Arsenic behavior in SW Florida ASR systems and its expert modeling

Pieter J. Stuyfzand1,2 and R. David G. Pyne3
1KWR Watercycle Research Institute, PO Box 1072, 3430 BB Nieuwegein, Netherlands
 Email: pieter.stuyfzand@kwrwater.nl.
2VU University, Amsterdam, Netherlands
3ASR Systems LLC, Gainesville, Florida USA, Email:  dpyne@asrsystems.ws

Abstract
Aquifer Storage Recovery (ASR) wellfields in karstified limestone have been operating at many locations in Florida since 1983. Elevated arsenic concentrations (up to 220 ug/L) became evident at some sites in 2001. When in 2005 the water quality standard for As was lowered from 50 to 10 ug/L, ASR became a much disputed water storage technique, instigating several investigations. Pyrite oxidation during injection and storage was identified as the main driver behind the As problem, with subsequent As desorption from neoformed ferric (hydr)oxides (HFO) and reductive dissolution of HFO especially during recovery.
Nevertheless, many uncertainties remained due to the complex As behavior, heterogeneity of karstic limestones, and doubts about the representativity of the hydrochemical data collected.

     A thorough data analysis revealed new As behavior patterns at Florida ASR wellfields, which are presented here. Comparison with ASTR and ASR systems in the Netherlands, showed the particular role of low Fe2+ concentrations and high Cl-, H2S and SO42- concentrations in Florida.

     The improved insights stimulated to construct an expert model for predicting As behavior in SW Florida ASR systems. This model is presented, tested against field data (from 6 ASR well fields with a total of 52 wells, each with 3-17 ASR cycles) and discussed including its merits and limitations.
It consists of an analytical model which calculates the As concentration by a sine function with decreasing amplitude during subsequent ASR cycles. The amplitude is determined by the As peak calculated via a mass balance equation which multiplies the average As content of pyrite with the total amount of oxidants in the recharge water oxidizing pyrite (after subtracting competing oxidation reactions).

     The resulting model provides a means to predict trends in As concentrations on each site, old or new, and the effects of changes in ASR operation, like reducing the oxidant input or recovery efficiency.

Subject Area:             Geochemistry during Infiltration and Flow
Preferences:              Oral presentation. If competitive with other contribution by Stuyfzand [Multitracing], which also aims at oral, then this one (Arsenic) poster, the other oral.