ECCOMAS Thema ic Con e ence on Compu a ional Me hods
in S uc u al Dynamics and Ea hquake Enginee ing
M. Papad akakis, D.C. Cha mpis, N.D. Laga os, Y. Tsompanakis (eds.)
Re hymno, C e e, G eece, 13–16 June 2007
RETROFITTING OF BRIDGE HOLLOW PIERS WITH CFRP
Ped o Delgado1*, Pa ício Rocha1, João Ped osa2, An ónio A êde2, Nelson Vila Pouca2,
Miguel San os3, Anibal Cos a4, Raimundo Delgado2
1 Poly echnic Ins i u e o Viana do Cas elo
Apa ado 574, 4901-908 Viana do Cas elo, Po ugal
[email p o ec ed], (pdelgado, p ocha)@es g.ip c.p
2 Facul y o Enginee ing o Uni e si y o Po o
R. D . Robe o F ias, s/n 4200-465 Po o, Po ugal
(aa ede, nelson p, delgado)@ e.up.p
3 S.T.A.P. – Repa ação, consolidação e modi icação de es u u as, S. A.
www.s ap.p
4 Depa men o Ci il Enginee ing, Uni e si y o A ei o
Campus Uni e si á io de San iago, 3810-193 A ei o, Po ugal
[email p o ec ed]
Keywo ds: Expe imen al es s, Non-linea cyclic beha io , Nume ical s uc u al modelling,
Seismic e o i , B idge pie s, CFRP.
Abs ac . Hollow b idge pie s gene ally ha e la ge sec ion dimensions, wi h ein o cemen ba s
sp ead along bo h wall aces. Unlike common solid sec ion columns, qui e o en he shea e ec has
g ea impo ance on he pie beha io . The e o e, i is o pa icula ele ance ha special a en ion is
gi en o his issue when he assessmen and e o i o RC hollow sec ion pie s is en isaged. Rep esen-
a i e o ypical b idge cons uc ion, RC pie s we e es ed a LESE – he Labo a o y o Ea hquake
and S uc u al Enginee ing o he Facul y o Enginee ing o Uni e si y o Po o. Expe imen al es s o
hollow sec ion pie s wi h squa e and ec angula c oss sec ions unde cyclic loading a e being ca -
ied ou in o de o compa e he esul s o he o iginal pie s and CFRP e o i ed pie s, ega ding
bene i s on hei s uc u al beha io and compa ing he esul s wi h analy ical p edic ions. The
adop ed nume ical me hodologies a e based on ini e elemen analysis using 3D elemen s wi h a Con-
inuum Damage Mechanics model o he conc e e unde ensile and comp essi e e e sals and uss
elemen s wi h he Menego o-Pin o model o he cyclic beha io o s eel ein o cemen . Fo di e en
ec angula c oss sec ions, he in e ac ion be ween pie walls is likely o a ec hei global beha io
and damage. The main pu pose o his pape is he e o e o p esen se e al s a egies o e o i wi h
CFRP in o de o p e en shea o lexu al collapse mechanisms, o bo h. Di e en amoun s o s ip
laye s we e applied o shea e o i and jacke con inemen nea he pie base sec ion was adop ed
o inc easing duc ili y. I is in ended o assess he s uc u al beha io and sa e y imp o emen due o
he adop ion o di e en CFRP e o i echniques and o illus a e he ex e nal and in e nal damage
pa e n. The need o in e io e o i is discussed on he basis o expe imen al e idence om some o
he es s; he e o e a s a egy o in e nal con inemen is also p esen ed in o de o p e en in e io
conc e e spalling and longi udinal eba buckling.
Ped o Delgado, Pa ício Rocha, Miguel San os, João Ped osa, An ónio A êde, Nelson Vila Pouca, Anibal Cos a
and Raimundo Delgado
2
1 INTRODUCTION
B idges and iaduc s a e, amongs all he s uc u es, hose ha sus ain he mos damage, as
clea ly demons a ed in se e al epo s o ecen ea hquakes. In compa a i e e ms, hese
consequences o b idge ulne abili y a e ound g ea e han hose obse ed in building s uc-
u es and, in mos cases, he b idge sa e y is limi ed and condi ioned by pie capaci ies. Se -
e al s udies and wo ks ha e been ca ied ou on solid pie s and can be applied o building
s uc u es [1]; howe e , o hollow pie s much less esea ch is ound in he li e a u e. Usually,
hollow pie s ha e la ge sec ion dimensions, wi h ein o cemen ba s sp ead along bo h wall
aces. Unlike common solid sec ion columns, qui e o en he shea e ec has g ea impo ance
on he pie beha io [2]. The e o e, i is o pa icula ele ance ha special a en ion is gi en
o his issue when he assessmen and e o i o RC hollow sec ion pie s is en isaged.
The main pu pose o his pape is o p esen an expe imen al campaign o ein o ced con-
c e e hollow sec ion pie s unde cyclic loading in o de o compa e he esul s o he o iginal
pie s wi h hose ob ained a e CFRP seismic e o i , o e alua e bene i s conce ning hei
s uc u al beha io and o compa e he esul s wi h analy ical model. Rec angula RC hollow
pie s ha e a pa icula s uc u al beha io , close o s uc u al walls, and he e o e a e mo e
di icul o simula e wi h simple nume ical ools; his ac is indeed a s ong mo i a ion o
his wo k. The adop ed nume ical me hodology eso s o sophis ica ed cons i u i e models,
simula ing he non-linea cyclic beha io o he conc e e by a cons i u i e model based on he
Con inuum Damage Mechanics and using 3D ini e elemen disc e iza ions o he conc e e,
while uss elemen s a e used o he s eel wi h a cyclic beha io model. Mo eo e , o di e -
en ec angula c oss sec ion dimensions, he in e ac ion be ween he pie walls can s ongly
in luence hei global beha io and obse ed damage. The e o e, a se o ec angula hollow
sec ion RC pie s, ep esen a i e o ypical b idge cons uc ion, we e es ed a LESE – he
Labo a o y o Ea hquake and S uc u al Enginee ing o he Facul y o Enginee ing o Uni-
e si y o Po o, [3][4]. The es se up was designed o impose cyclic ho izon al op displace-
men s wi h axial load and he possibili y o using wo o hogonal ac ua o s wi h a sliding
de ice ha allows pie op displacemen s and o a ions, ela i e o he e ical ac ua o ixed
o a s eel po al ame. The basic aim is, he e o e, o con ibu e o he s uc u al beha io and
sa e y imp o emen assessmen o di e en e o i echniques adop ed, while add essing he
ac ually obse ed ex e nal and in e nal damage pa e n. In addi ion, de elopmen and calib a-
ion is sough conce ning analy ical ools sui able o he e alua ion o cyclic esponse o ec-
angula hollow pie s.
2 EXPERIMENTAL CAMPAIGN
2.1 Tes ing se up
The es se up, shown in Figu e 1, makes use o a 500 kN ac ua o o apply la e al loads
and a 700 kN ac ua o o apply axial loads. The specimen and eac ion ame a e bol ed o he
s ong loo wi h high s eng h p es essed ods. A cons an axial load was applied du ing he
es s, he ein desc ibed, while he la e al loads we e cycled, unde displacemen con olled
condi ions. A special sliding de ice consis ing o wo s eel pla es, shown in Figu e 2, was
used o minimize he ic ion c ea ed by he axial loads. The lowe pla e is bonded o he
specimen op, whe eas he uppe is hinged o he e ical ac ua o , allowing op-end displace-
men s and o a ions on he specimens o ake place when la e al loading is imposed du ing he
es . The uppe pla e is also connec ed o a load cell o measu e he esidual ic ional o ce
be ween he wo pla es. Du ing he es s, he hyd aulic sys em o he e ical ac ua o was de-
signed o keep cons an he oil p essu e, in o de o main ain cons an he axial o ce. The
Ped o Delgado, Pa ício Rocha, Miguel San os, João Ped osa, An ónio A êde, Nelson Vila Pouca, Anibal Cos a
and Raimundo Delgado
3
ho izon al ac ua o con ol is done using a PXI con olle sys em om Na ional Ins umen s
(NI) and speci ically home de eloped con ol ou ines based on he LabVIEW so wa e pla -
o m (also om NI). The da a acquisi ion is also based on ano he PXI sys em equipped wi h
acquisi ion and signal condi ioning ca ds and allows di ec eading o da a om s ain gauges,
load cells, LVDTs (Linea Vol age Displacemen T ansduce s) and o he ypes o ampli ied
analogical o digi al senso s.
Figu e 1: Schema ic layou and iew o he es se up a LESE labo a o y.
Figu e 2: The sliding de ice used o apply he axial load.
2.2 Specimens and Ins umen a ion
The specimens p esen ed in his pape co espond o he second g oup o pie s es ed
wi hin his amewo k, being he esul s o he i s g oup al eady shown in p e ious epo s
[3], [4]. This se o specimens, consis ing on ec angula hollow sec ion RC pie s wi h 450mm
x 900mm ex e io dimensions and 75mm hick walls, was based on squa e pie s es ed a he
Labo a o y o Pa ia Uni e si y, I aly,[5] and is being es ed in o de o unde s and he in lu-
ence o he c oss sec ion geome y o ec angula hollow pie s on he cyclic beha io , bea ing
in mind he pu pose o assessing e o i ing solu ions. The uncon ined conc e e comp essi e
s eng h is 35 MPa and o bo h longi udinal and ans e sal ein o cemen he yielding
s eng h is 450 MPa. The model schemes shown in Figu e 3a co espond o ¼ scale ep esen-
a ions o hollow sec ion b idge pie s, he ein e e ed o as PO: PO1 o squa e sec ion and
PO2 o ec angula sec ion. Ins umen a ion o measu e cu a u e and shea de o ma ions
Ped o Delgado, Pa ício Rocha, Miguel San os, João Ped osa, An ónio A êde, Nelson Vila Pouca, Anibal Cos a
and Raimundo Delgado
4
was included along he pie heigh , because impo an shea de o ma ions we e expec ed in
hese es s. The LVDT con igu a ion used in bo h specimens is shown in Figu e 3b.
1-1 (1s se - PO1)
1-1 (2nd se - PO2)
(a) (b)
Figu e 3: Hollow RC pie s: a) model schemes and b) la e al LVDT layou .
3 CYCLIC TESTS AND NUMERICAL ANALYSIS
Two o iginal specimens we e es ed, namely he pie s PO2-N2 and PO2-N3 unde cons an
axial load o 250 kN and 440 kN, espec i ely. In o de o cha ac e ize he pie s cyclic beha -
io , h ee cycles we e applied o each o he ollowing peak d i a ios: 0.1%, 0.2%, 0.35%,
0.7%, 0.3%, 1.0%, 1.2%, 0.5%, 1.8%, 2.1% and 2.4%.
3.1 Pie damage
Conce ning he es esul s o o iginal specimens PO2-N2 and PO2-N3, he g ea es
amoun o damage was mainly obse ed a he la e al sides, eas and wes , whe e he conc e e
co e c ushed wi hin he en i e pie heigh (see Figu es 4 and 5) and s ong shea damage was
achie ed due o conc e e deg ada ion caused by lack o ans e se ein o cemen e iciency.
Li le damage was obse ed in he no h and sou h sides, wi h well dis ibu ed c acks. How-
e e , he c acks obse ed in hose sides a e no only ho izon al, as o he squa e pie s es ed
be o e [4], bu ins ead hey show an angle ha inc eases along he pie heigh , due he shea
lag e ec ha occu s o his wid h/heigh a io (2:1).
(a) No h side (b) Sou h side (c) Eas side (d) Wes side
Figu e 4: Pie PO2-N2 damage o 2.4% d i .
Ped o Delgado, Pa ício Rocha, Miguel San os, João Ped osa, An ónio A êde, Nelson Vila Pouca, Anibal Cos a
and Raimundo Delgado
5
(a) No h side (b) Sou h side (c) Eas side (d) Wes side
Figu e 5: Pie PO2-N3 damage o 2.4% d i .
3.2 Cyclic esul s and nume ical analysis
Figu e 6a shows he expe imen al esponses o bo h pie s PO2-N2 and PO2-N3 in e ms o
op o ce-displacemen diag ams; as expec ed, hese diag ams show la ge ini ial s i ness and
maximum o ces o he pie wi h highe axial load (PO2-N3). Howe e , ailu e o bo h pie s
was eached by he i s cycles o 25mm ampli ude (1.8% d i ), wi h isible shea ailu e
mode and he shea lag e ec e idenced in he damage pa e n shown in Figu es 4 and 5.
Conce ning he nume ical analysis o cyclic loading he CAST3M compu e code [6] was
adop ed, a gene al pu pose ini e elemen based p og am, whe e a wide a ie y o non-linea
beha io models a e a ailable and, pa icula ly, a damage model de eloped a FEUP [7] and
ecen ly implemen ed in CAST3M [8], ha has al eady p o ed o be sui able o seismic be-
ha io analysis o RC b idge pie s [9]. This la e modeling s a egy hus in ol es: he abo e
men ioned Con inuum Damage Mechanics based cons i u i e model o he conc e e zone
disc e ized in o 3D ini e elemen s and inco po a ing wo independen scala damage a iables
ha accoun o he deg ada ion induced by ensile o comp essi e s ess condi ions; he Giu -
é-Menego o-Pin o model [10] o he cyclic beha io simula ion o he s eel ein o cemen
disc e ized ia uss elemen s. Resul s o he nume ical model o pie PO2-N2 a e included in
Figu es 6b and 7.
-300
-200
-100
0
100
200
300
-50 -40 -30 -20 -10 0 10 20 30 40 50
Top Displacemen (mm)
Top Fo ce (kN)
Pie PO2_N2
Pie PO2_N3
-300
-200
-100
0
100
200
300
-50 -40 -30 -20 -10 0 10 20 30 40 50
Top Displacemen (mm)
Top Fo ce (kN)
Pie PO2-N2
Damage model - cyclic
(a) Expe imen al compa ison o PO2-N2 and PO2-N3 (b) Nume ical analysis o pie PO2-N2
Figu e 6: Expe imen al and nume ical esul s compa ison.

Ped o Delgado, Pa ício Rocha, Miguel San os, João Ped osa, An ónio A êde, Nelson Vila Pouca, Anibal Cos a
and Raimundo Delgado
6
(a) Sec ion disc e iza ion (b) Tensile damage o 5mm (c) Comp. damage o 25mm (d) Sec ion comp. damage
(e) De o med mesh ( ) T ans e se ein . s eng h (g) Longi udinal ein . s ain
Figu e 7: Nume ical esul s o PO2-N2 o 1.8% d i .
As al eady e e ed be o e, a shea ailu e mechanism was obse ed on he pie esponse;
hese e ec s we e sa is ac o ily cap u ed by he damage model as e idenced by he cyclic
cu es shown in Figu e 6b. The 3D ini e elemen disc e iza ion used in he damage model is
shown in Figu e 7a, bu in ac only hal o he c oss sec ion was modeled wi h adequa e
symme y condi ions. The esul o ensile damage pa e n is illus a ed in Figu e 7b o he
ini ial cycles, a ound 5mm, whe e i s c acks we e ound; he comp essi e damage pa e n is
also shown (Figu es 7c and 7d) o 1.8% d i , whe e some damage is obse ed a he pie
base. The de o med mesh o he ho izon al displacemen o 25mm (abou 1.8% d i ) is also
included in Figu e 7e. F om he esul s shown in Figu es 7 and 7g, i is possible o obse e
he s ess and s ain dis ibu ion, espec i ely, along he ans e sal and longi udinal ein-
o cemen ba s, o 1.8% d i , whe e s eel yielding is e idenced in ed. The ans e se ein-
o cemen s ess pa e n allows iden i ying he s u -and- ie shea mechanism de elopmen
and he longi udinal eba s ain dis ibu ion shows yielding in he ou e ba s abo e he oun-
da ion.
Wi h hese nume ical esul s i is possible o con i m he shea ailu e a ound he 25mm
cycle, bu wi h p e ious yielding o some longi udinal eba s nea he pie base. When com-
pa ed o he expe imen al es s, la ge o ces we e ob ained om nume ical esul s, possibly
due o mo e longi udinal eba s al eady wi hin he yielding phase.
4 THE RETROFIT PROCESS
A e he cyclic es s o he o iginal specimens, hey we e epai ed and e o i ed by an ex-
e nal con ac o (S.T.A.P.) acco ding o he ollowing s eps: 1) delimi a ion o he epai a ea;
2) emo al and cleaning o he damaged conc e e; 3) inside e o i wi h ans e sal s eel ba s;
Ped o Delgado, Pa ício Rocha, Miguel San os, João Ped osa, An ónio A êde, Nelson Vila Pouca, Anibal Cos a
and Raimundo Delgado
7
4) applica ion o o mwo k and new conc e e (Mic obe on, a p e-mixed mic o conc e e, modi-
ied wi h special addi i es o educe sh inkage in he plas ic and hyd aulic phase); 5) ou side
e o i wi h he CFRP shee s. In o de o p o ide a gene al idea o he pie damage and o he
e o i p ocess, he ollowing pho og aphs show he pie s du ing epai and a e ha ing been
e o i ed wi h CFRP shee jacke s (Figu e 8).
Figu e 8: Hollow pie s be o e and a e he shea e o i ing wi h inside s eel ba s and ou side CFRP shee .
The inside ans e sal s eel ba s (only o pie PO2-N3-R2) we e designed aking in ac-
coun he easibili y o u u e eal e o i s; such ba s we e concen a ed a he bo om, in co -
espondence wi h he ou e CFRP jacke s, o imp o ing he plas ic hinge con inemen . In
o de o design he ou side shea e o i wi h CFRP jacke s, he au ho s adop ed he P ies ley
app oach [11] o e alua e he hickness o he ec angula hollow pie jacke o inc easing he
shea s eng h abo e he maximum lexu al o ce while keeping he ini ial sec ion condi ions.
Acco ding o P ies ley me hodology he shea s eng h can be con eyed by Eq. (1) [11]:
sjpscd VVVVV
+
+
+
=
(1)
whe e Vc, Vs and Vp a e he shea o ce componen s accoun ing, espec i ely, o he nominal
s eng h o conc e e, he ans e se ein o cemen shea esis ing mechanism and he axial
comp ession o ce; he e m Vsj co esponds o he possible e o i con ibu ion wi h CFRP o
me al jacke s and can be es ima ed acco ding o Eq. (2) [11]
θθ
co 004.0co ⋅⋅=⋅⋅= hE
s
A
h
s
A
Vj
j
j
j
sj (2)
whe e h is he o e all pie sec ion dimension pa allel o he applied shea o ce, j is he
adop ed design jacke s ess le el co esponding o a jacke s ain o 0.004, Aj is he ans-
e se sec ion a ea o he jacke shee s spaced a dis ance s and inclined o he angle θ ela i e
o he membe axis. This condi ion is in oduced o a oid la ge dila ion s ains and hence ex-
cessi e deg ada ion o he conc e e, as well as o p o ide adequa e sa e y agains he possibil-
i y o jacke ailu e. The e o e, using Eq. (1) o bo h PO2 specimens e o i , wo s ip laye s
o CFRP shee we e adop ed wi h 0.117mm hickness by 100mm wid h and spaced a 100mm
along each pie heigh in o de o inc ease he shea capaci y. This e o i was doubled a he
pie base o imp o ing he conc e e con inemen and, he e o e, he o e all pie duc ili y.
Ped o Delgado, Pa ício Rocha, Miguel San os, João Ped osa, An ónio A êde, Nelson Vila Pouca, Anibal Cos a
and Raimundo Delgado
8
5 CYCLIC TEST OF THE RETROFITTED SPECIMENS
The e o i ed pie s ha e been es ed ollowing he same cyclic displacemen his o y o he
o iginal specimens, bu h ee addi ional cycles we e pe o med wi h inc eased op displace-
men ampli ude co esponding o 2.9%, 3.2% and 3.6% peak d i a ios. Resul s o bo h
specimens a e included in he ollowing sec ions.
5.1 Pie PO2-N2-R1
The damage e olu ion du ing he expe imen al es o he e o i ed specimen PO2-N2 is
illus a ed in Figu e 9, o he pie wes side. Fo small ampli ude cycles, co esponding o
d i below 0.7%, i s c acks abou 0.1mm wide occu ed (Figu e 9a). In he subsequen cy-
cles he c ack wid hs inc eased and new c acks we e also de eloped along he pie heigh . Fo
17mm op displacemen (1.2% d i ), he damage pa e n shown in Figu e 9b was mainly
cha ac e ized by diagonal c acks in he in e al zones be ween he CFRP s ips. Shea damage
con inued o inc ease in he nex cycles and, o 2.4% d i (Figu e 9c), c acking was con-
cen ed a he pie base wi hou isible damage on he CFRP shee s. Failu e o some o he
ibe s a he base was audible om his s age on. Du ing he las cycles o 33mm op dis-
placemen (2.4% d i ) he c acks on he conc e e and CFRP de eloped a li le. Figu e 10
shows he damage e olu ion on he pie in e nal aces also isualized du ing he es , in co e-
spondence wi h he ex e io damage, p esen ed in Figu e 9. When he CFRP s ips collapsed
a he pie base, o 2.4% d i , he in e io c acks inc eased conside ably, as shown in Figu e
10d. In he las cycles, gene alized damage became isible, wi h in e io collapse o he walls
and buckling o he longi udinal ein o cemen (Figu e 10d), ha caused a as educ ion o
he pie capaci y as e idenced in he esponse diag ams included in Figu e 12. The inal dam-
age s age is also shown in Fig. 11. As can be seen in Figu es 9 - 12, he e o i ed specimen
showed good beha io in compa ison wi h he o iginal one, exhibi ing well dis ibu ed c ack-
ing along he CFRP spacing. The shea e o i design, as used o his pie , showed excellen
pe o mance since he shea ailu e mechanism was p e en ed and he collapse was achie ed
a e he CFRP ailed a he pie base unde a lexu e mechanism ha occu ed u he a e
he o iginal specimen esponse. Figu e 12 shows he compa ison be ween he o iginal and e -
o i ed pie , whe e abou 40% inc ease o he op o ce is eached and abou 100% mo e o
he maximum displacemen is ob ained, wi hou signi ican educ ion o he esis an o ce.
(a) 0.7% d i (b) 1.2% d i (c) 2.4% d i (d) 3.6% d i
Figu e 9: Re o i ed pie PO2-N2. Damage om wes side iew du ing he es .
Ped o Delgado, Pa ício Rocha, Miguel San os, João Ped osa, An ónio A êde, Nelson Vila Pouca, Anibal Cos a
and Raimundo Delgado
9
(a) 0.7% d i (b) 1.2% d i (c) 2.4% d i (d) 3.6% d i
Figu e 10: Re o i ed pie PO2-N2 damage om in e nal eas side iew.
(a) Eas side (b) Sou h side (c) Wes side (d) In e nal and ex e nal
Figu e 11: Final damage iews o he e o i ed pie PO2-N2 co esponding o 3.6% d i a io.
-300
-200
-100
0
100
200
300
-60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60
Top Displacemen (mm)
Top Fo ce (kN)
Pie PO2-N2
Pie PO2-N2-R1
-300
-200
-100
0
100
200
300
-60-50-40-30-20-10 0 10 20 30 40 50 60
Top Displacemen (mm)
Top Fo ce (kN)
Pie PO2-N2-R1
Damage model
(a) PO2-N2 expe imen al cu es be o e and a e e o i . (b) Nume ical analysis o pie PO2-N2-R1
Figu e 12: Expe imen al and nume ical esul s compa ison.
Figu e 12b includes he compa ison o nume ical and expe imen al esul s o he es o
PO2-N2-R1, in e ms o op o ce-displacemen esponse, showing ha he nume ical simula-
ion model sligh ly o e es ima ed he s i ness and peak s eng h o he specimen. Nume ical
simula ions show also ha comp essi e damage o 1.8% d i (Figu es 13a and b) is g ea e
han o he o iginal pie because shea capaci y inc eased by CFRP shee s and a lexu e ype
de o ma ion mode was ac i a ed. F om he esul s illus a ed in Figu e 13e, i can be seen ha
a plas ic hinge was o med, since ed colo s o he longi udinal ba s a e concen a ed a he
pie base; his ac ag ees wi h he de o med mesh shown in Figu e 13c. Al hough ans e se
s eel has eached he yielding phase (Figu e 13d), he CFRP shee s, simula ed as high s eng h
s eel ba s and shown in Figu e 13 , a e esis ing well bellow he ul ima e s eng h (abou