Alternatives to Piloting in Bracksih Waters: Accuracy and Precision of Commerical Reverse Osmosis Membrane Design Model Projections

In a time of economic recession and drought, the Texas Water Development Board is reviewing alternatives to the current Texas Commission on Environmental Quality requirement of demonstration-scale pilot testing membrane desalination to decrease permitting costs and project construction delays. One alternative is to use engineering judgment with membrane manufacturer\u27s software models to predict membrane performance and ensure that product water quality requirements will be met. To comply with state permitting requirements, the first few months of successful full-scale operation can be submitted in lieu of pilot demonstration. The goal of this research is to characterize the accuracy and precision of commercial membrane design software models. The objective of this research is to compare actual membrane performance with computer software model predictions and evaluate the accuracy and precision of these models with respect to hydrodynamic and water quality parameters. The accuracy of membrane desalination models was analyzed by adjusting modeling parameters such as feed, boost, and throttling pressures to match full-scale membrane system flux and then comparing real and simulated operating pressures, bulk salt (total dissolved solids) rejections, and specific contaminant rejections (e.g., sodium, calcium, chloride, sulfate, carbonate, silica, etc.). The precision of membrane desalination models was analyzed by comparing the model output results from a set of membrane manufactures\u27 models for equivalent membrane systems. Membrane manufactures included in this comparison are Dow Chemical, GE, Toray, Koch, CSM, and Hydranautics with their respective software models: ROSA8.0.3, Winflows 3.1.2, Toray DS2 2.0.1.26, ROPRO 8.05, CSMPRO 4.1, and IMSdesign 2011.19. A list of specifications of brackish water, fouling resistant, and low energy reverse osmosis was compiled from the six manufacturers that detailed surface area, permeate flow rate, and testing conditions. Equivalent membranes among the six manufactures were selected based on flux and rejection performance, as well as testing conditions. Manufactures were also consulted to verify the selection of equivalent membranes. These commercial software models are compared for several full-scale desalination systems (including the Kay Bailey Hutchison Desalination Plant in El Paso, TX), demonstrating variation in brackish groundwater quality/composition (1500-5000 mg/L TDS) and overall system recovery (75%-85%). Conclusions indicate that membrane desalination manufactures\u27 software models are relatively accurate with respect to the prediction of operating pressures and salt/contaminant rejection. Based on this research, the models tend to conservatively over-estimate the required feed pressures, and they tend to slightly over-estimate the salinity removal of the membranes. Furthermore, the simulation of equivalent membrane systems among different manufacturers reveals differences in predicted operating pressures and contaminant rejections, but self-categorized membrane types generally perform relatively similarly