1、4755 A New Local Ventilation System Using a Vortex Flow Generated with a Finned Rotating Annular Disk Sang-Min Lee Jin-Won Lee ABSTRACT In a local exhaust ventilation system, the exhaust air velocity decreases very rapidly with distance from the exhaust inlet. One possible way of maintaining a high
2、capture velocity farfrom the exhaust inlet is the use of vortexflow, usually called a “vortex ventilation system,” which has a rotating annular disk installed in the exhaust inlet. Through numerical and experimental analysis, an optimal shape of the disk is devel- oped and its performance is represe
3、nted based on the operat- ingparameter 0, which is the ratio of the displacementflow rate by thefins on the disk to the exhaustflow rate. Compared with the simple exhaust, the vortex ventilation system gives more than twice as large a capture depth and more than ten times as large a capture volume.
4、The main mechanism for the improved performance is the displacement flow propagating radially around the exhaust inlet and thus inducing a second- ary airflow under the hood. INTRODUCTION The main purpose of local exhaust ventilation (LEV) is to reduce or avoid exposure of workers to contaminants. F
5、or good performance, the air velocity should be high enough to carry the contaminants toward the exhaust hood in the pres- ence of disturbing side flows or inertial settling. In a simple exhaust system the air velocity decreases very rapidly with distance from the exhaust inlet, approximately to the
6、 second power of the distance, so the capture velocity at a distance of one diameter of the exhaust inlet may get reduced to less than 10% of that at the exhaust inlet. For this reason, many local exhaust hoods are placed close to the operating zone, but this close placement prevents a workers free
7、movement or a flex- ible layout of the equipment. This limitation restricts the use of the local ventilation systems (Goodfellow and Thti 2001). Several ideas have been investigated to overcome the limitations of the traditional LEV systems, a few typical exam- ples of which are REEXS, ATLEVS, and V
8、EER. Though there are other kinds of improved designs, such as the air curtain, using combined exhaust hoods and supply inlets, and the push- pull ventilation system, these will not be considered here because they require a substantial amount of extra space for the installation of extra devices. REE
9、XS (reinforced exhaust systems), developed by Aaberg (1968,1977), uses a radial jet around the exhaust inlet, which causes an axial flow toward the exhaust inlet, resulting in an extension of the capture region. Pedersen and Nielsen (1991) figured out, through a series of experiments on the performa
10、nce characteristics of REEXS, that the ratio of the momentum fluxes of the jet and the exhaust flow is the most important parameter. Hunt and Ingham (1996) analyzed the two-dimensional and three-dimensional flow patterns in the REEXS using mathematical models. Recently Guber (2002) and Gubler and Mo
11、ser (2000) developed a prototype system using an extensive CFD study. The REEXS consumes quite a large amount of extra flow for the jet, about 70% of the exhaust flow, and a high pressure loss is caused in the generation of the jet and also a strong noise is generated. The viscous induction mechanis
12、m is not energy-eficient, and the need for an addi- tional duct for the jet flow is considered a shortcoming from a practical point of view. Another means of improving the performance of the LEV system is the use of a strong vortex flow such as the tomado found in nature. The low-pressure zone gener
13、ated by the centrifugal force along the vortex core in a strong vortex flow can be more effective to induce an axial flow than that gener- Sang-Min Lee is a doctoral student and Jin-Won Lee is a professor in the Division of Mechanical and Industrial Engineering, Pohang Univer- sity of Science and Te
14、chnology (POSTECH), Pohang, South Korea. 02005 ASHRAE. 149 atea by a viscous induction as in the REEXS. Iwo typical examples of using a vortex flow are the ATLEVS (Artificial Tornado Local Exhaust Ventilation System) by Yarnaguchi et al. (1991) and the VEER, using a swirling flow, by Spotar et al. (
15、1994). Yarnaguichi et al. (1991) made an artificial tornado-like flow structure using four separate tangential jets issuing normal to each other from four parallel vertical tubes. In this arrangement, a strong vortex column of square cross section can be generated, but additional tube structures hav
16、e to be installed in the space, which can be a great shortcoming in practical applications. Spotar et al. (1 994) used, for the generation of a vortex flow, a single annular air jet swirling spirally. The swirling jet is issued from a single annular nozzle formed around the exhaust inlet, and the ex
17、haust flow rate is kept higher than the jet flow rate. The supplied swirling flow plays the role of an air curtain and thus increases the capture range, but a rather large volume rate of air has to be supplied to form an effective vortex. In case the exhaust control is not perfect, part of the suppl
18、y air does not get exhausted but escapes to the surround- ing space, resulting in a rapid spread, not of exhaust, but of contaminants. Spotar and Terekhov (1987) showed, by experimental means, that the VEER can have unstable flow states where two different flow patterns alternate depending on the fl
19、ow veloc- ity conditions. Shtem and Hussain (1 996) proved by mathe- matical models the existence of unexpected jumps between flow regimes and bi- or multi-stability in a swirling flow and predicted proper operating conditions for the VEER. As explained above, the vortex flow can increase the ventil
20、ation efficiency, but, due to the unique characteristics of a vortex flow such as the flow instability, it is necessary to generate a proper vortex flow fieid and also to find out proper operating conditions for an optimum performance with the flow instability eliminated. One effective way of genera
21、ting a vortex flow is the use of a rotating disc. Flow around a rotating disc has been studied extensively, aiming at a variety of applications. The most fundamental study was initiated by von Karman, who proved the existence of a mathematical solution that is assumed to be self-similar (Schlichting
22、 1979). Richards and Graebel(l967) studied numerically the flow around a rotating disc with a sink or a source at the center. In a vortex flow around a rotating disc, most of the vorticity resides inside the boundary layer over the disc surface, and axial propagation of vorticity is difficult becaus
23、e the boundary layer thickness gets reduced with increased rotational velocity. This implies that viscous prop- agation of vorticity in the axial direction is very inefficient. Then some sort of advection is required for efficient propaga- tion in the axial direction, for which the use of fins on th
24、e disc will be of great advantage. In this study a new vortex ventilation system (W) is developed using a swirler, a rotating annular disk with fins, for enhanced generation of vortex, where extra flow and space are not needed at all. Flow instability is eliminated over the whole operational range t
25、hrough CFD analysis and experimental verification, and the swirler shape is optimized, resulting in better ventilation efficiency than in any of the existing venti- lation systems. EXPERIMENTAL FACILITY Test System The test system is as shown in Figure 1. The exhaust tube of 95 mm (3.74 in.) diamete
26、r (D) is fixed on a steel frame of o z-direction Q Vortex Vent. O Rotating Disk (swirler) 0 Motor Blower 6 Blower Controller 8 Flow Meter 6 Pitot tube 0 Lower Surface Figure I Schematic diagram andphoto for the test facility. 150 ASHRAE Transactions: Research about 2 m (6.56ft) height, and a flat ac
27、rylic plate is placed beneath the exhaust in order to check the effect of the floor. Predictions by a CFD analysis were confirmed by the experi- ment that the Coanda effect, making the radial flow attach to the ceiling, manifests itself when the exhaust inlet is placed below the ceiling at a distanc
28、e of less than 2D, and that a tomado-like flow field is formed due to the floor effect when the distance between the exhaust inlet and floor is less than 5D. In order to eliminate the secondary effects induced by the ceil- ing and the floor, the distance from the exhaust inlet to the ceil- ing and t
29、he floor is fixed at 3D and 10D, respectively. A variable-speed motor to run the rotating disc is situated on the axis of the exhaust inlet, and the primary exhausting is done by a fan blower of O. 1 17 m3/s (4.1 1 cfs). Test Procedure The exhaust flow rate is always fixed at O. 1 17 m3/s (4.1 1 cfs
30、), and a variety of swirlers are tested with the rotational speed varied in the range of 0-3000 rpm. Then the average flow velocity at the exhaust, Uah, is 16.4 m/s (53.1 fps). The air velocity in the axial direction is measured with a pitot tube connected to a micro-manometer at an interval of 25 m
31、m (0.98 in.). At each measuring point the velocity data are accumu- lated for two minutes and averaged. CFD SIMULATION The RSM Model To get a numerical solution for the turbulent flow field, the continuity equation (i) and the Reynolds-averaged Navier- Stokes equation (2) are solved. *+-(pui) a = o
32、at ax; aui au. au/ - - + 2 - -6- + -(-puru.) axj axi 3 IJax) axj 1 J where t = time p = air density ui = three components of Bow velocity xi = coordinates P= pressure p = viscosity - In Equation 2 above -pujuj denotes the Reynolds stresses for which a proper turbulence model has to be used. For swir
33、l- ing flow, the Reynolds stress model (RSM), which directly solves the transport equations for the Reynold stresses, is known to be more accurate than the well-known k-E model. In this study it was confirmed by two-dimensional simulation that the results with the RSM model are in far better agreeme
34、nt with the experimental results. Flow fields are obtained for a steady-state condition. Numerical Technique The governing equations are solved numerically. Since the required computing time and capacity are very large when the three-dimensional system is analyzed as a whole, includ- ing the exhaust
35、 hood and the swirler, the computational domain was reduced to only one-quarter of the system and a “periodic boundary condition” is applied. The total number of grids is about 95,000, which seems sufficient for a reasonable resolution of the flow field. When a single case is solved again for a much
36、 finer grid system with about 215,000 grid points, the maximum difference in velocity caused by the grid refine- ment was less than 2%. The hexahedral grid is used for a good convergence, and the “rotating reference frame model” was used to easily solve the rotating flow field around the swirler. Th
37、e second-order upwind scheme was used for discretization, but for some cases that correspond to very unstable flow states, the first-order upwind scheme was used. The convergence criterion was set such that the sum of absolute residuals of sources for velocities, turbulence kinetic energy, turbulenc
38、e dissipation rate, and Reynolds stresses are all less than In CFD simulation the exhaust tube of 100 mm (3.94 in.) diameter was used, but the exhaust flow rate was the same as in the experiment; then the average flow velocity becomes 14.9 m/s (48.7 fps). SWILRER MODEL The swirler is the most import
39、ant part in VV, consisting of an annular disk, fin, and hub for connection (Figure 2). The base model with 1 O0 111111 (3.94 in.) inner diameter and 200 mm (7.87 in.) outer diameter was developed through two-dimen- sional CFD analysis for an efficient vortex generation with a small space requirement
40、. The base model has 12 fins of 10 mm (0.39 in.) height and 50 mm (1.97 in.) length at 30” interval. Also analyzed are the swirlers with different fin shapes as shown in Figure 2. In the CFD analysis, all the models were tested, but in the experiment, only two models, Base Model and Model A, were te
41、sted. Table 1 summarizes the main specifications for the swirler models. Also studied are the conical swirlers. Models D and E are slanted at an angle of +i5” and -15” each. These two models were numerically analyzed and experimentally tested. Through numerical analysis and experimental verification
42、 for a number of swirler designs, an optimum shape of swirler was decided. Then a prototype W system with the optimum swirler was manufactured, and its fluid-dynamic characteris- tics and exhaust performance were tested with flow visualiza- tion techniques and the like. RES U LTS Figure 3 shows the
43、measured profile of axial velocity with distance from the inlet for the base swirler. The case of zero rpm corresponds to the simple suction or exhaust, and it is noted that the measured profile is in good agreement with the theoretical results of DallaValle (1952). Axial velocity always ASHRAE Tran
44、sactions: Research 151 Figure 2 Model Base A B C I Swirler designs studied. Width, W Height, H mm (in.) mm (in.) 50 (1.97) 10 (0.39) 25 (0.98) 17 (0.67) 25 (0.98) 10 (0.39) 37.5 (1.48) 12.5 (0.49) Table 1. Dimensions of the Fins for Each Swirler Model decreases monotonically with distance from the e
45、xhaust inlet, but the slope of the velocity profile differs depending on the rotational speed. The velocity profile curves for different rota- tional speeds remain almost similar up to about 2200 rpm, but the curve for 3000 rpm becomes much flatter. So, when compared with the case of 2200 rpm, the a
46、xial velocity at 3000 rpm is lower in the near region, z/D 1.5. Also, it is visually and audially observed that noise and turbulence intensity increase with rotational speed up to 2200 rpm but decrease at higher speeds. That is, the flow pattern for the base model seems to undergo a sudden change ov
47、er 2200 rpm, which may be called a sort of critical speed. It is also observed that an unstable flow pattern and a stable one alternate even at a fixed rotational speed when it is close to the critical speed. These flow features are generally reproduced in a CFD analysis. Figure 4 shows the change w
48、ith rotational speed of the flow pattern and tangential velocity contour in the merid- ional plane. The capture velocity, vcap,defined in Equation 3, is given in Figure 4a-4d and the tangential velocity in Figure 4e and 4f, both normalized with the respective maximum tip tangential velocity and at a
49、n interval of O. 1. 2 2 1/2 - cap - (axial + radial) “ah (3) ! Model D l ! Model E I i I I I I 0.4 DallaValle -w- O000 rpm -o- 1500 rpm -A- 2200 rpm O 1 2 3 4 z/D Figure 3 Experimental projle of axial velocity with distance from the inlet for the base model. The flow field at 1500 rpm (4b) does not look very differ- ent from that of the simple exhaust (4a), but a secondary recir- culating flow is formed beneath the swirler surface due to an interaction between the exhaust flow and the radial flow induced by the swirler rotation. The secondary recirculating flow remains extending ax
copyright@ 2008-2019 麦多课文库（www.mydoc123.com）网站版权所有
备案/许可证编号：苏ICP备17064731号-1