Professional Documents
Culture Documents
New Reinforcement Detailing in Concrete Frame Corners of Civil Defence Shelters
New Reinforcement Detailing in Concrete Frame Corners of Civil Defence Shelters
net/publication/338112896
CITATIONS READS
3 91
1 author:
Morgan Johansson
Chalmers University of Technology
37 PUBLICATIONS 117 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Blast and Fragment Impacts: Strengthening of reinforced concrete structures View project
All content following this page was uploaded by Morgan Johansson on 22 December 2019.
in ConcreteFrameCornersof
Detailing
NewReinforcement
CivilDefenceShelters
FiniteElementAnalysesand Experiments
Non-linear
MorganJohansson
Licentiatethesis
GdteborgNovember1996
CHALMERSUNIVERSITYOFTECHNOLOGY
OF STRUCTURALENGINEERING
DEPARTMENT
DIVISION OF CONCRETESTRUCTURES
Morgan Johansson
GoteborgNovember1996
Gciteborg,
November1996
MorganJohansson
CONTENTS
PREFACE
CONTENTS
ABSTRACT
NOTATIONS
INTRODUCTION I
l.l Background I
1.2 Aim of theStudy 2
LITERATURESURVEY ^
2.1 FrameCorners A
EXPERIMENTS l0
3.1 TestSpecimens l0
3.1.1 Dimensions andreinforcement t0
3.1.2 Materialproperties t2
3.2 TestSet-upandTestProcedure l4
3.3 FailureDevelopment l5
3.3.1 General Observations l5
3.3.2 Specimens with highreinforcement
ratio t5
3.3.3 Specimens with low reinforcement
ratio l5
3.4 TestResults l7
3.5 Discussion 2l
4.1 General 24
4.2 MaterialModels 24
4.2.1 Modellingof theconcrete )4
4.2.1.1Thecrackmodelfor tension .A
4.4.1 General JJ
4.4.2 The FE modelfor analysesof generalresponse )+
4.4.3 The FE modelfor detailedanalyses 36
4.4.3.I New reinforcementdetailing 36
4.4.3.2Conventional reinforcementdetailins 39
4.5 Resultsof theAnalvses 40
4.5.I General 40
4.5.2 FE analyses of generalresponse 4)
4.5.3 DetailedFE analyses 44
4.5.3.I Objectives andpreconditions 44
4.5.3.2New reinforcement detailing 45
4.5.3.3Conventional reinforcementdetailine 52
4.5.4 Comparisons of the FE analyses 55
4.5.4.1Comparison of conventional
andnew reinforcement 55
detailings
4.5.4.2The effectof fractureenergy 58
4.5.4.3Comparisonof planestressandplain strainanalyses 60
4.6 Discussion 6l
4.7 Analysesof a CantileverBeam 65
4.1.1 General 65
4.1.2 The finite elementmodel 67
4.7.3 Resultsof theanalyses 68
4.7.3.1General 68
4.7.3.2Effectof the bond-sliprelation 68
4.1.3.3 Effectof the reinforcementtvpe 70
4.1.4 Discussion 71
CONCLUSIONS 73
5.1 General 73
5.2 Suggestions
for FutureResearch 1^
REFERENCES 15
u
NOTATIONS
A, crosssectionareaof reinforcement
Du componentin stiffnessmatrix
E, Young'smodulusfor concrete
E, Young'smodulusfor steel
F load
I., force camed by reinforcement
Gf fractureenergy
I length
P projectionmatrix
c cohesion
dF, differentialquantityof force carriedby reinforcement
f(w) softeningfunction
f, compressive strengthof concrete
f,,,ub" compressive cubestrengthof concrete( 150 x 150x 150mm)
f,.,vt compressive cylinderstrengthof concrete(@150x 300 mm)
f,u ultimate strengthof reinforcement
f,, yield strengthof reinforcement
t tensilestrengthof concrete
ft.spt,r cubesplittingstrengthof concrete( 150x 150x 150mm)
i longitudinal direction
lbo, lengthof reinforcementbar
l"b*nt lengthof finite elementrepresenting bar
reinforcement
n normaldirection
r radiusof reinforcementloop
s slip
sm meancrackspacing
.r) slip at the pointwhereyieldingis obtained
t traction(stressvectoractingon a planeor surface),tangentialdirection,time
u displacement
w crackopening
wu ultimatecrack opening
r co-ordinatealongreinforcementbarsin cornerregion
) co-ordinatealongreinforcementbarsin beamand column
N
Greek letters
d stressblock factor
as scalarquantityusedto describethe Drucker-Prager yield surface
ds scalarquantityusedto determinethe hardeningparameter
P yield
stressblock factor,scalarquantityusedto describethe Drucker-Prager
surface
y ratioof ultimatestrengthto yield slrengthof the reinforcement
Tnod modifiedvalueof 1
lnt,,mt valueof l corresponding to "normalratio" steel
A incremental, increment of
A€^na modifiedincremental strain
6 displacement
8,. concretestrain
€n strainat steelhardening
€r steelstrain
e, ultimatesteelstrain
tP,r,^,or plasticstrainin the directionof uniaxialstress
,c hardeningparameter
tt factordescribingthechangein ductilityof reinforcement
n projectionvector
p rcinforcement ratio
o stress
6t, oz, 6: principalstresses
oc concretestress
o, steelstress
T bond stress
! bondstrengthin frictionalphase
Tw maximumbondstrength
Ty bondstressat thepointwhereyieldingis obtained
v Poisson's ratio
0 bu diameter,internalangleof internalfriction
Qo initial angleof internalfriction
VI dilatancyangle
INTRODUCTION
1.1 Background
When a concretefranrestructureis constructed, it is usuallycastin two separatestages:first
the casting of the wall, then the casting of the slab. After the casting of the wall, the
reinforcement bars,which will laterbe bentandsplicedinto the slab,stick up into the air, see
Figure la. Accordingto the regulationsof anchoragelength and splicing methodsusedin
Boverket'sHandbookfor ConcreteStructuresBBK 94, Boverket(1994),thesebarsmay be
severalmeterslong. This type of frame cornerreinforcementdetailing can be difficult to
realiseat a building site and is complicatedto carry out correctly.Therefore,it would be an
advantage to usea simplifiedreinforcement detailing.If all the reinforcement
could be spliced
within the comer a.reaa simplerprocedure,with lessrisk of incorrectdetailing,would be the
result,seeFigure 1b.
a)
The present Swedish regulationsfor the design of civil defence shelters allow the
reinforcementto be splicedin the immediatevicinity of the frame corner,with partsof the
reinforcementspliceextendedinto the corner,SwedishRescueServiceAgency(1994).This
results in a building procedure similar to that described above (Figure 1a) with a
reinforcementdetailing thar is time consumingand quite difficult to carry out correctly.
Consequently, the SwedishRescueServiceAgencywanteda simplerreinforcement detailing
to be usedin frame cornersin civil defenceshelters.Therefore,a new designproposal,in
which all the reinforcement barsare splicedwithin the comerruea,was workedout. To study
the behaviourof splicingin the reinforcedframe cornersof shelters,a researchprojectwas
initiatedar the Division of ConcreteStructuresat ChalmersUniversityof Technology,see
P l o s( 1 9 9 4 ab, ) .
To study the structuralbehaviourof the frame corner more thoroughly,the finite element
methodwas used.Four of the test specimenswere analysedusing materialmodelsbasedon
non-linearfracturemechanicsand plasticity.By usingthis approach,the progressive cracking
and the strainand stressstatescan be followed underincreasedload; which allows a better
understanding of the structuralbehaviourof the framecorner.Onceresultsobtainedusingthe
finite elementmodelshavebeenconfirmedby testresults,the finite elementmethodprovides
a valuabletool for further studies.Accordingly,in combinationwith experiments,finite
element analysessignificantly increasethe feasibility of carrying out parametricstudies.In
addition to the differencein the reinforcementdetailing, the effects that different parameters
haveon the load and deformationcapacityof the frame corner,were studiedfor framecomers
with a low reinforcement ratio,usingthe non-linearfinite elementmethod.The parameters of
interestwere:
. the weakness of theconstruction joint betweenthe first andsecondcastings,
r the bond-sliprelationbetweenthe reinforcementbarsand the sunoundingconcrete,and
r the mechanicalpropertiesof the reinforcingsteelbars.
Furthermore,the consequences of incorrectpositioningof the reinforcementloops in the new
reinforcementdetailing were examined.Detailedanalysesof the effect of the pararneterslisted
above have not been carried out for frame corners with a high reinforcement ratio.
Due to numerical difficulties, it was not possible to study the effect of differencesin the
mechanical properties of the reinforcing steel bars using finite element models of the frame
corner. lnstead,a simpler model of a cantilever beam was used. This model was then also
used to study more thoroughly the effect of drfferent bond-slip relations for structures with
both high and low amounts of reinforcement.
A limitation of the study carried out so far in this project is that all tests and analyseshave
been carried out for static loads. However, a civil defence shelter must withstand impulse
loading such as explosronsand falling buildings. Consequently,the behaviour of the new
reinforcementdetailingwhen subjectedto impulse loading needsto be studied.
) LITERATURE SURVEY
2.r Frame Corners
The bearing capacity of a frame structure dependson the strength of its independentstructural
members. To obtain a ductile behaviour in the structure. considerable redistributions of forces
and deformationsmust be possible.The capacityfor this relies heavily on the detailingof the
joint connections;i.e., the connectionsbetweendifferent members (e.g. beams and columns)
are of great importance for a sound structural behaviour. Thus, a joint connection must be at
least as strong as the structural members connected to it and show a ductile behaviour in the
ultimate limit state. ln this section, studies of different reinforcementdetailings in frame
corners are briefly presented; the term "frame corner" is used to describe a corner joint
connecting two structural members, such as a beam and a column or a slab and a wall, at an
angle of 90o.
Concrete frame comers can be separatedinto two principal types: those subjectedto a positive
moment (opening of the corner) and those subjected to a negative moment (closing of the
corner), see Figure 2. lt has been found by testing that the reinforcement detailing in frame
corners subjectedto positive moment is more sensitive than that in frame corners subjectedto
negativemoment, see Mayfield et al. (1911), Nilsson and l-osberg (1976); consequently,the
main effort of experimental studies has been concentratedon positive moment. Extensive
experimental studies on frame corners subjectedto positive moment have been conducted by
several researchers,see Swann (1969), Mayfield er al. t.1972), Nilsson (1973), Skettrup
et al. (1985). Many different reinforcement detailings with different reinforcement ratios have
been investigated; the experimental work done by Nilsson has resulted in detailing
recommendations, Nilsson (1973),Nilsson and Losberg (1976).
Some studies of frame corners subjected to negative moment have also been reported, see
Swann (1969), Mayfield er al. (1971), Yuan et al. (1982),Znuzou and Haldane (1993),
P l o s ( 1 9 9 4 b ) a n d L u o e t a l . ( 1 9 9 4 ) . A l i t e r a t u r e s u r v e y o fw o r k d o n e , b e f o r e l 9 T 3 , o n c o r n e r s
and joints subjected to positive and negative moment can be found in Nilsson (1973).
However, only tests on frame corners sublectedto a negative moment and with reinforcement
detailings similar to that examined in this study are mentioned here, see Table l. The work of
Plos is of special interest since it has functioned as a basis for the study presented in this
thesis.
a) b)
l=/ J
Figure2 Frame corner subjectedto: a) a positive bending moment and b) a negative
bendingmoment.
TableI Test resultson frame comers with reinforcementdetailingssimilar to that
presented in this study.Efficiencyis definedas the ultimateloadobserved
in the
testdividedby theestimatedloadcapacity.
11111 lttlt
*T:-----l
/-r------1
tlt*-tt.---l
tlt--it--i
L | | (t+e.)r l- - -l (t+e.)r l--t=l
tlttt-l
ltt--lt*-l t
l_L---__--_) l_' i_L____)
,'JJJJ JJJJT
a) b) c)
6, = f(w,) d.=0
(0<wcx.',) (w > w,)
f-lt t-*:l
t11
l-- -l
t--tl--l (l+e )Ltx
(l-e, )L+r' l-.-i ] -{ J - , , ,>---_
l * - r t' t- - - l
lt-ttl
i-t l- -l
-
vvVv.,
d) e)
fo' 6.
I - Unloadingre
li at maxlmum
li, -6, = f(w)
-T-
I
AL€cwuw wuw
In finite element modelling of cracks in concrete, there are two common concepts for treating
cracks: the discrete crack approach and the smeared crack approach. In this section, only the
principal differences in the two approachesare mentioned. More thorough descriptions of the
discrete and the smeared crack approacheshave been published by for instance Rots (1988)
and Plos (1995). A review ofprevious studiesofconcrete structuresusing the finite element
method can be found in Kwak and Filippou ( 1990).
In the discrete crack approach, the crack is modelled as a geometrical discontrnuity and
separateelements are used to simulate the cracks and the material between the cracks. ln the
fictitious crack model presentedby Hillerborg et al. (1916), the fracture zone is modelled as a
fictitious crack of initial width equal to z.ero.The behaviour of the crack is then describedby a
stress-crackopening relation. The crack band model of Bazant and Oh (1983) is a similar
approach;here the fracturezone is modelled with a band of a given width. However, in this
model, the localiseddeformationsare smearedout within the band, resultingin a responsethat
can be describedby a stress-strainrelation. Thus. the continuum of strainsand stressesare
preserved in the model. Since separate elements are used to model a crack in the discrete
crack approach, the possible crack path must be assumed in advance and the finite element
mesh arranged so that the crack path follows the element boundaries. This is a serlous
drawback of the approach: a great amount of work is required to establish the FE mesh since
the user has to decide where and how the cracks may arise. It also imposes a limitation on the
spontaneouscrack pattem.
According to Rots (1988), the smeared crack approach is the counterpart ofthe discretecrack
approach. Here, the localised non-lineanty of the crack is "smeared" out over the finite
element, i.e. all the matenal deformations, including the crack, is considered in the same
element. Accordingly, a cracked solid is modelled as a continuum allowing the cracked
material to be described with a stress-strainrelation. As this means that the crack pattem need
not be taken into account in advance, the smeared crack approach is a more attractive
procedurethan the discrete crack approach.
[',-l [','
I Ar,l= | D,,
D,,
D,,
D,,l[Ar,
I
," ll^.I (l)
l Lo,,D., ,,,11^,.1
Lo,,
whereindex / denoteslongitudinaldirection,indexn normaldirection,and index , tangential
directionof the interfacewhich is orientedalongthe meansurfaceof the reinforcement bar.
The tractionand the slip in the tangentialdirectionare often negligible,which resultsin an
approximaterelationaccordingto equation(2). Variousbond-sliprelations,D71in equations
(1) and(2), basedon work carriedout by Eligehausen e/ a/. (1983),can be found in the CEB-
FIP ModelCode.CEB (1993).
[r,, ,,,'l[*,.l
[o,,'l= (2)
LAr,l 14, 4,.lLAr,.j
Researchon the effectof bond in reinforcedconcreteusingnon-linearfracturemechanicshas
been carried out by, among others, Rots (1988), Kwak and Filippou (1990), and
Noghabai( 1995);a reviewof theuseof fracturemechanicsin modellingbondcanbe foundin
Noghabai. For a more thorough description of the bond concept, see CEB (1981),
(1992)andMagnusson(1997).
Engstrcim
a) b)
<_
F,
3.1.1 Dimensions
andreinforcement
To gain a better understandingof the behaviour of frame corners under loading to failure, two
test series, each containing of four full-scale specimens, were carried out. [n this thesis the
emphasis is on the second test series;more thorough information about the first test series can
be found in Plos (1994a,b). The dimensionsoithe specimensin the secondtest serieswere
the same as in the test series carried out by Plos, and are shown in Figure 6. The test
specimens were reinforced with deformed bars of reinforcement type K500; this type has
higher strength but lower ductility than the Ks 40 S reinforcement used by Plos, see Figure 7.
All four specimenswere cast with the new reinforcement detaiiing shown in Figure 8; that is
all reinforcement bars were spliced within the frame corners. This was accomplished by using
reinforcement loops. The legs of the loops were spliced to the main reinforcement in both the
column and the beam. To compensate for the risk of lower structural strength at the frame
corner due to the construction joint, the reinforcement ratio of the loops was increased by
25 Vo, in accordance with the Swedish Shelter Regulations, Swedish Rescue Service
Agency (1994). However, forthe specimenswith new detailing testedby Plos, the amount of
reinforcement was unequal in the sections adjacent to the frame corner, see Figure 8 and
Table 2.
2150
i l_J,*
2 1 5 0I
A-A
f---..-l Construction
I*l joint
10
Tensileforce [kN]
l 50-
150 200
Strain [0-3]
A, (RV2 andRV4)
1.25A, (RV5-RV8)
0.25A, 1.2A
5,
a) b)
Two of the specimens(denoted RV5 and RV6) had a large amount of longitudinal
reinforcement, 5 Ql6, approximatelyequal to the maximum allowed reinforcementratio in
agreement with the SwedishShelterRegulations. The othertwo specimens (denotedRV7 and
RV8) had a longitudinal reinforcementamount of 3 pl0, approximatelyequal to the
correspondingminimum allowed reinforcementratio. The same amount of longitudinal
reinforcementwas used on both the compressiveand tensile sides of the beam and the
column,seeFigure8 andTable2. For eachreinforcement ratio.one specimenwas reinforced
with the splicedreinforcementloopsplacedin contactwith eachother(RV5 and RV7); in the
other specimens, the loops were placedwith spacebetweeneachother (RV6 and RV8), see
Figure9. Drawingsof the specimensin the secondtest seriesis shownin AppendixA; for
furtherinformation,seeJohansson (1995).
lt
T a b l el Reinforcenrent amount and configurations for the test \peclmen\. Thc- sanre
n u m b e r o l r e i n f o r c e m e nbt a r s r i a s p l a c e coi n b o t h t h e c o m p r e s s r r ca n d t h e t e n s i l e
s r d e so f t h c ' b e a r na n d t h e c o l u m n . T h e s p e c i n r e n rse s r e db r p l o : r l ! 1 9 - h . b r a i c
rncluded.
RV6: s p a c e b e t r , ,e c n nc\\
RV7.] l\+ il] c0ntact llc\\
3.1.2 N'Iaterialproperties
T h e s p e c i m e n s\ \ ' e r ec a s t u i t h t h e c o l u m n i n a v e r t i c a lp o s i t i o n .s e e F i g u r e 6 . T o o b t a i n I
constructlon l o i n t . t h c i r l n t e c o l u m n s\ \ e r e c a s tf i r s t a n d l o u r d a y s l a t e rt h c b e a r n sa n t l c o r r r e r .
. l o l n t s$
. ' e r ec a s t .T h e s p e c i m e n s* e r e m a d ei n p a r r s :t h e t \ \ ' o s p e c i m e n st h a t u o u l i i b e d i r e c t l t
c o m p a r e du r t h e a c ho t h e r \ \ ' e r ec a s t a t t h e s a m et i m e a n d u i t h c ( ) n c r . - t ier r r r nt h e s a r n c ' b l t t c h .
C o n c r e t eq u a l i t r K - 1 0 .a c c o r d i n st o B o r e r k e t . H a n d b o o kf r t r c o n c r e t cS t r u c t u r e \B . B K 9 - 1 .: c , r
B o r " e r k e (t 1 9 9 ' 1 ) u. i t h : t t a r g e tc r l i n d e r c o m p r e s s i r es t r e n c t ho i 3 0 \ 1 P a t a s
c h o s e n .F g r t h c -
12
specimens in the secondtest serieswith low reinforcement ratio, a differentconcretequality
wasdelivered for thecastingof the beams.This qualityexhibiteda somewhat highersplitting
and compressivestrengthbut a lower value of Young'smodulusof elasticity.However,the
fractureenergywas only abouthalf that of the ordinaryconcreteof qualityK30. The strength
of the concrete used was determined by tests on cubes ( 150 mm) and cylinders
(0150x 300 mm), according to Swedishstandard, BST Byggstandardiseringen (1987),andis
presented in Table3. The strengthof the was
concrete determined at the ageof 28 daysandon
the day that the specimens were tested.The fractureenergywas determinedaccordingto the
recommendations of RILEM (1985) at a concreteage of between29 and 32 days. The
mechanical propertiesof the reinforcement arepresentedin Figure 10.
Table3 The strengthof the concreteused in the test specimens(mean value of three
specimens).
specimen f
J t.split f
" c.cuDe
f ,,qr Ert Age f f
" t , s p t t l " c.cuoe f ,.ryt E c1 Energyr
RV7, RV8 column J.Z 36.6 /- t.) z+.J 34 3.2 J I.J 21.8 24.6 108
beam J.+ 40.1 34.0 23.s 30 3 . 5 4 1 . 6 3 3 . 8 23.5 66
Not determined for the concrete used in the soecimensin the first test series
o, [MPa] Reinforce- En E,
f'u f'u cu
l3
3.2 TestSet-upand TestProcedure
The frame cornerspecimenswere testedin a verticaltest rig, seeFigure I l. The specimens
were braced in the horizontal direction at the loading and support points, allowing
displacements only along the loadingline. The load was appliedby a hydraulicjack and the
magnitudeof the load was measured by a loadcell. The total deflectionalongthe loadingline
was measured by electronicdisplacement transducers.Straingaugeswereusedto measurethe
strainin the reinforcementloops(length6 mm) as well as on the concrete(length60 mm) at
theinsideof theframecorner,seeFigure12.
The load was initially appliedin load incrementsof 5 kN for the specimenswith the low
reinforcementratio and in load incrementsof l0 kN for the specimenswith the high
reinforcementratio. To make it easierto follow the behaviourof the frame cornernearthe
maximum load, the load incrementwas halved when a non-linearstructuralresponsewas
observed(at 25 kN and 120 kN for the specimenswith low and high reinforcementratios,
When large time dependentdeformationsstartedto occur,the load level was
respectively).
keptconstantuntil thedisplacement waslessthan0.01mm/s.
Hydraulic jack
[.oad cell ;Displacement
Pendulum brace -. transducer
Construction
loint
-l
aw | 2
l50-lP! c = straingaugeon
200 lT lf- rernf
orcement
-[ - = Straingauge
on
200 lf
.-p .oncrere
ll
_u_\__r
14
3.3 Failure Development
3.3.1 Generalobservations
During the initial loading, two pnmary cracks were observedclose to the frame corner in all
the specimens,one in each section adjacentto the corner, see Figure 13. In all specimens
exceptRV5, the first crack was observedat the construction.yoint. When yielding was reached
in the reinibrcement bars, the deformations were concentrated to the frame comer region and
plastic hingesdevelopedon both sidesof the corner for all specimens.AII specimensshowed
ductile behaviour.
,, Crack section II
Construction
.yoint
Figure13 Sectionswhere the two primary cracks were first observedduring the initial
loading.
For specimens RV5 and RV6, very few crackswere observedoutsidethe immediatevicinity
of the framecorner.The behaviourof the two specimenswas similar and the maximumload
wasdeterminedfor both specimens by spallingof the concretesidecoverin the framecorner,
seeFigure14.Accordingto Boverket'sHandbookfor concreteStructures,BBK 94, Boverket
(1994),this failure should not have occuredfor the combinationof bar diameter,bending
radiusof the reinforcement loopsandconcretecoverusedin the specimens. That the concrete
side cover spalled off anyway indicatesthat the recommendations in BBK 94 are not
detailing.Before the spallingoccurred,the largest
applicablefor this kind of reinforcement
cracksfor bothspecimens wereobservedin cracksectionII (accordingto Figure13).
l5
appeared in crack section II for specimen RV7. For specimen RVE. the decisive crack
a p p e a r e d i n s e c t i o n l . a l o n g t h e c o n s t r u c t i o n j o r n t . s e e F i1g-u5r.eF o r h o t h s p e c i m e n s . r h e f i r s t
c r a c k w a s o b s e r v e di n t h e c o n s t r u c t r o n j o i n t a t a l o a d l e v e l l o * e r t h a n e x p e c t e dr a t a b o u t
l 0 k \ c o r n p a r e dt o a n e r p e c t e dl o a d o f a b o u t 2 0 k \ t . T h i s i n d i c a t e st h a t t h e t e n s t l e\ t r e n g t h
a c r o s st h e c o n s t m c t i o nj o i n t u ' a s l o w e r t h a n i n t h e c o n c r e t ec l o s e t o r t . r v h i c h r e s u l t c di n a
l o c a l i s e du e a k n e s si n t h e c o l u m n a d i a c e n t o t h e c o r l c r a r e a .
For the specimens with lou reinforcement ratio the marrmum load u'as reache{ aftcr.
c 0 n s t d e r a b l ep l a s t r c r o t a t r o n i n t h e \ e c t i o n s a d i a c e n tI o t h e f r a l l e c o r n e r . T h e n t i t x i r n u n t
( l c l i ) l r l l i l l l ( ) ltli l r h t l t h t h c s e s p e c r r l c l ) s\ \ l r s i i r n r t e d b r r u p t u r r ( ) t l h c r c i r ) 1 ' ( ) f c c t l c n
i ll l c c
r r - r n l r ) r c c l r c nl oI o p si n e a c hs p e c i n t e n\ \ e r e t o r n o f f ) .
t6
3.4 TestResults
The structuralbehaviourof the framecomer specimens is describedby the load-displacement
relationand the distributionof tensileforcesalongthe reinforcementbarsin the corner.The
relationsbetweenloadand verticaldisplacement for testspecimens RV5 to RV8 areshownin
Figuresl6 and 17.Exceptfor RV5, all specimenshad someof their reinforcement barstorn
off. A plateaucan be seenclearlyin the load-displacement curvesfor the specimenswith the
low reinforcement ratio.Due to spallingof the concreteside cover,this was not the casefor
the specimenswith high reinforcementratio. The load-displacement relations for test
specimens RVI to RV4, presented by Plos(1994a,b), areshownin Figure18.Thatthe load
capacitiesobtainedfor the specimenswith conventionaldetailingwere higherthan thosefbr
the specimenswith the new detailingis explainedby the unequalcapacitiesof the sections
adjacentto the cornerwith the new detailing,Figure8 and Table 2. The maximumload and
the maximumdisplacement at the end of the testareshownfor all specimens in Table4.
r00t
80t
60 1.
40
20
t1
Load,F [kN] - speclmen Kv /
fu-
+ Specimen RV8
40
30
20
l0
200 250
Displacement,
6 [mm]
140- r - SpecimenR-V4-
"
ffi+r----....
120. ! | 'J- ^t lr
100. ,,f u='4 t/
fl
80- I /a
a
OU* r
\.
a0-!
i
u-t"rrt*' i3$43;.=.-.-.--.c - € trG.'-]?.+.n
L
t
I
Figure18 l,oad-displacement relationsfor the frame corner specimensof the test series
carriedout by Plos(1994a,b).
18
Table4 Testresultsfor theeightspecimens. RVI to RV4 havebeen
Resultsfor specimens
takenfrom Plos( 1994a.b).
19
Tensileforce[kN] Specimen
RV5
-.-
120- F = 1 4 7k N , 6 = 2 9 m m
+f=l3lkN,6=20mm
100r /^\.] +F=100kN,6=12mm
*F=60kN.5=5mm
80r
! _ \---\
l _\ -\ \
-----
,/ 1
/?
/
a,, ,/,^"---_r-.
Strain gauge on
0r reinforcement
0 200 r000 1200 1400
Position
x [mm]
Specimen
RV6
Tensileforce[kNl
'
F=150kN,6=33mm
120. + F = 1 3 0k N , 6 = 2 5 m m
'-F = 1 0 0k N , 6 = 1 5m m
100l
-F=60kN,6=7mm
80 I
60 t ,.'
'""
""'
40 r.''
:"
20r-- '
Strain gauge on
0 +--- reinforcement
0 200 400 600 800 1000 1200 1400
Position
x [mm]
20
Tensileforce[kN] Specimen
RV7
)u- -"-
F = 42 kN, 6= 68 mm
+p=40kN,6=28mm
-'-" +
.\ F = 30 kN, d= 12mm
"t...
\ i*F=2okN,6=4mm
\
\ '\
L\\
---\
-\.'\i'
x
,./t-.t.-..------.
\:
\
\ - c = Strain gauge on
*-'
0r ---
0 200 1000 1200 1400 reinforcement
Positionx [mm]
= Straingaugeon
200 400 600 800 1000 1200 1400 reinforcement
x [mm]
Position
3.5 Discussion
In the first test series,carriedout by Plos, the specimenswith conventionalreinforcement
detailingwerefound to havea somewhathigherload capacitythanthosewith new detailing.
This was becauseof the greaterarnountof reinforcement,prescrib€dfor the cross-sectionwith
the constructionjoint (sectionI in Figure 13),that continuedthroughthe other criticalcross-
section(sectionII in Figure 13) for the conventionalreinforcementdetailing.For the new
detailing,the amountof reinforcement crossingsectionII waslessthanthat crossingsectionI,
see Figure 8, which resultedin unequal strengthin the sectionsadjoining the corner.
Consequently, a plastichinge developedonly in the weakercrosssectionof the specimens
2l
with new detailing,and at a lower load than for the specimenswith conventionaldetailing.
Therefore,in the secondtestseries,the adjoiningsectionsof the framecomerweredesigned
to be of equalstrength,seeFigure8. With this modification,plastichingesdevelopedat both
sidesof the framecorner,i.e. a similarbehaviourwasexhibitedfor specimens reinforcedwith
eithertypeof detailing.
Threeof the specimens in the secondtestserieshad somereinforcement barstorn off. The use
of a lessductile reinforcement type, K500, contributedto this behaviour.Thus, if the more
ductilereinforcement typeKs 40 S had beenused,a greaterdeformationbeforecollapseof the
specimens would havebeenobtained.No testswith the conventionalreinforcement detailing
andthe new,lessductile,reinforcement type werecarriedout.
22
Table5 Comparison of load capacity between the specimens in the second test series and
the specimens with the conventional reinforcement detailing tested by Plos.
Efficiency is defined as ultimate load observedin test divided by the estimated
maximum load capacity.
tkNl tkNl
RVI t57 115 l.ll
RV3 40 44 l.l0
The test specimenswere analysedusing the finite element programme DIANA, TNO (1993).
Two-dimensional plane stressmodels were used to simulate the concrete.The cracking of the
concrete was modelled using the smeared crack concept with fixed cracks. The non-linearity
of concrete in compression and the steel reinforcement were accounted for by plasticity
models. The specimens were modelled at two different levels of detail. A relatively coarse
mesh, assuming perfect bond between the reinforcement bars and the concrete, was used to
simulate the general responseof the specimens. To compare the new and the conventional
reinforcement detailings a refined element model, taking into account the interaction between
reinforcement and concrete. with a more dense mesh was used. This model was also used to
examine the influence of such parametersas the interaction between the reinforcement and the
concrete, the weakness of the construction joint, and the mechanical properties of the
reinforcing steel. Thorough information about the material data used in the FE analysescan be
found in Johansson(1995. 1996).
.)A
appearat the samelocationbut their formationis restrictedto a minimum angle(hereset to
60") to previouscracks.Whenthe crackedconcreteis unloaded,the secantunloadingmodulus
is usedas tangentstiffnessso that the strainacrossthe crackis reducedlinearlyto zeroas the
stressapproaches zero, see Figure 22. Thus, in the model used,a crack closescompletely
whenthe stressreacheszero.
f, =o.to(f,.,r,)''' (3)
where /].""1 was the cylinder compressive strength of the specimens on the testing day, see
Table 3.
a)
yield
Compression
surface Compression
Tensioncut-off yield surface
criterion
z)
andtension
Fisure22 Stiffnessusedin the analysesfor unloadedconcretein compression
wul6
openingrelationsimulatingthe concretesofteningcurve.The
Figure23 Bilinearstress-crack
shapeof thecurvewasbasedon recommendations givenin Gylltoft ( 1983).
I - sin0
c = J r.n 1(€ f,n,^,ot) -=------:- (4)
zcos@
26
where/..,,1(€Pun,*,ot)
was the compressivestrengthas a functionof the plasticstralnin the
directionof the uniaxial evaluated
stress, from standarduniaxial testson cylinders,see
Table 3. Poisson'sratio was. accordinqto recommendations in BBK 94. Boverket(1994).
approximatedtov=0.20
The reinforcement bars in the specimens were modelled with either the DIANA option
"embeddedreinforcements" or separatesteel elements using truss elements. In the embedded
reinforcement option, the reinforcement does not have separatedegrees of freedom; instead
the strength and stiffness of the concrete elements are increased in the direction of the
embedded reinforcement. With this model, perfect bond is assumed between the
reinforcement and the surrounding material. When the interaction between the reinforcement
and the concrete was taken into consideration, the reinforcement bars were modelled by
separateelements, using truss elements in combination with interface elements, see Section
4.2.3.The von Mises yield criterion with associatedflow and isotropic strain hardening was
used to describethe constitutive behaviour of the reinforcement. The modulus of elasticity and
the mechanical properties of the reinforcement used in the FE analyses are shown in
Appendix B. The Poisson'sratio was set to 0.3.
When testing a reinforcement bal to obtain its material properties, the plastic deformations
will localise to a short length of the bar once the ultimate strength of the steel material has
been reached. However, the strain in the stress-strainrelation obtained from such a test is
calculated from the extension of the bar divided by the length of the bar. Thus, the locaiised
deformations are smearedout over the whole length of the bar tested. Therefore, to take this
into considerationwhen determining the stress-strainrelation of the steel used in the analyses,
the localised deformations were smeared out over one reinforcement element, which resulted
in a less steepstress-sffainrelation for the softening branch. The strain exceeding the strain at
maximum stresswas modified according to equation (5), see Figure 24.
27
Stress,o, [MPa]
800 - ,A€
T--1
600
--o- Originalstress-straini
-r- Modifiedstress-strain
300 350
Strain,e, [l0-3]
f
'f
(6)
^, mod
tt - -:-- (7)
I
28
o l M Pal
I Stress-strain f" f,,
f',),
relation lMPal lMPal
f
High ratio Jt3 I1.3 1.30 1.1
Normalratio )/J 675 1.18 1.0
Low ratio <7? 608 r.06 0.9
e, [Vo]
29
Bond-slip
14-nodeinterfaceelement
)
Concrete
(4-nodeplanestresselement)
Reinforcementbar
(2-nodetrusselement)
Figure 26 Finite element idealisation in DIANA of reinforced concrete elements using truss
and interface elements.
The bond-slip relation between the reinforcement and the concrete used in the detailed
analyses was approximated according to the CEB-Ftr Model Code, CEB (1993) where the
bond stress,r, is given as a function of the relative displacement,s.
The CEB-FIP Model Code accountsfor the effect of splitting cracks by giving different bond-
slip relations for confined and unconfined concrete.Since no splitting cracks were observedin
the test specimens, "confined concrete" was assumed in the FE analyses. According to
Engstrom (1992), the bond stressdecreasesconsiderably when the reinforcement steel yields,
see Figure 27. ln the bond-slip relation proposed by Engstrom, the bond stress is not given
explicitly, but depends on when the steel reinforcement yields. However, the principal bond-
slip relation for this yield case is similar to the relation given when "unconfinedconcrete"is
assumed; therefore, as an approximation, "unconfined concrete" was assumed for the
reinforcementbars where yielding of the reinforcing steel was likely to occur.
i
I
I
L - -
.M
CEB-FIPModelCode
30
r IMPa]
Bond-stress,
+ "Good"bondcondition
l0
+ "Other"bondcondition
+ "Bad"bondcondition
3
Slip,s [mm]
t^F
'-F
5.!
t+NF t+ar
F
,F
f
a) b)
I+41..
u
c)
4,4.1 General
The frame corner specimenstestedin the secondtest serieswere analysedusing the finite
elementmethod.The analyses werecarriedout at two differentdetaillevels.In the analyses
of
the general response,perfect bond between the reinforcementbars and the concrete was
assumed,using the embeddedreinforcementoption. In the more detailedanalyses,truss
elementsin combinationwith interfaceelementswereusedto simulatethe reinforcement bars
in the region ciose to the frame corner area;thus, the interactionbetweenthe reinforcement
and the sunoundingconcretewas included. Embeddedreinforcementwas used in the
remainingpart of the model, to model the reinforcementbars. In the testsof specimenswith
high reinforcementratio, the side concretecover in the corner area spalledoff. As this
behaviourcouldnot be simulatedin the FE modelsused,thesespecimens wereanalysed using
only the lessdetailedmodel.The specimenswith low reinforcement detailingwere analysed
JJ
using both models.The FE models used do not take into accountthe placementof the
reinforcement bars,i.e. whetherthey arein contactwith eachotheror not.Therefore,the same
modelswereusedin the analyses of the specimenswhetheror not the reinforcement
barswere
in contactwith eachother;i.e. one modelwas usedto simulatespecimens RV5 andRV6, and
anothermodelwas usedto simulatespecimens RV7 andRV8.
J+
Figure 3l The finite element mesh and the position of steel reinforcement in the model for
analysing the general responseof the frame corner.
v
F-------->
Numberof effectivereinforcement
barswith respectto the anchorage
I capacrty
vl
Numberof @16
Numberof @10
Positiony Positiony
Numberof reinforcement barsin Numberof reinforcementbarsin
modelwith hieh reinforcement
ratio modelwith low reinforcement
ratio
35
4.4.3 The FE modelfor detailedanalvses
To model the geometry of the test specimens, a total of 322 two-dimensional plane stress
elements were used, see Figure 33. Truss elements in combination with interface elements
(a total of 137 each) were used to model more accurately the reinforcement bars within 1.0 m
of the corner, where cracking of the concrete was expected. Outside this area, the
reinforcement bars were modelled with embedded reinforcement, see Figure 34.
The test specimenswere cast with a construction joint, see Figure 6. The test results indicated
thar this joint exhibited a zone of weakness in the matenal in which cracking first occurred,
see Section3.3.3.To model the weaknessof the constructionjoint, a thin row of elementswas
used in which weaker material parameters were given, see Figure 35. In this way, the
constnlction joint was smeared out in the FE model, meaning that any reinforcement bars
placed in the element row simulating the constructionloint would be affected.The straight
reinforcement bars in the column did not reach into the comer area and were not, therefore,
affected by the construction joint. Thus, the straight tensile reinforcement bars in the columns
of the frame cornerstestedwere modelled to end just below the row of elementsmodelling
the constructionjoint, seeFigure 35.
36
- = Trusselement
(4@10)
- = Trusselement
(3d10)
Figure34 Modelling of the reinforcementbars in the frame comer with new reinforcement
detailing.
Row of elements
modellingthe
joint
construction
The straightreinforce-
mentbarsendbelow
the constructionjoint
JI
On the constructionsite,it sometimeshappensthat the reinforcement barsare not positioned
accordingto the designers'drawings. In this work, it is of interestto studythe consequences of
incorrectlypositionedreinforcementloops. Therefore,an analysisof reinforcementloops
incorrectlypositionedin the column was carried out, see Figure 36. When using truss
elementsin combinationwith interfaceelements,the elementmesh modellingthe concrete
dependson the reinforcementdetailing, see Section 4.2.3. To use truss elementsin
combinationwith interfaceelementsfor modelling the reinforcementdetailing shown in
Figure36, quite a complicatedfinite elementmeshfor the concreteelementsis necessary. In
the analyseswhere all reinforcementbars in the corner area were modelled with a
combinationof trusselementsand interfaceelements, the crackpropagationin the cornerwas
limited;thus,the slip of the reinforcement in the bendof the reinforcement barswas tolerably
small.This meansthat the assumptionof perfectbond for the bendpart of the reinforcement
loops is appropriate. Therefore,as an approximation,embeddedreinforcementwas usedto
model the loop of the reinforcementbars in the corner area.Parallelto this, a new, less
complicatedelementmeshin the cornerareawasusedto studyits effecton the crackpattern
in thecornerandthe structuralbehaviourof the specimen. seeFigure37.
= Embedded
reinforcement
- =Truss element
38
4.4.3.2 Conventionalreinforcementdetailing
39
- = Trusselement(3d10)
= Trusselement( 1@10)
= Embedded (3@10)
reinforcement
Figure 39 Modelling of the reinforcement bars in the frame corner with conventional
reinforcement detailing.
4.5.1 General
Two different iteration methods,the Modified Newton-Raphson method and the BFGS secant
stiffness method, were used in the FE analyses,see Section 4.3. ln the analysesof the general
response,only the Modified Newton-Raphson method was used. However, in the anaiysesof
the refined FE model it was found that fewer numerical problems were encounteredwith the
BFGS secant method; accordingly, this was the main iteration method used. Although the
Modified Newton-Raphson method gave a somewhat smoother load-displacement relation
than that achieved with the BFGS secant method. the difference in the effect of these iteration
methodson the resultswas negligible.seeFigure 40.
40
Load,F [kN]
50-
40*
-BFGS
30 T
- Modified
Newton-Raphson
20
10
Tl l
l0 15 20 25 30
Displacement,
6 [mm]
Figure40 Comparisonof the FE analysesof the frame corner with new reinforcement
detailinewhenusinqdifferentiterationmethods.
t-oad, F [kN]
50T
:
ooT
Concretein compression
30 'r usedin the analyses
(0 = 30", K= EP untuiat)
Concretein compression
that should/ravebeenused
( 0 = l o " , r = 1 . 1 6€ P , n , - , o t )
20 30
Displacement,
5 [mm]
A1
+l
lever arm remainedapproximatelythe same,independentof the strengthof the concrete.
Unlessotherwisestated,the resultspresentedin this study are from analysesin which the
incorrect modelling of the concretein compressionwas used. However, the negligible
differencein the behaviourof the structureshownin Figure41, whenusingdifferentvaluesof
theconcretein compression. showsthatthe resultsarestill valid.
Load,F [kN]
120
100 SpecimenRV5
80 * SWcrnen
nv6
.
60 ]
- FE analysis
40 l
z0
40 50
Displacenent,5
[mm]
A'
[nad tkNl
50 I
40+
JU* -- SpecimenRV7
+ Specimen
RV8
20i
l analysis
l0{
60 80
Displacement,
6
Figure43 Comparisonof the FE analysesof the generalresponseand the test resultsfor the
with low reinforcement
specimens ratio.
SpecimenRV5
F = 1.47kN, 6= 29 mm
Tensileforce [kN] SpecimenRV6
120 F = 1 5 0k N , d = 3 3 m m
-'- Planestressanalysis
r00
F = l'76kN, 6= 24 mm
80
60
40
20
Straingaugeon
reinforcement
200 400 t000 1200 1400
Positionx [mm]
43
-._ SpecimenRV7
Tensionforce [kN] F=40kN, d=28mm
* SpecimenRV6
50-
F=40kN,6=56mm
-"- Planestressanalysis
40
F=40kN,6=23mm
30
)o
10
f, - Strain gauge on
reinforcement
200 400 600 800 1000 1200 1400
P o s i t i o xn I m m l
4.5.3.1 Objectivesandpreconditions
As in the analysesof the generalresponse, it was not possibleto predictthe behaviourof the
frame corner throughouttotal failure. All the detailedanalysesof the frame corner were
disrupteddue to numericalproblemswhen the ultimatecompression strengthof the concrete
was reachedat one integrationpoint in the corner area. At this stage,yielding of the
reinforcement barshad occunedand,in somecases,alsostartedto harden.To studythe effect
that different mechanicalpropertiesof the steel reinforcementhave on the frame corner,
substantialhardeningof the steelis presumed.However,in the analysesof the framecorner,
sufficient strain of the reinforcingsteel was not reachedand, therefore,the mechanical
propertystudy was not made with the frame cornermodel. Insteada simpler model of a
cantileverbeamwas usedfor this studv.seeSection4.7.
AA
4.5.1.2 \eureinforcementdetailing
Figure ;16 crack patternsobtarned for test specimen s RV7 ( left ) and RV8 ( n-qhr).
l' ,l tl
tl ' ll tIl t I I
t, l
Figure 47 Crack pattern at the end of the analysis for a frame corner with the new
reinforcement detailing when assuming "good" bond condition.
Figure48 Crack pattern at the end of the analysisfor a frame corner wirh the new
detailingwhenassuming"other"bondcondition.
reinforcement
46
Figure49 Crack pattern at the end of the analysis for a frame corner with the new
reinforcementdetailingwhenassuming"bad"bondcondition.
Load, F [kN]
50 r
40t
I -+ SpecimenRV7
30* +
I
SpecimenRV8
"Good"bond
)o+
! condition
"Other"bond
condition
10
"Bad" bond
condition
20 25 30
Displacement,
6 [mm]
Figure50 Comparison
of the load-displacement
relationsfor threedifferentbondconditions.
The distribution of the tensileforces along the reinforcementloops, in the testsand analyses,
whenyieldingofthe steelreinforcement hasoccurred,arepresented in Figures51 and52.The
differencein the distributedtensileforces,observedfor the differentbond-sliprelations,was
47
dueto the appearanceof majorcracksin the columnand the beamnearthe cornerarea.In the
analysiswhere "bad" bond condition was assumed,the variationof the tensile force was
approximatelylinear.This was becauseof the largespacebetweenthe cracksin the column
andthe beam.For the "good"and "other"bond-conditions, a shift in the tensileforceresulted
dueto the appearanceof a cracknearthe corner,seeFigures47 to 49.
x,^.
c - Strain gauge on
reinforcement
1000 1200 1400
Position
x [mm]
Tensile
Force[kN] "Good" bond condition
50- F=43kN,6=18mm
"Other" bond condition
F = 43 k-|{,6 = 18rnmi
40- / "Bad" bond condition l
F=41 kN, 6=20mm
Figure52 Distribution of the tensile forcesalong the reinforcementbarsin the frame comer
for the FE analyses of differentassumptions of the bondcondition.
48
The effect of the weaknessin the construclionjoint on the structuralbehaviourof the frame
cornerwas examined.By modellingthe construction joint with a slightweakness, the first
crack was madeto form at the sameplaceas in the tests.Therefore,an FE analysiswith a
tensilestrengthand fractureenergyof 90 Vo,of that usedin the restof the model,wascarried
out. The load-displacement relationsof the FE analysesusing different strengthof the
constructionjoint are comparedin Figure 53. The changesin the crack pattern and the
distributionof tensile forces in the frame comer area, causedby the weaknessof the
constructionjoint,areshownin Figures54 and55,respectively.
Load,F [kN]
50r
40+
30r
+ SpecimenRV8
20*
- 50 7ostrength
- 90 7ostrength
l0 t5 20 25 30
Displacement,
d Imm]
'r--l
5{ f +
.|vr
I
|i- l
. ' -\ - \\ U4 lri
\VJZ
Itt
\\\t-
t ,l
/
F 'ilrl
\
49
TensileForce[kN]
50 7ostrenglh
5ol F=44kN,6=26mm
90 7ostrength
40+ F=44kN,6=26mm
30*
2O'
l0
1- ---
Figure55 Comparisonof the tensile forces in the frame corner when using different
joint.
for theconcretein the construclion
strengths
Load,F [kN]
50-
40*
- Reinforcement looDs modelled
usingtrusselements
JU-
(onginalmesh)
Reinforcementloopsmodelled
20 using embedded reinforcement
(originalmesh)
Reinforcementloopsmodelled
usingembedded reinforcement
(simplifiedmesh)
70 30
Displacement,
6 [mm]
of
relationsfor differentmodellingschemes
Figure56 Comparisonof the load-displacement
thecornerarea.seeSection4.4.3.1.
50
reinforcement loops was preventedfrom slipping,a somewhathighertensilestrainwith
consequent highertensilestressin the reinforcementbarswas the result.Parallelto this.the
effectof differentelementmeshesfor the crackpatternin the cornerarea,was examined.In
Figure57,theresultingcrackpatterns in thecorner,whenusingtheoriginalandthesimplified
meshshownin Figure37, arecompared.It canbe seenthatthecrackpatternis affectedby the
elementmesh and that the directionof the crackstendsto be parallelto the edgesof the
elements. The effectof the load-displacement relationwhen the reinforcement loopsin tbe
columnarepositioned incorrectis shownin Fieure58.
\t l--t r-.\\*ljf -l
{n-1 r r
\\ \\ \ll I
\ F - iI - \ \\ llF-
lr-
IFI] ill
lr {t ll..i
x:=i ll v/
\ \\ v r \ \\\\)- I
) l)
__*.{i
Figure57 Crackpatternin the framecomer when usingthe original FE mesh(left) and the
simplified mesh (right), see Figure 37. Embeddedreinforcementwas used to
modelthe bendof the reinforcement loops.
Load, F [kN]
5or
40+
Correctpositionof the
30* rehforcementloops
(simplifiedmesh)
l
20 30
Displacement,
6 [mm]
51
;1.5.3.3 Conventionalreinforcementdetailine
The crack pattern ol a test specimen with low, reinfrtrcement rarro tspeclmen RV-li.
P l o s t l 9 9 ' 1 a .b ) . c a r r i e c ol u t n r t h t h e c o n l e n t i o n a ld e t a r l i n - ies s h o w n i n F i g u r e 5 9 . A l t h o u g h
the rtrength of the concrete and the steel reinforcementused in thrs specimen were not thc
\arlle as ln the FE analtses. a comparison ofthe crack paftern can sttll be ntade. The crack
pattern (at the end of the analvses)and the load-displacementrelation tbr rhe 'good" and rhc
"other" bond conditions are shorvn in Figures60 to 62. The denser
crack pattern obtaineclin
l l t c r t n l i t s c ' so f t h e f r a m ec o r n e ru i t h c o n v e n t i o n arle i n t - o r c e m e dn et t a i J i n gc. o n t p a r c dr r i t h t h c
I t , t l l t ce r r r n c rt t i t h t h c n e u c l e t a r i i n cn. a s d u e t o t h e g r e a t e ra n r o u n to f r c r n i t l r c e r n c nrtn t h r '
\ l ! l r l l l \ o l t h e c o r n e ra r e a .\ e e F i g u r e s- l - 1a n d 3 9 . T h e d i s t r r b u t i o no t ' t h c .t e n s i l el i r r c e su i o n g
the retttlitrcenlcntbars ln the frame corner tbr the FE analvseswhere "good' and "othcr' borrtl
c o n d i t i o n su e r e a s s u m e c al .r e s h o u n i n F i e u r e6 3 .
Filur.r'-5t) Crack pattcrn oi' a test specimen {specinten RV j r carried out u ith thc
c o n v e n t l o n arl e i n f o r c e n t e ndte t a i i i n g p
. i o s t 1 9 9 , 1 ab.) .
52
Figure 60 Crack pattern at the end of the analysis for a frame comer with the conventional
reinforcement detailing when assuming "good" bond condition.
Figure6l Crackpattemat the end of the analysisfor a frame cornerwith the conventional
detailingwhenassuming"other"bondcondition.
reinforcement
53
I-oad,F [kN]
60 -r
50+
40:
30 -1-
:(
1,, "Good"bond
condition
"Other"bond
20 c91qq9t
l0 \
^
20 30 40
Displacement,
d [mm]
As for the frame corner with new reinforcementdetailing,a comparisonof the effect of
differentweaknessesof theconstructionjoint on the structuralbehaviour,wascarriedout.The
load-displacementrelationsfor different strengthsof the constructionjoint are shown in
Fieure64.
54
tnad,F [kN]
60-
50r
40-
% strength
30-
- 90 70strength
l0 20 30 40
Displacement,
6 [mm]
55
toad,F [kN]
60-
50-
40:
- Conventionalreinforcement
30t detailing
- New reinforcement
detailing
t0 20 30 40
Displacement,
d [mm]
Figure 65 Comparison of the load-displacementrelation for frame comers carried out with
new and with conventionalreinforcementdctailing.
Conventionalrein-
TensileForce[kNl
forcement detarling
50-
F=50kN,6=20mm
)b
\ \l
(l
\( si
\\/l I
il
\ NI
\t\ .
,.
l l
|
-
//
- tN
/t- |
+#l - +#t
-x)q I
----+b-l
/ \ \\ \ I
. \\ \({l li {ll!llJl l: I
j
\\\\ \ r +) l!{i it l "-
.\\)l$#)l l"
'\\ \tlx'rlI
I
I
Niiiii#ii]1'
-\\\\ \\ \+lvl lt I
^-\\\\+\\\\\
---::\\\\\\\\
ll
\\\\\\ /t /t ,l
\\\\\\ /t / \\f\X\ \\\\\ / /
\ \\\\\ / / \.\\\)\N\ /l
\Nl . \ \ \ \\N|2
++s+-)\\-
==I:F; f
{--
:' -l * - - - - z
I lF=49
i- ',il l l
/*- tt
lI d = 3 0
- -
----x>*-r + # t l I _ _ _ ? x ) a r _+l # l
;-TfiY]1 .- \ \tf-l
5'l
TensileForce[kN] = 55 kN, 6= 26 mm
50 F=50kN,6=28mm
F=49kN,6=30mm
40 F = 49 k'-f{,d= 32 mm
r0-
0* -
0 200 400 600 800 1000 1200 1400
Position
x [mm]
58
L,oad,F ikNl
r 8 0-
1 6 0r
1 4 0*
t20 - -:-
100 * lr TestspecimenRV5
T, + TestspecimenRV6
80+ -/K;.
- Full fractureenergy
60* K
40 I z \\
- Halvedfractureenergy
20
^
0
r0 ^!-ll 30 40 50
Displacement,6
[mm]
t oad,F [kNl
50
40-
30i
+ TestspecimenRV7
+ TestspecimenRV8
- Full fractureenergy
- Halvedfractureenergy
20 30
Displacement,
S [mm]
)v
Table6 Comparisonof the maximum load capacitiesand correspondingdisplacements
obtainedin the FE analyses
andobservedin the tests.
60
Lnad,F [kN]
50=
40-
30+
+ TestspecimenRV7
+ TestspecimenRV8
- Plare stressanalysis
'ot - Planestrainanalysis
l0 l5 20 25 30
Displacement,6
[mm]
4.6 Discussion
The resultsof the FE analyses conesponded quite well with the resultsfrom the experiments,
as can be seenin Section4.5; the behaviourwas similarfor both the maximumloadcapacity
obtained and the stiffness of the structure.Furthermore,the agreementbetween test
observations and FE analyseswas quite good for the crack patternand the distributionof
tensileforcesalong the reinforcementbarsin the frame corner.
bi
conventionaland the new reinforcementdetailingsbehavedapproximatelythe samefor these
This statementis still moreevidentfor the specimens
specimens. with low reinforcement
ratio
becauseof the similaritiesin the load-displacement relationsobservedin the tests with
conventionalandnew reinforcement detailings.
62
beam into the column did not have any effect on the critical sectionadjacentto the corner,the
bent pan of them was removed,seeFigure 72. Thereby,the propagationof a largecrack above
the bent bars was simulated in an approximate way. As can be seen in Figure 73, this
modification had a notable effect on the stnrctural behaviour after yielding of the
reinforcement,resulting in a load-displacement relation similar to that obtained in the frame
corner with new detarling. Furthermore, the distribution of tensile forces along the
reinforcement bars in the comer was similar to that obtained when using the new
reinforcement detailing. Consequently,plastic hinges may develop differently in frame
corners made with the new and the conventional reinforcementdetailings. With the new
reinforcementdetailing, plastic hinges always develop adjacent to the corner in both the
column and the beam. However, in a frame comer with the conventional reinforcement
detailing, the main plastic hinge may develop, dependingon the crack propagation.rn the
section below the construction joint where the bent bars, reaching out from the beam, end.
The statementsmade above, treating the effect of the construction joint and the bond-slip
relations on the structural behaviour, hold true also for a frame corner made with the
conventionalreinforcementdetailing.
The analysescarried out with different element meshes in the corner region have shown that
the cracks tend to propagate parallel to the mesh lines. This phenomenon has also been
observedby Rots (1988), who explainsthat it is causedby interlockingbetweenthe elements
and coupling between the integration points. lmprovements can be made by rotating the
element mesh and thereby adapting it to the expected crack directions. However, this
procedureis undesirablesince it may severelydecreasethe simplicity of the mesh generation
when using the smeared crack approach. Therefore, Rots suggestsusing triangular elements
placed in a cross-diagonalmesh, thus increasingthe number of lines which the cracks can
follow, see Figure 74a. However, for the analyses carried out in this study, the use of a
triangular bisectional mesh in the comer probably would have been sufficient, see Figure 74b.
Thesepossibleimprovementsof the elementmesh was not examinedin the analvses,
Possible
crack
path
63
Load,F [kN]
50-
40
30 -- - Conventionalreinforcement
detailing(modified)
20+ - New reinforcement
detailins
10 -r
20 30
Displacement,
6 [mm]
Figure 73 Load-displacement relation for frame corners carried out with new and with
conventional(modified as shown in Figure 72) reinforcementdetailing.
a) D)
The effect of the fractureenergyused in the analyseswas examined.It was found that a
reductionto 50 Voin the fractureenergyhad relativelylittle effect of the structuralbehaviour
in the analysesof the specimenswith either the high or low reinforcementratio; its
importancewas proportionallyhigherfor the specimenswith the low reinforcement ratio. In
conjunctionwith the slightlyhigherloadcapacitiesobtainedin the detailedanalyses compared
to that observedin the tests,this indicatesthat the fractureenergyusedin the analysesmight
have been somewhattoo high. However, it does not affect the results presentedand
conclusions drawnin this study.
4.7.1 General
To examinethe effect of the bond conditionon the load and the deformationcapacityof
structuresthat have failed from differentcauses,the analysesof the cantileverbeam were
carried out with two different reinforcementratios. A low amount of reinforcement,
conesponding to 4 010 1p = 0.20),was usedto obtainruptureof the reinforcementbarsanda
higheramount,conesponding to 4 Q16 (p = 0.52), was usedto reach ultimatecompressive
stressin theconcreteshortlyafteryieldingof the reinforcementhad occurred.
65
aa aa aa aa
i
, 600 \,
l---_--'''___-
A-A
(- Trusselement
I 4 Qtl (4 Qr6)
300I
,
7-
i
300I 700(9s0)
t
300
/
ln the analysesof the beamwith low reinforcementratio, the concretein compression never
reachedits ultimatestrength;consequently the use of the incorrectstress-strain
relationhad
negligibleeffecton the results.However,for the cantileverbeamwith higherreinforcement
ratio,crushingof the concretewas the causeof failure,which showsthat it is moreimponant
to simulatethe compressed concreteaccurately.This is why the more accuratestress-strain
relationfor concretein compressionwas used.Further,to examinethe importanceof the
descending branchfor concretein compression,analysesusing a more gradualstress-strain
relation(denotedmodifiedstress-strainreiation)werealsocarriedout, seeFigure76.
66
Stress.
o, [MPal
30-
20
Strain,
e. [10'3]
r = Node constrainedin
x and y directions
L_
Figure77 Finiteelementmeshandboundaryconditionsof thecantileverbeam.
o/
4.7.3 Resultsof the analyses
4.7.3.1 General
Unlessotherwisestated,thefollowing assumptions
weremadein the FE analyses:
r bondcondition - "good"(seeSection4.2.3),
r reinforcementtype = "normalratio"(seeSection4.2.2),and
o iterationmethod = BFGSsecantmethod(seeSection4.3).
t oad,F [kN]
40_
- BFGS secant
method
- Modrfied Newton-
Raphson
r00 150
Displacement,
6 [mm]
Figure 78 Comparison of the effect of the iteration methods used in the FE analysesof the
cantileverbeam.
68
Load,F [k]Jl
40-
- ,'Good,i
bond
condition
- ,,other,'bond
condition
50 100 150
Displacement,
6 [mm]
l-oad,F [kN]
i
100 : modrtled stress-
straln relatlon L
80:
l,/-
60* originalstress-
strainrelation - ,'Good"bond
40' condition
t0 15 20 25
Displacement,
6 [mm]
69
4.7.3.3 Effect of the reinforcement type
tnad, F [kN]
40-
- ,'High
rario',
- 'rNormalratio''
"[,ow ratio
r00 150
Displacement,
6 [mm]
Figure8 I load-displacement
relationfor the FE analysesof thecantileverbeamfor different
mechanicalpropertiesof the steelreinforcemenr.
10
4.7.4 Discussion
When ruptureof the reinforcement barsis the causeof final farlure,the valueof the ultimate
probablyaffectsthe total displacement
strainof the steelreinforcement in the analysesof the
cantileverbeam.For a stnicturewherecrushingof the concreteis the causeof failure, the
importanceof the steelductilitv dependson what strarnvalueshave been reached.If the
11
reinforcement barshavejust startedto yield whenthis happens(asin the cantileverbeamwith
the high reinforcementratio analysedabove)the ductility of the steelhas no effect at all.
However,when the reinforcementbars have startedto harden,the mechanicalpropertiesof
the steelmay be important,sinceit is thesethat determinethe capacityfor redistributionof
forcesin a concrete seeOberg( 1976).
structure,
72
\ CONCLUSIONS
5.1 General
The tests and the FE analysesconductedhave shown that the conventionaland the new
reinforcement detailingsfor practicalpurposesareequivalentfor a framecornerstructurewith
a low amountof reinforcement. Comparisons of testsandFE analyses indicatethat this is also
the casefor a framecornerwith a high amountof reinforcement. Thus,the testsand analyses
supportthe idea that the new alternativeis suitableto use insteadof the conventional
reinforcement detailing.
To obtain a ductile structure,the most important factor is the yielding of the steel
reinforcement; the longerthe reinforcementbarshave yieldedbeforefinal failure,the more
ductile is the behaviourobtained.However, the bond-slip relation and the mechanical
propertiesof the steel reinforcementcan also have significanteffect on the deformation
capacity.The FE analyseshave shown that the stiffnessof the structureis affectedby the
bond-sliprelationup to the point at which the steelreinforcementstartsto yield: the higher
the stiffnessof the bond-sliprelation,the higherthe stiffnessof the structure.Dependingon
the causeof final failurethe bond-sliprelationcanhavea noticeableeffecton the deformation
capacityof a structure.When ruptureof the reinforcement barsis the causeof final failure,a
weak bond-sliprelationhas a positiveeffect;when crushingof the concreteis the causeof
failureandhardeningof the steelreinforcement is reached,a stiff bond-sliprelationmay have
a positiveeffect. However,when hardeningof the steel reinforcementis not reachedand
crushingof the concreteis the causeof failure,the bond-sliprelationhasnegligibleeffecton
the deformationcapacityof the structure.The maximumloadcapacityis relativelyunaffected
by the bond-sliprelation,independentof the amountof reinforcement.However,a stiffer
bond-sliprelation does have a positive effect on the crack width, resultingin more, but
smaller,cracksin a structurewhich can be positiveconcerningthe risk of corrosionof the
't3
steel reinforcement.It was also shown that the mechanicalproperties of the steel
reinforcement effecton how largethe regionof yieldedreinforcement
can havea considerable
will be; consequently,it can have a significanteffect on the deformationcapacityof a
structure.
tn this study, tests and FE analyseshave been carried out only for a static loading. However, a
civil defencesheltermust withstand impact loading such as explosionsand falling buildings;
consequently, the behaviour of the new reinforcement detailing when subjected to impact
loading needs to be studied.
The use of reinforcement loops has been examined only in frame comers connecting two
structural members, i.e. a beam and a column. Therefore, an examination of the use of
reinforcement loops in other t)?es of connections,e.g. T-joints or corners with an angle wider
than 90o, would be valuable.
aA
6 REFERENCES
and CementaAB (1990):Betonghandbok
AB SvenskByggtyanst Konstruktion,utg6va2
(Concrete Design,second
Handbook 791pp.
AB SvenskByggqiinst'
edition.In Swedish).
(1987),Betongprovning
BST Byggstandardiseringen med SvenskStandard, BST handbok12,
(ConcreteTesting accordingto Swedish Standard,BST Handbook 12. In Swedish).
SIS- standardiseringskommisionen 272 pp.
i Sverigeoch SvenskByggtliinst,
CEB (1993): CEB-FIP Model Code 1990, Design Code. Thomas Telford, l-ausanne,
437 pp.
Switzerland,
/)
EngstrdmB. (1992):Ductility of Tie Connectionsin PrecastStructures.
Division of Concrete
ChalmersUniversityof Technology,Publication92:1,Gdteborg,452pp.
Structures,
GambarovaP. G. (1989):Steel-to-concrete
bondafterconcretesplitting:testresults.Materials
1989,22,pp. 35-47.
andStructures,
GylltoftK. (1983):Fracture
MechanicsModelsfor Fatiguein Concrete
Structures.
Divisionof
Structural
Engineering,LuleAUniversityof Technology,
DoctoralThesis1983:25D,
210pp.
76
ConcreteComersandJointsSubjected
NilssonI. H. E. andLosbergA. (1976):Rernforced to
BendingMoment.ASCEJoumalof theStructural Division,102(6),pp. 1229-1254.
M. J. N. (1978):Reinforced
PaulayT., ParkP. Priestley ConcreteBeam-Column JointsUnder
SeismicActions.ACI Journal,Proceedings V. 75,No. I 1,Nov.,pp. 585-593.
fcirramhorniskyddsrum(New Reinforcement
PlosM. (1994a):Ny armeringsskarv Splicefor
Frame Cornersin Civil Defence Shelters.In Swedish).Division of ConcreteStructures,
Chalmers Report94:2,21pp.
Universityof Technology,
in FrameComers-ExperimentalStudies.Nordic
PlosM. (i994b): Splicingof Reinforcement
ConcreteResearch,Publ.no. 14,The NordicConcrete pp 103-121.
Federation,
't'7
TNO (1993): DIANA-5.I. TNO Building and ConstructionResearch- Departmentof
Mechanics,P.O.Box 49,2600AA Delft, The Netherlands.
Computational
78
APPENDIX A Drawingsof the TestSpecimens
in the
SecondTestSeries
The dimensions of the framecomerspecimens are shownin FigureA-1. The amountand
posrtion
of thereinforcementbarsareshownin FiguresA-2 andA-3.
50
ol0
'\ 150
\tl , / l -----t---t
,z
0 3oF ---1'r
<ni300
45y
Construction
Jolnt 2150
0lQ o
150
| # ----jt. -^
,,9 +r)u
1,/ L
so Y I l rso
^^^l
6-- s2 7 Ql6K500(l=1720)
r=118/>------------
l. 1380 L
5 dl6 K500( I = 1720\
-5p16K500t/=1830)
5 d l 6 K 5 0 0( / = 1 8 3 0)
5016K500(l=1'720)
Concretecover24 mm
FigureA-2 Reinforcement
barsusedin specimens
RV5 andRV6.
AI
s 2 4 p l 0 K 5 0 0/ r= 2 3 8 0 )
n
r=l
97
Concretecover l5 mm
FigureA-2 Reinforcement
barsusedin specimens
RV7 andRV8
A2
APPENDIX B Concrete in Compressionin the
FE Analyses
The responseof the concretein compression plasticity
was modelledby a Drucker-Prager
yield surfaceis givenby
model.The tbrmulationof the Drucker-Prager
ft-
J @ , K I= eo - u,ft'o - kkl ( Bl )
l-o'
2-r-1000 I
-12-1000 I
-1-12000 I
P: and (82)
0 0 0 600 0
0 0 0 060 0
0 0 0 006 U
2 s i n0 ( r ) ^ 6cos@n
(83)
u r, = # and P- _
3-sinp(r) J-slnpo
l-dr
c = J(.qt(e;^tu,.,) (B4)
p
Here, /...r(e P,n,^,or)is the compressive strength as a function of the plastic strain in the
direction of the uniaxial stress,evaluated from standardtests on cylinders. In the analysesthe
angle of internal friction was constant; thus, @(r) = Qo= Q and the expressionin equation (B4)
can be written as
l-sin@
c = J,..n1(€ini^Mt )-:--- - (Bs)
zcosp
The hardeningparameter
r is definedas
^F;;a
6 = -J---a t!,n,^,oi (86)
l- dt
BI
where
2 sinry(r)
A, '^ =------------ 187)
3-sinty(r)
1 . 1f6,
Equalbiaxialstress
f, state(d/ I o2= 11
Uniaxialstressstate
(O2=Or=0)
ffin
Hr \when biaxial
L4
B2
correspondsto that from the uniaxial test; however, the choice of p also affects the stress-
strain relation in the other stressstates.To obtain an rncreasein compressivestrengthin an
equal biaxial compressionstate,correspondingto that suggestedby Chen (16 Vc),the angleof
internal friction, 0, should be set to approximately 10'. However, according to
recommendations in the DL{NA manual,TNO ( 1993, 1996),the angle of internalfriction was
approximatedto be 30o in the FE analyses.This resulted in a 200 7c increaseof the
compressivestrengthat an equal biaxial compressionstate compared to that in an uniaxial
stressstate. Further, due to a misunderstanding,the hardening parameter r was set to be
equivalentto the plastic strain €t,,,-,o1 when it should have been r = l.9lEtu,,,*iolaccording
to equation(86); thus, the softeningof the concretein compressionoccurredat a lower strain
also for the uniaxral stressstate. To examine the effect of these inconect material parameters
(i.e. @= 30o and K= €Pun,^,ot), an analysisof a frame comer specimenusing more accurate
materialparameters(0= l0', K= 1.16€p,n,-,o1) was carried out, see Section4.5.1.The stress-
strain relations for this analysis and for the concrete used in the other FE analyses are
compared for uniaxial and equal braxial compression states,see Figure B-2.
Concrete in compression
biaxial that should have been used
60 ( 0 = 1 0 " ,r = l . l 6 t P u n i * , a t )
sressstate
-----;
\
40 biaxial (Chen)
ffin
2O+
uniaxial
HD \when
biaxial
stressstate
6
e.[103]
B3
Tryckt& Bunden
Vasastadens AB
Bokbinderi
r996
View publication stats