Formatted Title
Computer Modelling Optimizes PFAS Barrier Performance and Commissioning
Background/Objectives
Background/Objectives. The in situ application of colloidal activated carbon (CAC) in a reactive barrier configuration is now an established remediation strategy for reducing risk posed by PFAS-contaminated groundwater. Computer modelling provides a valuable design tool enabling contaminant flux, dynamic transfer between aquifer compartments, and competitive interactions between PFAS species for activated carbon sorption sites to be quantified. Post-application monitoring data allow the model to be further calibrated and refined. Its use may then be extended through other project phases, and beyond the design utility to subsequent project phases such as performance tracking and optimization. Deviation of field performance data from the modelled trajectory may be observed at certain points within a barrier. This exposes zones that may require engineering attention. Pre-installation design modelling and post-installation model-supported adjustment are analogous to design and commissioning of mechanical systems such as hydraulic containment installations. This presentation provides an illustration of the model-supported commissioning process drawing from a case study of a midwestern PFAS site.
Approach/Activities
Approach/Activities. A former fire training area was impacted with PFAS contamination. The plume was travelling radially from a central location within an oxbow of a moderately-sized river. A CAC barrier was installed in 2019 to stop migration of the PFAS plume into the river. Modelling software was used to interrogate the post-application data. Initial input estimates of governing aquifer parameter values were adjusted based on field data to calibrate the model to the measured performance observations. Example overlay graphs of modelled trends and observed data are presented for time-series and individual sampling event data sets. These were used for communication with the client and for informing appropriate commissioning and optimization adjustments. Deviation of performance from the design trajectory was noted in two areas of the barrier. These were rectified as commissioning adjustments.
Results/Lessons Learned
Results/Lessons Learned. Overall agreement between trendlines predicted by the modelling software and observed performance data was high (Average r2 =0.90; range 0.51 – 1.00). This is despite the natural variability characteristic of a real-world data set that would reduce the r2 coefficient even if the trendline was perfect. The quantitative synthesis of advection, retardation and equilibratory dialog between aqueous, CAC-sorbed, natural organic carbon-sorbed and low-transmissivity zone compartments was adequate to describe the data set. The modelling revealed three distinct performance phases that become more pronounced as distance from the barrier increases – a lag, a principal fast decline and a residual slower decline. Model calibration to performance data and subsequent model interrogation allowed PFAS fate and transport at the site to be better understood and informed predictions of performance timing and extent. In addition, the approach exposed areas where commissioning adjustments were warranted and enabled modelled explorations of potential adjustments to determine the optimum engineering course. Model-supported post-application analysis provides a powerful tool to improve understanding, projection and inform commissioning of PFAS barriers. The engineering control and management capabilities afforded are key predictors of project success.