CONDENSATION OF STEAM ON SHORT TUBES IN HORIZONTAL ROWS PLACED IN A VERTICAL GRID

Abstract

The present investigation has been carried out to study experimentally the condensation of low pressure steam on four horizontal short tubes(L/d = 13.6), placed in a vertical grid. The tubes were housed in a rectangular vessel of 300 mm length, 310 mm breadth, and 750 mm height. Each tube of the condenser was 341 mm long with an inside diameter of 25 mm and outside of 28.8 mm. The tubes were held in perfect horizontal position by means of specially fabricated check-nuts. The clearance between two nearest tubes was 97 mm. In order to measure the temperature of outer surface of each tube, twelve copper-constantan thermocouples were embedded at four locations at distances of 21, 121, 221, and 321 mm, respectively measured from the leading edge of the tube. Every location had three thermocouples, one placed at each of the top-, the side-, and the bottom-regions of the tube. The thermo-e.m.f. signals from all the tube wall thermocouples on analysis showed periodic variation with time. Due to the large amplitude of variation a hardware-based signal integrating system comprising of Keithley programmable DMM with GPIB interface and Z-80 micro-processor was adopted to find out time-averaged value of the fluctuating signals. This scheme was thought adequate in view of its being fast in processing large number of data points. A specially designed and developed home-made mechanical hand was employed to measure the temperature of jsteam bulk around the tubes in a vertical plane by means of thermocouples. It had two arms each containing a copper-constantan thermocouple. It was possible to control the vertical and horizontal positions of the thermocouples in a vertical plane, containing tubes, from outside the test condenser with a minimum accuracy of 1.0 mm. The steam used in this investigation was raised in an oil fired boiler. To ensure that the steam entering the condenser was dry and saturated, extra precaution was exercised by employing an upright U-loop having a vertical drain pipe with steam trap at its free end in the pipe line connecting the boiler to the test condenser. In fact this helped in removing the condensate coming along with steam from the boiler before entering the test condenser. The temperature of steam was monitored by copper-constantan thermocouples attached to mechanical hand. An air vent with continuous steam purging was used to keep the test condenser free from noncondensables. To check the absence of noncondensables, the temperature of the steam was measured. When the temperatures maintained by the thermocouples placed near the top and bottom tubes were found equal to saturation temperature corresponding to the pressure of steam, it was taken that the steam was free from noncondensables. To maintain a quiescent environment of steam around the tubes of the condenser the incoming steam through a steam nozzle placed off-line to the vertical grid of tubes was allowed to impinge on a cup and then passed through a perforated plate. The above arrangement reduced the velocity of steam to a practical possible minimum value. The cooling water was pumped into the tubes through rotameters. The provision of inverted U-bend at the exit of each •ii tube ensured that the tubes were always full with cooling water, which was very necessary. The temperature distribution' of cooling water was determined theoretically for the estimation of heat flux distribution along different segments of each tube. This was done by iterating the heat rate equation with heat balance equation and finally matching the calculated exit temperature of cooling water with experimentally determined value. For this purpose each tube was considered to be made of four isothermal segments of 71, 100, 100, and 70 mm length, respectively. Experiments were conducted for steam pressures ranging from 146.75 to 269.38 kPa and cooling water flow rate from 11.6 to 17.1 1pm. The present investigation has shown the expected behaviour that the temperature of the wall of the tubes varies circumferentially. It is also observed that due to the continuously decreasing values of cooling water-side heat transfer coefficient, unlike that of long tube (L/d > 50) having high flow rate of cooling water, the value of the average circumferential wall temperature continuously increases throughout the length of the tubes irrespective of the row in which the tube lies. Thus the surfaces of the tubes become nonisothermal when condensation occurs on short tube condensers. It is established that for a given tube-row the average circumferential wall temperature of the tube is found to possess a specific functional relationship with the cooling water flow rate, it's inlet temperature, steam pressure, and distance from iii the leading edge of the tube. For the calculation of condensing heat transfer coefficient from an available correlation for the first row tube, a relationship for the weighted wall temperature for short tube, in terms of operating parameters is also obtained. These experimentally determined weighted wall temperatures of the first row tube of the bundle show a good agreement with the model of Bromley within a maximum deviation of +35%. The experimental data for the condensation of quiescent steam on first row tube of the bundle are correlated best by the Mikheyev correlation within a maximum deviation of -18.0 % to 10 %, followed by the correlation due to Henderson and Marchello within a maximum deviation of -26 % to 6%. For the determination of weighted condensing heat transfer coefficient for short tubes in second-, third-, and fourth- rows, based on the present experimental data, an empirical correlation is recommended within a maximum deviation of +10%, which establishes that heat transfer coefficient of a tube in rows other than the first row possesses a functional relationship with the heat transfer coefficient of first row tube and the row number in which the tube lies. It is also found that the predictions from Kern's correlation and the experimental values of weighted condensing heat transfer coefficient of second-, third-, and fourth-row tubes agree excellently within a negligible deviation. The finding of the present investigation is that Jakob's correlation for heat transfer coefficient for tubes in different rows of the iv tube bundle is a conservative one is in conformity with the observation reported by Marto.