There was no visual evidence of a liquid phase of any form (droplets, mist, film, etc.) in the vapor throttle when the boiler operated under normal conditions. However, a dense swirl of liquid mercury always appeared in the vapor throttle exit when lowquality vapor entered the throttle from the superheater. The liquid swirl appeared only in the condenser inlet where there was an abrupt enlargement of the tube diameter. Figure 15 shows mercury flow through the vapor throttle under various conditions. It is considered significant that, during metastable oscillations such as those represented in figure 10, no visual evidence of liquid choking or clogging of the valve was observed. This is considered significant in lieu of conclusions mentioned hereinafter concerning instability effects.

PREVIOUS FLOW PATTERN OBSERVATIONS

A number of previous mercury loops were constructed and operated under flow, temperature, and pressure conditions similar to those of the loop described heretofore. The X-ray system described in the APPARATUS section was used with the previous loops to observe flow patterns. Some of the observations are described briefly in appendix B. The appendix mentions flow phenomena seen in straight-tube boilers in which the problem of liquid carryover was demonstrated. Also described are the prototype helically swaged boiler tubes that demonstrated the improvements obtainable with liquid swirl.

DISCUSSION OF RESULTS

Some aspects of helical liquid motion are discussed first. In a succeeding section flow pattern correlations are discussed with respect to predictability and reproducibility of the patterns described herein. The discussion ends with comments concerning the instability effects and interactions promoted by nucleation and corrosion effects in the boiler inlet region.

Wall Contact and Phase Separation

The object of the groove in the boiler wall was to induce phase separation (i.e., stratification of the liquid phase) by conducting the liquid flow along a continuous helical path. The liquid flow was thus diverted from a purely axial to an essentially circumferential direction. Concomitantly, in this case the area of liquid contact per unit (axial)

length of the boiler is greater than that for an axially flowing streamlet having the same cross section. These ideas underlie the boiler designs employing twisted metal tapes and wire helix inserts (refs. 2 to 4). The use of helically grooved tubing for improving wall contact and thus heat transfer in heat exchangers and condensers has been previously proposed (refs. 11 to 13).

In the loop described herein, the groove certainly imparted some helical motion to the vapor flowing axially through the boiler-superheater. Therefore, some phase separation probably occurred because of centrifugal separation of liquid droplets that were entrained in the vapor stream. However, it should be expected that a guarantee for good phase separation farther downstream was the initial formation of the helical streamlet in the nucleation zone. When this streamlet has a sufficiently great tangential velocity in the nucleation zone, it tends to persist; but in a boiler, the downstream conditions (e.g., vapor phase velocity) are constantly changing with distance. Therefore, the initial momentum and radial acceleration of the liquid phase are important in ensuring continuance of the helical streamlet (see appendix C).

It was concluded, after a number of prototype mercury boilers were operated, that some sort of swirling device was required throughout the boiler (see appendix B). This requirement applies particularly if the boiler material is one that is not readily wetted by mercury. In this case, it is necessary to force intimate, sustained wall contact to attain wetting and prevent de wetting under two-phase conditions.

Flow Pattern Correlation

The stratified flow patterns that were observed in the mercury boiler can be explained in terms of the Baker flow pattern chart (ref. 14). The chart and an explanation of its usage and terminology are presented in appendix D. It is noteworthy that this chart was derived from adiabatic flow pattern studies. It should be expected that, with heat addition, flow patterns will be influenced by the rate of vapor generation and therefore heat flux. Difficulty in its application to the present case should also be expected because the chart is based on data obtained with straight cylindrical tubes.

The Baker chart for the boiler is shown in figure 16. Loci of G/A against Lay/G are plotted in figure 16 for the insert zone and the remainder of the boiler. The x-scale shows the estimated vapor quality for each value of Lλ/G. (See appendix E for definitions of symbols.)

In the boiler insert region, the diameter of the helical channel was approximately 0.04 inch (1 mm). In the remainder of the boiler and superheater, the hydraulic diameter was bracketed between 0.13 and 0.25 inch (3 and 6 mm). The plots of figure 16 are based on these diameters. The calculations for plotting the loci in the figure did not take

Figure 16. - Predicted flow pattern variations in mercury loop at 1075° F (583° C) with mass flow rate of
78 pounds per hour (9.8x10 kg/sec).

into account the fact that the liquid flowed on the boiler wall helically instead of axially, as a strict comparison with the Baker chart would require. Nor did the calculation account for the radial acceleration induced by the swirling action.

According to figure 16, a stratified-plug-slug-annular transition is predicted for the insert zone wherein the vapor quality increased from nil to approximately 0.03. Along the insert channel, the predicted variations between stratified and annular were not observed. Because of the strong swirling in the insert region, a radial acceleration several times that of Earth gravity was produced. It is therefore reasonable to expect that stratified flow was stable over a wider range, and thus the transition threshold was raised for stratified flow to higher values of G/λ for a given value of LX/G. Beyond the end of the insert, in the remainder of the boiler, the Baker chart predicts only stratified and wave regimes. The observations showed this to be the case even without corrections for the swirl effects.

The conclusion suggested by the preceding correlation is that the stratified helical

flow was not necessarily unique to the particular boiler described herein. Similar flow patterns can probably be expected for a range of boiler sizes and flow rates by designing in accordance with the Baker equations. However, the applicability of the Baker correlations appeared to depend on the induced liquid-to-wall contact. Previous experience with mercury flowing in straight tubes (appendix B) showed that anomalous flow patterns are more likely to occur with two-phase mercury flow in straight-tube boilers.

Corrosion and Instability Effects.

The corrosion and mass transfer in the nucleation zone could account for the increase in metastable fluctuations as the duration of the loop operation increased beyond several hundred hours.

Evidence that nucleation occurred in the insert zone is seen in figure 17, which shows a region downstream of the zone of nucleation inception after exposure to corrosion, erosion, and pitting damage for approximately 1147 hours.

Except for the pitted areas, the insert surface had remained intact, as evidenced by the original tool marks and scratches on its surface. By contrast, the inside surface of the boiler tube had been subject to severe corrosive and erosive attack. It is evident, therefore, that, because of radial acceleration, liquid flowed against the inside surface of the boiler wall. Separation of liquid from the insert surface was enhanced by bubbles