Search

Friction factors were determined for fully developed turbulent flow of mercury in smooth concentric annuli under conditions where either both walls were unwetted, or both were wetted, or the inner wall was wetted and the outer one unwetted. Three radius ratios (r2/r1) were used, i.e., 2.09, 2.78, and 4.00. Unwetted walls gave the lowest friction factors, which were practically independent of the r2/r1 ratio over the limited range tested. The factors were 10 ± 1 percent higher than the commonly accepted values for smooth pipes (at the same Reynolds number). The highest friction factors were obtained with the inner wall wetted and the outer wall unwetted, and the greater the r2/r1 ratio the greater was the effect. For example, at r2/r1 = 4.00, the friction factors were 9.9% greater than for the situation when both walls were unwetted. The wetting conditions affected the location of the radius of maximum velocity (rm); and it was found that the nearer rm approached r2, the higher was the friction factor.

An experimental heat-transfer program has been under way for some time at the Brookhaven National Laboratory, the general purpose of which is to obtain and correlate heat-transfer coefficients for liquid metals flowing outside of tube banks. This paper covers one phase of the program; i.e., the measurement of heat-transfer coefficients for mercury flowing normal to a staggered tube bank. The factors studied were (a) linear velocity, (b) tube location in the bank, (c) circumferential variation of the local coefficient for a single tube, (d) type of contact, i.e., “wetting” versus “nonwetting,” and (e) pressure drop.

Using sodium, an experimental study was carried out to determine the effect of rate of temperature rise on the incipient-boiling superheat for fully developed, forced turbulent flow. Three different methods of achieving boiling inception were used. The magnitude of the incipient-boiling superheat was found to be quite reproducible and independent of the experimental method as long as the rate of temperature rise (temperature ramp) was kept constant and kept the same. Moreover, the value of the superheat was found to have a strong dependency on the temperature ramp—the greater the ramp, the larger the superheat. The failure to carefully measure and control temperature ramp is apparently one reason why incipient-boiling-superheat data generally show so much scatter, and those from different laboratories so much disagreement.

Experimental results were obtained for the case of in-line flow of mercury through an unbaffled bundle of circular rods, and they were compared with theoretical predictions. The bundle consisted of 13 one-half-in-dia rods arranged in an equilateral triangular pattern, the pitch:diameter ratio being 1.750. Measurements were taken only on the central rod. Six different rods were tested. All rods in the bundle were electrically heated to provide equal and uniform heat fluxes throughout the bundle. The rods were of the Calrod type. The test rods had copper sheaths with fine thermocouples imbedded below the surface for measuring surface temperatures. Some rods were plated with a layer of nickel, followed by a very thin layer of copper, to provide “wetting” conditions, while others were chromeplated to provide “nonwetting” conditions. Heat-transfer coefficients were obtained under the following conditions: (a) Prandtl number, 0.02; (b) Reynolds number range, 7500 to 200,000; (c) Peclet number range, 150 to 4000; (d) “Wetting” versus “nonwetting”; (e) Both transition and fully established flow; (f) Variation of Lf/De ratio from 4 to 46. The precision of the results is estimated to be within 2 to 3 percent. An interesting finding, consistent with earlier predictions, was that the Nusselt number, under fully established turbulent-flow conditions, remained essentially constant, at the lower end of the turbulent flow regime, until a Reynolds number of about 40,000 was reached.

As part of a continuing liquid-metal heat-transfer program at the Brookhaven National Laboratory, both local and tube-average heat-transfer coefficients have been obtained for the flow of mercury normal to a staggered tube bank. The bank consisted of sixty ½-in. tubes, six wide and ten deep, arranged in an equilateral-triangular array. The paper presents results showing the effects of flow rate, Prandtl number, wetting, gas entrainment, and tube location on the tube-average coefficients. The angular variation of the local coefficient is not considered here, owing to the fact that the values have not been calculated completely from the original data. The mercury results are compared with a few results obtained with water in the same equipment. The heat-transfer runs cover the Reynolds number range of 20,000 to 200,000, whereas the water runs cover the range 8000 to 20,000. Pressure-drop results for both water and mercury flow through the tube bank are reported also. The mercury coefficients for tubes in the interior of the tube bank are well represented by the equationNu=4.03+0.228(Pe)0.67

An experimental study of heat transfer to mercury flowing in line through an unbaffled rod bundle was carried out. The “rods” were special electrical heaters whose claddings had different thicknesses and thermal conductivities. The experiments were carried out under a thermal boundary condition approaching that of uniform heat flux in all directions at the inner wall of the rod cladding. It was found that displacement of a rod from its symmetrical position can result in a large reduction in its average heat transfer coefficient. This reduction increases exponentially with the amount of displacement. For a given direction and amount of displacement, the reduction is little affected by variations in cladding thickness and conductivity but is affected considerably by flow rate. Not only does the displaced rod suffer a reduction in its own average heat transfer coefficient, but so do those toward which it is displaced. At the same time, the average coefficients of the rods from which it is displaced remain about the same. Thus the overall average coefficient of the group of affected rods goes down when a single rod is displaced.