Sunteți pe pagina 1din 8

PLASTIC-DAMAGE MACRO-MODEL FOR NON-LINEAR MASONRY STRUCTURES SUBJECTED TO CYCLIC OR DYNAMIC LOADS

Andrzej WAWRZYNEK Prof., DSc, PhD Silesian University of Technology, Poland andrzej.wawrzynek@polsl.pl Andrzej CICIO PhD Silesian University of Technology, Poland andrzej.cincio@polsl.pl

Summary
In the paper the plasticdamage material model for concrete (based on continuum damage mechanics and the theory of elasto-plasticity), proposed by J. Lubliner et. al. (1988) and modified by L. Fenves et. al. (1998), is adapted to masonry structures. A manner of determining model parameters determination for homogeneous masonry structures is explained. Some results of the numerical analysis of a prismatic specimen subjected to cyclic loading are presented, which show a good compatibility with laboratory tests (Singha et. al.). Additionally, a seismic simulation of the response (time history analysis) of a masonry structure to tremors is submitted. Keywords: dynamics of structures, masonry constitutive macro-model, non-local damage mechanics, theory of plasticity, time history analysis.

1. Introduction
The task of modelling masonry structures belongs to relatively complex numerical problems. The mechanical behaviour of masonry depends on the composite nature of the material, constituted by the fine dimensions, of natural or clay blocks connected by cement or cement-lime mortar joints. Both these mentioned materials differ from each other, among others: in the compression modulus, tensile (compressive, shear) strength. The emergency state of a masonry wall is manifested by crack expansions mainly near the joints. The complex mechanical behaviour of masonry structures requires so-called twomaterial micro-models with discontinuous discrete elements (modelling the blocks) and interface elements (modelling the mortar joints). However this type of these models (see eg. [5]) have a restricted range of application, limited to constructions which are not too large and whose geometry is rather simple. In numerical analyses, discontinuities of displacement fields are taken into account in FEM by interface elements. Also some new methods, based on the discrete element method (DEM) [6], is becoming more and more popular. In the case of modelling geometrically complicated structures significant computational complications leads to the formulation of a simplified description of the mechanical properties of masonry structures, using a macromodel. The macro-modelling approach makes no distinctions between blocks and joints, e.g. masonry material is treated as a fictitious homogeneous and continuous equivalent material, the mechanical parameters of which can be determined on the basis of laboratory tests. Additional difficulties of modelling masonry structures, especially subjected to cyclic and dynamic loads, are connected with taking into consideration a progressive degradation of brittle materials, as a result of the progressive lost of material integrity, caused by

crack/microcrack opening and closing. Also an increase of inelastic strains must be described by that model. It can be done by applying more and more popular models of non-local damage mechanics [7], in which the local damage of the material is expressed by some additional strain softening, as well as the reduction of mechanical parameters. The connections between the description of the material destruction and non-elastic softening are described by coupled damage-plasticity models, which are based on the theory of incremental elastoplasticity and continuum damage mechanics (e.g. for the concrete [8]). The paper deals with the task of adaptating the nowadays plasticity-damage model for concrete, called as Barcelona Model (BM), to masonry materials. The paper presents also a proposal of specifying of the model parameters basing on laboratory tests of masonry specimens [12-13]. Additionally, a seismic simulation (time-history analysis) of a twomaterial (concrete & masonry) spatial model of a building structure is submitted. The result of the analysis with applications of the plastic-damage material model is compared with results of the analysis where a linear-elastic model was used.

2. Basic concepts of the Barcelona Model


The plastic-damage model for concrete, proposed For the first time by a group of researchers viz. J. Lubliner, J. Oliver, S. Oller and E. Oate (1989) and modified by G. L. Fenves & J. Lee (1998) was used in numerical analysis of damages of concrete structures [9-10] as well as a numerical simulations of seismic tremor influences on a 2D plane model of a concrete dam [4]. The Barcelona Model is characterized by a bi-dissipative, isotropic degradation of material described by two separate damage variables dt i dc for tension and compression. These variables determined on the basis of independent material damage functions can be coupled, expressing the confirmed influence of the compression variable dc on the value of the tension degradation variable dt, after changing of the stress sign. The coupling of elasticplastic material characteristics and description its damage is realised by means of constitutive equations of the theory of plasticity by the effective stresses.

1
1

1 2

(2)

t0

t0

biaxial tension uniaxial tension

c0
3

t0

(4)
1 3

2 1 2

(3) (b0 ,b0)


biaxial compression uniaxial compression

c0

Fig. 1. Yield surface of BM for the plane stress space

From the point of view of the incremental theory of plasticity the BARCELONA Model is characterized by: Yield surface which is an extension of the classical Drucker-Prager model with a noncircular deviatoric cross section; Non-associative flow rule; Nonlinear isotropic hardening rules different and independent one for tension and the other for compression based on the hypothesis of strain hardening.

Figure 1 presents the shape of the yield surface of the BM in the plane stress space that consists of four functions (1): the 1st quarter of the co-ordinate system: one quarter of the circle with a radius equal to uniaxial tensile strength t 0 : the 2nd quarter:
f ( , ) 1 (q 3 p + 1 ) c ( ) = 0 1

(1) (1)

the 3rd quarter:


f ( , )

and the 4th one:


f ( , )

1 ( q 3 p ) c ( ) = 0 1
1 (q 3 p + 2 ) c ( ) = 0 1

where i are dimensionless parameters of the model eqs. (2) - which are expressed by the following yield stresses: for uniaxial compression c 0 = c (0) , for biaxial compression b 0 and for the tension (the same one for uni- and biaxial acting) t 0 = t (0) . The denotations p and q are stress tensor invariants. The overline (e.q. q ) denotes the effective value which depends on the actual level of material degradation.
= b0 c0 , 2 b 0 c 0 ( ) = c ( ) (1 ) (1 + ) t ( )

(2)

The evolution of the yield surface is described by two hardening variables


T % % = { c , t } which are expressed by equivalent plastic strains cp and tp , respectively for

compression and tension ( = {%cp , %tp } ).


T

a)

cu
c0

c
ultim ate stress plasticity lim it

1D
stress hardening

b)

t0
stress softening

t
Failure stress curve E0

1D

tension stiffening

% d c , d c d c ( cp )
(1- d c ) E 0
(1- d t) E0

E0

(1 - d c ) E 0

% d t, d t d t ( tp )
(1- d t) E0

%cp
%cin

%cd

%ce

%c

% % tp td

% te

% tin

% t

Fig.2. Description of the cyclic uniaxial test of BM for a) compression, b) tension

Material degradation (and changes of elastic stiffness) - that starts when the stress path reaches the yield surface - is described by two non-decreasing functions % % d c = d c ( cp ) , d t = d t ( tp ) which ideas are presented in Fig. 2.

3. Adaptation of BM for masonry


As mentioned above, masonry is an anisotropic composite material. Laboratory experiments show that the shape and dimensions of the biaxial envelope of the load capacity of masonry is strongly dependent on the angle between the stress principal directions and bed joints. So it is not possible to adapt concrete parameters of BM for masonry only by simple rescaling of its yield stresses. In particular, it is a great problem to describe the biaxial envelope for anisotropic masonry by the isotropic envelope (the yield surface) of the BARCELONA Model. A new proposition has been presented by A. Cicio in [1]. There it is assumed that in the biaxial area of compression (see Fig. 3a) the shape of the strength envelope is similar to a half of an elongated ellipse. Additionally, the diversification between uni- and biaxial strengths is greater in comparison to concrete. Simultaneously, on the basis of Eq. (2) the parameters of the model for masonry are evaluated (for assuming that cu = 0.5 fmx bu = 0.5 fmx - Fig. 3 where uniaxial compressive strength fmx is 8 MPa): =0.33 i = 9.84.
a)
1.0 0.75

2 / fmx
0.50 0.25 0

b)

2 / fmx
1.0 0.75 0.50 0.25 0

cu
2 1 1 2

0.25

0.25

2 = 22.5
0.50

1 2
0.75

0.75

bu

1 / fmx

1.0

1.0

1.25 A A

1.25

c)

2 / fmx
1.0 0.75 0.50 0.25 0

Biaxial compressive strength of masonry


Ganz, Thurlimann (1982), Page (1981), Lurati, Graf (1989)

0.25

= 45

1 / fmx

0.50
1

Soild clay brick masonry ( fmx= 8.0 MPa ) Description of strength envelope for BM proposal for cu = 0.50 fmx fmx

0.75

1.0

1.25 A

Fig. 3. Proposition of the biaxial strength envelope for masonry (dashed line) The envelope of the elastic states for biaxial compression of masonry (yield surface) was evaluated on the basis of experimental results obtained by Naraine & Singha [12, 13]. There were assumed the following values of uniaxial yield stresses has been assumed: for compression co = 0.67 cu = 0.34 fmx , and for tension tu =t0 = 0.03 fmx The hardening rule and degradation functions are determined for uniaxial material loading. Their specification can be found on the basis of the results of the cyclic compression

1 / fmx

0.50

of the masonry wall obtained by Naraine & Singla w [11] Fig. 4a (full line). Real hysteresis loops of loading should be changed by straight lines (dashed straight lines on Fig. 4a). It was assumed that the compressive yield stress is equal to co = 2.25 MPa for an elastic strain equal ce = 0.1 percent. The hardening function for compression was expressed according to an implementation of the BM in the finite element software ABAQUS by non-elastic strains cin = c ce (see fig. 2). The function of material degradation dc was determined on the basis of changes of Youngs modulus (different slope of the dashed line in Fig. 4a) during successive cycles of loading. The discrete values and their approximation are presented in Fig. 4b).
a)
-6.0

dc
0.5 0.4 0.38 0.22 0.35 0.28 0.40 0.43
dc

b)

-5.0

-4.0

0.3 0.2

stress [MPa]

-3.0

0.1 0.05 0.10 0.0 0.0 0.02 2.0 4.0

c in [00/0]
6.0 8.0

c0
-2.0

-1.0

0.0

e c = 1.0

2.0

3.0

Strain c [0/00]

4.0

5.0

6.0 7.0

8.0

Fig. 4. a) Experimental results used for the calibration of BM parameters [12], b) Degradation function for compression (dc) These parameters and functions of the BM are used during the numerical reconstruction of Naraines [12] experiment. Results of this process are presented in Fig. 5. and seem to be good the exactness of the reconstruction is very high.

[MPa]
5.0

Cyclic compression test (Naraine & Sinha 1989)


5.00 5.20 5.25 5.00

4.70 4.30 3.65 3.15 2.75 c 0=2.25

4.0 3.0 2.0 1.0 0.0 0.0 1.0 2.0

laboratory test numerical result (BM)

[0/00]
3.0 4.0 5.0 6.0 7.0 8.0 9.0

Fig. 5. Comparison between the laboratory [12] and the numerical results

4. Simulation of seismic tremor in building


The main aim of this simulation is: 1) to test the explicabilities of BM in non-elastic seismic or paraseismic analyses, 2) comparison between building responses in two cases elastic and non-elastic models of wall material. A 3D finite element shell model of two-storey building was analyzed with brick walls and floors and headers made of concrete (Fig. 6).
Front view

a)

g= 20 cm

10

b)

5.6

mur 3

0 cm

20 x 30 cm

40 x 50 cm

6.0 m

4.0

ax
front wall

Material code concrete masonry

Fig. 6. a) Geometry of the building model, b) The FE model The assumed values of yield points (c0 , t0), ultimate stresses (cu , tu) and initial modulus of elasticity E are as follows for concrete: c0 = 5.33MPa, cu = 16.0 MPa, t0 = tu = 1.60 MPa, E = 27.0 GPa for masonry: c0 = 1.99 MPa, cu = 3.0 MPa, t0 = tu =0.10 MPa, E = 2.21 GPa. The deadweight of the building is taken into account at the first step of the history analysis then the influence of the known seismic tremors is added as a kinematical forcing second step (step time from 0 to 3.73 sec) imposed horizontally - on the bottom edge of the model (see Fig. 6a where ax is a component of acceleration). The course of the real acceleration of the ground (recorded in the Polish copper mining district) which was used in simulation is presented in Fig. 7a). To receive a distinct plastic response of the building, this acceleration is multiplied by a coefficient equal to 2.5. The time history method is used to find numerical results - 752 integration steps are needed and a total number of iterations amounts to 776. The diagram of displacements of one point (No 10 on fig. 6a) from the upper edge of the building model, concerning plastic-damage and elastic models, are presented in Fig. 7b). Differences between these two curves result from taking into consideration the dissipation of the energy caused by material plasticity and its degradation. Additionally, in Fig. 8 the final pictures of degradation are described by two variables: dt for tension degradation and d - for global degradation.

5. Conclusions
The presented BM of masonry can be used for non-linear dynamic or cyclic analyses. There is a relatively good agreement between numerical simulations and laboratory experiments (see [1, 3, 9]). One of the main disadvantage of the BM is the assumption concerning the isotropy of masonry. Nowadays attempts are made ones work to modify the plane isotropic BM model for orthotropic material. Results which are received with the application of masonry BM in comparison with elastic solutions the following features: the field of displacement is characterised by a lower instantaneous value of the displacement amplitude;

there are no fictitious stress concentration which exceed the ultimate yield stresses for compression and tension (see Fig. 9); local damages of material are clearly described by degradation variables (Fig. 8).
a)
1.5 acceleration [m/s2] 1.0 0.5 0.0 -0.5 -1.0 0 0.5 1 1.5 2 2.5

b)
displacement [mm]

step time [s]


3 3.5 4

4.0 3.0 2.0 1.0 0.0 -1.0 -2.0 -3.0 -4.0 0.0 0.5 1.0 1.5 2.0

Barcelona Model Elastic model

step time [s] 2.5 3.0 3.5 4.0

Fig. 7. a) Recorded horizontal acceleration of the ground, b) comparison between building responses for elastic and plastic-damage models
tensile damage dt
0.45
Increase

scalar stiffness degradation d

0.11

0.00

0.00

a) layer n-

b)

layer n-

Fig. 8. Final damage image described by two variables d i dt


a) x [Pa] Elastic model b) x [Pa] Barcelona model

0.2 MPa

0.1 MPa

> t0 = 0.1 MPa

(non-realistic values)

Fig. 9. The horizontal tensile stresses result from elastic and plastic-damage models (for masonry part of model only)

6. Acknowledgements
The financial assistance of the Ministry of Scientific Research and Information Technology within the grant number 7 T07E 021 28 is gratefully acknowledged herewith. The numerical calculation were carried out in the Academic Computer Centre CYFRONET-AGH within the grant number KBN/SGI2800/Plska/023/2003.

7. References
CICIO A., Numeryczna analiza dynamicznej odpornoci niskiej zabudowy na wstrzsy parasejsmiczne z zastosowaniem przestrzennych modeli wybranych obiektw. PhD Thesis, Silesian Univ. of Technology., 2004 (in Polish). [2] CICIO A., WAWRZYNEK A.: Obliczenia dynamiczne budowli z zastosowaniem plastyczno-degradacyjnego modelu materiau, Proc. 3rd International Conference on New Trends in Static and Dynamic of Buildings, Bratislava 21-22.10.2004, pp. 293-296 (in Polish). [3] CICIO A., WAWRZYNEK A.: Zastosowanie plastyczno-degradacyjnego modelu materiau w obliczeniach dynamicznych budowli, Materiay 50-tej Konferencji Naukowej KILiW PAN i KN PZITB, Krynica 12-17 09.2004, t.II, pp. 29-36 (in Polish) [4] FENVES L., LEE J., A plastic-damage concrete model for earthquake analysis of dams. Earthquake Eng. and Structural Dynamics, vol. 27, 1998, pp. 937-956. [5] GIAMBANCO G., RIZZO S., SPALLINO R.: Numerical analysis of masonry structures via interface models, Computer Method in Applied Mechanics and Engineering (Elsevier), vol. 190, 2001 pp.6493-6511. [6] JING L.: Formulation of discontinuous deformation analysis (DDA) an implicit discrete element model for brick systems, Eng. Geology (Elsevier), vol. 49, 1998, pp. 371-381. [7] JIRSEK M.: Nonlocal models for damage and fracture: Comparison of approaches, Int. J. of, Solid and Structures, vol. 35, 1998, p. 4133-4145. [8] LEE J., FENVES G.L., Plastic-damage model for cyclic loading of concrete structures, Journal of Eng. Mechanics, vol 124, No.8, 1998, pp. 892-900. [9] LUBLINER J., OLIVER J., OLLER S., OATE E., A plastic-damage model for concrete. International Journal of Solids and Structures, vol. 25, 1989, pp. 299-329. [10] LUBLINER J., OLIVER J., OLLER S., OATE E., Finite element nonlinear analysis of concrete structures using a plastic-damage model. Eng. Fracture Mechanics, vol. 35, No.1/2/3, 1990, pp. 219-231. [11] NARAINE K. SINHA S.: Behavior of brick masonry under cyclic compressive loading, J. of Construction Eng. and Management (ASCE), vol. 115, No. 2, 1989, pp. 1432-1443. [12] NARAINE K. SINHA S.: Cyclic behaviour of masonry in biaxial compression, J. of Structural Eng. (ASCE), vol.117, No.5, 1991, pp.1336-1355. [13] NARAINE K. SINHA S.: Stress-strain curve for brick masonry in biaxial compression, J. of Structural Eng. (ASCE), vol.118, No.6, 1992, pp.1451-1461. [1]

S-ar putea să vă placă și