Challenges for inland desalination: Assessing the lifetime expectancy of injection
wells using geochemical modeling

Elizabeth Keating1, June Fabryka-Martin1, Patrick Brady2, Richard Kottenstette2

1 Los Alamos National Laboratory, ekeating@lanl.gov, fabryka@lanl.gov, Los Alamos, NM, USA
2 Sandia National Laboratory, pvbrady@sandia.gov, rkotten@sandia.gov, Albuquerque, NM, USA

ABSTRACT

One of the numerous challenges in designing a successful inland desalination plant is management of the concentrate produced by the reverse osmosis (RO) process. One possible option is re-injection into deep wells, but chemical incompatibility between the concentrate and receiving formations could cause an unacceptable reduction in aquifer permeability. Assessing this potential problem would require not only site-specific hydrologic and geochemical data but also an interpretive analytic tool that could be used to predict performance. In this case, the goal is not only to maximize recovery of product water but also to minimize problems with concentrate injection. Here, we report our progress on building such a tool using PHREEQC (Parkhurst and Appelo, 1999) to evaluate the geochemical processes expected to occur during re-injection.

In the arid southwestern U.S., the brackish aquifers that are potential feed waters for desalination exhibit remarkable spatial variation in chemistry. Of waters with total dissolved solids (TDS) within an acceptable range for desalination (< 10,000 mg/l), most are Ca-SO4 type waters; a smaller number are Na-Cl or Na- SO4 type waters. PHREEQC calculations show that these waters are typically saturated or supersaturated with respect to carbonate minerals and many iron oxides. Using simple assumptions about the operating efficiencies of the desalination process, and typical pre-treatment methods such as acidification, addition of anti-scalants, and degassing excess CO2, we simulate the geochemical character of RO concentrates derived from the feed waters in our database. Many of these concentrates are supersaturated with respect to gypsum and are not suitable for injection. Of the waters that are suitable for injection, many are predicted to precipitate significant amounts of gypsum upon mixing with certain receiving formations. We show that the impact of mineral precipitation on formation permeability, however, will be strongly influenced by hydrologic factors and reaction kinetics, which are not addressed here. Detailed study of these factors is worthy of future analysis.