EM 1110-2-2901

30 May 97

Chapter 6 (3) Determine plausible and possible failure modes

Design Considerations including construction events, unsatisfactory long-term per-
formance, and failure to meet environmental requirements.
Examples include instability problems or groundwater
inflow during construction, corrosion or excessive wear of
6-1. Fundamental Approach to Ground ground support elements, excessive leakage (in or out), and
Support Design settlements that may cause distress to adjacent existing
structures.
a. Underground design must achieve functionality,
stability, and safety of the underground openings during (4) Design initial and final ground supports. Initial
and after construction and for as long as the underground support includes all systems that are used to maintain a
structure is expected to function. There is no recognized stable, safe opening during construction. Final supports are
U.S. standard, practice, or code for the design of under- those systems that need to maintain a functional opening for
ground structures. Many designers apply codes such as the design life of the project. Initial supports may consti-
ACI's Codes and Practices for concrete design, but these tute a part of the final supports, or they may be the
were developed for structures above ground, not for under- final support (e.g., precast segmental liner installed behind
ground structures, and only parts of these codes apply to a TBM).
underground structures.
(5) Prepare contract documents. This is the synthesis
b. Designers often approach tunnel design by search- of all design efforts and may include provisions to modify
ing for modes of failure that can be analyzed (e.g., com- construction procedures based on observations. The con-
bined bending and compression in a lining), then apply tract documents also contain all information necessary for a
them to more-or-less realistic but postulated situations competitive bidding process, and means to deal with claims
(loading of a lining). While bending and compression are and disputes.
applicable failure modes for linings, many other modes of
failure must be analyzed. In principle, all realistic modes e. The following subsections describe functional
of behavior or failure must be defined; then means by requirements of tunnels and shafts, typical and not so typi-
which these can be analyzed and mitigated must be found. cal modes of failure of tunnels and shafts, including corro-
sion and seismic effects. Selection and design of initial
c. Failure modes are modes of behavior that could be ground support are described in Chapter 7, and final lining
considered unacceptable in terms of hazard, risk to cost or selection and design in Chapter 9.
schedule during construction, environmental effect, or
long-term failure of function. For underground structures, 6-2. Functional Requirements of Tunnels
failure of function means different things for different and Shafts
kinds of structures: a certain amount of leakage in an
urban highway tunnel might be a failure of function, while Most USACE tunnels are built for water conveyance, either
for a rural water conveyance tunnel such leakage might be for hydropower, fresh water transport, or flood control.
perfectly acceptable. Underground hydraulic structures may include drop and
riser shafts, inclines, tunnels, intakes, outlets, intersections,
d. The five basic design steps are outlined below: bifurcations, energy dissipators, venturi sections, sediment
control, surge chambers, gates, and valves.
(1) The functional requirements are defined in a broad
sense. They include all hydraulic and geometric require- a. Types of flow in underground hydraulic
ments, ancillary and environmental requirements and limi- structures.
tations, logistics, and maintenance requirements.
(1) Flow in underground hydraulic structures will be
(2) Collect geologic and cultural data including all either open-channel flow or pressurized flow. Pressurized
information required to define potential failure modes and flow is usually under positive pressure, but negative pres-
analyze them, field and laboratory data, and cultural data to sures can also be encountered.
define environmental effects and constraints. These data
may include ownership of right-of-way, the possibility of
encountering contaminants, and sensitivity of structures to
settlements.

6-1
EM 1110-2-2901
30 May 97

(2) If it is desired to maintain gravity flow conditions b. Hydraulic controls.

in a tunnel, then the size and grades must be designed to
accomplish this. Usually, the variable flow quantities and (1) Hydraulic controls are placed in a flow channel to
input pressures (minimums and maximums) are given and regulate and measure flow and to maintain water levels
cannot be adjusted. In some cases, geologic conditions upstream of a section. Over the full length of a tunnel, a
may limit adjustments to grade. On the other hand, it may variety of flow conditions may exist in each of the seg-
be desired to generate pressurized flow, for example in a ments. Discharge and flow depth are determined by the
hydropower intake tunnel to spin the turbines, in which slope, geometry, and lining of a tunnel and by the locations
case size and grade are selected for that purpose. Trade- of hydraulic controls such as gates, weirs, valves, intakes,
offs can be made between size and grade to determine and drop structures. Within each segment of a tunnel, the
whether pressurized or gravity flow will occur and which segment inlet or outlet can serve as the control section.
is more desirable for a specific facility. Inlet control will exist when water can flow through a
tunnel segment at a greater rate than water can enter the
(3) Short tunnels of 100 m (330 ft) or less can be inlet. Headwater depth and inlet geometry determine the
driven level, but longer tunnels are usually constructed at a inlet discharge capacity. Segments of a tunnel operating
minimum slope of 0.0001 (0.01 percent) to facilitate under inlet control will generally flow partially full.
drainage.
(2) Outlet control occurs when control sections are
(a) Open-channel (gravity) flow hydraulic structures. placed at or near the end of a tunnel segment and water
In open-channel flow, the water surface is exposed to the can enter the segment at a faster rate than it can flow
atmosphere. This will be the case so long as the rate of through the segment. Tunnel segments flowing under outlet
flow into the structure does not exceed the capacity as an control will flow either full or partially full. The
open channel. For a gravity flow tunnel with multiple flow capacity of a section flowing under outlet control
input sources or changes in cross section or grade, various depends on the hydraulic factors upstream of the outlet.
points along the alignment must be analyzed to ascertain
the flow volume and velocity to make certain that this (3) Weirs are one form of hydraulic control com-
condition is met. Hydraulic jumps can form within open monly used to regulate and measure flow in open channels.
channels if the slope of the channel is too steep or the Many variations in weir design exist, most of which are
outlet is submerged. If the hydraulic jump has sufficiently accompanied by their own empirical equations for the
high energy, damage to the structure can result. This con- design of the weir. Weir equations and coefficients are
dition should be avoided. found in most textbooks dealing with open-channel flow.

(b) Pressurized structures. When the flow rate c. Transient pressures.

exceeds the open-channel capacity of the structure, it
becomes pressurized. This may be a temporary condition (1) Transient pressures are a form of unsteady flow
or may be the normal operating configuration of the facil- induced whenever the velocity of moving water in a closed
ity. Cavitation occurs in flowing liquids at pressures below conduit is disrupted. Causes include changes in valve or
the vapor pressure of the liquid. Because of low pressures, gate settings, pump or power failures, lining failures, and
portions of the liquid vaporize, with subsequent formation filling of empty lines too quickly. One type of transient
of vapor cavities. As these cavities are carried a short flow is known as water hammer. This phenomenon is a
distance downstream, abrupt pressure increases force them significant design consideration in water tunnels because of
to collapse, or implode. The implosion and ensuing inrush the structural damage that can occur with excessive high or
of liquid produce regions of very high pressure, which low pressures. There are many other types of transient
extend into the pores of the hydraulic structure lining. flows in tunnels that can be caused by unequal filling rates
Since these vapor cavities form and collapse at very high at different locations along the tunnel: air entrainment, air
frequencies, weakening of the lining results as fatigue releases, and hydraulic jumps. For structural analysis,
develops and pitting appears. Cavitation can be prevented lower safety or load factors are used when designing for
by keeping the liquid pressure at all points above the vapor transient pressures.
pressure. The occurrence of cavitation is a function of
turbulence in the water flow and increases with tunnel (2) Transient pressure pulses arise from the rapid
roughness and flow velocity. conversion of kinetic energy to pressure and can be

6-2
 EM 1110-2-2901
30 May 97

positive or negative depending on position with respect to of water tunnels are restricted by potential cavitation dam-
the obstruction. Pressure pulses will propagate throughout age depending on the liner material used, sediment deposit,
a tunnel or pipe system being reflected at the ends and and flushing characteristics.
transmitted and reflected where cross sections change. The
magnitude and propagation speed of a pressure pulse are (2) The determination of tunnel friction factors for
determined by the elastic characteristics of the fluid and the use with the Manning or Darcy-Weisbach flow equations is
conduit and the rate at which the velocity is changed. All complicated by changes in flow depth, irregular channel
other factors being equal, the more rapid the velocity geometries, and the wide range of roughnesses that occur
change, the more severe the change in pressure. when multiple lining types are used. Friction coefficients
for the Manning and Darcy-Weisbach equations are each
(3) Transient pressures are managed by careful place- affected, but to different degrees, by changes in velocity,
ment of surge tanks, regulated valve closure times, surge depth of flow, lining material, tunnel size, and tunnel
relief valves, or a combination of these methods. shape. The Darcy-Weisbach approach is technically the
more rigorous of the two equations; however, the Manning
(4) Transient pressures should be analyzed for each equation survives in practice because of its reasonable
and every tunnel by the hydraulic engineering staff for use accuracy as an approximation for typical tunnel sizes and
in the design of pressure tunnels. For preliminary use, a its relative simplicity.
transient pressure 50 percent higher than the operating
design pressure is often used. (3) In practice, fluid velocities are limited so that
turbulent conditions and the possibility of damage to the
d Air relief. structure are limited. Velocities of less than about 3 m/s
(10 ft/s) are considered safe in tunnels with no lining.
(1) Air that occupies an empty or partially filled tunnel Velocities between about 3 and 6 m/s (10-20 ft/s) usually
can become trapped and lead to operating difficulties rang- necessitate concrete linings. For velocities greater than
ing from increased head loss and unsteady flow to severe 6 m/s (20 ft/s), the risk of cavitation increases, and special
transients and blowouts. Air can enter a tunnel system by precautions like steel or other types of inner lining must be
entrainment in the water at pump inlets, siphon breakers, taken to protect the inside of the structure. Where the
drop structures, and hydraulic jumps. It can also form water will carry sediments (silt, sand, gravel) the velocity
when pressure and temperature conditions cause dissolved should be kept below 3 m/s (10 ft/s).
air to be released.
(4) A study on friction losses in rock tunnels by
(2) Engineering measures to reduce air entrainment Westfall (1989) recommends friction factors (Manning's
include thorough evaluation of drop structures under all roughness coefficient, n) for different excavation methods
foreseeable flow conditions, elimination of hydraulic jumps and lining types as follows:
by reducing channel slopes or other means, and dissipation
of flow vortices at inlets. Drill and blast excavation, unlined n = 0.038
Tunnel boring machine excavation, unlined n = 0.018
(3) Air entrapment can lead to increased head losses Lined with precast concrete segments n = 0.016
caused by a constricted flow cross section, and more sig- Lined with cast-in-place concrete n = 0.013
nificantly, severe transient pressures when trapped air is Lined with steel with mortar coat n = 0.014
allowed to vent rapidly. Air entrapment at changes in Lined with steel (diam > 3 m (10 ft)) n = 0.013
tunnel cross sections are avoided by matching tunnel crown Lined with steel (diam < 3 m (10 ft)) n = 0.012
elevations rather than matching the inverts. Vents to the
ground surface frequently are used for air pressure relief. (5) Factors that can adversely affect friction include
overbreak and rock fallout in unlined tunnels, misalignment
e. Roughness. of precast segments and concrete forms, sediment, and age.
Westfall (1989) emphasizes the value of presenting several
(1) The roughness of a tunnel lining relative to its tunnel diameter and lining alternatives in final contract
cross-sectional dimensions is fundamental to the efficiency documents. Huval (1969) presents a method for computing
with which it will convey water. Tunnel excavation meth- an equivalent roughness for unlined rock tunnels that is
ods, geometry, and lining type affect flow capacity and employed for different tunnel stretches in an example by
play important structural and economic roles in water Sanchez-Trejo (1985). Figure 6-1 shows the basic
tunnel design. The allowable velocities in different kinds

6-3
EM 1110-2-2901
30 May 97

(1) Drop shaft components. A drop shaft has three

essential elements: an inlet structure, a vertical shaft bar-
rel, and a combination energy dissipator and air separation
chamber. The inlet structure's function is to provide a
smooth transition from horizontal flow to the vertical drop
shaft. The drop shaft barrel then transports the water to
the lower elevation and in the process dissipates as much
energy as possible. At the bottom of the drop shaft, a
structure must be provided that will withstand the impact
forces, remove any entrained air, and convey the water to
the tunnel.

(2) Basic consideration in drop shaft design. Several

factors must be considered in the design of drop shafts.
These factors are variable discharge, impacts on the drop
shaft floor, removal of entrained air, and head loss associ-
ated with the drop shaft. The selection of an appropriate
drop shaft for a particular use involves determining which
of these factors are most important. When the difference
in elevation between the upper level flows and the tunnel is
small, impacts on the drop shaft floor may be alleviated
with a simple plunge pool. As the difference in elevation
increases, removal of entrained air is necessary and floor
impact becomes more severe. In cases where the tunnel
hydraulic gradient can rise all the way up to the hydraulic
gradient for the upper level flows, head loss also becomes
a critical factor.

(3) Variable discharge. A drop shaft may be oper-

ated for steady-state flows, only during storm discharge
periods, or as a combination of the two. The flow variabil-
ity of a drop shaft has a considerable influence on the
Figure 6-1. Roughness factor calculations for unlined design. For instance, for steady-state flow the water sur-
tunnels face elevation in the tunnel may be below the base of the
drop shaft. In that case, a plunge pool is required at the
equations utilized by this method. Manning's n for com- drop shaft floor to dissipate energy. A shaft that handles
posite linings of different roughness can be estimated as a only storm flows will not normally require a plunge pool
weighted average of the friction factors for each surface because the water surface in the tunnel will submerge the
where length of wetted perimeter of each surface is used drop shaft base and cushion the impacts.
for weighting. Figure 6-2 illustrates the variation in fric-
tion factor versus flow depth in a shotcrete-lined tunnel (4) Impact on the drop shaft floor. The impact of the
with a concrete-paved invert. water on the floor of the drop shaft can be high, and steps
should be taken to minimize it. This is accomplished by
f. Drop shafts for vertical conveyance. Drop shafts forcing a hydraulic jump within the shaft, by increasing the
are used in water conveyance tunnels to transfer flows from energy dissipation due to wall friction as the water
a higher elevation to a lower elevation. Such drop descends, by entraining sufficient air to cushion the impact,
shafts are typically used in flood control and CSO systems. or by providing a plunge pool at the bottom of the shaft.
Drop shafts should be designed to dissipate the energy The plunge pool may be formed by a depressed sump or
increase associated with the elevation drop; to remove any by the use of a weir located in the chamber at the base of
air that mixes or entrains with the water as it descends; and the shaft and downstream of the shaft barrel. The required
to minimize hydraulic head losses when the tunnels are depth of the plunge pool can be determined by the use of
surcharged. the Dyas formula:

6-4
 EM 1110-2-2901
30 May 97

Figure 6-2. Friction factors for composite lined tunnel (see Figure 6-1 for definition of symbols)

6-5
EM 1110-2-2901
30 May 97

1/3 sewer to a lower sewer. These drop shafts are designed to

Depth = 0.5h 1/2dc (6-1)
minimize turbulence, which can release odorous gases and
where damage the shaft. A typical design has a personnel access
upstream of the shaft that allows maintenance personnel to
h = height of drop, ft enter the lower sewer without climbing down the wet shaft.

dc = critical depth in inlet, ft (9) Vortex drop shafts. Flow enters the vortex-flow
drop shaft tangentially and remains in contact with the drop
(5) Removal of entrained air. As the water falls shaft wall, forming a central air core as it descends. Since
through the drop shaft, it entrains, or mixes, with air. the flows through the inlet are spun against the shaft wall,
There are several advantages and disadvantages associated the entry conditions are relatively smooth. Vortex drop
with air entrainment. The advantages are as follows: shafts are effective for a wide range of discharges. The air
core helps to evacuate the entrained air and to provide near
$ Presence of air minimizes the possibility of subat- atmospheric pressure throughout the shaft, so as to prevent
mospheric pressures and thus negates the harmful any cavitation. Vortex drop shafts generally entrain less
effects of cavitation. air than other types of drop shafts for two reasons. First,
the flows are highly stable due to the entry conditions.
$ Impact of the falling water on the drop shaft floor Second, a reverse flow of air occurs in the core of the
is reduced by the cushioning effect of the air vortex, which causes much of the air entrained in the flow
entrained in the water. to be released and recirculated in the zone above the
hydraulic grade line. Below the hydraulic grade line, the
Disadvantages of air entrainment are as follows: helical flow has a pressure gradient, which forces bubbles
to move toward the center of the drop shaft where they are
$ Flow volume is bulked up and requires a larger able to rise against the relatively slower moving water.
drop shaft. Therefore, most air entrained by the flow is allowed to
dissipate before it enters the tunnel. As the flows are spun
$ In order to prevent the formation of damaging against the walls of the drop shaft, significant energy is
high-pressure air buildups, entrained air must be dissipated before the flow reaches the floor of the drop
removed before entering the tunnel. shaft. The dissipation is a consequence of the wall friction
as the flows spiral down at high velocity. The remainder
(6) Head loss associated with the drop shaft. Under of the energy is dissipated in the air separation chamber by
certain conditions the tunnel hydraulic gradient may rise to either a plunge pool or by the formation of a hydraulic
levels equal to those of the upper level inflow. In these jump. Several inlet configurations have been adopted to
circumstances, the head losses become important because a create a vortex flow down a drop shaft (see Figure 6-3).
large head loss may cause severe flooding in the upper Based on various model studies, a vortex drop shaft is
level flow delivery system. For example, if this upper highly efficient when the turned gradient does not
level delivery system is a sewer, large drop shaft head approach the level of the upper incoming flow. It is a
losses will result in flow backups into streets and/or good energy dissipator and has a high air removal rate.
basements.
(10) Morning glory drop shafts. Morning glory drop
(7) Types of drop shafts. Various types of drop shafts shafts employ a circular crested inlet structure. They are
have been designed and constructed based on hydraulic often used as outflows for reservoirs. Model studies have
laboratory model studies. Drop shafts as deep as 105 m determined that the flow characteristics are controlled by
(350 ft) have been constructed. The smaller structures, three conditions: weir control, orifice control, and differ-
normally used for drops of less than 21 m (70 ft), are ential head control. The capacity of the morning glory
divided into several categories. These categories are drop drop shaft is limited by the size of the circular crest. No
manholes, vortex, morning glory, subatmospheric, and cavitation is expected in this type of drop shaft. Induced
direct drop, air entraining. head losses could occur if the circular crest is inadequately
designed. The U.S. Bureau of Reclamation recommends
(8) Drop manholes. Drop manholes are generally used that the outlet tunnel be designed to flow 75 percent full to
in local sewer systems to transfer flows from a higher eliminate instability problems.

6-6
 EM 1110-2-2901
30 May 97

back up to the air vent side of the vertical shaft and rises
to the surface, some of it being recirculated through the
slots into the drop shaft. If the drop shaft is to be used for
steady-state flows, a plunge pool is built directly beneath
the shaft barrel to dissipate the energy.

(d) This structure requires a rather large air separa-

tion chamber. For larger drop shafts, this requires a high
chamber roof. During the design of the TARP system
(Chicago) in rock, it was determined that this type of shaft
was economical up to shaft diameters of 2.7 m (9 ft) with
a maximum discharge capacity of 17 m3/s (600 cfs).

(e) Another drop shaft design is suitable for drop

shafts larger than 2.7 m (9 ft) in diameter. This drop shaft
shown in Figure 6-5 has a separate shaft for the air vent
downstream from the downcomer and connected to the
downcomer above the crown of the incoming pipe. The air
separation chamber has a horizontal roof. The air vent
recycles air into the downcomer. This design can be used
in much larger drop shafts, up to 6 m (20 ft) in diameter
with a maximum discharge capacity of 127 m3/s
(4,500 cfs).
Figure 6-3. Five types of inlets for vortex-flow drop
structures (f) Both structures handle a wide range of discharges
and have head losses only one-fifth of those for vortex
type shafts. These shafts are the only commonly used drop
(11) Direct drop air entraining drop shafts. shafts that can adequately handle variable discharges,
impacts on drop shaft floors, remove entrained air, and
(a) Flow enters these drop shafts radially and descends have minimum head losses to prevent backflow problems
through the shaft. The shaft diameter is designed to flow when tunnel gradients reach the levels of incoming flows.
full with air entrained in the water, bulking it up enough to
fill the drop shaft. The air entrained also provides a cush- (g) The large dimensions of both of these types of
ion for the water, reducing the floor impact. A large sepa- drop shafts, particularly the air separation chambers, neces-
ration chamber is used at the base of the shaft and an air sitate mining a major chamber in rock with attendant rock
vent is necessary to allow the air to vent before entering reinforcement and lining. Larger sized versions of these
the tunnel. This type of structure is very effective in dissi- drop shafts can be overexcavated and used as construction
pating energy and removing entrained air. shafts.

(b) Two types of direct-drop air entraining drop shafts g. Air removal. High-velocity streams of water may
are discussed below. The first of these consists of a sump entrap and contain large quantities of air. Air entrainment
chamber with a sloping top, as shown in Figure 6-4. The causes the flow to be a heterogeneous mixture that varies
air vent is located inside of the drop shaft downcomer bar- in bulk density throughout the flow cross section and
rel, separated by a vertical slotted wall. The slots in the exhibits pulsating density variations.
wall allow air to be recirculated into the falling water in
the drop shaft resulting in the reduction of large air slugs (1) Potential problems.
and providing a more homogeneous mixture of air and
water. (a) The engineer should eliminate the harmful effects
brought about by the formation of high-energy hydraulic
(c) At the bottom of the shaft is the sloped-roof air jumps within the tunnel; transient phenomena induced by
separation chamber. As the air is released from the rapid filling of the downstream end of a tunnel without
mixture, it follows the sloping wall of the air collector

6-7
EM 1110-2-2901
30 May 97

Figure 6-4. Direct-drop air entraining drop shaft

provisions for adequate surge shafts; the formation of air $ BlowoutsChigh-pressure releases of air and water
traps within the tunnel system; the introduction of entrained in the same direction of the flow.
air into the tunnel from drop shafts; and the formation of
vortices, which may enter the tunnel through shafts. In $ GeyseringCair/water venting above the ground
addition, the design should provide for the easy egress of surface through shafts located at any point along
air from a tunnel while it fills with water. Improper design the tunnel.
can lead to one or more of the following phenomena,
which may lead to structural damage: $ Transient and surging flows causing rapid
dynamic instability and possible tunnel collapse.
$ BlowbacksChigh-pressure releases of air and
water in the opposite direction of the flow.

6-8
 EM 1110-2-2901
30 May 97

(e) Inlet No. 2 of the Oroville Dam Diversion Tun-

nels experienced the development of vortex. The vortex
grew in size and strength as the reservoir filled during the
December 1964 flood. After the flood, the tunnel was
dewatered and inspected throughout its entire length.
Although the observed damage was relatively minor, it did
consist of many rough scoured surfaces throughout the
entire tunnel length.

(2) Solutions. The above-mentioned problems can be

prevented by proper precautions during design. The fol-
lowing steps should be taken:

(a) Check the tunnel slopes for the development of

supercritical flow and calculate whether a hydrau-
lic jump can occur for any conceivable discharge.
A hydraulic jump may not occur during the maxi-
mum design discharge but can occur for some
lesser discharges. The tunnel slopes should be
reduced if the check shows the potential for a
hydraulic jump.

Figure 6-5. Direct-drop air entraining drop shaft with (b) Provide surge shafts of diameters at least equal to
separate air vent the diameter of the tunnel at both upstream and
downstream ends of the tunnel. A transient
(b) As long as the depth downstream of a hydraulic analysis should be made during the design phase
jump does not reach the tunnel crown, jumps within tun- to determine how high these surge shafts should
nels are not a severe problem. When the downstream be.
depth seals against the roof of the tunnel, the shock effects
of air trapped downstream of the jump can create violent (c) Whenever branch tunnels or drop shaft exit
impacts and associated damage. High-energy hydraulic conduits meet another tunnel and whenever a
jumps have caused both blowouts and blowbacks. These tunnel changes diameter, always match tunnel
rapidly escaping air pockets result in water rushing in to crowns rather than inverts, to prevent the forma-
fill the voids, creating loud noises and pressure waves, tion of air pockets.
which have resulted in stripping the lining from tunnels
and shafts, partial tunnel collapse, and severe erosion. (d) Prevent entrained air from entering the tunnel
from drop shafts.
(c) Even without the formation of hydraulic jumps,
blowbacks, blowouts, and geysering, dynamic instability (e) Provide a splitter wall to suppress the develop-
due to transients can take place whenever the downstream ment of vortices in the inlet to tunnels whenever
end of a tunnel is filling rapidly while air trapped within it is apparent that strong vortex development may
the system cannot escape at a reasonable rate. When the occur.
pressurization surge reaches an upstream end of the tunnel
during the filling process, water will rise rapidly in shafts (f) Provide some form of inlet control to regulate or
near the upstream end. Water levels in other shafts will completely shut off all flows into each inlet tribu-
also rise as the surge reflected by the upstream end travels tary to the tunnel. This may usually be accom-
downstream. plished by the use of remotely controlled gates at
each shaft inlet.
(d) In pressure tunnel flows, an air void can form at a
bend connecting a vertical shaft to a horizontal tunnel. A
sudden reduction in the flow rate can cause this void to
vent back up the shaft and cause geysering.

6-9
EM 1110-2-2901
30 May 97

h. Control of infiltration and exfiltration. must then be installed around the in-line diversion pipes on
both sides of the proposed connection to prevent water
(1) The phenomena of infiltration and exfiltration are from flowing along the backs of the pipes into the connec-
of critical importance to water conveyance tunnels. Infil- tion. With the in-line diversion in place, the new tunnel
tration during construction should be reduced to acceptable connection can be made in the dry while the existing tun-
levels in all types of tunnels. Significant infiltration after a nel is fully pressurized. When the connection is com-
water conveyance tunnel is completed is unacceptable. pleted, the existing tunnel may be dewatered again and the
Inflows can cause loss of ground into the tunnel and result diversion pipes and cutoffs removed and the project
in surface settlements and damage to neighboring struc- completed.
tures. The inflows may cause the adjacent groundwater
table to be seriously lowered with resulting adverse impacts (3) Open-piercing method, lake taps.
on water supply, trees, and vegetation. In flood control
tunnels, groundwater infiltration can reduce the carrying (a) The method is restricted to the construction of a
capacity available to handle peak flows. Infiltration in connection in rock. In this method, the new tunnel is
water supply tunnels may lead to pollution of the supply. advanced as close to the existing high-pressure source as
In sewer tunnels, infiltration contributes to increased water possible, leaving a rock plug in place above the tunnel
reclamation and pumping costs. crown. The tunnel near the connection should be con-
structed such that, when filled with water, a compressed air
(2) Exfiltration from water conveyance tunnels also cushion will be created below the plug. This air cushion
has potential for undesirable effects. In flood control and should be maintained until the final connecting blast is
sewer tunnels, exfiltration may cause pollution of the adja- made. A rock trap is provided in the invert of the new
cent groundwater. Exfiltration from water supply and tunnel below the plug. A shaft from ground surface to the
power tunnels can result in serious reductions in available new tunnel invert is also required as close as possible to
drinking water and energy supplies as well as revenue loss. the connection. A gate is provided on the side of this shaft
furthest from the rock plug to seal any water from entering
(3) The extent to which infiltration and exfiltration the tunnel beyond the shaft-rock plug section. The rock
should be reduced must be determined before the design of plug is then drilled and prepared for blasting to make the
the tunnel commences. It may be appropriate to apply final connection. Next, the gate is closed and the tunnel
different standards of water tightness to different sections (on the rock plug side of the shaft) and the shaft are filled
of the tunnel. It is common practice to specify in the with water to a depth slightly below the water level in the
contract documents permissible inflows both during and live tunnel or lake to be tapped. At this point, the air
after the construction of water conveyance tunnels. cushion below the plug should be checked for adequacy by
remote monitoring and additional air pumped in if neces-
i. Lake taps and connection to live tunnels. Con- sary. The charge is then detonated and the air cushion
necting a new water conveyance tunnel to an existing below the plug interrupts the water column to dampen the
high-pressure water tunnel or tapping a lake or reservoir is pressure shock and prevent damage to the new tunnel.
a task that requires careful advance planning. Obviously Since the water pressure at the time of the blast is less
such connections are best made in the dry, but in certain inside the newly constructed tunnel, most of the rock
cases this is not economically feasible. The following blasted in the connection will collect in the rock trap. In
discussion highlights some alternatives. this procedure, the final connection is left unlined.

(1) Cofferdam. For tunnels that are to connect to a (b) There are several other methods to execute lake
relatively shallow lake, a ring cofferdam can be constructed taps. In 1988, the Alaska District employed the Adry
from tunnel level below the bottom of the lake to an method@ for a lake tap for the Snettisham project near
appropriate elevation above the water surface. The Juneau, Alaska. The final plug was about 3.3 by 3.3 by
enclosed area can then be dewatered in order to make the 3.6 m (11 by 11 by 12 ft) and blasted using a double burn-
connection between the lake and the future shaft and tunnel hole cut pattern. A buffer was made of a large plug of ice.
in the dry. Two rock traps were employed.

(2) In-line tunnel diversion. To connect a new tunnel (c) The design and construction of lake taps and
to a live high-pressure tunnel, an in-line diversion pipe or other high-pressure taps must be carried out with the help
series of pipes can be installed within the existing tunnel of specialists experienced in this type of work.
after it has been temporarily dewatered. A flow cutoff

6-10
 EM 1110-2-2901
30 May 97

j. Other requirements. The hydraulic requirements examples of failure modes are encountered primarily dur-
of underground structures are of primary importance to ing construction, but some of them may apply to finished
design and construction. Other secondary considerations tunnels left unlined or with insufficient ground support.
are listed below. The second series of examples apply to finished, lined
tunnels. Failure of environmental nature, such as
(1) Construction tolerances. With open-channel flow, detrimental groundwater drawdown or damage due to set-
tunnel grade elevation must be established with some preci- tlements are discussed in Section 5-14.
sion to maintain the hydraulic properties of the facility.
Accurate grade also provides better drainage during con- a. Tunnel and shaft failure modes during
struction and avoids accumulation of water in depressions construction.
during construction. Grade tolerance for the finished tun-
nel is usually set at ±13 mm (0.5 in.) for relatively short (1) Failures controlled by discontinuities.
tunnels, ±25 mm (1.0 in.) for large tunnels. A greater
tolerance is given, for constructibility reasons, to tunnels (a) Rock masses are usually full of discontinuities,
lined with one-pass concrete segments. The centerline bedding planes, fractures and joints, or larger
tolerance for the finished cast-in-place tunnel is often set at discontinuities, faults, or shears that may form zones of
±25 mm (1.0 in.). However, this tolerance is often irrele- weakness. These are planes of weakness where the rock
vant for functional purposes, and a much greater horizontal mass may separate or shear during excavation. Whether or
tolerance, up to ±150 mm (6 in.) or more can usually be not they will separate or shear and cause a rock fall into
accepted. For a cast-in-place lining, the tolerance on the the tunnel is largely a matter of geometry, and of the
inside diameter can be set at 0.5 percent, provided the tensile and shear strength of the discontinuity.
lining thickness is not less than designated. For a precast
segmental one-pass lining, a maximum out-of-roundness of (b) The tensile strength across a bedding plane is
0.5 percent is usually acceptable. Surface irregularities often poor or nonexistent. The shear strength, however,
should be kept below 6 mm (0.25 in.). can be close to that of the adjoining materials, depending
on the normal stress across the plane, as well as joint
(2) Unlined sections may need rock traps. If a tunnel roughness and other surface characteristics. Because the
or shaft is unlined and may collect small pieces of rock or excavation of a tunnel results in a general unloading of the
debris, traps are recommended to collect the debris so that tunnel environment, the shear strength of a bedding plane
it has a minimal effect on flow area, velocities, and friction is often greatly reduced, depending on the orientation of
losses, and so that it will not enter turbines or valves. the bedding plane relative to the opening. Therefore, bed-
ding planes often participate in forming blocks of rock that
6-3. Modes of Failure of Tunnels and Shafts can fall from tunnel roof, wall, or face.

It is convenient to distinguish between modes of failure (c) Shaley beds in a sandstone or limestone formation
that occur during construction and those that occur some- may appear to be sound at first exposure, but the unloading
time during the operating life of the structure. Some fail- due to excavation combined with access to air and water
ure mechanisms observed during construction may be can soften and cause slaking in such beds in hours or days
present throughout the operating life if not properly con- such that they lose most of their tensile and shear strength
trolled. Some construction failure modes were discussed in and participate in the formation of rock falls. It is com-
the earlier subsection on tunneling hazards (flooding, mon in such bedded formations to experience rock falls
gases); others more related to the mechanics and chemistry days after excavation.
of rock masses are discussed in this subsection. This dis-
cussion is not exhaustive because combinations of natural (d) Joints and fractures have no tensile strength,
forces and the effects of construction can lead to events unless they have been healed by secondary deposition of
that cannot readily be categorized. Nonetheless, an under- minerals. The shear strength of a joint depends on a num-
standing of the forces of nature working in a tunnel envi- ber of factors: Width, infilling (if any), local roughness,
ronment is helpful in preparing for design work. Failures waviness on a larger scale, the strength of the joint wall
of tunnels and shafts range from collapse or complete (affected by weathering), and the presence of water.
inundation with water and silt to merely disfiguring cracks.
They all have underlying causes, and if these causes are (e) One discontinuity across or along the tunnel can-
understood, the potential exists to discover them ahead of not form a block that will fall from the roof, wall, or face.
time and prevent or prepare for them. The first set of

6-11
EM 1110-2-2901
30 May 97

Figure 6-6. Examples of discontinuities (in part after Proctor and White 1946) (Continued)

It usually takes three intersecting discontinuities to form a unfavorable fractures through intact rock, causing rock falls
loose block. However, gravity can help cause a cantile- even with only two (sets of) discontinuities. Figure 6-6
vered block to fail by bending or tension, and stress shows several examples of how fractures and bedding
concentrations around the opening can result in other planes can affect tunnel stability.

6-12
 EM 1110-2-2901
30 May 97

Figure 6-6. (Concluded)

6-13
EM 1110-2-2901
30 May 97

(f) If orientations and locations of discontinuities were come in patterns, with one to three sets of joints, each set
known before tunnel construction, stability of blocks could containing mostly subparallel joints but the joint sets
in theory be predicted using graphic techniques or block intersecting each other at angles. These kinds of rock are
theory (Goodman and Shi 1985). For a long tunnel, this is often called blocky or very blocky.
not feasible. If only orientations are known, with an idea
of the spacing or frequency of the discontinuities, then an (4) Interlocking rocks. Interlocking, jointed rock
assessment can be made of the probability or frequency of masses can be moderately or highly jointed, but the joints
potential rock falls. On this basis, a rational determination are tight and contorted such that their inherent shear
can be made of the need for ground reinforcement (e.g., in strength is high. Examples are some basalts, welded tuffs
the form of systematic or spot rock bolts or dowels) and and rhyolites, and other rock masses where the jointing is
the most effective orientation of such ground support. largely the result of tension fracture from cooling soon
after original deposition. Interlocking, jointed rock is often
(g) When tunnels are excavated by blasting, excess stable with a minimum of ground support.
blasting energy at the perimeter will cause damage to the
surrounding rock. This damage manifests itself as a loos- (5) Blocky and seamy rocks. Blocky and seamy
ening and weakening of the rock mass. With poor, rocks combine jointing with weak bedding planes or schist-
uncontrolled blasting practices, the zone of damage can osity. In sedimentary rocks, one or more joint sets are
reach a distance of one to several meters. Joints and other often seen at roughly right angles to the bedding planes.
planes of weakness may open temporarily or permanently
due to the pressure of escaping gases or the dynamic, (6) Shattered or crushed rock. This consists of
mechanical effect of the blast, thus eliminating any tensile mostly chemically intact fragments of rock, which may or
strength that might have been available and reducing the may not be interlocking; the fractures are sometimes partly
shear strength. The blast will also create new fractures. rehealed. Fault zones often contain rock that has been
Combined with the stress reduction due to the excavation completely sheared into a silty or clayey material of low-
of the opening, these effects greatly increase the opportun- strength, fault gouge. Such gouge is often responsible for
ity for rock falls. An opening that would otherwise have squeezing conditions. The Karawanken case history (see
been stable could require considerable ground support due Box 6-1) is a dramatic example of tunnel collapse in a
to effects of poor blasting. fault zone. Missing in these descriptions is an indication
of the degree of alteration and weathering. As earlier
(h) Jointed and otherwise flawed rocks can be classi- noted, weathering can have a profound effect not only on
fied in many ways. One method of classification is the strength of the joints but also on the intact rock
described in Section 4-4, Terzaghi's classification of rock strength. Recommendations for ground support based on
conditions for tunneling purposes. Additional comments these descriptions, intended for the design of steel sets,
are presented below. were formulated originally by Terzaghi. These recommen-
dations are found in Chapter 7.
(i) For purposes of underground design, intact rock
may be described as rock in which discontinuities are (7) Rock failures affected by stresses.
spaced such that, on the average only about three to five
discontinuities intersect the tunnel. Examples are massive (a) Before excavation of an underground opening, the
igneous rocks, marbles, or quartzites with widely spaced stresses in the rock mass are in a state of equilibrium.
joints, and sedimentary rocks that have been left largely Excavation will reduce or eliminate the stress normal to the
unaffected by tectonics, dolomites, limestones, shales, and wall of the opening, while at the same time increase the
sandstones sometimes qualify. stresses in the tangential direction through stress concentra-
tion, an effect similar to the development of stress concen-
(2) Stratified rocks. Stratified rocks are sedimentary trations around holes in plates. The effect of the increase
or metamorphosed rocks with distinctive layering, where in tangential stress depends on the strength of the rock, its
bedding planes are potential planes of weakness. Schistose ductility, and the stress distribution in the surrounding rock.
rocks are typically metamorphosed rocks with layers or
planes of weakness that are often greatly contorted. (b) If the rock is overstressed, it will yield or fail. A
plastic, ductile rock (e.g., shale), behaving similar to a
(3) Moderately and highly jointed rocks. These rocks clay, may yield without losing coherence while the yield
display few, if any, bedding plane weaknesses, but joints zone sheds load to deeper, unyielded rock. A fractured
crossing the tunnel may number 10 to 100. Joints often

6-14
 EM 1110-2-2901
30 May 97

Box 6-1. Case History: Karawanken Tunnel Collapse

The Karawanken Motorway Tunnel between Austria and

Slovenia was built in 1987-91. The tunnel is 7.6 km
long, with a 90-m2 cross section and a maximum cover
of nearly 1 km. This tunnel experienced a very large
collapse during construction in 1988.

The tunnel traverses a variety of sedimentary rocks,

ranging from dolomites and limestones to marls, clay
shales, and conglomerates. The strata are severely
folded and cut by a number of fault zones.

Excavation was by blasting methods, pulling 0.8-3.5 m

with each blast, with a crown heading followed by bench
removal at 80-150 m from the crown face. In some poor
areas, two benches were employed. Ground support
consisted of shotcrete varying in thickness from 50 to
250 mm, supplemented with rock bolts and steel mats
as well as steel arches, based on a rock classification
system. Where squeezing ground was encountered,
open slots were left in the shotcrete application at the
crown to permit displacements and rock relaxation. The
construction procedure relied upon stabilization by pre-
drainage, using horizontal bore holes from the face of
the tunnel.

The collapse occurred at Sta. 3028, close by the

Slovenia-Austria border and near the greatest amount of cover. Here is an abbreviated version of the series of events (see
figure).

1. Dec. 23, 1988: At Sta. 3010, two exploratory borings encounter water, and large quantities of water and sand are
released.
2. Dec. 27, 28: Five relief holes carry 60 l/s of water but soon collapse.
3. Jan. 3, 1989: Recommence driving after break; 1.2 m advance per round, relief drainage.
4. Jan. 7: At Sta. 3028 a crown borehole releases a water inburst, carrying 150 m3 of material. Later a 500-mm drainage
hole is drilled. This hole caves and delivers 200 m3 of material, followed later by an additional 400 m3 of material.
5. Jan. 8: The caved 500-mm hole is reopened by a small explosives charge. This is followed by more water and material
inburst. Later on the same day the face collapses suddenly, releasing about 4,000 m3 of water and material.

The causes of the failure were diagnosed to be a combination of at least the following factors:

1. Wide fault zone consists of crushed dolomite with sand and clay joint infill.
2. Removal of sand and clay material from the joint fillings result in loosening of the rock mass and loss of confining
pressure.
3. Strength of the rock mass is reduced due to water softening, high water pressures (up to 35 bar), and reduced confining
pressure.
4. Supporting pressure at the face was removed by excavation.
5. A contributing factor was the lengthy New Years break, during which water and fines were permitted to drain from the
face.

Remedial measures consisted of placing a concrete bulkhead in the tunnel, constructing a bypass, placing a 5-m-thick ring of
grout by injection, and careful remining.

If the potential seriousness had been recognized in time, the failure might have been prevented by grout injection into the entire
width of the fault zone to make the zone impermeable and stable.

Reference: Maidl and Handke (1993).

6-15
EM 1110-2-2901
30 May 97

rock, held in place by a nominal support of dowels or $ Stress gradient, which can be calculated just as
shotcrete, may yield with small displacements along frac- the induced stresses.
tures, perhaps with some fresh fractures, again shedding
load to more distant, stronger rock. On the other hand, if $ Effect of fractures on strength and ductility, not
the fractured rock is not held by a nominal force, pieces measurable and barely possible to guess.
may tend to loosen, resulting in a stress-controlled raveling
situation. A stronger, brittle rock will fracture and spall. $ Effects of stratification.
A very strong rock can store up a great deal of elastic
energy before it breaks, resulting, then, in occasionally (g) Box 6-2 shows a method of assessing modes of
violent rock bursts. failure based on induced stress level, rock strength, and
rock quality. Box 6-3 describes various manifestations of
(c) The strength of intact rock as well as that of a stress-induced failure based on rock type and rock strength.
fractured rock mass usually depends on the confining pres-
sure. Just like a frictional soil material, the strength (h) As discussed later in this section, one type of
increases with the confining pressure or the minimum stress-controlled failure is squeezing. This is a slow or
principal stress. Around an opening, the minimum princi- rapid encroachment of rock material into the tunnel, with-
pal stress is the pressure in the radial direction. Zero at the out change in water content. In a soil, this would be lik-
wall of an unlined opening, it increases rapidly when the ened to the squeezing or flow of a soft clay into the face
wall curves but not when it is straight; the sharper the of a shield, when the overburden pressure exceeds about
curve, the more rapid the increase in confining pressure. six times the undrained shear strength of the clay. In a
rock tunnel, squeezing conditions are often found in fault
(d) As it turns out, the highest stress concentrations zones with altered or weathered material of low strength.
are usually at the sharpest curves, such as the lower cor- At great depth where the stresses are high, a low-strength
ners of a horseshoe-shaped opening, but here the confining fault-zone material can result in a great deal of squeeze,
pressure increases so rapidly with distance that a little local and loads on a lining can approach the overburden pres-
yielding tends to stop the process of failure. On the other sure.
hand, low-stress concentrations are often found around flat
surfaces, such as flat roofs or floors (inverts). Here the (8) Failure modes affected by mineralogy.
stress gradients are small and stress fracture, when it
occurs, can be very extensive. This is exacerbated in a (a) Some modes of failure in tunnels are largely
rock formation that is horizontally stratified with little bond controlled by properties of the intact material. The concept
between the strata; here such stress conditions can lead to that the strength of a massive rock affects stress-controlled
buckling. failures such as spalling, rock bursts, or squeeze has
already been discussed. Properties other than the rock
(e) On occasion, tangential stresses induced over the strength also can result in failure or unacceptable behavior.
crown of a tunnel will help confine blocks of rock that
might have loosened in the absence of such a confining (b) Poorly consolidated shales or marls or shaley and
stress. marly layers in a limestone can slake when exposed to air
and moisture. This is a phenomenon brought about by the
(f) Stress effects, then, depend on (at least) the fol- stress relief combined with drying and wetting, and it
lowing factors: appears in the tunnel as loosening of flakes or chunks of
material, sometimes partly controlled by bedding. As
$ Induced stresses, which depend on in situ stresses pieces of the rock fall off, more rock gets exposed; slaking
and opening shape, and the distance from the with time can result in the loosening and removal of sev-
advancing face of the excavation. eral feet of rock. Slaking is greatly accelerated if water is
permitted to enter the latent fractures of the rock and
$ Rock strength; the intact rock strength can be soften the rock. The risk of slaking can be assessed by
measured; the operating parameter is the ratio means of laboratory tests, as discussed in Section 4-4.
between induced stress and rock strength, or if the
induced stress is undetermined, the in situ overbur- (c) Saturated clay-like materials, when unloaded, will
den stress to rock strength ratio. often generate negative porewater pressures (suction).

$ Rock modulus and ductility, also measurable.

6-16
 EM 1110-2-2901
30 May 97

Box 6-2. Assessing Mechnaical Modes of Failure

1. Behavior of Strong and Brittle Rock Based on RQD and Induced Stresses

The following method of assessment was developed for nuclear

waste repository design (Schmidt 1988) and is applicable to brittle,
jointed, interlocking rocks, such as basalt, welded tuff or rhyolite, as
well as other massive or jointed rocks, such as quartzite, marble,
and most igneous and metamorphic rocks.

The method is based on the premise that massive rocks subjected

to high stresses will suffer stress failure, but that flaws in the rock
mass will permit relaxation of high stresses, leading to the potential
for other modes of failure. The method requires the calculation of
the stress/strength ratio, defined as the ratio between maximum
tangential stress induced around an opening (calculated by closed
solutions or numerical methods) and the unconfined compressive
strength of the intact rock. The effect of flaws is assessed using a
modified RQD, as follows:

Modified RQD = RQD F1F2F3F4F5,

where

F1 = factor for joint expression on a large scale (waviness), on a small scale (roughness), and continuity. Range 0.9 to 1.0 (1.0 for
very wavy, rough, and discontinuous joints)

F2 = factor for joint aperture and infilling, and joint wall quality. Range 0.92-1.0 (0.92 for soft or weakened joints)

F3 = factor for joint orientation, favorable, random, unfavorable. Range 0.9-1.0

F4 = factor for blast damage. Range 0.8-1.0 (1.0 for TBM tunnel, 0.8 for poor blasting)

F5 = scale factor, function of ratio between opening size and joint spacing. Range 0.85-1.0

Ratio: Opening span/Joint Spacing <4 4-10 10-30 >30

Factor F5 1.0 0.96 0.88 0.85

The figure shows the predicted types of ground behavior based on stress/strength ratio and modified RQD. As most such charts, it is
conceptually accurate, but the bounds between regions of behavior are imprecise and subject to judgment. For example, a jointed rock
mass with joint blocks that are not interlocking (most tectonic joints) would most likely display a larger region of structurally controlled
behavior.

These materials will absorb water either from the air in the saturated clay in the underground is an entirely different
tunnel or from distant regions in the clay mass, resulting in phenomenon than the swelling of an unsaturated clay at the
swelling. If unsupported, the clay mass will encroach on surface, these tests are useless for the purpose. Such
the tunnel profile; if lined, the tendency to swell will be underground swelling pressures, in theory, can be predicted
halted but will result in lining pressures. Tertiary clays in by soil-structure analysis, but the necessary data to perform
Europe have been known to produce lining pressures these analyses are difficult to obtain. Experience shows
greater than the overburden pressure. This is possible that the amount of swell of a clay or clay shale depends on
because these types of overconsolidated clays are usually the degree of cementation between clay particles; however,
subjected to in situ horizontal stresses greater than the ver- hard and fast general rules have not yet been established.
tical stresses.
(e) Unsaturated clays or clay-shales are sometimes
(d) Prediction of swelling pressures in saturated clay found in tunnels. These can be more prone to swelling
or clay-shale has often been attempted using swell tests of than are the saturated materials, and standard swell tests
the types used to predict swelling of unsaturated clays at performed on unsaturated samples can be useful. When
the ground surface. However, because the swelling of a

6-17
EM 1110-2-2901
30 May 97

Box 6-3. Assessing Mechanical Modes of Failure

2. Manifestations of Stress-Controlled Failure

Unconfined
Compressive
Strength Typical Rock Types Overstressed Behavior
ksi MPa For Massive Rock For Jointed Rock

64 440 dense basalt, quartzite, violent regional, local

diabase, gabbro rock bursts

32 220 granite, most igneous rocks, breakouts in boreholes

gneiss, strong metamorphic lesser rock bursts combined failures
marble, slate spalling, popping (joints, intact rock)

16 110 hard, dense, sedimentary,

welded tuff, dolomite, limestone spitting, hour-glass
pillars

8 55 schistose rocks

flaking
4 28 phyllite

2 14 lower density sedimentary, stress slabbing

chalk

1 7 tuff slow slabbing

0.5 3.4 marl, shale squeezing swelling accelerated by

slaking of poorly water access to joints
cemented shales

0.25 1.7 weak clay shale swelling when cementation

destroyed

0.13 0.8 weathered and altered rock ravelling of fissured clays

0.06 0.4 hard clay yielding of nonfissured clays

Other effects: · Stress-induced creep in halite, potash

· Swelling of anhydrite (up to 2 Mpa swell pressure with access to water)
· Dissolution of soluble materials

Note: Approximate lower limit for violent rock bursts: 18-24 ksi (125-165 MPa)

such materials are exposed to water during tunneling or (f) A common failure in weak, shaley rock, partic-
due to leakage from the tunnel after completion, they can ularly in tunnels with a flat floor (horseshoe-shape) and
generate substantial swelling pressures. Such modes of high in situ horizontal stresses, is excessive floor heave.
behavior are accelerated by preexisting fractures (common This type of failure is the result of several factors:
in such materials) or fractures resulting from excavation
and stress redistribution. The Peace River diversion tunnel $ For most in situ stress conditions, a flat floor
case history (see Box 6-4) is an illustration of the effect of results in very low vertical and often high
water on a silty shale.

6-18
 EM 1110-2-2901
30 May 97

Box 6-4. Case History: Diversion Tunnel in Soft Shale, Peace River

For the Site 3 hydroelectric project in British Columbia, three diversion tunnels through the left abutment were proposed.
Confidence in the behavior of the soft shale was not great, and a test chamber in the shape of a truncated cylinder, 11.1 m
wide, 7.5 m high, and 45 m long was excavated. The chamber is at a 107-m depth and connected to the canyon wall
through an adit. Most excavation was by roadheader, but part of the adit and part of the chamber were excavated by con-
trolled blasting.

The geologic material is a Cretaceous, horizontally bedded silty shale with about 10 percent smectite, with unconfined com-
pression strength 6 Mpa (900 psi) and modulus 3-4 GPa (440-580 ksi) perpendicular to bedding, 6-8 Gpa (870-1,160 ksi)
parallel to bedding. The material is prone to slaking and weathering when exposed. Bedding plane fractures are common,
as are steeply dipping relaxation joints parallel to the canyon wall.

Ground support included two layers of fiber shotcrete and tensioned resin dowels spaced 2 m. The chamber was instru-
mented with convergence gages, multipoint extensometers, and stress cells.

The chamber was successfully excavated and supported, using heading and bench. Shotcrete in the roadheader section
was generally sound, with minor shrinkage cracking, but in the blasted section up to 65 percent of the shotcrete was
drummy.

After completion, the chamber filled with water, 5.5 m deep, for about 2 years; it was then pumped dry and inspected. Shot-
crete in the crown, which remained dry, had remained virtually unchanged and sound. Below the water line, the shotcrete
was badly cracked and spalled, and drummy throughout. Two block falls of 100-150 m3 each had occurred, bounded by
clean joints parallel to the tunnel wall. Cores were taken, and shale from the wet zone was found to be soft and fissile.
Ground movements in the dry crown were about 0.3 mm, but in the wet zone, ground movements amounted to 50-120 mm.

Conclusion: Shotcrete-shale bond was a problem if the shotcrete was not applied quickly; more so in the blasted than the
mechanically excavated parts. Water found its way through cracks and voids in the shotcrete into existing and latent fissures
in the shale, where it caused softening and swelling, and resulted in displacement and spalling of shotcrete. The diversion
tunnels are to be designed with a circular shape and a cast-in-place concrete lining over the initial shotcrete support.

Reference: Little (1989)

horizontal stresses in the floor, conducive to swelling of and accelerating dissolution. Voids can cause surface sub-
the floor material. sidence or irregular loading and loss of support for
tunnel lining and the ground support system. Removal of
$ Seepage water finds its way to the floor, causing the gypsum cement in a sandstone by seepage water has
swelling. caused the failure of at least one major dam (San
Franciscito Dam in California, in 1928). When such mate-
$ The floor is subject to construction traffic, which rials are present, particular attention must be paid to the
causes softening in the presence of water. watertightness of the tunnel.

Swelling also occurs when geologic materials such as (c) In the longer term, limestone is also subject to
anhydrite or shales containing anhydrite absorb water. dissolution. In this case, however, the concern is more for
the likelihood of encountering voids and caverns than the
(9) Effects of water. prolonged effect of dissolution on the tunnel structure.

(a) As discussed earlier, groundwater contributes to (d) Flowing water will erode unconsolidated material.
modes of behavior such as swelling and slaking. Water Piping phenomena are common in soils, where backward
can contribute to many other modes of behavior and erosion by seepage water can cause failure of dams and
failure. excavations as well as cut slopes. In rock masses, joint
fillings and crushed fine materials in faults and shear zones
(b) Some rocks or minerals are soluble in water. are particularly susceptible. In a construction situation,
These include, most notably, halite (rock salt) and gypsum. prolonged water flow out of joints and shear zones can
Moving water will carry away salt and gypsum in solution cause serious weakening of the rock mass by removal of
and leave behind voids that can cause increased water flow fines, resulting in loosening and potentially collapse (see

6-19
EM 1110-2-2901
30 May 97

Box 6-1 on the Karawanken Case History). Contributing $ Prevents blocks of rock from falling out by shear
factors in such situations are the weakening effect of the and bond strength; prevents smaller fragments
water on the strength of intact rock and joints, joint fill- from falling and start a raveling sequence.
ings, and gouges; the hydrostatic pressure reducing the
effective stress across joint surfaces; and the seepage forces $ By shear, bond, and bending to withstand local
of the flowing water. forces or forces of limited extent (local blocks,
seams subject to squeezing or swelling).
(e) Inflow into tunnels loaded with silt and sand will
cause maintenance problems for dewatering pumps. An $ As a compression arch or ring, to withstand
open TBM is not greatly affected by water inflow, but a more-or-less uniform loading from squeezing,
shielded TBM often suffers problems when inflows exceed swelling, or creeping ground.
several tens of liters/second (several hundred gpm), espe-
cially when the water brings in fines. Often the mucking $ Provide some degree of water inflow control.
system, whether by rail cars or conveyor, is overloaded by
the water, and water with fines escapes the system, result- $ In combination with rock bolts or dowels, provide
ing in deposition of fines at locations where they will be overall stabilization and ground movement con-
troublesome. As an example, silt deposited in a telescop- trol.
ing shield joint will cause wear in the joint and may
destroy waterproofing gaskets. Silt deposited in the invert (b) Overall, by inhibiting ground motions and sup-
can seriously hamper placement of invert segments. plying a confining pressure for the rock mass, the shotcrete
Excess water can also affect the electrical system and acts to retain and improve the strength of the rock mass
cause corrosion of tunneling machinery, especially if the and to help in creating a self-supporting ground arch in the
water is saline or otherwise corrosive. rock mass.

(f) Inflow into tunnels will tend to drain the rock mass (c) Where shotcrete is a part of initial ground sup-
and any overburden. This, in itself, may be unacceptable, port, to be followed by subsequent installation of a final
especially if existing flora or operating wells are dependent lining (whether by cast-in-place concrete or additional
on maintenance of the groundwater table. Lowering the shotcrete), performance requirements are less stringent than
groundwater table can also result in consolidation of when shotcrete is the final support. Shotcrete as initial
unconsolidated materials, especially soft clays, resulting in ground support can be repaired and even replaced as
unacceptable surface settlement. required, and even significant flaws can be tolerated, pro-
vided they do not impair the safety of personnel. The
(g) A particular type of failure mode applies to water principle of controlled deformation of initial shotcrete
tunnels in which the water pressure fluctuates, such as in support is discussed further in the section on the New
power tunnels with surges and water hammer effects. If Austrian Tunneling Method (Section 5-5).
the tunnel is unlined or supported only by rock bolts or
dowels, the fluctuations in water pressure can result in (d) Some failure modes of shotcrete result from
water flushing in and out of rock fissures, eventually clean- imperfections in its application, others from properties and
ing out joint fillings. This also happens if there are cracks nonuniformities of the rock mass, or the action of form-
in a tunnel concrete or shotcrete lining that permit the ation water. Some examples follow:
flushing of joints. More than one power water tunnel has
failed by collapse in this way. $ Shear failure resulting from loss of (or lack of)
bond between rock and shotcrete, usually initiated
(10) Particular failure modes for shotcrete. by nonuniform loading combined with an incom-
plete ring of shotcrete.
(a) Before reviewing failure modes for shotcrete
ground support, it is useful to recapitulate the various func- $ Shear failure from local block load or load from a
tions and actions of shotcrete support, when applied to a seam of squeezing material.
minimum thickness of 50-75 mm:
$ Compression failure from excessive external uni-
$ Sealing coat to prevent atmospheric deterioration, form or nonuniform load, sometimes a combined
slaking, drying, wetting, swelling. bending and compression failure.

6-20
 EM 1110-2-2901
30 May 97

$ Fracture due to excess external water pressure, (c) Individual dowels or bolts can fail in either shear
resulting in excessive water inflow, sometimes or tension in the steel, or yield can occur along the bond
resulting from plugging of geofabric strips and between grout and rock, or between metal and grout.
piping provided for draining the rock mass. Sometimes failure occurs due to faulty installation (insuffi-
cient grout, grout not properly set, improper anchoring).
$ Shear failure of shotcrete around a rock bolt or
dowel plate resulting from excessive displacement (d) A systematic bolt or dowel installation can fail by
(squeeze) of the rock mass. loosening, raveling, or block fall between individual bolts;
this depends on joint spacing relative to bolt spacing and
(e) Loss of rock-shotcrete bond can result from the degree of interlock between rock blocks. If bolts are
incomplete preparation of a wet, partly deteriorated rock too short to anchor a large wedge, such a wedge can fall
surface or one covered with grime, dust, or mud. Other out, bringing down one or several bolts with it.
common flaws are areas with too little or too much
aggregate, too high water/cement ratio, imperfect applica- (e) A systematic bolt or dowel installation forming an
tion of admixtures resulting in slow curing, or too thin an arch or a beam can fail due to overstress of the reinforced
application. Application of shotcrete in a location with rock mass. This usually indicates that the bolt length
flowing water can result in washouts or imperfect bonding chosen was too short.
or curing.
(f) In a soft, squeezing ground, bolt face plates can
(f) The case history in Box 6-4 shows failure modes fail by overload in the metal or by punching failure into
of shotcrete exacerbated by fractures in the shotcrete and the rock.
softening of the rock.
(g) Whether any of these modes of performance have
(g) Many potential modes of failure of a shotcrete serious consequences depends on the permanency of their
application are functions of flaws in shotcrete application installation. Systems installed for temporary purposes only
and local variations in geology and loading, generally not are considered to perform acceptably as long as there is no
subject to analysis but usually controllable during applica- hazard to personnel and the permanent lining can be
tion. Where the shotcrete forms a structural arch or ring installed without problem. The temporary installation is
bonded to the surrounding medium and subject to external employed to arrest ground movements before permanent
loads, the shotcrete structure is amenable to analysis. lining installation.

(11) Failure modes of rock bolt or dowel installations. (h) When the bolt installation is considered as part of
the permanent installation, some of these modes of failure
(a) Rock bolts or dowels can control or reduce dis- may still be acceptable. Yielding of part of the system
placements, both initially and in the long term, by prevent- (shear, tension, bond) may be acceptable as long as the
ing loosening of the rock mass and increasing the rock rock mass is coherent and deformations are under control.
mass modulus to hold rock blocks or wedges in place. In However, their value may have to be discounted for the
a pattern, they act to form a reinforced arch or beam design of the final lining. Any behavior mode that can
capable of sustaining loads that may be uniform or nonuni- result in future corrosion, however, usually requires that
form. By preventing loosening of the rock mass and by the element is ignored for final design consideration.
increasing the rock mass modulus, bolts and dowels control
or reduce displacement in the short or long term. (12) Particular failure modes for shafts.
Prestressed bolts induce compression in the rock mass,
further increasing its strength and carrying capacity and (a) Because shafts are oriented 90E from tunnels,
reducing displacements. Bolts and dowels are often sup- some modes of failure are more or less common than for
plemented by metal straps, wire fabric, or shotcrete. tunnels. There are several reasons for that. First, since a
shaft penetrates the geologic strata in a vertical direction, a
(b) Bolt or dowel installations may be considered shaft is likely to encounter a greater variety of conditions,
permanent parts of the underground structure, or they may including overburden and weathered rock. Second, gravity
be temporary and not counted on for permanent support. acts on the shaft wall like on a tunnel wall, much less
The installation may be supplemented at any time with severely than on the crown of a tunnel. Third, methods of
additional ground support elements. shaft construction are generally very different from

6-21
EM 1110-2-2901
30 May 97

methods of tunnel construction, as discussed in Sec- responsible for the failure modes, however, can also affect
tion 5-7. The following are a few examples of shaft failure long-term performance, especially if they are not dealt with
mechanisms. properly. Following are additional modes that apply, typi-
cally, to the finished, lined structures.
(b) Shaft bottom failure is usually caused by water
pressures. With an impervious plug above an aquifer at (1) Failures due to water pressure.
the bottom of the shaft, the plug can fracture or burst if it
is too thin and cannot hold the pressure, whether by bend- (a) Internal water pressure can result in fracture of a
ing failure or shear along the sides, or some combination. concrete lining and escape of the water into the formation.
Of course, sinking the shaft and ignoring the aquifer alto- If these formation water pressures cannot dissipate (as in a
gether could result in flooding of the shaft, if the perme- permeable formation), the formation may be fractured by
ability in the aquifer is sufficiently great. hydraulic jacking, with the potential for tunnel damage, or
worseCinstability of adjacent slopes or valley walls. This
(c) Grouting or freezing is often used to control phenomenon is discussed in Chapter 9. Such failures can
groundwater inflow and the effect of groundwater pressures occur if the lining is not designed for the hoop tension
during shaft construction. It is difficult to ascertain the caused by the internal water pressure and the formation
quality of grouting, and ungrouted zones can be left that (and formation water) pressure on the exterior is lower
would result in excess inflow of water, perhaps carrying than the internal pressure.
solids, when encountered during sinking. A freeze-wall
occasionally fails, also resulting in inrush of water, often (b) The principal failure mode of concern for external
because flowing groundwater brings caloric energy to the water pressure is the buckling of steel-lined tunnels. Dur-
site and thaws the wall. ing operation a steel-lined pressure tunnel is not in danger
due to external water pressure, but the empty tunnel must
(d) Another shaft failure mode has nothing to do with accept the full external pressure without internal balancing
rocks or groundwater but with the site arrangement: pressure. Not infrequently, leakage from the pressure
flooding of the shaft from surface waters. This type of tunnel causes the formation pressure to rise to a value
incident is inexcusable; shafts constructed anywhere near a close to that in the tunnel. When the tunnel is then
floodplain must be equipped with a collar tall enough to emptied, it has to withstand an external pressure equivalent
prevent flooding. to the internal pressure.

(13) Particular failure modes at portals. Portals are (c) A tunnel lining is often furnished with an imper-
typically cut into the hillside and preferably expose sound vious membrane to control groundwater inflow that would
rock. The portal cut is exposed to all of the failure modes otherwise be excessive. As a general rule, this impervious
of any man-made cut into soil, colluvium, talus, or rock, membrane must accept the full external water pressure and
including slope failure on a discontinuity plane, rock falls, be supported by an internal structure capable of withstand-
deterioration due to exposure, deep-seated failures, sliding ing this pressure.
of overburden materials on top of bedrock, etc. Fractures
are often opened in the ground due to the excavation, and (2) Tunnel lining failure caused by external loads.
if filled with rain water, the water pressure can result in
failure initiation. Rockfalls can be hazardous to personnel (a) The failure of a concrete tunnel lining has to be
moving in and out of the tunnel. In addition to the typical viewed in terms of its functional requirements. A tunnel
slope failure phenomena, the portal is also the intersection lining may crack or leak or deteriorate, but as long as it
between the tunnel and the portal cut. Tunnel excavation serves its function for the expected lifetime, it has not
by blasting, if not carefully controlled, can result in very failed.
large overbreaks. For these reasons, the ground surround-
ing the tunnel must be carefully supported, and the initial (b) The following discussion, for the most part,
tunnel blasting performed with low energy, as discussed in applies equally to cast-in-place and precast, segmental
Chapter 5. lining. Tunnel linings in rock are externally contained; they
are different from aboveground structures for at least the
b. Failure modes of tunnels and shafts during oper- following reasons:
ation. Most of the modes of failure discussed above apply
to the construction environment; once they are dealt with, $ Stresses and strains are governed not so much by
they pose no further threat. Some of the conditions loads as by interaction between the lining

6-22
 EM 1110-2-2901
30 May 97

structure and the ground requiring compatible $ Excess side pressure on walls of horseshoe-shaped
displacements. tunnel, resulting in gross bending of the
walls or buckling of the floor, or both. This can
$ Except for water pressure, loads on the lining often also result from loss of floor strut due to exces-
relax upon displacement and yield; they are not sive floor heave.
conservative or following loads.
$ External factors, such as effects of adjacent new
$ Radial fractures in a concrete lining do not usually construction, slope failure at a portal or in the
form a mechanism of instability, witness voussoir vicinity of the tunnel.
arches without bonds between blocks. The com-
pressive stress between adjacent blocks combined 6-4. Seismic Effects on Tunnels,
with friction between the blocks suffices to main- Shafts, and Portals
tain the stability of the arch, even with a substantial
external load. It is generally acknowledged that underground structures
are inherently less sensitive to seismic effects than surface
$ Because of net hoop compression (in a circular structures. The good performance of underground struc-
tunnel, often also for other shapes), a tension frac- tures was demonstrated during the 1986 Mexico City earth-
ture from the inside face due to bending does not quake, where subway structures in soft and very soft
usually penetrate the thickness of the lining. ground went undamaged and the subway served as the
principal lifeline, once power was restored. In contrast,
$ The rock surrounding a tunnel lining is usually buildings and other surface facilities suffered severe dam-
under relatively strong compression, and the bond age. Nonetheless, underground structures can suffer
between lining and rock is usually good. There- damage in an earthquake under particularly unfavorable
fore, tendencies to generate external tension frac- conditions. In most cases, however, the vulnerability of a
tures due to bending are greatly resisted. particular structure can be assessed and a design prepared
that will eliminate or minimize the effects of earthquakes.
$ The usual circular shape is inherently strong and The vulnerability of underground structures is examined in
forgiving and, with usual dimensions, resists buck- Box 6-5.
ling. Horseshoe and other shapes are not as for-
giving. a. Effect of earthquake shaking on tunnels and
shafts.
(c) Structural failure of a concrete lining does occur
on occasion. When it does it is usually for one of the (1) Earthquake waves traveling through the ground
following reasons: are displacement waves, generally compression (P) or shear
(S) waves. Due to scattering and other effects, the seismic
$ Loss of support around part of the lining due to displacement waves can vary nearly randomly in space and
inadequate concrete placement or contact grouting, time. The response of a tunnel or shaft is either axial
especially in the crown of the tunnel, or due to compression or extension, horizontal or vertical curvature,
washout of fines, dissolution, or rotting of timber, or ovalizing (racking), or usually a combination of all.
resulting in uneven loading and support.
Unrelieved differential hydrostatic pressures can (2) A tunnel or shaft structure subjected to axial and
also exist in such void spaces during filling or curvature motions may be compared with a beam under
emptying of the tunnel. combined compression (extension) and bending-maximum
and minimum stresses occur at the extremities. The
$ Excessive or nonuniform load on a circular lining, resulting stresses depend on the initial static stresses, upon
causing large distortions, sufficient to create com- which the dynamic motion is superimposed.
pressive failure in bending (rarely by uniform
thrust); nonuniform load may be caused by strati- (3) Ovaling may occur due to a shear wave imping-
graphic or structural geologic differences across ing nearly at a right angle to the tunnel or shaft. While
the tunnel section and by nonuniform swelling or one diameter is increased, the perpendicular diameter is
squeezing. reduced a similar amount, and moments are created

6-23
EM 1110-2-2901
30 May 97

Box 6-5. Case History: Tang-Shan Eearthquake, 1976; Performance of Underground

Structures.

The Tang-Shan Earthquake was of Magnitude 7.8, with surface Mercalli intensities of X to XI. It occurred in an industrial area
with several coal mines. Surface faulting extended for more than 10 km, and fault traces with displacements up to 1.5 m
traversed underground mine facilities. On the surface, destruction was nearly 90-percent complete, and several hundred thou-
sand lives were lost. Damage to underground structures, however, was relatively minor, and all miners, some 1,000 in num-
ber, were evacuated safely.

An incline provides access to the Tang-Shan mine, located in the area of greatest surface destruction. The inclined tunnel
passes through 4 m of clay and a 62-m strata of limestone before reaching shale and coal strata. The tunnel is horseshoe
shaped (arch and straight walls) and lined with bricks or stone blocks, with an unreinforced concrete floor. The tunnel is
1.8-2.5 m high and 1.2-2.5 m wide. Tunnel enlargements for electrical and pumping gear are 2-3 m high and 3-5 m wide.
These structures remained essentially intact and passable after the seismic event.

The first 15 m of tunnel through the clay experienced circumferential cracks 1-3 m apart and 10-50 mm wide; a horizontal
crack, 20 mm wide, also occurred. Down to a vertical depth of 30 m, the spacing of cracks decreased to more than 10 m,
with up to a 10-mm crack width. Beyond this depth, there were occasional cracks. The concrete floor of the pump station
at a 30-m depth heaved up to 300 mm and experienced a crack 10 m long, and a few bricks and pieces of plaster loosened and
fell. The station at a 230-m depth experienced a floor heave of 200 mm along a length of about 7 m. The station at a 450-m
depth showed a 50-mm floor heave in a 1-m area; only small pieces of plaster fell off roof or walls. Damage was noted mostly
at weak spots, such as at changes in cross section or lining material, or at bases of arches. There was clearly a great reduc-
tion in damage as a function of depth; but on the whole, the tunnel remained intact and passable.

In contrast, pumps and transformers in the underground were damaged; many transformers toppled over. Rail cars tipped on
their wheels and lifted up to 30 deg off their rails. People in the mine corridors were thrown into the air up to more than 0.3 m
or along horizontally several meters, indicating accelerations greater than one g.

Production drifts in the coal mines, designed and built for a limited lifetime through weak rocks, saw effects such as excessive
loading of hydraulic mine struts, breaking of support timber, loosening and fallout of chunks of coal, dust filling the air, squirt-
ing of water out of fractures during the earthquake motion, and increased water flow through fractures in general. Most of this
behavior occurred within a distance of some 100-150 m from the faults actually observed being displaced. Beyond this range,
the mine openings, though violently shaken, showed little permanent damage.

This case history demonstrates the survivability of even poorly supported tunnels and other underground openings through
relatively weak rock when subject to violent earthquake motions.

Reference: Wang (1985)

around the opening. Maximum and minimum stresses or 15 percent higher than pseudostatic solutions. This is
occur at four points around the opening, at the inside or different from typical surface structures (buildings,
outside surface of the lining, or tangential to the rock sur- bridges), whose natural frequency often falls within the
face in an unlined tunnel. typical seismic wave frequency band, and where amplifica-
tion can be large.
(4) Regardless of the motion induced by an earth-
quake, the result is manifested as extension or compression (5) In an unlined tunnel, shaped to have its circum-
at points around the tunnel or shaft opening. Tensile ference generally in compression, the additional seismic
stresses can occur if the initial tangential stress (usually stresses are generally inconsequential. Blocks of rock that
compression) is small. These transient stresses can usually are almost ready to fall can loosen and fall out due to the
be considered as pseudo-static superposition on the existing shaking. Even when tension cracks occur, or existing
stresses, because the seismic wavelength is almost always cracks open, they will typically close again in a fraction of
much longer than the dimension of the typical underground a second, without consequence. Similar arguments apply
structure. There is little dynamic amplification, because to a tunnel supported with spot bolts and occasional shot-
the resonant frequency of an underground opening is much crete support.
higher than the typical frequency band of seismic waves.
Studies suggest that dynamic stress amplification at the (6) Where a pattern of tensioned bolts has been
tunnel opening generally gives stresses that can be up to 10 applied as ground support, the bolts create a compression

6-24
 EM 1110-2-2901
30 May 97

ring around the tunnel or cavern arch, preventing tension (2) For rail tunnels, the strategy has been to build the
and holding blocks in place. Similar conditions prevail tunnel oversized through the fault zone, sufficient to
with untensioned pattern dowels and shotcrete support, realign the track with acceptable lateral and vertical curves
where ground motions have induced some tension in the after the event, while reinforcing the ground in and around
dowels to form a compression arch. the shear zone sufficient to prevent collapse. A ground
reinforcement system of great ductility is required, such as
(7) A concrete lining will be subject to compression a combination of lattice girders, wire mesh, rock dowels,
and extension at points on the exterior and interior of the and shotcrete. Tunnel damage is expected; however,
lining. As discussed in Chapter 9, exterior extension is of repairs can be quickly accomplished.
no consequence. In the event that tension cracks appear on
the interior surface, they will close again after a fraction of (3) For shallow water tunnels, the most effective
a second. Such cracks do not usually extend through the solution may be to plan for excavation and replacement of
thickness of the concrete and cannot, in themselves, form a the damaged structure after the event. In a deeper tunnel,
failure mechanism. A simplified method of analyzing repair and replacement may not be so easy. In this case,
tunnels in rock for seismic effects is shown in Box 6-6. the tunnel may be oversized through the fault zone and a
This simplified method ignores the effect of ground- relatively flexible pipe constructed within the tunnel, pro-
structure interaction and provides an upper-bound estimate viding enough space to avoid shearing the pipe due to the
of strains induced in the lining. The method permits a fault motion. The pipe must be supported or suspended to
quick verification of the adequacy of the lining design in permit motion in any direction.
reasonably competent ground. In very weak ground,
ground-structure interaction should be considered to avoid c. Other permanent displacements of the ground.
overdesign of the lining. Portals are particularly vulnerable to permanent displace-
ments during earthquake events. Slope stability in the
b. Effects of fault displacement. event of an earthquake can be analyzed using dynamic
slope stability analyses, and portal slopes can be rein-
(1) Tunnel alignments should avoid active faults when- forced, using tieback anchors or other devices as necessary.
ever possible; however, if faults cannot be avoided, the Another potential problem is falling rocks, loosened by the
design must include fault displacement. It is not possible earthquake. Large blocks of rock loosening may be
to build a structure that will resist the fault displacement. secured individually, or shotcrete may be applied to pre-
If the tunnel structure is to remain functional after the vent loosening.
earthquake, strategies must be planned to mitigate the
effects of fault displacement.

6-25
EM 1110-2-2901
30 May 97

Box 6-6. Seismic Analysis of Circular Tunnel Linings (Continued)

1. Longitudinal Bending and Extension or Compression

Obtain seismic input parameter from seismologist:

Vs = maximum particle velocity from shear wave

As = maximum particle acceleration from shear wave

Obtain effective shear wave propagation velocity Cs of rock medium from in situ seismic survey or from relationship with
effective shear modulus G (under earthquake shear strain level):

Cs = %G/D

where D = specific gravity of rock mass. Shear modulus is related to Young's modulus Er by

G = Er/2(1 + <r)

where <r is Poisson's ratio for the rock mass

With the assumption that the tunnel structure is flexible relative to the ground, then the tunnel structure will conform to the
free-field motion of the ground, and the maximum and minimum (compression, extension) strain of the tunnel structure is

Emax/min = ±( Vs/Cs) sin 2 cos 2 ± (AsR/Cs2) cos3 2,

where R = tunnel radius (strictly speaking, R = distance from extreme compression fiber to neutral axis) and 2 = angle of
incidence of seismic shear wave. The greatest/smallest strain is usually found for 2 = 45E:

Emax/min = ± 0.5 Vs/Cs ± 0.35 AsR/Cs2 -

2. Ovaling or Racking

A seismic shear wave impinging on a circular tunnel structure at a right angle will cause the structure to rack or ovalize,
shortening one diameter D by ) D and lengthening the orthogonal diameter by an equal amount. In the free field rock mass,
the shear strain can be approximated by

( max = Vs/Cs,

and an unlined hole driven through the rock mass would suffer an ovalizing distortion of

)D / D = ± ( max (1 - <r)

The maximum strain in the lining, then, is

Emax = Vs/Cs [(3(1 - <r)t/R + 1/2 R/t Er/Ec {(1 - <c2)/(1 + <r)}]

where t = lining thickness, R = tunnel radius, Ec = concrete modulus, <c = Poisson's ratio for concrete.

3. Notes

Ovalizing strains are superimposed on strains pre-existing from static loads.

For a maximum earthquake design, usable compressive strain is about 0.003.
Tension cracks due to excessive extension dynamic strains usually cannot be avoided. They will, however, generally close
again after the seismic event. Tension cracks can be reduced in size and distributed by appropriate crack reinforcement.

6-26
 EM 1110-2-2901
30 May 97

Box 6-6. (Concluded)

4. Example - Los Angeles Metro, Circular Tunnel in San Fernando Formation

As = 0.6g, Vs = 3.2 ft/sec, Cs = 1360 ft/sec

R = 10 ft, t = 8.0 in., Ec/(1 - <c2) = 662,400 ksf, Er = 7200 ksf, <r = 0.33

1. Longitudinal:

Emax/min = ± 0.5 x 3.2/1360 ± 0.35 x 0.6 x 32.2 x 10/13602

= ± 0.00118 ± 0.000037 = ± 0.00122 < 0.003 - ok

2. Ovalizing:

) D/D = + 2 * 3.2/1360 (1 - 0.33) = 0.0031

Emax/min = ± 3.2/1360 [3(1 - 0.33)(8/120) ± 1/2 * 120/8 * 7200/ (1 + 0.33) x 1/662,400)]

= ± 3.2/1360 (0.134 + 0.122) = 0.0006 < 0.003 - ok

This example is for a concrete tunnel through a weak, soil-like material. Tunnels through stronger, rock-like materials would
be subjected to lower seismic strains.

Reference: Wang (1985)

6-27