experiment for tall building with various corner shapes. ... Recently, the construction of large structures, such as long span bridges and tall buildings, is on the.
Comparison between the CFD and Wind Tunnel Experiment for Tall Building with Various Corner Shapes Chang-Koon Choi 1), Won-Jin Yu 2), Dae-Kun Kwon 2) 1) Professor, Department of Civil Engineering, KAIST, Korea 2) Graduate Student, Department of Civil Engineering, KAIST, Korea
Abstracts The flow around a rectangular cylinder has been an important subject in building research. There are various kinds of unstable aerodynamic phenomena with regard to a rectangular cylinder. In order to understand the physical mechanism about aerodynamic and aeroelastic instability of a bluff body, it is required to investigate the relations between flow structures and the body motion in detail. The technique of computational fluid dynamics(CFD) is a powerful tool to do it. Thus far a large amount of research has been done to investigate the effect of change of the section shape by several authors. This paper discusses the comparison between the CFD and wind tunnel experiment for tall building with various corner shapes. First we carried out wind tunnel test for measurement of aerodynamic response, where two–dimensional square section cylinder with various corner shapes in uniform flow is selected as an experimental model. And then, we perform the numerical analysis using the technique of CFD for comparing between test and analysis results.
Introduction Recently, the construction of large structures, such as long span bridges and tall buildings, is on the increase that relatively light and flexible. These structures tend to have low levels of intrinsic damping and can be especially prone to high resonance when subject to wind loading, like vortex-induced vibration phenomena, which may cause occupant discomfort or hinder the function of motion sensitive equipment. As such, the suppression of these vibrations has become an important design consideration in recent years. The recent advocation of the structural control technology has brought about quite a few practical control devices, which have been successfully applied onto some tall buildings. This paper discusses the performance for the reduction of wind loading on a tall building due to the modification of corner shape, and the CFD method is used for comparing between test and analysis results.
Experimental Procedure Experimental equipment The experiments were performed in an Eiffel type wind tunnel of KAIST that working section is 1m 1m and the length is 4m. We measured aerodynamic response, namely, displacements caused by vortexinduced vibration and galloping using two–dimensional model in the smooth flow for checking the aerodynamic stability with various corner shapes. In this study, five types of corner shapes were employed(fig. 1), a square cylinder(B/D=1.0) with sharp, rounded, chamfered, half-squared and squared corner shapes. Base model is cross-shaped section(square cut), and other models are described as attaching various corner shapes to this. To perform the tests at the same condition, the attachments are used as light as possible. Considering blockage effects, we determined the size of the model is basically 50mm and applied the same 1/10D(or B) corner size to all cases except half-square corner(1/20D). The test models have very low logarithmic damping ratio of 0.0015 and 5.0 Hz natural frequency measured by position sensor and signal analyzer. The aerodynamic responses were tested under various angle of attack, from 0 to 45 increasing 5 degrees, and measured in the range of 0~5.0m/s wind velocity increasing 0.2 m/s.
Wind Tunnel Test Results of various corner shapes Fig. 2 shows experimental results for the various corner shapes ; shortly, N,R,T,H, and S type respectively. There is no torsional vibration mode in these models–only occurs across-wind(vertical) vibration mode. No corner shapes Within the 15 attack angle(shortly, ), that is the critical occurrence point of galloping phenomenon as we know, occurs in =0,5,10(Fig.2(a)). As attack angle increases, the onset velocity of the galloping also increases from 1.2 m/s(0) to 1.4 m/s(5), 3.2 m/s(10). When is more than 15, there is no galloping phenomenon, but VIO occurs, having somewhat large amplitudes more than =20. The onset velocity of vortex induced oscillation(shortly, VIO) is 1.6 m/s(reduced velocity(U/f nD) is 6.4) for all attack angles except 15, 20(1.4 m/s, U/fnD=5.6). Modified corner shapes The general trends of the results are similar to each other even though there are slight differences in the onset velocity and the amplitude of the vortex induced oscillation and the galloping. In the case of modified corner shapes, the galloping only occurs in the range of 0, 5 except H, S types. Especially, S type is good improvement for the galloping with any attack angles except =0. There is peculiar phenomenon at 5 of H-type, occurred two vortex induced oscillation in the low and high wind velocities(Fig.2(f)). The onset velocity of galloping is all about 1.4m/s ; U/f nD=5.6. When an attack angle
increases larger than 15, strong vortex induced oscillation occurs at the same onset velocity of the galloping, but larger VIO amplitudes than no corner shapes. VIO occurs at small velocity range, and has sharp peak as attack angles increase. This trend was inevitable even though various corner shapes were used. From a viewpoint of stability of the structures, modified corner shapes are good improvement for the aerodynamic behavior. On the contrary, the serviceability and fatigue failure isn’t good due to strong vortex induced vibration.
NUMERICAL ANALYSIS The numerical analyses of the effect of corner shapes on wind flow have been performed. The finite element method is used to describe the geometry around corners and analysis the behavior of wind. The details of the finite element formulation can be found from a reference(Choi 1997). The wind is assumed as incompressible viscous flow without heat exchange. The governing equations consist of mass conservation equation and momentum conservation equation. The penalty function method, which can reduce the number of independent variables, is adopted for the purpose of computational efficiency and the selected reduced integration is carried out for the convection and pressure terms to reserve the stability of solution. The purposes of numerical analysis are to observe the wind, which flows around various corner shapes and to find the reason of the structural vibration. To view the wind flow around corner shapes, the five models are considered as shown in Fig.3. In the steady state, the symmetric wind flow is generated and the half models are used to decrease the computational cost. The problem statement is given in Fig.3. The inlet velocity of wind is 1 m/s, the viscosity and the density of wind are 1.8 kg/(ms) and 1.2 kg/m3. The time step for time history analysis is selected as 0.0001 second to achieve non-oscillation solution. In Fig.4, streamlines and pressure contours at the steady state are obtained. In the case of none corner shape, the serious separation of wind flow occurred at the front corner as shown in Fig.4(a). Furthermore, the size of wake at the rear side of the structure is larger than those of other corner shapes. Thus, when the original structure(without corner shape) is constructed, the vibration of the structure may happen easily. In other cases, the separation at the front corner has not developed, although the flow past sharp corner should be separated. The reason is that the size of elements is not as small as the separation can be developed or the size of separation is smaller than the size of element. The size of wake is smallest in S–type. This means that the reattachment of separated flow is better than other cases.
Conclusion This paper discusses the performance for the reduction of wind loading on a tall building due to the modification of corner shape, and the CFD methods is used for comparing between test and analysis results. In the results of wind tunnel test, general trends are similar to each other even though there are
slight differences in the onset velocity and the amplitude of the vortex induced oscillation and the galloping. Modified corner shapes have relatively good results for the galloping, however VIO has sharp peak as attack angles increase. This trend was inevitable even though various corner shapes were used. The CFD method is performed to investigate flows around various corner shapes, and to find the reason of the structural vibration. In the case of none corner shape, the serious separation of wind flow occurred at the front corner. Furthermore, the size of wake at the rear side of the structure is larger than those of other corner shapes. In other cases, the separation at the front corner has not developed, although the flow past sharp corner should be separated. The size of wake is smallest in S–type. This means that the reattachment of separated flow is better than other cases.
Reference C.K. Choi, W.J. Yu, “Variable Node Element for the Adaptive FE Analysis for Incompressible Viscous Flow”, Seventh International Conference on Computing in Civil and Building Engineering, 19–21, Aug. 1997, Seoul, Korea H. Kawai, “Effect of Angle of Attack on Vortex Induced Vibration and Galloping of Tall Buildings in Smooth and Turbulent Boundary Layer Flows”, The Third Asia–Pacific Symposium on Wind Engineering, 13–15 December, 1993, Hong Kong T. Tamura, T. Miyagi, “The Effect of Turbulence on aerodynamic forces on a square cylinder with various corner shapes”, The Fourth Asia–Pacific Symposium on Wind Engineering, 14–16 July, 1997, Australia
Acknowledgment The authors are pleased to acknowledge support from the advanced STructure RESearch Station (STRESS).
B=D
Wind
D=50mm
sharp corner(N type)
rounded corner(R type)
chamfered corner(T type)
half-square corner(H type)
1/10D 1/10D
square corner(S type)
Figure 1
Types of corner shapes
w / o C . S .
w / R o u n d C . S .
2 A ( m m )
2 A ( m m )
7 0
7 0 N 0
R 0
N 5 6 0
R 5
N 1 0
6 0
N 1 5 5 0
R 2 0
N 2 5
5 0
N 3 0
R 2 5 R 3 0
N 3 5 4 0
R 1 0 R 1 5
N 2 0
R 3 5
N 4 0
4 0
N 4 5
R 4 0 R 4 5
3 0
3 0
2 0
2 0
1 0
1 0
0
0
0
1
2
3
4
5
0
1
2
3
4
V ( m / s )
(a) N-type
(b) R-type
w / T r i a n g u l a r C . S .
w / H a l f C . S .
2 A ( m m )
2 A ( m m )
7 0
7 0 T 0
H 0
T 5 6 0
H 5
T 1 0
6 0
T 1 5
H 2 0
T 2 5
5 0
T 3 0
H 2 5 H 3 0
T 3 5 4 0
H 1 0 H 1 5
T 2 0 5 0
5
V ( m / s )
H 3 5
T 4 0
4 0
T 4 5
H 4 0 H 4 5
3 0
3 0
2 0
2 0
1 0
1 0
0
0
0
1
2
3
4
5
0
1
2
3
4
V ( m / s )
5
V ( m / s )
(c) T-type
(d) H-type
w / S q u a r e C . S .
w / H a l f s q u a r e C . S . a t 5 d e g r e e s
2 A ( m m )
2 A ( m m ) m
7 0
7 0
S 0 S 5 6 0
S 1 0
6 0
S 1 5 S 2 0 5 0
S 2 5
5 0
S 3 0 S 3 5 4 0
S 4 0
4 0
S 4 5
3 0
3 0
2 0
2 0
1 0
1 0
0 0
0
1
2
3
4
5
0
1
2
3
4
(e) S-type
(f) H-type at 5 Figure 2
5
6
V ( m / s ) m
V ( m / s )
Results of various corner shapes
v=0 u=1 v=0 0.25m v=0 1m Figure 3
The problem statement
(a) streamline(N–type)
(b) pressure contour(N–type)
(c) streamline(R–type)
(d) pressure contour(R–type)
(e) streamline(T–type)
(f) pressure contour(T–type)
(g) streamline(H–type)
(h) pressure contour(H–type)
(i) streamline(S–type)
(j) pressure contour(S–type)
Figure 4
Streamlines and pressure contours for various corner shapes
