Predictive Models to Aid in Design of Membrane Systems for Organic Micropllutant Removal
Year Released: 2013
Funding Partner: Bureau of Reclamation
Total Investment: $467,000 (Cash: $343,000, In-Kind: $124,000)
Principal Investigator: Jörg E. Drewes, Ph.D., Colorado School of Mines
Reverse osmosis (RO) and nanofiltration (NF) membranes are becoming increasingly widespread in water treatment. For drinking water augmentation projects in the United States, Singapore and Australia treatment using microfiltration (MF) pretreatment followed by RO is the industry standard. Past research projects have demonstrated that a limitation of RO and NF is the incomplete removal of various organic solutes, such as certain disinfection by-products, pharmaceutical residues, household chemicals, and endocrine disrupting compounds.
Goals and Objectives
The project develops models that can be used, a priori, to predict the rejection of a wide variety of organic compounds by NF and RO membranes. The objectives of this research project were (a) to evaluate molecular modeling approaches and determine method independent and reliable molecular descriptors for the development of quantitative structure activity relationship (QSAR) models, (b) to identify, develop, and optimize membrane modeling strategies and develop models that can be employed to predict the rejection of organic solutes, and © to evaluate the efficiency at which membranes employed at full scale remove trace organic chemicals and to successfully predict the removal rates with the developed model(s).
The research study consisted of three major phases. The project was initiated with the development of a roadmap for membrane rejection modeling, including a comprehensive literature review and the determination and calculation of reliable, accurate, and relevant molecular descriptors for a wide range of trace organic chemicals. The second phase of the project addressed the development, optimization, and validation of viable membrane rejection models. Evaluation and validation of the membrane rejection models were conducted at pilot- and full-scale membrane installations.
Findings and Conclusions
Current limitations in predicting the rejection of organic chemicals during water reuse applications exist because an understanding of viable modeling approaches is lacking, many past modeling approaches utilized unrealistic experimental set-ups and produced data that is inappropriate to predict rejection at larger scales. In addition, modeling membrane treatment is an inherently difficult problem to solve due to system complexities and numerous factors affecting rejection.
A significant portion of this study evaluated rejection at pilot-scale under carefully controlled laboratory experiments and in the field at a water reclamation facility. This approach provided valuable information on comparing and up-scaling bench-scale experimental rejection results to larger membrane systems. Findings from these studies suggest that bench-scale rejection results can be used to describe the rejection at large-scale, however, the hydrodynamic conditions, and flux and concentration gradients for a large-scale system need to be characterized. The differential element approach combined with the phenomenological model was an effective modeling approach to describe rejection at pilot-scale. Bench-scale derived phenomenological model coefficients could be used to estimate pilot-scale permeate concentrations and rejection. QSPRs and the rejection diagram approach developed with bench-scale rejection data could estimate rejection at pilot-scale.