1 views

Uploaded by Chien Chieu

© All Rights Reserved

- Smex1045 Strength of Materials 2 1 0 3[1]-1
- BOLTGRP
- CE6306 Strength of Materials(SOM)
- shells
- A Layered Limit Analysis of Pinned-joints Composite Laminates Numerical Versus Experimental Findings
- Intersections V06 No04 04
- Calculation Sheet
- Beams Membranes Plates and Shells BMPS Statements of Competence
- Truss problem 6.1 Johnston and Beer book-Static 1-1.docx
- JL-94-January-February Horizontal Shear Strength of Composite Concrete Beams With a Rough Interface
- Implant Dentar Personalizat BioMicron Transilvania-Study 1-1
- Importance of shear assessment of concrete structures detailed to different capacity design requirements
- 11 IJAERS
- Theories of Stress and Strain
- 9A01301 Mechanics of Solids
- Form Finding of Tensile Membranes
- Discussion
- CE IES 2012 Conventional Paper I
- 11-82-13 (vol.66 no1).pdf
- art02

You are on page 1of 30

of elastic infilled frames under monotonic loading.

I.N. Doudoumis

Aristotle University of Thessaloniki, Greece

e-mail: doud@civil.auth.gr

DOI:10.1016/j.engstruct.2006.07.011

ABSTRACT:

In this paper the behaviour of single-storey one-bay infilled frames under monotonic lateral loading is studied

and investigated analytically, assuming linearly elastic material laws. To this purpose a precise finite-element

micromodel is formulated and used, by considering contact interface conditions between the frame and the in-

fill. The analytical investigation is carried out through an extended comparative parametric study and is fo-

cused on the quantitative influence of certain simplified modelling assumptions and several critical modelling

parameters on the response of the system. In the analytical investigation, the effects of the alternative model-

ling of the interface conditions, the density of the discretization mesh, the size of the friction coefficient, the

relative beam/column stiffness, the relative frame/infill size and the orthotropy of the infill panel are evalu-

ated, providing a detailed insight and a better quantitative understanding of the basic features of the system

response. It is concluded that all these modelling parameters considerably affect the behaviour of the exam-

ined infilled frames, except for the parameter describing the infill’s orthotropy. The conclusions obtained can

be further extended in the case of multi-storey systems.

KEYWORDS: Infilled frames, finite element micromodels, analytical modelling, contact interface conditions

1 INTRODUCTION

It has been commonly recognized that the existence of infill panels within the interior of the building frames

considerably improves their lateral stiffness, strength and energy dissipation capacity against earthquakes. In

the modern practice the infill panels are fabricated after the construction of the building frames and they are

not firmly connected with the surrounding frames. This frame-infill connection is usually attained by simple

contact, or by the presence of some mortar with low and unreliable tensile strength and adhesion capacity,

which obviously could be ignored in the analytical modelling of the system. Consequently, only compressive

normal and limited frictional contact stresses can be considered at this interface, while separation or slipping

along any parts of it may occur.

Due to their construction method, the infill panels remain almost inactive when subjected to gravity loads,

while they are activated when subjected to lateral loading. In the last case, the actual contact area between

frame and infill is strongly localized at the compressive corners of the infill, but the lengths of contact are not

a priori known. As a consequence, the response of an infilled frame to wind and seismic loading is non-linear,

even in the case of linearly elastic material laws. Furthermore, additional non-linearities may be present due

to inelastic material behaviour.

Concerning the analytical models developed for the infilled frames, two different approaches have been

2

followed so far: a) the micromodels, based on a finite element representation of each infill panel with a fine

mesh of finite elements, and b) the macromodels, based on a physical understanding of the behaviour of each

infill panel as a whole, usually represented by a single structural member (like the “equivalelent” diagonal

strut [1], etc.). The micromodels are the most accurate approach, because they can take into consideration the

contact interface conditions, various material laws for the frame and infill (elastic, elastoplastic, fracturing,

etc.) and several other modelling parameters, while their results tend to the "exact" response of the physical

structure as the discretization mesh becomes finer.

One of the first applications of micromodels for the analysis of infilled frames was presented in 1968 by

Mallick & Severn [2], for the calculation of the elastic stiffness of several one-bay single-storey infilled

frames. The model included 16 rectangular plane-stress and 12 axially rigid beam elements, while for the cal-

culation of the active contact area and contact forces at the frame-infill interface, the effects of the frictional

forces on the beam elements were neglected and a trial-and-error scheme was used. Riddington and Stafford

Smith [3] carried out an elastic analysis of a micromodel with plane-stress elements for the frame and infill

and concluded that the bending moments of the frame were considerably reduced when an infill existed. A

similar model, with the addition of friction elements at the interface, was presented by King and Pandley [4].

In the above cases iterative procedures were adopted for allowing separation at the interface, but the slipping

possibility was not consistently taken into account.

Elastic analysis of refined micromodels with openings in the infill panel was presented by Achyutha et al

[5], together with an iterative procedure allowing for separation and slipping at the interface. Comparing the

elastic [6] and inelastic behaviour [7] of a proposed macromodel with the results of corresponding precise

micromodels, Doudoumis et al ([6], [7]) proposed and used micromodels consisting of 36 plane-stress 4-node

elements for the infill, together with frictional gap elements for the interface conditions.

Liauw and Kwan ([8], [9]) studied analytically and experimentally the non-linear behaviour of a pair of

four-storey single-span infilled frames both with shear connectors at their interface and without shear connec-

tors (allowing slipping and separation). In their analytical micromodels they used inelastic material laws for

the frame and infill, a fine mesh of triangular plane-stress elements for the infill panel and an iterative solu-

tion procedure with incremental imposed displacements.

May & Naji [10] proposed a finite element model to simulate the nonlinear behaviour of steel frames in-

filled with concrete panels under monotonic or cyclic loading. The steel frame was modelled as plastic mate-

rial with strain hardening, the infill panel as concrete model with cracking, yielding and crushing, while sepa-

ration and sliding at the frame-infill interface were properly accounted for.

El Haddad [11] suggested a micromodel for analyzing infilled frames allowing cracking of the frame. The

frame elements were divided into standard beam and cracked beam elements near the corners, while elastic 4-

node plane stress elements were used for the infill panel. The active contact lengths at the frame-infill inter-

face were considered as a priori known and the relative slipping at this part of the interface has been con-

strained.

Singh et al [12] presented a simplified micromodel with only 4 plane-stress (8-node) elements for the in-

fill and inelastic material laws for the RC frame and the masonry infill. The interface conditions have been

modeled approximately to permit sliding and separation, while a predictor-corrector scheme for the solution

of the non-linear problem was used.

By comparing existing experimental results with analytical studies on infilled frames, and in order to

3

simulate their failure mechanism, Mehrabi & Shing [13] tested 5 masonry-infilled concrete frames by using a

smeared-crack plasticity micromodel with 280 elements for the infill panel, together with a nonlinear consti-

tutive model for the mortar joints. Similar smeared-crack plasticity micromodels with fine mesh of plane-

stress elements for the infill panel and the surrounding frame have been used by Ghosh & Amde [14], to-

gether with a non-associated interface model (of Coulomb type) for the masonry and interface joints. Also

these authors verified analytically previous findings of experimental researchers [15], according to which the

cracking of the mortar joints in the infill does not considerably influence the stiffness and the strength of the

infilled frame. Consequently, the effects of modelling the mortar at the frame-infill interface with separate

elements are negligible and contact boundary conditions with Coulomb friction could be used at this interface

instead. In addition, the masonry infills could be modelled as homogeneous materials [14].

From all the above mentioned literature it is evident that several precise micromodels have been suggested

so far, which are sufficiently efficient to carry out a reliable elastic or inelastic analysis. However, none of

them has been systematically used for a detailed investigation of the behaviour of the infilled frames, in de-

pendence with the large number of parameters that influence this behaviour. It must be noted that the number

of these parameters is considerably decreased in the case of an elastic behaviour of the materials. In practice,

a perfectly elastic behaviour in all the extent of an infilled frame (with unchanged values of the elastic mate-

rial constants), takes place only if the developing stress state has never exceeded the strength capacity of the

materials at any point, something that seldom happens during the loading history of the real structures. Never-

theless, the importance of the elastic analysis should not be downgraded, because it gives the first reliable pat-

tern of the deformation and stress distribution within the examined structure and constitutes the first neces-

sary step for an effective inelastic analysis. Therefore, even for these reasons only, the elastic analysis contin-

ues to be widely acceptable in principle by the modern Codes.

The main object of this paper is to investigate quantitatively the influence on the results of the response of

the infilled frames, of certain simplified modelling assumptions and several critical modelling parameters, by

using refined finite element micromodels. Such a systematic investigation provides a detailed insight on the

basic features of the elastic behaviour and a better understanding of the system’s response, giving rise to a va-

riety of original results.

The quasistatic two-dimensional contact problem at the interface between an infill panel and the surrounding

frame is considered, under the assumptions of small strains and displacements and linearly elastic material

laws. At this interface, unilateral contact conditions (of Signorini's type) together with Coulomb's law of dry

friction are assumed to hold (Fig. 1). The frame and infill are properly discretized by the finite element

method in order for the contact area to consist of pairs of nodal points. It is further assumed that, for zero ex-

ternal loading of the system, no initial gaps or initial contact stresses exist along the frame-infill interface,

while certain springs with infinitesimal stiffness exist in order to prevent a rigid body motion of the infill.

4

Figure 1: (a) Signorini's law of unilateral contact, (b) Coulomb's law of dry friction.

We denote by sNi and sTi the normal and tangential contact reactions at any pair i of contact points (tension

positive), by uNi and uTi the corresponding relative displacements (penetration positive) and by fTi the slipping

function fTi = |sTi|+μ∙sNi, where μi is the friction coefficient. In the general case, the contact state of a pair of

contact points may be one of the following types:

• sticking contact

• slipping contact

• separation

The present study limited to the case that the initial conditions at any contact pair i correspond to zero initial

gaps, together with zero contact stresses due to zero external loading. These initial conditions are written as:

and the following incremental contact states may take place ([16], [17]):

a) sticking contact:

δfTi = |δsTi|+μi∙δsNi < 0, δsNi < 0, δuTi = 0, δuNi = 0 (2a)

b) slipping contact:

δfTi = |δsTi|+μi∙δsNi = 0, δsNi< 0, δuNi = 0 (2b)

c) separation:

δsTi = 0, δsNi = 0, δuNi ≤ 0 (2c)

T

−δfi ≥ 0, δλi ≥ 0, (−δfi) ∙δλi = 0, (3)

where

Ni = 1 μi ,

δfi = δ fTi+ , δλ i = | δ uTi+ |

0 1 δ s | δ u |

Ni Ni

−1 1 0 δ s δu

Vi = , δsi = Ti , δui = Ti (4)

0 0 1 δ s Ni δ uNi

Also, for the total number m of contact pairs, relations (3) can be written as

5

δf = N∙δs δu = −V∙δλ

T

−δf ≥ 0, δλ ≥ 0, (−δf) ∙δλ = 0, (5)

where

δλ = [δλ1, δλ2,.. ., δλm]T

N = diag[N1, N2,.. ., Nm]

V = diag[V1, V2,.. ., Vm] i=0,... m (6)

Fictitious 2-node contact-link (frictional gap) elements of infinitesimal size, which represent the contact

interface conditions, can be introduced at the pairs of contact points. These contact-link elements are consid-

ered as absolutely rigid and they are able to undergo imposed relative displacements between their ends. For

each contact-link element i, we denote by si=[sTi, sNi]T the end reactions and by ui=[uTi, uNi]T the corre-

sponding relative displacements of their ends due to imposed internal strains. The introduction of contact-link

elements defines a new structural system with fixed-support interface conditions, which is absolutely equiva-

lent [17] to the initial one (with unilateral boundary conditions). In this ″equivalent structural system″ all the

contact-link elements behave as bilateral ones, subjected to unknown imposed relative displacements

δui=−Vi⋅δλi between their ends, while at the same time 3∙m complementarity conditions are fulfilled, which

are equal in number with the non-negative unknowns δλi (see relations 3 and 4). Therefore the further study

may be performed exclusively in the equivalent structural system.

It is also known ([16], [17], [18]) that the work-corresponding pair of vectors δs and δu of any stable

and linearly elastic structure, are related by the equations

which can be obviously applied to the respective vectors of the contact-link elements as well. Equations (7)

mean that the increments of the reactions δs can be expressed by the superposition of the part K⋅δu due to

the increments δu of the imposed relative displacements (yet unknown), and the known part δso due to the

increments of the external actions. The square matrix K is the influence matrix (of stiffness type) of the im-

posed relative displacements of the contact-link elements on their corresponding mutual reactions. Substitut-

ing in relations (5) the value of δs obtained by the relations (7), the following incremental Linear Comple-

mentarity Problem (L.C.P) arises:

Find δf and δλ such that

T

−δf = (N∙Κ

Κ∙V)∙δλ−N∙δso) =M∙δλ+δq −δf ≥ 0 δλ ≥ 0 (−δf) ∙δλ = 0 (8)

Κ∙V is a given square matrix which depends only on the initial conditions (i.e. zero initial gaps

where M = N∙Κ

and internal stresses), and δq = −N∙δso is a given load vector which depends linearly on the external loading

and on the initial conditions. Since the infilled frame has been supposed to be a stable structure, the matrix K

is positive definite and consequently the L.C.P described by relations (8) has a unique solution, provided that

the size of the friction coefficient μ does not exceed certain limiting values [19]. The solution of this problem

defines the incremental contact state of the contact-link elements, i.e. which pairs of contact points are in

sticking contact, or slipping contact or have been separated, due to the incremental values of the external ac-

tions [17]. Furthermore, the stiffness matrix that corresponds to the above defined contact state of the contact-

link elements, is the tangential stiffness matrix KT of the structure.

6

λ,P) is the unique solution of the L.C.P. due to the external actions δP,

Let us consider now that (δf,P δλ

which produce the stress vector δso,p at the contact-link elements of the ″equivalent″ structural system. We

shall prove that increasing proportionally the external actions δP by a factor ρ>0 (i.e. for external actions

λ,P).

ρ∙δP), the solution of the corresponding L.C.P. is (ρ∙δf,P, ρ∙δλ

λ,P) is the unique solution of the L.C.P. due to the external actions δP, the relations

Proof: Since (δf,P δλ

(8) are written as

−δf,P = (N∙Κ

Κ∙V)∙δλ,P −N∙δso,P = M∙δλ,P + δq,P (9a)

T

−δf,P ≥ 0 δλ,P ≥ 0 (−δf,P) ∙δλ,P = 0 (9b)

− ρ∙δf,P = ρ∙(N∙Κ

Κ∙V)∙δλ,P − ρ∙N∙δso,P = ρ∙M∙δλ,P + ρ∙δq,P (10)

and since δf,p, δλ,p and δso,p are refferred to the "equivalent" structural system, the principle of superposition

holds, which leads to

− δf,ρP = (N∙Κ

Κ∙V)∙δλ,ρP − N∙δso,ρP = M∙δλ,ρP + δq,ρP (12a)

Also, multiplying the first two relations (9b) by ρ and the third by ρ2 it follows that

T

−δf,ρP ≥ 0 δλ,ρP ≥ 0 (−δf,ρP) ∙δλ,ρP = 0 (12b)

Therefore, since relations (12a) and (12b) hold, the values (δf,ρP δλ,ρP) constitute the unique solution of the

L.C.P. (8) corresponding to the external action ρ⋅P (∀ρ >0).

The numerical solution of the above incremental L.C.P. may be achieved by using several solution algo-

rithms of direct ([20], etc.) or iterative type [21]. Alternatively, if the numerical solution of this problem can

be achieved by using other iterative solution strategies (Newton-Raphson etc.), this is the unique solution to

the problem.

From the aforementioned analysis, which is valid only for the special case of zero initial gaps and zero ini-

tial contact stresses at the interface, the following conclusions can easily be drawn:

a. For a given external action P, the solution of the corresponding L.C.P. (8) defines which pairs of contact

points are separated by losing their initial contact (λi =0), or still remain in contact (sticking or slipping)

by developing compressive normal reactions (sNi ≤0).

b. For a proportional variation ρ⋅P (ρ>0) of the external action, the pattern of nodal pairs loosing their ini-

tial contact and the pattern of those pairs that are still in contact (sticking or slipping) remain unchanged.

In particular, the relative displacements of the separated or slipping nodal pairs, as well as the reactions of

the nodal pairs still in contact, vary proportionally (f,ρP =ρ⋅f,P λ,ρP =ρ⋅λ,P).

3. METHOD OF ANALYSIS

The above conclusions can be directly utilized in the investigation of the elastic behaviour of infilled frames

which follows. Although the contact problems are highly nonlinear and dependent on the loading history in

7

general, the assumption of elastic material laws and zero initial interface conditions (zero gaps and contact

stresses) for zero external loading, makes the system's response independent of the existence of any previous

loading history.

The basic analytical study was carried out on a typical single-storey one-bay infilled frame (Figure 2a),

assuming linear elastic behaviour of the materials. This system was subjected to proportionally increasing

quasi-static horizontal loads P acting at the ends of the top beam, thus providing results which are propor-

tional to the loading magnitude.

The knowledge of how a single-storey one-bay frame-infill interaction system behaves is a matter of

prime importance for the understanding of the static behaviour of multi-storey and multi-column systems.

This is because the single-storey one-bay system constitutes the smallest possible typical substructure in

which a more complex structure can be analyzed, while at the same time it includes all the basic characteris-

tics of the interaction between the infill panel and the surrounding frame.

For the infilled frame under consideration (Fig. 2a), the micromodel shown in Figure 2b was formed and in-

vestigated. In particular, each column was modelled with 12 frame (beam-column) elements located along the

respective column axis, while rigid zones at the ends of the columns were considered. The section properties

of the columns (moment of inertia J1 and axial area F) correspond to a square section 40×40 cm and remain

constant during this study.

Each beam was modelled with 12 frame elements located along the respective beam axis, also considering

rigid zones at the ends of the beams. The cross-section area F was assumed to be large, in order to take into

consideration the existence of slab-diaphragms at the levels of the beams. The moment of inertia J2 of the

beams can take several values during the analysis.

The infill panel was modelled with 4-node plane-stress elements with 2 D.O.F. per node and a mesh of

12×12=144 elements. The finite section height of the beams and columns was taken into consideration by in-

troducing 52 rigid zones connecting the nodes of the frame axis with the respective nodes on the internal face

of the frame. Also 52 unilateral contact-link elements (as mentioned above) were introduced along the inter-

face contact area.

Finally the stability of the system was ensured (in case that a separation at all the contact pairs of the inter-

face would take place), by connecting the frame and infill with additional springs of high flexibility. This mi-

cromodel was assumed to represent with sufficient accuracy the "exact" response of the infilled frame, since

the displacements at the corner nodes of an alternative model, which was obtained by a twice finer mesh of

24×24=576 plane-stress elements, were almost identical with those of the examined micromodel (difference

less than 0,5%).

8

P 1 J2, F2=∝, E 2 P P 1 2 P

J1 J1

h F1 Ei, t F1 h' h h'

E E uy

w ux

4 J2, F2=∝, E 3 4 3

L'

(a) (b)L'

Figure 2: a) The examined infilled frame, b) the proposed micromodel with 144 plane-stress elements

In the above micromodel, the effects of certain critical modelling parameters on the system's response were

investigated through a detailed analytical parametric study. For every different value of the parameters under

consideration, a separate structural model was analyzed. During this procedure, several hundreds models

were examined and characteristic results of their elastic response were compared and evaluated. In particular,

the influence of the variation of the following parameters was considered:

a) The ratio L/h of the width to the height of the infilled frame, which is equal to the ratio L'/h' of the infill

panel (Figure 2). The height h of the frame takes the constant value h=3,0 m while the length L of the

frame can take the values L=2,25 m (L/h=0,75), L=3,0 m (L/h=1,0), L=4,50 m (L/h=1,50) and L=6,00 m

(L/h=2,0). The examined values of L/h cover a considerable percentage of the infill panels used in practice

in building structures.

b) The ratio J2/J1 of the inertia moments of the beams to the columns, which takes values from 0,25 (flexible

girders) to 8,0 (stiff girders).

c) The relative lateral stiffness of the infill, which, for constant values of J2/J1 and L/h, can be expressed by

the ratio (Ei∙t)/Kc, where Kc is the lateral stiffness of the respective fixed-end column and Ei and t are the

elastic modulus and the thickness of the infill respectively.

d) Τhe alternative modelling of the interface conditions, by taking into consideration 4 different types of

them.

e) The density of the discretization mesh, by examining several coarser mesh densities of the infill panel.

f) The size of the friction coefficient μ, which was considered to take values between 0 and 1.

g) The orthotropy of the infill panel, which can be described by the ratio Ey/Ex of the elastic modulus along

the vertical and horizontal direction respectively.

h) The relative size of frame and infill, which can be described by the ratio r'=(h-h')/h', where h and h' are

the height of the frame and infill respectively.

During the evaluation and comparison of the results, aiming at a larger generality of the conclusions, non-

dimensional variables are used when possible. Such a non-dimensional parameter is the relative stiffness of

9

the infill to the frame, which has been already used from the earlier studies on this subject ([1], etc.).

In the present study, since the section and the height of the columns remain constant during the parametric

analysis, the column stiffness parameter Kc=12EJ1/(h')3 (which equals to the lateral flexural stiffness of the

respective fixed-end column) is further used. The lateral stiffness of the bare frame is proportional to Kc and

also depends on the ratio J2/J1 and on the ratio L/h. The lateral stiffness of the infill panel is proportional to

the elastic modulus Ei, to the thickness t of the infill and also depends on the ratio L'/h'. Thus, for constant

values of J2/J1 and L/h, the relative lateral stiffness of the infill with respect to the bare frame can be ex-

pressed by the non-dimensional stiffness ratio (Ei∙t)/Kc, in a way quite similar to that used in [1].

When the applied horizontal loads are progressively increased from 0 to their final value P, the infill and

frame separate over a large part of their interface. The pattern of this separation area, as well as of the area

remaining still in contact, is independent of the magnitude P of the applied loads and depends only on the

mechanical characteristics of the structural system, as it has been proved earlier in Section 2. Such a de-

formed shape of a typical square infilled frame is shown in Figure 3a, which was obtained from the results of

the proposed micromodel of Figure 2b. We can see that the active contact area of the interface is located

around the opposite corners of the shortened diagonal of the system (left up to right down), a fact that is al-

ready well-known from previous studies on the subject ([1], [2], etc.).

(a) (b)

(c) (d)

Figure 3: a) Deformed shape, b) Moment diagram (on compression side), c) Shear force diagram, d) Axial

force diagram, of a typical square infilled frame.

10

The patterns of the diagrams of the bending moments M, shear forces V and axial forces N of the typical

square infilled frame of Figure 3a, are shown in the Figures 3b, 3c and 3d respectively. In the separation area

of the interface (left down and right up), the moment and shear diagrams are very similar to the respective

diagrams of the bare frame. In the active contact area of the interface (left up and right down), and due to the

development of normal contact stresses, the bending moments M and shear forces V show a strong variation

and take their extreme values just at the corner nodes. In the same area, and due to the development of tan-

gential contact stresses, the tensile axial forces are considerably reduced near the left up corner of the frame,

while the compressive axial forces are concentrated and take their extreme values near the right down corner

of the frame.

In Figure 4, the distribution pattern of the membrane forces (membrane stresses multiplied by the thick-

ness t) within the infill panel of a typical rectangular infilled frame is shown. It is observed that along the

shortened diagonal of the infill panel and especially at the corners that remain in contact with the frame, there

is a high concentration of the normal forces Fxx, Fyy and of the minor principal normal forces Fmin. All of them

take their extreme negative values (compressive) just at the interface around the corner nodes, while they

progressively vanish when moving towards the left down and right up corners of the infill.

The effective (Von Mises) membrane forces FVM, show a similar stress pattern, while the pattern of the

shear forces Fxy is quite similar. In contrast, the major principal normal forces Fmax take their maximum value

(tensional) at the center of the infill.

Figure 4: Distribution of membrane forces within the infill panel of a typical infilled frame with µ=0

(darker=higher)

11

It must be noted that, due to different grading scales, the magnitude of the extreme minor principal normal

forces Fmin is about 10 times greater than that of the major principal normal forces Fmax. As a result, there is a

compressive stress state in nearly the whole area of the infill panel, which is characterized by the negative

value of the hydrostatic stresses σo=(Fxx+Fyy)/(3t) and of the corresponding membrane forces Fo=σo⋅t, as it

can be seen in Figure 4g. Lastly, the distribution of the maximum shear forces maxF12 =(Fmax−Fmin)/2, that are

defined by the half difference of the principal normal forces, is shown in Figure 4h. The hydrostatic forces

Fo, as well as the maximum shear forces maxF12, take their extreme values at the compressive (still in contact)

corners of the infill panel. The quantitative influence of several modelling parameters on the displacement

values and stress state of the examined structural system is presented in the next paragraphs.

Because neither the contact-friction interface conditions can be easily modelled analytically, nor can a con-

stant value for the static friction coefficient be determined accurately (considering that it depends on the

roughness between the contact surfaces, which actually varies), many times in the past analytical models with

rather simplified assumptions concerning the modelling of the interface conditions have been adopted (either

fixed-support conditions [22], or frictionless contact conditions). The effect of these simplified assumptions

on the response results of the infilled frames is investigated below by examining the following 4 types of in-

terface conditions: unilateral contact frictionless (μ=0), unilateral contact with μ=1, bilateral contact fric-

tionless (μ=0) and bilateral contact with μ=∞ (fixed support).

30 40

System's relative stiffness Ks/Kf

L/h=1,0 35 L/h=1,5

25 J2/J1=2 J2/J1=2

30

20

25

15 20

15

10

10

5

5

0 0

0 20 40 60 80 100 0 20 40 60 80 100

Infill's stiffness parameter: Ei.t/Kc Infill's stiffness parameter: E i.t/K c

(a) (b)

System's relative stiffness Ks/Kf

40 J2/J1=2

bilat. contact, µ=0

35 bilat. contact, fixed

30

25

P 1 2 P

20

15 Uy

10 h

5 W Ux

0 4 3

L

0 20 40 60 80 100

Infill's stiffness parameter: Ei.t/Kc

(c)

Figure 5: Relative system's stiffness Ks/Kf versus infill's stiffness parameter Ei⋅t/Kc

12

Figure 5 shows the variation of the relative lateral stiffness Ks/Kf for 3 different shapes of infilled frames

with respect to the stiffness of the corresponding bare frames, when the ratio J2/J1 equals 2 and the infill's

stiffness parameter Ei⋅t/Kc takes values between 0 and 100. Comparing the results of each type of interface

conditions with the results of the unilateral contact with μ=0, we can see that the models of unilateral contact

with μ=1 are about 25% stiffer, the models with frictionless bilateral contact are about 75% stiffer, while the

models with fixed support interface conditions are at least 300% stiffer. It is evident that the details of the

modelling of the interface conditions definitely affect the lateral stiffness of the system and that the substitu-

tion of the unilateral contact conditions with simplified ones leads to significant modelling errors.

Figure 6 shows the distribution of the effective stress (Von Mises) within the infill panel of a square in-

filled frame, for 3 different types of interface conditions and 2 cases of infill's stiffness (a "flexible" and a

"stiff" infill respectively). It is noted that the two models with unilateral contact interface conditions provide a

very similar stress pattern, in which the considerable stresses are distributed along the shortened diagonal of

the infill panel, and that the maximum stresses in the case of frictional contact are about 20% lower than those

of the frictionless case. In contrast, the models with fixed-support interface conditions provide a very differ-

ent stress pattern in which the large values of the stresses are located along both of the diagonals of the infill

panel, while the extreme values of the stresses are considerably lower that those of the unilateral contact mod-

els. It is evident that the effect of a realistic modelling of the interface conditions on the stress pattern and the

stress values in general, is very important indeed. At this point it must be noted that several studies using

plane-stress finite elements for the infill panels exist, in which the interface conditions are not properly mod-

elled ([2], [3], [4]) or have been considered to be fixed-supported ([23], 24, etc).

Figure 6: Distribution of effective stress (Von Mises) within a square infill panel, for several interface con-

ditions. (a), (b), (c): Flexible infill (Ei⋅t/Kc=6.5) (d), (e), (f): Stiff infill (Ei⋅t/Kc=100)

13

The accuracy of the results of any analytical model increases, in general, with the increase of the mesh den-

sity. On the other hand, an excessive mesh density increases the size of the numerical problem without im-

proving substantially the accuracy of the results. In order to survey this effect, alternative mesh densities of

the infill panel with 6×6 (4-node) elements, 4×4 elements and 2×2 elements are briefly examined, as well as

the 1-element macromodel.

Figure 7 shows the system's relative stiffness Ks/Kf of several models with J2/J1=2 and μ=0, where Ks and

Kf are the lateral stiffnesses of the infilled frame and the bare frame respectively. It also shows the percentage

difference between the stiffness of the examined models and the micromodel with 12×12 elements. We can

see that the alternative models with 6×6 and 4×4 elements, in all the examined cases of L/h ratio, provide re-

sults that are very close to the results of the micromodel (difference < 4,2%). The results of the alternative

model with 2×2 elements differ from those of the micromodel from 7,6% to 21,9% for the "stiff" infill, while

the results of the 1-element macromodel are rather inaccurate (difference from 44,2% to 83,8% for the "stiff"

infill). We can see also that the results of all the alternative models come closer to the results of the micro-

model with 12×12 elements, as the infill's stiffness decreases and, regarding the 2×2 elements and 1-element

models, as the ratio L/h increases.

Figure 7: Relative system's stiffness Ks/Kf and % difference from micromodel, for several mesh densities

(4-node elements).

14

Figure 8: Distribution of minor principal normal forces Fmin within a stiff infill panel (Ei⋅t/Kc=100), for sev-

eral mesh densities and 2P=375 kN (L/h=1,5 µ=0).

Figure 8 shows the distribution of the minor principal normal forces Fmin within a stiff infill panel

(Ei⋅t/Kc=100) with L/h=1,5 and µ=0, for several mesh densities. We observe that in the alternative model

6×6, the stress distribution is in good accordance with that of micromodel 12×12, while their extreme values

have a difference of about 10%. The stress distribution in the alternative micromodel 4×4 shows a similar pat-

tern, with the only noteworthy difference being the reduction of the peak stresses by about 20% in the area

very near to the compressive corners of the infill panel.

The pattern of the stress distribution changes significantly in the case of the alternative model 2×2 and the

same stands for their extreme values, which now differ from those of the micromodel by almost 55%. Never-

theless, a broad area along the compressive diagonal continues to be discerned quite clearly, where a rela-

tively higher concentration of stresses appears. The stress distribution that appears in the 1-element macro-

model is totally inaccurate, because the constant stress value of this model differs by more than 80% from the

extreme stress values in the corners of the micromodel and more that 20% from their values in its center.

From all the above, it becomes evident that a mesh density of 4×4 gives satisfactory results for the total

stiffness of the system and it is quite satisfactory for the stress distribution within the infill panel, while mesh

densities of more than 4×4 just define more accurately the extreme stress values in the small area near the

compressive corners of the infill panel.

Lastly, it is pointed out the fact that, although the macromodels with one 4-node plane stress element and

contact interface conditions give the poor results mentioned above, there are several publications where the

even more inaccurate macromodels with one 4-node element and fixed-support interface conditions are used

([23],[24], etc.).

15

The parameter of the friction coefficient μ at the interface between frame and infill is considered to take val-

ues between μ=1 (friction angle φ=45°) and μ=0 (φ=0°), thus covering sufficiently the cases that appear in

the practical applications. This variation of the friction coefficient affects the lateral stiffness Ks of the sys-

tem, along with the stress distribution in the infill panel and the frame.

The change in the system’s lateral stiffness can be expressed as a percentage of the lateral stiffness of the

respective system without friction, as it is shown in Figures 9 and 10. These diagrams show the change of the

system’s stiffness versus the variation of friction coefficient μ for the case of “flexible” (Ei⋅t/Kc=6,5) and

“stiff” (Ei⋅t/Kc=100) infill panels, when the ratio J2/J1 between the column and beam moments of inertia takes

4 different values from 0,25 to 8,0.

The curves in the diagrams 9a and 9b correspond to a ratio L/h=0,75 of the infilled frame, the curves in

the diagrams 9c and 9d to a side ratio L/h=1, the curves in the diagrams 10a and 10b to a ratio L/h=1,5 and

the curves in the diagrams 10c and 10d to a ratio L/h=2. Observing these diagrams the following conclusions

can be drawn.

30 40

L/h=0,75 35 L/h=0,75

% Stiffness Ks variation

% Stiffness Ks variation

25 Ei.t/Kc=6,5 Ei.t/Kc=100

30

20

25

15 20

15

10

10

5

5

0 0

0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

Friction coefficient µ (J2/J1=0,25) Friction coefficient µ

(a) (J2/J1=1,0) (b)

(J2/J1=2,0)

(J2/J1=8,0)

30 40

L/h=1,0 35 L/h=1,0

% Stiffness Ks variation

% Stiffness Ks variation

25 Ei.t/Kc=6,5 Ei.t/Kc=100

30

20

25

15 20

15

10

10

5

5

0 0

0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

Friction coefficient µ Friction coefficient µ

(c) (d)

Figure 9: Variation of system's stiffness versus friction coefficient, for several values of J2/J1 and infill's

stiffness:. (a), (b): For L/h=0,75, (c), (d): For L/h=1,00

16

30 40

L/h=1,5 35 L/h=1,5

% Stiffness Ks variation

% Stiffness Ks variation

25 Ei.t/Kc=6,5 Ei.t/Kc=100

30

20

25

15 20

15

10

10

5

5

0 0

0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

Friction coefficient µ (J2/J1=0,25) Friction coefficient µ

(a) (J2/J1=1,0) (b)

(J2/J1=2,0)

(J2/J1=8,0)

30 40

L/h=2,0 35 L/h=2,0

% Stiffness Ks variation

% Stiffness Ks variation

25 Ei.t/Kc=6,5 Ei.t/Kc=100

30

20

25

15 20

15

10

10

5

5

0 0

0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

Friction coefficient µ Friction coefficient µ

(c) (d)

Figure 10: Variation of system's stiffness versus friction coefficient, for several values of J2/J1 and infill's

stiffness: (a)-(b): For L/h=1,50, (c)-(d): For L/h=2,00

• The existence of friction at the interface increases the lateral system’s stiffness with respect to the fric-

tionless case. This stiffness increase is almost proportional to the increase of the value of the friction co-

efficient μ.

• For a given value of the ratio L/h the increase in the system’s stiffness, due to the increase of the friction

coefficient μ, is greater in the case of stiffer infill panels.

• The increase in the system’s stiffness depends definitely on the values of the ratios L/h and J2/J1.

• For the value μ=1 of the friction coefficient, the size of the increase in the system’s stiffness varies from

8% up to 26% in the case of “flexible” infill panels (Ei⋅t/Kc=6,5), and from 18% up to 37% in the case of

“stiff” infill panels (Ei⋅t/Kc=100).

The effects of the variation of the friction coefficient on the stress distribution within a stiff infill panel

(Ei⋅t/Kc=100) with L/h=1,5 is shown is Figure 11. In particular, Figure 11a shows the variation of the extreme

values of the minor principal stresses (per unit thickness) Fmin for three different cases of the ratio J2/J1. We

can see that the three curves of the diagram differ only a little from a linear variation and that the extreme

stress values for μ=1, are less than those for μ=0, from 33% up to 40%. In Figures 11b,c,d the distribution of

the minor principal stresses Fmin and their extreme values are shown for three different values of the friction

coefficient μ. We can see that the three patterns are very similar to each other in the range of small and me-

17

dium values and also take very similar values. Only in the area of the large values, which is located near the

compressive corners of the infill, these stresses vary more rapidly (thus taking different extreme values) as

the friction coefficient μ decreases. In conclusion the variation of the friction coefficient affects the stress pat-

tern less than the system's stiffness, while it affects the extreme stress values almost as much as the system's

stiffness.

Figure 11: Minor principal stresses Fmin within a stiff infill panel (Ei⋅t/Kc=100) for L/h=1,5 and 2P=375 kN

(a) Extreme values versus friction coefficient μ, (b)-(c)-(d) Distribution of stresses for several val-

ues of friction coefficient and J2/J1=8

18

Denoting by Ei and Gi the elasticity and shear modulus of the infill panel, several cases of orthotropic materi-

als are examined in which the ratio Ey/Ex takes values between 0,5 and 2, where Ex=Ei and Gxy=Gi. In these

notations the subscripts x and y refer to the orthotropic infill and the subscript i to the corresponding isotropic

infill with Ex=Ei. The effects of the infill's orthotropy on the system's response is examined by comparing the

results of each different case of the ratio Ey/Ex, with the results of the case Ey/Ex=1, which corresponds to the

respective isotropic infill panel.

Figure 12 shows the system's relative stiffness Ks/Kf of several "flexible" and "stiff" orthotropic models

with J2/J1=2 and μ=0, where Ks and Kf are the lateral stiffnesses of the infilled frame and the bare frame re-

spectively. It also shows the percentage difference in the stiffness of the orthotropic models with respect to

the isotropic case (Ey/Ex=1). Observing these diagrams the following conclusions can be drawn.

• The effect of the infill's orthotropy on the system's lateral stiffness increases as the infill's stiffness and or-

thotropy increase and as the ratio L/h decreases.

• When the ratio L/h takes values L/h≥1,5 the effect of the infill's orthotropy on the system's lateral stiff-

ness becomes negligible (difference < 5,5%) in both cases of "flexible" (Ei⋅t/Kc=6,5) and “stiff” infills

(Ei⋅t/Kc=100), and in both the extreme cases of the examined infill's orthotropy (i.e. Ey/Ex=1/2 and

Ey/Ex=2).

• When the ratio L/h equals 1, the effect of the infill's orthotropy on the system's lateral stiffness remains

less than 8% in the cases with Ey/Ex=1/1,5 and Ey/Ex=1,5 while it remains less than 15% in both the ex-

treme cases of the examined infill's orthotropy (i.e. Ey/Ex=1/2 and Ey/Ex=2).

Figure 12: Horizontal system's stiffness for orthotropic infill panel and several values of L/h and Ey/Ex

19

The effects of the variation of the infill's orthotropy on the stress distribution within several "flexible" and

"stiff" orthotropic models with J2/J1=2 and μ=0, are shown in Figures 13 and 14. In particular Figures 13a,b

show the maximum values of the effective stress Fvm (per unit thickness) for 2 differents cases of the ratio

L/h, while Figures 14a,b show the respective maximum values of the shear stress Fxy (per unit thickness).

These figures also show the percentage difference in the maximum values of the orthotropic models with re-

spect to the isotropic case (Ey/Ex=1). It is evident that:

• The effects of the infill's orthotropy on the maximum values of these stresses increase as the orthotropy

increases and as the infill's stiffness and the ratio L/h decrease.

• When the ratio L/h equals 2, the effects of the infill's orthotropy on the maximum values of the above

stresses becomes negligible (difference <5%) in both cases of "flexible" (Ei⋅t/Kc=6,5) and “stiff” infills

(Ei⋅t/Kc=100), and in both cases of the examined infill's orthotropy (i.e. Ey/Ex=1/2 and Ey/Ex=2).

• When the ratio L/h equals 1, the effect of the infill's orthotropy on the maximum values of the above

stresses varies from 11% to 19% in the case of effective stresses and from 3% to 12% in the case of shear

stresses.

Figure 13: Effective (Von Mises) stress in orthotropic infill panels, for several values of Ey/Ex

(a), (b): max values and percentage difference relative to isotropic infill (2P=375 kN).

(c), (d): Stress distribution within a square "flexible" infill panel (2P=375 kN).

20

Finally Figures 13c,d and 14c,d comparatively show the distribution of the above stresses within a "flexi-

ble" infill panel (Ei⋅t/Kc=6,5), for the extreme values of the infill's orthotropy (i.e. Ey/Ex=1/2 and Ey/Ex=2). We

can see that the patterns of the effective stresses for Ey/Ex=1/2 and Ey/Ex=2 in Figures 13c and 13d are very

similar to each other and that the contour lines of these diagrams show stress states that are proportional to

the extreme values of the diagrams. Similar observations can also be made for the distributions of the shear

stresses in Figures 14c and 14d.

In conclusion and within the examined limits of the Ey/Ex ratio (0,5≤ Ey/Ex ≤2), the infill orthotropy does

not significantly affect the behaviour of the rectangular infilled frames with L/h≥1,5. The same conclusion is

also valid for the case of L/h≥1, if the infill's orthotropy varies between (1/1,5)≤ Ey/Ex ≤1,5. Consequently, in

all theses cases the infill's orthotropy could be ignored (thus considering an isotropic infill panel) if the pa-

rameters of the infill's orthotropy were not explicitly known (provided that they vary between the above men-

tioned limits).

Figure 14: Shear stress Fxy in orthotropic infill panels, for several values of Ey/Ex

(a), (b): max values and percentage difference relative to isotropic panel (2P=375 kN)

(c), (d): Stress distribution within a square flexible infill panel (2P=375 kN)

21

1 2 u x=1

J2, F∞

J1, F1 J1, F1

h h' h h'

bare frame

uy

J2, F∞ K fx

w ux

4 3

L' L'

r'=(h-h')/h'

L L

Figure 15: Infilled frame and respective bare frame with rigid end-offsets.

The lateral flexural stiffness of the bare frame in Figure 15 is given by the formula:

where E is the elasticity modulus of the frame, h' and L' are the clear lengths of the columns and beams (infill

dimensions) and J1 and J2 are the inertia moments of the respective sections. Also the values of the bending

moments at the ends of the clear lengths of the columns and beams are respectively given by:

The above relations show that the lengths h and L do not affect the infill's stiffness, nor do the bending mo-

ments within the clear lengths of columns and beams of the bare frame. They only affect the size of the rigid

offsets (section heights (L-L')/2 and (h-h')/2) between the neutral axes of the columns and beams and their in-

ternal faces (see also Fig. 2b).

Regarding the infilled frames, these rigid offsets are responsible for the presence of distributed moment

loads on the frame neutral axis, due to the existence of tangential friction forces at the frame-infill interface.

Consequently, the ratio r'=(h-h')/h'=(L-L')/L' which measures the relative size of the frame to the infill, also

measures the size of the distributed moment loads on the frame, and therefore it can be effectively used for

further studies of the infilled frames.

Figure 16: Deformed shape and disribution of minor principal stresses within a stiff infill panel with μ=0

and h'=2,6 m. (a) for r'=(h-h')/h'=0,15 (b) for r'=(h-h')/h'=0,45

22

Figure 17: Deformed shape and disribution of minor principal stresses within a stiff infill panel with μ=1

and h'=2,6 m. (a) for r'=(h-h')/h'=0,15 (b) for r'=(h-h')/h'=0,45

In the special case of frictionless contact (μ=0) the developed moment loads vanish, thus giving stress

states within the clear lengths of the columns and beams (and consequently within the infill panel) which are

independent of the size of the rigid offsets (L-L')/2 and (h-h')/2. This fact is clearly shown in Figure 16, where

the compared infilled frames have the same clear lengths h' and L'. The result is an identical distribution of in-

ternal frame moments within these lengths and consequently an identical distribution of stresses within the in-

fill panel, although their respective lengths h and L are considerably different. In contrast, in the case of fric-

tional contact (μ>0) these moment loads can significantly affect the stress distribution within the frame and

the infill. This fact is also clearly shown in Figure 17, which refers to infilled frames that are identical to those

shown in Figure 16 except for the size of the friction coefficient which takes the value μ=1.

The developed moment loads along the axes of the frame elements are proportional to the ratio r' and to

the size of the friction coefficient μ, thus causing an increase in the lateral stiffness of the frame and the sys-

tem with respect to the frictionless case. The dependence of the system’s stiffness Ks(r') on these parameters

is shown in Figure 18a for the case of a “stiff” infill panel with L/h=1,5 and J2/J1=2. From this figure we can

see that the increase of the normalized system’s stiffness Ks(r')/Ks(0) is almost proportional to the value of r'.

The normalization is performed with respect to the lateral stiffness of the corresponding infilled frame with

zero rigid offsets (r'=0) and this normalization is identical to that performed with respect to the frictionless

case (μ=0), the value of r' remaining unchanged.

For the above case of infilled frame, Figure 18b shows the variation of the normalized values of the maxi-

mum frame moments M1s(r') as a function of the ratio r'. Also Figures 18c and 18d respectively show the

variation of the normalized extreme values of the minor principal stress Fmin and the effective stress FVM (per

unit thickness) within the infill panel. The normalization is performed with respect to the corresponding in-

filled frame with zero rigid offsets (r'=0). It is evident that, for a given friction coefficient and a given size of

the infill panel, the system’s stiffness increases as the relative size of the frame increases (increasing r').

Moreover, the maximum frame moment and the extreme values of the stresses within the infill panel de-

crease, as the relative frame's size increases.

23

1.80 1.80

1.00 1.00

0.60 0.60

0.20 0.20

0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5

r'=(h-h')/h' µ=0 r'=(h-h')/h'

(a) µ=0,5 (b)

µ=1

1.80 1.80

extr infill's stress Fmin(r')/Fmin(0)

1.40 1.40

1.00 1.00

0.60 0.60

0.20 0.20

0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5

r'=(h-h')/h' h'=2,6 m J2/J1=2

r'=(h-h')/h'

(c) L/h=1,5 Ei.t/Kc=100 (d)

Figure 18: Effects of the infill's relative size (ratio h'/h) on the system's response (normalized values):

(a) horizontal system's stiffness, (b) maximum frame moment, (c) minimum principal stress, (d)

maximum effective (Von Mises) stress.

The ratio L/h defines the geometric shape of the rectangular infilled frames, and the effects of its variation on

the system’s response is investigated by considering four different values of L/h, between 0,75 and 2,0. The

high dependence of the system’s response on the L/h ratio has been already recognized from the first studies

that have been conducted on the infilled frames ([1], etc.), as well as from almost all the subsequent ones.

The ratio J2/J1 of the inertia moment of the beam section to the column section, which, along with the ratio

L/h, describes the beam/column relative stiffness k=(J2/J1)⋅( h'/L')=(J2/J1)⋅( h/L) of the bare frame, also char-

acterizes its deformed shape, which further affects the way in which the frame and the infill panel interact

with each other. In the bare frame the flexural deformation of the columns, for small values of the ratio J2/J1,

turns up to be a lot smaller than that of the beams (almost straight columns), while for larger values of the ra-

tio J2/J1 the opposite takes place. The influence that the ratio J2/J1 has on the system’s response has not been

thoroughly studied until now, possibly because in the first experimental studies on the infilled frames ([1],

etc.) it is mentioned that no noteworthy dependence of the results on the variation of J2 has come up. As a re-

sult, the J2 parameter has not been considered into the semi-empirical formulas which have been suggested at

that time for the "equivalent" compressive strut model and have been widely used ever since.

24

In the present study the ratio J2/J1 takes four different values from 0,25 ("flexible" beams) to 8,0 ("stiff"

beams). The effects of the J2/J1 ratio, together with the effects of the L/h ratio on the system’s response, are

shown in the diagrams that follow.

8 12

L/h=0,75 L/h=1,00

System's stiffness Ks/Kc

µ=0 µ=0

6 9

4 6

2 3

0 0

0 20 40 60 80 100 0 20 40 60 80 100

Infill's stiffness parameter E i.t/K c J2/J1=0,25 Infill's stiffness parameter E i.t/K c

(a) J2/J1=1,0 (b)

J2/J1=2,0

J2/J1=8,0

16 16

L/h=1,50 L/h=2,00

System's stiffness Ks/Kc

System's stiffness Ks/Kc

µ=0 µ=0

12 12

8 8

4 4

0 0

0 20 40 60 80 100 0 20 40 60 80 100

Infill's stiffness parameter E i.t/K c Infill's stiffness parameter E i.t/K c

(c) (d)

Figure 19: Normalized system's stiffness Ks/Kc versus infill's stiffness parameter Ei⋅t/Kc, for

several values of J2/J1 (μ=0).

Figure 19 shows the variation of the normalized lateral system's stiffness Ks/Kc (with respect to the column

stiffness Kc) versus the infill's stiffness parameter Ei⋅t/Kc, for all the examined values of the ratios L/h and

J2/J1 for the case of frictionless contact. From these diagrams the following conclusions can be readily drawn:

• The increase in the system's stiffness is almost proportional to the increase of the infill's stiffness.

• The increase in the system's stiffness is definitely dependent on both the L/h ratio and the J2/J1 ratio.

• The diagrams 19c and 19d are very similar, which means that for values of the L/h ratio greater than 1,5

and μ=0, the system's stiffness tends to be independent of the L/h ratio.

• The system's stiffness increases as the ratio J2/J1 increases.

• For a given L/h ratio, the increase in the system's stiffness, due to the increase of the J2/J1 ratio, becomes

greater as the infill's stiffness decreases. It must be mentioned that for a flexible infill panel with

Ei⋅t/Kc≤10, the system's stiffness corresponding to a stiff beam (J2/J1=8) is more than 80% greater than the

system's stiffness corresponding to a weak beam (J2/J1=0,25), regardless of the L/h ratio. Moreover, for a

stiff infill panel with Ei⋅t/Kc=100, the system's stiffness corresponding to a stiff beam (J2/J1=8) is about

20% greater than the system's stiffness corresponding to a weak beam (J2/J1=0,25).

25

An alternative representation of the dependence of the system's stiffness on the L/h and J2/J1 ratios is pre-

sented in Fig. 20, which shows the variation of the relative lateral system's stiffness Ks/Kc with respect to

stiffness Kf of the bare frame. It is obvious that the existence of an infill panel within a bare frame increases

the system's stiffness several times. In the case of L/h=0,75 this increase varies from 4 times (beams with

J2/J1=8) to 12 times (beams with J2/J1=8), while in the case of L/h=2,0 this increase varies from 9 to 50 times

respectively.

20 30

System's relative stiffness Ks/Kf

L/h=0,75 L/h=1,00

16 µ=0 24 µ=0

12 18

8 12

4 6

0 0

0 20 40 60 80 100 0 20 40 60 80 100

Infill's stiffness parameter E i⋅t/K c J2/J1=0,25 Infill's stiffness parameter Ei⋅t/Kc

(a) J2/J1=1,0 (b)

J2/J1=2,0

J2/J1=8,0

40 50

System's relative stiffness Ks/Kf

System's relative stiffness Ks/Kf

L/h=1,50 L/h=2,00

µ=0 µ=0

32 40

24 30

16 20

8 10

0 0

0 20 40 60 80 100 0 20 40 60 80 100

Infill's stiffness parameter Ei⋅t/Kc Infill's stiffness parameter Ei⋅t/Kc

(c) (d)

Figure 20: Relative system's stiffness Ks/Kf versus infill's stiffness parameter Ei⋅t/Kc, for

several values of J2/J1 (μ=0).

The effects of the variation of the L/h and J2/J1 ratios, on the values of the maximum bending moments

within the frame, are shown in Figure 21. The values of these diagrams are normalized with respect to the

maximum moment of the corresponding bare frame. It is evident that:

26

1.2 1.2

max frame moment M1s/M1f

L/h=0,75 L/h=1,00

1.0 µ=0 1.0 µ=0

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0

0 20 40 60 80 100 0 20 40 60 80 100

Infill's stiffness parameter E i .t/K c J2/J1=0,25 Infill's stiffness parameter E i.t /K c

(a) J2/J1=1,0 (b)

J2/J1=2,0

J2/J1=8,0

1.2 1.2

max frame moment M1s/M1f

L/h=1,50 L/h=2,00

1.0 µ=0 1.0 µ=0

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0

0 20 40 60 80 100 0 20 40 60 80 100

Infill's stiffness parameter E i .t /K c Infill's stiffness parameter E i.t /K c

(c) (d)

Figure 21: Normalized maximum bending moment M1s/M1f of infilled frame versus infill's stiffness pa-

rameter Ei⋅t/Kc, for several values of J2/J1 and μ=0.

• The decrease in the maximum bending moments is mainly dependent on the J2/J1 ratio.

• The decrease in the maximum bending moments, due to the decrease of the J2/J1 ratio, becomes greater as

the infill's stiffness increases.

• For Ei⋅t/Kc≥5, the maximum bending moments of the curve J2/J1=8 is greater than those of the curve

J2/J1=0,25 by about 50% for L/h=0,75 and by more than 100% for L/h=2,0.

Finally in Figure 22 the effects of the values of the J2/J1 and L/h ratios on the stress state of a stiff infill

panel (Ei⋅t/Kc=100) are shown. In particular, Figure 22a shows the extreme values of the minor principal

stresses Fmin (per unit thickness) for 2 different values of the J2/J1 ratio and 4 values of the L/h ratio. Figures

22b,c,d,e show the distribution of the minor principal stresses within the infill panel for the case of flexible

beams (J2/J1=0,25) while Figures 22f,g,h,i show the distribution of these stresses for the case of stiff beams

(J2/J1=8). We can see that the respective patterns (for the same value of L/h) of these distributions are not af-

fected by the value of J2/J1, and only the stress values are affected.

In conclusion, the lateral system's stiffness and the stress state within the frame and the infill panel are

definitely affected by the variation of the J2/J1 ratio, as well as of the L/h ratio.

27

Figure 22: Minor principal stresses (kN/m) within a stiff infill panel (Ei⋅t/Kc=100) for horizontal loading

2P=1 kN and several values of L/h. (a): Extreme values for J2/J1=0,25 and J2/J1=8

(b)÷(e): Stress distribution for J2/J1=0,25, (f)÷ (i): Stress distribution for J2/J1=8.

28

400000

micromodel with µ=0

System's stiffness Ks (kN/m)

L/h=1,00

J2/J1=2,0 micromodel with µ=1

300000 diagonal strut (Smith)

200000

100000

0

0 1000000 2000000 3000000

Infill's stiffness Ei⋅t (kN/m)

Figure 23: Comparison of the lateral system’s stiffness Ks obtained by the proposed micromodel and the

diagonal strut model of Smith and Carter [1].

An indirect comparison of the proposed micromodel with a large amount of existing experimental data can be

performed, by comparing characteristic results of the micromodel with those obtained by the well-known di-

agonal-strut model. In the diagonal-strut model considered herein, the "equivalent width w" has been calcu-

lated by a semi-empirical formula suggested in [1] by Smith and Carter. In this reference various checks have

been made, which verified that the suggested formula gave results very close to existing experimental ones.

Figure 23 shows the lateral system’s stiffness of the compared models that correspond to a square infill panel

with J2/J1=2,0. We observe that a very close agreement exists between the stiffness of the diagonal-strut

model and the examined micromodel when μ=1, but a considerable difference exists when μ=0. It must be

also noted that the semi-empirical formula suggested by Smith and Carter cannot take into consideration the

effects of the friction coefficient μ and of the J2/J1 ratio on the system’s lateral stiffness, thus concluding that

the accuracy and applicability range of this formula is questionable in general.

6. CONCLUSIONS

The evaluation of the results of the extended analytical investigation showed that several critical modelling

parameters exist, which considerably affect the behaviour of infilled frames subjected to lateral loading. The

following conclusions summarize the discussion on the results and findings presented above:

• The details of the modelling of the interface conditions definitely affect the lateral stiffness of the system,

the stress pattern and the stress values in general, while the substitution of the unilateral contact interface

conditions with simplified ones can lead to significant modelling errors.

• Regarding the necessary mesh density, it is noted that a 4×4 density with 4-node elements for the infill

panel (in the assumption of a linear material behaviour), gives satisfactory results for both the total stiff-

ness of the system and the stress pattern within the infill panel. Finer mesh densities define more accu-

rately the stress pattern and the extreme stress values in the area just near to the compressive corners of

the infill, and must be generally preferred in case of inelastic material behaviour.

• The existence of friction at the interface increases the lateral system’s stiffness with respect to the fric-

29

tionless case. This stiffness increase is almost proportional to the increase of the value of the friction co-

efficient μ, is greater in the case of stiff infill panels and definitely depends on the values of the L/h and

J2/J1 ratio.

• Within the limits (1/1,5)≤ Ey/Ex ≤1,5) of the variation of the infill's orthotropy, the orthotropy does not

considerably affects the behaviour of the rectangular infilled frames and can be ignored. In such a case, an

isotropic infill panel could be considered.

• The relative frame/infill size affects the system’s stiffness almost proportionally to the size of the friction

coefficient. For large values of the friction coefficient the influence on the system’s response is signifi-

cant, while in the frictionless-contact case the influence is negligible.

• The system’s response is definitely affected by the variation of the L/h ratio, as well as of the J2/J1 ratio. It

is mentioned that the influence of the J2/J1 ratio has been systematically neglected in the various semi-

empirical formulas suggested for the "effective width" of the widely used "compressive-strut" model, thus

making the reliability of these formulas questionable.

The above conclusions are obviously valid not only for the case of the examined single-storey one-bay in-

filled frames, but also for nearly any case of multi-storey ones. This is due to the fact that the single-storey

one-bay systems constitute the smallest possible typical substructure in which a multi-storey infilled frame

can be decomposed, while preserving at the same time all the basic features of a frame-infill interaction sys-

tem. All these conclusions can be utilized for the design and reliable analytical modelling of infilled frames in

general, as well as for the formulation and verification of various types of simplifying models and macro-

models for the frame-infill interaction system, taking into consideration the quantitative effects of the exam-

ined modelling parameters on the system’s behaviour. They can be also utilized for the investigation of ine-

lastic response, where a detailed parametric study is a much more complicated matter, due to the large num-

ber of parameters introduced. In that case, it is very useful to know which simplifying assumptions can be

made, in order to reduce the number of parameters that the analytical models use and to preserve only the

critical ones which considerably affect the accuracy of the results.

REFERENCES

1. Smith B.S., Carter C. A method of analysis for infilled frames. Proc. Inst. Civ. Engrs. 1969; Vol. 44.

2. Mallick D.V., Severn R.T. The behaviour of infilled frames under static loading, Proc. Inst. Civ. Engrs,

1968; 20:639-656.

3. Riddington J.R., Smith B.S. Analysis of infilled frames subjected to racking with design recommenda-

tions. Struct. Engr, 1977; 55:263-268.

4. King G.J., Pandley D.C. The analysis of infilled frames using finite elements. Proc. Inst. Civ. Engrs, 1978;

65:749-760.

5. Achyutha H., Jagadish R., Rao P.S., Shakeebur Rahman S. Finite element simulation of the elastic behav-

iour of infilled frames with openings. Computers and Structures, 1986; 23(5):685-696.

6. Doudoumis I.N., Mitsopoulou E.N., Nikolaidis, G.N. A macroelement for the simulation of the infill pan-

els in multistorey frames under horizontal seismic actions. In: Proceedings of the 10th European Confer-

ence on Earthquake Engineering (Vienna, 1994): A.A. Balkema publ., Rotterdam, 1995. p.1371-1376.

7. Doudoumis I.N., Mitsopoulou E.N. Analytical modelling of infill panels using inelastic macroelements

with contact interface conditions. In: Proceedings of the 11th European Conference on Earthquake Engi-

neering (Paris, 1998): A.A. Balkema publ., Rotterdam, 1998. CD version.

8. Kwan K.H., Liauw T.C. Nonlinear analysis of integral infilled frames. Engng Struct., 1984; 6:223-231.

9. Liauw T.C., Kwan K.H. Nonlinear behaviour of non-integral infilled frames. Computers and Structures,

1984; 18(3):551-560.

30

10. May I.M., Naji J.H. Nonlinear analysis of infilled frames under monotonic and cycling loading. Com-

puters and Structures, 1991; 38(2): 149-160.

11. El Haddad M.H. Finite element analysis of infilled considering cracking and separation phenomena. Com-

puters and Structures, 1991; 41(3): 439-447.

12. Singh H., Paul D.K., Sastry V.V. Inelastic dynamic response of reinforced concrete infilled frames. Com-

puters and Structures, 1998; 69: 685-693.

13. Mehrabi A.B., Shing P.B. Finite element modelling of masonry-infilled RC frames. J. Struct. Engr, 1997;

123(5): 604-613.

14. Ghosh A.K., Amde A.M. Finite element analysis of infilled frames. J. Struct. Engr, 2002; 128(7): 881-

889.

15. Riddington J.R. The influence of initial gaps on infilled frame behavior. Proc. Inst. Civ. Engrs, 1984;

part2, 77: 295-310.

16. Klarbring A.: Contact problems with friction - using a finite dimensional description and the theory of lin-

ear complementarity. Linkoping Studies in Science and Techology, Thesis No 20, Linkoping Institute of

Technology, Sweden, 1985.

17. Doudoumis I.N. A mathematical programming incremental formulation for unilateral frictional contact

problems of linear elasticity. Applicable Analysis, 2003; 82 (6): 503-515.

18. Maier G. Incremental plastic analysis in the presence of large displacements and physical instabilizing ef-

fects. Int. J. Solids Structures, 1971; 7: 345-372.

19. Doudoumis I.N., Mitsopoulou E.N. A class of sufficiently small friction coefficients for the uniqueness of

the solution of the quasistatic contact problem. Mathl. Comput. Modelling, 1998; 28 (4-8): 309-321.

20. Lemke C. E. On Complementary Pivot Theory. Mathematics of the Decision Sciences, Part 1, 1968; ed-

ited by G. B. Dantzig and A. F. Veinott, Jr. American Mathematical Society.

21. Mitsopoulou E.N., Doudoumis I.N. A contribution to the analysis of unilateral contact problems with fric-

tion. Solid Mechanics Archives, 1987; 12(3); 165-186.

22. Doudoumis I.N., Mitsopoulou E.N. The deficiency of macromodels with fixed-support interface condi-

tions in the analytical modelling of infill panels. In: Proceedings of the 5th SECED conference - European

Seismic Design Practice (Chester U.K 1995): A.A. Balkema publ., Rotterdam, 1995. p. 427-434.

23 Kappos A.J., Stylianidis K.C., Michailidis C.N. Analytical models for brick masonry infilled R/C frames

under lateral loading. Journal of Earthquake Engineering, 1998; 2 (1): 59-87.

24 Dymiotis C., Kappos A.J., Chryssanthopoulos M.K. Seismic reliability of masonry-infilled RC frames. J.

Struct. Engr, 2001; 127(3): 296-305.

- Smex1045 Strength of Materials 2 1 0 3[1]-1Uploaded bysankarsuper83
- BOLTGRPUploaded byjembray
- CE6306 Strength of Materials(SOM)Uploaded bygoldencomet
- shellsUploaded byravi1214
- A Layered Limit Analysis of Pinned-joints Composite Laminates Numerical Versus Experimental FindingsUploaded bykhudhayer1970
- Intersections V06 No04 04Uploaded byBloom Bloom Iasi Romania
- Calculation SheetUploaded byShivendra Kumar
- Beams Membranes Plates and Shells BMPS Statements of CompetenceUploaded byShanice McAlister-Bishop
- Truss problem 6.1 Johnston and Beer book-Static 1-1.docxUploaded bymahfuzul_haque
- JL-94-January-February Horizontal Shear Strength of Composite Concrete Beams With a Rough InterfaceUploaded byKTMO
- Implant Dentar Personalizat BioMicron Transilvania-Study 1-1Uploaded byAlexandrina Buga
- Importance of shear assessment of concrete structures detailed to different capacity design requirementsUploaded byyasser_goldstone
- 11 IJAERSUploaded byJasmine Rai
- Theories of Stress and StrainUploaded byLutful Hasan Sayed
- 9A01301 Mechanics of SolidsUploaded bysivabharathamurthy
- Form Finding of Tensile MembranesUploaded bydjrexy
- DiscussionUploaded byGowtham Ragunathan
- CE IES 2012 Conventional Paper IUploaded byGaurav Pawar
- 11-82-13 (vol.66 no1).pdfUploaded byHassaan Ullah Khan
- art02Uploaded byJuan Andres
- 1-s2.0-S0141029611003725-main.pdfUploaded byparth daxini
- Strength computation of pin connection through 3D FE idealisationUploaded byFabio Okamoto
- Full TextUploaded byPramod Pamu
- IASS Lienhard Knippers Hybrid TextilesUploaded byRaghu Sagar
- MOFabRRD 1994Uploaded byRufus Dinrifo
- Q bankUploaded byNitin Sharma
- 503_ftpUploaded byToan Pham
- MINGGU KE 1 KONSEP TEGANGAN - REGANGAN.pptUploaded byreza aditya
- Huang SubgradeUploaded byLorea Elvis
- Is.456.2000 - Plain & Reinforced Concrete_Part13Uploaded bypandey71087

- foundation design sampleUploaded byArun Kumar
- bdp-16.pdfUploaded byČhhït
- Group pileUploaded byChien Chieu
- Brian Simpson-8th Lumb Lecture2014Uploaded byyuan
- Handout 9 13Uploaded bymoganna73
- En Midas Design Catalog LowUploaded byArnoldoAyalaSalamanca
- Basic+Soil+Mechanics +Volume+2+(R.whitlow)Uploaded byDu Nguyen Hoang
- Etabs Rc Slab DesignUploaded byEmanuelRodriguezElera
- Japanese-H-Sections.pdfUploaded byChien Chieu
- ASCE710EUploaded byArianBaghi
- Table of Concrete Design Properties (Fcd,Fctm,Ecm,Fctd) - Eurocode 2Uploaded byCheng Jiang
- bs.na.en.1998.1.2004Uploaded byno_milk_today
- Analysis of Wind Loads on Buildings and Signs a Computer Program Based on Asce 7Uploaded bynouman
- Using Matlab as Modelling Tool for Civil Engineering Design ProjectsUploaded bynelsonsainz
- PDFUploaded byChien Chieu
- SCD-AISC-360-10Uploaded byGiurcanas Andrei
- tv_19_2012_3_643_652Uploaded byQAISARNADEEM
- AFES-Tutorial_for_Tank_Foundation_English_2006_07_31.pdfUploaded byFakhrurozi Fani
- Doan Code Tinh Noi SuyUploaded bytrung1983
- Seismic Load AnalysisUploaded byMANDARAW
- Norma CMAA 70Uploaded byMarcelo Navarro
- Comflor Decking NewUploaded byCristina Capota
- Handout-10-2017Uploaded byChien Chieu
- Shear WallUploaded byChien Chieu
- Dynamic AnalysisUploaded byBijoy Koroth
- Wind LoadingUploaded byChien Chieu
- 14. Psc Design Aasht-lrfd 12 TutorialUploaded byChien Chieu
- ARUP_Design of Deep Beams in Reinforced Concrete CRIA 2 OAUploaded byChien Chieu

- Bluetooth Board Installation GuideUploaded byMarco Antonio Iribarra Vergara
- Circuit Design for ReliabilityUploaded byGábor Sörös
- CT-SHS Joint Design Examples 15-02-06Uploaded bypich222
- Comparative Study on Braking Wave Forces on Vertical Walls With Cantilever SurfacesUploaded bysethhoffman85
- ElementUploaded byHajar Yusop
- NSlookupUploaded byrahul_2554
- NOx Fluidized Bed-9Uploaded byJoão Diego Feitosa
- Best Builder FinalUploaded byJhon Walter Ortega Conde
- Thermal Runaway DetectionUploaded byarjmandquest
- Unsteady Operation of New Type Turbofan Engine with Aerodynamic Torque.pdfUploaded bybaharmarine
- Enhance the AIS data availability by screening and interpolationUploaded byMia Amalia
- Amazing rubber band cars.pdfUploaded byShauna Clark
- Oracle. SCM Cloud Implementing Product ManagementUploaded byManu Sethumadhavan
- MODFLOWSURFACT_V3.pdfUploaded byMuhammad Zarkasih
- ASHRAE ANSI IES Final Quantitative Analysis Report 90.1 - 2010 Determination Oct2011_v00Uploaded byronychaves
- UT Dallas Syllabus for cs6383.001.10s taught by Ying Liu (yxl059100)Uploaded byUT Dallas Provost's Technology Group
- 19.2-fluid_density_viscosity___drag-edexcel_ial_physics-qp.pdfUploaded bymrstalib
- 14313_ch6 M61A1 GUN INSTALLATION.pdfUploaded bySeth Pulgar
- ansys Chapter 4Uploaded byavinashj18
- Bucklimng of Composite PlatesUploaded bykiran_wakchaure
- Ficha tecnica octal GP 01.pdfUploaded byEsteban Floreano
- NESCC 13-045 - Comparison on PWHT Requirements of ASME CodesUploaded bygkvmit
- Environmental Barrier CoatingsUploaded byAnonymous p0mg44x
- Enrico Fermi.. and the Revolutions in Modern PhysicsUploaded byilyesingenieur
- ACR-TrafoUploaded byCarlos Parra
- Transformasi Koordinat RobotUploaded byRama Gian Hendraloka
- Design of Proposed Auditorium-project ReportUploaded byJoe Ps
- Detailed Resume Nilesh MungalaUploaded byNilesh Mungla
- New Control Structure for Feed-Effluent Heat Exchanger-Reactor SystemsUploaded bycymy
- Types of Shellz TheoryUploaded byDhiraj Dhariwal