Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory (1999)

The high-level waste (HLW) calcine consists of powdered ceramic particles of various sizes and chemical compositions. The range of properties is the result of compositional differences of the inputs (namely, the reprocessing waste and chemicals added to this waste "feed" solution) to the calciner. During fluidized bed calcination, particles (usually dolomite) in the fluidized bed were coated multiple times with layers of compounds from the waste stream.¹ The high heat caused gaseous species (e.g., water and some nitrates) to evolve. The output is granular calcine particles with a range of physical, chemical, and radiological characteristics.

These properties affect the operations discussed in this chapter. Any treatment of the HLW calcine requires adequate characterization and retrieval from the bin sets where it is currently stored. Blending of various calcine types is projected to occur. Aqueous-based processing can only proceed if the calcine is dissolved in acid solution. These important characterization, retrieval, blending, and dissolution steps are discussed in this chapter.

As stated in Chapter 1, after the HLW was convened to calcine, it was transferred pneumatically to large cylindrical stainless steel bins for storage. Several such bins comprise a bin set located within a single concrete containment structure, or vault. There are six bin sets containing calcine and one that is empty. The total volume of the six bin sets in use is approximately 180,000 ft³ (5,100 m³), and the volume of calcine contained in them is in excess of 130,000 ft³ (3,700 m³)² (Palmer, 1996; INEEL, 1997; Lopez and Kimmett, 1998). The calcine varies substantially in composition, both between bins and in different regions within a given bin. However, most calcine composition data are deduced from calciner feed data. Some calcine characterization data are based on pilot tests with cold simulants. There are few data for actual calcine, and only two samples of stored calcine have been retrieved from bins, collected in 1979 (Staples et al., 1979; Garcia, 1997).

The primary risk from retrieval, with respect to both environmental cleanup and health effects, appears to result from the possible release of radioactive and toxic particles. Dry calcine contains particles of various sizes, some of the right size to constitute a significant respirable hazard. Therefore, appropriate measures will have to be taken to ensure that releases cannot occur during sampling, installation of equipment for retrieval, and the actual

retrieval and subsequent handling. These operations are complicated by the fact that the bins were not specifically designed to provide for sampling or retrieval operations.

The characterization data provided to the committee were primarily from a report (Garcia, 1997) that was found by the committee to contain numerous errors (see the Adequacy of Existing Information section later in this chapter), to the point that its credibility and usefulness are questionable. However, because of the lack of alternative data, those in Garcia (1997) are used for most of the following discussion. Other data sources are specifically noted where discussed.

Calcine properties are deduced almost entirely from three sources of data: (1) samples of the solutions fed to the calciner, (2) cold (i.e., nonradioactive) calciner pilot plant testing with simulant feed, and (3) grab samples of real calcine. The majority of the work was done several years ago and does not include aging effects.

Regarding the third source mentioned above, the only samples of actual calcine that have been retrieved from bins consist of two core samples collected from the second bin set (Calcined Solids Storage Facility (CSSF) #2) in 1979. Staples et al. (1979) indicate that the two cores were full-length, with the sampler (a string of 28 sample tubes on a drill rig) in excess of 10 m long; another source (Garcia, 1997) indicates (most likely erroneously) that the cores were 1 m long. The calcination heat source was changed from indirect liquid-metal heating to in-bed combustion at about the time this waste was calcined, and no samples of calcine produced by in-bed combustion have been retrieved from bins. In-bed combustion generates both oxidizing and reducing chemical environments in different regions, which could affect calcine properties.

A calcine sample from the output of the calciner was collected in 1993 (Garcia, 1997: p. 14). This is also available for testing.

Although physical properties are reported (i.e., Table 2-1), they appear to be based on pilot tests with cold surrogates. Few data are available for long-aged calcines. Chemical compositional data (i.e., Tables 2-2 and 2-3) appear to be based largely on material from pilot-scale tests for alumina-based and zirconia-based feeds, and may not represent the full range.

The term "zirconia-type calcine" is misleading because it contains two to four times as much Al, Ca, and F as Zr.³ Chemical data for the 1979 samples of actual calcine (Garcia, 1997: Table 5) are in limited agreement with Table 2-2, and show that there is significant variation in composition vertically within the core sample. Data collected in 1993 for calcines, before the material was sent to the bin, are reported more extensively (Garcia, 1997: Tables 5-10), but some of the data columns are inconsistent with radioactive decay times. Analyses were not made for many components of interest, even in the feed solutions.

In general, the available data, analyzed for chemical, physical, and radiological properties, are not adequate for any type of calcine, for either process control or environmental considerations. Further, there appears to be no realistic sampling or characterization plan at this time to improve the database. However, it is necessary to know the approximate composition of the feedstock to be able to design applicable chemical processes. Representative sampling and analysis of the current bins is needed to determine the concentrations of the critical radioisotopes (i.e., Cs, St, and transuranic (TRU)) and other species of interest (i.e., those that would potentially interfere with radioisotope separations processes) in all of the bins.

A carefully defined sampling and limited characterization effort would significantly reduce the programmatic risk of inadequate retrieval. Data also are needed for (a) process testing, evaluation, and definition; and (b) environmental and regulatory purposes. To use limited resources most effectively, it is important to define what data are really needed and for what purpose, and then to tailor the requirements for each sample to the actual need.

Several successive calcining campaigns produced calcines of different compositions (e.g., alumina-and zirconia-based calcines, each differing in the relative abundance of aluminum, zirconium, and other elemental constituents such as calcium) that were injected into the bin sets. There appears to be a presumption that the calcines were deposited generally in layers with horizontal uniformity, as indicated by the common use of horizontal lines to delineate different calcine types in figures showing bin contents (as in Garcia, 1997: Figures 4b and 4c; see also Berreth, 1988: p. 2-1). However, different forms of calcine are probably commingled, because the pneumatic transfer systems deposited calcine in a cone at a significant and probably variable angle of repose (Garcia, 1997: p. 5), with inevitable avalanching of the conical deposits.⁴ Therefore, it is unlikely that calcine of one composition (reflective of the liquid feed stock composition) could be retrieved separately from the other calcines deposited nearby, and the extent to which different calcines can be separately retrieved and segregated is questionable. Therefore, it would be unwise to develop processing specifications that assume the intact recovery and segregation of these distinct types of calcines.

Before processing or disposition of the calcine can occur, it must be retrieved from the bins and transferred to a holding rank that will provide feed to the process. The primary issue addressed here is whether or not any problems are likely to occur during such retrieval operations. A secondary matter is the possibility of using retrieval to blend calcines of different compositions to provide feed that is more amenable to processing than, for example, some variable sequences of compositions. Both issues are discussed below.

Calcine is normally transferred from the calciner to the bins by entrainment (fluidization) in flowing air. Presumably, it can be transferred out of the bins in the same manner. This is a common and large-scale method used with dry particulate solids (e.g., in grain storage and transfer). It is expected to be successful so long as caking or sticking of the calcine does not occur. The question, then, is whether there are any mechanisms that can lead to cak-

ing. Two examples are (1) moisture entering the calcine, being absorbed by the hygroscopic components and causing the particles to stick together, and (2) a slow sintering bonding process.

The first caking mechanism, moisture effects, appears to be a possibility that would increase with time. The atmosphere in the bins should be protected from moisture, as by purging with dry air or protecting any openings to the outside with dryers.

The second caking mechanism, sintered bonding, is strongly influenced by calcine temperature. Although calcine temperatures are below normal sintering temperatures, sintering is a thermally activated process, which means that it could in principle occur over the long times (decades or more) that the calcine might be stored.⁵ The effect of sintering, if any, would of course likely be concentrated in the thermally hottest pan of the bins of calcine.

Should caking occur, it appears that this would not pose an unmanageable problem. It should be practical to break up the calcine by various mechanical means and then transfer it pneumatically. This would be a complication, but not likely a showstopper. An exception would be extreme caking, resulting, for example, from large amounts of water entering a bin. The committee believes this event to be unlikely. However, samples from many bins should be acquired to establish the capability of the calcine particles to flow freely.

The only data for production calcine retrieved from storage bins was obtained from two core samples, one of alumina-and one of zirconia-type calcines (Staples et al., 1979). The zirconia-type calcine had been stored for approximately 10 years with the maximum temperature decreasing from 220 °C initially to 190 °C. This material (13.2 kg of samples) flowed readily (i.e., it could be poured out of the sample tubes) and had a few small agglomerates. It contained < 9 percent by weight fines (< 200 mesh) near the bottom and about 3 percent by weight elsewhere. The sampler appeared clean when withdrawn. The alumina-type calcine had been stored for about 12 years with the temperature decreasing from 525 °C initially to 440 °C when sampled. This material (10.8 kg of samples) contained 18 wt percent fines and had "considerable cohesive strength" (Staples et al., 1979: p. iii). The sample tubes were coated with a visible film of dust, and resulted in air contamination. The calcine did not pour readily from the samplers, but required vigorous agitation and prodding. However, it was stated that this would not preclude pneumatic retrieval. Considerable dust was generated during sieving and pouring. There had been problems from plugging of transfer lines and the cyclone separator when this calcine was made. It is clear that this particular alumina-type calcine presented more potential problems than the zirconia-type calcine sample. The data show that calcine properties vary significantly, such that operational problems should be anticipated.

As reported in Staples et al. (1979), the force required for insertion of the sampler (a string of 28 sample tubes on a drill rig) did not cause a problem. There was some evidence for an easily penetrated crest at the top of the zirconia-type calcine but not the alumina-type. This may be significant since the zirconia-type contains hygroscopic substances. As stated above, the degree to which the calcine in bins is protected from atmospheric moisture was not made clear.

The quantity of calcine left in a bin after retrieval (the "heel") will depend on the nature of the calcine, the retrieval method, and the extent to which retrieval must be pursued. Certainly, as a minimum, there will be some fines left on the bin surfaces, and there might be

considerable calcine that is difficult to retrieve for any of several reasons. The committee is not aware of any specification on the amount of calcine that must be retrieved from each bin, although the committee thinks that there should be one. For example, if a bin or tank is to be refilled with a Class A solid waste, or must meet some closure requirement, the residual calcine (left unretrieved) must not be in sufficient quantity to cause the appropriate limit to be exceeded. (This subject is treated in greater detail in Chapter 8.)

With only this limited database with real calcine, and in view of the variability of behavior between the two calcines examined, it would be difficult to conclude that there would be no problem with pneumatic retrieval. Indeed, the committee believes that there will be problems but that they can probably be handled. However, this eventually might require mechanical operations to aid particle flow and more elaborate retrieval methods (e.g., a manipulator arm) than simple pneumatic transfer. Clearly, characterization of additional calcine from other bins is required to better define this problem.

The concept presented to the committee for retrieval was vague with respect to how the calcine will be retrieved and blended to provide a somewhat uniform feed to the subsequent steps (whether treatment or direct solidification). Some degree of blending of different compositions of calcine will inevitably occur during retrieval, and it may be difficult to predict the extent of this blending or what part of the bin is being retrieved at any given time. Because calcine compositions are different, and the requirements for further treatment are different for different calcines, a considerable degree of feed uniformity will be required for efficient operation of most calcine treatment options. This can be accomplished by retrieving uniform calcine for feed (which may be difficult or impossible to assure), by retrieving a large volume into one or several bins in which it is mixed to provide a large quantity of feed of a reasonably uniform composition, or by similarly blending a large volume after dissolution. Blending to provide large volumes of uniform feed is commonly used in successful radioactive waste treatment operations.

Because some degree of blending is going to occur, it will be advantageous to design the system to benefit from this phenomenon, insofar as possible. Substantial storage capacity probably will be required (surge capacity or lag storage) to provide continuous availability of consistent feed. Indications from other DOE sites (and ongoing foreign radiochemical operations) are that such storage capacity is essential for operational reasons and for minimization of the final waste volume.

This blending raises the issue of whether it is feasible to sufficiently characterize the waste after retrieval from storage containers, rather than before. Subject to regulatory approvals, a "quasi-batch mode" such as the following may have to be adopted:

1. Perform adequate testing on a range of surrogate compositions in order to understand the important process criteria and to develop suitable treatment requirements.

2. Remove the waste in batches, homogenize to the degree necessary, characterize the batch, define the processing conditions, and then process the batch.

This "quasi-batch mode" might be slower (e.g., due to sampling and analysis requirements) and slightly more involved than a complete "front-end" characterization of the initial waste feed into a continuous process. However, for heterogeneous and potentially complicated processing, this quasi-batch mode might reduce the complexity of the treatment method or might improve the efficiency and reliability of operations. If this or another approach is adopted, some broad knowledge of the range of initial feed characteristics (i.e., in

the current tanks and/or bins) would still be needed, because in order to fine-tune the chemical separations and/or immobilization processes, they must be developed to be adequate over the right range of conditions.

The basic dissolution approach proposed by INEEL is batch dissolution in nitric acid followed by use of settling tanks and filtration systems to remove undissolved solids (UDS). Data obtained from tests on a very limited number of samples suggest that as much as 98 percent dissolution of actual aged calcine might be achieved (assuming multiple (up to ten) recycles of UDS) under the following conditions: 5 M HNO₃, 10 ml acid/g calcine; 90 °C; and 30 to 60 minutes of contact time with heel recycle in each stage (shorter contact times also are suggested) (Brewer et al., 1997a: pp. 1, 7, 13; Fluor Daniel, Inc., 1997: p. 4-19). On blended calcine without recycle, dissolution greater than 95 percent was reported in some tests (Brewer et al., 1995). Dissolution for simulated calcine ranged between 71 and 97 percent (Herbst et al., 1995) and 46 and 95 percent (a function of stirring) (Brewer et al., 1997a). Accompanying the use of these dissolution values for planning purposes are the following assumptions:

• no calcine size reduction will be required prior to dissolution,

• no independent off-gas treatment system will be required,

• solid-liquid separation will be carried out by a combination of settling tanks and a filtration system yet to be defined,

• dissolver tank heels and filtered solids will be recycled to the batch dissolver tank, and

• residual UDS ultimately will be returned to the HLW stream.

Test parameters have varied in different experiments described in various technical reports. Table 2-4 below summarizes the dissolution performance that has been reported on three types of test materials: actual aged calcine retrieved from bins, radioactive calcine samples collected from the calciner during operations (without having been deposited in the bins), and simulants.

Although a high degree of dissolution (>98 percent) was observed in some of the experiments summarized in Table 2-4, these results were obtained for a limited number of tests. Indeed, the observation of much lower and highly variable dissolution percentages in other tests indicates that reliably achieving a high dissolution percentage in a large-scale process is questionable for the range of calcine feeds to be dealt with. The data cited in Brewer et al. (1997a), Herbst et al. (1995), and Brewer et al. (1995) demonstrate that a wide variety of dissolution challenges are to be expected. Therefore, large-scale dissolution tests on representative samples of typical (mixed) actual aged calcine, covering the full range of anticipated compositions, are needed to adequately demonstrate feasible dissolution conditions.

As Table 2-4 shows, the amount of UDS after acid leaching (and prior to any subsequent processing or separations) is uncertain. The only data with actual aged calcine gave 89 to 95 percent dissolution (Staples et al., 1979). Other tests involved selected calcine samples and therefore may not be indicative of the range of dissolution behaviors that will be encountered in practice. The committee concluded from Table 2-4 that a total UDS content of 1 to 10 percent is a reasonable estimate during full-scale calcine dissolution operations, with dissolution of Al calcine near 90 percent and dissolution of Zr calcine near 98, assuming ample contact time. Dissolution values of 90 to 99 percent result in a total volume of UDS of 63.5 to 635

m³ for processing approximately 3,000 m³ of zirconia-type and approximately 800 m³ of alumina-type calcine (Brewer et al., 1995).

Table 2-4 data infer that the 2 percent value for a UDS assumed by Fluor Daniel (1997) may be low. The data for actual aged calcine are extremely limited, and vary from around 11 percent UDS down to about 5 percent. UDS in the 1 to 2 percent range were reported for a few cases in which multiple batches of actual (but not aged) calcine were leached through the same dissolver (heel recycle). Multiple leaching of dissolution heels left in the dissolver appears to be somewhat effective, but the test data are too limited to draw a firm conclusion. It also appears that 98 percent total dissolution might be attained for most calcines using 5 M HNO₃ and heel recycle, although outliners clearly are to be expected as demonstrated by one test in which only about 50 percent alumina calcine dissolution was achieved. The unknown factor may be the amount of alpha alumina within various calcines, as this does not dissolve even after prolonged treatment. Unfortunately, the testing program has been too restricted to obtain data for a representative variety of the full range of calcine samples present in the bins.

While the dissolution data available to the committee are subject to considerable uncertainty for reasons outlined above, the data do suggest that parametric studies have identified the master variables of acid/calcine ratio, the final nitric acid concentration, contact time, and temperature.

The characterization data also strongly suggest that TRU constituents and strontium-90 (⁹⁰Sr) are preferentially retained in the UDS. For example, Table VI of Brewer et al. (1995) shows that Al-type calcine dissolved in the 78 to 96 percent range, while Am, Cm, and Pu are significantly enriched in the heels compared to the initial calcine. The UDS contains Al₂O₃, calcium-stabilized zirconia, and CaF₂. These results are consistent with the well known tendency of tri-and tetravalent actinides and Sr(II) to follow insoluble calcium and zirconia phases. However, not enough information is readily available to confidently predict the partitioning of actinides and ⁹⁰Sr between the UDS and solutions delivered to subsequent radio-chemical separation. This gap in knowledge translates directly into uncertainty as to the ultimate fate of these key radionuclides in the flowsheet, including required capacities of the separations units and amount of HLW due to solid residuals.

While Brewer et al. (1995) includes a citation (reference 6) that reports that actinide buildup in UDS does not occur with heel recycle that report apparently was never published and this assertion therefore could not be evaluated.

The degree to which the calcine can be dissolved affects the quantity and composition of the final waste form(s). The UDS component would probably be included in the high-activity fraction for immobilization because of its content of Sr and actinides. This high-activity fraction likely will have significant CaF₂ content coming either from solution or from

UDS. If such waste streams with significant calcium or fluoride content is to be vitrified, a potential problem arises because borosilicate glass has limitations in incorporating constituents such as calcium fluoride (as discussed in Chapter 5). In this case, two remedies are readily identified:

1. modify the dissolution process to dissolve a larger fraction of CaF₂, which then would be removed from the HLW fraction in separation steps, or

2. produce an alternative glass (e.g., a phosphate glass) that is more tolerant of CaF₂ (this is treated more fully in Chapter 5).

The first remedy requires that dissolution requirements be derived based on considerations of the impact of undissolved components on the downstream processes.

As stated earlier, the amount of characterization and testing data on actual calcine is inadequate for the selection of large-scale retrieval and treatment processes. Little direct information is available for actual aged calcine retrieved from bin sets. Conclusions on dissolution behavior are based largely on extrapolations from surrogate materials rather than the calcines currently existing in the bin sets. These surrogates and the few samples of actual aged calcine available for testing are not fully representative of all the calcine contents in all six bins. Additional characterization studies are needed to provide definitive conclusions about the full range of calcine characteristics and bounds on important calcine properties.

A second issue is that of the reliability of the current quantitative information on calcine characteristics, some of which comes from calculations that contain errors. Large fields of numbers in tables do not necessarily indicate an abundance of actual measured data. For example, Table 2-3 presents calculations based on data and flowsheet information (Garcia, 1997, p. 19). The "data" of tables 9 and 10 of Garcia (1997) are also calculated quantities, some of which contain errors. These tables give isotopic abundances after 500 years for many isotopes present in the calcine. For some of the isotopes of critical importance, the committee recalculated, these abundances using straightforward half-life decay methods to confirm their accuracy. Many were found to be in error. Some examples are the values for ²³⁸Pu, ²⁴¹Pu, ²⁴⁴Cm, ¹³⁷Cs, ¹⁵¹Sm, and ^166mHo in Table 9, and ²¹⁰Pb and ²²⁶Ra in Table 10 of Garcia (1997).

In other key reports, the quality of analytical data that is available is clearly inconsistent and contradictory. For example, Tables 4.3-3 and I-17.2 in Fluor Daniel-Hanford, Inc. (1997), which describe average alumina and zirconia content in the six bin sets, are in conflict. Bin set numbering appears to be scrambled for bin sets 2 through 4, and the zirconia content is listed as 0 percent ZrO₂ in one case and 10 percent Zr in the other. Another example of significant analytical inconsistency is found in Table 11 of Brewer et al. (1997a), where radio-chemical analyses of UDS from a zirconia calcine dissolution test are listed. In this table, the gross alpha content is listed as 7.1 × 10⁸ dps/g, yet the sum of the individual TRU constituents is approximately four orders of magnitude less. A similar discrepancy exists in Tables 13 and 15 of Brewer et al. (1997a). While there may be reasonable explanations for such discrepancies, they are not explained in the reports. The committee therefore is unwilling to base firm conclusions on such questionable analytical data.⁶

The committee believes that substantial and sustained testing with a full range of actual aged calcine spanning all production variations is required to provide adequate confidence in the understanding of partitioning of radioactive constituents between UDS and liquids. Even more critical is the need to demonstrate that solid-liquid separations (Chapter 3) will be adequate to meet the demanding separation factors required by the proposed flow-sheets, particularly in meeting Class A waste levels. The key question, which is one of judgment as well as adequacy of resources devoted to the problem, is whether the required information can be obtained in parallel with design efforts, as suggested in Fluor Daniel, Inc. (1997). Because of the expense and difficulty in obtaining the required bin set materials, the judgment of the committee is that, even if adequate resources are provided, it is unlikely that the required information can be obtained in the timeframe specified by the baseline plan.

This recommendation for further testing regarding dissolution is consistent with conclusions of earlier external review groups, as listed in Appendix G of Murphy et al. (1995). In addition, the reports the committee reviewed repeatedly point out the requirement for additional testing. For example, Table 12.3-1 in Fluor Daniel, Inc. (1997) recommends as a high priority a 2-year dissolution study with actual aged calcine material, and recommends that this be carried out prior to early conceptual design. As another example, Brewer et al. (1997a) states that additional testing of the quantity and nature of radionuclides in UDS must be carried out with actual calcine. The committee concurs that the tests should include dissolution rates, temperature, acid requirements, end-point acid concentration, and residue characterization as well as settling/filtration tests.

Chapter 13 presents recommendations based on the material in this chapter. As discussed above, (1) the characterization data on actual aged calcine are limited, and (2) testing of planned operations should be done on materials that span the range of the different properties of calcine that are stored in the bin sets. These characterization, retrieval, blending, and dissolution operations are important, insofar as they will determine the ultimate fate of key radio-nuclide constituents. That is, those radionuclides that are not retrieved are left in situ, and those that are dissolved and partitioned from the high-activity fraction are the only candidates for a possible low-activity fraction.