Get the results of TWRI-funded research through technical and special reports. Find abstracts and full-text online for our reports.


Potential Impacts of Desalination Concentrate on Salinity of Irrigation Water: A Case Study in the El Paso Valley

S. Miyamoto

Winter returnflow has not been fully utilized for crop irrigation in the El Paso Valley. There are, however, emerging interests in utilizing it for urban water supply through desalting. This study examined the potential impact of concentrate discharge on salinity, sodicity, and ionic composition of irrigation water supply, using historical or published records. The analyses performed consisted of the estimate of riverflow rates on river water quality, a review of concentrate and permeate quality from nanofiltration (NF) and reverse osmosis (RO), and the impacts of dilution or blending on water quality.

Riverflow and quality data from the U.S. Section, International Boundary and Water Commission (US-IBWC) were examined first. This analysis has shown that salinity and ionic composition of riverflow can be described by a simple power function as related to the momentary riverflow rate when water samples were taken for chemical analyses. This method provides more accurate estimates of monthly salinity than the use of monthly average flow which has a high degree of variation. In addition, this approximation technique allows for the estimation of river salinity and ionic compositions at any riverflow rates of interest.

A review of published articles on NF processes indicates that there are essentially two types of membranes: one has a low rejection rate for Na and Cl, and the other has a high rejection rate. If the objective is to minimize Na and Cl ions while maximizing Ca and Mg concentrations in the concentrate, the first type is preferred. However, the sodium adsorption ratio (SAR) of concentrate from the first type of NF membrane is also influenced by feed water quality. Typically, the SAR of the concentrate does not change appreciably in water that is rich in SO4, as the rejection rate of SO4 is high, and SO4 ions remain in the concentrate along with accompanying cations. The SAR of the concentrate is not necessarily lower than that of feed water, due to the salt concentration effect on SAR. The SAR value which directly impacts the cation exchange reaction in soils decreases with dilution, but increases due to the increased formation of sulfate-divalent cation ion-pairs. Sodicity of the concentrate from the second type is higher than the sodicity of feed water or that of the concentrate from the first type, and approaches the concentrate composition from a RO process. The most significant changes that take place in the concentrate composition from the first type are an increase in TDS and divalent cations and anions, whereas sodicity and chloride concentrations remain more or less the same as those of feed water.

Permeate from the first type of NF membrane is likely to be higher in Na, Cl, and TDS than from the second type. These elevated salt levels limit the opportunity for blending with the river water, which has elevated salinity and SO4 concentrations, especially at a low riverflow of 5 Mm3/mo or less. Sodicity and the concentrations of Na and Cl in the permeate could also exceed the unofficial water quality guidelines for irrigating urban landscape. If the RO process or the second type of NF membrane is used, the permeate can be blended with river water at nearly a 1:1 ratio. This means that a lesser quantity of water needs to be treated when a RO process is used. If river water high in Na and Cl concentrations is used for blending, the salt load of the concentrate from the NF process can actually be greater than that from the RO process, because of the limited blending possibility. If the NF option is to be retained, a NF membrane with some rejection of Cl ions may be warranted, unless blending water low in Cl is available at or near the site.

Assuming that flow and salinity monitoring data at the Courchesne Bridge are realistic, the disposal of NF concentrate from 5 and 10 MGD membrane processes at a riverflow rate of 5 Mm3/mo may increase salinity of riverwater by around 7 and 16%, respectively, over the existing salinity. This estimate is for a NF membrane with a low rejection for Cl ions, and applies to the situation of low flow periods, around 5 Mm3/mo. At a riverflow rate of 10 Mm3/mo (which is close to the median flow), salinity increases associated with 5 and 10 MGD plants are estimated at 3.5 and 7.0%, respectively. Salinity increases from RO processes or the NF membrane with high Na and Cl rejection rates would be somewhat higher. The use of a two-stage process (the NF first, then the RO process) for the permeate increases the potential for blending, provided that a cost-effective means of disposing the RO reject (dominated by NaCl) can be found at the site in question. The quantity of salts, which may be removed in the second stage, is about ¼ of the total salt loading or about 15 tons of NaCl/day (460 tons/mo) at a 5 MGD processing capacity. Salinity of the blend is reduced when it is mixed with reclaimed municipal effluent, which has lower salinity. However, the effect of concentrate disposal to the mixture of river water and reclaimed water will persist. Impacts of these water quality changes on soil salinity and winter crop production are yet to be analyzed. A possibility exists in some crops that crop response may not be proportional to the projected increases in TDS, as ion activities and species are going to be altered by the discharge of NF concentrate. This aspect is scheduled to be studied in the follow-up phase of this project.

Back to Top