You are on page 1of 12

Three-Dimensional Strut-and-Tie Analysis

for Footing Rehabilitation


Chun S. Cai, P.E., M.ASCE1

Abstract: A 3D indeterminate strut-and-tie model is developed for the analysis of a footing system to explain the cause of cracking and
to develop a rehabilitation solution. The performance predicted with the model, simulating the sequence of construction, is well correlated
Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

with the field observations. The prediction thus explains the cause of cracking and concludes that phased construction is its main source.
From numerical prediction, posttensioning forces are recommended to close the existing cracks and to prevent new ones. In developing
an approach to rehabilitating a cracked structure, the stiffness of the analytical model needs to be properly selected. For this reason, five
different levels of stiffness are used to cover the lower and upper bounds for both cracked and uncracked situations. A 3D finite-element
solid modeling is also conducted.
DOI: 10.1061/共ASCE兲1084-0680共2002兲7:1共14兲
CE Database keywords: Footings; Rehabilitation; Three-dimensional models; Struts; Ties; Cracking.

Introduction explain the cause of cracking and to design a rehabilitation pro-


cedure. A typical combined footing system, pier 15, was used in
The A1A-Lytle Avenue bridge 共#790152兲, located in Volusia this study 共Fig. 2兲. A beam theory is no longer valid for this
County, Florida, was built in 1990 to replace an old existing bas- footer-pile cap system, whose force flows are much more compli-
cule bridge. This high-level bridge has an overall length of 745.31 cated than a beam theory would predict. Therefore, this particular
m consisting of 23 spans varying from 27.1 to 39.62 m in span footer-pile cap system is a very good candidate for a strut-and-tie
length. Piers 2 through 9 and 18 through 23 are founded on indi- analysis. A three-dimensional 共3D兲 indeterminate strut-and-tie
vidual footings, whereas piers 10 through 17 rest atop combined analysis was conducted to serve the objective.
footings. The sketch of the superstructure is shown in Fig. 1. For a structure where beam theory does not apply, the strut-and-
Fig. 2 shows a typical combined footing system consisting of tie method is a very good design tool that has gained popularity
piles, a pile cap, a column footer, three columns, two column for the ultimate strength design of ‘‘disturbed’’ zones, such as
struts, and a pier cap. The bridge was constructed in two phases to deep beams, beam ends, anchorage zones, and brackets
meet the requirements of traffic control during the construction 共AASHTO 1998兲. However, the applications of the strut-and-tie
process. Phase 1 of the bridge was constructed on the south side model in the rehabilitation of existing structures are limited
of the existing bascule bridge, and phase 2 was built after the 共Kesner and Poston 2000兲, let alone a 3D indeterminate strut-and-
existing bridge was demolished. tie model, which is the focus of the present study.
The latest bridge inspection reported severe cracking on the top
and side faces of the column footers for piers 10 through 17.
These severe cracks occur mainly on the portion constructed in 3D Strut-and-Tie Model
phase 1; only minor cracking was reported on the portion con-
structed in phase 2. Fig. 3 shows a typical cracking pattern for The successful application of a strut-and-tie model depends on a
pier 15. Engineering surveys did not find a significant differential reliable visualization of the paths of force flows. In a typical
settlement between the portions constructed in phases 1 and 2. strut-and-tie analysis, the force distribution is visualized as com-
The possibility of uneven settlement that would cause the crack- pressive and tensile force flows that are modeled as compressive
ing was thus excluded. The phased construction that may cause struts and tensile ties, respectively. As shown in Fig. 2共b兲, the
uneven force distribution in the footing was suspected as the pos- footer and pile cap form an inverted T-shape section. To consider
sible source of the cracking. the effect of this geometry, a 3D strut-and-tie model, shown in
The objective of this study was to develop an analytical model to Fig. 4, was developed to include both the column footer and the
pile cap in the model. The force flows need to be visualized in
1
Asst. Prof., Dept. of Civil and Environmental Engineering, Louisiana both the transverse and longitudinal directions of the footing. The
State Univ., 3502 CEBA, Baton Rouge, LA 70803; formerly Sr. Engineer, reasoning of the development of this model is given below.
Structures Research Center, Florida Dept. of Transportation. The end view, Fig. 4共c兲, shows the force flows along the trans-
Note. Discussion open until July 1, 2002. Separate discussions must verse direction of the footing. All the loads eventually need to go
be submitted for individual papers. To extend the closing date by one
to the piles. However, since the pile cap is wider than the footer,
month, a written request must be filed with the ASCE Managing Editor.
The manuscript for this paper was submitted for review and possible the column force cannot go directly to the top of the exterior
publication on February 20, 2001; approved on February 20, 2001. This piles, as represented by nodes 24 and 60. Instead, a portion of the
paper is part of the Practice Periodical of Structural Design and Con- column force goes vertically to the top of the pile cap represented
struction, Vol. 7, No. 1, February 1, 2002. ©ASCE, ISSN 1084-0680/ by nodes 9 and 12 and is then distributed to the top of both the
2002/1-14 –25/$8.00⫹$.50 per page. exterior and interior piles represented by nodes 24, 36, 48, and 60.

14 / PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002

Pract. Period. Struct. Des. Constr. 2002.7:14-25.


Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Sketch of superstructure

The remaining force in the column goes directly to the interior known to develop a rehabilitation method. Since the structure is
piles represented by nodes 36 and 48. indeterminate, the dimensions of each member affect the internal
The elevation view, Fig. 4共a兲, shows the force flows in the longi- force distribution. The stiffness of the struts and ties is loading/
tudinal direction of the footing. It can be seen that the column cracking dependent. In a typical design process that considers
forces are distributed among a few piles, either directly to the top only the strength limit, the sizes of struts and ties are first as-
of the piles or to the top of the footer first and then to the top of sumed and then optimized. Several iterations are usually con-
the piles. ducted, and the final design provides detailings to match the as-
The isometric view is given in Fig. 4共b兲. For clarity, only a partial sumptions. For this existing structure, whose structural
model is shown. The possible tension members are defined as performance from service to strength limits needs to be investi-
tension ties 共dashed lines兲, and compression members are defined gated, the stiffness 共or size兲 of each strut and tie member varies,
as compression struts 共solid lines兲. These assumptions were veri-
depending on the loading level, i.e., the service or strength limit
fied by analysis. In other words, if the analysis predicted a com-
states. By assuming no cracking for the piles, the actual sizes of
pressive force in a tension tie, then the tie would be changed into
piles, 0.61⫻0.61 m (24⫻24 in.), were used to model the pile
a compression strut and vice versa.
In the present analytical model, all 48 piles that are 24.4 m 共80 ft兲 elements. However, the sizes of struts and ties were varied to
in length are assumed pinned at the bottom tips. In addition, two consider the effect of cracking.
piles are restrained at the top in the X and Y directions for nu- Three models were used to simulate the uncracked situation
merical stability. The piles are modeled with beam elements. 共strictly speaking, cracking is required to form a strut-and-tie
The existing force distribution in the footing system needs to be model兲. The objective of simulating an uncracked structure was to
predict the distribution of internal forces before cracking, which
may help explain the cracking patterns and the cause of cracking.
In the model designated as F12, all ties and struts were assumed
as 0.3⫻0.3 m (12⫻12 in.). In the model designated as F120, the
struts and ties were assumed to be 100 times as stiff as model F12
to consider the large stiffness of concrete footer and pile cap
before cracking. These two models, F12 and F120, should cover
the lower and upper bounds of stiffness before cracking. Given
the pile size of 0.61⫻0.61 m (24⫻24 in.), and the footer of 2.44
m 共96 in.兲 in width, the strut members were assumed as 0.61
⫻0.61 m (24⫻24 in), and the ties as one-half the footer width,
i.e., 1.22⫻1.22 m (48⫻48 in.) in the model Tie 48. The stiffness
of this model lies between the F12 and F120 models.
Two more models were used to simulate the cracked section and
to predict the existing internal force distribution of the footing
system. In model Tie3.3, the transformed area of each tie was
assumed as 84⫻84 mm (3.3⫻3.3 in.) that corresponds to the ac-
tual area of reinforcement 共3.11 in2兲 of the footer. In another
model, Tie01, which simulates the situation after the reinforce-
ment is broken, the sectional area of each tie was assumed as
2.5⫻2.5 mm (0.1⫻0.1 in.), a small nonzero number to prevent
numerical instability. In all analyses, the material properties of all
Fig. 2. Sketch of substructure 共pier 5兲

PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002 / 15

Pract. Period. Struct. Des. Constr. 2002.7:14-25.


Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Cracks 关共a兲 on top of; 共b兲 on side兴 of typical column footer 共pier 5兲

members were specified as concrete with f c⬘ ⫽28 MPa. As stated analysis of phase 1 loads, only the right portion of Fig. 2 that was
earlier, transformed areas were used for those ties that simulate constructed in phase 1 was modeled. For the analysis of phase 2
the reinforcement. loads, the whole structure was modeled as a unit, i.e., there is no
separation between the phase 1 and phase 2 structures. The loads
from phase 2 construction were applied on the phase 2 structure,
Application of Load
i.e., the left part in Fig. 2.
To simulate the two-phase construction, the application of loads Since loads can only be applied to the nodes in a strut-and-tie
to the model followed the sequence of construction. For the model, appropriate assumptions should be made to distribute the

16 / PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002

Pract. Period. Struct. Des. Constr. 2002.7:14-25.


Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. 3D strut-and-tie model with node numbering 共partial for clarity兲

Fig. 5. Loading definition and application

PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002 / 17

Pract. Period. Struct. Des. Constr. 2002.7:14-25.


Table 1. Force Calculation and Definition
Load per
Total loads Nodes carrying node
Group Components 关kN 共kips兲兴 loads 关kN 共kips兲兴 Load number
Phase 1 self-weight Pile Cap 3,959 共890兲 28 nodes at the top 141 共32兲 P1
of piles
Footer 3,051 共686兲 4 nodes 共8, 9, 11 762 共172兲 P2
and 12兲
Column and pier 2,397 共539兲 4 nodes 共2, 3, 5, 600 共135兲 P3
cap and 6兲
Superstructure 7,517 共1690兲 4 nodes 共2, 3, 5, 1,879 共422兲 P4
and 6兲
Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

Phase 2 self-weight Pile Cap 3,082 共693兲 24 piles 128 共29兲 P6


Footer 2,500 共562兲 4 nodes 共7, 8, 10, 625 共140兲 P7
and 11兲
Column 636 共143兲 2 nodes 共1 and 4兲 318 共71兲 P8
Pier Cap 818 共184兲 4 nodes 共1, 2, 4, 204 共46兲 P9
and 5兲
Superstructure 7,517 共1,690兲 4 nodes 共1, 2, 4, 1,879 共422兲 P 10
and 5兲

Live load Phase 1 1,748 共393兲 4 nodes 共2, 3, 5, 437 共98兲 P5


and 6兲
Phase 2 1,748 共393兲 4 nodes 共1, 2, 4, 437 共98兲 P 11
and 5兲

Posttensioning Top of footer 2,224 共500兲 2 nodes each end 1,112 共250兲 PT t1
force for 共1, 4兲 and 共3, 6兲
service Bottom of footer 4,448 共1,000兲 2 nodes each end 2,224 共500兲 PT b1
共7, 10兲 and 共9, 12兲

Posttensioning Top of footer 2,224 共500兲 2 nodes each end 1,112 共250兲 PT t2
force for 共1, 4兲 and 共3, 6兲
ultimate
Bottom of footer 6,672 共1,500兲 2 nodes each end 3,336 共750兲 PT b2
共7, 10兲 and 共9, 12兲

Table 2. Load Case Definitions


Structures
Load cases Loads applied Source of loads modeled
SubDL1 P 1⫹ P 2⫹ P 3 Phase 1 substructure dead load Phase 1
SupDL1 P4 Phase 1 superstructure dead load Phase 1
SupLL1 P5 Phase 1 superstructure live load Phase 1
DL1 SubDL1⫹SupDL1 Total dead load effects after phase 1 construction Phase 1
DLL1 DL1⫹SupLL1 Service (live⫹dead) load effects after phase 1 construction Phase 1
UDLL1 1.3•DL1⫹2.17•SupLL1 Ultimate load effects after phase 1 construction Phase 1
SubDL2 P 6⫹ P 7⫹ P 8⫹ P 9 Phase 2 substructure dead load Entire structure
SupDL2 P 10 Phase 2 superstructure dead load Entire structure
SupLL2 P 5 ⫹ P 11 live load of the whole structure Entire structure
DL2 DL1⫹SubDL2⫹SupDL2 Total dead load effects after phase 2 construction Phase 1 and entire structure
DLL2 DL2⫹SupLL2 Service (live⫹dead) load effects after phase 2 construction Phase 1 and entire structure
UDLL2 1.3•DL2⫹2.17•SupLL2 Ultimate load effects after phase 2 construction Phase 1 and entire structure
DL ( P 1 to P 4 )⫹( P 6 to P 10) All dead load Entire structure
PT – DLL2 PT t1 ⫹ PT b1 ⫹DLL2 Posttensioning force and DLL2 Phase 1 and entire structure
PT – UDLL2 PT t2 ⫹ PT b2 ⫹UDLL2 Posttensioning force and UDLL2 Phase 1 and entire structure

18 / PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002

Pract. Period. Struct. Des. Constr. 2002.7:14-25.


Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Member definitions 共partial for clarity兲

loads to the nodes. As will be seen below, the self-weight of the Possibility of Cracking
substructure is about 47% of the total service load. The method of
applying the self-weight of the substructure in the strut-and-tie According to the construction drawings, the reinforcement pro-
model significantly influences the member forces. In this study, vided at the top of the footer is 2,006 mm2 共3.11 in.2兲, which
the self-weights of the pile cap, footer, and column were applied results in a capacity of 830 kN 共186.6 kips兲. The predicted total
to the nodes at the top of piles, the top of the pile cap, and the top tension force at the top of the footer constructed at phase 1 is
of the column footer, respectively, following the construction pro- calculated as twice the force of member 2 共along the footing tie兲
cess. The load calculations are summarized in Table 1, and load plus the along-footing component of member 11 共diagonal tie兲.
applications are shown in Fig. 5. For example, for the lower-bound force model F12 under service
After phase 1 construction, the structure experienced the self- load case DLL2 共column 7, Table 3兲, the total tension force at the
weight as well as live load ( P 5 ) shown in Fig. 5共a兲. After phase 2 top of footer is
construction, in addition to the self-weight of phase 1 that had 2x 共 367.2⫹365.5x cos共 7.13兲兲 ⫽1,459.8 kN
already been locked into the phase 1 structure, the entire structure
experienced the self-weight from phase 2 construction as well as As discussed earlier, models F12, F120, and Tie48 were intended
the live load 共P 11 and P 5 兲 shown in Fig. 5共b兲. The recommended to predict the precracking stresses. All these models predicted
posttensioning forces are also listed in Table 1 and shown in Fig. larger forces under the service load 共DLL2兲 at the top of the
5共b兲. These posttensioning forces were applied to the top and footer than the provided capacity. A similar calculation for the
bottom of the column footer to close the cracks and prevent new footer constructed at phase 2 gives a total tensile force of 903 kN
ones. Further discussion of these posttensioning forces will be that is also larger than the capacity of 830 kN. This observation
conducted later. indicates an underdesign of reinforcement and explains the trans-
It is noted that the live loads were assumed to be evenly dis- verse crackling at the top of the footer.
tributed to the column nodes in the calculations shown in Table 1. Model Tie3.3, intended to model the cracked section, predicted
Therefore, the middle column nodes 2 and 5 take the same live tension forces for members 174 and 175 under both dead load and
load as nodes 3 and 6 in phase 1 analysis, but take twice ( P 11 service load 共columns 6 and 7兲. This indicates that cracking on the
⫹ P 5 ) the live load of side column nodes 1, 3, 4, and 6 in the side faces of the footer is also possible, following the cracking of
phase 2 analysis. This is not completely accurate since the live the top of the footer. Predicted tension forces in members 6, 7,
load may shift more closely to one side. However, since the live and 11 also explain the along-footer cracking at the top of the
load plus impact 3,496 kN 共786 kips兲 is a small part of the total footer.
service load 34,975 共7,863 kips兲, the above approximation is ac-
ceptable for the substructure analysis in this study. Phased Construction
Two observations were made from Table 3. First, for models F12,
Analysis Results F120, and Tie48, the tie forces of members 2, 11, and 22 共built in
phase 1兲 are significantly larger than those of their counterparts,
The load cases examined in this study are summarized in Table 2. members 1, 8, and 19 共built in phase 2兲 under dead load DL2,
For convenience of discussion, the numbering definitions of service load DLL2, and ultimate load UDLL2. This explains why
members are shown in Fig. 6. Again, only a partial model is the cracking of the footer built in phase 1 is much more severe
shown for clarity. The predicted forces of selected members are than that of phase 2.
summarized in Table 3. As discussed earlier, the true stiffness of Second, for typical phase 1 tie members 2, 11, and 22, the dead-
each member is not known for a section with cracks. Therefore, load case DL 共column 9, which assumes a nonphased construc-
the member forces of all five models discussed earlier were ex- tion兲 predicts less tension forces than does the corresponding load
amined under all the load cases to predict the lower and upper case DL2 共column 6, which considers phased construction兲. This
bounds of forces. A few observations were made as follows. indicates that phased construction is more critical for cracking

PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002 / 19

Pract. Period. Struct. Des. Constr. 2002.7:14-25.


Table 3. Member Forces 共kN兲 for Different Models and Load Cases
Load Cases
Member number 共1兲 Model DL1 DLL1 UDLL1 DL2 DLL2 UDLL2 DL PT – DLL2 PT – UDLL2
1 F12 0.00 0.00 0.00 193.18 228.60 328.00 241.85 ⫺320.72 ⫺274.36
1 F120 0.00 0.00 0.00 210.36 242.23 342.63 247.27 ⫺422.76 ⫺389.40
1 Tie48 0.00 0.00 0.00 199.88 233.53 332.87 240.38 ⫺362.14 ⫺292.62
1 Tie3.3 0.00 0.00 0.00 37.74 43.34 61.22 40.36 ⫺56.70 ⫺55.54
1 Tie01 0.00 0.00 0.00 0.04 0.05 0.07 0.04 ⫺0.06 ⫺0.06

2 F12 288.89 330.82 466.55 331.70 367.24 508.33 242.69 ⫺183.24 ⫺95.72
2 F120 324.24 368.99 518.61 378.39 410.05 560.62 245.82 ⫺254.81 ⫺171.21
2 Tie48 313.75 358.38 504.73 356.28 389.74 535.78 239.03 ⫺205.36 ⫺88.86
⫺14.43 ⫺0.55
Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

2 Tie3.3 78.60 89.56 125.96 80.58 86.23 117.03 40.73


2 Tie01 0.09 0.11 0.15 0.09 0.10 0.14 0.04 ⫺0.01 ⫺0.01

8 F12 0.00 0.00 0.00 190.20 225.16 323.12 237.89 ⫺315.51 ⫺269.81
8 F120 0.00 0.00 0.00 206.79 238.04 336.64 241.59 ⫺415.35 ⫺382.31
8 Tie48 0.00 0.00 0.00 196.54 229.44 326.89 234.06 ⫺355.27 ⫺286.57
8 Tie3.3 0.00 0.00 0.00 37.39 43.01 60.79 39.85 ⫺55.50 ⫺54.05
8 Tie01 0.00 0.00 0.00 0.04 0.05 0.06 0.03 ⫺0.04 ⫺0.04

11 F12 286.09 327.67 462.15 329.45 365.50 506.50 246.03 ⫺186.31 ⫺102.59
11 F120 319.27 363.43 510.88 373.99 405.92 555.48 246.38 ⫺247.88 ⫺164.12
11 Tie48 309.73 353.88 498.45 352.52 386.55 532.12 242.23 ⫺201.59 ⫺86.47
11 Tie3.3 78.49 89.49 125.91 80.49 86.29 117.21 41.17 ⫺14.41 ⫺0.54
11 Tie01 0.10 0.11 0.16 0.10 0.11 0.15 0.05 ⫺0.01 0.01

19 F12 0.00 0.00 0.00 103.90 125.95 182.91 164.18 ⫺442.50 ⫺642.81
19 F120 0.00 0.00 0.00 263.42 311.56 446.91 348.23 ⫺391.73 ⫺575.06
19 Tie48 0.00 0.00 0.00 198.10 239.00 346.28 302.06 ⫺660.79 ⫺988.87
19 Tie3.3 0.00 0.00 0.00 13.09 15.51 22.26 17.72 ⫺29.31 ⫺36.54
19 Tie01 0.00 0.00 0.00 0.01 0.01 0.02 0.01 ⫺0.02 ⫺0.02

22 F12 123.40 137.72 191.49 165.59 186.90 261.52 158.70 ⫺374.04 ⫺553.31
22 F120 423.97 477.90 668.19 473.05 520.60 718.14 344.02 ⫺182.33 ⫺303.26
22 Tie48 266.39 298.20 415.33 337.35 377.33 525.31 295.47 ⫺519.69 ⫺805.70
22 Tie3.3 21.79 24.65 34.54 25.26 27.70 38.13 17.92 ⫺17.46 ⫺21.11
22 Tie01 0.03 0.03 0.05 0.03 0.04 0.05 0.03 ⫺0.02 ⫺0.02

174 F12 ⫺148.82 ⫺170.32 ⫺240.13 ⫺217.63 ⫺270.35 ⫺397.32 ⫺319.73 ⫺392.67 ⫺580.35
174 F120 ⫺112.66 ⫺128.36 ⫺180.53 ⫺143.08 ⫺190.98 ⫺289.94 ⫺271.38 ⫺243.82 ⫺347.81
174 Tie48 ⫺161.11 ⫺183.88 ⫺258.86 ⫺229.67 ⫺285.47 ⫺419.66 ⫺343.08 ⫺312.99 ⫺462.18
174 Tie3.3 363.08 413.35 581.10 205.56 158.47 165.06 ⫺275.80 ⫺35.47 ⫺107.62
174 Tie01 514.40 586.25 824.64 330.09 283.45 327.90 ⫺268.16 75.21 39.89

175 F12 ⫺429.62 ⫺491.39 ⫺692.55 ⫺513.00 ⫺565.74 ⫺781.35 ⫺365.30 ⫺693.52 ⫺871.48
175 F120 ⫺417.23 ⫺474.00 ⫺665.59 ⫺501.43 ⫺535.22 ⫺725.19 ⫺284.63 ⫺573.51 ⫺753.23
175 Tie48 ⫺491.07 ⫺559.66 ⫺787.22 ⫺578.82 ⫺630.35 ⫺864.30 ⫺376.80 ⫺635.31 ⫺850.07
175 Tie3.3 143.05 162.83 228.88 159.58 181.62 255.28 152.20 ⫺1,134.43 ⫺1,218.95
175 Tie01 334.85 381.62 536.80 355.32 391.07 539.50 249.47 ⫺1,157.67 ⫺1,209.39

than nonphased construction and also that phased construction is force provided, designated as load case PT – DLL2 in Table 2, are
the main cause of the severe cracking of phase 1 structure. given in column 10, Table 3. In the second case, to obtain com-
pression forces under the ultimate-load case, posttensioning
forces of 2,224 kN 共500 kips兲 and 6,672 kN 共1,500 kips兲 were
Posttensioning Force applied at the top and bottom of the footer, respectively 共Table 1兲.
To close the cracking on the top of the footing, i.e., to obtain This case is designated as load case PT – UDLL2 in Table 2, and
compression force in members such as 2 and 11, two levels of the results are shown in column 11, Table 3.
posttensioning forces were examined. In the first case, to gain To avoid future cracking due to overloading and other environ-
compression under service load, 2,224 kN 共500 kips兲 and 4,448 mental effects, such as temperature change, the higher values of
kN 共1,000 kips兲 posttensioning forces were applied at the top and 2,224 kN 共500 kips兲 and 6,672 kN 共1,500 kips兲 were recom-
bottom of the footing, respectively 共Table 1兲. The total member mended. These forces correspond to two and six tendons 共with
forces resulting from the service load plus the posttensioning 7-0.6⬙ strands兲 at the top and bottom of the footer, respectively.

20 / PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002

Pract. Period. Struct. Des. Constr. 2002.7:14-25.


Table 4. Pile Reaction Forces 共kN兲 for Load Cases DL2 and DL
Load Case DL2 Load Case DL
Pile number Member number F12 F120 Tie48 Tie3.3 Tie01 F12 F120 Tie48 Tie3.3 Tie01
1 195 ⫺443.13 ⫺684.99 ⫺582.02 ⫺748.17 ⫺772.57 ⫺447.39 ⫺659.37 ⫺560.09 ⫺785.10 ⫺812.61
2 196 ⫺366.84 ⫺617.29 ⫺508.92 ⫺524.83 ⫺527.57 ⫺387.11 ⫺657.86 ⫺541.22 ⫺562.46 ⫺566.12
3 197 ⫺208.28 ⫺536.66 ⫺331.41 ⫺232.56 ⫺218.63 ⫺237.56 ⫺641.15 ⫺398.89 ⫺264.48 ⫺249.58
4 198 ⫺160.80 ⫺418.50 ⫺219.74 ⫺184.12 ⫺178.85 ⫺193.55 ⫺582.49 ⫺275.64 ⫺240.13 ⫺236.57
5 199 ⫺203.07 ⫺386.60 ⫺276.39 ⫺249.59 ⫺245.90 ⫺289.73 ⫺631.90 ⫺419.59 ⫺402.63 ⫺402.41
6 200 ⫺729.03 ⫺1,013.40 ⫺905.75 ⫺1,062.20 ⫺1,099.43 ⫺495.31 ⫺658.93 ⫺618.31 ⫺627.37 ⫺631.63
7 201 ⫺548.06 ⫺845.54 ⫺729.13 ⫺794.81 ⫺811.44 ⫺496.02 ⫺658.94 ⫺618.66 ⫺627.84 ⫺631.64
8 202 ⫺281.04 ⫺658.48 ⫺423.56 ⫺388.04 ⫺382.87 ⫺289.03 ⫺631.72 ⫺419.12 ⫺402.28 ⫺402.40
9 203 ⫺178.05 ⫺469.87 ⫺249.87 ⫺175.69 ⫺162.07 ⫺192.79 ⫺582.38 ⫺275.28 ⫺239.72 ⫺236.56
Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

10 204 ⫺258.85 ⫺785.38 ⫺448.08 ⫺280.05 ⫺250.92 ⫺238.33 ⫺641.17 ⫺399.33 ⫺264.90 ⫺249.59
11 205 ⫺388.78 ⫺653.47 ⫺535.99 ⫺560.79 ⫺568.12 ⫺387.58 ⫺657.40 ⫺541.47 ⫺562.59 ⫺566.12
12 206 ⫺410.42 ⫺502.43 ⫺479.29 ⫺757.28 ⫺811.57 ⫺446.77 ⫺658.26 ⫺559.32 ⫺784.49 ⫺812.60
13 207 ⫺1,089.60 ⫺702.08 ⫺862.45 ⫺1,111.22 ⫺1,160.36 ⫺1,041.50 ⫺674.01 ⫺814.64 ⫺1,132.70 ⫺1,185.50
14 208 ⫺948.59 ⫺629.49 ⫺778.49 ⫺823.12 ⫺832.06 ⫺1,004.39 ⫺672.33 ⫺829.38 ⫺902.30 ⫺914.86
15 209 ⫺689.70 ⫺557.16 ⫺669.25 ⫺506.80 ⫺474.26 ⫺848.04 ⫺671.36 ⫺819.03 ⫺658.24 ⫺631.58
16 210 ⫺556.67 ⫺486.18 ⫺577.94 ⫺344.09 ⫺294.32 ⫺778.81 ⫺671.15 ⫺806.02 ⫺566.42 ⫺526.63
17 211 ⫺514.53 ⫺416.38 ⫺506.12 ⫺423.22 ⫺403.66 ⫺863.86 ⫺673.55 ⫺837.70 ⫺778.28 ⫺766.03
18 212 ⫺1,722.84 ⫺989.92 ⫺1,300.24 ⫺1,551.43 ⫺1,623.35 ⫺1,284.27 ⫺677.20 ⫺953.19 ⫺947.55 ⫺946.08
19 213 ⫺1,485.43 ⫺914.17 ⫺1,167.08 ⫺1,248.75 ⫺1,269.90 ⫺1,287.23 ⫺677.31 ⫺954.38 ⫺949.50 ⫺946.12
20 214 ⫺917.27 ⫺836.47 ⫺948.24 ⫺837.39 ⫺804.62 ⫺869.63 ⫺673.84 ⫺840.30 ⫺783.58 ⫺766.15
21 215 ⫺850.10 ⫺764.61 ⫺895.64 ⫺452.93 ⫺329.55 ⫺794.80 ⫺671.69 ⫺812.17 ⫺576.16 ⫺526.86
22 216 ⫺845.92 ⫺692.95 ⫺839.96 ⫺496.05 ⫺416.86 ⫺831.86 ⫺670.36 ⫺810.63 ⫺647.40 ⫺631.32
23 217 ⫺974.67 ⫺623.20 ⫺791.66 ⫺845.16 ⫺860.19 ⫺995.16 ⫺671.56 ⫺823.68 ⫺898.35 ⫺914.77
24 218 ⫺967.66 ⫺554.06 ⫺712.19 ⫺1,141.24 ⫺1,240.31 ⫺1,038.82 ⫺673.43 ⫺811.39 ⫺1,135.16 ⫺1,185.58
25 219 ⫺1,089.62 ⫺701.87 ⫺861.94 ⫺1,111.86 ⫺1,160.35 ⫺1,041.56 ⫺673.45 ⫺811.79 ⫺1,135.31 ⫺1,185.48
26 220 ⫺946.91 ⫺629.20 ⫺777.51 ⫺822.80 ⫺832.06 ⫺996.71 ⫺671.57 ⫺823.90 ⫺901.00 ⫺914.88
27 221 ⫺686.38 ⫺556.78 ⫺667.77 ⫺505.57 ⫺474.28 ⫺832.85 ⫺670.36 ⫺810.77 ⫺653.15 ⫺631.66
28 222 ⫺559.50 ⫺486.37 ⫺578.99 ⫺345.46 ⫺294.34 ⫺791.75 ⫺671.66 ⫺811.86 ⫺572.08 ⫺526.72
29 223 ⫺515.51 ⫺416.48 ⫺506.54 ⫺423.87 ⫺403.67 ⫺868.37 ⫺673.82 ⫺840.04 ⫺780.96 ⫺766.05
30 224 ⫺1,723.46 ⫺990.04 ⫺1,300.20 ⫺1,553.87 ⫺1,623.38 ⫺1,286.12 ⫺677.27 ⫺954.08 ⫺947.19 ⫺946.03
31 225 ⫺1,482.24 ⫺914.26 ⫺1,166.26 ⫺1,246.63 ⫺1,269.85 ⫺1,283.24 ⫺677.16 ⫺952.90 ⫺945.27 ⫺945.98
32 226 ⫺912.13 ⫺836.56 ⫺946.93 ⫺831.11 ⫺804.49 ⫺862.61 ⫺673.53 ⫺837.44 ⫺775.70 ⫺765.93
33 227 ⫺836.87 ⫺764.68 ⫺893.04 ⫺440.96 ⫺329.33 ⫺775.76 ⫺671.12 ⫺805.69 ⫺562.41 ⫺526.50
34 228 ⫺858.32 ⫺693.10 ⫺842.90 ⫺510.00 ⫺417.17 ⫺849.03 ⫺671.36 ⫺819.17 ⫺664.04 ⫺631.92
35 229 ⫺981.88 ⫺623.32 ⫺793.65 ⫺850.33 ⫺860.29 ⫺1,005.82 ⫺672.34 ⫺829.58 ⫺904.91 ⫺914.97
36 230 ⫺970.17 ⫺554.18 ⫺713.34 ⫺1,138.34 ⫺1,240.21 ⫺1,043.94 ⫺674.03 ⫺814.99 ⫺1,132.73 ⫺1,185.40
37 231 ⫺443.06 ⫺684.62 ⫺581.96 ⫺748.14 ⫺772.57 ⫺447.07 ⫺658.40 ⫺559.72 ⫺784.98 ⫺812.61
38 232 ⫺367.00 ⫺617.12 ⫺508.99 ⫺524.85 ⫺527.57 ⫺387.82 ⫺657.44 ⫺541.60 ⫺562.58 ⫺566.11
39 233 ⫺208.48 ⫺536.68 ⫺331.51 ⫺232.63 ⫺218.63 ⫺238.47 ⫺641.20 ⫺399.43 ⫺264.75 ⫺249.58
40 234 ⫺160.68 ⫺418.46 ⫺219.68 ⫺184.07 ⫺178.85 ⫺192.99 ⫺582.38 ⫺275.32 ⫺239.95 ⫺236.57
41 235 ⫺203.00 ⫺386.53 ⫺276.33 ⫺249.57 ⫺245.90 ⫺289.41 ⫺631.71 ⫺419.26 ⫺402.54 ⫺402.41
42 236 ⫺728.91 ⫺1,013.50 ⫺905.67 ⫺1,062.11 ⫺1,099.43 ⫺495.04 ⫺658.77 ⫺618.04 ⫺627.26 ⫺631.63
43 237 ⫺547.23 ⫺845.58 ⫺728.86 ⫺794.09 ⫺811.43 ⫺494.18 ⫺658.76 ⫺617.63 ⫺626.75 ⫺631.62
44 238 ⫺281.39 ⫺658.53 ⫺423.68 ⫺388.35 ⫺382.87 ⫺289.55 ⫺631.89 ⫺419.47 ⫺402.69 ⫺402.41
45 239 ⫺178.65 ⫺469.89 ⫺250.01 ⫺176.20 ⫺162.08 ⫺193.68 ⫺582.49 ⫺275.64 ⫺240.35 ⫺236.58
46 240 ⫺258.22 ⫺785.40 ⫺447.92 ⫺279.49 ⫺250.91 ⫺237.51 ⫺641.17 ⫺398.85 ⫺264.27 ⫺249.56
47 241 ⫺388.25 ⫺653.56 ⫺535.88 ⫺560.52 ⫺568.11 ⫺386.84 ⫺657.89 ⫺541.15 ⫺562.28 ⫺566.11
48 242 ⫺411.50 ⫺502.67 ⫺479.73 ⫺758.31 ⫺811.59 ⫺448.82 ⫺659.52 ⫺560.90 ⫺785.89 ⫺812.63

Max⫽ ⫺160.68 ⫺386.53 ⫺219.68 ⫺175.69 ⫺162.07 ⫺192.79 ⫺582.38 ⫺275.28 ⫺239.72 ⫺236.56
Min⫽ ⫺1,723.46 ⫺1,013.50 ⫺1,300.24 ⫺1,553.87 ⫺1,623.38 ⫺1,287.23 ⫺677.31 ⫺954.38 ⫺1,135.31 ⫺1,185.58

PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002 / 21

Pract. Period. Struct. Des. Constr. 2002.7:14-25.


Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Pile reaction distribution, model tie48, load case DL2

Fig. 8. Pile reaction distribution, model F120, load case DL

22 / PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002

Pract. Period. Struct. Des. Constr. 2002.7:14-25.


Reaction Force of Piles
The pile reaction forces for all models under dead load consider-
ing both phased 共DL2兲 and nonphased 共DL兲 constructions are
summarized in Table 4. It can be seen by comparing the minimum
and maximum reactions that the distributions of reactions are far
from uniform except for model F120 with load case DL. In other
words, a uniform distribution of pile reactions will be true only
under the conditions that the footing is very rigid and that a non-
phased construction method is used. As an example of the phased
construction under dead load DL2, Fig. 7 shows the nonuniform
distribution of pile reactions predicted with model Tie48. In com-
parison, for the nonphased construction under load case DL,
model F120 predicts an almost uniformly distributed pile reac-
Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

tion, as shown in Fig. 8. Therefore, a uniform distribution of the


reactions is not always true, depending on the stiffness of the
strut-and-tie model and the construction sequence.

Discussion of Load Placement in Strut-and-Tie


Model
Fig. 9. Strut-and-tie model with known and unknown reactions
As discussed earlier, the 3D strut-and-tie model established in the
present study is an indeterminate structure whose distribution of
member forces depends on the relative stiffness of the members.
It is noted that although the tension forces of members at the top Since the member stiffness of this particular structure is very
of the footer are larger than those at the bottom, larger postten- difficult to determine, one would wonder why not to develop a
sioning forces are needed at the bottom of the footer because the determinate strut-and-tie model.
bottom is stiffer than the top. When the bottom is posttensioned, a One choice would be the same as that in the KCA study 共1998兲,
large amount of the posttensioning force goes to the diagonal i.e., the model was fixed at the bottom of the columns, and the
struts instead of the ties. This phenomenon is called a secondary pile reactions were applied as loads from the bottom of the pile
effect. For an indeterminate structure, secondary effects can be cap. It is thus possible to reduce the redundance or formulate a
very significant. determinate model by further simplifying the footing as a 2D

Fig. 10. Stress in along-footing direction under substructure dead load 共nonphased construction兲

PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002 / 23

Pract. Period. Struct. Des. Constr. 2002.7:14-25.


Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Stress in along-footing direction under total dead load 共nonphased construction兲

Fig. 12. Stress in along-footing direction under total dead load 共phased construction兲

24 / PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002

Pract. Period. Struct. Des. Constr. 2002.7:14-25.


structure. This would be acceptable if the reaction forces of piles footing direction兲 under substructure dead load is shown in Fig.
are known and the self-weight of the substructure is small. First, 10. As expected and stated earlier, substructure dead load causes
however, the reactions are unknown and far from uniformly dis- very low tension stress at the top of footing. The superstructure
tributed among the piles due to the elastic deformation of struts dead load and live load do cause tension at the top of the footing.
and piles and to the phased construction, as discussed earlier. The S XX stress contour 共along-footing direction兲 of nonphased
Second, the self-weight of the substructure is a significant part construction under the total dead load 共DL兲 is shown in Fig. 11,
共47%兲 of the total load. Applying this significant self-weight of compared with its counterpart under phased construction 共DL2兲
the substructure from the bottom of the pile cap would cause large shown in Fig. 12. The difference in stresses between nonphased
tension forces in the ties at the top of the footer, as in the KCA and phased construction is obvious. While the nonphased load
study 共1998兲. In reality, substructure dead load causes much less case DL predicts a uniform distribution with each pile reactions of
tension force on the top of the footing, if any. For example, since 620 kN, the phased-load case DL2 predicts a nonuniform distri-
the self-weight of the pile cap goes to the piles before the column bution of pile reactions varying from 327 to 826 kN.
footer is built, it should not cause tension force near the top of the
Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 06/20/14. Copyright ASCE. For personal use only; all rights reserved.

footer. This was verified in the finite-element analysis, as dis-


cussed below. Conclusions
Furthermore, applying load from the top of piles may change the
indeterminate structure into a determinate one whose internal A 3D indeterminate strut-and-tie model was developed for the
forces do not depend on the relative member sizes. Take a two- analysis of the footing system. Five different levels of stiffness
span continuous deep beam shown in Fig. 9, for example. An were used to cover the lower and upper bounds for both cracked
elastic analysis of the continuous beam shown in Fig. 9共a兲 pre- and uncracked situations. By following the sequence of construc-
dicts the reactions as R 1 ⫽R 3 ⫽10P/32 and R 2 ⫽22P/16. By ap- tion, the model predicted the stresses of the members at different
plying these reactions to the bottom of the strut-and-tie model in stages. Finite-element solid modeling was also conducted to
Fig. 9共b兲, the model becomes a determinate structure that gives verify the strut-and-tie model. The following conclusions are
the strut force as N c ⫽R 3 / cos ␪⫽10P/(32 cos ␪). In comparison, drawn from the study:
for an indeterminate strut-and-tie model shown in Fig. 9共c兲, the 1. The tensile force on the top of the footer is larger than the
strut force depends on the member stiffness of the struts and ties. capacity of the reinforcement. This indicates an underdesign
In an extreme case where the top tie is broken, i.e., its stiffness is of reinforcement and explains the transverse crackling at the
zero, the strut force N c would be P/(2 cos ␪). The difference in top of the footer.
the strut forces between the two approaches is significant. 2. The tensile force on the top of the footer constructed in
phase 1 is significantly larger than that constructed in phase
2. This explains why the former has much more severe
Confirmation of 3D Finite-Element Analysis cracking than the latter and indicates that phased construc-
tion is the main source of the cracking.
In addition to the strut-and-tie model analysis conducted above, a 3. To close the existing cracks and prevent new ones, two and
3D finite-element solid modeling was performed with the GT- six tendons 共with 7-0.6⬙ stands兲 are recommended at the top
STRUDL program. While the finite-element method 共FEM兲 with and bottom of the footer, respectively.
solid modeling would be a very powerful method for complicated 4. In developing a retrofitting approach for cracked structures,
structure systems, it is not the most appropriate one to model this the stiffness of the model needs to be properly selected.
particular structure with cracking. To predict the existing force
distribution of a cracked structure, FEM analysis tracing the
cracking process would be materially nonlinear. This kind of References
analysis would be very time-consuming.
For the reason discussed above, only a linear finite-element solid AASHTO. 共1996兲. ‘‘Standard specification for highway bridges.’’ Wash-
modeling was performed. The purpose of this analysis was to help ington, D.C.
AASHTO. 共1998兲. ‘‘LRFD bridge design specifications.’’ Washington
understand the behavior of the structure. One needs to be very
D.C.
cautious when one correlates the results between the solid mod-
Kesner, K. E., and Poston, R. W. 共2000兲. ‘‘Evaluation and strengthening
eling and the strut-and-tie analysis since these two models may of structural concrete using strut-and-tie models.’’ Concr. Int., 22共11兲,
have different stiffness and represent two different stages of the 48 –54.
structures. The correlation should be more qualitative than quan- Kissinger, Campo, and Associates 共KCA兲. 共1998兲. ‘‘SR A1A over intrac-
titative. oastal waterway pier analysis.’’ Report prepared for Florida Depart-
For a nonphased construction, the S xx stress contours 共along- ment of Transportation, District 5, Deland, Florida.

PRACTICE PERIODICAL OF STRUCTURAL DESIGN AND CONSTRUCTION / FEBRUARY 2002 / 25

Pract. Period. Struct. Des. Constr. 2002.7:14-25.

You might also like