Escolar Documentos
Profissional Documentos
Cultura Documentos
www.elsevier.com/locate/engstruct
Received 17 April 2002; received in revised form 29 January 2003; accepted 3 February 2003
Abstract
The box system that consists of reinforced concrete walls and slabs is adopted for many high-rise apartment buildings recently
constructed. In an apartment building, shear walls may have one or more openings for functional reasons. Some researches on the
analysis of shear wall with openings were performed, but some restrictions prevent those research results from being applied to
the practical analysis and design procedure. It is necessary to use fine mesh finite element models for an accurate analysis of a
box system structure with openings. But it would take a significant amount of computational time and memory if the entire building
structure were subdivided into a finer mesh. An efficient analysis method that can be used regardless of the number, size and
location of openings is proposed in this study. The analysis method uses super elements developed using the matrix condensation
technique. Static and dynamic analyses of example structures having various types of openings were performed to verify the
efficiency and accuracy of the proposed method. The analyses of example structures demonstrated that the proposed method can
be used for the analysis of a shear wall structure with openings. Outstanding accuracy in analysis could be achieved with drastically
reduced computational time and memory.
2003 Elsevier Science Ltd. All rights reserved.
Keywords: Shear wall with openings; Super element; Matrix condensation; Fictitious stiff beam
0141-0296/03/$ - see front matter 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0141-0296(03)00041-5
982 H.-S. Kim, D.-G. Lee / Engineering Structures 25 (2003) 981–991
K0 ⫽ t 冕冕1
⫺1
1
⫺1
BTEB兩J兩dxdh (1)
K1 ⫽ t 冕冕b2
b1
a2
a1
BTEB兩J兩dxdh (2)
almost the same shear stress distribution to the exact sol- For an accurate analysis of shear wall with openings,
ution, even if only one element per story was used. it is necessary to use a fined mesh model, but if the entire
Fig. 6. Comparison of shear stress distribution from Q4 elements and Lee elements(A–A’).
984 H.-S. Kim, D.-G. Lee / Engineering Structures 25 (2003) 981–991
Fig. 7. Procedure composing stiffness matrix of plane stress element with openings.
building structure were subdivided into a finer mesh with number of nodes. Therefore, if all of the DOFs except
a large number of elements, it would cost a large amount the four nodes at the corners of a wall in a story are
of computational time and memory. And also, it is eliminated using the matrix condensation technique as
inconvenient to use inconsistent grids for the shear wall shown in Fig. 8(d), the number of nodes in a super
and frame in modeling the structure. Therefore, it would element is identical to that of a conventional plane stress
be very useful if the analysis of a shear wall structure element. This procedure is represented using the matrix
with openings could be performed without such prob- form in Eqs (3) and (4). The static equilibrium equation
lems. for a super element can be rearranged by separating the
DOFs for the corners from those for the inner area of a
3.1. Development of super elements for shear wall super element as follows;
冋 册冋 册 冋 册
with openings
Sii Sic Di Ai
⫽ (3)
Some of the high-rise apartment buildings adopt the Sci Scc Dc Ac
box system that has repeated arrangement of residential
units with the same architectural plan on each floor. Fig. where subscripts i and c represent the inner area and
8(a) shows a three story shear wall structure with win- corners of a super element respectively. Eliminating the
dow type openings and Fig. 8(b) shows a fine mesh DOFs for the inner area by matrix condensation, the
model for this structure using many plane stress static equilibrium equation can be represented for the
elements. This fine mesh model can be separated into DOFs only at the corners only as follows;
three substructures having the same configuration in ∗
[Scc]{Dc} ⫽ {Ac∗} (4)
each story as shown in Fig. 8(c). Super elements illus-
trated in Fig. 8(d) could be generated if all of the DOFs where, and
except those for the node at the corners of the shear wall
in a story are eliminated using the matrix condensation {A∗c } ⫽ {Ac}⫺[Sci][Sii]⫺1{Ai} and [Scc
∗
] ⫽ [Scc]
technique,. Then, the shear wall structure is constructed ⫺[Sci][Sii]⫺1[Sic]
by joining the corner nodes of a generated super element
as shown in Fig. 8(e). The matrix S∗cc is the stiffness matrix for the super
For the efficiency in the analysis, the stiffness matrix element having nodes only at corners of a shear wall and
of a super element needs to be developed with the least Eq. (4) is the equilibrium equation for a super element.
Fig. 8. Modeling procedure of a shear wall with openings by using super elements.
H.-S. Kim, D.-G. Lee / Engineering Structures 25 (2003) 981–991 985
Fig. 9. Use of fictitious beams for shear wall with window type opening.
986 H.-S. Kim, D.-G. Lee / Engineering Structures 25 (2003) 981–991
Fig. 12. Deformed shape of a shear wall structure with window type openings.
fictitious beam should be infinitely stiff as shown in Fig. fictitious beams are added to the boundaries of a shear
11. However, if the fictitious beam is too stiff, it may wall as shown in Fig. 13(b) and then all of the DOFs in
cause numerical errors in the matrix condensation pro- the shear wall except those at the ends of the fictitious
cedure. On the contrary, the compatibility condition at beams are eliminated as illustrated in Fig. 13(c). At last,
the boundaries is not satisfied if the fictitious beam is the flexural stiffness of the fictitious beams is subtracted
too flexible in comparison to the shear wall. A rule of to eliminate the surplus stiffness introduced by the fic-
thumb to determine the stiffness of the fictitious beams titious beams as shown in Fig. 13(d). Models in Fig. 14
is to use a stiff beam having the same thickness with the were used to verify the efficiency of the fictitious beams
corresponding shear wall and the depth assumed to be for the model with door type openings. It can be easily
the story height. The deformed shape of the upper and noticed that the proposed analysis method also satisfies
lower sides of a super element generated using the the compatibility condition at boundaries of super
matrix condensation technique is governed by the elements by introducing the fictitious beams.
behavior of the fictitious beams. This effect is main-
tained, even if the surplus stiffness introduced by the
fictitious beams is eliminated. 5. Analysis of example structures
Fig. 12 illustrates the effects of the fictitious beams
used for the development of the super element. The fine
Static analyses of four example structures were perfor-
mesh model is assumed to provide the most accurate
med to verify the efficiency and accuracy of the pro-
result as shown in Fig. 12(a) and the deformed shape of
posed analysis method. Example structures with various
the model using super elements are shown in Fig. 12(b)
type, number, location and size of openings were used
and (c) respectively. Fig. 12(b) illustrates that the
in the analyses. A computer code MIDAS [12] was used
deformed shape of the model using the super element
for the analysis of example structures.
derived without fictitious beams does not satisfy the
compatibility condition at the interfaces. It can be easily
noticed that the deformed shape of the model using fic- 5.1. Models used in the analyses
titious beams is almost identical to that of the fine
mesh model. Static analyses of various models as shown in Fig. 15
were performed. Model A is a fine mesh model subdiv-
4.2. Use of fictitious beams for analysis of shear wall ided into a large number of finite elements as shown in
with door type opening Fig. 15(a). Analysis results of this model are assumed
to be the most accurate. The effect of openings was
The use of fictitious beams in the case of a shear wall ignored in Model B as shown in Fig. 15(b) and Model
with a door type opening is similar to that of shear wall C, that is used by most of the engineers in practice,
with window type opening as shown in Fig. 13. Two replaces the link beam above the opening with an equiv-
Fig. 13. Use of fictitious beams for shear wall with door type opening.
H.-S. Kim, D.-G. Lee / Engineering Structures 25 (2003) 981–991 987
Fig. 14. Deformed shape of a shear wall with door type openings.
alent stiff beam as shown in Fig. 15(c). Model D and height of the opening and b is that of the shear wall
employs the super element proposed in this study while in each story. Lateral displacements of each model are
Model E is using the shear wall element with opening shown in Fig. 16(b), (c) and (d). When a / b is 0.1, lateral
proposed by Choi and Bang [2]. displacements were similar in all cases since the effects
of openings are not significant. If a / b was increased to
5.2. Shear wall with window type openings 0.3, Model C resulted in somewhat larger displacements
while the other models provided similar lateral displace-
The example structure 1 adopted from the work by ments. However, the difference in the lateral displace-
Choi and Bang has window type openings as shown in ments was noticeable in the case of a / b = 0.5 because
Fig. 16(a). Analysis was performed for the cases of a / the effect of the opening became more significant. As
b = 0.1, a / b = 0.3 and a / b = 0.5 where a is the width expected, the lateral displacement of Model B was
smallest because the openings were not considered. Lat- respectively while the height of the opening was fixed
eral displacements of Model C that uses equivalent as 210 cm, example structure experienced lateral dis-
beams were larger than those of other models. Although placements as shown in Fig. 19. As the opening become
equivalent beams were assigned with the stiffness equiv- larger, the lateral displacements of Model C using equiv-
alent to that of link beams in the case of Model C, the alent beams were larger than Model A while the pro-
lateral displacements were significantly larger than those posed Model D could provide an almost identical result
of Model A. to that of Model A in all cases. The results from Model
This overestimation of displacements was introduced E for example structures, except the first one, were not
by the additional deformation in the upper part of shear provided because they were not available since it is not
walls that is restrained by link beams in the fine mesh applicable to such cases.
Model A. In the case of Model E, which used a plane
stress element with an opening, lateral displacements 5.4. Shear wall with openings connected to frame
were smaller than those of Model A. Even when a / b
was 0.5, Model D proposed in this study provided almost Framed structures with shear wall core are frequently
the same results to those of Model A that is expected to adopted as the structural system for high-rise building
provide the most accurate results. The Von-Mises stress structures. Therefore, analyses of example structure 3
distribution of Models A, B and D are shown in Fig. consisting of shear wall with door type openings that
17(a), (b) and (c) respectively for the case of a / b equals was connected to a frame as shown in Fig. 20 were per-
to 0.33. Model D provided stress concentration at cor- formed. The results from Model B are not provided here-
ners openings with accuracy comparable to that of after because they are not so meaningful. As shown in
Model A. Fig. 20(b), the lateral displacements of the proposed
Model D were very close to those of Model A while
5.3. Shear wall with door type openings Model C overestimated lateral displacement signifi-
cantly.
Lateral displacements of example structure 2 with the
height of the openings 66, 77 and 88% of the story
5.5. Shear wall with window and door type openings
height are shown in Fig. 18(b), (c) and (d) respectively
while the width of the opening was fixed. As could be
expected, Model B without considering openings Example structure 4 has window and door type open-
resulted in identical lateral displacements in all cases and ings with a frame on one side as illustrated in Fig. 21. As
the error in the lateral displacement increases as the could be expected, the proposed method provided almost
opening was higher. The lateral displacements of Model identical result to that of Model A. In a similar trend
C using equivalent beams were larger than those of with example structure 3, Model C overestimated the
Model A when the opening was small. But if the opening displacements significantly.
was higher, the lateral displacements of Model C were
getting closer to those of Model A. The proposed Model 5.6. Dynamic analysis
D could provide almost identical results to those of
Model A regardless of the height of the opening. Eigenvalue analysis and time history analysis were
When the width of the opening in the example struc- performed to verify the accuracy in prediction of seismic
ture was 11, 33 and 55% of that of the shear wall response of building structures. The ground acceleration
time history of the El Centro earthquake (NS, 1940) was
used as the input ground motion. The computational time
and the number of DOFs used for the analyses of
example structures were compared to verify the
efficiency of the proposed method.
Example structure 2 with door type openings with the
size of 90 cm × 210 cm is shown in Fig. 22(a). The
results of eigenvalue analysis and time history analysis
indicate that the analysis results of Model A and D are
almost identical while Model C provided longer periods
of vibration resulting in somewhat deviated displace-
ment time history.
Similar trend was observed in the periods of vibration
and displacement time histories of the example structure
Fig. 17. Von-Mises stress distribution in a shear wall structure with 4 as shown in Fig. 23. The computational time and the
window type openings (a / b = 0.33). number of DOFs used for the analyses are compared in
H.-S. Kim, D.-G. Lee / Engineering Structures 25 (2003) 981–991 989
Fig. 18. Lateral displacements of shear wall structure with door type openings for different opening height.
Fig. 19. Lateral displacements of shear wall structure with door type openings for different opening width.
Fig. 20. Shear wall with door type openings connected to a frame.
Fig. 22. Natural period and displacement time history of a shear wall structure with door type openings.
Fig. 23. Natural period and displacement time history of a frame structure with shear wall with window door type openings.
Table 1. The analysis for Model D required less than 1. The lateral displacements of the model using a plane
5% the computational time used for Model A. stress element with opening proposed by Choi and
Bang were similar to those of a fine mesh model when
the opening was small, but the error in lateral dis-
6. Conclusions placement increased as the opening became larger.
2. The lateral displacements of the model using equival-
An efficient analysis method for the analysis of shear ent beams to represent the link beam above the open-
wall with openings was proposed in this study using ings were similar to those of the fine mesh model
super elements derived by introducing fictitious beams. when the depth of the link beams was small. But the
The accuracy and the efficiency of the proposed method lateral stiffness of the structure was underestimated
were investigated by performing analysis of various as the depth of the link became larger resulting in
example structures. The major observations and findings larger lateral displacements compared to the fine
are summarized as follows: mesh model.
Table 1
Number of DOFs and computation time for analysis
Models Number of DOFs Assembly M & K Solution of equation Eigenvalue analysis Time history analysis Total