1、OR-05-1 6-3 Design Tool for Display Case Evaporators Ramesh Chandrasekharan Student Member ASHRAE ABSTRACT A model capable of simulating evaporators with multiple modules under frosting conditions was developed, validated, and used to optimize a display case system. Effects ofgeometry on fan and com
2、pressor power consumption were quantij?ed, and some general guidelines for designing display case evap- orators were developed. Thesegeneral guidelines were used to design more eficient prototypes for a given external load, keeping in mind the packaging restrictions and constraints on tube and Jin a
3、vailabilities. The performance of these new prototypes was compared to the original baseline, and on the basis of these comparisons, recommendations for further improvement of display case systems are made. INTRODUCTION This paper presents the development of energy- efficient evaporators for medium-
4、temperature open multi-deck display cases. The most significant aspect of current display case evaporators is the frost accumulation and the resultant change in air-side heat transfer and pressure drop characteristics. The change in air-side pressure drop due to frosting leads to higher fan power re
5、quirements and also results in lower airflow rate through the air curtain wherever constant speed fans are used. Any drop in air-curtain flow leads to higher infiltration and, hence, higher latent and sensible loads. Hence, to reduce energy consumption it is necessary to design for low frosting rate
6、s and to minimize the effect of frost on air-side pressure drop and heat transfer. Most early research on frosting was experimental, focus- ing on frost growth or its effect on coil performance. Research focused on properties of frost growing on typical structures such as flat plates. (e.g., Mao et
7、al. 1999, Hayashi et al. Clark Bullard, PhD Fellow ASHRAE 1977, Storey and Jacobi 1999, Ostin and Andersson 1991, Yonko and Sepsi 1967). A comprehensive review of the effect of frosting on heat exchangers is provided by Kondepudi and ONeal (1987), who also investigated frost growth effects on the pe
8、rformance ofheat exchangers with flat and louvered fins. Tassou and Datta (1 999) studied frosting of multi-deck display case evaporator defrost intervals as a func- tion of ambient humidity. Carlson et al. (200 1) experimentally studied frosting of a secondary refrigerant-based evaporator of geomet
9、ry similar to those used in display cases. The main aim was to understand the effect of ambient temperature, humidity, and other operating conditions on frost deposition and distri- bution. Ogawa et al. (1993) studied ways to improve heat exchanger performance under frosting conditions, including st
10、aging, cutting, and/or extending the fins. Most of these results focused on heat exchangers very different from those typically used in supermarket display cases. Most of the early experiments were conducted on single-row heat exchangers, providing no insight into frost distribution along deeper coi
11、ls. Also, they lacked information regarding the effect of ambient conditions on frost deposition. A lot of modeling work has been done on water conden- sation on air-conditioning coils, but modeling of frost distri- bution in heat exchangers is much more limited. Models of the frosting processes hav
12、e focused mainly on frost deposition on cold surfaces, such as flat plates or single cylinders (Tao et al. 1993; Padki et al. 1989). A few researchers did look at frost modeling of heat exchangers, but apart from Verma et al. (2002), most dealt with geometries typical of heat pumps or refrigerators
13、and not display cases. Recently, Seker (2004) presented the development of a semi-empirical model to simulate frosting in commercial Ramesh Chandrasekharan is a graduate student and Clark Bullard is a research professor in the Mechanical Engineering Department, University of Illinois at Urbana-Champ
14、aign. 02005 ASHRAE. 1 O71 refrigerators. Although the numerical model simulates heat and mass transfer equations similar to those used by Verma et al. (2002), it was semi-empirical and developed for refrigera- tors. It does not concern itself with the intricacies of display case system issues and ho
15、w they relate to evaporator simula- tion. It does not consider the change in frost thickness in the airflow direction and, hence, is not capable of simulating heat exchangers with multiple modules and/or different fin pitches. The model developed here overcomes these difficulties, and this paper ill
16、ustrates the power of such a numerical tool. The most significant issues in modeling display case evaporators is the fact that the geometries of most currently used heat exchangers lie outside the range of available corre- lations for heat transfer and pressure drop. Also, it is necessary to use fin
17、ite volume techniques to model local frost deposition within the heat exchanger. An added benefit of using such a finite volume technique is that modeling of variable fin spac- ing and heat exchangers with more than one module in the air flow direction is a trivial task. The model presented here als
18、o includes simple equations describing ali thermal loads, the air curtain, and ambient temperature and humidity. By considering the evaporator as part of a larger system, the evaporator inlet conditions are properly defined so the detailed optimization of the evaporator geometry can proceed uncouple
19、d from the details of air curtain and load optimization. The models structure and validation are addressed briefly, The validated design tool is used to arrive at general recommendations for evaporator design when the inlet condi- tions are known. Finally, the design tool and the recommen- dations a
20、re used to arrive at improved designs for evaporators. MODEL AND VALIDATION The numerical model uses a quasi-steady finite-volume approach to simulate cross-counterflow DX evaporators in frosting conditions. Within the finite-volume framework, the model can handle variable fin and tube properties th
21、rough a modular approach. It is implemented using a Newton-Raph- son-based nonlinear simultaneous equation solver. (Klein and Alvarado 1995) A few simpliing assumptions are used in the modeling: Frost growth modeling is simplified by considering only mature frost growth and by assuming an initial fr
22、ost layer of negligible (5 pm) thickness on the fins and tubes. This assumption eliminates the need to model the complex early crystal growth period. The frost surface roughness effect, which increases the surface area, is also neglected because it is significant only in early frost growth. 2. 3. Fi
23、n and tube frost thickness is computed for each finite volume, typically one per tube row for a typical 8-12 row coil. As temperature increases, frost density grows to that of ice at 0C; then density remains constant while thickness increases. Frost thickness thus increases monotonically. 4. The fre
24、eze-thaw cycles observed in laboratory experiments (Raju and Sherif 1993) are neglected. The justification for neglecting this complication lies in the validation of the numerical model. Only the leading edge finite volume is assumed to contain a superheated segment because of the small values of su
25、per- heat normally employed in display case evaporators. The heat and mass transfer modeling and governing equa- tions are given by Verma et ai. (2002) and not reproduced here. The essence of the model is to determine the frost surface temperature by solving simultaneously the separate equations for
26、 the convective heat (sensible) and mass (latent) transfer between ambient air and the frost surface. The correlations used are also discussed in Verma et al. (2002). The main corre- lations are also listed in the references (Hayashi et ai. 1977; Gnielinski 1976; Souza and Pimenta 1995; Wattelet et
27、al. 1994). Fin efficiency effects are addressed by using smaller finite elements to calculate heat transfer through fins, frost, and tubes. The model was validated against data taken at two-minute intervals from a well-instrumented medium-temperature vertical display case, for two evaporators having
28、 very different configurations (the original baseline coil and an interim coil referred to as Prototype-1). The data from these experiments (Faramarzi 2003) provided both the input data for the model and also the data to compare the modeling results. Redundant data were recorded, and used to help qu
29、antify experimental uncertainties. The general structure of the validation process is presented here. For details see Chandrasekharan and Bullard (2004). The variables chosen as input to the model were the geometry of the coil, air inlet temperature, initial air flow rate (determined from an energy
30、balance), the fan curve, duct pres- sure loss coefficient, refrigerant inlet enthalpy, refrigerant exit pressure, air curtain infiltration rate, and store conditions. Data from both the baseline (0.5 in. diameter tubes) and Prototype-1 (3/8 in. diameter tubes) coils lead to the same conclusions. Fol
31、lowing is a gist of the validation process and the strengths and weaknesses of the model: 5. The data available included air-side temperatures that showed a variation of -3OC from one end of the inlet to the other. Given that airflow rates werent directly mea- sured, these data had to be used to cal
32、culate the initial air- flow rates (using an energy balance), which were later used as an input to the model. This uncertainty propagates in the model, affecting the values of the calculated refng- erant- and air-side temperatures. Allowing for this uncer- tainty in some model inputs (-10-15% in air
33、flow rates and -1-2C in the measured outlet air temperatures), plus the instrumentation errors affecting other measured inputs to the model, it is not possible to estimate refrigerant and coil surface temperatures closer than the 1-1.5“C differ- ences observed for the three coils tested. To validate
34、 the 1072 ASH RAE Transactions: Symposia model more precisely would require a display case with more uniform air flow or increasing model complexity to simulate the nonuniformities. The overall conclusion was that to achieve a particular discharge air temperature, the model predicts a slightly lower
35、 evaporating and surface temperature than suggested by the experimental values. The mass of frost matches within 4%. Although some of the variables couldnt be predicted quantitatively with high accuracy, the predicted and observed trends for all the variables always matched. Complete validation was
36、not possible due to the absence of direct airflow measurements and the uncertainties in the fan curves. Since the calculated refrigerant-side resistance is small (DT-0.75“C from the refngerant to the outer surface of the frost) and fairly certain, the air-side heat transfer coeffi- cient would have
37、to be underestimated by 40% or more to explain the difference between the measured and predicted. Therefore, it is likely that the uniform air flow assumption is the major factor contributing to the difference between measured and predicted evaporator performance. DESIGNING DISPLAY CASE SYSTEMS This
38、 section examines the extent to which the evaporator coil optimization can be decoupled from the larger process of designing a display case. Other components, such as the ducts and fans, must meet the air flow requirements of the case and its air curtain as well as the coil. There are two important
39、aspects of using an evaporator simulation to design energy-efficient display cases. The first deals with how the inputs to the evaporator simulation are chosen when the evaporator is viewed as part of a display case system. The second aspect involves extending the modeling to an optimization. Figure
40、 1 gives a brief outline of the steps involved in simulating/designing the display case. The basic idea is that the coil simulatioddesign process and the air flow rate selection can be viewed as essentially separable if a few minor approximations are made. Airflow Rates and Inputs to the Evaporator
41、Simulation The airflow rates are key to any effort to minimize energy consumption. Figures 2a and 2b show the composition of the total airflow rate and how the individual airstreams interact with the display case system. The two airstreams can be considered separately for a first order analysis. The
42、 flow via the back panel over the products is the medium of heat transfer interaction between the loads and the air flowing through the refrigerated space. The loads on a display case include the radiation load (-lo%), lights, anti-sweat heaters, other accessories, and other effects, such as conduct
43、ion (-10-15%) and the load due to infiltration (75- 80%). Only the back panel flow deals with the non-infiltration load, so the required back panel flow rate can be determined by knowing the non-infiltration loads on the display case. A 7 Figure 1 Outline of simulatioddesign steps. Entrainment Air c
44、urtain Ducting RAG +spillover iT-r Shelvesproduct Figure 2a Schematic of airflow paths in the display case. Back now Ambient air entminment Figure 26 Cross section of a display case showing airflow components. simple first order relation governing the effects of back panel flow rate on the DAT (disc
45、harge air temperature) is presented in Equation 1. The heat transfer between the product and the air around it is given by - Qnon-entrainment - hair . Aproduci (Tproduci- Tdischarge) . (1) When the air flows over the products fast enough for forced convection, the heat transfer coefficient depends o
46、n the velocity. A higher heat transfer coefficient allows a higher ASHRAE Transactions: Symposia 1073 DAT to maintain the same product temperature, which, in turn, enables raising the evaporating temperature of the coil. This lowers the amount of frost forming on the coil. Moreover, the warmer coil
47、surface reduces the rate of frost formation, further increasing the COP of the display case system. Thus, it makes sense to improve the heat transfer between the products and air by redesigning the airflow path andor by increasing the back panel flow rate. The only other factor that might have to be
48、 considered while selecting the back panel airflow rate and velocity is its potential for affecting air curtain stability or increasing turbulence. The other stream is the air curtain flow through the DAG. The main consideration while designing this stream is mini- mizing infiltration while providin
49、g a stable barrier between the refrigerated space and the ambient. Inertial and downward buoyancy forces stabilize, whereas entrainment and viscous shear destabilize the air curtain, placing a lower bound on air curtain flow. High flow rates cause turbulence that increases infiltration. The optimum value for the air curtain flow and the design of the corresponding discharge air grille are done with the help of CFD analysis (Walker et al. 2003). Once the loads and the air-curtain stability criteria are known, the airflow rates and accompanying infiltration rate are calculate
