Don’t Inject Blind: Using 3D Statistical Modeling and Python to Create Targeted Injection Plans

Register to day at: https://register.gotowebinar.com/register/7607039797659747415

Background/Objectives

According to the USEPA Superfund Remedy Report published in January 2023, in-situ treatment remedies have become the most popular form of active treatment on sites with impacted groundwater. Further, both bioremediation and chemical treatment are, by far, the most widely implemented form of in-situ treatment, and both are typically applied through injection techniques.

However, many in-situ injection projects still rely on monitoring well data to design the remedy. Overreliance on this type of data can be problematic since monitoring wells typically provide concentration data that is weighted by the different hydraulic conductivities of the soils within the screened interval, and most monitoring well data tells us nothing about the vertical distribution of the impacts.

Using both high resolution data collection techniques and 3D statistical modeling technology, targeted injection plans can be created to optimize injection strategies.

Approach/Activities

Using both Visual Basic programming language in Excel and Earth Volumetric Studio (EVS) by C-Tech, targeted injection plans were created at three sites from different types of data sets:

• Groundwater Analytical Data from Discrete Samples;
• Membrane Interface Probe (MIP) Data; and
• Membrane Interface Probe and Hydraulic Profiling Tool (MiHPT) Data

Specifically, the analytical data or data from the MIP/MiHPT were statistically interpolated in 3D to define both the horizontal and vertical extent of the impacts at several different concentration levels. The coordinates of the extent of the impacts were then exported to Excel where the locations of the proposed injection wells could be calculated at any spacing using Visual Basic macros.

Using Python scripting, cross section maps were automatically created to visualize injection depths/elevations at every injection well location. Additionally, where available, geologic information from boring logs or HPT data was statically modeled to determine the hydrogeologic distribution of the impacts.

Results/Lessons Learned

• Understandable and actionable injection plans were created from all three data sets. Final deliverables included:
• A fully interactive 3D model where any components could be toggled on/off, including injection well locations, contaminant plumes, and geologic units;
• A series of cross sections depicting the injection depths and elevations of every injection well location; and
• A field-ready injection plan detailing the coordinates of each injection well, injection depths and elevations, anticipated concentrations (for amendment dosing), and the anticipated geologic setting (for injection pressure.

The plans offered a far better solution for targeting subsurface impacts than using monitoring well data alone and because automation technology was implemented (using pre-built applications in EVS, Visual Basic macros, and Python scripts), the plans were produced efficiently and economically.