C. A. Nash, S. W. Rosencrance, and W. R. Wilmarth
Westinghouse Savannah River Company
Aiken, SC 29808

P. S. Townson
BNFL, Inc.
Richland, WA 99352

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

The Department of Energy is sponsoring the River Protection Project (RPP) at the Hanford site in order to pretreat and vitrify nuclear waste. Processing of Hanford complexant-bearing aqueous waste requires pretreatment to remove specific radionuclides. This removal is accomplished by a strontium nitrate/sodium permanganate precipitation and filtration process. Rheology, particle size, and filtration measurements that were made with actual Hanford waste are presented in this paper.

A shielded and remotely operated Cells Unit Filter was used to provide crossflow filtration with a Mott 0.1 micron 0.61-meter long 0.00953-meter ID filter tube. Entrained solids alone (0.1 weight percent) provided poor filter fluxes, of the order 0.0068 l/(m²*s) under the range of conditions tested. The mixture of strontium carbonate and manganese oxide/hydroxide solids formed by the precipitation acted as a filter aid and allowed practical filtration rates.

Filtration fluxes for the precipitate with entrained solids included ranged from 0.027 l/(m²*s) to 0.1 l/(m²*s). Transmembrane pressures were 207 to 483 kPa and crossflow velocities were 3 to 5 m/s. Length of operation (cumulative shear) was found to be the most important variable affecting filter flux. Shown in the figure below are the runs in chronological order. The lowest filter fluxes were obtained during the first run containing entrained solids only. All subsequent runs were performed using the batch after precipitation and demonstrated improved filter fluxes.

Particle size measurements show that the slurry as precipitated had a wide range of particle sizes with a volume average size of 40 microns. Flow shear during crossflow filtration reduced the average to 10 microns with further shear not reducing the average much more than that.

Waste was precipitated and concentrated by filtration to various insoluble solid loadings. Rheology measurement were performed between 10 and 40 Celsius for the initial entrained solids mixture and various precipitate slurries up to 16 weight percent insoluble solids. The data are well described by the Bingham plastic model. Apparent viscosity’s ranged from 2 to 28 centipoise with yield stresses less than 100 dynes/cm2 at shear rates between 250 and 1500 reciprocal seconds. The data can be well described by a single empirical function of temperature, shear rate, and weight percent insoluble solids.

This work provides important confirmation of the new process to achieve both acceptable filterability and decontamination for wastes to be treated by the Hanford River Protection Project. Work continues on the bench and pilot scales to further understand and improve processing of Hanford waste.

Keywords: Filtration, Cross-flow, Radioactive, Hanford, River Protection Project

The U.S. Department of Energy is sponsoring the River Protection Project (RPP). Its goal is to pretreat and vitrify nuclear waste currently stored at the Hanford site in the state of Washington. The Washington Group International is under contract to support the RPP. Some of the research and development including the current work is being done at the Savannah River Technology Center at the Savannah River Site in South Carolina.

Some of the alkaline liquid wastes at Hanford contain complexants like ethylene diamine tetraacetic acid (EDTA) and gluconate ion, causing transuranic species like plutonium and americium to be soluble. These alpha-emitting species need to be removed so that the bulk of the waste volume can be vitrified as low-activity glass. The process for removing and concentrating these species involves co-precipitation into a slurry of manganese oxides that is then separated by crossflow filtration. Rheology, particle size, and filtration measurements that were made with actual Hanford waste are presented in this paper.

Applications of crossflow filtration have reached an industrial prominence in only the past few decades (Porter, 1972). Crossflow filtration is an applicable tool for the slurry concentration and washing steps because it is able, unlike the more familiar deadend filtration, to handle high solids loadings on an almost continuous basis. In crossflow filtration a concentrate flow at high pressure to the filtrate side of a porous medium is used to sweep away a continuously forming cake of solids. For most applications prediction of the filter performance comes must be taken from experiments with the specific feed under consideration (Murkes and Carlsson, 1988). Advantages of crossflow filtration include applicability to difficult applications, suitability for slurry thickening, and total particle rejection (Murkes, 1986). While crossflow applications are often run without filter aids, the use of filter aids can increase filter cake permeability (Milisic and Bersilion, 1986).

The work reported here shows the utility of crossflow filtration for the precipitation of actual liquid waste from Hanford tank 241-AN-102. The particle size and rheological characterization of the slurry from the precipitation process is also presented. Similar work with this chemistry and with simulants is reported separately (Duignan, 2000a). The Savannah River Technology Center (SRTC) is one of only a few locations in the country where such characterizations and testing can be performed with highly radioactive samples.

Crossflow filtration was performed with a Cells Unit Filter (CUF) rig that was set up in the Shielded Cells at the Savannah River Technology Center. The unit was designed to process a small sample of radioactive material (inventory of less than 1 liter) at plant-typical pressures and crossflow velocity. Figure 1 shows the unit without cooling tubes connected to the heat exchanger for clarity. Feed from the reservoir at the left went to a progressive cavity pump. The pump was operated at variable speed by controlling air pressure to the air motor that drives it. Liquid was pumped through a magnetic flowmeter and heat exchanger that removes pump heat. It then passed down the center of a crossflow filter of 0.61 m (2-ft) porous length. A throttle valve downstream drops fluid pressure back to atmospheric.

Filtrate was measured with a sightglass and stopwatch. A simple backpulse system could be charged with filtrate. Compressed air stored in the filtrate chamber forced reverse flow upon the filter medium. Standard Bourdon-type pressure gauges indicated pressure. A thermocouple mounted near the bottom of the reservoir measures slurry temperature directly. Details of the CUF are documented on engineering drawings.

The filter unit used in this test was manufactured by the Mott Metallurgical Corporation. The filter and housing are a single piece of welded construction. It had the following characteristics:

Figure 2 shows the filter and housing arrangement. Concentrate at several hundred kilopascals relative to the housing shell was sent down the tube side of the porous filter at 3 to 5 m/s. Filtrate weeping through the porous tube wall was recycled back to the feed reservoir most of the time, though filtrate was also sampled and flow measurements were taken.

The waste sample to be processed had been taken from Hanford Tank 241-AN-102. This is one of the tanks containing complexants like EDTA that cause increased solubility of transuranics. A small subsample was diluted to 6.4 M total sodium for analysis. Hay et. al., 2000, performed the characterizations shown here. Tables 1 and 2 show the main species. The sample contained less than 0.1 wt% insoluble solids.

The following steps cover dilution of the larger sample to approximately 6 M sodium, caustic adjustment, and precipitation. The portion of larger sample received here was initially at 7 M sodium. It was run in the crossflow filter as is before the recipe below was applied.

1. Measure 1.2 liters of 241-AN-102 sample into a large Erlenmeyer flask.
2. Heat and stir the liquid, target temperature being 50°C.
3. Slowly add 200 ml dilution water to bring the sodium level to 6 M.
4. Slowly add 66.65 ml of 17 M NaOH solution.
5. Heat and stir the vessel with a Teflon magnetic stir bar and hot plate. After 50°C is reached, slowly add 108 ml of 1 M strontium nitrate. Actual addition time was 4 minutes.
6. Wait 10 minutes during stirring at the 50°C temperature.
7. 72 ml of 1 M sodium permanganate solution was added over a period of 5 minutes.
8. Continue stirring for four hours while maintaining 50°C.
9. Slurry was added to the CUF very slowly while the CUF heat exchanger was on to provide cooling. The material was cooled to room temperature in a single pass because of the slow addition.

The finished slurry is about 2 wt% insoluble solids. The caustic addition was meant to keep the pH high for this solution to ensure that aluminum does not precipitate. The strontium nitrate addition causes the precipitation of strontium carbonate. The permanganate oxidizes organic matter in the waste feed, causing manganese oxides/hydroxides such as manganese dioxide solids to form. The strontium carbonate and manganese oxides slurry is ready after cooling for crossflow filtration.

The purpose of the precipitations are to reduce activity of strontium-90 and transuranics in the liquid phase of the slurry but further discussion of the chemistry is beyond the scope of this paper.

The rig was initially cleaned and flushed with prefiltered water containing 0.01 M NaOH several times. The caustic was present to prevent precipitation of aluminum hydroxide when feed was introduced. The rig was then drained.

Initial testing was done using the 241-AN-102 sample as received, before precipitation, to see how easily the entrained solids (fines) filtered. The rig was then cleaned and prepared for precipitate slurry. The steps below apply to any feed, either before or after precipitation.

Feed was introduced to fill the reservoir about 2/3 of the way up as indicated by the sight glass on the reservoir. Initial flow was established with the slurry throttle valve fully open.

Operation of the CUF for each set of filter flow and pressure conditions involved the following routine:

1. Filtrate was generated to fill the backpulse chamber. The chamber was air-pressurized to provide 300 kPa (gauge) overpressure. Filter concentrate pressure was reduced to a less than 30 kPa gauge for best backpulse effectiveness.
2. Two backpulses were performed before each set of conditions was run.
3. Conditions were set after the second backpulse while the filtrate valve was kept closed. Pump speed and slurry throttle valve position are coordinated to reach desired pressure and flow rate.
4. Each run started when the filtrate valve was opened slowly. Slowness was controlled by watching reduction of filtrate side pressure while the needle valve was opened – slow pressure reduction was desired. Filtrate side gauge pressure is zero during filter operation.
5. Table 3 below shows the set of conditions that were run. Conditions were maintained for at least30 minutes after the initial backpulsing.

Particle size of slurries were found by exiting small slurry samples from the Shielded Cells and using a Microtrac instrument. Particle size was determined by a laser scattering technique.

The rheology measurements were performed using a Haake RV30/M5 control/head combination and the NV sensor. The complete unit had been installed into the Shielded Cells. Installation was done in a way to allow remote operation in the high radiation field (all electronic parts were outside of the cell). An aliquot of 9ml of material was placed in the thermally jacketed sample cup and allowed to equilibrate for 30 minutes at each temperature of analysis. The sample was ramped from 0-2700 s^-1 over three minutes followed by a 30-second constant shear-rate hold time. The final segment of the analysis sequence was a three-minute linear ramp from 2700-0 s^-1.

As a matter of practice all data analysis involved a blank subtraction wherein the rotor was operated and a rheogram collected with all parameters identical to the sample analysis except the rotor remained suspended in the air. The shear stress and viscosity as the result of air is negligible and thus the average slope of the blank stress-strain curve was subtracted from all subsequent sample analysis. As a matter of record this method yields reproducible results (precise) as well as an accurate representation of low consistency samples. This method is not necessary for samples having higher feedback.

Figure 3 shows the flux data for the conditions that were run in the campaign. The first run was of feed before precipitation. It was found that filter fluxes were very low because of the presence of the entrained insoluble solids. While the solids level was less than 0.1 wt% of the feed their impact on the filter flux was severe. Some of these solids were captured on a deadend filter in a separate analysis. They were found to contain 28 wt% aluminum on an elemental basis (Hay et. al., 2000). Chromium (2.1 wt%), iron (1.4 wt%), and sodium (2 wt%) were the other elements found by inductively coupled plasma-emission spectroscopic analysis of acid-dissolved material. Analytical error was 10% for these measurements. The small amount of entrained solids available prevented particle size measurement.

Before the current process of using reduced permanganate to remove transuranics there was work done on a precipitation process using ferric nitrate as a precipitant instead. The resulting ferric hydroxide slurry made with complexant organic-containing caustic feeds was found to be unfilterable by either crossflow or deadend filtration after the slurry experienced shear. The shear provided by a standard or progressive cavity pump was enough to cause the problem. Fluxes from attempted filtration of such material were identically zero (Nash et. al, 2000). It is believed that the combination of organic materials with the caustic solution caused the ferric hydroxide to be unfilterable. Similar entrained solids in alkaline solutions without the organic content were found to filter well (Duignan, 2000b).

Figure 3 above shows that the strontium-manganese slurry filtered much faster than the entrained solids alone despite the higher solids loading for the precipitate. It is clear that the precipitate acts like a filter aid and masks the filter fouling that entrained solids alone will cause. The data also show that the flux from slurry filtration is consistently above 0.03 l/(m²*s) and that backpulsing can increase it above 0.06 l/(m²*s).

Figure 3 shows how filter flux is mostly a function of run time rather than conditions. The legend is listed in chronological order with the first run at the top. The entrained solids data are the lowest curve on the graph. The first slurry run data form the highest curve. A linear statistical analysis of average flux data within each condition also shows this to be true. Runtime was significant and a meaningful linear model was:

Average flux (l/(m²*s)) = 0.68*[(0.14+/-0.022) – (0.0109+/-0.00169)*(runtime, hours)]

The linear coefficients have standard estimates of error as shown. The first slurry run was omitted because it was so much higher than following runs. Nonlinearity it introduced would not have led to a statistically significant linear model.

Particle size measurements were made on the precipitate slurry before pumping or filtration (initial), after approximately four hours of filter testing (intermediate), and after 7 hours of filter testing (final). Figure 4 shows the results.

It is clear that the strontium carbonate/manganese oxide slurry particles had a bimodal distribution before pump shearing. A significant amount of the solids were in particles of 50-100 microns with the rest being less than 7 microns in size. The average micron size calculated was 40. The hours of pumping and filtration that followed collapsed the size distribution into one broad with average calculated micron sizes of 10. The sample exposed to further pumping exhibits a small amount of further size reduction.

Smaller particle sizes are expected to reduce filterability (Tarleton and Wakeman, 1993, Lojkine et. al, 1992). However, nature of the particle played an important role since the entrained solids as well were so hard to filter. Size distribution and nature of the particles were found to be the major factors in time-dependent flux decline of mineral slurry by Baker et. al. (Baker et. al., 1985).

A program to measure the rheology of the slurry from the precipitation process used material that had been run in the CUF (Rosencrance et. al., 2001). This was considered to produce more realistic data partly because particle size would have the single distribution that was characteristic of sheared material.

Figure 5 shows data from Tank 241-AN-102 feed before precipitation, slurry without concentration (2 wt%) and slurry that had been concentrated in the CUF. All the slurry data that is shown provided the following equation:

Error bars on each data point were calculated for a total of nine rheometer runs at each condition. The slurry behavior is best described as Bingham plastic. The feeds before precipitation were Newtonian.

Crossflow filtration and characterization of Hanford Tank 241-AN-102 feed, a caustic liquid with soluble organics, provided data to conclude the following:

1. Feed with 0.1 wt% or less of entrained solids produced relatively low filter fluxes.
2. Strontium carbonate/manganese oxide slurry was much easier to filter despite the higher solids loading.
3. Runtime (shearing) was the only significant crossflow filter parameter for average flux under the range of conditions tested. However, the fact that the backpulsing between run conditions caused some flux restoration showed that some filter cake fouling also played a role.
4. Particle size data quantitatively showed how pumping shear reduced particle size.
5. Rheological data followed expected trends and show that the strontium carbonate/manganese oxide slurry was about 7 cP under the crossflow conditions that were run.

1. Baker, R. J., Fane, A. G., Fell, C. J. D., and Yoo, B. H., 1985 Factors affecting Flux in Crossflow Filtration, Desalination, 53, 81-93.
2. Duignan, M.R., 2000 Final Report: Pilot-scale Cross-flow Ultrafiltration Test using a Hanford Site Tank 241-AN-107 Waste Simulant – Envelope C + Entrained Solids + Strontium-Transuranic Precipitation, Westinghouse Savannah River Company, Report number BNF-003-98-0226.
3. Duignan, M.R., Nash, C.A., Townson, P.S., 2001 Cross-flow filtration with a shear-thinning organic-based slurry. 4^th Int. Conf. Multiphase Flow, New Orleans, May 27 – June 1, 2001. [Paper 303]
4. Hay, M. S., Bronikowski, M. G., Hsu, C. W., and White, T. L., 2000 Chemical Characterization of an Envelope C Sample from Hanford Tank 241-AN-102, Westinghouse Savannah River Company, Report number BNF-003-98-0250.
5. Lojkine, M. H., Field, R. W., and Howell, H. A., 1992 Crossflow Microfiltration of Cell Suspensions: A Review of Models with Emphasis on Particle Size Effects, Transactions of Industrial Chemistry and Engineering, 70, Part C, 149-164.
6. Milisic, V., and Bersillon, J. L., 1986 Anti-Fouling Techniques in Crossflow Microfiltration, Filtration & Separation, Nov/Dec, 347-349.
7. Murkes, J. 1986 Low-shear and High-shear Crossflow Filtration Filtration & Separation, Nov/Dec, 364-365.
8. Murkes, J. and Carlsson, C.G., 1988. Crossflow Filtration. John Wiley & Sons Ltd., (ISBN-0-471-92097-5).
9. Nash, C. A., Rosencrance, S. W., Walker, B. W., and Wilmarth, W. R., 2000 Investigation of Varied Strontium-Transuranic Precipitation Chemistries for Crossflow Filtration, Westinghouse Savannah River Company, Report number BNF-003-98-0171.
10. Porter, M. C., 1972 Concentration Polarization with Membrane Ultrafiltration Industrial and Engineering Chemistry Product Research and Development, 11, 3, 234-248.
11. Rosencrance, S. W., King, W., and Nash, C. A., 2001 Physical Characterization for Part B1, Westinghouse Savannah River Company, Report number WSRC-TR-2000-00026.
12. Tarleton, E. S., and Wakeman, R. J., Understanding Flux Decline in Crossflow Microfiltration Part 1: Effects of Particle and Pore Size, Transactions of Industrial Chemistry and Engineering, 71, Part A, 399-410.