Prevention of Solids Formation

Transcription

1 Dilution, DIAL TR98-1.2, Diagnostic Instrumentation and Analysis Laboratory, Mississippi State University, Mississippi State, MS, (October 1998). 23. Lindner, J. S., and R. K Toghiani, Thermodynamic Simulation of Tank 241-SY-101 Dissolution, Part 3:Crust Solids Dissolution Modeling and Associated Gas Release, DIAL TR98-1.3, Diagnostic Instrumentation and Analysis Laboratory, Mississippi State University, Mississippi State, MS, (April 1999). 24. Herting, D. L., personal communication, June Toghiani, R.K., Lindner, J.S., DIAL/MSU Saltcake Dissolution: Fiscal Year 2000 Status Report, DIAL TR00-1, Diagnostic Instrumentation and Analysis Laboratory, Mississippi State University, Mississippi State, MS, (2000). 26. Herting, D.L., personal communication, April Prevention of Solids Formation J. S. Lindner, A. Antonyraj, T. Durve, and R. K. Toghiani INTRODUCTION Tank farm operations at Hanford include the interim stabilization program where the supernatant and interstitial liquor in the singleshell tanks is reduced. Benefits from this process include the minimization of leakage from aging tanks, thereby limiting migration of waste into the soil, and the temporary reduction of waste within the tank. The process consists of jet-pumping the liquid in a given tank, obtained through a screen or salt well to a double-shell holding tank and then to an evaporator. Dilution water is added at the pump head. Recently, solids formation and plugging have been noted during transfers from tanks 241-SX-104, 241-U-103, and 241-BY The primary solid responsible for the plugs from the first two tank wastes DIAL

2 has been tentatively assigned, through experiments conducted on the waste liquid in the laboratory, as Na 3 PO 4. 12H 2 O. The plug formed during salt well pumping of BY-102 was believed to arise from sodium carbonate. Other solids may participate in the plug formation process and this will largely depend on the solid-liquid equilibrium of the species contained in the waste stream. Little information, aside from the laboratory screening experiments is known regarding the mechanisms of plug formation and, more importantly, the required change in pressure that would indicate the beginning of plug formation. From operations measured records, the time needed for a plug can be determined and by knowing the pressures and flow rates the approximate location of the plug can be estimated; however, prevention of inadvertent plugs may be possible based on a suitable engineering tool that will allow operators to tailor waste transfers. Long-term site operations will involve the dissolution of saltcake in the single-shell tanks and the removal of the salt liquor. Site plans are currently in progress for the retrieval of all of the salt in Hanford tank 241-S-112. Diluent will be added to the top of the saltcake followed by dissolution and permeation of the supernatant through the saltcake to the saltwell pump already located in the tank. Thus, some of the same problems that have already been addressed regarding the transport of high concentration phosphate solutions will remain. In addition, questions involve the rate and mechanisms associated with the saltcake dissolution process. These include the distribution of chemical species as retrieval proceeds (dissolution, dilution, and reprecipitation) and the physical characteristics of the saltcake bed, such as the porosity and permeability, which govern the rate at which the tank can be emptied. Work has started in this area with the development of an experimental approach that can provide the necessary information. DIAL

3 WORK ACCOMPLISHED Support of 241-S-112 Saltcake Retrieval Background, chemical compositions and ESP results. Previous experiments in these laboratories have focused on the dissolution of a surrogate saltcake (Surrogate #1) developed by R. Hunt at ORNL.28,29 The majority of the results for these experiments, presented at the May Workshop, will not be discussed in detail here. 30 In accordance with the time line for the upcoming experiments at Florida International University (FIU) we began to examine the dissolutions behavior of a second saltcake surrogate composition developed by Hunt. 29 The surrogate was prepared according to the composition in Table 7. The various constituents were added to water at 50 C and thereafter heated to around 110 C to force evaporation. Once the desired volume was attained the solution was allowed to cool. The procedure is similar to tank farm operations reflecting evaporator operation followed by transfer of the stream to a waste tank and cooling. Anion and cation loadings for the two recipes (Table 7) indicate that the nitrate, phosphate and aluminum loadings are the same for both compositions and Surrogate #2 contains oxalate along with fluoride and chloride. The hydroxide, sodium, and nitrite loadings are lower in Surrogate #2 as compared to Surrogate #1, however, the feed levels for sulfate and carbonate anions are increased. The amount of water used in each recipe is not important owing to the evaporation step in the procedure. TABLE 7. Constituent base composition loadings derived for Surrogate #1 and Surrogate #2. Constituent Surrogate #1 Surrogate #2 Mol Mol Al CO DIAL

4 TABLE 7. Constituent base composition loadings derived for Surrogate #1 and Surrogate #2. Na NO NO OH PO SO C Cl F H 2 O A presumption of the behavior of the two surrogates can be made based on the chemical compositions and the previous results for Surrogate #1. Specifically Surrogate #2 contains sodium oxalate. Earlier core sample dissolution experiments and ESP calculations have indicated that up to % dilution by weight using water is required to dissolve this solid. 31 Also, the lower hydroxide concentration in Surrogate #2 would be expected to result in a lower ph. This, in turn, would be expected to result in the formation of gibbsite, Al(OH) 3. As opposed to the composition of Surrogate #1, Surrogate #2 contains solids that will not undergo dissolution with water or that are expected to remain in the column for a long period of time. Environmental Simulation Program (ESP) results for the two recipes shed additional light on the differences. These calculations were carried out using the Trona and Na2Snacl databases with version 6.5 of ESP. Figure 38 provides comparisons of the supernatant viscosity and ionic strength at various levels of dilution. DIAL

5 25.00 ESP Predictions Abs. Viscosity (cp) and Ionic Strength Surrogate #2 Ionic Strength Surrogate #1 Ionic Strength Surrogate #1 Viscosity Surrogate #2 Viscosity % Dilution by Weight FIGURE 38. Surrogate supernatant ionic strengths and absolute viscosities. The ionic strength of the liquid for Surrogate #2 is greatly reduced compared to Surrogate #1 and this is also reflected in the values predicted for the absolute viscosities. Measurements of the viscosity for the second surrogate were performed with an average value of ca. 12 cp. Associated ph values for Surrogate #2 are also lower than for Surrogate #1, Figure 39. ESP Predictions ph and Ionic Strength % Dilution by Weight Surrogate #2 ph Surrogate #1 ph FIGURE 39. ph values predicted by ESP for Surrogate #1 and Surrogate #2. DIAL

6 The model predictions for the masses of gibbsite, present in both surrogates, and for sodium oxalate, present only in Surrogate #2, were added together and then the total mass from these solids was converted to a volume using the associated densities (2.43 and 2.34). These data are shown in Figure 40. The amount of insoluble or slightly soluble solids is considerably larger in Surrogate #2 than in Surrogate #1 and this prediction is especially apparent at small dilutions. 40 Mass (g) and Volume (cm 3 ) of AlOH and Na C O Mass of Insoluble Solids, Surrogate #2 Volume of Insoluble Solids, Surrogate #2 Mass of Insoluble Solids, Surrogate #1 Volume of Insoluble Solids, Surrogate # Weight % Dilution Water FIGURE 40. Insoluble (gibbsite and sodium oxalate) masses and volumes from ESP calculations against the percent dilution (water) by weight. Based on the initial compositions and the ESP predictions it was believed that the dissolution behavior of Surrogate #2 would be somewhat different than Surrogate #1. In the later case all of the solids would undergo dissolution; with Surrogate #2, however, the amount of diluent needed to complete dissolution of the soluble or slightly soluble components would be expected to be larger, owing to the sodium oxalate. Formed gibbsite would be expected to remain in the column. DIAL

7 Column experiment physical data. Figure 41 compares the saltcake bed level heights and the liquid level above the saltcake for the SC-9 experiment (Surrogate #1) and for the run with Surrogate #2, B-2. For both of these experiments the prepared saltcake was poured into a level column at room temperature followed by draining the supernatant through the bed until the liquid was at or below the saltcake level bed height. Diluent (water) was then added to the top of the bed using a peristaltic pump. A circle of filter paper was placed over the salt bed so as to distribute the diluent evenly across the surface. Height (cm) SC-9 Salt Bed Level SC-9 Liquid Level Above Salt Bed B-2 Salt Bed Level B-2 Liquid Level Above Salt Bed SC-9 After Re-pack SC-9 Liquid Above Bed After Re-pack Poly. (B-2 Salt Bed Level) Elapsed Run Time (min) FIGURE 41. Saltcake layer height and liquid layer height above the bed for column experiments for Surrogate #1 (SC-9) and Surrogate #2 (B-2). The length of time necessary to dissolve the soluble solids in the saltcake bed was about a factor of 3 longer for Surrogate #2 (5000 minutes) as for Surrogate #1 (ca minutes). Both of these experiments were performed in the 3-inch round, 3-inch tall acrylic col- DIAL

8 umn. In point of fact there was a larger mass of saltcake in the SC-9 experiment using Simulant #1 and this can be seen in the initial saltcake bed heights of Figure 41. FIGURE 42. Image collected at t = 1570 minutes into the run. FIGURE 43. Collected at 2610 minutes into the run. Channeling is observed at the scale marks etched on the inside of the acrylic column. Selected images from the B-2 experiment are presented in Figures 42 and 43 and both show two distinct layers in the saltcake bed. Layering or stratification was not observed during experiments with Surrogate #1. It is believed that the layering reflects segmentation in DIAL

9 chemical composition and that these solids may, in fact, be insoluble or sparingly soluble molecules such as gibbsite or sodium oxalate. Images were analyzed from the beginning to the end of the experiment. Details of the physical manifestation of the formed layer and the change in the saltcake dissolution rate are provided in Figure 44. The height of the top (formed) layer increased slowly during the first portion of the experiment. The layer then attained a constant and maximum height before channeling into the other constituents of the bed at an elapsed run time of around 3100 minutes. 5 Saltcake, Formed Layer Heights & Liquid Level Above Saltcake (cm) B2 SC layer height total cm B2 SC top bright layer height cm B2 liquid layer above saltcake Linear (B2 SC layer height total cm) y = - 3.4E - 04x + 4.7E + 00 R 2 = 9.4E Elapsed Run Time (min) FIGURE 44. Data from the images collected during the B-2 experiment. The change in the height of the saltcake bed when the layer was forming until channeling occurred was calculated as Saltcake Bed Height = -3.4 x 10-4 cm/min (elapsed run time) cm (EQ 1) DIAL

10 with a correlation coefficient (goodness of linear fit, r) value of A similar calculation for the change in saltcake height with time for the SC-9 experiment (data shown in Fig. 41) was determined as (r = ) Saltcake Bed Height = -1.3 x 10-3 cm/min (elapsed run time) cm. (EQ 2) Clearly larger changes of the saltcake height with time are observed at longer elapsed run times (see Fig. 41). Nonetheless, it appears that the gross dissolution rate, as inferred from the decrease of the matrix height with time, is much reduced in the presence of the formed layer. On the formation of the layer. In an attempt to reproduce the observed behavior about 40 g of saltcake was loaded into a modified plastic sample cone and, following draining of the supernatant, water was carefully added to the column using a pipette. Images were collected throughout the course of the experiment. Figure 45 shows that the formation of the layer, observed earlier in the 3-in. ID acrylic column, was reproduced. FIGURE 45. One of the images collected during the small-scale experiment. Total saltcake, secondary layer and liquid layer heights determined from the collected images are plotted against elapsed run time in Figure 46. The secondary layer began to form around 900 minutes. DIAL

11 Thereafter the layer grew until the experiment was stopped around 3500 minutes. 6 5 Height (cm) y = - 2.0E - 04x + 5.0E + 00 R 2 = 9.7E - 01 Formed layer height Test col SC height Saltcake layer & water above saltcake Linear (test col SC height) Time (min) FIGURE 46. Results of image analysis for the test column experiment. Very little channeling was observed in the test experiment and consequently the layer remained intact. The change of the height of the saltcake can be described by Saltcake Bed Height = -2.0 x 10-4 cm/min (elapsed run time) cm (EQ 3) The correlation coefficient for the linear fit was The slope for the test experiment, -2 x 10-4 is of the same magnitude and range as that found for the B-2 experiment, -3.4 x Using the former value the time needed to attain a saltcake height of 0.5 cm, or a 90% change in volume, in the 40 g test experiment would be 15.7 days. Following stoppage of the experiment the supernatant liquid above the column was carefully withdrawn and a small sample of the solid/interstitial liquid from the top layer was collected. The material from the formed layer had the appearance of a concentrated white DIAL

12 latex paint, Figure 47. The viscosity of the layer is large, slow flow of the layer was only observed when the petri dish was positioned vertically. FIGURE 47. Sample extracted from the secondary layer. FIGURE 48. Polarized light microscope image (20x) of the particles in the formed layer. Small rectangular particles are sodium oxalate. Large amorphous particles are gibbsite. Small colored rods (gold or blue) are sodium phosphate dodecahydrate. Small white rods are sodium carbonate monohydrate. White rhomboids are sodium nitrate. Attempts were made to separate the solid and liquid constituents within the layer sample using filtration, however, this was largely unsuccessful. Some indication of the particles present was possible by placing a small sample of the neat layer on a microscope slide. Figure 48 illustrates the particles present in the formed layer and Figure 49 shows the particles present near the bottom of the saltcake matrix that had not yet undergone dissolution. The particles are observed to be somewhat smaller in the sample taken from the DIAL

13 formed layer. A reduction in particle size is expected on dilution; however, the layer image (Fig. 48) also contains loose amorphous particles of gibbsite along with a large number of small sodium oxalate particles. FIGURE 49. PLM image (20x) of the particles present in the saltcake near the bottom of the column. The particles are as identified in Figure 48. Samples of the formed layer were also diluted with water and then subjected to IC, ICP, and TIC/TOC analysis. Upon initial dilution the sample formed a turbid solution indicating the presence of a large number of very small particles. This is consistent with the PLM images collected on the formed layer. Initial results for the layer composition are provided in Table 8 along with the predictions of ESP calculations performed in an attempt to characterize the associated chemistry. TABLE 8. Experimental and predicted anion and cation concentrations (in parts-per-million) for the formed layer. ESP Supernatant From Solids From Solids Total ESP Experiment #1 Cl E E+02 NO E E+02 Experiment #2 C 2 O E E E E E+04 Al E E E E E+03 CO E E E E+04 DIAL

14 TABLE 8. Experimental and predicted anion and cation concentrations (in parts-per-million) for the formed layer. NO E E E+05 F E E E E+02 PO E E E E E+04 SO E E E+04 Both chloride and nitrite anion were found, in low concentration, upon characterization of the layer. Model calculations as well as the PLM images of the base saltcake composition indicate that both Cl -1 and NO 2-1 are completely partitioned into the liquid or supernatant phase; consequently the layer must contain a small fraction of the initial supernatant. From the data in Figure 46 it is possible to determine the volume of the saltcake dissolved and the volume of the layer formed. In addition it is possible to determine the volume of the water that was added to the column. Subsequent conversion of these volumes into densities indicated a percent dilution by weight of the interacting saltcake (i.e., the top layer that was dissolved or converted to the formed layer) was around 100% or 1 g water per g saltcake. The base supernatant composition predicts chloride and nitrite anion concentration of 12,200 and 27,400 ppm, respectively so at the point where the experiment was stopped most all the original supernatant had passed from the top to lower portions of the column. The resulting composition of the formed layer is therefore primarily from solids that were dissolved upon the addition of water. The experimental results given in the table further reflect a 400x dilution of the formed layer. Two sets of experiments on the formed layer were performed. For the first, the turbid solution prepared by adding water to the layer material was used directly for the analysis. Thus any solids that had not dissolved or had formed upon dilution would be accounted for. In the second experiment a portion of the formed layer was diluted to DIAL

15 400x and this solution was filtered. The reduced aluminum loadings, Table 8, in the filtered sample reflect the filtration of gibbsite particles from the solution. Companion ESP calculations were performed through the knowledge that a small portion of the original saltcake surrogate was present in the layer along with the 100% by weight dilution of the solids. The amount of the original supernatant was determined by recursion to the experimentally determined concentrations of the chloride and nitrite anions. Concentrations determined for the resulting supernatant phase, from the solids, and the total ion concentration are given in the table. Agreement between the calculated and experimental values is excellent for carbonate, nitrate, and sulfate. The experimental result for fluoride is somewhat larger than that calculated; however, both values are small owing to the original composition, Table 7. Predicted aluminum concentrations are somewhat larger than the value determined experimentally; in this case some gibbsite solids were observed at the bottom of the (diluted) sample vial and were not accounted for in the ICP measurements. Deviations between the calculated and measured concentrations for oxalate and phosphate anions are under investigation. Predicted physical properties of the layer composition are collected in Table 9. TABLE 9. ESP predictions for the physical properties of the supernatants. Supernatant Original Saltcake Predicted Layer Density, g/l Abs Visc, cp ph Ionic Strength The liquid phase resulting from water addition to the solids of the saltcake composition along with a small portion of the base supernatant is predicted to have a smaller density, reduced values for the ph DIAL

16 and ionic strength, and a markedly lower value for the viscosity. This later result may appear to be at odds with earlier comments regarding the viscosity of the layer proper, here, however, the layer observation also reflects those solids, such as gibbsite and sodium oxalate, and the other solids observed in the PLM images that contribute to the overall layer viscosity. Implications on layer formation during saltcake waste retrieval. The formation of the layer with the Surrogate #2 composition was observed to be reproducible. The original composition for the recipe was derived by Hunt using waste loadings from the Best Basis Inventory. Specifically, the composition of the record for tank 241-S-112 was reduced to the principle components. These were then examined by the customer and modifications made. The final composition of Surrogate #2 is more representative of tank 241-S-102 than of S-112 but is also representative of many of the tanks that will require pretreatment and retrieval. Based on the results above it is likely that secondary layers will form during single 0-shell tank retrieval activities. For this reason additional experiments are in progress to further characterize the formed layer and to ascertain the associated physical properties. CONCLUSIONS Saltcake dissolution experiments using a new surrogate waste composition revealed the formation of a secondary layer as dilution proceeded. Small-scale experiments were developed to reproduce the observed behavior and to obtain samples allowing characterization. The formation of the layer was found to result in extremely long dissolution times as compared to previous work using a surrogate designed to undergo complete dissolution. Run times in excess of 5000 minutes were observed with the modified simulant as compared to 1500 minutes for the baseline composition. The increase in time is related to the chemical composition of the surrogate and in the imper- DIAL

17 vious nature of the formed layer. A principle component in the formed layer was determined to be gibbsite particles. Other solids such as sodium oxalate, sodium nitrate and sodium phosphate dodecahydrate were also observed in images of the formed layer. Companion ESP model calculations have been performed in an attempt to model the behavior. A comparison of the model predictions to the experimental results indicates that the model accurately described the behavior of most all of the major cations and anions with the exception of oxalate and phosphate. Work on these species is continuing. Secondary layer formation, such as that first reported here, has one significant implication on single-shell tank saltcake retrieval. Extensive dissolution times are likely unless a means to mobilize the layer, through for example, directed sprays or nozzles, are incorporated in the retrieval plan. If mobilization of the viscous layer can be accomplished then dissolution with water can proceed but eventually the layer will have to be dissolved. The primary constituent of the layer is Al(OH) 3. Re-dissolving this material can be accomplished using 2 or 3 M sodium hydroxide and depending on the extent of the formation of the layer it may be necessary to alternate the use of water and of caustic. WORK PLANNED Additional experiments on the formation and implications of the formed layer are in progress. REFERENCES 27. D.A. Reynolds. May Status of waste transfers, criteria, and plans. Presented at the 3rd Saltcake Dissolution and Feed Stability Workshop, Richland, WA, and J.R. Jewett. Personal communication. January Tank Farm Informa- DIAL

18 tion for TFA Workers and Saltwell Pumping from Tank SX-104. Numatec Hanford Corporation, Richland, WA. 28. J.S. Lindner, T. Durve, V. Raju, and R.K. Toghiani Prevention of Solids Formation in DIAL Quarterly # R. Hunt, personal communication. 30. J.S. Lindner, A. Antonyraj, T. Durve, T. and R.K. Toghiani. May Bench-scale saltcake dissolution tests and modeling. Presented at the Saltcake Dissolution Workshop, Richland WA. 31. R.K. Toghiani and J.S. Lindner. December DIAL/MSU saltcake dissolution project. FY 00 Status Report, DIAL TR00-1 Tank Focus, Diagnostic Instrumentation and Analysis Laboratory, Mississippi State University. DIAL