Vervelidis - Seismic Response of Self-centering Precast Concrete Dual-shell Steel Columns

7/25/2019 Vervelidis - Seismic Response of Self-centering Precast Concrete Dual-shell Steel Columns

1/181

ALMA MATER STUDIORUM UNIVERSITA DI BOLOGNA

FACOLTA DI INGEGNERIA

International Master Course in Civil Engineering

D.I.C.A.M.

Dipartimento di Ingegneria Civile, Ambientale e dei Materiali

Tesi di Laurea Specialistica in

Earthquake Engineering

SEISMIC RESPONSE OF SELF-CENTERING PRECAST

CONCRETE DUAL-SHELL STEEL COLUMNS

Laureando:

Athanassios Vervelidis

Relatore:

Chiar.mo Prof. Ing. Marco Savoia

Correlatori:

Chiar.mo Prof. Ing. Jos I. Restrepo

Dott. Ing. Gabriele Guerrini

Sessione III

Anno Accademico 2010/2011


2/181


3/181

1

Contents

Chapter 1 Introduction.............................................................................................. 3

Chapter 2 Test Setup ................................................................................................ 6

2.1 Construction of the test units ............................................................................. 8

2.2 Specimens components ................................................................................. 16

2.2.1 Footing ..................................................................................................... 16

2.2.2 Load stub ................................................................................................. 19

2.2.3 Column..................................................................................................... 21

2.2.4 Energy dissipating devices ....................................................................... 23

2.2.4.1 Unit 1A External energy dissipators ............................................... 23

2.2.4.2 Unit 1B Internal energy dissipators ................................................ 26

2.2.5 Mortar bed ................................................................................................ 28

2.2.5.1 Unit 1A Embeco 885 grout ............................................................. 29

2.2.5.2 Unit 1B Embeco 885 grout with polypropylene fibers .................... 29

2.2.6 Post-tensioning system ............................................................................ 302.2.6.1 Unit 1A Square rubber pads .......................................................... 31

2.2.6.2 Unit 1B Circular Fyfe discs ............................................................. 32

2.3 Material properties .......................................................................................... 34

2.3.1 Concrete .................................................................................................. 34

2.3.2 Grout ........................................................................................................ 36

2.3.2.1 Plastic mortar .................................................................................... 36

2.3.2.2 Fluid grout ......................................................................................... 37

2.3.3 Steel ......................................................................................................... 38

2.3.4 Bearing pads ............................................................................................ 40

2.4 Instrumentation ............................................................................................... 42

2.4.1 Strain gauges ........................................................................................... 42

2.4.2 Potentiometers and inclinometers ............................................................ 46

2.4.3 Load cells ................................................................................................. 49

2.4.4 Other instrumentation .............................................................................. 49

2.5 Test protocol ................................................................................................... 50


4/181

2

Chapter 3 Experimental Results............................................................................ 54

3.1 Test unit 1A ..................................................................................................... 54

3.2 Test unit 1B ..................................................................................................... 64

Chapter 4 Numerical Modeling of the Tests.......................................................... 714.1 Material models ............................................................................................... 74

4.1.1 Concrete 01 ............................................................................................. 74

4.1.2 Steel 02 .................................................................................................... 78

4.2 Model components .......................................................................................... 81

4.2.1 Column..................................................................................................... 81

4.2.2 Energy dissipating devices ....................................................................... 82

4.2.3 Mortar bed ................................................................................................ 864.2.4 Post-tensioning bars ................................................................................ 87

4.3 Analysis parameters ........................................................................................ 89

Chapter 5 Comparison between Experimental and Numerical Results.............. 91

5.1 Test unit 1A ..................................................................................................... 91

5.2 Test unit 1B ..................................................................................................... 98

Chapter 6 Conclusions ......................................................................................... 103

Appendix ACompressive Cylinder Strengths of Cementitious Materials........ 106

Appendix B Specimens Drawings....................................................................... 109

Appendix C OpenSees Script for the Numerical Model..................................... 140

References ............................................................................................................. 177

Acknowledgements............................................................................................... 179


5/181

3

Chapter

Introduction

The impact and cost of the consequences of earthquake-induced damage on

structures have raised serious questions on whether current seismic design

philosophies are satisfying the needs of modern society. Most seismic design

standards are based on a life-safety approach, where structural damage is accepted

provided that collapse is avoided (Restrepo, 2002).

Designing a structure to respond elastically under severe earthquakes has generally

been considered impractical for economical reasons. Current seismic design

philosophies for ductile reinforced concrete bridges allow them to respond beyond

the elastic limit, with the inelastic behavior localized within plastic hinge regions at the

bottom and/or top of the pier elements. The system is therefore built with regions that

will be sacrificed during moderate and strong earthquakes and may require from

minor to expensive repair work. In addition, the occurrence of severe damage orpartial collapse of a bridge system can lead to critical consequences associated to

the interruption of a fundamental road path, such as obstruction of rescue and

recovery operations and economical losses related to business interruption,

displacement of people, and goods (Palermo, et al., 2008).

This has led to focus on cost-effective seismic structural systems capable of

providing a high level of protection, low structural damage, and reduced downtime

after a design-level seismic event. An example of such technology is the combination

of unbonded post-tensioning techniques with rockingdissipating connections

(Marriot, et al., 2008). The non-linear demand is accommodated within the

connections themselves, through the opening and closing of a gap at the column-

footing and column-bent cap interfaces. No or negligible damage is induced in the

main structural elements, as opposite to what would occur in a monolithic solution

with the development of plastic hinges, and self-centering behavior is achieved. A

particularly efficient jointed ductile solution is given by the hybrid system, based on

an adequate combination of self-centering/rocking properties (provided by unbonded


6/181

4

post-tensioning and gravity forces) with energy dissipation capacity (localized within

additional dissipating devices) (Palermo, et al., 2008).

The typical hysteretic behavior of monolithic, purely rocking, and hybrid structures is

schematized in Figure 1.1. The conventional system (Fig. 1.1-a) offers large energy

dissipation, represented by fat hysteretic loops, at the expenses of structural integrity

and significant residual deformations. The purely rocking solution (Fig. 1.1-b) is

characterized by non-linear elastic behavior, due to gap-opening at the base, with

self-centering properties but no o very little dissipation. The hybrid system (Fig. 1.1-c)

provides a trade-off between these two extremes: balancing re-centering forces and

energy dissipation leads to a flag-shaped response, with small residual

displacements but relatively fat loops.

The innovative bridge column technology developed and tested at the University of

California, San Diego (UCSD) combines a precast concrete hollow-core column with

on-site post-tensioning and added energy dissipation. The column consists of two

cylindrical steel shells (dual-shell technology) that run for the full-height of the

column, with concrete sandwiched in between. The outer shell acts as permanent

formwork, and provides longitudinal and transverse reinforcement. The inner shell

also behaves as permanent formwork, and prevents concrete implosion under large

compressive strains. Constructability is enhanced by the use of a precast element of

reduced weight (hollow-core section) without a reinforcing cage (Guerrini, et al.,

2011).

Fig 1 1 Hysteretic response of various structural systems (Holden, et al., 2003)


7/181

5

Large inelastic rotations can be attained with minimal structural damage: the

column-footing joint is allowed to open in tension under severe lateral displacement

demand, and to close subsequently upon load reversal. Re-centering behavior is

ensured by post-tensioned (PT) bars, designed to respond elastically. A special

connection between column, bent cap, foundation, and PT bars allows for eventual

bar replacement, should corrosion or other damage to the bar be a concern.

Energy dissipation takes place through extensive yielding in tension and

compression of internal dowel bars or external buckling-restrained devices,

preventing the main structural members (column, footing, bent cap) from suffering

significant damage. Under a strong-intensity shake only these devices are expected

to undergo multiple cycles within the inelastic range of response, with possible need

of replacement, but the structure is expected to remain functional overall.

In order to obtain a flag-shaped hysteretic response, the self-centering forces

(gravity and post-tensioning) must be large enough to overcome the overstrength

capacity of the energy dissipators, thus forcing them in compression and closing the

gap at each cycle. Care must be paid to prevent post-tensioning losses due to

yielding of the PT bars or crushing of the mortar at the precast elements joints: these

issues are addressed with connection details and material choices.

The main objective of this thesis is the development of a reliable numerical model

for hybrid self-centering dual-shell columns, able to reproduce the behavior of the

specimen recorded during tests. It is expected that such numerical model can be

used in future for implementation in broader structural systems and for dynamic

analyses. This task was addressed with the aid of the open-source software

OpenSees, developed by the Pacific Earthquake Engineering Research (PEER)

Center at the University of California, Berkeley (UCB).

Following this brief introduction, Chapter 2 describes the experimental work done at

UCSD on two hybrid column specimens, including external (test unit 1A) and internal

(test unit 1B) energy dissipating devices. The experimental results for both test units

are presented and discussed in Chapter 3. Chapter 4 focuses on the modeling of the

two specimens; the test results were used to calibrate the numerical model. The

accuracy of the model is addressed in Chapter 5, where experimental and numerical

results are compared. Final considerations regarding the behavior of the overall

system and the modeling procedure are reported in Chapter 6, together with some

recommendations for future research.


8/181

6

Chapter 2

Test Setup

The behavior of the innovative bridge column technology, introduced in the previous

chapter, was assessed by means of laboratory testing at the Powell Structural

Engineering Laboratories of the University of California, San Diego. The experimental

program consisted of unidirectional quasi-static tests conducted on two specimens

intended to represent 1:2.4 scale recentering hybrid bridge columns. In both casesthe column diameter was 0.51 m (20 in) while the total cantilever span above the

footing, to the point of lateral load application, was 1.13 m (44.5 in). A low aspect

ratio of 2.2 was chosen to induce a large ultimate base shear for a given cross-

section, thus leading to a more critical condition for sliding at the base. Moreover, a

short element can accommodate short post-tensioning bars, which have a low elastic

axial flexibility and need to be protected against yielding.

The components of the test units are listed below:

Test Unit 1A:

Footing

Load stub

Column

External energy dissipating devices

Mortar bed

Post-tensioning system including

rubber bearings

Test Unit 1B:

Footing

Load stub

Column

Internal energy dissipating devices

Fiber-reinforced mortar bed

Post-tensioning system including

adiprene bearings

In the first section of this chapter the construction of the specimens is described,

while their components are presented in detail in the second section. The third

section reports the specimens material properties. The fourth section deals with the

instrumentation used to record the behavior of the two units. Finally the test protocol

is presented in the last section.


9/181

7

(a) (b)

Fig 2 1 Meridian sections: (a) test unit 1A and (b) unit 1B.

(a) (b)

Fig 2 2 Specimens before testing: (a) test unit 1A and (b) unit 1B.


10/181

8

2 1 Construction of the Test Units

As the first stage of specimens construction, reinforcing cages were assembled for

the footing and the load stub (Fig. 2.3). The cages were then placed within woodenformworks, together with anchorages for the post-tensioning bars, corrugated ducts

for the energy dissipators, and PVC ducts for the connection with tie-down rods and

actuator (Fig. 2.4)..

Two steel shells, having the inner and outer diameters of the hollow-core column,

were positioned using a wooden template at the base (Fig. 2.5). Six steel brackets

were welded to the outer shell, to be used for the external dissipator devices of unit

1A. Six corrugated steel ducts were welded to the inner shell at the upper end, to be

used for the internal energy dissipators of unit 1B. The shells acted as permanent

formworks for the column casting.

Concrete was poured for the footing, the load stub, and the column, thought to be

precast elements in real applications. Once the concrete had hardened, the

temporary formworks were removed from footing and load stub, and the interfaces

between the three elements were roughened (Fig. 2.6) to increase friction.

Fig 2 3

Reinforcement for the footing and the load stub.


11/181

9

(a) (b)

Fig 2 4 Formworks and reinforcement prior to concrete pouring: (a) footing and

(b) load stub. Corrugated and PVC ducts are visible.

(a) (b)

Fig 2 5

(a) Steel shells positioned with a wooden template. (b) Ducts for the

internal dissipators.


12/181

10

Fig 2 6 Column end surface roughened to increase friction.

The first assemblage step for unit 1A consisted of screwing four post-tensioning(PT) bars into their anchorages within the footing. These bars were then sleeved with

metal ducts for the entire height of the column, and grouted, to simulate protection

against corrosion (Fig. 2.7)

After having protected the ducts for the internal dissipators of test Unit 1B with rugs

and foam, a 12.7-mm (0.5-in) thick mortar bed was cast over the footing and the

column was placed on top of it (Fig. 2.8). A wooden template was used to center the

column. Before mortar hardening, the column weight was supported by six 12.7-mm(0.5-in) diameter polyurethane sticks.

Fig 2 7

Post-tensioning bars within grouted ducts for corrosion protection.


13/181

11

Fig 2 8

Realization of the mortar bed.

Fig 2 9 Grouting of external dissipator anchorages within ducts in the footing.


14/181

12

With the column positioned, six steel dowel bars providing anchorage for the

external dissipators were placed inside ducts in the footing and grouted (Fig. 2.9).

Steel brackets had previously been welded to the upper end of these bars, and were

left sticking out from the footing surface. Steel angles and C-clamps were used to

align these anchorages with the brackets on the outer shell.

Prior to positioning the load stub, two post-tensioning bars were instrumented with

straing gauges right above the grouted portion. Then a 25.4-mm (1-in) thick layer of

hydrostone was prepared on the column top, and. the load stub was seated above it

(Fig. 2.10). A temporary scaffolding was build to support the load stub during

hydrostone hardening.

Fig 2 10 Placement of the load stub on scaffolding and realization of thehydrostone joint.


15/181

13

Six external buckling-restrained energy dissipators were assembled, starting from

dog-bone milled steel bars, grouted within steel tubes to prevent buckling (Fig. 2.11).

The milled portion of the bars was greased to minimize friction with the grout, while

the taper was covered with mastic tape to reduce interaction with the grout in

compression. These devices were then welded to the steel brackets of the column

outer shell and of the footing anchorages (Fig. 2.12).

Fig 2 11 Assemblage of the external energy dissipators.

Fig 2 12 Welding of the external dissipators to thei anchorages in the footing and

on the column outer shell.


16/181

14

The last construction stages for unit 1A consisted of stressing the post-tensioning

bars and completing the instrumentation. To provide additional flexibility to the post-

tensioning system and prevent yielding of the bars, stacks of rubber pads alternated

with steel plates were placed in series with the PT bars on top of the loading stub

(Fig. 2.13-a). Rubber pads and steel plates were bonded together using an epoxy

adhesive (Fig. 2.13-b).

After testing specimen 1A, the unit was disassembled and its main components

column, footing, and load stub were used for unit 1B. In fact, only the lower portion

of the cantilever column was subject to critical stresses and strains during lateral

deflection: the remaining part of the column and the other elements were designed to

remain elastic.

(a) (b)

Fig 2 13 Stack of rubber pads and steel plates: (a) location between the load stub

and PT bearing plates, (b) epoxying operations.


17/181

15

Similar steps were followed in the assemblage of test unit 1B with few differences.

Six stainless-steel internal dowels were used as energy dissipating devices instead

of external buckling-restrained ones. Before column placement the dowels were

grouted within prearranged ducts in the footing (Fig. 2.14-a); the column was then

positioned, with the dowels matching its prearranged ducts (Fig. 2.14-b); finally these

ducts were grouted through pre-drilled holes.

The mortar at the column-footing joint was reinforced with polypropylene fibers, and

Fyfe polyurethane discs were used instead of rubber pads for the PT system, to

improve the recentering behavior at large lateral drifts.

In the following section the specimens components are presented in detail.

(a)

(b)

Fig 2 14 Internal energy dissipators: (a) dowel bars grouted within ducts in the

footing, (b) dowel bars matching column ducts during column placement.


18/181

16

2 2 Specimens components

2.2.1 Footing

High-strength normal-weight concrete was used to cast the footing. The specified

concrete compressive strength at 28 days was 62 MPa (9.0 ksi), the ones measured

at 28 days, 49 days (day of testing Unit1A) and 96 days (day of testing Unit1B) were

66 MPa (9.5 ksi), 70 MPa (10.2 ksi), and 72 MPa (10.4 ksi) respectively; each value

was obtained as the average from tests on three standard cylinders. Detailed results

from material tests are presented in Appendix A.

The footing consisted of a 1.83-m (6-ft) long, 1.22-m (4-ft) wide, 0.61-m (2-ft) high

reinforced concrete beam. Ten #6 (19-mm diameter) top and bottom U bent bars

provided longitudinal reinforcement, while ten #3 (9.5-mm diameter) 4-legs stirrups

constituted the transverse reinforcement; two #3 bars were placed along each side

face of the beam as skin reinforcement. All the reinforcing steel in the footing

consisted of Grade 60 rebars conforming to the ASTM designation A615, with a

specified yield strength of 420 MPa (60 ksi)

Two series of six 50.8-mm (2-in) diameter steel ducts for the energy dissipators,

four anchorage devices for the PT bars, and four 63.5-mm (2.5-in) diameter PVC

pipes for the tie-down rods were provided in the footing. A #3 (9.5-mm diameter)

spiral reinforcement, with a 76.2-mm (3-in) pitch and a diameter of 660 mm (26 in),

was placed around the dissipator ducts for concrete confinement. Footing details are

shown in Figures 2.15 and 2.16.

Each PT-bar anchorage was obtained from a vertical PT-bar segment, with a

diameter of 34.9 mm (1-3/8 in) and a length of 0.5 m (19.5 in). A plate and a nut were

located at the lower end, while a coupler was provided at the upper end to be

connected with the main bar (Figs. 2.17 and 2.18). In order to rely only on the

bearing effect of the bottom plate and to avoid friction along the bar segment

embedded in the concrete, this segment was wrapped with duct tape. This would

prevent post-tensioning losses due to subsequent slippage, since all movements

take place during stressing operations.

The footing was connected to the strong floor by means of four 4.45-cm (1.75-in)

diameter tie-down rods tensioned at 979 kN (220 kips) each.


19/181

17

Fig 2 15 Construction drawings of the footing.

Fig 2 16 Footing details and casting.


20/181

18

Fig 2 17

Construction drawings of PT-bars anchorage devices.

Fig 2 18 PT-bar anchorage devices in the footing.


21/181

19

2.2.2 Load stub

The load stub on the top of the column consisted of a 2.44-m (8-ft) long, 0.76-m (30-

in) wide, 0.51-m (20-in) high reinforced concrete beam, cast with the same concrete

used for the footing and having therefore the same characteristics.

Four #6 (19-mm diameter) top and bottom U bent bars provided longitudinal

reinforcement, while sixteen #3 (9.5-mm diameter) 2-legs stirrups at 150-mm (6-in)

spacing constituted the transverse reinforcement. Grade 60 rebars conforming to the

ASTM designation A615, with a specified yield strength of 420 MPa (60 ksi), were

used for the load stub.

Four horizontal 63.5-mm (2.5-in) diameter PVC pipes were provided in the load stub

to connect the actuator. Two vertical 63.5-mm (2.5-in) diameter PVC pipes were

placed close to the member ends in order to apply the vertical load by tie-down rods

and jacks. Four vertical 50.8-mm (2-in) diameter PVC ducts allowed the post-

tensioning bars to run through the load stub. Figures 2.19 and 2.20 show details of

the member.

Fig 2 19

Construction drawings of the load stub.


22/181

20

Fig 2 20 Load-stub details and casting.


23/181

21

2.2.3 Column

The column element had a length of 0.84 m (33 in) and a hollow circular cross

section, with a 0.51-m (20-in) outer diameter and a 0.36-m (14-in) inner diameter.

The column consisted of two concentric circular steel shells running for its full length,

with high-performance concrete poured in between the shells. The concrete had the

same characteristics of the one used to cast the footing.

Both shells acted as permanent formwork during concrete casting. The outer shell

substituted longitudinal and transverse reinforcement, with stresses being transferred

between steel and concrete through weld beads on the internal surface of the shell.

The inner shell prevented concrete implosion under large compressive strains. Both

shells were obtained folding and welding steel plates of Grade 50 material

conforming to the ASTM designation A572, with a specified yield strength of 345

MPa (50 ksi). The outer steel shell had a diameter of 0.51 m (20 in) and a thickness

of 6.4 mm (0.25 in), while the inner shell had a diameter of 0.36 m (14 in) and a

thickness of 3.2 mm (0.125 in).

The column was equipped with six radially distributed 12.7-mm (0.5-in) thick steel

plates, welded to the outer shell: they acted as brackets for the installation of the

external dissipators of unit 1A (see 2.2.4.1) and were designed to withstand the

ultimate strength of these devices. Six 50.8-mm (2-in) diameter, 0.46-m (18-in) long,

corrugated metal ducts were left in the concrete for the installation of the internal

dissipators of unit 1B.

All welds were realized with an E70 electrode (480-MPa ultimate strength)

conforming to AWS D1.1 Structural Welding Code specifications.

Column details are shown in Figures 2.21 and 2.22.


24/181

22

Fig 2 21

Construction drawings of the column.

Fig 2 22

Column details and casting.


25/181

23

2.2.4 Energy dissipating devices

2.2.4.1 Unit 1A External energy dissipators

Six external energy dissipating devices were incorporated in test Unit 1A, radially

distributed around the perimeter. They were obtained from steel bars with a reduced

diameter over a specific length (dog-bone milled bar), where the dissipation was

provided by hysteresis of the material. The 343-mm (13.5-in) long steel bar had a

virgin diameter of 25.4 mm (1 in), which was reduced to 14.3 mm (9/16 in) in the

milled part for a length of 165 mm (6.5 in). Hot-rolled Grade 1018 steel, conforming to

the ASTM Designation A576, with a measured yield strength of 290 MPa (42 ksi),

was used for the external dissipators. Figures 2.23 and 2.24 show details of these

devices.

In order to prevent buckling phenomena and to allow for full plasticity development

on both tensile and compressive ranges, the entire milled part was encased in a steel

pipe and grouted. BASF Embeco 885 grout with fluid consistency was used at this

scope. Furthermore, the tapered segments between the virgin and milled parts were

wrapped with mastic tape to avoid interactions between bar and grout under

compression and allow for shortening of the bar (Fig. 2.24). The entire milled part of

the bar was specifically greased to reduce friction with the grout.

Grade 60 steel rebars, with a specified yield strength of 420 MPa (60 ksi)

conforming to the ASTM Designation A706, were welded to steel connectors and

positioned inside the footing to provide anchorage for the external dissipators of unit

1A (Figs. 2.25-a and 2.26-a). Brackets were welded on the column outer shell to

connect the external dissipators (Figs. 2.25-b and 2.26-b). All brackets and

connectors were obtained from A572 Grade 50 steel plates, with a specified yield

stress of 345 MPa (50 ksi), and were proportioned to resist the ultimate force carried

by the devices.

The dissipators were welded to the connection brackets after column, footing, and

load stub had been assembled (Fig. 2.12). All welds were performed with an E70

electrode (480-MPa ultimate strength) conforming to AWS D1.1 Structural Welding

Code specifications, and were designed to withstand the ultimate strength of the

dissipators. The final assemblage is shown on Figure 2.27.


26/181

24

Fig 2 23 Construction drawings of external energy dissipators.

Fig 2 24 Details of external energy dissipators.


27/181

25

(a) (b)

Fig 2 25 Construction drawings of the steel connectors for external energy

dissipators: (a) connector for footing anchorage, (b) bracket for column outer shell

(a) (b)

Fig 2 26 Details of the steel connectors for external dissipators: (a) anchorage inthe footing, (b) bracket on the column outer shell.


28/181


29/181

27

Fig 2 28 Construction drawing of internal energy dissipators.

(a) (b)

Fig 2 29

Grouting of the internal dissipators in the footing: (a) wooden template

and C-clamp holding a dowel in place, (b) dowel after grouting. Half of the unbonded

length is visible above the footing surface.


30/181

28

(a) (b)

Fig 2 30

Grouting of the internal dissipators in the column: (a) Detail of the grout

inlet, (b) detail of the air outlet.

2.2.5 Mortar bed

Given the precast nature of the Dual-Shell technology, it is important to take into

account construction tolerances during assemblage of the precast components. To

compensate for lack of precision in the construction and positioning of the members,

it is necessary to leave some space between them to be filled with grout or mortar

once they have been assembled.

However, rocking of the column relies on the opening of a gap between the column

itself and the footing and, in the case of multiple-column bent, also between the

column and the cap beam. Upon rocking, large compressive strains and stresses

arise on the mortar, making a high-performance material necessary for this

application. In fact, crushing of the mortar may lead to a loss of post-tensioning,

which in turn would result in partial or total loss of recentering behavior.

In the specific case, the mortar joint had a thickness of 12.7 mm (0.5 in). It was

obtained mixing BASF Embeco 885 grout with a low percentage of water in order to

have a plastic consistency. This product is characterized by a well-graded blend of

metallic and quartz aggregate, which gives it high strength and good impact

resistance, and therefore makes it ideal for use under cyclic loading. The external


31/181

29

diameter of the mortar bed was smaller than the diameter of the outer steel shell, in

order to avoid contact between the two materials and to prevent the shell from

initiating mortar crushing (Fig. 2.31).

2.2.5.1 Unit 1A Embeco 885 grout

For unit 1A the mortar was obtained from Embeco 885 grout without adding any

fibers. The producer specifies a compressive strength of 76 MPa (11 ksi) at 28 days,

while the one obtained from material tests after 29 and 49 days (day of testing of unit

1A) were 46.4 MPa (6.7 ksi) and 49.16 MPa (7.1 ksi), respectively.

2.2.5.2 Unit 1B Embeco 885 grout with polypropylene fibers

For test Unit1B the mortar was reinforced with Durafiber polypropylene fibers. This

inhibited the growth of cracks once formed, thus preserving the integrity of the mortar

for longer. The compressive cylinder strength obtained from material tests after 28

and 35 days (day of testing Unit1B) were 53.4 MPa (7.8 ksi) and 52.9 MPa (7.7 ksi),

respectively.

Fig 2 31 Detail of the mortar bed (test unit 1B): external diameter smaller than the

outer-steel-shell diameter to retard mortar crushing.


32/181

30

2.2.6 Post-tensioning system

Post-tensioning was provided by four 34.9-mm (1-3/8 in) diameter DSI Threadbars,

made of hot-rolled Grade 150 steel conforming to the ASTM Designation A722, with

a specified ultimate strength of 1034 MPa (150 ksi). Each bar was post-tensioned to

a force of 200 kN (45 kips) from the top, for a total of 801-kN (180-kips) initial post-

tensioning.

The 2.87-m (113-in) long bars were running within the column hollow core, and

were sleeved in metal (unit 1A) or PVC (unit 1B) ducts filled with fluid BASF Embeco

885 grout to simulate corrosion protection. The bars were screwed into the

anchorage devices prearranged in the footing (see 2.2.1) as shown in Figure 2.33, to

allow for subsequent replacement. The anchorage on top of the load stub was

realized by means of 3 steel plates, to uniformly distribute the force on rubber (unit

1A) or polyurethane (unit 1B) bearings underneath.

Fig 2 32

Post-tensioning bar size and length.

Fig 2 33 Post-tensioning bars screwed into footing anchorages (test unit 1A).


33/181

31

In order to preserve the recentering capacity of the system it is important that, upon

gap opening, the tensile strain increment on the PT bars does not lead them to

yielding: this would result in a prestress loss, impairing the functionality of the rocking

system. For this reason, additional deformability was provided to the post-tensioning

bars by placing rubber or polyurethane bearings between the top anchorage plates

and the load stub. With this configuration the tensile deformation demand on the bars

was partially transformed into compressive deformation of the rubber pads. The

bearings acted therefore as springs in series with the bars, increasing the elastic

deformability of the group and avoiding yielding of the bars.

2.2.6.1 Unit 1A Square rubber pads

A bearing made of five square rubber pads alternated with six square steel plates,

with central holes to accommodate the bars, was provided to each PT bar in test unit

1A (Figs. 2.34 and 2.35). Square rubber pads had a side of 191-mm (7.5-in) and a

thickness of 25.4 mm (1 in). Square steel plates had a side of 197 mm (7.75 in) and a

thickness of 3.2 mm (1/8 in). The steel plates were intended to limit the lateral

expansion of the rubber pads under compression, and for this reason they were

joined to the pads by means of an epoxy adhesive; however this practice proved to

be ineffective, with the pads squeezing between the plates.

Fig 2 34 Construction drawing of the top PT anchorage of test unit 1A.


34/181

32

Fig 2 35 Rubber pad and PT anchorage of test unit 1A.

2.2.6.2 Unit 1B Circular Fyfe discs

A bearing made of four Fyfe discs alternated with five circular steel plates, with

central holes to accommodate the bars, was provided to each PT bar in test unit 1A

(Figs. 2.36 and 2.37). Fyfe discs are made of adiprene, a special polyurethane

material developed by Edward Fyfe. Urethane discs had a diameter of 191 mm (7.5

in) and a thickness of 47.6 mm (1-7/8 in). Cirular steel plates had a diameter of 191

mm (7.5 in) and a thickness of 3.2 mm (1/8 in). The steel plates were intended to limit

the lateral expansion of the urethane discs under compression, and for this reason

they were joined to the discs by means of an epoxy adhesive; however this practice

proved to be ineffective, with the discs squeezing between the plates.


35/181

33

Fig 2 36 Construction drawing of the top PT anchorage of test unit 1B.

Fig 2 37

Fyfe disc and PT anchorage of test unit 1B during bar tensioning.


36/181

34

2 3 Material properties

The materials employed for the construction of the specimen have been

characterized in the university laboratory; in the followings the results obtained arereported.

2.3.1 Concrete

High-performance concrete was used to cast the column, the load stub and the

foundation. Concrete compressive strength was measured on standard cylinders

(Fig. 2.38) with a diameter of 152 mm (6 in) and a height of 305 mm (12 in). Tests

were performed at 28 days after casting, at 49 days (day of test of unit 1A) and at 96

days (day of test of unit 1B). Results are listed in Table 2.1. Each value presented is

the average from three specimens tested. The measured values were larger than the

specified strength of 62 MPa (9 ksi) at 28 days.

The concrete modulus of elasticity Ec, defined as the slope of the secant line at 45%

of the compressive cylinder strength, was obtained from the ACI relation below and

summarized in Table 2.1:

= 571000[]

Tab 2 1 Concrete compressive strength and modulus of elasticity at various ages.

material age

[days] [MPa] [ksi] [MPa] [ksi]

28 65.5 9.5 38317 5557

49 (DOT, Unit1A) 70.2 10.2 39643 5750

96 (DOT, Unit1B) 71.8 10.4 40102 5816

f' c E c


37/181

35

Fig 2 38

Preparation of concrete cylinders for material testing.

The consistency and workability of the concrete were assessed by the standard

slump test (Fig. 2.39). The resulting slump was equal to 216 mm (8.5 in), indicating

high workability of the concrete. A value larger than 200 mm (8 in) was required to

ensure adequate consolidation, given the column small thickness.

Fig 2 39 Concrete slump test.


38/181

36

2.3.2 Grout

2.3.2.1 Plastic mortar

The mortar layer between column and footing for both test units was realized with

BASF Embeco 885 cementitious grout. This material is a prepackaged hydraulic

cement-based, metallic-aggregate, high strength, non-shrink grout, with an extended

working time. It is characterized by a well-graded blend of metallic and quartz

aggregate, which gives it high strength and good impact resistance, and therefore

makes it ideal for use under cyclic loading. This grout meets the requirements of

ASTM C 1107. For the specific application it was mixed with a low percentage of

water (5 volumes of grout for 1 volume of water) to obtain a plastic consistency.

For test unit 1B the mortar was reinforced with Durafiber polypropylene fibers,

developed to conform to the requirements of ASTM C 1116-00. This inhibited the

growth of cracks once formed, thus maintaining the integrity of the mortar and

retarding crushing under large compressive strains.

The compressive strengths measured on standard cylinders are reported in tables

2.2 and 2.3. Each value presented is the average from three specimens. Standard

cylinders have a diameter of 51 mm (2 in) and a height of 102 mm (4 in). The

specified strengths for the grout with plastic consistency at 7 and 28 days after

casting were 62 MPa (9 ksi) and 76 MPa (11 ksi), respectively.

Tab 2 2 Mortar strength for test unit 1A at various ages.

Tab 2 3 Mortar strength for test unit 1B at various ages.

material age

[days] [MPa] [ksi]

21 45.8 6.6

23 (DOT, Unit1A) 50.4 7.3

29 46.4 6.7

f' c

material age

[days] [MPa] [ksi]

28 53.4 7.8

35 (DOT, Unit1B) 52.9 7.7

f' c


39/181

37

2.3.2.2 Fluid grout

BASF Embeco 885 was also employed to grout the external dissipators pipes, the

ducts of the post-tensioning bars, the external dissipators anchorages inside the

footing, and the ducts for the internal dissipators. In this case, however, it was mixed

with a higher percentage of water (4.2 liters of water for a 25-kg bag) to obtain a high

fluidity, meaning ease of placement and self-consolidation.

The compressive strengths measured on standard cylinders are reported in tables

2.4 to 2.6. Each value presented is the average from three specimens. Standard

cylinders have a diameter of 51 mm (2 in) and a height of 102 mm (4 in). The

specified strengths for the grout with fluid consistency at 7 and 28 days after casting

were 48 MPa (7 ksi) and 62 MPa (9 ksi), respectively.

Tab 2 4 Strength of the grout used to connect the external dissipators anchors in

the footing of test Unit 1A.

Tab 2 5 Strength of the grout used to connect the internal dissipator dowels to the

footing of test unit 1B.

Tab 2 6 Strength of the grout used to conenct the internal dissipator dowels to the

column of test unit 1B.

material age

[days] [MPa] [ksi]

18 (DOT, Unit1A) 47.1 6.8

f' c

material age

[days] [MPa] [ksi]

41 (DOT, Unit1B) 52.7 7.6

f' c

material age

[days] [MPa] [ksi]

27 (DOT, Unit1B) 59.2 8.6

f' c


40/181

38

2.3.3 Steel

Hot rolled, ASTM A576 Grade 1018 carbon steel bars were used for the external

dissipators. A 25.4-mm (1-in) diameter bar was tested in monotonic tension to

characterize the steel (Fig. 2.40). The values of the yield stress and strain, elastic

modulus, and ultimate stress and strain are reported in Table 2.7. A detailed stress-

strain curve (Fig. 2.41) was obtained using a clip extensometer.

The internal dissipators were made of 316LN, Grade 75 stainless steel. A tensile

test on a 12.7-mm (0.5-in) diameter bar was performed to characterize the steel. The

yield stress and strain, elastic modulus, and ultimate stress and strain are reported in

Table 2.7, while a detailed stress-strain curve is shown in Figure 2.41.

Fig 2 40 Steel bar testing setup.


41/181

39

Tab 2 7 Steel properties of external and internal energy dissipators.

Fig 2 41

Experimental stress-strain relationship for the external and internal

dissipators steel.

y u

[MPa] [ksi] [MPa] [ksi] [MPa] [ksi] [%] [%]

A576 331 48 490 71 220632 32000 0.15 33.90

316LN 834 121 889 129 220632 32000 0.38 25.70

material fy fu E s


42/181

40

2.3.4 Bearing pads

SA-47 rubber pads, produced by Fabreeka International, were used to assemble

the bearings mounted in series with the post-tensioning bars in unit 1A. These pads

are made from masticated rubber using a blend of recycled rubber compounds and

synthetic fiber reinforcement. The random-oriented-fibers structure of the material

provides enhanced compressive strength, stiffness, and tensile strength when

compared to unreinforced or virgin rubber pad materials.

The design of the bearings, i.e. number and thickness of the pads to be stacked in

each bearing, was initially based on the stiffness provided by producers

specifications; then it was refined on the basis of cyclic compressive tests performed

on two rubber pads (Fig. 2.42). The resulting elastic modulus under cyclic loading

was 524 MPa (76 ksi). During testing the material underwent significant creep

deformations, even under cyclic loading: its behavior therefore depends on the load

rate.

Fig 2 42 SA-47 rubber pad during testing.


43/181

41

Fyfe discs were used to assemble the bearings mounted in series with the

post-tensioning bars in unit 1B. They are made of adiprene urethane rubber, a

thermosetting material characterized by unusual toughness. It combines resilience

and high load bearing capacity with resistance to impact, abrasion and compression

set and degradation by oxygen, ozone and oil. In addition to the properties

distinguishing it from conventional elastomers, Adiprene also differs in its physical

form: it is a liquid polymer which is mixed with a curing agent and can be fabricated

by casting.

The design of the bearings, i.e. number and thickness of Fyfe discs to be stacked in

each bearing, was initially based on the stiffness provided by producers

specifications; then it was refined on the basis of a cyclic compressive test performed

on one urethane disc (Fig. 2.43). The resulting elastic modulus under cyclic loading

was 331 MPa (48 ksi). Also in this case creep deformations and load-rate dependent

behavior were evident.

Fig 2 43 Adiprene Fyfe disc during testing.


44/181

42

2 4 Instrumentation

The test specimen was instrumented to measure deformations, displacements,

rotations, and forces during testing. Strain gauges, potentiometers, inclinometers,and load cells were used at this scope, respectively.

2.4.1 Strain gauges

The instrumentation included strain gauges mounted on the outer steel shell of the

column (Figs. 2.44 and 2.45). 5-mm horizontal strain gauges, placed on the north

and south faces of the specimen, were used to monitor the tensile stresses induced

on the shell by the lateral expansion of compressed concrete, and thus to capture the

confinement effects. 5-mm vertical strain gauges were placed on the north face of the

column to detect compressive stresses on the shell at the base level due to eventual

contact with the mortar (theoretically precluded, being the external diameter of the

mortar bed smaller than the diameter of the steel shell). Finally, 5-mm strain gauges,

arranged in a delta-rosette configuration, were employed to determine shear

deformations along the east and west sides of the specimen.

In Unit1B additional 5-mm vertical gauges were placed on the nort-east and south-

west faces of the outer shell in front of the internal dissipators, in order to monitor the

stress transfer from the dowel bar to the steel shell.

Strain gauges were also used to instrument two diametrically opposite post-

tensioning bars in both units 1A and 1B (Fig. 2.46). Two sensors were applied to

each bar, in order to monitor axial extension and bending of the bar.

Two diametrically opposite external dissipators in unit 1A were instrumented with

two strain gauges each (Fig. 2.47). These sensors, applied to the virgin portion of the

bar which was expected to remain elastic, were used to obtain the dissipator forces

through elasticity relationships.

Strain gauges were also applied to two diametrically opposite internal dissipators in

unit 1B. Four sensor were mounted on each bar, as shown in Figure 2.48, to

measure axial and buckling deformations of the dowels.

More details about strain gauges locations can be found in Appendix B.


45/181

43

Fig 2 44

Column outer-shell strain gauges for unit 1A.


46/181

44

Fig 2 45

Column outer-shell strain gauges for unit 1B.


47/181

45

Fig 2 46 PT-bar strain gauges for units 1A and 1B.

Fig 2 47

External dissipators instrumentation for unit 1A.


48/181

46

Fig 2 48

Internal dissipators strain gauges for unit 1B.

2.4.2 Potentiometers and inclinometers

A cable-extension position transducer, referred to as horizontal string potentiometer,

was connected to the load stub at the lateral loading point, in order to record the

lateral displacement of the column and control the actuator (Fig. 2.49).

Four vertical potentiometers were installed on the north face and four on the south

face of the column (Figs. 2.50, 2.51, and 2.52). These spring-loaded potentiometers

were mounted on steel rods sticking out from the column shell, in order to monitor the

average rotation between the rods and evaluate the column curvature.

Two horizontal spring-loaded potentiometers were added at the column base of unit

1B, to monitor eventual shear sliding.

A spring-loaded potentiometer was mounted on two diametrically opposite external

dissipators of unit 1A (the same instrumented with strain gauges, Fig. 2.47), in order

to record their axial deformation and capture their hysteretic response.

An inclinometer was mounted on the foundation and one on top of the load stub to

record the rotation of these elements (Figs. 2.50 and 2.51).

The layout of external instrumentation is shown in detail in Appendix B.


49/181

47

Fig 2 49

Horizontal string potentiometer installation on the load stub of unit 1A.

Fig 2 50

Potentiometers and inclinometers for unit 1A.


50/181

48

Fig 2 51

Potentiometers and inclinometers for unit 1B.

Fig 2 52 Potentiometers on the column south face of unit 1B.


51/181

49

2.4.3 Load cells

Load cells were used to monitor the force on two tie-down rods, tensioned by

vertical jacks, intended to apply a gravity-equivalent load to the specimen. They were

mounted between the load stub and the jacks, as shown in Figure 2.53.

2.4.4 Other instrumentation

A data acquisition system was used to interpret and record the signals provided by

each instrumentation channel and convert them into digital information. In addition to

active measurement provided by the instrumentation described above, three video

cameras were mounted on site: one was zooming on the column-footing joint from

north; one was recording the full column from the east; the remaining one was

installed on the contrast wall and was recording the full unit. Furthermore the

specimen response was documented through digital photos and notes.


52/181

50

2 5 Test protocol

The preliminary stage of loading consisted of applying a vertical load to the

specimen, simulating the gravity forces during a seismic event. The vertical load wasapplied by two vertical jacks (Figs.2.53 and 2.54), positioned on the load stub and

connected to the strong floor by one 31.8-mm (1-1/4 in) diameter tie-down rod each.

Keeping the vertical force constant, the test unit was subjected to quasi-static

reversed cyclic loading by a 734-kN (165-kips) MTS horizontal actuator contrasting

on the strong wall (Figs. 2.53 and 2.55).

Fig 2 53 Test layout.


53/181

51

Fig 2 54

Vertical jack (yellow) and load cell (red) for the application of gravity

load.

Fig 2 55 Actuator for the application of lateral forces.


54/181

52

The first part of lateral loading was run in force control. Each load increment

consisted of three cycles of push and pull excursions. Two load increments were

applied as follows:

3 cycles up to 111 kN (25 kips);

3 cycles up to 231 kN (52 kips).

The first lateral load value (111 kN or 25 kips) was intended to induce

decompression at the base section, thus incipient gap opening. The second

increment (231 kN or 52 kips) corresponded to 75% of the base nominal moment

capacity, according to a preliminary prediction on unit 1A. The force-controlled part of

the loading protocol is graphically shown in Figure 2.58.

Fig 2 58

Force-controlled cycles.


55/181

53

Upon completion of the force-controlled loading cycles, the test proceeded in

displacement control. The following drift ratios, defined as the ratio between the

lateral displacement of the loading point and the height of this point above the

footing, were targeted:

3 cycles up to 0.5% drift ratio

3 cycles up to 0.75% drift ratio

2 cycles up to 1% followed by 1 cycle up to 0.75% drift ratio

2 cycles up to 1.5% followed by 1 cycle up to 1% drift ratio


2 cycles up to 3% followed by 1 cycle up to 2% drift ratio

2 cycles up to 5% followed by 1 cycle up to 3% drift ratio

2 cycles up to 7.5% followed by 1 cycle up to 5% drift ratio


For unit 1A the test ended with only one cycle at 10% drift ratio since three out of

six dissipators had already failed and incipient failure of the system was almost

reached, whereas for unit 1B the full displacement history was run. The displacement

control loading protocol is graphically shown in Figure 2.59.

Fig 2 59

Displacement-controlled cycles.


56/181

54

Chapter 3

Experimental Results

Data collected during testing were post-processed and reviewed to provide

additional insight into the response of the two units. Cyclic lateral loading was applied

in the north-south direction; northward displacements and forces are taken as

positive.

3 1 Test unit 1A

Testing of unit 1A was performed on October 21st, 2011. The test assemblage

together with a close-up view of the external buckling-restrained energy dissipators,

connecting the column to the footing, is shown in Figure 3.1.

The full hysteretic response of the unit is plotted in Figure 3.2, while Figure 3.3

zooms on the cycles up to 3% drift ratio. On the x-axis the lateral displacement is

divided by the height of the column thus defining the drift ratio; on the y-axis the

applied lateral load is divided by the weight (i.e. the applied gravity load) thus

defining the base shear coefficient.

The stress-strain hysteretic responses of the external dissipators are shown in

Figures 3.4 and 3.5 In this case data collection was interrupted due to flexural

distortion of the dissipators, which caused the tips of the spring-loaded

potentiometers to run off their correct position. The different behavior in tension and

compression of the dissipators is due to the additional compressive strength provided

by the grout in the buckling-restrained portion of the bars. Fictitious stresses are

plotted, since the dissipator force is divided by the area of the bar reduced section;

these are not real stresses, because the area of grout is not properly accounted for.

Figure 3.6 shows the strain histories for the two instrumented post-tensioning bars:

as it can be noticed, the strain starts from the value corresponding to the initial post-

tensioning. Data are interrupted due to damage or saturation of the strain gauges.


57/181

55

Fig 3 1 Unit1A: test assemblage (left), and close-up view of the external buckling-

restrained energy dissipators, connecting the column to the footing (right).

Fig 3 2 Unit 1A: column full hysteretic response.


58/181

56

Fig 3 3 Unit 1A: column hysteretic response up to 3% drift-ratio cycles.

Fig 3 4 Unit 1A: north-west dissipator stress-strain response.


59/181

57

Fig 3 5 Unit 1A: south-east dissipator stress-strain response.

Fig 3 6 Unit 1A: post-tensioning bars strain histories.


60/181

58

Cracks started forming at the column-mortar bed interface during the cycles at

231 kN (52 kips); due to this a first loss of stiffness was observed on the diagram of

Figures 3.2 and 3.3. During the 0.5% drift-ratio cycles the cracks opened and

closed with the cyclic load and expanded laterally. Cracks extended all around the

interface perimeter during 0.75% drift-ratio cycles (Fig. 3.7).

The gap at the column-mortar bed interface was visibly opened during 1% drift-

ratio cycles; upon each reversal it extended until half the column depth.

The mortar bed started flaking during 1.5% drift-ratio cycles, and showed

extensive flaking and permanent compressive deformation on the north and south

sides (extreme fibers) after 3% drift-ratio cycles, as shown in Figure 3.8. This could

be one of the reasons for the sudden loss of strength and stiffness at about 2.5%

drift ratio during the first positive and negative cycles to 3% drift ratio (Figs. 3.2 and

3.3).

Mortar bed started crushing extensively during 5% drift-ratio cycles, with

significant loss of stiffness.

Fig 3 7 Unit 1A: cracking at the column-mortar bed interface during 0.75% drift-

ratio cycles.


61/181

59

Fig 3 8 Unit 1A: mortar bed during 3% drift-ratio cycles.

External dissipators subjected to tension visibly extended outside the grouted pipe

during 1.5% drift-ratio cycles and started bending between the buckling-restrained

central portion and the end connections during 3% drift-ratio cycles (Figs. 3.9 and

3.10). This distortion was due to the rotation imposed by the rocking kinematic on the

dissipators. It could be another reason for the loss of strength and stiffness at about

2.5% drift ratio during the first positive and negative cycles to 3% drift ratio (Figs.

3.2 and 3.3).

Fig 3 9 Unit 1A: dissipator extending outside the grouted pipe during 3% drift-

ratio cycles.


62/181

60

Fig 3 10 Unit 1A: dissipator distortion during 5% drift-ratio cycles.

The north-west dissipator fractured during the first negative cycle to -7.5% drift ratio,

nearly at peak displacement; the fracture was exposed during subsequent cycles,

and is visible in Figure 3.11. This corresponded to a jump on the graph of Figure 3.2.

A second dissipator fractured on the south side during the second positive cycle to

+7.5% drift ratio, nearly at peak displacement, with another jump visible in Figure 3.2.

The loops remained however stable. A third dissipator fractured on the south side

during the first positive cycle to +10% drift ratio, when +7.5% drift ratio was reached,

with the same effect on the graph of Figure 3.2.


63/181

61

Fig 3 11 Unit 1A: fractured north-west dissipator during 10% drift-ratio cycles.

Figures 3.12 and 3.13 show the condition of the specimen at the end of testing.

Distorted and fractured dissipators are visible on the north (left) and south (right)

sides of the column in Figure 3.12. The east and west dissipators, closer to the

neutral axis, did not present this type of behavior. Crushing of the mortar bed is

evident in Figure 3.13.

Fig 3 12 Unit 1A: specimen at the end of the test.


64/181

62

Fig 3 13 Unit 1A: mortar bed crushed at the end of the test.

As an additional detail, it was observed that the three dissipators that fractured

during the test experienced a flexural distortion towards the exterior of the column;

instead the north-east dissipator, the only one that didnt fracture among the four

devices subjected to major solicitations, showed a lateral inflection (Fig. 3.14).

Moreover, when the column was lifted from the footing it was possible to notice a

residual deformation on the concrete, causing a shortening of about 13 mm (0.5 in)

with respect to the steel shells on both the north and south sides. The high

longitudinal compressive stresses experienced by the concrete caused a lateral

expansion that permanently deformed the steel shells (Fig. 3.15).

Fig 3 14 Unit 1A: distortion of the dissipators on the north side of the column.


65/181

63

Fig 3 15 Unit 1A: permanent deformations on concrete and steel shells at the

base of the column (south side).

The overall specimen performance was satisfactory. Self-centering behavior of the

specimen was lost after reaching 3% drift ratio (Fig. 3.2).

The reference yield displacement was estimated to correspond to 0.5% drift ratio

(Fig. 3.3). In correspondence to incipient flaking of the mortar bed, which happened

at a drift ratio of 1.5%, a displacement ductility of about 3 was developed. Adisplacement ductility larger than 6 was obtained when bed mortar started crushing

(drift ratio larger than 3%), and a ductility of nearly 15 was reached when the first

dissipator fractured (nearly 7.5% drift ratio).


66/181

64

3 2 Test unit 1B

Testing of unit 1B was performed on December 6th, 2011. The test assemblage

together with a close-up view of the internal stainless steel energy dissipators is

shown in Fig. 3.16.

The full hysteretic response of the unit is plotted in Figure 3.17, while Figure 3.18

zooms on the cycles up to 3% drift ratio. As for unit 1A, the drift ratio and the base

shear coefficientare plotted.

The strain histories of the two instrumented internal dissipators are shown in Figure

3.19. Data recording stopped when the strain gauges applied to the bar were

damaged or saturated.

Fig. 3.20 shows the strain histories for the two instrumented post-tensioning bars:

as it can be noticed, the strain starts from the value corresponding to the initial post-

tensioning. Data are interrupted due to damage or saturation of the strain gauges.

Fig 3 16 Unit 1B: test assemblage (left), and close-up view showing the internal

energy dissipators (stainless steel dowel bars), the PVC-sleeved PT bars, and the

fresh mortar bed, during placement of the column on the footing.


67/181

65

Fig 3 17 Unit 1B: column full hysteretic response.

Fig 3 18 Unit 1B: column hysteretic response up to 3% drift-ratio cycles.


68/181

66

Fig 3 19 Unit 1B: dissipators strain histories.

Fig 3 20 Unit 1B: PT bars strain histories.


69/181

67

Cracks started forming at the column-mortar bed interface on the north and south

sides, during cycles at 231 kN (52 kips); due to this a first loss of stiffness was

observed on the diagram of Figures 3.17 and 3.18. The crack extended all around

the interface perimeter during 0.75% drift-ratio cycles (Fig. 3.21). However, along

the north-west sector the crack propagated along the mortar bed-footing interface

instead of the column-mortar one.

The gap at the column- mortar bed interface opened visibly during 1% drift-ratio

cycles, following the crack pattern described above; upon each reversal it extended

until half the column depth.

The mortar bed started flaking during 2% drift-ratio cycles and showed extensive

flaking and permanent compressive deformation on the north and south sides

(extreme fibers) after 3% drift-ratio cycles, as shown in Fig. 3.22. This could be the

reason for the sudden loss of strength and stiffness at about 2.5% drift ratio during

the first positive and negative cycles to 3% drift ratio (Figs. 3.17 and 3.18).

Mortar bed crushing progressed during 5% drift-ratio cycles, but not abruptly, and

became extensive under 7.5% drift-ratio cycles; the stiffness was evidently

reduced.

Fig 3 21 Unit 1B: cracking at the column-mortar bed interface during 0.75%

drift-ratio cycles.


70/181

68

Fig 3 22 Unit 1B: mortar bed during 3% drift-ratio cycles.

The first dissipator fractured on the north side during the second negative cycle to -

7.5% drift ratio, nearly at peak displacement; the bang-like noise was clearly heard

by the audience. This corresponded to a jump on the graph of Figure 3.17. A second

dissipator fractured on the south side during the first positive cycle to +10% drift ratio,

when about +7.5% drift ratio was reached, with another jump visible in Figure 3.17. A

third dissipator fractured on the south side during the second positive cycle to +10%

drift ratio, when about +8.5% drift ratio was reached, with the same effect on the

graph of Figure 3.17. A fourth dissipator fractured on the north side during the

second negative cycle at -10% drift ratio, when about -7.5% drift ratio was reached,

producing a fourth jump in Figure 3.17.

Figure 3.23 show the specimen during the last cycles of testing. Crushing of the

mortar bed at the end of the test is evident in Figure 3.24.


71/181

69

Fig 3 23 Unit 1B: column during 10% drift ratio cycles.

Fig 3 24 Unit 1B: mortar bed crushed at the end of the test.


72/181

70

As an additional detail, when the column was lifted from the footing it was possible

to notice a residual deformation on the concrete, causing a shortening of about 13

mm (0.5 in) with respect to the steel shells on both the north and south sides. The

high longitudinal compressive stresses experienced by the concrete caused a lateral

expansion that permanently deformed the steel shells (Fig. 3.25).

Unit 1B offered an improved response compared to unit 1A: the polypropylene

fibers gave to the mortar bed a more ductile behavior, and crushing became

extensive under a larger drift ratio. The recentering capacity was then preserved up

to greater displacements (7.5% drift ratio).

The reference yield displacement was estimated to correspond to 0.5% drift ratio

(Fig. 3.18). Flaking of the bed mortar joint was observed in the cycles to 2% drift

ratio: this corresponds roughly to a displacement ductility of 4. A displacement

ductility larger than 10 was developed when bed mortar was significantly crushing

(drift ratio larger than 5%), and a ductility of nearly 15 was reached when the first

dissipator fractured (nearly 7.5% drift ratio).

Fig 3 25 Unit 1B: permanent deformations on concrete and steel shells at the

base of the column (north side).


73/181

71

Chapter 4

Numerical Modeling of the Tests

A 3D numerical modeling of the tests was performed and validated with the

experimental results. To this purpose the software OpenSees was used. OpenSees,

the Open System for Earthquake Engineering Simulation, is an object-oriented, open-

source software framework. It allows users to create both serial and parallel finiteelement computer applications for simulating the response of structural and

geotechnical systems subjected to earthquakes and other hazards. OpenSees is

primarily written in C++ and uses several Fortran and C numerical libraries for linear

equation solving, and material and element routines

(http://opensees.berkeley.edu/wiki/index.php/Main_Page). The user interfaces with

applications such as OpenSees.exe, which interpret the input written in an extended

Tcl programming language.

A good physical understanding of the problem is required in order to correctly

implement the model and the associated parameters, and to be confident with the

output. Simple sketches of the models are presented in Figure 4.1.

The elastic cantilever behavior of the column member was modeled with elastic

frame elements, connected to the mortar joint at the base and to the point of

application of the load at the top.

Multiple non-linear truss elements represented the mortar at the interface between

column and footing.

The post-tensioning bars were modeled as non-linear truss elements, with an initial

stress equal to the effectively applied prestress. They were fixed at the base and

connected to the top node of the column by rigid elements.

External energy dissipating devices for test unit 1A were represented by non-linear

frame elements, as well as internal dowel bars for unit 1B. In both cases, energy

disispators lower ends were fixed to the footing, and their upper ends were

connected to a column intermediate node by rigid links.


74/181

72

Fig . 4.1 South-east view of the proposed 3D model for test unit 1A (left) and unit

1B (right).

Externaldissipator

Internaldissipator

Mortar bed

PT bars

Column

Loading point


75/181

73

The analysis was performed in two stages, First, the gravity load was applied and

held constant. Then a cyclic quasi-static displacement history was assigned to the

load stub centroid, modeled as a node positioned just above the column, at a

distance equal to half the load stub height. Since the deformations within the load

stub member are expected to be negligible, the node created in that position was

linked to the top of the column with a rigid element.

All the rigid elements mentioned in this chapter were realized with elastic elements

with a very high stiffness compared to the elements they were connecting.

The analysis was performed under the hypothesis of small displacements, i.e. linear

geometryin OpenSees language. In fact the vertical load was applied using tie-down

rods, which were leaning together with the column under lateral displacement; as a

consequence, the applied force was always directed along the column axis and

interaction between axial load and lateral displacement was prevented.

The results of the analysis were recorded and used for the validation of the models

which is presented in chapter 5.

Special care was given to correctly modeling the system behavior for drift ratios up

to 5%, since larger values of displacements would be too demanding for a bridge

superstructure. Moreover, modeling the material behavior near collapse requires the

implementation of complicated routines and the knowledge of many parameters that

were not available from experimental tests.

In the first section of this chapter the material models are introduced with some

theoretical background. The second section is about the modeling of each single

component of the specimens. In the third section the parameters defining the

analysis are discussed. A print of the whole Tcl script is attached as an appendix to

this work.


76/181

74

4.1 Material models

4.1.1 Concrete 01

Among the concrete models implemented in OpenSees, Concrete 01 is the simplest

one but gives the possibility to model the material non linearity with an accurate

response. It is based on the Kent-Scott-Park concrete material model, with degrading

linear unloading/reloading stiffness according to the work of Karsan and Jirsa;

concrete tensile strength is neglected. The required input parameters are listed in

Figure 4.2, while the material behavior is shown in Figure 4.3.

Fig . 4.2 Command line for the definition of Uniaxial Material Concrete01.

Fig. 4.3 Typical backbone stress-strain relation for Concrete 01.


77/181

75

The first branch of the backbone curve is a second degree parabola, up to the point

where the maximum stress, fpc, is attained with a corresponding strain c0. The initial

tangent slope is given by 2*fpc /c0. The descending branch is assumed to be linear.

The last branch, starting from the strain u, assumes a constant value of the stress

equal to the residual strength fpcu: this assumption is based on the expectation that

member failure will occur before the concrete strain becomes unrealistically large.

The stress-strain relationship is expressed by the following equations:

for cc0

= [2

(

)]for c> c0

= [1 ( )]where:

= 0.5+ The strain c50u corresponds to a stress equal to 50% of the maximum concrete

strength, and c50h is the additional strain observed for confined concrete in

correspondence to the same stress. In OpenSees c50his taken equal to zero, since

the confinement effect is accounted for when defining the peak point (fpc , c0) and the

other parameters of the material routine.

The degrading unloading/reloading stiffness for Concrete 01 is based on the work of

Karsan and Jirsa. They defined these branches as second degree parabolas passing

by three points, shown in Fig. 4.4. However, in the OpenSees model points A and D

are coincident and the unloading/reloading behavior is not defined by second degree

parabolas, but by a linear branch connecting point A with point B. The equation for

the strain at point B has been provided by Karsan and Jirsa as:

= 0.145+ 0.13


78/181

76

Fig . 4.4 Unloading and reloading curve proposed by Karsan and Jirsa.

The effect of confinement has been taken into account according to the model

proposed by Mander, et al. (1988). The stress-strain model is illustrated in Figure 4.5

and is based on an equation by Popovics.

Fig . 4.5 Stress-strain model for confined and unconfined concrete (Mander, et al.,

1988).


79/181

77

The confinement compressive strength is given by:

= 1.254+ 2.2541 +7.94

2

where fc0 is the unconfined concrete compressive strength and fl is the effective

lateral confining stress on the concrete, equal to:

=12 with sindicating the ratio of the volume of transverse confining steel to the volume

of confined concrete core, fyh being the yield stress of the confining steel, and ke

defining the efficiency of confinement.

The strain corresponding to fccis obtained as:

= 1 + 5 1where the unconfined peak strain c0is given by an empirical relationship with fc0:

= 0.0017 + 60000[]The ultimate concrete compressive strain, cu, has been empirically related to the

volumetric confining ratio:

= 0.004 + 1.5The ultimate stress in concrete can be obtained as:

= 1 + where:

=

=


80/181

78

In the above equations, Ec is the initial elastic modulus of concrete and Esec the

secant modulus at peak, equal to:

=

When modeling a confined concrete with Concrete 01 material, the following

correspondence can be established between OpenSees parameters and Manders

parameters: at peak fpc is taken equal to fccand c0equal to cc ; at ultimate fpcu is

taken equal to fcuand uequal to cu.

4.1.2 Steel 02

OpenSees material model Steel 02 is used to construct a uniaxial Giuffr-

Menegotto-Pinto steel material object with isotropic strain hardening. The Giuffr-

Menegotto-Pinto stress-strain relationship, well described in the report by Filippou, et

al. (1983), offers numerical efficiency and good agreement with experimental results

from cyclic tests on reinforcing steel bars.

The required input parameters are listed in Figure 4.6, while the hysteretic behavioris represented in Figure 4.7.

Fig . 4.6 Command line for the definition of Uniaxial Material Steel 02.


81/181

79

Fig. 4.7 Stress-strain relationship for Steel 02.

The model is described by the following equation:

= + (1 )(1 ) where:

= and:

= It represents a smooth curved transition from an asymptotic straight line with slope

E0to another asymptotic straight line with slope E1; bis the ratio between E1 and E0,

referred to as the bilinear factor. and are the current stress and strain; * and *

are normalized stress and strain; yand yare the coordinates in the stress-strain

plane of the intersection of the two asymptotes (point A in figure 4.7); r and r,

initially set to zero, are the stress and strain at the point where the last strain reversal


82/181

80

took place, with stress of equal sign (point B in figure 4.7); R is a parameter

describing the curvature of the transition curve between the asymptotes. The

parameters y, y,r, rand R are updated at each strain reversal.

The parameter Ris given by the following relationship:

= + where R0is the value of Rduring first loading; is the ratio of the maximum plastic

strain over the initial yield strain and is updated following a strain reversal; a1and a2

are experimentally determined parameters to be defined together with R0.

Filippou et al. also proposed a stress shift sh in the linear yield asymptote, to

account for the isotropic hardening, depending on the maximum plastic strain:

= ( )where maxis the absolute maximum strain at the instant of strain reversal; y0and

y0 are the initial yield stress and strain; a3 and a4 are experimentally determined

parameters.


83/181

81

4.2 Model components

4.2.1 Column

The column was modeled with two elastic elements in series: one spanning from

the top of the mortar bed to the connection with the dissipators, and another one

running from this point up to the column top at the connection with the load stub. One

reason of this choice is the need of an intermediate point to connect the dissipators

to the column axis. Moreover, in the lower part the column outer shell is not effective

in carrying any tensile stresses, since they are carried by the energy dissipating

devices.

Therefore area, second moment of area, and torsional moment of inertia of the

cross section were calculated accounting only for the concrete in the lower element,

while the transformed section has been considered for the upper segment where the

outer shell is contributing.

Based on the particular technology adopted the column is believed to remain in the

elastic field under the applied load and to rock about the foundation at the mortar bed

level. Therefore the element type used to model it is the Elastic Beam Column

Element. For a three-

Documents

Vervelidis - Seismic Response of Self-centering Precast Concrete Dual-shell Steel Columns