, that many develop-

ments and discussions in Volume 1 and 3 (introduced below)

also contribute to improved understanding of the ground-structure

interaction around tunnels.

Volume 3 reports the application of the three-dimensional

finite element program ADINA to the Peachtree Center Station

in Atlanta. In particular, the excavation sequences for the

4
Atlanta research chamber and the main station cavern are simula-

ted and the instrument readings are predicted. This application

of the three-dimensional finite element model is intended to

illustrate some of the advantages and limitations of such

methods when used for design or to compare predicted with

measured movements. The latter objective has not yet been

achieved due to problems with the monitoring program.

Volume 4 documents and evaluates the information gathered

in Austria and Germany. A large number of transmountain tunnels

(Austria) and subway tunnel sections (Germany) were visited,

and many discussions were held with owner-authorities, design

firms, and contractors. The report provides a detailed review

of technical, operational, and contractual aspects of Austrian

and German tunneling practice and discusses the reasons why tun-

neling there is often more economical and technically innovative

than it is in the U.S.

Volume 5 evaluates empirical methods in tunneling. This

evaluation provides the practioner with a guide to the advantages,

limitations, and especially the ranges of applicability for the

best known empirical methods. In addition, improvements are

made to arrive at empirical approaches best suited to observa-

tional methods.

5
 The executive summary provides the reader with

extended abstracts of each of the aforementioned volumes.

6
 2 . SIMPLIFIED ANALYSIS FOR GROUND-STRUCTURE

INTERACTION IN TUNNELING

This simple, design-oriented analysis method has been

developed for determining the loads on tunnel supports. It

is generally intended for circular tunnels with closed-ring

primary support systems in ground masses that can, for practical

purposes, be treated as time-independent continua. The method

is also aimed at tunnels that are excavated full-face under

free air at depths greater than about tv;o tunnel diameters.

Examples of practical tunneling situations suitable for analysis

using the simplified method include--but are not limited to--

circular steel ribs, prefabricated segmented liners, con-

tinuous shotcrete supports in soil, heavily jointed rock, and

massive rock formations.

It is doubtful that the complex interrelationships

among the nearly countless variables in any tunneling

problem can ever be rigorously analyzed, even using the most

sophisticated numerical techniques. As an alternative

approach, the simplified method focuses on the essential

elements of very complicated physical phenomena in order to

isolate the few major factors that have an overriding influence

on the support loads. These three major factors are explicitly

considered: (1) the relative stiffness of the support and

7
the ground mass, (2) the spatial lag or delay of support

construction behind the tunnel face, and (3) the yielding of

the ground mass as its shear strength is exceeded. The method

is structured where each of these major factors is handled

in a separate step, with subsequent steps building on the

preceding ones.

Step 1. Relative Stiffness

The effects of relative support stiffness on the tunnel

support loads is incorporated into the simplified analysis

through closed-form relative stiffness solutions. These

solutions assume: (1) plane strain conditions, (2) elastic

behavior for the ground and support, and (3) simultaneous

excavation and support of the tunnel. They explicitly consider

the effects of the support stiffness and the ground stress

conditions (the lateral in situ stress ratio, in particular)

on the support thrusts, moments, and displacements at all

points around the circumference of the opening. However,

they do not consider the effects of support delay and ground

yielding^ these effects are treated in subsequent steps of

the analysis. The loads calculated from the relative stiff-

ness solutions can be considered as the "basic" loads in the

sim^plified method.

The input parameters required for this step of the

simplified analysis method are:

8
 . .

- Elastic constants (Modulus and Poisson's

ratio) for the ground and for the support.

- Geometry of the support (radius, cross-

sectional area and moment of inertia)

- In situ stress field (vertical and lateral

stresses)

Some of these parameters are used to calculate the dimension-

less stiffness ratios for the ground-support system, the com-

pressibility ratio, C*, (a measure of the circumferential stiff-

ness) , and the flexibility ratio , F* , (a measure of the flexural

stiffness) . The circumferential support thrusts are primarily

related to C* and the support bending moments, to F*; decreas-

ing values for C* and F* imply an increasingly stiffen support

(or softer ground)

The analysis considers the two limiting conditions "full-

slip" and "no-slip" for the shear transfer at the ground- support

interface. Also, and very importantly, the correct "excavation

unloading" condition for tunnels is incorporated into the solu-

tion, i.e., the tunnel is excavated in an initially stressed

ground mass. The results from the analysis are the support

thrusts, bending moments, and displacements at any point around

the circumference of the tunnel. These quantities can be easily

calculated either by hand or with a programmable pocket or desk-

top calculator; program documentation for a Hewlett-Packard HP97

calculator is provided.

9
 From parametric studies, the following conclusions can be

drawn regarding the sensitivity of the thrusts and moments to

variations in the input parameters:

(1) Thrust is strongly dependent on C* only within the

range 0.05 < C* < 50 and is relatively insensi-

tive to variations of F* (for practical values of

F*)

(2) Moments are near zero for F* > 100 and are insensi-

tive to variations of C*.

(3) For excavation unloading conditions, both thrust

and moments are insensitive to variations in

Poisson's ratio for the ground.

(4) Thrusts and moments vary linearly with K.

The overall bending moments due to the nonuniform in situ stresses

are usually small; support moments due to factors like faulty

erection, inadequate blocking, or incomplete grouting are usually

more critical. The tunnel designer must check for these "local"

critical moments in his particular support system.

Step 2. Support Delay

Support delay refers to the spatial lag of support construc-

tion behind the tunnel face. Increasing the support delay usu-

ally decreases the tunnel support loads. The reduction of the

support load can be represented by the multiplicative support

delay factor

10
in which T^, are the thrusts and moments derived with the

relative stiffness solution in Step 1 and T

2
/ M 2 are thrusts

and moments corrected for support delay. Step 2 of the analysis.

The primary input parameter in this step is the normalized

support delay length, L^/R. The delay length is defined as

the distance between the face and the midpoint of the leading

active support element.

The relationship between and L^/R was determined from

parametric finite element studies for K=1 stress conditions:

X^ = 0.98 - 0.57 (L^R)

Based on physical consideration, X^ varies between the extremes

of 1.0 (no support delay effect) and 0 (full support delay ef-

fect) . The equation can be used with reasonable accuracy for

values of b^/R greater than approximately 0.15 and less than

about 1.5.

If in addition to the ground movements arising from the

support delay there are other ground movements that develop

before the support contacts the ground (e.g., from an incom-

pletely grouted tailpiece void behind a shield) , these must

also be incorporated into the delay factor X^. Defining u^' as

these additional pre-support ground movements and u^ as the

total elastic radial displacement of the wall of an unlined

tunnel, the support delay correction factor is expressed as:

L
d
0.98 - 0.57
d R

11
 .

The factor is still limited to the range 0 £ £ 1.

The sensitivity of A^ to K^^l stress conditions was investi-

gated using a modified plane strain relative stiffness solution

that approximately models the effects of support delay. Based

on the analyses for both K=1 and conditions some general

conclusions can be drawn about the applicability of A^:

1) A^ can be used with reasonable accuracy to

modify both support thrusts and overall bending

moments

2) A^ is substantially independent of the relative

support stiffness, as expressed by C* and F*.

3) A^ is independent of the lateral stress ratio, K.

The support delay correction factor A^ is a very important

parameter in the analysis. Unfortunately, it is also a parameter

that is very difficult to calculate in practice. The support de-

lay length in a real tunnel is usually only approximately

known; small variations in blocking or grouting procedures, for

example, may significantly move the point at which the support

becomes "active" — i.e., comes into contact with the ground.

Step 3 Ground Yielding

Yielding in the ground mass surrounding the tunnel tends to

increase the support loads. The quantitative effects of this

yielding were investigated using both axisymmetric and plane

stress elas to-plastic finite element analyses. A ground yield

12
 ;

factor was devised to represent the effects of ground yield-

ing:

,
A = P*
s
y ^s

The term P' is the equilibrium support pressure in the elastic

s

ground case, reduced for the effect of support delay, and P* is

the support pressure in the yielding case; P* includes the ef-

fects of both support delay and ground yielding . The yield factor

A^ has a physical lower limit of 1 (corresponding to completely

elastic ground behavior) but no upper bound.

The third step in the simplified analysis method, then, is

the calculation of A^. This factor is primarily a function of

the strength parameters of the ground mass, although it is also

indirectly dependent upon the support delay and the relative

support stiffness. The input parameters required for this step

are

- The input parameters for Step 1 (elastic constants

for support and ground, geometry of support, and

stress state)
- The support delay factor from Step 2

- Strength (or yield) parameters for the ground,

e.g., c (cohesive strength parameter) and ({)

(frictional strength parameter)

The yield factor can be calculated from approximate plane strain

plasticity solutions. These solutions can be easily programmed

on a pocket or desk-top caluclator; program documentation for

13
a Hewlett-Packard Model HP97 is provided.

Because of the interactions among the relative support

stiffness, support delay, and ground yielding variables, X

will be indirectly dependent on C* (and F*) and this fact

was substantiated by the findings from the axisymmetric finite

element
1) analyses. The primary purpose of the plane stress fin-

ite element analyses was to investigate the sensitivity of A^

to variations in the lateral stress ratio K. The principal con

elusions from these analyses were:

2)
For small to moderate amounts of yielding (A^

less than or equal to 2, approximately) , A^ is

reasonably insensitive to variations in K between

3)
0.5 and 1.5. As the level of yielding increases,
4)
this range for K decreases; at very high levels of

yielding, A^ can only be calculated for K=l.

The insensitivity of A^ to variations in K is

the same whether the yielding is due to low ground

strength properties or a long support delay, or both.

A^ is not significantly affected by the shear

transfer conditions at the ground-support interface.

Stiff supports will generally produce values for

Ay that are small and reasonably independent of K.

The calculations of A^ represents the final step in the

simplified analysis method. The support thrust, which is pro-

portional to the magnitude of the pressure P*, is calculated as

s

T* = A A. T
y d

14
In this equation T is the "basic" thrust from Step 1; it in-

cludes the effects of relative support stiffness and the lateral

stress ratio . The term is the support delay factor from

Step 2; this factor, which is independent of the support stiff-

ness and the lateral stress ratio, modifies T for the effects

of support delay . The last term, A^, is the ground yield factor

from Step 3; this factor, which is a function of the ground

strength, relative support stiffness, and support delay but

which is independent of the lateral stress ratio, at least for

small amounts of yielding, modifies A^T for the effects of ground

yielding . The term T* is then the final design thrust.

In order to verify the accuracy of the simplified analysis

method, it was applied to five tunnel projects (a total of 11

tunnel sections) reported in the literature in which the actual

support loads were measured during construction. These cases

span a representative range of tunneling situations for which

the simplified analysis is applicable. Two of the case studies

are tunnels in soft rock and three are in soil (various types of

clay). The supports in the case studies include steel ribs, pre-

cast concrete segments, and cast iron segments; several hand and

machine excavation systems are also represented.

Given the streamlined, approximate nature of the method, it

was surprisingly accurate in predicting the support loads measured

in the case studies. The errors in the predicted support thrusts

ranged between the extremes of -79% (unconservative) and +62%

(conservative) . However, for 7 of the 11 tunnel sections anal-

15
yzed, the errors varied between the more limited bounds of

±30%. On the whole, there was a slight conservative bias in

the predictions.

In all, the simplified analysis method is an appropriate

and accurate design technique in many types of tunneling situ-

ations. It reduces the tunneling problem to its bare essentials,

capturing the fundamental aspects in a few variables. Since

the method does require only a few, relatively easily determined

input parameters, it is very simple to apply. No large computers

are needed, as they are for the more sophisticated numerical

techniques; the support loads can be readily computed using sim-

ple desk-top pocket programmable calculators. However, the en-

gineer must be aware of the deviations between the actual and

the assumed conditions and, as in any practical engineering

problem, judgement and ingenuity must be exercised when applying

this simplified analysis to an actual tunnel design. Neverthe-

less, since the support loads can be computed with little ef-

fort, the simplified analysis is perfectly suited for parametric

sensitivity studies when designing for uncertain conditions.

This computational ease, coupled with accuracy sufficient for

any practical tunnel analysis, makes the simplified analysis

method a valuable and effective design tool.

16
 .

3. ASPECTS OF YIELDING IN GROUND-STRUCTURE INTERACTION

The present research on improved design procedures for tun-

nel supports, as well as preceding and parallel research on

underground construction and rock and soil behavior, has led to

a better understanding of ground-structure interaction. Studies

on ground-structure interaction were an integral part of the

review of observational and analytical methods for tunneling

reported previously (Einstein et al., 1977). These studies were

intensified during the development of the simplified analytical

method (Vol. 1 of this report), the three dimensional

numberical analysis of the Atlanta Research Chamber (Vol. 3),

and the review and extension of empirical methods (Vol. 5)

They also benefited substantially from the review of tunnel

design and construction procedures in Austria and Germany,

since the application of the principles of ground-structure

interaction has been a significant factor responsible for the

advances in tunneling practice in these countries.

Improved understanding of ground-structure interaction

has led to and is implicitly contained in the specific

technical developments presented in this series of reports.

The design profession may benefit, however, from a more basic

review of ground-structure interaction in tunneling, particu-

larly if a better understanding of some of the complex phenomena

and an insight into some controversial aspects can be provided.

Previous work (Chapter 8 in Einstein et al., 1977) attempted

17
to review the entire ground-structure interaction phenomena

and to highlight the most important factors, which were then

taken into consideration in the development of analytical and

empirical methods (Vols. 1 and 5 of this report). The dis-

cussion in this volume concentrates on a particularly complex

and often ill understood aspect of ground-structure inter-

action, that of ground yielding. In addition to providing the

reader with some basic concepts that will increase his under-

standing of ground yielding, this report will also describe and

compare analytical solutions for plastic ground behavior.

The excavation of an underground opening leads to an in-

crease of the shear stresses acting on elements at or near

the surface of the opening. If these increased stresses lead

to yielding, two major types of ground behavior can be visualized.

As the ground is excavated, the stresses acting on the circum-

ference of the future opening (the virtual "internal pressure")

approach zero. The corresponding strains may increase beyond

the elastic limit, eventually reaching a large but finite value.

The unsupported opening will be stable, albeit with possibly

unacceptably large displacements. Alternatively, the strains

may initially increase beyond the elastic limit as described

above, but instead of stabilizing at some point, they will in-

crease even if the stress at the circumference (internal

pressure) is kept constant or actually raised. This phenomenon

is commonly referred to as loosening behavior , and it has been

the subject of intense discussion and dispute among tunnel de-

signers .

18
 This report attempts to clarify this problem by first

examining possible conceptual mechanisms and then by applying

analytical methods to study their quantitative effects. The

most obvious mechanism is that of strain softening ground behavior.

Such behavior may, in some circumstances, lead to increasing in-

ternal pressure with increasing strains, i.e., to loosening.

However, it can be shown conceptually that the occurrence or

non-occurrence of loosening behavior depends strongly on the

assumed material behavior, i.e., the precise relation between

strength deterioration and strain. As a consequence, some of

the existing analytical methods that model loosening are incon-

sistent. We therefore developed a closed-form solution for

predicting displacements, strains and stresses around a circular

opening in a strain softening medium. The solution is based on

the following idealizations: 1) plane strain; 2) in situ prin-

cipal stresses are equal, 3) gravity effects are neglected;

4) material obeys the Tresca yield criterion; and 5) the ground

mass behavior can be represented by a linear three-segment stress-

strain relationship, namely elastic, strain softening and per-

fectly plastic segments. Extensive parametric studies with this

analytical method did not produce any loosening behavior. The

often advanced opinion that strain softening ground necessarily

leads to loosening is thus incorrect; nevertheless, it still

cannot be concluded that strain softening never leads to loosen-

ing .

Loosening can also be examined using the concept of a

stable ground arch around the tunnel. Given a certain displace-

ment at the opening circumference, a ground arch will form at

19
 .

a certain distance and the ground between the arch and the

opening can be held in place by a certain counterpressure.

If the displacement at the opening circumference is increased,

the ground arch will form at a greater distance from the opening,

more ground between this arch and the opening has to be held in

place, and the counterpressure has to increase — i.e., loosen-

ing occurs. The concept is appealing and can be substantiated

by some simple models like Rabcewicz's shear body and an ex-

tension of Terzaghi's arching theory. Analytical approaches

that model the stable ground arch mechanisms described above

do not seem to exist; in practical terms, the only situation

that approaches such a mechanism is the behavior of a blocky

rock mass in a low stress field.

The purpose of the above-mentioned analysis method for

strain softening ground goes beyond the examination of loosen-

ing behavior. It, and the graphs developed from the parametric

study, is intended to give the designer a convenient tool to

investigate various characteristics of strain softening be-

havior, for instance the magnitude of the stress drop. Using

the analysis, the designer can easily develop ground charac-

teristic curves and determine the extent of yield zones, in-

formation that can then be used to determine ground-structure

interaction as it is described in Volume 1 of this report

series

The reader will thus obtain from this volume a better

understanding of ground-structure interaction, especially of

some aspects of ground yielding. In addition, an analytical

tool for treating strain softening behavior is provided.

20
 4 . FINITE ELEMENT ANALYSIS OF THE PEACHTREE

CENTER STATION IN ATLANTA

In this volume, the application of the three-dimensional,

3-D, finite element program ADINA to the Peachtree Center Station

in Atlanta is described. The study is intended to shed some

light on the practical usefulness of 3-D finite element methods.

In particular, it was intended to show where such methods have

an advantage over simpler ones and to determine what efforts

and costs are involved. To be practically meaningful, such a

study had to be conducted for an actual case, even if this meant

a loss of generality. The Atlanta Research Chamber at the

Peachtree Center Station provided a good opportunity for this

study for three reasons: (1) it involved a relatively complex

geometry; (2) the ground conditions were well established through

thorough testing; and (3) the extensive monitoring program •

would make comparisons between predicted and actual behavior

possible. The prediction of the ground movements and stresses

caused by the enlargement of the pilot tunnel to form the test

chamber and by the excavation of the main station cavern were

of particular interest in the study. Plots of calculated

stresses and deformations were generated in a form suitable

for practical comparisons with instrument readings.

21
 Before starting the analysis, some slight modifications

of ADINA were necessary, and pre- and post- processor programs

with plotting routines were developed to construct the finite

element mesh and to produce the output in practically usable

form. A three-dimensional mesh consisting of 1093 3-D iso-

parametric elements and 2915 nodal points was devised to model

the geometry of the research chamber, the running tunnels,

and the adjacent part of the main cavern, including two major

side drifts. Elastic isotropic material properties and an

anisotropic stress state were found to reasonably approximate

the real conditions. The finite element model simulated the

research chamber enlargement and the main station excavation;

the ground deformations and stresses were determined at each

step in this simulation. The major findings regarding the

ground deformations are:

The rock mass displacements are very small. At the

walls of the main station cavern, the rock displacements reach

a maximum of 0.025 inches radially and 0.06 inches longitudinally.

At the walls of the research chamber, the respective movements

are 0.0043 and 0.018 inches. These small displacements are a

consequence of the low deformability of the rock and the

comparatively low in situ stresses.

The excavation of the main station cavern induces rela-

tively large longitudinal movements in the rock surrounding the

research chamber. The research chamber does not tilt measurably

22
in the longitudinal plane as a consequence of the main station

cavern excavation; instead, it settles uniformly.

The comparisons between rock displacements measured

during the excavation of the pilot tunnel and movements predicted

by this analysis are unsatisfactory. The discrepancies in

these results may be due either to the use of incorrect rock

mass properties and in situ stresses in the analysis or to

malfunctions of the field instruments resulting from excessive

blasting. More field measurements are needed to determine the

exact causes.

Comparisons between the predicted and measured move-

ments due to the research chamber enlargement cannot be made

for two reasons: (1) the excavation sequence employed by the

contractor was different from the one originally planned by

the designer and assumed in this analysis, and (2) no measure-

ments are yet available from the research chamber instruments.

Regarding the stress state in the rock mass, the 3-D analysis

shows that:

The maximum compressive stresses around the openings

are low, relative to the strength of the rock mass.

A tensile stress zone forms at the south wall of the

main station cavern which may cause problems, such as excessive

overbreak, during construction of the running tunnels at this

location. In addition, radial tensile stresses form at the

flat parts of the main station cavern's crown, sidewall,

and invert.
23
 In the transverse cross-sections through the research

chamber, the major and minor principal stresses are approxi-

mately tangential and radial with a maximum tangential stress

concentration (relative to the in situ vertical stress) of

about two.

In addition to the specific predictions discussed

above, this study has shed some light on the use of three-

dimensional finite element analysis as a tool in the analysis

and design of underground structures. Several features of

the analysis and results could not have been modelled or

predicted using two-dimensional finite element model:

1. The complex geometry of the station.

2. The longitudinal in situ stresses. In the 2-D

model, these stresses are dictated by the plane strain assump-

tion. In the 3-D model, the measured values of stresses are

incorporated into the analysis. These longitudinal stresses

have an overriding effect on the instrument movements, and

they are also largely responsible for the tensile stresses

at the south wall of the main station cavern.

3. The excavation sequence. Because of the plane

strain assumption, the 2-D model cannot include any effects

from the main station cavern excavation.

4. Inclinometer movements. Because the inclinometers

are approximately located in the longitudinal plane of symmetry,

a 2-D model is incapable of predicting their movements.

24
 The effectiveness of the 3-D finite element analysis

for design is constrained by the significant time and high

cost required to prepare the input, conduct the analysis, and

interpret the results. However, this cost is not usually

a major limitation when compared to the total cost of a sub-

way station. For example, for this case it is estimated

that the computer and manpower cost of a 3-D analysis would

represent a maximum of 0.25% of the total project cost.

In cases where ground-structure interaction in complex geometries

must be modelled, the potential design savings will very

likely outweigh these analysis costs.

25/26
 5 . TUNNELING PRACTICES IN AUSTRIA AND GERMANY

Information on European tunneling practice has been

collected and evaluated and is documented in Volume 5.

Since the documented material is so voluminous, an extended

summary is necessary.

The information was received from the literature

obtained during a trip to Austria and Germany in January

1978, and from professional contacts in Europe. Without

the generous and continuing assistance of many professional

colleagues in these countries, it would not have been possible

to write this report. The assembled information covers cost

(economical), contractual, and technical aspects of tunneling.

This report, as well as this summary, will be structured along

these lines. However, as seen in this summary and the report,

there is significant interaction amongst these aspects and

the role that integration of these aspects plays is one of

the major findings of this study.

Cost information consists of: general cost data and costs

of specific tunnels in Austria and Germany; studies conducted
in Germany in which various effects on subway tunneling costs

were investigated; similar, but less extensive, investigations

on the data collected by the authors and a cost comparison

between the U.S. and Europe,

The major factors affecting construction costs of shallow

27
and deep tunnels are: ground conditions, material availability,

site accessability and other restrictions, market conditions,

and size (cross-sectional area) of the tunnel. Ground conditions

affect shallow and deep tunnels differently; consideration of

surface effects in shallow tunnels leads to a more conservative

design and thus higher costs than for deep lying tunnels in the

same ground. The effect of market conditions has been illus-

trated dramatically during the past few years. Tunnels of the

same size and located in similar, or less favorable, ground

have recently been bid, or completed, at the same or up to

20% lower prices than five years ago, mainly due to increasing

competition in a somewhat depressed economy. Very important

in this respect is the possibility that contractors can submit

alternate proposals in addition to, or instead of, bidding

for the official design. This permits the contractor to adapt

design and construction to his expertise, crew capabilities,

and equipment, and thus optimize his approach. One result, which

is probably due to a combination of market conditions, the

possibility of submitting alternate proposals, and the

restrictions in inner cities, is the price equivalence of open-

cut and mined tunnels in such locations.

Tunneling costs per lineal unit in the U.S. seem to be

between 30 and 80% higher than in Europe for mined tunnels

and 100 to 300% for stations; the cost per unit volume is

between 50 and 100% greater. Major causes for this difference

seem to be the smaller support quantities, smaller crew sizes.

28
 .

flexible equipment, and a substantial, if not full, integra-

tion of design and construction.

Contractual information is based on a detailed review of

contractual standards and procedures in Austria and Germany

and of contractual setups at individual projects. This is

supplemented by opinions of owners, engineers, and contractors

and a consideration of recent and future developments in the

contractual area. In addition, special attention is given to

payment and incentive aspects, as well as to dispute settlement

procedures

The first impression that one gains in reviewing contractual

standards and procedures in Austria and Germany is that the

recommendations of the report "Better Contracting for Under-

ground Construction" (National Academy of Sciences, 1974) are

standard practice; even more important, however, is how these

procedures are realized in all phases of design and construc-

tion. Most notable are the detailed contractual and technical

standards according to which the specifications, schedules,

and bids have to be prepared. Although there are national

standards, the individual owners have the possibilities to

modify them. The detailed contractual and technical provis-

ions (amongst the technical provisions there is frequently a

detailed ground classification) facilitate adaptation during

construction. These detailed provisions, together with the

usual unit price bidding, make it possible to cover all

contingencies by assembling bid items corresponding to the

29
 .

actual design construction procedure for the particular

contractor, ground conditions, and other boundary conditions.

Thus, rather than making the design-construction procedure

rigid, the detailed setup seems to make it more flexible.

In most cases, these procedures also spell out in detail how

disputes have to be settled; usually through mediation-

arbitration or through an administrative appeal process. The

inherent flexibility, the dispute settlement procedures, and

a very careful legal preparation, and naturally also the

riskier European trial procedures*, make court cases very

rare (for instance, there were only two court cases in Munich^

where 30 km of subway tunnels have been constructed so far)

Payment is usually for work performed based on a very

detailed price system. Since "all possible" conditions have

been anticipated in the bid and contract documents, payment

can be uniquely established. The aforementioned dispute

settlement procedures take care of most remaining disagree-

ments. It is important to realize that the payment system is

not a simple cost reimbursement. Considerable incentives are

built-in to make it financially interesting for the contractor

to work carefully, particularly by letting the contractor

reap the full benefit of effective construction procedures.

Contractors have significant possibilities in this respect

due to the partial or full integration of design and construction.

*
The losing party has to pay the court and trial
costs of the winning party.

30
 Technical information consists of general information on

design philosophy and construction procedures and of detailed

information on analytical and empirical methods and on design

aspects. In addition, crew and equipment aspects in con-

struction-planning and execution are also discussed. A

substantial part of the technical information involves data

on geotechnical conditions and related performance of tunnels.

Most important in the technical areas is the fact that, in one

sense or another, only alternate proposals are actually

constructed and that these proposals include considerable

changes compared to the owner's official design and construction

procedures. The extreme consequence is the situation in Munich

where the specifications basically provide only the alignment

and required cross-section (and references to the detailed

contractual and technical standards) and where the contractor

performs the detailed design and construction planning. This

integrated design-construction approach is facilitated by the

usual in-house design capabilities of the contractor, and it

assures the most effective use of the contractor's expertise,

crews, and equipment. The liberal use of alternate proposals

or even full design construction approaches does not involve

technical risks since any design has to be examined and approved

by a designated inspection engineer. Since the official

design, which is usually prepared by the owner's staff, also has to

be approved in this manner, no extra time or cost is involved

in accepting alternate proposals.

31
 in addition to the integration of design and construction,

it is very significant that design is flexible and construction

procedures are adaptable. In this context, ground classifica-

tion systems that relate geotechnical conditions to design and

construction procedures are frequently used. The adaptation is

practically achieved by assigning appropriate ground classes.

Flexible design and adaptable construction procedures lead to

considerable economic benefits where there is a great variety

of tunnel shapes and sizes in shallow tunnels and in uncertain

ground conditions in deep tunnels. A prerequisite for the

successful use of design flexibility and construction adapta-

tion, and proper conservation of safety aspects, is the ex-

tensive use of performance monitoring during construction. The

advantage of flexibility is reflected in the fact that, at the

time of information collection in Munich, the somewhat less

flexible shield tunneling is only economical if the tunnel

length is greater than 1400 m. Below this length, the possibility

of having several points of attack and the liberal use of mobile

equipment in non-shield tunnels overcome the higher production

rates of a semi- or fully mechanized shield. The emphasis on

flexibility is also obvious in the preference of several pieces

of smaller equipment instead of a single, highly productive

piece. This permits the contractor to easily adapt the pro-

cedures to varying conditions and to have redundancy in his

operations that increases the overall productivity. The

technical and related contractual conditions are not only

32
reflected in the design and construction approaches in general,

but also in the fact that practically every new tunnel or

tunnel section has innovative design and construction features.

All this is only possible with the generally outstanding

quality of the personnel, from the owner's staff to the miners

in the tunnel.

As far as analysis methods are concerned, some interesting

details of finite element methods have been developed in Germany.

In general, the analysis methods are not advanced and they have

not been fully integrated in the design-construction approaches.

This can be explained by the presence of personnel on all levels

who have a great amount of experience and are using it effect-

ively. Nevertheless, a trend toward greater use of analysis

can be observed.

Finally, in the context of technical information, it was

possible to receive two large sets of geotechnical and

performance data from the Tauern and Arlberg Tunnels.

Together with data from several mountain and subway tunnels^

they will serve as a basis for improving empirical design

approaches and will make possible realistic sensitivity

studies with analytical methods.

What has been summarized above is discussed in detail in

the text and the appendixes of this volume. It seems useful

to conclude this summary by emphasizing the major findings on

tunneling practice in Austria and Germanyo

- Technical, contractual, and cost (economic) aspects

33
are closely related. The lower tunneling costs

compared to the U.S. are to a significant extent due

to contractual setups that make technical innovations

and contractual changes easily possible.

Most important amongst the contractual conditions seems

to be the possibility of submitting alternate proposals

that allow the contractor to use his expertise,

personnel, and equipment most effectively.

The contractual and technical documentation as well as

the bids are so detailed that most conditions can be

anticipated, work performed correspondingly, and

payments made without dispute.

The three items above show that flexibility in con-

tractual procedures is the underlying philosophy that

in turn makes flexibility in design and construction

possible. This flexibility exists on many levels from

the overall design to the adaptation of detailed con-

struction procedures; in the latter it is frequently

based on well developed ground classification systems.

The technical and contractual flexibility is the pre-

requisite for the most important finding, the integra-

tion of design and construction. The most significant

cost and time savings and technical innovations have

been achieved by fully considering construction aspects

in design. This is reflected in the technical features

of design and construction as well as in the contractual

setups of particular projects.

34
 6 . EMPIRICAL METHODS FOR ROCK TUNNELING
REVIEW AND RECOMMENDATIONS

The complexity of ground and structure behavior and

particularly of ground-structure interaction in tunneling

makes it difficult to understand the underlying principles.

Consequently behavioral models are only approximate and analy-

tical predictions that have to be based on such models are

either generally inaccurate or accurate over very limited

ranges of application. Empirical methods that avoid the use

of an explicit model by relating ground conditions to observed

prototype behavior have thus always played a major role in

tunneling. As a matter of fact, the relative significance of

empirical methods in tunneling is much greater than in any

other civil engineering area. Interestingly, the last few

years has seen the creation of additional empirical methods.

As knowledge and understanding of tunnel behavior increases

analytical methods will become more accurate provided the

input parameters can be determined at a corresponding level.

However, given the usual limitation in parameter determination

it can be foreseen that empirical methods will continue to be

used in tunnel design albeit at a reduced level.

In reviewing tunnel design practice and establishing areas

that need improvement it became readily apparent that empirical

methods had to be considered. There is understandable uncertainty

35
amongst designers and contractors as to what method to use.

Comparative predictions made with different methods lead to

substantially different results. Some methods require a

minimum of easily determinable parameters and others require

a large set of relatively complex ones. Some methods are based

on qualitative descriptions of the soil and others are highly

quantitative. Subjective assessment by the user may be required

or not. This list could be further extended to show that - decid-

ing on a suitable empirical method may be a difficult task. An

additional problem arises in the context of this research project

and report series. With the exception of the empirical procedure

for the New Austrian Tunneling Method (NATM) none of the methods

fully integrate design and construction.

It was decided that the tunneling profession could gain

considerably from a systematic review of the best known empirical

methods for rock tunneling. The review describes the correct

application procedure of each method; it compares each empirical

method with a set of carefully determined and structured criteria

and based thereupon identifies advantages and limitations of each

method. This review is the main objective of the report and

can serve as a guideline for tunnel designers and contractors.

Nevertheless, recommendations are also made as to what require-

ments an empirical method should fulfill if it is to be used

in an observational (adaptable) tunneling procedure - since

such procedures provide at the present time an optimum inte-

gration of design and construction. Based on these requirements

and the preceding review, modifications of existing empirical

methods can be recommended.

36
 Empirical methods in tunneling all incorporate the

same basic structure. Ground conditions are quantitatively

or qualitatively characterized in form of niv'erical parameters

and their states or in form of verbal descriptors. Usually

combinations of verbal descriptors or combinations of

parameter states are grouped into ground classes -

thus the term ground classification is often used in this context.

These combinations of quantitative or verbal descriptors are

then related to support (and excavation) requirements either

directly or via rock load (support pressure).

A direct relation is, for instance, between a ground class and

a certain support type and quantity (e.g., steel sets, bolts,

shotcrete, concrete) ; the other involves applying the rock

load in standard structural analysis to obtain support quantities

and dimensions.

It is possible, and has been illustrated in this report, to estab-

lish criteria that any correct and practical empirical method (be

it for tunneling or otherwise) should satisfy and, built there-

upon, to formulate requirements that empirical methods in

tunneling should fulfill. Specifically empirical methods have

to fulfill several user requirements :

they should be economical and safe ; i.e.

supports should not be overly conservative nor should they

fail, ideally the factor of safety should be known

- they should be generally applicable and robus t ; i.e.,

a method should be applicable to a wide range of conditions

37
or as a minimum clearly define to what range it can be applied.

Robustness means that the method should be insensitive to

vagaries in judgement.
- the parameters or descriptors should be readily deter -

minable. Before construction accessibility is limited but time

is not; during construction the opposite is usually true.

Methods can only be based on parameters that can be obtained

under these restrictive conditions.

In addition to satisfying user requirements empirical methods

have to be basically correct, they have to fulfill requirements

regarding methodology and derivation :

- The underlying models should be accurate. Although not

based on explicit formal models (in contrast to analytical

methods) empirical methods rely on an implicit model. Such

a model should represent the inherent variability of ground and

construction processes as closely as possible. However, there

are trade-offs, to be made. A detailed model is more accurate

but requires more parameters than a simplified one. Since

measurement and inference of parameters involves (statistical)

uncertainty, the more (interdependent) parameters there are,

the greater the uncertainty in inferences of each. Model

uncertainty and parameter uncertainty have to be balanced.

- Empirical methods have to be representative and complete .

If a method is based on cases in which a parameter was pre-

dominant generalization may be erroneous. All relevant

influence factors have to be considered.

38
 .

The sub j ective character of empirical methods has to

be recognized. Because models are abstractions of reality and

because parameter estimates involve uncertainty no procedure

for establishing support requirements can be "objective", they

are all substantially subjective. Subjectivity may enter mostly

on the part of the user, for instance in classifying geology,

or it may enter on the part of the developer of the method in

making hypotheses in test case evaluation (for instance assump-

tions underlying regression analysis) . Subjectivity cannot

be avoided and the method should clearly state where subjectivity

occurs

These requirements make it possible to assess existing

empirical methods. The review of empirical methods for rock

tunneling, which is the main purpose of this report considers

the methods by Ritter, Bierbaumer, Stini, Terzaghi (Proctor

and White), Deere, Barton et al., Bieniawski , Wickham

et al. and by Louis and Franklin. For each method an extensive

guideline is given, to serve two main purposes, 1) to set the

stage for the review and 2) to provide potential users with

information on how the method should be used. This latter

point is practically very important since many methods are

applied in an inconsistent and incorrect manner. The guide-

line does not intend to replace the background information

given by the developers of the method, it rather emphasizes

the necessity to carefully study the basis of whatever method

one wants to use. Related to these guidelines and providing

39
the major substance for the review is a detailed description

of the development and underlying philosophy of each method.

Many aspects can only be understood by examing in detail the

thoughts and assumptions that went into the development. A

critique is then made of each method by comparing the afore-

mentioned requirements with the capabilities and the basis of

each method.

A complete review cannot only rely on an abstract considera-

tion of criteria but has to include a comparative application

of the methods. Example applications performed by the authors

of this report as well as from the literature are used for this

purpose. The result of the review is a detailed description

of advantages and disadvantages of each existing empirical

method and a series of general conclusions:

The result of any 'tunnel design method' should be an

optimum combination of support features and construction pro-

cedures for the particular ground conditions, opening dimensions

and use. Due to the variety and variability of influencing

factors and the interdependent character of support and con-

struction, the problem is usually very complex. The use of

analytical methods relying on prior creation of a model is

thus limited. Empirical methods that relate ground conditions

to tunnel design and construction are a very appropriate tool

substituting for or complementing analytical methods. However,

the complexity of the problem affects empirical methods also —

none of the methods completely represents all the influencing

factors. This has two consequences.

40
 .

• A particular method can only provide accurate predictions

conditions similar to those of cases by which it v/as

calibrated (base cases)

• Applying different methods to the same case will usually

lead to different predictions.

No method is generally and consistently more accurate than

others; however, optimum methods can be defined or developed

for a limited range of applications. Present methods are limited

in their consideration of construction procedures and of some

characteristics of geologic structure.

This leads to the first conclusion :

The user has to carefully study the base cases or (as a

minimum) the assumptions, comments on development and limita-

tions formulated by the developer of an empirical method. This

should be done for every application.

An accurate method should not only predict what has been

done under similar conditions but what is adequate, i.e., the

optimum support features and construction procedures. Empiri-

cal methods depend on base cases and since these were usually

designed conservatively the empirical methods will also lead

to overdesign. This can only be corrected if the degree of

overdesign in the base cases is adequately known.

The second conclusion is:

Present empirical methods over-estimate support requirements;

the degree of overestimation is usually not known.

While predictions of support requirements (dimensions,

materials) are often not accurate for the reasons discussed

41
 .

above, support pressure predictions are usually even less re-

liable. Support pressures based on design assumptions or analy-

tically backfigured from the design have often no similarity to

the actual support pressures (ground support pressure relations

have been inferred by analyses but this introduces further

questions)

The third conclusion is:

Ground-support pressure relations should not be used unless

they are based on measurements or analogous observations that

include all components of ground and structure. Ground-support

pressure relations may be necessary if support systems different

from the base cases are to be used, but the additional uncer-

tainty has to be taken into account.

Even if the three previous conclusions are taken into con-

sideration and a relatively accurate method chosen, the user

has to be aware that the prediction will not be precise. This

is due to the inherent uncertainty of the problem caused by

the variability of the influencing factors and the difficulty

determining them. Precision can be improved by setting an

optimum between parameter uncertainty. Since our knowledge

on tunnel behavior is still limited, it seems more appropriate

to rely on empirical methods whose underlying models are simple.

The fourth conclusion is:

Methods with a small number of parameters and a large data

base per parameter may provide more precise results at present

than very complex methods do.

42
 Another consideration is always the practicality of par-

ameter determination. Boreholes, outcrops and observations in

the tunnel provide substantially different information; the

time required prior to or during construction is also important.

The fifth conclusion is:

The selection of an empirical method has to reflect the

availability of information on parameters and the time limita-

tions affecting information collection.

The last conclusion relates to the subjective character of

empirical methods. The inherent uncertainty and complexity

of the tunneling problem makes empirical methods inevitably

subjective. Subjective aspects may exist in the parameter

determination by the user or in the formulation of the method

by the developer or both. One relies thus in other words to

varying degrees on one's experience or judgement and on that

of the developer. Relying on somebody elses judgement may be

wise but does not make the method "objective." Subjective

character is not unique to empirical methods. In a complex

problem like tunneling analytical methods cannot be built on

first principles alone but also involve many subjective hypotheses.

Empirical methods are useful tools if informed judgement

is applied to their use. Users of such methods might be con-

stantly aware that the methods can be improved only by

enlarging the data base and by monitoring and incorporating

actual behavior in future predictions.

Although the main purpose of this report is to provide the

tunneling profession with a review on empirical methods it

should also give some guidelines as to what empirical methods

43
are best suited for observational (adaptable) tunneling procedures.
If necessary, existing methods have to be modified to be applicable

in observational tunneling.

Observational tunneling methods integrate design and con-

struction. Pre-construction design consists of developing a

number of design alternatives for the anticipated ground condi-

tions. During construction the appropriate design alternatives

are implemented based on extensive observation of ground condi-

tions and performance monitoring. Usually the design alternatives

are further modified to optimally suit the encountered condi-

tions. In other words support and construction procedure are

physically adapted to the encountered conditions through updated

empirical (very rarely analytical) relations between ground

conditions and support-construction procedure. The updating

is often implicit and based on judgement.

With a few exceptions, notably the New Austrian Tunneling

Method, none of the existing empirical methods is readily appli-

cable to observational tunneling procedures. It is however

possible to specify requirements that empirical methods should

satisfy and to describe updating procedures for these methods,

such that they can be applied in observational tunneling. One has

to distinguish the preconstruction and construction phase. Empiri-

cal relations prior to construction have to be based on few

representative parameters (descriptors) , include an assess-

ment of uncertainty, and fulfill the standard (previously described)

requirements for empirical methods. Also the simple parameters

should be easily relatable to the more detailed information

44
 . .

obtained during construction. In this latter phase a procedure

has to be provided that updates the relations between ground

and support (construction procedures) and between ground,

support (and construction procedures), and performance. The main

requirement for the updating procedure is the observation and

identification of (now more detailed) ground parameters that

significantly affect support (construction procedure) and perfor-

mance. Based on the updated empirical relations and on observed

trends in ground conditions it is then possible to predict support

and construction procedure for the next round (excavation step)

Existing empirical methods are modified or combined to

satisfy these requirements. For the preconstruction phase

empirical methods rely now on two parameters describing ground

conditions, one representing a normalized 'rock mass strength'

(including strength of intact rock and discontinuities,

water, stress state, and to some extent attitude of discontinuities)

and one representing discontinuity spacing (including the effect

of several sets and of attitude) . These two parameters are

related to the parameters used in existing methods. However,

the "strength" parameter may require substantial judgement

at the present state of the art, if it is to be directly

determined. Performance predictions based on empirical re-

lations from the literature are also made in the preconstruction

phase

During construction it is mainly necessary to systematically

record and scan the information on ground conditions,

support and construction procedures, and performance. This

45
makes the identification of significant ground parameters,

support characteristics and performance values possible which

in turn serve to update the empirical relations. Guidelines

for this updating procedure and for the required information

collection and documentation are given in the report. It should

be emphasized that these modifications and procedures are

recommendations and not practically proven. Their main purpose

is to show that empirical methods can be modified to suit

observational procedures. This would have the great advantage

of eliminating many of the above-mentioned limitations of

empirical methods in tunnels: economy and safety, represent-

ativity, model accuracy, and correct derivation are all properly

considered through updating the empirical relations and by

adapting the tunneling process to fit actual conditions.

46
 .

7. REFERENCES

Einstein, H.H., et al. (1977), "Improved Design for Tunnel

Supports," Report Prepared for U.S. Department of
Transportation, Contract No. DOT-OS-601 36

National Academy of Sciences (1974). "Better Contracting

for Underground Construction," Report by Subcommittee
No. 4 of U.S. National Committee on Tunneling
Technology, 143 pp.

47/48
 APPENDIX

REPORT OF NEW TECHNOLOGY

The work performed under this contract, while leading

to no new technology, has led to the development of improved

practical design tools to provide more accurate

representations of the ground-structure interaction in

tunneling. They range from simple analytical and empirical

methods to sophisticated finite element techniques which

satisfy the requirements of the various phases within the

design construction process.

49
 X.
Form 3 E
FORMERLY
C T3
T5 ^
UMTA-79-49

DOT
o O
O < 18.5

0)
FORM
F ci- a
DOT
1720.;
••
a
(p
.A37