1998 JEE Elnashai Pinho
1998 JEE Elnashai Pinho
1998 JEE Elnashai Pinho
net/publication/239794310
CITATIONS READS
51 755
2 authors:
Some of the authors of this publication are also working on these related projects:
Croatian Science Foundation project IP-06-2016-5325: Seismic base isolation of a building by using natural materials - shake table testing and numerical modeling View
project
All content following this page was uploaded by Rui Pinho on 16 January 2018.
In the context of capacity design philosophy, where a desired failure mode exhibiting
adequate levels of energy absorption capacity is envisaged, control must be exercised
on the member behaviour to safeguard the achievement of the target overall response.
Therefore, local repair and retrofitting methods that result in unquantifiable effects on
seismic response characteristics should be re-assessed. In contrast, techniques to affect,
in a controlled and easy-to-monitor fashion, individual design response parameters, i.e.
stiffness, strength and ductility, may provide a new framework for repair and retrofitting
earthquake-damaged structures to mirror ‘capacity design’ principles used for new struc-
tures. Such an approach is discussed in this paper and possible scenarios where selective
intervention may be required are identified. A number of tests on RC walls are also
reviewed to confirm the feasibility of the proposed intervention techniques. Finally, ex-
tensive parametric studies are carried out, using verified analytical models, leading to
the derivation of selective re-design expressions and guidelines.
1. Introduction
Conventionally, the philosophy behind repair/retrofitting schemes has been one of
over-strengthening. However, in the light of modern ‘capacity design’ concepts used
in seismic codes in which a significant role is assumed by the control of the overall
response of RC structures, such an approach can be counter-productive. Structures
designed to conform to a particular failure mode rely heavily on individual design
response parameters of its members, such as stiffness and strength in shear and
flexure, as well as ductility. Accordingly, repair or retrofitting methods which will
affect these characteristics should consider not only the local but also the global
effects of the intervention. In fact, local over-strengthening of individual members
may cause stiffness irregularities or disturb the sequence of plastic hinge formation,
thus jeopardising the whole retrofitting process.
Moreover, several recent studies [Moehle, 1992; Calvi and Pavese, 1995; Kowal-
sky et al., 1995] have introduced the concept of ‘displacement-based design’ as a
525
526 A. S. Elnashai & R. Pinho
logical and rational alternative to the currently used ‘force-based design’ method-
ology. In this new approach, a structure is designed to meet a target deformation
criterion whilst strength and stiffness become end-products of the design procedure.
Since capacity design requires control on the deformational demand and supply at
dissipative zones, it blends perfectly with displacement- (or more generally,
deformation-) based design. Such an approach to seismic design clearly requires
a higher level of control over the local behaviour of members and their global effect
on the seismic response of a structure.
Therefore, methods of structural intervention, especially of RC structures, have
to address these new requirements of local deformational behaviour control by pro-
viding tools for affecting individual design response parameters: stiffness, strength
and ductility. This would allow the fine-tuning of the local response characteristics
in favour of a desired global performance, enabling tighter control of the behaviour
and failure modes of RC structures, thus leading to a more rational assessment and
repair/retrofitting solution.
The three parameters governing displacement-based design (DBD) are period,
equivalent damping and global displacement. Since the period in DBD is the se-
cant value at the target displacement, it becomes a function of the initial stiffness,
strength and maximum displacement. The equivalent damping is a function of the
level of ductility, for it is mainly contributed to by hysteretic energy dissipation. If
separate control of stiffness, strength and ductility is afforded to the designer (in
a displacement-based assessment leading to repair/retrofitting design), then more
than one solution to achieve the target DBD objective is availed of. Hence, selective
intervention blends well with displacement-based assessment and re-design.
Such intervention philosophy is conceptually described in Fig. 1. The original
structure may have a ratio between capacity (∆u ) and demand (∆d ) displacements
that does not meet the target safety requirements. The three intervention scenar-
ios present different solutions to address such a situation, either by targeting the
structure initial stiffness (Ki ), its strength (Fy ) or ductility (µ). Which solution is
more economical and feasible will depend on the case under consideration, keeping
in mind both the structure characteristics and the displacement spectrum. Whilst
an increase in the level of equivalent damping (ξ) will decrease the displacement
demand, changes in the effective period of vibration of the structure (Tef f ) may
or may not be beneficial depending on the spectrum shape. It is clear, though,
that this view provides more degrees of freedom for the designers of intervention
schemes.
In this paper, tests performed at Imperial College [Elnashai and Salama, 1992]
to investigate the feasibility of affecting one response parameter with little or
no effect on others are re-assessed. An analytical model is assembled to repre-
sent the experimentally-verified intervention mechanisms for stiffness-, strength-
and ductility-only scenarios. The model realism is confirmed by comparison with
the available test data and further employed in a parametric investigation. Sim-
ple design-oriented expressions are derived, then analytically tested, to provide
Repair and Retrofitting of RC Walls Using Selective Techniques 527
Fig. 1. Variation in DBD design parameters due to selective intervention: (a) stiffness; (b) strength;
and (c) ductility.
relationships between the physical characteristics of the intervention and the per-
centage effect on the target response parameter. Finally, worked numerical examples
are given to demonstrate the potential for application of the derived expressions.
cyclic loading. The test results show significant increase in the ductility of the re-
inforced concrete members, whilst minimum change in the initial stiffness of the
members was observed. The ductile flexure-dominated response of the retrofitted
models enabled their full flexural capacity to be developed, hence a significant in-
crease in strength was also recorded.
Aviles et al. [1996] conducted a similar set of experiments on 18 column spec-
imens. Some of the models were deficient in the level of concrete confinement at
the foundation level. These were retrofitted with 1.2 mm thick steel plate wrapping
combined with anchor bolts. The effects of such interventions were visible only in
the increase of deformation capacity of the specimens, whilst no change in either
stiffness or strength was verified.
Saadatmanesh et al. [1997] carried out experimental work on the application
of high-strength fibre-reinforced plastic (FRP) composites straps to retrofit rectan-
gular bridge columns. Five 1/5-scale specimens were tested under reversed cyclic
loading. Both oval and rectangular shapes were used, and active confinement was
applied in one of the specimens by means of pressure injection of epoxy resin. The
increase in the level of ductility of the upgraded models is clearly shown in this
work. Shear failure was avoided, thus enabling full attainment of the initial flexural
capacity of the members.
Ghobarah et al. [1998] tested three 1/3-scale columns to investigate the effec-
tiveness of corrugated steel jacketing in the rehabilitation of reinforced concrete
columns. The jackets were constructed from commercially available corrugated
steel sheets and the gaps between the concrete and the steel jacket were filled with
grout to provide continuity between the two elements. Furthermore, the undulated
shape exhibits an out-of-plane stiffness which increases its efficiency in providing
external passive confinement to the rehabilitated members. These exhibited a sig-
nificant increase in deformation capacity, without any significant change in their
initial stiffness.
The above research shows excellent results in the increase of ductility of re-
inforced concrete columns. Since a gap is provided between the steel/composite
jackets and the foundation, no change in the original flexural capacity of the mem-
ber is introduced. Similarly, significant modifications in the stiffness of the columns
are prevented by the use of thin steel plates or high-strength composite materials.
The latter will induce changes in initial stiffness up to a maximum of 5% as opposed
to a 75% change if conventional concrete jacketing is used [Priestley et al., 1996].
17.5
15.0
Stiffness (kN/mm)
Intact
12.5
Repaired
10.0
7.5
5.0
2.5
0.0
0.0 5.0 10.0 15.0 20.0 25.0 30.0
Displacement (mm)
Fig. 2. Experimental results of intact and epoxy resin repaired walls [Elnashai and Salama, 1992].
to make it comply with new modern code design criteria, a selective intervention
that will enable the designer to concentrate uniquely on the parameter in question
is required. Otherwise, if repair or retrofitting methods which affect not only the
stiffness of the members but also their flexural capacity are applied, then a com-
plete structural redesign is needed, resulting in a more costly and time-consuming
solution.
For such design scenarios, where only lack of member ductility needs to be
addressed, the most effective and straightforward way of intervening is through a
selective approach that will controllably affect the defective parameter alone.
Steel Plates
Discontinuity gap
A minimum gap between the plates and the foundation or loading beam needs to
be imposed in order to avoid plate bearing due to flexural deformation. Since there
is no continuity between the plates and the loading beam or foundation, the critical
section remains the same, hence the flexural capacity is unaltered. Furthermore,
534 A. S. Elnashai & R. Pinho
the increase in the level of stiffness is totally controlled by the position of the plates,
their height, width and thickness, as demonstrated later.
The photograph of the specimen tested at Imperial College (shown in Fig. 4),
corresponds to a model repaired with 3.6 mm steel plates after being previously
damaged through cyclic loading with limited post-yielding deformation. The cor-
responding experimental results indicate that, unlike epoxy resin injection, the use
of EBSP enables full reinstatement of the initial stiffness of damaged walls (Fig. 5).
Moreover, steel plate adhesion is often used in repair and strengthening following
earthquakes, hence this scheme is practical.
20.0
17.5
15.0
Stiffness (kN/mm)
12.5
Intact
10.0
Repaired
7.5
5.0
2.5
0.0
0.0 5.0 10.0 15.0 20.0 25.0 30.0
Displacement (mm)
Fig. 5. Experimental results of intact and selectively repaired walls [Elnashai and Salama, 1992].
The presence of the steel plates will evidently affect the spread of inelasticity,
hence the relationship between local and global ductility. This can be overcome by
using plates with slits and/or thin (yielding) plates, to avail the development of a
plastic hinge along the height. This issue needs further investigation.
Slotted
Steel Plate
Mechanical
Couplers
Gusset Plates
External
Re-bars
(a) (b)
Fig. 6. Strength-only intervention - tested specimens: (a) EURB; (b) EUSP [Elnashai and Salama,
1992].
In the first scheme, illustrated in Fig. 6(a), steel bars are fixed to the top and
bottom beams and a mechanical coupler with an initial slack is used to allow for
a tolerance in displacement, beyond which the bars effectively contribute to the
flexural strength of the wall. The bars were fixed to the foundation beam using
high-strength epoxy patching mortar and to the top loading beam by means of a
steel assembly fixed to the concrete wall.
In the second scheme, shown in Fig. 6(b), steel plates are bolted to the wall
by steel angles and gusset plates. The IDM was achieved through the inclusion of
slotted holes at the gusset plates located on the loading beam, which allowed the
bolts to travel freely within the designed groove before bearing on the steel plate.
Hence, additional resistance will only be provided after the designed displacement
gap is exhausted.
The level of strength increase can be easily controlled by the area and posi-
tion of the steel plates or re-bars. Moreover, the displacement tolerance used in
the delay mechanism can also be manipulated to provide the designer with the
536 A. S. Elnashai & R. Pinho
possibility of choosing the ideal point from which the external mechanism should
start contributing to the flexural strength. This is demonstrated in subsequent
sections of the present paper.
The force vs. maximum cyclic displacement plot resulting from the test of the
scheme presented in Fig. 6(b) is shown in Fig. 7, where it is compared with the
behaviour of the reference model. The point where the external plates start in-
teracting with the wall is clearly visible, as was intentionally designed. The in-
crease in the specimen strength is also evident, whilst its initial stiffness remains
unaltered.
100
80
60
40
20
Force (kN)
0
Strengthened
-20
Original
-40
-60
-80
-100
-30 -20 -10 0 10 20 30
Displacement (mm)
Fig. 7. Experimental results of original and selectively strengthened walls [Elnashai and Salama,
1992].
This scheme can be successfully applied in practice by fixing the external bars
or plates to the top and bottom slabs. These should then be encased in a ductile
material or a plastic duct that will protect the steel from corrosion, provide fire
protection and allow for the free movement of the delay mechanism. Furthermore,
couplers are frequently used as an effective means of splicing reinforcement bars,
hence the EURB are only one step ahead of currently applied practice. Also, there
may be other more practical means of achieving the same end, e.g. by cutting bars
of longer length than required for immediate load transfer.
Finally, it is worth noting that this intervention can only be successful if the
wall has sufficient capacity to withstand the increased demand on both shear and
concrete compression triggered by the strengthening scheme. This issue is further
analysed later in this paper.
Repair and Retrofitting of RC Walls Using Selective Techniques 537
U-Shaped
Steel Plates
Fixing
Bolts
This technique was experimentally tested on a scaled wall initially designed and
built without any confinement reinforcement (Fig. 8). The 3.6 mm thick ECSP were
designed to provide the specimen with the same ductility capacity of an identical
model confined with internal steel hoops. Both walls were subjected to similar cyclic
loading regime until failure. The results from these two experiments are compared
in Fig. 9, where it can be seen that the target ductility level was achieved with the
application of the external plates, without any significant changes in either stiffness
or strength.
538 A. S. Elnashai & R. Pinho
80
60
40
20
Force (kN)
0
External Confinement
-20
Internal Confinement
-40
-60
-80
-30 -20 -10 0 10 20 30
Displacement (mm)
Fig. 9. Experimental results of internal and externally confined walls [Elnashai and Salama, 1992].
Fig. 10. Shear-strength intervention - tested specimen [Elnashai and Salama, 1992].
responded in a more ductile mode and its ultimate flexural strength was fully de-
veloped (Fig. 11). Concrete crushing was recorded at the boundary elements rather
than the web, as was originally the case.
These test results show that the shear strength of RC structural walls can be fully
reinstated and, if required, enhanced by bonding external horizontal steel plates to
the damaged wall without the need of epoxy resin injection. However, the use of
the latter would improve the shear stiffness further, increase the energy absorption
capacity and ensure higher durability for the retrofitting scheme.
The application of the shear-strength intervention technique may assume a vital
role in guaranteeing that retrofitted structures respond to dynamic loading in a
flexural-dominated behaviour, consistent with the re-design philosophy introduced
previously. However, since the analytical work developed in this paper focuses solely
on the flexural component of the seismic response of reinforced concrete walls, this
selective intervention scheme is not studied in the following sections.
150
120
90
60
30
Force (kN)
0
-30
-60
-90
-120
-150
-40 -30 -20 -10 0 10 20 30 40
Displacement (mm)
(a)
150
120
90
60
30
Force (kN)
-30
-60
-90
-120
-150
-40 -30 -20 -10 0 10 20 30 40
Displacement (mm)
(b)
Fig. 11. Experimental results of tested walls: (a) intact; (b) selectively repaired [Elnashai and
Salama, 1992].
accurately as possible these selective schemes. These are then verified by com-
parisons made with the test results and by conceptual understandment of their
behaviour.
with a large library of 2D and 3D elements which can be used with a wide choice
of typical pre-defined steel, concrete and composite section configurations.
For the purpose of this work, ten cubic plastic elements capable of modelling
distributed inelasticity are used to represent the test specimens. Each element has
two Gaussian sections, subdivided into 150 fibres where strains and stresses are
calculated. The concrete material representation is based on the uniaxial constant
confinement model proposed by Mander et al. [1988]. However, the cyclic rules have
been improved to enable the prediction of continuing cyclic degradation of strength
and stiffness, as well as better numerical stability under large displacement analy-
sis [Martinez-Rueda and Elnashai, 1997]. The behaviour of the reinforcement bars
is represented using the multi-surface plasticity model for steel, suggested by Pe-
tersson and Popov [1977]. This complex model is suitable for comparisons with
experimental results since it has the ability to represent a yield plateau and non-
linear strain hardening, as well as cyclic degradation. Experimental monotonic and
cyclic steel tests results available in the literature are used for the calibration of its
parameters [Martinez-Rueda, 1997]. The external plates and re-bars are modelled
using a bi-linear elastoplastic steel model with kinematic strain hardening since they
are not expected to undergo significant inelastic load reversals.
The modelling of the stiffness-only intervention scheme is carried out using an
eccentric steel section, whilst gap elements are utilised for the representation of the
strength-only intervention. As for the case of ductility-only intervention, a physical
representation of the intervention scheme is not sought. Instead, the effects of this
selective method are introduced directly in the concrete stress-strain relationship
through the use of a confinement factor, described later.
80
60
40
20
Force (kN)
-20
Analysis
-40
Experiment
-60
-80
-20 -15 -10 -5 0 5 10 15 20
Displacement (mm)
In Fig. 12, the envelopes of the flexural cyclic response of the intact wall ob-
tained in the experiment and using ADAPTIC, are shown. The results are in good
agreement thus leading to the conclusion that the FE modelling adopted accurately
represents the flexural cyclic behaviour of the tested specimens (similar agreement
was obtained for a number of wall analyses in this study, and for other struc-
tural members in previous studies [Elnashai and Elghazouli, 1993; Broderick and
Elnashai, 1994]).
The hardening effect obtained in the analysis at 12 mm (1% drift), which was
not observed during the test, may be attributed to the model used for the steel bars,
since this was not specifically calibrated to the steel employed in the experiments.
80
70
60
50
Force (kN) 40
Stiffness-only
30
Original
20
10
0
0 10 20 30 40 50 60 70 80
Displacement (mm)
Furthermore, analyses carried out in the parametric study described in Sec. 6.1,
undoubtedly demonstrate the effectiveness of the modelling adopted by providing
results which can be physically interpreted and understood. It is therefore possi-
ble to conclude that the developed FE model accurately represents this selective
scheme.
100
80
60
40
20
Force (kN) 0
-20
Analysis
-40
Experiment
-60
-80
-100
-20 -15 -10 -5 0 5 10 15 20
Displacement (mm)
Fig. 14. Force vs. flexural-displacement hysteresis envelope of wall retrofitted with EUSP.
2
IDM – Displacements (mm)
-2
-4
Left IDM
-6
Right IDM
-8
-10
0 2 4 6 8 10 12 14 16 18 20 22 24
No. of cycles
proposed model appears to simulate this selective technique rather well, providing
a useful tool for the upcoming parametric study.
60
50
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20 22 24
No. of cycles
80
60
40
20
Force (kN)
-20
Analysis
-40
Experiment
-60
-80
-20 -15 -10 -5 0 5 10 15 20
Displacement (mm)
Fig. 17. Force vs. flexural-displacement hysteresis envelope of wall retrofitted with ECSP.
In Fig. 18, plots of force vs. displacement of a monotonically loaded wall with
different intervention schemes are shown. The effects of the intervention on ductility
are clear whilst no change in the initial wall stiffness is noticed, confirming the
experimental observations.
Furthermore, in the extreme case of an increase in displacement ductility
by a factor of about 10, only a 30% increase in the member strength is
observed.
546 A. S. Elnashai & R. Pinho
120
100
80
Force (kN)
60
40
20
0
0 10 20 30 40 50 60 70 80 90
Displacement (mm)
Fig. 19. Typical load-displacement relationship for a reinforced concrete element [Paulay and
Priestley, 1992].
Repair and Retrofitting of RC Walls Using Selective Techniques 547
level, where knowledge of the real non-linear behaviour of the member is not gen-
erally pursued. It is therefore most suitable to the objectives of the present work
which, as mentioned earlier, focus on the development of practical analytical tools
for re-design of RC structures.
The slope of the idealised linear elastic response, K = Sy /∆y , is used to quan-
tify stiffness. It should be based on the effective secant stiffness to the real load-
displacement curve at a load of about 0.75Sy . An acceptable approximation is to
calculate this point based on the first yield of reinforcement [Paulay and Priestley,
1992]. The yield value of the equivalent elasto-plastic response, i.e. the ideal or
yield strength of the member, can thus be estimated as being Sy = 1.33Sy0 . Finally,
a convenient quantity to evaluate the member ductility capacity is the ultimate
displacement ductility, defined as µ∆ = ∆u /∆y . In the following sections, these
definitions are used to calculate the design response parameters of the walls.
ep
tp
lp
bw
lw/2 lw/2
Fig. 20. Plate position and section dimensions for stiffness-only intervention.
h/hw
1.0 FK
2.0
0.8
1.8
0.6
1.6
Extent of inelasticity
0.4
1.4
0.2 1.2
0.0 1.0
0.0000 0.0001 0.0002 0.0003 φ (1/mm) 0.0 0.2 0.4 0.6 0.8 1.0 hp/hw
(a) (b)
Fig. 21. (a) Curvature at maximum response; (b) Stiffness improvement vs. height of steel plates.
tp=6.0 mm
FK tp=1.8 mm
tp=0.3 mm
0
0
ep
0
lp
of FK are obtained in the range of smaller values of lp . Such effect can be clearly
observed in Fig. 22, particularly for the surface corresponding to tp = 6.0 mm.
Moreover, because flexure cracking is concentrated at the wall edges, having a
thicker but narrower plate shifted to the sides will be a more efficient way of gaining
stiffness than using a wider but slimmer plate. Nevertheless, care should always be
taken to ensure that the plates possess a contact surface with the wall sufficiently
wide to avoid problems of lack of bonding.
In the following parametric study, two different approaches are adopted for the
development of analytical tools for this intervention. The first approach aims at
the development of a single design expression, derived by regression analysis, which
returns the value of FK . This methodology is, however, too simplistic given the
complex influence of the plates geometric characteristics on the outcome of the
intervention. Therefore, alternatively, design curves which consider the influence of
each variable separately, are also developed. These give the designer better insight
and control on the effects of the intervention. Nevertheless, the simplicity of the
first method perhaps renders it more appropriate for design situations.
Ap ep 4tp lp ep tp lp ep
SK = constant × × = 100 × × = 400 2
, (6.2)
Aw lw b w lw lw b w lw
where Ap and Aw represent the total cross-section area of the plates and wall,
respectively.
550 A. S. Elnashai & R. Pinho
FK
4.5
4.0
3.5
2.0
FK = 1.2SK + 1 , 0 < SK ≤ 1
1.5 2
R = 0.96
1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 SK
A large number of analyses were carried out, varying the thickness, width and
position of the plates. The values of FK are plotted against the corresponding SK
in Fig. 23, where it is shown that a bi-linear polynomial fit to the data provides
excellent results with considerably high global correlation factors (R2 ).
It was also verified that an accurate representation of the role played by the
existence of different levels of longitudinal reinforcement or the presence of axial load
would require these variables to be included in the design expression. Hence, further
studies were carried out to include two additional parameters — reinforcement ratio
(ρs ) and normalised axial force (ν). The former is defined by Eq. (6.3), where
As represents the total area of longitudinal reinforcement in the wall, and was
considered in the range of 0.4 ≤ ρs ≤ (4%), as is generally adopted in design
applications. The variable ν is defined by Eq. (6.4), where N stands for axial force
and fc0 is the concrete compressive strength. For the purpose of this study, the value
of ν was varied in the range of 0 ≤ ν ≤ 0.10, as it is commonly observed in practice.
As
ρs = , (6.3)
b w lw
N
ν= . (6.4)
bw lw fc0
The final design expressions for this intervention can, therefore, be expressed as
follows:
The procedure followed to define the expressions shown above, with the inclusion
of axial force and reinforcement ratio effects, involved extensive parametric studies
and regression analysis, similarly to that outlined above for parameter SK . However,
for the sake of brevity, details of these have been omitted and can be found elsewhere
[Pinho, 1999] together with a more in-depth discussion of the effects that these
parameters have on the outcome of the intervention.
tp
α= , (6.6)
bw
lp
β= , (6.7)
lw /2
lp
ep − 2 2ep − lp
γ= = . (6.8)
lw
2 = lp lw − 2lp
γ=0
4.2
α = 0.10
3.8
α = 0.08
3.4
α = 0.06
3.0
α = 0.04
FK
2.6 α = 0.03
2.2 α = 0.02
1.8
α = 0.01
1.4 α = 0.005
1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
β
γ = 0.25
4.2
α = 0.10
3.8
α = 0.08
3.4
α = 0.06
3.0
α = 0.04
2.6
FK
α = 0.03
2.2 α = 0.02
1.8
α = 0.01
1.4 α = 0.005
1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
β
γ = 0.5
4.2
α = 0.10
3.8
α = 0.08
3.4
α = 0.06
3.0
α = 0.04
2.6 α = 0.03
FK
2.2 α = 0.02
1.8 α = 0.01
1.4 α = 0.005
1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
β
γ = 0.75
4.2
α = 0.10
3.8
α = 0.08
3.4
α = 0.06
3.0
α = 0.04
2.6
α = 0.03
FK
2.2 α = 0.02
1.8
α = 0.01
1.4 α = 0.005
1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
β
γ=1
4.2
α = 0.10
3.8
α = 0.08
3.4
α = 0.06
3.0
α = 0.04
2.6
FK
α = 0.03
2.2
α = 0.02
1.8
α = 0.01
1.4 α = 0.005
1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
β
level of a plate with a particular thickness is reached for smaller values of width
(Fig. 28). This confirms the statement made previously, where it is proposed that
maximum of stiffness-only intervention effect can be achieved by using relatively
narrow plates located as close as possible to the wall extremities. Further, for
values of γ = 1, Fig. 28 seems to suggest that ideal solutions should be obtained for
bp ≤ lw /4 (β ≤ 0.5).
554 A. S. Elnashai & R. Pinho
0 < FK ≤ 2 ρ,ν
FK = FK [1.1 − 3.24(FK − 1)ν]ρs−0.6 ln(FK ) , (6.9a)
FK > 2 ρ,ν
FK = FK [1.1 − (0.2FK + 2.9)ν]ρs0.03FK −0.48 . (6.9b)
Both the design curves and expressions derived by regression analysis revealed
a maximum coefficient of variation of 15% when compared with a wide range of
results obtained by finite element analyses on a large variety of walls. These results
confirm that the parameters defined to characterise the selective intervention and
the derived design expressions (or curves) appropriately account for all the variables
that most influence the effectiveness of this repair scheme.
S upgraded
FS = (6.10)
S original
dp lw dp
A p , f yp
100
90
80
70
Force (kN)
60
50
gIDM = 0 mm
40
gIDM = 5 mm
30
Original
20
10
0
0 10 20 30 40 50 60 70 80
Displacement (mm)
From section analysis, the value of curvature at the point of first yield (φ0y )
and the corresponding neutral axis depth (c0y ) can be easily determined at the
foundation level. At this loading stage, it is also possible to assume the latter as
constant throughout the height of the wall, whilst the curvature varies linearly.
Finally, taking into consideration the relationship between Sy and Sy0 , suggested
556 A. S. Elnashai & R. Pinho
earlier, a lower and upper bound limit for the design gap of the delay mechanism
is defined by:
hw hw
(dp + lw − c0y )φ0y ≤ gIDM ≤ (dp + lw − c0y )φ0y . (6.12)
2 1.5
Alternatively, simplified methods of seismic design in which a behaviour factor is
used to obtain the design load, i.e. the ideal or global yield strength of a reinforced
concrete member (Sy ), can also be used to obtain the upper bound limit for the
IDM gap value. In this way, the design of the strength-only mechanism would be
consistent with the wall original design philosophy. By characterising the curvature
at the base of the wall as
Md Fd hw Fe hw
φd = = = , (6.13)
Ec Ie Ec Ie qEc Ie
where Fe represents the elastic horizontal load, q stands for behaviour factor and Ie
is the equivalent moment of inertia of the cross-section at first yield in the extreme
fibre, and substituting (6.13) in (6.12), the following alternative expression can be
used to define the value of gIDM :
hw Fe h2w
(dp + lw − c0y )φ0y ≤ gIDM ≤ (dp + lw − cd ) . (6.14)
2 2qEc Ie
It is, nevertheless, important to notice that these expressions are derived from
principles which apply to monotonic loading only. Under cyclic regimes, cumulative
plastic deformations, mainly associated with cracks that remain open after load
reversal, will raise uncertainties regarding the vertical elongation of the wall. These
factors cannot be addressed by the expressions suggested above. Further, because
these expressions are derived assuming linear curvature distribution at point of first
yield, their application in design situations where such an assumption is not valid
(for instance, high axial load), will produce slightly inaccurate results. Still, they
do provide a first insight into the control of this intervention parameter, and should
be used at design level provided there is awareness of its limitations.
120
100
80
Force (kN) 60
Fp = 36 kN ; dp = 400 mm
40 Fp = 72 kN ; dp = 0 mm
Original
20
0
0 10 20 30 40 50 60 70 80
Displacement (mm)
smaller external plates or re-bars are placed at greater distances from the wall, the
same strength can be obtained with a smaller ductility decrease.
In Fig. 31, the plots of analyses carried out for two identical walls with different
configurations of strength-only interventions are shown. Although both walls ex-
hibit identical strength enhancement, their ductility is completely different. The use
of larger distances between the external mechanism and the wall proves extremely
efficient in avoiding high levels of ductility degradation. Hence, such configura-
tion should be applied in the upgrading of walls in which the available ductility
does not comply with the demands arising from a ‘short-distance’ strength-only
intervention. Moreover, its applicability is still feasible, particularly if there are
infill panels by the sides of the wall, since the external mechanism can be located
at a greater distance embedded in the infill panel without adverse architectural
effects.
Ap fyp dp + lw /2 Ap fyp dp + lw /2
Ss = constant × × = 10 , (6.15)
As fy lw As fy lw
where fy represents the yield strength of the longitudinal reinforcement of the wall.
558 A. S. Elnashai & R. Pinho
FS
2.05
1.90
1.75
1.60
FS = 0.22SS + 1
1.45
R2 = 0.97
1.30
1.15
1.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 SS
In Fig. 32, the polynomial fit to the results of the parametric investigations with-
out axial load is shown. A global correlation factor of 0.97 indicates the adequacy
of the linear regression used.
The final design expression, which includes the variables discussed above to-
gether with the existent level of axial load, becomes
The results obtained by Eq. (6.16) compare extremely well with the values given
by finite element analysis of several different RC walls, revealing a maximum coef-
ficient of variation of 4%. Hence, it is concluded that all the wall properties and
retrofitting scheme characteristics, which are most relevant for the outcome of the
intervention, are considered in the proposed methodology.
is defined as the ratio between the externally confined and unconfined concrete
0
compressive strength, fcc and fc0 , respectively.
0
fcc
kD = (6.17)
fc0
Effectively
confined core Pb
45°
dc tp
hp
Pb
sp bc
(i) the U-shape steel plates are stiff enough to avoid being pushed outwards due
to lateral confinement pressure, therefore reduction in plan of the effectively
confined concrete core due to arching occurs only in the vicinity of the
closing bolt (see Fig. 33);
560 A. S. Elnashai & R. Pinho
Effectively
confined core
s'p 45 ° sp
Consequently, the effectively confined concrete core area (Ae ) at the most un-
favourable location (midway through two consecutive plates) is given by
d2 sp + s0p s0p
Ae = bc dc − c 1− 1− , (6.18)
6 4bc 2dc
with ρcc representing the ratio of longitudinal reinforcement to the confined concrete
core area (Acc ).
The prestressed bolt is designed to work as the fourth side of the external hoop,
capable of transmitting the lateral stresses to the steel plates. To guarantee equilib-
rium of confining stresses, the bolt should be designed so as to assure that its axial
capacity equals that of the opposite plate. This may involve the use of a prestress
force such that
fyp Ap = fyb Ab + Pb , (6.20)
Repair and Retrofitting of RC Walls Using Selective Techniques 561
where Ab and Ap represent the cross-section area of the bolt and steel plate, respec-
tively, whilst their yield strength values are introduced in the expression by fyb and
fyp .
Since ke takes into account all the discontinuities present in this scheme, and
the bolt is designed to ensure a closed hoop behaviour of the whole assembly, the
effective section area ratios of transverse reinforcement (ρx and ρy ) can be defined
by
2Ap 2hp tp
ρx = = (6.21a)
dc sp dc sp
2Ap 2hp tp
ρy = = . (6.21b)
bc sp bc sp
By virtue of force equilibrium between the lateral confinement stress resultants and
0 0
the tension forces in the plates, the effective lateral confining stresses (flx and fly )
are determined by
0
flx = ke ρx fyp (6.22a)
0
fly = ke ρy fyp . (6.22b)
Finally, by using the general solution curves described in the work of Mander et
al. [1988], the ductility-only selective intervention parameter kD can be determined,
and the stress-strain relationship of the concrete adjusted to incorporate the effect
of the intervention.
As mentioned earlier, this approach is equivalent to methods currently used in
the design of reinforced concrete members confined with internal steel hoops. Hence,
its application in design situations is equally practical and straightforward.
kD
2.4
2.2
2.0 hp = 25 mm
hp = 50 mm
1.8
hp = 100 mm
1.6
1.4
1.2
1.0
25 50 75 100 125 150 175 200 225 sp (mm)
2.4
tp = 10 mm tp = 5 mm tp = 3 mm
2.2
2.0
Smaller Cross
1.8 Section
kD
1.6
1.4
1.2
1.0
0 10 20 30 40 50 60 70 80 90 100
hp (mm)
380, 275 and 207 mm2 , respectively. Therefore, using high plates closely spaced is
the most efficient way of achieving a confinement factor that will satisfy a target
ductility.
7. Closing Remarks
The effects on structures of currently used methods of repair and retrofitting re-
inforced concrete members are generally not easy to quantify or control and their
influence on the global behaviour of RC structures may not optimise the overall
seismic response. New earthquake-resistant design trends require that the global
response and failure mode are fully controlled, an objective not readily achieved
Repair and Retrofitting of RC Walls Using Selective Techniques 563
Further tests are planned for the immediate future under the auspices of the
European network ICONS (Innovative Concepts for Seismic Design of New and Ex-
isting Structures) and supported by ECOEST II (European Consortium of Earth-
quake Shaking Tables). These will focus on testing the design expressions under
simulated earthquake action and assessing the efficiency of mixed intervention types
under both repair and retrofitting design scenarios. Moreover, in the strength-only
scheme, an interaction device such as a spring mechanism will be developed in or-
der to reduce the abrupt stress increase caused by earthquake loading in the delay
mechanism.
564 A. S. Elnashai & R. Pinho
Acknowledgements
The experimental work was carried out by Dr. A. I. Salama, as part of his doc-
toral thesis at Imperial College. The tests were funded by the UK Science and
Engineering Research Council. The current work is supported financially by the
European Community through the research network ICONS (Innovative Concepts
for Seismic Design of New and Existing Structures) and by the Calouste Gulbenkian
Foundation.
Appendix
Examples of application of the design expressions
For the purpose of this example, let us consider a 200 × 1500 × 3000 mm wall.
The concrete compressive strength is fc0 = 30 MPa and the yield stress of the steel
re-bars is fy = 400 MPa. The longitudinal reinforcement in the boundary elements
is made of six 12 mm bars, whilst in the web 10 mm bars spaced at 200 mm are
used as vertical reinforcement (four bars at each face). Confinement reinforcement
is not available.
Stiffness-only intervention
The objective of the intervention is to increase by 50% the stiffness of the wall
using 4-mm thick steel plates. It is assumed that due to architectural constraints
the external plates can only be applied at a distance of 100 mm from the wall edges.
An axial load of 200 kN is also considered.
200 × 103
ν= = 0.022; FK = 1.5; tp = 4 mm;
200 × 1500 × 30
lp =?; ep =? (free edge ≥ 100 mm)
The value of the selective intervention parameter SK is determined using Eq. (6.5a):
4 × lp × (650 − lp /2)
0.21 = 400 ⇔ lp ≈ 100 mm
200 × 15002
Hence, the solution is to use 100 × 4 mm plates at a distance of 600 mm from the
centre of the wall.
Strength-only intervention
√
Fe = 350 kN (q = 2); Ec = 4.7 30 ≈ 26 GPa; FS = 1.3;
The value of the selective intervention parameter SS is determined from Eq. (6.16),
φ0y = 1.6 × 10−6 mm−1 ; c0y = cd = 240 mm; Ie = 9.46 × 109 mm4
and using Eq. (6.14), lower and upper bounds for the IDM gap design value are
Ductility-only intervention
The objective of the intervention is to achieve a ratio of five between the ultimate
and peak confined concrete strain, by means of external U-shaped plates with yield
strength of 275 MPa. These are 4 mm thick and spaced by 20 mm to avoid crack
arrest. The wall boundary elements are 250 mm wide and 200 mm deep.
6 × π × 122 /4
ρcc = = 0.014; εcu /εcc = 5; tp = 4 mm; s0p = 20 mm; hp =?
250 × 200
An initial value of 100 mm is arbitrated for the plates height in order to initiate
the trial procedure. The spacing between plates sp thus becomes equal to 120 mm
(see Fig. 34). By substituting in Eq. (6.19) the confinement effectiveness coefficient
is evaluated:
200 120 + 20 20
1− 1− 1−
6 × 250 4 × 250 2 × 200
ke = = 0.72
(1 − 0.014)
Using the general solution curves described in the work of Mander et al. [1988], the
ductility-only selective intervention parameter kD can be determined:
0
flx 275
2 × 100 × 4
ρx = = 0.033 f 0 = 0.72 × 0.033 ×
30
= 0.22
200 × 120 c Mander et al. [1988]
→ 0 −−−−−−−−−−−−−−−−−−→ kD ≈ 2
2 × 100 × 4 fly
275
ρy = = 0.027
250 × 120 0 = 0.72 × 0.027 × = 0.18
fc 30
The peak and ultimate confined concrete compression strains are estimated by the
following expressions [Paulay and Priestley, 1992]:
Thus, the ratio between peak and ultimate compressive strain becomes equal to
0.062/0.012 = 5.2, which satisfies the proposed target. The closing bolts are de-
signed to satisfy Eq. (6.20):
Hence, the retrofitting solution consists of utilising U-shaped steel plates with a
thickness of 4 mm, and a height of 100 mm, together with 40 mm bolts.
References
Aboutaha, R. S., Engelhardt, M. D., Jirsa, J. O. and Kreger, M. E. [1996] “Retrofit of
concrete columns with inadequate lap splices by the use of rectangular steel jackets,”
Earthquake Spectra 12(4), 693–714.
Repair and Retrofitting of RC Walls Using Selective Techniques 567
Priestley, M. J. N., Seible F. and Calvi, G. M. [1996] Seismic Design and Retrofit of Bridges
(John Wiley & Sons, Inc., New York).
PWRI [1986] Public Works Research Institute, Manual for Repair Methods of Civil Engi-
neering Structures Damaged by Earthquakes, Ministry of Construction, Japan.
Saadatmanesh, H., Ehsani, M. R. and Jin, L. [1997] “Seismic retrofitting of rectangular
bridge columns with composite straps,” Earthquake Spectra 13(2), 281–304.
Xian, D. [1992] “Inelastic response and effective design of asymmetric buildings under
strong earthquake loading,” Ph.D. Thesis, University College, University of London,
London.
UJCEERP [1981] US/Japan Cooperative Earthquake Engineering Research Program,
Proc. Second Seminar on Repair and Retrofit of Structures, Tsukuba, Japan.