The Seventh Asia-Pacific Conference on Wind Engineering, November -2, 29, Taipei, Taiwan WIND TNNEL TEST ON THE FLOW AND DISPERSION OF AIRBORNE POLLTANTS IN THE COMPLEX TERRAIN OF COASTAL REGION OF SHIEHE POWER PLANT OF TAIWAN Bao-Shi Shiau and Ben-Jue Tsai 2 Professor, Department of Harbor and River Engineering, National Taiwan Ocean niversity, Keelung 22, Taiwan, and Institute of Physics, Academia Sinica, Taipei 5, Taiwan, bsshiau@gate.sinica.edu.tw 2 Ph.D. student, Department of Harbor and River Engineering, National Taiwan Ocean niversity, Keelung 22, Taiwan, d9252@mail.ntou.edu.tw ABSTRACT In this study, experiments of flow and dispersion of multiple stacks emissions of power plant in the coastal region with complex terrain were conducted in the wind tunnel. Wind tuft observation and concentration tracer measurement were employed to yield the wind flow and dispersion characteristics in the complex terrains. Tuft observations indicate the reversed flow regions. Such regions are the leeside of mountains or hills. We find that the flow turbulence is largely increased when the wind over the mountains. This is shown by the unsteadiness of tufts. Tracer measurement results exhibit two regions of equal concentration contours in the near filed of emission plume, since there are multiple stacks emission. Such pairs of plume finally are emerged in the far downwind distance of sources. The vertical dispersion parameter increases as increasing the downwind distance of the source. Terrain complexity has a favorable effect on the increase of value for the vertical dispersion parameters. The more complexity of the terrain features becomes, the larger the values for vertical dispersion parameters are. KEYWORDS: COMPLEX TERRAIN, PLME, DISPERSION, WIND TFT VISLIZATION Introduction Keelung is a harbor city with about four hundred thousands population. It is also the second largest commercial harbor in Taiwan. The harbor city is located in the mountainous region of northeastern coast of Taiwan. The coastal region is of the complex terrain with varying elevations form m in the harbor city center region to near by mountainous region of 2 m. ShieHe power plant with three high stacks was very closing to the city (see Fig. ). To assess the air pollutant dispersion, it is necessary to study the flow and dispersion characteristics. In the past decades, many numerical and experimental studies on the gas diffusion and dispersion have been studied. Typical numerical study on the gas dispersion in the coastal region, like Jin & Raman (996), they made numerical study on the air pollutant dispersion from an elevated accidental release in the coastal region. The air pollutant numerical simulation of dispersion mostly focused on the open country and flat terrain. Wind tunnel experimental studies such as: Kato and Hanafusa (996) conducted wind tunnel simulation of atmospheric turbulent flow over a flat terrain. Literature on wind tunnel experiments for the air pollutant dispersion over the coastal region with complex terrain like Duijm (996). He proposed an analysis technique to investigate the wind tunnel modeling dispersion of air
The Seventh Asia-Pacific Conference on Wind Engineering, November -2, 29, Taipei, Taiwan pollutants over complex terrain in the Hong Kong Territory. The study was lack of probing thoroughly the terrain effect on the dispersion characteristics. MacDonald et al. (99) succeeded to perform wind tunnel modeling of dispersion in arrays of obstacles, and they compared the wind tunnel results with the measurements of full-scale field trials. To offer the assessment of pollution dispersion for such coastal region with complex terrain, wind tunnel modeling of the pollution dispersion was carried out in this study. The objective is mainly on the investigation for the wind flow variation and plume dispersion characteristics in coastal region with complex terrain. Figure : Location of power plant Experimental Set-up Experiments were conducted in the environmental wind tunnel. The test section of the wind tunnel had the cross section of 2m by.m and 2.6m long. Four spires were placed at the entrance of the test section and roughness elements succeeded to be arranged 9 m long. This arrangement created a fully developed turbulent boundary layer flow, which was used as the approaching flow. The Reynolds number of the simulated approaching flow is about ~ 5. This is sufficient for the turbulent flow similarity requirement of the critical number of ~ (Snyder (9)). An X-type hot-wire incorporating with the TSI IFA-3 constant temperature anemometer was applied to measure the turbulent flow signals. Topographical model was constructed at a scale of to 2. The model was 2 m long and 2 m wide. It was made layer by layer through a number of vertically mounted polystyrene plates of cm thickness around the isopleth of height. The roughness length in the model keeps about 2 mm. As suggested by Snyder (9), it assures flow to avoid laminar sublayer occurring in the model for flow simulation. Exact simulation of three stacks discharge effluents required rigorous similarities between the prototype and model. But it is difficult to meet the rigorous requirements, consideration of relaxed similarities for the momentum length scale, l m and buoyancy length scale, l b between the prototype and model is sufficiently for present study of approximated simulation (Snyder (9)). The heights of power plant three stacks are 2 m, 22 m and 25 m, respectively. All three stacks are with inner diameter of 6. m. The prototype discharging velocities of three stacks are 9.9 m/s,.9 m/s and.6 m/s, respectively. The densimetric Froude numbers for three stacks of discharge effluent are about 3.9, 3.7, and 29.35. For the model simulation experiments, we choose the free stream velocity of 2.56 m/s. For flow direction visualization of the wind over complex terrain in coastal region, tufts were used as indicators of flow direction. Tufts were mounted on many locations of the
The Seventh Asia-Pacific Conference on Wind Engineering, November -2, 29, Taipei, Taiwan terrain model, and each tuft was at a height of 7 cm from the model surface of its location. The height corresponds to m full scale. Visualization of the flow direction was conducted by employing the charged couple detector (CCD). The top view image of wind over the complex terrain model was grabbed at a sampling rate of frames per second. We took an average of sampled frames of images for each run of flow direction visualization with a period of 3 seconds. Methane was used as tracer gas, and it mixed with the standard gas. The mixed gas emitted from two stacks as the discharge of sources in the experiments. The rake of sampling tubes was placed at the sampled position. The tube-rake is composed of ten tubes. A cam mechanism is employed for design of the pump to suck the tracer sample to the airbags through the ten tubes simultaneously and efficiently. Each sample was taken for 5 minutes. The sampled tracer gas in airbag was analyzed with FID (Flame Ionization Detector) to count the methane concentration. Results The flow and dispersion of pollution plume in the complex terrain of coastal region were measured and analyzed. The effect of topography change of the complex terrain on plume dispersion characteristics, such as the plume average height, vertical concentration profiles, and vertical dispersion parameter were discussed. The northeasterly wind is the prevailing wind direction from statistical analysis of the long period of wind record (969-999) of Keelung city meteorological station. And the southeastern and southern regions of the model are areas near the city center. Therefore, we conducted experiments of flow direction visualization with tufts for such three wind direction cases, that is northeasterly, northwesterly, and northerly winds over the complex terrain of model. Approaching flow A turbulent boundary layer flow is generated as approaching flow, with power law type of mean velocity profile, which the power exponent, n is.222. The friction velocity of the simulated boundary layer is about u* =.23 m/s. The roughness length of the simulated boundary layer is about z =.236 cm. This corresponds to the full scale of 3.5 m which lies in the range of to m for urban or complex terrain as indicated by Dyrbye and Hansen (997). Fig. 2 shows the comparison of the present simulated longitudinal turbulent velocity spectrum of approaching flow at the height of Z/Z ref =.76 with the Von Karman spectrum equation. Here Z ref is the boundary layer thickness of approaching flow. Maeda and Makino (9) rewrote the Karman power spectrum equation and it was expressed as following form. ' ux 2u L Su ( n) = () ux 5 2cnL 2 6 [ + ( ) ] In Fig. 2, the spectrum density, S u (n) and frequency, n are normalized, and they are ux denoted by S u ( n) / u' L and nl u x /, respectively. Here u ' denotes the mean square of 2 ux longitudinal velocity fluctuation, u ; c is coefficient of.265; L is the integral length scale of longitudinal velocity in x direction; is the longitudinal mean velocity at the height of Z. The integral length scale is obtained by multiplying the integral time scale, T E with the longitudinal mean velocity,. The integral time scale, T E is computed by integrating the longitudinal velocity autocorrelation coefficient function, R u(τ ). It is found that a satisfactory agreement is achieved for the turbulent approaching flow structure simulation. Visualization of wind flow direction over the complex terrain model
The Seventh Asia-Pacific Conference on Wind Engineering, November -2, 29, Taipei, Taiwan Fig. 3 shows the tuft observation for flow direction of northerly wind over the coastal complex terrain. In the figure the arrows point in the direction of the tufts. The wind flow direction can be observed from the arrows. The tuft indicator observational results for northeasterly and northwesterly winds over the terrain model are obtained. Topography in some regions of terrain model is shown to be more irregular, therefore the wind accordingly is flowing irregularly. Reversed flow regions are seen from the tuft indicators observations for three cases. In general, such regions are the leeside of mountains or hills. Examined from the tuft observational results for three cases, we find that the flow turbulence is largely increased when the wind over the mountains. This is shown by the unsteadiness of tufts. The flow turbulence differs from place to place and with the highest turbulence occurs on the leeside of the hills or mountains as expected. The reversed flow is viewed at some regions which are generally the leeside of main hills or mountains. E+ E+ E- Su(n) u ' Lu x E-2 E-3 =.76 experimental data von Karman E-... nl u x Figure 2: Longitudinal turbulence velocity power spectrum at height of =.76 Figure 3: Tuft observation of wind flow for northerly wind over the complex terrain Flow and turbulence characteristics of wind over complex terrain The mean wind velocity and longitudinal turbulence intensity profiles for different downwind stations of sources for northerly wind over the complex terrain are shown in Fig..
The Seventh Asia-Pacific Conference on Wind Engineering, November -2, 29, Taipei, Taiwan Results indicate that mean wind speed increased (i.e. speed-up phenomenon) at the hill or mountain crest. The turbulence intensity increases as measured locations shift to leeside of hills or mountains. Fig. 5 is the mean wind velocity and longitudinal turbulence intensity profiles for different downwind stations of sources for northeasterly wind over the coastal complex terrain. For northwesterly wind over the coastal complex terrain, the mean velocity profiles and longitudinal turbulence intensity profiles at different downwind stations of sources are shown as Fig. 6. In summary, the speed-up phenomenon occurred when the wind flow over the topography with hill or mountain crests. And the turbulence intensity increases as the topography strongly change..5.5.5.5.75.5.5.75.5.75.5.75.5.75 2 6 2 2 3 6 6 6 6 6 2 6 Iu(%).2.2...6.6...2.2 2 6 2 5 5 2 25 3 35 5 2 6 2 5 5 2 25 3 35 5 Figure : Mean wind velocity and longitudinal turbulence intensity profiles for different downwind stations of sources for northerly wind over the coastal complex terrain.5.5.5.5.5.75.5.25.25.5.75.25.5.75.25.25.5.75.25.5.75 2 6 2 2 3 2 3 2 6 2 3 2 3 2 Iu(%) 6.2.2...6.6...2.2 2 6 2 5 5 2 25 3 35 5 2 6 2 5 5 2 25 3 35 5 Figure 5: Mean wind velocity and longitudinal turbulence intensity profiles for different downwind stations of sources for northeasterly wind over the coastal complex terrain.5.5.5.5.75.25.5.5.75.5.75.75.5.75 2 3 2 6 6 2 6 2 6 2 6 2 6 2 Iu(%) 2 6.2.2...6.6...2.2 2 6 2 5 5 2 25 3 35 5 2 6 2 5 5 2 25 3 35 5 Figure 6: Mean wind velocity and longitudinal turbulence intensity profiles for different downwind stations of sources for northwesterly wind over the coastal complex terrain Concentration distribution Fig. 7 shows the concentration contours of horizontal plane at the height of m for northerly wind. In the figure the tracer concentration C is scaled by the initial concentration C. Results exhibit two regions of equal concentration contours in the near filed of emission plume. It is the interaction of multiple plumes from stacks. Such pairs of plumes finally are emerged in the far downwind distance of sources. This is similar with the observational result of MacDonald et al. (22) who had done water flume study on the buoyant rise in pairs of merging plumes.
The Seventh Asia-Pacific Conference on Wind Engineering, November -2, 29, Taipei, Taiwan The vertical concentration contours along the downwind distance for northerly wind are shown in the Fig.. Two layers of equal concentration are significantly exhibited. For northerly wind, three stacks are aligned in angles with the crosswind. Contini and Robins (2) had also shown similar phenomenon in their experiments on the plumes from two identical sources for different wind directions. Figure 7: Concentration contours of horizontal plane at full scale of m high for the plume with northerly wind flow Figure : Vertical concentration contours along the downwind distance for northerly wind Topographical effect on the vertical dispersion parameter of plume Plume vertical dispersion parameter σ z is adopted to characterize the extent of spread for plume in vertical directions. It is estimated from the measured concentration distributions and here is defined as: 2 2 2 σ z = [( z Cdz) / Cdz) zc ] (2) surface surface where C is the measured concentration; z is the vertical ordinates of Cartesian coordinates. And z c defined as the plume average height, is calculated by, z = ( zcdz) / Cdz) (3) c surface surface
The Seventh Asia-Pacific Conference on Wind Engineering, November -2, 29, Taipei, Taiwan To characterize the elevation variation of hilly terrain, the topography change parameter is proposed. The cumulative root mean square of elevation fluctuation for topography along the downwind distance is designated as, s ( z rms ). Here z rms is the root mean square of topography elevation fluctuation. The slope of s( z rms ) is defined as the topography change parameter, i.e. d[ s( zrms )]/ dx. It is obviously that the topography change parameter for flat terrain is zero. The topography change parameter is applied to analyze the effect of topography of hilly terrain on the plume vertical dispersion. Fig. 9 plots the cumulative root mean square of elevation fluctuation for topography as functions of downwind distances for winds over various complex terrains. The slope of the line shown in Fig. 9 is the topography change parameter. Results indicate that the averaged topography change parameter for northeasterly wind case is the largest among the three cases. And the averaged topography change parameter for northwesterly wind case is the smallest. Fig. shows the plume averaged height, z c as function of downwind distances of source for wind over various complex terrains. As revealed from Fig. 9, the averaged topography change parameter for northwesterly wind case is the smallest among three cases. So the plume averaged height along the downwind distance of source is the smallest among three cases. Figure 9: Topography change parameter along the downwind distance of source for winds over various complex terrains Figure : The plume averaged height as function of downwind distances of source for winds over various complex terrains Fig. plots the vertical dispersion parameter, z σ as functions of downwind distance of sources, x for northerly wind, northeasterly wind, and northwesterly wind over the coastal complex terrains. The plots are scaled with the averaged of three stacks height, h. It is seen that vertical dispersion parameter increases as increasing the downwind distance of the source. The terrain topography feature for northerly wind direction changes dramatically and is the most complexity among three cases. Results also reveal that the complex terrain has a favorable effect on the increase of value for the vertical dispersion parameters. The more complexity of the terrain features becomes, the larger the values for vertical dispersion parameters are.
The Seventh Asia-Pacific Conference on Wind Engineering, November -2, 29, Taipei, Taiwan Figure : Vertical dispersion parameter as functions of downwind distance Conclusions Conclusions are drawn from the analysis of the measurements and shown as follows: () Tuft observations indicate the reversed flow regions. Such regions are the leeside of mountains or hills. We find that the flow turbulence is largely increased when the wind over the mountains. This is shown by the unsteadiness of tufts. (2) Tracer measurement results exhibit two regions of equal concentration contours in the near filed of emission plume, since there are multiple stacks emission. Such pairs of plume finally are emerged in the far downwind distance of sources. (3) The plume averaged height increases with increasing the topography change parameter of the complex terrain. () The vertical dispersion parameter increases as increasing the downwind distance of the source. Terrain complexity has a favorable effect on the increase of the vertical dispersion parameters. The more complexity of the terrain features becomes, the larger the values for vertical dispersion parameters are. References Contini, D., and Robins, A., (2), Experiments on the Rise and Mixing in the Neutral Crossflow of Plumes from Two Identical Sources for Different Wind Directions, Atmospheric Environment, 3, 3573-353. Duijm, N.J., (996), Dispersion over Complex Terrain: Wind-tunnel Modeling and Analysis Techniques, Atmospheric Environment, 3(6), 239-252 Dyrbye, C. and Hansen, S.O. (997), Wind Loads on Structure, p.2, John Wiley & Sons Ltd, New York Jin, H., and Raman S., (996), Dispersion of an Elevated Release in a Coastal Region, Journal of Applied Meteorology, 35, 6-62 Kato, M., and Hanafusa, T., (996), Wind Tunnel Simulation of Atmospheric Turbulent Flow over a Flat Terrain, Atmospheric Environment, 3(6), 253-25 MacDonald, R.W., Griffiths, R.F., and Hall, D.J., (99), A Comparison of Results from Scaled Field and Wind Tunnel Modeling of Dispersion in Arrays of Obstacles, Atmospheric Environment, 32(22), 35-362 MacDonald, R.W., Strom, R.K., and Slawson, P.R., (22), Water Flume Study of the Enhancement of Buoyant Rise in Pairs of Merging Plumes, Atmospheric Environment, 36, 63-65. Maeda, J., and Makino, M., (9), Power Spectra of Longitudinal and Lateral Wind Speed Near the Ground in Strong Winds, Journal of Wind Engineering and Industrial Aerodynamics, 2, 3-. Snyder, W.H., (9), "Guideline for Fluid Modeling of Atmospheric Diffusion," EPA Report 6/--9, SA.