FREE-FIELD RACKING DEFORMATION METHOD FOR SEISMIC ANALYSIS OF
UNDERGROUND STRUCTURES
The free – field racking deformation methodology has been agreed as the most reliable
and rational approach for the design of tunnel infrastructures. Since its publication in the
FHWA and AASHTO Tunnel Manual in 2010, is quickly becoming an international standard,
as few other code references are available. The actual application of this method in shallow
box – shaped tunnel infrastructure projects entails some considerations that contribute to its
practical utilization. The main issue is related with the clear definition of the boundaries where
the traditional dynamic earth pressure methods (i.e. Mononobe – Okabe and Wood) will still
be applicable. Additionally, the global effect of the structure racking distortion should be
completed with the local effect that determines the earth pressure on the lateral walls. The key
parameter appears to be the flexibility ratio, after which a simple procedure is proposed, in
order to unify the results of the deformation method with the earth pressure approaches. On the
other hand, the design of extremely flexible underground structures has to be accompanied
with additional measures at least to avoid the joints decompression and consequences at the
ground surface. Finally, the direct extrapolation of the successfully adopted criteria for the
aboveground structures has to be observed carefully. Considering tunnel infrastructures,
different issues would show up when relying on the plastic hinges formation, as the ulterior
required reparation will present unaffordable problems in most cases.
The racking deformation method, as proposed by Wang and included in the FHWA Report,
assumes that buried structures will experience transverse distortion deformations due to its
interaction with the surrounding soil during an earthquake event. The structure racking stiffness
modifies the ground free – field deformation, and therefore an R factor is calibrated to
determine the actual racking distortion to be considered in the box structure design.
Figure 1 – Racking Deformation of Box Structure
The racking coefficient R is the ratio of the racking distortion of the structure s (diff in Fig.1),
to that of the free – field soil ff. The flexibility coefficient F is the ratio of the surrounding
ground stiffness to that of the racking structure. Wang obtained the relationship between the
racking coefficient R and the flexibility ratio F (Eq. 1). The free – field deformation ff is
obtained from the ground strain γ caused by the vertically propagation of the seismic shear
waves (Eq. 3).
F = (Gm/Ks) · (W/H) (1)
4∙(1−vm )∙F
R= (2)
3−4vm +F
σ PGA
γ = Gv ∙ ∙ Rz (3)
m g
∆ff = γ ∙ H (4)
∆s = R ∙ ∆ff (5)
Where:
F = Flexibility ratio
Gm = Average Shear Strain Modulus
vm = Poisson Coefficient
γ = Ground Strain due to seismic waves
σv = Vertical Stress
PGA = Peak ground acceleration
g = gravity acceleration
Rz = depth reduction factor (Refer to Figure 2)
Ks = racking stiffness of the box
W,H = width and height of the box
Figure 2 – Depth reduction factor, with z in meters
Design approach considerations
Aboveground structures analyses have quickly evolved to a two level seismic design – criteria
approach, as far as it allows a sound and economic risk management. The first level (Operation
Design Earthquake ODE) will ensure the facility operational condition, with minor damages;
the second level (Maximum Design Earthquake MDE) will ensure the life –saving condition,
with plastic hinges formation and energy dissipation. It is said the same approach should apply
to underground structures.
The plastic hinges formation in aboveground structures is limited to elements and locations
that could be easily inspected and repaired after an earthquake event. The application of the
same approach to tunnels presents some issues:
• The plastic hinges will be irreparable in the outer elements, except for special cases.
• The energy dissipation at the plastic hinges will be only a little fraction of the total energy
dissipated in the ground itself. Therefore, some of the main assumptions usually considered
in the aboveground structures should be thoroughly restudied (i.e. seismic displacement
remains constant even when plastic hinges develop).
• The effects on other near buildings should be considered in every single case, especially in
urban areas.
Therefore, a quasi – elastic approach, with minor damages, is suggested for the maximum
seismic design scenario. Details providing ductility to the critical sections would be sensible,
but the ductility should be expected to mobilize only over the MDE event. Some owners (e.g.
San Francisco BART) have begun requiring their facilities to remain operational after MDE
level shaking.
Besides, the risk level adopted for the facility life should take into account the catastrophic
effects if the tunnel collapses, even larger than those for the aboveground structures, especially
for shallow structures. Obviously this is critical for massive transportation systems like subway
stations. Usually 5% to 10% failure risk level in 50 years of lifetime is considered for bridges
and buildings, corresponding with 500 to 1000 year return period. For public underground
facilities could be proposed 2% to 5% (1000 to 2500 year return period), depending on the
number of people expected simultaneously and the effects on near buildings, if any.
Relative Structure Flexibility influence
Major influence of the flexibility ratio F on the structure racking distortion behavior can be
summarized as follows:
o F = 0; the structure is rigid, so it will not rack regardless of the ground distortion.
o F < 1; the structure is considered stiff relative to the medium and therefore distorts less.
o F = 1; the structure and medium have equal stiffness, so the structure will undergo
o approximately free – field distortions.
o F > 1; the structure racking distortion is amplified relative to the free – field, though not
o because of dynamic amplification, but because of the cavity – like presence.
o F → ∞; the structure has no stiffness, so it will undergo deformations identical to the
o perforated ground.
Hereafter design practical strategies will be analyzed for both cases, relatively rigid and flexible
structures.
Relatively Rigid Structures F < 1
Relatively rigid structures could be seen for small facilities in soft to very soft soils. The racking
distortion method leads to a racking coefficient R lower than unity (Eq. 2) as expected.
Obviously, this methodology is not applicable for F values near to zero, where no effects would
be predicted in a non – realistic and non – conservative scenario.
Relatively rigid structures are reducing the ground distortion, and therefore some soil thrust
would be expected at the lateral wall in the push side, and reduced accordingly in the opposite
side. Two additive effects on the structure can be considered separately (Fig. 3):
• Racking distortion considering the soil – structure interaction. It could be considered as
pseudo concentrated force P.
P = R · m · W (6)
• Local thrust on the lateral wall including the effect of the surrounding ground reduced
distortion. It can be evaluated as an inverted triangular distributed force, totaling P.
= (1 – R) · m = (1 – R) · σv · PGA/g · Rz (7)
P= · W (8)
Where:
P = Psuedo Concentrated force equivalent to the racking distortion effecct
R = Racking Soil – Structures Interaction Coeff. (Eq. 2)
m = Shear stress in free – field medium
= the amount of shear stress not assumed by the structure distortion
because of its stiffness,
Figure 3 – Additive load assumptions for relatively rigid structures
It should be pointed out that Eq. 6 and 7 are equivalent to the Wood’s expressions for rigid
structures when a value near zero is adopted for the racking coefficient R. In this way, the two
methods find a common expression: as the structure racks, the earth thrust predicted by Wood
for absolutely rigid structures is consequently reduced.
Relatively flexible structures F > 1
Most of the large underground facilities, like subway stations and cut and cover road tunnels,
in medium to rigid soils could be predicted as relatively flexible structures. Furthermore, as
the racking coefficient becomes almost constant for F values over 10 (Fig. 4), most of them
could be treated as relatively no – stiff without much difference in the numerical results.
Figure 4 – Racking coefficient, considering vm equal to 0.2, 0.3 and 0.4.
Mononobe – Okabe method is not able to predict the expected distortion; therefore, racking
distortion method appears as the most sensible and rational approach. As far as the structure is
more flexible than the surrounding ground, the structure distortion can only be imposed through
the shear stresses. What is more, a certain reduction of the static earth pressure on the lateral
walls is expected; this reduction could be estimated with the same expressions in Equation 6
and 7, with R > 1.0.
To properly evaluate the local effects at the structure lateral walls, the static “at rest” earth
pressure is reduced because of structure flexibility; but it should be checked that the total earth
pressure has to be kept not less than the Mononobe – Okabe earth pressure prediction.
= (R – 1) · m = (R – 1) · σv · PGA/g · Rz (9)
P = · W = (R – 1) · σv · PGA/g · Rz · W (10)
qmax = 2 · P / H (11)
q(z) = σv(z) · K0 – Δq(z) ≥ σv(z) · KMO (12)
Where:
q(z) = the final earth pressure including static and pseudo dynamic effect as
a function of the structure height
Δq(z) = the reduction of the static “at rest” earth pressure as a function of
the structure height
K0 = the static “at rest” earth pressure coefficient
KM0 = the Mononobe – Okabe dynamic active earth pressure coefficient
The above procedure offers a consistent methodology to evaluate the resultant earth pressure
including the static and the dynamic effect. Mononobe – Okabe predicted pressure is kept as a
minimum, assumed as the dynamic yielding active earth pressure if the enough strain is
developed from the “at rest” level. These local effects must be superposed to the racking
distortion effect.
In any case, the seismic forces induced in structural members’ decrease as the structure
flexibility increases. In such cases, the security of very flexible structures relies on the
structural details. It cannot be forgotten that structures with no moment resistance, such as
unreinforced brick arches, could totally collapse. For flexible box structures, some design
considerations are included hereafter, that could impose some limitations to the flexibility at
the design stage.
• The structural joints between roof and walls have to be thoroughly analyzed. The racking
distortion is imposed to the structure mainly through the shear stresses at the top of the
roof; therefore, to avoid their decompression, the joints have to be able to assume loads
Pjoints not less than
Pjoints = R · m · W (13)
• Where pinned connections are considered, the maximum rotation has to be checked and
validated. The support devices have to assume a dynamic rotation d, to be added to the
static requirements, not less than
tag θd = R · τm/Gm (14)
• The structure racking distortion stiffness Ks is supposed to be evaluated considering the
cracked concrete condition. However, very flexible structures would remain elastic even
with medium to high seismic event. Assuming reduced values of the structure stiffness will
overestimate the displacements and racking distortion (in the safe side); but on the other
hand it will underestimate the local earth pressure at the walls, or the loads at the structural
joints. Sound engineering judgment is required to establish in each case what it is the more
appropriate approach.
• Where the flexibility ratio F is much larger than 1.0, the additional distortion deformation
could be reflected to the earth surface, especially for shallow and very shallow facilities. If
there are near buildings, as usual for subway infrastructures in urban areas, the maximum
induced differential settlement has to be checked. The expected settlement can be estimated
from the value of the racking distortion, considering (R – 1)·Δff as the horizontal
deformation. As a general rule, differential settlement should be kept under the 1/500 limit
to prevent the occurrence of a serviceability limit state. It should be pointed out that the
existing buildings have had to be designed with their own Earthquake Resistant Strategy
(ERS), but not considering additional deformations from other infrastructures.
Conclusions
The free – field racking deformation method is the most reliable approach in practical terms
for the design of underground infrastructures. Some contributions to the applicability of this
methodology have been explained for shallow box – shaped tunnel structures. They are
applicable where the racking deformation is, as it is known:
• Bi – dimensional analysis is carried out. Three dimensional issues are not included.
• Surrounding ground is supposed uniform and isotropic.
• Liquefaction, faults or other geological hazards are not considered.
• Interaction with other tunnels, buildings or infrastructures is not included.
• Special geometries are not included; the overall shape is considered as a rectangular box,
whatever its inner structure elements distribution is.
A consistent and rational approach to unify the results from the racking deformation method
with Wood method for rigid structures and Mononobe – Okabe for flexible structures are
described, discussing on the flexibility factor as the most significant parameter. It is concluded
that the racking deformation method is able to determine not only the general racking distortion
effects, but the local effects of the earth pressure on the lateral walls with very simple
expressions.
Additionally, some considerations have been pointed out about the requirements of the flexible
structures. The roof and bottom slabs joints with the lateral walls have to endure the
compatibility forces and, if pinned connections are adopted, the rotations. Simple expressions
are given to use in the design. Moreover, other remarks related with the effects that could be
generated at the ground surface are included.
Finally, the plastic hinges formation strategy is discussed. It is not recommended for tunnel
facilities unless special measures could be adopted to repair the plastic joints after an
earthquake event, or the operational condition could be ensured. The risk level assumed should
be adopted considering the risk to human lives and the cost of the re – investment in case of
collapse; therefore, a minimum of 1000 year return – period is suggested for road tunnels, and
2500 year return – period for subway and other massive transportations systems.