Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
UNIVERS ITE IT •STELLENBOSCH •UNIVERS ITY
j ou kenn i s v ennoo t • you r know ledge pa r tne r
INTERFACIAL BOND PROPERTIES FOR ECC
OVERLAY SYSTEMS
HEINRICH STANDER
THESIS PRESENTED FOR THE DEGREE OF MASTER OF SCIENCE
AT THE DEPARTMENT OF CIVIL ENGINEERING OF
THE UNIVERSITY OF STELLENBOSCH
SUPERVISOR: PROF. G.P.A.G. van ZIJL
March 9, 2007
i
DECLARATION
I, the undersigned, declare that the work contained in this dissertation is my own original work and has
not previously in its entirety or in part been submitted at any University for a degree.
Signature:
Date:
Heinrich Stander University of Stellenbosch
ii
SUMMARY
Bonded overlays are increasingly used in concrete and reinforced concrete repair and rehabilitation appli-
cations, despite the high probability of interfacial debonding. Reasons for such failures include inefficient
substrate surface preparations, inappropriate overlay materials, poor curing conditions and time depen-
dent influences.
The introduction of engineered cement-based composite (ECC) as an overlay or repair material, does
not only address durability aspects but also structural performance. The associated ductility of the ma-
terial induces a high performance aspect where applied. It is crucial to execute reliable design methods,
especially at interfacial level, in order to harness the ductility at hand. The fact of the matter is that
through identifying the required performance, one can engineer an optimal bond through implementation
of reliable substrate surface preparation techniques (SSPT’s).
ECC is a material which exhibits ductile mechanical behaviour. The material matrix is reinforced with
synthetic fibres, in the case of this study, poly vinyl alcohol (PVA) fibres were used. The introduction of
fibres induces strain-hardening behaviour when in tension. Strain-hardening occurs from the first crack
onwards and is accompanied by ductile behaviour, due to a multiple cracking phenomenon. Multiple
cracking continues until the increased tensile load incurs localising of an existing crack.
The literature study investigates bond properties and bond model parameter test methods. A review of
composite design, mainly concrete to concrete, in local and international codes discloses design specifi-
cations towards calculating interfacial shear bonds. The interfacial transition zone (ITZ) between the
aggregate and cement matrix of concrete is used to define the interfacial bond characteristics and pro-
cesses. The next step is to investigate a variety of interfacial shear and tensile test methods, in order to
implement the most suitable tests.
An experimental design stage follows the literature review. The push-off method is chosen for the
interfacial shear parameter test. Numerical models are employed to analyse the experimental test set-
up, modifying its geometry and boundary conditions for optimal shear model parameter characterisation.
The essence of the interfacial characterisation experimental program is to find correlations between the
SSPT and the corresponding interfacial properties. This research investigates uncomplicated and prac-
tical preparation methods, which leads to characterisation of the respective interfacial bonds and their
methodical preparations. This is a necessary step towards developing design tools for overlay systems.
Numerical simulation of experimental results with the appropriate material models and parameter val-
ues, is an important stage in characterising the interfacial properties towards a computational predictive
Heinrich Stander University of Stellenbosch
iii
capacity. Both the shear and tensile parameter tests are analysed and a comparison with the experimen-
tal data displays satisfactory agreement.
An application of a thin bonded overlay system and its numerical simulation validate the results pro-
duced by the parametrical research stage. A comparison between a concrete and an ECC overlay system
indicates superior system behaviour of an ECC in terms of strength and ductility for the particular set-up.
The sandblasting SSPT shows the strongest interfacial bond in shear and tension. A short substrate
moistening period produces stronger bonds than long (24 hour) submersion in water. The experimental
programme was successful in delivering parameter values for the numerical models, because the numerical
responses agreed with that of the experiments.
Heinrich Stander University of Stellenbosch
iv
OPSOMMING
Die gebruik van gebonde bolae vir herstel en rehabilitasie in beton en gewapende betonwerk neem toe,
ondanks die hoe waarskynlikheid van tussenvlak ontbinding. Daar is verskeie oorsake van hierdie soort
falingsmeganisme, byvoorbeeld oneffektiewe voorbereiding van die substraat oppervlak, onvanpaste bo-
laag materiale, swak nabehandeling en tydsafhanklike invloede.
Die bekendstelling van “engineered cement-based composite” (ECC) as ’n bolaag en/of herstel materiaal
spreek duursaamheidsaspekte sowel as strukturele werksverrigting aan. Die vervormbaarheid van die
materiaal induseer ’n hoe werksverrigtingsaspek waar ook al dit aangewend word. Dit is noodsaaklik
om betroubare ontwerp metodes toe te pas, veral op die gebied van tussenvlak verband. Sodoende kan
die beskikbare materiaal vervormbaarheid nuttig ingespan word. Die feit van die saak is dat ’n optimale
verband ontwerp kan word deur die identifisering van die benodigde werksverrigting. Voorsiening word
daarvoor gemaak deur die implementering van ’n betroubare oppervlak voorbereidingsmetode vir die
substraat.
ECC is ’n materiaal wat duktiele meganiese gedrag openbaar. Die materiaal matriks is deur sintetiese
vesels (poliviniel alkohol - PVA) versterk, wat lei tot vervormingsverharding wanneer ECC onder trek
verkeer. Vervormingsverharding geskied vanaf die eerste kraak en word vergesel deur duktiele gedrag as
gevolg van die verskynsel van meervoudige krake. Die ontstaan van meervoudige krake gaan voort totdat
die verhoogde treklas lokalisering van ’n bestaande kraak veroorsaak.
Die literatuurstudie ondersoek verbandseienskappe asook verbandstoetsmetodes. ’n Oorsig van saamgestelde
beton ontwerp, in plaaslike en internasionale ontwerpskodes, voorsien ontwerp spesifikasies vir die
berekening van tussenvlak skuif verband van beton bolae. Die tussenvlak oorgangsone tussen die aggre-
gaat en matriks in gewone beton word gebruik om die verbandseienskappe en prosesse te verduidelik. ’n
Ondersoek van verskeie bestaande verband skuif- en trektoetsmetodes volg om sodoende die mees paslike
toetsmetodes te implementeer.
’n Eksperimentele ontwerpfase volg op die literatuur oorsig, waartydens die “push-off” metode gekies
word om gebruik te word vir die tussenvlak se skuif karakterisering toetse. Die eksperimentele op-
stelling, geometrie en randwaardes word aangepas volgens numeriese analises. Dus geskied optimering
van die toetsmodel vir skuif karakterisering.
Die hoofsaak van die eksperimentele program vir tussenvlak karakterisering is om verbintenisse tussen die
toegepaste oppervlak voorbereiding en gevolglike tussenvlak eienskappe te vind. Ongekompliseerde en
praktiese voorbereidingsmetodes word ondersoek wat lei tot die karakterisering van die betrokke tussen-
vlak verbande. Dit is ’n noodsaaklike stap tot die ontwikkeling van ontwerpsgereedskap vir ECC bolaag
Heinrich Stander University of Stellenbosch
v
sisteme.
Die gebruik van numeriese metodes met die noodsaaklike materiaal modelle en parameter waardes, is
’n belangrike stadium tydens die karakterisering van die tussenvlak. Dit lei tot verbeterde numeriese
voorspellingskapasiteit. Analises van beide die skuif en trek parameter toetse lewer bevredigende verge-
lykings met die eksperimentele data.
’n Toepassing van ’n dun gebonde bolaag sisteem en die numeriese simulasie daarvan, onderskryf die re-
sultate wat tydens die parametriese ondersoeke geproduseer is. In terme van sterkte en duktiliteit, dui ’n
vergelyking tussen ’n bolaag van beton en een van ECC, verbeterde sisteem gedrag vir die ECC opsie aan.
Die gebruik van sandstraling om die substraat se ruheid te verhoog, lei tot die sterkste skuif en trek
verband in vergelyking met die ander nagevorsde metodes. ’n Kort substraat bevogtigingsperiode van tien
minute lei tot ’n hoer verband as ’n langer bevogtigingsperiode van 24 uur. Die eksperimentele program
het suksesvolle parameter waardes geproduseer, omdat die resultate daarvan in numeriese simulasies
ooreengestem het met gedrag en eienskappe van die eksperimente.
Heinrich Stander University of Stellenbosch
vi
ACKNOWLEDGEMENTS
Several individuals and institutions have contributed to enable the conduct of this research. An initial
return on their investment is presented here in terms of acknowledgement and expression of gratitude.
Most of the gratitude goes to my promoter, Prof. Gideon van Zijl. His assistance, motivation and great
knowledge contributed tremendously to my understanding of concepts and problem-solving.
I would like to thank Dr Billy Boshoff, who became a mentor during this two year research period. His
unselfish assistance and dedication towards research was of great help and inspiration.
I would also like to thank Prof. Jan Wium for the role he played in my understanding of concepts
and assistance in design details, for work conducted during the first six months of research. This work
assisted towards understanding the requirements of overlay design better.
Everyone knows that experimental work can not be conducted without the assistance of laboratory
personnel. Experimental tests can not be conducted without the delicate manufacturing of equipment
for proper test set-ups and configurations, which was executed to precision by the workshop personnel.
Many thanks to Mr Arthur Layman and Mr Ashley Lindoor for their work in the laboratories and to Mr
Dion Viljoen and Mr Louis Fredericks for their work in the workshop.
A special thanks to Mrs Marle Lotter, a very sweet person and inspiring secretary who assisted me with
meticulous precision in organisational aspects.
Many thanks to Mr Cole Olsen for the research that he conducted during his two months visit to the
University of Stellenbosch. It assisted and supplied necessary information to the research programme.
We acknowledge the support of INFRASET Infrastructure Products, especially Mr Johan Kleynhans
and Mrs Santie Gouws, for their support and financial backing. I would like to thank the various com-
panies that supplied products to the research programme, Kuraray, Chryso and Sika. Thanks to the
Technology and Human Resources in Industry Programme (THRIP) of the South African Ministry of
Trade and Industry. A debt of gratitude to the Department of Civil Engineering at the University of
Stellenbosch for allowing me to conduct the research over the time period of two years and supplying all
the necessary tools and materials.
A special thanks to my friends and family for their love and support during this research period. Thanks
Kosie, Etienne and Teaan for keeping me sane at times and supplying refreshments on the way. Thank
you, Jessica, for your loving support and becoming my wife in this period.
Heinrich Stander University of Stellenbosch
TABLE OF CONTENTS
DECLARATION i
SUMMARY ii
OPSOMMING iv
ACKNOWLEDGEMENTS vi
TABLE OF CONTENTS x
LIST OF FIGURES xiii
LIST OF TABLES xv
NOMENCLATURE xviii
1 INTRODUCTION 1
1.1 SCOPE OF THIS REPORT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 BACKGROUND TO INVESTIGATION . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 OBJECTIVES OF REPORT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.4 LIMITATIONS AND SCOPE OF INVESTIGATION . . . . . . . . . . . . . . . 1
1.5 PLAN OF DEVELOPMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2 CONCEPTUAL DESIGN PROPERTIES 4
2.1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 PROPERTIES OF ECC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.1 Material properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.2 Matrix constituent properties . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2.2.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2.2.2 Fibres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.2.3 Admixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.2.4 Binder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.2.5 Fine Aggregate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 OVERLAY AND SUBSTRATE BOND PROPERTIES . . . . . . . . . . . . . . . 10
2.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
vii
TABLE OF CONTENTS viii
2.3.2 Code based design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3.2.1 South African national standard . . . . . . . . . . . . . . . . . . 11
2.3.2.2 International standards . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.3 Interfacial bond characteristics . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.3.2 Definition and classification . . . . . . . . . . . . . . . . . . . . . 15
2.4 PARAMETER TEST METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4.2 Test methods for shear parameter values . . . . . . . . . . . . . . . . . . . 18
2.4.3 Test methods for tensile parameter values . . . . . . . . . . . . . . . . . . 19
3 EXPERIMENTAL DESIGN 22
3.1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2 COMPUTATIONAL MATERIAL MODELS . . . . . . . . . . . . . . . . . . . . 23
3.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2.2 ECC model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2.3 Composite interface model . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3 NUMERICAL REFINEMENT OF SHEAR TEST METHOD . . . . . . . . . . . 27
3.4 EXPERIMENTAL SET-UP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4 EXPERIMENTAL PROGRAM 39
4.1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2 SPECIMEN PREPARATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2.1 Composite material mix designs . . . . . . . . . . . . . . . . . . . . . . . 39
4.2.1.1 Substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2.1.2 Overlay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2.2 Substrate surface roughening techniques . . . . . . . . . . . . . . . . . . . 40
4.2.3 Substrate moistening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2.4 Overlay casting and curing procedures . . . . . . . . . . . . . . . . . . . . 43
4.3 EXPERIMENTAL TESTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.3.1 Material tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.3.2 Interfacial tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.3.2.1 Shear tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3.2.2 Tensile tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.3.2.3 Condensed results . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5 NUMERICAL SIMULATION 65
5.1 Material models and input data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.1.1 Concrete material model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.1.2 Steel material model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.1.3 ECC material model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.1.4 Composite interface model . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Heinrich Stander University of Stellenbosch
TABLE OF CONTENTS ix
5.2 Numerical simulation of shear parameter tests . . . . . . . . . . . . . . . . . . . . 68
5.2.1 Global response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.2.2 Interfacial response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.3 Numerical simulation of tensile parameter tests . . . . . . . . . . . . . . . . . . . 74
5.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6 APPLICATION OF A THIN BONDED OVERLAY 78
6.1 EXPERIMENTAL PROCEDURE . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.1.2 Experimental programme . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.1.2.1 Specimen preparation . . . . . . . . . . . . . . . . . . . . . . . . 79
6.1.2.2 Test set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.1.3 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.2 NUMERICAL VALIDATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.2.1 Numerical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.2.2 Numerical results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.3 APPLICATION COMPARISON . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.4 SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7 CONCLUSIONS AND RECOMMENDATIONS 92
7.1 CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
7.1.1 Substrate surface preparation . . . . . . . . . . . . . . . . . . . . . . . . . 92
7.1.2 Interfacial bond . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
7.1.3 Interfacial shear test refinement . . . . . . . . . . . . . . . . . . . . . . . . 93
7.1.4 Numerical modelling of parameter tests . . . . . . . . . . . . . . . . . . . 94
7.1.5 Bonded overlay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
7.1.6 Design of composite interface . . . . . . . . . . . . . . . . . . . . . . . . . 95
7.2 RECOMMENDATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
BIBLIOGRAPHY 99
A STRUCTURAL DESIGN CODES 100
A.1 SABS - Code of practice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
A.2 EUROCODE 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
B RESULTS: INTERFACIAL EXPERIMENTS 108
B.1 Reference surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
B.1.1 Moistening period: 24 hours . . . . . . . . . . . . . . . . . . . . . . . . . . 108
B.1.2 Moistening period: 10 minutes . . . . . . . . . . . . . . . . . . . . . . . . 109
B.2 Scrape surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
B.2.1 Moistening period: 24 hours . . . . . . . . . . . . . . . . . . . . . . . . . . 110
B.2.2 Moistening period: 10 minutes . . . . . . . . . . . . . . . . . . . . . . . . 110
B.3 Sandblast surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Heinrich Stander University of Stellenbosch
TABLE OF CONTENTS x
B.3.1 Moistening period: 24 hours . . . . . . . . . . . . . . . . . . . . . . . . . . 111
B.3.2 Moistening period: 10 minutes . . . . . . . . . . . . . . . . . . . . . . . . 112
B.4 Drill holes surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
B.4.1 Moistening period: 24 hours . . . . . . . . . . . . . . . . . . . . . . . . . . 113
B.5 Precast grooves surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
B.5.1 Moistening period: 24 hours . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Heinrich Stander University of Stellenbosch
LIST OF FIGURES
1.1 Systematic flow of the research program. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1 The tensile stress-strain behaviour of cement-based composites. . . . . . . . . . . . . . . . 5
2.2 Steady-state crack analysis presents two crack propagation scenarios: (a) The Griffith
crack where the fibres are shown as springs which slide out or rupture in the mid-crack
section where δm exceeds δp. (b) The steady-state flat crack where the fibres remain intact
as the crack propagates under a constant σss, with the opening δss less than δp [22]. . . . 6
2.3 Differences in ITZ for concrete containing aggregates with (a) a dense outer shell and (b)
a porous outer surface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 The influence of mechanical interlocking on interfacial shear and tension, at micro-scale. . 16
2.5 The different interfacial surface areas. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.6 A schematic illustration of various test methods to determine τxy at the interface of two
different cementitious materials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.7 (a) Direct uniaxial tensile test, (b) Wedge splitting test and (c) Cylinder splitting test. . . 21
3.1 The unloading behaviour of the damage formulation. . . . . . . . . . . . . . . . . . . . . . 24
3.2 The uniaxial stress-strain response of ECC, represented by a tri-linear curve. . . . . . . . 25
3.3 Two-dimensional interface model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.4 The geometry of the interfacial shear parameter specimen (dimensions in mm). . . . . . . 30
3.5 Shear stress-displacement curves for the different boundary conditions. . . . . . . . . . . . 31
3.6 A comparison of contour plots for the three different boundary conditions is depicted by
each column. Shear stress contours are illustrated from (a) to (l) and principle strain
contours from (m) to (o). The first four rows compare the deformation and stress dis-
tributions at the four stages of the stress-strain relationship. The last row compares the
highest respective strain distributions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.7 A comparison of the σxx and τxy distributions over the interface for each boundary con-
dition, indicated respectively by the each column. The rows represent the four stages of
the stress-displacement curve in sequential form. . . . . . . . . . . . . . . . . . . . . . . . 35
3.8 Experimental set-up for the interfacial tensile parameter test (dimensions in mm). . . . . 37
3.9 Experimental set-up for the interfacial shear parameter test (dimensions in mm). . . . . . 38
4.1 Application of the scrape roughening technique. . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2 Illustrations of (a) sand, (b) drill and (c) groove roughened surfaces. . . . . . . . . . . . . 42
4.3 Casting of ECC on a concrete substrate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.4 Systematic flow of the specimen preparation. . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.5 Specimen geometry of the ECC direct uniaxial tensile test. . . . . . . . . . . . . . . . . . 46
4.6 Test set-up for ECC uniaxial tensile testing. . . . . . . . . . . . . . . . . . . . . . . . . . . 46
xi
LIST OF FIGURES xii
4.7 Uniaxial tensile test results at different ages. The coloured responses indicate different
tests of the same specimens. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.8 The development of Young’s modulus over time. . . . . . . . . . . . . . . . . . . . . . . . 49
4.9 Shear test set-up. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.10 Shear-bond and Mode II fracture energy development of sandblasted specimens. . . . . . 52
4.11 Shear stress plotted against vertical slip and related to normal uplift for (a) unreinforced
SB1410, (b) SB1424 and (c) reinforced SB1410 specimens. . . . . . . . . . . . . . . . . . . 54
4.12 Post-fractured surface of a unreinforced SB1410 specimen. . . . . . . . . . . . . . . . . . . 55
4.13 Post-fractured surface of a reinforced SB1410 specimen. . . . . . . . . . . . . . . . . . . . 55
4.14 Shear stress plotted against vertical displacement and related to normal displacement
(dilatation) for (a) S1410 and (b) S1424 specimens. . . . . . . . . . . . . . . . . . . . . . . 56
4.15 Post-fractured surface of a S1410 specimen subjected to shear loading. . . . . . . . . . . . 57
4.16 Experimental set-up of the interfacial tensile test. . . . . . . . . . . . . . . . . . . . . . . . 58
4.17 Tensile stress versus vertical displacement for (a) TSB1410 and (b) TSB1424 specimens. . 59
4.18 Interfacial tensile test of a sandblasted specimen, showing bridging fibres. . . . . . . . . . 60
4.19 The concrete part of a sandblasted interfacial tensile test, showing ECC material on its
fracture surface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.20 Post-fractured surfaces for a sandblasted tensile specimen, showing ECC on the left and
concrete on the right. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.21 Tensile stress plotted against vertical displacement for (a) S1410 and (b) S1424 specimens. 61
4.22 Post-fractured surfaces of a scraped tensile specimen, illustrating ECC on the left and
concrete on the right. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.23 Interfacial parameter results at 14 days for tension and shear tests. . . . . . . . . . . . . . 64
5.1 The exponential softening curve and mode I fracture energy for the concrete material model. 66
5.2 Numerical results: A comparison of the experimental and numerical shear responses for
(a) τxy versus v and (b) u versus v for unreinforced SB1410 specimens. . . . . . . . . . . . 70
5.3 Numerical result: A comparison of the experimental and numerical responses for τxy versus
u of SB1410 specimens. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.4 Numerical result: Contour plots of τxy at pre-peak values of 20% and 100%. . . . . . . . . 72
5.5 Numerical results: Contour plots of τxy at post-peak values of 50% and 20%. . . . . . . . 72
5.6 Numerical result: The interfacial σxx at four stages of the shear test on a SB1410 specimen. 73
5.7 Numerical result: The interfacial εxx at four stages of the shear test on a SB1410 specimen. 74
5.8 Numerical result: The interfacial τxy at four stages of the shear test on a SB1410 specimen. 74
5.9 Numerical result: A comparison of the experimental and numerical tensile responses for
ft,i versus u of SB1410 specimens. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.10 Numerical results: Peak σyy,i and εyy,i values for a direct uniaxial tensile test on a SB1410
specimen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.11 Numerical results: Post-peak σyy,i and εyy,i values for direct uniaxial tensile test on a
SB1410 specimen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.1 (a) Sawing of concrete substrate and (b) casting of beam overlay. . . . . . . . . . . . . . . 79
6.2 Geometrical aspects of the composite beam specimens (dimensions in mm). . . . . . . . . 80
6.3 Experimental set-up for the three-point bending test. . . . . . . . . . . . . . . . . . . . . . 80
6.4 Experimental result: The applied force versus nominal displacement for both SB1410 and
S1410 in three-point bending. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Heinrich Stander University of Stellenbosch
LIST OF FIGURES xiii
6.5 Experimental results: Interfacial delamination responses for three-point bending tests on
(a) SB1410 and (b) S1410. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.6 The formation of a localised crack in the middle of the bonded overlay of a SB1410 specimen. 82
6.7 Numerical results: Three-point bending response curves demonstrating (a) the applied
force versus deflection and (b) interfacial delamination at the inner points of a SB1410
specimen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.8 A comparison of experimental and numerical data, demonstrating (a) the applied force
versus deflection and (b) interfacial delamination at the inner points of a SB1410 specimen. 85
6.9 Numerical result: Contour plots of the σ1 at 50% (pre-peak) and 100% of the peak load. . 86
6.10 Numerical results: Contour plots of the ε1 at 50% (pre-peak) and 100% of the peak load. 86
6.11 Numerical results: Contour plots of the equivalent strain at 50% (pre-peak) and 100% of
the peak load. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.12 Numerical results: Contour plots of the equivalent strain at 100% and 50% (post-peak) of
the peak load, showing the localised areas. . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.13 Numerical results: Comparison of three point bending responses (* fully bonded). . . . . . 90
A.1 Indented construction joint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
A.2 Shear diagram representing the required interface reinforcement . . . . . . . . . . . . . . . 107
B.1 Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimens
with a 10 minute moistening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
B.2 Interfacial responses in (a) shear and (b) tension, at 28 days for the reference specimens
with a 10 minute moistening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
B.3 Interfacial response in shear at 28 days for the reference specimens with a 24 hour moist-
ening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
B.4 Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimens
with a 24 hour moistening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
B.5 Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimens
with a 10 minute moistening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
B.6 Interfacial responses in (a) shear and (b) tension, at 7 days for the reference specimens
with a 24 hour moistening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
B.7 Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimens
with a 24 hour moistening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
B.8 Interfacial responses in (a) shear and (b) tension, at 28 days for the reference specimens
with a 24 hour moistening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
B.9 Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimens
with a 10 minute moistening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
B.10 Interfacial response in shear at 28 days for the reference specimens with a 10 minute
moistening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
B.11 Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimens
with a 24 hour moistening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
B.12 Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimens
with a 24 hour moistening period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Heinrich Stander University of Stellenbosch
LIST OF TABLES
2.1 Properties of PVA-fibre [15]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 The composition ratio of binder constituents for ECC. . . . . . . . . . . . . . . . . . . . . 8
3.1 Initial parameter values for the ECC material model. . . . . . . . . . . . . . . . . . . . . . 25
3.2 Parameter variables for the composite interface material model. . . . . . . . . . . . . . . . 27
3.3 Parameter values for the concrete material model . . . . . . . . . . . . . . . . . . . . . . . 28
3.4 Interfacial parameter values provided by research conducted by E.C. Olsen [28] . . . . . . 29
4.1 Concrete mix design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2 ECC mix design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.3 Material properties of the concrete substrate. . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.4 ECC compressive mean values. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.5 Experimentally determined parameter values for the ECC material model. . . . . . . . . . 48
4.6 Number of shear specimens tested for each SSPT. . . . . . . . . . . . . . . . . . . . . . . . 50
4.7 Shear parameter values obtained from experiments on sandblasted specimens. . . . . . . . 52
4.8 Shear parameter values obtained from experiments on scraped specimens . . . . . . . . . 56
4.9 Number of tensile specimens tested for each SSPT. . . . . . . . . . . . . . . . . . . . . . . 58
4.10 Tensile parameter values obtained from experiments on sandblasted specimens. . . . . . . 59
4.11 Tensile parameter values obtained from experiments on scraped specimens. . . . . . . . . 60
4.12 Shear parameter values obtained from experiments performed on the various SSPT’s. . . . 62
4.13 Tensile parameter values obtained from experiments performed on the various SSPT’s. . . 62
4.14 The bond type induced by the particular roughening technique. . . . . . . . . . . . . . . . 63
5.1 Tensile strengths for various concrete grades [9]. . . . . . . . . . . . . . . . . . . . . . . . . 66
5.2 Fracture energy base values (Gfo) and standard values (GIf ) [9]. . . . . . . . . . . . . . . . 67
5.3 Parameter values for the concrete material model . . . . . . . . . . . . . . . . . . . . . . . 67
5.4 Parameter values for the ECC material model. . . . . . . . . . . . . . . . . . . . . . . . . 68
5.5 Parameter values for the composite interface material model, characterising the SB1410
interface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.6 Shear parameter values extracted from masonry tests [41]. . . . . . . . . . . . . . . . . . . 69
5.7 The final parameter values for the numerical interface model, characterising SB1410. . . . 69
5.8 Outstanding interfacial parameter values after an initial iteration. . . . . . . . . . . . . . . 77
6.1 ECC material model properties for the three-point bending tests. . . . . . . . . . . . . . . 83
6.2 The σfc and εsh values used for each analysis. . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.3 A comparison of peak loads and deflections for three point beam tests. . . . . . . . . . . . 90
A.1 Design ultimate horizontal shear stresses at interface (SABS 0100-01). . . . . . . . . . . . 103
xiv
NOMENCLATURE
¯De Elastic stiffness matrix
δ Residual friction coefficient
δm Griffith crack mid-opening width
δp Plastic fibre strain length
δss Steady-state crack width
γ Shear strain
κp Equivalent plastic relative displacement
µ Coefficient of friction
ν Poisson’s ratio
ω Damage indicator scalar
Φ Interface friction coefficient
Φr Residual interface friction coefficient
Φ0,exp Experimental initial interface friction coefficient
Ψ Dilatancy
Ψ0 Interfacial dilatancy at zero confining stress and shear slip
Ψ0,exp Experimental initial dilatancy
σ1 Principle stress
σt Tensile stress
σcon,i0 Confining normal stress on the interface causing Ψ = 0
σcon,i Confining stress normal to the interface
σfc First crack tensile strength
σss Steady-state stress
σtu Ultimate tensile strength
σxx Stress in normal direction on normal plane
σyy,i Interfacial normal stress for a direct uniaxial tensile test
σyy Stress in tangential direction on tangent plane
τRd,i Interfacial design shear strength
τxy Shear stress
υp Plastic interfacial shear slip
xv
NOMENCLATURE xvi
ε1 Principle strain
εsh Strain-hardening strain
εu,f Ultimate fibre elongation
εxx Strain in normal direction on normal plane
εyy,i Interfacial normal strain for a direct uniaxial tensile test
εyy Strain in tangential direction on tangent plane
Cs Shear traction contribution to compressive failure
dmax Maximum aggregate size
fc Compressive stress
fs Tensile strength of the splitting test
ft Tensile strength
fu Ultimate load
fck-cube Characteristic cube strength of concrete
fcm Mean compressive cylinder strength
fctd Design tensile strength of concrete
fctk,min Minimum characteristic tensile strength
fcu Ultimate compressive strength
ft,c Tensile strength of concrete
ft,f Tensile strength of fibres
ft,i Tensile strength of interface
ftu Ultimate tensile strength of ECC
Gf Fracture energy
GIIf Mode-II fracture energy
Gf,i Interfacial fracture energy
GfcCompressive fracture energy
Ls Theoretical pull-out length of fibres equal to half the fibre length
P Force/Area
u Interfacial normal uplift
utop Experimental interfacial normal uplift at the top point of the interface
up Plastic interfacial normal uplift upon shear-slipping
up,exp Experimental plastic interfacial normal uplift upon shear-slip
v Interfacial shear-slip
vp,exp Experimental plastic interfacial shear-slip
A/B Aggregate/Binder ratio
ASTM American Society for Testing and Materials
BC Boundary condition
Heinrich Stander University of Stellenbosch
NOMENCLATURE xvii
C Chemical
CH Calcium hydroxide
COV Coefficient of variance
CSH Calcium silica hydrate
DH1424 Holes drilled and moistened for 24 hours; tested at 14 days.
DOF Degree of freedom
E E-modulus
E-modulus Elastic modulus
ECC Engineered cement-based composites
F-F Fixed-fixed boundary condition
F-R Fixed-rotational boundary condition
FA Fly Ash
FE Finite element
FEA Finite element analysis
FEM Finite element modelling
FRC Fibre reinforced concrete
G-modulus Shear modulus
GGCS Ground granulated Corex slag
IBP Interfacial bond properties
ITZ Interfacial transition zone
LVDT Linear variable displacement transducer
M Mechanical
PG1424 Precast grooves, moistened for 24 hours; tested at 14 days.
PVA Poly Vinyl Alcohol
R-R Rotational-rotational boundary condition
R/C Reinforced concrete
R1410 Reference surface, moistened for 10 minutes; tested at 14 days.
R2810 Reference surface, moistened for 10 minutes; tested at 28 days.
R2824 Reference surface, moistened for 24 hours; tested at 28 days.
RD Relative density
RILEM The International Union of Testing and Research Laboratories for Materials and Structures
S1410 Scraped and moistened for 10 minutes; tested at 14 days.
S1424 Scraped and moistened for 24 hours; tested at 14 days.
SB1410 Sandblasted and moistened for 10 minutes; tested at 14 days.
SB1424 Sandblasted and moistened for 24 hours; tested at 14 days.
SB2810 Sandblasted and moistened for 10 minutes; tested at 28 days.
Heinrich Stander University of Stellenbosch
NOMENCLATURE xviii
SB2824 Sandblasted and moistened for 24 hours; tested at 28 days.
SB724 Sandblasted and moistened for 24 hours; tested at 7 days.
SP Superplasticiser
SSMT Substrate surface moistening technique
SSPT Substrate surface preparation technique
SSRT Substrate surface roughening technique
VA Viscous Agent
W/B Water/Binder ratio
W/C Water/Cement ratio
Heinrich Stander University of Stellenbosch
Chapter 1
INTRODUCTION
1.1 SCOPE OF THIS REPORT
This research concerns the characterisation of the interfacial bond properties (IBP) between
reinforced concrete (R/C), the substrate, and Engineered Cement-Based Composite (ECC),
the overlay. Different substrate surface preparation techniques (SSPT’s) were researched in
order to produce varying interfacial bond properties. Implementation of the IBP through finite
element analysis, indicates an optimal SSPT as needed for a specific structural application.
1.2 BACKGROUND TO INVESTIGATION
The need for this study arose out of the necessity to describe the interfacial debonding nature
of ECC and R/C overlay systems and consequently supply the most suitable SSPT. An earlier
investigation of bond preparation methods conducted with simpler experimental procedures,
suggested that less surface preparation, i.e. a weaker interfacial bond, amounted to more desir-
able and durable repairs [21]. A performance demanding application conducted experimentally
during the early stages of this research program, illustrated a contradiction. In order to activate
composite action, some form of bond should exist. The need for better understanding of the
interfacial bond was thus implicated.
1.3 OBJECTIVES OF REPORT
The main objective of this study is to report on IBP determined from parameter tests and
to indicate procedures to achieve the corresponding SSPT’s. The product of this work is a
numerical modelling technique by which composite behaviour can be predicted. This enables a
reliable and affordable development of ECC/Concrete overlay strategies.
1.4 LIMITATIONS AND SCOPE OF INVESTIGATION
Properties of the interfacial bond are limited to the influence that a single type of ECC material
has on concrete surface preparations. The ECC material at hand, was developed at the Univer-
1
1.5. PLAN OF DEVELOPMENT 2
sity of Stellenbosch. Capturing of interfacial post-crack behaviour depended on the capabilities
of the testing apparatus at hand, which were not optimal but useable. Tests were conducted
with displacement control between the steel platens of the testing machine. In order to achieve
accurate post-crack behavioural mapping, control needs to be exerted just over the interface
crack. Conducting tests this way, diminishes instability due to the contribution of the elastic
strain in specimen deformation.
The investigation stipulated some pre-examined conclusions. Correlations to results of previous
studies were found. The research led to new ideas on how and where to implement ECC as an
overlay. However, conclusions drawn from experimental and numerical work should be verified
by further tests using a broad range of different ECC overlays and concrete substrates.
1.5 PLAN OF DEVELOPMENT
Following the introduction, this report embarks on a continuous flow of methodical research
as depicted by the flow chart in Figure 1.1. In chapter two the conceptual design properties
will be conveyed. This includes properties and micromechanical behaviour of ECC material
and theories behind it. A section which discusses code based overlay design follows, with a
discussion on both local and international standards. Several parameter test methods are also
illustrated and discussed.
Chapter three hosts the discussion of the experimental design, referring to and building on the
foregoing literature review. The choice and workings of the numerical models, used to simulate
the experimental data, are then presented. The chapter further proceeds into the developmental
stage of the parameter test set-ups, through finite element analysis (FEA).
In chapter four the interfacial bond characteristics are discussed via the disclosure of the ex-
perimental data. During this chapter the different SSPT’s are examined along with their shear
and tensile parameter values.
The report then proceeds in chapter five with the numerical simulation stage, the parameter
values used for the applicable material models and the eventual computational results. As ap-
plication and validation, a thin bonded overlay is studied experimentally and computationally
in chapter six. It is an important stage which succeeds on chapters four and five.
Conclusions are drawn in the final chapter, together with recommendations for further work.
It is succeeded by the bibliography and two appendices. The two appendices convey extracts
from structural design codes and interfacial experimental results respectively.
Heinrich Stander University of Stellenbosch
1.5. PLAN OF DEVELOPMENT 3
Problem Statement
Literature Review
Experimental Procedure
Experimental Design
Experimental Tests Numerical Simulation
Numerical Refinementof
Test Methods
Existing Methods+
Numerical Tools
Numerical Refinementof
Test Methods
Validation of Results
Experimental Application Numerical Validation
Conclusions + Recommendations
Figure 1.1: Systematic flow of the research program.
Heinrich Stander University of Stellenbosch
Chapter 2
CONCEPTUAL DESIGN
PROPERTIES
2.1 INTRODUCTION
Concepts about the development of ECC as a cost effective building material will be presented
in the subsequent sections. Attention is given to the unique material properties and the con-
stituents used to achieve the material behaviour. Most research on the application of ECC as
a structural repair material, provide vague reference to the implemented SSPT. Conceptually,
standardised methods are necessary to promote certainty and reliability when designing ECC
overlay systems. Research prior to the start of this study, revealed various SSPT’s, inconsistent
in application and functioning.
The attention of this literature study will be directed towards characterising ECC, its con-
stituents and properties. Secondly the attention will focus on interfacial bond properties be-
tween substrate and overlay and methods to produce sufficient bonds. This information is
mostly gathered from research based on concrete overlays. The third section discusses different
options for experimental test set-ups in order to produce shear and tensile parameter values.
2.2 PROPERTIES OF ECC
2.2.1 Material properties
The most fundamental mechanical property of ECC is the ductile tensile behaviour after the
occurrence of the first crack. The fundamental difference between ECC and fibre reinforced
concrete (FRC), is the strain-hardening that accompanies the ductile behaviour of ECC. FRC
experiences a diminishing in strength with an increasing strain after first crack, which is referred
to as tension-softening. Strain-hardening is a behaviour of increased post cracking strength,
which exhibits a formation of multiple cracks which are not always visible to the eye. Multiple
cracks are the only phenomenon through which a rigid cement-based material manifests strain
larger than elastic strain, with an increase in strength. It is accomplished via the capacity of
4
2.2. PROPERTIES OF ECC 5
fibres to bridge the applied stresses over the matrix cracks. The fibres implemented for the
research were poly vinyl alcohol (PVA) fibres. Only when strain-hardening reaches its peak
at about three to six percent strain, ECC resolves in a tension-softening trend, also exhibited
by FRC. The cumulative interfacial bond strength between PVA fibres and the matrix over a
section controls the ultimate tensile strength of ECC (ftu). A stress-strain curve similar to that
of a ductile metal is achieved and illustrated by Figure 2.1.
ECC
FRC
concrete
post-crack region
strain-hardening
tension-softening
σ
ε
Figure 2.1: The tensile stress-strain behaviour of cement-based composites.
Crack widths in the matrix are controlled by the elastic strain and slip of fibres bridging the
matrix cracks. The ability of ECC to control crack width, occurs in the envelope starting from
first cracking strength (σfc) and ending at ftu. ECC, like concrete, is a dynamic material in
which the hydration process continues over very long periods until all the free water is incor-
porated in hydration products. When cracks occur as they do in all cement-based materials,
surfaces are exposed to water particles. The ability of ECC to limit crack widths, induces the
phenomenon of autogenous healing (formation of C-S-H over fine cracks) [18].
The elastic modulus (E-modulus) of ECC plays a crucial role in the choice of an application.
Load-bearing composites that utilise overlays with a difference in stiffness, incur higher stresses
to stiffer parts. This happens only if total bond between the two parts are obtained, otherwise
one produces debonding of the substrate and overlay. From elastic theory of composite sec-
tions, it can be shown that the stress peaks may arise in the section. The stresses at coinciding
material points, but across the connecting interface in the two different materials in the case of
an overlay and substrate, differ by a factor equal to the ratio of the E-moduli. Through this
mechanism, there is a shift in neutral axis towards the part with high E-modulus, which may
Heinrich Stander University of Stellenbosch
2.2. PROPERTIES OF ECC 6
partly reduce the effect of stress peak. Nevertheless, combination of parts with significantly
differing E-moduli should be avoided.
Another important consideration when developing a ductile ECC, is the amount of fracture
energy (Gf ) exhibited by the cementitious matrix. Gf is an attribute influenced by the micro-
structure [23]. It is measured as the amount of energy dissipated per unit crack surface. High
Gf values for the material matrix, adversely affects the strain-hardening property. It may lead
to fibre pull-out or break, instead of the crack extending in its length. Figure 2.2 shows two
types of cracks that might occur in ECC, the Griffith and steady-state cracks [22]. In a Griffith
crack the fibres slide out or rupture in the midsection, where the crack mouth opening (δm)
exceeds the ultimate fibre deformation (δp). The fibres for the steady-state crack remain intact
as the crack propagates under a constant steady-state stress (σss), with the steady-state crack
opening (δss) less than δp.
(a)
(b)
δm > δpbroken or softening “springs”
σss
δss < δp
Figure 2.2: Steady-state crack analysis presents two crack propagation scenarios: (a) The Griffithcrack where the fibres are shown as springs which slide out or rupture in the mid-crack section where δmexceeds δp. (b) The steady-state flat crack where the fibres remain intact as the crack propagates undera constant σss, with the opening δss less than δp [22].
One way of manipulating Gf in cement-based composites is to vary the aggregate content.
Higher aggregate content leads to more tortuous crack paths and associated high Gf . Higher
aggregate content also increases the E-modulus [44]. Thus, increased E-modulus and improved
substrate-overlay stress distribution can be achieved by higher aggregate content, but at the
risk of reduced tensile ductility.
2.2.2 Matrix constituent properties
2.2.2.1 General
ECC utilises the same ingredients as those in fibre reinforced cements (FRC), such as water,
cement, sand, fibre and other common chemical additives. Coarse aggregates are not used as
they tend to increase fracture toughness which adversely affects the unique ductile behaviour
of the composite. Unlike some high performance FRC, ECC does not utilise large amounts of
Heinrich Stander University of Stellenbosch
2.2. PROPERTIES OF ECC 7
fibres. Rather, the combination of ingredients based on micromechanical principles, is what
makes the mechanical properties of ECC products so unique. In general 2% or less by volume
of discontinuous fibre is adequate. This relatively small amount of fibre with its short length
contributes to the fact that the mixing procedure is similar to that of ordinary concrete.
Consolidated ECC consists of three components namely fibres, cement-based matrix and the
interface. Proportioning of each component with the correct mechanical and geometric prop-
erties is necessary to attain the unique ductile behaviour. Material design of ECC is guided
by micromechanical principles. Admixtures are incorporated into the mix design, because they
promote the properties of fresh ECC, which in turn result in a beneficial consolidated state.
The beneficial condition is defined by the uniform spread of constituents, causing a resem-
blance to isotropic material properties. Uniformity of constituent rheology is imperative for the
functioning of micromechanical properties.
2.2.2.2 Fibres
Several different types of fibres exist, which can be used in ECC. PVA-fibres were used in this
research. These fibres were chosen because of their relative high tensile strength (ft) and E-
modulus (E). Thereby the fibre breakage at the crack regions, which will lead to premature
brittle behaviour of the composite, can be avoided or restricted. A fibre length of 12 millimetre
(mm) was chosen to ensure better fibre dispersion and a more workable material in its fresh
state. Longer fibres tends to coagulate. The ft and E-modulus are shown in the Table 2.1.
Table 2.1: Properties of PVA-fibre [15].
Type Diameter [mm] Length [mm] ft,f [GPa] E [GPa] εu,f [%]
PVA-REC15 0.04 12 1.6 37 6
2.2.2.3 Admixtures
Admixtures are chemicals that are added to the concrete immediately before or during mixing
and significantly alter its fresh, early age or hardened state to economic or physical advantage.
Only small quantities are required, typically 0.1 to two percent by weight of the binder.
Superplasticiser (SP)
Known as workability aids, it increases the fluidity or workability of cement paste or concrete.
They are also referred to as high-range water reducers. It has a high molecular weight and is
manufactured to high standards of purity and can therefore achieve substantially greater pri-
mary effects without significant undesirable side-effects.
The addition of fibres and methyl cellulose lead to an increase of viscosity, thus needing an
additive (SP) to improve the workability.
Heinrich Stander University of Stellenbosch
2.2. PROPERTIES OF ECC 8
The mode of action induced by SP, is purely physical, a combination of mutual repulsion and
steric hindrance between cement particles, creating less friction when the particles move. The
behaviour of any particular combination of SP and cement will depend on several factors other
than the admixture type, including the cement composition, the cement fineness and the wa-
ter/binder (W/B) ratio.
Substantially increased performance can be obtained if the SP is added a short time (1-2 min-
utes) after the initial contact between the mix water and binder [19]. This reference notes that
prior research depicts an appearance that when SP and mix water is added simultaneously, a
significant amount of SP is incorporated into the rapid (CaO)3.Al2O3/gypsum reaction, hence
reducing that amount available to attain workability. The SP action only occurs for a limited
time period.
The SP used in this research is a product of Chryso, namely Premia 100. The relative density
(RD) is 1.2.
Methyl Cellulose (Viscous Agent)
Chryso Aquabeton ZA is a powder additive necessary to prevent the segregation of the fresh
concrete. Also known as viscous agent (VA), it causes an increase of intermolecular shear force
of the fresh mix. Thus it acts as a dispersion agent and assists the uniform dispersion of the
fibres in the mix. The addition of the Chryso Aquabeton ZA will tend to reduce the workability
of a concrete or cement-based mix. The starting mix should therefore be designed to take
account of this, based on experience from trials. Workability is compensated for, by the use of
correct water-binder ratio and an optimal amount of SP.
2.2.2.4 Binder
The binder consists of three materials, Cement (CEM I 42.5N), Fly Ash (FA) and Ground granu-
lated Corex slag (GGCS). A constant ratio was used throughout the research. Micromechanical
considerations, structural mechanics and economical viability were influential factors concerning
a feasible composition for the systematic study of the ECC material. The composition of binder
constituents used in this research is shown in Table 2.2.
Table 2.2: The composition ratio of binder constituents for ECC.
CEM I 42.5N FA GGCS
0.4 0.5 0.1
Cement (CEM I 42.5N)
CEM I 42.5N is hydraulic cement consisting essentially of hydraulic calcium silicates. Since the
cement is composed of a heterogeneous mixture of several compounds, the hydration process
consists of reactions of the anhydrous compounds with water, occurring simultaneously. As
the hydration reaction of cement compounds is exothermic, the compounds of cement are none-
Heinrich Stander University of Stellenbosch
2.2. PROPERTIES OF ECC 9
equilibrium products of high temperature reactions and are therefore in a high-energy state. The
cultivated heat of hydration could lead to cracks in some applications and affect the structural
strength and durability. The 42.5 in the name corresponds to the strength in MPa achieved at
28 days with a W/B ratio of 0.5. The letter N for normal depends on the strength at 2 days,
in this case equal to or more than 10 MPa.
Fly Ash (FA)
FA is collected, by electrostatic precipitator, from the flues of power stations that burn finely
ground coal. Being a by-product of an industrial process, it has cost advantages and is readily
available in the Gauteng province of South Africa, but not in the coastal regions.
FA is a pozzolanic material which, when mixed with portland cement and water, reacts with
calcium hydroxide (product of cement hydration) to produce C-S-H gel. Other reasons for the
use of this product are that it gives a variety of useful enhancements to the concrete properties.
The spherical shape of the particles causes an increase in mix workability. The relative density
is also less than that of cement, and therefore the substitution on a weight-for-weight basis will
lead to an increased volume. It increases the durability of cement-based composites in the sense
that it increases the density of the material and improves the mechanical behaviour. FA also
limits the heat exerted during hydration because it prolongs the hydration process. The type
of FA implemented in the research was Dura-Pozz, an ungraded FA product provided by Ash
Resources, South Africa.
Ground Granulated Corex Slag (GGCS)
Corex slag is a by-product of the steel manufacturing process at Saldanha steel plant in the
Western Cape region of South Africa. Slag is a latent hydraulic binder in that it hardens very
slowly in water, but becomes much more reactive when activated by alkalinity of calcium hy-
droxide, a product of cement hydration. The introduction of Corex slag as a cement extender
does not only improve the durability and workability of cement-based materials, but also re-
duces the cost of the material.
The RD is also less than that of cement and therefore, the substitution on a weight-for-weight
basis will lead to a volume increase. The particles are finer than cement particles, which results
in better workability, lower bleed rates and shorter setting times than conventional slag [1].
Increased durability is achieved via an increased density of the composite. Slag also limits the
heat exerted during hydration.
2.2.2.5 Fine Aggregate
An important aspect of a cement-based mix is to ensure the uniform spread of different sand
grain sizes. In short we call this spread of grain size the grading of the aggregate. It is important
to ensure that the particle sizes are not all the same and that they yield a dense packing of the
Heinrich Stander University of Stellenbosch
2.3. OVERLAY AND SUBSTRATE BOND PROPERTIES 10
aggregate, which in turn cultivates workability and durability.
A F95 grading is a very fine grading and is used to ensure a good spreading of the sands,
ensuring that the role of aggregate is optimal. The finer grading ensures a lower matrix fracture
toughness, which conforms to micromechanical models of strain-hardening. This suggests that
a matrix with lower toughness (in comparison to concrete) should require a smaller number
of fibres to make the transition from brittle to pseudo-strain-hardening mode of failure. The
process of grading the sand is not economical, but is used to ensure the occurrence of the multiple
cracks phenomena, which yields high ductility. The RD of the sand used in this research is 2.7.
It is a blended sand, mixed from dune sand and crusher dust, graded to F95.
2.3 OVERLAY AND SUBSTRATE BOND PROPERTIES
2.3.1 Introduction
Ageing of concrete structures create the need for their repair. The thin bonded overlay technique
is widely used; it is particularly suitable for the repair of large concrete areas. The primary
purpose of overlays is the extension of the service life of the composite structure by increasing its
thickness, or to provide additional cover for corrosion protection. The durability of such repairs
or overlays depends on the durability of their bond with the base structure; it is usually the
critical aspect. Cement-based overlays are widely implemented either as repair or retrofitting
strategies. In the case of retrofitting, the overlay modifies the original design capacity of the
structure to protect against natural disasters.
Bonded overlays are increasingly used in practice as repair and rehabilitation methods, despite
the high probability of failure incurred by interfacial debonding. Reasons for such failures in-
clude inefficient substrate surface preparations, inappropriate overlay materials, poor curing
conditions and extended time dependent behaviour.
South African codes for the structural use of concrete, have no design specifications for overlays.
A broad description of surface preparation at concrete joints is presented. International stan-
dards and specifications for the design of bonded overlays, other than ordinary concrete, are
also generally considered as deficient, refer to CEB-FIP Model code 1990 [9] and the following
codes [11], [2] and [8]. The greatest problems with creating strong bonds between overlays and
substrates, are little knowledge of what surface preparations to implement and the execution
thereof.
The quality of an interfacial bond between the substrate and overlay plays an intricate role in
the successful performance of the composite. Special attention to capture optimal bond char-
acteristics are of utmost importance. Optimal interfacial bond characteristics are case specific,
in some instances the transfer of stresses is crucial and in others not. The fact of the matter is
that through identifying the need, one can engineer an optimal bond through implementation
Heinrich Stander University of Stellenbosch
2.3. OVERLAY AND SUBSTRATE BOND PROPERTIES 11
of reliable SSPT’s.
The introduction of ECC as an overlay or repair material, does not only address durability
aspects but also structural performance. The associated ductility of the material induces a high
performance aspect wherever it is applied. It is thus crucial to execute reliable design methods,
especially at interfacial level, in order to harness the ductility at hand.
2.3.2 Code based design
The existing specifications are limited to concrete overlays. Design values for interfacial tension
and shear stresses are given in accordance to surface preparation conditions. Specifications
on application methods for surface preparation techniques are restricted to a visible approach.
Overlays constructed with important high performance materials such as high-strength, self
compacting and fibre-reinforced cement-based materials are omitted. New technologies imple-
menting these specialised materials are gaining importance in the field of concrete repair [17].
The necessity of new technologies are driven by higher aspirations and demands of the user.
Based on the consideration that if composite members are designed with adequate interfa-
cial shear strength, they can be modelled following the same procedures as those developed
for monolithic members. Therefore codes define limitations for design shear resistance at the
interface between substrate and overlay.
2.3.2.1 South African national standard
The only relevant composite design specifications found in the South African Standard for the
structural use of concrete, SABS 0100-1 (part 1), are in the subclause: “Composite concrete con-
struction”. This subclause applies to flexural composite elements consisting of precast concrete
units acting in conjunction with added concrete, with provision for the transfer of horizontal
shear stresses at the interfacial zone.
Calculations for the horizontal interfacial shear are governed by the ultimate limit state. Pre-
scribed methods ensure that composite action is retained under serviceability limit states and
that the design shear strength is adequate for the ultimate limit state [38].
The primary prescription for arresting the development of debonding, is obtained when the
horizontal design shear stress is less than the design ultimate horizontal shear stresses at the in-
terface. The design ultimate horizontal shear stress values correspond to substrate surface types
and the grade of overlay concrete. Three surface types are stipulated, it includes no surface
preparation, artificial roughening and cleansing. The design values for these surface types can
be increased by a factor of three, through the incorporation of links that bridge the interface. A
very tough mechanical interlocking is gained with the application of steel links. The interfacial
design ultimate horizontal shear stress values are presented in Table A.1.
Heinrich Stander University of Stellenbosch
2.3. OVERLAY AND SUBSTRATE BOND PROPERTIES 12
An important prescription is to account for relative stiffnesses in the design, when the strength
of the materials differ more than 10 MPa. Detailed specifications given by SABS 0100-1 are
summarised in Appendix A.1.
Extra information concerning the execution of the substrate surface preparation at construction
joints is supplied in the SABS 0100-2 [34] code of practise. This section elaborates on four me-
thods to attain a working bond at a construction joint. The first two methods incorporate the
use of an adhesive and a chemical retarder. The remaining two methods of interest are artificial
roughening of the substrate surface and the use of metal stops. Both these methods integrate
a mechanical interlocking aspect.
The following details are prescribed in the SABS 0100-2, for roughening of the concrete substrate
surface at construction joints. Acceptable, uniform exposure of the aggregate without laitance
or loosened particles on the surface, is necessary. Various options are given in order to dampen
the substrate after it has been roughened, depending on the age of the substrate. The two
dampening options prescribed, differ in duration of application. A short moistening period and
a longer moistening period extending over 24 hours are supplied, which create a difference in
substrate saturation levels. Dampening is followed by the application of a cement grout, which
is left to dry before the fresh concrete is cast onto the old. An important contrasting difference
between the two dampening methods is that the application of the cement grout follows the
shorter dampening period directly, while the 24 hour dampened surface is first left to surface
dry. Both these methods conform to the fact that a suitable substrate surface saturation level
is needed to ensure some sort of capillary suction on the fresh concrete. Detailed specifications
given by SABS 0100-2 are enclosed in Appendix A.1.
2.3.2.2 International standards
Existing international design specifications for concrete repair patches are limited in source
and detail [17]. Updated versions of the codes contain more comprehensive information on the
topic than older versions. It is due to the increasing demand of structural repair in developed
countries. Two standards are more comprehensive than other, the Eurocode 2 [8] and the DIN
1045-1:2001-07 [11]. Details on substrate surface preparations, general rules for the application
of repair materials, as well as limiting values for material strength, shrinkage, thermal coeffi-
cients and tensile bond strength are some of the raised issues. A partial summary of Eurocode
2 follows in Appendix A.2.
The following shear bond strength values for the selected codes, result from a similar defined
roughness and 25 MPa concrete:
• ACI 318-02 [2]: The interfacial design shear resistance is 0.55 MPa for clean surfaces, free
of laitance and artificially roughened to a full amplitude of approximately 6.4 mm. Prior
to casting, the substrate surface must be wetted and standing water should be removed.
Wet blasting is recommended as a substrate dampening method.
Heinrich Stander University of Stellenbosch
2.3. OVERLAY AND SUBSTRATE BOND PROPERTIES 13
• CEB-FIP Model Code 90 [9]: The interfacial design shear resistance (τRd,i) is calculated
in (2.1). The multiplication factor (β) depends on the surface category. The design tensile
strength (fctd) of the weakest concrete part (substrate or overlay) is calculated using the
minimum characteristic tensile strength (fctk,min) divided by the material factor (γm) as
follows: fctk,min/1.5. The formula also accounts for shear resistance attained from normal
confining pressures on the interface. For example, an overlay with a characteristic strength
(fck) of 25 MPa (on a category II roughened, higher strength substrate) with no confining
stress (σcon,i) has a design shear as follows:
τRd,i = (β × fctd) − (µ× σcon,i)
= (0.4 × 1.167) − (0.9 × 0)
= 0.467 MPa
(2.1)
• DIN 1045-1:2001-07 [11]: The interfacial design shear resistance is calculated similarly to
the method obtained from the CEB-FIP Model Code. It takes into account a β factor
which represents interfacial surface roughness. τRd,i is proportional to the fck of the weaker
material. Shear friction due to confining stress over the interface is also considered. For
the same example used above, the design shear strength is:
τRd,i = (0.042 × β × (fck)13 ) − (µ× σcon,i)
= (0.042 × 2.4 × 2.924) − (1.0 × 0)
= 0.295 MPa
(2.2)
• Eurocode 2 [8]: The interfacial design shear resistance is proportional to the design tensile
strength of the weakest concrete. The method in obtaining τRd,i, corresponds to both the
previously mentioned European standards. The coefficients c and µ, relate the degree of
substrate roughness which is calibrated against four types of surfaces. Using the same
concrete as above on a rough surface, yields:
τRd,i = (c× fctd) − (µ× σcon,i)
= (0.45 × 1.167) − (0.7 × 0)
= 0.525 MPa
(2.3)
The interfacial design shear resistance values illustrated by the codes above, fall in a relatively
small range (0.3 - 0.55 MPa). The magnitudes of these values can be considered to be conser-
vative. The essential reason for the conservative design values is the development of long-term
material phenomena, such as differential shrinkage and creep that cause deterioration of the
interfacial bond. Short-term bond strength values are usually substantially higher than the
prescribed values. The bond is sensitive to workmanship, which can cause lower values.
Heinrich Stander University of Stellenbosch
2.3. OVERLAY AND SUBSTRATE BOND PROPERTIES 14
2.3.3 Interfacial bond characteristics
2.3.3.1 Introduction
The performance of composite members can mainly be attributed to interfacial resistance
against cracking and debonding. Interfacial bond properties are primarily related to mate-
rial composition, interface textures and substrate moisture conditions, curing procedures and
environmental influences. Large volumes of theoretical data and practical experience exist in
the field of concrete repair, but overlay debonding is still a major concern in practice, mostly
due to the lack of precise and concise methods for specific applications. A lack of knowledge and
experience exist in the field of overlay execution with high performance materials. It is believed
that knowledge concerning concrete-to-concrete repair can be translated and implemented for
ECC-to-concrete repair. The deformable nature of ECC tends to support ideas of harnessing
experience gained on interfacial bonds from concrete overlays.
Little data exists on specialised applications involving ECC as overlay material. With the in-
troduction of this ductile material, the strain at the interface increases due to the heightened
demand in performance of the newly defined composite. Extreme applications such as retrofitted
members subjected to seismic loads must endure extreme deformations. Load transfer between
materials across the interface is of utmost importance. Implementing the correct interfacial
bond properties for structural analysis of design are crucial for the success of the retrofitted
members. Existing specifications are not enough, thus the reason for proper interfacial bond
characterisation that supports adequate application methods.
Defining the processes influencing the chemical bond at the interface is conceived by analogy to
existing knowledge on the interfacial transition zone (ITZ) between aggregate and the cement
matrix [42]. The analogy is made to indicate the importance of moisture transport at the inter-
face and the effect a porous state can have on the moisture transport. The substrate consists
primarily of fine and coarse aggregate, which further supports the use of the analogy. In the
following analogy, reference to aggregate and its properties must be interpreted correspondingly
for the substrate.
Van Mier [42] states that the bond mechanism between aggregate and cement paste depends
largely on the porosity of the aggregate. Figure 2.3 depicts the influence of aggregate porosity
on the bond mechanism. Generally a thin layer of CH forms at the physical boundary between
the two materials. Next is a porous layer of CSH , CH crystals and ettringite, which together
form a transition layer of high porosity. The effective W/B ratio increases as a result of physical
absorbtion of water by the aggregate surface during the mixing process. The increased W/B
ratio causes the increase in porosity and along with this, a wall-effect is present. The wall-effect
influences the fibre orientation, thus the casting and testing direction is of importance.
Heinrich Stander University of Stellenbosch
2.3. OVERLAY AND SUBSTRATE BOND PROPERTIES 15
ITZ
bulk cement paste
aggregate particle
(a) (b)
Figure 2.3: Differences in ITZ for concrete containing aggregates with (a) a dense outer shell and (b)a porous outer surface.
Research has shown that fracture surfaces generally exist not directly at the physical bound-
ary between the aggregate and matrix. Fracture normally occurs slightly removed from the
interface, in the porous transition zone. Experimental work on concrete-concrete composites
conducted at the University of Cape Town (UCT) has revealed that failure occurs predomi-
nantly in the overlay, close to the interface [4].
Whether the substrate mostly exposes aggregate or not, the idea behind mechanical alteration
is to increase the“porosity”, specific surface area of the substrate, on a macro and micro level.
The increased “porosity” of the substrate at a micro scale increases capillary suction which
results in water being drawn from the overlay, thus increasing the effective W/B ratio in the
contact area. Suction occurs in a heightened state for a substrate/ovelay, when compared to
what happens in the ITZ.
Determination of interface properties is not straightforward. As has become clear, the interface
structure is very small and the behaviour seems very much affected by moisture transport in
the cement matrix itself, but also by moisture transport between the substrate and overlay.
The following sections analyse the fundamental bond mechanisms, material and environmental
influences on bond properties.
2.3.3.2 Definition and classification
Adhesion is the coupling force at the interface between two materials. It consists of two different
contributing mechanisms mechanical interlocking and chemical bonding [16].
Mechanical interlocking is achieved on a macroscopical and microscopical level. The macro-
scopical level represents physical roughness and indentations in the substrate, measured on a
scale of 10−3 m and larger. This form of bond between two different materials, is achieved
through physical processes applied to the exposed surface of the substrate material. Physical
roughness and indentations are produced at different stages of material processing and through
a number of different application processes. In the fresh state of the cement-based substrate,
artificial roughness can be produced during extrusion with specialised forms. Sandblasting and
drilling are examples of methods used to induce artificial roughness in the consolidated state.
Heinrich Stander University of Stellenbosch
2.3. OVERLAY AND SUBSTRATE BOND PROPERTIES 16
Application methods relate to the time at which it is needed in the design of such a structural
member, whether implemented from the start or as part of a repair strategy.
Chemical bonding at the interface is regarded as the combination of two processes. The first
process refers to Van der Waals-type forces between the hydrate fibres of both materials. Van
der Waals forces are obtained through exposing unhydrated cement particles of the substrate to
the fresh overlay material. Hydration across the interface produces chemical bonding through
Van der Waals type forces. The toughness of such a chemical bond correlates to the quantity
of unhydrated cement particles exposed by roughening, on the substrate surface. Moistening
of the roughened substrate and the duration thereof has an effect on the bond strength, due to
the hydration process.
The second process governing chemical bonding is defined by mechanical interlocking at a mi-
croscopical level. Such physical anchorage is achieved through the hardening of overlay material
in micro-cavities of the roughened substrate surface. In this thesis, this type of bond is regarded
as chemical. The degree of substrate surface saturation relates to the magnitude of capillary
absorption, i.e. the suction of overlay paste into micro cavities. The moisture condition of the
substrate surface plays an important role in achieving mechanical interlocking at a micro-scale.
Micromechanical adhesion values in tension and shear primarily differ because of the angle
at which the physical anchorage is tested. When the interfacial zone is subjected to uniax-
ial tension, these physical anchors pull out. In the case where the interface is subjected to
shear, these physical anchors need to be fractured in order to overcome the bond. The result
is a higher bond strength in shear, compared to tension. Figure 2.4 illustrates this phenomenon.
Figure 2.4: The influence of mechanical interlocking on interfacial shear and tension, at micro-scale.
The bond between two materials composing a composite, depends on the contact surface area
at the interface. The interfacial surface is conceptually divided into three types [4]:
1. The specific surface area depicts the total surface area that incorporates everything down
to micro level. It is primarily influenced by the roughness of the substrate surface.
2. The effective surface area portrays the actual interfacial contact area between the two
Heinrich Stander University of Stellenbosch
2.4. PARAMETER TEST METHODS 17
materials. Workability, compaction and thermodynamic properties of the overlay in its
fresh state are the main influences of the consequential contact surface area.
3. The geometrical surface area is used in the quantification of the ultimate interfacial bond
strength and represented by a plane through the interface.
specific surface effective surface geometrical surface
overlay
substrate
Figure 2.5: The different interfacial surface areas.
Research thus far has hardly focussed on characterising chemical bonding at the interface,
because investigation thereof is complicated by the scale at which it must be conducted. Exper-
imental testing of the interfacial bond strength result in conclusions that are phenomenological.
The subsequent result is the indirect relation of macroscopic phenomena to microscopic events.
2.4 PARAMETER TEST METHODS
2.4.1 Introduction
In recent years, the importance of the fracture mechanical characterisation of cement-based com-
posites has been recognised by researchers and the industry. It was found that parameters such
as the compressive strength and the elastic modulus do not adequately describe the behaviour
of cement-based materials under different loading conditions. Several fracture test methods
have been proposed to international institutions like RILEM (The International Union of Test-
ing and Research Laboratories for Materials and Structures) and ASTM (American Society for
Testing and Materials). However, there is no general agreement within the fracture mechanics
community about the parameters, suitable to describe and characterise the fracture behaviour
of cement-based materials. Important problems facing the field of concrete engineering and
design, can be solved by applying fracture mechanics [30]. This can only be accomplished if one
or more standard fracture test method, in conjunction with an accepted procedure to calculate
fracture parameters, are in place.
Test Methods for ascertaining interfacial parameter values of composite materials are not stan-
dardised in local and international codes as of yet. Many variations of shear and tensile test
methods exist. Results produced by the different test methods vary accordingly. Different
parameters such as specimen size, loading rate, test set-up, etc. culminate to this variation.
Careful consideration needs to be undertaken to find the best suited test method for the specific
materials at hand. The following subsections illustrate an array of different test methods for
Heinrich Stander University of Stellenbosch
2.4. PARAMETER TEST METHODS 18
both shear and tensile values. As stated earlier, the bond mechanism at the interface results
in different fracture mechanisms for mode I and II failures. It is important to implement the
best possible test methods to characterise tensile and shear behaviour. Observations can be
converted to relations between the two phenomena for each of the tested SSPT’s.
Stress fields in general structural applications are complicated, making it essential for the dif-
ferent mechanisms to be identified, objectively modelled and their parameters characterised,
to enable accurate prediction and design. The tests have the above objective characterising
function but also serve for quality assessment.
2.4.2 Test methods for shear parameter values
Determination of ultimate strength values for pure shear loading, is a controversial issue regard-
ing theory and experimental implementation thereof. Pure shear, i.e. shear testing of materials
which implicate loading that generates only shear stresses, without normal stresses of any kind.
Theoretically this must account for both the elastic and inelastic ranges until total fracture oc-
curs. Subsequently, the acquisition of a pure shear state along the interface requires conditions
expressed in (2.4):
τxy 6= 0, σxx = 0, σyy = 0, σzz = 0 (2.4)
Composite beams and slabs are designed to withstand loads, functioning as a monolithic mem-
ber in which case interfacial shear plays a crucial role. A number of important aspects need to be
taken into account when designing a shear parameter test method for such a composite. Firstly
it is important to achieve little or no bending moment over the interface. When compression
and tensile stresses accompany shear, pure shear values are distorted and in extreme cases, the
mode of fracture can differ. Secondly it is also crucial to capture post-crack behaviour which
depicts the amount of interfacial fracture energy (Gf,i). This means that such a test needs to
be controlled throughout the duration of the test procedure, especially at the point of fracture
and from then onwards until total fracture.
Numerous test methods have been proposed and used for a variety of materials [12]. Figure
2.6 shows eight specimen geometries and loading configurations. An ideal situation of pure
shear stresses along an interface, which is conceptualised for testing but cannot be realised, is
indicated by Figure 2.6 (a).
The axisymmetrical punch-through specimen (Figure 2.6 (b) [29]) has been used on mortar and
concrete due to ease of use. Large tensile stresses occur at the crack tip. A recent study has
shown that the tensile stresses can be reduced by choosing four notches instead of two and
varying its depth [10]. A practical test method (Figure 2.6 (c)) correlating with the previous
two, conducted on cylindrical core samples with four notches, yields mixed mode results [25].
Another test procedure without any notches, correlating to that of Figure 2.6 (b), is proposed
Heinrich Stander University of Stellenbosch
2.4. PARAMETER TEST METHODS 19
in [14]. All these test methods, have two interfaces which cause impractical casting obligations.
The Ohno shear beam [3] configuration is illustrated in Figure 2.6 (d). A state of pure shear
exists, but only at the centroid of the specimen. The measured shear strength is a value close
to that of its pure shear capacity [13]. A modified version of this test shown in Figure 2.6, was
developed by Iosipescu [20]. The specimen is weakened by means of angular notches at the
interface and results in the most correct test method investigated. The reduction of interfacial
area reduces the probability of bending failure and leads to uniform τxy in the notch section.
Large enough specimens are necessary to create a proper, representative interfacial size, thus
posing size issues for the testing apparatus.
Figure 2.6 (f) illustrates a push-off specimen, proposed by Hawkins and Mattock [26] and mod-
ified later by Nooru-Mohamed [27] as shown in Figure 2.6 (g). Compressive stresses exist at
the top and bottom of the interface, increasing the shear strength in these zones.
Degradation of the interfacial bond can be observed nondestructively by measuring the contact
electrical resistance of the joint under cyclic loading [7]. This method only indicates a decrease
in bond strength. The set-up is illustrated in Figure 2.6 (h). Implementation of this method
to investigate the composite shear characteristics, will result in an overestimation of post-crack
resistance, due to the confining shape of the surrounding material.
2.4.3 Test methods for tensile parameter values
The occurrence of tensile forces perpendicular to the interface is a rare phenomenon in structural
design. However, in extreme cases such as seismic loading, tension normal to the interface is
possible, which may lead to debonding. Restraint to differential shrinkage also leads to normal
stresses perpendicular to the bonded interface in composites due to curling. Some researchers
even believe that debonding is always initiated through tensile stresses at boundaries [17]. Char-
acterising the tensile parameter values is of great importance for quantification purposes with
computational modelling methods.
The three most common tensile test methods are shown in Figure 2.7. The first and most
common test is the direct uniaxial tensile test of which the geometrical aspects are flexible, but
a large enough cross-section is required in order to be a representative sample of concrete. A
couple of difficulties are associated with the direct tensile test, the first being a particularly
sensitivity to eccentric loading. The fact that the concrete is brittle, results in grip difficulty.
Eccentric loading and failure at or in the grips, are difficult to diminish. Most direct tensile
methods make use of adhesive to grip the specimen to metal platens. Introducing an adhesive
to the testing process, elongates the test preparation time but lessons the impact of geometri-
cal inconsistencies at the adhered surfaces. The fabrication of such testing systems is relative
uncomplicated and fast. Special specimen geometries are not needed and straight forward rect-
angular specimens can be used, depending on the adhesive quality. Other methods utilise clamps
Heinrich Stander University of Stellenbosch
2.4. PARAMETER TEST METHODS 20
(a) Ideal (b) Punch-through (c) Modified punch-through
(d) Ohno (e) Iosipescu
(f) Push-off (g) Modified push-off (h) Cyclic punch-through
Figure 2.6: A schematic illustration of various test methods to determine τxy at the interface of twodifferent cementitious materials.
Heinrich Stander University of Stellenbosch
2.4. PARAMETER TEST METHODS 21
to anchor specimens. In such cases the specimen geometry is modified to resemble a dog bone
like shape, which fits into complex gripping systems. The design of such systems is elaborate and
the accompanying fabrication thereof is time consuming. On the other hand it induces produc-
tive test programmes because the eventual test preparation is easier and consistent in execution.
Two attributes are crucial for the Compact-Tension method or the so-called Wedge Splitting
Test method [5] [40] which is illustrated in Figure 2.7 (b). A sufficiently large fracture sur-
face area is necessary to obtain size-independent fracture values and stable crack propagation,
until complete separation of the specimen is obtained [30]. Fulfilling both requirements simul-
taneously is very difficult, because relatively high testing loads are necessary to fracture large
specimens, which often leads to unstable crack propagation. This is an indirect method for
ascertaining interfacial tensile strength values. More information on this method is available in
[30].
The third tensile test method is based on the concrete cylinder splitting test and is shown
in Figure 2.7 (c). It is an indirect method preferred for routine purposes and performed in a
normal compression testing machine. The theoretical distribution of horizontal stresses over the
vertical interfacial plane is close to uniform tension, with local compression peaks at the contact
points. The cylinder splitting strength (fs) is defined as the magnitude of near-uniform tensile
stress on the interfacial plane. The state of stress in the specimen is biaxial and together with
the high compression zones, result in fs being higher than interfacial tensile strength (ft,i).
(a) (b) (c)
Figure 2.7: (a) Direct uniaxial tensile test, (b) Wedge splitting test and (c) Cylinder splitting test.
Heinrich Stander University of Stellenbosch
Chapter 3
EXPERIMENTAL DESIGN
3.1 INTRODUCTION
In order to investigate interfacial bond characteristics, suitable test methods are necessary to
obtain both the tensile and shear capacities of the interfacial bond. A number of different pa-
rameter test methods were studied, as noted in section 2.4.
For the selection of a particular test method, various factors were considered. These factors are
listed below:
• Available testing equipment and measuring instruments,
• Time constraints, restricting the duration of the parameter study,
• Accuracy of method and centricity of loading,
• Ease of use and versatility of method,
• Displacement control from the initial crack until total fracture,
• Test application on the correct casting direction of the overlay,
• Application at the specified testing age of the overlay.
The issues raised above, entailed shear and tensile interfacial test methods with a versatile and
simplified set-up, incorporating large enough specimen sizes with the necessary room for instru-
ment connections.
In the end it was decided to implement the shear-test method proposed by Hawkins and Mattock
[26]. For the tension behaviour, a direct uniaxial tensile test with flexible geometrical aspects
was chosen. Effort was placed into refining the shear test method, in order to secure the best
possible uniform shear stress-field. The geometrical flexibility of the tensile test resulted in a
simple, rectangular shape that fitted into the shear test set-up. The set-up was developed to
suite both test methods with the highest possible accuracy. This way time was saved during
construction of the set-up and execution of tests.
22
3.2. COMPUTATIONAL MATERIAL MODELS 23
The following sections will present details on tools used to develop the parameter test methods
and also the process instituted to achieve the final geometrical aspects thereof.
3.2 COMPUTATIONAL MATERIAL MODELS
3.2.1 Introduction
The ever increasing complexity of structures drives the need for new building materials to ad-
dress these structural demands. Code based design of structures is becoming inadequate in
supplying the necessary design equations, especially for new materials. Incorporating design
methods into existing international building standards and codes is a long process. With accu-
rate material models it becomes possible to test structural elements without physically testing.
Accurate material models shorten the execution periods of experimental research, that are
conducted to ascertain knowledge about material characteristics and structural performance.
Eventually it leads to the inclusion of new material design aspects into design codes at earlier
stages.
Concrete constitutive models have been used in finite element analyses for the past 40 years.
Improvement in modelling the mechanical response and fracture has occurred and is an ongoing
process. Whether or not it is a feasible design tool, is not the purpose of this research, but as
a numerical tool its contribution in product development and characterisation is substantial.
In this project computational modelling was conducted with DIANA, which is a general purpose
finite element code, based on the displacement method. DIANA offers several material models
for structural analysis. A user-supplied model can be used when the supplied material models
are inadequate. The following sections provide background theory and input details on two
crucial material models utilised in this research program.
3.2.2 ECC model
An user-supplied constitutive model [6] was implemented to simulate the nonlinear behaviour of
ECC. It is a plane stress model, developed by Boshoff (2006) at the University of Stellenbosch.
This model approaches the numerical modelling of the mechanical behaviour and fracture of
ECC as a continuum material with continuum damage mechanics. It is implemented with the
smeared cracking approach, i.e. the inelastic strains associated with the crack openings are con-
sidered to be smeared/averaged over a finite element and added to the total strain of a material
point. Another option exists in discrete crack modelling whereby interfacial elements act as
crack openings and need to be placed in appropriate positions. The latter type of modelling is
inadequate for ductile materials such as ECC, because crack positions and paths are not known
beforehand.
A damage mechanics approach model is employed in this smeared cracking model. It implements
a damage approach, which results in a material that exhibits full unloading, i.e. no residual
Heinrich Stander University of Stellenbosch
3.2. COMPUTATIONAL MATERIAL MODELS 24
loading
unloading
σ
ε
Figure 3.1: The unloading behaviour of the damage formulation.
strains when the applied force is unloaded, as indicated by Figure 3.1. The result is that all
crack openings close when the material is unloaded. It is crucial to understand the implications
of this damage approach when inspecting computational results and behaviour.
A basic relation (3.1) is used for the elastic computational model. The elastic stiffness matrix
( ¯De) relates strain to stress, the mathematical relations of these concepts are illustrated by
equations (3.2), (3.3) and (3.4). These equations indicate that this model was developed for
two dimensional, plane stress analysis. Further information regarding the workings of this
material model is enclosed in [6].
σ = ¯Deε (3.1)
σ = {σxx, σyy, τxy}T (3.2)
ε = {εxx, εyy, γxy}T (3.3)
¯De =E
1 − ν2
1 ν 0
ν 1 0
0 0 (1 − ν)/2
(3.4)
The damage approach is based on the reduction in material stiffness when damage occurs. The
reason for illustrating these basic material relations, is to convey how and where the damage
approach fits into the mathematical model and what experimental values are used to predict
global material behaviour. A damage indicator scalar, ω, ranging from zero to one, refers to
zero and total damage and is applied mathematically as shown in (3.5).
σ = (1 − ω) ¯Deε (3.5)
Modelling of ECC behaviour is based on a one-dimensional tensile response, i.e. a scalar which
represents the strain vector of any material point. For evaluation whether damage evolution
occurs, this scalar called equivalent strain, is required. The implication stemming from the
one-dimensional representation of the strain vector, is the use of uniaxial tensile test data as
Heinrich Stander University of Stellenbosch
3.2. COMPUTATIONAL MATERIAL MODELS 25
characterisation of the damage evolution. This is achieved by linearisation of the uniaxial stress-
strain curve of ECC, as indicated in Figure 3.2. The respective parameters are used as input
values for the material model. The required input parameters are summarised in Table 3.1 and
the set of values correspond to values retrieved from work done by E.C. Olsen [28]. These values
were used for the ECC material model in initial numerical studies, discussed in section 3.3. The
remaining parameters are self-explanatory and defined by Figure 3.2, except Ls which is the
theoretical pull-out length of fibres at the fracture point, equal to half the fibre length.
σtu
σfc
εsh
εtu
εfc
ε
σ
Figure 3.2: The uniaxial stress-strain response of ECC, represented by a tri-linear curve.
Table 3.1: Initial parameter values for the ECC material model.
σfc [MPa] σtu [MPa] εsh [mm/mm] Ls [mm] E [MPa] ν
2.23 3.3 0.045 6 9200 0.2
3.2.3 Composite interface model
Another essential material model supplied by DIANA, the composite interface model, was
utilised to simulate the interfacial behaviour of the composite. This plane stress model was
formulated by Lourenco and Rots [24] and enhanced by Van Zijl [43]. It is a discrete model that
simulates fracture, frictional slip as well as crushing within a material, or between two separate
continuum materials with interfacial elements.
Based on multi-surface plasticity, it comprises a Coulomb friction model combined with a ten-
sion cut-off and an elliptical compression cap (Figure 3.3). Softening occurs in all three modes
and is preceded by hardening in the case of the cap mode.
Heinrich Stander University of Stellenbosch
3.2. COMPUTATIONAL MATERIAL MODELS 26
intermediateyield surface
initial yield surface
residual yield surface
tension mode
Coulombfriction
modecap mode
σ
φ
τ
Figure 3.3: Two-dimensional interface model.
Parameter values needed for the model are illustrated in Table 3.2. The definitions of the
interfacial parameters are as follows:
• GAPVAL is the interfacial tensile strength (ft,i) which is ≤ c/tanφ
• MO1VAL is the fracture energy (GIf ) for Mode-I
• FRCVAL describes the friction criterion which consist of cohesion (c), i.e. the unconfined
shear strength, the friction coefficient (Φ), i.e. the tangent modulus of the friction angle
φ (Φ = tanφ) and dilatancy coefficient (Ψ) (Ψ = tanψ).
Non-consistent friction and dilatancy require three more parameters: the residual friction
coefficient (Φr), the confining normal stress (σcon,i0) for which the normal uplift is 0, and
the exponential degradation coefficient δ of the dilatancy coefficient with shear-slipping
displacement. Φ and Ψ are used as initial values for this case.
• MO2VAL defines the shear fracture energy (GIIf ) for which a and b are parameters in
GIIf = (a × σ) + b. This equation determines the linear relation between fracture energy
and normal confining stress, a constant GIIf for σ 6= 0 will be achieved when taking a as
zero.
• CAPVAL conveys the cap criterion, in which fc and Cs respectively represent compressive
strength and shear traction contribution to the compressive failure.
• MOCVAL describes the compressive inelastic law with Gfc, compressive fracture energy,
and κp which is the equivalent plastic relative displacement corresponding to the peak
compressive stress.
Heinrich Stander University of Stellenbosch
3.3. NUMERICAL REFINEMENT OF SHEAR TEST METHOD 27
Table 3.2: Parameter variables for the composite interface material model.
GAPVAL MO1VAL FRCVAL MO2VAL CAPVAL MOCVAL
ft,i GIf c a fc Gfc
Φ b Cs κp
Ψ
Φr
σcon,i0
δ
Derivation of the interface model is in terms of generalised stress and strain vectors
σ = {σxx, τxy}T
ε = {u, υ}T(3.6)
with σ and u representing stress and relative displacement in the normal direction of the in-
terface. τ and υ are the shear stress and shear-slip displacement parallel with the interfacial
direction. The constitutive behaviour is described by equation (3.1) on page 24. A full mathe-
matical description of this material model is shown in both [24] and [43]. The following extract
conveys the mathematical relations of plastic normal uplift, or dilatancy upon shear-slipping:
up =
0 if σ < σcon,i0
Ψ0δ
(1 − σσcon,i0
)(1 − e−δυp) if σcon,i0 ≤ σ < 0
Ψ0δ
(1 − e−δυp) if σ ≥ 0
(3.7)
The derivation of (3.7) results in (3.8).
Ψ =
0 if σ < σcon,i0
Ψ0(1 − σσcon,i0
)e−δυp if σcon,i0 ≤ σ < 0
Ψ0e−δυp if σ ≥ 0
(3.8)
Several parameters are required for the interface model, as indicated in Table 3.2. Various tests
must be performed to obtain these parameters. However, not all parameters can be measured
directly and require inverse analysis to obtain best estimates. In this thesis unconfined shear
tests are performed, ruling out accurate, direct determination of the friction parameters (Φ, Φr
and a). The three dilatancy parameters (Φ0, σcon,i0 and γ) are determined by inverse analysis
of shear tests, during which normal uplift perpendicular to the interface was measured.
3.3 NUMERICAL REFINEMENT OF SHEAR TEST METHOD
Finite element modelling (FEM) of the test method was a necessary step towards rendering per-
spective about the set-up. It made allowance for informative decisions on specimen geometry
Heinrich Stander University of Stellenbosch
3.3. NUMERICAL REFINEMENT OF SHEAR TEST METHOD 28
and the necessary boundary conditions (BC). Important aspects considered included the stress
distribution over the interface, not only tangential but also perpendicular. Stresses normal to
the interface, might either increase or decrease the effective interfacial shear stress and had
to be minimised as best possible. Geometrical aspects were not only influenced by material
strain capacities but also by the necessity to include enough space for displacement measuring
instruments.
A two-dimensional plane stress model was compiled of the selected shear test with four node
square elements of five millimetres side length, resulting in 20 elements along a 100 mm long
shear interface (Figure 3.4). The model made use of four different material models, each re-
presenting the four different parts of the model, namely concrete, ECC, steel platens and the
interface. A linear elastic material model was used for the steel platens. All the other material
models captured nonlinear behaviour of the respective materials. Nonlinear analyses were per-
formed on the models in order to consider nonlinearities, reveal loading effects on the respective
material strain capacities and the influence they may have on objective parameter characteri-
sation.
Concrete’s nonlinear behaviour was simulated using the Total Strain Rotating crack model. Ta-
ble 3.3 summarises the different parameters needed for this model as well as the corresponding
values or curve types used for the analysis.
Table 3.3: Parameter values for the concrete material model
E-modulus [MPa] ν Tensile curve ft,c [MPa] Compression curve Shear curve
30000 0.2 Brittle 3 Elastic Uniformly distributed
Boshoff’s [6] constitutive material model for ECC was applied with the parameter values as
indicated in Table 3.1 on page 25. The composite interface model, described in subsection
3.2.3 was used for the ECC/concrete interface. Initial values for the parameters were estimated
from the research conducted by E.C. Olsen [28] (Stanford University, U.S.A.), during his two
month (July-August 2005) visit to the University of Stellenbosch. The initial values used for
the numerical refinement of the shear test method, which are stipulated in Table 3.4, were
obtained by inverse analysis of three-point bending tests by Olsen. The tests were of a similar
nature to what is documented in the penultimate chapter. He modelled the experimental tests
in DIANA, using typical bond values of masonry units with mortar joints. Subsequently, during
this trial and error method, the values of the required parameters for the interfacial material
model were changed to improve agreement between computed and measured global behaviour
of the three-point bending composite beams. However, no directly measured parameter values
were obtained through his research and no claim to uniqueness of the set of parameter values
he finally proposed, was or could be made. Nevertheless, for lack of other information, these
values were implemented for the refinement of the shear test.
Heinrich Stander University of Stellenbosch
3.3. NUMERICAL REFINEMENT OF SHEAR TEST METHOD 29
Table 3.4: Interfacial parameter values provided by research conducted by E.C. Olsen [28]
SSPT ft,i [MPa] GIf [N/mm] τxy = c [MPa]
Control 0.25 0.018 0.225
Sandblasted 0.25 0.018 1.35
Drill Holes 0.35 0.18 1.0
An initial round of analyses consisting a trial and error method were conducted. It entailed a
change in geometry at each iteration in order to achieve the best possible stress distribution
over the interface. A specimen geometry illustrated in Figure 3.4 was produced. After the
geometry was finalised, the top and bottom BC’s were investigated. It had to be established
whether a free rotating, or a non-rotating/fixed platen should be used in the push-off shear
test. The purpose of this numerical investigation was to perceive the shear stress distribution
over the interface for three different BC’s at four different test stages. The three different BC’s
examined were:
1. A fixed-fixed (F-F) set-up, i.e. the bottom as well as top platen are rotationally fixed,
2. A fixed-rotational (F-R) degree of freedom (DOF) set-up,
3. A rotational-rotational (R-R) DOF set-up.
Unlike Figure 2.6 (f), the load was applied over the full top and bottom surfaces.
Heinrich Stander University of Stellenbosch
3.3. NUMERICAL REFINEMENT OF SHEAR TEST METHOD 30
ECC
concrete
Figure 3.4: The geometry of the interfacial shear parameter specimen (dimensions in mm).
Numerical analysis with data from Table 3.4 [28], indicated that the sandblasted interfacial
parameter values result in the highest load capacity. The highest load capacity indicates the
highest state of stress in the overall system and thus the reason for using the sandblasted pa-
rameter values estimated from Olsen’s work in the analyses of the boundary conditions. The
interfacial stress distributions at elastic, peak, post-peak and advanced nonlinear stages were
obtained. The stages were represented by 20%, 100%, 50% and 20% of the ultimate load, re-
spectively. The percentage values used, are arbitrary. Two post peak points are illustrated
because of the difference in normal uplift and energy state in the post-fracture stage.
Figure 3.5 depicts the stress-displacement curves for the three different BC’s. The average shear
stress, obtained by dividing the applied force by the geometrical interfacial shear area, is plotted
on the vertical axis. Vertical displacement of the top node is shown on the horizontal axis and
it represents the displacement of the load application point. This way it incorporates the total
elastic deformation of the whole specimen. From this graph the differences in elastic stiffness,
strength and softening curve are apparent for the different BC’s. No definite explanations for
these differences can be concluded, implying the need to investigate specimen deformation as
well as stress and strain patterns.
Heinrich Stander University of Stellenbosch
3.3. NUMERICAL REFINEMENT OF SHEAR TEST METHOD 31
0 0.2 0.4 0.6 0.8 10
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Vertical Displacement [mm]
τ xy [
MP
a]
Fixed−fixedRotational−fixedRotational−Rotational
Figure 3.5: Shear stress-displacement curves for the different boundary conditions.
In order to make an objective conclusion about the effects of the BC’s, comparisons of the
interfacial stress distributions are needed at all four stages. An acceptable interfacial stress
distribution is not the only requirement for a proper test set-up. Adequate global behaviour
is a requirement as well. Focussing away from the interfacial zone and considering the whole
specimen and its global behaviour is the next step of inspection. By inspection of global per-
formance, secondary influences caused by the BC’s can be studied.
The analysis results are firstly presented in terms of contour plots of shear stresses and secondly
principle strains. Figure 3.6 illustrates a comparison of the contour plots for the three different
BC’s. The three columns represent the different BC’s with shear stress contours from Figure
3.6 (a) to (l) and principle strain contours from (m) to (o). Stress contours for all the cases
are plotted in the same envelope of 0 to 2.5 MPa, to establish objective correlations. The first
four rows compare the deformation and stress distributions at the four stages of the stress-
strain relationship. The last row compares the highest respective strain distributions. All the
animations were plotted with exaggerated deformations, the scale of which is indicated at the
bottom right-hand corner of each graph. This feature depicts global deformation trends and
strain localisation.
Figure 3.6 (d) indicates a higher shear distribution for the F-F BC, which correlates to the
higher load capacity indicated in Figure 3.5 on page 31. Two different scenarios affecting the
interfacial stiffness are present here, both a result of the applicable BC, as depicted by the scaled
deformations. The first scenario is explained by the F-F BC, by which the platens constrain the
Heinrich Stander University of Stellenbosch
3.3. NUMERICAL REFINEMENT OF SHEAR TEST METHOD 32
(a)
Fixed-fixed Fixed-rotational Rotational-rotational
100x
(b)
100x
(c)
100x
(d)
100x
(e)
100x
(f)
100x
(g)
100x
(h)
100x
(i)
10x
Heinrich Stander University of Stellenbosch
3.3. NUMERICAL REFINEMENT OF SHEAR TEST METHOD 33
(j)
50x
(k)
50x
(l)
5x
(m)
100x
(n)
100x
(o)
10x
Figure 3.6: A comparison of contour plots for the three different boundary conditions is depicted byeach column. Shear stress contours are illustrated from (a) to (l) and principle strain contours from (m)to (o). The first four rows compare the deformation and stress distributions at the four stages of thestress-strain relationship. The last row compares the highest respective strain distributions.
specimen from rotation and interfacial dilatancy, adding to the stiffness. The second scenario
is explained by both rotational set-ups, which allows interfacial rotation. The rotational free-
dom lessens the compressive normal forces over the interface and increases the tensile normal
forces. Lesser confining forces, lower the interfacial stiffness and strength. In extreme cases, the
presence of tensile forces may lead to mode I initiated shear fracture. Neither scenario can be
conclusively deducted from by Figure 3.6 nor 3.5.
The strain contours in Figures 3.6 (m), (n) and (o) show evidence that is conclusive in illustrat-
ing the development of a concrete crack in the latter two. For the F-R BC, the crack initiates at
an earlier stage, depicted by the gradient deflection in Figure 3.5, it is not detrimental and in-
Heinrich Stander University of Stellenbosch
3.3. NUMERICAL REFINEMENT OF SHEAR TEST METHOD 34
terfacial shear failure still occurs. The crack in the R-R BC set-up initiates at a later stage, but
continues and causes specimen failure. It is well substantiated by the sudden failure in Figure
3.5, which is a result of the brittle failure curve assigned to the Total Strain Rotating material
model. Together it indicates that F-R and R-R BC’s are not well suited for the geometry at
hand, without steel reinforcement.
Further investigation of the stress distributions at interfacial level is necessary and conducted in
Figure 3.7. A comparison of the σxx and τxy distributions over the interface for each boundary
condition are indicated respectively by each column. The rows represent the four stages of the
stress-displacement response in sequential formation.
Figure 3.7 (a) and (b) represent the elastic regime. Similar stress distributions are conveyed
for the different BC’s. The shear distribution in (b) is not uniform, except for the middle part.
The heightened shear stress at both ends is caused by the localised confining normal pressure
as depicted in (a). The fact that little or no normal stress acts on most of the interface in (a),
is an important observation which promotes conditions for pure shear.
Heinrich Stander University of Stellenbosch
3.3. NUMERICAL REFINEMENT OF SHEAR TEST METHOD 35
−15 −10 −5 00
10
20
30
40
50
60
70
80
90
100
σxx
[MPa]
Ver
tical
Pos
ition
alo
ng In
terf
ace
[mm
]
Fixed−fixedFixed−rotationalRotational−rotational
(a)
0 0.5 1 1.5 20
10
20
30
40
50
60
70
80
90
100
τxy
[MPa]
Ver
tical
Pos
ition
alo
ng In
terf
ace
[mm
]
Fixed−fixedFixed−rotationalRotational−rotational
(b)
−15 −10 −5 00
10
20
30
40
50
60
70
80
90
100
σxx
[MPa]
Ver
tical
Pos
ition
alo
ng In
terf
ace
[mm
]
Fixed−fixedFixed−rotationalRotational−rotational
(c)
0 0.5 1 1.5 20
10
20
30
40
50
60
70
80
90
100
τxy
[MPa]
Ver
tical
Pos
ition
alo
ng In
terf
ace
[mm
]
Fixed−fixedFixed−rotationalRotational−rotational
(d)
−15 −10 −5 00
10
20
30
40
50
60
70
80
90
100
σxx
[MPa]
Ver
tical
Pos
ition
alo
ng In
terf
ace
[mm
]
Fixed−fixedFixed−rotational
(e)
0 0.5 1 1.5 20
10
20
30
40
50
60
70
80
90
100
τxy
[MPa]
Ver
tical
Pos
ition
alo
ng In
terf
ace
[mm
]
Fixed−fixedFixed−rotational
(f)
−15 −10 −5 00
10
20
30
40
50
60
70
80
90
100
σxx
[MPa]
Ver
tical
Pos
ition
alo
ng In
terf
ace
[mm
]
Fixed−fixedFixed−rotational
(g)
0 0.5 1 1.5 20
10
20
30
40
50
60
70
80
90
100
τxy
[MPa]
Ver
tical
Pos
ition
alo
ng In
terf
ace
[mm
]
Fixed−fixedFixed−rotational
(h)
Figure 3.7: A comparison of the σxx and τxy distributions over the interface for each boundary condition,indicated respectively by the each column. The rows represent the four stages of the stress-displacementcurve in sequential form.
Heinrich Stander University of Stellenbosch
3.4. EXPERIMENTAL SET-UP 36
Figure 3.7 (c) and (d) compares peak stress distributions. Little difference is observed between
the normal stress distributions which consist of two parts, the end compression peaks and the
inner uniform tensile zone. The inconsistent wiggles present in the normal stress distributions,
are a result of numerical inaccuracies in the interfacial elements. The comparison of the shear
stress distributions portrays variation, which indicates the impact of the selected BC’s. It is
obvious that the F-F BC produces the most uniform τxy distribution over the middle interfacial
zone. It is also the only distribution similar to that of the preceding stage, because global
deformation is restricted by the F-F BC.
Information gathered from Figure 3.6 (o) on page 33 indicates crack initiation which eventually
leads to total fracture of the substrate in the R-R specimen. This is the reason why the inter-
facial shear distribution for this specimen in the last two stages is omitted.
Post-peak interfacial stress distributions are captured in Figure 3.7 (e) and (f). Higher com-
pression stresses at the top interfacial element are visible for the F-R BC. The effect thereof is
visible in the corresponding τxy distribution.
3.4 EXPERIMENTAL SET-UP
The final parameter test set-up, based on work done by Hawkins and Mattock [26], was a
push-off system. Based on the interfacial stress distributions and global deformation from the
previous section, a rotationally F-F set-up was chosen. The design produced a versatile method,
used for both shear and tension parameter tests. Figures 3.8 and 3.9, illustrate the set-up con-
figurations along with the tensile and shear specimen geometries.
The choice of the test set-up was not to mimic specific practical applications. The reason why
the parameter tests and their set-ups were chosen the way they were, was to obtain obtain
objective parameter values for the numerical model (Coulomb-friction, along with tensile and
compression limits) for generic application. It is the primary aim to model any boundary value
problem. This is why uniform stress fields in all stages of the response are needed. The chosen
set-up is superior to most others, while practical reasoning ruled out the better test, namely the
Iosipescu test, which has been shown to produce uniform shear stress along the plane of failure
[33] (in this case the interface). Large specimens would have had to be prepared for this test,
in order to have interface areas of representative size in the notched region. Another limitation
of the Iosipescu test is that normal pressure cannot be applied, in order to establish the friction
coefficient, while this may be possible in the current configuration in future work.
Heinrich Stander University of Stellenbosch
3.4. EXPERIMENTAL SET-UP 37
ECC
concrete
load cell
Figure 3.8: Experimental set-up for the interfacial tensile parameter test (dimensions in mm).
Heinrich Stander University of Stellenbosch
3.4. EXPERIMENTAL SET-UP 38
ECC
concrete
load cell
Figure 3.9: Experimental set-up for the interfacial shear parameter test (dimensions in mm).
Heinrich Stander University of Stellenbosch
Chapter 4
EXPERIMENTAL PROGRAM
4.1 INTRODUCTION
The experimental program was subdivided into two important phases. The aim of the interfacial
experimental program was to find correlations between the substrate surface preparation and
the corresponding interfacial properties. An important objective of this research was to pursue
preparation methods that are practical and uncomplicated when implemented. Preparation of
the specimens is described in section 4.2 and represents the first experimental phase. All the
different SSPT’s will be conveyed along with the preparation procedures up to the testing phase.
The next phase shows the parameter testing environment which describes the test methods and
results achieved. The first phase is indicative of the results of the second, which highlights the
importance of a correctly employed procedural substrate preparation method. Characterisation
of the interfacial bond and its methodical preparation is an important step towards developing
design tools for overlay systems.
4.2 SPECIMEN PREPARATION
4.2.1 Composite material mix designs
4.2.1.1 Substrate
A concrete mix design appropriate for slab design, was used for the substrate. Certain con-
straints were imposed on the design of the concrete substrate. A time constraint existed on
the experimental program which entailed fast-tracking of the procedure, i.e. shortening the
specimen curing and testing ages to accelerate retrieval of experimental results. Mechanical
roughening processes were performed on the substrate at three days after casting, at compres-
sive strengths ranging between 10 and 15 MPa. Early age tests implied that the mix design
had to ensure concrete compressive strengths in an envelope ranging between 35 and 45 MPa
at the earliest of testing ages, namely 10 days.
Klipheuwel sand and crushed Greywacke were used as aggregates. A basic mixing process was
instituted for the concrete. Table 4.1 illustrates the implemented concrete mix design.
39
4.2. SPECIMEN PREPARATION 40
Table 4.1: Concrete mix design.
W/C Water content [l] A/B [by mass] Sand/B [by mass] Stone size [mm]
0.5 200 4.54 2.03 13
4.2.1.2 Overlay
The ECC mix design was developed and improved over a number of years as part of a research
program at the University of Stellenbosch. Material optimisation is an ongoing process that
still continues. The mix proportions for the ECC employed during the experimental program,
were acquired from a previous research project conducted by Stander (2004) [39] and are shown
in Table 4.2.
Table 4.2: ECC mix design.
W/B A/B CEM I/B FA/B GGCS/B SP VA Vf
[by mass] [by mass] [by mass] [by mass] [by mass] [% of B mass] [% of B mass] [% by volume]
0.4 0.5 0.4 0.5 0.1 1 0.3 2
The mixing process was conducted in a Hobart-type mixer, with a capacity of 10l. The process
consists of the following steps:
• Premixing - All the dry constituents except the fibres are added into the mixing bowl.
These constituents are then mixed at the lowest mixing speed until the colour is uniform.
At this point the water is slowly added while the mixing speed is gradually increased.
• Addition of SP - The SP is only added at a time of one to two minutes after the addition
of water. The reason is that if the SP is to be added with the water, it would take part
in the initial hydration action [19], resulting in less SP to attain workability. The speed is
gradually increased throughout this stage until a relative high speed. The stage continues
for four minutes to ensure that the physical process of SP initiates.
• Addition of fibres - The speed is turned down for this phase. The reason for adding the
fibres to the mix this late, is to reduce the possibility of fibre damage incurred by harder
mix constituents. It continues a further two minutes, until the process is completed.
The mixing process discussed above has a duration of about eight minutes.
4.2.2 Substrate surface roughening techniques
Four different surface roughening strategies together with a reference surface, were investigated.
All these roughening methods are easily applicable to the substrate and were conceived with
this in mind. Characterisation of these methods follows in section 4.3. The necessity of having
various preparation methods, is to obtain different bond strengths, suitable for different appli-
cations and their specific requirements. Variety in choice is beneficial towards optimisation of
composite designs, especially when trying to harness the ductile capacity of ECC. The different
substrate surface roughening techniques (SSRT’s) exploited during tests, are as follows:
Heinrich Stander University of Stellenbosch
4.2. SPECIMEN PREPARATION 41
1. Reference surface - This is a mechanically untouched surface for reference purposes, rep-
resented by the substrate surface as cast against the mould.
2. Scrape surface - A sharp edged scraping plate is used to scrape the thin, weaker outer
layer of the substrate surface. It presents the least effort of mechanical roughening, only
removing what is necessary to expose the tougher inner material and to increase roughness
on a micro level. By definition, this method only embarks on creating a better chemical
bond. The method is illustrated in Figure 4.1.
3. Sandblast surface - Roughening of the substrate surface with sandblasting, not only in-
creases the chemical adhesion but also creates mechanical interlocking on a meso-scale
between the two materials. Sandblasting commences until the coarse aggregate is exposed
to about one millimetre clearance. An illustration of a sandblasted surface can be seen in
Figure 4.2 (a).
4. Drill holes - This method mostly contributes to increasing the mechanical bond, causing
a dowel effect of the overlay into the substrate. Holes with a diameter of ten mm are
drilled to five mm depths and cover ten percent of the surface area. The dimensions and
a roughened surface are illustrated in Figure 4.2 (b).
5. Precast grooves - All the above mentioned surface roughening techniques can be imple-
mented for repair and retrofitting strategies, except the precast grooves. This is only
applicable for new construction methods, utilising precast or extruded elements. The
grooves are situated perpendicular to the principal shear direction, and have dimensions
as illustrated in Figure 4.2 (c).
Figure 4.1: Application of the scrape roughening technique.
Heinrich Stander University of Stellenbosch
4.2. SPECIMEN PREPARATION 42
(a)
(b)
10 mm
5 mm
(c)
10 mm 10 mm
6 mm
6 mm
Figure 4.2: Illustrations of (a) sand, (b) drill and (c) groove roughened surfaces.
Heinrich Stander University of Stellenbosch
4.2. SPECIMEN PREPARATION 43
4.2.3 Substrate moistening
Sufficient moistening of the substrate, before casting of the overlay commences, is an important
phase of the SSPT. Omitting this step of the preparation procedure is detrimental to achieving
a proper interfacial bond. The degree of substrate surface saturation plays a substantial role in
acquiring the best possible chemical bond. It is an important aspect with little explanation in
the South African and international building codes.
Two different procedures were introduced, one addresses the time constraint in which a repair
or overlay strategy must be completed and the other induces simplicity and uniformity of ex-
ecution. For both methods, the fresh concrete was sealed and left to set in its moulds for one
day, after which it was demoulded and cured for at least two days in water curing tanks at
23◦C. At the age of three days the substrate was taken out of the curing tanks for the SSRT
to be applied. After the roughening application, the moistening procedure was applied and a
choice between two methods available.
Submerging the roughened substrates in the curing tanks for another 24 hour cycle was one of
the available procedures. Though difficult to execute, this timely procedure is specified by the
SABS 0100-2 as stated earlier in subsection 2.3.2.1. It ensured high saturation levels. After-
wards the specimens were taken from the curing tanks and left to dry for 30-45 minutes before
casting of the overlay took place. This procedure induces uniformity and simplicity to practical
execution.
The alternative procedure, was to leave the roughened substrates exposed for at least 24 hours
in order to simulate an unsaturated, exposed surface, ready for repair in practise. The exposed
conditions were that of room temperature, approximately 20◦C. Then the specimens were moist-
ened for 10 minutes and wiped dry with a damped cloth prior to the overlay casting.
Both procedures have the common goal of trying to achieve an optimal degree of saturation.
Neither saturation levels was measured, but an impact thereof is seen in the experimental
results. Accurate measuring of the saturation levels will definitely be beneficial in producing a
better moistening procedure.
4.2.4 Overlay casting and curing procedures
Directly after the ECC mixing process was concluded, casting of the overlay occurred. The
casting direction is an aspect which has an influence on the interfacial bond strength, thus it is
of importance to enforce the correct direction when simulating what occurs in practise. Figure
4.3 illustrates the casting process of a shear specimen.
Heinrich Stander University of Stellenbosch
4.2. SPECIMEN PREPARATION 44
ECC overlay
concrete substrate
mould
shake table
Figure 4.3: Casting of ECC on a concrete substrate.
Each specimen was vibrated on a shake table for two minutes, to ensure that interfacial contact
was uniform and good. All the specimens were sealed and left for three days to set, after which
they were stripped from their moulds. Experimental tests followed at different intervals to
perceive the ageing effect on interfacial bond strength. Testing at different ages, influenced the
curing time of the specimens accordingly. All specimens were cured in water curing tanks at
23◦C for the duration until the particular testing age, i.e. a 14 day old test specimen was cured
for 11 days. The testing procedure initiated directly after the specimens were removed from
the water. A full description of the specimen preparation is illustrated systematically in the
flowchart, shown in Figure 4.4.
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 45
SUBSTRATE CASTING
Setting Time: 1Day
SSRT
Water Curing @ 23 C
Moistening Procedure:
1. Mixing Procedure2. Vibration for 2 min
Demoulding
a. 24h Submerged period followed by 30 minute dryingb. 24h Drying period followed by 10 minute moistening
OVERLAY CASTING
Setting Time: 3 Days Water Curing @ 23 CDemoulding
TESTING
◦
◦
Figure 4.4: Systematic flow of the specimen preparation.
4.3 EXPERIMENTAL TESTS
4.3.1 Material tests
Uniaxial compression tests were performed on concrete specimen cubes of size 100 x 100 x
100 mm. The tests were executed in a Contest Grade A compression test apparatus. The
ultimate compressive strength (fcu) values were obtained for each batch of substrate material.
The remaining material parameters, indicated in Table 4.3 were estimated using the CEB-FIP
Model Code 1990 [9]. The concerning methods are detailed in subsection 5.1.1.
Table 4.3: Material properties of the concrete substrate.
fcu [MPa] E [MPa] ν ft,c [MPa] GIf [N/mm]
35-45 29000 0.2 2-3.8 0.05-0.075
Two different material test methods were performed to obtain the necessary ECC material pa-
rameter values. The first and most simple test was the uniaxial compression test, performed in
a Contest Grade A compression test apparatus. Cubes of size 50 x 50 x 50 mm were prepared
to retrieve the fcu values. All the specimens were prepared from the same batch as the corre-
sponding interfacial parameter tests. This experimental data is necessary for the ECC material
model. Mean values of the compression results at the respective ages, are listed in Table 4.4.
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 46
Table 4.4: ECC compressive mean values.
Age [Days] fcu [MPa] COV [%] Specimen amount
7 14.72 3.83 3
14 22.39 10.14 35
28 31.03 6.56 11
Direct uniaxial tensile tests were also performed. Specimens with dog bone-like geometry,
illustrated in Figure 4.5, were prepared for testing in the Zwick Z250 Universal Testing machine.
A displacement rate of 0.5 mm/minute was applied in order to conform to loading rates specified
in South African concrete testing codes [35], [36] and [37], which specify a three to ten minute
test duration limit for cement-based materials. Two LVDT’s were implemented to measure
deformation along the uniform middle part. They were spaced diagonally across the specimen
for averaging purposes. A special frame was used in the set-up as indicated in Figure 4.6.
260 mm
80 mm
60 mm
30 mm
15 mm
Figure 4.5: Specimen geometry of the ECC direct uniaxial tensile test.
Figure 4.6: Test set-up for ECC uniaxial tensile testing.
Results of tests differ according to the specimen testing age. Figure 4.7 on page 48 shows the
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 47
tensile test data for the ECC material at different test ages. The tensile test data produced four
parameter values which are given in Table 4.5. All four parameter values are mandatory for the
user-supplied material model. Special considerations were necessary to obtain the first crack
tensile strength (σfc). It is imperative that the trilinear material model curve in Figure 3.2 on
page 25, represents the actual tensile behaviour as best possible. Reflecting this statement, σfc
was determined at the intersection of the elastic and linearised strain-hardening curves. This
method produced accurate outcomes for the two lower testing ages. The first crack values for
the 28 and 150 day periods, were obtained at their first crack peaks. A quad-linear material
model curve is necessary for simulation of the 28 and 150 day actual responses. Use of a trilinear
response curve for the two older materials, in the same way as for the younger materials, results
in an overestimation of material strength in its early damage state.
The fifth parameter value in Table 4.5, namely Poisson’s ratio (ν), was taken from [32]. It is just
an indication, used because it is the only relevant value available for ECC. According to [32],
both the E-modulus and shear modulus (G-modulus) were determined by calculation methods
from RILEM, as shown in (4.1) and (4.2) respectively. ν was determined indirectly with (4.3),
using the E-modulus and G-modulus values generated from their respective test methods. More
information concerning the shear behaviour of ECC and the accompanying test method can be
found in [32].
E −modulus =∆P
∆ε
=Pb − Pa
εb − εa
= 7.4 GPa
(4.1)
G−modulus =∆P
∆γ
=Pb − Pa
γb − γa
= 2.8 GPa
(4.2)
Where Pa = 0.1 MPa, Pb=1
3× σtu; ε and γ are the respective normal and shear strain values
at a and b.
ν = (E
2 ×G) − 1
= (7.4
2 × 2.8) − 1
= 0.32
(4.3)
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 48
0 1 2 3 4 5 6 7 80
1
2
3
4
5
6
ε [mm/mm %]
σ t [MP
a]
Age: 7 Days
0 1 2 3 4 5 6 7 80
1
2
3
4
5
6
ε [mm/mm %]
σ t [MP
a]
Age: 14 Days
0 1 2 3 4 5 6 7 80
1
2
3
4
5
6
ε [mm/mm %]
σ t [MP
a]
Age: 28 Days
0 1 2 3 4 5 6 7 80
1
2
3
4
5
6
ε [mm/mm %]
σ t [MP
a]
Age: 150 Days
Figure 4.7: Uniaxial tensile test results at different ages. The coloured responses indicate different testsof the same specimens.
Table 4.5: Experimentally determined parameter values for the ECC material model.
Age [Days] σfc [MPa] σtu [MPa] COV [%] εsh [mm/mm] E [MPa] COV [%] ν
7 1.42 1.88 6.7 0.028 6730 15.7 0.32
14 2.42 3.44 17.6 0.05 10610 19.6 0.32
28 2.58 4.02 5.2 0.052 14023 14.1 0.32
150 3.21 5.11 18.7 0.020 19197 0.6 0.32
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 49
5 10 15 20 25 300.4
0.6
0.8
1
1.2
1.4
1.6x 10
4
Age [Days]
Ela
stic
mod
ulus
[MP
a]
0 20 40 60 80 100 120 140 1600.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2x 10
4
Age [Days]
Ela
stic
mod
ulus
[MP
a]
Figure 4.8: The development of Young’s modulus over time.
E-modulus, i.e. the material stiffness, is an important material attribute. When two materi-
als, with different stiffnesses, are used as monolithic structural members, higher loads will be
transferred to the stiffer material. In cases where repairs are made with a new material, it is
important that both materials have similar stiffnesses. A stiffer overlay for repair of a damaged
structural element would attract higher loads, relieving the substrate material to some extent
or taking over the structural function at the position of repair. The extent of damage, i.e.
micro-cracks in brittle concrete, is not known or seen in many repair cases. In such cases, a
preferable scenario would be a slightly stiffer repair material. This will enhance the repaired
structure through lowering the stress state in the damaged material.
This research utilised an ECC material with a slightly lower E-modulus when compared to
concrete, but one with superior ductility. Altering the ECC mix design in order to increase the
E-modulus, would lower the ductility. Figure 4.8 shows the short and long term development
of the ECC’s E-modulus. The fact that ECC contains no coarse aggregate, reflects on the
obtained E-modulus and the development thereof. The gradient of the demonstrated stiffness-
evolution, is different compared to ordinary concrete. It is a result of the high portion of FA in
the binder composition, which causes a lower E-modulus at first, but one that develops over a
longer period.
4.3.2 Interfacial tests
Data concerning the interfacial properties of the composite was retrieved from experimental test
results. A variety of different substrate surface preparations were performed and investigated
through a systematic approach. Shear and tensile tests were conducted to extract the necessary
interfacial parameter values. The following subsections will report in detail on the two most
promising surface roughening techniques. All the results are summarised to a condensed format
in the last subsection. All the experimental results are shown in Appendix B.
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 50
4.3.2.1 Shear tests
Data concerning the interfacial shear properties was obtained through shear parameter tests.
Different SSPT’s were examined for reasons previously mentioned. A list indicating the number
of specimens tested for each SSPT, is shown in Table 4.6. The number of test specimens are
limited due to a time limitation. A larger set of specimens for each parameter would have
increased the statistical value significantly. Nevertheless, a minimum of three or more test were
performed for the tests considered more important. This section reveals data concerning the
experimental set-up, data captured and visuals of post-test interfacial surfaces.
Table 4.6: Number of shear specimens tested for each SSPT.
Roughening Age [Days] Moistening Abbreviation Number of Specimen
Reference 14 10min R1410 3
28 10min R2810 3
24h R2824 1
Scrape 14 10min S1410 3
24h S1424 2
Sandblast 7 24h SB724 2
14 10min SB1410 9∗
24h SB1424 4
28 10min SB2810 1
24h SB2824 2
Drill holes 14 24h DH1424 3
Precast grooves 14 24h PG1424 3
*Four of these specimens were reinforced with rebar in the substrate
Subsection 4.2.2 presented eight different SSPT’s as shown in Table 4.6, that were prepared for
investigation. All the specimens were subjected to a fixed-fixed shear parameter test, conducted
in the Zwick Z250 Universal Material Testing machine. Tests were executed with displacement
control. The rate of displacement was converted from force-based loading rates, provided by
the SABS Methods 863:5-1994 [36], which specifies a time envelope of three to ten minutes for
concrete material tests. Tests performed outside the time envelope, addresses either dynamic or
long-term material behaviour. The displacement loading rate was 0.1 mm/min, corresponding
to test durations of five to twenty minutes, depending on the post-crack behaviour. Interfacial
fracture usually occurred in the prescribed envelope of three to ten minutes. The displacement
loading rate was measured between the base platens of the Zwick testing machine and thus
governed by the deformations of the whole specimen. A more correct way of loading control, is
displacement measurements over the interface which excludes the rest of the specimen and thus
its elastic deformation. Unfortunately, this was not an available option for the experimental
program and the effect thereof was occasional snapback behaviour at the point of fracture as
illustrated in Figure 4.11 on page 54.
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 51
Specifications for the design of the test set-up were gathered from the numerical investigation
conveyed in section 3.3. The specimen geometry is illustrated in Figure 3.4 on page 30. Figure
4.9 illustrates the actual test set-up that includes the connections and the necessary measure-
ment instruments. Displacement measurements were performed with five LVDT’s and the data
was logged with a Spider 8 data logger. The five LVDT’s were placed in strategic places around
the interface. Three were used to measure horizontal displacements, two at the top on both
sides and one at the bottom in the middle. The other two were used for vertical displacements
on both sides of the interface.
Figure 4.9: Shear test set-up.
The specimens were connected to the base plates with an epoxy adhesive which differed ac-
cording to availability. HBM X60 and Sika Anchorfix 1 were the two adhesives used to achieve
specimen fixity on the steel platens. Specimen fixity facilitated the capturing of post-fracture
behaviour. Applying the correct amount of adhesive ensured full contact between the specimen
and platens and it also lessened the occurrence of stress concentrations between the specimen
and platens.
All the specimens had a drying period of 30 minutes after they were taken out of the curing
tanks, after which they were glued to the steel platens under a constant load of 400 N for a
period depending on the epoxy curing time. The epoxy curing times varied from 30 minutes for
HBM X60 to 60 minutes for Sika Anchorfix 1, after which the test commenced. This process
reduced the occurrence of drying shrinkage and its possible effects on the interface.
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 52
Results of two different, easily applicable SSRT’s are of interest. These two methods are scraping
and sandblasting. The results thereof, for both the 24 hour and 10 minute moistening periods
are showcased in the following paragraphs.
Sandblasted surface
Sandblasting the concrete substrate occurred after it was collected from the curing tanks. It
endured until coarse aggregates were exposed up to one millimetre. The duration of sandblast-
ing relates to an exposure time of one to two minutes per 1000 cm2. This was easily achieved,
because the substrates were all cast with the concerning surface facing downwards. Settlement
of the aggregate occurred close to the surface, as gravity imposes. It is difficult to illustrate the
freshly sandblasted surface with a two dimensional picture, because the substrate removal is so
little, but Figure 4.2 (a) on page 42 depicts the exposure of coarse aggregates.
The averaged interfacial shear test results are shown in Table 4.7. Both types of SSMT’s were
investigated, i.e. the 24 hour and the 10 minute moistening periods. The three batches subjected
to 24 hour moistening prior to casting, were tested at different ages in order to investigate the
shear-bond development over 28 days. Figure 4.10 shows the shear-bond development for both
moistening periods. It is obvious from Table 4.7 and Figure 4.10 that the shorter moistening
period improves both the interfacial bond strength and fracture energy.
Table 4.7: Shear parameter values obtained from experiments on sandblasted specimens.
Moistening Age [Days] Rebar τxy [MPa] COV [%] GIIf [N/mm] Ψ0,exp up [mm] vp [mm]
24h 7 - 1.138 66 0.546 0.46 0.31 0.55
10min 14 - 2.73 19.2 0.964 0.98 0.5 0.8
10min 14 X 3.239 37.9 1.534 1.11 1.08 0.92
24h 14 - 2.238 11.2 0.751 1.08 0.7 0.61
10min 28 - 3.67 n/a 2.539 1.53 2.2 1.05
24h 28 - 2.376 34.1 0.904 1.25 1.38 0.59
5 10 15 20 25 301
1.5
2
2.5
3
3.5
4
Age [days]
τ xy [M
Pa]
10min24h
5 10 15 20 25 300.5
1
1.5
2
2.5
3
Age [days]
GfII [N
/mm
]
10min24h
Figure 4.10: Shear-bond and Mode II fracture energy development of sandblasted specimens.
A flaw in the design of the shear test set-up was not detected by the numerical investigation
conducted in section 3.3 to refine the test set-up, because the applied interfacial bond parameter
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 53
values were greatly underestimated. This flaw is the initiation of a crack in the lower back zone
of the concrete, due to heightened tensile stresses in this region. It was only exposed during the
experimental programme. The occurrence of such a crack in the substrate, was predicted by
the preceding numerical simulations, but only for the F-R and R-R DOF options. Both these
DOF options incurred heightened tensile stresses in the particular region at lower loads. Crack
growth stopped before interfacial fracture occurred, resulting in little rotation of the interface.
Rotation of the interface, favours the development of tensile forces over the interface, because of
the newly adapted geometry. The presence of tensile forces is not the only factor that reduces
the interfacial shear capacity. The damage incurred by the substrate crack, lowers the speci-
men stiffness. The interfacial confinement caused by the F-F boundary condition is lowered as
a result of the lower specimen stiffness. Less interfacial confinement is thus the other factor
reducing the interfacial shear capacity.
Over the entire scope of the programme, this crack was detrimental to two specimens. Rein-
forcement bars were placed in the substrate as a solution for crack development in this area. A
comparison between normal and reinforced specimens was able after a batch of reinforced sand-
blasted specimens were tested. Table 4.7 depicts this increase in shear-bond, confirming both
the effect of the rotation induced interfacial tensile stresses and the lower interfacial confinement.
A detailed comparison of experimental results for both the 14 day unreinforced sandblasted
SSMT’s are showcased in Figures 4.11 (a) and (b). Two attributes of the shear stress graphs
are obvious, firstly the stronger bond attained by the 10 minute moistening method and sec-
ondly the graph kink at about 1.6 MPa. The kink is a result of crack initiation in the back
of the substrate. The specimen deformation resulting from the substrate crack, contributes to
the measured vertical displacement before interfacial fracture occurs. The kink is visible when
investigating the normal-tangential slip graphs.
Inspection of these graphs, reveal three different gradients. The first which is zero, represents
elastic specimen deformation with no dilatation. The end of the first gradient marks the in-
ception of chemical bond degradation. The second gradient marks the initial dilatancy of the
measured point (Ψ0,exp), and the values thereof are shown in Table 4.7. It represents the initial
interfacial debonding period. This is a rapid occurring phenomenon that not only represents
total degradation of the chemical bond but also degradation of the physical bond. The speed
at which this debonding occurs, along with relaxation of the total elastic deformation of the
whole specimen, culminates in occasional snapback behaviour, captured by the LVDT’s at the
end of the chemical bond fracture. Snapback behaviour is indicated in Figure 4.11 (b). The
third gradient ensues with lesser dilatation than the foregoing, this last stage represents a com-
bination of mechanical friction and fibre pull out until total fracture and BC induced normal
uplift. Due to the BC’s of the set-up, dilatation and normal uplift are ongoing measurements
in the experiment. Dilatation is suppose to reach zero as the up moves towards its asymptote
at large shear-slipping, when zero confining pressure exists on a state of pure shear over the
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 54
interface. The values of up and vp are averaged values obtained at the end of the second gradient.
Figure 4.11 (c) shows the comparative shear responses for reinforced specimens. The compari-
son indicates a stronger interfacial shear bond as well as very little elastic deformation without
the kink response. The reinforced specimen succeeds in preventing the substrate crack and the
accompanied interfacial rotation. It leads to a higher shear bond and higher response conformity.
0 0.5 1 1.5 20
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 20
0.5
1
1.5
2
u [m
m]
υ [mm]
(a) (b)
snapback
snapback
0 0.5 1 1.5 20
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 20
0.5
1
1.5
2
u [m
m]
υ [mm]
(c)
Figure 4.11: Shear stress plotted against vertical slip and related to normal uplift for (a) unreinforcedSB1410, (b) SB1424 and (c) reinforced SB1410 specimens.
Upon inspection of the post-fractured interfacial surfaces, minor differences between the two
moistening techniques were found. Post-fracture surface roughness and material rupture are
the two main differences. Rupture of composite material adjacent to the interface is depicted
either by aggregate exposure or manifestation of ECC and its fibres on the substrate surface.
In many cases, both materialised simultaneously. The higher fracture energy demonstrated
by the 10 minute moistened specimens is further substantiated by its rougher post-fractured
surfaces, when compared to the 24 hour moistened specimens. A stronger discrepancy is visi-
ble between the post-fracture surfaces of unreinforced and reinforced specimens. Figures 4.12
and 4.13 illustrate the difference and degree of aggregate exposure between the two specimens.
The reinforced specimens do not convey any visible traces of ruptured ECC material on the
substrate, which is not the case for the unreinforced version. Results of the tensile tests will
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 55
indicate fracture of ECC adjacent to the interface, which supports the probability of fibres on
the substrate for the unreinforced specimens. This is because of the rotation induced interfacial
tensile stresses.
Figure 4.12: Post-fractured surface of a unreinforced SB1410 specimen.
Figure 4.13: Post-fractured surface of a reinforced SB1410 specimen.
Scraped surface
Scraping of the substrates were performed with a flat sharpened plate, four centimetres in width
(Figure 4.1). Application thereof was brief in duration, only two to three scrapes per section.
A difference in appearance between the pre- and post-scraped areas was difficult to see with
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 56
the naked eye. The only visible phenomenological difference was the little cementitious powder
coming off as a result of the scraping, which indicates the exposure of a new substrate layer.
Unhydrated cement particles were thus exposed, which influenced the chemical adhesion.
Application of the physical procedure occurred prior to the moistening procedure. A higher
degree of hydration for the newly exposed layer is a result of the longer moistening period, prior
to overlay casting. A stronger chemical bond is thus established in the 10 minute moistening
method when compared to the 24 hour moistened specimen, due to the different level of satura-
tion present at the time of overlay casting. Figure 4.14 illustrates good interfacial shear-bonds
for both SSMT’s , the shorter moistening period enhances the bond. In one case the bond was
stronger than allowed by the substrate, which led to substrate failure as depicted by the red
line in Figure 4.14 (a).
0 0.5 1 1.5 20
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 20
0.5
1
1.5
2
2.5
u [m
m]
υ [mm]
(a)
0 0.5 1 1.5 20
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 20
0.5
1
1.5
2
u [m
m]
υ [mm]
(b)
Figure 4.14: Shear stress plotted against vertical displacement and related to normal displacement(dilatation) for (a) S1410 and (b) S1424 specimens.
Table 4.8 shows a higher shear-bond achieved with the shorter moistening period as a result
of the lesser hydrated cement particles on the newly exposed surface of the substrate. Again
a lower variation in shear strength is seen for the longer moistening period. This is similar
to the response exhibited by the sandblasted specimens. As suspected the fracture energy is
more or less the same because no mechanical interlocking was achieved or addressed by the
method. This method only addresses the chemical bond, which is substantiated by the almost
unblemished fracture surfaces, illustrated in Figure 4.15. The result of the pure interfacial
fracture is captured in test data, when considering the third horizontal gradient of dilatation
in Figure 4.14. The approximately zero horizontal gradient depicts the presence of vertical slip
without further dilatation.
Table 4.8: Shear parameter values obtained from experiments on scraped specimens
Moistening Age [Days] Rebar τxy [MPa] COV [%] GIIf [N/mm] Ψ0,exp up [mm] vp [mm]
10min 14 - 2.287 29.0 0.988 1.05 1.5 0.9
24h 14 - 1.773 13.5 0.9 0.74 1.05 0.97
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 57
Figure 4.15: Post-fractured surface of a S1410 specimen subjected to shear loading.
4.3.2.2 Tensile tests
It was required that the design of the tensile test method conforms to geometric specifications
of the shear test method. Fulfilling this requirement, resulted in an accelerated experimental
program. This implied that the geometry of the tensile specimens had to fit into the steel
platens, which were constructed for the shear tests. Using the same test set-up, resulted in a
fixed-fixed set-up i.e. no DOF’s present except axial elongation. The eventual shape of the
specimen was rectangular, with the interface in the middle of the composite. Figure 3.8 on page
37 illustrates the geometry of the tensile test and its set-up.
A list of the amount of correctly performed tensile tests is shown in Table 4.9. Evidently,
not the same amount of tensile tests were performed when compared to the shear tests. This
was due to the difficulty associated with performing these tests. A detailed discussion of both
sandblasted and scraped specimens is given in this section and the results of the remaining
SSPT’s are graphically illustrated in Appendix B.
The same experimental procedure, used for the shear tests, was instituted for the tensile tests
and executed in a Zwick Z250 Universal Materials Testing Machine. All the tensile specimens
were left to dry for 30 minutes after they were taken from the curing tanks. The specimen was
then glued to steel platens with an epoxy adhesive namely Sika Anchorfix 1. The gluing process
endured approximately 45 to 60 minutes. A major concerning factor for this method was one
of eccentric loading. Care was needed for the gluing process, to ensure the highest possible
centricity. Directly after the glue process, the actual test started. Displacement control was
applied at 0.1 mm/min movement between the Zwick base platens, effectively incorporating
global deformation of the specimens. As discussed in the previous subsection, this is not an
optimal set-up, but it is conclusive in revealing reliable data. Four LVDT’s were employed, one
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 58
Table 4.9: Number of tensile specimens tested for each SSPT.
Roughening Age [Days] Moistening Abbreviation Number of Specimen
Reference 14 10min TR1410 1
28 10min TR2810 2
24h TR2824 -
Scrape 14 10min TS1410 3
24h TS1424 2
Sandblast 7 24h TSB724 2
14 10min TSB1410 3
24h TSB1424 2
28 10min TSB2810 -
24h TSB2824 3
Drill holes 14 24h TDH1424 2
Precast grooves 14 24h TPG1424 3
at each corner of the specimen, measuring over as small as possible length across the interface.
Figure 4.16 illustrates the configuration of the experimental test set-up.
Figure 4.16: Experimental set-up of the interfacial tensile test.
Sandblasted surface
The experimental investigation of the sandblasted substrate contains results for two different
moistening periods. Tensile tests were conducted in coherence with the shear tests, thus reflect-
ing ageing influences as well. Table 4.10 shows an increase in tensile bond strength, occurring
as time ensues. Differentiating between the two moistening periods at 14 days is difficult, for
they lack obvious differences, except larger Gf and higher variability for the 10 minute period.
A larger COV is exhibited by the shorter moistening period for both the tensile and shear bond
strengths. It depicts the influence of variation in the duration of the execution for the shorter
moistening period and substantiates the conformity in saturation levels achieved by the longer
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 59
moistening method. The 28 day tests revealed an increase in bond strength for the shorter
moistening period. It is not possible to give the exact bond strength at this age, because some
of the specimens were too strong, resulting in adhesive failure.
Table 4.10: Tensile parameter values obtained from experiments on sandblasted specimens.
Moistening Age [Days] ft,i [MPa] COV [%] GIf [N/mm]
24h 7 0.43 6.9 0.052
10min 14 0.798 22.5 0.662
24h 14 0.835 16.1 0.46
10min 28 > 1.1 -* -*
24h 28 > 0.919 -* -*
*Adhesive failure
A comparison of the 14 day tensile results is given in Figure 4.17. An interesting observation is
exhibited here by the interfacial post-fracture toughness. After the initial fracture occurs, there
is an extra resistance present, formed by fibres bridging a newly defined interface. Figure 4.18
on page 60 clearly illustrates the presence of fibres across the interface. Fracture occurred into
the ECC and not on the originally defined interface depicted by the line, as shown in Figure
4.19. The tensile strength of ECC is three to five times that of the interfacial bond, stipulating
that the ECC near the interfacial contact zone is weakened by inter-material moisture trans-
port processes. Various reasons might explain this phenomenon, but without deeper research,
no conclusive reason can be stated. The degree of capillary suction prior to casting, created by
the moistening procedure and the enlarged specific surface area of the substrate, may have a
considerable effect in producing this phenomenon. The hypothesis, explaining that the height-
ened capillary suction of the substrate draws cement paste from the overlay at the contact zone,
causing a weaker, aggregate dominated area behind the interfacial contact zone, is promising
[42].
0 0.5 1 1.5 2 2.5 30
0.2
0.4
0.6
0.8
1
u [mm]
σ t [MP
a]
(a)
0 0.5 1 1.5 2 2.5 30
0.2
0.4
0.6
0.8
1
u [mm]
σ t [MP
a]
(b)
Figure 4.17: Tensile stress versus vertical displacement for (a) TSB1410 and (b) TSB1424 specimens.
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 60
Figure 4.18: Interfacial tensile test of a sandblasted specimen, showing bridging fibres.
Figure 4.19: The concrete part of a sandblasted interfacial tensile test, showing ECC material on itsfracture surface.
ECC Concrete
Figure 4.20: Post-fractured surfaces for a sandblasted tensile specimen, showing ECC on the left andconcrete on the right.
Scraped surface
Again both moistening methods were implemented, but no substantial difference in terms of
strength and fracture toughness was measured. The COV for ft exhibited by the 24h moistening
period, indicates higher reliability than the shorter period. A clean break on the interface
occurred at each failure with almost no fracture toughness present. Figure 4.22 substantiates
the little fracture toughness exhibited in Table 4.11 for the respective methods, by showing the
clean surfaces on either side of the interface. A very small COV was attained with the 24h
moistening period as seen in Figure 4.21 (b), which arises from the reliable degree of saturation
gained from the 24h moistening period.
Table 4.11: Tensile parameter values obtained from experiments on scraped specimens.
Moistening Age [Days] ft,i [MPa] COV [%] GIf [N/mm]
10min 14 0.6 13.2 0.11
24h 14 0.625 2.2 0.06
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 61
0 0.5 1 1.5 2 2.5 30
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
u [mm]
σ t [MP
a](a)
0 0.5 1 1.5 2 2.5 30
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
u [mm]
σ t [MP
a]
(b)
Figure 4.21: Tensile stress plotted against vertical displacement for (a) S1410 and (b) S1424 specimens.
ECC Concrete
Figure 4.22: Post-fractured surfaces of a scraped tensile specimen, illustrating ECC on the left andconcrete on the right.
4.3.2.3 Condensed results
The results acquired throughout the experimental programme are summarised in Tables 4.12
and 4.13, illustrating the full spectrum of applied SSPT’s. A comparison of results between
the various SSPT’s and the reference tests, reveals a considerable improvement in strength. A
comparison of the bond strength achieved and its associated COV for the two SSMT’s, demon-
strates stronger bonds in shear and tension for the 10 minute moistening but lower variability
for the 24 hour moistening period.
Strength, COV and fracture toughness are important when judging the results from a structural
perspective, because brittle failure of structural elements must be avoided as best possible. An
important aspect of all the SSPT’s is to enlarge the specific surface area of the substrate, to which
an overlay can adhere. Properties of the overlay material in its fresh state, such as workability
have a significant effect on creating an interfacial bond. Workability together with vibration
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 62
and the duration thereof at casting, significantly influences the effectively obtained interfacial
surface area. These parameters were kept constant throughout the research programme and the
ECC mix design with an optimum workability was collected from previous work [39].
Table 4.12: Shear parameter values obtained from experiments performed on the various SSPT’s.
SSPT Rebar τxy [MPa] COV [%] GIIf [N/mm] Ψ0,exp up [mm] vp [mm]
SB724 - 1.138 66.0 0.504 0.46 0.31 0.55
R1410 - 0.776 12.2 0.097 1.7 0.45 0.24
S1410 - 2.287 29.0 0.996 1.05 1.5 0.9
S1424 - 1.773 13.5 0.9 0.74 1.05 0.97
SB1410 - 2.73 19.2 0.964 0.98 0.5 0.8
SB1410 X 3.24 37.9 1.534 1.11 1.08 0.92
SB1424 - 2.238 11.15 0.751 1.08 0.7 0.61
DH1424 - 1.588 34.4 0.394 1.53 0.83 0.54
PG1424 - 1.759 4.0 0.826 1.94 2.25 1.1
R2810 - 1.024 21.6 0.85 0.59 1.5 1.77
R2824 - 0.225 - 0.025 1.3 1.05 0.5
SB2810 - 3.67 - 2.539 1.53 2.2 1.25
SB2824 - 2.376 34.1 1.008 1.25 1.38 0.59
Table 4.13: Tensile parameter values obtained from experiments performed on the various SSPT’s.
SSPT ft,i [MPa] COV [%] GIf [N/mm]
SB724 0.43 6.9 0.052
R1410 0.1 - 0.0065
S1410 0.6 13.2 0.11
S1424 0.625 2.2 0.06
SB1410 0.798 22.5 0.662
SB1424 0.835 16.1 0.46
DH1424 0.664 10.1 0.624
PG1424 0.71 29.0 0.079
R2810 0.087 0.21 0.077
R2824 - - -
SB2810 > 1.1 - -
SB2824 > 0.919 - -
Examination of the results is classified into two subclasses, represented by preparation tech-
niques addressing either the chemical bond or the mechanical bond. By definition a mechanical
bond is achieved at a macro-level via mechanical interlocking, thus the bond obtained with
sandblasting should be classified as chemical but because of the substantial mechanical altering
that borders the macro-level, it is not. Table 4.14 distinguishes between chemically (C) and
mechanically (M) induced bonds. The bond type created by the precast grooves, differs from
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 63
tension to shear. In the uniaxial tensile direction, the grooves only increase the interfacial area,
which is unscathed like the reference surface. In the shear direction, mechanical interlocking is
achieved on a macro-scale. On the other hand, the scrape roughening method removes a very
fine powdery substance from the substrate surface, too little to be classified as increasing the
mechanical bond. This fine powder is believed to be hydrated cement from the top outer layer
of the substrate.
Table 4.14: The bond type induced by the particular roughening technique.
Method Reference Scrape Sandblast Drill Holes Precast Grooves
Bond type C C M M C/M
All the interfacial parameter results at 14 days are summarised as a bar chart in Figure 4.23,
displaying differences in bond strengths and fracture toughness for the two moistening methods.
The shear bond strength produced from scraping the substrate surface, is higher than that of
both the drill holes and grooves and near to that of the sandblasted specimens. When one
addresses only the chemical bond as in the case of scraping, a tension fracture toughness prob-
lem arises. Post-fracture behaviour is well addressed by the sandblasting option, especially in
tension, because the ECC fibres come into play. The sandblasting option definitely addresses
brittle failure, a phenomenon associated with concrete overlays. Both scraping and sandblasting
display stronger bonds and tougher failures for the shorter moistening period. Precast grooves
result in mechanical interlocking when in shear, but not when in direct tension, which is shown
by the low mode I fracture energy values.
Drilled and grooved specimens resulted in weaker than expected bond strengths. In tension,
both methods performed better than the scraped specimens. In shear, both methods performed
under expectations. No mechanical scraping or sandblasting were performed on either of the
specimens. This resulted in a reference surfaces with larger macroscopic areas. The fact that
the ECC studs caused by the drill holes sheared off, means that the stud area was not enough
to increase the interfacial shear capacity to higher levels. Both methods methods show promise
and can be implemented as is to achieve weaker interfacial bonds where required. In shear, the
PG1424 specimens produced the lowest variability, which generate reliable performance pre-
diction. The difference between the scraping and the other indented surface is the amount of
post-fracture toughness. This toughness, depending on the application and situation, can arrest
the growth of delamination along an interface.
Throughout the results it is seen that an increase in strength occurs with age, there is a differ-
ence in the gradient of change for the two moistening methods. Figure 4.10 on page 52 depicts
this difference and shows that the shorter moistening period, induces a stronger bond, not only
in shear but also in tension.
Heinrich Stander University of Stellenbosch
4.3. EXPERIMENTAL TESTS 64
1 2 3 4 50
0.2
0.4
0.6
0.8
1
Reference Scrape Sandblast Drill holes Grooves
f t,i [M
Pa]
/ G
fI [N/m
m]
1 2 3 4 50
0.5
1
1.5
2
2.5
3
Reference Scrape Sandblast Drill holes Grooves
τ xy [M
Pa]
/ G
fII [N/m
m]
ft,i
@ 10min
GfI @ 10min
ft,i
@ 24h
GfI @ 24h
τxy
@ 10min
GfII @ 10min
τxy
@ 24h
GfII @ 24h
Figure 4.23: Interfacial parameter results at 14 days for tension and shear tests.
Inspection of the sandblasted and scraped post-fracture surfaces in Figures 4.20 and 4.22 on
pages 60 and 61 respectively, indicates that the moistening methods alone did not cause the
material fracturing of ECC in sandblasted specimens. Both roughening methods were subjected
to the same moistening procedures. The post-fracture analysis delivered two totally different
images: A clean fracture for the scraped surface and a newly defined fracture surface for the
sandblasted specimens. Thus by increasing the substrate specific area, one definitely enlarges
the capillary suction present when casting, which causes material fracture next to the interface.
The dilatancy values indicated in Table 4.12, depict experimental values captured at the top
part of the interface. Ψ0,exp does not represent the required Ψ0, which is the dilatancy at zero
confining stress on an interface in a state of pure shear. It actually underestimates Ψ0, because
of interfacial confining stresses due to set-up configuration. Ψ0 is thus calculated by means of
a numerical iterative process until the instant where the numerical model produces the same
dilatancy response at the concerning material point. The implemented Ψ0 value producing the
correct result, denotes the expected experimental Ψ at zero confining stress. This iterative
process and the corresponding values will be discussed in section 5.2.
Heinrich Stander University of Stellenbosch
Chapter 5
NUMERICAL SIMULATION
Numerical simulation of experimental results with the appropriate material models, is an im-
portant stage in characterising the interfacial properties towards a computational predictive
capacity. The interfacial properties, which vary for each of the applied substrate preparation
techniques, were retrieved from the experimental programme. Modelling the whole experimen-
tal shear set-up, ensured that the numerical prediction incorporated secondary influences caused
by the connections. It also relates to a more precise reflection of values expressed by the mea-
suring instruments. Parameter values depicting the properties of the sandblasted interface were
implemented and used for illustration purposes in this chapter.
5.1 Material models and input data
Four different material models were used to simulate the parametrical experiments. A discussion
of the two models for concrete and steel are given in this section. The other two models,
describing ECC and the interface have been defined in section 3.2.
5.1.1 Concrete material model
A constitutive model was implemented to simulate the substrate behaviour, it is known as the
Total Strain Rotating Crack model. It describes the tensile and compressive behaviour of con-
crete with one stress-strain relationship. This model consists of two parts: A basic part, defining
elastic behaviour and a complex part, defining nonlinear relations in tension, compression and
shear. The softening curve defining nonlinear tensile behaviour, was chosen as an exponential
function and it is illustrated in Figure 5.1. Precision is of utmost importance, but sometimes
the effectiveness of the result needs to be weighed against the duration to achieve it. With
insignificant differences in the eventual results, it was decided to use the elastic compressive
and constant shear curve.
65
5.1. Material models and input data 66
up
σ
ft
GIf
Figure 5.1: The exponential softening curve and mode I fracture energy for the concrete material model.
Values for tensile strength, fracture energy, elastic modulus and Poisson’s ratio are necessary
input values for the material model. No such data was generated during the experimental pro-
gramme. The CEB-FIP Model Code 1990 [9] provided the necessary mathematical relations to
estimate the required values from the concrete compressive strength. The latter was the only
experimental data. Using this knowledge, the characteristic cube strength (fck-cube) values
were calculated and the concrete was classified as a C30 concrete.
The variability of concrete tensile strength values is higher than the variability of its compressive
strength values. The shape and surface texture of aggregates influence the ft values to a larger
extent than the fc values. The eventual ft may also be substantially reduced by environmental
influences. Caution needs to be exerted when accounting for ft in design. In the absence
of more accurate data about this particular concrete, the lower (fctk,min) and upper bound
(fctk,max) values of the characteristic tensile strength may be estimated from the characteristic
compressive strength, using equations (5.1) and (5.2). The results for these calculations can be
seen in Table 5.1, the applicable values are printed in bold.
fctk,min = fctko,min(fck
fcko
)23 (5.1)
fctk,max = fctko,max(fck
fcko
)23 (5.2)
where
fcko = 10 MPa
fctko,min = 0.95 MPa
fctko,max = 1.85 MPa
Table 5.1: Tensile strengths for various concrete grades [9].
Concrete grade C12 C20 C30 C40 C50
fck 12 20 30 40 50
fctk,min 1.1 1.5 2.0 2.4 2.8
fctk,max 2.1 2.9 3.8 4.7 5.4
Heinrich Stander University of Stellenbosch
5.1. Material models and input data 67
Gf = Gfo(fcm/fcmo)0.7 (5.3)
where
fcmo = 10 MPa
Table 5.2: Fracture energy base values (Gfo) and standard values (GIf ) [9].
dmax [mm] Gfo [N/mm] GIf [N/mm]
8 0.025 0.065
16 0.03 0.075
32 0.058 0.095
Base values of fracture energy depend on the maximum aggregate size (dmax), as shown in Table
5.2. The stone size was 13 mm, which fall between the bold printed GIf values shown in Table
5.2. A parametrical study was conducted, using both the upper and lower bound values for GIf
as shown in Table 5.3. The GIf envelope was extended to ensure that the numerical simulation
accounts for brittle concrete failure. The reasoning will become apparent after the results are
disclosed.
Table 5.3: Parameter values for the concrete material model
E [MPa] ν Tensile curve ft,c [MPa] GIf [N/mm] Compression curve Shear curve
29000 0.2 Exponential 2-3.8 0.05-0.075. Elastic Constant
Table 5.3 shows the parameter values for the concrete material model. Young’s modulus and
Poisson’s ratio were obtained from [9]. It was decided to extend the parametric study and
investigate the effects caused by the tensile strength interval. Such a parametric study is
important in order to determine the influences of parameter values which have a relative high
uncertainty.
5.1.2 Steel material model
A linear elastic material model was used to predict the elastic behaviour of the steel connections
and load cell. The experimental loads did not incur stresses beyond the the elastic regime of
the steel elements. Only two elastic parameter values were necessary, the E-modulus of 200000
GPa and Poisson’s ratio of 0.3, both were extracted from [31].
5.1.3 ECC material model
Refer to subsection 3.2.2 for detailed information about the material model. The parameter
values used in numerical simulations, are depicted in Table 5.4. These values are gathered from
material tests, discussed in subsection 4.3.1.
Heinrich Stander University of Stellenbosch
5.2. Numerical simulation of shear parameter tests 68
Table 5.4: Parameter values for the ECC material model.
σfc [MPa] σtu [MPa] εsh [mm/mm] Ls [mm] E [MPa] ν
2.34 3.44 0.05 6 10610 0.32
5.1.4 Composite interface model
Refer to subsection 3.2.3 for detailed information about this material model. Most of the
parameter values used in this material model were extracted from the experimental results. It
was not possible to measure all the necessary parameter values experimentally. Some of the
values were generated indirectly by applying equations (3.7) and (3.8) on page 27 along with
a numerical sensitivity study. Table 5.5 distinguishes between the experimentally determined
values and the remaining parameters to be calculated indirectly. The values shown in Table
5.5 are that of the sandblasted surface. The indirect methods for determining the unmeasured
parameter values are discussed in the following section.
Table 5.5: Parameter values for the composite interface material model, characterising the SB1410interface.
GAPVAL MO1VAL FRCVAL MO2VAL CAPVAL MOCVAL
ft,i = 0.798 GIf = 0.662 c = 2.73 a = 0 fc Gfc
Φ b = 0.964 Cs κp
Ψ
Φr
σcon,i0
δ
5.2 Numerical simulation of shear parameter tests
The numerical simulation of the shear parameter test, incorporated the total set-up and enforced
the applicable BC’s. Several external factors influenced the eventual experimental results. In
order to gain perspective on the external influences, it was necessary to effectively model all
the contributing components. Two dimensional plane stress elements were used for the simu-
lations and it imposed certain geometrical considerations. The actual set-up connections and
the accompanying load cell were circular units, which entailed the conversion thereof to square
units with the correct moment of inertia. All the connections and the load cell were modelled
as steel elements, with the corresponding material properties. The specimen itself was modelled
using the appropriate material models with properties as discussed in the foregoing subsection.
Certain property values for the composite interface model were incomplete prior to this iterative
stage.
The fact that no experimental tests with a confining force over the interface were executed,
resulted in no physical data existing for the friction coefficient, the residual friction coefficient
and the confining normal stress that causes zero normal uplift. Values for these parameters
Heinrich Stander University of Stellenbosch
5.2. Numerical simulation of shear parameter tests 69
were based on similar research conducted on masonry units. The masonry units consisted of
two brick elements, connected by mortar, which represents the interface. Table 5.6 shows the
parameter values extracted from [41].
Table 5.6: Shear parameter values extracted from masonry tests [41].
Φ Φr σcon,i [MPa]
≈ 1 0.75 -(1 to 2)
These values are applicable to calcium silicate bricks with a mortar connection layer. An or-
dinary mortar mix design results in a weaker material when compared to either the substrate
or overlay used in the research here. This is then the reason for adjusting the σcon,i to (-) five
MPa, because it resulted in the most suitable numerical response.
The continuation of the iterative process resulted in estimates for the remaining two parameter
values namely Ψ0 and δ. The experimental work produced a dilatancy coefficient (Ψ0,exp) for
each of the shear tests. It underestimated the Ψ0 value necessary for the numerical model, be-
cause confining forces existed over the interface at the measured point. Confining forces reduce
the normal uplift, thus decreasing Ψ = up/vp. The numerical iterative process started with
estimates, higher than the measured Ψ0,exp values. The estimated values were altered until
the computed value, at the same material points, expressed a response fitting the experimen-
tal result. During this iterative process, δ was determined using equations (3.7) and (3.8) on
page 27. For each iteration, δ was altered in order to attain suitable response curves. Table
5.7 contains the final parameter values for the numerical model, implemented to simulate the
shear behaviour of the SB1410 specimen. Note that the interfacial compressive inelasticity was
de-activated, by selection of the shown values.
Table 5.7: The final parameter values for the composite interface material model, characterising theSB1410 interface.
GAPVAL MO1VAL FRCVAL MO2VAL CAPVAL MOCVAL
ft,i = 0.798 GIf = 0.662 c = 2.73 a = 0 fc = 1E10 Gfc
= 1E10
Φ = 1 b = 0.964 Cs = 9.0 κp = 1E10
Ψ = 1.2
Φr = 0.75
σcon,i = −5.0
δ = 2.3
5.2.1 Global response
The global response curves for the numerical simulation of the shear test with the properties of
SB1410, correspond very well to the actual experimental results. A comparison of the τxy versus
shear-slip is shown in Figure 5.2 (a). In this comparison the numerical response simulates the
Heinrich Stander University of Stellenbosch
5.2. Numerical simulation of shear parameter tests 70
mean shear strength and elastic deformation with high accuracy. A kink in the elastic part of
the numerical data corresponds to that of the experimental data. It indicates the occurrence of
a crack in the substrate and the influence thereof on the global deformation pattern. The range
of both ft and GIf for the substrate, only influenced the first branch of the shear response. The
main difference between the actual and simulated responses, is observed in the post-fracture
response. The numerical model utilises an exponential degradation coefficient to simulate the
post-fracture behaviour, where the experimental data reveals an almost linear degradation.
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20
1
2
3
4
υ [mm]
τ xy [M
Pa]
EXP1EXP2EXP3EXP4FEA
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20
0.5
1
1.5
2
u [m
m]
υ [mm]
(a)
(b)
Figure 5.2: Numerical results: A comparison of the experimental and numerical shear responses for (a)τxy versus v and (b) u versus v for unreinforced SB1410 specimens.
Figure 5.2 (b) draws a comparison for the normal uplift upon shear-slip. The numerical re-
sponse shows three different stages rather than three gradients as in the experimental case. The
first, elastic part corresponds to a slope of zero angle. The second part starts with a similar
angle to its experimental counterpart, but diminishes as it advances towards an asymptotic
part, which represents the third part of the numerical response. The asymptotic part rep-
resents a phase where no additional dilatancy occurs, just vertical shear slip. This is not the
case for the experimental responses, because the designated BC’s prohibited the overlay to move
Heinrich Stander University of Stellenbosch
5.2. Numerical simulation of shear parameter tests 71
freely away from the substrate and it induced a rotation that results in progressive normal uplift.
An additional comparison reveals the accuracy of the numerical model. Shear stress and normal
uplift for both the experimental and numerical responses are illustrated in Figure 5.3. The only
contradiction is observed in the latter stage of the numerical data where no additional normal
uplift occurs due to its asymptotic nature.
0 0.5 1 1.5 20
0.5
1
1.5
2
2.5
3
3.5
4
utop [mm]
τ xy [M
Pa]
EXP1EXP2EXP3EXP4FEA
Figure 5.3: Numerical result: A comparison of the experimental and numerical responses for τxy versusu of SB1410 specimens.
Contour plots of global shear stress patterns in a deformed state (magnification 20x) are shown in
Figures 5.4 and 5.5 on page 72. Both illustrations show two plots each, that represent a sequence
of the stress state at 20%, 100%, 50% and 20% of the τu, respectively. The τxy contours in all
four plots, are restricted to an envelope between zero and 2.6 MPa. The post-peak contour
plots show an exaggerated normal uplift between the interfaces, due to the deformation factor
chosen for visualisation.
Heinrich Stander University of Stellenbosch
5.2. Numerical simulation of shear parameter tests 72
X
Y
Z
Model: S2Deformation = 20LC1: Load case 1Step: 1 LOAD: .5E-1Gauss EL.SXX.G SXYMax = 2.79Min = -3.47Results shown:Mapped to nodes
0.1.2.3.4.5.6.7.8.911.11.21.31.41.51.61.71.81.922.12.22.32.42.52.6
X
Y
Z
Model: S2Deformation = 20LC1: Load case 1Step: 122 LOAD: .34Gauss EL.SXX.G SXYMax = 15.5Min = -15.3Results shown:Mapped to nodes
0.1.2.3.4.5.6.7.8.911.11.21.31.41.51.61.71.81.922.12.22.32.42.52.6
1 2
22-NOV-2006 12:32 SXY_1_2iDIANA 8.1.2-02 : Univ. Stellenbosch
Figure 5.4: Numerical result: Contour plots of τxy at pre-peak values of 20% and 100%.
X
Y
Z
Model: S2Deformation = 20LC1: Load case 1Step: 166 LOAD: .428Gauss EL.SXX.G SXYMax = 7.47Min = -8.24Results shown:Mapped to nodes
0.1.2.3.4.5.6.7.8.911.11.21.31.41.51.61.71.81.922.12.22.32.42.52.6
X
Y
Z
Model: S2Deformation = 20LC1: Load case 1Step: 276 LOAD: .648Gauss EL.SXX.G SXYMax = 2.92Min = -4.11Results shown:Mapped to nodes
0.1.2.3.4.5.6.7.8.911.11.21.31.41.51.61.71.81.922.12.22.32.42.52.6
1 2
22-NOV-2006 12:36 sxy_3_4iDIANA 8.1.2-02 : Univ. Stellenbosch
Figure 5.5: Numerical results: Contour plots of τxy at post-peak values of 50% and 20%.
Heinrich Stander University of Stellenbosch
5.2. Numerical simulation of shear parameter tests 73
5.2.2 Interfacial response
The interfacial responses are depicted by the stress and strain distributions over the interfacial
surface. Figure 5.6 presents the normal stress distribution over the interface at the four relevant
stages. Both the elastic and peak stages, reveal heightened confining stresses at the top and
bottom of the interface and a uniform state of tensile stress over the rest of the interface. Both
the post-peak responses are relatively uniform around zero.
The second graph (Figure 5.7) demonstrates the normal strain distribution over the interface
for the four representative stages. In the initial stage (20 % of peak shear) a perfect uniform
strain is found across the interface height. Upon increasing the shear load level, a slight strain
gradient is observed. An increasing strain behaviour is illustrated, which corresponds well to
the progressive reduction in confining stresses over the interface, illustrated by the σxx response.
A value of (-) five MPa is conceded to σcon,i0, and the influence thereof is observed when the
relation between the confining stresses and normal strains at the top and bottom of the interface
is investigated at the peak state. In this state the confining stress at the interfacial bottom is
less than σcon,i0, which results in zero normal strain. The confining stress at the top is greater
than σcon,i0, thus the reason for the non-zero strain.
The last interfacial response curves (Figure 5.8), exhibit the shear stress distributions for the
representative stages. The elastic response demonstrates heightened shear peaks as a result of
the confining pressures at the outer limits. The response curve at peak shear is complex, where
more than one factor contributes to the response. The role of confining pressures diminishes to
an extent, cf. equations (3.7) and (3.8) on page 27, because shear-slip initiates here. Both the
post-peak responses are inverse in distribution to that of the elastic behaviour, because relatively
little normal stresses exist over the interface and interfacial shear-slip is at a progressive stage.
−16 −14 −12 −10 −8 −6 −4 −2 00
10
20
30
40
50
60
70
80
90
100
σxx
[MPa]
Ver
tical
Pos
ition
alo
ng In
terf
ace
[mm
]
ElasticPeakPost−peakAdvanced
Figure 5.6: Numerical result: The interfacial σxx at four stages of the shear test on a SB1410 specimen.
Heinrich Stander University of Stellenbosch
5.3. Numerical simulation of tensile parameter tests 74
−0.45 −0.4 −0.35 −0.3 −0.25 −0.2 −0.15 −0.1 −0.05 0 0.050
10
20
30
40
50
60
70
80
90
100
εxx
[mm/mm]
Ver
tical
Pos
ition
alo
ng In
terf
ace
[mm
] ElasticPeakPost−peakAdvanced
Figure 5.7: Numerical result: The interfacial εxx at four stages of the shear test on a SB1410 specimen.
0 0.5 1 1.5 2 2.5 30
10
20
30
40
50
60
70
80
90
100
τxy
[MPa]
Ver
tical
Pos
ition
alo
ng In
terf
ace
[mm
]
ElasticPeakPost−peakAdvanced
Figure 5.8: Numerical result: The interfacial τxy at four stages of the shear test on a SB1410 specimen.
5.3 Numerical simulation of tensile parameter tests
The tensile parameter test was modelled without the experimental set-up. This implies the
assumption of zero eccentricity. The model consisted of two material parts connected by the
interfacial material model. A steel platen was attached to the top ECC part, in order to apply
the displacement controlled load indirectly to the model. The parameter values used for this
model were gained from the foregoing shearing simulations. Table 5.7 on page 69 summarises
the values.
Several numerical results were produced by the simulations. The most important result, de-
picts the applied load against the interfacial normal opening. A graphical illustration of this
result follows in Figure 5.9. It conveys three experimental data sets and the corresponding
finite element analysis (FEA) result. The numerical response curve represents the average ft,i
Heinrich Stander University of Stellenbosch
5.3. Numerical simulation of tensile parameter tests 75
exactly, but differs from the experimental data in the post-peak region. Numerical post-fracture
toughness is modelled with an exponential function, which explains the numerical softening re-
sponse. The actual post-fracture toughness for the specific SSPT, is achieved by fibres bridging
a newly defined fracture surface. The numerical model can not simulate this response, but has
to portray the given GIf in its response. A better fit of the post-peak response can be found by
selection of another softening model, but generally at the cost of computational ease.
0 0.5 1 1.5 2 2.5 3 3.5 40
0.2
0.4
0.6
0.8
1
u [mm]
σ t [MP
a]
EXP1EXP2EXP3FEA
Figure 5.9: Numerical result: A comparison of the experimental and numerical tensile responses forft,i versus u of SB1410 specimens.
Figures 5.10 and 5.11 represent interfacial stress and strain values at the peak load and at
one mm interfacial displacement respectively. In both cases, the stress distribution over the
interface is uniform. The interfacial strain at peak load is higher on the left side, depicting
that fracture starts here because of an artificially induced imperfection. The imperfection is
achieved by reducing the size of the last interface element on the left side by 0.5% to avoid
simultaneous inelasticity in the whole interface. It eventually equalises and becomes uniform
over the interface in the post-fracture region, as indicated by Figure 5.11 (b).
Heinrich Stander University of Stellenbosch
5.4. Summary 76
0 20 40 60 80 100 1200
0.5
1
1.5
Horisontal position along interface [mm]
σ yy,i [M
Pa]
(a)
0 20 40 60 80 100 1200
0.005
0.01
0.015
Horisontal position along interface [mm]
ε yy,i [m
m/m
m]
(b)
Figure 5.10: Numerical results: Peak σyy,i and εyy,i values for a direct uniaxial tensile test on a SB1410specimen.
0 20 40 60 80 100 1200
0.5
1
1.5
Horisontal position along interface [mm]
σ yy,i [M
Pa]
(a)
0 20 40 60 80 100 1200
0.5
1
1.5
Horisontal position along interface [mm]
ε yy,i [m
m/m
m]
(b)
Figure 5.11: Numerical results: Post-peak σyy,i and εyy,i values for direct uniaxial tensile test on aSB1410 specimen.
5.4 Summary
The comparison of numerical and experimental results for both the shear and tensile parameter
tests, shows reasonable agreement. It is an indication of reasonable parameters, determined as
outlined. All mechanisms that contributed to the experimental behaviour were captured in the
numerical simulations. Based on the reasonable agreement, a predictive capacity exists.
The analyses were performed for sandblasted specimens. Numerical simulation of the other
preparation types and the determination of their outstanding parameters are to be obtained
through a repeat of the process followed in this chapter. Several parameters are directly acquired
from the test data and the others are fitted by indirect analysis. A summary of the parameter
values for all the other SSPT’s, after an initial round of computing, are shown in Table 5.8. The
values for SB1410 are used as basis for the determination of the remaining values. The correct
Φ for each preparation technique, is to be determined by future confined shear tests.
Heinrich Stander University of Stellenbosch
5.4. Summary 77
Table 5.8: The outstanding parameter values characterising the remaining SSPT’s, after an initialround of computing.
SSPT Rebar Φ0,exp vp,exp up,exp Φ0 δ vp Φ0,end up
SB1410 - 0.98 0.8 0.5 1.2 2.3 0.8 0.19 0.44
SB1410 X 1.11 0.92 1.08 1.31 2.1 0.92 0.19 0.53
SB2810 - 1.53 1.25 2.2 1.73 1.75 1.25 0.19 0.88
SB724 - 0.46 0.55 0.31 0.66 2.3 0.55 0.19 0.21
SB1424 - 1.08 0.61 0.7 1.28 3.1 0.61 0.19 0.35
SB2824 - 1.25 0.59 1.38 1.45 3.45 0.59 0.19 0.37
S1410 - 1.05 0.9 1.5 1.25 2.1 0.9 0.19 0.50
S1424 - 0.74 0.97 1.05 0.94 1.65 0.97 0.19 0.46
DH1424 - 1.53 0.54 0.83 1.73 4.1 0.54 0.19 0.38
PG1424 - 1.93 1.1 2.25 2.14 2.2 1.1 0.19 0.89
R1410 - 1.7 0.24 0.45 1.9 9.6 0.24 0.19 0.18
R2810 - 0.59 1.77 1.5 0.79 0.8 1.77 0.19 0.75
R2824 - 1.3 0.5 1.05 1.5 4.15 0.5 0.19 0.31
Heinrich Stander University of Stellenbosch
Chapter 6
APPLICATION OF A THIN
BONDED OVERLAY
6.1 EXPERIMENTAL PROCEDURE
6.1.1 Introduction
An experiment was designed to test the accuracy of the numerical interfacial model as well as
the performance of a thin bonded overlay. A successful numerical validation of the experimen-
tal results would substantiate the interfacial parameter values used for the applied SSPT. The
experimental set-up was designed to be simulated by a two-dimensional numerical model. The
computational material models available, inflicted this limitation on the experiment. Thus it
was decided to solve a two-dimensional problem, for instance a repair strategy for an industrial
floor under point loads.
Cracking and spalling of the top layer of industrial concrete floors is a problem that occur
commonly. In addition, imposed loads such as wheel loads of forklifts and other utility vehicles
may induce increased levels of tension, and they are abrasive because of high frictional forces.
A material such as ECC, that exhibits high fracture toughness and crack width control, is a
possible solution. In order to address this problem, the top layer of concrete needs to be re-
placed or covered by ECC. The bond between the materials has to function in such a way that
composite action occurs without interfacial delamination. It is important that the composite
acts as one and not as separate layers.
The experiment was constructed to test the interfacial bond between the materials by means
of a static load in order to assess the performance under heightened deflections. Note that an
extreme case is considered, with no support in the span length, and no rotational constraint
at the ends. The substrate does not contribute structurally to the composite because of an
artificial crack, as in the case of a damaged substrate. A repair strategy was thus tested in
which the substrate restrained the movement of the overlay because of its rigidity, enforced by
the interfacial bond. Thus, a simply executed three-point bending case was studied.
78
6.1. EXPERIMENTAL PROCEDURE 79
6.1.2 Experimental programme
6.1.2.1 Specimen preparation
Concrete beams were produced in moulds, measuring 500x100x100 mm. The class of concrete
namely C30, was used for the substrate and it is identical to the concrete implemented for the
interfacial experiments. After casting the substrate specimens, a 24h setting period was allowed
before demoulding occurred. Water curing commenced in a tank, at 23◦C for the next two
days. Application of the roughening techniques occurred next, after the specimens were taken
from the curing tanks. Prior to the respective roughening applications, each 500x100x100 mm
beam was cut in half to create two 500x100x50 mm beams, i.e. half the height of the original
beam. The cutting process was performed with a blade saw as shown in Figure 6.1. The newly
shaped beam represented the substrate, as used for the composite member. Again the substrate
was cut, but at half its length. This extra cut induced an artificial crack at the middle, right
through the substrate layer.
(a) (b)
Figure 6.1: (a) Sawing of concrete substrate and (b) casting of beam overlay.
Two substrate surface roughening techniques were applied on the saw surface, namely scraping
and sandblasting. After the roughening occurred, a 24 hour period passed during which the
specimen was exposed to room temperature. On both sides of the substrate “crack”, insulation
tape was wrapped around for 25 millimetres, inducing an artificial delamination on the interface.
A 10 minute moistening period was applied for both roughening techniques, after which casting
of the overlay followed. The same ECC mix was used, as in Table 4.2. The ECC overlay was
50 mm thick, having been cast in the 500x100x100 mm steel moulds, on top of the prepared
concrete substrate.
The freshly cast ECC was given a setting time of 3 days, after which the specimens were
stripped from their moulds. All the specimens were covered with wet burlap for the duration
of the setting stage. Water curing at 23◦C followed for the next 11 days until testing occurred
at an ECC age of 14 days. Testing was conducted on wet specimens, directly from the curing
Heinrich Stander University of Stellenbosch
6.1. EXPERIMENTAL PROCEDURE 80
tanks, in order to minimise specimen shrinkage. Tests were conducted in a climate controlled
room with a constant temperature of 23◦C and a relative humidity of ± 65%.
6.1.2.2 Test set-up
A set-up for the three point bending test was constructed in a Zwick Z250 Universal Testing
Machine. A displacement based load was applied over the artificial crack at a rate of 3 mm/min.
Figure 6.2 illustrates the geometry of the composite structure with its artificial crack and delam-
inated interface. A total of eight LVDT’s were used to measure delamination over the interface
and a Spider 8 logging system was utilised to capture the data. Data concerning the deflection
and applied load were gathered from the Zwick data logger. Both sets of data were correlated,
using a calibrated time line. Figure 6.3 illustrates the experimental set-up.
500
450
50
3.5
50
50
100
Figure 6.2: Geometrical aspects of the composite beam specimens (dimensions in mm).
Figure 6.3: Experimental set-up for the three-point bending test.
6.1.3 Experimental results
A total of four tests were executed, two for each of the two applied roughening techniques.
Both the sandblasted and scraped roughening methods, arrested the development of interfa-
cial delamination, i.e. propagation of the artificial induced delamination was restrained. The
Heinrich Stander University of Stellenbosch
6.1. EXPERIMENTAL PROCEDURE 81
fact that the weaker of the bonds also succeeded in preventing delamination, with similar re-
sults to that of the sandblasted bond, indicates that it is not necessary to use the strongest
possible bond. In this particular boundary value problem, a high frictional resistance is induced.
Eliminating delamination adds a constraint to the ductile movement of ECC. Allowance of some
delamination and no delamination could result in greater global ductility. Depending on per-
formance requirements, the bond can be altered to meet specifications.
Only one specimen of each SSRT type was fitted with the array of LVDT’s. The experimental
global results in terms of total force versus mid-span deflection for all four tests, remained in
a narrow response envelope. Such a result offers high design reliability, a crucial feature when
implementing new materials and repair strategies in structural design.
The response curves for nominal displacement are plotted against the applied force in Figure
6.4. The behaviour is one of deflection hardening, followed by a long tail, representative of
deflection softening. All four tests had similar strength and deflection results as shown in the
graph.
0 5 10 15 20 25 300
500
1000
1500
2000
2500
3000
3500
Vertical displacement [mm]
For
ce [N
]
Sandblasted 1Sandblasted 2Scraped 1Scraped 2
Figure 6.4: Experimental result: The applied force versus nominal displacement for both SB1410 andS1410 in three-point bending.
The data extracted from the 8 LVDT’s, fitted and spaced at four measuring positions on either
side of the specimen, is demonstrated in Figure 6.5. The LVDT’s were employed to log inter-
facial delamination at two points on either side of the artificial crack. The inner two measure
points were at the end of the artificial delaminated areas on both sides. The remaining outer
two measuring points were spaced a further 25 mm away from the inner positions. Figure 6.5
(a) depicts the response for SB1410. A larger interfacial separation/delamination is measured
on the right hand side of the specimen. Only the inner measuring points convey interfacial de-
lamination and little thereof. Figure 6.5 (b) reveals the response for S1410, with similar results
on either side of the notch. Negligible delamination is observed at the outer measuring points.
Heinrich Stander University of Stellenbosch
6.2. NUMERICAL VALIDATION 82
Figure 6.6 demonstrates the formation of a localised crack, directly above the substrate notch.
This particular crack leads to ultimate failure of the overlay system.
0 0.01 0.02 0.03 0.04 0.050
500
1000
1500
2000
2500
3000
3500
Interfacial delamination [mm]
For
ce [N
]
Outer leftInner leftInner rightOuter right
(a)
0 0.01 0.02 0.03 0.04 0.050
500
1000
1500
2000
2500
3000
3500
Interfacial delamination [mm]
For
ce [N
]
Outer leftInner leftInner rightOuter right
(b)
Figure 6.5: Experimental results: Interfacial delamination responses for three-point bending tests on(a) SB1410 and (b) S1410.
Figure 6.6: The formation of a localised crack in the middle of the bonded overlay of a SB1410 specimen.
6.2 NUMERICAL VALIDATION
6.2.1 Numerical model
The numerical model consisted of three material models and four node plane stress finite el-
ements. The Total Strain Rotating Crack, Composite Interface and the user-supplied ECC
models were implemented to simulate the behaviour of the concrete, interface and ECC respec-
tively. The fact that the ECC material model requires the element size to be restricted to the
crack spacing, resulted in a very fine mesh of 1.3 mm square elements in the middle region
over the notch area. Away from this region, the mesh size was increased in order to minimise
computational time. A numerical simulation was performed for the sandblasted interface, as
Heinrich Stander University of Stellenbosch
6.2. NUMERICAL VALIDATION 83
reported in section 6.1.
A sensitivity study was performed, varying the σfc and εsh values in a narrow band around
the applicable ECC uniaxial tensile test values. The round of direct uniaxial tensile tests was
omitted for this batch of ECC and the previously reported values of Chapter 4 were used. The
same ECC mix design was used for this part of the experimental programme and its material
properties are well known. Table 6.1 indicates the applied ECC material properties. The
sensitivity analyses were performed using the outer bound values of both σfc and εsh in four
simulations and a last one with the mean values. The fifth and last simulation is also the
one illustrated in the contour plots. Table 6.2 depicts the sensitivity values employed in each
analysis.
Table 6.1: ECC material model properties for the three-point bending tests.
σfc [MPa] σtu [MPa] εsh [mm/mm] Ls [mm] E [MPa] ν
2.3-2.54 3.44 0.045-0.055 6 10610 0.32
Table 6.2: The σfc and εsh values used for each analysis.
FEA 1 FEA 2 FEA 3 FEA 4 FEA 5
σfc 2.30 2.54 2.30 2.54 2.42
εsh 0.055 0.045 0.045 0.055 0.05
Heinrich Stander University of Stellenbosch
6.2. NUMERICAL VALIDATION 84
6.2.2 Numerical results
The influences of both the σfc and εsh on the bending behaviour of ECC and the composite
member, are clearly illustrated in Figure 6.7 (a). A higher first crack strength results in stiffer
behaviour for the deflection hardening period and an earlier peak strength, as observed for
FEA 2. A lower strain-hardening value corresponds also to an earlier peak strength as well as
less overall deflection, when comparing FEA 3 and FEA 1. Figure 6.7 (b) illustrates the in-
terfacial delamination at the inner measuring points and the ECC parameter sensitivity thereof.
0 5 10 15 20 25 300
500
1000
1500
2000
2500
3000
3500
Vertical displacement [mm]
For
ce [N
]
FEA 1FEA 2FEA 3FEA 4FEA 5
(a)
0 0.01 0.02 0.03 0.04 0.050
500
1000
1500
2000
2500
3000
3500
Interfacial delamination [mm]
For
ce [N
]
FEA 1FEA 2FEA 3FEA 4FEA 5
(b)
Figure 6.7: Numerical results: Three-point bending response curves demonstrating (a) the appliedforce versus deflection and (b) interfacial delamination at the inner points of a SB1410 specimen.
A comparison of the experimental and numerical results for SB1410 follows in Figure 6.8. In
terms of global response, the numerical response with the higher σfc and lower εsh values, i.e.
FEA 2, simulates the experimental response the best. Nevertheless the computed response is
relatively insensitive to the variation in these parameters. As for the analyses of Chapter 5, the
softening part of the response is computed less accurately, due to choice of smooth, exponential
softening functions.
The graph showing interfacial delamination responses only compares the inner measuring points
and not the curves of the outer points. The reason is that the numerical simulation did not
demonstrate any delamination at the outer points, corresponding to insignificant values logged
during the experiments. All of the numerical response curves, depicting the delamination at the
inner positions, fitted into the experimental envelope created by the difference in measurements
on the left and right-hand side of the notch. FEA 2, FEA 4 and FEA 5 correspond well to the
experimental mean response curve.
Heinrich Stander University of Stellenbosch
6.2. NUMERICAL VALIDATION 85
0 5 10 15 20 25 300
500
1000
1500
2000
2500
3000
3500
Vertical displacement [mm]
For
ce [N
]
EXP 1EXP 2FEA 1FEA 2FEA 3FEA 4FEA 5
(a)
0 0.01 0.02 0.03 0.04 0.050
500
1000
1500
2000
2500
3000
3500
Interfacial delamination [mm]
For
ce [N
]
EXP leftEXP rightEXP meanFEA 1FEA 2FEA 3FEA 4FEA 5
(b)
Figure 6.8: A comparison of experimental and numerical data, demonstrating (a) the applied forceversus deflection and (b) interfacial delamination at the inner points of a SB1410 specimen.
Contour plots were drawn to illustrate the deflection, principle stress, principle strain and
equivalent strain of the composite member at 50% (pre-peak), 100% and 50% (post-peak) of
the ultimate load. The contours of the principle stress plots are confined to an envelope ranging
from 0 to 3.44 MPa. Figures 6.9 and 6.10 demonstrate the contour plots of σ1 and ε1 respectively.
Figure 6.9 illustrates the state of σ1 in the composite member and also the area in which
structural mobility occurs. The strong interfacial bond between the ductile overlay and rigid
substrate, restricts the ductile behaviour of ECC. The rigid behaviour of the composite, outside
the delaminated zone, incurs stress on the substrate, as depicted by the graph. The ductility
of only a relatively small part of ECC outside the artificial delaminated zone, is activated.
Activating the ductility of a larger part of ECC is the key factor towards achieving a composite
with better ductile behaviour. Figure 6.10 indicates strain concentration points at 50% of the
peak load, which are visible at the ends of the artificially induced delamination. Localisation
of the material points, directly above the substrate notch, follows as indicated at peak load.
Heinrich Stander University of Stellenbosch
6.2. NUMERICAL VALIDATION 86
X
Y
Z
Model: 5Deformation = 1LC1: Load case 1Step: 41 LOAD: 1.56Gauss EL.S1... S1Max = 2.62Min = -8.26Results shown:Mapped to nodes
0.132.265.397.529.662.794.9261.061.191.321.461.591.721.851.982.122.252.382.512.652.782.913.043.183.313.44
X
Y
Z
Model: 5Deformation = 1LC1: Load case 1Step: 157 LOAD: 3.14Gauss EL.S1... S1Max = 3.44Min = -26.4Results shown:Mapped to nodes
0.132.265.397.529.662.794.9261.061.191.321.461.591.721.851.982.122.252.382.512.652.782.913.043.183.313.44
1
2
10-DEC-2006 12:08 s1iDIANA 8.1.2-02 : Univ. Stellenbosch
Figure 6.9: Numerical result: Contour plots of the σ1 at 50% (pre-peak) and 100% of the peak load.
X
Y
Z
Model: 5Deformation = 1LC1: Load case 1Step: 41 LOAD: 1.56Gauss EL.E1... E1Max = .903E-2Min = -.327E-3Results shown:Mapped to nodes
.326E-4
.393E-3
.753E-3
.111E-2
.147E-2
.183E-2
.219E-2
.255E-2
.291E-2
.327E-2
.363E-2
.399E-2
.435E-2
.471E-2
.507E-2
.543E-2
.579E-2
.615E-2
.651E-2
.687E-2
.723E-2
.759E-2
.795E-2
.831E-2
.867E-2
X
Y
Z
Model: 5Deformation = 1LC1: Load case 1Step: 163 LOAD: 2.74Gauss EL.E1... E1Max = 2.11Min = -.287E-4Results shown:Mapped to nodes
.813E-1
.163
.244
.325
.407
.488
.569
.65
.732
.813
.894
.9761.061.141.221.31.381.461.541.631.711.791.871.952.03
1
2
24-NOV-2006 09:29 e1iDIANA 8.1.2-02 : Univ. Stellenbosch
Figure 6.10: Numerical results: Contour plots of the ε1 at 50% (pre-peak) and 100% of the peak load.
Heinrich Stander University of Stellenbosch
6.2. NUMERICAL VALIDATION 87
In order to make an assessment on the amount of damage present, a contour plot of the equiva-
lent strain is required. The envelope of the contours must start with the value of εfc, this then
illustrates the areas where damage has occurred. Depending on the end value of the envelope,
an assessment on the severity of damage can be made. Figure 6.11 presents the amount of dam-
age in the overlay at 50% (pre-peak) and 100% of the peak load. The inception of damage, has
already occurred prior to the 50% load. The stress concentrations at the ends of the artificial
delamination, have incurred damage in both areas. Experimentally, the damage spreads in the
form of multiple cracks, as the load increases to its maximum. Numerically, the damage spreads
to neighbouring elements and covers an equivalent area in which multiple cracking would be
present physically, as is the case at peak load. The extent of damage progresses until localisation
occurs.
Localisation arises when the strain at a material point becomes equal to the value of (εfc +εsh).
Exceeding this strain value, results in the further widening of a single crack, which is detrimental
to the system. Figure 6.12 illustrates areas of localisation at 100% and 50% (post-peak) of the
peak load. At peak load, three areas have progressed beyond the strain-hardening stage. The
amount of localisation at the edges of the artificial delamination is small in comparison to the
middle part. It is this localisation, in the middle of the composite that continues further to
result in a single crack, which leads to ultimate failure.
X
Y
Z
Model: 5Deformation = 1LC1: Load case 1Step: 41 LOAD: 1.56Gauss EQUIEPSMax = .102E-1Min = -.836E-3Results shown:Mapped to nodes
.228E-3
.307E-3
.386E-3
.465E-3
.544E-3
.623E-3
.702E-3
.78E-3
.859E-3
.938E-3
.102E-2
.11E-2
.118E-2
.125E-2
.133E-2
.141E-2
.149E-2
.157E-2
.165E-2
.173E-2
.181E-2
.189E-2
.196E-2
.204E-2
.212E-2
.22E-2
.228E-2
X
Y
Z
Model: 5Deformation = 1LC1: Load case 1Step: 157 LOAD: 3.14Gauss EQUIEPSMax = .959Min = -.285E-2Results shown:Mapped to nodes
.228E-3
.11E-2
.196E-2
.283E-2
.37E-2
.457E-2
.544E-2
.631E-2
.717E-2
.804E-2
.891E-2
.978E-2
.106E-1
.115E-1
.124E-1
.133E-1
.141E-1
.15E-1
.159E-1
.167E-1
.176E-1
.185E-1
.193E-1
.202E-1
.211E-1
.219E-1
.228E-1
1
2
10-DEC-2006 13:04 EQ1iDIANA 8.1.2-02 : Univ. Stellenbosch
Figure 6.11: Numerical results: Contour plots of the equivalent strain at 50% (pre-peak) and 100% ofthe peak load.
Heinrich Stander University of Stellenbosch
6.3. APPLICATION COMPARISON 88
X
Y
Z
Model: 5Deformation = 1LC1: Load case 1Step: 175 LOAD: 1.52Gauss EQUIEPSMax = 5.02Min = -.24E-2Results shown:Mapped to nodes
.502E-1
.241
.433
.624
.8151.011.21.391.581.771.962.152.342.542.732.923.113.33.493.683.874.064.264.454.644.835.02
X
Y
Z
Model: 5Deformation = 1LC1: Load case 1Step: 157 LOAD: 3.14Gauss EQUIEPSMax = .959Min = -.285E-2Results shown:Mapped to nodes
.502E-1
.852E-1
.12
.155
.19
.225
.26
.295
.33
.365
.4
.435
.47
.505
.54
.575
.609
.644
.679
.714
.749
.784
.819
.854
.889
.924
.959
1
2
10-DEC-2006 13:17 eqlocaliDIANA 8.1.2-02 : Univ. Stellenbosch
Figure 6.12: Numerical results: Contour plots of the equivalent strain at 100% and 50% (post-peak)of the peak load, showing the localised areas.
6.3 APPLICATION COMPARISON
A numerical comparison of different SSPT’s and application strategies leads to informative
decision making when employing such a repair or retrofitting strategy in the field. This sec-
tion reports on numerical comparisons between concrete and ECC overlays as well as different
SSPT’s. A comparison between two application methods are made, i.e. a bonded overlay system
with and without an artificial delaminated area respectively. The latter was tested experimen-
tally and validated numerically for a SB1410 specimen in the foregoing two sections. Another
comparison is made to illustrate the difference between an overlay with a higher E-modulus,
the 28 day ECC and its younger, 14 day old counterpart.
Figure 6.13 illustrates the various numerical responses. A Total Strain Rotating Crack material
model with parameter values of the class C30 concrete (Table 5.3) was implemented to simulate
the behaviour of a concrete overlay. The interfacial parameters of the SB2810 specimen, was
used for this simulation. The eventual result indicates a poor response for such a concrete to
concrete system when compared to the ECC overlay systems. The concrete overlay generates
stiffer composite action than the ECC overlays, because its E-modulus is higher, but a weaker
response is also observed. The associated brittleness of concrete is responsible for the weaker
composite action.
Heinrich Stander University of Stellenbosch
6.3. APPLICATION COMPARISON 89
The numerical simulation of the SB2810 overlay system, produced a stiffer and stronger re-
sponse than the SB1410 overlay. The increased stiffness is due to the effect of the higher ECC
E-modulus. The stronger response is a result of the higher σfc and σtu values.
Also shown in Figure 6.13 is the numerical response of a single ECC beam with the same di-
mensions as the overlay, i.e. 500x100x50 mm. This particular analysis presents the three point
bending response of an ECC beam, without the substrate, or with the substrate fully debonded.
Note that in this boundary value problem, the substrate does not contribute to the structural
strength. This is due to the notch, dividing the substrate in order to simulate a crack and the
localised failure in the composite at this section. Depending on the strength of the interfacial
bond, the ductile movement of the overlay is restricted by the substrate. Thus when comparing
the three point bending response of a solitary ECC beam with that of the composite system,
similar peak resistance may be anticipated but the amount of ductility will vary. However,
due to strong bond, the ECC softening energy dissipation is restricted away from the substrate
crack, actually reducing the load bearing capacity.
The numerical response of the bonded overlay, with the strong SB1410 interfacial bond and
without an artificial delaminated area, results in an earlier peak. Compared to the solitary
ECC beam, the load carrying capacity diminishes fast with increasing deflection. Applying
the SB1410 SSPT in this fashion, restricts the ductile behaviour of ECC, but increases system
stiffness to values equivalent of the SB2810 bonded system.
The introduction of an artificial delamination as part of the application method, produces a
more ductile system. This is because less restriction on the movement of ECC is implied through
the artificial delaminated area. The response of this application method is similar to the re-
sponse of the solitary ECC beam, thus harnessing the ductile behaviour of ECC better than
the system without an artificial delamination.
Another method, not illustrated here, is the application of a weaker bond, which allows greater
delamination. Such an application may be easily studied with the supplied interfacial parameter
values of other tested SSPT’s.
Heinrich Stander University of Stellenbosch
6.3. APPLICATION COMPARISON 90
0 2 4 6 8 10 12 140
1000
2000
3000
4000
Vertical displacement [mm]
For
ce [N
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
1000
2000
3000
4000
Vertical displacement [mm]
For
ce [N
]
ECCSB1410SB1410*SB2810Concrete
Figure 6.13: Numerical results: Comparison of three point bending responses (* fully bonded).
Table 6.3 presents a peak load and deflection comparison for different three point bending
systems. It shows that the SB2810 specimen achieved the highest peak load and accompanied
deflection. The C30 overlay system achieved the lowest peak load and exhibited brittleness.
The comparison between the fully bonded and partially debonded SB1410 systems, indicates
similar peak loads (partially debonded is slightly higher). The associated deflection at peak
load for the fully bonded system, is 41% lower than the partially debonded system.
Table 6.3: A comparison of peak loads and deflections for three point beam tests.
Repair materials Type Delamination Method Peak load [N] Deflection at fu [mm]
Concrete C30 X NUM 1212.31 0.075
ECC S1410 X EXP 3069.36 4.806
SB1410 X EXP 3110.15 6.378
X NUM 3135.25 6.9
- NUM 3042.17 4.047
SB2810 X NUM 3638.18 7.68
Heinrich Stander University of Stellenbosch
6.4. SUMMARY 91
6.4 SUMMARY
Sufficient agreement between the computational and experimental results was found. This
agreement substantiates the implementation of numerical simulations, to investigate applica-
tion methods and their outcomes. Optimising the method in which a SSPT is applied, can lead
to improved behaviour of the composite system. Inducing enough room for overlay movement,
i.e. allowing the ECC to exhibit multiple cracking, results in improved system ductility and
energy dissipation. Improved ductility can either be instigated by supplying an artificial delam-
ination along with a strong bond or by using a weaker bond. Depending on the application, the
progress of the artificial delamination can be arrested by the strong bond. The application of a
weaker bond may lead to partial delamination, which supplies the necessary space for overlay
movement.
Figures 6.4 and 6.13 show a good outcome in that the superior behaviour of ECC is enabled.
Of course, if the substrate is not completely cracked through, it will contribute structurally.
Omitting the artificial crack would have resulted in a much stronger and tougher resistance
until the total fracture of the concrete substrate occurred at very low deflection values. From
there onwards, a similar response will yield as in the case for the specimen with an artificial
crack in the substrate.
Several methods can be employed in order to induce improved behaviour for an application
through a“weakened” bond. All these methods, drill holes, scarifying, etc. are still to be tested
experimentally. Another method, a modified version of the artificial delamination, supplies a
discrete bond. Stripe sandblasting or spaced shear connectors in combination with a de-bonding
agent are possible options for obtaining a discrete bond. The ultimate objective is to acquire
maximum ductility and composite stiffness, through supplying the required interfacial freedom,
necessary for the multiple cracking phenomenon to occur in ECC.
Heinrich Stander University of Stellenbosch
Chapter 7
CONCLUSIONS AND
RECOMMENDATIONS
The main aim of this research was to characterise certain SSPT’s for composite elements con-
sisting of ordinary concrete (the substrate) and ECC (the overlay). The applicable preparation
types were chosen in such a way, so that the application thereof can be executed with relative
ease and conformity. Characterisation of interfacial parameters required a rigourous test phase,
for which shear and tensile test methods were developed with the assistance of computational
tools. Composing a numerical design tool was another envisaged aim. Such a design tool is
valuable to study and design structural applications such as overlay strategies and composite
(R/C-ECC) elements.
In the end, the chosen preparation types showed much promise. Strong interfacial bonds were
achieved, in some instances the specimens fractured in other areas and not at the interface.
The experimental designs were successful in acquiring much needed parameter data, which
were simulated to reasonable accuracy by the employed computational model. The eventual
experimental and numerical work on a thin bonded overlay strategy led to validation of the
foregoing parameter results.
7.1 CONCLUSIONS
7.1.1 Substrate surface preparation
The interfacial experimental program successfully characterise the interfacial properties of var-
ious substrate surface preparations.
Mechanical roughening of the substrate leads to stronger interfacial bonds. Physical alteration
increases the specific area and thus the effective interfacial bond surface. Sandblasting of the
substrate, induces the highest specific area when compared to the other techniques implemented
in this research, which is reflected via the obtained bond strengths.
92
7.1. CONCLUSIONS 93
The degree of substrate saturation is another factor affecting the obtained bond strength. The
stronger interfacial bond achieved by the shorter moistening period, is indicative of a more
optimal degree of saturation when compared to the longer moistening period.
The duration of practical application gives the 10 minute SSMT the advantage of short execution
periods, but the method demands more attention from the worker because of higher strength
variability when compared to the longer SSMT. Variability in strength is the result of the shorter
SSMT’s sensitivity to the moistening duration. The 24 hour procedure induces uniformity and
simplicity to practical execution.
7.1.2 Interfacial bond
Strength, COV and fracture toughness are important when judging the results from a structural
perspective, because brittle failure of structural elements must be avoided as best possible. An
important aspect of all the SSPT’s is to enlarge the specific surface area of the substrate, to
which an overlay can adhere.
Physical alteration of the substrate surface induces either stronger chemical and/or mechanical
interfacial bonds. Scraping, although a mechanical process, increases the chemical bond and
results in a stronger shear bond than that of the drill holes and precast grooves, which depend
on mechanical interlocking. The strongest bond is achieved by sandblasting, which addresses
both the chemical and mechanical bond.
Alteration of the substrate surface saturation levels results in a stronger interfacial bond due to
the amount of capillary suction which influences the obtained effective interfacial contact surface.
Addressing only the chemical bond through scraping, results in low GIf . Post-fracture behaviour
is well addressed by the sandblasting option, especially in tension, because the ECC fibres come
into play. Both the increased specific surface area and degree of capillary suction present on
the sandblasted surface, cause a strong interfacial bond with a weaker area in the ECC next to
the interfacial contact zone, which leads to fracture in the overlay when in tension. The result
is a tough post-fracture behaviour controlled by the fibres.
Inducing a strong enough interfacial bond through sandblasting, results in interfacial fracture
slightly removed from the original interface, occurring predominantly in the ECC overlay. This
corresponds to research [42] on concrete that has shown that fracture surfaces generally exist
not directly at the physical boundary between the aggregate and matrix. Fracture normally
occurs slightly removed from the interface, in the porous transition zone [4].
7.1.3 Interfacial shear test refinement
The use of numerical models to investigate interfacial shear test methods, supplies the required
information in order to draw objective conclusions regarding the geometry and boundary con-
Heinrich Stander University of Stellenbosch
7.1. CONCLUSIONS 94
ditions of the considered method.
The F-F BC results in higher interfacial shear capacities than the rotation enabled set-ups.
Two different scenarios affecting the interfacial stiffness are present here, both a result of the
applicable BC. The first scenario is explained by the F-F BC, by which the platens constrain the
specimen from rotation and interfacial dilatancy, adding to the stiffness. The second scenario is
explained by both rotational set-ups, which allows interfacial rotation. The rotational freedom
lessens the compressive normal forces over the interface and increases the tensile normal forces.
Lesser confining forces, lower the interfacial stiffness and strength.
The push-off shear test induces heightened tensile stresses in the back regions of both materials.
The brittle nature of concrete initiates the development of a single crack in this region. Both
the rotation enabled set-ups result in substrate crack development at lower loads than the F-F
system. The F-R and R-R BC’s are not well suited for the geometry at hand, without steel
reinforcement. The addition of steel reinforcement to this area, prevents substrate cracks and
results in stronger interfacial shear bonds. A stronger shear bond is achieved by preventing the
crack which lowers the system stiffness and causes interfacial rotation.
7.1.4 Numerical modelling of parameter tests
The comparison of numerical and experimental results for both the shear and tensile parameter
tests, shows reasonable agreement. It is an indication of reasonable parameters, determined as
outlined. All mechanisms that contributed to the experimental behaviour were captured in the
numerical simulations. Based on the reasonable agreement, a predictive capacity exists.
In this thesis unconfined shear tests are performed, ruling out accurate, direct determination of
the friction parameters (Φ, Φr and a). Inaccuracies exist in the composite interface model due
to insufficient parameter values.
The experimental work produced a dilatancy coefficient (Ψ0,exp) for each of the shear tests.
It underestimated the Ψ0 value necessary for the numerical model, because confining forces
existed over the interface at the measured point. Confining forces reduce the normal uplift,
thus decreasing Ψ = up/vp. Contradiction is observed in the latter stage of the numerical data,
where no additional normal uplift is computed due to the model’s asymptotic nature, while the
measurements show continued dilatation.
Numerical post-fracture toughness is modelled with an exponential function, which explains the
numerical softening response. The actual post-fracture toughness response for a sandblasted
specimen in tension, is achieved by fibres bridging a newly defined fracture surface. The nu-
merical model can not simulate this response, but has to portray the given GIf in its response.
A better fit of the post-peak response can be found by selection of another softening model, but
generally at the cost of computational ease.
Heinrich Stander University of Stellenbosch
7.1. CONCLUSIONS 95
7.1.5 Bonded overlay
ECC applied as a thin overlay for repair strategies, shows promise. Numerical simulation indi-
cates superior performance when compared to concrete overlays in terms of ductility.
Higher σfc values result in stiffer composite behaviour for the deflection hardening period and
earlier peak strengths. Lower strain-hardening value corresponds also to an earlier peak strength
as well as lower overall deflection. Higher E-modulus values for ECC result in stiffer and stronger
composite responses.
Inducing enough room for overlay movement, i.e. allowing ECC to exhibit multiple cracking,
results in improved system ductility and energy dissipation. Improved ductility can either be
acquired by supplying an artificial delamination along with a strong bond or by using a weaker
bond.
7.1.6 Design of composite interface
The introduction of ECC as an overlay in repair or retrofitting applications, does not only
address durability aspects but also structural performance. The associated ductility of the ma-
terial induces a high performance aspect wherever it is applied. It is thus crucial to execute
reliable design methods, especially at interfacial level, in order to harness the ductility at hand.
Knowledge of the interfacial bond properties makes it possible to alter the SSPT in order to
optimise the eventual interfacial bond and to test structural applications through numerical
modelling. Numerical modelling serves as a design tool through which an optimal interface
with the required characteristics can be engineered.
The design of a bond depends on the application. Sometimes a weaker bond is necessary to
achieve ductility through partial interfacial delamination and vice versa for strength. In order
to achieve a weaker bond, two approaches towards an optimal design can be implemented. One
implement a SSPT which induces a weaker bond. The second option is the discontinuous use
of a strong bond.
Accurate material models make it possible to test structural elements numerically. Thus the
execution period of experimental research, conducted to ascertain knowledge about material
characteristics and structural performance is shortened. Eventually it leads to the inclusion of
new material design aspects into design codes at earlier stages.
The product of this work is a numerical modelling technique by which composite behaviour
can be predicted and optimised, enabling reliable and affordable development of ECC/Concrete
overlay strategies. Whether or not it is a feasible design tool, is not the purpose of this re-
search, but as a numerical tool its contribution in product development and characterisation is
imperative.
Heinrich Stander University of Stellenbosch
7.2. RECOMMENDATIONS 96
7.2 RECOMMENDATIONS
On the basis of the findings and conclusions, the following recommendations are made:
• Research the interfacial chemical bond on a microscopic level in order to understand the
involved chemical processes and to relate it to macroscopic results.
• Measure the substrate saturation levels accurately to assist in achieving and executing an
optimal degree of saturation in practice.
• Implement the modified shear test method and conduct displacement control just over the
interface.
• Conduct interfacial shear tests with confining forces over the interface in order to deter-
mine the friction parameters (Φ, Φr and a).
• Design a grip system for the tension test, omitting the use of adhesive.
• Standardise parameter tests used to characterise composite interface properties. Compar-
ison of results will become easier.
• Modify the ECC material model in order for the mesh to be size independent. Computa-
tional time will be saved.
• Research the influences of time-dependent behaviour on the ECC/concrete interface and
incorporate the effects thereof in the numerical model.
• Study the influence of overlay casting direction on the obtained bond.
• Optimise mix designs, thus material properties to produce ECC with a higher E-modulus.
• Conduct experiments, testing and optimising the method using links to acquire a discon-
tinues strong interfacial bond.
Heinrich Stander University of Stellenbosch
Bibliography
[1] J.R. Mackechnie, M.G. Alexander and Jaufeerally. Structural and durablity properties of
concrete made with corex slag. Technical report, University of Cape Town, University of
Witwatersrand, 2003. 9
[2] American Concrete Institute, P.O.Box 9094 Farmington Hills, Michigan 48333-9094. ACI
318-02, 2 edition, August 2002. 10, 12
[3] T. Arakawa and K. Ono. Transactions of the architectural institute of japan, 57:581–584,
1957. 19
[4] Hans-Dieter Beushausen. Long-term performance of bonded concrete overlays subjected to
differential shrinkage. PhD thesis, University of Cape Town, 2005. 15, 16, 93
[5] Anual book of ASTM standards. Standard test method for plane-strain fracture toughness
of metallic materials. ASTM. 21
[6] W.P. Boshoff. Time-dependant behaviour of Engineered Cement-based Composites. PhD
thesis, University of Stellenbosch, 2006. 23, 24, 28
[7] J. Cao and D.D.L. Chung. Degradation between old and new mortar under cyclic shear
loading, monitored by contact electrical resistance measurement. Cement and concrete
research, 31:1647–1651, 2001. 19
[8] CEN, Rue de Strassart, 36 B-1050 Brussels. Eurocode 2, December 2004. 10, 12, 13
[9] Comite Euro-International du Beton, Case Postale 88, CH 1015 Lausanne, Switzerland.
CEB-FIP Model code 1990, first edition, 1993. xiv, 10, 13, 45, 66, 67
[10] J. Davies. Study of shear fracture in mortar specimens. Cement and concrete research,
25:1031–1042, 1995. 18
[11] Deutches Institut fur Normung. DIN 1045-1:2001-07, July 2001. 10, 12, 13
[12] H.W. Reinhardt et al. Shear of structural concrete members and pure mode ii testing.
Advanced cement based materials, 5:75–85, 1997. 18
[13] Li et al. On the shear behaviour of engineered cementitious composites. Advanced cement
based materials, 1(3):142–149, March 1994. 19
97
Bibliography 98
[14] P. Chen et al. Improving the bonding between old and new concrete by adding carbon
fibres to the new concrete. Cement and concrete research, 25(3):491–496, 1995. 19
[15] T. Horikoshi et al. Properties of polyvinylalcohol fibre as reinforcing materials for ce-
mentitious composites. In V.C. Li G. Fischer, editor, International RILEM workshop on
HPFRCC in structural applications, page 147. Rilem, S.A.R.L., 2006. xiv, 7
[16] M.H. Fiebrich. Influence of the surface roughness on the adhesion between concrete and
gunite mortars overlays. Beuth, 1994. 15
[17] J.L. Granju. 193-rls rilem tc bonded cement-based material overlays for the repair,the
lining or the strengthening of slabs or pavements. Technical report, Rilem, 2004. 11, 12,
19
[18] J.M. Illston and P.L.J. Domone, editors. Construction Materials, chapter 13, page 104.
Spon Press, 2001. 5
[19] J.M. Illston and P.L.J. Domone, editors. Construction Materials, chapter 14, page 112.
Spon Press, 2001. 8, 40
[20] N. Iosipescu. New accurate procedure for single shear testing of metals. Journal of Mate-
rials, 2(3):537–566, Sept 1967. 19
[21] T. Kamada and V.C. Li. The effects of surface preparation on the fracture behaviour of
ecc/concrete repair system. Cement & Concrete Composites, 2000. 1
[22] V.C. Li. Reflections on the research and development of ecc. In Proceedings of the JCI
International Workshop on Ductile Fiber Reinforced Cementitious Composites (DFRCC),
Application and Evaluation (DRFCC-2002), pages 1–21, 2002 2002. xi, 6
[23] V.C. Li. On engineered cementitious composites (ecc) a review of the material and its
applications. Journal of advanced concrete technology, 1(3):215–230, November 2003. 6
[24] P.B. Lourenco and J.G. Rots. A multi-surface interface model for the analysis of masonry
structures. Structural engineering, ASCE 123(7):660–668, 1997. 25, 27
[25] M.P. Luong. Tensile and shear strengths of concrete and rock. Engineering fracture me-
chanics, 35:127–135, 1990. 18
[26] A.H. Mattock and N.M. Hawkins. Research on the shear transfer in reinforced concrete.
PCI Journal, 17(2):55–75, March-April 1972. 19, 22, 36
[27] M.B. Nooru-Mohamed. Mixed-mode fracture of concrete: An experimental approach. PhD
thesis, Delft University of Technology, 1992. 19
[28] E.C. Olsen. Comparison of surface preperation techniques on the bond strength between
ecc and tradisional concrete. Unpublished report on research during an exchange period
at Stellenbosch University, 2005. xiv, 25, 28, 29, 30
Heinrich Stander University of Stellenbosch
Bibliography 99
[29] H. Tada, P. Paris and G. Irwin. The stress analysis of cracks handbook. Paris productions
incorporated: St.Louis, 2nd edition, 1985. 18
[30] Klaus-Alexander Rieder. Determination of crack resistance curves of cementitious materials
from measurements of a wedge splitting test method. Fracture mechanics for concrete
materials: testing and applications, ACI SP-201:207–228, 2001. 17, 21
[31] SAISC. Southern African steel construction handbook, fifth edition, September 2005. 67
[32] Q. Shang. Shear behaviour of cement-based composites. Master’s thesis, University of
Stellenbosch, 2006. 47
[33] Q. Shang and G.P.A.G van Zijl. Characterising the shear behaviour of strain-hardening
fibre- reinforced cement-based composites. SAICE, 2007. 36
[34] South African Bureau of Standards. SABS 0100-2, 2 edition, 1994. 12, 100
[35] South African Bureau of Standards. SABS method 1253:1994, 1994. 46
[36] South African Bureau of Standards. SABS method 863:1994, 1994. 46, 50
[37] South African Bureau of Standards. SABS method 864:1994, 1994. 46
[38] The South African Bureau of Standards. SABS 0100-1, 2.2 edition, 2000. 11, 100
[39] H. Stander. The role of admixtures in cement-based composite materials. A final year
project at the University of Stellenbosch, 2004. 40, 62
[40] E.K. Tschegg. New equipment for fracture tests on concrete. Material testing, pages
338–343, 1991. 21
[41] R. van der Pluijm. Out-of-Plane Bending of Masonry, Behaviour and Strength. PhD thesis,
Technical University of Eindhoven, 1999. xiv, 69
[42] J.G.M. van Mier. Fracture processes of concrete. CRC Press, 1997. 14, 59, 93
[43] G.P.A.G. van Zijl. Computational modelling of masonry creep and shrinkage. PhD thesis,
Delft University of Technology, 2000. 25, 27
[44] G.P.A.G van Zijl. The role of aggregate in high performance fibre reinforced cement-based
composites. Concrete/Beton, 2005. 6
Heinrich Stander University of Stellenbosch
Appendix A
STRUCTURAL DESIGN CODES
A.1 SABS - Code of practice
The only relevant composite design specifications found in the South African Standard for the
structural use of concrete, SABS 0100-1 (part 1) [38], are in the subclause 6.4: “Composite con-
crete construction”. Only certain parts are extracted, the highlighted parts are of importance.
The subsequent extract is retrieved from SABS 0100-2 (part 2) [34], which focusses on materials
and execution of work. The only relevant part is in subclause 10.4: “Construction joints”.
The paragraph numbering used in this section refers to the numbering of the applicable SABS
code of practice.
6.4 Composite concrete construction
6.4.1 General
6.4.1.1 The provisions of this subclause apply to flexural composite elements consisting
of precast concrete units acting in conjunction with added concrete where provision
has been made for the transfer of horizontal shear at the contact surface. The precast units
may be of either reinforced or prestressed concrete. Analyse and design composite concrete
structures and elements in accordance with clause 4 or clause 5, modified, where appropriate,
in accordance with 6.4.3 and 6.4.4. Pay particular attention, in the design of both the compo-
nents and the composite section, to the effect of the method of construction, on stresses and
deflection, and to whether or not propping is to be used.
6.4.1.2 Base the relative stiffnesses of elements on the properties of the concrete, gross or trans-
formed sections, as described in 3.4.3.1. If the concrete strength in the two components
of a composite element differs more than 10 MPa, make allowance for this when
stiffness is being assessed.
6.4.1.3 Differential shrinkage of the added concrete and precast concrete units may require
100
A.1. SABS - Code of practice 101
consideration in analysing composite elements for the serviceability limit states (see 6.4.3.4);
differential shrinkage need not be considered for the ultimate limit state.
6.4.1.4 When precast units, having pre-tensioned tendons, are designed as continuous elements
and continuity is obtained with reinforced concrete cast in-situ over the supports, the compres-
sive stresses due to prestress in the ends of the units may be ignored over the transmission
length of the tendons when the strength of section is being assessed.
6.4.2 Shear
6.4.2.1 Carry out the analysis of the resistance of composite section to vertical shear due to
design ultimate loads in accordance with 4.3.4 for reinforced concrete and 5.3.4 for prestressed
concrete. However, when in-situ concrete is placed between precast prestressed units and the
composite concrete section is used in design, ensure that the principal tensile stress does not ex-
ceed 0.24√fcu anywhere in the prestressed units; calculate this stress by making due allowance
for the construction sequence and by taking into account only 0.8 of the compressive stress due
to prestress at the section under consideration.
6.4.2.2 Calculations for horizontal shear between the two components of a compos-
ite section are governed by the ultimate limit state. The methods given in 6.4.4.1
to 6.4.4.4 ensure that composite action does not break down for the serviceability
limit states and that the design shear strength is adequate for the ultimate limit
state.
6.4.4 Ultimate limit state
6.4.4.1 Horizontal shear force due to design ultimate loads
The interface of the precast and in-situ components occurs either in the tension zone or in the
compression zone affecting the horizontal shear force due to design ultimate loads so that this
shear force is either:
• where the interface is in the compression zone: the compression from that part of the
compression zone above the interface, calculated from the ultimate bending moment; or
• where the interface is in the tension zone: the total compression (or tension) calculate
from the ultimate bending moment.
6.4.4.2 Average horizontal design shear stress
The average horizontal design shear stress is calculated by dividing the design horizontal shear
force (see 6.4.4.1) by the area obtained by multiplying the contact width by the beam length
between the point of maximum positive or negative design moment and the point of zero mo-
ment.
Heinrich Stander University of Stellenbosch
A.1. SABS - Code of practice 102
The design horizontal design shear stress should then be distributed in proportion to the vertical
design shear force diagram, to give the horizontal shear stress at any point along the length of
the composite component. The horizontal design shear stress v so detained, should nowhere
exceed the appropriate value in Table A.1.
6.4.4.3 Nominal links
Where nominal links are provided, they should be of cross-section at least 0.15% of the contact
area. Spacing should not be excessive. The spacing of links in T-beam ribs with composite
flanges should not exceed the greater of four time the minimum thickness of the in-situ concrete
or 600 mm. Links should be adequately anchored on both sides of the interface.
Heinrich Stander University of Stellenbosch
A.1. SABS - Code of practice 103
Table A.1: Design ultimate horizontal shear stresses at interface (SABS 0100-01).
Design ultimate horizontal shear stresses atinterface
Precast unit Surface type Grade of in-situ concrete (MPa)
25 30 ≥ 40
Without links As-cast oras-extruded
0.4 0.55 0.65
Brushed,screeded orrough-tamped
0.6 0.65 0.75
Washed toremove laitanceor treated withretarder andcleaned
0.7 0.75 0.8
With nominal linksprojecting into in-situconcrete
As-cast oras-extruded
1.2 1.8 2.0
Brushed,screeded orrough-tamped
1.8 2.0 2.2
Washed toremove laitanceor treated withretarder andcleaned
2.1 2.2 2.5
NOTES
1 The description ’as-cast’ covers those cases where the concrete is placed and vibrated, leaving a roughfinish. The surface is rougher than would be required for finishes to be applied directly without afurther finishing screed but not as rough a would be obtained if tamping, brushing or other artificialroughening had taken place
2 The description ’as-extruded’ covers those cases in which an open-textured surface is produced directfrom an extruding machine.
3 The description ’brushed, screeded or rough-tamped’ covers those cases where some form of deliberatesurface roughening has taken place but not to the extent of exposing the aggregate.
4 For structural assessment purposes, it may be assumed that the appropriate values of γm (included inthe table) is 1.5.
6.4.4.3 Links in excess of minimum
Where the horizontal shear stress from 6.4.4.2 exceeds the values given in Table A.1, all the hor-
izontal shear force should be carried on reinforcement anchored on either side of the interface.
The amount of steel required, Ah (in mm2/m) should be calculated from the following equation:
Ah = 1000bvh
0.87fy
Where:
b is the contact width;
Heinrich Stander University of Stellenbosch
A.1. SABS - Code of practice 104
vh is the average horizontal design shear stress, as in 6.4.4.2; and
fy is the characteristic strength of links.
10.4 Construction joints
10.4.3 Bonding
When so required or permitted, bonding shall be achieved by means of one of the methods given
in 10.4.3.1 to 10.4.3.4.
10.4.3.1 The use of an adhesive
Joints to which an adhesive is applied shall be prepared, and the adhesive applied, in accordance
with the manufacturer’s recommendations, prior to the placing of fresh concrete.
10.4.3.2 The use of a retarder
Surfaces of joints that have been treated with a chemical retarder shall be prepared in accor-
dance with the manufacturer’s recommendations, prior to the placing of fresh concrete.
10.4.3.3 Roughening and dampening the surface of the concrete
Roughening the surface of the concrete in an acceptable manner shall uniformly expose the
aggregate and shall not leave laitance, loosened particles of aggregate or damaged concrete on
the surface.
The hardened concrete of construction joints and of joints between footings and walls or columns,
between walls or columns and beams or floors they support, and joints in other elements not
mentioned above shall be dampened (but not saturated) immediately prior to the
placing of fresh concrete.
The hardened concrete of horizontal construction joints
• in exposed work, or
• in the middle of beams, girders, joists and slabs, or
• in work designed to contain liquids
shall be dampened (but not saturated) and then thoroughly covered with a thin
coat of cement grout of similar proportions to the mortar in the concrete. Alter-
natively, application of a concrete layer of thickness approximately 250 mm, made
richer by reducing the amount of coarse aggregate, may be considered. The fresh
concrete shall be placed before the grout or the intermediate layer of concrete has
Heinrich Stander University of Stellenbosch
A.2. EUROCODE 2 105
attained its initial set.
10.4.5 Construction
10.4.5.3.1 Construction joints when concrete is less than 24h old
The surface of the concrete shall be brushed with a steel wire brush before new mortar
and concrete are placed as specified in 10.4.5.3.2.
10.4.5.3.2 Construction joints when concrete is more than 24h and less the 3d old
The surface of the concrete shall be sand-blasted, or chipped with a light hammer, and
swept clean. The surface dry concrete shall then be brushed with a thin layer of mortar
composed of cement and sand mixed in the same ratio as the cement and sand in the
concrete mixture, or a thin layer of cement / water mix of creamy consistence. This mortar or
mix shall be freshly mixed and placed immediately before the new concrete is placed. It shall
not be allowed to dry out before new concrete is poured.
10.4.5.3.3 Construction joints when concrete is more than 3d old
The old surface shall be cleaned as in 10.4.5.3.2 and kept continuously wet for at least 24
h and then allowed to become surface dry before the mortar or cement / water mix (
as in 10.4.5.3.2) and new concrete are placed.
A.2 EUROCODE 2
The following section is taken from Eurocode 2: Design of concrete structures - Part1-1: Gen-
eral rules and rules for buildings. Calculations and classifications are given as how to compute
the design shear resistance at the interface. A contribution to the shear resistance, made by re-
inforcement crossing the interface, are incorporated into the calculation of vRdi. In cases where
no reinforcement is present, this part of the equation is omitted. Part (2) of this particular
section, extracted from Eurocode 2, is of more importance to this research. This part indicates
certain surface roughnesses and the associated parameter values. Still no specific method of
how to ascertain such roughnesses are revealed.
The paragraph numbering used in this section refers to the numbering of the Eurocode 2.
6.2.5 Shear at the interface between concrete cast at different times
(1) In addition to the requirements of 6.2.1-6.2.4 the shear stress at the interface between
concrete cast at different times should also satisfy the following:
vEdi 6 vRdi
Heinrich Stander University of Stellenbosch
A.2. EUROCODE 2 106
vEdi is the design of the shear stress in the interface and is given by:
vEdi = βVEd/(zbi)
where:
β is the ratio of the longitudinal force in the new concrete area and the totallongitudinal force either in the compression or tension zone, both calculated forthe section considered
VEd is the transverse shear force
z is the lever arm of composite section
bi is the width of the interface
vRdi is the design shear resistance at the interface and is given by:
vRdi = cfctd + µσn + ρfyd(µsinα+ cosα) ≤ 0.5 νfcd where:
c and µ are factors which depend on the roughness of the interface (see (2))
fctd is as defined in 3.1.6 (2)P
σn stress per unit area caused by the minimum external normal force across theinterface that can act simultaneously with the shear force, positive for compres-sion, such that σ < 0.6fcd, and negative for tension. When σn is tensile cfctd
should be taken as 0.
ρ As/Ai
As is the area of reinforcement crossing the interface, including ordinary shearreinforcement (if any), with adequate anchorage at both sides of the interface.
Ai is the area of the joint
α is defined in Figure A.1, and should be limited by 45◦ ≤ α ≤ 90◦
ν is a strength reduction factor (see 6.2.2 (6))
≤ 30◦
h2 ≤ 10d
h1 ≤ 10d
α
NEd
VEdd ≥ 5mm
A - new concrete, B - old concrete, C - anchorage
Figure A.1: Indented construction joint
(2) In the absence of more detailed information surfaces may be classified as very smooth,
smooth, rough or indented, with the following examples:
• Very smooth: a surface cast against steel, plastic or specially prepared wooden moulds:
c = 0.25 and µ = 0.5
• Smooth: a slipformed or extruded surface, or a free surface left without further treatment
after vibration: c = 0.35 and µ = 0.6
Heinrich Stander University of Stellenbosch
A.2. EUROCODE 2 107
• Rough: a surface with at least 3 mm roughness at about 40 mm spacing, achieved by
raking, exposing of aggregate or other methods giving an equivalent behaviour: c = 0.45
and µ = 0.7
• Indented: a surface with indentations complying with Figure A.1: c = 0.50 and µ = 0.9
(3) A stepped distribution of the transverse reinforcement may be used, as indicated in Figure
A.2. Where the connection between the two different concretes is ensured by the reinforcement
(beams with lattice girders), the steel contribution to vRdi may be taken as the resultant of the
forces taken from each of the diagonals provided that 45◦ ≤ α ≤ 90◦.
(4) The longitudinal shear resistance of grouted joints between slab or wall elements may be
calculated according to 6.2.5(1). However in cases where the joint can be significantly cracked,
c should be taken as 0 for smooth and rough joints and 0,5 for indented joints (see also 10.9.3
(12)).
(5) Under fatigue or dynamic loads, the values for c in 6.2.5 (1) should be halved.
vEdi ρfyd(µsinα+ cosα)
cfctd + µσn
Figure A.2: Shear diagram representing the required interface reinforcement
Heinrich Stander University of Stellenbosch
Appendix B
RESULTS: INTERFACIAL
EXPERIMENTS
A total of 57 successful tests were conducted. This amount included shear and tensile specimens.
The tests were conducted to perceive the interfacial properties of various preparation methods.
The preparation methods are defined by two primary stages, namely the roughening and moist-
ening applications to the substrate. The following sections contain graphical illustrations of all
the results, which are categorised by its respective roughening applications.
B.1 Reference surface
B.1.1 Moistening period: 24 hours
Testing age: 14 days
0 0.2 0.4 0.6 0.8 10
0.2
0.4
0.6
0.8
1
υ [mm]
τ xy [M
Pa]
SB1SB2
0 0.2 0.4 0.6 0.8 10
0.2
0.4
0.6
0.8
1
u [m
m]
υ [mm]
(a)
0 0.2 0.4 0.6 0.8 10
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
u [mm]
σ t [MP
a]
(b)
Figure B.1: Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimenswith a 10 minute moistening period.
108
B.1. Reference surface 109
Testing age: 28 days
0 0.5 1 1.5 20
0.5
1
1.5
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 20
0.5
1
1.5
2
u [m
m]
υ [mm]
(a)
0 0.2 0.4 0.6 0.8 10
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
u [mm]
σ t [MP
a]
(b)
Figure B.2: Interfacial responses in (a) shear and (b) tension, at 28 days for the reference specimenswith a 10 minute moistening period.
B.1.2 Moistening period: 10 minutes
Testing age: 28 days
0 0.2 0.4 0.6 0.8 10
0.2
0.4
0.6
0.8
1
υ [mm]
τ xy [M
Pa]
0 0.2 0.4 0.6 0.8 10
0.2
0.4
0.6
0.8
1
u [m
m]
υ [mm]
(a)
Figure B.3: Interfacial response in shear at 28 days for the reference specimens with a 24 hour moist-ening period.
Heinrich Stander University of Stellenbosch
B.2. Scrape surface 110
B.2 Scrape surface
B.2.1 Moistening period: 24 hours
Testing age: 14 days
0 0.5 1 1.5 20
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 20
0.5
1
1.5
2
u [m
m]
υ [mm]
(a)
0 0.5 1 1.5 2 2.5 30
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
u [mm]σ t [M
Pa]
(b)
Figure B.4: Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimenswith a 24 hour moistening period.
B.2.2 Moistening period: 10 minutes
Testing age: 14 days
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20
0.5
1
1.5
2
2.5
u [m
m]
υ [mm]
(a)
0 0.5 1 1.5 2 2.5 30
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
u [mm]
σ t [MP
a]
(b)
Figure B.5: Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimenswith a 10 minute moistening period.
Heinrich Stander University of Stellenbosch
B.3. Sandblast surface 111
B.3 Sandblast surface
B.3.1 Moistening period: 24 hours
Testing age: 7 days
0 0.5 1 1.5 2 2.50
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 2 2.50
0.5
1
1.5
2
2.5
u [m
m]
υ [mm]
(a)
0 0.5 1 1.5 2 2.5 30
0.2
0.4
0.6
0.8
1
u [mm]
σ t [MP
a]
(b)
Figure B.6: Interfacial responses in (a) shear and (b) tension, at 7 days for the reference specimenswith a 24 hour moistening period.
Testing age: 14 days
0 0.5 1 1.5 20
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 20
0.5
1
1.5
2
u [m
m]
υ [mm]
(a)
0 0.5 1 1.5 2 2.5 30
0.2
0.4
0.6
0.8
1
u [mm]
σ t [MP
a]
(b)
Figure B.7: Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimenswith a 24 hour moistening period.
Heinrich Stander University of Stellenbosch
B.3. Sandblast surface 112
Testing age: 28 days
0 0.5 1 1.5 2 2.50
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 2 2.50
0.5
1
1.5
2
2.5
3
3.5
u [m
m]
υ [mm]
(a)
0 0.5 1 1.5 2 2.5 30
0.2
0.4
0.6
0.8
1
u [mm]
σ t [MP
a]
(b)
Figure B.8: Interfacial responses in (a) shear and (b) tension, at 28 days for the reference specimenswith a 24 hour moistening period.
B.3.2 Moistening period: 10 minutes
Testing age: 14 days
0 0.5 1 1.5 20
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 20
0.5
1
1.5
2
u [m
m]
υ [mm]
(a)
0 0.5 1 1.5 2 2.5 30
0.2
0.4
0.6
0.8
1
u [mm]
σ t [MP
a]
(b)
0 0.5 1 1.5 20
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 20
0.5
1
1.5
2
u [m
m]
υ [mm]
(c)
Figure B.9: Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimenswith a 10 minute moistening period.
Heinrich Stander University of Stellenbosch
B.4. Drill holes surface 113
Testing age: 28 days
0 0.5 1 1.5 2 2.50
1
2
3
4
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 2 2.50
0.5
1
1.5
2
2.5
u [m
m]
υ [mm]
(a)
Figure B.10: Interfacial response in shear at 28 days for the reference specimens with a 10 minutemoistening period.
B.4 Drill holes surface
B.4.1 Moistening period: 24 hours
Testing age: 14 days
0 0.5 1 1.5 2 2.50
0.5
1
1.5
2
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 2 2.50
0.5
1
1.5
2
2.5
u [m
m]
υ [mm]
(a)
0 0.5 1 1.5 2 2.5 30
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
u [mm]
σ t [MP
a]
(b)
Figure B.11: Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimenswith a 24 hour moistening period.
Heinrich Stander University of Stellenbosch
B.5. Precast grooves surface 114
B.5 Precast grooves surface
B.5.1 Moistening period: 24 hours
Testing age: 14 days
0 0.5 1 1.5 2 2.50
0.5
1
1.5
2
υ [mm]
τ xy [M
Pa]
0 0.5 1 1.5 2 2.50
1
2
3
4
5
u [m
m]
υ [mm]
(a)
0 0.5 1 1.5 2 2.5 30
0.2
0.4
0.6
0.8
1
u [mm]
σ t [MP
a]
(b)
Figure B.12: Interfacial responses in (a) shear and (b) tension, at 14 days for the reference specimenswith a 24 hour moistening period.
Heinrich Stander University of Stellenbosch