(211b) Successes and Emerging Issues in Simulating the Mixing Behavior of Liquid-Particle Nuclear Waste Slurries at the Savannah River Site

Legacy nuclear waste generated at the Savannah River Site (SRS) during Cold War production of enriched uranium and plutonium is currently being processed into a stable borosilicate glass waste form for long term storage. The majority of the legacy waste is stored as a mixture of hydroxide and hydrous oxide insoluble solids in large cylindrical storage tanks at SRS. Over 90% of the waste solids are non-radioactive chemical byproducts derived from the fuel rod targets and purification chemistry. The 3.5-4.5 thousand cubic meter (900,000-1,200,000 gallon) carbon steel waste storage tanks also contain 5-7M sodium solutions rich in hydroxide, nitrate, and nitrite anions that are contaminated with soluble radioactive isotopes of cesium and strontium. Insoluble solids have been allowed to settle to the bottom of the tanks and have been aging for 25-50 years. These solids must be mobilized and transferred between tanks as the first step in waste treatment.

The list below summarizes some of the issues:

1) Rheological properties of suspended waste solids degrade with time under shear (the slurry becomes more viscous).

2) Rheological properties need to be understood as a function of the insoluble solids content and particle size distribution.

3) Slurry pump flow fields in the large, cooling coil filled, tanks have numerous stagnant zones where radioactive waste mounds form creating closure issues.

4) Settling times during large tank washing operations are not predictable. Settling times also do not mimic those of small scale radioactive settling test samples.

5) Radiolytic hydrogen generation and bubble accumulation during gravity settling constrains washing volumes.

6) Addition of glass frit to the slurries prior to the waste glass melter produces a shift in the rheological character of the slurry.

SRNL has a program that produces non-radioactive ?simulants? of the tank farm wastes in sufficient volumes, 50 to 5,000 liters, to study multi-phase mixing and transport issues without the radiation hazards of the real waste. Simulant slurries are used in small to intermediate scale equipment to evaluate the potential feasibility of various proposed process operations. Although simulant preparation chemistry generally follows that of the original radioactive wastes, the physical properties (rheology, particle size, cohesiveness, foaming tendency, etc.) often do not adequately match those of the actual wastes. The radioactive and simulant slurries generally behave like pseudo-homogeneous, thixotropic (shear thinning) fluids. Typical particles sizes for the precipitated solids are less than 40 micrometers, including a significant fraction of sub-micron sized particles.

Significant progress has been made in producing rheologically similar simulants. Testing has shown that batch precipitated simulants tend to be more viscous than those produced using continuous stirred tank crystallizers. High-shear mixers have been used to increase the apparent viscosity of simulants as necessary. Unfortunately, this can come at the expense of creating a mismatch in particle size distribution that may impact studies of other phenomena such as foaming or bubble retention. Actual waste rheological properties must be adjusted (by decanting or dilution) into certain ranges for successful tank slurrying and pump inter-tank transfers while retaining sufficient viscosity to minimize segregation of particles during anticipated worst case pump outage scenarios.

A fairly successful method for predicting the Bingham plastic yield stress and consistency of waste slurry blends was developed. This involves taking individual slurry rheometer data in the form of shear-rate dependent apparent viscosities, i.e. shear stress/shear rate, and applying an empirical mixing rule. The Kendall-Monroe rule has been used locally on the apparent viscosities (μ_i = shear stress/shear rate) instead of the Newtonian viscosities it was developed for:

μ_m^1/3 = x₁ μ₁^1/3 + x₂ μ₂^1/3

The x_i's are the mass fractions of the two slurries being blended together. The apparent viscosities of slurry 1 and slurry 2 at a given shear rate produce a predicted apparent viscosity of the blend at that shear rate which is converted into a shear stress for the blend, shear stress = μ_m*shear rate. The calculations are performed repeatedly over a range of shear rates. This produces a model rheogram (flow curve) for the blend slurry that can be fit to a rheological model such as the Bingham plastic equation.

Properties of the liquid-particle waste slurry change significantly in the final processing step just prior to the waste melter. Glass frit is added at a mass equivalent to about 2-3 times the mass of the waste solids. The frit provides the chemical backbone that holds the waste species in a stable glass waste form after processing through the melter. The frit particles are essentially fine shards of ground glass with particles in the 80-200 mesh range (0.07-0.20 mm) based on the Tyler standard screen scale. The impact of these particles on rheology is more comparable to adding 0.1 mm glass beads to water than to adding additional fine particles like the precipitated wastes. A fairly viscous slurry is required to ensure that the frit particles do not segregate relative to the waste particles during mixer outages and inter-tank transfers. SRS contractors are required by the U.S.-Department of Energy to verify that they are producing a stable homogeneous melter feed slurry with acceptable waste form properties so that direct sampling of the poured glass stream is not required.

Progress has been made on modeling the melter feed rheology as a Bingham plastic fluid (with the fine solids) in two-phase flow with the larger frit particles using a variation of the extended Einstein equation. The rheological properties of the frit-free slurry are obtained in the form of an apparent viscosity (as a function of shear rate) and used with the volume fraction, φ, of the frit particles to calculate the shift in viscosity from the entrained large particles:

μ_{melter feed} = μ_sludge*(1 + 2.5*f*φ + 10*φ²)

Frit can occupy 7-14 volume percent of the melter feed slurry. The factor, f, was introduced to the Einstein equation to account for deviations from spherical particles. The primary use for this equation is in predicting the impact of changing the ratio of frit to waste solids on rheological properties without having to repeat series of rheometer measurements on different compositions of melter feed at different total solids loadings.

The above model was applied to simulated waste rheological data prior to frit addition. A reasonable match was obtained to rheological data following frit addition (11 volume percent frit) with f = 1.8. An on-going program is currently evaluating the impact of replacing the glass frit with comparably sized glass beads. Improved rheological properties in the slurries with beads may permit the melter feed to be successfully processed with reduced water content (higher waste content).

An emerging area of study is an effort to quantify and model the retention of bubbles in the settled solids in the waste tanks. The bubbles are H₂-O₂ mixtures from radiolysis of water. These gas molecules coalesce into bubbles within the settled liquid-particulate mass in the bottom region of the storage tanks. Molecular diffusion is not able to keep up with the radiolytic generation rate. The settled solid matrix appears to play a key role in retaining the bubbles. Rapid release of trapped bubbles occurs once slurry pumps are inserted and started in a tank containing settled solids. These bubbles can briefly lead to the formation of a flammable or explosive atmosphere over the tank contents, and that is unacceptable for nuclear waste storage. It is believed that the retention of bubbles is not solely due to the more viscous properties that would prevail in the settled slurry due to the higher volume fraction of solids content than when the tank is fully mixed. The particles themselves are thought to try to orient at the bubble-slurry interface in such a manner that the bubbles become anchored to the settled matrix. Fundamental studies are needed to better understand the forces acting to prevent bubble coalescence and release.