21768729-MIT

Embed Size (px)

Citation preview

  • 8/18/2019 21768729-MIT

    1/158

    STRUCTURAL BEHAVIOR OF FIBER

    REINFORCED

    MORTAR

    RELATED

    'IO

    MATERIAL

    FRACTURE

    RESISTANCE

    by

    ROBERT

    JAMES

    WARD

    B.E.

    University

    College

    (1986)

    Galway, Ireland

    SUBMITTED

    IN PARTIAL

    FULFILLMENT

    OF

    THE REQUIREMENTS FOR THE DEGREE OF

    MASTER

    OF

    SCIENCE

    IN

    CIVIL ENGINEERING

    at

    the

    MASSACHUSETTS

    INSTITUTE

    June

    1989

    OF TECHNOLOGY

    Massachusetts

    Institute

    of

    Technology

    1989

    All

    rights

    reserved

    Signature

    of Author

    Certified

    by

    Accepted

    by

    -------

    Ch

    Signature

    redacted

    Department

    of Civil

    Engineering

    Signature redacted

    Victor

    C. Li

    Associate

    Professor

    of Civil

    Engineering

    Thesis

    Supervisor

    Signature

    redacted

    J

    Ole

    S.

    Madsen

    airman,

    Departmental

    Committee

    on Graduate

    Students

    Department

    of Civil

    Engineering

    NSS.

    JIST.

    T d

    JU '89

      C

    ARCHIVES

    0

  • 8/18/2019 21768729-MIT

    2/158

    STRUCTURAL BEHAVIOR

    OF

    FIBER REINFORCED

    MORTAR

    RELATED TO

    MATERIAL FRACTURE

    RESISTANCE

    by

    ROBERT

    JAMES

    WARD

    Submitted to the Department of

    Civil Engineering

    on May 11, 1989, in partial fulfillment of the

    requirements for the Degree

    of Master

    of

    Science

    in

    Civil Engineering

    ABSTRACT

    An indirect J-integral

    technique

    for measuring

    the tension

    softening

    curve of non-yielding

    materials like

    concrete

    is

    presented and is

    applied

    to

    mortars

    reinforced

    with

    various short fiber

    types

    (steel

    and

    synthetic). The tension

    softening

    curve serves

    to

    characterize and

    measure material fracture

    resistance. The dependence

    of

    structural

    behavior

    on

    material

    fracture

    resistance is

    investigated

    through

    third-point

    loading tests on unreinforced fiber

    mortar beams

    which fail

    in flexure and through

    center-point loading tests on

    longitudinally

    reinforced

    fiber

    mortar

    beams

    without shear stirrups which

    fail

    in

    shear.

    Conclusive

    evidence,

    relating improved structural

    performance directly

    to

    improvements in material

    fracture

    resistance

    due to fiber reinforcement,

    is

    presented.

    Semi-empirical

    formulae are

    proposed which

    relate

    flexural toughness

    indices

    directly to parameters

    involving

    just

    the flexural strength,

    the

    splitting tensile strength and

    the fiber

    length. Also,

    semi-empirical

    design

    formulae

    are

    proposed

    which

    relate

    the

    ultimate shear strength

    of

    longitudinally

    reinforced beams

    with

    fibers as

    shear reinforcement,

    to

    the

    material fracture resistance

    (represented by a

    combination

    of

    flexural and

    splitting tensile

    strengths), the

    shear span/effective

    depth

    ratio,

    the

    longitudinal

    reinforcement ratio

    and the

    beam

    depth. All

    proposed formulae are suitable for

    the purposes of design and quality

    control.

    Thesis Supervisor: Dr.

    Victor

    C. Li

    Title:

    Associate Professor of Civil Engineering

    2

  • 8/18/2019 21768729-MIT

    3/158

    A

    I'DG ENT

    I wish to

    thank the

    following

    people:

    Professor Victor

    Li

    for

    introducing

    me

    to

    the

    subjects of fracture

    mechanics

    and

    fiber

    reinforced

    concrete,

    and for

    overseeing

    my

    entire

    research

    program.

    He

    provided

    me with

    invaluable

    guidance,

    support

    and

    encouragement

    and

    above

    all I

    thank

    him for

    making

    my

    stay at M.I.T.

    a

    truly enjoyable

    experience.

    My

    fellow graduate

    students

    who

    so

    often

    offered

    their

    advice

    and

    help,

    especially

    in the

    laboratory.

    Professor

    Stanley

    Backer

    who

    always

    ensured

    that I

    had

    a large

    and varied

    supply

    of synthetic

    fibers

    with

    which

    to work.

    Mr.

    Rick

    Smith of

    Ribbon

    Technology

    Corporation,

    Ohio,

    who

    provided

    me

    with

    all

    the steel

    fibers.

    Mr.

    Kevin

    Grogan of

    W.

    R.

    Grace and

    Co., Cambridge,

    Mass.,

    who

    provided

    me

    with

    the

    superplasticiser.

    The

    U.S.

    National

    Science

    Foundation

    and

    the Shimizu

    Corporation

    of

    Japan

    who

    both

    provided

    vital

    funding

    which

    made

    this

    research

    program

    possible.

    Miss

    Irene Uesato,

    who

    took

    time

    out

    of her

    very

    busy

    schedule,

    to

    type

    this

    thesis.

    3

  • 8/18/2019 21768729-MIT

    4/158

    TABLE

    OF

    CONTENTS

    Title Page

    Abstract

    Acknowledgements

    Table

    of

    Contents

    List

    of

    Figures

    List of

    Tables

    Chapter

    1:

    INTRODUCTION

    Chapter

    2: TENSION

    SOFTENING

    CURVE

    BY

    J-INTEGRAL

    TECHNIQUE

    2.1

    Introduction

    2.2

    Theoretical

    Basis

    of

    J-Integral

    Technique

    2.3

    Numerical

    Verification

    of

    Test

    Technique

    2.4

    Experimental

    Procedure

    2.4.1

    Specimen

    preparation

    2.4.2

    Testing

    procedure

    2.5

    Results

    and

    Data

    Analysis

    2.6

    Calculation

    of

    GF

    2.7

    Discussion

    and

    Recommendations

    for

    the

    J-Integral

    Test.

    2.8

    Current

    Status

    of

    J-Integral

    Test

    Technique

    Chapter

    3:

    FLEXURAL

    BEHAVIOR

    OF

    FIBER

    REINFORCED

    MORTAR

    3.1

    Introduction

    3.2

    General

    Behavior

    of

    Fiber

    Concrete

    in

    Flexure

    3.3

    Experimental

    Program

    3.3.1

    Specimen

    preparation

    3.3.2

    Testing

    procedure

    3.4

    Compressive

    Strength

    3.5

    Splitting

    Tensile

    Strength

    3.6

    Flexural

    Strength

    3.6.1

    Size

    dependence

    of

    flexural

    strength

    3.6.2

    Flexural

    strength

    related

    to

    fracture

    resistance

    3.7

    Flexural

    Load-Deflection

    Curves

    4

    1

    2

    3

    4

    6

    10

    11

    17

    17

    18

    21

    22

    22

    24

    25

    27

    28

    30

    40

    40

    42

    46

    46

    47

    48

    49

    50

    51

    52

    57

  • 8/18/2019 21768729-MIT

    5/158

    3.7.1

    Effect

    of

    fiber

    type

    and

    volume

    fraction

    on

    the

    flexural

    load-deflection

    curve

    57

    3.7.2

    Effect

    of beam

    size

    on the

    flexural

    load-deflection

    curve

    60

    3.8 Flexural

    Toughness

    Indices

    62

    3.8.1

    15'

    10'

    3

    and

    150

    indices

    62

    3.8.2

    Tmaxp

    T

    50

    and T

    10

    indices

    65

    3.9

    Simple

    Flexural

    Toughness

    Estimates

    68

    Appendix

    3.1

    Comparison

    of Size-Effect

    Predicted

    by

    the

    Weibull

    and

    Fictitious

    Crack Models

    73

    Chapter

    4: FIBERS

    AS SHEAR REINFORCEMENT

    IN

    LONGITUDINALLY

    REINFORCED

    BEAMS

    102

    4.1

    Introduction

    102

    4.2

    Advantages

    of

    Fibers

    as

    Shear

    Reinforcement

    104

    4.3

    Experimental

    Program

    107

    4.3.1

    Specimen

    preparation

    107

    4.3.2 Testing

    procedure

    110

    4.4 Observed

    Failure

    Modes

    111

    4.4.1

    Beam-action (a/d

    > 2.5)

    111

    4.4.2

    Arch-action

    (a/d

    < 2.5)

    113

    4.5 Test

    Results

    and

    Discussion

    115

    4.5.1

    First crack

    strength

    115

    4.5.2

    Shear

    span/effective

    depth

    ratio

    118

    4.5.3

    Reinforcement

    ratio

    121

    4.5.4

    Beam depth

    122

    4.6 Simple

    Design

    Formulae

    for

    the

    Shear Strength

    of

    Reinforced

    Mortar

    Beams

    with

    Fibers

    124

    Chapter

    5: CONCLUSION

    144

    Chapter

    6:

    RECOMMENDED

    FUTURE

    RESEARCH

    147

    References

    154

    5

  • 8/18/2019 21768729-MIT

    6/158

    List of

    Figures

    Fig.

    2.1

    Cohesive

    Zone

    Ahead of

    the

    Crack

    Tip

    Fig.

    2.2

    Specimen Configurations

    used

    with J-Integral

    Test

    Fig.

    2.3

    Inside

    of

    Omni-mixer

    Fig.

    2.4

    Specimens

    covered

    in

    Plastic

    just

    after

    Casting

    Fig. 2.5

    Notched

    Beam

    ready

    for

    Testing

    Fig.

    2.6

    Loading

    Machine

    Used for

    J-Integral

    Test

    Program

    Fig.

    2.7

    Average

    Load

    versus

    Load

    Point

    Displacement

    Curves

    Fig.

    2.8

    Average

    Load

    Point

    Displacement

    versus

    Crack

    Opening

    Curves

    Fig.

    2.9

    J-Integral

    versus

    Crack

    Opening

    Curve

    Fig.

    2.10 Deduced

    Tension

    Softening

    Curve

    Fig.

    2.11

    Comparison Between Deduced

    and

    Directly Measured Tension

    Softening

    Curves

    Fig.

    2.12

    Flexural

    Load

    Deflection

    Curve

    Corrected

    to Account

    for

    Energy

    Supplied

    by Beam

    Self-Weight

    Fig.

    2.13

    Tension

    Softening

    Curves

    Deduced

    by

    Indirect

    J-Integral

    Technique

    Fig.

    3.1

    Specimen

    Geometry

    and

    Loading

    Configuration

    for

    Flexural

    Test

    Fig.

    3.2

    Wooden

    Mold for

    228

    mm Deep

    Beam

    Fig.

    3.3

    Beams

    Covered with Plastic

    Just

    After Casting

    Fig.

    3.4

    Cylinders

    in

    Plastic

    Molds

    and

    Covered

    with

    Plastic

    Just

    After

    Casting

    Fig.

    3.5

    228

    mm

    Deep

    Beam

    ready for

    Testing

    Under

    Third

    Point

    Loading

    Fig.

    3.6

    Cylinder

    with

    LVDTs

    on

    either

    side

    ready

    for

    Compression

    Test

    Fig.

    3.7

    Cylinder

    ready for

    Splitting

    Tension

    Test

    Fig.

    3.8

    Effect

    of

    Fiber

    Reinforcement

    on

    Compressive

    Strength

    Fig.

    3.9

    Effect

    of

    Fiber

    Reinforcement

    on the

    First

    Crack

    Splitting

    Tensile

    Strength

    Fig.

    3.10

    Influence

    of

    Beam

    Size on

    Flexural

    Strength

    Fig.

    3.11

    Average

    Normalized

    Flexural

    Strength

    Values

    as

    a

    Function

    of

    Beam

    Depth

    Fig. 3.12

    Theoretical

    Curves

    Relating

    the

    ff/ft

    Ratio

    to

    the d/lch

    Ratio

    Calculated

    Using

    the

    Fictitious

    Crack

    Model

    [101

    6

  • 8/18/2019 21768729-MIT

    7/158

    Fig. 3.13 Empirical Relationship

    Between

    the Flexural/Tensile

    Strength

    Ratio

    and

    the Compressive

    Strength for Plain Concrete [401

    Fig. 3.14 Empirical Relationship Between the

    Flexural/Tensile

    Strength

    Ratio and

    the

    Material

    Characteristic

    Length

    for

    Plain

    Concrete

    [10]

    Fig.

    3.15 Effect of

    Fiber

    Reinforcement on the Ratio Between Flexural

    and Splitting Tensile

    Strengths

    Fig. 3.16

    Typical Tension Softening Curves

    of Mortars

    Reinforced

    with

    various

    Fibers

    Fig. 3.17

    Flexural Load-Deflection

    Curves for Mortars

    Reinforced with

    various Fiber Types

    Fig. 3.18 Flexural Load-Deflection Curves

    for

    Kevlar Fiber

    Reinforced

    Mortar Beams

    Fig.

    3.19

    Flexural Load-Deflection Curves

    for Steel

    Fiber (25 mm)

    Reinforced

    Mortar

    Beams

    Fig. 3.20

    Flexural Stress versus Normalized Deflection

    for

    Different

    Beam Sizes

    of

    Kevlar Fiber

    Reinforced Mortar

    Fig. 3.21 Flexural Stress versus Normalized Deflection

    for

    Different

    Beam Sizes of Steel Fiber Reinforced Mortar

    Fig. 3.22

    Calculation

    of Flexural

    Toughness

    Indices from

    Flexural

    Load-Deflection

    Diagram

    Fig.

    3.23

    Influence

    of

    Beam

    Size

    on

    Flexural

    Toughness

    Indices

    5

    I10'

    130 and

    150

    Fig.

    3.24

    Influence

    of

    Beam

    size

    on

    Flexural

    Toughness

    Index

    Tmax

    Fig.

    3.25

    Influence

    of

    Beam

    Size

    on

    Flexural

    Toughness

    Index

    T

    5m

    Fig.

    3.26

    Influence

    of

    Beam

    Size

    on

    Flexural

    Toughness

    Index

    T10

    Fig.

    3.27

    Various Flexural Toughness

    Indices

    for

    Mortars Reinforced

    with each Fiber Type

    Fig.

    3.28 Semi-Empirical

    Relationship

    Between

    the

    f /ft Ratio

    and

    the

    Flexural

    Toughness

    Index

    Tmax

    Fig.

    3.29 Semi-Empirical

    Relationship Between the

    Flexural Toughness

    Index

    T

    10

    and a

    Parameter

    involving

    the Flexural and

    Tensile

    Strengths

    and the

    Fiber Length

    Fig.

    3.30

    Theoretical

    Curves

    Relating

    the

    f

    /ft

    Ratio

    to

    the

    d/lch

    Ratio Calculated Using the

    Fictitious Crack

    Model

    [38]

    Fig.

    3.31 Comparison of

    Size

    Dependence of Flexural

    Strength

    Predicted

    by

    the Weibull

    and Fictitious Crack

    Models

    7

  • 8/18/2019 21768729-MIT

    8/158

    Fig. 4.1

    Specimen Geometry

    and

    Loading

    Configuration for

    Shear

    Beam

    Test

    Fig. 4.2 Rebars

    Fixed at

    Proper Spacing

    Using Short

    Steel

    Bars

    Fig.

    4.3 Rebars

    Fixed

    in

    Mold

    Ready

    for

    Casting

    Fig.

    4.4

    Shear

    Beam Ready for

    Center

    Point Loading

    Fig.

    4.5

    Loading

    System for

    Large

    Shear

    Beams.

    Bottom

    Beam is just a

    Support

    Fig.

    4.6

    (a) Typical

    Crack

    Shape in

    Plain Mortar

    Beam with

    a/d > 2.5

    (b)

    Typical

    Critical

    Crack

    Shape

    in

    Fiber Reinforced

    Beam

    with a/d

    > 2.5

    Fig. 4.7

    Force

    System in

    Reinforced Concrete

    Beam at

    a

    Diagonal

    Shear

    Crack

    Fig. 4.8

    Failure Patterns for

    Shear

    Beams

    with

    a/d

    =

    3.0

    and

    d =

    204

    mm

    (a)

    =

    Plain

    Mortar

    (c)

    Kevlar

    2%

    (b) =

    25

    mm Steel 1% (d)

    25 mm

    Steel

    2%

    Fig.

    4.9 Shear

    Failure Patterns

    in

    Beams with a/d =

    3.0, d = 204 mm

    and 25 mm

    Steel

    Fiber Reinforcement

    2%

    Fig.

    4.10

    Typical

    Shear Failure

    Patterns

    in

    Beams with

    a/d

    <

    2.5

    (a) Splitting Failure

    (b)

    Shear

    Compression

    Failure

    (c) Flexural

    Tension Failure

    Under Eccentric

    Compression

    Fig.

    4.11

    Typical Splitting Tension

    Like Failures

    of

    Plain Mortar

    Beams with

    a Reinforcement

    Ratio

    of 2.2%

    and

    Loaded with

    a/d

    = 1.0

    [18]

    Fig.

    4.12

    Typical

    Shear Compression

    Like

    Failures of

    Beams Reinforced

    with

    2%

    Acrylic Fibers.

    The Reinforcement

    Ratio

    is

    2.2% and

    a/d

    =

    1.0

    [18].

    Fig.

    4.13 Cracking

    and Ultimate

    Shear

    Strengths of

    various

    Fiber

    Reinforced

    Mortar

    Beams

    Fig.

    4.14 Influence

    of the

    Shear

    Span/Effective

    Depth

    Ratio on

    Ultimate

    Shear

    Strength

    of

    Fiber

    Reinforced Mortar

    Beams

    Fig.

    4.15

    Influence

    of

    the

    Shear

    Span/Effective

    Depth Ratio on

    the

    Maximum

    Bending

    Moment in

    Fiber

    Reinforced

    Mortar

    Beams with

    Longitudinal

    Steel

    Fig.

    4.16

    Influence

    of the

    Longitudinal

    Reinforcement

    Ratio

    on

    the

    Ultimate

    Shear

    Strength

    of

    Fiber

    Reinforced

    Mortar

    Beams

    8

  • 8/18/2019 21768729-MIT

    9/158

    Fig.

    4.17

    Influence

    of

    Beam

    Depth

    on the

    Ultimate

    Shear

    Strength

    of

    Fiber

    Reinforced

    Mortar

    Beams

    Fig.

    4.18

    Semi-Empirical

    Relationship

    Between

    Ultimate

    Shear

    Strength

    and the

    Material

    and

    Geometrical

    Properties

    for

    a/d

    >

    2.5

    Fig.

    4.19

    Semi-Empirical Relationship Between Ultimate

    Shear

    Strength

    and

    the

    Material

    and

    Geometrical

    Properties

    for

    a/d < 2.5

    9

  • 8/18/2019 21768729-MIT

    10/158

    List of

    Tables

    2.1 Summary of J-Integral Test Results

    3.1

    Fiber

    Properties

    3.2

    Flexural,

    Splitting Tensile

    and

    Compressive Strengths

    of

    Mortars Reinforced

    with various Fiber

    Types

    3.3 Flexural Toughness Indices of Mortars Reinforced

    with

    various

    Fiber

    Types

    3.4 Size Dependence

    of

    Flexural Strength Predicted

    by the

    Weibull

    and

    Fictitious

    Crack Models

    4.1

    First Crack and

    Ultimate

    Shear

    Strengths

    4.2

    Flexural,

    Splitting

    Tensile

    and Compressive

    Strengths

    10

    Table

    Table

    Table

    Table

    Table

    Table

    Table

  • 8/18/2019 21768729-MIT

    11/158

    1 INTRODUCTION

    Concrete

    is used as

    a structural

    material

    in

    vast

    quantities

    worldwide.

    Due to its low tensile strength

    and

    susceptibility

    to cracking

    designers

    have conventionally just considered

    its compressive strength and

    have used steel

    reinforcement to resist tensile

    stresses in

    structures.

    However,

    in

    recent years

    the

    inadequacy

    of just

    considering

    compressive

    strength has

    been

    realised. Reinforced

    concrete structures

    usually crack

    before the full service load is applied

    due

    to

    the

    low

    tensile strain

    capacity

    of the concrete.

    These

    cracks

    allow penetration

    of

    chloride

    ions

    and

    other

    environmental

    agents

    into

    the reinforcing

    bars,

    causing

    corrosion.

    Subsequent expansion

    of the

    reinforcing bar

    is followed by

    cracking and spalling

    of the concrete cover. Shrinkage and thermal

    stresses

    can lead to

    significant cracking even before any load is applied

    to

    a structure. Anchorage failure

    of

    reinforcing

    bars

    is due

    to

    cracks

    propagating

    along

    the

    steel-concrete interface.

    Prestressed

    concrete beams

    exhibit web cracking

    and

    surface

    spalling

    due to transverse tensile

    stresses. Reinforced

    concrete

    beams

    crack

    diagonally under principal

    tensile

    stresses

    caused

    by combined

    moment and shear. Even

    a concrete

    cylinder in uniaxial

    compression can

    fail

    due to local

    transverse

    tensile

    forces.

    The

    use

    of

    low volume

    fractions short fibers

    to reinforce plain

    concrete matrices

    has led

    to

    only small changes

    in

    pre-peak stress-strain

    behavior but

    very significant

    changes in the composite's

    resistance to

    crack propagation

    under tensile

    forces.

    The development of

    high strength

    concrete has produced

    a material

    which has

    very

    high

    compressive

    strength

    but

    which

    is perceived

    as being

    even more brittle

    than

    ordinary concrete

    due

    to a less than

    proportional

    increase in its crack

    resistance.

    All the

    11

  • 8/18/2019 21768729-MIT

    12/158

    above

    phenomena

    illustrate

    the

    important

    role

    which

    tensile

    properties

    may

    play

    in

    determining

    the

    structural

    performance

    of

    concrete

    in

    numerous

    applications.

    This

    realization

    has

    led

    to

    dramatically

    increased

    research

    in

    this field

    in

    recent

    years

    (eg.

    NATO

    Advanced

    Research

    Workshop

    on

    Application

    of

    Fracture

    Mechanics

    to

    Cementitious

    Composites,

    Evanston,

    1984;

    International

    Conference

    on

    Fracture

    of

    Concrete

    and

    Rock

    sponsored

    by

    RILEM

    &

    SEM,

    Houston,

    U.S.A.,

    1987;

    American

    Concrete

    Institute

    Symposium

    on

    Applications

    of

    Fracture

    Mechanics

    to

    Concrete

    Structures,

    Seattle,

    U.S.A.,

    1987;

    MRS

    International

    Meeting

    on

    Advanced

    Materials:

    Symposium

    on

    Fracture

    of

    Rock

    and

    Concrete,

    Tokyo,

    Japan,

    1988;

    International

    Conference

    on Fracture

    and

    Damage

    of

    Concrete

    and

    Rock,

    Vienna,

    Austria,

    1988;

    International

    Workshop

    on Fracture

    Toughness

    and

    Fracture

    Energy

    --

    Test

    Methods

    for

    Concrete

    and

    Rock,

    Sendai,

    Japan,

    1988;

    International

    Workshop

    on

    Applications

    of

    Fracture

    Mechanics

    to

    Concrete

    Structures,

    Lulea

    University

    of

    Technology,

    Lulea,

    Sweden,

    June

    28-30

    1989;

    International

    Conference

    on

    Recent

    Developments

    in

    Fiber

    Reinforced

    Cements

    and

    Concretes

    and

    International

    Conference

    on

    Recent

    Developments

    in

    the

    Fracture

    of

    Concrete

    and

    Rock,

    both

    Cardiff,

    Wales,

    1989).

    Since

    concrete

    is

    a

    non-yielding

    brittle

    material

    which

    fractures

    by

    crack

    propagation,

    many

    researchers

    have

    adopted

    a fracture

    mechanics

    approach

    to the

    study

    of

    its

    tensile

    behavior.

    A

    general

    consensus

    has

    been

    reached

    that

    whilst

    linear

    elastic

    fracture

    mechanics

    is not

    directly

    applicable

    to

    concrete

    structures

    due

    to

    the

    relatively

    large

    non-elastic

    zone

    in

    front of

    a growing crack

    [1],

    it is

    possible

    to

    use

    a non-linear

    fracture

    mechanics

    theory

    which

    accounts

    for

    the

    growth

    of

    a

    fracture

    process

    zone

    ahead

    of

    the

    crack

    tip

    [1,2,3,4,5].

    Carpinteri

    [6]

    related

    such

    structural

    phenomena

    as

    the

    size-effect

    and

    the

    transition

    from

    brittle

    to ductile

    behavior

    to an

    energy

    brittleness

    number

    sE

    defined

    by:

    12

  • 8/18/2019 21768729-MIT

    13/158

    (1.1)

    sE

    = GF

    ft

    d

    where

    GF

    is the

    fracture

    energy,

    f t

    is the

    tensile

    strength

    and d

    is the

    beam

    depth.

    Hillerborg

    [1]

    defined

    the

    material characteristic length

    1

    ch

    as:

    1ch

    = GF

    E

    / f

     

    t

    (1.2)

    where

    E

    is the

    modulus

    of

    elasticity,

    for

    similar

    purposes.

    Li and

    Liang

    [4]

    related

    the

    process

    zone

    size

    to

    the

    characteristic

    length.

    They

    showed

    the

    importance

    of

    1

    ch in

    determining

    the

    transition

    from

    brittle

    to

    ductile

    behavior

    and

    the

    limitation

    of

    LEFM

    (linear

    elastic

    fracture

    mechanics)

    and

    strength

    concepts

    in

    their

    application

    to

    grain

    brittle

    materials

    such as

    concrete

    and

    FRC

    (fiber

    reinforced

    concrete).

    However

    two

    major

    problems

    blocking

    the

    widespread

    use

    of

    fracture

    mechanics

    ideas

    in

    concrete

    design

    codes

    is

    first

    the

    lack

    of

    a standard

    test

    method

    to

    characterize

    and

    measure

    fracture

    resistance

    and

    secondly

    the

    lack

    of

    adequate

    experimental

    data

    relating

    changes

    in structural

    behavior

    directly

    to

    changes

    in

    material

    fracture

    resistance.

    It

    is

    the

    purpose

    of this

    thesis

    to

    address

    both

    of

    these

    problems.

    Concrete

    structural

    members,

    unlike

    their

    counterparts

    in

    steel,

    are

    more

    often

    than

    not

    made in

    situ,

    and

    their

    quality

    is

    almost

    exclusively

    dependent

    on the

    workmanship

    of

    concrete

    making

    and

    placing

    and

    on

    the

    curing conditions which

    exist

    while

    the

    concrete gradually

    gains

    strength.

    The

    exact

    influence

    of

    each

    of

    these

    factors

    on

    the

    quality

    of

    the

    final

    product

    is

    usually

    not

    very

    well

    understood.

    Also,

    in

    many

    cases,

    when

    we

    measure

    some

    property

    of

    concrete

    we

    find

    it

    very

    difficult

    to

    determine

    exactly

    how

    this

    property

    affects

    the

    behavior

    of

    concrete

    in

    various

    structural

    applications.

    For

    these

    reasons

    much

    of

    the

    concrete

    design

    13

  • 8/18/2019 21768729-MIT

    14/158

    code

    is

    based

    purely

    on

    empirical

    data.

    We

    make

    a

    specimen

    and test

    it and

    we

    hope that

    if

    we build

    a

    somewhat

    similar

    structure

    in the

    field

    its

    behavior

    under

    load

    will

    not

    differ

    greatly.

    The

    availability

    of

    a

    huge

    base of

    experimental

    data

    together

    with

    the

    use

    of

    conservative

    safety

    factors

    allows

    us

    to

    design.

    However

    modern

    trends

    towards

    structures

    which

    are

    novel

    and

    more

    complicated

    and

    also

    subjected

    to

    more

    severe

    loading

    conditions

    and

    environments

    dictate

    the

    need

    for

    a

    design

    code

    which

    has

    a

    greater

    basis

    in

    understanding

    how

    and

    why

    concrete

    structures

    behave

    as

    they

    do

    and

    how

    individual

    material

    characteristics

    influence

    structural

    behavior.

    Concrete structural members

    are

    increasing

    in

    size due

    to

    advances

    in

    materials

    and

    improvements

    in

    design

    and

    construction

    techniques

    [7].

    It

    is

    well

    known

    that

    larger

    size

    structures

    fail

    in

    a more

    brittle

    and

    'fracture'

    like

    mode.

    Presently

    the

    ACI

    code

    does

    not

    account

    for

    size-effect

    in

    the

    ultimate

    shear

    strength

    of

    longitudinally

    reinforced

    concrete

    beams.

    This

    may mean

    that

    the

    code

    is

    unconservative

    for

    large

    beam

    sizes.

    Another

    trend

    in

    the

    construction

    industry

    is

    the

    gradual

    adoption

    of

    high

    strength

    concrete

    as

    a

    construction

    material.

    LeMessurier

    [8]

    suggested

    that

    very

    tall

    buildings

    (say,

    height

    =

    800

    m)

    could

    be

    economically

    constructed

    by

    transferring

    all

    the

    loads

    to

    very

    large

    reinforced

    concrete

    columns

    around

    the

    perimeter

    of

    the

    building.

    Sidesway

    due

    to

    wind

    loading

    governs

    the

    design

    of

    such

    tall

    buildings

    and this

    would

    be

    inversely

    proportional

    to

    the

    concrete

    modulus.

    Thus

    high

    strength concrete with

    high

    modulus

    is

    ideally

    suited for

    this

    application.

    However,

    there

    is

    relatively

    little

    experimental

    data

    available

    relating

    to

    behavior

    in

    large

    structures

    of

    a material

    which

    is probably

    much

    more

    brittle

    than

    plain

    concrete.

    The

    development

    of

    fiber

    reinforced

    concrete

    over

    the

    past

    two

    decades

    or

    so

    has

    ignited

    the

    dream

    of

    a non-brittle

    cement

    based

    material.

    It

    has

    also

    presented

    a

    great

    challange

    to

    14

  • 8/18/2019 21768729-MIT

    15/158

    researchers

    and designers

    alike. Due

    to

    the huge variety

    of fibers

    available,

    steel

    and synthetic,

    long

    and short, straight,

    crimped,

    hooked,

    deformed,

    etc., and the option

    of using various volume fractions

    up to

    about

    2 ,

    or of

    combining different

    fiber types

    in

    a single

    mix,

    a

    whole

    new dimension

    of

    cement based

    materials

    have been

    created. The

    possibility

    of

    overcoming

    the extra brittleness

    of high strength concrete

    due to

    aggregate fracture

    by providing

    stiff

    fibers which slip

    out of

    the matrix

    may allow us to use

    similar design rules for high strength

    concrete as

    ordinary strength concrete

    if we

    first agree

    to reduce

    the

    brittleness

    with

    some fiber

    reinforcement.

    It will

    not be possible to

    have

    experimental data for all possible

    material and structural variations

    which we will want to

    use.

    We

    do not

    want

    our ability to creatively

    use new

    materials and

    new

    structural

    forms

    to be

    drastically limited

    by our lack of

    understanding

    of

    basic

    structural

    behavior. Fracture

    Mechanics ideas

    are very

    useful for

    explaining

    certain

    trends in

    the

    structural

    behavior

    of

    a

    brittle material such as

    concrete.

    It predicts the

    size-effect

    in

    structural

    strength. It can

    also

    explain

    why

    the

    strength

    of

    members,

    made

    with

    high strength concrete,

    in which

    the

    concrete

    is subjected to

    tensile stresses such as the diagonal tensile

    stresses

    in

    a reinforced concrete beam,

    may increase at

    a lesser rate

    than

    the compressive strength. Hawkins [9] argued in favor of

    fracture

    mechanics

    playing

    an

    important role in

    a revised reinforced concrete design

    code.

    Fracture mechanics

    ideas

    may allow

    us

    to understand the behavior

    of

    a wide

    range of new

    materials

    and

    structural

    forms

    with

    only

    a

    limited

    number

    of

    experimental

    tests.

    Economic considerations

    often play a powerful role

    in shaping our

    design

    procedure. Designers

    will

    not use

    a

    complicated

    design

    procedure

    which can

    save the client

    say

    1%

    due

    to reduced material

    costs

    but adds 2%

    to

    his

    costs

    through greater

    design effort.

    Many designers

    and concrete

    15

  • 8/18/2019 21768729-MIT

    16/158

    producers do

    not believe

    that the

    advantages

    of

    considering

    a fracture

    resistance

    material

    property

    outweigh the disadvantages

    associated

    with

    more troublesome design

    procedures.

    At the outset it was the hope of

    the

    work

    presented

    in

    this

    thesis

    to find

    a

    way

    of

    producing conclusive

    evidence

    of a link between material fracture

    resistance and structural

    performance

    and also to develop semi-empirical

    design formulae

    which

    have

    a

    rational

    basis

    in material

    properties and at the

    same

    time

    represent

    experimental data reasonably well.

    It was hoped

    that

    these formulae could

    take

    account of

    material

    fracture resistance

    in

    a simple and

    efficient

    manner which

    would be

    acceptable

    to both researchers, designers and

    builders alike.

    Initially

    the

    tension

    softening curve

    is presented as

    a

    material

    property

    which characterizes fracture

    resistance.

    A

    relatively simple

    experimental procedure

    is then proposed

    to measure

    this

    curve.

    Using

    this

    procedure the

    tension softening

    curves

    of

    mortars

    reinforced with

    various

    fibers

    is measured. In

    chapters 3

    and

    4

    the

    fracture resistance

    is

    manipulated

    through fiber reinforcement

    and

    the

    effects

    on

    structural

    performance

    are

    examined.

    Tests

    on unreinforced

    beams failing in

    flexure

    and longitudinally

    reinforced

    beams

    failing

    due to diagonal

    tension cracks

    establish

    conclusive

    experimental

    evidence

    of

    the

    link between

    greater

    fracture

    resistance

    and better

    structural

    performance.

    Simple

    semi-empirical

    formulae

    are developed

    which relate fracture

    resistance

    to

    structural

    behavior

    in a manner

    which

    is suitable for the

    purposes

    of

    design

    and

    quality

    control.

    16

  • 8/18/2019 21768729-MIT

    17/158

    2

    TENSION

    SOFTENING

    CURVE

    BY

    J-INTEGRAL TECHNIQUE

    2.1

    Introduction

    The

    fracture

    toughness

    KIc,

    used in linear elastic

    fracture mechanics,

    cannot

    be

    used

    for

    analysis of

    most

    concrete

    structures. This is because

    of

    the size-dependence

    of

    toughness values measured using

    laboratory

    size

    specimens, caused by the

    large

    process

    zone

    size in front

    of the

    crack tip

    relative

    to

    the

    specimen dimensions. Hillerborg

    [1] showed that 3-point

    bend concrete specimens would

    need to

    be

    at least

    1

    - 2 m in depth to

    give

    a valid

    KIc measure. Thus

    it

    is

    necessary

    to

    look outside

    LEFM

    to

    obtain a

    fracture

    resistance

    parameter

    which

    is

    a

    true

    material property. To

    do

    this

    it is necessary

    to

    appreciate

    the physical processes

    which lead to

    fracture in

    a non-yielding

    material like

    concrete.

    When

    loaded

    in

    tension

    the

    material

    follows a stress-strain curve

    which is

    almost

    linear

    initially and then

    becomes increasingly

    non-linear

    due to

    distributed microcracking

    throughout

    the

    material volume. When the

    maximum

    load

    is

    reached cracking

    concentrates on

    a

    narrow

    fracture zone

    across the

    material leading

    to the development of

    a macrocrack. If the

    specimen

    is

    loaded

    in a displacement

    controlled

    machine

    the load gradually

    drops

    as the

    macrocrack

    width increases.

    Energy is absorbed

    across

    the

    fracture

    plane by

    microcracking in

    the cement

    paste and the

    cement-aggregate

    interface

    as

    well as by

    fiber

    pull-out

    or

    fiber breakage

    where

    applicable. The

    constitutive relationship

    between the

    tensile

    stress

    transferred

    across

    a

    crack plane and the

    separation

    distance of

    the

    crack

    faces

    is a material property

    and

    is

    generally

    referred to

    as the

    tension

    softening

    (a-6)

    curve. A

    number

    of material

    parameters

    may be calculated

    from this

    curve.

    The

    maximum

    stress

    is

    the

    tensile

    strength,

    the

    crack

    separation

    at

    which

    the stress

    transferred

    drops

    to

    zero

    is

    the

    critical

    17

  • 8/18/2019 21768729-MIT

    18/158

    crack

    separation

    Sc,

    and

    the

    total

    area

    under

    the

    curve

    represents

    the

    energy

    required

    to

    form

    a

    unit

    area

    traction

    free

    crack

    and

    is called

    the

    fracture

    energy.

    The a-6

    relationship

    is

    a basic

    material

    property

    which

    characterizes fracture resistance

    in

    a

    material

    which

    exhibits

    tension

    softening

    behavior

    and

    can

    be

    used

    for

    numerical

    simulations

    of

    crack

    formation

    and

    propagation

    in

    structures

    constructed

    from

    such

    materials

    [1,4].

    The

    a-S

    curve

    can

    be

    obtained

    from

    a displacement

    controlled

    direct

    tension

    test.

    However

    an

    inherent

    difficulty

    with

    this

    test

    is the

    stability

    of

    loading

    the

    specimen

    during

    the

    softening

    process.

    Usually

    a

    very

    stiff

    testing

    machine,

    which

    is

    not

    available

    in

    most laboratories,

    is

    required

    to obtain

    a

    completely

    stable

    load-deformation

    curve.

    Successful

    tests

    have

    been

    performed,

    using

    mechanisms

    such

    as

    parallel

    steel

    bars

    in

    the

    direction

    of

    loading

    and

    closed

    loop

    feedback

    systems,

    by

    Petersson

    [10],

    Gopalaratnam

    and

    Shah

    [11]

    and

    Reinhardt

    [5].

    However

    it

    is unlikely

    that

    this

    will

    become

    a standard

    test

    method

    due

    to the

    need for

    these

    intricate

    modifications

    of

    the

    loading

    machine.

    This

    chapter

    focuses

    on

    an

    indirect

    J-integral

    technique,

    first

    proposed

    by

    Li [12],

    to experimentally

    determine

    the

    a-S

    curve.

    This

    test

    procedure

    is

    simple

    and

    requires

    only

    a simple

    testing

    machine.

    It

    is

    believed

    that

    any

    standard

    test

    procedure

    must

    possess

    these

    characteristics.

    2.2

    Theoretical

    Basis

    of

    J-Integral Technique

    The

    path

    independent

    J-integral

    is

    defined

    as:

    j

    r (Wdy

    -

    T @u/ax

    ds)

    2.1)

    18

  • 8/18/2019 21768729-MIT

    19/158

    where

    r

    is

    a

    curve

    surrounding

    the

    notch

    tip,

    W

    is the

    strain

    energy

    density,

    T is

    the

    traction

    vector

    in the

    direction

    of

    the

    outward

    normal

    along

    r

    u

    is the

    displacement

    vector

    and

    ds

    is

    an arc

    along

    r.

    From

    Eq.

    (2.1)

    Rice

    [13,14]

    produced

    two

    alternative

    definitions

    of

    J.

    He

    used

    the

    Barrenblatt

    approach

    which

    considers

    a cohesive

    zone

    ahead

    of

    the

    crack

    tip

    in

    which

    the

    restraining

    stress

    a(S)

    is

    viewed

    as

    a

    function

    of

    separation

    S.

    If

    the

    J-integral

    is

    evaluated

    along

    a contour

    rl,

    shown

    in

    Fig.

    2.1,

    which

    runs

    along

    beside

    the

    cohesive

    zone,

    then

    we get:

    J

    = f

    a(S)

    (dS/dx)dx

    (2.2)

    cohesive

    zone

    This

    definition

    may

    be

    interpreted

    as

    follows.

    If

    the

    crack

    opening

    at

    each

    point

    in the

    cohesive

    zone

    increases

    by

    an

    amount

    dS

    then

    the

    profile

    of

    the

    cohesive

    zone

    boundary

    extends

    a

    distance

    dx.

    The

    quantity

    a(S)dx

    is

    the

    force

    over

    each

    infinitesimal

    area

    and

    a(S)

    dx

    dS

    is the

    energy

    absorbed

    during

    increased

    separation

    dS.

    Thus

    Eq. (2.2)

    defines

    J

    as

    the

    rate

    of

    energy

    absorption

    with

    respect

    to

    cohesive

    zone

    propagation.

    Eq.

    (2.2)

    may

    also be

    expressed

    as:

    J

    =

    fS

    a(S)

    dS

    (2.3)

    0

    where

    St

    is

    the

    separation

    distance

    at

    the crack

    tip.

    When

    St reaches

    c

    the

    real

    crack

    propatates

    and

    a

    critical

    J-integral

    value,

    J

    is reached:

    j

    =

    f

    cr(8)

    dS

    (2.4)

    S0

    Jc

    is

    equal

    to

    the

    total

    area

    under

    the

    a-8

    curve

    and

    may

    be

    interpreted

    as

    the

    rate

    of

    energy

    absorption

    in

    the

    cohesive

    zone

    with

    respect

    to

    crack

    tip

    propagation.

    19

  • 8/18/2019 21768729-MIT

    20/158

    The

    second

    interpretation

    of

    J

    may be

    given

    as:

    J =

    - [a(PE)/aa]

    (2.5)

    where

    PE

    is the

    potential

    energy

    of

    a body

    with crack

    length

    a.

    Thus

    J

    is

    equal

    to the

    rate

    at

    which

    the

    potential

    energy

    of

    a

    cracked specimen

    decreases

    as the

    crack

    propagates.

    The

    basis of

    the

    indirect

    J-integral

    technique

    of

    finding

    the

    a-S

    relationship

    is

    to find

    J

    experimentally

    using

    Eq.

    (2.5)

    and

    then

    to

    substitute

    into Eq.

    (2.3)

    and

    find

    a(S).

    Potential

    energy

    may be

    calculated simply

    from

    a load-displacement

    curve. However, since

    the

    crack

    tip

    position

    is difficult

    to

    locate

    accurately,

    it would

    be

    almost

    impossible

    to

    directly

    evaluate

    Eq.

    (2.5)

    by

    propagating

    a

    crack

    in

    a

    single

    specimen.

    One

    approximate

    procedure

    for

    getting

    around

    this

    problem

    is

    to

    use two

    cracked

    specimens

    identical

    in every

    respect

    except

    that

    there

    is

    a

    slight

    difference

    in

    their

    initial

    crack

    lengths.

    If

    the

    load-load

    point

    displacement

    (P-a)

    curves

    are

    measured

    for

    each

    specimen,

    then

    the

    area

    A(A)

    between

    the two

    curves

    up

    to a load

    point

    displacement

    A

    represents

    the

    difference

    in

    energy

    and Eq.

    (2.5)

    may be

    interpreted

    as:

    J(A)

    = A(A)

    /

    bia

    (2.6)

    where

    Aa

    is

    the

    difference

    in

    crack

    lengths

    and

    b

    is the

    ligament

    width.

    If,

    during

    the experiment,

    the

    crack

    tip

    separation

    8,

    is

    also

    measured

    then

    it

    is possible

    using

    the

    A-8

    relationship

    to convert

    J(A)

    to

    J(8).

    Differentiation

    of

    Eq.

    (2.3)

    then

    gives:

    a 8)=

    J(8)/98

    2.7)

    20

  • 8/18/2019 21768729-MIT

    21/158

    and

    the tension

    softening

    curve

    may

    be

    determined from

    the slope

    of

    the

    J(8)

    curve.

    2.3

    Numerical

    Verification

    of Test Technique.

    This

    method has

    been

    verified numerically

    for

    both

    beam bending and

    compact tension configurations.

    A. Hillerborg (private communications,

    1985) provided verification by employing his

    fictitious crack

    model in a

    finite element

    scheme to simulate

    the load-load

    point

    displacement curves

    and load-crack tip separation curves

    of

    a pair

    of

    three-point

    bend

    specimens

    of

    slightly different

    crack lengths. He used an

    artificial

    bi-linear curve

    as input for the

    tension

    softening behavior

    in the

    material

    ahead

    of

    the crack

    tips.

    The objective

    of

    the

    exercise was to

    extract

    the

    same curve using

    the

    indirect

    J-integral

    technique

    with his numerically

    derived

    test

    results. The extracted

    curve essentially

    overlapped the

    initial

    assumed curve,

    thus

    verifying the theoretical

    basis.

    Reyes

    [15]

    used

    a boundary

    element method to

    carry

    out

    a similar procedure with a

    compact tension

    configuration. Again the input

    curve and

    the

    extracted

    curve showed

    excellent

    agreement.

    The theoretical basis

    and numerical verification

    confirm that this

    test

    technique

    is independent of specimen

    geometry

    and

    should

    also

    be

    independent of

    specimen

    size. The

    only

    restriction on specimen

    size

    is

    that the smallest

    specimen dimensions

    should be a number of times

    larger

    (maybe

    four or five times) than

    the

    largest

    single particles

    in the

    material.

    Thus

    minimum

    dimensions depend

    on

    material properties

    such as

    aggregate size

    and fiber length.

    21

  • 8/18/2019 21768729-MIT

    22/158

    2.4 Experimental

    Procedure

    This

    J-integral

    technique

    has

    been

    used

    by

    Li and

    co-workers

    [12,16,17,18,19]

    with

    both

    the

    compact

    tension

    and

    four-point

    bend

    beam

    loading

    configurations

    illustrated

    in

    Fig.

    2.2.

    Hashida

    [20]

    also

    used

    a

    compact

    tension

    configuration

    and

    Chong

    et al

    [21]

    used this

    technique

    on

    semi-circular

    specimens

    from

    rock

    cores.

    Because

    the

    experimental

    procedure

    is very

    similar

    irrespective

    of

    the

    specimen

    configuration

    or

    the

    material

    type,

    a

    thorough

    description

    of

    specimen

    preparation,

    testing

    procedure,

    data

    analysis

    and

    results

    will

    be

    given

    for

    just

    one

    geometry

    and

    one

    material.

    The

    geometry

    used

    is

    the

    beam shown

    in

    Fig 2.2 and

    the

    material

    is

    a Kevlar

    fiber

    reinforced

    mortar.

    2.4.1 Specimen

    preparation

    The

    following

    materials

    were

    used:

    (i)

    Type

    III

    rapid

    hardening

    portland

    cement

    (ii)

    Sand

    passed

    through

    a #8

    sieve

    (iii)

    Kevlar

    fibers

    with

    length

    =

    6.4

    mm,

    diameter

    =

    12

    pm,

    modulus

    =

    130

    GPa,

    tensile

    strength

    =

    2.8

    GPa,

    and

    density

    =

    1.45

    g/cc.

    (iv)

    A

    high

    range

    water

    reducing

    admixture

    named

    Daracem-100

    and

    classi-

    fied

    as ASTM

    C-494

    Type A.

    The

    cement:sand:water

    ratio

    was

    1:1:0.5

    by

    weight.

    The volume

    fraction

    of

    fibers

    was

    2%

    with

    a superplasticiser

    volume

    fraction

    of

    0.75%.

    An

    Omni-mixer

    in which

    random

    movement

    of

    the

    particles

    is

    induced

    to

    occur

    by

    a

    wobbling

    flexible

    drum

    bottom

    was

    used.

    A

    photograph

    of the

    inside

    of

    the

    mixer

    is

    shown

    in

    Fig.

    2.3.

    The

    cement,

    fibers

    and

    sand

    were

    22

  • 8/18/2019 21768729-MIT

    23/158

    added

    to

    the mixer in that order. The plasticiser was mixed with

    the water

    and

    added

    last.

    Mixing was

    then

    carried

    out

    for

    approximately

    3

    minutes.

    Some preliminary trial mixing suggested that there was

    no

    advantage to be

    gained

    by

    dry mixing the cement, fibers and sand

    before

    adding water.

    Plexiglass

    molds

    (63.5

    x

    114

    x

    432

    mm)

    were used. Casting

    was done in a

    direction such that

    the

    63.5

    mm side

    was vertical and

    thus

    all

    specimens

    were rotated

    through

    900 for testing.

    Compaction was achieved using

    a

    tamping

    rod and

    a table vibrator.

    For synthetic fiber

    mortar tamping is

    probably

    much

    more effective than

    vibration.

    It

    proved relatively

    difficult

    to achieve a low

    air content

    comparable

    to

    that

    which

    could be

    achieved with

    just plain mortar.

    After casting

    the

    specimens were covered

    with

    plastic and left

    in the

    molds

    for

    approximately 16

    hours. Fig. 2.4

    shows

    some

    specimens

    just

    after

    casting.

    They

    were then

    removed

    from the

    molds and

    placed in water

    at 22 0C. After 13

    days

    the specimens

    were

    removed from

    the water and

    all the

    notches

    were

    cut

    using a 1.5 mm thick

    diamond blade.

    Two beams had a through the

    thickness

    notch,

    57.15 mm deep

    on the

    tension

    side and two beams had a similar notch, 63.5

    mm

    deep. Crack

    guides,

    6.35 mm

    deep

    were

    also

    cut on

    each vertical side of the

    specimen,

    extending from

    the

    tip of the notch

    to

    the top of

    the

    beam.

    This left a

    ligament

    50.8

    mm wide and either 56.85

    or 50.5

    mm

    deep,

    depending on

    the

    beam, at

    the center of the span.

    All the

    specimens were

    then

    left in

    air

    until testing the following day

    at

    14 days

    of

    age.

    Some preliminary

    experimental work

    showed

    that

    cast-in-notches

    could

    be used instead

    of

    saw-cut notches

    if

    a suitable

    saw

    is

    not

    available.

    Thin

    aluminium plates (0.8 mm

    thick)

    were placed in

    the mold before

    casting

    and

    coated with mold

    release. They

    were easily

    removed from the specimen

    about 10

    hours after casting.

    This

    technique

    assures

    very

    accurate

    notch

    length.

    No

    evidence of

    fiber bundling

    at the

    plate tip was found.

    Either

    notching

    technique

    can

    be used

    depending on

    preference

    and available

    23

  • 8/18/2019 21768729-MIT

    24/158

    facilities.

    2.4.2

    Testing

    procedure

    Figs. 2.5 and 2.6 are

    photographs

    of the

    experimental

    set

    up.

    The

    beam was

    supported on a roller

    at one end and a ball

    at the other. The

    load

    was

    applied via a universal

    joint onto

    a

    steel plate

    which was

    supported

    on the

    top of the

    beam

    by a

    roller

    at one

    side

    and

    a

    ball at the

    other. A displacement controlled

    Instron loading

    machine

    was used with

    load being

    applied

    by

    a

    screw

    mechanism.

    LVDTs (Linear

    Variable

    Displacement Transducers) with a displacement measuring capacity

    of

    5.1

    mm

    were used to

    measure the load point displacement and

    the

    crack opening.

    One

    LVDT,

    placed at the front of

    the beam measured

    the

    average load point

    displacement. A

    second LVDT was placed

    on the front of the

    beam

    at the

    notch tip

    level to measure the crack tip opening.

    The third LVDT was

    placed

    on the

    back

    of the beam,

    63.5 mm from

    the

    bottom, corresponding to

    the

    position

    of the

    notch tip in beams with

    a long

    notch and

    being 6.35 mm

    (equal to Aa) above the notch tip in beams with a short notch. The LVDT

    holder

    and

    its target

    were

    placed

    14 mm apart (inside to

    inside)

    on

    opposite

    sides

    of the notched plane. Fig.

    2.5 shows the positions

    of

    the

    two

    LVDTs

    on

    the front of the

    beam.

    The horizontally

    placed

    LVDTs have

    the

    shaft spring loaded

    against

    the target to ensure

    permanent contact when

    the

    crack opens.

    The vertical LVDT

    shaft stayed

    in

    contact

    with the target

    under gravity.

    The

    cross-head

    speed was

    initially set

    at

    0.0254 mm/min. and

    was

    changed

    to

    0.0508 mm/min.

    after

    the load

    dropped

    to

    30% of its

    peak

    value.

    It

    was later changed

    to

    0.127

    mm/min. to

    ensure

    a reasonable test time.

    Tests

    were continued

    until

    the

    beams failed

    under self-weight

    with

    total

    test time

    being

    about 50 minutes.

    The data

    was

    passed

    through a

    fluke

    data

    24

  • 8/18/2019 21768729-MIT

    25/158

    acquisition

    system

    and

    stored

    on

    an

    IBM

    PC.

    A

    Basic

    Program

    was

    used

    to

    convert

    the

    output

    voltage

    readings

    into

    load

    and

    displacement

    values.

    2.5

    Results

    and

    Data

    Analysis

    Fig.

    2.7

    shows

    the

    average

    load

    versus

    load

    point

    displacement

    curves

    for

    each notch

    length.

    Each

    average

    curve

    was calculated

    by

    taking

    a

    displacement

    value

    and

    calculating

    the

    average

    load

    corresponding

    to this

    displacement

    for

    a set

    of

    beams

    with

    the

    same

    notch

    length.

    Fig.

    2.8

    shows

    average

    crack

    opening

    values

    61,

    62

    and

    63

    expressed

    as

    a function

    of

    the

    load point

    displacement.

    S

    is

    the

    average crack opening

    at

    the

    notch

    tip

    in the

    short

    notched

    beams,

    62

    is

    the

    average

    crack

    opening

    at the

    notch

    tip

    in

    the

    long

    notched

    beams

    and

    63

    is

    the

    average

    crack

    opening

    in

    the

    short

    notched

    beams at

    a position

    6.35

    mm

    directly

    above

    the

    notch

    tip.

    The

    average

    crack

    opening

    value

    6 was

    defined

    as:

    6

    =

    (6

    1

    +

    63

    )/2

    (2.8)

    Some

    previous

    experimental

    work [12,16,17,18]

    had

    used 6

    =

    (8i

    + 62)

    /2

    as

    the

    definition

    of

    the

    average

    crack

    opening.

    However,

    since

    the

    uncracked

    ligament

    Aa,

    representing

    the

    difference

    between

    the

    short

    and

    long

    notched

    beam,

    exists

    in

    the

    short

    notched

    beam,

    it

    is

    considered

    more

    accurate

    to

    define

    the

    average

    crack

    opening

    in

    this

    ligament

    by

    measuring

    the

    crack opening

    at

    each end (61 and

    63)

    and

    taking

    the

    average

    of

    these

    two values.

    From

    Fig.

    2.8

    it can be deduced

    that using

    63 instead

    of 62

    produces

    an

    average

    6

    value

    which

    initially

    increases

    slower

    and

    later

    increases

    faster

    with

    respect

    to

    the

    6 values.

    Eqn.

    (2.7)

    shows

    that

    this

    produces

    a

    tension

    softening

    curve

    which

    is

    higher

    initially

    and

    becomes

    lower

    after

    some

    crack

    opening.

    It

    will

    be

    seen

    later

    that

    this

    produces

    25

  • 8/18/2019 21768729-MIT

    26/158

    better

    correlation

    between the

    deduced

    tension softening

    curve

    and

    the

    softening

    curve measured

    by a direct

    tension

    test.

    Numerical integration was

    used to calculate

    A(A)

    which

    was

    then used

    with Eqn. (2.6) to calculate

    J(A).

    Using

    the S(a) relationship

    obtained

    from Fig. 2.8

    and Eqn. (2.8), the J(6)

    curve was calculated and

    is

    illustrated

    in

    Fig.

    2.9.

    Numerical

    differentiation

    of

    the

    J-6

    curve

    in

    accordance

    with

    Eqn.

    (2.7) was

    achieved using

    Taylor expansions at

    five

    consecutive

    points J(S-g), J(S-h),

    J(6), J(S+j), J(S+k)

    and solving for

    J'().

    Fig. 2.10

    shows the

    deduced tension

    softening

    curve.

    It shows an

    initial rise from a =0 to a = ft

    before descending back down

    to

    a 0.

    This

    initial

    ascending

    part may be

    interpreted

    as

    the sum of

    recoverable

    elastic

    deformation,

    and distributed

    inelastic

    deformation in

    the form of

    microcracking,

    which occurs between the

    LVDT holder and

    its

    target,

    prior

    to the localization

    of

    inelastic

    deformation

    onto the fracture

    process

    zone.

    The dashed line

    in Fig. 2.10

    shows

    the

    unloading

    path

    of

    the

    material

    in this zone.

    It is clear that

    the area

    between

    the ascending

    path and the

    dashed line

    represents

    inelastic

    energy absorbed

    outside

    the

    fracture

    plane and thus

    should not

    be counted

    as

    part of

    the

    tension

    softening

    curve.

    Fig.

    2.11

    shows

    the

    corrected

    a-8

    curve,

    accounting

    for

    this

    energy loss

    away from

    the

    fracture zone,

    together with a

    tension

    softening

    curve

    for

    the

    same material

    measured by

    Wang

    [22]

    using

    a direct

    tension

    test.

    Considering

    the

    limited number

    of

    tests

    carried out

    (just two beams

    for each

    notch

    length)

    the

    agreement

    between

    the

    deduced softening

    curve

    and

    the

    directly measured

    curve is

    reasonably good.

    The deduced

    tensile

    strength

    is 3.35

    MPa

    compared

    to

    a

    directly

    measured

    value of

    3.95

    MPa.

    This

    underestimation

    is

    consistent

    with

    previous

    test

    results

    and some

    explanation for

    this

    trend

    will be

    presented

    in

    Section 2.7.

    The critical

    J-integral

    value

    Jc

    , defined as

    the

    total

    area

    under

    the

    deduced a-S

    curve,

    26

  • 8/18/2019 21768729-MIT

    27/158

    is 1485

    N/m

    compared

    to

    a

    directly

    measured

    value of

    1250 N/m.

    This

    error

    can

    be

    attributed

    to the

    small

    number

    of tests

    and

    would probably

    be

    eliminated

    by

    carrying

    out

    four or

    five

    tests

    for

    each

    notch

    length.

    2.6

    Calculation

    of

    GF

    Fracture

    energy

    was

    also calculated

    using

    a method

    similar

    to

    that

    recommended

    by

    RILEM

    Technical

    Committee

    50

    [23].

    The RILEM method

    is

    based

    on

    the total

    area under

    the

    load-load

    point

    displacement

    curve

    for

    a

    three-point

    bend

    test

    on

    a

    notched beam.

    However

    the

    results

    should

    be

    similar for

    the

    four-point

    bend

    geometry.

    The fracture

    energy GF

    is

    given

    as:

    GF

    = A

    /

    b(d-a)

    (2.9)

    where

    A

    is

    the

    total

    area

    under the

    load

    versus

    load

    point

    displacement

    curve,

    including

    a

    correction

    for

    beam

    self-weight.

    The method

    of

    estimating

    energy

    supplied

    by

    beam

    self-weight

    was

    as proposed

    by

    Petersson

    [10]

    and

    is

    illustrated

    in

    Fig.

    2.12.

    The calculated

    GF

    values

    were

    1181

    N/m

    and

    1143

    N/m

    for

    the

    short

    notched

    and

    long notched

    beams

    respectively.

    These

    values

    are

    less than

    that

    calculated

    from

    the

    direct

    tension

    test

    and

    this

    finding

    is

    the

    opposite

    of

    test

    results

    obtained

    by

    Horvath

    and

    Persson

    [241

    which

    showed

    GF

    values

    to be

    about

    20% greater

    than

    direct

    tension results.

    Their

    tests were

    on

    plain

    concrete

    specimens whose

    load-deflection

    curves

    have

    a

    relatively

    short

    tail.

    The self-weight

    correction

    may

    not

    be

    very

    accurate

    for

    fiber

    reinforced

    concrete

    curves

    which

    have

    a

    much

    longer

    tail.

    This

    may

    account

    for

    some

    of

    the

    above

    discrepancy.

    It

    should

    be

    remembered

    that

    GF

    values

    are

    usually

    expected

    to

    27

  • 8/18/2019 21768729-MIT

    28/158

    overestimate

    the true

    fracture

    energy because of

    inelastic

    energy absorbed

    in

    the

    material

    which lies

    close to the

    fracture

    zone and

    also in the

    material

    close to the

    supports

    and loading

    points

    where

    concentrated

    loads

    are applied.

    Using

    a relatively

    long notch

    length

    helps to

    reduce this

    inelastic

    energy

    absorption.

    Another

    interesting

    point

    concerning

    the

    GF

    calculation

    method

    is

    that as

    the

    beam

    size is

    increased,

    while the

    geometry

    is kept constant,

    the

    loads on

    the

    beam before the

    maximum load

    is

    reached

    increase

    approximately

    in proportion

    to the

    beam

    volume,

    and thus

    inelastic

    energy absorbed

    away

    from the

    fracture

    zone

    increases at a

    similar rate.

    However

    the fracture

    energy

    absorbed, only

    increases in

    proportion

    to

    the cross-sectional

    size. Thus

    as

    beam

    size increases,

    the

    ratio

    of

    energy

    absorbed outside

    the fracture

    zone to that

    absorbed

    within,

    increases,

    leading

    to greater

    overestimation

    by

    GF

    of

    the true

    fracture

    energy.

    Horvath and

    Persson

    [24]

    found calculated GF

    values to increase

    by

    as much

    as 50%

    when

    1455

    x 300

    x

    100

    mm beams

    were

    used

    instead of

    840

    x

    100

    x 100

    mm beams.

    Thus the

    GF fracture

    energy

    value

    cannot

    be considered

    a

    material

    property,

    but

    is dependent

    on

    specimen

    geometry.

    GF should

    be

    closer

    to

    the

    true

    fracture

    energy

    value when

    small size specimens

    are

    used.

    However

    specimen

    dimensions

    should be

    at least

    3 or

    4 times

    larger

    than the

    largest

    particle

    size in

    the material

    such as

    aggregate

    diameter

    or

    fiber length.

    2.7 Discussion

    and

    Recommendations

    for the

    J-Integral

    Test

    A

    summary

    of all

    test

    results

    obtained

    by Li

    and

    co-workers

    [12,16,17,18,19]

    is

    given

    in

    Table

    2.1 and

    the

    deduced

    tension

    softening

    curves

    are

    illustrated

    in

    Fig.

    2.13.

    There

    is close

    agreement

    between

    Jc

    and GF*

    However

    the

    deduced

    tensile

    strength

    is consistently

    less

    than

    the

    actual

    strength.

    A

    possible

    explanation

    for

    this

    tendency

    and

    a possible

    28

  • 8/18/2019 21768729-MIT

    29/158

    means

    of

    avoiding

    it

    is

    now

    presented.

    The J-integral

    method examines a portion

    of

    the

    crack

    plane

    which

    extends a distance Aa

    between the positions of the

    short and

    long

    notch

    tips.

    The actual

    ligament

    exists

    in the

    beam with

    the

    short notch. The

    deduced a-S

    curve

    represents the average stress

    transferred across this

    area

    (b

    Aa),

    as a

    function

    of the average crack

    separation. The

    loading

    configuration

    ensures that at

    any

    given time

    during

    the test the actual

    crack opening

    at

    any point within the specified zone gets smaller as the

    point moves from

    the

    initial

    notch

    tip

    towards the neutral

    axis. When the

    Aa

    zone is

    transferring

    maximum

    load,

    the

    stress

    along

    some

    line

    within the

    zone is equal to

    the

    tensile

    strength.

    At

    all

    other

    points

    the

    stress

    transferred must

    be less than

    ft. Thus when

    the

    average stress throughout

    the zone is

    calculated, it

    must be less than ft.

    Theoretically

    this

    problem is solved by letting Aa approach

    zero.

    In practice, however, there

    is a lower

    limit on Aa

    so that

    errors

    in

    the

    measurement of

    initial crack

    lengths and in the loading machine

    and instrumentation

    are

    not

    significant.

    Experience suggests

    a lower limit of about

    6

    mm

    for

    the beam bending

    configuration

    in Fig.

    2.2.

    A possible extension of

    the approach

    to date

    may

    be to

    use

    different

    Aa values

    in a given test

    program

    and to

    extrapolate the J-S or

    the a-S

    curve

    to

    obtain a curve as a-0. A

    variation

    of

    this possibility

    may be to

    use similar Aa

    values, to

    vary

    the

    ratio Aa/(d-a),

    and to extrapolate the

    resulting

    curves to find a solution

    as this

    ratio

    approaches

    zero.

    The

    problem

    with these

    approaches

    is

    that

    the

    number

    of

    test

    specimens

    required

    is

    considerably

    increased.

    Another

    source of

    error

    in

    the

    final result

    is from the raw

    data

    curves. Because this

    technique

    relies

    on

    measuring the

    difference

    between

    two load

    versus

    load point

    displacement curves

    obtained from notched

    specimens

    with

    only a small difference

    in

    initial notch lengths, any

    error

    in

    the

    measured

    curves results

    in

    a

    magnified

    error in the

    difference

    29

  • 8/18/2019 21768729-MIT

    30/158

    between the curves.

    Concrete, and

    especially

    fiber

    reinforced

    concrete,

    is

    a

    heterogeneous

    material

    and

    so

    there

    is

    always some

    variation in

    test

    results

    obtained

    from

    supposedly exactly

    similar

    specimens. In

    order

    to

    guard

    against the

    possiblity

    that

    a single exceptional

    test result may

    significantly

    affect

    the

    final

    deduced

    tension softening

    curve,

    a

    standardized

    approach

    is

    suggested

    for

    examining

    individual

    raw

    data

    curves

    and combining

    them

    to produce

    average

    curves which

    are likely

    to give

    a

    good

    result.

    Using

    each load

    versus load

    point

    displacement

    curve, two

    values

    are

    calculated,

    namely GF

    and ff(net)

    where

    ff(net)

    is

    defined

    as:

    f

    f(net) =

    6 M

    max

    b(d

    - a)

    (2.10)

    were

    Mmax is the

    maximum

    bending

    moment

    resisted

    by the

    notched

    section.

    It is reasonable

    to

    expect that

    these

    values

    are

    similar

    for the

    short

    notched

    and

    long

    notched beams.

    Thus

    it is

    also reasonable

    to assume

    that

    if

    the

    raw data

    curves for

    each

    notch

    length

    are chosen

    in

    a

    manner

    which

    ensures

    that

    the two

    average

    curves

    have similar

    GF and

    ff(net)

    values,

    then

    they

    are likely

    to produce

    an

    accurate

    tension

    softening

    curve.

    The

    ff(net)

    values

    significantly

    affect the

    deduced

    tensile

    strength

    and

    there

    is

    a direct

    relationship

    between

    the

    two GF

    values

    and the J c

    value.

    This

    approach

    is especially

    useful when

    the

    number of

    tests performed

    is small.

    2.8

    Current

    Status

    of J-Integral

    Test

    Technique

    The J-integral

    test

    technique

    just outlined

    has

    undergone

    extensive

    development

    and

    refinement

    in the

    last

    few

    years

    both

    here at

    M.I.T.

    and

    elsewhere.

    The

    testing

    details

    outlined

    here

    represent

    the

    most recent

    procedure

    used

    at

    M.I.T.

    for the

    beam

    bending

    configuration.

    Successful

    application

    of this

    test method

    to an

    increasingly

    wide

    range

    of

    materials

    30

  • 8/18/2019 21768729-MIT

    31/158

    including

    mortar

    and

    fiber

    reinforced

    mortar,

    granite

    [20]

    and basaltic

    rocks

    [211

    together

    with

    the possibility

    of

    using

    various

    specimen

    geometries

    such as

    beams,

    compact

    tension

    specimens

    and semi-circular

    specimens,

    suggests

    its

    usefulness

    as

    a standard

    test for

    the

    fracture

    properties

    of many

    non-yielding

    materials.

    This test

    is relatively

    simple

    and

    can

    be

    performed

    using

    any standard

    testing

    machine.

    Through

    the

    tension

    softening

    curve

    it

    provides

    us with

    a

    basic

    measure

    of

    the

    fracture

    resistance

    properties

    of

    any non-yielding

    material

    such as

    concrete.

    Given

    that

    the

    fracture

    resistance

    can

    be

    quantified

    by

    this

    test

    it

    is now

    essential

    to examine

    why we

    need

    to

    measure

    this

    material

    property

    and

    how

    we can use

    it

    to

    achieve

    improved

    and

    more

    efficient

    material

    performance

    in

    structural

    applications.

    It

    is

    imperative

    that

    not

    only

    does

    a standard

    test

    provide

    a good

    measure

    of

    some

    material

    property

    but

    also

    that

    that

    measure

    is

    capable

    of

    contributing

    to

    our

    understanding

    of

    how

    the

    material

    will

    behave

    in

    practice.

    The

    aim of

    chapters

    3

    and

    4

    is

    to establish

    fracture

    resistance

    as

    a

    material

    property

    which

    can

    dramatically

    affect

    structural

    performance

    and

    which

    therefore

    must

    be

    considered

    in any

    accurate

    design

    code.

    31

  • 8/18/2019 21768729-MIT

    32/158

    Table

    2.1 - Summary

    of J-Integral

    Test

    Results

    Matial

    Deduced

    Jc

    8C

    Dict

    GF

    #

    ft

    ft

    (MPa)

    (N/m)

    ( Im)

    (M~a) (N/m)

     

    2

    3

    4

    5

    6

    7

    8

    2.09

    2.80

    1.78

    2.09

    2.08

    2.01

    2.11

    3.35

    83.8

    1200

    187

    81

    205

    404

    543

    1485

    140

    3700

    700

    190

    1100

    1520

    1340

    2300

    3.0

    2.6

    2.6

    2.6

    2.6

    3.95

    78

    209

    414

    607

    1162

    32

  • 8/18/2019 21768729-MIT

    33/158

    Fig.

    2.1

    1 a

    J

    J

    127

    r

    ds

    5t

    S

    =

    separation

    distance

    :)

    =

    stress

    transferred

    Cohesive

    Zone

    Ahead of

    the Crack

    Tip

    -4

    -4

    ii

    a

    127 5

    Specimen width

    =

    63.5

    19.1

    82.5

    Specimen

    width

    = 19.1

    Dimensions

    in

    mm

    Fig.

    2.2

    Specimen

    Configurations

    used

    with

    J-Integral

    Test

    33

    j

    i

    25

    127

     

    I

    G

  • 8/18/2019 21768729-MIT

    34/158

    Fig.

    2.3 Inside of Omni-Mixer

    Fig.

    2.4

    Specimens Covered in

    Plastic just after Casting

    34

  • 8/18/2019 21768729-MIT

    35/158

    Fig

    2.5 Notched Beam

    ready

    for

    Testing

    Fig. 2.6 Loading Machine

    Used

    for J-Integral Test Program

    35

  • 8/18/2019 21768729-MIT

    36/158

    3000 r

    2000

    Crack

    length

    = 57.15mm

    1000-

    Crack

    length

    =

    63.5mm

    1000

    2000

    3000

    4000

    5000

    Load

    point

    displacement

    (pm)

    Fig.

    2.7

    Average

    Load

    versus

    Load Point

    Displacement

    Curves

    3000.

    M

    82

    2000-

    83

    0

    o -

    S

    1000-

    1000

    2000

    3000

    4000

    Load

    point

    displacement

    (pm)

    Fig.

    2.8

    Average

    Load

    Point

    Displacement

    versus

    Crack

    Opening

    Curves

    36

  • 8/18/2019 21768729-MIT

    37/158

    I

    - I ~

    1000

    2

    Crack

    opening

    (pm)

    Fig. 2.9 J-Integral

    versus Crack Opening

    Curve

    correct

    8

    500 1000

    1500

    Crack opening (pm)

    Fig.

    2.10

    Deduced

    Tension

    Softening

    Curve

    37

    2000

    r

    1500

    [

    E

    S..-

    w

    4.

    C

    1000

    500

    3000

    3.0

    (U

    4.-

    (I,

    2.01

    1.0

    2000

    2500

  • 8/18/2019 21768729-MIT

    38/158

    4.0-

    3.0-

    2.0-

    Deduced

    curve

    1.0.

    Directly

    measured

    curve

    500

    1000 1500 2000

    2500

    Crack

    opening

    (pm)

    Fig. 2.11

    Comparison

    Between

    Deduced

    and

    Directly

    Measured

    Tension

    Softening

    Curves

    Shaded

    area

    =

    energy

    absorption

    recorded

    in test

    A

    + A2 =

    energy

    supplied by

    beam

    self

    weight

    0

    A A

    (Petersson,

    10]

    F0

    =

    applied load

    which

    gives

    same

    moment

    at

    mid-span

    F

    0

    A

    A as

    beam

    self-weight

    ] . 2

    d

    0

    Load

    Point Displacement

    Fig. 2.12

    Flexural

    Load

    Deflection

    Curve

    Corrected

    to

    Account for

    Energy

    Supplied

    by

    Beam Self-Weight

    38

  • 8/18/2019 21768729-MIT

    39/158

    4.0

    3.0.

    #8,

    Kevlar

    FRM, 6.4 mm,

    2%

     7, Acrylic

    FRM,

    6.4 mm, 3%

    2 0

    #6, Acrylic

    FRM,

    6.4 mm,

    2%

    #2,

    Steel FRM, 9.53

    mm, 1%

    #4, Plain Mortar

     1,

    Plain

    Mortar

    1.0-

    #5,

    Acrylic

    FRM,

    6.4

    mm, 1%

     3 ,

    Acrylic

    FRM,

    12.7

    mm,

    1%

    C

    = 3700Pm

    500

    1000

    1500

    2

    25 0

    Crack

    opening

    (m)

    Tension

    Softening Curves

    Deduced by Indirect

    J-Integral

    Technique

    ig.

    2.13

  • 8/18/2019 21768729-MIT

    40/158

    3

    FLEXURAL

    BEHAVIOR

    OF FIBER

    REINFORCED MORTAR

    3.1 Introduction

    Is

    there

    any

    need

    to use

    fracture resistance

    as

    a material property

    in

    the design of concrete

    structures? For many years the majority of

    Engineers,

    if

    asked, would

    answer

    a definite no

    to

    this question. The

    design code