A new study investigating spills from hydraulically fractured oil and gas uncovered 6,648 spills in just four states over a ten-year period. Part of the SNAP Partnership, the study examined data from Colorado, New Mexico, North Dakota, and Pennsylvania. Significant differences in reporting requirements across states made this analysis difficult. For example, North Dakota required spills of 42 gallons or more to be reported, whereas Colorado and New Mexico required reporting only if spills exceed 210 gallons. Lauren Patterson, a water policy specialist at Duke University, and her coauthors say making this kind of state-level data more uniform and transparent would help regulators and industry reduce the number of incidents. We spoke with her to learn more.

ResearchGate: What motivated your study?

Lauren Patterson: We wanted to look at the impacts of hydraulic fracturing on water, including the risk spills pose to surface and groundwater resources. In the process of gathering and analyzing spills data, we found there was an additional story to be told about state reporting requirements and spills data.

Having a multi-disciplinary team enabled us to not just quantify the risk of having spills, but also to look at the interplay between state reporting requirements and results from the data. We explored state spills data from four states—Colorado, New Mexico, North Dakota, and Pennsylvania—as well as how reporting requirements have changed in those states over time.

RG: Can you tell us what you found?

Patterson: Our study identified 6,648 spills at 31,481 unconventional oil and gas wells in these states between 2005 and 2014. On average, that’s equivalent to 55 spills per 1,000 wells in any given year. We also found that across all states, over 75 percent of spills at these wells occurred within the first three years of a well’s life. This is when drilling, hydraulic fracturing, and the heaviest production occurs. Most spills at these sites occurred when storing materials in tanks and pits, and moving fluids around via flowlines.

Our study concludes that making state spill data more uniform and accessible could provide stakeholders with important information on where to target efforts for locating and preventing future spills. States would benefit from setting reporting requirements that generate actionable information—that is, information regulators and industry can use to identify and respond to risk “hot spots.” It would also be beneficial to standardize how spills are reported. This would improve accuracy and make the data usable to understand spill risks.

RG: What were the most common causes of the spills?

Patterson: Fifty percent of spills, including those spills whose cause was unknown, occurred at tanks or pits, and flowlines. In tanks and pits, the cause of those spills can vary widely, from equipment failure that manifests in a tank overflow due to corrosion, to human error, to lightning strikes. Many of the flowline leaks were due to corrosion or being punctured by equipment. Spill prevention in these instances might include marking flowlines clearly, replacing or inspecting flowlines more frequently, or using more corrosion resistant materials.

RG: How else can spills be prevented? 

Setting reporting requirements and standardizing data collection at the state-level would enhance opportunities for precision management by industry and regulators. For example, where data reveals that spills occur during the transport and transfer of materials, caused by human error, industry could target loading and unloading training for drivers. Or, if industry finds that a lot of valves are freezing in a particular location, and then failing when temperatures rise, they could replace those valves with freeze-resistant valves. Meanwhile, regulators could calibrate inspection efforts to wells in the first three years of life, or wells that have already experienced a spill and to pieces of equipment most at risk for failing.

    Meta Center
    Transforming realities one thought at a time.

    Leave a Reply

    %d bloggers like this: