A.F.

pruijssers AGGREGATE INTERLOCK AND

DOWEL ACTION UNDER MONOTONIC AND
CYCLIC LOADING
 ^4U i *

TA. i n s A\~>
AGGREGATE INTERLOCK AND DOWEL ACTION
UNDER MONOTONIC AND CYCLIC LOADING
AGGREGATE INTERLOCK AND DOWEL ACTION
UNDER MONOTONIC AND CYCLIC LOADING

PROEFSCHRIFT
ter verkrijging van de graad van doctor
aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus Prof.dr. J . M . Dirken,
in het openbaar te verdedigen ten overstaan van
een commissie aangewezen door
het College van Dekanen
op dinsdag 14 juni 1988
te 16.00 uur door

ADRIANUS FRANS PRUIJSSERS,

geboren te Rotterdam,
civiel ingenieur

Delft University Press / 1988

TR diss
1643
Dit proefschrift is goedgekeurd door de promotoren
Prof.dr. - Ing. H.W. REINHARDT
en
Prof.dr.ir. J.C. WALRAVEN

ACKNOWLEDGEMENT

The experimental part of this research was performed in the Stevin Laboratory of the
Delft University of Technology with financial support and under the supervision of the
CUR (Netherlands Centre for Civil Engineering, Research, Recommandations and
Codes), which is greatly appreciated.
The author wishes to record his thanks to all members of the "Concrete Structures
Group", who have contributed to this research project.
I would like to express my gratitude to Dirk Verstoep b.v. for giving the opportunity to
complete this thesis.
The financial support received from the "Stichting Professor Bakkerfonds" for this
publication is gratefully acknowledged.

CIP GEGEVENS KONINKLIJKE BIBLIOTHEEK, DEN HAAG

A.F. Pruijssers

ISBN 90-6275-451-1

Copyright © 1988 by A.F. Pruijssers.

All rights reserved. Published 1988.
No part of the material protected by this copyright notice may be
reproduced or utilized in any form or by any means, electronic or
mechanical, including photocopying, recording or by any
Information storage and retrieval system, without written permission
from the publisher: Delft University Press.

Printed in The Netherlands

CONTENTS

1. INTRODUCTION
1.1 Scope of research.
1.2 Aim of the research program.

2. SURVEY OF THE LITERATURE

2.1 Introduction.
2.2 Aggregate interlock; monotonie loading.
2.3 Aggregate interlock; cyclic loading.
2.4 Dowel action; monotonie loading.
2.5 Dowel action; cyclic loading.
2.6 Contribution of axial steel stress.
2.7 Shear strength of cracked reinforced concrete;
monotonie loading.
2.8 Shear strength of cracked reinforced concrete;
cyclic loading,
2.9 Conclusions.

3. EXPERIMENTAL STUDY
3.1 Introduction.
3.2 Reinforced specimens; repeated loading.
3.2.1 Test arrangement.
3.2.2 Test variables.
3.2.3 Experimental results.
3.3 Externally reinforced specimens; repeated loading.
3.3.1 Test arrangement.
3.3.2 Test variables.
3.3.3 Experimental results.
 page

A. THEORETICAL MODELLING OF THE RESPONSE OF CONCRETE TO MONOTONIC SHEAR

LOADING
4.1 Introduction. 59
A.2 The mechanism of aggregate interlock. 59
A.3 The mechanism of dowel action. 6A
A.A The combined mechanism of aggregate interlock and dowel 81
action.
A.5 Influence of the normal restraint stiffness upon the 89
shear stiffness.
A.6 Additional detailed tests. 93
A.7 Concluding remarks. 95

5. THEORETICAL MODELLING OF THE RESPONSE OF CRACKED CONCRETE TO REPEATED

AND REVERSED SHEAR LOADING
5.1 Introduction. 97
5.2 The mechanism of aggregate interlock. 97
5.3 The mechanism of dowel action. Ill
5.A The combined mechanism of aggregate interlock and dowel 120
action.
5.5 Influence of the normal restraint stiffness upon the 129
shear stiffness for the case of repeated loading.
5.6 Concluding remarks. 131

6. IMPLEMENTATION OF THE CYCLIC AGGREGATE INTERLOCK MODEL INTO

NUMERICAL PROGRAMS
6.1 Introduction. 13A
6.2 Simplified expressions for the static two-phase model. 136
6.3 Rheological model for an element with the smeared out 138
crack concept.
6.A The stress-strain relation for the case of cyclic 1A8
loading.
6.5 Implementation of the dowel action mechanism. 151
6.6 Concluding remarks. 153
Stellingen behorende bij het proefschrift:
"Aggregate interlock and dowel action under monotonie and cyclic loading"
van A.F. Pruijssers

1. De openingsrichting van een scheur in gewapend beton belast door een

monotoon toenemende schuifkracht wordt aanvankelijk bepaald door de
vervorming van de wapeningsstaven. Na het volledig ontwikkelen van
plastische scharnieren in deze staven, wordt het scheuropeningspad
opgelegd door de haakweerstand van de toeslagkorrels.

2. Het gedrag van een scheur in gewapend beton onderworpen aan een zeer
groot aantal lastwisselingen met een kleine amplitude van de schuif-
spanning, kan quasi-statisch worden beschreven.

3. Het twee-fasen model van Walraven voor de beschrijving van de haak­

weerstand van de toeslagkorrels onder monotoon toenemende belasting
kan op eenvoudige wijze worden aangepast voor het geval van herhaal­
de- en wisselende schuifbelastingen.

4. De samenwerking van staal en beton leidt tot een verhoging van de

deuvelsterkte van op afschuiving belaste wapeningsstaven.

5. Ten aanzien van de haakweerstand van de korrels onder wisselende

schuifbelasting met constante amplitude kan worden gesteld dat de
belastingsgeschiedenis volledig ligt besloten in de eindscheurver-
plaatsingen van de laatste wisseling.

6. Er treedt geen herverdeling van de belasting op tussen de mechanis­

men van de haakweerstand van de korrels en van de deuvelwerking ten
gevolge van het wisselen van belasting.

7. Het 'gebruik van een schuif-reductiefactor met een constante waarde

gaat voorbij aan het fysische gedrag van een scheur, maar leidt on­
der monotoon toenemende belastingen niet tot onrealistische schuif-
spanningen in een gescheurd element. Indien de richting van de
hoofdspanningen zich sterk wijzigt gedurende het belasten, bijvoor­
beeld door het wisselen van de belasting, wordt met een constante
reductiefactor een fysisch onjuist scheurgedrag verkregen.
 8. Een goede toets voor de toepasbaarheid van een numeriek elementen-
programma is wellicht het simuleren van een proef met een vooraf
bekende, doch gefingeerde en fysisch onmogelijke uitkomst.

9. Het afschuifdraagvermogen van niet op afschuiving gewapende liggers

berust nagenoeg geheel op de som van de schuifspanningen in de druk-
zone, de ongescheurde trekzone en de tension-softening zone.

10. Problemen zijn als een muur, men dient op eikaars schouders te staan
om vooruit te komen. Veelal komt men echter niet verder dan op el-
kaars tenen te staan.

11. De te verwachten zeespiegelrijzing noopt het laaggelegen en dichtbe­

volkte Nederland tot een zeer actief beleid ten aanzien van de Euro­
pese eenwording.

12. De juiste oplossing voor een probleem is vaak zo eenvoudig dat het
niet meevalt om te verklaren waarom deze niet eerder gevonden is.

13. Het is niet dom om iets slims niet zelf te bedenken, wel om het om
die reden niet te gebruiken.
 page

7. RETROSPECTIVE VIEW AND CONCLUSIONS 154

8. SUMMARY 157

9. NOTATION 163

10. REFERENCES

11. APPENDICES
 - 1 -

1. INTRODUCTION

1.1. Scope of the research.

Todays offshore industries demand offshore platforms enabling the ex­

ploitation of large oil and gas reservoirs in the Arctic and the deep
sea up to about 300 m. Heavily reinforced high-strength concrete struc­
tures are very effective in withstanding the severe loading conditions
in the Arctic environment, dominated by icefields and icebergs, and by
the deep sea, characterized by extreme wave and wind attacks. The safety
against failure of such complex structures is analysed by idealizing the
structure as an assemblage of basic elements. The interactions of these
elements and their redistribution of the applied loads and deformations
can be simulated in advanced finite element programs. As a consequence,
the problem of designing a large-scale structure with sufficient safety
against failure is shifted towards a thorough understanding of the mate­
rial behaviour of the basic elements and towards efficient numerical so­
lution techniques.
It was for this reason that the Netherlands Centre for Civil Engineering
Research, Recommendations and Codes (CUR) started the project 'Concrete
Mechanics'. This project comprises experimental research and material
modelling on the one hand and implementation of these models in numeri­
cal programs on the other hand. The Concrete Mechanics project is a co­
operation of a division of the Netherlands Ministry of Transport and
Public Works (Rijkswaterstaat), the Institute for Applied Scientific
Research on Building Materials and Building Structures (IBBC-TNO) and
the Universities of Technology of Delft and Eindhoven.
Due to the applied loads and deformations, structural elements are sub­
jected to tensile stresses causing cracking of the concrete. Although
offshore structures are generally designed to remain uncracked under
service conditions, colliding ships or icebergs might cause cracks. In
1980 five severe ship collisions were reported in the English part of
the North Sea [59], resulting in damage of the structure.
Apart from the 'special circumstances" such as collisions, a structure
can possibly be designed in a more economic way when the stiffness of
cracked reinforced concrete, which is still considerable, is utilized in
withstanding the applied loads and deformations. In bridge design, it
 - 2 -

appeared that such an approach is especially favourable for the case of

large settlements. A partially prestressed concrete structure can easily
follow such settlements, whereas a fully prestressed structure cannot.
As a consequence of the development of cracks, the response of the ele­
ments to severe loading conditions becomes highly non-linear with large
irreversible deformations. This non-linear material behaviour must be
thoroughly understood and modelled. Therefore, the first phase of the
'Concrete Mechanics' project focussed upon the experimental and theore­
tical investigation of the static shear strength and stiffness of crack­
ed concrete, the bond behaviour of the reinforcing bars and the funda­
mental material behaviour, such as multiaxially loaded concrete.
As numerical tools, two basically different non-linear finite element
programs were developed. In the first program (MICRO), the development
of cracks is taken into account by defining additional crack displace­
ments within an element, the so-called 'discrete crack' concept. This
program is particularly suitable for analyzing structural details. The
second program (DIANA) is based upon the concept of 'smeared-out'
cracks, in which the effect of cracking is accounted for by reducing the
stiffness of the 'cracked' elements.

a. Typical offshore structure b. Base of the structure.

Fig. 1.1. Typical offshore structure and loads acting upon the structure.

Offshore structures are designed in such a way as to transfer the cyclic

loads due to wave and wind attacks to the subsoil by means of in-plane
 - 3 -

stresses [25,26]. The walls of the base of such a structure will be sub­
jected to in-plane shear, see Fig. 1.1. Thermal deformations due to the
storage of hot oil and unequal settlements might cause additional crack­
ing of the walls of the base. For this reason, the current study, which
forms part of the second phase of the Concrete Mechanics project, focus-
ses upon the response of cracked reinforced concrete to cyclic in-plane
shear loads. Experiments with cyclic in-plane shear loading provide vi­
tal information on the response degradation of the cracked elements due
to cycling.
A large number of tests [33,37,43,78] has been conducted with a rather
large initial crack width and a relatively high shear load, the so-call­
ed 'high-intensity low-cycle' experiments. These tests especially re­
flected the case of a nuclear containment vessel, which is cracked due
to an internal explosion and subsequently subjected to earthquake mo­
tions. For the case of offshore structures, those tests provide informa­
tion on the response of the structure to severe loading conditions. How­
ever, offshore structures are generally subjected to millions of load
cycles due to wind, wave and ice attacks. These load cycles have a rela­
tively low amplitude with respect to the static strength, but might
cause gradually increasing irreversible deformations, thus influencing
the strength and stiffness of the structure in the case of subsequent
higher loads. Therefore, apart from the 'high-intensity low-cycle'
tests, experiments of the 'low-intensity high-cycle' type are of special
interest for offshore structures.

1.2. Aim of the research program.

The aim of the research program is the determination of the relationship

between the stresses and displacements occurring in the crack plane. The
results of previous experimental investigations [45,76,81] showed that
the transfer of stresses across a crack in concrete depends upon the me­
chanisms of the axial and lateral stiffness of the bars crossing the
crack and upon the roughness of the crack faces. With regard to the
roughness of the crack faces, the first phase of the Concrete Mechanics
project yielded a physical model describing the response of cracked
plain concrete to monotonie shear loading [81]. According to this model,
this roughness is caused by the contact between the matrix material and
 - A -

the aggregate particles protruding from the crack faces. Because of the
nature of this mechanism, the particle distribution, the maximum par­
ticle size, the strength of the matrix material and the coefficient of
friction between the particles and the matrix affect the shear stiffness
of the crack.
The contribution of the bars to the transfer of shear stress across the
crack is characterized by a strong interaction of the axial steel force
and the lateral force (dowel force). Therefore, a physically sound de­
scription of the shear stiffness must incorporate the interaction with
the normal restraint stiffness of the crack. Hence, the relationship
between the stresses and displacements in a crack has to be expressed

SM S12 A6 n (1.1)

S 21 S22 A« t

with AÓ , A6 = increments of crack displacements, (see Fig. 1.2.)

Fig. 1.2. Stresses and displacements in the crack plane.

However, in numerical programs generally only Sli is taken into account.

Therefore, the existing static model [81] will first be simplified for a
proper implementation of eq. (1.1) in numerical programs. Second, the
effect of cyclic loading on the crack response will be experimentally
investigated and incorporated in the existing model.
The response of cracked concrete to shear loading has been subject of
numerous experimental studies. Therefore, the information obtained in
these surveys concerning shear transferring mechanisms will be briefly
reviewed in the following Chapter. Furthermore, additional tests have
been carried out since it appeared that there was a lack of experimental
information.
 - 5-

2. SURVEY OF THE LITERATURE

2.1. Introduction.

The major mechanisms affecting the transfer of stresses across cracks in

reinforced concrete are:
a. Aggregate interlock of the crack faces; Due to the roughness of the
crack faces, stresses can be transferred from concrete to concrete. This
mechanism, denoted aggregate /interlock by Fenwick [18], is based upon
the fact that in low to medium strength gravel concrete the particles
have a much higher strength than the matrix material. Therefore, a crack
runs through the matrix and along the interface between particles and
matrix, see Fig. 2.1a. As a consequence, the stiff particles are pro­
truding from the crack plane, thus providing a contact with the matrix
material of the opposing crack face when shear sliding occurs.

a. Aggregate interlock. b. Dowel action. c. Axial steel stress.

Fig. 2.1. Transfer mechanisms in cracked reinforced concrete.

b. Dowel action of the reinforcing bars; Dowel action is based upon the
response of the concrete supporting a steel bar, which is forced to a
lateral displacement, see Fig. 2.1b.
c. Axial steel stress in the reinforcing bars; The reinforcing bars ge­
nerally cross the crack plane at different angles. The component of the
steel stress parallel to the crack plane contributes to the shear stress
transfer across the crack, see Fig. 2.1c.
These mechanisms will be discussed separately. Finally the interaction
of the mechanisms is reviewed. In this Chapter most attention is paid to
the experimental results reported in the literature. Information on the
available empirical and physical models will be reviewed in Chapter A.
 - 6 -

2.2 Aggregate interlock; monotonie loading.

Taylor (in 1959, [67]) and Moe (in 1962, [48]) paid some attention to
the role of aggregate interlock in the load transfer in cracked concre­
te. It was, however, Fenwick [18], who first carried out a detailed ex­
perimental study into the aggregate interlock mechanism. The scope of
this investigation was to determine the relationship between the shear
resistance of cracked plain concrete and the crack displacements. The
variables were the crack width, ranging from 0.06 mm to 0.38 mm, and the

nrplnrm ed crack
-
r-

(
125
on
50 A s 0
£BK!IS ■

125
/a
/O

/ ■ "*"

Fig. 2.2. Test arrangement used by Fenwick [18]

shear stress T n {MPal shear stress t . I MPal

2.0
\, 0.05 mm
6n =
ti
, , 5 6 MPa
/■«MPa
0.13 mm 6 n = 019 m

0.19 mm

/\ OjSmrn. 0.32 mm
,.33MPa

*^ . 0.38 mm

, . 19 MPa
rv crnr.kin^
'ccy1=
oLissa men
05

( „ y l = 33 MPo

0.05 0.15 0.20 0.25 005 010 015

sheor displacement 6\ [mml shear displacement of (mm)

Fig. 2.3. Shear stress - shear slip Fig. 2.4. Shear stress - shear slip
relation as function of the as function of the concrete
crack width for Fenwick's grade for Fenwick's tests
experiments [18]. [18].

concrete strength varying from 19 MPa to 56 MPa. Fig. 2.2. presents the
test specimen and testing rig used by Fenwick. The specimens were pre-
cracked providing a relatively small shearing area of 7900 mm 2 . The
influence of the crack width was investigated in the first test series,
 - 7 -

in which the crack width was kept constant during the stepwise shear
load application. Unfortunately, the normal force, which was used to
adjust the crack width, was not measured during the tests. For this test
series with a concrete cylinder strength of 33 MPa, the mean test re­
sults are shown in Fig. 2.3. All the specimens failed due to secondary
cracking. Each test was repeated five to six times to reduce the scatter
of the readings.
A second test series with a constant crack width of 0.19 mm and varying
concrete grade was used to determine the influence of the concrete
strength upon the shear resistance. The average experimentally obtained
curves are shown in Fig. 2.A. The following empirical expression was de­
rived from the experimental results:

T = l[üilél _ o.658)(/f
y ~- 1.447ÏÏ6 - 0.0446 1 [MPa] (2.1)
a 6 ' ccyl '^ t n'
n '

with T , f , in l[MPa], 6 , 6 in [mm]

a ccyl ' n t

In addition to this experimental study, Houde and Mirza [311 performed

32 experiments in a testing rig, which was quite similar to the equip­
ment used by Fenwick. Apart from the crack width and the concrete
strength, the maximum particle size was a variable, ranging from 9.5 mm
to 19 mm. The extremities of the concrete specimens were reinforced pre­
venting any secondary cracks. Fig. 2.5 shows some average test results,
which are comparable with Fenwick's results. It was found that the maxi­
mum particle size hardly influenced the shear resistance. The shear
strength was found to be proportional to /Ê ~ for concrete strengths
ranging from 16.5 MPa to 51 MPa. The experimental results of Fenwick and
Paulay [18] and Houde and Mirza [31] might be influenced by the test
set-up allowing flexural cracking and the relatively small shearing
area, giving rise to a considerable scatter.
Therefore, Paulay and Loeber [50] performed tests with an improved type
of specimen, see Fig. 2.6. Now, the shear plane of the pre-cracked push-
off specimen was 21660 mm 2 . The upper part of the specimen could slide
along the shear plane of the lower part, which was fixed. The authors
performed 44 tests exploring the nature of shear transfer. A part of
this test series was carried out with a cyclic shear load and will be
 - 8 -

_ shear stress \Q iMPal

V
2.0
0 05 mm
r
U13 mm 0.25 rgm flgrmol Stress
/>"
1.5
/° 0.36 mm

tl "0.51mm

0.5 7
> 4
A
A
Icrock olone 1K»190
f c c y , = 31.5MR)
\-'^
0,05 0.10 0.15 0.20 025 0.30
shear displacement 6j Imm)

Fig. 2.5. Test results of Fig. 2.6. Test specimen used by

Houde and Mirza [31]. Paulay and Loeber [50],

discussed in Section 2.3. The variables were the type of aggregate (9.5
mm and 19 mm round maximum size and 19 mm crushed maximum size), the
crack width (0.13 mm, 0.25 mm and 0.51 mm) and the way of load applica­
tion. A concrete cylinder strength of 37 MPa was used. The experimental­
ly obtained relationship between the monotonically increased shear
stress and the shear displacements is shown in Fig. 2,7 for constant
crack widths. It appeared that neither the aggregate size nor its shape
strongly influenced the shear resistance.
Because of the improved type of specimen, the shear strength exceeded
the maximum values obtained by Fenwick [18] and Houde [31]. During the
constant crack width tests the magnitude of the normal restraining force
was measured, see Fig. 2.8. The test results yielded an average coeffi­
cient of friction equal to 1.7. An important observation was the insen-
sitivity of this value to the crack width and the aggregate type.
A second test series focussed upon the influence of an increasing crack
width upon the shear transfer in cracked plain concrete. During these
tests, the ratio of the shear load to the crack width was kept constant
at a value of 1.38 MPa to 0.1 mm. The experimental results of these
tests are compared with the results of the constant crack width tests,
see Fig. 2.9. The dotted line in Fig. 2.9 represents the theoretical re­
sults according to the tests with constant crack width. This curve has
the same shape as the mean experimental curve for the variable crack
 - 9 -

shear displacement 6 , [mm] normal stress aa IMPal

Fig. 2.7. Relation between shear stress Fig. 2.8. Shear stress as function
and shear displacement [50]. of normal stress [50].

shear stress x a (MPal

e I 1 1 1 1 ] 1 1
0.51 mm

shear displacement 6. [mm!

Fig. 2.9. Comparison of tests with constant

and with variable crack width [50].

width tests. This indicated that the load history or crack opening path
hardly influenced the crack response to shear loads.
Taylor [66] pointed out that a crack actually opens simultaneously with
the shear sliding. Therefore, he performed tests with a constant ratio
of the crack width to the shear displacement. A schematic presentation
of the test equipment is shown in Fig. 2.10a.
The specimens were pre-cracked with a shearing area of 17780 mm2.
 - 10 -

A total of 32 tests was carried out, exploring the influence of type of

aggregate, the aggregate size, the concrete strength and the ratio of
crack width to shear displacement. The influence of the crack opening
direction is shown in Fig. 2.10b for crack width to shear displacement
ratios ranging from 0.27 to 2.15. The influence of the concrete strength

shear strength t ou lMPo| shear strengthToUtMPal

5
i
j ;

• J

strain gauges

0 0.5 1.0 1.5 20 2.5 40 60

lower crossheod/ 6 n / 6 , 1-] „IMPol

a. Test arrangement. b. Shear strength versus c. Shear strength versus

crack opening direction. concrete strength.

Fig. 2.10. Test arrangement and experimental results of Taylor [66]

is presented in Fig. 2.10c showing a nearly linear relation between the

shear strength and the .concrete strength, although a large scatter is
observed. From a test series with weak aggregate particles, it appeared
that the particle strength with respect to the matrix strength strongly
influenced the shear strength. A relatively weak aggregate particle com­
pared with the matrix material will allow the crack to run through the
particles, thus yielding a smooth crack plane. During these tests, the
crack opened simultaneously with the shear sliding. It must, however, be
doubted whether the constant ratio of crack width to shear slip provided
a suitable description of the actual crack behaviour. From [80] it is
known, that for beams the ratio of crack width to shear slip increased
with increasing crack width. Therefore, Walraven [81] performed tests on
precracked push-off specimens with external restraint bars, see Fig.
2.11.

For these specimens the crack opening was restrained passively, so that
the crack opened according to the internal equilibrium in the crack
plane. The crack displacements were measured by means of plate spring
gauges. The displacements were recorded at three locations on both sides
of the specimen. The external normal force was measured using strain
gauges stuck to the external bars. The tests were performed in a dis-
 - 9 -

shear stress t„ IMPol

sheer stress t n [MPal

6
n° 0.13mm 0.25 mm
1 / 0.51 mm
A/ /

j
'/A
ê
§>.r\

'/
/ /

v^>
A
!20V.

\L lAuWf.
. '45%

0.2 0.3 04 0.5 0.6 07 08 2 4 6

shear displacement 6 t Imml normal stress ffQ IMPal

Fig. 2.7. Relation between shear stress Fig. 2.8. Shear stress as function
and shear displacement [50]. of normal stress [50].

shear stress x a IMPol

04 0.5 0.6 0.7 0.8

shear displacement 6, (mm!

Fig. 2.9. Comparison of tests with constant

and with variable crack width [50],

width tests. This indicated that the load history or crack opening path
hardly influenced the crack response to shear loads.
Taylor [66] pointed out that a crack actually opens simultaneously with
the shear sliding. Therefore, he performed tests with a constant ratio
of the crack width to the shear displacement. A schematic presentation
of the test equipment is shown in Fig. 2.10a.
The specimens were pre-cracked with a shearing area of 17780 mm2.
 - 11 -

restraint plate

crock plane 120 » 300

plote spring gouges

restraint bar

strain gauges

Fig. 2.11. Test specimen as used by Walraven [81].

placement-controlled manner, so that the post-peak behaviour could be

recorded. The variables were the concrete strength and composition, the
external restraint stiffness and the initial crack width.
The mix composition was varied using gap-graded mixes and mixes accord­
ing to Fuller's ideal curve, normal and lightweight concrete and varying
the maximum particle size (16 mm and 32 mm). The cube crushing strength
was ranging from 13.4 MPa to 59 MPa. For the tests the initial crack
width varied between 0.01 mm and 0.40 mm. The Figs. 2.12a-c present some
typical test results for a 150 mm cube strength equal to 37.6 MPa and a
maximum particle diameter of 16 mm.
n
shear displacement 6tlmm) <> specimen
1 1 /.0/68
- a •■■ 2 1/0/36
3 W. 4/1.0

th
12 I 4 1/2/1.6
2.0 5 1/2/U
6 1/2IM
7 U . t /.3

f 1
normal stress <JQ IMPol
h

I
1.5
/

J- 2

f
'■3 2
1.0

V,4 1/
5
/

f %
^
^ --'" 6 0.5

m ///
/
^
Ê•
%A yi

0.5 1.0 t.5 2.0 2.5 0 05 1.0 1.5 0.5 1.0 1.5
sheor displacement 6. Imml crack width 6 n [mml crack width 6 n [mm)
o. Shear stress - shear b. Crack opening path. c. Normal stress-crack
displacement relationship. width relationship. -
Fig. 2.12. Typical test results of Walraven for the plain concrete
specimens[81].
 - 12 -

The identifying number of the individual specimens consists of the mix

number, the initial crack width and the restraint stress at a crack
width equal to 0.6 mm respectively. Due to the increasing normal force,
there still was a slight increase in shear strength for very large shear
displacements. Although the crack was allowed to open simultaneously
with the shear sliding, the curves for constant crack width were derived
in the same manner as was done in Fig. 2.9. These curves are shown in
Fig. 2.13. It appeared that the maximum particle size had only a slight
influence upon the shear strength in the range tested. Therefore, simple
bilinear expressions were derived empirically from test results, ignor­
ing the influence of the maximum particle size:

1.8 0.234
T = 0 80 ♦ (-™- - 0.20)f c c j6 t (T > 0) (2.2)
a
a 30 o
s n

n 0.191
+ (- 0.15 )f K (o > 0)
a
(2.3)
20 1.35 ccnr t
0.63 n
with T , o , f in [MPa], ó , 6 in [mm]
a a ccm ' ' n t

sheor stress T0 [MPo]

normo! stress crQ [MPa!

Fig. 2 . 1 3 . Comparison of the experimental r e s u l t s and e q s . ( 2 . 2 ) - ( 2 . 3 ) 181],

These b i l i n e a r r e l a t i o n s a r e compared with the experimental r e s u l t s in

Fig. 2 . 1 3 .
Daschner and Kupfer [12] performed 52 tests on normal- and lightweight
concrete specimens varying the concrete cube strength (f = 25 and 55
ccm
MPa) and the maximum particle diameter (8 mm and 16 mm). The test equip­
ment was an improved version of the test arrangement used by Fenwick et
al. [18] and Houde et al. [31]. In a first test series, the crack width
 - 13 -

was kept constant on a preset value ranging from 0.05 mm to 0.40 mm. A
second test series focussed on tests with a constant normal restraint
stress during shear sliding. It emerged from the test results that for
very high initial normal stresses the crack width changed sign, which
indicated that the readings of the displacement transducers were influ­
enced by the deformation adjacent to the crack. Indeed, Daschner and
Nissen [13] suggested that the high normal restraint force had caused
elastic and plastic deformations of the test specimens thus influencing
the deformation between the measuring points. Because of the questions
left open, these tests will not be discussed here. In addition to these
tests, Nissen [49] improved the test arrangement, which was used by
Daschner and performed 42 push-off tests, see Fig. 2.14a. He investi­
gated the influence of the crack opening path upon the shear stiffness
of the crack. Tests with constant crack width and tests with constant
normal stress were performed. Some typical results are shown in Fig.
2.14b.

relative stresses To'lccmffi'ccml-l

crack plane £00 » 200 mm' Oflüxm ,
29.6
/ .normol force
02
rV„t='6 mm
JSLIB
SP1F 3_1__.
02 tU 0'B B'8 10 13
shear slip (Imml.

crack width n S
fmm]
a. Test arrangement. b. Test results.

Fig. 2.14. Test arrangement and test results of Nissen [49].

The cube concrete strength was varied between 27-31 MPa and between 54-
57 MPa (cube 200x200x200 m m 3 ) . The water cement ratio was rather high
(w/c 0.51-0.80). The maximum diameter of the gravel particles was varied
between 8, 16 and 32 mm. Nissen found that the ratio x li was hardly
' a ccm ■"
affected by the concrete strength and maximum particle diameter. It
appeared that the stresses transferred across a crack for any given
combination of the crack displacements are strongly influenced by the
 - 14 -

crack opening path followed during the tests.

Millard and Johnson [45] carried out tests on pre-cracked specimens of
the push-off type. The equipment used was very much alike the testing
rig used by Walraven [81]. However, now the normal restraint bars could
be tensioned before application of the shear load, see Fig. 2.15. The
test variables were the initial crack width ranging from 0.063 mm to
0.75 mm, the cube crushing strength varying between 29 MPa and 52 MPa
and normal restraint stiffness. The Figs. 2.16a-c present some typical
test results for a cube crushing strength of 36 MPa and a normal re­
straint stiffness of 6.2 MPa.

The experimentally obtained results are in agreement with the test re­
sults found by Paulay and Loeber [50] and Walraven [ 8 1 ] .
sheor lood

distribution beom

knife-edge bearing

adjusting turnbuckle

flexible strop

crock plane 7Q»270

Fig. 2.15. Test arrangement used by Millard and Johnson [45],

Vintzeleou [76] carried out push-off experiments exploring the influence

of the surface roughness (smooth, sand blasted, rough) upon the shear
strength. Furtheron, the concrete cylinder strength was varied between
16 and 40 MPa and the normal restraint stress was kept constant at val­
ues of 0.5, 1.0 and 2.0 M P a . Fig. 2.17 shows the test arrangement used
by Vintzeleou. For a cylinder strength equal to 25 MPa and maximum par­
ticle diameter of 30 mm, some typical test results are presented in
Fig. 2.18a-b. Note that the crack opened faster the higher the normal
restraint stress. Vintzeleou stated that this was due to the large scat­
ter. Apart from the crack displacements, the roughness of the shear
plane was measured before and after the actual shear test. For the rough
 - 15 -

shear stress la tMPol normol stress a 0 (MPol

no. t^. öpg axiol stiffn. 15L
2S 36MRj.06mm 5 . 5 - ^ -
9S 41 .25 63 m m
15L 35 .50 7.2 9S /
19L 31 .75 5.7
/ i //

19L

/ /
f

0 1 2
shear displacement 6( [mml
a. Shear stress - shear
0
/ 1
0.5 1.0
crock width 6 n [mml
b. Crack opening path.
0
/ / /

- 1

c. Normal stress - crack

width relationship.
2
crack width 6 n Imml

displacement relationship.

Fig. 2.16. Typical test results of Millard [45].

crock plone
A /jUal_jocJs.
'///,//sss,

m ..
>////»///$>
300 , 300 \ \
T //////»;
. 300
' sheor l o o d \

Fig. 2.17. Test arrangement used by Vinzeleou [76],

interface, the roughness, defined as half the height of the protruding

aspertities, was 1.75 mm before and 1.45 mm after testing due to the de­
terioration of the crack faces.
Divakar and Shah [14] also performed push-off tests with constant normal
stress. Using a dead-weight, see Fig. 2.19a, a constant normal stress
was applied to the crack plane. For displacement-controlled tests, it
was found that for increasing normal stress, the increment of the shear
displacement becomes larger relative to the crack width increment, see
Fig. 2.19b. The concrete strength was 35 MPa. Note that the shear area
 - 16 -

1.2
crock width 6 n [mm]

rough interfc 6 t U '; mm

" 0 Qa = [MPO]
12 - 'ccyl= 25 MPa

0.8 • .V *
^r
1.0
•
08
"A
0.6

0.1 id. ^ - ^ 0 . 5 f- •■

0.2

0
0 1 2
shear displacement 6 ( Imm]
L
0 0.2 04 0.6 0.8
»t ' 6 t u H
1.0

a. Crock opening path b. Relative shear stress as function of

the relative shear displacement.

Fig. 2.18. Some typical test results of Vintzeleou [76].

was very small with respect to the maximum particle diameter of 12.7 mm
(crushed angular aggregate). In fact, the results showed a remarkable
consistency related to those small specimen dimensions.

152*.
crack width 6n [mm\
specimen
rack plane 152.4 » 25.4 mm*

restraint rod

0 005 010 0.15 0.20 0.25 0.30

shear displacement 6f (mm)

a Test arrangement. b. Test results.

Fig. 2.19. Test specimen and results of Divakar [14].

 - 17 -

2.3. Aggregate interlock; cyclic loading.

Colley and Humphrey [9] performed cyclic loading experiments on joints

in plain concrete in pavements. The specimen consisted of two slabs
based upon a subsoil, see Fig. 2.20. The test variables were the joint
width, the load level, the aggregate type and the quality of the sub­
soil. The applied load simulated the approach and departure of a wheel
by subsequently unloading the approach slab and loading the departure
slab. Fig. 2.20b presents the loading rate and joint response in time.
The joint resistance to shear load was expressed by the effectiveness as
defined by Teller and Sutherland [68].:

c-xrr - • - departure slab ,.„ .„, ,„ ,-

c
Effectiveness = — . 100 [%) (2.4)
approach departure

. „ sheor lood [kNl

deflection [mml

Fig. 2.20a. Test arrangement of Colley Fig. 2.20b. Loading rate and slab
and Humphrey [9]. deflection [9].

An effectiveness less than 100 percent indicated that shear slip oc­
curred in the joint. Some test results are shown in Fig. 2.21a-c. It
appeared that the aggregate type influenced the joint effectiveness. The
tests were of the 'high-cycle low-intensity' type with a low shear
stress (0.1-0.2 MPa) and a high number of cycles (up to one million
cycles).
Other experimental work focussed on the 'low-cycle high-intensity' be­
haviour, exploring the response of cracked nuclear containment vessels
subjected to shear. Such tests with a relatively high stress intensity
were conducted by White and Holley [88]. A total of sixteen precracked
specimens was loaded as to transmit shear by the crack roughness. The
 - 18 -

.effectiveness (%] effectiveness 1%I effectiveness I'M

v 6 n = 0.62 mm
~x^=0.10 MPa

crushed aravel

75

K
crushed stone
015 MPa

- 0.90 mm
50 50
natural gravel

\ ^ 1.13 mm
SO.20 MPa
25
•v^JMmm
\2.15mm

0 0.5 1.0 0 0.5 1.0 ' 0 0.5 10

loading cycles N 110'cyctes] loading c y c l e s N [10* cycles] loading cycles N |10« cycles!

a. Influence of j o i n t opening b. Influence of lood level c. Influence of aggregate type

on effectiveness. on effectiveness. on effectiveness.

Fig. 2.21. The joint effectiveness found by Colley and Humphrey [9].

shearing area was 180645 mm 2 . The parameters investigated were the size
and gradation of the aggregate, the normal restraint stiffness provided
by external bars, the shear stress level, the number of cycles and the
initial crack width. The tests were used to try out the test equipment
and to make a first assessment of the crack response to cyclic shear
loading.
On the basis of these results, further tests were performed by Laible,
White and Gergely [37]. The type of specimen used was similar to the
specimen as used by White et al. [88], see Fig. 2.22. Now, the shearing
area was 194000 mm 2 . The concrete cylinder strength for the major series
was 20.7 MPa, the maximum particle size was equal to 38 mm. Apart from
the variables in White's test series [88], the specimen geometry and the
strength and the age of the concrete were varied in the tests. The
specimen was precracked by applying line loads halfway the specimen.
Next, the crack width was set to the desired value of 0.25, 0.51 or 0.76
mm by adjustment of nuts on the restraint bars. The applied shear stress
of 1.24 MPa was fully reversed. Fig. 2.23a-c presents a test result,
which is representative of the generally observed behaviour. The number
of cycles was 25. For the cycles No. 1 and No.15 the load was applied
stepwise, during the other cycles the full load was applied in one step.
The first loading cycle showed a nearly linear relationship between the
crack displacements and the shear stress, whereas this relation became
 - 19 -

■a. * A.
qpp'.ied shear lood

external restraint rods

o o
V concrete specimen

crack plane/
HP ¥
*^J
Fig. 2.22. Test arrangement used by Laible et al. [37].

shear stress in IMRJ shear stress i n IMPal shear stress i q [MPol

n r
' ' ' ' n=1 15

a. Shear stress-shear slip relationship.

K^
b. Shear stress-normal c. Shear stress-crack
stress relationship. width relationship.

Fig. 2.23. Experimental results of Laible et al. [37]

highly non-linear for the later cycles. During unloading the recovery of
the shear displacement was about 20 percent of the maximum slip, which
was probably due to local irreversible deformation of the contact areas.
Fig. 2.23a shows that the stiffness increases with increasing shear dis­
placement, which supports the assumption of deformed contact areas. Due
to the crushing of the matrix material in the previous cycles, the ini­
tial stiffness is very low, because a 'contactless' free slip can occur
before any contact between the opposing crack faces is possible.
Paulay and Loeber [50] carried out both static (see Section 2.2) and re­
peated shear loading tests. Fig. 2.24a-c shows the experimental results
for a maximum shear load of 6 MPa. The crack width was kept constant
during the tests. A surprising result was the low stiffness during un­
loading compared with the stiffness during loading. This result deviated
from the low recovery in shear displacement during unloading found in
the tests of Laible [37]. The major difference between both test series
was the constant crack width in Paulay's tests, where the crack width
 - 20 -

shear stress t a IMPa] shear stress l Q IMPol shear stress t a (MPal

7.0 ■ 1 -—i

a. 6 n = 0.13 m m . b 6n=0.25mm. c 6n= 0.51 mm.

Fig. 2.24. Test results of Paulay and Loeber [50].

increased with increasing shear sliding in Laible's tests. The high nor­
mal stress required to maintain the constant crack width, probably in­
fluenced the unloading of the specimen in Paulay's test series.
In addition to the static test series, Vintzeleou [75] performed cyclic
tests with a fully reversed shear displacement. Due to the large applied
displacements only a few cycles were used. For various normal stresses,
the test results are presented in Fig. 2.25. It was found that for a
high normal stress no degradation of the response occurred. The follow­
ing empirical expression was derived describing the decrease in shear
strength:

= 1 0.12 (2.5)
o 6
n=l a tu

with n = number of cycles and 6 = 2 mm.

tu

Chung [8] carried out impact tests on push-off specimens with a shearing
plane of 18750 mm2, which consisted of a joint between precast and cast
in situ concrete. Apart from a test series with a single impact load, a
test series was performed, in which the specimens were preloaded with a
low intensity shear load during two million cycles. For a load intensity
of 55 percent of the static strength no degradation of the response was
recorded. For an intensity of 66 percent a decrease of 14-20 percent was
observed. It was found that the impact shear strength for a loading rate
of 12000 MPa/s was 80 percent higher than the static a shear strength.
 - 21 -

shear s ress t a [MPa]

n= 1
jhear stress T« IMPal
r t = - 1
2
3
2
T - 30.0.50/2.0 T - 30.2.0/0.5
1 2 2 ^ 5

8 1
_- ,4 -2
u = = r .1 2 . ^ 0 . 8 — r a t " /
?—-^— ^—
shear slip 5. (mm] ,-\ shear slip 6| [rnrnj
7
5 K
-2
2
2 /
r
1 u"'

a. o =0.50 MPa b. ai = 2.00 MPa

Fig. 2 . 2 5 . Test r e s u l t s of Vintzeleou [ 7 5 ] .

 - 22 -

2 . 4 . Dowel action; monotonie loading.

Reinforcing bars crossing a crack will counteract the crack displace­

m e n t s . For bars perpendicularly crossing the crack plane, this response

can be subdivided in an axial and a lateral stiffness of the bar. The

axial stiffness is provided by the bond between steel and concrete. The

lateral stiffness is due to the reaction stresses of the surrounding

concrete and is called dowel stiffness.

Fd
crushing failure .—

y y ^ splitting failure

f ^-
57
a. Splitting failure b. Crushing failure c. Load - displacement
relationship.

Fig. 2.26. Failure modes for dowels.

Several failure modes can occur in dowel action, including splitting

failure of the concrete cover, see Fig. 2 26a. This type of failure ge­
nerally occurs in the case of bottom bars in a beam, when the concrete
cover is too small to make equilibrium with the dowel force. This fail­
ure mode will not be discussed here. If adequate confinement is provided
to prevent splitting of the concrete, concrete crushing around the bar
may occur, see Fig. 2.26b. Now, the concrete reaction force is relati­
vely high with respect to the concrete strength due to the multi-axial
stress condition in the surrounding concrete. Fig. 2.26c presents the
difference in response for both failure modes.
As for the aggregate interlock mechanism, the first dowel tests focussed
on joints in concrete pavements based upon a silt loam subgrade [65,68].
Teller and Sutherland (68] showed that the effectiveness in load trans­
fer of a dowel depends on the slab thickness, the joint width, the dowel
spacing and the load application with respect to the location of the
dowels. From [65] it was found that the slab deflection was directly
proportional to the magnitude of the load on the slab, see Figs. 2.27a-
b.
Paulay, Park and Philips [51] performed dowel action tests on a fixed
corbel, which was connected to a concrete block by means of reinforcing
 - 23 -

deflection 6 Imml
1.25 corner load
1.00
// t Imml s tmmi
203 686
/"

w
corner load edge load. 0.75 228 457
zzi 0.50
025 /
} f-
12192 \
deflection 0 /
25 50 75 100
_x
load IkN!
a. Test arrangement. b. Test results.

Fig. 2.27. Test arrangement and slab deflection as function of the load
and slab thickness [65].

. shear stress TH IMPal

specimen
.lauding. ram for cyclic

0.5 1.0 1.5 2.0 2.5

join!/ Z-Jr.nrh»! shear displacement 5^ Imml

a. Test arrangement b. Shear stress-shear displacement

relationship.

Fig. 2.28. Test arrangement and resuLts of Paulay et al. [51]

bars perpendicularly crossing the smooth contact area, see Fig. 2.28a.
Now, there was no subsoil influencing the response of the dowels to the
shear load. The test results are presented in Fig. 2.28b.
Rasmussen [57] performed tests on dowels perpendicularly protruding from
a large concrete block, see Fig. 2.29a. He found that plastic hinges de­
veloped in the bar accompanied by a considerable crushing of the con­
crete under the bar. The experimentally obtained ultimate dowel force is
presented in Fig. 2.29b and can be expressed by the following relation:

Fdu = C * 2 ■/£ccylif sy [N] (2.6)

with C = 1.3 for the case of no load eccentricity.

<l> in [mm], f , , f in [MPa]

ccyl sy
 - 24 -

ultimate dowel force F^lkN]

.dowel |oad
«251/225

«16/439
£10

■,,m 40

20

350 0 10 20 30 40 50
concrete strength t ccy( [MPal
-t- 6»
a. Test arrangement. b. Dowel strength as function of
the concrete strength.

F i g . 2.29. Test arrangement and r e s u l t s of Rasmussen [57].

dowel force
d. ultimate dowel force Fq^lkNl

/
/
/' i
/ 2919
-?2»16

\sheor plane with plostic 446 ■ 2413

sheet
, ^2»6
iff?""?)/ ) 17S 350 525 700
steel area A s [mm21
a. Test orrongement. b. Dowel strength as function
of the steel area.

Fig. 2.30. Test arrangement and results of Bennett et al. [4].

Rasmussen's test results were in agreement with experimental results ob­

tained by Bennett and Banerjee [4]. The specimen used is shown in Fig.
2.30a. Tests were performed with bottom bars or top bars only and with
the combination of top and bottom bars. For the Lests with bottom bars,
the results showed that the dowel strength is directly proportional to
the cross-sectional area of the bar, see Fig. 2.30b.
In practice, bars cross the crack plane at various angles. Dulacska [15]
explored the influence of the angle of inclination upon the dowel
strength. The specimen used was of the push-off type, in which the ag­
gregate interlock mechanism was prevented by means of two 0.2 mm thick
brass plates, see Fig. 2.31a. The experimental results are shown in Fig.
 25

2.31b. The ultimate dowel force can be expressed by the following empir­
ical relation:

F, = 0.2 f s in(9)
du sy ( '* + 0.03 HinO)* -1 [N] (2.7)
sy

with 0 = angle of inclination (normal to the crack plane 0 = 0°).

f , f in [MPa]
ccm sy

ultimate dowel force FdufcN]

i i

ImmHMPa]
o 10° 10 295
A 10° 6.5 247
□ 10° U 257
& • 30° 10 295
1 ■ 40° 10 295
8
q|\fOQmpfi
< j
Plastics

30 40

a. Test arrangement. b. Dowel strength versus

concrete cube strength.
Fig. 2.31. Testing rig and experimental results of Dulacska [15].

For 0 is equal to 0°, Rasmussen's formula is obtained with C equal to

1.25. The experimentally found shear displacement as a function of the
applied dowel load, can be expressed by:

11.35 10 F, F
d *
tan(-—x)
/ta [mm] (2.8)
FJ 2' f
du ccm

with $, S in [mm], F,, F^ in [N]

t d du

Mills [A7] performed three dowel tests with an angle of inclination of

45°. For a bar with a diameter of 38 mm and yield strength of 210 MPa
and a concrete cylinder strength of 36 MPa, an average dowel strength of
76 kN was obtained.
Utescher and Herrmann [73] performed a large number of dowel tests, ex­
ploring the influence of the bar diameter and load eccentricity upon the
dowel strength. Fig. 2.32a-b presents the experimental results. The load
 - 26 -

eccentricity was varied by applying the load at distances of 5, 10, 20

and 50 mm from the concrete surface. It was found that the load eccen­
tricity strongly influenced the ultimate dowel force. The testing rig
used was very similar to Rasmussen's test arrangement. Therefore, it
must be doubted whether Rasmussen's tests were carried out with zero-
eccentricity, as was reported in [57]. Fig. 2.32b shows that the dowel
strength was proportional to the steel area. Utescher et al. observed a
considerable crushing of the concrete close to the crack plane, see Fig.
2.32c.
ultimate dowel force F^jIkNl ultimate dowel force Fdu [kNl

10 20 30 40 50 100 200 300 «D 500

load eccentricity e [mm] steel area As [mm2]
a. Dowel strength versus eccentricity. b. Dowel strength versus steel area. c. Spalling-olf of the concrete

Fig. 2.32. Experimental results of Utescher and Herrmann [73].

In practice, dowels cross cracks. Therefore, the load eccentricity is

caused by the crack width. Due to the bond between the steel bar and the
concrete, the load eccentricity is accompanied by an axial steel stress.
Eleiott [16] performed dowel tests with pretensioned bars. A cyclic
dowel force was applied. It appeared, that already in the first -static-
cycle the dowel stiffness was strongly decreased by the axial steel
force, see Fig. 2.33. Unfortunately, no detailed information on the
crack width was reported in [16].

cross beom
l^-O- aggr. int. specimen
dowel forceiFrj IkSfl

^external restraint rods

n
greased plates

i dowel action specimen

reinforcing bar 025 050
dowel displacement 6f Imm]

T
Fig. 2.33. Influence of the axial stress upon the dowel force [16],
 - 27 -

Vintzeleou [75] carried out dowel action tests with a reinforced version
of the specimen shown in Fig. 2.17. The bars perpendicularly crossed a
joint of 4 mm, thus preventing the aggregate interlock mechanism. Fig.
2.34 presents the experimental results for a steel yield strength of 420
MPa, showing that the dowel strength is approximately proportional to
the square root of the concrete strength.
r„ ultimate dowel force FdufoN
o «8

•
a *18
••
• •
•
o

0 25 50
concrete s t r e n g t h f , - ^ iMPa]

Fig. 2.34. Experimental results of Vintzeleou [75],

Millard [45] performed dowel action tests, exploring the influence of

bar diameter, concrete strength and axial steel stress upon the dowel
strength.
The testing rig shown in Fig. 2.15 was used. The experimental results as
shown in Fig. 2.35 were in agreement with those of Rasmussen [57],
Bennett [4], Utescher [73] and Vintzeleou [75].

, shear slip 6| [mm]

« 12(mm! p l%l
.20 21L
01 [MPa]
s
0
2
|55L
21L
16 2.13 26L 0.1 175
24L
8 D.53 27L 0.3 3 U
25L
2iL
60
^ - 21L _21l

A
27L
!^6L_
_25L

05 1.0 1.5 2.0 0.5 1.0 1.5 20 "0 0.5 TO

s h e a r displacement 6 t (mm] shear displacement 6^ [mm] crack widthOplmm]

a. Influence bar diameter. b. Influence axial stress. c Crack opening path.

Fig. 2.35. Test results of Millard et al. [45].

 - 28 -

2.S. Dowel action; cyclic loading.

Numerous cyclic dowel action tests are performed at Cornell University.

In [33] experimental results of Eleiott, Stanton and Jimenez [32] are
briefly reviewed. Eleiott [16] carried out tests exploring the influence
of axial steel stresses upon the dowel stiffness. Fig. 2.36a presents a
test result for a bar diameter of 12.6 mm and a concrete strength of 21
MPa. As a result of the steel stress of 175 MPa, the crack width in­
creased, thus reducing the dowel stiffness by up to fifty percent with
respect to a test with an unstressed bar (see Fig. 2.33). In cycle
No.16, the steel stress was increased to 350 MPa, which again strongly
increased the crack width and reduced the dowel stiffness. Stanton [64]
and Jimenez [32] performed tests on large concrete blocks interconnected
by several bars perpendicularly crossing the crack plane. Fig. 2.36b
presents the experimental result for a specimen with four 29 mm diameter
bars. It was found that the energy absorption capacity decreased with
cycling (up to 50 cycles). During these tests, the load was fully re­
versed, showing a similar response in both loading directions.

a. Experiment of Eleiott. b. Experiment of Jimenez. c Experiment of Vintzeieou

Fig. 2.36. Experimental results of cyclic dowel action tests [33,78].

Vintzeleou and Tassios [77,78] performed tests focussing on structures

subjected to earthquakes. As earthquakes cause cyclically imposed dis­
placements, the tests were performed in a displacement-controlled man­
ner. The test arrangement was similar to the one described for the stat­
ic tests, see Section 2.2. Fig. 2.36c presents a test result for a bar
with a cover of 260 mm in the positive direction and a cover of 40 mm in
 - 29 -

the negative direction. Obviously, the response of the bar to lateral

displacements is asymmetrical due to the splitting failure in the nega­
tive direction. The decrease in dowel force at maximum shear displace­
ment can be expressed by the following expression:

d
P
r,
' "=" = 1 --a f^ï
,
(2.9)
d, n=l

with n = number of cycles (n < 7 cycles)

a = 7 for fully reversed loads
14 for repeated loads

2.6. Contribution of axial steel stress.

Reinforcing bars generally cross a crack at different angles. Shear

stress is transferred across the crack by means of the component of the
steel stress parallel to the crack plane. This contribution to the shear
stress transfer can easily be determined when the axial steel stress and
the angle of inclination are known. The magnitude of the axial steel
stress depends upon the bond characteristics.
For bars perpendicularly crossing a crack, the relationship between the
magnitude of the axial steel stress and the crack width is known from
pull-out experiments. However, the bond characteristics obtained in
these tests cannot be applied to the case of bars at different angles to
the crack plane or to bars subjected to both axial and lateral dispace-
ments. Due to the lateral displacement, the bond between the steel bar
and the concrete is broken. Therefore, it is expected that for these
cases the bond capacity will decrease with decreasing angle of inclina­
tion. This was experimentally confirmed by Klein et al. [35], who per­
formed displacement-controlled tests with bars at various angles to the
crack plane, see Fig. 2.37a. Each bar was prepared with strain gauges
stuck to the bar over a length of 360 mm, thus recording the variation
of steel strains over the bond length. Test variables were the bar diam­
eter (10-16 mm) and the angle of inclination (45°,60° and 90°). Some
typical results are presented in Fig. 2.37b, showing that no systematic
variation in bond behaviour for several angles of inclination was ob­
tained in these tests. Due to the lack of proper bond characteristics,
the magnitude of the axial steel stress must be derived from the equi-
 - 30 -

bond stress Tb IMPol

5.0
e= 90°

^0»

ft
-1 v 45°

I t -10 mm

0 003 006
slip £^ [mm]
Q. Test specimen. b. Bond stress versus slip.

Fig. 2.37. Test arrangement and experimental results of Klein et al. [35].

librium condition for the normal force on the crack plane due to the ag­
gregate interlock mechanism (the dowel force is defined here as the bar
force parallel to the crack plane).

2.7. Shear strength of cracked reinforced concrete; monotonie loading.

For design purposes, a simple shear-friction model was introduced by

Mast [39] and Birkeland [5]. According to this model the shear strength
of cracked reinforced concrete was provided by the friction in the shear
plane. The shear strength can then be calculated by multiplying the nor­
mal compressive stress due to the reinforcement by the tangent of the
angle of friction t|>. The ultimate shear stress is reached at the onset
of yielding of the reinforcing bars, thus:

T = pf tan(il)) [MPa] (2.10)

u sy

The angle of friction was empirically derived from tests yielding ty

equal to 55° (tan(t|>) = 1.4) for bars normal to the shear plane. From
tests on corbels [39], it was found that an applied normal tensile
stress could be subtracted from the contribution of the steel pf,
sy'
yielding: (tensile stress has a negative sign)
 31

T = (pf + o ) tan(i|i) [MPa] (2.10a)

u sy n

with T , o , f in [MPa]
u n sy

ultimate shear stress I u IMPa]

too

J'

5.0
uncracked
cracked

0 2.5 5.0 7.5 10.0

mechanical reinforcement ratio pf sy [MPa)
a. Shear strength versus reinforcement ratio. b. Diagonal cracking.

Fig. 2.38. Shear failure of uncracked and cracked reinforced concrete [29],

This was confirmed by tests performed by Hofbeck, Ibrahim and Mattock

[29], who performed an experimental study, exploring the applicability
of the shear-friction analogy. They carried out tests on uncracked and
precracked reinforced push-off elements. Fig. 2.38a shows some test re­
sults indicating that the failure mechanism for the uncracked specimens
was basically different from the failure mode of the cracked specimens.
Fig. 2.38b shows that the uncracked specimens failed forming short diag­
onal cracks across the shear plane. For heavily reinforced precracked
specimens, the shear plane locked up and a similar type of failure was
found. For moderately reinforced precracked specimens, the shear
strength was determined by the response of the crack plane and could be
expressed by the shear-friction analogy. However, it was found that the
angle of friction was 39° (tan(ili ) = 0.8). Also a cohesion was added,
representing the dowel action. Therefore, the average ultimate shear
stress can be expressed by:

T = 2.8 + 0.8(pf + o ) [MPa] (2.11)

u sy n
 - 32 -

w i t h T . o , f i n [MPa]
u' n' sy

. ultimgte shear stress t u |MFa] u l t i m a t e shear s t r e s s * u [MPol

if
0 30 60 90 0 t5 90 135 160
angle 6 [degrees! angle 9 Idegrees]
a. Orthogonal reinforcement. b. Parallel reinforcement.

Fig. 2.39. Experimental results of tests with inclined bars [40].

In further tests, Mattock [40] investigated the shear capacity of cracks

crossed by parallel and orthogonal reinforcement with an angle of incli­
nation to the shear plane. Fig. 2.39a-b presents the experimental re­
sults, showing that there was little influence of the bar inclination
for the specimens with orthogonal bars. However, a strong influence of
the bar inclination on the shear capacity was obtained for the parallel
bars. A maximum was found for 6 equal to approximately 120°. Apart from
tests exploring the restrictions of the shear-friction analogy, Mattock
[41] performed tests to obtain information on the fundamental crack re­
sponse to shear loads. The aggregate type was varied thus yielding a
sand-gravel concrete, a sanded lightweight and an all-lightweight con­
crete. The number of bars perpendicularly crossing the crack plane was
varied yielding reinforcement ratios of 0.4% lo 2.3%. For a concrete
cylinder strength of 28 MPa and an average initial crack width of 0.25
mm, test results are presented in Fig. 2.40a-c. Mattock found that the
sand-gravel and the sanded lightweight concrete exhibited the same crack
opening path, whereas the all-lightweight concrete exhibited a steeper
crack opening path. Therefore, the small particles must have a large in­
fluence upon the crack opening direction.
In another test series, Mattock [42] performed experiments with a ten-
 - 33 -

sile force perpendicular to the crack plane. No systematic differences

in crack opening paths were found for the tensile forces investigated,
see Fig. 2.41.

crack width <Sn Imm) crack width 6nlmm]

Fig. 2.40. Experimental results of tests Fig. 2.41. Results of tests with a
with various mixtures [41]. normal force [42].

Walraven [81,85] conducted tests on push-off elements similar to the

specimens used by Mattock. In addition to the tests on plain concrete
(see Section 2.2), Walraven carried out displacement-controlled tests on
reinforced specimens. Fig. 2.42a presents the specimen used. The test
variables were the bar diameter ranging from 6 to 16 mm, the reinforce­
ment ratio varying between 0.56% and 3.36% and the concrete cube
strength ranging from 19.9 MPa to 56.1 MPa. The maximum particle diame­
ter was 16 mm, except for mix 5, in which a maximum particle of 32 mm
was used. The initial crack width remained small (< 0.1 mm). The steel
yield strength was 460 MPa. The measured crack opening paths appeared to
be insensitive for variations of the bar diameter and reinforcement ra­
tio, see Fig. 2.42b. However, the number of bars strongly influenced the
ultimate shear strength, see Fig. 2.42c.

In addition to the tests on plain concrete and on dowels, Millard [46]

performed push-off tests on cracked reinforced concrete. The test vari­
ables were the initial crack width and the bar diameter ranging from 8
to 16 mm. The steel yield strength was 485 MPa, the concrete cube
strength was in the range of 25.5 MPa to 45.4 MPa for a maximum particle
diameter of 10 mm. The variation of the initial crack width yielded a
 - 34 -

v a r i a t i o n in the i n i t i a l axial s t e e l s t r e s s . The t e s t r e s u l t s are shown

in Fig. 2 . 4 3 a - b . The experimentally obtained crack opening paths showed
a constant angle to the crack p l a n e . However, t h i s crack opening d i r e c ­
t i o n deviated from the average crack opening path found by Walraven, see
d o t t e d l i n e in Fig. 2.43b.

shear stress I IMRil

12.5
crock plane 120 « 300
i i6 37 / y/
3 16 56 / // \ M «

p%E3l ^ *e
10
t
5
16
32
20 ///
38 / A '
j
/
10.0
\v«e

] Hi.

50
OS "V^»»

r ^ 2.S

l ^ 0 05 1.0 0 0.5
Mix 3

1.0
crack width c^lmm) crack width 6n(mm]
a. Test specimen. b. Crack opening path. c Shear stress versus
crack width.

F i g . 2 , 4 2 . E x p e r i m e n t a l r e s u l t s of Walraven [85].

shear slip 6t Iroml

shear stress I |MPa|

0 0.5 1.0 1.5 20 0.5 1.0

shear slip 6, [mm) crack width Öplmml
a. Shear stress-shear slip relationship. b. Crack opening path.

Fig. 2.A3. Experimental results of Millard [46].

 - 35 -

2.8. Shear strength of cracked reinforced concrete; cyclic loading.

Apart from tests on plain concrete and on dowel action, Eleiott [16] and
Jimenez [33] performed cyclic push-off tests on cracked reinforced con­
crete. Fig. 2.44a-c presents some test results of Jimenez, showing that
the crack response to cyclic loading depends on the initial crack width
and the applied shear stress level. It must be noted that an increase in
the initial crack width is accompanied by an increasing axial steel
stress. Fig. 2.44d presents an experimental result, for which the shear
stress was increased in the 15th cycle. It can be seen that the response
in the 15th cycle tended to the static envelope, which would have been
obtained if the shear load reached this level in the first cycle.

sheor'stress I IMPq) shear stress T JMRa] shear stress I MPal shear stress x IMBa]
n=1 15 I 15 115
2445
h
i1 I
075 0.75

-0.375 -0375
/ *7, ' 0.375 ƒ0.375
slip^lmm] slipöjlmml
/ -075 -0.75

IV
a 4*22 .brcf 0.50 mm. b. 4329,6^=0 50 mm. c. 4# Hfbo = 0.2S mm d. Increasing stress level.

Fig. 2.44. Experimental results with cyclic loading of Jimenez [33].

Mattock [43] carried out cyclic tests on the specimen shown in Fig.
2.45a with a reinforcement ratio ranging from 0.60% to 1.32%. The con­
crete strength was approximately 41.6 MPa for the normal weight concrete
and 28.3 MPa for the lightweight concrete with a maximum particle diame­
ter of 16 mm and 13 mm respectively. Some schematic presentations of the
relations between the shear stress and the shear displacement are shown
in Fig. 2.45b. During unloading the response was characterized by a re­
tention of almost all the shear displacement under maximum shear stress
until the shear stress was reduced to approximately 50 percent of its
maximum value. In all precracked specimens a decrease in crack width at
zero shear stress was observed. This decrease accounted 0.08-0.13 mm
with respect to an initial crack width of 0.25 mm. This crack width
 - 36 -

remained constant until shortly before failure. Fig. 2.46a-b shows some
experimental results for both normal weight and lightweight concrete.

shear stress x
shear plane 25t » 127 cvcle shortly before failure

'I
-> (h=— i n * = j i = ! s 1

1
1 I
1
i
II
i I fsheor displacement fy

L__
U\ 1
f /intermediate
1 / cycle
L «*
»—u190.5 d

a. Test specimen. b. Schematical presentation of the shear stress

versus shear displacement.

Fig. 2.45. Test specimen and experimental results of Mattock [43].

shear load f IkNl shear load F IkNl

250 250

200 200
AO O 1

— monotonie monotonie Hl&P

o cyclic o cyclic
- U^t - ^2 L MPn tccfl-- 28.9 MPa
fsy = 375 MPa L.y = 455 MPa
p = 0.( 8 % p = 06 %
0 I 2 0 1 2
shear displc
displacement 6, Imm] shear displc
a. Sand/gravel concrete b. Lightweight concrete.

Fig. 2.46. Shear stress versus shear displacement for tests of Mattock [43].

The result for a cyclic loading test is compared with the static test
performed with a similar specimen. Fig.2.46a shows Lhat for the case of
normal gravel concrete, the maximum slip during cycling with a low shear
stress level was approximately equal to the shear displacement occurring
in the monotonie test at the same stress. However, at a shear stress
level of 80 percent of the static shear strength, the slip rapidly in­
creased. The same held true for the crack width.
For the cyclic test on lightweight concrete, the shear slip was larger
and the crack width was smaller than those occurring in the monotonie
test at the same shear stress.
 - 37 -

2.9. Conclusions.

With regard to the mechanism of aggregate interlock, it was found that a

parallel displacement of the crack faces does not only generate shear
stresses o ( T ) but also normal stresses o . The dependency between
nt nn '
stresses and displacements can be expressed by the following relation:

a S
l 1 S
12 (2.12)
nn
=
°nt
s21 s 2 2

The static shear strength of plain concrete was found to be proportional

to the square root of the concrete compressive strength. Furthermore,
the type of aggregate influences both the shear strength and the crack
opening path (for a constant normal restraint stiffness). The maximum
particle diameter, ranging from 8 to 32 mm hardly influences the static
shear strength, but affects the crack opening direction.
For cracks in plain concrete subjected to cyclic loading, it was found
that there is a basic difference in the behaviour in the first and in
the subsequent cycles. The response in the subsequent cycles becomes
highly non-linear with an initially low stiffness, which is abruptly
changed in a high stiffness approaching the maximum shear slip obtained
in the previous cycle. The shear stress level was found to be an impor­
tant parameter.
Most of the experimental work focussed upon the response to a high in­
tensity loading. No tests, with a very large number of cycles (10 3 -10 6 )
were carried out yet.
The static shear stress-shear displacement relationship for dowel action
is characterized by a linear line up to approximately 40 percent of the
dowel strength. Next, the relation between dowel displacement and dowel
force becomes nonlinear until the maximum dowel capacity is reached.
Now, the concrete strength largely influences the ultimate dowel
strength. The type of failure was found to be characterized by the for­
mation of plastic hinges in the bar and local crushing of the concrete
under the bar.
The dowel strength is strongly influenced by the initial crack width and
the initial axial steel stress, although these parameters might inter­
fere.
 - 38 -

As for the aggregate interlock mechanism, the dowel response to cyclic

loading becomes highly non-linear for subsequent cycles with respect to
the response in the first cycle, which is approximately linear.
All dowel action tests found in literature were of the 'high-intensity
low-cycle' type.
For the contribution of the axial steel stress to the static shear
stress transfer for inclined bars, it was found that the ordinary bond
characteristics obtained in pull-out experiments can not be applied to
the case of inclined bars or bars subjected to the combined action of
axial and lateral forces. For bars subjected to a lateral displacement,
the bond between the bar and the concrete is broken at one side of the
bar, thus decreasing the over-all bond capacity.
The angle of bar inclination and the number of bars were found to be
important parameters for the static shear strength of cracks in rein­
forced concrete.
The type and size of the aggregate have some influence upon the crack
opening direction, whereas no influence of the number of bars was found.
The response of cracked reinforced concrete to cyclic shear stresses is
influenced by the initial crack width and the shear stress level. As for
the tests on plain concrete and on dowel action, the available experi­
mental knowledge is restricted to the range of high-intensity loads.
Therefore, it can be concluded that there is still a lack of experimen­
tal information on the response of cracked concrete subjected to a very
large number of load cycles with a relatively low shear stress with re­
spect to the static shear strength.
 - 39 -

3. EXPERIMENTAL STUDY

3.1. Introduction.

The survey of the literature yielded the conclusion that there still is
a lack of experimental information with respect to the response of
cracks in concrete to 'low-intensity high-cycle' loading. Therefore, an
experimental study focussing on 'low-intensity high-cycle' fatigue of
offshore structures was started in the Stevin laboratory. For offshore
structures, the Arctic and deep sea environments provide intense dynamic
forces with a loading frequency of approximately 1 cycle per second
[25]. In a substructure placed on the seabed, there generally is a stat­
ic load apparent, to which the cyclic load is superimposed. In conse­
quence, the shear walls in the base are predominately subjected to re­
peated loads. Therefore, the first test series was performed with a re­
peated shear load on precracked push-off specimens. In the second test
series, the dowel action was eliminated by using external restraint
bars, enabling a quantification of the contribution of the aggregate in­
terlock mechanism to the shear transfer. The magnitude of the contribu­
tion of the dowel action is then known. These experiments will be brief­
ly described in this Chapter. A full survey of all the experimental re­
sults has been given in [56].

3.2. Reinforced specimens; repeated loading.

3.2.1. Test arrangement

The geometry of the test specimen in this experimental program was simi­
lar to the push-off specimen used by Walraven in his static tests [81].
The shear area was 36000 (120 x 300 m m 2 ) , see Fig. 3.1. Previously,
shear tests were performed on specimens with a similar shear area, so
the test results can be compared without scale-factors. The specimens
were cast in a steel mould placed horizontally, so that at the time of
casting the shear plane was in a vertical position. The cantilevers of
the specimen were prestressed preventing preliminary failure of these
cantilevers due to secondary cracking. Prior to the actual shear test,
 - 40 -

the crack was made in a three-point bending test by pushing a steel

knife-edge into a V-shaped groove along the shear plane. Both sides of
the specimen were subsequently cracked in this manner, resulting in an
initial crack width ranging from 0.01 mm to 0.08 mm. Next, the specimen
was centred in the test frame, see Fig. 3.3. The specimen was supported
by a roller bearing thus preventing restraining forces being transmitted
during crack opening. The shear load was provided by means of a 1000 kN
hydraulic jack placed on the foot of the frame. A knife hinge induced
the load at the top of the specimen.

prpstressina duct

»K
as
^
^
—
i»8
£

300* 120 mm? I

S\\
:%^==M-
200 200

Fig. 3.1. Test specimen.

crack plane |

Fig. 3.2. Measuring system. Fig. 3.3. Testing rig.

The crack displacements were recorded by means of linear voltage dis­

placement transducers, attached to steel footings glued on the concrete
on both sides of the specimen, see Fig. 3.2. The transducers, Hewlett-
Packard type 7-DCDT-100, had a 0.01 mm measuring accuracy at 5 mm range.
The shear load was measured by means of a load cell with a measuring ac-
 - 41 -

curacy of 0.25 kN. All the signals were led to a micro-computer for
storage and monitor display. In order to reproduce the sinusoidal sig­
nals each measured cycle was scanned nine times. A trigger level was
adjusted to the maximum load by means of a special circuit. By sampling
this trigger level it was possible to start the first scan after each
call on the peak value of the applied load.
During the actual test, the specimen was subjected to a shear load,
which alternated between a minimum shear stress level t and a maximum
o
shear stress level T . The crack displacements were not recorded for the
m
first few cycles due to the adjustment of the shear stress levels, the
load frequency and the trigger level.

3.2.2. Test variables.

The test variables were the reinforcement ratio and steel yield
strength, the concrete compressive strength, the initial crack width,
the number of cycles and the applied shear stress level.
The normal restraint stiffness depends on the reinforcement ratio p. For
the tests, four and six 8 mm diameter stirrups were used, yielding a
reinforcement ratio of 1.12% and 1.68% respectively. The steel yield
strength of the ribbed bars f was 460 MPa (denoted low-strength) and
550 MPa (denoted high-strength) with a rib coefficient f„ (approximately
the rib heigth/rib distance) equal to 0.050 and 0.059 respectively. The
use of two steel grades provided an opportunity to investigate whether
the reinforcement yielded or not.
The concrete grade f had an average 28-day cube crushing strength of
50 MPa (Mix A) and 70 MPa (Mix B) reflecting the high-strength concrete
used in offshore structures. It can be expected that with increasing
concrete strength an increasing number of particles is fractured during
cracking of the concrete, thus reducing the aggregate interlock mecha­
nism. Both mixes had a maximum particle diameter of 16 mm and almost
complied with the Fuller grading curve. Detailed information is given in
Appendix I.
The initial crack width & varied from 0.01 mm to 0.08 mm to ensure a
no
small crack width (< 0.25 mm) in order to simulate offshore service con­
ditions. For the tests, the initial crack width was not an adjusted, but
a measured parameter.
 - 42 -

The number of cycles ranged from 118 to 931731 cycles for the test se­
ries. The large number of cycles interferred with a good planning of the
tests, in general the experiments did not start at an age of 28 days.
Therefore, the concrete strength at the start of the test was obtained
from Fig. 3.4.

relative concrete strength tcnJ" ' tern' 2 8 '

MhsJL
Mix B /
"5^*^"^

OS

56 84 112
age t Idays]

Fig. 3.4. The concrete strength versus the concrete age.

The applied shear stress level is referred to the static shear strength
obtained in the static tests of Walraven [81]. According to the shear
friction analogy, the shear strength can be expressed as a function of
the yield strength and the reinforcement ratio [84]:

T = a(pf )c (3.1)
u sy

0 106
with a = 0.822 f
ccm
0. 303
b = 0.159 f
ccm
T , f , f in [MPa]
u ccm sy

During the tests, the shear stress level i was in range of 46 to 90

m
percent of the static strength and a minimum shear stress T equal to
0.3 MPa. Despite of the shear stress of up to 90 percent of the static
strength, the tests were still of the 'low-intensity' type because of
the very small crack width.
 - 43 -

3.2.3. Experimental results.

A total of 42 repeated push-off tests was carried out. The specimens

have been asigned an identifying code consisting of 5 characteristics.
The first character denotes the concrete grade (Mix A or Mix B ) , the
second the number of 8 mm diameter stirrups and the steel yield strength
(Low or High). The next character represents the shear stress level
T /T , followed by the actually applied shear stress. Finally, the
initial crack width forms part of the identifying code.
The Tables 3.1 and 3.2 list a review of the experimental results.

Table 3.1. Test results of Mix A.

Code f T X T /T No. of failure

ccm 2 u 2
m m u
[N/mm ] [N/mm ] cycles during cycl ing

A/4L/.61/6.1/.03 54.47 10.00 6.1 .610 455000 no

A/4L/.63/6.0/.06 50.20 9.47 6.0 .634 263337 no
A/4H/.64/7.0/.06 54.30 10.97 7.0 .638 368000 no
A/4L/.65/6.0/.01 46.83 9.29 6.0 .646 34000 no
A/4H/.66/6.9/.02 50.99 10.50 6.9 .657 550000 no
A/4L/.70/7.0/.01 54.49 10.00 7.0 .700 2410 no
A/4L/.73/7.2/.05 53.70 9.90 7.2 .727 14000 yes
A/4L/.74/7.0/.01 49.85 9.42 7.0 .743 435 no
A/4L/.76/7.0/.02 48.20 9.20 7.0 .761 5925 yes
A/4H/.76/7.7/.03 48.50 10.15 7.7 .759 4762 yes
A/4L/.77/7.2/.04 49.48 9.38 7.2 .768 1785 yes
A/4H/.78/8.0/.04 49.30 10.26 8.0 .780 5198 yes
A/4L/.79/8.6/.02 53.47 10.86 8.6 .792 895 yes
A/4L/.80/7.3/.03 47.20 9.09 7.3 .803 1192 yes
A/4L/.80/7.5/.05 49.60 9.39 7.5 .799 299450 yes
A/4L/.82/7.4/.05 46.43 8.99 7.4 .823 996 yes
A/4L/.90/9.0/.05 54.50 10.00 9.0 .900 118 yes

A/6L/.51/6.0/.04 51.37 11.89 6.0 .505 508000 no

A/6L/.56/6.7/.05 51.41 11.90 6.7 .563 386000 yes
A/6L/.58/6.8/.01 50.40 11.73 6.8 .578 194000 no
A/6L/.61/7.2/.04 51.30 11.88 7.2 .606 160300 no
A/6L/.62/8.0/.04 57.30 12.89 8.0 .621 410781 no
A/6H/.66/7.9/.03 45.10 11.84 7.9 .663 1750 yes
A/6L/.66/8.6/.02 58.60 13.10 8.6 .656 40722 yes
A/6L/.67/8.2/.08 53.20 12.20 8.2 .672 21000 yes
A/6L/.68/8.0/.03 51.04 11.83 8.0 .676 1000 yes
 - 44 -

Table 3.2. Test results of Mix B.

Code f T T T /l No. of failure

ccm U
m u
m
[N/mm2] [N/tim2] cycles during cycling

B/4L/.57/7.0/.03 73.54 12.27 7.0 .570 665000 no

B/4L/.59/7.0/.20 69.90 11.85 7.0 .591 1331 yes
B/4L/.60/7.0/.06 69.46 11.74 7.0 .596 512660 no
B/4L/.60/7.4/.08 73.54 12.27 7.4 .603 82500 yes
B/4L/.61/7.3/.04 70.80 11.95 7.3 .611 23500 yes
B/4H/.61/8.5/.04 75.34 13.89 8.5 .612 355716 no
B/4L/.63/7.3/.04 67.89 11.61 7.3 .629 931731 yes
B/4L/.65/8.0/.07 73.47 12.26 8.0 .653 62000 yes
B/4H/.66/9.0/.04 74.00 13.69 9.0 .657 2224 yes
B/4L/.75/8.4/.05 65.10 11.28 8.4 .745 219029 yes
B/4L/.79/8.8/.08 63.88 11.11 8.8 .792 1150 yes
B/4L/.81/9.1/.04 64.70 11.23 9.1 .810 1441 yes

B/6L/.46/6.9/.04 70.70 15.11 6.9 .457 550000 no

B/6L/.52/7.9/.02 71.70 15.26 7.9 .518 290000 no
B/6L/.53/8.0/.06 71.40 15.21 8.0 .526 250000 no
B/6L/.56/8.9/.02 75.10 15.81 8.9 .563 325900 no

Because of the large number of load cycles, it is hardly possible to

show the crack response for each measured cycle. Fig. 3.5 presents a
typical relation between the crack displacements in a specific cycle as
a function of the shear stress related to the maximum applied shear
stress. For this test, a maximum of 1785 cycles till failure was ob­
tained. Fig. 3.5 shows that even for this relatively small number of
cycles the increase in crack displacements between two subsequent cycles
remained smaller than the 0.01 mm accuracy of the displacement transduc­
er.

, shear stress/moximum shear stress T/Xm

025 0.50 075 1.00

crack displacement 6 Imm]

Fig. 3.5. Typical crack response during cycling.

 - 45 -

Furthermore, it was observed that the shear displacement increase was

initially smaller than the crack width increase, but exceeded the crack
width increment with increasing crack displacements.

crack width on (mmj shear displacement h\ [mm!

1.00 1.00
/
/ /
0.75 / ƒ 0.75 / /
/ /
/ - A/4H/.78/8.0/0.04//
A/ 4LA77/7.2/0.0 4 / / A/tH .78/80/004
/ /AHU.77n.2IOOi
0.50 050 7
/ //
0.25 025
— *"" ■

n /
10' I0 1 103 10' XT 10° 0 0.25 0.50 075 1.00
number of cycles login) [cycles! crack width o^mm]
a. Crack width versus number of cycles. b. Crack opening path.

Fig. 3 . 6 . Influence of the s t e e l yield s t r e n g t h ,

The influence of the steel yield strength upon the crack response during
cycling is shown in Fig. 3.6a-b. The specimens Nos. A/4L/.77/7.2/.04 and
A/4H/.78/8.0/.04 were subjected to a nearly equal percentage of the
static shear strength (77% and 78% respectively). Both specimens exhib­
ited a similar behaviour with respect to the crack displacement increase
during cycling and followed an identical crack opening path. This indi­
cates. that the influence of the steel yield strength can be properly
taken into account by eq. (3.1) for the static strength. The actual max­
imum shear stress was 7.2 MPa for specimen No. A/4L/.77/7.2/.04 and 8.0
MPa for specimen No. A/4H/.78/8.0/.04 with a calculated static shear
strength of 9.38 MPa and 10.26 MPa respectively.

Fig. 3.7a-b presents the effect of the variation of the reinforcement

ratio. It was found for this parameter, that specimens with a different
number of bars exhibited a nearly similar response to cycling. The
specimens shown in Fig. 3.7, Nos. A/4L/.61/6.1/.03 and A/6L/.61/7.2/.04,
had four and six 8 mm stirrups respectively and were both loaded at 61
percent of their static strength. Therefore, it can be concluded that
the influence of reinforcement ratio is accounted for by the eq. (3.1).
It is obvious that an increasing concrete strength yields a stiffer
 - 46 -

crack width 6n [mm] sheof displacement 6( |mm]

1.00

0.75

050 050

A/6U.61 7.2/.04 - AML/.61/6.1/.03

0.25 025
V /
,^i^Z- ^\A(4U6)/6.1/.03
V7A/6L/.61/7.2/0i

/ 1 1
10' 102 103 10' O5 10* 0.25 0.50 0.75 1.00
number of cycles login) Icyclesl c r a c k width ö^mm]

o. Crack w i d t h versus number of c y c l e s . b. Crack o p e n i n g path.

Fig. 3.7. Influence of the reinforcement ratio.

crack w i d t h Op Imml sheor d i s p l a c e m e n t 6 j |mm)

1.00

075 0.75

//
A/4L/.65
^5^ BKL/.6S/8.0/.C 1
0.25
— - " * " " _——"1

0.50 0.75 100

10' 10' 10 3 10' O5 K)
number of cycles log(n) [cycles] c r a c k w i d t h Onimm]

a. Crock w i d t h versus number of c y c l e s . b. Crack o p e n i n g path.

Fig. 3.8. Influence of the concrete grade.

crack response to cyclic loading. In Fig. 3.8a-b it is shown that also

the effect of the concrete strength upon this response is satisfactorily
accounted for by the influence of the concrete strength upon the static
shear strength. The specimens Nos. A/4L/.65/6.0/.05 and B/4L/.65/8.0/.07
exhibited a similar relation between the crack width and the number of
cycles and a similar crack opening path. It must be noted that the rela­
tion between the shear displacement and the number of cycles had the
same shape as the relation shown in Fig. 3.8a.
 - 47 -

1 0 0 crack w i d t h 6n (mmt sheor displacement 6( I m m l

10' 10* 10J 10' tO5 I06 0.50 0.75 1.00

number of cycles log(n) [cyctesl c r a c k width 6 ^ [mm]
o. Crack w i d t h versus number of cycles b. Crack opening path.

Fig. 3.9. The influence of the initial crack width

Contrary to the previously mentioned parameters, no influence of the

magnitude of the initial crack width was found within the range investi­
gated. The specimens Nos. A/6L/.66/8.6/.02 and A/6L/.67/8.2/.08 had a
difference in the initial crack width of 0.06 mm. Both specimens fol­
lowed nearly the same crack opening path except for the small crack dis­
placements.
For mix A, the Fig. 3.10a shows the relationship between the crack width
and the number of cycles for various shear stress levels. Similar rela­
tions were obtained for the shear displacement as a function of the num­
ber of cycles, see Fig. 3.10b. The shear stress level is referred to the
static shear strength according to eq. (3.1). Fig. 3.11 presents similar
results for mix B.
As expected, the increments of the crack displacements increased with
increasing shear stress level. The influence of the parameters investi­
gated .in this test series are satisfactorily taken into account by means
of the magnitude of the static shear strength. Therefore, the relations
presented in the Figs. 3.10-3.11 can be approximated by the following
empirical expressions, see Fig. 3.12:

T T T T
« = 0.10(—)2 + [0.15(—)3 + O.A0(—)9Jlog(n) + 0.20(—)* 2 [log(n) ]'»
n T T T T
(3.2a)
 - 48 -

T T T T
6 t = 0.10(^2)2 + [0.07(—)3 + 0 . 7 0 ( ^ ) 9 ] l o g ( n ) + 1 . 1 7 ( ^ ) 2 6 [ l o g ( n ) ] 5
(3.2b)

with T according to eq. (3.1) and 6 , 6 in [mm].

u t n

. crack width 8n Imml shear displacement 5t Imm

x
u *

/o9° L la.il

l;
'073

MI

s*7^
10* W* O* O4 nl «'
number of cycles login) (cycles) number of cycles login) [cycles)

a. Crack width versus number of cycles. b. Shear displacement versus number of cycles.

Fig. 3,10. The influence of the applied shear stress level for mix A.

. crock width 6n |mm| -rock displacements o (mm)

1
»n
— - 6,

075
i f ■
i f
i 1
t f
t 1
0.50 j /
i

•
/ »07
_^>£*
025

^rT-
■^^ZlS— r r r m —"2.0 w
■ZZZ^L
10' O* 0
number of cycles login) Icyclesl number of cycles login) (cycles)

a. Crack width versus number of cycles. a. Crack width versus number of cycles.

Fig. 3 . 1 1 . The influence of the applied Fig. 3.12* The crack response
shear s t r e s s l e v e l for mix B. a c e . to eq. ( 3 . 2 ) .

In the expressions (3.2) for the crack displacements, the minimum shear
stress T is not apparent. One should assume that the ratio of the ap-
 - 49 -

plied maximum shear stress x to the applied minimum shear stress T

m o
will strongly influence the relations expressed by the eqs. (3.2a-b).
However, no information on this variable was obtained in this test se­
ries.

shear stress level W'u

too
■.
l
1 ■
075 *■ *
■
<'
^ê
♦A
* ' %
0.50 ^ T
Mix stirrups failure no failure V-~
A 4 ■ o—
A 6 ♦ c—
0.25
B U A A-
B 6 V-
n
10' ioJ to' to 5 to'
number of cycles till failure logtnJicycles]

Fig. 3.13. The shear stress level versus the number of cycles till failure.

Fig. 3.13 presents the relationship between the applied shear stress
level and the number of cycles till failure. Despite of the large scat­
ter, the following mean relation was derived by means of a regression
analysis:

T
■— = 1.00 - 0.07361og(n ) (3.3)
u

The specimens, which did not fail during cycling, were subsequently
sheared-off in a static test. Table 3.3 lists the ratio of the shear
strength according to eq. (3.1) to the experimentally obtained strength.
An average ratio of 1.01 with a coefficient of variation of 0.08 was
found. Therefore, it can be concluded that pre-loading with a low shear
stress has no measurable influence upon the crack response to higher
static shear loads. A similar conclusion can be drawn with respect to
higher repeated shear loads. The specimens Nos. B/4L/.57/7.0/.03 and
B/6L/.53/8.0/.06 were firstly loaded to 57 and 53 percent of their
static strength respectively. Both specimens endured more than 250000
cycles. Subsequently, they were subjected to a repeated load of 65 per-
 - 50 -

cent of their static strength. Shear failure occurred after 88500 and
20631 cycles respectively. These numbers of cycles were in reasonable
agreement with the 57500 cycles according to eq. (3.3).

Table 3.3. Comparison of the theoretical and the experimental static

shear strength.

Code T T
U u,th u u, l h
[N/mm2] [N/mm?]

A/AL/.61/6.1/.03 10.00 10.17 0.98

A/AL/.63/6.0/.06 9.A7 10.10 0.9A
A/AH/.64/7.0/.06 10.97 11.77 0.93
A/AH/.66/6.9/.02 10.50 12.AA 0.8A
A/6L/.51/6.0/.0A 11.89 12.10 0.98
A/6L/.58/6.8/.01 11.73 11.21 1.05
A/6L/.61/7.2/.0A 11.88 12.30 0.97
A/6L/.62/8.0/.0A 12.89 12. A8 1.03
B/AL/.60/7.0/.06 11.7A 10.60 1.11
B/AH/.61/8.5/.0A 13.89 12.66 1.10
B/6L/.A6/6.9/.0A 15.11 13.50 1.12
B/6L/.52/7.9/.02 15.26 IA.27 1.07
B/6L/.56/8.9/.02 15.81 15.99 0.99

average ratio = 1 .01, s = 0 .08

 - 51 -

3.3. Externally reinforced specimens; repeated loading.

3.3.J. Test arrangement

Contrary to the previous test series, the normal restraint stiffness was
not applied by means of embedded reinforcement, but by means of four
external restraint bars. This test series comprised 14 repeated loading
tests focussing upon the aggregate interlock mechanism. The specimen is
shown in Fig. 3.14.

shear load

prestressinq strand

=d
Fig. 3.14. Test specimen with external restraint bars.

The dimensions of the specimen were the same as in the previous series.
Now, no bars crossed the crack plane. The auxiliary reinforcement was
still apparent preventing preliminary failure of the specimen. At the
small sides of the specimen steel plates were placed interconnected by
four 20 mm diameter bars. A thin layer of rapidly hardening sand-cement
paste placed between the steel plates and the concrete surface of the
specimen ensured an almost linear interaction between crack-opening and
restraint force. However, the restraint stiffness remained low compared
with the specimens having embedded reinforcement. To ensure a small
crack width during the first cycles all the specimens were prestressed
with an initial normal stress on the crack plane ranging from 0.8-3.6
MPa.
The instrumentation used in this part of the experimental program was
nearly the same as used for the reinforced specimens, see Section 3.2.1.
The addition made for this test series was the recording of the steel
strains of the external restraint bars by means of strain gauges.
 - 52 -

3.3.2. Test variables

The concrete strength, the initial crack width, the number of cycles and
the applied maximum shear stress level were variables as already de­
scribed in Section 3.2.2. Now, the initial normal stress was an addi­
tional variable instead of the reinforcement ratio for the specimens
with embedded reinforcement. The initial normal stress was relatively
high (0.8-3.6 MPa) to ensure very small crack widths. Unfortunately,
Walraven [81] performed static tests on similar specimens without such a
high initial normal stress. Therefore, the static shear strength ob­
tained in his tests provided only global information for the static
shear strength in the present test series. Furthermore, for plain con­
crete no actual shear failure occurred due to the increasing normal
stress. In consequence, no shear stress level could be determined
without performing static shear tests with a high normal stress. Because
of the better fundamental insight in the mechanism of aggregate inter­
lock as a result of Walraven's theoretical work (see Chapter A and
[81]), it was decided to perform repeated loading tests with various
maximum shear loads in spite of a lack of knowledge about the actual
static shear strength. As for the previous test series, the applied mi­
nimum shear stress was 0.3 MPa.

3.3.3. Test results

A total of 14 tests was performed. The identifying code, which was

assigned to the specimens consisted of the concrete grade (mix A or mix
B ) , the initial normal stress, the applied maximum shear stress and the
initial crack width. Contrary to the experiments on specimens with em­
bedded reinforcement, the applied maximum shear stress T was not the
m
initially applied shear stress. Due to the fact that for low shear
stress levels no significant crack displacements were recorded, the
applied shear stress was raised to such a level that significant crack
displacements occurred. This stress level was then subsequently main­
tained and is adopted in the code of the specimen. Table 3.4 lists a
review of the experimental results.
 - 53 -

Table 3.4. Experimental results of plain concrete specimens.

Code f 0 T No. of failure

ccm 0 m
[N/mm2] [N/mm2] cycles during cycling

A/1.3/4.G7.12 51.53 1.30 4.0 30800 yes

A/1.2/5.0/.01 52.06 1.22 5.0 20326 no
A/1.3/5.0/.01 54.60 1.27 5.0 3080 yes
A/1.9/5.0/.19 48.40 1.90 5.0 9756 yes
A/0.8/5.5/.01 49.56 0.80 5.5 1520 yes
A/2.1/6.1/.01 54.53 2.14 6.1 8640 yes
A/1.3/6.2/.01 48.09 1.34 6.2 283549 yes
A/1.3/6.2/.02 52.57 1.26 6.2 3120 yes

B/l.1/5.0/.04 71.64 1.07 5.0 254369 no

B/1.2/5.6/.01 68.36 1.15 5.6 89500 no
B/l.0/6.2/.04 72.25 0.99 6.0 2736 yes
B/2.0/6.5/.01 69.14 2.02 6.5 343 yes
B/3.6/6.9/.19 70.20 3.58 6.9 82378 no
B/l.5/7.7/.01 67.54 1.99 7.7 3970 yes

Fig. 3.15 presents some typical relations of the crack displacements

versus the shear stress related to the applied maximum shear stress,
showing the crack response during a cycle. As for the reinforced spe­
cimens, the hysteresic loop indicated that dissipation of energy oc­
curred in the shear plane. Again it was observed that the shear dis­
placement increment was initially smaller than the crack width incre­
ment. For increasing crack displacements, the shear slip increment
exceeded the crack width increment. In fact, no shear failure occurred
during the repeated tests, because even for large crack displacements
(6 > 2mm) the shear plane was still capable of transferring the applied
shear load. This was probably due to the increase in normal force with
crack-opening. The specimens were unloaded when the shear displacement
exceeded 2 mm.
The influence of the concrete strength upon the shear stress was propor-
0,56
tional to fccm for the static tests of Walraven [811. If this proporti-
r r
onality could be applied to the case of a repeated shear loading, this
must be reflected in the range of the applied maximum shear stress T .
The range for which significant crack displacements were recorded, was
5.0-6.2 MPa for mix A and 5.6-7.7 MPa for mix B. The average ratio of
the range for mix A to the range for mix B was equal to 1.18, which was
0.56
in good agreement with (70/51) " = 1.19.
 - 54 -

sheor stress/maximum shear St ess T/x m ,. normal stress Oh IMFta]

1.00
JffilQA 234Q h i, JQm
Specimen No.
i /
075
irn
It i
II i
i
/
/
l
tl
i
i
A/1.9/5.0/.I9
l „ r 5.0 MFta
ii / Ji i
tem ' 8 ' MPo
1

025
il
f M' li
i
o-no= 1.90 MPo
— = 6„
■i i
it ii
M
1 / i'
n
0.25 0.50 0.75 100 050 0.75 1.00
crack displacement 6 Imml crock width 6n[mm]

F i g . 3 . 1 5 . Crack response during Fig. 3.16. The normal stress versus

cycling. the crack width.

The relationship between the normal stress and the crack width was in
the same range for both mixes, see Fig. 3.16.
The influence of the magnitude of the initial crack width upon the crack
response to repeated shear loading is shown in Fig. 3.17a. The specimens
Nos. A/1.3/5.0/.01 and A/1.9/5.0/.19 had an initial crack width equal to
0.01 mm and 0.19 mm respectively.
crack width 6n Imml . normal stress dn IMRal

AH.9/5.0/.I9

050

/ A/1.3/50 fll

10' 10' »' W' 0.50 • 0.75 100

number ot cycles login) [cycles) crack width 6n|mm]

o. Crock width versus number of cycles. b. Normal stress versus crock width.

Fig. 3.17. The influence of the initial crack width.

The last mentioned specimen had a very large crack width during pre-
cracking due to a bad fitting of the steel restraint plates to the con­
crete surface. Although a high normal stress was applied to the crack
plane, it was not possible to completely re-close the crack. Therefore,
this specimen had an initial crack width, which was not in the range re-
 - 55 -

fleeting offshore conditions. However, the result is of interest for the

influence of the initial crack width. The normal stress as a function of
the crack width was similar for both specimens, see Fig. 3.17b. Fig.
3.17a. shows that the. initial crack width hardly influenced the crack
response for the range investigated. The influence of the normal stress
is shown in Fig. 3.18. The specimens Nos. A/0.8/5.5/.01, A/1.3/5.0/01
and A/2.1/6.1/.01 had initial normal stresses of 0.8 MPa, 1.3 MPa and
2.1 MPa respectively. There was a slight difference in the applied shear
stress. Fig. 3.18a shows that the increase in crack width with cycling
was similar for both the specimens with the high initial normal
stresses. The specimen with the relatively low initial normal stress
exhibited initially smaller crack widths, which increased much more rap­
idly than for the other specimens. Obviously, an increase of the initial
normal stress up to about 1.3 MPa influenced the crack response. A fur­
ther increase of the normal stress however, had little effect on the in­
crease in crack width. The effect of the normal stress upon the crack
opening path was only slight. Fig. 3.18b shows that there was no system­
atic variation of the crack opening path for the initial normal stress
investigated.
sheer displocement 6) IrrtmI
"•»•" ™'"'" "n
!

A/I.3/S.0/.0I . 1
i i A/2.V6.I/.0I
< r
0.50 // !i
A;O8/5.5/.0I / |

J/
^P /

0.50 0.75 100

10' 10' »' »'
number of cycles login) (cycles] crack width OnEmm]
a. Crack width versus number o( cycles. b. Crack opening path.

Fig. 3.18. The influence of the initial normal stress.

The influence of the magnitude of the applied shear stress T is shown

in Fig. 3.19a-b, representing the crack width versus the number of cy­
cles for both mixes and various shear stresses. Contrary to the spe­
cimens with embedded reinforcing bars, the increase in crack displace-
 - 56 -

ments appeared to be hardly affected by the applied shear stress. An

exception was obtained for the specimen with the relatively low initial
normal stress of 0.8 MPa, as is already shown in Fig. 3.18.

crack width 6n Imml crock width on Imm)

1.00

B/1.5/77/.0I
' n r 7.7MRJ

B/3.6/6.9/.I9
"E9MRÏ

B/1.2/5.6/.0I . ■ — 5"6 MPö

10' 10! 101 10' O5 10' 10' 10' 105

number of cycles log In) [cycles! number of cycles log (n) (cycles!.
b. Mix B

Fig. 3.19. The influence of the applied maximum shear stress.

sheof displocement 6^ !mml sheor displocement b\ |mm]
1.00

i M!
Mix A i Ms
/
/
/,
0.50 1 A

0.25

0.25 0.50 0.75 1.00 0.25 050 0.75 100

crock width Onlmm! crack width On (mm)
o. Mix A. b Mix B.

Fig. 3.20. Crack opening paths for both mixes.

Shear failure occurred when the crack faces became unable to transfer
the applied shear stress and was characterized by an abrupt increase in
the crack displacements, instead of the gradual increase observed for
the reinforced specimens. The crack opening paths for mix A varied only
within a small range, see Fig. 3.20a. For mix B, a somewhat wider range
was found, with a crack opening path slightly deviating from the mean
 - 57 -

crack opening path of mix A, see Fig. 3.20b.

The number of cycles till failure was not only influenced by the applied
shear stress, but also by the normal stress. Although no ultimate shear
stress is clearly defined for experiments with an increasing normal
stress, a rough approximation of the static shear strength can be made
using the experimental results of Walraven [81] and Daschner [12].
Daschner's results were questionable [13] with respect to the measured
crack displacements, but the relation between the (constant) normal
stress and the static shear strength was properly recorded, see Fig.
3.21. The following expression for the static shear strength was derived
by means of a regression analysis 182]:

0 321 0 "(27
1.647 f * o (3.A)
can n

1
reinforce d specimens

,0.56
0.75
2.< A
0
0 •
i
—" o
0.50 •
•
J*^ 0

08 0.25
l(o • Mix A
a Walraven
A Mix B
o Daschner

045 1.B0 10' 10] 10' »' 10*

„0.56 number ot cycles till failure log(nfl (cycles)

Fig. 3.21. Shear strength versus Fig. 3.22. Shear stress level versus
normal stress [82]. number of cycles till
failure

For the present test series, the normal stress at the onset of shear
failure was presented in [56]. Inserting this value of the normal stress
into eq. (3.A), the number of cycles till failure is related to the
(approximated) shear stress level, see Fig. 3.22. Despite of the scatter
of the results, it appeared that most of the plain concrete specimens
endured less cycles till failure than the reinforced specimens at the
same shear stress level. In Fig. 3.22 this is shown by means of the
 - 58 -

dashed line according to eq. (3.3).

As discussed in Section 3.2.2., the aggregate interlock mechanism is
strongly influenced by the amount of particles with interfacial bond
fracture. For four plain concrete specimens, the area of fractured par­
ticles after pre-cracking and load-cycling was determined. Typical re­
sults are shown in the Figs. 3.23a-b. The mean ratio of the fractured
area of the particles to the total cross-sectional area of the particles
in the crack plane was 20.4 percent for mix A and 25.1 percent for mix
B. This was in reasonable agreement with the expectation, although the
amount of particles fractured in mix A was larger than expected. The
difference between the areas of fractured particles of the two mixed was
not significant.

fractured particles =
* . m *• . . ■••'
black area

• *'** / .

120

Fig. 3.23. Fractured area of particles.

 - 59 -

4. THEORETICAL MODELLING OP THE RESPONSE OF CRACKED CONCRETE TO

MONOTONIC SHEAR LOADING

4.1. Introduction.

In the Chapters 4 and 5, all the attention will be devoted to the physi­
cal understanding of the shear transfer mechanisms in cracks in plain
and reinforced concrete. Therefore, models referred to in the litera­
ture, which are mainly based upon empirically derived relations between
the stresses and displacements in the crack plane, such as the shear
friction-analogy, are not discussed here.
Physical models properly describing the response of the mechanisms of
aggregate interlock and dowel action are basically derived for the case
of increasing static shear loads [22,57,81]. These models wil be ex­
tended to the case of a repeated or reversed shear load with a constant
amplitude. Therefore, the existing models will be briefly discussed and
their presentation will be adapted to the case of constant shear loads.
Furthermore, newly observed material behaviour is incorporated into
these models. Next, the response of cracks in reinforced concrete to
static shear loads is described on the basis of the crack opening path.
It is shown how the transfer mechanisms affect the crack opening direc­
tion.
Finally, in Chapter 5 the existing models will be adapted to the case of
repeated and reversed shear loads. A distinction is made between the
description of 'high-intensity low-cycle' fatigue on the one hand and
'low-intensity high-cycle' fatigue on the other hand.

4.2. The mechanism of aggregate interlock.

Walraven [81] developed a physical model, based upon the assumption that
concrete can be conceived as a composition of two basically different
materials; the strong and stiff glacial river aggregate particles and
the matrix material consisting of hardened cement paste with a much
lower strength and stiffness. If a crack is formed in the concrete, it
wil run through the matrix material and along the interface of the ma­
trix and the particles. Therefore, the crack plane exhibits a global
undulation caused by the irregular shape of the crack faces and a local
 - 60 -

roughness due to the particles protruding from the crack plane. The
roughness due to the protruding particles dominates the roughness caused
by the global undulation. The crack plane can therefore be approximated
by a flat plane intersected by stiff particles. Next, the irregularly
shaped particles are randomly orientated. The most accurate simplifica­
tion of the particles is to consider them as spheres. The crack plane
according to Walraven's assumptions is shown in Fig. 4.1. His schematic
two-phase presentation of the actual crack plane provides a physically
close approximation of the experimentally observed crack response to
static shear loads. The particles are regarded as rigid spheres embedded
in the matrix material, which is considered as a rigid material with
crushing strength a . In consequence, the particles are undeformably
crushing the matrix during shear sliding.

Fig. 4.1. Crack plane according to Walraven's two-phase model [81].

Whether a particle makes contact with the opposing crack face depends
upon the particle size, its embedment depth, the crack width and the
shear displacement. An interesting aspect of Walraven's schematic pre­
sentation of the crack plane is that the total contact area of all the
particles in a unit area of the crack plane can be determined analyti­
cally. Considering a gradation according to Fuller's ideal curve, Wal­
raven quantified the projected contact areas a and a for any given
x y
particle during shear sliding, see Fig. 4.2a. For a thin slice of the
crack plane the particles reduce to circles, which apparance in the
crack plane is described by a probability density function. The projec­
ted contact areas in this thin slice can be determined analytically
 - 61 -

con toct area,

a. Contact areas. b Projected contact areas

Fig. 4.2. Contact area during shear sliding.

being the distance between the intersection point of two circles and the
intersection point of a straight line and a circle. All the projected
contact areas a and a are summed up numerically yielding the total
projected
r J contact areas Ax and Ay for a unit area of the crack rplane.
Now, the equilibrium condition for this unit area can be described by,
see Fig. A.2b:

T = o (A + uA ) (4.1a)
a pu y x

o=o (A - uA ) (4.1b)
a pu x y

in which T = shear stress

a
o = normal stress
a
o = strength of matrix material
pu
A = total projected contact area per unit area of the crack
plane parallel to the crack plane
A = total projected contact area per unit area of the crack
plane normal to the crack plane

The strength of the matrix material o and the coefficient of friction

pu
were derived from the experimental results of Walraven's static tests
0.56
a[81]:
= 6.39 f [MPa] (4.1c)
pu ccm
 - 62 -

= 0.4 4.ld)

Fig. 4.3 presents a comparison of the model with some typical test re­
sults [81]. Apart from his own experimental results, Walraven's model
provides good predictions for tests of Paulay el: al. [50] and Millard et
al. [45].

shear d isplccement 6t Imml shear d i s p l a c e m e n t Imml

o a »100 _
„shear stress i Q I MPa]
Opo °"pu
20 12 10
2.0

'V 1.5
11/I
v\
/ max

/
V,* JC
—10 mm
-16 mm
32 mm
1.0 L f /
/f

k
0.5
/ u .
-7>
0.1 mm M*?- ^
hormal stress OQ [MR]) 0 0.5 1.0 1.5 0 0.5 1.0 15
crack width C^ [mm! crack width 6^ [mm)
a. Shear stress. b. Normal stress.

Fig. 4.3. Comparison of the model with Fig. 4.4. Crack opening path ace.
experimental results [81]. to the model.

The two-phase model as presented in Fig. 4.3 describes a unique rela­

tionship between the stresses and the displacements in the crack plane.
For the use in this document, it was preferred to present this relation­
ship as crack opening paths for constant shear or constant normal
stresses. These stresses are related to the matrix strength according to
eq. (4.1c), see Figs. 4.4a-b. According to the model the maximum parti­
cle size has a slight influence upon the crack response. This is shown
in the Fig. 4.4 for particle diameters of 10 mm, 16 mm and 32_ mm. The
lower the size of the maximum particle the steeper the crack opening
path (6 -increment > 6 -increment).
t n
It should be noted that the two-phase model is based upon the assumption
that the contact areas between the particles and the matrix material for
a given combination of the crack displacements can be calculated ne­
glecting the previously followed crack opening path, see Fig. 4.5a. Gen­
erally, the actually followed crack opening path causes a gradual in­
crease of the contact areas, thus incorporating the previously formed
contact areas in the newly formed contact areas, see Fig. 4.5b.
 - 63 -

a. Actually deformed matrix. b. Crushed matrix according to the model.

Fig. 4.5. T h e contact areas for a given combination of the crack

displacements.

. shear disptacement6t Imml

°TOX ■ 16 mm

; / 1
1.0
r
/ / /
/ / / /
as
' y

crack width 6 n l m m '

a. Critical crack opening direction. b. Average crack opening

direction.

Fig. 4.6. T h e additional condition to the two-phase model.

However, the condition of the gradually increasing contact areas must be

fulfilled for the applicability of the two-phase m o d e l . F i g . 4.6a shows
that for a given particle diameter this condition is just fulfilled when
the maximum value of the slope of the crack opening path in each point
is equal to the tangent of the upper intersection point of the contact
area. Now, for a given combination of the crack width and the shear d i s ­
placement, an average crack opening direction can be determined taking
 - 64 -

into account all the particles intersecting the crack plane. In fact
this calculation is performed similarly to the numerical solution of the
total contact areas of the two-phase model of Walraven using the proba­
bility density function. For a maximum particle diameter of 16 mm, the
additional condition to Walraven's model is presented in Fig. A.6b.
Other maximum particle diameters yield only slightly different results.
The model is valid when the crack opening direction is according to Fig.
4.6b or steeper. The experimentally observed crack opening paths in Wal­
raven's static test series generally fulfilled this condition. It must
be realized that the matrix strength and the coefficient of friction
were determined from these results, thus accounting for some influence
of the previously followed crack opening path.
For the case of a less steep crack opening direction the transferred
stresses will be less than according to the two-phase model.

4.3. The mechanism of dowel action.

The mechanism of dowel action is based upon the response of a bar and
the surrounding concrete to a lateral bar displacement. As described in
Section 2.4, failure of a dowel occurs due to crushing of the concrete
and yielding of the bar when the concrete cover of the bar is suffi­
ciently large to prevent splitting failure. For the case of crushing
failure three mechanisms can be distinguished, according to Paulay [51]:
a. bending; the dowel force is transmitted due to bending of the bar.
For this mechanism the ultimate load is reached due to yielding of
the bar.
b. pure shear; it is expected that the transfer of dowel force by means
of pure shear is unlikely because of the deterioration of the con­
crete at the vicinity of the bar. Therefore, the resulting dowel
forces at both sides of the crack plane have a relatively large ec­
centricity resulting in yielding of the bar due to bending.
c. kinking; for a considerable lateral bar displacement the axial bar
force in the crack plane has a component parallel to the crack direc­
tion, see Fig. 4.7. For the case of cracked concrete the crack width
remains small relatively to the bar diameter. Hence, the effect of
kinking of the bar will be small (except for the case of large crack
widths in combination with small bar diameters).
 - 65 -

Fig. 4.7. Kinking of a bar according to Paulay [51].

The ultimate dowel force is reached when plastic hinges develop in the
bar. Therefore, this mechanism is affected by both the properties of
concrete and steel. The most important parameters are the bar diameter,
the concrete strength, the axial steel stress and the steel yield
strength.

jifillllllfliliiite
JtT^-'-~'~• .\- ■■ ■ v.~v7\-v~'"v-As',■.'.'??syZ/

F
d - -r-r--i-T- -
reaction stresses ~- -J

l>
Fig. A.8. Bar considered as a beam on elastic foundation.

The dowel load-lateral displacement relation can be described using the

theory of beams on elastic foundation, as published by Timoshenko and
Lessels [70]. Now the bar is considered as a flexible beam of infinite
length supported on an elastic foundation, see Fig. 4.8. This mechanism
is first used by Friberg [22] and accounts for the bar diameter and the
concrete strength.
According to this mechanism, the following relation can be derived:

6 =F L_ [mm] (4.2)
t 'd 28 3 El
 - 66 -

wlth B = ,/__

K f = foundation modulus of concrete [MPa/mm]

<j> in [mm], E in [MPa], I in [mm1*]

According to Finney [19] the value of the foundation modulus of concrete

is in the range of 200 to 2400 MPa/mm with an average value of approxi­
mately 700 MPa/mm, thus showing a large scatter. This scatter is probab­
ly due to some deterioration of the concrete at the vicinity of the bar,
although it is assumed that for the overall behaviour the concrete is
still uncracked. In consequence, the way of load application and the ec­
centricity of the dowel force (distance of dowel load to the concrete
surface) may influence the magnitude of the foundation modulus obtained
in an experimental program.
Finney described the bar displacement accounting for the lateral dis­
placement caused by the deformation of the 'free' length of the bar.
This free length comprises the crack width for bars perpendicularly
crossing the crack plane. However, for bars with an inclination to the
crack plane, Schafer [63] suggested to enlarge the free length to take
account for the inclined cracking of the concrete at the vicinity of the
crack plane.
In [81, page 42] Walraven combined the analytical solutions of Friberg
[22], Finney [19] and Schafer [63], thus yielding:

1 75 0 75 ,
F = 3.56 * ♦ " * Kr * 6 [N] (4.3)
d i t

with K f = foundation modulus of concrete [MPa/mm]

<(>, 5 in [mm]

1.75 . .
So the dowel force is proportional to <t> . Jimenez [34] based a simi­
lar relation also on a beam on elastic foundation:

F = 190 <t> * « [N] (4.4)

d t

with $, 6 in [mm]

Vintzeleou [77] derived an expression for the foundation modulus assum-

 - 67 -

ing that the concrete supporting the bar is deformed by the dowel force
up to a distance of twice the bar diameter. Assuming linear elastic ma­
terial behaviour it was found that:

E
K r = ^T [MPa/mm] (A.5)
t zq>

foundation modulus Kf iMPol

* 6.3 mm
6 12.7 mm
« 9.5 mm

AM
\>\
\.\
»\
^ > ^

0 0.5 1.0 1.5 2.0 2.5

lateral bar displacement 6[ (mm]

Fig. 4.9. Foundation modulus as function of the lateral bar displacement.f 51]

Equation (4.5) appeared to be valid for dowel loads less than 50 percent
of the ultimate dowel force. In fact, for very low dowel forces, the
concrete stressed by the dowel load is situated close to the bar. For
that case the distance will be smaller than twice the bar diameter. In
consequence, the foundation modulus will increase. Indeed, according to
experimental observations reported by Paulay [51] the foundation modulus
must decrease with increasing lateral bar displacements, see Fig. 4.9.
Millard [45] found experimentally that the initial foundation modulus of
concrete was equal to 750 MPa/mm for moderate strength concrete. For
high-strength concrete the value of the foundation modulus was found to
be proportional to the square root of the concrete strength.
It must be noted that the stress distribution according to Timoshenko's
theory does not agree with the real distribution of the reaction
stresses in the concrete, see Fig. 4.10.

With increasing dowel force the concrete stresses at the vicinity of the
bar exceed the uniaxial compressive strength. However, the surrounding
concrete provides a considerable confining pressure, thus yielding a
triaxial compressive zone under the bar. Therefore, the concrete
strength can be several times as high as the uniaxial strength.
 - 68 -

w^ ■r -T — T-T-.
., i'i Y
i-'T>-.
--.i
u /

a Stress distribution according to the model. b. Actual distribution.

F i g . 4 . 1 0 . The t h e o r e t i c a l and r e a l s t r e s s distribution.

Now, the bar itself becomes the weakest link and the ultimate dowel
force is reached when the bar yields.
Rasmussen [57] performed tests on dowels protruding from a large con­
crete block, see Section 2.4. Apart from his experimental study, Rasmus-
sen modelled the dowel action according to the behaviour of the steel
dowels in timber structures. Fig. 4.11 presents the failure mechanism in
which a plastic hinge is. situated at some distance to the 'crack plane'.
In the plastic hinge the plastic moment of the bar is reached, which is
equal to 0.167 f <t>3. Now, the equilibrium condition yields:
sy '
Fd = B 4>2 / f ccyli fsy [N] (4.6a)

Lth B = C ( / l+[eC] 2 - E C ) (4.6b)

lA ccyl
(4.6c)
sy
e = eccentricity of the dowel load [mm]
C = empirical constant
f ,, f in [MPa], <t> in [mm]
ccyl' sy ' '

It was found experimentally that C was equal to 1.3, assuming a zero-

eccentricity of the dowel load.
Dulacska [15] also derived an expression for the ultimate dowel force,
see eq. (2.7).
Vintzeleou [77] derived an expression for the ultimate dowel force,
which is in fact similar to Rasmussen's formula. However, the derivation
 - 69 -

r
Fig.4.11. Failure mechanism according to Rasmussen [57].

is based upon a failure criterion, which was used by Broms [7] to de­
scribe the ultimate lateral force for a pile in a cohesive soil. In
Vintzeleou's approach, no empirical constant is used for the calculation
of the ultimate dowel force. Fig. 4.12 presents Broms' failure mecha­
nism. The soil at the vicinity of the surface reacts less stiff than the
soil situated at some distance from the surface, where the compressive
strength of approximately five times the uniaxial strength is obtained.
According to Vintzeleou, there is no decrease in stiffness at the vicin­
ity of the crack plane for a bar embedded in concrete.
For the failure mechanism presented in Fig. 4.13, Vintzeleou derived the
following expression for the ultimate dowel force:

Fl + (10 f . e *) F. -1.7 ♦" f , f = 0 (4.7)

du ccyl du ccyl sy

F_, in [N], f , , f in [MPa], *, e in [mm]

du ccyl sy

For zero-eccentricity eq. (4.7) becomes equal to Rasmussen s formula.

Despite the close fit to experimental results it must be doubted whether
the failure mechanism shown in Fig. 4.13 is valid. Experimental observa­
tions of Dulacska [15] and Utescher [73] showed a considerable spalling-
off of the concrete close to the crack plane, see Section 2.4.
Due to this spalling-off of the concrete, Rasmussen's and Broms' de­
scriptions seemed to be in closer agreement with the actual stress dis-
 - 70 -

reaction stresses

-I 5»
approximated reaction stresses.

c = cohesion = 0.5 t c c y l

Fig.4.12. Failure mechanism for a Fig. 4.13. Failure mechanism for a

pile in a cohesive soil dowel in concrete according
according to Broms [7]. to Vintzeleou [77],

tribution than Vintzeleou's approach. Therefore, Rasmussen's stress dis­

tribution will be used for a further analysis of the dowel action mecha­
nism.
Fig. 4.14 presents a bar protruding perpendicularly from a concrete
block. At a distance X from the 'crack plane' a plastic hinge had been
developed. The magnitude and the distribution of the reaction stresses
in the concrete under the bar are not known beforehand. According to the
theory of plasticity the plastic moment of the bar is equal to
0.166 <(>3 f . However, this plastic moment can only be applied to a bare
bar. For the case of a dowel embedded in concrete, the concrete support­
ing the bar in the cross-section situated at the plastic hinge of the
bar is deformed due to the lateral bar displacement. In consequence, the
bond between the bar and the concrete above the bar must be broken. Con­
trary to this, the bond between the bar and the concrete supporting the
bar will be extremely good due to the high reaction stresses. For the
cross-section, in which the plastic hinge is situated, the equilibrium
condition is shown in Fig. 4.15.
The resulting force due to the bond stresses between the bar and the
concrete supporting the bar is situated at a distance z from the
neutral axis of the bar. Due to the bond force the neutral axis of the
 - 71 -

loss of bond

D.A[
M.4M
A°
M

z:{ r_~": / shifted axis

tetÜAt-

'CD

Fig. 4.IA. Failure mechanism of bar Fig. A.15 The equilibrium condition
due to plastification. for the plastic hinge.

bar is shifted over a distance z, see Fig. A.16. To obtain equilibrium

the bond force N must be equal to 2A .f . Now, according to the equi­
librium condition it follows:

Fig. A.16. Shift of the neutral axis

uVN = 0: (A + A ) f = (A,+ 2A ) f (A.8a)

a c sy b a sy

7M
L = 0: (A + A ) f z = A, f zu+ 2A f (z - z) (A.8b)
a c sy c b sy b a sy n

with A = T r ! -A. (A.8c)

a 2 b

Ab= r2 [0 - 0.5 sin(20)] (A.8d)

 - 72 -

A = 0.5 nr2 (4.8e)

c

z = 0.67 3
r. sin3(0)
z (4.8f)
b A
b

0.67 r3 sin3(0)
2 + 2 (4.8g)
c ATI
a c
The plastic moment can now be calculated as a function of z according
n °
to e q s . (4.8a-g). This is shown in Fig 4.17, in which the distance z is
n
related to the radius of the bar r. The shift of the neutral axis of the
bar can be determined if the eccentricity of the bond force is known.
The eccentricity of the bond force is related to the distribution of the
bond stresses. This distribution is not exactly known, but can be deter­
mined with reasonable accuracy.

2.2.Vr,,ir

"0 ' 0.6 07 0.8 0.9 1.0

relative eccentricity z n / r 1-1

Fig.4.17. The plastic moment as function of the eccentricity of the bond

force z .
n

First, the bond stress is related to the steel strain. Therefore, the
bond stress is proportional to r cos(a), see Fig. 4.18. Second, the
bond stress is influenced by the normal pressure on the bar. Untrauer
[72] found that the bond stress is proportional to / o . According to a
linear elastic response, the normal stress can be approximated by:

AD , v
a = — cos(a) (4.9)
n rtr
 - 73 -

Thus yielding for the bond stress:

= f(cos',5(a)) (4.10)
bond

The magnitude of the eccentricity of the bond force can now be calcu­
lated according to, see Fig. A.18:

IT/2-0
f T, , cos(0)d0 ƒ cos ' (a) cos(0)d0
bond
o o
Z r (4.11)
n= r
it/2-B ^ 7r"2 ^ 6 1 5
J J cos ' (a)d0
bond

with a
Tt - 26 relative eccentricity z n /r

neutral axis
I P
-shifted n.a.

cosl5(ol 0.25

0 « 90
angle ' Y (degreesi

Fig. 4.18. The distribution of Fig. 4.19. The eccentricity of the bond
the bond stresses. as function of y.

The steel close to the shifted neutral axis of the bar will slip rela­
tively to the concrete due to the reduced bond as a result of the low
normal stress. Therefore, the concrete at the vicinity of the shifted
neutral axis will provide only a minor contribution to the bond force.
This is shown by the dashed line in Fig. 4.18, assuming that for an
angle y there is little bond between steel and concrete. Now, eq. (4.11)
becomes:

1./2-B-Y s
J cos * (a) cos(0)dO

Z
(4.12)
n/r = ,/2-B-Y 1 > s

J cos " (a)d6

 - 74 -

Eq. (4.12) is shown in Fig. 4.19 for 6 equal to 10 degrees.

With the eccentricity of the bond force taken equal to the average value
of 0.93r, it is found in Fig. 4.17, that the plastic moment is equal to:

M = 1.78 r3 f = 0.223 *3 f (4.13)

u sy sy

Note that the plastic moment is 34 percent higher than the plastic mo­
ment of the bare bar according to the theory of plasticity. For this
case the neutral axis is shifted over a distance equal to 0.14r. In con­
sequence 6 is equal to 8 degrees, which is in good agreement with the
assumption of 10 degrees.
Now, the equilibrium according to Fig. 4.14 becomes:

FJ (e + aX) = 0.223 4>3 f (4.14)

du sy

and

F = X f ,4. (4.15)
du ccyl
-it
with f = mean compressive strength of the multi-axially loaded con­
crete

Combining eqs. (4.14)-(4.15) yields:

FJ = 0.5 C I / 4x0.223 +(Ce) 2 - Ce I <t>2 f \ f ~ [N] (4.16)

du ' ' sy ccyl

with C = / n/a = empirical constant

n = f ,/f .
ccyl ccyl

e / ccyl
e * ' f
sy
<t>, e in [mm], f , , f in [MPa]
ccyl sy

The constant C can be solved empirically by means of the experimental

results of Rasmussen. Rasmussen found from his test results, assuming
zero-eccentricity:
 - 75 -

0.5 C / 4x0.223 = 1.3 (A.17)

Thus C is equal to 2.75. However, as stated in Chapter 2, some eccentri­

city was inevitable during these tests. Therefore:

0.5 C [ AxO.223 + (Ce) 2 - Ce] = 1.3 (4.17a)

According to Rasmussen's formula the ultimate dowel force is equal to:

F. = 1.3 * 2 / f f 7 [N] (4.18)

du sy ccyl
Assuming that the contact zone between the bar and the loading frame is
loaded beyond its yielding strain up to the ultimate strain, the length
of the contact zone can be determined. For the steel used in Rasmussen's
tests the ratio between the yield strength and ultimate strength was ap­
proximately 0.6. Now, the eccentricity becomes, see Fig. 4.20:

0.5 F
e = 0.5 L = A . i- , [mm] (4.19)
T~t sy/0.6

thus

± | mm]
= o.39 4 /f ^
sy

The average value of the ratio between steel strength and concrete
strength was equal to 10.77 for the tests of Rasmussen. N o w , combination
of the eqs. (4.17a) and (4.19) yields:

3.1

Thus:

F . = 1.55 I / 0.892 + ( 3 . 1 E ) 2 - 3 . l e l ^ f~l f ~ [N] (4.20a)

l
du ' sy ccyl
with

sy
 - 76 -

F. = 1.35 [ / 1 + 9e 2 - 3e j <t>2 / l f [N] (4.20b)

l
du ' sy ccm

with

e /f "
E = / / ccm
♦ f—
sy
L
H 1-

loading frame concrete

°s
1!
bar )
+i o

Fig. 4.20. Load eccentricity in Rasmussen's tests.

In Fig. 4.21 eq. (4.20b) is compared with available test results of

Bennett [4], Paulay [51], Rasmussen [57], Vintzeleou [75] and Utescher
[73]. It was found that for 76 experimental results the average ratio
between the theoretical and the experimental dowel force was equal to
0.998 with a coefficient of variation of 17.2 percent. Appendix II
presents detailed information on these tests.
Now, the model will be extended to the case of a combined axial and
lateral load. In practice, axial steel stresses develop because embedded
reinforcing bars crossing the crack plane are strained due to the crack
opening during shear sliding. In consequence, this increasing axial
steel- tensile force will influence the equilibrium of forces and
bending moments in the bar. The axial steel force; will make equilibrium
with axial stresses at the vicinity of the (shifted) neutral axis of the
bar, see Fig. 4.22.
Fig. 4.22 is similar to Fig. 4.16 except for the influence of the axial
force. The contribution of the axial force to the equilibrium can be
easily taken into account in the eqs. (4.8a-g). Fig. 4.23 presents the
influence of the axial force upon the magnitude of the ultimate dowel
force. For small values of the axial force the shift of the neutral axis
is hardly influenced. For increasing axial force the shift of the neu­
tral axis decreases, to become zero for an axial force equal to the
yield force, see Fig. 4.24.
 - 77 -

theoretical dowel force Fq-U, t tkNl

100 /
Fdu = U 5 ( ^ / K 9 ^ - 3 E ] ^ ^ s y f c c m /
s
/
. e
/*ccm /
£
T/f /
80
/
/
/
o<
7
A g o /
9°
/
/
nB
P

•IF A
a
Bennett
Pa u lay
□ 7 Rasmussen
*
9 o Vintzeleou
o Utescher
3T=0998 ; c.ov = 17.2%

60 100

Fig. 4 . 2 1 . Comparison of the experimental and t h e o r e t i c a l results.

due to bending due to axial force.

Fig. 4.22. Equilibrium of the forces in the plastic hinge.

The interaction between the axial steel force and the ultimate dowel
force can be satisfactorily predicted by the eq. (4.21) proposed by
Vintzeleou [75] with n and m equal to 2. See Fig. 4.23:

F. F / F
r " in i s \m , n/, i s im (4.21)
du sy sy
with F.
a = dowel force
F, = ultimate dowel force according to eq. (4.20b)
 - 78 -

axial steel force

steel yield force
sy
F /F
d' du

Fd ' F * .

'S
•J^N. model
\ \ \X
\\
\\
0.5

PW- \
l'dull'syl \\

Fig. 4.23. The dowel force as a func­ Fig. A.24. The shift of the neutral
tion of the axial steel axis as a function of
force. the axial force.

Eqs. (4.20b) and (4.21) can be combined for the ultimate dowel force,
thus yielding an expression for bars with an axial steel force. This
expression is compared with experimental results of Millard [45], see
Fig. 4.25. The experimental result of test 25L is obviously disturbed.
This was probably due to the fact that for the small 8 mm diameter bar
used in this test, the strain gauges were stuck to the surface of the
bar thus influencing the bond of the bar to the concrete. Neglecting
this result, an average ratio of the predicted to the experimentally ob­
tained dowel strength equal to 1.02 with a coefficient of variation of
8.1% is found. Detailed information is presented in Appendix II.
Bars generally cross a crack plane at different angles. For inclined
bars, the angle of inclination influences the magnitude of the ultimate
dowel force. Two cases of inclined bars can be distinguished. First, for
small angles of inclination, the concrete supporting the bar will re­
spond less stiff to lateral bar displacements than is the case for bars
perpendicular to the crack plane. This is caused by the less favourable
shape of the concrete, which might cause inclined cracking. Second, for
large angles of inclination, the concrete will provide a stiff response
 - 79 -

theoreticol dowel force F du |lkN]

/
/ Spec men ♦ 0%
No. Imm] IMFbl
• A3 12 0
21L 0
22L 12 0
23L
•
D 12 0
/ 24 L ■ 16 0
25L A 8 0
/ 26L 1 12 175
/
/ 27L O 12 3«

experimental dowel forceEJkN]

Fig. 4.25. The experimental and theoretical results for Millard's

tests (45].

to lateral bar displacements. However, the axial steel force, which is

inevitable for this case, will influence the magnitude of the dowel
force according to eq. (4.21).
Next, both types of bar inclinations will be treated separately.

a. Angle of inclination in the range 0" to 90°.

Fig. 4.26 presents a bar with an angle of inclination 0 to the crack
plane. The bar is subjected to a load F parallel to the crack plane.
According to Vintzeleou [75], the concrete reaction force is provided by
a layer of concrete with a depth equal to twice the bar diameter. For
small bar inclinations, the concrete supporting the bar at the vicinity
of the crack plane has a depth, which is far less than twice the bar di­
ameter. Apart from that, inclined cracking can occur in the concrete,
see Fig. 4.26.
In consequence, the ultimate dowel force will be less than the dowel
resistance in the case of a bar perpendicular to the crack plane. The
actual stress distribution in the concrete is not accurately known, so
that the influence of the angle of inclination upon the concrete re­
sponse to a lateral bar displacement must be roughly estimated. To ac­
count for this influence, the dowel resistance is related to the incli­
nation according to:

F
du= Fdu,9oSin(6) (4.22)
 - 80 -

r
exl"dul
— - siniei
— model
• exp. Dulacska

i5° 90"
angle of inclination 0 [degrees!

Fig. A.26. Bar with a small angle Fig. 4.27. Dowel strength related to
of inclination. the angle of inclination.

In consequence the externally measured force is equal to F, „„.

du, 30
Furthermore, the dowel force in the crack has an eccentricity e to the
concrete. This eccentricity is equal toS

e = 0.5 $ cotan(O) + 0.5 6 /sin(0) = appprox. 0.5 * cotan(O) (4.23)

Now, this eccentricity can be inserted in eq. (4.20b). The result is

shown in Fig. 4.27 for the ratio f /f equal to 10. The dowel force
sy ccm
is taken proportional to the dowel force for a bar perpendicular to the
crack plane with zero-eccentricity. It is shown, that for the range of
0 investigated by Dulacska [15], eq. (4.23) provides a reasonable pre­
diction of the experimental results. Eq. (4.23) is in close agreement
with the reduction according to sin2(0) as proposed by Mattock [40]. It
can be concluded that the influence of small inclinations can be taken
into account in eq. (4.20b) with an additional eccentricity and multi­
plying the resulting dowel force with sin (t)).

b. Angle of inclination in the range of 90° to 180°.

Fig. 4.28 presents a bar with an angle of inclination 0 in the range of
90° to 180°. Now, the additional eccentricity has a negative sign. It
can be expected, that the response of the concrete to lateral bar dis­
placements is stiffer than for bars perpendicularly crossing the crack
plane. However, because of the fact that this increase is limited to the
concrete close to the crack plane, the increase in stiffness is less
pronounced than the decrease in stiffness for small angles of inclina­
tion. The minor increase of concrete stiffness will be neglected here.
 - 81 -

" \
90° 135° 180°
angle of inclination 6 [degrees]

Fig. 4.28. Bar with a large angle Fig. A.29. Dowel strength related to
of inclination. the angle of inclination.

The eccentricity of the dowel load can be expressed by:

0.5 ♦ tan(0 - 0.5 IT) (A.24)

Simultaneously with the dowel force, an axial force deveLops in the bar,
influencing the magnitude of the dowel force according to eq.(4.14). The
axial force is equal to F tan(0 - 0.5 n ) . Thus an implicit expression
is obtained for the dowel force by inserting this axial force into eq.
(4.21) and combining it with eq. (4.20b). Therefore, this expression is
solved numerically. Fig. 4.29 presents the dowel force as a function of
the angle of inclination. The ratio f It is taken equal to 10. The
sy ccm
dowel. force is related to the dowel resistance of a bar perpendicularly
crossing the crack plane with zero-eccentricity. The externally measured
force consists of contribution of the dowel force and axial steel force.
It can be concluded, that the dowel force of a bar with a large angle of
inclination can be calculated according to eq. (4.20b) with an addition­
al eccentricity.

4.4. The combined mechanism of aggregate interlock and dowel action.

Vintzeleou [75] and Millard [46] have already demonstrated that the com­
bined mechanism of aggregate interlock and dowel action is suitable for
predicting the shear resistance of cracked concrete according to the
equilibrium condition shown in Fig. 4.30. However, for push-off experi­
ments with a very small initial crack width as performed by Walraven
 - 82 -

[85], their models underestimate the measured shear strength of the

crack.

*. -
4 l external shear load ^s^
2 dowel force ^ s ^
3 normal force due lo '' 4
aggregate interlock.

U shear force due to

^j aggregate interlock
5 axial steel force. •+ -1

Q. Bors perpendicularly crossing b. Inclined bars.

the crack plane.

Fig. 4.30. The equilibrium condition for cracked reinforced concrete.

Furthermore, for practical use the equilibrium presented in Fig. 4.30

can only be determined if the relation between the axial bar force and
the crack width is known. The magnitude of the axial bar force can be
calculated by means of the equilibrium with the normal force due to ag­
gregate interlock. The bond characteristics for a bar subjected to the
combined action of axial and dowel forces are not yet determined experi­
mentally. Due to the lack of knowledge about the actual bond behaviour,
empirical relations are still used describing the crack opening path.
In this Section, the previously described mechanisms will be combined
taking into account the magnitude of the initial crack width. In order
to accurately predict the crack response, the effect of these mechanisms
and of their interactions upon the crack opening direction must be
known.

The crack opening path obtained in push-off tests on pre-cracked speci­

mens showed a large scatter, see Section 2.7. However, it was found by
Walraven [85], that the crack opening path is hardly influenced by the
reinforcement ratio, nor by the bar diameter. There was some influence
of the concrete strength on the direction of the crack opening (see Fig.
2.41b; mixes 1, 3 and 4 ) .
Furthermore, some influence of the maximum particle diameter was ob­
served (mixes 1 and 5). This indicates that the aggregate interlock
mechanism determines (partially) the crack opening path, because the
dowel action is not related to the maximum particle size. Therefore, it
is important to know how the aggregate interlock mechanism influences
 - 83 -

shear displacement 6, Imml crack width 6„lmml

Fig. 4.31. Shear stress versus crack Fig. 4.32. Theoretical and experi-
width for plain concrete [85]. mental crack opening path
for reinf. specimens [85].

the crack opening direction in a plain concrete push-off specimen. For

the experimental test series on cracked plain concrete performed by
Walraven [85], some typical results are shown in Fig. 4.31. In these
tests a small initial crack width (< 0.1 mm) was used. It appeared that
the crack opening path followed during the tests resulted in a more or
less constant shear stress after some shear sliding. The tests were
carried out load-controlled. Apparently, the increase of the crack
displacements is accompanied by a decreasing increase of the shear
stress. A slight decrease of the shear stress is observed only in a few
tests.

Assuming that this holds true for reinforced specimens, the crack open­
ing path followed during the tests will provide a constant contribution
of the aggregate interlock to the transferred shear stress. According to
this, the crack opening path for different shear stress levels (trans­
ferred by aggregate interlock) can be drawn, see Fig. 4.4 in Section
4.2.
Indeed, the crack opening paths obtained in Wal raven's tests fit reason­
ably with the calculated crack opening paths, see Fig. 4.32. Further­
more, the calculated crack opening direction explains the difference in
crack opening paths obtained in the experiments of Walraven [85] and
Millard [46]. The maximum particle diameter used by Walraven and by
 - 84 -

5heor displacement öt Imml

1.5

PU
i /
1.0 i / '

0.5
■ft u
max
10 Imml
16
32
n
0 0.5 1.0

crack width 6nlmm]

Fig. 4.34. Deformation of the bar
Fig. 4.33. The crack opening paths for
due to the dowel force.
different maximum partiele
diameters according to the
two-phase model.

Millard was 16 mm and 10 mm respectively. Accounting for the particle

diameter, the crack opening paths can be calculated, see Fig. 4.33,
which is in good agreement with Fig. 2.42.
However, it appeared from the test results that the aggregate interlock
mechanism provides a prediction for the crack opening direction for con­
stant shear stress contribution only. For the phase with an increasing
contribution of the aggregate interlock to the shear transfer, the plas­
tic hinge in the reinforcing bars is still developing. Therefore, the
bars will also influence the crack opening direction. To account for
this influence, the deformation of the bar due to the dowel force is
roughly approximated as presented in Fig. 4.34, in which the Lateral bar
displacement (shear sliding) is assumed to be caused by a rotation of
the bar around the plastic hinge. Now, the crack opening direction can
be expressed by:

A6 X + 0.56 2X
t _ n jpprox. — (4.25)
A6 0.5 ♦
n

The magnitude of X is expressed empirically on basis of Rasmussen's

experiments:
 - 85 -

0.318 4> fi Ti (4.26)

sy ccm
Thus,

AS .
-r^- = 0.736 J i l l (4.25a)
AS sy ccm
n '

Eq. (4.25a) is a rough approximation of the actually occurring crack

opening direction. It is obvious, that for instance the initial crack
width or the initial steel stress will influence the crack opening path.
Therefore, an empirical relation is derived describing the crack opening
path for increasing shear stress. Indeed, it was found that the crack
opening path is determined by the ratio f li and by the initial
' sy ccm '
crack width:

6
t 2f
A
= /12°ISZ ( 4 - 8 ) C
n no
[mm] (4.27)

ccm
with S = initial crack width
no
6„, S , & in [mm], f , f in [MPa]
t' no' n ' ccm' sy

sheor d i splaceme n t 6t IromI shear displacement 6| Imml shear displacement 6|lmml shear displacement 6t Imml
19mm mix
D^lOmm W "max 'ccn
-IS. 100. Imm] IMPi
°pu t^OSMfti 1 16 37
10 IB k 3 16 56
4 16 20
5 32 38

M
1.0
determined — exp — exp
by oggr int. - - - model — model
J//r /

0.5 0.5

determined
by dowel act.

0 0.5 1.0 0 0.5 1.0 0 0.5 1.0

crack width 6nlmml crack width öntmml crack width 6 n [mm!
a. Theoretical opening path. b. Experiments of Mattock c Experiments of Walraven. d. Experiments of Millard.

Fig. 4.35. Theoretical and experimental crack opening paths.

Application of eq. (4.27) provides combinations of the crack displace­

ments. Now, the contribution of the aggregate interlock mechanism to the
shear transfer can be calculated according to the two-phase model, see
Section 4.2. At the onset of a decrease in shear stress, the crack open-
 - 86 -

ing direction is fully determined by the aggregate interlock mechanism,

as is shown in Fig. A.35a. During this phase, the equilibrium condition
in the bar is determined by the axial steel force. The bar itself does
not influence the crack opening path, due to the plastic hinge, which is
now fully developed. In the plastic hinge each combination of the axial
bar force and the dowel force according to eq. (A.21) is possible. Fig.
A.35c presents a comparison of the calculated crack opening paths with
experimentally obtained opening paths of Millard [46], Mattock [41] and
Walraven [85]. There is a reasonable agreement between the calculated
and experimental crack opening paths. During the phase of increasing
shear stress, the contribution of the dowel action mechanism to the
shear also increases. An empirical relation is given by:

r- - ^-rh— < * (A 28)

-
du no t,e
with 6 = maximum slip occurring during the elastic deformation.
t ,e

For very small values of the initial crack width a minimum value of 0.1
mm is used, accounting for the larger crack width, which occurs during
pre-cracking the specimen. From the experimental results of Vintzeleou
[75], it was found that 6 was obtained for a dowel force equal to
t ,e
41 percent of the ultimate dowel force. Inserting this value into eq.
(4.2) yields:

[mtt] (A 29)
«t,."-2-p-Sr -
A*
with 6 = »/m

According to Fig. 4.9 the foundation modulus of the concrete can be

approximated by: (Note that 6 = 26 in Fig. 4.9)
t ,e t

K f = 390 6~ ' [MPa/tnm] (4.30)

1 t ,e

Now, eq. (4.29) becomes:

-7 0.60 1 2
6 = 1.31 10 * (f f ) * [mm] (4.31)
t,e ccm sy
 - 87 -

For a large initial crack width, eq. (A.31) becomes of minor importance
proportional to the crack width.
Now, the crack opening path can be described according to the transfer
mechanisms. However, the bars crossing the crack plane restrain the
crack opening due to the normal stress caused by the mechanism of aggre­
gate interlock. For increasing crack displacements this normal stress
can become so high,, that the restraining force in the bars is equal to
the yield force. A further increase in crack displacements should ful­
fill the equilibrium condition according to Fig. 4.29. Therefore, it was
expected that the crack opening direction was now determined by a con­
stant contribution of the aggregate interlock mechanism to the normal
stress. However, it was found experimentally [85], that the crack open­
ing path was hardly affected by the yielding of the bars. This can be
easily explained by the additional condition to the two-phase model as
presented in Fig. 4.6. In the case of yielding of the bars, the increase
in crack width exceeds the increase in shear slip, thus causing a dra­
matic decrease in the magnitude of the contact areas. In consequence,
the normal stress decreases and makes equilibrium with the yield
strength in the bars. Due to the decrease in contact areas the shear
stress decreases also. According to the two-phase model this decrease is
less pronounced than the decrease of the normal stress. However, for an
easy calculation method, the reduction in shear stress is related to the
decrease in normal stress. This yields:

f P f
_li =
Y = 52 (4.32)
a o o
s a
Now, the combined mechanism will be shown by means of an example of
Walraven's test specimen No. 110208c For this test, the cube compres-
sive strength was 35.9 N/mm2. The maximum particle diameter was 16 mm.
The steel yield strength was equal to 460 N/mm2. Four 8 mm diameter bars
were used, corresponding to a reinforcement ratio of 0.0056. The exter­
nally measured shear stress consists of the contributions of dowel ac­
tion and aggregate interlock according to:

Y (4.33a)
cal a a 'd d

o = Y o (4.33b)
cal a a
 - 88 -

with x according to eq. (4.1a), o according to eq. (4.1b)

Y according to eq. (4.32),
T. according to eq. (4.20b) (F./shear area)
u d
Y. according to eq. (4.21)

Table 4.1 lists the results of the calculation for test No. 110208t. The
theoretical crack opening path is calculated according to eq. (4.27) and
according to Fig. 4.4 for x equal to 3.8 MPa. The theoretical results
a
are in reasonable agreement with the experimental results. This holds
true for both the crack opening path and the shear stress - crack width
relation. Both relations are shown in Fig. 4.36a-b.

Table 4.1. The calculated results of test No. 110208t.

5 6 6 X 0 T Y Y T
n t t a a
d d a cal exp
exp • calc. calc .
mm mm mm N/mm2 N/mm2 N/mm 2
N/mm 2
N/mm2

.02 .00 .00 0.0 0.0 0.0 1.0 1.0 0.0 0.0
.05 .03 .03 0.9 0.0 0.5 1.0 1.0 1.4 2.5
.10 .07 .07 3.3 0.2 0.7 .99 1.0 4.0 3.8
.20 .17 .17 3.8 0.7 1.1 .96 1.0 4.9 4.5
.30 .28 .28 3.8 1.0 1.2 .92 1.0 4.9 5.0
.40 .40 .40 3.8 1.1 1.2 .90 1.0 4.9 5.1
.50 .57 .53 3.8 1.3 1.2 .86 1.0 4.8 5.0
.60 .71 .66 3.8 1.5 1.2 .81 1.0 4.8 4.9
.70 .88 .84 3.8 1.7 1.2 .75 1.0 4.7 4.7
.80 1.10 1.00 3.8 1.8 1.2 .72 1.0 4.7 4.4
.90 1.30 1.20 3.8 2.1 1.2 .58 1.0 4.5 3.9
1.00 1.50 1.40 3.8 2.2 1.2 .52 1.0 4.4

Fig. 4.36c presents the comparison between a few experimental and cal­
culated results. The calculations for these tests are carried out in the
same way as has been done for specimen No. 110208t. The agreement
between the calculated and experimental results is quite satisfactory.
 - 89 -

. shear displacement 6t [mm] . shear stress t [MPa] . shear stress i |MPal

D 32mm
max = 16 mm
II
W
f t t m =36Mft. ll f„„=38Mfti
ccm
1,46 mm II
II
'/~'
1608
/I ^ -
— exp II
II
n '/\- —a-Ü*8
li
— model /I i
ll exp
// i qr 8*8^
li _ - - model

exp

Na II02081 r
i
No. n0206t
F model

0 0.5 10 0 0.5 1.0 0 0.5 tO

crack width 6nImml crack width 5_ (mm) crock width 6 n Imml
a. Crack opening path. b. Shear stress versus crack width. c. Shear stress versus crock width.

Fig. 4.36. Comparison of the experimental and calculated results.

4.5. Influence of the normal restraint stiffness upon the shear

stiffness.

In the previous Section, it is shown that for reinforced push-off speci­

mens the crack, opening path and the shear stresses transferred across
the crack can be calculated according to the model of the combined
action of aggregate interlock and dowel action. However, the equilibrium
condition in such a specimen generally deviates from the equilibrium
condition in practical cases, see Fig. 4.37a-c.

I i X
I
._l
a
'.'.Z f*>
IW w *- Fj,=constant

7 t 7
a. Push-off element b. Restrained deformation c. Constant normal force

Fig. 4.37. Schematical presentation of restraint conditions.

 - 90 -

Case a in Fig. 4.37a represents the restraint condition of a push-off

specimen as used by Walraven [85] and Millard [46]. Case b represents a
shell element, which is cast between two large elements. Due to shrink­
age and thermal deformation, cracks will arise in the element. When a
shear load is subsequently applied to the element, shear displacements
will occur in the cracks. Due to the action of the bars, the cracks will
simultaneously open. Due to this crack opening, the concrete between two
cracks is unloaded, reducing the steel stress in the bars. As a conse­
quence, the crack opening has a crack closing effect.
Case c represents a large constant normal force on a crack. In fact this
is a special situation of case b.
Because of the fact, that case b represents most of the practical cases,
it will be shown in what way the model can be applied to this case.

top toads

a. Front view

Fig. 4.38. Wall cast between two storage tanks.

Fig. 4.38 presents a wall, which is cast between two storage tanks. Due
-6
to shrinkage, the strain in the element is 300.10 . A subsequent drop
-6
in temperature of 20 degrees causes an additional strain of 240.10 .
This thermal deformation is not followed by the tanks, due to the tem­
perature of the fluid in the tanks.
Due to this shortening, the element will have an almost fully developed
crack pattern, see Fig. 4.39. The model, which is used to describe the
bond behaviour, is not discussed here, because of its minor importance.
The calculated crack width is 0.106 mm due to an axial steel stress of
152 MPa. The mean crack distance is equal to 190 mm.
 - 91 -

normal stress op IMFta)

Fig. A.39. The normal stress-strain relation.

After cracking, the wall is subjected to a top load causing shear

stresses in the cracks. As a consequence, the crack faces slide with
respect to each other. From the model presented in the previous Sec­
tions, it is found that shear sliding in the crack is accompanied by a
crack opening according to eq. (4.27). According to the mechanism (eqs.
(4.28) and (4.31)), the plastic hinges in the bars are fully developed
after a shear displacement equal to 0.106 + 0.088 = 0.194 mm. Such a
crack response is obtained for case a. However, for case b any increase
in the crack width causes a decrease in the tensile stress in the con­
crete between the cracks.
To determine the crack response for case b, first the initial crack
response for case a is determined, see Table 4.2.
In Table 4.2 the decrease in steel stress Ao due to the decrease in
concrete stress between the cracks is presented. This decrease is calcu­
lated according to Fig. 4.39a. Furthermore, the increase in steel stress
Ao due to o is presented.
s,cr a
For any crack width and constant bond characteristics of the reinforcing
bars, the normal stress in the crack must be in equilibrium with
Ao + Ao . The equilibrium with Ao is necessary because otherwise
s,c s,cr s,c
the element is unloaded and the crack closes. The equilibrium with
Ao is necessary according to the model of dowel action. It must be
noted that the bond behaviour of the bars changes during shear sliding,
as is already shown in Section 2.6. Due to this, the axial steel stress
in the cross-section of the bar situated in the crack decreases. A gen-
 - 92 -

erally applicable description of the bond behaviour of a bar subjected

to both axial and dowel forces is not yet derived. In the example
presented here, the change in bond behaviour is not taken into account.
For this particular case, only slight differences with the actual
response will occur, due to the very high stiffness of the springs
representing the concrete between the cracks.
Because of this relationship between the crack width and the total
change in steel stress as listed in Table 4.2, the crack opening path
for case b can be calculated, see Table 4.3.

Table 4.2. Crack response for case a.

6 6 T a T Y Ao Ao
n
t a d \ d s,c s,cr
a
[mm] [mm] [MPa] [MPa] [MPa] I ~ ] [ - ] [MPa] [MPa]

0.11 0.02 0.32 -0.04 1.55 1.00 0.89 45 -2

0.12 0.05 1.42 -0.12 2.10 1.00 0.89 157 -8
0.13. 0.07 2.24 -0.08 2.36 1.00 0.88 269 -6
0.14 0.08 2.82 0.06 2.47 1.00 0.86 382 4
0.15 0.10 3.28 0.17 2.65 1.00 0.85 494 11
0.16 0.11 3.59 0.27 2.74 1.00 0.85 606 18

Table 4.3. Crack response for case b.

6 Ao Ao 0 T A6 T
n \ s,c s,cr
a a
no d
[mm] [mm] [MPa] [MPa] [MPa] [MPa] [mm] [MPa]

0.11 0.10 45 -2 0.67 5.0 0.03 3.28

0.12 0.16 157 -8 2.24 8.0 0.04 3.30
0.13 0.24 269 -6 3.95 10.4 0.04 3.30

The decrease in external normal stress on the crack plane can be taken
into account by reducing the initial crack width with A6 . This change
in initial crack width can be calculated because the crack displacement
in Table 4.3 must be compatible with the crack displacements according
to eq. (4.27). As a consequence, the ultimate dowel force is reached at
a shear displacement equal to 6 + 6 - A6 . According to the model,
^ t,e no no
any further crack opening will be determined by the crack opening direc­
tion for a constant contribution of the aggregate interlock mechanism to
the transfer of shear stress. However, in this case the influence of the
 - 93 -

external normal stress is larger than the normal stress according to the
dowel action mechanism (Ac ). As a consequence, the crack opening
s , cr
path is determined according to the equilibrium between the internal
normal stress due to aggregate interlock o and the external normal
stress according to line A-B in Fig. 4.39. The crack response for the
cases a and b is shown in Fig. 4.40.
shear stress T iMPo] shear displacement 6f (rnml

failure due to

"/ ƒ
incline d cracks *"

i
t

case h
/case b . case a

r p
-fase a
010

0 0 05 0.10 0.15 0.20 0 0.05 010 015 0.20

crack width ^ Imm] crack width 6 n [rnml

a. Shear stress-crack width relation. b. Crack opening path

F i g . 4 . 4 0 . C a l c u l a t e d crack response.

According to the calculation method presented in this Section, the ef­

fect of the external springs in case b can be accounted for by means of
a decrease or increase (for an increase in tensile stress) of the ini­
tial crack width. For most of the practical cases the stiffness of the
external springs is so large that the crack opening path is fully deter­
mined by the equilibrium of the normal stresses in the crack.
It must be born in mind that the proper bond characteristics must be
known.

4.6. Additional detailed tests.

In addition to the main test program described in Chapter 3, some push-

off tests were performed to verify experimentally the dowel action mech­
anism, which is developed in Section 4.3. These tests are fully de­
scribed in [21]. The specimen geometry and the test procedure were as
described in Chapter 3.
However, now the steel strain of the reinforcing bars was recorded by
means of a bolt gauge, type BTM-8 of Tokyo Sokki Kenkyojo. Half of all
the bars was prepared with a gauge of this type, which was cemented at
 - 94 -

crock plane

I 100 ! 30 | 100 ,

Fig. 4.41. Bar prepared with bolt gauge.

the neutral axis of the bar near the crack plane, see Fig. 4.41. The
overall dimensions of the gauge were 20 m x 2 mm with a working length
of 8 mm x 1 mm [71].
A total of six push-off tests was performed. Three series of two speci­
mens were subjected to static, sustained and repeated shear loading. The
results of the static tests will be briefly discussed here.
In theory, the bolt gauges, which were in the neutral axis of the bare
bar, reflected the strain due to the equilibrium with the normal stress
caused by aggregate interlock. However, according to the model of dowel
action the neutral axis was shifted to the supported side of the bar. In
consequence, the ends of the strain gauges were situated in regions of
the bar, which were highly strained. According to the model, the plastic
hinges are situated rather close to the crack plane, so that the ends of
the gauge nearly reached the yield strain. Thus, the average recorded
strain according to the model will be considerably larger than is neces­
sary to make equilibrium with the normal force due to aggregate inter­
lock. Indeed, this phenomena was observed in the experiments. Taking
into account the yield strength at the ends of the gauge, the average
normal stress according to the model agrees reasonably with the experi­
mental results, see Fig. 4.42.
Due to the plastic hinges, which are apparent in the bar according to
the dowel action model, a marked localization of the elongation of the
steel occurred during shearing-off. Therefore, a microscopic examination
of the steel crystals at the vicinity of the crack plane was performed.
The theoretical.shift of the neutral axis could be roughly determined by
means of this localized orientation. The bars, which were not prepared
with a bolt gauge, were carefully removed after testing. A total of 22
specimens was obtained from the six push-off specimens. Ignoring the
 - 95 -

normoi stress q |MPg]

model sf experimen
plastic hinges ,/^~
.-r-t
s /
1 /
1 /
1 /
r
s.aggr. int. 1/ ^ . - r . d u e to aqqr. int.
I I
0 0.25 0.50 075 1.00
crack width ön[mm!
a. Stresses in the gauge. b. Normal stress versus crack width.

Fig. 4. 42. Normal stress versus crack width for test Dl [21].

results indicating yield of the whole cross-sectional area of the bars,

the majority of the observations (67 percent) supported the shift of the
neutral axis and thereby the model developed. There is, however, more
detailed information necessary for a solid physical proof of this
mechanism.

4.7. Concluding remarks.

The transfer of static shear stress in cracked reinforced concrete de­

pends upon the aggregate interlock mechanism and dowel action. The in­
teraction between both mechanisms is determined by the equilibrium of
forces normal to the crack plane.
The dowel action is depending upon the bond between the bar and the con­
crete, thus causing a shift of the neutral axis of the bare bar. This
results in a strong increase of the plastic moment relative to the plas­
tic moment of the bare bar.
For an increasing axial steel force, this shift decreases, thus reducing
the dowel strength.
The crack opening path is determined by the deformation of the bar until
the plastic hinges in the bar have fully been developed. Subsequently,
the crack opening direction is related to a constant contribution of the
aggregate interlock mechanism to the shear transfer.
The combined model presented in this Chapter can be applied to cracked
reinforced concrete subjected to static shear loads irrespective of the
magnitude of the initial crack width. The effect of external springs
 - 96 -

normal to the crack plane can be accounted for by means of an increase

or decrease of the initial crack width.
 - 97 -

5. THEORETICAL MODELLING OF THE RESPONSE OF CRACKED CONCRETE TO

REPEATED AND REVERSED SHEAR LOADING

5.1. Introduction.

The static models of aggregate interlock and dowel action, which are de­
scribed in Chapter 4, will be adapted to the case of a repeated or a re­
versed shear load. Although the crack response remains essentially the
same, a distinction is made between 'high-intensity low-cycle' fatigue
and 'low-intensity high-cycle' fatigue. This distinction is made for
practical reasons. For 'low-intensity' tests as described in Chapter 3,
the increments of the crack displacements per cycle can be far less than
the numerical accuracy of any numerical program. Therefore, this type of
test cannot be analysed by calculating all subsequent load cycles. In
consequence', the physical models must describe the over-all response de­
gradation and irreversible deformations of the concrete due to cycling.
The same holds true for analyzing the response of large-scale structures
to millions of load cycles with a low amplitude. Over-all characteris­
tics, such as reduced shear stiffness and increased crack displacements,
will then be used in numerical programs for analyzing the response of
the structure to subsequent load cycles with a very high amplitude. For
these 'high-intensity' cycles, the displacement increments can be deter­
mined accurately in a numerical program. Therefore, in this Chapter
'low-intensity' fatigue will be treated by means of the description of
the over-all behaviour. On the other hand, 'high-intensity' fatigue will
be analysed in detail by means of the physical transfer mechanisms.
As in Chapter 4, first the transfer mechanisms will be dealt with sepa­
rately. Finally, the response of cracked reinforced concrete to repeated
and reversed shear loading is analysed.

5.2. The mechanism of aggregate interlock.

The analyses of the crack response to 'high-intensity' fatigue provides

a deep insight in the physical behaviour of cracked concrete, because of
the relatively large crack displacements in a load cycle. Therefore,
this type of fatigue will be discussed firstly. In 1980 [81], Walraven
already described qualitatively the response of a crack in plain con-
 - 98 -

crete to cyclic shear loading. According to his two-phase model, the

crack response can be roughly described monitoring the displacements of
a single particle during sliding, see Pig. 5.1.

a. Ascending branch

Ó.
V
c Descending branch.- zero-stress. d. Fully reversed loading.

Pig. 5.1. Qualitative description of the response to cyclic shear loads

in the first load cycle.

Fig. 5.1a shows that the ascending branch of the first cycle can be de­
scribed with the static two-phase model. After reaching the maximum
applied shear stress, the shear load is decreased. Consequently, the
normal restraint force will re-close the crack. However, the friction in
the contact area will counteract this displacement to a certain extent.
Walraven derived the following expression for the reduction in shear
stress before any displacement backwards can occur, see Fig. 5.1b:

T A - uA A - uA
_ _ _X £ _%_ X (5.1)
T A + uA A + uA
m x y y x
A further decrease in shear load forces the crack to close until the
initial crack width is obtained (then the normal force becomes zero).
Simultaneously, the shear displacement decreases, but reaches not its
original value due to the deformed matrix material, see Fig. 5.1c. This
shear displacement can be determined by means of the static two-phase
model, inserting the initial crack width and zero-shear stress. Reducing
the shear displacement to its original value only some friction due to
 - 99 -

rubble in the crack plane causes a low shear stiffness. The crack re­
sponse during sliding in the opposite direction is similar to the pre­
viously described, see Fig. 5.Id. During re-loading in the following
cycle, the shear displacement increases without (hardly) any shear load,
see Fig. 5.2a. This free slip is caused by the already deformed matrix
material. A further increase in shear load brings the particle in firm
contact with the opposing crack face, see Fig. 5.2b. The contact area
can now, however, not be calculated according to the analytical two-
phase model.

a. Free slip. b. Ascending branch.

Fig. 5.2. Qualitative description of the response to cyclic shear

loads in the second cycle.

The numerical contact model.

Walraven modified the calculation of the contact areas of the particles
(1986, [83]) by replacing the analytical solution by a numerical solu­
tion. Now, the shape of the contact area of each particle in the crack
plane is monitored by means of several points situated on the surface of
the deformed matrix material, see Fig. 5.3.
For a specific combination of the crack displacements, the highest and
the lowest point, which are situated in the contact area, can be deter­
mined. Now, the projected contact areas a and a can be determined as
r J x y
the distance between those points in parallel and normal direction re­
spectively. Taking into account several particle diameters and embedment
depths, the total contact areas per unit area of the crack plane
A and A can be calculated as in the original analytical two-phase
x y "
model. Again, the presence of a specific particle is accounted for by a
probability density function [81]:
 - 100 -

JÏS-vo
x2.y7

' H-A

° y = H "^o
x
10' y IO
«n

4-^4

Fig. 5.3. Representation of the contact area by means of points situated

in the contact area [83].

r\, 0 5 't 6 8 10
p(D ) = ^ (0.532x " - 0.212x - 0.075x - 0.036x - 0.025x ) (5.2)
O D
o

with p = volume of particles/total volume

This modification of the two-phase model will be denoted here as the nu­
merical contact model. A listing of the program is presented in Appendix
III.
With this model, Walraven simulated the cyclic push-off tests of Laible
[36,37]. The particles used in this test series had a moderate strength.
Consequently, the number of particles fractured during cracking of the
concrete was large with respect to the number of fractured glacial river
particles used by Walraven. Hence, a reduction of the total contact
areas must be taken into account. Furthermore, the coefficient of fric­
tion must be adjusted to the proper value for this type of aggregate.
From the experimental results of the first cycle of the test, Walraven
derived a reduction factor of 0.75 and a coefficient of friction equal
to 0.20. Inserting these values into the model, the subsequent cycles
are simulated, see Fig. 5.4. For reasons of symmetry, only the results
in the positive direction are shown. A good agreement between the exper-
 - 101 -

imental and the calculated results is found. It appeared from the calcu­
lation that no reduction of the matrix strength due to cycling has to be
taken into account. This is probably due to the fact that the high con­
tact stresses are only apparent in the contact areas. In Laible's tests,
the displacement increments in each cycle are large enough to crush the
previously loaded matrix material and to subsequently load new material.

shear stress la IMPo] crack w i d t h 6 n l m m |

n
1.25 125
experiment k1 1.00
II i
model
'1 jn=1 n=lS
075
'
ll !\ 050
ll
0.25
II
0
II 0 5 10 15 20 25 30
1 n
I shear slip 6( (mm)
125
fl *
1
J, i
1 / 1.00
// ll *-
ll 0.75 -U

f
/ 1

i
/ i
' 1
II ll -/' 0.50
II
II >' 'l 0.25
It ^ ^
/ 0
0.50 0.75 1.00 1.25 5 10 15 20 25 30
shear slip 6 { Imml

F i g . 5 .4. Comparison of the experimental result of test No. Al of Laible

with the numerical contact model (83].

In the calculation process, ten different particle diameters are used,

reflecting the gradation curve. Each particle diameter was embedded at
ten different depths, thus yielding one hundred particles to be taken
into account. The contact area of each particle was monitored by means
of ten points. Hence a total of two thousand coordinates determines the
total contact areas. The experimentally obtained normal restraint
stiffness was input data for the program.
Walraved demonstrated that the numerical contact model is suitable for
describing the crack response to cyclic and repeated shear loads. This
model can be used to perform sensitivity studies, but is too complex for
implementation in advanced finite element programs.
Because of the fact that Walraven already solved the problem of simulat­
ing the crack response to cyclic 'high-intensity' shear loading, most
effort will be devoted here to a sound simplification of the model in
order to speed up the calculation process.
 - 102 -

The analytical contact model.

As already stated in Section 5.1, the physical model must describe the
over-all response degradation and the irreversible deformations due to
'low-intensity' load cycles. In consequence, only a part of the load
history will be accounted for in subsequent 'high-intensity' load
cycles. Therefore, it is assumed that the load history is fully incorpo­
rated in the contact areas, which are formed in the last load step of
the previous cycle. This assumption is shown in Fig. 5.5.

Fig. 5 . 5 . The load history incorporated in the end-deformation of the

crack plane.

An important consequence of this assumption is that it holds true for

all the load steps in a cycle, see Fig. 5.6.
When this assumption is valid, the magnitude of the contact areas can be
determined analytically by means of the intersection points of three
circles, see Fig. 5.6. Now, only three pairs of coordinates of the ori­
gin of the circles determine the contact area. Again ten particle diame­
ters and ten embedment depths are used. This version of the model is de­
noted as the analytical contact model, which is listed in Appendix III.
For test No. Al of Laible [36], the model is compared with the, experi­
mental results in Fig. 5.7.

Fig. 5.6. The assumption applied to a given load step.

 - 103 -

., crock width 6 n Ifnml

1.25
/

li
experiment
/1 1.00
/
-—model
/ i i 075
^
i i
n=l / 050
! 2! 15
1 1 025
1 il
1 0
1 0 5 10 15 20 25 30
1 1 n
1
1
1 shear slip f>\ |mml
050 I.2S
i // ~T
100 f,
1
//
0.75
!«*
' f
t
1
1 0.50

*'t

0.75
il100 1.25
0.25
0
0 5 10 15 20 25 30
shear slip 6t (mm)

Fig. 5.7. Comparison of the experimental result of test No. Al of Laible

with the analytical contact model.

Input in the program was the end displacement of the first load cycle.
Therefore, the first cycle is not simulated. It appeared that the crack
response is satisfactorily described even if the three pairs of coordi­
nates of the three circles remain the same for all particle diameters.
Therefore, six coordinates determine the total contact areas in normal
and parallel direction. The calculation process is about twice as fast
as for the numerical contact model. However, the model is still too com­
plex for implementation in finite element programs.

The reduced contact model.

A further simplification can be obtained by reducing the number of par­
ticle diameters and embedment depths. There is, however, a more simple
method. During cycling, the stresses in the crack plane for a given com­
bination of the crack displacements are as large as or less than in the
case of a static test. This is due to the reduced size of the contact
areas, thus:

T = o (X A + iiX A ) (5.3a)
a pu y y x x

o=o (XA-pXA) (5.3b)

a pu x x y y
 - 10A -

with A , A = total contact areas for the static case.

x' y
X , X = reduction factors.
x y
.. X„(-l Xyl- 1
100
t
ml, t
0.75 ' 075
/ /
/ l>
60 6tm
0.50

//
s& //
025
< L * I > ^
^ry *<r «J6i -5o
rr 6 0 | ^< ' lOtm-OoJ
1 .. n
0.50 075 1.00 0 025 0.50 0.75 1.00
l6,-60)/|6,m-60) {6r6^n6lm-6J
a. X„ b. \y

Fig. 5.8. The contact area reduction factors.

For test No. Al, these reduction factors are derived on basis of the
calculation process with the analytical contact model, see Fig. 5.8. It
appeared that these factors can be approximated by:

6 -6
X = 0.8 (, l ,°)2 (5.4a)
x 6 - 6
tm o
6 -6
X = 0.7 (. * °)3 (5.4b)
y1 6„ - 6
tm o
2 2
with 6o= 6nm- / 6nm- 6no < 0.67 tm
6t (5.4c)

6 , 6 = end displacements of previous cycle.

nm' tm '
6 = initial crack width.
no

With the eqs. (5.4a-c), the calculation process becomes quasi-statical-

ly. Now, the problem is shifted towards an easy calculation method of
the contact areas for the static case. Because the derivation of these
expressions forms part of the numerical implementation, these expres­
sions are presented in Chapter 6 (eqs. (6.1a-i)).
Substitution of these equations into the model yields a simple version
of the model, which is denoted the reduced contact model. Again test No.
Al is simulated, showing good agreement with the experimental results,
see Fig. 5.9.
 - 105 -

shear stress t Q IMPa) crack width 6 n Imml

1.25 1.25
/i
1 .. •'.s
-—experiment

— model
A
/
i 1.00

0.75
^
100 II

n=1 / 1
21 1: i
lis
0.50

025

075 / 1 0
// 0 5 10 15 20 25 30
, i 1 n
i 1 shear slip 6t |mm]
050 1.25
i
i / 1.00
/
075
'il i
//
025
/ i
i
1 0.50

il
0.25
n - ^ ^ r, 0
0.75 100 1.25 5 10 15 20 25 30
shear slip 6 t Imml

Fig. 5.9. Comparison of the experimental result of test No Al of Laible

with the reduced contact model (6 = 0.76 mm).
no

shear stress Tq [MPal crack w d t h 6 n 1mm]

1.25 1.25
experiment
1'! 1.00
model / i I 0.75 ^ ^
100 i / 1
050
n=l/ ':■
0.25
"/ /
075 0
i 0 5 10 15 20 25 30
n

050
/ / shear stip 6t (mml
1.25
1
/ / / 1.00

025
1/

iLJ
'J
y / l
/ 0.75
0.50
0.25
n / 0
025 0.75 1.00 1.25 5 10 15 20 25 30
shear slip 6t Imm]

Fig. 5.10. Comparison of the experimental result of the test No. Cl of

Laible with the reduced contact model (6 = 0.51 m m ) .
no

The calculation process is now approximately one hunderd times as fast

as for the numerical contact model. The Figs. 5.10-5.11 show that the
reduced contact model also provided good predictions of the tests with
smaller crack widths and higher normal restraint stiffnesses.
 - 106 -

shear stress i n IMPol crock w dlh

1.25 «n' mml
\
1 I experiment
model
100
0.75

I 1
1
050
n= 1 j 15
0.25
|2
075 0
0 5 10 15 20 25 30
n
shear slip 6( (mm]
\ 1.25
1 100
1
0.75

t 0.50
025
lu' h 0
050 0.75 100 1.25 0 5 10 15 20 25 30
shear slip 6 t Imml

Fig. 5.11. Comparison of the experimental result of the test No. C2 of

Laible with the reduced contact model (6 = 0.51 mm).
no
shear stress TQ [MPal crack width 6 n [mml
1.25
experiment 1
1.00
model
/ - — = *•"
1/ 0.75
11
21 r" 0.50
1
'/ 0.25
II
ii 0
0 5 10 15 20 25 30
n=1 20J n

i sheor slip 6f 1mm]

1.25
1.00
r^
/ 0.75
i 0.50

0 0.25
7.
050 0.75 1.00 1.25
shear slip 6 t Imml
0 25

0 5 10 15 20 25 30
n
Fig. 5.12. Comparison of the experimental result of test No. El of
Laible with the reduced contact model (6 = 0.76 mm).
no

In Fig. 5.12, the results of test No. El of Laible is shown. In this

test, twenty load cycles with a maximum shear stress up to 0.69 MPa were
applied to the specimen, followed by five cycles with a maximum shear
stress of 1.2A MPa. Because of the assumption, that the end deformation
of the contact areas incorporates the load history, only the cycles Nos.
21 to 25 are simulated. Although the calculated crack response is some­
what too stiff, the over-all behaviour is satisfactorily predicted.
It must be noted, however, that both the analytical and the reduced con-
 - 107 -

tact model are only valid for load cycles, in which the maximum shear
stress is at least equal to the maximum shear stress in any previous
cycle. However, cyclic deformation-controlled shear tests, such as per­
formed by Vintzeleou [75], can be easily simulated with these models.
Now, the calculation is performed similarly to the case of a stress-
controlled test. Afterwards, the last part of each cycle is neglected,
see Fig. 5.13a. There is of course a difference with the actually ob­
tained deformation of the contact areas. This difference has, however,
only a very small effect upon the observed crack response. The case of
repeated loading can also be treated as a fully-reversed shear loading.
Now, the first part of the cycle is neglected, see Fig. 5.13b. Again,
the actual response is slightly different. For the case of repeated
loading, the crack width will not reduce to its initial value, because
of the fact that some normal stress can be transferred due to the fric­
tion in the remaining contact areas. In consequence, re-loading will
cause an initially stiff response of the crack, which is satisfactorily
simulated by the method presented in Fig. 5.13b.

shear stress xQ shear stress t Q

1
cycle 1 A '
/'
1,1
III

/'
n

1
1

shear displacement 6| shear displacement öt

a. Displacement-controlled. b. Repeated loading.

Fig. 5.13. The crack response for the case of deformation-controlled

tests and of repeated shear loading tests.

Simulating the above mentioned shear tests, it emerged from the calcula­
tion that there was no decrease in matrix strength due to cycling. This
is probably due to the fact that the crack displacement increments are
relatively large for the case of 'high intensity' fatigue. The high
stresses causing fatigue of the matrix material are restricted to the
volume close to the contact areas. Due to the increasing displacements,
 - 108 -

the previously loaded matrix material is crushed and the matrix material
lying behind is then subjected to high contact stresses. This matrix ma­
terial was subjected to low stresses in the previous cycles, so no fa­
tigue of the material has occurred. For the case of 'low-intensity high-
cycle' fatigue, the crack response can theoretically be simulated with
the proposed models. Decreasing the initial crack width to 0.15 mm and
using a normal restraint stiffness of 7.5 MPa/mm, a shear stress of 3
MPa can be transferred by the crack plane. The crack displacement in­
crease in each cycle rapidly diminishes to become smaller than the nu­
merical accuracy, indicating that the actual displacement increase is
even smaller. For this case, the crack displacement increments are very
small, so now a reduction of the matrix strength occurs due to cycling.
On the other hand, 'high-cycle' tests are generally performed with a
rather high loading frequency, which might cause rate-effects and thus
increases the matrix strength.

Because of the small displacement increment, the fatigue of the sug­

gested matrix material and the rate effects, empirical expressions must
be used to describe the increase in crack displacements due to 'low-
intensity' cycles.
For the test series on cracked plain concrete described in Section 3.3,
no straightforward relationship between the stress level, the number of
cycles and the crack displacements was obtained. Generally, the crack
response was initially stiffer than could be expected on basis of the
two-phase model, even if the repeated shear load was treated quasi-stat-
ically, see Fig. 5.14. In this figure shaded area corresponds to the
T to o ratios used in the experiments.
a pu
This phenomenon is probably due to the relatively high initial normal
stress, which was used in this test series. During cracking of the con­
crete prior to the actual push-off test, some matrix material and small
particles will be completely torn out of the crack faces, as is shown
theoretically by Termonia and Meakin [69]. Due to this material, the
crack cannot be re-closed to its original value. This rubble transfers
the initial normal stress and is thereby pushed into the crack faces. A
subsequent shear sliding will force this rubble to act like stiff struts
transferring both normal and shear stresses. Five different stages can
be distinguished. Initially, the struts only transfer normal stress, see
Fig. 5.15a. A shear displacement will cause a rotation of these struts.
 - 109 -

! 5 shear displacement 6t 1mm) , ^ shear displacement 6t Imm)

crack width6 n !mm! crack widthöplmml

a. Mix A b. Mix B.

Fig. 5.14. The stiff response of plain concrete to repeated loading.

In consequence, the struts provide a positive contribution to the

transfer of shear stresses and determine the crack opening direction,
see Fig. 5.15b. Simultaneously, the transfer of stresses due to aggre­
gate interlock increases. For a given shear displacement, the average
orientation of the struts is perpendicular to the crack plane, see Fig.
5.15c. Now, only normal stress is transferred by the struts. A further
increase in the shear displacement will cause a rapidly decreasing nega­
tive contribution of the struts to the shear transfer. A steep crack
opening path is then obtained, see Fig. 5.15d. Finally, the normal
stress due to aggregate interlock is in equilibrium with the externally
applied normal stress. Now, the struts become inactive, similarly to the
case of a very low initial normal stress.

on °n 6n cy,
a. No slip. b Stiff response. c. Strut transfers d. Negative shear
no shear stress. stress contribution
— — experimental — — — ace. to aggr. int for t m —-— ace. to aggr.int. for C—...

F i g . 5 . 1 5 . The s t r u t mechanism due t o the i n i t i a l normal s t r e s s .

 - 110 -

However, the crack opening direction is already determined by the

struts. According to the additional condition to the two-phase model,
see Section 4.2, a less steep crack opening direction will cause a sharp
decrease of the shear stress transferred by the crack plane. Therefore,
the steep crack opening path is followed despite the inactivity of the
struts.
Some experimental proof of this mechanism can be obtained from the tests
on cracked plain concrete. According to the strut mechanism, the ratio
of the shear stress to the normal stress transferred by the struts as a
function of the shear displacement must be similar for all tests.
Indeed, such a constant relationship was found, see Fig. 5.16.

, ltexp-Jaggr[/to"e>cp -%ggr.l

A/1.3/5.0/.01

A/I.9/5.0/.19
_■ A/1.3/4.2/12

A/0.8/S.S/.01
— A/1.3/6.2/.02

Fig. 5.16. The ratio of the shear stress to the normal stress as func­
tion of the shear slip for the plain concrete specimens.

The calculated average length of the struts was approximately 0.7 mm,
which is a realistic value for the small particles and matrix material
in the crack plane.
The higher the initial normal stress, the deeper the particles are
pushed into the crack faces. In consequence, the initial crack opening
direction will become less steep for higher normal stresses. Foj- static
tests, this phenomenon was observed by Vintzeleou [76], see Fig. 2.17.
For cyclic and repeated shear loads, the struts will cause a nearly con­
stant crack opening path with small crack displacement increments in
each cycle. However, when the struts become inactive, the aggregate in­
terlock has to transfer the full shear load, thus causing a rather
abrupt and strong increase in crack displacements, as was indeed ob­
served experimentally, see Fig. 3.19.

It can be concluded that for the case of cracked plain concrete with a
relatively high initial normal stress, stiff struts contribute to the
 - Ill -

stress transfer and initially determine the crack opening direction.

Therefore, this type of test cannot be simulated with the proposed
cyclic aggregate interlock models alone.
For cracked plain concrete with a low initial normal stress, empirical
expressions for the relationship between the shear stress, the number of
cycles and the crack displacements can be derived on basis of the tests
on cracked reinforced concrete specimens.

5.3. The mechanism of dowel action.

Although numerous cyclic dowel action tests are performed by among

others, Eleiott [16], Jimenez [32] and Vintzeleou [75], the large number
of observations did not yield a fundamental insight in the physical be­
haviour of dowels embedded in concrete and subjected to cyclic loads.
Therefore, empirical expressions for the dowel stiffness are derived for
practical use. Vintzeleou derived an expression for the maximum dowel
force for the case of imposed shear displacements, see eq. (2.9). On
basis of this relation, she proposed a formalistic model for fully re­
versed shear displacements as presented in Fig. 5.17. [77].

Dowel force Fg [kN]

Fig. 5.17. Formalistic model for fully reversed shear displacements as

proposed by Vintzeleou [77].

Vintzeleou's formalistic model can be easily implemented into numerical

programs. There still is, however, a need for a more theoretical ap­
proach of the behaviour of a dowel subjected to cyclic loads. Therefore,
an attempt is made to derive a model, which is to a large extent based
 - 112 -

upon material behaviour. Because of the fact, that previously mentioned

dowel action tests did not provide information of the local response de­
gradation of the concrete underneath the bars, this model will also be
to some extent based upon empirically derived expressions.
For the case of a monotonie loading, the dowel response can initially be
predicted by regarding the bar as a beam on elastic foundation, see Sec­
tion A.3. With increasing dowel force a plastic hinge develops. This
plastic hinge is situated at a distance to the crack plane, which is ap­
proximately equal to the bar diameter. Now, the bar displacement is
largely determined by the response of this part of the bar. Underneath
the bar, high triaxial stresses support the bar. After reaching the max­
imum dowel force, the force decreases. Now, the triaxial stress state
first rapidly looses its confinement. In consequence, the dowel force
can decrease without (hardly) any restitution of the concrete deforma­
tion, see Fig. 5.18.

Dowel force F^ [kNI

F
d ~Z^\
075Fd --/*-- J

shear displacement 6. [mm]

Fig. 5.18. Mechanism of loading and unloading in the first cycle.

The drop in concrete stress is strongly related to the loading path of

the concrete. For triaxially loaded concrete, Van Mier [44] found that
at the moment of unloading a sudden stress drop occurs. This will cause
a drop in dowel force, which appeared to be approximately 25 percent of
the maximum dowel force. After that, the shear displacement decreases
with decreasing dowel force. Some non-recoverable shear displacement re­
mains for zero-dowel force, which is about 25 percent of the maximum
shear displacement. For a fully reversed dowel force, the deformed
matrix material after unloading is shown in Fig. 5.19.
Upon re-loading in the positive direction, the bar initially responds as
a beam partially fixed at one side, see Fig. 5.20. For this loading
 - 113 -

Fig. 5.19. Matrix deformation after the first fully reversed load cycle.

case, the dowel stiffness is determined by linear elastic material be­

haviour and can be expressed by:

9 è" E
K, [N/mm] (5.5)
120 *L2 + 40 L3

Fig. 5.20. The initial dowel stiffness K, for subsequent cycles.

Unfortunately, the magnitude of L cannot be derived theoretically.

Therefore, Vintzeleou [75], performed dowel action tests on specimens
having a cylinder strength of 30 MPa, a steel yield strength of 420 MPa
and maximum dowel force equal to 80 percent of the dowel strength. For
imposed displacements, the response degradation of the concrete causes
an increase of the length L with cycling. Eq. (2.9) accounts for this
decrease. The following empirical relation was derived for L:

3.5 * fi (n-1) [mm] (5.6)

t ,max

with & = maximum shear displacement in previous cycle.

t,max
n = number of cycle; Note that the first load cycle in the
negative direction is cycle 2. In consequence, the second
 - 114 -

cycle in the positive direction is cycle 3. For the case,

that the imposed displacement in the negative direction is
for instance 50 percent of the displacement in the positive
direction, the cycle in the negative direction is cycle
1.5. The second cycle in the positive direction is then
cycle 2.5 and so on.

0,1
The term (n-1) ' in eq. (5.6.) accounts for the fatigue of the concrete
underneath the bar. Because of the fact that for increasing dowel
displacements, previously strained matrix material is crushed and unaf­
fected material is loaded, the number of cycles can be reset to 1 for
the case of load-controlled tests.
The stiffness Kj determines the dowel response until the shear displace­
ment & is reached, see Fig. 5.20. For this shear displacement, the
t ,o
bar is supported by the concrete over the length L-$, see Fig. 5.21.

Fig. 5.21. The dowel stiffness K 2 .

Due to the support of the concrete, the dowel stiffness increases to be­
come equal to stiffness K 2 . The response of the concrete is approximated
regarding the bar as a beam on an elastic foundation. According to eq.
(4.30), the coefficient of the subgrade reaction is:

* -0 78
390 (« t ) ' [MPa/mm]

However, in this expression the shear displacement does not include the
*
slip related to Kj. Substitution of 6 by an average value of
0.3 6,t,max yields:

K f = 998 U ) [MPa/mm] (5.7a)

f t,max
 - 115 -

Because the subgrade reaction is proportional to the modulus of elasti­

city, which is approximately proportional to the square root of the con­
crete grade, the following expression is obtained:

K= 168 /I (6. )" " [MPa/mm] (5.7b)

f ccm t,max

Combination of eq. (5.7b) and eq. (4.3) yields:

1 75 0 375 -0 59
F = 166 * " f * & ' A6^ [N] (5.8)
d ccm t,max t

However, according to Fig. 5.21, there is an eccentricity e equal to the

bar diameter. For this case, the shear displacement is expressed by:

F.
«t= 3 B° E I (3 + 6Be + 6B 2 e 2 + 263e3) [mm] (5.9)

For an average value of Kf = 300 MPa/mm, it was found that Be is ap­

proximately equal to 0.6. Now, the total shear displacement is about
three times the slip according to eq. (5.8). Accounting for this, the
dowel stiffness K2 becomes:

1 75 0 375 -0 59
K,= 55 ♦ f 6 [N/mm] (5.10)
2
ccm t,max

This stiffness is valid until the plastic hinge develops. Because of the
fact, that in this stage there is not yet a contribution of the concrete
to the section modulus of the bare bar, the dowel force at the onset of
yielding is:

F, = M /e = 0.1 * 2 f [N] (5.11)

d,sy sy sy

Due to the development of the plastic hinge, the bar makes contact with
the concrete close the crack plane, see Fig. 5.22. For the coefficient
of subgrade for this part of the concrete again eq.(5.7b) can be used.
This part of the concrete is now determining the dowel stiffness.
According to the observation, that about 25 percent of the maximum shear
displacement is non-recoverable, it is assumed that the deformation of
the concrete over the length L-* contributes 25 percent of the total
shear displacement. Thus, the dowel stiffness K3 is expressed by:
 - 116 -

1 75 O 375 _0 59
K,=
3 0.75 . 168 <j> " f ó ' = (5.12)
ccm t,max
1 75 0 375 -0 59
= 126 4> * f 6 [N/mm]
ccm t,max

plastic hinge

Fig. 5.22. The dowel stiffness K 3 .

An additional condition is that the total dowel force is always less

than 85 percent of the dowel force for the monotonie case at the same
shear displacement. The total loading path according to the proposed
model is shown in Fig. 5.23. Unloading causes a drop in dowel force of
25 percent of the maximum value. Next, the unloading stiffness is equal
to K,
3 until F. is reached. Finally, the stiffness K,z is used to
d,sy •"
obtain zero-dowel force. For the case of a repeated dowel force, the
response is calculated according to the model for reversed dowel loads.
However, the stiffness K 3 is used to connect the zero-stress state to
the re-loading branch, see Fig. 5.23a. Apart from that, another restric­
tion should be made. The dowel force is not allowed to exceed the magni­
tude of 85 percent of the dowel force for the monotonie case, see Fig.
5.23b. The stiffness is then equal to the stiffness K,,, which was found
for the monotonie case for the given shear displacement.

Dowel lorce F^ Dowel force hj

fa

0.75Fd
K
/ /
y/
1h
-ow\, II
*<i/s'
K
l^^y / KT.repeated load
-" IS
shear displacement 6i shear displacement 6.

a. Proposed model. b. Additional condition,

Fig. 5.23. Loading and unloading according to the model.

 - 117 -

Pig. 5.24 presents a comparison of this model with experimental results

of Vintzeleou [75] and Jimenez [32]. There is a reasonable agreement
with the experimental results. Jimenez's test result shows that the
model can also be applied to the case of imposed dowel loads.
From the monotonie dowel tests of Eleiott [16], it is known that the
dowel stiffness is strongly influenced by the magnitude of the axial
steel stress, see Fig. 2.32. For the case of an axial steel force, the
following expression is proposed to reduce the dowel stiffness:

U (5.13)
V - F^>
sy

dowel force Fd 1 k(s| dowel force Fg- [kN)

'ccyl 21MPO
f5y =455MPa
»29

K,

/
n=1 /
K ? /
It

IM
0.25 0 50
shear displacement
0.75 100
Of [mmI
0
H. 0.25
/
0.50 0.75 100
shear displacement 6^ [mml
a. Experiment of Vintzeleou b. Experiment of Jimenez.

Fig. 5.24. Comparison of experimental results with the proposed model.

For the case of cyclic dowel action tests with an initial axial steel
force, the dowel stiffnesses K,, K2 and K3 are simply multiplied by the
dowel stiffness reduction factor y. • In Fig. 5.25 a cyclic dowel test of
Eleiott (see Fig. 2.35) is simulated with the proposed model. Cycle No.
16 showed a underestimation of the energy-dissipation. The over-all
response is, however, satisfactorily simulated.
It must be born in mind that this model is partially based upon empiri­
cal expressions, thus limiting its application.
With respect to the proposed model, a special situation arises when the
 - 118 -

, dowel force FH [kN]

0.25 0.50 075 100

shear displacement 6. (mm]

Fig. 5.25. Comparison of Eleiott's test result with the proposed model [16].

dowel strength F, is obtained in a cycle. Now, the whole cross-sec-

du
tional area of the bar will yield. For the case of a reversed dowel
force, the only restriction is that the force cannot exceed the dowel
strength. However, for the case of a repeated dowel force, the yield of
the bar strongly influences the bar stiffness. For this case, the bar
remains in close contact with the supporting concrete, thus approximat­
ing the situation in the first load cycle. Because of the already de­
formed bar, 'this can be regarded as a shift of the initial stiffness of
the bar, see Fig. 5.26. Now, the force-displacement relation for the
static case must be applied.

Dowel force Fc

shear displacement 6(

Fig. 5.26. The model in case of reaching the dowel strength.

 - 119 -

For very large fully reversed shear displacements, the bond between the
bar and the concrete will diminish. The magnitude of the ultimate dowel
force is, however, strongly related to the cooperation of the bar and
the supporting concrete. Generally, the high radial contact stresses
will prevent slip of the bar relative to the concrete. For the case of
large reversed shear displacements, the residual elongation due to the
decrease in bond increases. In consequense, the bar shape and the shape
of the supporting concrete will be different, thus reducing the coopera­
tion of the bar and the concrete. Although there is no experimental
proof found yet, the magnitude of the ultimate dowel force can decrease
due to this 'lack of fit', see Fig. 5.27. According to the proposed
mechanism presented in Chapter 4, a strength reduction up to approxima­
tely 15 percent is possible. For the case of repeated shear loads, the
bar remains more or less in close contact with the concrete, thus pre­
venting this reduction.

Fig. 5.27. Strength reduction due to 'lack of fit'.

Till now, no cyclic dowel tests have been performed with a 'low-intensi­
ty' dowel force. Because of the fact, that for 'high-intensity' dowel
tests the response degradation during cycling is very similar to the
behaviour for the aggregate interlock mechanism, it is expected that
this holds true for 'low-intensity high-cycle' fatigue. For practical
use, however, the response of both interacting mechanisms in cracked
reinforced concrete is of much more interest. The combination of both
the transfer mechanisms will be discussed in Section 5.A.
 - 120 -

5.4. The combined mechanism of aggregate interlock and dowel action.

The mechanism of aggregate interlock for the case of cyclic shear loads,
as presented in Section 5.2 and the mechanism of dowel action for the
cyclic loads as presented in Section 5.3, will be combined to describe
the case of cyclic shear loads applied to cracked reinforced concrete.
In Chapter 4, it was shown how these mechanisms influence the crack
opening direction and the relation between Che stresses and displace­
ments in the crack. It appeared that the bond characteristics obtained
in an ordinary pull-out experiment, cannot be applied to this case.
Fortunately, the static crack opening path can be determined without
exactly knowing the bond characteristics of the reinforcing bars.
In Section 5.2, Laible's cyclic aggregate interlock tests have been
satisfactorily described by the proposed aggregate interlock model.
However, the relationship between the normal stress and the crack width
was input in the calculations. Therefore, this relationship must also be
known for the case of cyclically loaded cracked reinforced concrete. The
normal stress was not recorded during the tests reported in the
literature.
normal stress c**n

'ccm » 50 MPa
«8

crack width 6 n

Fig. 5.28. The relation between normal stress and crack width during
cycling [21].

In order to obtain information on this normal restraint stiffness for

reinforced concrete, some additional tests were performed with bars
prepared with a bolt gauge situated in the crack plane. The testing pro­
cedure is roughly discussed in Section 4.5 and in detail in [21]. Apart
from the static tests, two specimens were subjected to a repeated shear
 - 121 -

load. From these tests, it appeared that the relation between the normal
stress and the crack width is almost linear, see Fig. 5.28. However, as
stated before in Section A.5, the magnitude of the measured normal
stress is largely influenced by the yield strength in the plastic
hinges. Therefore, a linear relationship between the normal stress and
the crack width is assumed. The magnitude of the normal restraint stif­
fness is, however, derived from Fig. 4.38b. It was found, that the nor­
mal stress can be approximately related to the crack width according to:

o = a pf (S -6 ) [MPa] (5.14)
n sy n n,o

with a = 0.25 [mm ] for 'low-intensity high-cycle' fatigue

bond stress [MPol stiffness Kh IMPo/mml

90
■
\
\
\
J^

'"•-' ...
30

06 08 6 8 10 12 14
bond slip Imml number of cycles n [cycles)

Fig. 5.29 The response degradation of a bar in a cyclic pull-out test [21].

For the case of 'high-intensity' fatigue, the normal stiffness will be

initially higher due to the large displacements, but will decrease in
the subsequent cycles to the value of 0.25 pf . Cyclic pull-out tests
on similar bars [21], show a decrease of the bond behaviour during cycl­
ing, see Fig. 5.29. Assuming a similar response degradation of the bond
for the bars used in the push-off elements, a is expressed by:

3/(n+2) > 0.25 (5.15)

with n = number of cycles (fully reversed)

 - 122 -

The normal stiffness according to the eqs. (5.14)-(5.15) was applied to

the tests of Jimenez et al. [37]. The cylinder compressive strength of
the concrete used was approximately 23 MPa, the steel yield strength was
455 MPa. In test No. 5, the initial axial steel stress was 331 MPa. The
specimen was reinforced by means of four 22 mm diameter bars (p = 1.08%).
The experimentally obtained and the calculated response for cycle number
15 is shown in Fig. 5.30. The calculated response is in reasonable agree­
ment with the experimental result, although the calculated response is a
little bit too soft. The end-displacements in each cycle are satisfacto­
rily predicted, see Fig. 5.30. The load was fully reversed, but only the
response in the positive direction is shown. The crack, response in the
opposite direction might be slightly different due to the position of the
bars with respect to the casting direction. Because of this, the bars can
have a different support of the concrete in both directions.

shear stress x IMPal

cracW width Bp (mm!
1.0
n=t h$
0.8
0.6 25**? I I I
! „ xm t
m
0.4
0.2

|i 0
5 10 15 20 25
number of cycles n [cycles]
Id
tm

shear displacement 6t 1 nm)

i 1.0 050
i
0.8
i i *m
0.6
0.25
/
ƒ / ''If'
/fi
experiment
- - - model
0.4
0.2
0
0.2 04 0.6 0.8 1.0 5 10 15 20 25 5 10 15 20 25
shear displacement 6^ (mml number of cycles n [cyclesl , n [cycles]

Fig. 5.30. Test No. 5 of Jimenez [37] compared with the proposed model.

The calculation yielded information about the contributions of both

transfer mechanisms to the externally applied load. It appeared that the
contributions of both mechanisms remained nearly constant during cycl­
ing. For this experiment, the aggregate interlock mechanism transfers
approximately two-third of the total shear load.
An important criterion for a model is its sensitivity to the magnitude
 - 123 -

of the load or displacement increments used in the calculation process.

For the calculation shown in Fig. 5.30, the displacement increment was
0.01 mm. The sensitivity of the model was investigated by performing the
same calculation with two different displacement increments,
A6 = 0.02 mm and AS = 0.002 mm respectively. The results are shown in
Fig. 5.31. The differences between both calculations are in fact negli­
gible.

shear,stress T [MPal
crack width 6 n [mml

la.Id.

T
d
Im
p 5 10 15 20 25 °75
number of cycles n [cycles]

shear displacement of [mm] n

1.0
0.8 'a
tm
0.6 ■kTtSl

0.4
^"*H
0.2
H
0.2 0A 0.6 0.8 1.0
shear displacement fc^ [mm)
0
1
5 10 15 20 25
number of cycles n Icyctesl
0 5 10 15 20 25
n [cyctesi

Fig. 5. 31. Test No. 5 with different displacement increments.

Although the model is apparently rather intensitive for the magnitude of

the displacement increment A6, the increment must be small with respect
to the actual maximum displacements. Fig. 5.32 shows the comparison of
the model with the experimental result of test No. 6 of Jimenez. Now,
the initial crack width was 0.51 mm. The bar diameter was equal to 29 mm
(p = 1.82%), the initial steel stress was 227 MPa. Because of the ex­
pected small increase in displacement in each cycle, the displacement
increment A6 was chosen equal to 0.005 mm. Again, the end-displacements
are satisfactorily predicted by the model. However, the crack response
in the 15th cycle was initially too stiff. For this test, the contribu­
tion of the dowel action mechanism to the transfer of the externally
applied shear stress was approximately 60 percent and remained nearly
constant during cycling. Due to the small contribution of the aggregate
 - 124 -

interlock mechanism, the calculation process became even more insensi­

tive for changes in the displacement increment than was the case for
test No.5.
shear stress x IMPol
crack w dth 6n [mml
1.0
0.8
n= 1 15
n 0.6
1.00
iTf l . i l
m *m
i 0.2
i 0
i 5 10 15 20 25 0.75

/ 1
i number of cycles n [cycles] td
tm
I 1 i shear displacement 5( [mm]
i 1.0
/ 1
0.8
ƒi
i 1 0.6
0.25
02 ' /) 0.4
\lr
vff '
— experiment
model 0.2
;tm

0
0.2 0.4 0.6 0.8 1.0 5 10 15 20 25 5 10 15 20 25
shear displacement 6, (mm) number of cycles n [cyclesl n [cycles]
Fig. 5.32. Test No. 6 of Jimenez [37] compared with the model.

A further decrease of the initial steel stress will provide a stiffer

response of the crack to cyclic shear loads. Fig. 5.33 presents the
comparison of the experimental and calculated result of test No. 7 of
Jimenez. The initial steel stress in the 29 mm diameter bars was 151
MPa. The calculated crack response is in good agreement with the ex­
perimentally obtained response. The contribution of the dowel action
mechanism, is 50 percent of the total shear stress. Again, this contri­
bution remained approximately constant during cycling. Because of the
o.i
very small displacements, a factor (n-1) was applied to the dowel
stiffness to account for the fatigue of the supporting concrete. Each
fully reversed cycle was counted as one cycle.
The combined model proposed in this Section can of course be applied to
the case of 'low-intensity high-cycle' fatigue. This type of experiments
is described in Chapter 3. The experimentally obtained crack displace­
ments must however be larger than the smallest displacements, which can
accurately be predicted by the model. Therefore, some cycles approaching
shear failure are simulated to ensure a sufficient increase in the crack
displacements in each cycle. Again, the normal restraint stiffness ac­
cording to eq. (5.14) was used in the calculation.
 - 125 -

shear stress t [MPa!

crack width 6 n [mm]

n= IS
la,id.

j f
1

1.0
0.8
0 5 10 15

shear displacement Bt (mml

20 25
number of cycles n (cycles!
0.75
tm

0.6 a
m
0.25
0.2 0.4
— experiment
0.2
f1 model
0
0 0.2 0.4 06 0.8 1.0 0 5 10 15 20 21 0 5 10 15 20 25
shear displacement 6^ [mm] number of cycles n (cycles) n [cycles!

Fig. 5.33. Test No.7 of Jimenez [37] compared with the model.

First, the cycles 640 and 1620 of test No. A/6H7.66/7.9/.03 are consid­
ered (page 57, [56]), see Fig. 5.34. The maximum applied repeated shear
stress during cycling was 7.9 MPa. The cube compressive strength was
48.0 MPa. The crack plane was reinforced by means of twelve 8 mm diame­
ter bars with a yield strength of 550 MPa. For the cycles considered,
the shear displacement largely exceeds the shear displacement, for which
the ultimate dowel strength is obtained for a monotonie increasing shear
load. Therefore, it is expected that a large plastic deformation has oc­
curred in the plastic hinges in the reinforcing bars. For the calcula­
tion process, this plastic deformation is accounted for by applying the
static dowel action model to the measured shear displacement at zero
stress according to Fig. 5.26. Furthermore, the measured crack width at
zero stress was input in the calculation. The theoretical results are
also presented in Fig. 5.34, showing a reasonable agreement with the
experimentally obtained results. Because of the large contribution of
the dowels to the transfer of shear stress, restitution of the shear
slip during unloading starts at 75 percent of the shear load, as is
predicted by the dowel action model.
 - 126 -

shear stress I IMftj) shear stress t (MFtal

1680 n = 640 1680

11
'n lb

exp.
r 1
ril
fit
A
>/}
is
7 1
jI 1
ill
/111
/1' 1
11 r
'II

1
1
III 2.5

1
hi ft
1
jl model m
1
0 0.25 0.50 075 1.00 0 0.25 0.50 0.75 1.00
crack width 6 n ImmJ shear displacement 6j Imml
a. Shear stress versus crack width b. Sheor stress versus shear slip

Fig. 5.3A. Test No. A/6H/.66/7.9/.03 compared with the proposed model [56]

Fig. 5.35 presents the comparison between the model and the experimental
result for test No. B/AL/.81/9.1/.04 (page 72, [56]). Now, the cube com-
pressive strength was 68.0 MPa. Eight 8 mm diameter bars with a yield
strength of 460 MPa were used. The maximum applied repeated shear load
was 9.1 MPa. Again, it was found that the dowel stress reaches its ulti­
mate value. As for the computations on plain concrete test results, it
appeared that there is no decrease in matrix strength. For this spe­
cimen, the restitution of the shear slip during unloading started at 55
percent of the shear load being the average of 75 percent according to
the dowel action mechanism and 40 percent of the aggregate interlock
mechanism.

For this type of tests, in which the contribution of the dowel mechanism
to the shear transfer is equal to its ultimate value, an interesting
scenario for the tests can be found. During the first few cycles, the
mechanism of aggregate interlock transfers the difference between the
applied shear stress and the ultimate dowel stress. The combination of
the end-displacements in these cycles will be determined by the matrix
strength and the maximum particle size according to Fig. 4.4. Because of
the fact, that the contribution of aggregate interlock to the shear
stress transfer remains constant during cycling, it can be expected that
the crack opening path is determined by a constant ratio of i to o
r
' a pu
 - 127 -

shear stress I IMftil shear stress T IMFtal

n = 1020 wo

ill
;'1
. h\
In,
I'V
\\'
/;/'
lit
Ilk

ItIII
1
W'

0 0.25 050 0.75 1.00 0 0.25 050 075 1.00

6 [mm] shear displacement 6t 1mm 1
a. Shear stress versus crack width. b. Shear stress versus shear slip.

Fig. 5.35. Test No. B/4L/.81/9.1/.04 compared with the proposed model [56],

With this assumption, it can be easily checked whether the tests de­
scribed in Chapter 3 are in agreement with the proposed model. For all
the tests, the crack opening path should follow the theoretical opening
path for a constant ratio (T -T )/O . In Fig. 5.36 a few typical test
m du pu 'r
results are compared with this assumption. It was found that for a total
of 42 repeated loading tests, 16 experimental crack opening paths are in
close agreement with the theoretical opening palh. There is a reasonable
agreement for three tests, while the difference between the result of
six tests and the model can easily be explained. The measured displace­
ments of 12 experiments were too small to drawn any conclusions. In con­
sequence, only five measured crack opening paths showed large deviations
from the theoretical crack opening paths. From these results, it can be
concluded, that also in the case of 'high-cycle' repeated shear loading
the crack opening direction is determined by a constant ratio between
T and the matrix strength o
a pu
According to the model, it can be concluded that the crack displacement
increments are in fact to T la
related
rather than to T /o . which
a pu m pu'
was used in the eqs. (3.3a-b). Therefore, empirical results similar to
the eqs. (3.3a-b) are derived relating to the ratio of T to the matrix
a
strength:
 128

sheor displacement 6t 1mm] sheor displocement 6t Imml

0.75

050
exp f K
model \
075

050
model
,/
//
exp
//
//

//
//
//
0.25 0.25
//
s' /
•/

0 025 0.50 075 1.00 0 0.25 0.50 0.75 100

crock width 6 n ImmJ crock width on Imm!
a. Test No. A/6H/.66/79I.03. b. Test No. AKH/.78/B0I.04

Fig. 5.36. The experimental crack opening path compared with the opening
path according to the model.

T T
a 1 6 n 2 9 6 T- 6 9
«n= 3.34 (^_) * + (130(^_) * + 2.6.10 (^-) ' ) log(n)
pu pu pu
33 a 3M 10
(y-)
+ 2.1.10 (log(n)) (mm) (5.16a)
°pu
T
a ^ . 6 . ,,,,Xa 2
N . 9 . , „ , „ s / a .6.9.
< = 3.34(-2-) * + (61(-5-) ' + 1.2.10 (-5-) • ) log(n)
L O 0 fj
pu pu pu
T
19 a ,19 * 5
+ 2.2.10 (^_) (log(n)) [mm] (5.16b)
pu
. n _ crock displocements 6 Imm)

10' I05 10°

number of cycles log(n) [cycles!

Fig. 5.37. The relationship between the number of cycles and the crack
displacements as a function of the stress level.
 - 129 -

with o pu = A.5 f ° ^ ? [MPa] (5.17)

Eq. (5.17) is a slight modification of eq. (A.lc), to ensure that the

matrix strength is equal or higher than the compressive strength of the
concrete for very high concrete strengths. Fig. 5.37 presents the rela­
tion between the stress level and the crack displacements during cycling
according to the eqs. (5.16a-b).

5.5. Influence of the normal restraint stiffness upon the shear

stiffness for the case of repeated loading.

In this Section, a description will be given of the influence of the

normal restraint stiffness upon the shear stiffness of cracks in rein­
forced concrete subjected to a repeated shear loading.
In addition to the example presented in Section A.5, the same case will
be used here. Because of the small increase of the crack width during
the first load cycle (S : 0.106 mm to 0.122 m m ) , the development of the
inclined cracks is ignored. It is assumed that unloading starts at a
crack width of 0.13 mm. Now, it is possible to investigate the effect of
a crack width increment during re-loading.
The top load placed upon the wall between the two storage tanks is re­
moved. Due to the elastic deformation of the bars and the high normal
stress upon the cracks, the crack faces will partially slide back. For
the static case (cycle 1 ) , it was shown that the plastic hinges in the
reinforcing bars crossing the cracks have fully been developed. Because
of the large shear displacement, which occurred after these hinges have
been developed, the residual shear displacement according to eq. (5.Ac)
cannot be applied to this case. The shear displacement at zero shear
stress is therefore estimated to be equal to 0.16 mm (approximately
equal to & - 6 ) . Because of the plastic hinges, the static dowel
r
^ t,max t,e °
action model is applied with a reduced value of the shear displacement
(6 - 0.16 mm) according to Fig. 5.26. The crack width at zero shear
stress is taken equal to the initial crack width. Furthermore, it is
assumed that the normal restraint stiffness is not influenced by the
unloading and re-loading of the wall. Therefore, the normal stress -
crack width relation is the same as for the static case.
 - 130 -

Table 5.1 presents the calculated results for the crack response during
re-loading the wall. The reduced contact model is applied for the aggre­
gate interlock mechanism. The shear stress - crack width relationship
and the crack opening path are shown in the Figs. 5.38a-b.

Table 5.1. Crack response according to the reduced contact model.

6 T 0 T
n \ a a d
[mm] [mm] [MPa] [MPa] [MPa]

0.11 0.19 0.57 0.64 2.03

0.12 0.22 4.55 2.24 2.49
0.13 0.25 10.35 3.95 2.86 ,

shear stress t (MFtal shear displacement 6t Imni

0.30
/
/
/ i
j
I
0.20
n//,i
l
n = 1 cX
i
i
0 005 0.10 015 0.20 0 0.05 0.10 0.15 020
crack width 6 n Imml crack width Gp (mm)

a. Shear stress-crack width relation. b. Crack opening path.

Fig. 5.38. Calculated response of the cracks in the wall during re­
loading.

Table 5.2. Crack response according to the analytical contact model.

6 T 0 T
n
\ a a d
[mm] [mm] [MPa] [MPa] [MPa]

0.11 0.20 0.58 0.69 2.03

0.12 0.23 2.37 2.44 2.49
0.13 0.24 8.91 3.95 2.86

The increments of the crack displacements during re-loading deviated

from the results, on which the empirical relations for the retention
 - 131 -

factors X and X are based. In order to investigate the sensitivity of

x y ° . '
these retention factors for different crack opening paths, the crack
response is also calculated according to the analytical contact model,
see Table 5.2. The agreement between the results of both methods is
satisfactory.

5.6. Concluding remarks.

In this Chapter, the static aggregate interlock and dowel action models
are adapted to the case of repeated and reversed shear loads.
Walraven already showed that the two-phase aggregate interlock model can
be applied to the case of cyclic shear tests on cracked plain concrete.
Therefore, all effort is paid to a sound simplification of the two-phase
model in order to speed up the calculation process. For increasing crack
displacements during cycling, the proposed reduced contact model pro­
vides good predictions of the crack response during cycling. However, a
simplified model cannot be used being as general as the original two-
phase model.

For practical use, the response of cracked reinforced concrete is of

much more interest than the response of cracked plain concrete. Now, the
contribution of the dowels to the shear stress transfer must be taken
into account. However, even the simple reduced contact model is rather
sophisticated with respect to the available models describing the cyclic
dowel action behaviour. Therefore, the use of the reduced contact model
to describe reinforced concrete tests will introduce smaller deviations
from the actual response than the deviations caused by the dowel action
model.

The existing cyclic dowel action models are formalistic. In Section 5.3,
a dowel action model is proposed, which is to some extent based upon the
physical dowel behaviour. Due to a severe lack of detailed information
on the actual bar behaviour, this model is still very simple and there­
fore limited in its applications.
The reduced contact model and the proposed dowel action model are com­
bined to describe the response of cracked reinforced concrete. The com­
bined model satisfactorily predicts the experimentally obtained crack
response for the case of 'high-intensity low-cycle' fatigue and for the
case of 'low-intensity high-cycle' fatigue.
 - 132 -

All the simulated tests were subjected to shear loads with a stress ra­
tio R (T . /T ) being equal to 0 (repeated load) and -1 (fully re­
r
min max '
versed load) respectively. It was found from the calculations, that
there was no strength degradation due to fatigue of the matrix material.
This can be explained by the large increase in crack displacements. The
previously loaded matrix material is then crushed. For stress ratios R
in the range of -1 to 1, fatigue of the matrix material will affect the
crack response. A sustained shear load (R = 1) in the range of 60 to 90
percent of the static shear strength causes a gradual increase of the
crack displacements in time, Frenay [20]. This increasing displacements
must be caused by strength reduction and material flow in the contact
areas between the particles and the matrix material.
A stress ratio of 0.4-0.5 will not cause a restitution of the displace­
ments during unloading, see Fig. 5.1. Any increase in the displacements
due to re-loading must then be caused by strength reduction due to fa­
tigue.

°nx]x^ 'ccm °fno ' 'ccm

1.0
\
^
^\
^. " V , A 08 ^ fc^ ~"i
0.8
R=
s8

1
"'s. .6 —6
V4
-~.2
* V ? \
^v •^.0 0.6
V *.0 Vo
04

02
freq. 6 cps freq. 6 cps freq 07cps
cured wet cured wet cured wet
0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 -2 3 4 5 6 7
log(nfl Icyclesl logtrif I [cycles] logtrif) [cycles]

Fig. 5.39. The fatigue strength of concrete [38],

The phenomenon, of concrete fatigue is investigated by among others Van

Leeuwen and Siemes [38]. For various types of concrete grade, curing
condition and age, Wöhler curves were determined for several stress
levels and stress ratios R, see Fig. 5.39a-c. It was found that the
fatigue strength increases with increasing concrete strength, but to a
lesser extent than the concrete grade. The frequency of the stress
 - 133 -

cycles was found to affect the fatigue strength, see Fig. 5.38a and c.
The lower the frequency, the lower the fatigue strength. For very high
stress levels (> 75% of the static strength) the strength reduction
increases progressively. This phenomenon is probably caused by creep
effects. These findings are in agreement with the experimental results
of Holmen [30] and Graf et al. [27]. It must however be noted that the
experimentally obtained material characteristics of concrete cannot be
applied directly to the matrix material. Fatigue of concrete is partial­
ly due to the interfacial bond between the stiff particles and the ma­
trix material. The matrix material has a relatively large amount of air
voids and water inclusions, which affects fatigue. However, the overall
characteristics of concrete and matrix material will be similar.
Even if a proper description of the fatigue of the matrix material due
to the high cyclically applied compressive stresses is implemented in
the model for the crack response to cyclic shear loads, it is doubted
whether such a model can be used to describe the crack response degrada­
tion for small stress ratios R. The increase in crack displacements dur­
ing cycling will decrease with decreasing shear stress level and with
decreasing stress ratio R. These small crack displacements cannot be
simulated with the proposed model. Therefore, further experimental in­
vestigations into this field are necessary.
 - 134 -

6. IMPLEMENTATION OF THE CYCLIC AGGREGATE INTERLOCK MODEL INTO

NUMERICAL PROGRAMS

6.1. Introduction.

In the previous chapters, the mechanisms of aggregate interlock and

dowel action are described for both cases of monotonie and cyclic shear
loads on cracked reinforced concrete. Now, an attempt is made to imple­
ment the models developed into advanced finite element programs. It must
be noted that the implementation of the mechanism of dowel action is
strongly related to the way reinforcing bars are treated in numerical
programs. In these programs, the bars are tied to plain concrete ele­
ments. This can be done directly or by using a boundary layer between
the bars and the ordinary concrete elements [61,79]. The numerical de­
scription of the static bond stresses between the steel bar and the con­
crete is rapidly improving [61], but a very fine mesh is needed for such
a detailed description. For practical use in elements intersected by re­
inforcing bars, the element strain is also applied to the bars. In such
programs as proposed by among others, Bazant and Gambarova [1] and
Vecchio and Collins [74], the combined stiffness of steel and concrete
is accounted for by a tension-stiffening parameter. The contribution of
dowel action to shear stiffness of the cracked element is completely ne­
glected. Models developed by Fardis and Buyukozturk [17] and by
Perdikaris et al. [54,74] account for dowel action by means of a dowel
stiffness based upon the concept of a beam on an elastic foundation. The
interaction between an axial steel force and a dowel force is, however,
not yet considered. Apart from that, the applied bond characteristics
are derived for bars subjected to an axial force only. The bond capacity
is, however, reduced by a shear displacement of the crack faces, the
same holds true for bars inclined to the crack direction, if the crack
only widens. The static dowel action model described in Chapter 4 can
easily be implemented in any numerical program, when an appropriate bond
model is used. The cyclic dowel action model developed in Chapter 5 has
been already expressed by means of various dowel stiffnesses, which can
directly be implemented. Obviously, the implementation of the dowel ac­
tion model must be accompanied by the implementation of a proper bond
model for bars subjected to both axial and lateral forces. Such a bond
 - 135 -

model is, however, still lacking. Furtheron, for the case of monotonie
loading, the crack opening path is no longer related to the bond
stresses after the plastic hinges in the bars have fully been developed,
see Chapter 4.
Finally, the plain concrete on the one hand and the reinforcing bars on
the other hand represent two different mechanisms, which must be de­
scribed by means of two different stiffness relations.
Therefore, the aggregate interlock mechanism and the dowel action mecha­
nism will be treated separately. For the mechanism of dowel action a
rough description of the stiffness relation will be given because of its
dependency on the (unknown) bond characteristics.
Although the static two-phase aggregate interlock model satisfactorily
describes the physical behaviour, it is still too complex for imple­
mentation into advanced finite element programs. Therefore, first em­
pirical relations will be derived based on the two-phase model. These
expressions will be used to describe the stiffness relation between the
stresses and the displacements in a concrete element. Numerical programs
can be subdivided into two types. The first type, denoted as discrete
crack program, accounts for the development of cracks by defining new
element boundaries along the crack or by adding the crack displacements
to the displacements of the whole element. The second type of program
uses the smeared crack approach. Now, the entire cracked element is
still considered as a continuum. With this concept crack displacements
are converted into strains of the entire element or Gauss point. There­
fore, the expressions, which are based upon crack displacements, should
also be converted to strain based formulas.
In a finite element program, the calculation starts linear elastically.
With increasing external loads, the tensile strength is reached in a
given Gauss-point. Then the uncracked element becomes partially cracked
due to the development of micro-cracks. On increasing tensile strain the
damage will affect the whole area; the element is then fully cracked. In
this chapter, an attempt will be made to describe all these three stages
with only one stiffness relation. It will be shown how the interaction
between tension-softening and shear-softening affects the element stiff­
ness for the case of monotonie loads. Finally, the stress-strain rela­
tions are adapted to the case of cyclic loads.
 - 136 -

6.2. Simplified expressions for the static two-phase model.

In order to keep close to the physical model of Walraven [81] and to

keep the expressions as simple as possible, expressions are derived for
the contact areas instead of for the stresses. With the magnitudes of
the contact areas and eqs. (A.la-d), the stresses in the crack can be
determined. The curves, which fitted closely to the theoretical results
according to the two-phase model are described by the following expres­
sions)

Pk K-l+exp(-K)
A = -^ (a / „si„^, + b.p) [mm2/mm2] (6.1a)
75 exp(-K)/K+l

with K = b/a . 6 (6.1b)

for A : a = 4 (6.1c)
X
0 056 -1 07
b = 7.00 D * 6 (6.Id)
max n
p = 0 (6.1e)

for A : a = 2 (6.If)
^ 0 280 m
b = 3.00 D 6 (6.1g)
max n
-o 063
m = -1.47 D (6.1h)
max
2
p = 0.5 ([6 -6 ]-abs[6 -6 1) . exp(-l-D /32-0.5 6 ) (6.1i)
n t n t max n

The limitations of the equations are:

- <S < 1 . 2 &

t n

- Particle distribution according to Fuller.

Because of the fact, that in finite element calculations, the crack dis­
placements are very small, the eqs. (6.1a-i) are adapted to provide a
closer fit at small shear displacements. The eqs. (6.Id) and (6.1g) are
altered:

7.74 D * 6'" (6.1j)

max n
 - 137 -

-o o i
m = -1.07 D (6.Ik)
max
o. 2 i m
A : b = 4.50 D 6 (6.11)
y max n
-0 03
m = -1.21 D " (6.lm)
max

Now, the limitations are:

- & < 0.2 mm

t

- & < &

t n

In the case of settlements large shear displacements can occur. Then,

the original formules according to the two-phase model must be used.

Next, the eqs. (6.1a-m) will be converted into strains in order to use
these expressions in numerical programs based upon the concept of
smeared out cracks.
The strains due to cracking can be expressed as:

e = 6 /h (6.2a)
nn,cr n

Ycr = «t/h (6.2b)

with h = size of the element normal to the crack.

The smeared out deformation can represent one large crack, but also two
or more smaller cracks, see Figs 6.1a-c. Both systems transfer the same
stresses. So, the concept of smeared cracks implies constant stresses
T and a for a given ratio of e to Y This can be expressed
a a nn,cr cr
by:
.h e
n _ nn.cr _ nn,cr _ ,, .. *
— = '—r— = ' — = constant (6.3)
S_ Y -h Y
t cr cr

then T = constant and o = constant

a a
 - 138 -

b. Single crack. c Two cracks.

Fig. 6.1. Philosophy of smeared out cracks.

Therefore, it is quite essential that the relations for T and o are

a a
dependent upon the ratio between the normal strain e and the shear
nn,cr
s0
sliding Y > that they will be independent of the element size. With
this restriction, it was found empirically on basis of the two-phase
model, that:
For A : K = 2.17 D * — (6.4a)
x max e
nn,cr
0.13 Yrr
For A : K = 3.74 D (6.4b)
y max enn,cr

The eqs. (6.1a-k) can directly be used in programs of the discrete crack
concept. For elements of the smeared-out crack type, the eqs. (6.4a-b)
have to be implemented into the stiffness of the whole element. There­
fore, a rheological model will be presented in the next Section, in
which the crack strain can be related to the element strain.

6.3. Rheological model for an element with the smeared out crack concept.

In an element a crack zone is formed, when the strain in this element

exceeds the tensile fracture strain of concrete. The behaviour of this
crack zone is determined by the development of small micro-cracks and
can be described by a tension-softening model. Due to this crack zone,
the stiffness of the whole element is strongly reduced. In order to
adapt the stiffness of the element to this reduction a normal retention
factor v is applied to the modulus of elasticity of concrete E , see
 - 139 -

Fig. 6.2. Actually, the softening of the element is localized at the

crack zone, see Fig 6.3. The shape of the descending branch depends upon
the element size L. Therefore, the stiffness of the descending branch
E t and the normal retention factor y are no real material parameters. In
fact, the observed softening probably is a structural effect due to non-
homogeneous deformations during cracking [6]. The stiffness E is used
to describe the incremental stress-stain relation of a partially cracked
element, the normal retention factor is used to account for the reduced
stiffness of the element. E can be expressed |i.
ssed as a function of u.
iffnn i°nn

tensile stress CTct(MPo] ' t ' I

^ _ — — ; —--& (micro) cracks

tensile strain Enn

a. Continuum. b. RheologiCQl model.

Fig. 6.2. Tension-softening behaviour Fig. 6.3. The rheological model.

of cracked plain concrete.

The reduction factor for the crack zone can be obtained from the
rheological model shown in Fig. 6.3b:

LAe = L AE + L Ae (e. s}
V0,J
nn co nn,co cr nn,cr '
with L = L ♦ L , see Fig. 6.3.
co cr °

Now, the normal retention factor for the crack zone r, can be expressed
by:

L L /(L + L ) p L
„ - c r / co co cr . c ,
cr
n
~ Lco (
~ 1-L co/(L co-Lcr) p ) =
—L ( 1-u+u L 7L (6.6)
cr

However, the length of the crack zone L is not accurately known. A

cr by Bazant et al.' [3]. This
length of approximately 3 D ^ ^ is suggested
length of approximately
length will be used here.
 - 140 -

It must be born in mind that eq. (6.6) is no longer valid when the
length of the crack zone is nearly equal to the length L (this can be
the case when very small elements are used).
Using the factor n, the weakening of the element is numerically local­
ized in the crack zone. Now, the rheological model will be extended us­
ing the philosophy of smeared out cracks. According to this philosophy,
the crack strain caused by the integrated action of the micro cracks can
be considered as the strain caused by a single large crack, so that the
crack zone can be regarded to be divided in a fully cracked part and an
uncracked linear elastic part with a reduced cross-sectional area, see
Fig. 6.4.

uncracked zone

tA

A uncracked zone

a. Continuum. b Reduced cross-seclion. c. Rheological model.

Fig. 6.4. Extended rheological model.

For this extended version of the rheological model, the incremental

stress-strain relation for each part of the element will be described;
finally the relation for the entire element will be derived.

Stiffness matrix for the uncracked section of the element.

To the uncracked part of the element, the linear elastic theory can be
applied, yielding:

Ao 1_U2 !_u2 Ac
nn,co
uE _E
Ao\ 1-u* 1-u2 "C (6.7)
tt tt.co
Ao 0 0 A
Y_
nt 2(l+u2)

Stiffness matrix for the uncracked part of the crack zone.

The stiffness matrix for the uncracked part of the crack zone corre­
sponds to the matrix of the uncracked section adapted to the softening
 - 141 -

by means of the parameter n!

nE vnE
Ao 1-nv2 1-riv2 Ae
nn,cr,co nn,cr
vnE E
Ao 1-nv2 1-nv2 Ae.
tt,cr,co tt,cr (6.8)
0 0 nE A
Ao 2(l+v2) Y,
nt,cr,co

The normal retention factor n is related to the normal retention factor

p of the whole element. This relation is expressed by eq. (6.6).

Stiffness matrix for the cracked part of the crack zone.

As shown in Fig. 6.1, the external shear sliding and normal strain can
be caused by a single large crack as well as a large number of very
small (micro-)cracks. According to this philosophy, the micro-cracks are
replaced by a single crack. Now, the two-phase model can be applied to
this crack. Because of the rather complex relations describing this
behaviour, the actual matrix will be represented by the following matrix
for use here!

Ao J
13 Ae
nn,cr,cr nn,cr
A<rtt,cr,cr 0 0 Ae
tt.cr (6.9)
Ao AY.
nt ,cr,cr

in which S n , S,3, S 3 1 and S 3 3 represent the stiffness relations

according to the two-phase model.

The stiffness relation for the entire cracked element.

Both springs in the rheological model representing the crack zone are
subjected to the same state of deformation. Hence, the eqs. (6.8) and
(6.9) can be directly added, thus yielding:
nE vnE
Ao ♦ S ii i. '* i - n v 2 U'S. Ae
nn,cr l-nv2 nn,cr
vnE . nE
Ao^ 1-nv 2 Ae.
tt,cr 1-nv 2 tt,cr (6.10)
Ao 0 s +
nt , c r
*s31 * 33 iffc^T AY.

in which I|I is a reduction factor for the cracked area, which is

expressed by:
 - 142 -

(f -o )
ctm nn
< 1 for partially cracked element (6.11)
f
„ ctm
for uncracked element

In [55], a detailed description of the derivation of the stiffness ma­

trix of the whole element is given. This stiffness matrix, which in­
cludes the stiffness relation of the uncracked section and of the crack
zone is presented by eq. (6.12). This matrix is derived by directly
adding the strain increment of the uncracked zone to the strain incre­
ment of the crack zone. This yields the strain increment of the entire
cracked element AE :

a,E
Ao 1-OjV 2 l-a,v2
AE

vcijE
(6.12)
Ao AE^
1-OjV 2 l-a,v2 tt
Ao BiG AY
nt

The factors a,, a 2 , 6, and B 2 are expressed by the following (approxi­

mated) relations:

nE + 4.S, i
a (6.13a)
'= (l+n)E + «jiSjj

pG + ipS33
B = (6.13b)
2 (l+n)G ♦ 4.S33
l-o. E nE + 41S.,
|_ nn . (6.13c)
2
° ~ 1-B2 Y (l+n)E + *S 1 !
1-8, Y nG + «pS3 3
8l= (6.13d)
l-o,2 1 nn (l+n)G + t
T|iS,,
33

Now, the stiffness relation for the entire plain concrete element is de­
scribed, except for the exact description of I lie terms S M > S 1 3 , S31
and S 3 3 . In [55], the derivation of proper expressions for these terms
is given according to the two-phase model. The following expressions are
obtained:

S,,=do /d (6.14a)
1 1 nn
nn nn

S,1 3J= <)° nn/dY (6.14b)

 - 143 -

S,,= do /de (6.14c)

31
nt nn
S„= do /dY (6.14d)
3J
nt

This yields:

S = o (dA /de - p dA /de ) (6.15a)

1
! pu x nn y nn

S13=-^nSll (6.15b)

S
3.= --T- S 33 <6-15c>
nn
S„= o (dA /dy + pdA /dy) (6.15d)
33
pu y x
in which: a, K according to the eqs. (6.19 and (6.4).
u = coefficient of friction = 0.4

dA/de =-aexP(-K)[K-lW-K)]^^__C_
2
nn [exp(-K)+K] e e
nn nn
r
nn

For A : C = 2.17 D°'° (6.15g)

x max °
A : C = 3.74 D * (6.15h)
y max
Now, an accurate description of the real crack, behaviour can be made. In
the following example, the interaction between tension-softening and
shear-softening is shown. An element is subjected to a given normal ten­
sile strain. Reaching a specific post-peak normal stress o (3.0, 2.0,
1.0 MPa), the normal strain is kept constant at increasing shear slid­
ing. The simple normal tensile stress strain relation used is presented
in Fig. 6.5a. The calculated results for the shear stress- relative
strain relation are shown in the Figs. 6.5b-d. In the figures, two dif­
ferent curves are drawn. The first is based upon the calculation ac­
cording to the expressions (6.15a-h). Second, the response is calculated
with the advanced models reported in the literature. In fact, till now
the interaction between tension-softening and shear-softening is ne-
 - 144 -

glected in the numerical programs. The shear softening is described by a

constant shear retention factor 62i with a value between 0.05 and 0.20.
More sophisticated expressions for 62 are derived by Rots [60], see the
eqs. (6.16a-b). The first expression is derived empirically from the ex­
perimental results of Paulay and Loeber [50]. The second is derived on
basis of the theoretical model of Bazant and Gambarova[3]:

1
B
(2) 1+4447E (6.16a)

(2) 4762e 1346 fl— (6.16b)

nn nn

normal stress a IMPOl , stress i , a iMPa) , stress t.oiMPo] , stress T, <r [MPa)
f ^ 30 MPa

"k <W,9mm
'iVi fig ' b
fd

1 Vfig. c.

1
\ f ig d

! \
/
001 III " 0 1.0 0 0.5 0 °ö
normal strain Enn'03 Y'Enn V'Enn
Tension-softening. b. O* : 3.0 MPa. c. rr = 2.0 MPa. d. 0" = 1.0 MPa.

F i g . 6.5. The calculated r e s p o n s e of a p a r t i a l l y c r a c k e d e l e m e n t to

shear sliding.

shear retention factor (3

1.0 1 1
1 eqs. of Ro s:
08

Bazant et al.
0.6

0.4
\l

0.2

0 2 1 J ! 10
normal strain C^IO 3

Fig. 6.6. The shear retention factor 6 [60],

 - 145 -

Both expressions are shown in Fig. 6.6. The calculated response of the
cracked element according to these formulas is also shown in the Figs.
6.5b-d. The shear stress is still reasonably well predicted. However, as
is shown in Fig. 6.5d, the description according to Rots still allows
the transfer of a constant tensile stress across the element . The model
presented in this chapter indicates that the stress normal to the crack
plane decreases for increasing shear sliding.

Walraven [80] performed shear tests on reinforced beams without

stirrups.
During the tests, cracks developed in the beam. The displacements in
these cracks were recorded and a typical result is shown in Fig. 6.7a.
The calculated response of one crack is presented in Fig. 6.7b. The re­
corded crack displacements are converted into strains using a measuring
length of 100 mm. First, the shear stress- shear sliding relation is de­
termined according to Walraven 's two-phase model, thus according to the
matrix presented in eq. (6.12). Second, the crack response according to
the eqs. (6.16a-b) is shown. It is found that the response according to
Rots overestimated the shear stiffness of the element. Finally, the re­
sponse according to the strain based formulas (eqs. (6.4a-b)) is shown,
which only slightly deviated from the result of Walraven's model. The
overestimation of the result according to the formulas of Rots is mainly
caused by the fact that the expressions neglect the effect of the off-
diagonal terms in the stress - strain relation upon the transferred
shear stress. Therefore, contrary to the statement of De Borst [17, page
115], the off-diagonal terms have to be taken into account even if small
crack strains are considered.

The eqs. (6.16a-b) cannot provide a proper description of the real crack
behaviour because they are not related to a possible increase in crack
width. For a strong increase in crack width without shear sliding, the
physical two-phase model predicts a sharp drop in the transferred shear
stress, whereas the eqs. (6.16a-b) only predict a less strong increase
of the transferred shear stress. These formulas fit, however, very well
with the solution methods, which are commonly used in numerical pro­
grams. Therefore, an attempt is made to derive an expression for a shear
retention factor based upon the physical two-phase model instead of upon
experimental results.
 - 146 -

shear slip Ö. [mm! shear stress i 0 [MPai shear stress l 0 [MRal

2.0 T
/ I
/ Rots

strain based eqs.

I
\two phase model

0 0.05 0)0 0.15 "0 0 025 0.050 0.075 0.025 0.050 0.075
crack width ö n [mm] shear slip 6( [mm] shear slip 6) [mml
a. Crack opening path 'b. Shear stress - shear slip c. Shear stress-shear slip
relation ace. to the model relation ace. to Rots

Fig. 6.7. The crack response for a measured crack opening path [3],

Re-arranging eq. (6.13b), it was found that the following expression can
be used for the case that the ratio of normal strain to shear strain
decreases with increasing deformations:

0 P e + 4>
(6.17)
'(2) (1+n) P enn + *

2500
with P
0 1
D " [0.76-0.16 e /Y(l-exp[-6Y/e ])]

shear retention factor

Fig. 6.8. The shear retention factor 0 according to eq. (6.17).

 - 147 -

An important conclusion is that 6 is independent of the concrete

quality, so 6 is a real damage parameter. Eq. (6.17) is shown in Pig.
6.8 for a fully cracked element (n = 0; il> = 1 ) . For D equal to 19 mm
max
and the ratio of the normal strain to the shear sliding equal to 3, eq.
(6.17) fits with the expressions derived by Rots.
The physical crack behaviour indicates that there is hardly any increase
in the magnitude of the transferred stresses for constant ratios of the
normal strain to the shear strain. If the increase in normal strain is
larger than the increase in shear sliding, the transferred normal and
shear stress must become smaller. To account for this, eq. (6.17) is
extended to the following expression:
{n P e n n + *} abs(Xn.2-Xn.1)
6 (6.18)
(2)={(l*n)Penn+W(xn_2-Xn_1)

with X = e / Y for step n-2 in the calculation.

n-2 nn

X = e /v for stepr n-1 in the calculation.

n-1 nn
shear s ip Ö, Imm] shear stress i n IMftO shear stress i a [Mftil
025

020

0.15
3 : 1=100 mm
1.0 1.0
0.10
L=50mrt.
/;\<L5a
005
two phase model L = 100 mm

0.05 0.10 0.15 0 0 025 0050 0.075 0.025 0.050 0.075

crack width 6 n Imml shear slip 6f Imml shear slip 6g [mm]

a. Crack opening path. b. Shear stress - shear slip c. Shear stress-shear slip
relation ace. to Pmodel relation ace. to P =005.

Fig. 6.9. The shear retention factor for mixed mode fracture problems.

For the crack opening path shown in Fig. 6.7, the shear stress - shear
sliding relationship according to the e q s . (6.17-6.18) is shown in Fig.
6.9b. Now, there is a reasonable agreement between the calculated result
and the result obtained with the two-phase model of Walraven. Variation
of the measuring length (element length) yielded only very small differ­
ences. In Fig. 6.9c, the response of the element is calculated with a
 - 148 -

constant 8-value of 0.05 for the cracked element. Although there is a

rather strong dependency upon the measuring length, the calculated re­
sult agrees reasonably well with the result according to the physical
model. The very low constant value of the shear retention factor aver­
ages the slightly increasing shear stress according to the physical mod­
el. Because of the fact, that the crack opening path shown in Fig. 6.7a
is rather typical for shear tests on beams, it is obvious that for this
case a constant value of 6 yields good results. A constant value of
6 has, however, no general applicability.

This was also recognized by Rots and De Borst [62] for mixed-mode frac­
ture problems. They found that a constant positive value for 8 yields an
overestimation of the shear strength of such an element. Therefore, they
proposed a shear softening behaviour as shown in Fig. 6.10. Although the
shear stiffnesses 01 and D 2 are largely based upon trial-and-error
methods, the over-all shear stiffness is to some extent similar to the
relation shown in Fig. 6.9b. However, they also proposed a linear rela­
tion between shear stress and shear strain for unloading and re-loading
of the element. According to the cyclic model presented in Chapter 5,
this relation cannot be true either for unloading or for re-loading the
element. Therefore, an improved expression for the shear stiffness for
the case of cyclic shear loads will be derived in the next Section.

shear stress T,

/ X?

/ ^ <unloodinq >v
A DJ* reloading \.

0 Yu
shear sliding Y

Fig. 6.10. Shear softening relation according to Rots and De Borst [62].

6.4. The stress-strain relation for the case of cyclic loading.

Unloading can be described by means of a linear relation between the

shear stress and the shear strain. However, the shear strain must be
reduced by the residual shear sliding y , which can be calculated
 - 149 -

according to eq. (5.Ac). In this expression the shear displacement can

be replaced by the shear sliding.
Re-loading the element, the crack response is determined by the cyclic
aggregate interlock model, as is developed in Chapter 5. In this Section
the simplest model, the reduced contact model, will be implemented. In
this model, the projected contact areas, which would develop in case of
a monotonie loading, are reduced by means of the reduction factors
X and X , see eqs. (5.4a-b). These reduction factors are determined
x y
empirically from the calculated results based upon the analytical con­
tact model and are therefore limited in their applications. With regard
to the average increments in the crack displacements, it is found that
the reduction factors can be applied when the crack width increment is
approximately equal to the increment of the shear displacements. Conver­
sion to strains yields the assumption:

de = dy (6.19)
nn

For the case of cyclic loading, the eqs. (6.15a~d) become:

S,,= o (d(X A )/de - v d(X A )/de ) (6.20a)

1
' puv xx nn y y nn'

S
' 3 = " E nn / Y S'i (6
'20b)

S, = - y/e S„ (6.20c)

S,,= o fd(X A )/dy + ji d(X A )/dy) (6.20d)

33 pu1, y y xx '

in which:

|AA_ = d^ A + A dA_ (6
de de de
nn nn nn
!}*A B ijX A ♦ x ij* (6.20f)
dy dy dy
However, X and X are not related to the normal strain. Because of eq.
x y
(6.19), it can be stated that: dx/de = dX/dy
nn

Now, the eqs. (6.20e-f) become:

 - 150 -

ML = ff* A ♦ X - ^ - (6.20e)
de dY de
nn nn
i]M=dXA+x5JA (620f)
dy dy dY

with A expressed in the eqs. (6.1) and (6.A).

dA/dy and dA/dc expressed in the eqs. (6.15e-f).
nn
\ expressed in the eqs. (5.4a-b).

Finally, the expressions for the reduction factors must be converted to

strains and differentiated with respect to Y- Using the eqs. (5.4a-b),
the following expressions are obtained:

dX (Y-Y )
-T-l = 1.6 7 % (6.20g)
dY (y -
m 0
L\,21 2 . 1
dY 2 A
i'-v;
7 m 2~VÏ
o
(Y - Ï ) *
(6.20h)

With these expressions, the incremental stress - strain relation for a

cracked element can be described. It must be noted, that for the case of
unloading and subsequent re-loading of an element, the off-diagonal
terms of the stiffness matrix are as important as the diagonal terms.
This is due to the fact that upon re-loading the direction of the prin­
cipal tensile stress will generally strongly deviate from the direction
normal to the crack plane. To reduce the calculation process, the effect
of the off-diagonal terms can be partially accounted for when I(I in eq.
(6.17) is reduced. Whereas the shear stress largely depends upon the
projected contact area in the y-direction, i|> is reduced according to eq.
(6.20h). This is, however, a very rough approximation. For the previous­
ly given example, the calculated response of the crack in the second
cycle is shown in Fig. 6.11 (in the experiment no cyclic loading was
applied). It is obvious, that this cyclic behaviour strongly deviates
from the response suggested by Rots and De Borst [62], see Fig. 6.10.
 - 151 -

, - sheer stress In IMPol

n_L'^

5 0025 0.050 0075

shear slip 6^ [mml

Fig. 6.11. Calculated response for the second cycle.

6.5. Implementation of the dowel action mechanism.

The implementation of the mechanism of dowel action is strongly related

to the description of the bar-elements in a numerical program. A physi­
cally sound description is provided by using three independent elements,
see Fig. 6.12:
- a steel bar element,
- the interface element or slip layer element,
- and the plain concrete element.

plain concrete element

t slip layer element

i-frfcm^ bar element

Fig. 6.12. Element types used for reinforced concrete.

Such an approach allows for slip between the steel bar and the concrete.
The major disadvantage of this method is the fine mesh, which is re­
quired to account for the effect of the splitting cracks close to the
bar.
In this section, the slip layer element and the bar element will be con-
 - 152 -

sidered as being one element. In the Chapters 4 and 5, it is shown that

the dowel action mechanism depends upon the cooperation of the steel bar
and the concrete under the bar. The empirical and theoretical expres­
sions derived for this mechanism are based upon this cooperation of the
bar element and the slip layer element. This element wil be denoted here
as the dowel element.
The dowel element will provide the normal restraint force for the
(cracked) plain concrete elements.
The stiffness matrix for the dowel element is presented schematically by
eq. (6.21):

Ao J
\ l J
l 3 Ae

Ao\ 0 0 Aet (6.21)

tt
Ao AY
nt

In this matrix D J 3 represents the dowel stiffness. For the case of a

monotonie dowel load,
Loac the dowel stiffness K can be used, which can be
' o '
derived from eq. (4.28):

3
K = 0.33 ?_, I /TM6 +X ) (6.22)
o du t no t,e

For the case of cyclic loading, the dowel stiffnesses Kj- K,^ as derived
in Chapter 5, can be used for D 3 3 .

The term DJJ represents the axial bond characteristics of the bar. Un­
fortunately, this stiffness is related to both the normal strain and the
shear sliding. Therefore, the bond characteristics obtained in a pull-
out experiment can not be used here. Detailed tests in this field are
necessary. The same holds true for the terms D 1 3 and D 3 1 . The first
represents the decrease in bond capacity with increasing shear sliding.
The second represents the decrease in dowel capacity with increasing
normal strain. Because of the fact, that the stiffness relation pre­
sented in eq. (6.21) is related to both the dowel mechanism and the (un­
known) bond mechanism, no further description of the stiffness relation
is possible here.
 - 153 -

6.6. Concluding remarks.

The implementation of the reduced contact model, which describes the

cyclic aggregate interlock mechanism, yielded complex expressions for
the stiffness relations of a partially cracked element. It appeared that
both the diagonal and off-diagonal terms of the matrix are non-zero.
For a monotonically increasing load on an initially uncracked element,
it is expected that the assumption of zero off-diagonal terms is rather
close to reality. The direction of the principal stresses for this type
of loading is only slightly deviating from the stress direction during
cracking.
However, for closing and re-opening cracks (cyclic loading), the effect
of the off-diagonal terms must be accounted for directly or indirectly
by means of the eqs. (6.17) and (6.20h).
The dowel action model is not valid for implementation in numerical pro­
grams. This mechanism is strongly related to the bond characteristics of
the bar. Unfortunately, a generally applicable bond mechanism is not yet
described in the literature.
 - 154 -

7. R E T R O S P E C T I V E V I E W A N D CONCLUSIONS

T h e safety of large-scale structures, such as offshore platforms, might

depend upon the stiffness of cracked reinforced elements subjected to
in-plane loads. Previous studies [34,40,43,76,81] showed that the trans­
fer of in-plane static and cyclic shear stresses across cracks in rein­
forced concrete largely depends upon the combined mechanism of aggregate
interlock, dowel action and components of the axial steel stress in the
reinforcing bars. In fact, the experimental studies concerning cyclic
shear loads w e r e restricted to a relatively small number of cycles with
a very large load amplitude.

force

\hioh-cvc1e fatigue

Mi .
lex-cycle tolique/
, ■

Fig. 7 . 1 . Scheme of the load cycles for an offshore structure.

An offshore structure endures millions of load cycles during its econom­

ical l i f e - t i m e . Generally, the amplitude of these cycles is far less
than the magnitude of the static load, see F i g . 7 . 1 . The subsequent few
cycles with a very large amplitude, representing a super-storm, are of
special interest for the designer. The (partially) cracked structure has
to withstand this severe loading conditions, even if there is a stiff­
ness d e g r a d a t i o n d u e to the millions of 'low-intensity'-cycles. There
are no numerical tools available to simulate all these cycles in order
to determine the response degradation. Furtheron, there was no experi­
mental knowledge w i t h respect to 'low-intensity high-cycle' fatigue.
Therefore, first an experimental program was devoted to this type of
test, yielding empirical expressions for the increase in crack displace­
ments due to cycling. It is shown that 'high-cycle' fatigue can b e
treated quasi-statically in order to obtain the crack displacements at
the onset of the design load ('high-intensity low-cycle' f a t i g u e ) .
 - 155 -

Second, the static models for aggregate interlock and dowel action are
adapted to the case of 'high-intensity' cyclic loading. Both mechanisms
are described as to fit to the crack displacements obtained with the
empirical relations for 'low-intensity' load cycles. An important con­
clusion is that for the aggregate interlock mechanism, the load history
is fully incorporated in the end crack displacements of the previous
cycle. Furtheron, it is shown, that the contributions to the shear
transfer of both dowel action and aggregate interlock remain nearly
constant during cycling. This observation was used to prove that the
cyclic models are also valid for the case of 'high-cycle' fatigue.
Although valuable experimental and theoretical information is obtained
with respect to 'low-intensity high-cycle' fatigue, further study is
necessary in this field. First, only one stress ratio (R = 0) was in­
vestigated in combination with a constant loading frequency. Both para­
meters might largely influence the stiffness degradation of the crack
due to cycling.
Second, only reinforcing bars perpendicularly crossing the crack plane
were used. In practice, orthogonal reinforcing webs cross cracks at
various angles to the crack plane. For this case, the contribution of
the axial steel stresses and the bond strength degradation also influ­
ence the crack response. Furtheron, the cyclic aggregate interlock model
is strongly simplified with respect to the physical reality. This model
is, however, still applicable to a wide range of tests. Contrary to
this, the cyclic dowel action model is to a large extent based upon em­
pirical relations, thus limiting its application. Therefore, further
theoretical work into this field is necessary.
Finally, according to the static dowel action mechanism there is no
longer a relationship between the axial steel stress and the crack width
when the plastic hinges in the bar fully have been developed. Because of
this lack of a relation between the crack width or normal strain and the
normal restraint stiffness, the implementation of the dowel action mech­
anism into numerical programs is not yet fully described in this report.
With respect to the mechanism of aggregate interlock, further experimen­
tal study is necessary in the following fields:
a. 'low-intensity high-cycle' experiments with a reversed shear load (R
< 0).
b. cyclic push-off tests on pre-cracked specimens with water or oil in
 - 156 -

the crack. Due to the opening and re-closing of the crack water (oil)
is pumped in and out the crack, thus transporting crushed matrix ma­
terial. Furthermore, the pressure of the fluid transfers stresses
normal to the crack plane.
The mechanism of dowel action can provide an important contribution to
the transfer of shear stress across cracks in reinforced concrete. In
most cases, however, the direction of the principal tensile stress after
cracking only slightly deviates from its direction during cracking. For
these cases, the dowel action mechanism is of minor importance due to
the relatively high axial steel stress.

With respect to the mechanism of dowel action, some questions remained

unanswered:
a. The dowel capacity decreases with increasing initial crack width.
Yet, it is not known to what extent this is caused by the crack width
itself or by the initial steel stress necessary to obtain this crack
width.
b. The dowel mechanism is strongly related to the bond mechanism. This
interaction is, however, poorly understood.
A experimental study on push-off elements in which the aggregate inter­
lock mechanism is prevented by means of smooth crack faces, can provide
the detailed information needed in this field. In such a test series,
the dowel capacity must be measured for several combinations of the ini­
tial crack width and axial steel stress. Furthermore, several load paths
must be investigated, such as:
- first a dowel force is applied, subsequently an axial steel force is
increased monotonically until failure occurs.
for an initial axial steel stress, the dowel force is increased. Then
the axial steel stress is removed. What will be the crack displace­
ments due to this unloading.
It must be noted that especially any improvement in modelling the bond
between the steel bar and the concrete will also improve the physical
understanding of the dowel action mechanism. It is the author's opinion
that bond tests are necessary to make the dowel model generally applica­
ble.
 - 157 -

8. SUMMARY

Offshore platforms, used for the exploitation of the oil and gas reser­
voirs in the Arctic and the deep sea, are designed to withstand severe
loading conditions, characterized by wave and wind attacks. Such struc­
tures are so configured as to transfer the applied cyclic loads to the
subsoil by means of in-plane shear and normal stresses. The walls of the
base of such a structure might be cracked due to unequal settlements and
thermal deformations. As a consequence, these cracked reinforced panels
will respond highly nonlinear to the applied stresses.
The transfer of in-plane stresses across cracks in reinforced concrete
is based upon the interaction of several mechanisms:
a. the axial stiffness of the reinforcing bars crossing the crack,
b. the lateral stiffness of the bars, called dowel action, and
c. the interlocking of the aggregate particles protruding from the crack
faces, denoted as aggregate interlock.

For the case of cyclic loads, usually a distinction is made between on

the one hand 'low-intensity high-cycle' loading, reflecting the load
history of millions of small wave attacks. On the other hand, high-in­
tensity low-cycle' loading is considered, which forms another severe
loading condition a structure has to withstand. From literature research
it has become clear that there was a lack of experimental knowledge on
the response of cracked concrete subjected to shear loading, especially
for the case of a large number of cycles with a low shear stress rela­
tive to the static shear strength.
Therefore, first an experimental study was carried out on push-off spec­
imens. For reinforced concrete specimens, a repeated shear load was ap­
plied ranging from 46 to 90 percent of the static shear strength. The
number of cycles ranged from 118 to 931731 cycles. The increase in crack
displacements due to cycling was recorded and expressed in empirical re­
lations. Apart from this test series, similar tests were performed on
plain concrete specimens in order to determine the contribution of the
aggregate interlock mechanism alone.
Second, the crack response under monotonie loading was discussed. Where­
as the aggregate interlock mechanism was satisfactorily described by
means of the two-phase model of Walraven, the mechanism of dowel action
 - 158 -

was not yet fully understood. Therefore, a physical description of the

static dowel action mechanism based upon cooperation of the steel bar
and the supporting concrete was given. It was found that the combined
mechanism of aggregate interlock and dowel action could be used to sim­
ulate static shear tests on reinforced concrete push-off specimens. Ini­
tially, the deformation of the bars determines the crack opening direc­
tion. After plastic hinges in the bar have fully been developed, the
crack opening path is determined by the aggregate interlock mechanism.
Next, it was shown that Walraven's extended version of his two-phase
model could be applied to the case of cracked plain concrete subjected
to cyclic shear loading. This numerical model was then simplified yield­
ing an analytical solution method, the analytical contact model. The re­
sults of this model were used to derive empirical expressions for reten­
tion factors, which could be applied to the contact areas according to
the static two-phase model. Several 'high-intensity low-cycle' experi­
ments were simulated with this reduced contact model.
As for the static dowel action mechanism, the response of a bar to
cyclic dowel forces was not yet described by a physical model. There­
fore, a rather simple model is proposed, which is based upon physical
material behaviour. Because of a lack of detailed experimental informa­
tion, this model is still to a large extent based upon empirical expres­
sions. It enables the determination of the effects of bar diameter,
steel and concrete strength and initial crack width. The results of test
specimens subjected to a small number of cycles with a large amplitude
were satisfactorily simulated with the dowel action model.
For practical use, the combined model of aggregate interlock and dowel
action under cyclic shear loading is much interest. Again, it was found
that the calculated crack response according to the combined model
agrees very well with the results of 'high-intensity low-cycle' experi­
ments. Purtheron, a few cycles of 'high-cycle' tests were satisfactorily
predicted. An important conclusion was that the contributions of both
transfer mechanisms remained almost constant during cycling. With this
observation, it was possible to give a reasonable prediction of the
crack opening path of 'low-intensity high-cycle' experiments.
Finally, the aggregate interlock model for monotonie and cyclic shear
loading was made valid for implementation in numerical programs. It was
shown that the commonly used shear retention factor, which has a con-
 - 159 -

stant value, has no general applicability and neglects the physical be­
haviour of the crack. However, for the case of a monotonie loading, a
constant retention factor gives reasonable predictions of the shear
stresses. Contrary to this, for the case of mixed-mode fracture problems
or cyclic loading, a constant shear retention factor will largely over­
estimate the shear loading capacity of the crack for a given combination
of the crack displacements. Furthermore, for this case the interaction
between the tension-softening and the shear-softening behaviour must be
accounted for.
 - 160 -

SAMENVATTING

De booreilanden, welke worden gebruikt voor de exploitatie van de olie-

en gas reserves in de Poolzee en de diepzee, zijn ontworpen om de ex­
treme belastingen te weerstaan, welke worden gekenmerkt door golf en
wind belastingen. Dergelijke constructies zijn zo samengesteld dat de
opgelegde (wisselende) belastingen naar de ondergrond worden overgedra­
gen door middel van spanningen in het vlak. De wanden van de funderings­
plaat van een booreiland zijn mogelijk gescheurd ten gevolge van onge­
lijkmatige zettingen en temperatuurspanningen. Als een gevolg hiervan
reageren deze gescheurde schijven sterk niet-lineair op de opgelegde
spanningen.

De overdracht van de spanningen in het vlak over de scheuren in gewapen­

de betonnen schijven berust op de interactie van verschillende mechanis­
men, te weten:
a. de axiale stijfheid van de staven, welke het scheurvlak doorsnijden,
b. de deuvelweerstand van de wapening en
c. de haakweerstand van de toeslagkorrels, die uit het scheurvlak ste­
ken.

Voor het geval van wisselbelastingen wordt er meestal onderscheid ge­

maakt tussen aan de ene kant belasting met een lage intensiteit en een
groot aantal wisselingen, welke staan voor de belastinggeschiedenis van
miljoenen golfaanvallen. Aan de andere kant worden belastingen beschouwd
met een grote amplitude gedurende een beperkt aantal wisselingen. Deze
wisselingen vormen feitelijk de ontwerpbelasting. Uit een literatuur­
onderzoek is gebleken dat experimentele gegevens ten aanzien van de re­
actie van gescheurd gewapend beton op wisselende schuifkrachten ontbra­
ken, met name ten aanzien van een groot aantal wisselingen met een lage
schuif spanning in verhouding tot de statische schuif sterkte.
Teneinde deze informatie te verkrijgen is een experimenteel programma
uitgevoerd op afschuif-elementen. Op gewapende proefstukken is een her­
haalde belasting variërend van 46 tot 90 procent van de statische
schuifsterkte aangebracht. Het aantal wisselingen varieerde daarbij van
118 tot 931731 wisselingen. De toename van de scheurverplaatsingen ten
gevolge van de lastwisselingen is gemeten en uitgedrukt in empirische
relaties. Daarnaast zijn soortelijke proeven uitgevoerd op proefstukken
 - 161 -

van ongewapend beton teneinde de bijdrage van het mechanisme van de

haakweerstand van de korrels te bepalen.
Vervolgens is de reactie van de scheur op monotoon stijgende belasting
bestudeerd. Daar waar het mechanisme van de haakweerstand van de korrels
goed werd beschreven door het twee-fasen model van Walraven, was het me­
chanisme van de deuvelwerking nog niet volledig verklaard. Daarom is een
beschrijving van het fysische gedrag van de staaf onder een deuvelkracht
gegeven, waarbij is uitgegaan van de samenwerking van de stalen staaf en
de beton direct onder de staaf. Het bleek dat het gecombineerde mecha­
nisme van deuvelwerking en haakweerstand kon worden toegepast op experi­
menten met gewapend betonnen proefstukken. Aanvankelijk wordt het
scheuropeningspad bepaald door de vervorming van de staven. Echter na
het ontwikkelen van plastische scharnieren in deze staven, bepaalt de
haakweerstand de scheuropeningsrichting.

Daarna is aangetoond dat een uitgebreid twee-fasen model zoals dat is

voorgesteld door Walraven, kan worden toegepast voor het geval van wis­
selende schuifspanningen op ongewapend beton. Dit model is vervolgens
vereenvoudigd tot een model met een analytische oplossing voor de groot­
te van het kontaktvlak tussen de korrels en de matrix. De resultaten van
de berekeningen met dit model zijn op hun beurt weer gebruikt om uit­
drukkingen af te leiden voor reductie-factoren. De reductie-factoren
kunnen worden toegepast op de kontaktvlakken volgens het statische twee­
fasen model. Dit gereduceerde kontakt-model is gebruikt om verschillende
proeven met een grote amplitude en een beperkt aantal wisselingen door
te rekenen.
Net als voor het statische deuvelmodel was de reactie van een staaf op
wisselende deuvelkrachten nog niet volledig fysisch verklaard. Daarom is
een relatief eenvoudig model voorgesteld, welke is gebaseerd op fysisch
materiaalgedrag. Een ontbreken van gedetailleerde experimentele informa­
tie veroorzaakte echter dat ook dit model tot op zekere hoogte is geba­
seerd op empirische uitdrukkingen. Het model maakt het mogelijk om de
effecten van staafdiameter, staal- en betonkwaliteit en initiële scheur-
wijdte in rekening te brengen. De resultaten van meerdere deuvelproeven
met een grote amplitude gedurende een gering aantal wisselingen zijn re­
delijk gesimuleerd met dit model.
Voor toepassing in de praktijk is met name het gekombineerde model van
de haakweerstand en deuvelwerking van belang. Ook nu werd gevonden dat
 - 162 -

de berekende reactie volgens het gekombineerde model een goede voorspel­

ling geeft voor het experimenteel verkregen scheurgedrag in proeven met
een gering aantal wisselingen. Verder zijn enkele wisselingen van proe­
ven met een zeer groot aantal wisselingen nagerekend met bevredigend
resultaat. Een belangrijke conclusie was dat de bijdragen van de afzon­
derlijke mechanismen nagenoeg konstant blijven gedurende de lastwisse-
lingen. Op grond van deze waarneming was het mogelijk een betrouwbare
voorspelling te doen van het te volgen scheuropeningspad voor proeven
met een kleine amplitude en een groot aantal wisselingen.
Tenslotte is het model van de haakweerstand voor monotone en wisselende
schuifbelasting geschikt gemaakt voor implementatie in numerieke pro­
gramma's. Aangetoond is dat de vaak gebruikte reductiefactor voor de
schuifweerstand, welke een konstante waarde heeft, niet algemeen toe­
pasbaar is en geen relatie heeft met het werkelijke scheurgedrag. Toch
zal een konstante waarde voor de reductiefactor in de regel geen over­
schatting van de schuifspanningen geven voor het geval van een monotoon
stijgende belasting. Daarentegen wordt de schuifsterkte wel sterk over­
schat indien een konstante waarde voor de reductiefactor wordt gehan­
teerd in het geval van .mixed-mode scheurproblemen en in het geval van
wissel belastingen voor gegeven combinaties van de scheurverplaatsingen.
Tevens moet voor dit geval worden meegenomen de interactie tussen de
afnemende stijfheden onder enerzijds de normaal spanning (tension-
softening) en anderzijds de schuifspanning (shear-softening).
 - 163 -

9. NOTATION

a,b numerical constants

a projected contact area in x-direction [mm2]
a projected contact area in y-direction [mm2]
e eccentricity [mm]
f . cylindrical concrete crushing strength [MPa]
f cube concrete crushing strength [MPa]
ccm ° °
f steel yield strength [MPa]
h element size [mm]
m numerical constant
n number of cycles
nf number of cycles till failure
p, volume of the particles/total volume
r radius [mm]
x length [mm]
x direction in the two-dimensional space
y direction in the two-dimensional space
z distance to neutral axis [mm]
A cross-sectional area [mm2]
A total projected contact area in x-direction [mm2]
A total projected contact area in y-direction [mm2]
C numerical constant
D maximum rparticle diameter l [mm]
max
D specific particle diameter [mm]
E modulus of elasticity of concrete [MPa]
E modulus of elasticity of steel [MPa]
F, dowel force [kN]
d
F. ultimate dowel force [kN]
du
G shear modulus [MPa]
I moment of inertia [mm'']
K numerical constant
K. ,, dowel stiffness [N/mm]
»• • i

Kf foundation modulus of concrete [MPa/mm]

L length [mm]
M ultimate bending moment [Nraro]
P numerical constant
 - 164 -

R cyclic stress ratio = T . /T

min max
S.. ,, terms of the stiffness matrix
> • >3J
a numerical constant
a angle of inclination
at 2 normal retention factors
t • t
8 angle
6[ 2 shear retention factors
r • t
Y shear sliding
Y maximum shear sliding
m
Y residual shear sliding
Y shear sliding of the crack zone
cr °
Y shear sliding of the uncracked zone
co °
Y retention factor of the stresses due to aggregate interlock
Y. retention factor of the stresses due to dowel action
a
Y. retention factor of the dowel stiffness
6. shear displacement [mm]
& residual shear displacement
r [mm]
o
6 maximum shear displacement [mm]
r
m
Jt shear displacement due to elastic deformation [mm]
t,e
6 crack width [mm]
n
& initial crack width [mm]
n,o '
c numerical constant
e normal strain
nn
e normal strain of the crack zone
nn,cr
£ normal strain of the uncracked zone
nn,co
ri normal retention factor
X retention factors of the contact areas
u coefficient of friction
u normal retention factor
u Poisson's ratio
p reinforcement ratio
o normal stress [MPa]
o normal stress due to aggregate interlock [MPa]
a
a. normal stress due to dowel action [MPa]
d
o steel stress [MPa]
s
o strength of material [MPa]
pu
o normal stress acting upon the crack plane [MPa]
 - 165 -

T shear stress due to aggregate interlock [MPa]

T. bond stress [MPa]
b
T, shear stress due to dowel action [Mpa]
T minimum applied shear stress [MPa]
T maximum applied shear stress [MPa]
T shear strength [MPa]
i|> angle of inclination
i|) retention factor of the cracked area
0 angle of inclination
<t> bar diameter [mm]
i|) angle of friction
 - lit.1 -

10. LITERATURE

[I] Bazant, Z.P., Cambarova, P.G.,

Rough cracks in reinforced concrete,
ASCE, Journal of Structural Division, Vol. 106, No. 4,
April 1980, pp. 819-842.
[2] Bazant, Z.P., Cambarova, P.C.,
Microplane model for concrete subject to tension and shear,
Int. Conference on Concrete under multiaxial conditions, Toulouse,
22-24 May 1984, pp. 240-250.
[3] Bazant, Z.P., Cambarova, P.G.,
Crack shear in concrete, crack band microplane model,
ASCE, Journal of Structural Engineering, Vol. 110, No.9,
Sept. 1984, pp. 2015-2035.
[4] Bennett, E.W., Banerjee, S.,
Strength of beam-column connections with dowel reinforcement,
The Structural Engineer, Vol. 51, No. 4, April 1976, pp. 133-139.
[5] Birkeland, P.W., Birkeland, U.W.,
Connections in precast concrete constructions,
ACI-journal, Vol. 63, No.3, March 1966, pp. 345-368.
[6] De Borst, R.,
Non-linear analysis of frictional materials,
Dissertation Delft University of Technology, 1986, 140 pp.
[7] Broms, B.B.,
Lateral resistance of piles in cohesive soils,
ASCE, Journal of soil mechanics, Vol. 90, No.2,
March 1964, pp. 27-59.
[8] Chung, U.W.,
Shear strength of concrete joints under dynamic loads,
Concrete, Vol. 12, March 1978, pp. 27-29.
[9] Colley, B.E., Uumprey, U.A.,
Aggregate interlock at joints in concrete pavements,
Highway Research Record, No. 189, 1967, pp. 1-18.
[10] Collins, M.P.,
Memorandum to the participants in the University of
Toronto's International prediction competition,
October 1984.
[II] Collins, H.P.,
Shear design of complex high strength concrete structures,
Proceeding of the conference on high strength concrete,
Stavanger 1987, pp. 345-365.
[12] Daschner, P., Kupfer, U.,
Versuche zur Schubkraftuebertragung in Rissen von Normal-
und Leichtbeton,
Bauingenieur 57, (1982), pp. 57-60.
[13] Daschner, F., Nissen, I.,
Schubkraftuebertragung in Rissen von Normal- und Leichtbeton,
Betonwerk und Fertigteiltechnik Vol.53, Heft 1, 1987, pp. 45-51.
[14] Divakar, M.P., Fafitis, A., Shah, S.P.,
A constitutive model for shear transfer in cracked concrete,
submitted for publication, ASCE, Structural division 1987.
[15] Dulacska, U.,
Dowel action of reinforcement crossing cracks in concrete,
ACI-Journal, Vol. 69, Dec. 1972, pp. 754-757.
 - lit.2 -

[16] Eleiott, A.F.,

An experimental investigation of shear transfer across cracks in
reinforced concrete,
M.S. Thesis, Cornell University, Ithaca, June 1974.
[17] Fardis, M.N., Buyukozturk, 0.,
Shear stiffness of concrete by finite elements,
ASCE, Journal of the Structural Division, Vol. 106, No. 6,
June 1980, pp. 1311-1327.
[18] Fenwick, R.C., Paulay, T.,
Mechanisms of shear resistance of concrete beams,
ASCE, Structural Division, Vol. 94, No. 10, Oct. 1968,
pp. 2325-2350.
[19] Finney, E.A.,
Structural design considerations for pavement joints,
Subcommittee III, ACI-committee 325, ACI-journal,
Proceeding Vol. 53, No 1, 1956, pp. 17-30.
[20] Frenay, J. W.,
Shear transfer across a single crack in reinforced concrete
under sustained loading, Part I. Experiments
Report 5-85-5, Stevin Laboratory, Delft University of
Technology, 1985, pp. 114.
[21] Frenay, J.W., Liqui Lung, G., Pruijssers, A.F.,
Shear transfer across a single crack in reinforced concrete
Additional detailed tests,
Report 5-86-5, Stevin Laboratory, Delft University of
Technology, 1986, pp. 106.
[22] Friberg, B.F.,
Design of dowels in transverse joints of concrete pavements,
Transactions, ASCE, Vol. 105, 1940, pp. 1078-1080.
[23] Gambarova, P.G.,
On aggregate interlock mechanism in reinforced concrete plates
with extensive cracking, Transactions of IABSE colloquium Delft
1981 on Advanced Mechanics of Concrete Delft, June 1981, pp. 99-
120.
[24] Gambarova, P.G.,
Crack shear in concrete: Rough crack model and micro-plane model,
Presented at the 1983 meeting of the Italian Society for normal
and prestressed concrete, Bari, May 26-29 1983.
[25] Cerwick, B.C.,
High strength concrete, key to the Arctic and deep sea,
Proceeding of the conference on high strength concrete,
Stavanger 1987, pp. 393-404.
[26] Gerwick, B.C.,
High-Amplitude Low-cycle fatigue in concrete sea structures,
PCI-journal, Vol. 26, Sept-Oct. 1981, pp. 82-96.
[27] Graf, O., Brenner, E.,
Versuch zur Ermittlung der Wiederstandfahigkeit
von Beton gegen oftmals wiederholte Belastung,
Deutscher Ausschuss für Stahlbeton, Heft 76 und 83,
Berlin, 1934 and 1936.
[28] Uansen, R.J., Nawy, E.G., Shah, J.M.,
Response of concrete shear keys to dynamic loading,
ASCE-journal of Structural Div., Vol. 32, No. 11, May 1961,
pp. 1475-1490.
 - lit.3 -

[29] Hofbeck, J.A., Ibrahim, I.O., Mattock, A.U.,

Shear transfer in reinforced concrete,
ACI-journal, Vol. 66, Febr. 1969, pp. 119-128.
[30] Uolmen, J.O.,
Fatigue of concrete by constant and variable amplitude
loading,
Bulletin No. 79-1, Division of Concrete Structures,
NTH-Trontheim, 1979, pp. 218.
[31] Uoude, J., Mirza, M.S.,
A finite element analysis of shear strength of reinforced
concrete beams,
ACI-Special Publication 42, 'Shear in reinforced concrete',
pp. 103-128.
[32] Jimenez, B., Gergely, P., White, 8.N.,
Shear transfer across cracks in reinforced concrete,
Report No. 78-4, dept. of Structural Eng., Cornell University,
Ithaca, Aug. 1978, pp. 357.
[33] Jimenez, R., Perdikaris, P., Cergely, P.,
Interface shear transfer and dowel action in cracked reinforced
concrete subject to cyclic shear,
Proceeding of the ASCE conference, Madison Aug. 1976, pp. 457-475.
[34] Jimenez, R., White, R.N., Gergely, P.,
Cyclic shear and dowel action models in reinforced concrete,
ASCE, Journal of Structural Engineering, Vol. 108,
No. 5, May 1982, pp. 1106-1123.
[35] Klein, D., Kristjansson, R., Link, J., Hehlhorn, G., Schaeffer, U.,
Zur Berechnung von dunnen Stahlbetonplatten bei Berucksichtigung
eines wirklichkeitsnahen Werkstoffverhaltens,
Forschungbericht No. 25, Inst. für Massivbau,
Technische Hochschule Darmstadt, 1975, pp. 1-28.
[36] Laible, J.P.,
An experimental investigation of interface shear transfer and
applications in the dynamic analysis of nuclear containment
vessels,
Dissertation, Cornell University, Ithaca, 1973, pp. 343.
[37] Laible, J.P., White, R.N., Gergely, P.,
Experimental investigation of seismic shear transfer across
cracks in concrete nuclear containment vessels,
ACI-Special Publication 53, Reinforced concrete structures
in seismic zones, 1977, pp. 203-226.
[38] Van Leeuwen, J., Siernes, A.J.M.,
Miner's rule with respect to plain concrete,
Heron, Vol. 24, No. 1, Delft, 1979.
[39] Mast, R.F.,
Auxiliary reinforcement in concrete connections,
ASCE, Journal of Structural Division, Vol. 94,
No. 6, June 1968, pp. 1485-1499.
[40] Mattock, A.U.,
Shear transfer in concrete having reinforcement at an angle
to the shear plane,
ACI-Special Publication 42, Shear in reinforced concrete,
Vol. 1, pp. 17-42, 1974.
 - lit.4 -

[41] Mattock, A.H.,

Effect of aggregate type on single direction shear transfer
strength in monolithic concrete,
Report SM 74-2, Dept. of Civil Eng., University of Washington,
Seattle, Washington, Aug. 1974.
[42] Mattock, A.H.,
Effect of moment torsion across the shear plane on single
direction shear transfer strength in monolithic concrete,
Report SM 74-3, Dept. of Civil Eng., University of Washington,
Seattle, Washington, Aug. 1974.
[43] Mattock, A.H.,
Cyclic shear transfer and type of interface,
ASCE, Journal of Structural Division, Vol. 107,
No. 10, Oct. 1981, pp. 1945-1964.
[44] Van Mier, J.G.M.,
Strain-softening of concrete under multi-axial loading
conditions,
Dissertation, University of Technology Eindhoven,
1984, pp. 349.
[45] Millard, S.C., Johnson, R.P.,
Shear transfer across cracks in reinforced concrete due to
aggregate interlock and dowel action,
Mag. of concrete research, Vol. 36, No. 126, March 1984, pp. 9-21.
[46] Millard, S.G., Johnson, R.P.,
Shear transfer in cracked reinforced concrete,
Mag. of concrete research, Vol. 37, No. 130, March 1985, pp. 3-15.
[47] Mills, G.M.,
A partial kinking yield criterion for reinforced concrete slabs,
Mag. of Concrete Research, Vol. 27, No. 90, March 1975, pp. 13-22.
[48] Moe, J.,
Discussion of 'Shear and diagonal tension',
by AC1-ASCE committee 326, Proceedings AC1, Vol. 59, No. 9, Sept.
1962, pp. 1334-1339.
[49] Nissen,I.,
Rissverzahnung des Betons, gegenseitige Rissuferverschiebungen und
übertragene Krafte,
Dissertation, Technische Universitat München, 1987, 214 pp.
[50] Paulay, T., Loeber, P.J.,
Shear transfer by aggregate interlock,
ACI-Special Publication 42, 'Shear in reinforced concrete',
pp. 1-15.
[51] Paulay, T., Park, R., Philips, M.H,
Horizontal construction joints in cast in place reinforced
concrete,
ACI-Special Publication SP-42, Shear in reinforced concrete, Vol.
II, pp. 599-616, 1974.
[52] Perdikaris, P.C., Hiltoy, S., White, R.N.,
Extensional stiffness of precracked reinforced concrete panels,
ASCE, Journal of Structural Engineering, Vol. Ill, No. 3,
March 1985, pp.487-504.
[53] Perdikaris, P.C., White, R.N., Gergely, P.,
Strength and stiffness of biaxially tensioned reinforced
concrete subjected to reversed shear loads,
Paper presented at the ASCE mini-conference on 'Civil
Engineering and Nuclear Power", Boston, April 1979.
 - lit.5 -

[54] Perdikaris, P.C., White, R.N.,

Shear modulus of precracked reinforced concrete panels,
ASCE, Journal of Structural Engineering, Vol. Ill,
No. 2. Febr. 1985, pp. 270-289.
[55] Pruijssers, A.P.,
Description of the stiffness relation for mixed-mode fracture
problems using the rough-crack model of Walraven,
Stevin Report 5-85-2, Delft University of Technology, 1985, pp.36.
[56] Pruijssers, A.P., Liqui Lung, G.,
Shear transfer across a crack in concrete subjected to repeated
loading, Experimental results, Part I,
Report 5-85-12, Stevin Laboratory, Delft University of Technology,
1985, pp. 178.
[57] Rasmussen, B.H.,
Strength of transversely loaded bolts and dowels cast into
concrete, Laboratoriet for Bugningastatik, Denmark Technical
University, Meddelelse, Vol. 34, No. 2, 1962, (in Danisch).
[58] Reinhardt, H.W., Walraven, J.C.,
Cracks in concrete subject to shear,
ASCE, Journal of Structural Engineering, Vol. 108,
No. 1, Jan. 1982, pp. 207-224.
[59] Roland, B., Skare, E., Olsen, T.O.,
Ship impact on concrete shafts,
Nordisk Betong 2-4, 1982, pp. 111-114.
[60] Rots, J.G., Kusters, CM.A., Nauta, P.,
Variabele reductiefactor voor de schuifweerstand van gescheurd
beton,
TNO-IBBC report BI-:84-3, 1984, pp. 44.
[61] Rots, J.C.,
Bond-slip simulations using smeared cracks and/or
interface elements,
Report Delft University of Technology, 1985, pp. 56.
[62] Rots, J.G., De Borst, R.,
Analysis of mixed-mode fracture in concrete,
Paper submitted to ASCE, 1987, pp. 25.
[63] Schaefer, H.,
Zur Berechnung von Stahlbetonplatten,
Dissertation, University of Technology Darmstadt, 1976.
[64] Stanton, J.P.,
An investigation of dowel action of the reinforcement of nuclear
containment vessels and their non-linear dynamic response to
earthquake loads,
M.S. Thesis, Cornell University, Jan. 1977.
[65J The structural design of concrete pavements,
Pt. 4, Public Roads, Sept. 1936.
[66] Taylor, H.P.J.,
Fundamental behaviour in bending and shear of reinforced concrete,
Thesis, London, 1971.
[67] Taylor, R.,
A note on the mechanism of diagonal cracking in reinforced
concrete beams without web reinforcement,
Mag. of concrete research, Vol. Ill, No. 31, March 1959, pp. 151-
158.
 - lit.6 -

[68] Teller, L.W., Sutherland, E.J.,

A study of structural action of several types of transverse and
longitudinal joint design,
Public Roads, Vol. 17, No. 7, Sept. 1936.
[69] Termonia, Y., Meakin, P.,
Formation of fractal cracks in a kinetic fracture model,
Nature, Vol. 20, April 1986, pp. 429-431.
[70] Timoshenko, S., Lessels, J.M.,
Applied elasticity,
Westinghouse Technical Night School Press, East Pittsburg, 1925.
[71] TML technical bulletin No. 1007,
Tokyo Sokki Instruments, The Hague, 1985.
[72] Untrauer, R.E., Henry, R.L.,
Influence of normal pressure on bond strength,
ACI-Journal, Vol. 62, No. 5, May 1965, pp. 557-586.
[73] Utescher, C , Herrmann, M.,
Versuche zur Ermittlung der Tragfahigkeit in Beton
eingespannter Rundstahldollen aus nichtrostendem austenitischem
Stahl,
Deutscher Ausschuss fiir Stahlbeton, Heft 346, Berlin 1983,
pp. 49-104.
[74] Vecchio, F.J., Collins, M.P.,
The modified compression-field theory for reinforced concrete
elements subjected to shear,
ACI-Journal, Vol. 83, March-April 1986, pp. 219-231.
[75] Vintzeleou, E.N.,
Mechanisms of load transfer along reinforced concrete interfaces
under monotonie and cyclic actions,
Ph.D. Thesis (in Greek), Department of Civil Eng., National
Technical University of Athens, December 1984, pp. 549.
[76] Vintzeleou, E., Tassios, T.P.,
Mechanisms of load transfer along interfaces in reinforced
concrete, prediction of shear force versus shear displacement
curves,
Studi e ricerche, Vol. 7, 1985, pp. 121-159.
[77] Vintzeleou, E., Tassios, T.P.,
Mathematical models for dowel action under monotonie and cyclic
conditions,
Mag. of concrete research, Vol. 38, No. 134,
March 1986, pp. 13-22.
[78] Vintzeleou, E., Tassios, T.P.,
Behaviour of dowels under cyclic deformations,
ACI, Structural journal, Vol. 84, Jan-Febr. 1987, pp. 18-30.
[79] Vos. E.,
Influence of loading rate and radial pressure on bond in
reinforced concrete. A numerical and experimental approach,
Dissertation Delft University of Technology, Oct. 1983, pp. 235.
[60] Walraven, J.C.,
Influence of depth on the shear strength of leightweigth concrete
beams without shear reinforcement,
Stevin report 5-78-4, Delft University of Technology, 1978.
[81] Walraven, J.C.,
Aggregate interlock, a theoretical and experimental analysis,
dissertation, Delft University of Technology, Oct. 1980, pp. 197.
 - lit.7 -

[82] Walraven, J.C.,

The behaviour of cracks in plain and reinforced concrete subjected
to shear,
Final report of the IABSE colloquium Delft, 1981, pp. 227-245.
[83] Walraven, J.C.,
Kornverzahnung bei zyklischer Belastung,
Mitteilungen aus dem Inst. für Massivbau der Techn.
Hochschule Darmstadt, Heft 36, 1986, pp. 45-58.
[84] Walraven, J.C., Frenay, J-, Pruijssers, A.F.,
Influence of concrete strength and load history on the shear
friction capacity of concrete members,
PCI-Journal, Vol. 32, No. 1, Jan/Febr. 1987, pp. 66-85.
[85] Walraven, J.C., Vos, E., Reinhardt, U.W.,
Experiments on shear transfer in cracks in concrete. Part I,
Description of results,
Report No. 5-79-3, Jan. 1979, Stevin Laboratory, Delft University
of Technology.
[86] White, R.N.,
Interface shear transfer and dowel action in cracked reinforced
concrete,.
Paper presented at the cooperative research group meeting on shear
in reinforced concrete, Delft, June 1981.
[87] White, R.N., Gergely, P.,
Design considerations for seismic tangential shear in reinforced
concrete containment structures,
Paper presented at the 4th International Conference on structural
mechanics in reactor technology, Aug. 1977.
[88] White, K.N., Uolley, H.J.,
Experimental study of membrane shear transfer,
ASCE, Structural Division, Vol. 98, No 8, aug. 1972, pp. 1835-1853..
 - App.I.1 -

APPENDIX I. Mix proportions

Mix code B1632550 strength f' = 51 N/mm*

— — cc

(mix A)
Components [kg/m3] Sieve analysis of aggregate
Sieve opening
(mm] [kg]

sand 877.2 8 16 623.7

gravel 1065.0 4 8 441.3
cement-B 325.0 2 4 312.1
1 2 220.9
water 162.5 0.5 - 1 156.2
0.25 - 0.5 110.3
2429.7 0.10 - 0.25 77.7

1942.2

Mix code B1642037.5 strength f' = 70 N/mm2

(mix B)

sand 857.3 8 16 596.5

gravel 1018.5 4 8 421.9
cement-B 420.0 2 4 298.3
water 147.0 1 2 212.0
superpi.2^% 10.5 0.5 - 1.0 148.6
0.25 - 0.50 105.0
2453.3 0.10 - 0.25 93.5

1875.8

Sieve analysis of aggregate

[cum.Z]

mix A mix B Fuller sieve open ing [mm]

100.0 100.0 100.0 8 - 16

67.9 68.2 70.7 4 - 8
45.2 45.7 50.0 2 - 4
29.1 29.8 35.4 1 - 2
17.7 18.5 25.0 0.5 - 1
9.7 10.6 17.7 0.25 - 0.5
4.0 5.0 12.5 0.1 - 0.25
 - App.II.1 -

APPENDIX II. Results of dowel action tests compared with the theoretical
results.

Specimen No. bar diameter eccentricity f f F. F. ,

ccm sy d,exp d,cal
[mm] [mm] [MPa] [MPa] [kN] [kN]
Experiments of Bennett and Banerjee [4]

1 6 0 47 410 11.0 6.7

2 6 0 47 410 8.6 6.7
3 6 0 47 410 6.7 6.7
4 6 0 47 410 7.0 6.7
5 13 0 36 410 23.3 27.7
6 13 0 36 410 24.0 27.7
7 13 0 36 410 27.6 27.7
8 16 0 39 410 39.5 43.7
9 19 0 36 410 43.7 59.2

Experiments of Paulay, Park and Philips [51]

1 6.3 0 33 317 6.0 5.4

2 9.6 0 33 317 11.7 12.6
3 12.7 0 33 317 19.2 22.1

Experiments of Rasmussen [57]

*) See eq. (4.19)

Dl 15.8 1.4*) 13 247 16.8 18.1

D2 25.1 2.2 12 225 38.5 41.4
D3 15.8 1.9 24 247 24.0 23.2
D4 25.1 3.7 32 225 62.5 61.0
D5 16.0 1.3 20 439 35.5 30.9
D6 25.9 2.0 16 408 70.5 69.9
D7 15.8 2.4 37 247 29.5 26.9
D8 25.1 3.8 35 225 69.5 62.8
D9 15.8 2.8 52 247 31.8 30.0
D10 25.1 4.7 52 225 79.2 70.2

Experiments of Vintzeleou [75]

DB-40,8/M 8 2 51 420 16.8 9.7

DB-40.14/M 14 2 28 420 21.0 25.5
DB-40,14/M 14 2 38 420 31.5 29.3
DB-40,14/M 14 2 55 420 32.5 34.5
DB-40,14/M 14 2 55 420 35.0 34.5
DB-40,14/M 14 2 55 420 37.5 34.5
DB-150,14/M 14 2 52 420 35.0 33.6
DB-40.18/M 18 2 49 420 42.8 55.9
 - App.II.2 -

Specimen No. bar diameter eccentricity f £ F, F,

ccm sy d,exp d,cal
[mm] [mm] [MPa] [MPa] [kN] [kN]
Experiments of Utscher and Herrmann [73]
*) According to Utscher the steel yield strength was approximately 84%
of the reported yield strength [73, page 60].

109 25 5 26 270*) 55.4 58.7

110 25 5 24 270 51.7 56.8
111 25 5 26 270 48.8 58.7
112 25 5 24 270 48.8 56.8
141 14 5 33 293 15.0 18.3
142 14 5 33 293 15.6 18.3
143 14 10 33 293 11.9 13.3
145 14 10 33 293 12.2 13.3
147 14 20 33 293 8.4 8.2
201 20 5 33 280 36.5 40.2
202 20 5 33 280 38.8 40.2
203 20 10 33 280 31.8 31.7
204 20 10 33 280 29.0 31.7
205 20 20 33 280 24.9 21.1
206 20 20 33 280 23.8 21.1
251 25 5 33 270 57.6 64.7
253 25 5 33 270 60.1 64.7
254 25 10 33 270 53.6 53.0
257 25 10 33 270 51.6 53.0
258 25 20 33 270 37.7 37.1
259 25 20 33 270 37.7 37.1
1143A 14 50 31 245 3.0 3.0
2141A 14 20 31 245 8.2 6.9
2141B 14 20 31 245 8.3 6.9
2143A 14 50 31 245 3.2 3.0
2144A 14 20 31 245 8.3 6.9
3141A 14 5 32 245 14.6 16.1
3141B 14 5 32 245 13.2 16.1
3143A 14 5 32 245 15.9 16.1
3143B 14 5 32 245 17.2 16.1
3144B 14 5 32 245 13.3 16.1
3202A 20 50 32 275 11.3 9.5
3203A 20 20 32 275 24.1 20.6
3204A 20 20 32 275 24.1 20.6
3253A 25 50 32 280 21.5 18.6
4201A 20 20 32 275 23.8 20.6
4202A 20 20 32 275 23.8 20.6
4203B 20 20 32 275 22.9 20.6
4251A 25 20 32 280 40.0 38.0
4251B 25 20 32 280 36.2 38.0
4252A 25 50 32 280 21.2 18.6
4253A 25 20 33 280 39.4 38.3
4253B 25 20 33 280 36.9 38.3
4255A 25 20 33 280 41.9 38.3
4255B 25 20 33 280 39.2 38.3
 - App.II.3 -

Specimen bar diam. eccent. init. steel f £ F, F. .

ccm sy d,exp d,cal
No. [mm] [mm] stress [MPa] [MPa] [MPa] [kN] [kM]
Experiments of Millard and Johnson [46]

21L 12 0 0 38 485 28.5 26.3

22L 12 0 0 39 485 25.6 26.6
23L 12 0 0 54 485 30.5 31.5
24L 16 0 0 28 485 40.8 40.0
25L 8 0 0 32 435 5.5 10.2
26L 12 0.10 175 37 485 20.8 24.2
27L 12 0.31 344 40 485 19.1 18.6
 - App.III.1 -

APPENDIX III. Contact models.

Numerical contact model.

'BEGIN'
'REAL' FC,R,RMAX, DMAX,Y,WO,W,DEL,SIGI,TAUI,SIGE,C,MU,SPU,PK
XCT,XSNO,XSNB,YSNO, YSNB,G,V,DO,PCDO,AAX,AAY,KUK,
N,POX,ROX,DOX,DELS,DW,DDEL,TAUMAX,TOX, WROX,QROX,PUNK,
DELE,WMAX,WE,SOX,SIGEX,TAUS,TAUO,KSI,KCQ,HAX,HAY,M,NN,Z,BD;
'INTEGER'A,B,T,RU,P,Q,U,CYC,NCYC,NMAX;
'REAL" ARRAY'XH(/1:10,1:20,0:20/),AX(/1:20/),AY(/1:20/),GAX(/1:10/),
GAY(/l:10/),PC(/l:10/);
PK:=0.75;
FC:=2O.7;TAUMAX:=1.24;WO:=0.76;NMAX:=5;DW:=.03;DDEL:=O.03;c:=3000;
WMAX:=WO;DELE:=0;
MU:=0.23;
SPU:=0.85*O.77*6.39*FC**O.56;
AX:=38.01;
OUTSTRING(1,'('PARAMETERS')');LINE(1,1);
OUTSTRING(l,'('SPU=')');FIX(l,2,l,SPU);LINE(l,l);
OUTSTRING(l,'('DMAX=')');FIX(l,2,2,DMAX);LINE(l,l)j
OUTSTRING(1,'('MU=')');FIX(1,1,2,MU);LINE(1,1);
OUTSTRING(1,*('PK=')');FIX(1,1,2,PK);LINE(1,1);
OUTSTRING(1,'(INMAX=')');FIX(1,3,0,NMAX);LINE(1,3);
OUTSTRING( 1,' ( ' ')') ;LINE( 1,1);
OUTSTRING(l,'('W DELTA SIGMA TAU')');LINE(1,1);
RMAX:=0.5*DMAX;
'COMMENT' PROBABILITY DENSITY FUNCTION
'FOR'A:=1'STEP'1'UNTIL'10'DO " BEGIN'
DO:=(A/10-0.05)*DMAX;BD:=DO/DMAX;
PC(/A/):=0.532*BD**.5-0.212*BD**4-0.075*BD**6-0.036*BD**8;
PC(/A/):=(PC(/A/)-0.025*BD**10)*PK/DO;'END';
TOX:=0;WROX:=0;SOX:=0;QROX;=o;A:=0;
'COMMENT' UNDEFORMED CONTACT AREAS
HED:A:=A+1;R:=(A/10-0.05)*RMAX;B:=0;
OL:B:=B+l;T:=-l;
LIL:T:=T+1;RU:=T-B;'IF'RU<0.1'THEN'
'BEGIN"
Y:=0.1*R*T;XH(/A,B,T/):=SQRT(2*Y*R-Y**2+.0000000001);
'END'
'ELSE'
'BEGIN'XH(/A,B,T/):=10000;XH(/A,B,T-l/):=XH(/A,B,T-l/)+R/20;
'END';
' IF'T<10'THEN" GOTO'LIL;
'IF,B<10'THEN"GOTO'OL;
'IF'A<10'THEN''GOTO'HED;
A:=0;B:=0;T:=0;NCYC:=0;
'COMENT' LOB= START OF CYCLE NCYC
LOB:NCYC:=NCYC+1;LINE(1,2);W:=WO;POX:=0;TOX:=0;ROX:=0;
WROX:=0;DOX:=0;SOX:=0;DEL:=DELE;SIGI:=0;
OUTSTRINGd, '('CYCLE = ' ) ' );FIX(1,2,0,NCYC);
TAUI:=0;W:=WO-DW;
 - App.III.2 -

'COMMENT' CRACK WIDTH INCREMENT AND EXTERNAL NORMAL STRESS

DEB:W:=W+DW;'IF*NCYC=1'THEN'SIGEX:=2.2*(W-WO);
'IF'NCYC>1'THEN'SIGEX:=(.6-.02*NCYC)*W**3.4;
'COMMENT' SHEAR DISPLACEMENT INCREMENT
WIL:DEL:=DEL+DDEL;A:=0;
'COMMENT' DETERMINATION OF TOTAL CONTACT AREAS
ANK:A:=A+1;R:=(0.1*A-.05)*RMAX;B:=0;
FLOR:B:B+1;XSNO:=0;XSNB:=0;YSNO:=0;P:=0;T:=0;
ILSE:T:=T+1;Y:=0.1*R*T;'IF'Y<W'THEN'XCT:=0'ELSE'
XCT:=DEL+SQRT(R**2-(Y-R-W)**2);
G:=XH(/A,B,T/)-XCT;U:=T-B;
'IF'G>0"THEN'
' BEGIN" IF'P=0'THEN "BEGIN'
'IF'U<0'THEN"GOTO'ILSE;
'END'
'END';
' IF 'G<0'THEN" BEGIN'
'IF'P=0'THEN"BEGIN'
Q:=T-l;M:=0.1*R/(XH(/A,B,T/)-XH(/A,B,Q/));
NN:=0.1*R*(T-1)-M*XH(/A,B,Q/);Z:=NN-R-W;
XSN0:=-(M*Z-DEL)-SQRT((M*Z-DEL)**2-(M*M+1)*(DEL**2+Z*Z-R*R));
XSNO:=XSNO/(M*M+1);YSNO:=M*XSNO+NN;P:=1;
'IF'T<B'THEN "GOTO'ILSE;
'END';
XH(/A,B,T/):=XCT;
'END';
'IF'G<0'THEN''BEGIN " IF'P=1'THEN'
' BEGIN" IF'T<B'THEN "GOTO'ILSE;'END';'END';
' IF' G>0' THEN " BEGIN" IF'P=l'THEN "BEGIN'
Q:=T-i;M:=0.1*R/(XH(/A,B,T/)-XH(/A,B,Q/));
NN:=0.1*R*(T-1)-M*XH(/A,B,Q/);Z:=NN-R-W;
XSNB:=-(M*Z-DEL)+SQRT((M*Z-DEL)**2-(M*M-1)*(DEL**2+Z*Z-R*R));
XSNB:=XSNB/(M*M+1);YSNB:=XSNB;M=NN;P:=2;
'END';
'END';,IF'T<B'THEN"GOTO,ILSE;
'IF'P=1'THEN"BEGIN'
XSNB:=DEL+SQRT(R*R-(O.I*R*T-W-R)**2);YSNB:=O.I*T*R;'END';
AX(/B/):=XSNB-XSNO;AY(/B/):=YSNB-YSNO;
'IF'B<IO'THEN"GOTO'FLOR;
H A X : = 0 ; H A Y : = 0 ; ' FOR' B : = 1 ' STEP' 1 ' U N T I L ' 1 0 ' D O " BEGIN'
HAX:=HAX+AX(/B/);HAY+AY(/B/);* END';
GAX(/A/):=HAX/10;GAY(/A/):=HAY/10;'IF'A<10'THEN"GOTO'ANK;
'COMMENT' SUMMATION O F C O N T A C T AREAS
A:=0;AAX:=0;AAY:=0;
CAR:A:=A+1;DO:=(A/10-.05)*DMAX;
N:=0.1275*PC(/A/)*DMAX/DO;
AAX:=AAX+N*GAX(Ikl);AAY+N*GAY(Ikl);
'IF'A<10'THEN"GOTO'CAR;
'COMMENT' CALCULATION OF INTERNAL STRESSES
TAUI:=SPU*(AAY+MU*AAX);SIGI:=SPU*(AAX-MU*AAY);
* IF'SIGKSIGEX'THEN "BEGIN*
' IF' TAUKTAUMAX' THEN " BEGIN'
POX:=SIGI;ROX:=TAUI;DOX:=DEL;
BLANK(1,3);FIX(1,1,2>W);BLANK(1,3);FIX(1,1,2,DEL);
BLANK(1,3);FIX(1,2,2,SIGI);BLANK(1,3);FIX(1,2,2,TAUI);LINE(1,1);
'IF'N>1'THEN''BEGIN" IF'DEL<DELE'THEN'DDEL:=0.05
'ELSE'DDEL:=0.01'END';
'GOTO'WIL;
'END';'END';
 - App.III.3 -

'COMMENT' FOR THE CASE THAT THE INTERNAL NORMAL STRESS

'COMMENT' EXCEEDS THE EXTERNAL STRESS, INTERPOLATION OF '
'COMMENT' THE CRACK DISPLACEMENTS
'IF'SIGI>SIGEX'THEN''BEGIN'
DELS:=DEL-(SIGI-SIGEX)/(SIGI-POX+.OOOOOl)*DDELJ
TAUS:=TAUI-(DEL-DELS)*(TAUI-ROX+.0000001)/DDELj'END'
'ELSE''BEGIN'
TAUS:=TAUMAX+0.000001;DELS:=DEL-(TAUI-TAUMAX)/(TAUI-ROX+.00001)*DDEL;
'END';
'IF"TAUS<TAUMAX'THEN''BEGIN'
'COMMENT" CALCULATION OF TOTAL CONTACT AREAS FOR THE
'COMMENT' INTERPOLATED CRACK DISPLACEMENTS
TOX:=TAUS; WROX:=W;SOX:=SIGEX;QROX:=DELS;A:=0;
HEL:A!=A+1;R:=(A/10-0.05)*RMAX;B:=0;
AS:B:=B+1;T:=0;
NEL:T:=T+1;Y:=0.05*T*R;'IF'Y<W'THEN'XCT:=0'ELSE'
XCT:=DELS+SQRT(R*R-(Y-R-W)**2);
'IF'XCT>XH(/A,B,T/)'THEN'XH(/A,B,T/):=XCT;
'IF'T<B'THEN''GOTO'NEL;
'IF'B<20'THEN''GOTO'AS;
'IF'A<10'THEN''GOTO'HEL;
BLANK(1,3);FIX(1,1,2,W);BLANK(1,3);FIX(1,2,2,DELS);
BLANK(1,3);FIX(1,2,2,SIGEX);BLANK(1,3);FIX(1,2,2,TAUS);
BLANK(1,3);FIX(1,2,2,SIGI);LINE(1,1);
'GOTO'DEB;
'END'
'COMMENT' DETERMINATION OF END - DISPLACEMENTS
'ELSE''BEGIN''IF'SIGI>SIGEX'THEN''BEGIN'
DELE:=(TAUMAX-TOX)/(TAUS-TOX+.00001)*(DELS-QROX)+QROX;
WE:=(TAUMAX-TOX)/(TAUS-TOX+.00001)*(W-WROX)+WROX;
SIGE:=(TAUMAX-TOX)/(TAUS-TOX+.00001)*(SIGEX-SOX)+SOX;
'END "ELSE "BEGIN'
DELE:=DELS;WE:=W-(TAUI-TAUMAX)/(TAUI-ROX+.00001)*DW;
SIGE:=SIGI-(TAUI-TAUMAX)/(TAUI-ROX+.00001)*(SIGI-POX);
'END';
BLANK(1,3);FIX(1,1,2,WE);BLANK(1,3);FIX(1,1,2,DELE);
BLANK(1,3);FIX(1,2,2,SIGE);BLANK(1,3);FIX(1,2,2,TAUMAX);LINE(1,1);
A:=0;
'COMMENT' CALCULATION " OF END - DEFORMATION OF CONTACT AREAS
HELEN:A:=A+1;R:=(A/10-.05)*RMAX;B:=0;
AST:B:=B+1;T:=0;
NELL:T:=T+1;Y:=0.05*T*R;'IF'Y<WE'THEN' XCT:=0'ELSE'
XCT:=DELE+SQRT(R*R-(Y-R-WE)**2);
'IF'XCT>XH(/A,B,T/)'THEN'XH(/A,B,T/):=XCT;
'IF'T<B'THEN''GOTO'NELL;
'IF'B<20,THEN''GOTO'AST;
' IF'A<10'THEN "GOTO'HELEN;
'COMMENT' CALCULATION OF TAU AT THE MOMENT OF
'COMMENT' RESTITUTION OF THE SHEAR SLIP DURING UNLOADING
KSI: = (AAX-MU*AAY)/(AAX+MU*AAY+.00001)*( AAY-MU*AAX)/(AAY+MU*AAX+.00001);
TAUO:=KSI*TAUMAX;LINE(l,l);FIX(1,2,2,TAUO);LINE(1,2);
WMAX:=W;DELE:=0.2*DELE;
'IF'NCYC<NMAX'THEN''GOTO'LOB;
'END';
'END'
 - App.III.4 -

ANALYTICAL CONTACT MODEL

0 rem analytical contact model
1 n=12:qm=.005
3 dim x(15),w(15)
S open4,4
10 readp$,dm,f,mu,wm,xm,wo
15 foru=ltolO:readp(u):wm(u)=wm:xm(u)=xm:nextu
20 su=6.39*f".56
1000 w=wo:dx=.5*xm+.005:wa=wo+.02:da=0: rem start of cycle
1002 print#4," wisseling = "n
1003 print#4," scheurwijdte slip Labda-x Labda-y tau sigma-in sigma-
ex"
1004 print#4," "
1005 dx=.35
1009 rem crack width increment and external normal stress
1010 gosub9500:w=w+.02:sb=0:qq=.02:se=(0.6-.02*n)*w"(8-.33*n):al=l:kk=0
1011 ifn=lthense=.98*w".48
1014 ifqq<qmandsi>sethenl058
1015 rem shear displacement increment
1016 dx=dx+qq:lx=l:ly=l
1017 foru=ltol0:dd=(u-.5)/10*dm:ly(u)=0:lx(u)=0: rem calculation of the
1019 £orwu=0to9:uu=wu*dd/18: rem contact areas
1020 ifw>=dd/2thenly=0:,lx=0:uu=.51*dd:gotol052
1021 if(w+uu)>=dd/2thenly=0:lx=0:uu=.51*dd:gotol052
1025 dw=wm(u)-w:xd=xm(u)-dx
1030 ww=dw:xx=xd:gosub5000:cd=cc
1031 ww=w:xx=dx:gosub5000:c=cc
1032 ww=w-wa(u):xx=dx-da(u):gosub5000:ca=cc
1033 ly=(ca*xx-cd*xd+.5*(ww+dw)):tt=(c*dx-.5*w-uu):
1034 lx=-(ca*ww-cd*dw-.5*(xx+xd)):ll=(.5*dx-c*w+.5*(dd"2-4*(uu+w)*2)*.5):
1040 ifll<=0ortt<=0thenly=0:gotol052
1041 ifly<0orlx<0thenly=0:lx=0
1042 ifdx<xmthenl050
1046 ww=w-wa(u):xx=dx-da(u):gosub5000:ca=cc
1047 ly=(ca*xx+.5*ww-w-uu):lx=-(ca*ww-.5*xx-.5*(dd"2-4*(uu+w)"2)*.5)
1048 ifly<0orlx<0thenly=0:lx=0
1050 ly=ly/tt:lx=lx/ll
1051 ifly>lorlx>lthenly=l:lx=l
1052 ly(u)=ly(u)*ly*tt:lx(u)+lx*ll:nextwu:nextu:gosub9000

REM CALCULATION OF INTERNAL STRESSES

1053 gosubl0330:print"w= "w" dx:print"ta= "ta" si= "sitprint" "se
1054 ifda=0andta>0thenl057
1055 ifsi<seandkk=lthenqq=qq/2:gotol015
1056 ifsi<sethenl015
1057 ifsi/se<1.2thankk=l:dx=dx-qq:qq=qq/2:gotol015
1058 gosub9900:kk=0:y$=le£t$(str$(ly)+"000000",6)
1059 x$=left$(str$(lx)+"000000",6)
1060 w$=le£t$(str$(w)+"000000",6)
1061 d$=left$(str$(dx)+"000000",6)
1062 t$=left$(str$(ta)+"000000",6)
1063 s$=leftS(str$(si)+"000000",6)
1070 print#4,w$" "d$" "y$" "t$" "s$,se
1071 si=si/al
 - App.III.5 -

1074 rem taumax = 1.24 MPa.

1075 ifta>1.24andsi/se>1.20thendx=dx-qq:qq=qq/2:gotol015
1080 ifta>1.24then:n=n+l:gosub9600:x(n-l)=dx:w(n-l)=w:xm=dx:wm=w:gotol000
1090 ifsi>sethenwa=w:da=dx:gotol010
2000 gotol015
5000 cc=.5*(dd"2/(ww"2+xx"2)-l)".5:return
9000 lx=0:ly=0:foru=ltol0:ly=ly+ly(u)*p(u)/10:lx=lx+lx(u)*p(u)/10:nextu: return
9500 foru=ltol0:
9510 iflx(u)>Othenwa(u)=wa:da(u)=da:
9520 nexturreturn
9600 foru=ltol0:
9610 i£lx(u)>0thenwm(u)=w:xra(u)=dx:goto9620
9615 wm(u)=wa(u):xm(u)=da(u)
9620 print#4,u,wm(u),xm(u)Inextu:return
9900 ifw=wo+.02thenreturn
9910 al=se/si:ta=ta*al:si=se:ly=ly*al:lx=lx*al:return
9980 rem determination of static contact areas and
9990 rem of retention factors
10000 foru=0.05to.96step.l:p=(.532*u*.5-.212*u"4-.075*u"6-.036*u"8-.025*u*10)
10010 p=p*.75*.1275/38/uA2:s=s+p:printu,p:prints:stop
10330 ifw<.2thenl0333
10331 bx=7*dnT.056:by=3*dnr.28:bb=-1.47*dm"-.063
10332 kx=bx*w"-1.07/4*dx:gotol0340
10333 bx=7.74*dm*(-1.07*dnT-.01):by=4.5*dm".21:bb=-1.21*dnr-.03
10334 kx=bx/4*dx
10340 ax=0.01*(4*(kx-l+exp(-kx))/(exp(-kx)/kx+l))
10350 ky=by*w"bb/2*dx
10360 p=.5*((w-dx)-abs(w-dx))*exp(-l-dm/32-.5*w"2)
10370 ay= 0.01*(2*(ky-l+exp(-ky))/(exp(-ky)/ky+l)+p*ky*2/dx
10375 ly=ly*.9:lx=lx*1.25
10376 rem calculation of internal stresses
10380 ta=su*(ly+mu*lx)
10390 si=su*(lx-mu*ly):lx=lx/ax:ly=ly/ay
11000 return
50000 data "al",38,15.8,.22,1.10,.899,.740
50010 data .120, .023,.0107,.0064,.0043,.0031,.0023,.0017,.0012,.0007
59999 end
 - App.III.6 -

REDUCED CONTACT MODEL

0 rem reduced contact model
1 n=l:gm=.001:du=.25:tm=0.69
5 open4,4
10 readp$,dm,f,mu,wm,xm,wo
20 su=6.39*f".56
1000 w=wo:wa=wo+.02:da=0
1002 print#4," wisseling = "n
1003 print#4," scheurwijdte slip Labda-x Labda-y tau sigma-in sigma-ex"
1004 print#4," "
1005 dx=du:vx=xm-du:ifn=lthendx=0
1006 ifn=21thentm=1.24
1010 w=w+.02:qq=.02:se=(3.2-.01*n)*(w-.759):al=l:rem crack width increment
1011 ifn=lthense=3.8*(w-.759)
1015 ifqq<qmandsi>sethenl058
1016 dx=dx+qq:lx=l:ly=l:ifn=lthenl053:rem shear displacement increment
1017 vd=dx-du:lx=.8*(vd/vx)"2:ly=.7*(vd/vx)A3:rem retention factors
1051 ifly>lorlx>lthenly=l:lx=l
1053 gosubl0330:print"w= "w" dx= "dx:print"ta="ta"si="si:print"- "se
1054 i£da=0andta>0thenl057
1055 ifsi<seandkk=lthenqq=qq/2:gotol015
1056 ifsi<sethenl015
1057 ifsi/se>1.2thenkk=l:dx=dx-qq:qq=qq/2:gotol015
1058 gosub9900:kk=0:y$=left$(str$(ly)+"000000",6)
1059 x$= left$(str$(lx)+"000000",6)
1060 w$=left$(str$(w)+"000000",6)
1061 d$=left$(str$(dx)+"000000",6)
1062 t$=left$(str$(ta)+"000000",6)
1063 s$=left$(str$(si)+"000000",6)
1070 print#4,w$" "d$" "x$" "y$" "t$" "s$,se
1071 si=si/al
1075 ifta>tmandsi/se>1.20thendx=dx-qq:qq=qq/2:gotol016
1080 ifta>=tmthenn=n+l:l:xm=dx:wm=w:gotol000
1090 ifsi>sethenwa=w:da=dx:gotol010
2000 gotol015
9900 ifw=wo+.02thenreturn
9910 al=se/si:ta=ta*al:si=se:ly=ly*al:lx=lx*al:return
10330 ifw<.2thenl0333:rem determination of static contact areas
10331 bx=7*dm".056:by=3*dm".28:bb=-1.47*dm"-.063
10332 kx=bx*w"-1.07/4*dx:gotol0340
10333 bx=7.74*dm".06*w*(-1.07*dm"-.01):by=4.5*dm".21:bb=-1.21*dm"-.03
10334 kx=bx/4*dx
10340 ax=0.01*(4*(kx-l+exp(-kx))/(exp(-kx)/kx+D)
10350 ky=by*w"bb/2*dx
10360 p=.5*((w-dx)-abs(w-dx))*exp(-l-dm/32-.5*w-2)
10370 ay=0.01*(2*(ky-l+exp(-ky))/(exp(-ky)/ky+l)+p*ky*2/dx)
10380 ta=su*(ly*ay+mu*lx*ax):rem calculation of internal stresses
10390 si=su*(lx*ax-mu*ly*ay):
11000 return
50000 data "el",38,12.8,.20,0.76,.000,.740
59999 end
60000 save"0:labda-korrelel",8
Curriculum vitae

Adrianus Frans Pruijssers

10 december 1958 Geboren te Rotterdam

1971 - 1977 Van Oldebarneveldt

Scholengemeenschap te
Rotterdam

mei 1977 Diploma Atheneum - B

1977 - 1982 Technische Universiteit

Delft

Faculteit: Civiele Techniek

Af studeerrricht ing:
Constructieve waterbouwkunde

Afstudeerproject:
Sluisverlenging te Maasbracht

november 1982 Diploma civiel ingenieur

15 augustus 1982 Werkzaam als wetenschappelijk

31 december 1987 ambtenaar bij de sectie
betonconstructies van de
Faculteit der Civiele
Techniek, Technische Univer­
siteit Delft.
Projekt: Aggregate interlock

1 januari 1988 Werkzaam bij ontwerpbureau

van Dirk Verstoep b.v.

25 augustus 1982 Gehuwd met Augustina

Josephina Maria van Wezel

25 augustus 1984 Geboren Francisca

23 juni 1986 Geboren Marianne