Sand use in nitrogen or carbon dioxide disclosures

Summary: The use of nitrogen or carbon dioxide gas for foam-based fracking fluid can increase the percent of proppant used, suggesting this might be an explanation for the 2000+ disclosures where fracking fluids report greater than 50% sand. We use FracFocus data to explore that suggestion. We conclude that 1) using those gases is indeed linked to a larger percentage of sand in disclosures, 2) but even in the gas-foam disclosures, the sand percentages over 50% of the fracking fluid are highly unusual.

Open-FF has found more than 2,000 disclosures that apparently misreported sand and water percentages in fracking fluid. These disclosures may seriously misrepresent chemical additive quantities. We will be posting a number of examples soon.

While investigating this issue, we were informed by FracFocus Support¹ that fracking jobs that use carbon dioxide (CO₂) or nitrogen (N₂) gas as foaming agents are able to tolerate higher percentages of sand. In this brief post, we use FracFocus data to examine how such foaming gas use correlates with the patterns of sand use.

Proppants, materials that “prop open” the fractures created during fracking, are essential to a fracking job’s success. Sand is by far the most common proppant. Although different sources and grain sizes are reported, this range of material is all classified chemically as “Quartz (silicon dioxide)²“. While other materials are also used as proppants, companies used sand in 94% of disclosures. In 2023 alone, companies used more than 200 billion pounds of sand³.

Evaluating sand use with CO₂/N₂

We separated all FracFocus disclosures into two groups: those disclosures with carbon dioxide or nitrogen gases and those disclosures without⁴. The “with CO₂/N₂” group consisted of 3,620 disclosures and the “without CO₂/N₂” group consisted of 165,251 disclosures. We divided each set into 2% bins and converted them to percentages. The first figure shows how the distributions of these two sets differ:

Indeed, it is clear that the set with CO₂/N₂ tended to have more sand than the set without. The mode of the “with” group is 18-19% whereas the mode of the “without” group is only 10-11%. Thus it makes sense for us to treat the “with CO₂/N₂” separately when we are looking at anomalously high sand percentages.

Relationship to sand dominance

However, the disclosures that we’ve examined and are highlighting in the sand-dominance blog series are quite different from the typical disclosures:

While the blue bars of the “without CO₂/N₂” describe the typical profile of sand in a disclosure, the target of our sand-dominated analysis is the long tail (highlighted in red) where we find more than 2,000 misreported disclosures⁵.

Even those disclosures with CO₂/N₂, sand percentages of 50% and above are also far into the distribution’s tail⁶:

In the series presented so far, we have excluded disclosures that reported some CO₂/N₂ use. It seems likely that even in those disclosures that include gas, reports of sand dominance are also errors.

---

1. Thanks to FracFocus Support for offering this suggestion when we asked about the sand-dominated disclosures. ↩︎
2. or by the CAS number 14808-60-7 ↩︎
3. Disclosures that are sand-dominated or otherwise don’t have water as the primary ingredient are excluded from this mass calculation. Furthermore, many FracFocus disclosures lack enough information to calculate masses. Therefore, this number is likely a substantial underestimate. We used the FracFocus data downloaded on July 18, 2024. ↩︎
4. The separation criteria was simply the presence of either carbon dioxide or nitrogen in the disclosure at any quantity, creating the “with CO₂/N₂” and “without CO₂/N₂” groups. Disclosures without chemical records of sand (CASRN: 14808-60-7) or without any chemical records were excluded. ↩︎
5. Statistically, a sand value of 42.5% in the without-gas distribution has a z-score of greater than 3.0, which is often used to flag outliers. ↩︎
6. Statistically, a sand value of 53.5% in the with-gas distribution has a z-score of greater than 3.0, which is often used to flag outliers. ↩︎