Upload
vokhue
View
219
Download
2
Embed Size (px)
Citation preview
Workshop 4 , June 9Workshop 4 , June 9
Two civil and industrial applications Two civil and industrial applications f 2D DIC tf 2D DIC tof 2D DIC measurements of 2D DIC measurements
combined with numerical simulationscombined with numerical simulations
Roberto FedeleRoberto Fedele and and coworkerscoworkers: : M. M. ScaioniScaioni, G. Rosati. M. , G. Rosati. M. FerrarisFerraris, V. Casalegno, V. Casalegno
D f Ci il d E i l E i i (DICA)Department of Civil and Environmental Engineering (DICA) Politecnico di Milano, Milan, Italy
Items to be discussed
Delamination tests on FRP-reinforced masonry pillar, optical monitoring by 2D DIC and FE predictionsoptical monitoring by 2D DIC and FE predictions
coworkerscoworkers: M. : M. ScaioniScaioni, G. Rosati , G. Rosati
Sh t t t l i bli
,,
Shear tests on metal-ceramic assemblies, identification of cohesive parameters for innovative joints
coworkerscoworkers: M. : M. FerrarisFerraris, V. Casalegno, V. Casalegno
Closing remarks and future prospectsClosing remarks and future prospects
D l i iD l i i i d FEi d FE d llid lliDelaminationDelamination experiments and FE experiments and FE modellingmodellingof FRPof FRP--reinforced masonry reinforced masonry
Roberto FedeleRoberto Fedele, M. , M. ScaioniScaioni, G. Rosati, G. Rosati
Ref: Fedele et alii, Cement & Concrete Composites, 45 (2014)
D t t f Ci il d E i t l E i i (DICA)
, p , ( )243–254.
Department of Civil and Environmental Engineering (DICA) Politecnico di Milano, Milan, Italy
CFRP-reinforced pillar: Historical bricks (XVII century) and high strength mortar( y) g g
single-lap shear test
Optical monitoring by 2D DIC
1 pixel footprint on the object 84 m metric object space
dcodco
1 pixel pitch on the sensor
image space
1 pixel pitch on the sensor = 6.1 m
Benchmarking
RMSE=15.9 m (0.19 Pz)
3D heterogeneous finite element modellingCFRP
masonry
Elastic-damageable model (Comi-Perego 2001)
“bi di i ti ” d l two isotropic damage variables:“bi-dissipative” model two isotropic damage variables:
in tension in compressiontD cDVumat (Abaqus explicit)
23(1 ) (1 ) 2ij t c v ij ijD D K G G
state equations
3( ) ( )ij t c v ij ij
l di l di diti(1 )- D
( , ) 0jt i tD f
loading-unloading conditions
0 0
j
t
t i t
ttD D
f
f
( , ) 0
0 0
jc i cD
D D
f
fdamage activation functions
0 0c tcD D f
3D FE modelling with perfect adhesionEffective elastic modulus of the CFRP reinforcement estimated by DICEffective elastic modulus of the CFRP reinforcement estimated by DIC
h dspoon shaped mechanisms
FOverall delamination response
sF
FE model with “ideal”constraints
“actual”
DIC corrected boundary conditions
“actual”
DIC-corrected boundary conditions
Local slip between FRP and masonry support
Characterization of innovative CFC/Cu joints b f ll fi ld t d fi it l tby full-field measurements and finite elements
Roberto Fedele*, Valentina Casalegno#, Monica Ferraris#
Ref: Fedele et alii, Materials Science & Engineering A, 595 (2014) 306–317.
*Dept. of Civil and Environmental # Dept of Applied Science d T h l (DISAT)Engineering (DICA)
Politecnico di Milano, Milan, Italyand Technology (DISAT)
Politecnico di Torino, Turin, Italy
Composites for aggressive environments
2760
oC
1371
540
ITER(International(International Thermonuclear Experimental pReactor)
brake discs
turbine engine blades
Single-lap shear tests on flat-tile joined samples
CFC / Cu
glue (Araldites AV 119, Ciba-Geigy)
Macroscopic responsei t i i CFC h t th 15 20 MP !
P
intrinsic CFC shear strength 15-20 MPa !
maxmax 23.6
joint
PA
[MPa]
uncertain boundary conditions due to the glue layersto the glue layers and compliant grips!
max 34 4 [MPa]comparative assessment by different tests
4-dofs motion for zoomed camera for optical monitoring
Z
X
X
Y
Finite element discretization of ROI1 pixel 1 pixel 3.0 m
pixe
lp
pixel
Joint collapse
pixe
lp
pixel
L l hLocal approach: FE simulations driven by boundary displacements
Cuboundary displacements
plane stress schematization
CFC
plane stress schematization
(anisotropic) elastic-plastic behavior
finite thickness joint
behavior
j
Displacement fields measured by DIC and computed
tangential component
Displacement fields measured by DIC and computed
tangential component
Displacement fields measured by DIC and computed
tangential component
Joint governing parameters to be identifiedVan der Bosch Schreursinterface “di l t j ”
2
exp expn n n tp
Van der Bosch, Schreurs and Geers, EFM, 2006
interface tractions
“displacement jumps”
2
2
exp exp
2 1 exp exp
n n n tn
n n n t
t t n n t
p
p
tr
21 exp expt t n n t
tt t n n t
p
TX t t id tif
nassumed a priori
, ,n n tX
tn
parameters to identifyminimization by Trust Region, reflective, i i i h d
u1
1/2
ˆ arg min ( )t
Tk k
k
k k k
XX X R R interior point Method
in a Matlab environment
1/2 exp comp exp( )k k kk k
R W U U X U , 1,..., tk k n
boundary conditions provided by DIC were deterministically prescribed without any regularization provision
Tangential stress evolution
maxMPa43tp traction predicted under pure mode II
Normal stress evolution
maxMPa58np traction predicted under pure mode I
Closing remarks and future prospects
DIC measurements especially suitable forcalibration/validation of FE modelswith special reference to :(i) accuracy of 2D/3D geometry assumptions ;(ii) boundary data estimation ;(iii) response of joint/interfaces ; (i ) i i l i hi(iv) constitutive relationships .
Information fusionfrom several sensorsand diverse testing configurations
Extension to High and Ultra High Temperature (UHT) testing (?)testing (?)
322 ij jiJ S S [MPa2]
1 tr ij jiI σ [MPa] )( ii Dh cti ,
2 2b h k hfmeridian plane
first invariant stress tensor second invariant stress deviator hardening/softening functions
2 22 1 1, 0c c c c c c cD J a I b h I k h σf
hardening/softeninghardening/softening
2 22 1 1, 0t t t t t tt D J a I b h I k h σf
IHaigh-Westergaad space
, 0t tD σf
II
III
= , ,i i I II IIIprincipal stresses
, 0c cD σf
ifi f
Fracture energy regularization
(m ) iGg
specific fracture energy
(e)ch
(m )i igl
“element” characteristic length
Local predictions of FE model
2 LVDT [ ] 0 1
2 mm
1.6
2 clip [mm]LVDT [mm]
0.08
0.1
1.2 0.06
0.8 0.04
0.4 0.02
0 1 2 3 40
di ti ti0 1 2 3 4
0
ordinative timeordinative time
Closing remarksClosing remarks
•• SingleSingle--lap shear tests were performed under cliplap shear tests were performed under clip--controlcontrol
•• DelaminationDelamination of CFRP strips from a small masonry pillar was of CFRP strips from a small masonry pillar was simulated under the hypothesis of a perfect adhesion.simulated under the hypothesis of a perfect adhesion.
•• 3D heterogeneous FE model with elastic3D heterogeneous FE model with elastic--damageable phases damageable phases was developedwas developedpp
•• Optical monitoring was validated with correction Optical monitoring was validated with correction f ti lf ti l di t idi t iof optical of optical distorsiondistorsion
•• Information fusionInformation fusionInformation fusionInformation fusion
•• Future prospects: combining interface and bulk damageFuture prospects: combining interface and bulk damage
Local traction predicted by FEMy1 2
3
y
3
x
1 7 MPamaxF .A
41
FEM
12 delamination front
A
B
42
12C
delamination frontfront
D
43
A priori assumed parameters (“diffuse load cell”)
Ceramic phase CFC SEP NB31: Mechanical parameters at room temperature
Hill parameters Ramberg-Osgood Elastic properties
pfor plastic anisotropy
g gparameters for the
incompressible strains
107 [GPa]xE ; 15[GPa]yE ; 12 [GPa]zE ; 0 8F ; 0 5G ; 100 [MP ][ ]x [ ]y [ ]z
0.10xy ; 0.20xz ; 0.20yz ; 10 [GPa]xyG
0.8F ; 0.5G ;
0.5H ; 10N ; 0 100 [MPa] ;
/ 2083 [GPa]RE ; 7n ;
Cu phase: Mechanical parameters at room temperature
Elastic properties Ramberg-Osgood parameters
125[GP ]E ; 300 [MPa] ;125[GPa]E ;0.34 ;
0 300 [MPa] ;0.06 ; 7n ;
Ref: ITER Final Report, Material Assessment Report, 2001
Post-mortem analyses of fracture surface by SEMFailure of CFC by interlaminar shear, carbon fibre pull-out and cracking of CCr
resolutionSEM resolutionSEM pictures
ITER Final Design Report (July 2001): Materials
Flat-tile mock-ups manufacturing
armour / plasma facing component
li h lcooling channel of heat sink
1) High heat flux applied on CFC surface
CFC NB31/Cu/CuCrZr
1) High heat flux applied on CFC surface up to 10 MW/m2 (3000 cycles)up to 20 MW/m2 during transient events (20 cycles)2) Neutron irradiation and radiation damage
Problems when joining metals and CFC
Large thermal expansion mismatchbetween Cu and CFC
20Thermal expansion at 300 °C
CFC=1,7-3,3 x 10-6 K-1, Cu=16,6 x 10-6 K-1
Cu
10
15
20
16 K10 Cu 6,6 0 high residual stresses CFC
0
5
10K10
Low wettability of molten copper on CFC (contact angle= 140°)
θ=139-145°Cu
sessile drop test θ 139-145
CFC
pat 1100 oC for 30 min under Argon CFCunder Argon
Joint manufacturing by one-step brazingF i t lii J N l M t1. Composite surface is modified by direct
reaction with chromium a carbide layer is formed: large reduction of
Ferraris et alii, J. Nucl Mat. 2008
a carbide layer is formed: large reduction of the C/C-Cu contact angle.
2 C i l b i ll (G ® ) i i ht2. Commercial brazing alloy (Gemco® ) is used to braze C/C to pure copper and pure copper to CuCrZr by the same heat
weight
pp ytreatment. Alloy does not contain any activating element ( h Ti d Si)
CFCC/C
(such as Ti and Si)
brazing alloy
pure Cu
C Cbrazing alloy CuCrZrbrazing alloy
2D- Digital Image Correlation“ i ” d ti f th l l t t“passive” advection of the local texture (optical flow conservation)
displacement vector
( ) ( [ ])gf x x xulocal form
displacement vector
reference imagedeformed image
weak form
deformed image
2arg min ( ) ( ) ( ) df g x uxu x
weak formsimilarity measure
g ( ) ( ) ( )f gu
variational problem
1 1i i i u = u u incremental form
Truncated expansion and stationarityEuler Lagrane d 0 LEuler-Lagrane stationarity condition
2 2 1 2grad , 0i i v v L
T 1 12 i
Ti
Tg g d v u x1 1( , )i ia u v
bilinear form
Tg g K
1( )iF v 12 ( ) ( )iT
ig g f d
xuxv xlinear form ( ) L L LL
2 2find : ( , ) ( )i i i i ia F u L u v v v L
linear form2 2 2 2( ) L L L L
semi-coercive variational problemmultiplicity of solution 0 Keru K
Galerkin finite-element discretization
u=NU(e)
1i K U Bpseudo-stiffness pseudo-load
Tangential traction and 95% confidence strip
Normal traction and 95% confidence strip
Parameter sensitivity of displacement field
Estimate dependence on a priori informationvariance assessment by -point strategy
3 99% confidence intervals
Confidence ellipsoids and Bonferroni’s domains0 106 [GP ] 1 2T0.106 [GPa m]n
0.665 [ m]n 10.7 [ m]t X
1 2X 1( ) ( )T
,n- - X X C X X
C X CX z X z X X1C X C
ii iii i iX z X z
1 2 3i , , X(2 )/ n equivalent parallelipedic domains
Engineering motivations
The behavior of masonry structures, strengthened with fiber-i f d l (FRP) hi h i f d i d breinforced polymer (FRP) thin sheets, is often dominated by
delamination of the FRP reinforcement from the support.
A further complication is the presence of mortar joints, where cracks may propagate preferentially.
This fundamental issue is relatively under-investigated for masonry especially from a numerical point of viewfor masonry, especially from a numerical point of view.
Items to be discussed
engineering motivations
shear tests on joined CFC/Cu assemblies
“optical” inverse problem: from pictures to displacementsfrom pictures to displacements through 2D Digital Image Correlation
“mechanical” inverse problem: from full field data to joint properties through Finite Element Model Updating
closing remarks and future prospects
Ceramic Matrix Composite SEP NB31NOVOLTEX
PAN=Poly-Acrylo-Nitrilex
NOVOLTEX
preform
yy
SEM (Philips 525 M)
Energy Dispersive X-ray spectroscopy
Cromium carbide (about 20 m thick) Cr7C3 and Cr23C6
Adherent nonlinear behavior under plane stressanisotropic extension of multiaxial Ramberg-Osgood relationship
anisotropic extension of multiaxial Ramberg-Osgood relationship
deviatoris stress tensor3 parameters
Műcke & Bernhardi CMAME (2003)
1
1
eqel pl
0R
n
nE
ε ε ε σ sC M 1/2
eqσ T s sM
(tr 3) σs σ 1
equivalent stress0R
([1 : 3],[1 : 3]) 0 M
1/ / / 0x yx y zx zE E E
M
G+H -H -G 0-H H +F -F 0
/ 1/ / 0/ / 1/ 0
0 0 0 1/
xy x y zy z
xz x yz y z
E E EE E E
G
C
M-G -F F+G 00 0 0 2N0 0 0 1/ xyG
/ /ij i ji jE E , , ,i j x y z incompressible strain pltr ( ) 0ε3 parameters
7 parameters 312 2; F = G = H Nfor isotropic Cu
First-order sensitivity analysis by Direct Differentiation Method
i ( ; ( ); )nn X X UF U 0
FE equilibrium eqs at the free dofs
unknown parameters
Here dependence only upon U!nUint ( ; ( ); ) X X UF U 0
at step nprescribed displacements along the boundary
only upon U!U
along the boundary
int intint T
n
T n T
X U
FFX X
F U 0int TT n T X U X Xpseudo-load vector
intn
FK Up
tan T T XK
Xbl d iff iassembled tangent stiffness matrix,
already available in a Newton scheme