Case Studies

Case Study: Reactive Transport Modeling of Catalyzed Hydrogen Peroxide ISCO at the Field Scale

Client:  Confidential Industrial Client ​
Location:  Northern California ​

Issue:  Catalyzed Hydrogen Peroxide (CHP) ISCO was used to remediate a chlorinated solvent plume at an industrial site. Monitoring well results showed little improvement in concentrations after two rounds of injections.

Action:  Develop a three-dimensional reactive transport model of CHP ISCO to diagnose why existing treatment was mostly ineffectual and to determine the number of injections required to reach remedial targets.

Site Overview:

Volatile organic compounds (VOCs), including tetrachloroethylene (PCE), trichloroethylene (TCE), cis-1,2-dichloroethylene (cis-1,2-DCE) and vinyl chloride were present in groundwater, soil, and/or soil vapor in a shallow groundwater aquifer comprised of saturated silty sand to poorly graded sand, approximately 18 to 24 feet bgs. The plume has approximately 1,000 to 4,000 μg/L of PCE and TCE. Cleanup goals were 530 μg/L for TCE, and 120 μg/L for PCE.

ISCO using Catalyzed Hydrogen Peroxide (CHP) had been selected as the remedy for this site to chemically degrade these VOCs. CHP consists of H2O2 + a chelated iron catalyst. CHP produces a suite of reactive oxygen species (hydroxyl radical, superoxide) which make it effective for many organic species. Limitations of using CHP include a short-lived hydrogen peroxide (H2O2) residual (a few hours to days) due to decay. After two rounds of injections, there was only a 18 to 29% reduction in TCE and PCE showed little to no change. These results cast doubt on future planned remedial actions at the site.

Mutch Associates was retained to diagnose why the first two rounds of ISCO were mostly ineffectual and to determine the number of future injections required to reach remedial targets. A three-dimensional reactive transport model was developed to help answer these questions. The model includes all of the relevant chemical reactions that govern CHP persistence and effectiveness including loss of residual H2O2. The model was constructed using MODFLOW + RT3D using six different chemical species including H2O2, bromide, and dissolved and sorbed TCE and PCE.


Figure 1: Model predicted hydrogen peroxide concentrations in groundwater, resulting from injection of 200 gal of 12% H2O2 at 3.5 gal/min at AC-139U.


Figure 2: Model predicted TCE concentrations in groundwater, resulting from a series of injections of 200 gal of 12% H2O2 at 3.5 gal/min on 25-ft centers. Initial conditions are 1,000 μg/LTCE with retardation coefficient of 3.5.

Model Simulations:

The decay rate coefficient of H2O2 was determined via calibration to monitoring data which indicated a radius of influence of 9 feet (Figure 1). Since reaction of H2O2 with dissolved PCE and TCE is rapid, there was excellent removal of VOCs (> 99%) within this radial distance. Retardation factors for TCE and PCE were estimated at 3.5 and 6.0, respectively from soil foc data. Desorption of TCE and PCE from soil was handled kinetically using a first-order process, based on bench-scale studies of adsorption to sand.

The injection schedule was simulated exactly as it was performed in the field, using the same locations, flow rates, volume of injected fluid and concentrations of CHP. Immediately after injection, TCE is degraded within the 9-ft ROI created by the injected CHP (Figure 2). However, the H2O2 residual is quickly lost and the TCE “rebounds” as a result of desorption from the soil. There is some destruction of TCE and PCE mass, but only a modest reduction in dissolved concentrations after two injection events.

Comparison to Field Data:

Model-predicted results at a monitoring well MW-1 (Figure 3) show only 28% decrease in dissolved TCE, consistent with the observed 18 to 29% reduction. The model was used to simulate six future potential injections using the same injection schedule as the first two. The model predicts TCE would reach clean-up levels after the 6th injection event, while PCE clean-up levels will not be met even after eight injection events (Figure 4). Based on these results, future injections were abandoned.


Figure 3: Model predicted PCE and TCE at MW-1 after two injection events.


Figure 4: Model predicted PCE and TCE at MW-1 after eight simulated injection events.

Summary and Conclusions:

ISCO designs are often “ad hoc” or based on engineering experience. This case study shows that reactive transport modeling can quantitatively assess their effectiveness, thereby providing a means to design, optimize and post-audit ISCO approaches. The best time to utilize modeling is at the remedy selection and/or design phase where it can save significant cost and prevent delays. Reactive transport modeling offers unparalleled insights into what is occurring in the subsurface and can help in-situ treatment systems from falling short of performance objectives or failing altogether.