1、OR-05-1 6-4 Past, Present, and Future Research Toward Air Curtain Performance Optimization Homayun K. Navaz, PhD Member ASHRAE Dana Dabiri, PhD Mazyar Amin Student Member ASHRAE ABSTRACT This paper presents a comprehensive discussion on past, present, and future research focused on display case air
2、curtain performance characterization and optimization. The past research mostly relies on simplified analytical solutions forjets. Thepresent approach takes a more comprehensivestep toward understanding and quantification of all majorparam- eters that afect the air curtain flow jeld by utilization o
3、f modern analytical/computational and experimental tech- niques. The goal of future work is to optimize air curtain performance as a function of the major design parameters by adoption a systematic approach. This approach would be inde- pendent of any particular display case design specifics and sho
4、uld be useful to all display case manufacturers. PAST RESEARCH Air curtains for open vertical refrigerated display cases are initiated at a supply cold air grille called the discharge air grille (DAG) that is basically a slot jet. Professor Ronald H. Howell and his associates have pioneered numerous
5、 and significant studies on air curtains. Initially, they investigated the transfer of heat and moisture through the plane of an air curtain (Howell et al. 1976). One of their most important find- ings was the direct proportionality of heat transfer across an air curtain to the discharge air velocit
6、y (DAV). Later studies by Howell and Shiabata (1 980) revealed that the ratio of the open- ing height (H) to the DAG width (w) and also the jet velocity (V) affect the “performance” of air curtains. This research was further extended to the turbulent flow formulation of a free jet. It examined the e
7、ffects of the turbulence intensity at the deliv- ery jet or DAG on the turbulence development process along the air curtain as it moves downward. Howell et al. (1983) Ramin Faramarzi, PE Associate Member ASHRAE showed that higher turbulence intensity (Ger) at the DAG or jet accelerates the widening
8、of the jet, causing a higher heat trans- fer across the air curtain. Their formulation was based on the incompressible boundary layer theory applied to shear layers. The analysis relied on the eddy viscosity model for turbulence flows. Howell and Adams (1 99 1) extended their analysis to the field.
9、They have shown that about 75% of the refrigeration load in an open vertical display case is a result of the warm air entrainment across the air curtain. Although the research above used simplistic formulations for air curtains, its importance lies in identiing most param- eters that impact “any” ai
10、r curtain performance. For instance, turbulence intensity at the DAG (Ger) as a boundary condition is a measure of mixing enhancement and the air curtain width. The more distance that the air curtain travels (H) also provides more opportunities for the air curtain to widen. The width of air curtains
11、 (w) provides the initial length for the flow to move laterally, which can enhance widening of the jet. The velocity at the jet (v) specifies how much kinetic energy is available at the boundary to be implemented toward the initiation and amplification of turbulence kinetic energy within the air cur
12、tain. These parameters are crucial to understanding air curtain performance. In terms of nondimensional quantities, these parameters can be grouped as (H/w), Reynolds number (Vw/v), and (re/V,. However, a free jet model is not quite applicable to an air curtain because of the presence of a return ai
13、r grille (RAG), the asymmetrical nature of display cases, non-aligned supply and return air passages, and usually complex geometry before the exit plane of the DAG that can affect the initial velocity profile at the DAG. Furthermore, the eddy viscosity model requires a mixing length model that is ba
14、sed on the definition of a boundary layer “edge.” This edge H.K. Navaz is an associate professor in the Mechanical Engineering Department, Kettering University, Flint, Michigan. M. Amin is a graduate student and Dana Dabiri is an assistant professor in the Aeronautics and Astronautics Engineering De
15、partment, University of Washington, Seattle. R. Faramarzi is manager of the Refkigeration and Thermal Test Center, Southern California Edison, Irwindale, California. Q2005 ASHRAE. i 083 is defined arbitrarily and its location significantly affects the turbulence viscosity and the extent of mixing. T
16、herefore, it can be concluded that although the earlier works of Howell et al. (1 976- 1991) provided information regarding the major parameters impacting the air curtain performance, but a more sophisticated model is required to “quanti” the dependency of the air curtain performance on the aforemen
17、tioned param- eters. Stribling et al. (1999) made an attempt to combine computational fluid dynamics (CFD) and experimental results to study the velocity and turbulence in a display case. In their CFD model they utilized a two-equation turbulence model that is better suited for free jet research. Th
18、is model does not utilize the boundary layer theory and therefore does not require a definition for the boundary layer “edge.” Their research indi- cated a good qualitative agreement but demonstrates some quantitative discrepancies between the experimental and computational results. Further applicat
19、ion of CFD codes to air curtains has been inconclusive due to nonmatching results between two CFD codes (Cortella and DAgar0 2002). They also found discrep- ancies among turbulence models within the same computer program. They recommended further investigation to identify the source of the inconsist
20、encies. One should realize that CFD provides a numerical solution to the conservation of mass, momentum, and energy equations, commonly known as the Navier-Stokes (NS) equations. It is mathematically known that there is no unique solution for these equations. So it is quite possible that a careless
21、implementation of a boundary condition (from a user or programmer) could propagate and yield inconsistent results. Above research may have benefited from addressing a simpler problem and then gradually intro- ducing complexities and comparing inconsistencies. Combining experimental and analytical me
22、thodologies in understanding air curtains dates back to the 1960s. Early works of Hetsroni et al. (1963) and Hetsroni and Hall (1963) were based on the laminar formulation of the boundary layer equations with body forces to study buoyancy effects. The analytical approach provided a correlation among
23、 nondimen- sional groups, such as Reynolds, Nusselt, Grashoff, and Prandtl numbers. Then experimental methods were used to curve-fit data and quantify the amount of air curtain heat trans- fer. It is evident that although the amount of heat transfer could be estimated, no detailed information could
24、have been obtained from this approach. A more modern analytical approach with the same basic goal, which took advantage of sophisticated tools such as CFD, was adopted by Axell and Fahlen (2002,2003). Their research resulted in development of a correlation for evaluating the Nusselt number for an ai
25、r curtain and evaluation of heat transfer and cooling load there- after. The effect of the Richardson and Reynolds numbers on the shape of the streamlines representing the entrained air at the DAG has been studied by Field et al. (2002). This work was quite valuable because it quantified the effects
26、 of the Richard- son number, Ri = Gr/Re2 (ratio of Grashoff to the square of Reynolds number), on the entrainment of ambient air into the cold air jet. The buoyancy effects that are represented by the Grashoff number demonstrate a controlling role on the entrainment. It was found that, for a Reynold
27、s number based on the DAG width of about 100, the buoyancy forces become significant and must be taken into consideration. Creating an air curtain at this very low Reynolds number requires either a rather small opening or low DAG velocity and may bring about issues ofpracticality. Furthermore, varia
28、tions in entrain- ment may not translate directly into infiltration of warm air into the display case. PRESENTRESEARCH The body of reviewed previous research work reviewed was focused on an atempt to understanding air curtain behav- ior and its controlling parameters. Most of the recent works intend
29、 to use modern techniques such as CFD and experimen- tal methods to better understand and quanti the behavior of air curtains. The application of CFD methods by itself could not be totally relied upon for the reason of existence of multi- ple solutions for the same problem. On the other hand, modern
30、 experimental methods are too time-consuming and expensive and they require a great deal of know-how. The best solution methodology appears to depend on an effective and careful combination of both technologies. Navaz et al. (2002) have demonstrated that a marriage between the digital particle image
31、 velocimetry (DPIV) experimental technique and CFD simulation can be quite effective. The DPIV can calibrate the numerical technique after which the CFD code can be used for parametric studies. They have shown that this hybrid approach can effectively produce curve fits similar to previous works tha
32、t can be useful for engineering calculations for heat transfer and entrainment rate. Furthermore, the wheel should not stop at just “engineer- ing calculations.” There is a need to identify, quantify, and optimize all the variables that can affect the air curtain perfor- mance. Recent works ofNavaz
33、et al. (2003,2004) take a more modem perspective of those issues that have previously been pointed out by Howell on entrainment rate as a function of Reynolds number and turbulence intensity at the DAG. It was found that the Reynolds number based on the DAG width, the shape of the velocity profile,
34、and turbulence inten- sity at the DAG, the length of the opening (vertical distance between DAG and RAG), and angle of throw will affect the entrainment rate. Based on simulation results, it is concluded that the turbulence level observed at the back panel flow inlet (if any) does not contribute muc
35、h to the overall entrainment. To demonstrate the importance of the DAG design, the origi- nal DAG geometry in a specific display case at the DAG was varied. The original DAG geometry of this case generated a two-peak velocity profile with relatively high turbulence intensity. To eliminate this doubl
36、e-peak velocity profile, the vertical surface in the original design was initially replaced by a 20”-and later 57”-slanted surfaces, postulating that a I 084 ASHRAE Transactions: Symposia 1 2“ Sutface, withHoneycnb TurbulentiGnetlcEnsrl8( - Jncg TU a8400 c; aima 0.3051 0.1838 0.1108 a0668 0.0402 34
37、0.0242 0.0146 0.0053 aom 0.0032 aoom % *2 O 0.0012 0.0007 TLcrbulentffinetk:Enery Jllm 67O Sutface, nith Honeycomb t Figure I Turbulent kinetic energy contours for a variety of geometries at the DAG region. more gradual change in the direction of flow would lessen turbulence at the DAG. The first an
38、gle represents the original design, and the second angle was suggested by CFD results. Turbulent kinetic energy contours for the original DAG geom- etry (actual geometry), the 20“ slanted surface design, and a 57“ slanted surface design are shown in Figure 1. After many simulations with different an
39、gles for this surface, it became clear that the 57“ with a wider throat provides the least turbu- lence intensity at the DAG for this particular case. In Figure 2, the velocity profile at the DAG exit for each geometry is shown and the two-peak profile of the original case is clearly seen. These two
40、 peaks cause a shear between two layers of fluid that can trigger mixing. The 20“ slanted surface profile seems to have a pronounced peak toward the outside of the case with another small peak to the right. It appears that this case may be less effective than the original design. However, as the ang
41、le is changed to 57“ and the flow passage area at the throat is widened, significant improvement Vertical Velocity Profiles at the DAG t Ij O 20 & T , TSUW where m = mass flow rate, Re, = vw I v = Reynolds number based on the DAG width and with being the average velocity at the DAG, - R mInfilfraied
42、/mToial Case 3 570 Slanted Surfacewith 10% Tuidence Intensity a the DAG 0.32250 0.19331 0.11594 0.06952 0.04168 0.02499 0.01499 0.00899 0.- 0.0wn 0.00194 0.00116 0.00070 O.ooo42 O.ooo25 0.00015 0.00009 Figure 4 Turbulence intensity contours for the proposed and actual cases with laminar and 10% turb
43、ulence intensity being imposed at the DAG. 1086 ASHRAE Transactions: Symposia normal vertical distance from the DAG to the RAG, DAG and RAG widths, DAG and RAG lengths, absolute average temperature at the DAG, absolute room temperature, turbulence intensity at the DAG angle between the line that con
44、nects the centers of the DAG and RAG and the vertical direction, offset angle when the RAG is shifted laterally, and throw angle, the angle between the surface normal to the DAG and RAG. Previous research indicated that for Reynolds numbers (based on DAG width) above 100 (Field et al. 2002), the tem
45、perature difference does not affect the entrainment and infiltration rates significantly. So, Equation 1 can be rewritten as R = f(Re, Hw, L/w, CI, , I). (2) It is evident that the design of a modular display case is necessary to perform all required parametric studies concern- ing Equation 2. A sch
46、ematic of this test air curtain is shown in Figure 5. A combination of DPIV and CFD experimental techniques will be used for the two following scenarios: A modular display case composed of only a DAG and RAG. The position of the RAG can be varied with respect to the DAG. In this geometry, the room a
47、ir is allowed to mix with the incoming air from the air curtain along the length of the DAG (H). The domain is bounded by two surfaces on the width of the DAG and RAG. This simple yet revealing setup would allow us to understand the behavior of the flow in relation to the parameters discussed above
48、without the added complexity of an enclosure that is Characteristic of a display case. The modular air curtain system will be modified to repre- sent an enclosure similar to a display case so that the air can be entrained from one side only. This configuration will allow us to extend our fundamental
49、 understanding acquired in step 1 (above) toward the more complex flow configura- tions found in real display cases. The quantity R in Equation 1 will be obtained by calcu- lating the entrainment rate from DPIV and CFD results. The infiltration rate will be directly measured by injecting a known amount of a tracer gas (such as carbon dioxide, CO,) into the incoming flow and measuring its infiltrated amount in the RAG. This method is expected to be far more accurate than the Figure 5 Schematic of the experimental air curtain setup. enthalpy method used previously (Navaz et al.
