Desalination of inland brackish groundwater and high salinity agricultural drainage water can be accomplished via reverse osmosis (RO) technology. High recovery desalting operation is critical to reduce the challenge of concentrate management. However, recovery is constrained by mineral scaling and thus the drive to develop RO process configuration in which the crippling impact of mineral scaling is alleviated. In the present work a process configuration and operational approach was evaluated for high recovery desalination of inland water of high mineral scaling propensity. A continuous process of chemically-enhanced-seeded precipitation (CCESP) was designed and evaluated under both laboratory and field conditions for its potential integration with reverse osmosis desalination of inland water of high gypsum and calcite scaling potentials. In this process the concentrate from primary RO (PRO) first undergoes a partial lime treatment (PLT) step to promote CaCO
3 precipitation (as calcite) for antiscalant (AS) scavenging. Subsequently, gypsum seeded precipitation (GSP) is accomplished in a fluidized bed crystallizer (FBC) for concentrate desupersaturation. The treated concentrate is then filtered and further desalted in a secondary RO (SRO) to enhance the overall product water recovery. The suitability of integrating GSP with PLT in a continuous manner requires: (1) effective removal of residual antiscalant from the PRO concentrate to avoid retardation of gypsum precipitation, and (2) solids management to ensure stable GSP operation in the FBC given the temporal change in the particle size distribution within the FBC. Accordingly, a novel FBC for GSP was developed along with a numerical model to: (a) describe the mineral scalant concentration profile and particle size distribution along the FBC, and (b) evaluate solids handling strategies for stable CCESP desupersaturation performance. Process model simulations along with laboratory and field studies were conducted for desalination of brackish water of salinity in the range of 11,000-16,000 mg/L total dissolved solids (TDS) in which the gypsum and calcite supersaturation indices were in the range 0.9-1.01 and 6.5-7.8, respectively. The developed FBC/GSP model was first experimentally validated under laboratory conditions using a small-scale CCESP system. The CCESP approach was evaluated with gypsum seed particles in the size range of 70-127 µm narrow size distributions (standard deviation of 10-14%) and FBC solids volume fraction in the range of 0.1-0.3. Model predictions were in excellent agreement with experimental data with respect to the attained level of gypsum desupersaturation attained for the treated GSP product water as quantified by the extent of reaction and gypsum supersaturation index. Controlling the specific surface area of gypsum particles (site/project-specific seed size and mass loading) in the FBC was shown to be critical for attaining steady-state operation with respect to desupersaturation, solids removal rate and the fluidize bed stability. Thus, an optimal protocol was established in which mineral particles were periodically purged from the FBC bottom while fresh seed particles (or recycled ground particles) were added to maintain the required active surface area for heterogeneous gypsum crystal growth and for maintaining a stable fluidized bed.
The CCESP approach was evaluated under field conditions at the UCLA RO desalination field plant at the Panoche Drainage District (PDD) housing. The mobile integrated membrane system (MIMS) located at PDD and housed in a 40 ft.-long ISO container had a plant water production capacity up to ~40,000 gallons/day. The MIMS RO feed pretreatment train consisted of a hydrocyclone, followed by microfiltration using a 300 µm self-cleaning disk filter and a subsequent ultrafiltration unit containing two parallel multi-bore inside-out hollow fiber modules. Raw agricultural drainage feed water was delivered at a pressure of up to 50 psi. The RO system consisted of two RO membrane trains. The first stage pump had a capacity of 24 gallons/min and the interstage pump capacity was 12.5 gallons/min. The first RO stage contained 14 brackish water RO membrane elements of 4 inch diameter and 40 inch length, while the second stage consisted of 7 seawater RO membrane elements of the same dimensions. In the present study, the PRO concentrate was generated from desalting of the agricultural drainage water at a recovery level of about 65%. The PRO concentrate was about a factor of 2.5 above saturation with respect to gypsum. The CCESP process reduced the PRO concentrate gypsum desupersaturation to only ~10% above saturation, which then enabled SRO desalting with antiscalant dosing in the range of 3-4 pm. The overall recovery that was achieved by the integrated PRO-CCESP-SRO desalination process was assessed to be in the range of with ~89%-93% with further enhancement to ~94-96% with a 0.4-3 concentrate recycle (to the CCESP) to FBR particles purge ratio. In addition to achieving high recovery desalting, the CCEPS process enabled highly pure salt (gypsum) harvesting of 2.6-3.6 kg per m3 of RO concentrate. Results of the present study suggest that scaleup of the approach is feasible but will require the development of self-adaptive control to manage potential variability of feed water salinity and pH.
