Cross-flow Ultrafiltration Scaling Considerations

One legacy of the nuclear age is radioactive waste and it must be stabilized to be stored in a safe manner. An important part of the stabilization process is the separation of radioactive solids from the liquid wastes by cross-flow ultrafiltration. The performance of this technology with the wastes to be treated was unknown and, therefore, had to be obtained. However, before beginning a filter study the question of experimental scale had to be addressed. Of course, carrying out experiments using full-size equipment is always ideal, but rarely practical when dealing with plant size processes. Flow loops that will handle millions of liters of slurries, which are either highly caustic or acidic, with flow rates of 10,000 lpm make full-scale tests prohibitively expensive. Moreover, when the slurries happen to be radioactive such work is also very dangerous. All of these considerations lend themselves to investigations at smaller scales and in many situations can be treated with computational analyses. Unfortunately, as scale is reduced it becomes harder to provide prototypic results and the two and three phase multi-component mixtures challenge accurate computational results. To obtain accurate and representative filter results the use of two scales were chosen: (1) Small-scale--would allow the testing with actual radioactive waste samples and compare results with simulated wastes that were not radioactive. For this scale the feed tank held 6 liters of waste and it had a single cross-flow filter tube 0.61 m long. (2) Pilot-scale--would be restricted to use simulated non-radioactive wastes. At this larger scale the feed tank held 120 liters of waste and the filter unit was prototypic to the planned plant facility in pore size (0.1 micron), length (2.29 m), diameter (0.0127 m inside and 0.0159 m outside diameter), and being multi-tubed. The small-scale apparatus is convenient, easy to use, and can test both radioactive and non-radioactive wastes; therefore, there is a larger database than at the pilot scale. In fact, the small-scale data are very useful to compare actual waste to simulated waste filter performance to validate a simulant, but data availability does not mean they accurately represent full-scale performance. Results indicate that small-scale filter fluxes to be significantly higher that those at the pilot scale. In an attempt to study the difference in filter performance at the two scales an experiment was done that used exactly the same simultant which was created at the same time so that issues of composition and aging would not compromise the results. This paper will discuss those experimental results, as well as those from a computational fluid dynamics model to better understand the small-scale limitations