Skip to Content

News

The Satartia CO2 Pipeline Rupture: 2-3 Times Larger Than Reported?

Photo of Paul Blackburn giving a presentation on Keystone Pipeline System.

By Paul Blackburn

News July 8, 2026

Site of 2020 Denbury (Exxon) CO2 pipeline rupture near Satartia, MS. (Image: Failure Investigation Report – Denbury Gulf Coast Pipelines, LLC by Pipeline and Hazardous Materials Safety Administration OPS Office of Pipeline Safety – Accident Investigation Division)

The Satartia CO2 Pipeline Rupture: 2-3 Times Larger Than Reported?

Paul Blackburn, Bold Alliance
July 8, 2026

On February 22nd, 2020, a carbon dioxide (CO2) pipeline operated by Denbury Gulf Coast Pipelines LLC (Denbury) ruptured approximately one mile from the center of Satartia, Mississippi. According to the 2020 Census, Satartia is reported to be the smallest incorporated municipality in Mississippi with a population of just 41, although additional people live in the unincorporated areas surrounding it. The word “satartia” may derive from a combination of Choctaw words meaning the “pumpkin place.” Before the pipeline rupture, one of the village’s claims to fame were reports that the nearby vertical lift bridge over the Yazoo River is haunted, perhaps by members of the Yazoo Tribe murdered by the French, or perhaps by soldiers from one of the 29 vessels sunk in the river during the Civil War. Unlike these ghost stories, the Denbury CO2 pipeline rupture was all too real – causing the evacuation of 248 people, hospital treatment for 49 of these, and long-term medical harm to those closest to the rupture. Its memory continues to haunt the community. 

Denbury ultimately claimed that the rupture released 31,405 barrels of liquid CO2.  But, is this number accurate? This blog reports on new engineering estimates showing how much CO2 spewed from the pipe. But, an investigation of this number is also helpful to increase understanding of CO2 pipeline ruptures, pipeline company accident reporting, and the efficacy of federal pipeline accident reporting requirements in the U.S. Code of Federal Regulations (49 C.F.R. § 195.50).  

Let’s start with some background on the rupture from the Failure Investigation Report (FIR) prepared by the Pipeline & Hazardous Material Safety Administration (PHMSA), a part of the U.S. Department of Transportation. Denbury’s recorded pipeline pressure data shows that the pipeline ruptured at 7:06 PM on Saturday, February 22, 2020, while it was transporting CO2 from Denbury’s Barksdale, Gluckstadt, and Trace CO2 production plants to a number of oilfields where the CO2 was used in enhanced oil recovery operations. Within minutes of the rupture, Denbury turned off its pumps, and began closing two valves to isolate the rupture site and prevent CO2 from continuing to flow from the processing plants and oilfields toward the rupture. Specifically, Denbury needed to close one upstream valve at the Tinsley Pump Station and a valve downstream from the rupture called the Satartia valve. While Denbury was able to close the Satartia valve by 7:15 PM (within 9 minutes), it was unable to close the Tinsley valve fully until 10:20 PM, one hour and 14 minutes after the rupture, thereby allowing a substantial volume of CO2 to flow from upstream CO2 sources to the rupture site.  

Here are a few basic concepts related to how much CO2 is released when a pipeline ruptures.  Pipelines have multiple valves typically located miles apart. If a pipeline ruptures, the operator remotely turns off any operating pumps and then closes all of the valves along the pipeline.  It cannot close valves until the CO2 stops moving. This procedure is intended to isolate the damaged pipeline segment and prevent the entire pipeline from draining out. There is no way to plug a pipeline rupture site or otherwise stop the release of all of the CO2 contained within a damaged pipeline segment. All of the CO2 in the damaged segment will flow out of the pipeline.  

If a pipeline were to rupture at a time when the valves are closed, then the volume of CO2 released would be simple to calculate, because it would be the same as the inside volume of the ruptured pipeline between the nearest isolation valves. This situation rarely happens, because valves are open when pipelines are operating. It takes time for the operator to confirm that a rupture has occurred, turn off the pumps, and close valves. During this period, CO2 will continue to flow through adjacent segments toward the rupture site. Thus, the total volume of CO2 released is a combination of the volume of CO2 that flows out of adjacent pipeline segments before valves are closed, plus essentially all of the CO2 remaining in the damaged segment after the valves are closed.  

While one may think that pipeline companies can easily and precisely calculate how much product is lost based on flow meters and other measuring equipment, estimating actual losses may require relatively complex modeling of how CO2 moves within a pipeline following a rupture, taking into account any failure to close either an upstream or downstream valve. This being said, for emergency planning purposes, it is entirely possible to generate accurate predictions of a worst-case release spill for any location based on: (a) the inside diameter and length of a pipeline segment between isolation valves; (b) the estimated time to identify a rupture; (c) the time required to turn off operating pumps and the maximum pumping rate; (d) the time required to close all open valves; and (e) a safety factor that accounts for a possible failure to close valves within expected timeframes.  

How much CO2 did Denbury claim was released during the Satartia rupture? Two hours after the rupture at 9:06 PM, Denbury told the federal National Response Center that an estimated 222 barrels of CO2 was released. On February 24 at 6:56 PM, about two days after the rupture, Denbury claimed that it spilled 21,873 barrels of CO2. On November 25, 2020, over nine months after the rupture, Denbury changed this figure to 31,405 barrels, and asserted that 9,532 barrels of this amount were released “unintentionally” and 21,873 barrels were released “intentionally.” FIR, Accident Report No. 20200087 – 34574, page 1 (signed November 24, 2020). However, in footnote 2 on page 3 of PHMSA’s FIR, the report states: “The actual release volume likely exceeded this amount due to a valve operation error, however, Denbury has not confirmed and reported any new release volume to PHMSA.”  

In order to understand why PHMSA believed that more CO2 was likely released, Bold Alliance submitted a Freedom of Information Act request to PHMSA, seeking documents about the volume of CO2 released by the Denbury rupture. In response, PHMSA provided a May 2, 2022, report (see below) by Trimeric Corporation to Denbury (Trimeric Report) containing three different estimates of CO2 losses using different engineering methodologies. Frankly, the data tables in the report are somewhat hard to follow, so here is a narrative description that should help those who wish to read it.

Initially, the report identified four time periods:

  • 19:06 to 19:17 (7:06 PM to 7:17 PM): the initial time period after the rupture, but before any valves were closed. During this period CO2 flowed toward the rupture from both upstream and downstream segments. This time period ends when Denbury closed the Satartia valve, stopping flows from downstream segments toward the rupture.
  • 19:18 to 19:30 (7:18 PM to 7:30 PM): the time period during which the downstream Satartia valve was closed, but the upstream Tinsley valve was partially open, allowing CO2 to flow from the Barksdale, Gluckstadt, and Trace CO2 production plants toward the rupture. This time period ends when Denbury closed valves to stop flows from the CO2 production plants.
  • 19:31 to 22:25 (7:31 PM to 10:25 PM): the time period during which the Tinsley valve remained open, but Denbury slowed the volume of CO2 flowing through it by closing additional upstream valves to isolate the Barksdale, Gluckstadt, and Trace CO2 production plants. This time period begins when Denbury closed valves at the CO2 production plants, and ended when Denbury workers manually closed the Tinsley valve.
  • 22:26 to 23:06 (10:26 PM to 11:06 PM): the time period during which both the Satartia and Tinsley valve were fully closed, thereby isolating the ruptured segment.  It begins when the Tinsley valve was fully closed, and continued until essentially all of the CO2 in the ruptured segment drained out.  

The report then estimated CO2 release volumes for each time period, using three different engineering methodologies (see tables attached to the report, paying particular attention to the time periods noted above). A description of these estimating methodologies is beyond the scope of this blog, but suffice it to say that they were based on pressure, density, and flow restrictions in the pipeline. The report then summed the volumes for each time period for each methodology, to calculate three different spill estimates. (See first page of the tables in Trimeric Report; totals on bottom right.)   

Here’s a summary of the Trimeric Report’s three estimates of the barrels of CO2 released by the rupture and, for comparison, Denbury’s November 2020 loss estimate:

Description Estimated Barrels Released
Method 1 Hydraulic Calculations Based on Pressure Profiles 65,925
Method 2 Density Profile in Pipelines 118,149
Method 3 Choked Flow through Pipelines 104,593
Denbury  11/25/2020 Estimate 31,405

Trimeric Method 1 estimated that more than double the amount of CO2 was released as compared to Denbury’s November 2020 claim. Methods 2 and 3 each estimated that Denbury released between three and four times more CO2 than reported.  

The Trimeric Report found that the pipeline segment that ruptured measured 34,795.2 feet from the upstream Tinsley valve to the rupture site, and 15,523.2 feet from the downstream Satartia valve to the rupture site, for a total length of 50,318.4 feet. Given that the inner diameter of the pipeline was 23.062 inches, it found that the interior volume of the damaged segment was 145,965 square feet, which equates to a liquid volume of 25,999 barrels of CO2. This amount is the total amount of CO2 that poured out of the damaged segment after Denbury closed the upstream and downstream valves. This is the minimum possible loss, assuming all the valves were closed at the time of the rupture, which they were not. All of this loss was unintentional. 

PHMSA also released one page containing a fourth spill estimate, entitled “Refill Calculations” (provided below), based in part on measuring the number of barrels of CO2 pumped into the pipeline after it was repaired to return the pipeline to the operating pressure at the time of the rupture. This methodology estimated that 49,207 barrels needed to be pumped into the upstream Tinsley segment, and then it added this figure to the Trimeric Report’s Satartia segment loss of 25,999 barrels, and downstream Redwood segment loss of 4,117 barrels, for a total loss of 79,323 barrels. The Refill Calculations document does not expressly say who authored it, but since Denbury is the only entity that would have known how much CO2 was pumped back into the Tinsley segment, we can assume that the Tinsley segment loss estimation came from Denbury.

That four methodologies estimated spill amounts much higher than the loss reported by Denbury to PHMA explains why the Failure Investigation Report stated that the actual release “likely exceeded” Denbury’s reported amount of 31,405 barrels. These estimates are also evidence that Denbury seriously low-balled its spill estimate.

What lessons can we learn? First, the more CO2 released, the more risk to impacted communities. The larger the release, the larger the kill and injury zone, the more people put at risk, and the harder rescue becomes. If the volume of CO2 released by the Denbury rupture was two to three times more than reported, this might explain why CO2 concentrations on the banks of the Yazoo River in Satartia – more than a mile from the rupture site – were high enough to cause people to pass out. It is important to have accurate estimates of spill size to allow for accurate post-accident modeling to better understand the potential real world spread of CO2 plumes and their impacts. We need to have a clear understanding of CO2 pipeline risks and impacts to ensure that federal safety standards are sufficiently robust to protect communities and avoid potential mass casualty events.

Second, Denbury’s legally required accident reports demonstrate that a pipeline operator may initially dramatically underreport release volumes, as Denbury did here with its initial estimate of just 222 barrels. When a pipeline ruptures, first responders should not rely on a pipeline company’s initial spill estimate, but rather should respond assuming a worst-case discharge that includes: (a) the release of all of the CO2 in the ruptured pipeline segment; (b) the volume of CO2 that would flow from upstream and downstream segments into the damaged segment before pumps are shut down and valves closed; and (c) an additional safety margin to take account of a possible valve failure. A worst-case discharge amount should be estimated for each pipeline segment then used as the basis for high quality computational fluid dynamic (CFD) plume modeling to estimate how far dangerous CO2 concentrations may spread from a rupture site.  Such plume modeling provides a reliable basis for emergency response planning. Trusting a pipeline company’s initial spill estimate may prove to be a fatal mistake.

Third, this case demonstrates a significant limitation in the PHMSA accident dataset: it’s based entirely on unverified pipeline operator spill estimates. Even when PHMSA has substantial evidence that a spill is much larger than reported by a company, PHMSA does not correct company-reported spill amounts. When a CO2 pipeline ruptures, all of the spilled CO2 evaporates, with the result that accurate spill loss estimates will come only from company data. Yet, pipeline companies may have an incentive to underreport the size of their releases, possibly to reduce administrative fines and civil liability. Unlike petroleum spills that result in visible impacts to water and land and hazardous waste disposal documentation, the physical evidence of CO2 spills, especially small ones, literally evaporates, creating opportunities for CO2 pipeline operators to lowball spill size estimates and hide the existence of spills, particularly small ones. The PHMSA accident data for CO2 pipelines should be assumed to underestimate spill sizes and to underreport the number of small CO2 spills. 

A better understanding of the real risks created by CO2 pipelines combined with active community participation in pipeline siting and emergency planning can help us protect our families and communities from potential CO2 pipeline disasters. Be engaged!

Trimeric Loss Estimate Report for Denbury CO2 pipeline rupture at Satartia, MS (obtained by Bold Alliance via Freedom of Information Act (FOIA) request):

Pipeline Fighters Hub