Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
Transcript of FYP Presentation
Department of Mechanical Engineering
INTRODUCTION The presence of atmospheric boundary layer.
The need to simulate it in the wind tunnel. Background: Objectives: • Turbulence intensity measurements around a building model with a scale of 1:500;
• Spectral analysis of longitudinal velocity fluctuations and compare it with theoretical spectra obtained from literature;
• Determine appropriate length scale to check whether the scale used (1:500) is correct.
EXPERIMENTAL PROCEDURES LITERATURE REVIEW Atmospheric Boundary Layer The power law developed by Davenport (1960): Typical α and δ values for various terrains (Scanlan & Simiu, 1996). Turbulence Intensity Walshe (1973) proposed: Power spectral analysis Power spectral density function: Kaimal spectrum: Eurocode I spectrum: von Kárman spectrum: ,f = z ,f = L Walshe (1973) predicted: Atmospheric Boundary Layer Simulation Methods Simple techniques: unevenly spaced rods with equal diameter (Owen & Zienkiewicz, 1957);
unevenly spaced rods of different diameters;
honeycomb with different length (Kotansky, 1996);
a “conveyor belt” like moving surface (Kiya, Tamura, & Arie, 1980) and;
a curved wire gauze screen (Elder, 1959). More accurate and sophisticated techniques: Counihan (1969):
a tandem of castellated barrier wall upstream of a row of constant
wedge angle quarter-elliptic shaped spires as his vortex generator;
square-mesh grid as a mixing device upstream of a plain barrier wall;
the flat triangular spires as vortex generator;
Downstream the vortex generator, there are roughness elements;
to achieve adiabatic simulated atmospheric boundary layer, wind
speed must be higher than 10m/s. Barrier creates momentum deficit. Square-mesh grid reproduces atmospheric turbulence spectra. Triangular spires overcome the initial lack of turbulence in the simulated boundary layer. Roughness elements represent the actual full-scale ground roughness. At this speed, thermal effects are overcome by the strong turbulence generated. Industrial Wind Tunnel (NUS) => wind speed at 15m/s
Combination of those techniques used by Counihan (1969), Cook (1973) and Irwin (1981) Wind tunnel set-up Velocity profile measurement Power law exponent Focus: Sub-urban terrain (α=0.30) Turbulence intensity measurement Power spectra analysis FFT to longitudinal velocity component
The result is plotted as 'magnitude spectrum':
y-axis => S x n
x-axis => n
Compare with Kaimal, Eurocode and von Karman spectra
Compute the appropriate length scale by fitting the experimental curve onto the theoretical curve
Conclude whether the expected scale (1:500) is appropriate RESULTS & DISCUSSIONS Simulation of sub-urban atmospheric boundary layer Lim's (2009) set-up for urban atmospheric boundary layer α value is supposed to be 0.3933.
However, α value obtained is 0.278. Desired α value = 0.30 By trial and error: Boundary layer thickness = 0.73m Slight gaps and steps between the planks do not affect the α value much.
Roughness cubes nearest the turntable affects α value the most.
Change in the height of the roughness cubes must be done gradually to obtain smoother flow. α value = 0.2933. Boundary layer thickness = 0.73m Turbulence intensity profile Power spectral analysis Comparison of the present experimental spectra with 3 theoretical spectra, namely Kaimal, Eurocode I and von Kárman spectra;
Spectral analysis was done at 3 different heights in the wind tunnel, namely at 0.24 m, 0.48 m and 0.72 m from the centre of turntable, with and without Low-Pass filter;
HP VEE program was used;
1st round of experiment: sampling rate = 1,000 Hz
no. of samples = 15,000
2nd round of experiment: sampling rate = 2,000 Hz
no. of samples = 20,000 Kaimal spectrum ,f = z Power spectral analysis (Kaimal spectrum), scale = 1 Power spectral analysis (Kaimal spectrum), scale = 25 Power spectral analysis (Kaimal spectrum), scale = 500 Eurocode I spectrum ,f = L Walshe (1973) predicted: ,f = L Walshe (1973) predicted: von Kárman spectrum Power spectral analysis (Eurocode spectrum), scale = 1 Power spectral analysis (Eurocode spectrum), scale = 500 Power spectral analysis (von Kárman spectrum), scale = 1 Power spectral analysis (von Kárman spectrum), scale = 500 Present experimental spectra are not comparable with the 3 theoretical spectra.
Possible misinterpretation of what happens to a real time data after going through an FFT in the HP VEE program.
Appropriate length scale for this experiment cannot be evaluated. Similarity of Eurocode and von Kárman spectra Comparison between 1,000 Hz and 2,000 Hz sampling rates 2,000 Hz sampling rate will give more accurate result. 2,000 Hz (red) is constantly higher than 1,000 Hz (blue).
This might be due to more background noise captured.
More accurate does not mean it is always better. Comparison of with and without Low-Pass filter Low-Pass filter is good to minimize or eliminate background noise.
In this experiment, the cut-off frequency is set at half the sampling rate. The red plot (with Low-Pass filter) is peakier than the blue plot (without Low-Pass filter).
The difference is not very significant.
Low-Pass filter is not necessary in this experiment. Comparisons of spectra at 3 different heights Variance of data of the blue plot (0.24m) is the most and it is the least for the green plot (0.72 m).
This is due to higher turbulence intensity at 0.24 m compared to at 0.72 m.
The velocity reading is more varied and unpredictable at 0.24 m. CONCLUSIONS Simulation of atmospheric boundary layer over sub-urban terrain
=> using square turbulence grid, flat triangular spires and roughness elements
Turbulence intensity measurement, compared with Walshe's (1973)
=> trend can be considered similar
=> the present data were reproducable and therefore found to be fairly accurate
Power spectral analysis
=> present experimental spectra are not comparable with 3 theoretical spectra
=> appropriate length scale for this experiment cannot be evaluated
Various comparisons of experimental data were carried out and discussed. RECOMMENDATIONS Try reproducing this year’s wind tunnel set-up => α value of about 0.2933?
Result of FFT to real time data in HP VEE program?
Usage of Low-Pass filter is not necessary. THANK YOU