Spinning PFAS Straw into Gold in an Age of Uncertainty: How a forensic conceptual site model can help you manage PFAS liabilities even when your data is limited

The Problem

Per- and Polyfluoroalkyl Substances, PFAS for short, pose significant potential liabilities to those dealing with PFAS contamination. Those liabilities might include risks to human health or the environment, which can lead to high cleanup costs, enforcement and/or litigation. In spite of their long use and occurrence nearly everywhere, there is still much uncertainty about the science, treatment, and regulation of PFAS. We don’t have a list of low-cost technologies to choose from to treat PFAS contamination. PFAS are extremely difficult to destroy. The cleanup concentration targets we have for soil and water can be difficult to achieve because they are extremely low. Consequently, we are faced with a contaminant that is:

• Widespread
• Complex, and
• Difficult to manage.

PFAS also have

• Extremely low cleanup concentrations
• Evolving regulations at the state and federal levels
• Only high-cost clean-up technologies currently available, and
• The potential for enforcement and litigation.

The result is a regulated community that is likely to hold off on assessing PFAS contamination as we wait for more certainty and better options. Unfortunately, that isn’t always possible when agencies or other third parties force an investigation of your site for PFAS in soil, groundwater, or surface water.

How can you realistically estimate how much PFAS contamination you have, where it came from, who you may have impacted, or who may have impacted you with their PFAS? By using a cautious, phased approach to that investigation you can manage the amount of data you collect to:

• Better control the process
• Ensure valid results
• Gain time for improved scientific developments, and
• Gain time for regulations to become more established so you at least know the “rules of the game”.

Once the ball is rolling and you have begun investigating your site, you need to know your potential liabilities so you can develop a data-driven strategic plan to manage them. But how?

The Solution

The answer is PFAS Forensic Conceptual Site Model Analysis using established machine learning algorithms and other techniques. PFAS Forensic Analysis is relatively new and developing. But it is still a powerful tool to identify significant differences in PFAS in soil, groundwater, and stormwater and relate those back to potential sources.

For example, at an industrial site, we used a variety of graphical and multivariate statistical analyses to separate out PFAS that may have come from different sources. Some of the specific techniques we used include Principal Components Analysis (PCA), Cluster Analysis, and Non-Metric Multidimensional Scaling (NMDS). These are rigorous mathematical techniques that identify similarities and differences between samples so we can group them into similar “buckets”. We also performed graphical analysis of digital chromatograms, radar plots, and other graphical tools to visually “check the math” to confirm we had the right “buckets”.

If we have laboratory analytical data for PFAS in potential sources, then we can also include them in the statistical and graphical analyses. This lets us directly match potential sources with samples. PFAS sources that land in the same “bucket” as PFAS samples may be a good match, leading to the next step where we look for a clear pathway for migration from that source to those samples.

For that migration pathway search, we added these groups of similar and different PFAS samples to maps. We used a site map and area map to visually see if the sample “buckets” make sense and were located near potential sources both on- and off-site. We also created a 3D model to look for potential pathways through soil and groundwater that could link samples to potential sources.

Some of the specific tools we used for this include 3D geologic and hydrogeologic modeling to build a three-dimensional picture of the soil, groundwater, surface water, and PFAS contamination. We plotted the “buckets” in this model to see if there were migration pathways between the “buckets” and potential sources. The model let us cross-check the match of samples to potential sources under real world conditions. In our experience, building a rigorous, data-driven visual story to pass the “common sense” test of practical believability can be just as important as passing the “Occam’s Razor” test of scientific validity.

The Payoff

We combined our knowledge of known or suspected PFAS sources with the PFAS sample groups from the statistical and graphical analyses, then evaluated this knowledge within the model of soil and groundwater to identify a combination of at least one on-site potential source and at least one off-site potential source. This information can then be used to develop a strategic risk management plan to guide decisions around further assessment, remediation, and potential enforcement and/or litigation, significantly improving your ability to control your potential PFAS liabilities and identify other potentially responsible parties.

If you are interested in learning more about how we can help you identify and manage your potential PFAS liabilities, even if you have only limited data, contact me at mhawthorne@geiconsultants.com.