Validation of Ultraﬁlter Performance Model Based on Systematic Simulant Evaluation R. L. Russell,* J. M. Billing, H. D. Smith, and R. A. Peterson Paciﬁc Northwest National Laboratory, P.O. Box 999, Richland, Washington 99352 Because of limited availability of test data with actual Hanford tank-waste samples, a method was developed to estimate expected ﬁltration performance based on physical characterization data for the Hanford Tank Waste Treatment and Immobilization Plant. This method relies upon the similarity between the gel concentration measured during crossﬂow ﬁltration and the slurry concentration measured after a sample is centrifuged. Testing was performed to determine the centrifuged-solids concentration, and then a subset of simulants was tested to determine the gel concentration during crossﬂow ﬁltration. These two approaches produced identical results, indicating the centrifuged-solids concentration can be used to represent the gel concentration for ﬁltration. This substitution will allow the expected ﬁltration performance to be characterized at a signiﬁcantly reduced cost. Introduction Most industrial ultraﬁltration processes are characterized by feed streams with small (<20%) changes in composition and that are readily available for testing at nominal costs. However, nuclear waste treatment is characterized by feeds that are highly variable in both chemical speciation and physical properties and that are relatively expensive to test (on the order of $1 million per test due to the requirement to work in a highly shielded environment). The goal of this work is to validate a model for ﬁltration performance to provide a basis for a less expensive method for characterizing the ﬁltration performance of these feeds. The need for this improved method is driven by the need to process and vitrify approximately 60 000 t of high-level waste (HLW) sludge contained in 177 underground storage tanks at the Hanford Site in Richland, Washington. This waste was generated during the more than 45 years that plutonium and other nuclear materials were produced at Hanford. Characteriza- tion studies have identiﬁed more than 150 different signiﬁcant sludge-bearing streams. 1 This sludge phase is typically a blend of metal hydroxides and oxides. The most common species include aluminum hydroxides (predominately gibbsite and boehmite), iron hydroxides (for example, goethite), uranium oxide, and a broad spectrum of other oxides and hydroxides. Average particle sizes of these solid particulates range from less than 1 µm for many of the precipitated hydroxides, such as iron hydroxide, to greater than 10 µm for the more crystalline solids, such as gibbsite. The process requires that 100% of these solids be rejected by the ﬁlter element to verify decontamination of the product efﬂuent. The ﬁrst step in treating this HLW sludge to prepare it for vitriﬁcation in the Hanford Tank Waste Treatment and Im- mobilization Plant (WTP) will involve ultraﬁltration to dewater the sludge phase. The planned ultraﬁltration process involves concentrating the feed streams from 4 to 17 to 20 wt % using porous stainless-steel ﬁlter elements. The full-scale unit will consist of ﬁve bundles of 1 / 2 -in. (i.d.) tubes with three bundles 10 ft in length and two bundles 8 ft in length. The tubes will have a nominal pore-size rating of 0.1 µm. After ultraﬁltration, the concentrated sludge solids are vitriﬁed while the clariﬁed supernate undergoes cesium removal by ion exchange before ﬁnal disposition. Because of the high cost of working with highly radioactive materials, only a limited number of actual waste ﬁltration trials have been performed. These samples have primarily been selected on the basis of the initial tank processing sequence and do not fully cover the broad range of feeds expected to be processed over the lifetime of the plant. Consequently, it is desirable to develop methods to estimate ﬁltration performance for the balance of the Hanford HLW based on available characterization data. The majority of prior work with Hanford HLW has been performed at relatively low solids concentra- tions. 2-4 However, these tests did not examine the regime where ﬁltration is dominated by cake formation. This work will focus on this ﬁltration regime. The primary thesis of this work is that gel concentration (i.e., the sludge concentration at which the ﬁlter ﬂux falls to zero) can be estimated using the centrifuged-solids concentrations for these particulate slurries. We assert that the gel concentration associated with crossﬂow ﬁltration can be estimated with centrifuged-solids experiments, and knowing this, solids con- centration associated with the gel allows us to avoid extensive crossﬂow ﬁltration experiments. This will be demonstrated in two steps. First, a simulant ﬁltration trial will be evaluated to demonstrate that ﬁltration of these types of slurries is consistent with a concentration polarization model. Actual waste data will then be analyzed to determine the efﬁcacy of using centrifuged- solids data to estimate the limiting concentration for ﬁltration. Once it has been proven that both phenomena produce the same results, centrifuged-solids experiments can be used as a surrogate for crossﬂow ﬁltration experiments. The ultimate experimental method will include some crossﬂow ﬁltration work, centrifuged- solids experiments, and a method to predict mass-transfer rates, which will save both time and money. Experimental Methods and Analyses HLW Simulant Preparation. The initial HLW simulant was composed of a basic tank-waste simulant having the components listed in Tables 1-3. Table 1 shows the components of the sludge solids consisting primarily of iron oxyhydroxide. This is produced when initially the KMnO 4 salt and the Mn(NO 3 ) 2 solution are prereacted in deionized water to produce insoluble * To whom correspondence should be addressed. Telephone: (509) 373-6235. Fax: (509) 376-3108. E-mail: renee.russell@pnl.gov. Ind. Eng. Chem. Res. 2009, 48, 10077–10086 10077 10.1021/ie901042w CCC: $40.75  2009 American Chemical Society Published on Web 10/07/2009