06.07.2008 Zimmermann-Becker GmbH Beratende Ingenieure TGA Heilbronn Hamburg Leipzig 06.07.2008 1 Dipl.-Ing. Frank Zimmermann, Heilbronn, D und Dipl.-Ing. Frank Kanters, Heerlen, NL Contents 1. Foreword 2. Introduction 3. Boundary conditions 4. Slender tower 4.1 Wind-tunnel measurements 4.2 CFD calculations 4.3 Comparison with DIN 1055-4 5. Cube-shaped buildings 5.1 Wind-tunnel measurements 5.2 CFD calculations 5.3 Comparison with DIN 1055-4 6. Rectangular buildings H = 0.5 B 6.1 Wind-tunnel measurements 6.2 CFD calculations 6.3 Comparison with DIN 1055-4 7. Summary of the results and conclusions References Validation of PHOENICS-2008 for the determination of wind pressures on buildings, by Frank Zimmermann, Heilbronn, D and Dipl.-Ing. Frank Kanters, Heerlen, NL; an English-language version by Brian Spalding. Validation of the CFD program PHOENICS 2008 for the determination of wind loads on buildings
39
Embed
06.07.2008Zimmermann-Becker GmbH Beratende Ingenieure TGA Heilbronn Hamburg Leipzig 06.07.20081 Dipl.-Ing. Frank Zimmermann, Heilbronn, D und Dipl.-Ing.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Validation of the CFD program PHOENICS 2008for the determination of wind loads on buildings
1.0 Foreword
DIN 1055-4 03/2005 permits the determination of wind loads on buildings of rectangular-plan shapewith sufficient accuracy. It gives however no guidance regarding buildings of more-complex shape.
Since knowledge of the wind loads is necessary for proper design, especially for tall buildings, recourse therefore has in the past had to be made to measurements in wind tunnels, in which the non-uniformity of wind profile and the presence of surrounding buildigs should be represented.
These requirements, together with the need for a large large number of measurement locations, make the method costly.
For these and other reasons, better accuracy then +/- 15 % is not to be expected.
The development of Computational Fluids Dynamics, and of sufficiently powerful computer hardware, now permit wind loads on buildings to be computed without great expense. The use of the large-eddyturbulence model together with parallel processing leads to especially good results.
In the present report, the results of calculations made by means of the PHOENICS-2008 CFD-softwarepackage are compared with experimental data and with the prescriptions of DIN 1055-4, and areshown thereby to be sufficiently accurate.
Such calculations can be made for buildings of any shape.
Dipl.-Ing. Frank Zimmermann, Heilbronn, D und Dipl.-Ing. Frank Kanters, Heerlen, NL
Validation of the CFD program PHOENICS 2008for the determination of wind loads on buildings
2. Introduction
A validation exercise requires knowledge of varied and reliable experimental data as standards of comparison.In the present investigation, the following sources were used:
1. NISTIR 6371, United States Department of Commerce, Technology Administration, National Institute of Standards and Technology Building and Fire Research Laboratory, Gaithersburg, [2].
2. Numerical Prediction of Wind Loading on Buildings and Structures, The working group for numerical prediction of wind loading on buildings and structures,
Subcommitee for wind engineering data unit for structural design, Archtectural Institute of Japan, 1998, [3].
3. DIN 1055-4, 03/2005, Einwirkungen auf Tragwerke, Teil 4: Windlasten, [4].
In [2] the numerical determination of wind loads by means of Large-Eddy Turbulence simulation iscompared with wind-tunnel measurements and with the large-eddy caculations by Shah und Ferziger [5]. Two building models were used: a tower of 10 x 10 x 80 m and a cube of 30 x 30 x 30 m.
A further comparison was made with the investigation of the AIJ – Report, Architectural Institute of Japan, [3]for a rectangular-shaped building of 30 x 30 x 15 m which compared CFD calculations with wind-tunnel measurements.
Dipl.-Ing. Frank Zimmermann, Heilbronn, D und Dipl.-Ing. Frank Kanters, Heerlen, NL
Validation of the CFD program PHOENICS 2008for the determination of wind loads on buildings
3. Boundary conditions
The large-eddy model was chosen because of its advantages [6] . Calculations with the k-epsilon model lead to more or less the same conclusions provided that one reaches convergence. The LE model is specially suitable for fine grids.
The LE – Model proved to be more robust, and to converge faster than K-epsilon.
The choice of grid is of great importance in wind simulations.
PHOENICS uses a basically Cartesian grid; and it has two especially valuable features for providing fine grids where they are needed and for handling surfaces which cut the grid planes obliquely. They are,respectively:
FINE-GRID EMBEDDING and PARSOL.
Dipl.-Ing. Frank Zimmermann, Heilbronn, D und Dipl.-Ing. Frank Kanters, Heerlen, NL
Comparison of the results of calculations made by PHOENICS With wind-tunnel measurements according to NIST 6371,for both shear- and uniform-flow wind profiles.
Validation of the CFD program PHOENICS 2008for the determination of wind loads on buildings
Dipl.-Ing. Frank Zimmermann, Heilbronn, D und Dipl.-Ing. Frank Kanters, Heerlen, NL
6.0 Summary of the results; and conclusions
There exists a multitude of CFD programs; and within them are to be found many different turbulence models Therefore to prove the general applicability of CFD for the prediction of wind forces on buildings will notbe possible until there exists a standardised validation process of the kind which VDI 6020, for example,provides for the simulation of buildings and apparatus.
Nevertheless it would be unwise to disregard a calculation method which has delivered excellent and verified results in an enormous number and variety of applications.
Indeed, publications from the USA, Japan and Switzerland have shown that CFD calculations of the wind loads on buildings have given acceptable results.
In the absence of the above-mentioned standardised validation procedure, there remains the possibility ofvalidating a specific CFD program under precisely-specified boundary conditions. Such a validation mustdemonstrate conformity with standards and with wind-tunnel measurements.
These measurements, of course, stand equally in need of validation.
Validation of the CFD program PHOENICS 2008for the determination of wind loads on buildings
Dipl.-Ing. Frank Zimmermann, Heilbronn, D und Dipl.-Ing. Frank Kanters, Heerlen, NL
6.0 Summary of the results; and conclusions (continued)
For buildings of complex form, CFD shows the critical locations instantly and without additional expense. In wind-tunnel investigations one must know, or guess, these locations beforehand so as to set a limit tothe number of measurement positions.
The picture on pages 18 and 19 show clearly that measuring positions in the guitter region were either omitted, or were in the wrong positions; whereas the CFFD calculations according to [2] bring to light the problems presend by sharp edges.
The absence from DIN 1055-4 of the low-pressure zone in the gutter region is probably to be explained by the desire of the investigator to simplify the measuring procedure.
Validation of the CFD program PHOENICS 2008for the determination of wind loads on buildings
Dipl.-Ing. Frank Zimmermann, Heilbronn, D und Dipl.-Ing. Frank Kanters, Heerlen, NL
6.0 Summary of the results; and conclusions (concluded)
This means that the use of PHOENICS for the present application, namely the Hamburg Central PlazCentral Plaza, Hamburg, guarantees conformity with DIN 1055-4 März 2005, if the following conditionsare observed:
1. Sufficiently large model with blockage < 5 %.2. Grid size on the wall surface <= 1m, by using FINE GRID EMBEDDING3. Representation of slanting surfaces by way of PARSOL4. Use of the Large-Eddy-Simulation turbulence model5. Calculation with varying wind direction (in steps of 15 degrees)6. Calculation of the wind velocity profile of ground-level category IV for plane surfaces and/or7. In order to determine the influence of surrounding buildings, calculation with
a wind-velocity profile of ground-level category II
Validation of the CFD program PHOENICS 2008for the determination of wind loads on buildings
References:
[1] Kiefer, H., Windlasten an quaderförmigen Gebäuden in bebauten Gebieten, Dissertation 2003, Universität Karlsruhe
[2] NISTIR 6371, United States Department of Commerce, Technology Administration, National Institute of Standards and Technology Erstellt vom Building and Fire Research Laboratory, Gaithersburg [2]
[3] Numerical Prediction of Wind Loading on Buildings and Structures, The working group for numerical prediction of wind loading on buildings and structures, Subcommitee for wind engineering data unit for structural design, Archtectural Institute of Japan, 1998 [3]
[4] DIN 1055-4, Einwirkungen auf Tragwerke, Teil 4: Windlasten, März 2005
[5] Shah, Kishan, B. and Joel H. Ferziger, A Fluid Mechanicians View of Wind Engineering: Large Eddy Simulation of Flow Past a Cubic Obstacle, J. Wind Engineering and Industrial Aerodynamics, Vols. 67 & 88, pp 211 – 224,1997
[6] Gary Easom, Improved Turbulence Models for Computitional Wind Engineering, Dissertation 2000, University of Nottingham
Dipl.-Ing. Frank Zimmermann, Heilbronn, D und Dipl.-Ing. Frank Kanters, Heerlen, NL