Industrial wastewater volume reduction through osmotic concentration: Membrane module selection and process modeling

Abstract

Osmotic concentration (OC), a form of forward osmosis (FO) but without draw solution recovery, can be applied for reducing wastewater disposal volumes in the oil & gas industry. Within this industry, wastewater is often disposed of by injection through disposal wells into deep underground reservoirs. By reducing wastewater disposal volumes, the sustainability of the disposal reservoir is improved. In this application of OC, seawater or brine from a desalination plant serves as the draw solution and the diluted seawater is discharged to the sea. This study compared 3 commercial hollow-fiber FO membranes (CTA, TFC, aquaporin proteins) for reducing the volume of low salinity wastewater generated during liquified natural gas (LNG) production. Additionally, a model was developed to predict the performance of commercial full-scale membranes by identifying optimum operating conditions, taking into consideration the trade-off between feed concentration factor and water flux. Bench-scale tests were conducted using synthetic and actual wastewater from an LNG facility to evaluate OC technology performance and validate model predictions.Based on model results with a feed mimicking the salinity of actual wastewater, a 4x concentration factor produced a reasonable compromise between feed recovery and draw solution dilution and was considered the optimum for future tests. At higher concentration factors, the increased dilution of the draw solution negatively impacted flux. In bench tests with real wastewater, the TFC chemistry had a ≈5x higher water flux (9.7 vs. 1.9 L/m2-h) and a ≈3x lower specific reverse solute flux (192 vs. 551 mg/L) compared to the CTA chemistry. However, both membranes showed less than 5% fouling and a specific forward organic solute flux of less than 0.5 mg/L of total organic carbon (TOC). Pilot testing for >50 h showed stable performance, comparable to bench scale data and model predictions.