H. N. Guerrero
Westinghouse Savannah River Company
Aiken, SC 29808

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

This report has been reproduced directly from the best available copy.

Available for sale to the public, in paper, from:  U.S. Department of Commerce, National Technical Information Service, 5285 Port Royal Road, Springfield, VA 22161,  phone: (800) 553-6847,  fax: (703) 605-6900,  email:  orders@ntis.fedworld.gov   online ordering:  http://www.ntis.gov/support/ordering.htm

Available electronically at  http://www.osti.gov/bridge/

Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy, Office of Scientific and Technical Information, P.O. Box 62, Oak Ridge, TN 37831-0062,  phone: (865 ) 576-8401,  fax: (865) 576-5728,  email:  reports@adonis.osti.gov

Siphon tests of molten glass were performed to simulate potential drainage of a radioactive waste melter, the Defense Waste Processing Facility (DWPF) at the Savannah River Site. Glass is poured from the melter through a vertical downspout that is connected to the bottom of the melter through a riser. Large flow surges have the potential of completely filling the downspout and creating a siphon effect that has the potential for complete draining of the melter. Visual observations show the exiting glass stream starts as a single-phase pipe flow, constricting into a narrow glass stream. Then a half-spherical bubble forms at the exit of the downspout. The bubble grows, extending upwards into the downspout, while the liquid flows counter-currently to one side of the spout. Tests were performed to determine what are the spout geometry and glass properties that would be conducive to siphoning, conditions for terminating the siphon, and the total amount of glass drained.

The pouring of a liquid from a tank to the atmosphere via a vertical tube is often times encountered in industrial applications. In the Defense Waste Processing Faciltiy (DWPF) Melter, molten glass, incorporating radioactive waste for eventual storage, is poured in this manner. A 432 mm long pour spout is connected to the bottom of the melter via an angled transition tube, just as in a tea pot. Pouring is initiated by applying a small over pressure on top of the molten glass pool. Under normal conditions, molten glass flows down one side of the pour spout tube as a falling film. Safe control of the pouring process is insured by removal of this overpressure. However, if the tube accidentally becomes full, e.g., due to a flow surge, there is a potential for siphoning the entire inventory of glass in the melter. There could be environmental risks in the event of such an accident.

Flow visualization with glycerin as a simulant for liquid glass has shown the downward flowing liquid interacts with a trapped air bubble at the top of the spout and an incipient air bubble at the bottom opening. Towards the later part of the draining event, a cylindrical slug of air rose up to the top of the spout to break the siphon. To prevent siphoning, it is important to know the effect of such two-phase flow conditions, along with design parameters, i.e., tube diameter and length, and glass operating conditions, i.e., viscosity, on the initiation of siphoning. It is also important to know what conditions end siphoning and affect the total amount of liquid transferred in order to assess the extent of potential environmental consequences. Since molten glass is rather unique in having both a high viscosity and high surface tension (e.g., 7 N-s/m² and 0.3 N/m, respectively at 1000^oC), an experimental program was initiated to simulate siphoning of glass. Further, due to the difficulty of viewing the glass flow through windows, flow visualization tests were performed with glycerin to interpret the results of the glass siphon tests.

The Full Scale DWPF Pour Spout Mockup is shown in the schematic drawing of Figure 1. This consists of an 203 mm diameter reservoir , which simulates the 1.83 m. diameter melter, a full scale 102 mm diameter riser, and the full scale pour spout section. The pour spout section was fitted with interchangeable inserts that made the flow passage into one of the following configurations:

• 50.8 mm ID, 432 mm long
• 50.8 mm ID, 228 mm long, 38 mm ID, 203 mm long,
• 50.8 mm ID, 432 mm long.

Glass was melted in place inside the reservoir. The reservoir, riser, and pour spout had external heaters for melting glass and maintaining a desired temperature. Glass frit was added to reservoir until molten glass started to spill over into the pour spout section. A plug was then inserted into the bottom opening of the pour spout and held in place by a spring mechanism. Additional glass frit was melted to fill the pour spout until the level was at least 12.5 mm above the bottom of the riser to pour spout transition pipe. The test was started by pulling the plug off the pour spout opening.

A load cell measured the weight of the collected glass inside a catch drum. The pressure in the air space at top of the pour spout was measured with a Rosemount pressure transmitter. The glass draining out of the bottom of the insert was imaged and recorded with a video camcorder.

Sixteen tests were performed. These are listed in Table 1, including the insert tested, glass temperature and total glass drained.

During the glass tests, a video camera was directed upwards at the spout bottom opening to record the downward flowing glass stream. Figure 2 provides a typical sequence of images from the 38 mm ID spout, Run 15. Figure 2a shows the bottom plug prior to initiation of the test. The initial high flow is shown in Figure 2b, giving the appearance of a symmetrical, constricting liquid stream. This was followed, Figure 2c, by movement of the stream to the side. In Figure 2d, a half-spherical bubble starts to form beside the offset cone of glass. Then in Figure 2e, the bubble penetrates all the way to the air space at the top of the pour spout. Figure 2f shows the glass collecting to close the insert bottom opening, which was followed by a series of openings and closings.

*includes 229 mm long upper section of 50.8 mm ID spout

Figure 2. Video Images of Glass Stream Exiting 36.1 mm ID Spout - Run 15

Mass of Glass Drained

The general trend from Table 1 is that total mass of glass transferred during siphoning, listed in the last column, decreases as the diameter of the pour spout increases. Also, the lower the temperature, and thus the viscosity, the larger is the amount of glass drained. This trend is not universal as shown for the case of Runs 10 and 11. These two runs had the same nominal test conditions (spout dimensions and glass temperature), but the mass drained in Run 11 was 50% higher than in Run 10.

Typical measurements of the draining mass flow rates and pressures at the top of the pour spout are given in Figure 3 for Runs 10, 11, and 15. The weight of the glass collected as a function of time was fitted with a polynomial curve and then differentiated to give the mass flow rate curves in Figures 3a, b, and c. The initial mass flow rate was higher for a large diameter spout as compared to a small diameter spout. However, the total amount of glass drained may be higher for the small diameter spout as the duration of draining was longer than for the larger diameter spout. The pressure transients for Runs 10 and 11 differed significantly for nominally the same test conditions, where the maximum negative pressure recorded in Run 11 was 53% higher than in Run 10.

Siphoning tests similar to the glass tests discussed earlier were performed in a geometrically similar test facility as that shown in Figure 1. The pipe components however were made of clear Plexiglas to afford the means to visualize the flow. The draining flow was collected in a drum and measured using a weight scale. The weight scale had an electronic output connected to a computer data acquisition system. The pressure at the top of the pour spout was also measured with the same pressure instrument as in the glass tests. The glycerin with a small amount of water had a density of 1200 Kg/m3 and a viscosity of 0.61 N/s-m².

Flow visualization with glycerin showed the following general trends during draining. The run with the small diameter spout had the largest mass of glycerin drained. For the same spout dimensions, the liquid level height, Hw, above the overflow point on the horizontal channel connecting the riser and spout had a significant effect on the total mass drained. The drained mass increased with increasing liquid level.at the start of the siphoning transient.

Flow visualization showed that for the run with small diameter spout (GLY-3), the liquid level above the overflow point stayed relatively constant until the lowering level in the reservoir was in the range of 75 mm above the exit of the spout. Thus, the liquid flowed without any influence of the trapped air volume above the overflow point. Then, as the flow rate decreased to a low value, a cylindrical air slug with a spherical upper surface rose up quickly penetrating the downward flowing liquid. When, the slug reached the top and merged with the upper air volume, a slugging action occurred until there was a clear air path from the top of the pour spout to the outside atmosphere. The siphoning then stopped.

For the larger diameter spout, visual observations are illustrated in Figure 4. This figure shows that immediately after the plug at the bottom of the spout was pulled, the liquid level, Hs, in the vertical spout dropped to a low value of 152 mm above the exit end of the spout in run GLY-5. The glycerin was falling from the overflow point down the side of the spout as a film with a free surface (Figure 4a). As the flow progressed, the liquid level in the spout started to rise. At a critical low value of the draining flow, a cylindrical slug started rising from the bottom end (Figure 4b). In essence, a liquid slug was rising ahead of the air slug. The length of the liquid slug was decreasing as liquid constantly drained counter-currently against the rising air slug. When the air slug broke through the liquid slug to the air space at the top of the pour spout, the siphon was broken. Similar behavior occurred in run GLY-6, where the minimum liquid level in the spout reached was 228 mm.

The measured mass flow rate for run, GLY-3, is given by Figure 5a, and the measured pressure in the top air space by Figure 5b. Knowing the mass flow rate, the pressure at the top of the pour spout, P_top, in the air space can be calculated assuming a liquid filled spout by Equation 1.

Figure 4. Flow Visualization Results for Glycerin Tests

Figure 5. Mass Flow Rate and Air Space Pressure Measurements for Glycerin Run GLY-3

Figure 6. Mass Flow Rate and Air Space Pressure Measurements for Glycerin Run GLY-5

This equation is plotted in Figure 5b for run GLY-3 together with the experimental pressures. The good agreement between experimental and calculated values confirms the assumption of a liquid filled spout. Draining of the reservoir therefore can be predicted on the basis of single phase flow. The trapped air volume at the top of the spout has no effect on the flow because the high frictional resistance of the small diameter spout keeps the spout full.

Figure 7. Observed Liquid Slug Height in the Pour Spout during the Run GLY-5

The measured mass flow rates and pressures for run GLY-5 for a larger spout diameter (40 mm) than for run GLY-3 (25.4 mm) are given in Figures 6a and 6b. Assuming a liquid filled spout would lead to a predicted spout top pressure that would deviate greatly from Figure 7. However, using the observed liquid height in Figure 7, the calculated pressure obtained with Equation 1 agrees reasonably well with the experimental values, as shown in Figure 6b.

A small trapped air mass after the postulated flow surge implies that its pressure can be reduced greatly under the action of the draining flow without a large absolute change in volume. The reduced spout top pressure is accommodated without dropping the liquid level into the spout region. Thus, the trapped air volume does not influence the draining process. The draining will proceed as in single phase flow until the level in the reservoir is slightly above the exit end of the spout.

A large trapped air mass expands under the action of the reduced pressures due to draining until a pressure equilibrium between the air space and draining liquid is reached. The expanding air mass intrudes into the spout region, reducing the static head in the spout. Thus, the overall driving head from the liquid surface at the reservoir to the bottom end of the spout is reduced and draining will end at a higher reservoir level. As the flow decreases to a low level, the air slug at the spout bottom entrance rises and with it the liquid slug. This explains why a larger amount of liquid is drained when the initial liquid level, Hw, is high than for the case of a low initial level. A low level implies a large trapped air volume and conversely, a high initial level implies a small trapped air volume.

The radically different amounts of glass drained during Run 10 and Run 11, where spout dimensions and operating temperature were the same, appear to be due to this effect. The pressure traces are similar to those in the glycerin test GLY-5. This can only be surmised to be the reason since the initial glass level above the overflow point was not measured.

The effect of an initial vacuum pressure applied to the top of the pour spout in Runs 6,7 and Run 16 also can be explained on a similar basis as above. The total glass drained in Runs 6, 7 were higher than for Run 5, without an initial vacuum. However in Run 16, the mass of glass drained did not differ significantly from that of Run 15, without a vacuum. It may be that the initial level in Run 15 was already sufficiently high that a higher level had no increasing effect. In fact the 28 Kg drained for these runs was the maximum obtained during the tests and is close to the mass in the reservoir with a height equal to the spout length.

A large spout diameter allows expansion of the trapped air volume into the spout without significantly reducing its pressure. Thus the differential pressure from the atmosphere to the top of the spout is small compare to when the spout is liquid filled. This again results in a reduced static head in the pour spout, which tends to end the siphoning.

The effect of viscosity appears primarily to be on the initiation of siphoning. The tendency of the spout to become full at a low flow is higher for a more viscous liquid compared to a less viscous liquid. This was not considered during these tests because a flow surge was not simulated. The assumption was a completely full spout for initial test conditions. Viscosity however does not appear to affect the total amount of liquid drained as seen from the test results with glass given by Table 1. Comparing results of similar test runs except for glass temperature is not valid because the liquid level above the overflow point for these tests is not known. From a theoretical standpoint, Equation1 shows that a flow will continue as long as the pressure at the top of the pour spout has not reached atmospheric conditions.

Siphoning tests were performed with molten glass to determine the total amount of glass drained from a melter in the event of a flow surge. Visual observations showed that after the initial high flow, the exiting glass stream flowed downwards against a rising cylindrical slug of air. The siphoning stopped when the continuous slug of air reached the top of the pour spout. A similar behavior was observed during siphoning tests with glycerin. These tests provided insight into the physical mechanisms going on during draining of a laminar fluid. Analysis of the glycerin tests showed the importance of the mass of air trapped at the top of the pour spout and also of the diameter of the spout. Viscosity of the liquid appears to be an important parameter only to a limited extent both from the experimental results and from a theoretical stand point. To prevent siphoning, a

large diameter spout and provision for insuring a large trapped air mass are required.