and/or a vapor blanket that formed as a result of buoyancy effects produced by centrifuging.

The pitting on the insert piece (fig. 17) was probably produced by liquid-entrained vapor bubbles that were formed on the boiler wall and then driven by the liquid stream into the helical crevice formed along the line of contact between the insert cylinder and the boiler wall. The bubbles should have collapsed in this region because the temperature of the insert was less than the temperature of the boiler wall where the bubbles formed (by perhaps 10° F or 5.5° C or more).

It was inferred from the temperature variations near the boiler inlet, that during metastable intervals, the location of the point of two-phase inception was changing. After the loop test ended, examination of the pitting and corrosion patterns in the insert zone seemed to confirm this. Examination of the deposits farther downstream in the boiler showed considerable evidence that granular particles were eroded from the insert zone. This could account for the cavities (i.e., nucleation sites) in the boiler surface in the nucleation zone (ref. 10).

During the first 500 hours of the test, there were only a few moderate metastable intervals. Thereafter, they became more frequent and more pronounced. This increase in metastable oscillations and excursions was apparently related to corrosion and mass removal events in the insert zone. The disturbances were probably triggered by and then interacted with changes in the nucleation pattern. This pattern was, in turn, determined in part by the availability of alternative nucleation sites in the insert zone. As corrosion progressed, more cavities and cracks were formed, and, hence, more alternative nucleation sites became available along the insert zone channel. Meanderings of the point of two-phase inception and changes in the nucleation pattern were probably triggered when these alternative nucleation sites became activated.

Instability Interactions

A comparison of calculated transit times and the period of metastable oscillations indicated that the pulse rate of these oscillations was closely related to the boilersuperheater geometry and heat flux. The amplitude of the metastable pressure oscillations corresponded closely to the pressure drop in the insert zone. This seemed to confirm the conclusion that the oscillations were associated with rhythmic relocations of the point of nucleation inception in the insert zone. This kind of link between oscillations and particulars of boiler geometry and heat flux is discussed by the authors of reference 15. There was no visual evidence indicating that metastable oscillations originated or continued because of liquid choking effects in the vapor throttle. The constancy of pump-developed and output pressure during flow and pressure oscillations seemed to

eliminate the possibility that metastable interactions were transmitted through the liquid leg. Boiler entrance conditions were rather completely decoupled from the condenser because the condenser exit pressure and pressure variations of the liquid condensate column were nearly zero.

Interactions occurring during metastable intervals were probably confined within the boiler-superheater. The X-ray observations indicated that there was a coupling between the nucleation phenomena and the flow patterns seen in the vicinity of the liquid film terminal. The interaction between these two regions was undoubtedly affected by the mode of thermal and power control used for the boiler-superheater.

CONCLUDING REMARKS

The X-ray observations made with prototype mercury boilers indicated that swirling devices greatly enhance phase separation. Without some kind of swirling device, appreciable liquid carryover was found in previous boilers examined by X-ray. In the boiler described herein, shallow helical grooves in the boiler wall served to keep the liquid flowing against the wall. The centrifugal forces thus produced resulted in the intimate, sustained wall contact apparently needed with mercury to induce wetting, prevent dewetting, and thus enhance phase separation.

It appears that good phase separation and prevention of liquid carryover is better accomplished if a helical liquid streamlet with high radial acceleration is produced in the nucleation inception region. The X-ray observations showed that this helical stream let will persist as an essentially continuous ribbon of liquid if the boiler contains suitable (e.g., helical) surface convolutions.

Flow pattern correlations indicated that the stratified helical flow pattern observed was not necessarily unique to the particular flow conditions and boiler described herein. Similar flow patterns can probably be reproduced for a range of boiler sizes and flow rates. However, anomalous flow patterns are likely with mercury unless there is adequate liquid-to-wall contact (i. e., wetting).

The X-ray observations reported herein were augmented by the corrosion results seen in post-test metallurgical examination of the mercury boiler-superheater. The corrosive attack and deposit patterns added information that enhanced the in situ observations made by the X-ray system.

It was concluded that corrosive attack of the boiler surface and consequent transfer of wall material were responsible for increasing periods of metastable performance after the boiler had operated stably for several hundred hours. Boiling instabilities were probably produced by the formation and activation of alternative nucleation sites in the region of two-phase inception. The corrosion results seem to verify that erosion

figured significantly in triggering changes in the nucleation pattern. The pulse rate of consequent flow and pressure oscillations was related to the transit times associated with the boiler-superheater geometry and heat flux. The transit times and amplitudes of pressure pulses were closely linked with the geometry of the nucleation zone of the boiler inlet.

Lewis Research Center,

National Aeronautics and Space Administration,

Cleveland, Ohio, January 3, 1970,

120-27.

APPENDIX A

LOOP CONDITIONING PROCEDURE

Surface contaminants could inhibit mercury wetting and thereby impair performance in the boiler and superheater. A chemical cleaning procedure was used to remove surface contaminants from all loop components. Cleaning was done both before and after fabrication and welding of various components. This ensured that all internal surfaces were free of contaminants detrimental to the experiment and loop performance. All welding was done with an argon purge to ensure against contamination. To ensure against subsequent contamination, the loop was kept under argon backfill and then was charged with triple-distilled mercury under vacuum.

An additional procedure, termed "thermal conditioning, was used to induce mercury wetting in the boiler. The first step in this procedure involved a vacuum pumpdown of the loop and loading of mercury into the loop only after the pressure was less than 10-4 10 torr. Thereafter, the mercury in the boiler was heated to approximately 1100° F (595° C). In the boiler, this heating was initially done under zero flow conditions, and when the temperature reached approximately 1100° F (595° C), further heating was done with flowing mercury.

Evidence of mercury wetting by this thermal conditioning method was most noticeable in the electromagnetic pump. As explained in reference 16, the developed pressure in the pump was directly related to the degree of wetting. In the boiler, wetting was detected by observing the liquid meniscus as the liquid was withdrawn from the boiler under ambient temperatures. About 2 hours of thermal conditioning were sufficient to secure a significant increase in wetting in the boiler at temperatures between 1050° and 1100° F (568 and 595° C).

After initial wetting had been accomplished, the loop could be shut down, cooled to room temperature, and restored to preshutdown conditions within 15 to 20 minutes. This reaffirmed the evidence indicating relatively complete wetting prior to shutdown. Later, when the loop experiment was over, the loop was dissected. Examination of the inside surface of the boiler showed the anticipated effects of wetting.

Only slight dewetting occurred during shutdown in the absence of air leak-in. Previous experience with prototype loops indicated that dewetting would invariably occur whenever the wetted interface was reexposed to the air.

APPENDIX B

FLOW PATTERN OBSERVATIONS IN PREVIOUS MERCURY BOILERS

Boilers Without Helical Swirling Devices

Previous mercury loops were constructed with both straight and slightly curved cylindrical boiler tubes. These loops were operated under approximately the same conditions as the loop described in the text, namely, peak liquid temperature in the boiler inlet, about 1100° F (595° C); maximum superheater temperature, 1300° F (705° C); and mercury flow rate, 100 pounds per hour (12×10-3 kg/sec).

With a straight 1/4-inch- (0.63-cm-) diameter tube, X-ray observations showed droplets and sluglets (large, elongated, nonspherical particles) of mercury that were invariably torn away from the main liquid stream. These were swept by the highvelocity vapor into the superheater and into the vapor throttle. Although the boiler tube was slightly inclined from the horizontal, the droplets and sluglets easily skimmed along the boiler surface. When these particles of liquid entered the hotter portions of the superheater, they did not vaporize rapidly but tended to persist both in size and form. This was primarily attributable to the Liedenfrost effect (ref. 17).

In one previous loop formed from a straight cylindrical tube, liquid mercury was seen streaming upward along the entire length of a vertical section of a 1/4-inch(0.63-cm-) diameter tube comprising the final 2 feet (61 cm) of the superheater. In addition, it was found that liquid carryover occurred even in 1/4-inch- (0.63-cm-) diameter superheater tubes bent on a 4-inch- (10-cm-) diameter helix and with a total length exceeding 10 feet (305 cm).

The previously mentioned boilers confirmed the observation that with mercury there was a strong tendency for liquid particles to become entrained in the vapor. As a result, boiler and superheater performance degenerated appreciably; and wet vapor, often laden with large liquid particles, entered the vapor throttle.

Boilers With Helically Swaged Grooves

One of the prototype loops had a helically grooved boiler tube but contained no insert in the inlet. Another prototype boiler tube was also helically grooved but had a cylindrical insert installed in the inlet. These helically grooved boilers demonstrated how a helical liquid flow pattern can be produced and liquid carryover can be essentially eliminated. The X-ray observations obtained with these boilers are described in the following paragraphs.