Accepted Manuscript

Strut-and-tie models of repaired precast concrete bridge substructures with

CFRP shell

Dylan N. Brown, Joel E. Parks, M.J. Ameli, Chris P. Pantelides

PII: S0263-8223(15)01029-6
DOI: http://dx.doi.org/10.1016/j.compstruct.2015.11.026
Reference: COST 6981

To appear in: Composite Structures

Please cite this article as: Brown, D.N., Parks, J.E., Ameli, M.J., Pantelides, C.P., Strut-and-tie models of repaired
precast concrete bridge substructures with CFRP shell, Composite Structures (2015), doi: http://dx.doi.org/10.1016/
j.compstruct.2015.11.026

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 STRUT-AND-TIE MODELS OF REPAIRED PRECAST

CONCRETE BRIDGE SUBSTRUCTURES WITH CFRP SHELL

Dylan N. Browna, Joel E. Parksb, M.J. Amelib, Chris P. Pantelidesb*

Abstract: Strut-and-tie models (STM) have been developed for original and repaired precast

reinforced concrete (RC) bridge assemblies tested under quasi-static cyclic loads. The

original assemblies, built using accelerated bridge construction techniques, are half-scale

column-to-footing and column-to-pier cap specimens connected using grouted splice sleeves.

The repaired assemblies are strengthened over the plastic hinge region using a unidirectional

carbon fiber reinforced polymer (CFRP) shell, post-installed headed steel bars, and non-

shrink concrete. The repair technique was successful and the load and displacement

capacities of the damaged assemblies were restored. Generic modeling parameters are

established for the STM procedure, enabling the models to be adapted to new repair

applications. Special attention is focused on the struts within the CFRP shell. All assemblies

are modeled using sectional analysis in conjunction with the predicted STM load to estimate

a bilinear force-displacement response. The STM and bilinear force-displacement response

envelopes obtained from the models show satisfactory agreement with the experiments of the

original and repaired bridge assemblies in terms of initial stiffness, lateral load and

displacement capacity.

Keywords: analysis; bridge; FRP composite; repair; seismic; strut-and-tie.

a
Michael Baker International, Madison, Wisconsin 53719, USA
b
Dept. of Civil and Env. Eng., Univ. of Utah, Salt Lake City, Utah 84112, USA
*
Corresponding author. Tel.: 801-585-3991; Fax: 801-585-5477; E-mail: c.pantelides@utah.edu

1
1. Introduction

Strut-and-tie models (STM) are used to analyze reinforced concrete (RC) members

with disturbed internal stress-strain distributions, known as D-regions, to ensure force

equilibrium at the ultimate limit state. The STM method is widely accepted in practice [1-3]

and in concrete design codes [4-7]. There is limited application of the STM method in

analysis of concrete structures rehabilitated with FRP composites [8-12]. This paper extends

the use of the STM method for cyclically tested precast bridge assemblies; this includes

original and repaired bridge assemblies with carbon fiber reinforced polymer (CFRP) shells,

headed steel bars, and non-shrink concrete. Techniques previously used for analyzing

concrete with CFRP composite jackets [8, 9] and headed steel bars [13] are implemented in

the STM models and are used to create a bilinear force-displacement response. The results

are compared with the hysteretic curves of all specimens to assess the validity of the STM

approach to model repaired concrete assemblies damaged under cyclic loads.

The specimens being modeled are precast RC bridge column-to-footing and column-

to-pier cap assemblies which are connected using grouted splice sleeve (GSS) connectors

[14]. Bridge piers constructed with GSS connectors are gaining recognition as a feasible

accelerated bridge construction (ABC) technique in regions of moderate-to-high seismicity

[14-17]. The GSS sleeve connection has recently been used to construct the bridge piers of

the Utah Frontrunner light rail, which is located in a seismic region. With increasing use of

GSS connections for bridge piers of both rail and vehicular bridges, a post-earthquake repair

technique is needed. Such a repair using CFRP shells has been developed and the test results

have shown satisfactory performance when the post-earthquake repair is implemented [18].

An analysis technique is developed herein to evaluate the performance of the original and

repaired precast RC bridge assemblies connected using GSS. The STM procedure and

2
bilinear force-displacement response analysis technique developed in this paper provide an

alternative to finite-element analysis and can be adapted to actual RC bridge assemblies.

2. Experimental Program

Two RC precast specimens representing half-scale bridge elements, conforming to

current seismic bridge design standards [19], were constructed utilizing two different GSS

sleeve systems. The precast concrete specimens were assembled and tested to failure under a

quasi-static cyclic lateral load and a constant axial load [14, 20]. Upon completion of the

tests, the specimens were repaired and retested under the same cyclic load to failure [18].

Each specimen was tested twice, first in the original condition and second after it was

repaired. The naming convention of the specimens is an abbreviation of the GSS system

used; sleeves with both bars grouted (NM) or sleeves with one bar grouted and the other bar

fastened (LE); this is followed by the stage of testing, original (O) or repaired (R).

2.1 Original Specimens

Specimen NM-O is a column-to-footing assembly connected with NM GSS sleeves

and high strength nonshrink grout on both sleeve ends to splice the footing and column bars,

as shown in Fig. 1(a). Specimen LE-O is a column-to-pier cap assembly connected with the

LE GSS sleeve system which uses a threaded connection on the factory dowel end of the

sleeve and a grouted connection on the field dowel end, as shown in Fig. 1(b). The GSS

connectors are precast into the footing for the NM-O and the pier cap for the LE-O

specimens.

The geometry and reinforcement of the original specimens is shown in Fig. 2. The

columns are 2.59 m high with a 0.53 m wide octagonal cross-section. The longitudinal

reinforcement consists of six 25 mm bars arranged in a circular pattern. A 13 mm spiral at a

64 mm pitch is provided for the column transverse reinforcement. The footing is 1.83 m long,
3
0.61 m deep and 0.91 m wide, as shown in Figs. 2(a) and 2(b). The pier cap is 2.74 m long,

0.61 m deep and 0.61 m wide, as shown in Figs. 2(c) and 2(d). The longitudinal and

transverse reinforcement in the footing and pier cap are also shown in Fig. 2. Both NM-O and

LE-O specimens were tested under quasi-static cyclic loads until a minimum 20% decrease in

lateral load carrying capacity was observed.

2.2 Repaired Specimens

The same repair was employed for both specimens NM-O and LE-O after the original

cyclic tests were completed. The repair was designed to strengthen the original plastic hinge

region, subsequently relocating the new plastic hinge region to the column section adjacent to

the repair. This was achieved by increasing the column cross-section from a 530 mm

octagonal section to a 762 mm circular section for a column length of 457 mm, as shown in

Fig. 3. After the repair, NM-O and LE-O are referred to as NM-R and LE-R, respectively.

The repaired cross-section consisted of a CFRP shell filled with nonshrink concrete and

reinforced with post-installed epoxy anchored headed steel bars. Six 25 mm mild steel headed

bars were placed around the column section and post-installed into the footing and pier cap

for NM-R and LE-R, respectively. Four layers of unidirectional CFRP composite were

provided, with the fibers oriented in the hoop direction to form a CFRP shell; the number of

layers was determined based on enhancement of flexural, shear and confinement

considerations of the enlarged cross section [21]. Non-shrink concrete was used to fill the

void between the original column and the CFRP shell.

2.3 Experimental Results for Original and Repaired Specimens

The test setup and loading for the original and repaired specimens is shown in Fig. 4.

A constant axial load equivalent to 6% of the axial load capacity of the column was applied.

A quasi-static cyclic lateral load in terms of a drift ratio was applied in the East-West

direction as shown in Fig. 4(a) with two cycles per increasing drift as shown in Fig. 4(b). Tie
4
down rods were used to secure the specimens to the strong floor, providing horizontal and

vertical reaction forces. Bearing plates were used to transfer the axial load to the column from

the post-tensioned axial rods. This information and the ultimate lateral load are used to

establish the external loading used in the STM models.

The failure mode of original specimens NM-O and LE-O was fracture of an extreme

longitudinal bar and crushing of the concrete, as shown in Fig. 5(a) for specimen NM-O. The

extreme east longitudinal bar fractured in NM-O and the extreme west longitudinal bar

fractured in LE-O. At failure of both original specimens, the lateral load capacity dropped

well below 20% of the ultimate load. The reserve lateral load capacity of the original columns

ranged from 44% to 65% of the maximum lateral load capacity. A very well developed

plastic hinge was formed at the footing-to-column and column-to-pier cap interfaces, as

shown in Fig. 5(a). Extensive spalling and cracking occurred in the plastic hinge region of

the original specimens. The original specimens experienced flexural cracking which extended

up to 356 mm away from the footing or pier cap interface.

The successful plastic hinge relocation of NM-R is shown in Fig. 5(b). The strength

and displacement capacity of the damaged bridge columns was restored for both specimens

as evidenced by the hysteresis curves shown in Fig. 6. Specimens NM-O (Fig. 6(a)), LE-O

(Fig. 6(b)), and NM-R (Fig. 6(c)), were tested under quasi-static cyclic loading. Specimen

LE-R (Fig. 6(d)) was tested monotonically prior to cyclic testing; results of the monotonic

pushover test are given in Fig. 7. The cyclic results for specimen LE-R will only be

compared to the bilinear force-displacement response model when displaced in the opposite

direction as the monotonic test.

From Figs. 6(a) and 6(c) it is shown that specimen NM-R achieved an 18% higher

lateral load than NM-O and had a similar displacement capacity. The failure mode of NM-R

was fracture of column longitudinal bars in the relocated plastic hinge region after yielding of
5
the headed bars. The failure mode of LE-R was facture of the extreme east longitudinal bar in

the relocated plastic hinge region which occurred after yielding of the headed bars; overall

specimen LE-R performed similarly to LE-O.

3. Strut-and-tie Model Formulation

Four STM models were established following ACI 318 [4], for the original and

repaired specimens to predict the ultimate lateral load. Strength reduction factors were not

used in the model so comparisons to the test results could be made. The STMs developed are

determinate truss models which do not rely on the geometrical or material properties of the

members to obtain member forces. Allowable design forces for compression struts and nodes,

and tension ties were established following ACI 318 guidelines. The STM procedure outlined

within ACI 318 is not intended for modeling lateral force resisting elements. However, due to

the conservative nature of STM models, ultimate load prediction matches well the cyclic

capacity of the elements; this agrees with previous research findings [22]. The lateral load at

which an allowable design force is exceeded for any member within the STM model is the

ultimate load capacity of the specimen.

3.1 Strut-and-tie Model Layout

Strut-and-tie models require designer input that can only be established through

understanding of the stress trajectories within the structure and from experience. This leads to

different model layouts from different designers. The present research has established generic

parameters to standardize the STM layout of the repaired assemblies, allowing the models to

be adapted to new applications in terms of geometry and reinforcement configuration.

The STM layouts developed for all specimens are shown in Fig. 8 and the generic

model inputs used to establish the models are presented in Table 1. The location of external

load application within the models was determined based on the experimental configuration.
6
The lateral cyclic load, Pcy, is applied to the column in the horizontal direction for each

specimen. The vertical axial load, Pax, is applied through external post-tensioned rods into

the column and footing or pier cap through bearing plates. The external support reactions in

Fig. 8, Rx and Ry, are located where tie down rods were used to secure the specimens to the

strong floor, as shown in Fig. 5.

3.1.1 B-regions

The STM layout for the columns represents the uniform force transfer which occurs in

Bernoulli, B-regions, and uses three generic layout parameters θ, xt and xc, shown in Fig. 8.

Angle θ is measured between the axis of the diagonal strut and transverse reinforcement; the

tension force centroid at first yield is xt, and the compression force centroid at first yield is xc.

Angle θ remains constant between all diagonal struts and transverse tension ties for a given

model. The recommended range of values for θ is between 31 and 59 degrees [6]. For the

present study, the largest value of θ, which remained within the recommended values and

produced an integer number of equally spaced transverse tension ties, n, was used. Eq. (1)

can be used to find n and θ by varying n until the value of θ is within recommended limits:

= tan  
ℎ −  −  
(1)

where Hd is equal to Hcol minus Hrep for the repaired specimens and is the height from the

point of lateral load application to the closest longitudinal bars in the footing or pier cap for

specimens NM-O and LE-O; Hd is shown in Fig. 8 and h is the column width.

The tension and compression force centroids at first yield, xt and xc respectively, are

found from a sectional analysis of the original column. The first yield state is defined by the

onset of extreme tension steel yielding, and is shown in Fig. 9(a). Once xt and xc are

determined, the nodes on the compression face of the original column are located a distance

xc away from the extreme compression fiber and the nodes on the tension face of the original
7
column are located a distance xt away from the extreme tension fiber. Any reinforcement

located within the tension zone at first yield contributes to the tension tie area, as shown in

Fig. 9(a). The tension ties for NM-O and LE-O represent the column longitudinal bars and

extend through the GSS sleeves into the footing and pier cap, respectively.

The pier cap layout for LE-O and LE-R is found by maintaining a constant angle

between all diagonal struts and transverse tension ties. This angle does not need to be the

same as θ, but should fall within the recommended bounds of 31-59 degrees. The

longitudinal compression struts and tension ties are located at the height of the top and

bottom longitudinal reinforcement centroids. All tension ties in the STM layout correlate well

with the actual reinforcement arrangement, which consists of transverse and longitudinal

steel, and the CFRP shell; concrete tensile capacity is neglected.

3.1.2 D-regions

The repaired column section, referred to as the repaired cross-section, was modeled

using four generic layout parameters xt’, xc’, HCFRP and Hhead. The tension and compression

force centroids of the repair at first yield, xt’ and xc’ respectively, are found from a sectional

analysis of the repaired cross-section and are shown in Fig. 9(b). The first yield state is

defined by the onset of steel yielding in tension for the extreme headed bar. The contribution

of longitudinal bars located in the column of the original specimen is neglected in the

sectional analysis of the repaired section due to the damage state of this reinforcement. In the

experiments, the extreme longitudinal column bars had fractured in both NM-O and LE-O.

The confined concrete compressive strength within the repaired section, fcc’, is used in the

sectional analysis due to the high confinement provided by the CFRP shell. Prediction of fcc’

can be determined from previous research [23-25]. The tension ties representing the headed

steel bars are located a distance xt’ away from the extreme tension fiber of the repaired

section. Headed steel bars located within the tension zone at first yield, contribute to the area
8
of the headed rebar tension tie shown in Fig. 9(b). The compression force centroid in the

repaired section at first yield, xc’, is used to locate the node which transmits the compressive

forces from the repair into the footing or pier cap. This node is located a distance xc’ away

from the extreme compression fiber of the repaired section and at the elevation of the tension

tie representing the longitudinal steel bars in the footing or pier cap, as shown in Fig. 8.

The elevation of the horizontal CFRP tension tie in the repair cross-section, HCFRP,

was determined from strain gauge data recorded during testing [21] and is shown in Fig. 8(e).

The centroid of the tensile force within the CFRP composite at first yield of the column steel

reinforcement was used to determine HCFRP. HCFRP was calculated to be 320 mm, correlating

to 70% of Hrep.

The tension tie representing the headed steel bars extends for a length into the footing

or pier cap equal to Hhead, which is taken as the embedment length of the headed steel bars

minus half the development length of the epoxied headed bars, as shown in Fig. 8(e). Tension

ties originating at half of their development length have been used in previous studies [2].

The node at the top of the headed steel bars is placed at the location of the head. This is due

to the drastically shortened development length of headed steel bars [13].

In the STM model of the repair, the tension tie representing the longitudinal column

reinforcement was terminated away from the GSS sleeve interface due to the fact that

longitudinal column reinforcement fractured in the tests of the original specimens near the

GSS sleeve interface. The longitudinal column reinforcement tension tie node N1 is located

169 mm and 166 mm away from the GSS sleeve interface, for specimens NM-R and LE-R

respectively. This distance is less than half the development length of the bars, as shown in

Fig. 8(e). The clamping force provided by the CFRP shell truncates the development length

of the longitudinal bars within the repair and is the reason for the placement of node N1. The

reduced development length was demonstrated during testing of NM-R where a column
9
longitudinal bar fractured for the second time 546 mm, or 50% of the design development

length, above the fracture location in the original NM-O specimen. The short development

length shows the significant clamping force on the longitudinal steel bars from the CFRP

shell and non-shrink concrete, which is represented in the STM by anchorage from diagonal

struts.

3.2 Nominal design strength

The lateral load at which the nominal unfactored design strength for any member

within the model is exceeded is defined as the ultimate load capacity of the specimen. The

nominal design strength of a member is found by multiplying the effective stress capacity of

the member, governed by the material properties and stress configuration, by the controlling

member area.

3.2.1 Effective stress capacity

Four different materials are present in the model of the repaired specimens: concrete,

mild steel and headed steel bars, CFRP composite and confined concrete. Only a few studies

have been carried out using an STM formulation for RC members strengthened with CFRP

composites [8-12]. There is one tension tie representing the contribution of the CFRP shell as

shown in Figs. 8(c), 8(d) and 8(e). The strength of the tension tie was found by taking the

cross-sectional area of the shell multiplied by the allowable effective tensile stress. The

effective stress of the tension tie representing the CFRP shell is the product of the CFRP

modulus and strain at failure, modified by the strain efficiency factor, ҡε. Values of ҡε have

been determined experimentally and account for the multiaxial state of stress acting on the

CFRP shell and stress concentration effects. ACI Committee 440 recommends a value of ҡε

between 0.57 and 0.61 [26] and for the current study, ҡε has been taken conservatively as

0.57.

10
 The effective stress capacity of concrete struts was determined by taking the smaller

value of the effective stress capacity of struts and nodes. The effective stress capacity of

struts and nodes is found by multiplying the effective stress by factors βs and βn, respectively,

as defined in ACI 318 [4]. The strength of CFRP confined concrete struts was found in the

same manner as other concrete struts except that the confined concrete strength, fcc’, was used

in place of the unconfined strength, fc’. The effective stress capacity of steel ties is the yield

strength of the reinforcement. Material properties for concrete, mild steel bars, headed steel

bars, and CFRP composite from the day of testing were used for all models, and are shown in

Table 2.

3.2.2 Member area

The area of a tension tie is the area of the reinforcement the tie represents. The ties

used in the model represent longitudinal steel, transverse steel and transverse layers of the

CFRP composite shell. The CFRP tension tie representing the contribution of the CFRP shell

is shown in Figs. 8(c), 8(d) and 8(e), and was described earlier. The areas of tension ties

representing the longitudinal steel reinforcement are found from sectional analysis performed

to establish tie location. The tie area is equal to the area of the longitudinal reinforcement

located in the tension zone at first yield, as shown in Fig. 9. The tension tie area representing

transverse steel reinforcement was found by multiplying the ratio of the tension tie spacing

and the transverse reinforcement pitch by the cross sectional area of the transverse

reinforcement.

The controlling strut width, used to calculate the controlling strut area, is defined as

the maximum strut width attained without overlapping struts, nodes or the physical

boundaries of the member, while maintaining a finite strut length between nodes. The

maximum strut widths, in a 2D plane parallel to the lateral load application, can be

determined graphically. Once the maximum 2D strut widths are found, the cross-sectional
11
area of the concrete struts can be determined from geometry. For non-prismatic struts, such as

struts found within the repaired region of specimens NM-R and LE-R and the diagonal struts

in the column, the minimum cross-sectional area along the length of the strut is the

controlling strut area.

Determining the strut areas in the repaired cross section is not trivial, due to the

varying angles of the struts and circular cross section of the repair, making the area of each

strut an elliptical slice. The controlling area for one strut located in the repair is shown in

Fig. 10. The ellipse in which the controlling area is bound is a cylindrical cross section. This

cylindrical cross section is formed by extending the controlling strut width to meet the

cylindrical projection of the repair cross section, shown in Section A-A of Fig. 10. The

controlling area of a strut in the repair cross section, As, can be found by integrating the

ellipse over the limits shown in Fig. 10. The equation of an ellipse and the length of its major

axis used to find the controlling strut area in the repair can be found as:

4 
 =  1 −
(2)


 = 2 ∗ cos α (3)

where r is the radius of the repair cross section; L is the length of the major axis of the ellipse;

and angle α is measured between the axis of the strut and horizontal, as shown in Fig. 10.

The definite integral used to find As is:

&
 '()
 4 
"# = 2 $ 1 − %

(4)
&
 '(*

where h1 and h2 are the distances from the edges of the controlling strut width to the edge of

the ellipse, as shown in Fig. 10. Length h 1 is greater than h 2 and Their difference gives the

controlling strut width.

12
4. Bilinear force-displacement response model

Bilinear force-displacement models are created to predict the response of the

specimens. Each model is created by connecting the effective stiffness of the specimen,

computed using elastic beam theory, to the ultimate bending moment predicted by the STM.

The ultimate displacement value is determined analytically and from experimental results.

4.1 Stiffness

Sectional analysis was applied to predict the effective stiffness of the bilinear force-

displacement response curves. Transverse shear deformations were ignored since the shear

span-to-depth ratio was large. Bond-slip was assumed to have a negligible effect on the

stiffness since the method is only valid for the linear portion of the response.

The effective flexural stiffness of the original column, EIcol, and the repaired region,

EIrep, were determined from moment-curvature analysis as:

./0
+,- =
1/0
(5)

where Myi and ϕyi represent the ideal yield moment and curvature for a bilinear moment

curvature approximation. In this study, the bilinear moment curvature approximation follows

provisions of bridge design codes [19, 27] where an elastic perfectly plastic response is used

to estimate the capacity of the member cross-section.

The effective stiffness of the original column was determined by standard stiffness

calculations for a cantilever column in single bending and is defined as:

3+,37
2345 =
37 8
(6)

For the repaired specimen, the effective stiffness was derived to include the effect of a non-

prismatic cross-section along a portion of the column and is defined as:

13
 1
24-9 =

8 4-9 4-9



+ ; + + 4-9 <
(7)
3+,37 +,4-9 3

4.2 Ultimate Displacement

Analytical methods to determine the ultimate displacement of original and repaired

precast bridge assemblies connected with grouted splice sleeves is a subject which requires

further research. The ultimate displacement of the original precast specimens is taken as the

value observed in the tests. The repaired specimens display characteristics which are a closer

match to what would be expected of cyclically tested RC monolithic specimens. Therefore

two ultimate displacement values are displayed on the bilinear force-displacement response

model curves of the repaired specimens: (1) ultimate displacement observed in the tests, and

(2) analytical ultimate displacement predicted using monolithic relationships.

The analytical ultimate displacement of the system is equal to the product of the

displacement ductility, µ∆, and yield displacement, ∆y. The equation used to determine µ∆ is

based on the curvature ductility of the member, which can be found from sectional analysis

used to lay out the STM and predict the stiffness of the bilinear force-displacement response.

For this study, µ∆ is obtained as [28]:

?@ 1 E9
=> = = .4 + C=B − 1D3E9 ;1 − <
?/ AB 2AB
(8)

where Mr is equal to the moment capacity ratio of the column defined as the ultimate bending

moment, Mu, divided by the yield moment, My. Eq. (8) has been used previously to predict

the ultimate displacement of CFRP confined RC columns under lateral cyclic loads with

satisfactory agreement compared to experimental results [29]. Conservatively, an elasto-

plastic assumption can be made where Mr is assumed equal to 1.0; Cϕ is the column curvature

coefficient equal to 1.0 or 0.5 for single and double curvature bending, respectively; µϕ is the
14
curvature ductility of the column, where the ultimate and yield curvatures of the column

cross-section, ϕu and ϕy, are determined from sectional analysis of the column cross-section.

The normalized plastic hinge length for a monolithic flexure-controlled column, λp, can be

determined as [30]:

9 I/- %J7
E9 = = 0.12AB + 0.014H# ; <
37 − 4-9 37 − 4-9
(9)

where Lp is the length of the plastic hinge; αs is the reinforcement bond-slip coefficient,

which is taken as zero since bond-slip in the plastic hinge region was not observed during

testing; fye and dbl are the expected yield strength of the reinforcement and diameter of the

longitudinal bars, respectively. For this study ∆ y was found using the yield displacement

relationship for single curvature bending as follows:

1/ 9
?/ =
3
(10)

4.3 Comparison of Model to Experimental Results

The bilinear force-displacement response model results are compared to the

experimental hysteretic responses for specimens NM-R, NM-O, LE-R and LE-O in Figs. 6

and 7. The STM model and experimental results are tabulated in Table 3.

The STM model reached failure due to crushing of a concrete strut in the column for

both original specimens, NM-O and LE-O, and due to yielding of the tension tie representing

the headed steel bars for both repaired specimens, NM-R and LE-R. Crushing of column

concrete in the plastic hinge region, which is represented by the longitudinal concrete strut in

the STM, was observed during testing of all four specimens. Also, from strain gauge data

recorded during testing of NM-R and LE-R it is known that the headed bars yielded. The

15
failure mechanisms predicted by the STM model show that the force transfer of the repair is

accurately predicted by the strut-and-tie method.

The ultimate load predicted by the STM models was 0.95, 0.92, 1.09 and 1.03 times

the average ultimate load, for both directions of cyclic testing, for specimens NM-R, NM-O,

LE-R and LE-O, respectively, as shown in Figs. 6 and 7. These results show very good

agreement between STM predictions and experimental results. The ultimate load predicted

by the STM was 0.94 times the ultimate load reached during the pushover of LE-R, showing

the excellent correlation of the STM method for both cyclic and monotonic results. No

strength reduction factors have been applied to the loads presented, therefore when the 0.75

strength reduction factor prescribed in ACI 318 [4] is applied, all of the results would provide

conservative estimates of the strength of the assemblies.

The stiffness values predicted for the four specimens show good correlation to the

initial stiffness recorded during testing. When compared to the initial stiffness of the

cyclically tested specimens the model predicts 1.32, 0.91 and 0.90 times the experimental

stiffness observed for specimens NM-R, NM-O and LE-O, respectively. The initial stiffness

of the cyclic specimens was found using the peak displacement and corresponding force

averaged for both directions of testing during the first displacement step. The over-prediction

of the model compared to the initial stiffness of specimen NM-R in Fig. 6(c) can be attributed

to the fact that NM-R is a repaired specimen which experienced significant cracking and

plastic deformations in the original test. The pre-damage is evident for the initial stiffness of

the specimen LE-R pushover when compared to the model stiffness. The model over-

predicts the initial stiffness of LE-R during the pushover test by 1.19 times the experimental

stiffness in Fig. 7, where the experimental stiffness of the pushover is defined at the

displacement equal to the first displacement step of the cyclic tests, which was equal to 12.7

mm.
16
 The ultimate displacement values predicted analytically for specimens NM-R and

LE-R in Fig. 6(c) and Fig. 6(d) under-predict the ultimate displacement observed in the

cyclic experiments, as shown in Table 3. The bilinear force-displacement response model

compared to the monotonic LE-R specimen test shows good agreement, as displayed in Fig.

7. Ultimate displacement relationships for repaired precast bridge assemblies require further

research, however the values predicted analytically assuming monolithic properties provide a

conservative estimate of the ultimate displacement observed during testing.

5. Conclusions

Strut-and-tie (STM) models have been developed to analyze precast reinforced

concrete (RC) bridge assemblies joined with grouted splice sleeves (GSS) both in their

original form and after a repair technique using CFRP shells has been employed. The STMs

were derived for the specimens following the recommendations of the American Concrete

Institute 318. The STM method of analysis was originally developed for monotonic static

loads and is not intended to model lateral force resisting elements subjected to cyclic loads;

however, the analytical results demonstrate that all four models predicted the unfactored

ultimate load capacity of the specimens within 9% of the ultimate cyclic load achieved during

the tests. After application of strength reduction factors, all models would provide

conservative estimates of the experimental strength of the assemblies, thus proving the design

capabilities of the STM method.

The monotonic load STM models successfully simulate the force transfer mechanism

using elements representing CFRP shells, confined non-shrink concrete, headed steel

reinforcement, mild steel and concrete. The failure mechanism of the STM models was

column concrete strut crushing for the original specimens and headed reinforcement yielding

for the repaired specimens, which match the behavior observed during the tests.
17
 Bilinear force-displacement response models were created in this paper to predict the

response of the specimens by utilizing their effective flexural stiffness, the ultimate load

predicted by the STM method, and the ultimate displacement from displacement ductility

considerations. The bilinear force-displacement response models developed extend the STM

sectional analysis procedures to provide a simplified alternative to finite-element analysis for

finding the response envelope of the original or repaired precast concrete assemblies. The

STM models have been developed using adaptable generic modeling parameters, eliminating

much designer subjectivity thus facilitating adaption for future applications of the STM

method for original and repaired precast concrete assemblies with CFRP shells.

Acknowledgments

The authors would like to thank the Utah, New York State, and Texas Departments of

Transportation and the Mountain Plains Consortium through grant MPC-405 for their

financial support. The authors would also like to thank NMB Splice Sleeve North America,

ERICO, Sika USA, Inc., and Headed Reinforcement Corp. for their donation of construction

materials. The authors also acknowledge helpful discussions with Professor L.D. Reaveley of

the University of Utah.

6. Notation

As = area of a strut in the repair

c, c’ = neutral axis location of original and repaired columns

Cϕ = column curvature coefficient

dbl = diameter of longitudinal bars

EIe, EIcol, EIrep = effective flexural stiffness, stiffness of original and repaired column

fc’ = unconfined concrete compressive strength

18
fcc’ = confined concrete compressive strength within the repaired section

ffu = ultimate tensile strength of CFRP composite

fy = yield strength of steel reinforcement

fye = expected yield strength of steel reinforcement

h = column width

h1, h2 = distance from edge of strut width to edge of ellipse

HCFRP = vertical distance from CFRP tie to GSS interface

Hcol = column height from point of lateral load application to GSS interface

Hd = column height defined in Fig. 8 for original and repaired specimens

Hhead = vertical distance from headed bar node to GSS interface

Hrep = height of repair

2345 = effective stiffness of original column

24-9 = effective stiffness of repaired column

L = length of major axis of ellipse

Lp = plastic hinge length

Mr = column bending moment capacity ratio

Mu = ultimate bending moment

My, Myi = yield and ideal yield moment

n = number of equally spaced transverse tension ties in column

Pax = vertical axial load

Pcy = lateral horizontal load

r = radius of repair cross section

Rx, Ry = external horizontal and vertical reaction

x = coordinate of ellipse

19
xc, xc’ = compression force centroid at first yield for original and repaired specimen

xt, xt’ = tension force centroid at first yield for original and repaired specimen

y = coordinate of ellipse

α = angle between axis of strut and horizontal

αs = bond-slip coefficient

βs, βn = effective stress factor for struts and nodes

∆y = yield displacement

θ = angle between diagonal strut and transverse ties in column

ҡε = CFRP strain efficiency factor

λp = normalized plastic hinge length

µ∆ = displacement ductility

µϕ = column curvature ductility

ϕy, ϕyi, ϕu = yield, ideal yield, and ultimate curvature

References

[1] Schlaich, J., Schäfer, K., and Jennewein, M. (1987). “Toward a consistent design of

structural concrete.” PCI J., 23(3), pp. 74-150.

[2] Xiao, Y., Priestley, M.J.N., and Seible, F. (1996). “Seismic assessment and retrofit of

bridge column footings.” ACI Struct. J., 93(1), 79-94.

[3] Sritharan, S., and Ingham, J. M. (2003). “Application of strut-and-tie concepts to

concrete bridge joints in seismic regions.” PCI J., 48(4), 66-90.

[4] American Concrete Institute Committee 318. (2014). “Building Code Requirements

for Structural Concrete (ACI 318-14) and Commentary (318R-14).” American

Concrete Institute, Farmington Hills, MI.

20
[5] American Association of State Highway and Transportation Officials. (2012).

“AASHTO LRFD Bridge Design Specifications.” AASHTO, Washington, D.C.

[6] International Federation for Structural Concrete, FIB. (2010). “FIB Model Code for

Concrete Structures.” Ernst and Sohn, Berlin.

[7] Canadian Standards Association. (2010). “Design of Concrete Structures: Structures

(Design)—A National Standard of Canada.” CAN/CSA Committee A23.3-04.

Toronto, ON.

[8] Saenz N., Pantelides, C.P., and Reaveley, L.D. (2004). “Strut-and-tie model for shear

friction of concrete with fiber-reinforced polymer composites.” ACI Struct. J., 101(6),

863–871.

[9] Park S., and Aboutaha R.S. (2009). “Strut-and-tie method for CFRP strengthened

deep beams.” J. Struct. Eng., 135(6), 632–643.

[10] Perera, R., Vique, J., Arteaga, A., and Diego, A.D. (2009). “Shear capacity of

reinforced concrete members strengthened in shear with FRP by using strut-and-tie

models and genetic algorithms.” Composites: Part B, 40, 714-726.

[11] Burningham, C.A., Pantelides, C.P., and Reaveley, L.D. (2015). “Repair of reinforced

concrete deep beams using post-tensioned CFRP rods.” Composite Structures, 125,

256-265.

[12] Qazi, S. Michel, L., Ferrier, E., and Limam, A. (2015). “Strut-and-tie model for a

reinforced concrete wall strengthened with carbon fibre-reinforced polymers.”

Composite Structures, 127, 87-99.

[13] Hong, S.G., Chun, S.C., Lee, S.H., and Oh, B. (2007). “Strut and tie model for

development of headed bars in exterior beam column joint.” ACI Struct. J., 104(5),

590–600.

21
[14] Ameli, M.J., Parks, J.E., Brown, D.N., and Pantelides C.P. (2014). “Grouted splice

sleeve connection alternatives for precast reinforced concrete bridge piers in

moderate-to-high seismic regions.” Proc. 10th Nat. Conf. Earthq. Eng., DOI:

10.4231/D3D50FZ10, EERI.

[15] Marsh M.L., Wernli M., Garrett B.E., Stanton J.F., Eberhard M.O., and Weinert M.D.

(2011). “Application of accelerated bridge construction connections in moderate-to-

high seismic region.” NCHRP Report 698, Washington, D.C.

[16] Belleri, A., and Riva, P. (2012). “Seismic performance and retrofit of precast concrete

grouted sleeve connections.” PCI J., 57(1), 97–109.

[17] Haber, Z.B., Saiidi, M., and Sanders, D.H. (2015). “Seismic performance of precast

columns with mechanically spliced column-footing connections.” ACI Struct. J.,

111(3), 639–650.

[18] Parks, J.E., Brown, D.N., Ameli, M.J., Pantelides, C.P., and Reaveley, L.D.

(2014). “Repair of damaged precast RC bridge columns with grouted splice sleeve

connections using CFRP shells and plastic hinge relocation.” Proc. 10th Nat. Conf.

Earthq. Eng., DOI: 10.4231/D38C9R473, EERI.

[19] American Association of State Highway and Transportation Officials. (2011).

“AASHTO Guide Specifications for LRFD Seismic Bridge Design.” AASHTO,

Washington, D.C.

[20] Ameli, M. J., Parks, J. E., Brown D. N., and Pantelides, C. P. (2015) “Seismic

Evaluation of Grouted Splice Sleeve Connections for Reinforced Precast Concrete

Column-to-Cap beam Joints in Accelerated Bridge Construction.” PCI J., 60 (2), 80-

103.

22
[21] Brown, D.N. (2014). “Post-earthquake repair of precast concrete column-to-footing

plastic hinges.” MS Thesis, Univ. of Utah, Salt Lake City, UT.

[22] Barbachyn, S.M., Kurama, Y.C., and Novak, L.C. (2012). “Analytical evaluation of

diagonally reinforced concrete coupling beams under lateral loads.” ACI Struct. J.,

109(4), 497–507.

[23] Moran, D.A, and Pantelides, C.P. (2012). “Elliptical and circular FRP-confined

concrete sections: A Mohr-Coulomb analytical model.” Int. J. Solids and Struct.,

49(6), 881–898.

[24] Seible, F., Priestley, M.J.N., and Hegemier, G. (1997). “Seismic retrofit of RC

columns with continuous carbon fiber jackets.” J. Compos. Constr., 1(2), 52–62.

[25] Lam, L., and Teng, J.G. (2003). “Design–oriented stress-strain model for FRP-

confined concrete.” Constr. Build. Mater., 17(6), 471-489.

[26] American Concrete Institute Committee 440. (2008). “ACI 440.2R-08 Guide for the

Design and Construction of Externally Bonded FRP Systems for Strengthening

Concrete Structures.” Farmington Hills, MI.

[27] California Department of Transportation. (2010). “Seismic Design Criteria.” Division

of Engineering Services, Sacramento, CA.

[28] Priestley, M.J.N., and Park, R. (1987). “Strength and ductility of concrete bridge

column under seismic loading.” ACI Struct. J., 84(1), 61–76.

[29] Pantelides, C.P., and Moran, D.A. (2013). “Design of FRP jackets for plastic hinge

confinement of RC columns.” J. Compos. Constr., 17(4), 433-442.

[30] Panagiotakos, T.B., and Fardis, M.N. (2001). “Deformations of Reinforced Concrete

Members at Yielding and Ultimate.” ACI Struct. J., 98(2), 135-148.

23
Figure 1 – Grouted Splice Sleeve connectors: (a) NM connector, both dowels grouted; (b)

LE connector, threaded factory dowel and grouted field dowel.

1
Figure 2 – Specimen reinforcement details: (a) Plan view, NM-O footing; (b) Elevation view

NM-O; (c) Elevation view, LE-O; (d) Plan view, LE-O pier cap.

1
Figure 3 – Repair details: (a) CFRP shell; (b) headed steel bars; (c) cross-section.

1
 (a) (b)

Figure 4 - Test setup: (a) lateral and axial load application; (b) lateral drift ratio protocol.
Figure 5- Specimens displaced to the east at their maximum displacement: (a)NM-O,

(b) NM-R.

1
Figure 6 – Bilinear force-displacement response model results compared to experimental

hysteretic response curves: (a) NM-O; (b) LE-O; (c) NM-R; (d) LE-R.

1
Figure 7 – Bilinear force-displacement response model compared to LE-R pushover

response.

1
Figure 8– STM model layout with standardized modeling parameters: (a) NM-O; (b) LE-O;

(c) NM-R; (d) LE-R; (e) repair detail.

1
Figure 9 – First yield stress state: (a) original column cross-section; (b) repaired cross-

section.

1
Figure 10 –Repair strut inside CFRP shell.

1
 Table 1 – Generic modeling parameters

Modeling Values
Method
Parameter NM-R NM-O LE-R LE-O
Hcol (mm) Geometry 2438 2438 2438 2438
θ (deg.) Eq. (1) 50.7 52.8 50.8 52.5
n (integer) Eq. (1) 5 6 5 6
Hrep (mm) Geometry 457 - 457 -
Hd (mm) Geometry 1981 2515 1981 2515
HCFRP (mm) 0.70 * Hrep 320 - 320 -
Development length of
Hhead (mm) 381 - 381 -
epoxied headed bars
h (mm) Geometry 533 533 533 533
Sectional analysis of
xc (mm) 63 68 68 68
column
Sectional analysis of
xt (mm) 149 147 145 145
column
xc' (mm) Sectional analysis of repair 74 - 79 -
xt' (mm) Sectional analysis of repair 115 - 115 -

24
 Table 2 – Model material properties

Concrete Steel CFRP

fc ' fcc' fy
ffu
(MPa) (MPa) (MPa)
(MPa)
Specimen Precast Repair ϕ25 ϕ13 Headed
NM-R 44.3 66.9 469 434 427 696
NM-O 37.6 - 469 434 - -
LE-R 41.4 64.1 469 434 427 696
LE-O 41.4 - 469 434 - -

25
 Table 3 – Bilinear force-displacement response model and experimental results
Ultimate drift
Ultimate load Yield drift ratio Displacement Initial stiffness
ratio
(kN) (%) ductility (kN/mm)
Specimen (%)

STM/ Mod./ Mod./ Mod./ Mod./

STM Exp. Mod. Exp. Mod. Exp. Mod. Exp. Mod. Exp.
Exp. Exp. Exp. Exp. Exp.

NM-R 189 200 0.95 0.78 1.16 0.67 5.52 6.96 0.79 7.1 6.0 1.19 9.99 7.56 1.32
NM-O 150 162 0.92 0.94 1.10 0.86 - 6.69 - - 6.1 - 6.53 7.15 0.91
LE-R 197 180 1.09 0.80 - - 5.70 7.20 0.79 7.1 - - 10.03 - -
LE-R PO 197 209 0.94 0.80 1.04 0.77 5.70 6.88 0.83 7.1 6.6 1.08 10.03 8.42 1.19
LE-O 163 158 1.03 1.02 1.12 0.91 - 6.50 - - 5.8 - 6.54 7.30 0.90

26