Home | Business News | Browse by Publication | A | ASHRAE Transactions

Performance prediction of adiabatic capillary tubes by conventional and ANN approaches: a comparison.

Article Excerpt
INTRODUCTION

The capillary tubes are the expansion devices used for expanding refrigerant from the condenser to the evaporator of a vapor compression system. Low cost, no maintenance requirement, and the ability to start the motor at low torque makes capillary tubes favored for use in low-capacity (up to 10 kW) refrigeration equipment, like household refrigerators and packaged air conditioners.

Capillary tubes have been investigated in detail for many decades. Staeblar (1948) pioneered the use of the capillary tube as a refrigerant controller. A capacity balance characteristic was obtained, and the factors affecting it were also explored. Mikol (1963) conducted another comprehensive experimental study to investigate the various phenomena associated with refrigerant flowing inside an adiabatic capillary tube.

Koizumi and Yokoyama (1980) conducted experiments on glass capillary tubes for flow visualization. They found that the flow inside the capillary tube is a homogeneous two-phase flow. An extensive work on adiabatic capillary tubes with refrigerant HCFC-22 was reported by Kuehl et al. (1990). They concluded that drawn capillary tubes could not be considered smooth tubes. Furthermore, they were probably the first to report the effect of coiling on the refrigerant mass flow rate. They concluded that, irrespective of the length of capillary tube coiled, the mass flow rate was reduced by no more than 5%. Based on the experimental data, several researchers have proposed a number of semi-empirical correlations for the prediction of mass flow rate of various refrigerants for a given size of capillary tube and given inlet and outlet conditions. Wolf et al. (1995) proposed a correlation for the prediction of mass flow rate through an adiabatic capillary tube for the flow of refrigerant HFC-134a. Another important study regarding the flow of newer refrigerants inside capillary tubes was carried out by Melo et al. (1999). They undertook the project of studying capillary tubes with CFC-12 and its alternatives HFC-134a and HC-600a. They proposed separate correlations for CFC-12, HFC-134a, and HC-600a and proposed a combined mass flow rate correlation for all three refrigerants inside an adiabatic capillary tube. These correlations were developed from the experimental data. Choi et al. (2003) modified the dimensionless parameters and proposed a correlation for the prediction of mass flow rate through the capillary tube, and Fiorelli et al. (2006) carried out an experimental analysis of refrigerant mixtures HFC-407C and HFC-410A through adiabatic capillary tubes. The survey was carried out for both subcooled and two-phase inlet conditions and characterized the influence of both refrigerants. The geometric parameters, such as capillary tube diameter and length, were also incorporated in the study. Jabaraj et al. (2006) recently investigated the flow characteristics of the newer ecofriendly refrigerant mixture HFC-407C/HC-600a/ HC-290 (termed [M.SUP.2]0) inside an adiabatic capillary tube. They also developed a nondimensional correlation to predict the mass flow rate.

Numerical modeling and development of correlation from the experimental data play an important role in the design of capillary tubes. Another novel and simple approach--an approximate analytical approach for analyzing the flow of refrigerants through an adiabatic capillary tube--was developed by Yilmaz and Unal (1996). Yilmaz and Unal (1996) did not consider choking at capillary tube exit leading to incorrect results. Zhang and Ding (2001) considered choking at the capillary tube leading to iterative calculation. The use of artificial neural networks (ANN) in the prediction of refrigerant mass flow rate through capillary tubes was recently reported in published literature. Yasar et al. (2004) used the ANN, based on a backpropagation algorithm, to predict mass flow rate and suction line outlet temperature of experimental data from Melo et al. (2002) and Bittle et al. (1995) for diabatic (nonadiabatic) capillary tubes. The error analysis suggests the ANN model by Yasar et al. (2005) predicts the output more accurately than do the proposed correlations of Melo et al. (2002) and Bittle at al. (1995). Zhang (2005) proposed a generalized correlation of refrigerant mass flow rate through an adiabatic capillary tube using ANN. He used the experimental data of various refrigerants from the published literature on adiabatic capillary tubes for training and testing using the ANN model.

It is quite common in a refrigeration system that refrigerant enters the capillary tube in a saturated state (zero inlet subcooling). One of the drawbacks associated with the non-dimensional correlations in nonlinear power law form is that such correlation fails to predict the mass flow rate for zero inlet subcooling. The biggest advantage of modeling data using the ANN approach is the ANN model predicts the mass flow rate accurately, even for zero inlet subcooling. Neither Yasar et al. (2005) nor Zhang (2005) conducted their own experimental study to compare the mass flow rate predicted by the regression correlation and that predicted by the ANN model. Therefore, an experimental study has been conducted to measure the mass flow rate of refrigerant HFC-134a through a capillary tube of different lengths and diameters for different inlet subcoolings. The data acquired in this experimental study facilitated the development of a semi-empirical correlation using regression analysis and in the development of an ANN model using a feed-forward backpropagation training algorithm to predict the refrigerant mass flow rate through the adiabatic capillary tube.

EXPERIMENTAL SETUP

The schematic...

View this article FREE - Now for a Limited Time, try Goliath Business News
Free for 3 Days!