Use of Reactive-Transport Models in Field Studies: Experience with the PHAST Simulator David Parkhurst and Ken Kipp U.S. Geological Survey Denver, CO
Topics The PHAST Simulator Field Studies –Arsenic in the Central Oklahoma Aquifer –ASR in Charleston, SC –Phosphorus at Cape Cod, MA B-Z oscillating reactions Summary
PHAST 3D Reactive-Transport Simulator HST3D—Flow and transport PHREEQC—Chemistry Operator splitting—Sequential Non-Iterative Approach Chemistry Transport Flow Chemistry Transport Flow
Flow and Transport Point-distributed finite-difference grid Boundary conditions –Flux –Leaky –Specified value –River –Well Constant temperature Constant density
Chemistry Ion-association or Pitzer aqueous model Mineral equilibrium Surface complexation Ion exchange Solid solutions Kinetics –Explicit ODE (Runge-Kutta) –Implicit ODE (CVODE)
Parallelization Single processor: Flow and transport Multiple processors: Chemistry Data passed using MPI processors Model grids up to 200,000 nodes hours of clock time Allows field-scale modeling Transport Flow Transport Flow Chemistry Cells
Arsenic in the Central Oklahoma Aquifer Arsenic mostly in confined part of aquifer Arsenic associated with high pH Flow: unconfined to confined back to unconfined
Arsenic in the Central Oklahoma Aquifer Chemical analyses Carbon-14 age dating Microscopic examination of sediments Cation-exchange measurements Selective extractions for arsenic Water levels (Ground-water flow model) Available data
Geochemical Reactions Brine initially fills the aquifer Calcite and Dolomite equilibrium Cation exchange 2NaX + Ca+2 = CaX2 + 2Na+ 2NaX + Mg+2 = MgX2 + 2Na+ Surface complexation Hfo-HAsO4- + OH- = HfoOH + HAsO4-2 Desorption at pH > 8.5
Simulated Arsenic Concentrations in Central Oklahoma
Charleston, South Carolina
Aquifer Storage Recovery— Charleston, SC Well logs 2 Aquifer tests 4 ASR cycles Conservative break-through data Periodic chemical analyses Quantitative X-ray mineralogy Available Data
Dispersion Constant dispersivity Dispersion adjusted by contrast in hydraulic conductivity
Simulation of an ASR Cycle
Predicted Recovery Efficiency
1, 10, 100 Year Bubbles
Phosphorus Transport at Cape Cod, MA
Column experiments—PO 4, cations, O 2 Flow and transport parameters Mineralogy Tracer tests Water chemistry with time and space Microbial processes Isotopes Available Data—Everything
Reactions Sorption—PO 4 Sorption—Cations Mineral equilibria –Fe oxyhydroxide –Mn oxide –Fe(3) phosphate –Fe(2) phosphate Kinetic decomposition of organic matter
PHAST Simulation of Column Experiments
Fit of Surface-Complexation Constants with UCODE Log K = 26.7 Sites = 3.0e-3 sites/L Log K = -1.8 Sites = 23.0e-3 sites/L Log K = 4.1 Log K = -7
Phosphorus, mol/L Evolution of Phosphorus Plume at Cape Cod Sewage disposal during years 1-60
MeasuredSimulated
Predicted P Load to Ashumet Pond
Belousov-Zhabatinskii Recipe SpeciesConcentration Malonic acid 0.2 M Sodium bromate 0.3 M Sulfuric acid 0.3 M Ferroin M
B-Z Definitions X [HBrO 2 ] Y [Br - ] Z[Ce(IV)] A [BrO 3 - ] B[Organic] P[HOBr] ReactionRate A + Y = X + P k 3 [H + ] 2 AY X + Y = 2P k 2 [H + ]XY A + X = 2X + 2Z k 5 [H + ]AX 2X = A + P k4X2k4X2k4X2k4X2 B + Z = 0.5 Y k 0 BZ Kinetic Rate Expressions
B-Z—Concentration with Time
B-Z Time Series of Petri Dish
Conclusions Modeling results –Understanding natural systems—Oklahoma –Designing engineered systems—South Carolina –Predicting long-term effects—Massachusetts Modeling has a weakest link –Flow—Oklahoma –Transport—South Carolina –Reactions—Massachusetts Data requirements –Field—Aquifer tests, tracer tests, logging, chemical samples –Laboratory—column experiments, extractions, mineralogy –Resolving uncertainties is expensive B-Z, Kindred and Celia link to biological processes