Skip Navigation


Annals of Occupational Hygiene Advance Access originally published online on December 9, 2004
Annals of Occupational Hygiene 2005 49(1):1-15; doi:10.1093/annhyg/meh065
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
49/1/1    most recent
meh065v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (9)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by HUANG, R. F.
Right arrow Articles by CHEN, C.-C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by HUANG, R. F.
Right arrow Articles by CHEN, C.-C.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?


© 2004 British Occupational Hygiene Society Published by Oxford University Press;

Aerodynamic Characteristics and Design Guidelines of Push–Pull Ventilation Systems

R. F. HUANG1,*, S. Y. LIN1, S.-Y. JAN1, R. H. HSIEH2, Y.-K. CHEN3, C.-W. CHEN4, W.-Y. YEH4, C.-P. CHANG4, T.-S. SHIH4 and C.-C. CHEN5

1 Department of Mechanical Engineering, National Taiwan University of Science and Technology, 43 Keelung Road, Section 4, Taipei, Taiwan, PRC; 2 Department of Mechanical Engineering, Kuan Wu Institute of Technology, 151 Yide St., Beitou Chiu, Taipei, Taiwan 112, PRC; 3 Department of Occupational Safety and Health, Chang Jung Christian University, Kway Jen, Taiwan, PRC; 4 Institute of Occupational Safety and Health, Council of Labor Affairs, No. 99, Lane 407, Hengke Rd., Shijr City, Taipei, Taiwan, PRC; 5 Institute of Occupational Medicine and Industrial Hygiene, National Taiwan University, 1 Jen-Ai Rd., Sec. 1, Taipei, Taiwan, PRC

* Author to whom correspondence should be addressed. Tel: +886-2-2737-6488; fax: +886-2-2737-6460; e-mail: rfhuang{at}mail.ntust.edu.tw


   ABSTRACT
TOP
 ABSTRACT
NOMENCLATURE
 INTRODUCTION
 EXPERIMENTS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 ACKNOWLEDGEMENTS
 REFERENCES
 
Aerodynamic characteristics such as the flow patterns, velocity field, streamline evolutions, characteristic flow modes and characteristic flow regimes of the push–pull ventilation system are cross-examined by using the laser-light sheet smoked-flow visualization method and laser Doppler velocimetry. Four characteristic flow modes, which are denoted as dispersion, transition, encapsulation and strong suction, are identified in the domain of the push-jet and pull-flow velocities at various open-surface tank widths and rising gas velocities. It is argued phenomenologically, from the aerodynamic point of view, that operating the system in the strong suction regime would be a better strategy than operating it in other characteristic regimes for the consideration of capture efficiency. Design guidelines are developed and summarized based on the results obtained from this study. The regression formulas for calculating the critical values of the push-jet and the pull-flow velocities are provided for easy access. The sulfur hexafluoride tracer gas validation technique is performed to measure the capture efficiency. The results of tracer gas validations are consistent with those obtained from the aerodynamic visualization and measurements. The operation points obtained by employing the American Conference of Governmental Industrial Hygienists design criteria are compared with the results obtained in this study for both the aerodynamics and the capture efficiency. Methods for improving the capture efficiency and energy consumptions are suggested.

Keywords: aerodynamic characteristics • push–pull ventilation • tracer gas


   NOMENCLATURE
TOP
 ABSTRACT
 NOMENCLATURE
INTRODUCTION
 EXPERIMENTS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 ACKNOWLEDGEMENTS
 REFERENCES
 
Ag
area of open-surface tank (=H x L)

Co,g
concentration of sulfur hexafluoride gas measured at 0.5 m upstream of end-opening of exhaust pipe when sulfur hexafluoride gas is injected into pull hood

Co,s
concentration of sulfur hexafluoride gas measured at 0.5 m upstream of end-opening of exhaust pipe when sulfur hexafluoride gas is released with air from open-surface tank

D
height of pull hood

E
height of push nozzle

H
width of open-surface tank

L
transverse length of open-surface tank

Qb
blowing volumetric flow rate of push nozzle

Qg
volumetric flow rate through open-surface tank

Qs
suction volumetric flow rate of pull hood

U
distance between lower edge of push nozzle/pull hood and liquid-surface-simulating screen of open-surface tank

u
local velocity component in x direction

v
local velocity component in y direction

vb
area-averaged flow velocity of push jet at exit of push nozzle (=Qb/(E x L))

push-jet velocity at minimum energy consumption point

v
g area-averaged rising gas velocity at exit of open-surface tank (Qg/(H x L))

pull-flow velocity at minimum energy consumption point

vs
area-averaged flow velocity of pull flow at inlet of pull hood (=Qs/(D x L))

S
slope ({equiv}vs/vb) of critical operation boundaries at

W
height of push-hood flange

x
coordinate along the direction of push jet that originated at the bottom edge of push nozzle in a symmetry plane

y
coordinate along the direction in height of push nozzle that originated at the bottom edge of push nozzle in a symmetry plane

z
coordinate along the direction in length of push nozzle that originated at bottom edge of push nozzle in a symmetry plane

{eta}
capture efficiency of hood (=Co,g/Co,s)

V
kinematic viscosity of air


   INTRODUCTION
TOP
 ABSTRACT
 NOMENCLATURE
 INTRODUCTION
EXPERIMENTS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 ACKNOWLEDGEMENTS
 REFERENCES
 
Pull hoods combined with push airflows have been widely applied in the industries that require the removal of contaminant vapors, fumes or aerosols from a large open surface during the half-century since Malin (1945)Go reported that a push–pull system can save airflow by about 50% when compared with using side exhaust alone.

Many investigators have studied the key parameters. Since the 24th edition of the ACGIH (2001)Go ventilation guide, Rota et al. (2001)Go have employed a numerical method to calculate the time-averaged flow field and have recommended some rules based on the calculated streamline patterns for the design of the push–pull system, and Marzal et al. published a series of papers (2002aGo,bGo,cGo, 2003aGo,bGo) on the development of an analytical model for the evaluation, determination and interpretation of capture efficiencies, influence of push-nozzle geometry on capture efficiency, methodologies for determining capture efficiencies and flow visualization results in the experimentally simulated surface-treatment tanks. Mathematical models and flow patterns on the open surface were addressed and discussed.

Most of the recommended design guidelines in the literature were based on the experimental results of tracer gas concentration measurements together with empirical rules for tunable parameters. There is still need for more study in the flow physics that governs the transport process of the contaminants, the push flows and the environmental air through the pull hood, and design strategy based on the flow physics is also a field where more effort is required. In this work, in order to understand the inter-correlation among the primary design parameters of the push-flow velocity, pull-flow velocity, rising gas velocity and open-surface width, and to obtain design guidelines based on detailed aerodynamic characteristics of the flow field, a series of studies were conducted in an experimentally simulated open-surface tank. The laser-light sheet flow visualization technique and a laser Doppler anemometer were applied to diagnose and analyze the physics embedded in the complex flows. A design procedure based on the systematic diagnostic results of the flow field is proposed. The SF6 tracer gas detection technique was subsequently used to evaluate the overall capture efficiency and local dispersion concentration and to validate the proposed design guidelines.


   EXPERIMENTS
TOP
 ABSTRACT
 NOMENCLATURE
 INTRODUCTION
 EXPERIMENTS
RESULTS AND DISCUSSION
 CONCLUSIONS
 ACKNOWLEDGEMENTS
 REFERENCES
 
Test rig
The system consists of three parts—the push nozzle, the pull hood and the open-surface tank, which is made of plexiglas and arranged as shown in Fig. 1. The coordinate system, the geometric parameters of the system and the flow parameters are defined in Fig. 2. The open-surface tank has a transverse length L = 0.5 m in the z direction. The width H of the open-surface tank is skillfully designed so that it can be easily varied between 0.5 and 1.5 m without causing any leakage of gas. We demonstrate in the Results and Discussion section that the governing principles, which dominate the embedded physics, are similar in either a short or a long tank if the aerodynamic phenomena are considered. We also demonstrate that using the lateral exhaust hood on a short-length tank does not result in better performance than using the push–pull system. The push–pull system is not installed on the longer side of the tank in our experimental system because the flow is basically two-dimensional in the symmetry plane and this arrangement of the experimental set-up can save expenditure. When using the design method proposed in this article for practical application, most of the push–pull system should be installed on the longer side of the tank if the spatial occupation is convenient for installation of the ducts, fans/blowers and nozzle/hood.



View larger version (25K):
[in this window]
[in a new window]
  		Fig. 1. Experimental setup.

 


View larger version (15K):
[in this window]
[in a new window]
  		Fig. 2. Definitions of geometric parameters.

 
The open-surface tank is in reality a large rectangular nozzle, where the airflows used to simulate the rising gas are driven by a centrifugal fan. The airflows go through a venturi flow meter, a series of flow conditioning devices (including a buffer tank, four flow distribution conduits, a perforated plate, a nozzle section, honeycombs and several layers of mesh screens) to reduce the turbulence and increase the flow uniformity, and a layer of liquid-surface-simulating screen, and then vent up to the atmosphere. The liquid-surface-simulating screen is made of fine metal grids with a mesh-size to wire-diameter ratio of about 2. The rising velocity of the air on the open surface was tested at 15 evenly distributed points using a calibrated fine-wire hot-wire anemometer. The results showed that the non-uniformity of the rising air velocity distributions is less than 1.5%.

Both the push nozzle and the pull hood have a rectangular cross-section with a cross-stream length L = 0.5 m, which is the same as that of the open-surface tank. The axes of the push and pull devices are arranged vertically to the axis of the open-surface tank. The distance between the lower surface of the push nozzle/pull hood and the liquid-surface-simulating screen of the open-surface tank, U, is 0.1 m. The push-nozzle height, E, is the same as the pull-hood height D with a fixed value of 0.05 m. Smaller push-nozzle heights are also tested in this study. Although the data are not shown because it is not the primary focus of this paper, the effects of the push- and the pull-hood heights are stated in the Design Guidelines section below. Honeycombs are inserted into the push nozzle and the pull hood to improve the uniformity of the jet velocity at the exit of the push nozzle and the suction flow at the opening of the suction hood. The uniformities of the jet and the suction flow are surveyed by a fine-wire hot-wire anemometer. The results show that the maximum non-uniformity of the jet and the suction flow are lower than 2%. The push jet and the suction flow are driven separately by two centrifugal fans. Venturi flow meters are used to measure the flow rates through the push nozzle and the pull hood. A pull-hood flange with a height W = 0.5 m is installed, while the push nozzle is not equipped with a flange. Determination of the subsidiary parameters U and W are made according to the experimental results of Shibata et al. (1982aGo,bGo), which show that when U/E ≥ 2 and W/E ≥ 5 the geometries of the liquid-surface height and the pull-flange height would have negligible effect on the flow patterns.

The flow parameters include the push-jet velocity vb (the area-averaged blowing velocity at the exit of the push nozzle), the pull-flow velocity vs (the area-averaged suction velocity at the opening of the pull hood) and the rising air velocity vg (the area-averaged gas velocity at the plane of y = 0). The width of the open-surface tank, H, together with the flow parameters, vb, vs and vg, are the primary influencing factors in this work. The values of vb, vs and vg are varied in the ranges of 0–3.5, 0–16 and 0–0.2 m/s, respectively.

Laser-light sheet flow visualization
The laser beam provided by a 5 W argon-ion laser is transmitted through an optical fiber and connected to a 20° laser-light sheet expander. The laser-light sheet expander is mounted on an adjustable block so that the light sheet can be flexibly aimed at any target plane. In this study, the light sheet is aligned across the symmetry plane of z = 0. The laser-light sheet has a thickness of ~0.5 mm. Mineral oil mist is continuously seeded through a homemade smoke generator into the conduit of the push nozzle and vented out with the push-jet flow to scatter the laser light. The diameter of the oil-mist particles, measured by a Malvern 2600C particle analyzer, is 1.7 ± 0.2 µm. The density is 0.821 g/ml. Ignoring the effect of turbulent diffusion, the relaxation time constant is estimated to be <7.7 x 10–5 s and the Stokes number is in the order of 10–6 within the range of experiment. Therefore, the seeding particles can properly follow the flow fluctuations at least up to 10 kHz (Flagan and Seinfeld, 1988Go). The particle images are recorded by both a Minolta Dynax 9xi still camera and a Hi-8 CCD camera. The still camera is equipped with an asynchronous variable electronic shutter, ranging from 1/12 000 to 30 s. The Hi-8 CCD camera can record images at 30 fps.

Laser Doppler velocimeter
The velocity field is measured with a two-component laser Doppler velocimeter. The whole-spectrum laser-light beam is provided by a 5 W argon-ion laser. Two color filters are used to allow blue and green laser beams of 514.5 and 488 nm, respectively, to pass through. They are subsequently split and focused through an optical system. The dimensions of measuring volumes of the green and blue components are 0.075 x 0.075 x 0.680 and 0.071 x 0.071 x 0.645 mm, respectively. The fringe separations of the green and blue components are 2.34 and 2.22 µm, respectively. The system is configured for backward scattering and installed with a Bragg cell and two electronic frequency shifters for the detection of direction ambiguity. Two electronic correlators are used to process the frequency of the signals. The digital outputs of the correlation processors are fed into a data-acquisition system. Each velocity data record consists of 1000 samples, about 3 s long. Uncertainty of the velocity measurements is estimated to be less than ±0.5% of reading. The seeding particles are the same as those used in the flow visualization.

Sulfur hexafluoride (SF6) detector
In the tracer gas experiment, a Miran 1 BX Portable Ambient Air Analyzer is used to measure the concentration of the sulfur hexafluoride gas. The Miran 1BX Portable Ambient Air Analyzer is a single-beam infrared spectrometer, using a pyroelectric lithium tantalite substrate as the detector that can detect concentrations of 150 toxic gases or vapors. The flow rate of pump for proper operation is 25–30 l/min. Sampling goes through a probe of a filter enclosure. The sampled gas is drawn through a hose and directed into a gas cell. The sample gas absorbs infrared energy from the infrared beam. The amount of absorption is measured by the detector, and subsequently amplified and converted to concentration units by the electronics. For SF6 concentration measurement, the span is from 0 to 99 999 ppm. After being calibrated in the range of 0–500 ppm, the accuracy of measurement is ±5% of reading. The lower limit of measurement is 0.6 ppm. In situ calibration was done each time, before the experiment was conducted.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 NOMENCLATURE
 INTRODUCTION
 EXPERIMENTS
 RESULTS AND DISCUSSION
CONCLUSIONS
 ACKNOWLEDGEMENTS
 REFERENCES
 
Gas transport process and aerodynamic characteristics
Characteristic flow modes. By using the laser-light sheet flow visualization technique with the assistance of smoke particles as the light-scattering media, the flow patterns in the symmetry plane are observed. Figures 3 and 4 show four fundamental flow patterns observed at different values of the pull-flow velocity vs when the push-jet velocity vb and the rising air velocity vg are set at 0.3 and 0.05 m/s, respectively. In Fig. 3 the smoke mists are issued along with the push-nozzle flow. While in Fig. 4, the smoke particles are issued along with the rising air flow from the open-surface tank.



View larger version (101K):
[in this window]
[in a new window]
  		Fig. 3. Photos of typical characteristic flow modes obtained by using smoked-flow visualization method assisted with laser-light sheet scattering. Smoke released with push-jet flow. vg = 0.05 m/s, vb = 0.30 m/s. (a) vs = 1.55 m/s, dispersion mode; (b) vs = 2.79 m/s, transition mode; (c) vs = 3.49 m/s, encapsulation mode; (d) vs = 4.17 m/s, strong suction mode. Shutter speed 1/30 s.

 


View larger version (112K):
[in this window]
[in a new window]
  		Fig. 4. Photos of typical characteristic flow modes obtained by using smoked-flow visualization method assisted with laser-light sheet scattering. Smoke released with rising air from open-surface tank. vg = 0.05 m/s, vb = 0.30 m/s. (a) vs = 1.72 m/s, dispersion mode; (b) vs = 2.49 m/s, transition mode; (c) vs = 2.83 m/s, encapsulation mode; (d) vs = 4.30 m/s, strong suction mode. Shutter speed 1/30 s.

 
In Fig. 3a where the pull-flow velocity is low, the push-jet going out of the push nozzle is tilted up by an angle by the rising current from the open-surface tank. After about the midpoint of the width of the open-surface tank, influence of the pull velocity becomes appreciable so that a small part of the jet flow is drawn down and goes into the pull-hood opening, while a large part of the jet flow disperses and goes across the upper edge of the pull-hood flange, because the pull velocity is not high enough. Along the jet boundary, the coherent structures (the vortices) are evolved because of the concentration of the vorticity that is induced by the shear effect developed in between the jet and the environmental flows (Tritton, 1988Go). During the evolution process of these coherent structures, some of the atmospheric air above the jet and the airflow under the jet, which is rising from the open-surface tank, are entrained into the jet by the rolling-and-traveling motions of the vortices. The coherent structures entrain neighboring air and the mixedness increases as the jet evolves downstream. When a large part of the jet fluids is dispersed into the atmosphere, as shown in Fig. 3a, the airflow issued from the open-surface tank and entrained by the coherent structures is inevitably dispersed into the atmosphere along with the jet fluid. This phenomenon can be seen clearly in Fig. 4a where the smoke particles released along with the rising air from the open-surface tank escape apparently to the atmosphere instead of being blocked under the bowed jet trajectory. This type of flows usually occurs at low-pull velocity vs and is called the dispersion mode. Operating a push–pull ventilation system at this mode would result in a serious dispersion of a large quantity of the rising contaminants from the open-surface tank to the atmosphere. A large push-jet velocity may cause such a condition if the corresponding suction strength developed by the pullhood is not properly matched.

In Fig. 3b, the pull velocity vs is greater than that of Fig. 3a. The highest point of the bowed jet trajectory moves to a location about three-quarters of the tank width from the push-nozzle tip. A large part of the jet fluids and the entrained environmental fluids are drawn into the pull-hood opening. A small part of the jet fluids goes up to a high level above the pull-hood along the pull-hood flange, detaches from the flange, turns around towards the jet, and is finally drawn by the suction hood. The air issued from the open-surface tank, as presented in Fig. 4b, is dispersed up to a high level due to the entrainment effect of the jet-evolution process and is finally sucked into the pull hood. Although not clear enough in the picture, it is observable in situ and from the continuous video images that some of the smokes issued from the open-surface tank recirculate around the area near the flange and above the pull-hood opening. This type of flow, which is called the transition mode, usually occurs at pull velocity vs higher than that of the dispersion mode.

At further higher suction velocity vs, the jet is still tilted up by the rising air from the open-surface tank, as shown in Fig. 3c. The tilt angle, however, is not as large as that found in the transition mode. Also, all of the jet fluids and the entrained environmental airs are drawn into the pull-hood opening. It seems that all the rising contaminants are isolated under the jet, as shown in Fig. 3c. However, traces of the smoke issued from the open-surface tank presented in Fig. 4c show that although all the smokes are finally drawn into the pull hood, the risk of turbulent and molecular dispersions across the jet to the atmosphere during the evolution process from the push-nozzle tip to the pull-hood opening is significant. Flows with this type of pattern are called the encapsulation mode.

When the suction velocity vs is increased further, the trajectory of the jet remains almost horizontal or even bowed toward the screen installed in the tank, as shown in Fig. 3d. The tendency of bowing down of the jet is possibly induced by the Coanda effect (Newman, 1961Go) when the jet momentum, as well as the distance between the jet and the wall, are matched in a proper range. Under this circumstance, the smoke particles issued from the open-surface tank are all drawn into the pull hood and isolated by the jet, as shown in Fig. 4d. The risk of dispersion of the tank fluids to the upper atmosphere across the jet due to the entrainment by the coherent structures is relatively low. Flows of this type of pattern are called the strong suction mode. If the push–pull ventilation system is operated in this characteristic mode, escape of the pollutants coming from the open-surface tank should be less than that operating at other characteristic modes. Besides, because the whole air curtain generated by the push and pull flows in the strong suction mode is below the level of the upper edges of the push–pull slots, it is possible to avoid the pollutants coming out of the width-sides of the open-surface tank by simply installing guard-plates along the sides of the open-surface tank.

Hand sketches shown in Fig. 5 summarize the observed typical flow patterns. There are three typical flow patterns included in the dispersion mode. The flow pattern (c) of the dispersion mode is the one corresponding to Figs 3a and 4b, which is usually observed at large push-jet velocities. The flow pattern shown in the subset (a) of the dispersion mode occurs when the push-jet velocity is extremely low. Under this situation the issued jet is pushed up by the rising air from the open-surface tank and rolls back towards the upstream area. The flow pattern shown in the subset (b) of the dispersion mode appears when the velocity of the push jet is in between those of pattern (a) and pattern (c). Under this situation, the strengths of the push jet and the pull sink are both insufficient because of which the rising air emitted from the open-surface tank ‘breaks through’ the air curtain and escapes to the atmosphere. There are also three typical flow patterns included in the transition mode. The flow pattern (c) of the transition mode is the one corresponding to Figs 3b and 4b, which is usually observed at large push-jet velocities. The flow patterns shown in subsets (a) and (b) of the transition mode are found at push-flow velocities lower than that for subset (c). The hand sketches shown for the encapsulation and strong suction modes in Fig. 5 delineate the flow patterns discussed in Figs 3c and 4c and in Figs 3d and 4d, respectively.



View larger version (13K):
[in this window]
[in a new window]
  		Fig. 5. Hand sketches delineating flow patterns of typical characteristic flow modes.

 
Velocity field. The velocity vectors and the streamline patterns of different characteristic flow modes in the symmetry plane at H = 1 m are superimposed onto the corresponding long-exposed pictures of smoke-flow patterns, as shown in Fig. 6. The velocity data are obtained by LDV measurements. The streamlines are obtained by using the shooting method based on the velocity vectors.



View larger version (143K):
[in this window]
[in a new window]
  		Fig. 6. Velocity vector field and corresponding streamline patterns of typical characteristic flow modes superimposed onto long-exposed smoke flow patterns. Exposure time of smoke pattern is 4 s. H = 1.0 m, vg = 0.05 m/s, vb = 0.3 m/s. (a) vs = 1.48 m/s, dispersion mode; (b) vs = 2.0 m/s, transition mode; (c) vs = 3.6 m/s, encapsulation mode; (d) vs = 5.0 m/s, strong suction mode.

 
In Fig. 6a for the dispersion mode, the suction strength of the pull hood is weak so that the rising air released from the open-surface tank pushes the jet issued from the push nozzle up and therefore forms a hump in the streamline that evolves from the bottom edge of the push nozzle (x, y) = 0. The boundary of the hump coincides almost with the lower boundary of the traces of the smoked jet. The existence of this boundary does not imply that the rising gas from the tank is isolated by the lower boundary of the deflected jet because the streamline and smoke patterns shown here are time-averaged ones. The dynamic behaviors of the flows discussed in the previous section clearly show the entrainment mechanism developed by the evolving vortices. Above the hump, the smoke particles disperse up to a wide region. On the pull-hood flange, the streamlines form a stagnation point at y/D {approx} 5. Above the stagnation point, the streamlines turn upwards along the flange and eventually vent out to the atmosphere. Below the stagnation point, the streamlines turn downward and eventually go into the suction hood. The streamlines going across the upper boundary of the jet indicates the entrainment of the atmospheric air by the deflected jet. In Fig. 6b for the transition mode, the hump appearing in Fig. 6a becomes insignificant. Also, the stagnation point on the flange disappears. Although all the streamlines near the flange above the upper edge of suction hood go to the suction opening, traces of the dispersed smoke can still be seen around that region. In Fig. 6c and 6d for the encapsulation and strong suction modes, the boundaries of the hump appearing in Fig. 6a and 6b become concave downwards. In Fig. 6d a recirculation bubble is formed in the area near and below the push nozzle. Apparently, it is difficult to identify the dynamic characteristic flow behaviors in this case from the time-averaged streamline patterns obtained from LDV measurements, because the entrainment, dispersion and transportation processes of the fluids induced by the complex interaction among the push jet, suction stream and the rising air are unsteady phenomena. From the fundamental principle of the fluid mechanics, the smoke particles in the pictures delineate the streaklines of the flows. The streaklines, however, would not coincide with the streamlines if the flow is unsteady. Therefore, the flow patterns delineated by the time-averaged flow-velocity measurements using the LDV or Pitot tube may not provide significant contributions to the diagnostics of dispersion characteristics of the unsteady flows.

Figure 7 shows the distributions of the axial velocity component u of four characteristic flow modes along the geometric centerline y/D = 0.5 at a fixed vb. These velocity profiles indicate two features. First, for the dispersion and transition modes, the suction velocity vs is not high enough so that the axial centerline velocities near the push nozzle (x/H < 0.1) are not significantly affected. In the region of 0.1 < x/H < 0.2, the velocity profiles drop sharply. Second, in the encapsulation and strong suction modes, the influences of the suction force on the axial centerline velocities are extended upstream to the push-nozzle tip. For the encapsulation mode, the values of u/vs remain almost constant downstream until x/H {approx} 0.7. In the region of x/H > 0.7, the velocity profile is sharply pulled up by the suction force provided by the pull hood. The distribution of the axial centerline velocities of the strong suction mode behaves similarly to that of the encapsulation mode except that the velocity profile of the strong suction mode is concave downwards. These two features in the centerline velocity distributions may be used for the diagnostics of the characteristic flow modes if a quantitative index is required. However, the smoke-flow visualization method assisted by the laser-light sheet technique may be a more convenient and rapid way for identification of the flow features.



View larger version (27K):
[in this window]
[in a new window]
  		Fig. 7. Axial velocity distributions of four characteristic flow modes along x direction in geometric centerline at y/D = 0.5 at vg = 0.05 m/s and H = 1.0 m.

 
Characteristic flow regimes
Figure 8 shows the characteristic flow regimes where the above-discussed characteristic flow modes are observed at H = 1.5 m. The bands of short slashed lines represent boundaries between different characteristic flow modes. They are obtained by fixing the push-jet velocity at various values and increasing the pull-flow velocity gradually from zero to high values. The characteristic flow modes are identified according to cross-reference of the in situ visual-flow patterns, the pictures and the video motion images. The thickness of the slash-lined boundary bands indicates the uncertainty of identification. Figures of the characteristic flow regimes for the open-surface tank width other than 1.5 m are also obtained, although not shown here. They all have similar patterns and characteristics.



View larger version (58K):
[in this window]
[in a new window]
  		Fig. 8. Characteristic flow regimes at H = 1.5 m. (a) vg = 0.03 m/s, (b) vg = 0.07 m/s, (c) vg = 0.15 m/s and (d) vg = 0.20 m/s. Bands of short slashed lines represent boundaries between different characteristic flow modes.

 
All the boundaries between the characteristic flow regimes follow similar general behaviors, although the critical values of (vb, vs) are different for different vgs, as shown in Fig. 8a–d. When the boundaries in Fig. 8a–d are compared with each other, it is apparent that the larger the value of vg is, the higher the value of critical vs will be for the same value of vb. The critical pull-flow velocity vs on the boundaries decreases first with the increase of the push-jet velocity vb when vb is increased a little from zero. The tendency of the decrease of the pull-flow velocity benefits from the push jet. When the push-jet velocity vb is increased to a value that is commonly lower than about 0.3 m/s, the corresponding critical value of vs reaches a minimum. When the push-jet velocity vb is further increased beyond the minimum critical value, the corresponding critical value of vs increases. The increase rate can be taken as approximately linear.

As discussed previously, according to the aerodynamic behaviors of the flow field shown in Figs 3 and 4, the push–pull ventilation system may attain a good performance if the operation parameters of vb and vs are set in the strong suction regime for the corresponding vg and H. Therefore, the boundary between the regimes of the encapsulation and the strong suction mode becomes the minimum operation requirement. That boundary is hereafter termed the critical operation boundary and the minimum point on this boundary is called the minimum energy-consumption point, as shown in Fig. 6. The values of (vb, vs) corresponding to the minimum energy-consumption point are denoted by (, ). It is obvious from Fig. 8 that a small velocity vb together with a proper value of vs would be enough to force the ‘air curtain’ to the strong suction regime. If the push-jet velocity vb is set at a high value, the pull-flow velocity vs will have to be correspondingly large to operate in the strong suction regime. In our experience, if the cross draft in the environment is not significant, it is recommended not to use high push-jet velocity if the energy-saving is taken into account. The primary design work by employing the aerodynamic observations, therefore, requires determination of the functional relationship among the parameters of vb, vs, vg, and H on the critical operation boundary.

Discussion on ACGIH design code
According to the ACGIH design code (ACGIH, 2001Go), the criteria for determination of the push- and pull-flow rates are listed as follows (the units have been converted into metric system).

The push-flow rate Qb can be calculated by

(1)
where Qb is the blowing volumetric flow rate of the push nozzle in cubic meters per minute, and L and E are the width and height of the push nozzle, respectively, in meter, as shown in Fig. 2.

The pull-flow rate Qs can be calculated by

(2)

(3)
where Qs is the suction volumetric flow rate of the pull hood in cubic meter per minute, Ag is the area of the open-surface tank ({equiv}L x H) in square meter, and Tg is the temperature of the liquid surface in °C.

By plugging the geometric parameters (H = 0.5–1.5 m, L = 0.5 m, E = 0.05 m) used in this study in equations (1)Go–(3), calculating for two cases: Tg < 65.5°C and Tg = 100°C, and converting the volumetric flow rates Qb and Qs into the push velocity vb and pull velocity vs, the results by employing the ACGIH design code are obtained and listed in Table 1. Since the dimensions of the push nozzle are fixed in all cases, the push-jet velocities vb for H = 0.5, 1.0 and 1.5 m have the same value of 3.1 m/s. The pull-flow velocities vs for H = 0.5, 1.0 and 1.5 m are 3.82, 7.63 and 11.45 m/s, respectively, for Tg < 65.5°C. They are 5.07, 10.15 and 15.22 m/s, respectively, for Tg = 100°C. These data for H = 1.5 m are marked in Fig. 6. For vg = 0.03 m/s, as shown in Fig. 8a, the design point of the ACGIH code falls in the regimes of strong suction and encapsulation for Tg = 100°C and Tg < 65.5°C, respectively. For higher vg, as shown in Fig. 8b–d, the ACGIH designs deteriorate and even fall into the regimes of transition or dispersion. This is because the critical operation boundary rises in vs with the increase of the rising gas velocity vg. The deterioration of the operation becomes even more serious when the open-surface tank width becomes smaller, e.g. H = 1.0 and 0.5 m, as shown in the last column of Table 1. Using vb = 3.1 m/s does not seem necessary because it requires a much higher pull-flow velocity vs in order to push the operation point into the strong suction regime. Using a vb lower than 3.1 m/s and matching with a higher pull-flow velocity vs may be a better strategy to operate the system in the strong suction regime.


View this table:
[in this window]
[in a new window]
  		Table 1. Values of (vb, vs) calculated from ACGIH design code for H = 0.5–1.5 m, L = 0.5 m, E = 0.05 m

 
Equation (1) indicates that the push-flow rate Qb is related solely to the size of the push nozzle. For a fixed nozzle width L, the push-flow rate Qb is proportional to the square root of the nozzle height E. The push-jet velocity vb therefore equals 40.9/E0.5. In other words, the push-nozzle height alone plays the role of determining the push-jet velocity. This design criterion does not seem logical because the effects of the pull velocity vs, the rising gas velocity vg, and the open-surface tank width H are not considered while determining push-jet velocity vb.

Equation (2) implies that the pull-flow rate Qs is a linear function of the open-surface tank area Ag with a proportionality coefficient of 22.9 m3/min/°C for Tg < 65.5°C. If the liquid-surface temperature Tg is greater than 65.5°C, then equation (3) takes the temperature effect (actually, it is the effect of the increased rising gas velocity) for consideration by adding a factor of 0.22Tg to the proportionality coefficient of 8.44 m3/min/°C. The pull-flow velocity vs is thus proportional to H/D. For a fixed pull-hood height D, it becomes that vs {propto} H. In other words, vs is simply linearly proportional to H. In the next section on Design guidelines, we show that vs is not a simple linear function of H. Instead, complex non-linear behaviors are embedded in these primary influential parameters vb, vs, vg and H.

Design guidelines based on aerodynamic characteristics
Analysis. As discussed in the earlier section on Gas Transport Process and Aerodynamic Characteristics, it is preferred to operate the push–pull ventilation system in the regime of strong suction. The design work employing the aerodynamic observations would require determination of the functional relationship among the primary flow parameters of vb, vs, vg and H on the critical operation boundary. The subsidiary geometric parameters, such as E, D, U and W, can be selected subsequently according to designers' experiences or the results of previous investigations. If the critical operation boundaries can be cast into a mathematical formula, determination of (vb, vs) for the operation of the system would be more convenient. By using all the experimental data of the critical operation boundary for various H's, Figs 9 Go–11 are obtained for convenient access of the design procedure.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 9. Normalized minimum suction velocity, , of pull hood at minimum energy-consumption point.

 


View larger version (25K):
[in this window]
[in a new window]
  		Fig. 10. Ratio between suction velocity and push velocity at minimum energy-consumption point.

 


View larger version (19K):
[in this window]
[in a new window]
  		Fig. 11. Slopes of critical operation boundaries.

 
In Fig. 9, the normalized minimum energy-consumption suction velocity, , of the pull hood at the minimum energy-consumption point increases non-linearly with the increase of H. This phenomenon is different from that implied by the ACGIH design code as shown in equations (2) and (3), which indicates that the design value of vs should increase linearly with the increase of H. For a fixed H, the higher the rising gas velocity vg is, the larger the value of . The following regression equation provides for the convenience of calculation:

(4)
where all variables adopt the metric units, i.e. , vg in meters per second and H in meters. Once the rising air velocity vg is estimated and the open-surface tank width H is known, Fig. 9 or equation (4) can be employed to locate or to calculate the value of . Maximum deviation of the fitted curve is 6.3% at H = 0.5 m and vg = 0.15 m/s.

Figure 10 shows the variations of the normalized parameter with the open-surface tank width H. Again, this variable increases non-linearly with the increase of H. The following regression equation is provided for the convenience of calculation:

(5)
where all variables adopt the metric units, i.e. , , and vg in meters per second and H in meters. Once the rising air velocity vg is estimated, the open-surface tank width H is known, and the is obtained by using Fig. 9 or equation (4), then Fig. 10 or equation (5) can be employed to locate or to calculate the value of . Maximum deviation of the fitted curve is 11.2% at H = 0.5 m and vg = 0.10 m/s.

The values of vs on the critical operation boundary for can be taken approximately as a linear function of vb, as shown in Fig. 8. The slopes S ({equiv}vs/vb) of the critical operation boundaries for at various Hs and vgs are collected in Fig. 11. It can be seen that the slopes of the critical operation boundaries at fixed H remain almost constant for different vgs. The slope S, however, increases linearly with the increase of the open-surface tank width H. The linear equation is provided below for calculation:

(6)
where H is in m and S is dimensionless. Once the open-surface tank width H is known, the slope S can be obtained by using Fig. 11 or equation (6).

Once the values of (, ) and S are obtained, the regime for the strong suction mode can be expressed by the following equation

(7)
The equality holds on the critical operation boundary. A safety factor may be applied to vs calculated on the critical operation boundary.

In our experience in practical applications, equations (4)GoGo–(7) can be confidently applied to the range of H smaller than 2.0 m. It is unknown if they would deviate to intolerable values when H is larger than 2 m.

Design guidelines. The design procedure for the primary parameters vb, vs, vg and H, as discussed above, is summarized as follows:

  1. For an open-surface tank with a width H, estimate possible maximum rising gas velocity vg by employing thermodynamic principles or by measurements.
  2. Based on the values of H and vg, use equation (4) or Fig. 9 to obtain .
  3. Based on the values of H, vg, and , use equation (5) or Fig. 10 to obtain .
  4. Based on the values of H, use equation (6) or Fig. 11 to obtain the slope S.
  5. Choose an operation push-jet velocity vb. If the cross draft is not so serious, pick up a low value for vb, e.g. taking ~1.0 m/s for H which is shorter than 2 m. The smaller the H is, the lower the vb should be taken. An excess vb requires a large vs to operate the system in the strong suction mode.
  6. Based on the vb predetermined in step (5) as well as the (, ) and S calculated in steps (2)–(4), use equation (7) to calculate corresponding vs on the critical operation boundary.
  7. Multiply the value of vs on the critical operation boundary, which is calculated in step (6), by a safety factor of ~1.1–1.2 to obtain the design value. This range of safety factor is obtained from the experience of the tracer gas concentration experiments.
  8. Put guard plates along sides of the open-surface tank to avoid lateral leakage of the contaminants. The length of the guard plates is the same as the tank width H. The height can be as large as the height D of the pull hood. The purposes of the installation of guard plates are 2-fold. First, when the system is operated in the strong suction regime, the guard plates can reduce the lateral dispersion (leakage) of the contaminants and the influence of the lateral draughts because the contaminants are completely enclosed by the upper ‘air curtain’ and the side guard-plates under the operation condition of the strong suction. Second, the push-jet velocity vb can be set at low value so that energy consumption can remain low.

Some of the authors' experiences to design the subsidiary geometric parameters, E, D, U and W, are provided as follows.

  1. The height of the push nozzle, E, can be designed as a value between 2 and 5 cm, depending on the seriousness of the cross-draft around the push–pull system. If the cross-draft is large, a larger value of push-nozzle height can be considered in order to increase the ‘bulk-momentum’ of the isolation air curtain. For E smaller than 1 cm, the pressure drop becomes large.
  2. The height of the pull hood D can be taken in the range of 1 ≤ D/E ≤ 2. In the authors' experience, taking D/E = 2 is of little help in lowering the level of the critical operation boundary than taking D/E = 1. This arrangement may lower a little the necessity of the critical suction velocity vs.
  3. The depth between the pollutant emission surface and the lower edge of the pull hood, i.e. U, can be determined according to Shibata et al.'s (1982b)Go suggestions: U/D ≥ 2.
  4. Aerodynamically, the suction-hood flange would not be necessary if the push–pull system is operated in the strong suction mode because the push jet and the entrained pollutants and atmosphere air would not disperse by convection to the area higher than the upper edge of the pull hood. However, installation of a short flange may be helpful to reduce extra energy consumption induced by the flows turning-around from the upper region of the suction hood.

SF6 tracer gas validation
In order to validate the effectiveness of the proposed design guidelines, tracer gas concentration experiments for measurement of the capture efficiency are conducted. When measuring the capture efficiency of the push–pull system, the sulfur hexafluoride gas of a certain flow rate is first injected into the pull hood. After evolving along a exhaust pipe 10 m long for good mixing, the concentration of the SF6 tracer gas, which is denoted as Co,s, is measured in the exhaust at the location ~0.5 m upstream of the end-opening of the exhaust pipe. The sulfur hexafluoride gas of the same flow rate as used in the measurement of Co,s is injected into the inlet piping that is connected the open-surface tank and is released with the rising air. Again, the SF6 tracer gas concentration is measured at the point ~0.5 m upstream of the end-opening of the pipe and is denoted by Co,g. The capture efficiency {eta} is defined as the quotient of Co,g divided by Co,s, i.e. {eta} {equiv} Co,g/Co,s.

Figure 12 shows the capture efficiency measured for H = 1.0 at vg = 0.05 and 0.15 m/s. Figure 12a,b shows the case without guard-plates. Figure 12c,d presents the case with guard-plates installed. Figures for H = 0.5 and 1.5 m at vg = 0.03, 0.10 and 0.20 m/s are not shown here because their characteristics are similar to Fig. 12. In the system either without guard-plates or with guard-plates installed, the capture efficiency {eta} increases with the increase of the pull velocity vs. In the regime of the dispersion mode, the capture efficiency is low and the increase rate is high. In the strong suction regime, the capture efficiency is generally increased to values approximately higher than 98%. When Fig. 12c and d are compared with Fig. 12a and b, the capture efficiencies of the push–pull system with guard-plates is obviously higher than those without guard-plates. In order to attain a capture efficiency higher than 99%, it requires the critical value of vs corresponding to a certain vb on the critical operation boundary to be raised by ~10~20%.



View larger version (46K):
[in this window]
[in a new window]
  		Fig. 12. Capture efficiency {eta} measured for H = 1.0 at (a,c) vg = 0.05 and (b,d) vg = 0.15 m/s.

 
The capture efficiencies measured in the cases of the operation conditions set according to the ACGIH design code, i.e. (vb, vs) = (3.05, 7.63) m/s are marked in Fig. 12 for comparison. Although ACGIH did not suggest the installation of guard-plates, the benefit is apparent by doing so. At vg = 0.05 m/s, the capture efficiencies of the ACGIH design fall in the transition regime. In this case, it is medium high to about 97 and 98% for no guard-plates and guard-plates installed, respectively, as shown in Fig. 12a and c. To attain the same capture efficiency as that of the ACGIH design, the energy consumption required using the present design method is much lower than the conventional ACGIH method. For instance, the capture efficiency shown in Fig. 12c is about 98% by setting (vb, vs) = (3.05, 7.63) m/s using the ACGIH design code. Lower velocity pairs of (vb, vs) = (1.0, 5.6) m/s is good enough to attain the same capture efficiency if the current design method is employed. If the velocity pair (vb, vs) = (1.0, 8.0)m/s is adopted, the operation point falls in the strong suction regime and the capture efficiency can even attain almost 100%. Obviously, an over-blowing by the push nozzle is the primary cause of downgrading the operation characteristic flow mode. This phenomenon can be observed more clearly in the following example. At vg = 0.15 m/s, as shown in Fig. 12b,d, the capture efficiencies measured for the ACGIH designs are low because the operation points, i.e. (vb, vs) = (3.05, 7.63) m/s, fall in the dispersion regime. If the push velocity vb is reduced from 3.05 to 0.15 m/s and the pull velocity vs is kept at the same value of 7.63 m/s, the capture efficiency would be increased to about 99%, as shown in Fig. 12d.


   CONCLUSIONS
TOP
 ABSTRACT
 NOMENCLATURE
 INTRODUCTION
 EXPERIMENTS
 RESULTS AND DISCUSSION
 CONCLUSIONS
ACKNOWLEDGEMENTS
 REFERENCES
 
The flow patterns and aerodynamic characteristics of the push–pull ventilation system were phenomenologically and quantitatively studied. The following conclusions are drawn from the results and discussion.

  1. Four characteristic flow modes, dispersion, transition, encapsulation and strong suction, can be identified in different regimes in domain of the push-jet velocity and the pull-flow velocity, according to the aerodynamic characteristics diagnosed by flow visualization and laser Doppler velocimetry. To obtain better capture efficiency, operating the push–pull system in the strong suction regime with the assistance of guard-plates is a better strategy than operating in other characteristic regimes.
  2. By operating the push–pull system on the critical operation boundary, a capture efficiency of higher than ~98% is possible if the guard-plates are installed. The capture efficiency can be further increased to higher than 99% if the critical pull velocity is increased by 10–20%. When the obstacle objects exist in the pathway of the push–pull flow, the present results may require some modifications.
  3. The proposed design guidelines suggest a step-by-step procedure for designers to determine the primary flow parameters, the push-jet velocity and the pull-flow velocity, if based on the open-surface tank width and estimated rising gas velocity. The design ranges of the subsidiary geometric parameters, such as the push-nozzle height, pull-hood height, distance between the pull hood and the free surface and the push-hood flange, are suggested according to the authors' experiences.
  4. The push–pull system by using the ACGIH design criteria can be further refined in both capture efficiency and energy consumption by considering the aerodynamic characteristics of the flow transport phenomena.


   ACKNOWLEDGEMENTS
TOP
 ABSTRACT
 NOMENCLATURE
 INTRODUCTION
 EXPERIMENTS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 ACKNOWLEDGEMENTS
REFERENCES
 
This research was supported by the Institute of Occupational Safety and Health, Council of Labor Affairs of Taiwan, PRC.

Received May 19, 2004; in final form July 12, 2004


   REFERENCES
TOP
 ABSTRACT
 NOMENCLATURE
 INTRODUCTION
 EXPERIMENTS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 ACKNOWLEDGEMENTS
 REFERENCES
 

ACGIH. (2001) Common industrial ventilation. Industrial ventilation—A manual of recommended practice. 24th edn. American Conference of Governmental Industrial Hygienists. Lansing, MI, pp. 10-108–10-112.

Flagan RC, Seinfeld JH. (1988) Fundamentals of air pollution engineering. Englewood Cliffs, NJ: Prentice Hall, pp. 290–357.

Malin BS. (1945) Practical pointers on industrial exhaust system. Heat Vent; 42: 75–82.

Marzal F, González E, Miñana A et al. (2002a) Analytical model for evaluating lateral efficiencies in surface treatment tanks. Am Ind Hyg Assoc J; 63: 572–7.

Marzal F, González E, Miñana A et al. (2002b) Determination and interpretation of total and transversal linear efficiencies in push–pull ventilation systems for open surface tanks. Ann Occup Hyg; 46: 629–35.[Abstract/Free Full Text]

Marzal F, González E, Miñana A et al. (2002c) Influence of push element geometry on the capture efficiency of push–pull ventilation systems in surface treatement tanks. Ann Occup Hyg; 46: 383–93.[Abstract/Free Full Text]

Marzal F, González E, Miñana A et al. (2003a) Methodologies for determining capture efficiencies in surface treatment tanks. Am Ind Hyg Assoc J; 64: 604–8.

Marzal F, González E, Miñana A et al. (2003b) Visualization of airflows in push–pull ventilation systems applied to surface treatment tanks. Am Ind Hyg Assoc J; 64: 455–60.

Newman BG. (1961) The deflection of plane jets by adjacent boundaries—Coanda effect. In Lachmann GV, editor. Boundary layer and flow control—its principles and application. Vol. 1. New York: Pergamon Press, pp. 232–64.

Rota R, Nano G, Canossa L. (2001) Design guidelinesfor push–pull ventilation systems through computational fluid dynamics modeling. Am Ind Hyg Assoc J; 62: 141–8.

Shibata M, Howell RH, Hayashi. (1982a) Characteristics and design method for push–pull hoods: Part 1—cooperation theory on airflow. ASHRAE Trans; 88: 535–56.

Shibata M, Howell RH, Hayashi. (1982b) Characteristics and design method for push–pull hoods: Part 2—streamline analyses of push–pull flows. ASHRAE Trans; 88: 557–70.

Tritton DJ. (1988) Physical fluid flows, 2nd edn. New York: Oxford University Press, pp. 289–94.


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 ANN OCCUP HYGHome page
S.-J. Tsai, R. F. Huang, and M. J. Ellenbecker
Airborne Nanoparticle Exposures while Using Constant-Flow, Constant-Velocity, and Air-Curtain-Isolated Fume Hoods
Ann. Hyg., November 20, 2009; (2009) mep074v1.
[Abstract] [Full Text] [PDF]

Home page
 ANN OCCUP HYGHome page
M.-J. CHERN and W.-Y. CHENG
Numerical Investigation of Turbulent Diffusion in Push Pull and Exhaust Fume Cupboards
Ann. Hyg., August 1, 2007; 51(6): 517 - 531.
[Abstract] [Full Text] [PDF]

Home page
 ANN OCCUP HYGHome page
R. F. HUANG, Y. D. WU, H. D. CHEN, C.-C. CHEN, C.-W. CHEN, C.-P. CHANG, and T.-S. SHIH
Development and Evaluation of an Air-Curtain Fume Cabinet with Considerations of its Aerodynamics
Ann. Hyg., March 1, 2007; 51(2): 189 - 206.
[Abstract] [Full Text] [PDF]

Home page
 ANN OCCUP HYGHome page
T. OGDEN
Annals of Occupational Hygiene at Volume 50: Many Achievements, a Few Mistakes, and an Interesting Future
Ann. Hyg., November 1, 2006; 50(8): 751 - 764.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
49/1/1    most recent
meh065v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (9)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by HUANG, R. F.
Right arrow Articles by CHEN, C.-C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by HUANG, R. F.
Right arrow Articles by CHEN, C.-C.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?