For decades, oil refineries have been required by law to report their annual emissions of benzene – a dangerous carcinogen that evaporates from petroleum – and other pollutants to state governments and the U.S. Environmental Protection Agency, based on company estimates.
Then, in 2018, EPA imposed regulations that required refineries for the first time to actually monitor the level of benzene at their fence lines, as the pollutant drifts from the plants out toward the surrounding community.
As it turns out, the data collected by the fenceline monitors suggest that some refineries may have been significantly under-estimating and underreporting their benzene emissions for years. That means neighborhoods downwind could have been facing higher risks of cancer than long thought.
The Environmental Integrity Project’s Center for Applied Environmental Science (CAES) worked with Dr. Andrew Gray and Dr. Ranajit Sahu, two experts in air pollution modeling, to compare the benzene concentrations measured at the fence lines of three refineries in 2019 – two in the Houston area, one in New Mexico – to benzene emissions estimated by each of those refineries and reported to EPA. Dr. Gray holds a doctorate in environmental engineering from the California Institute of Technology and Dr. Sahu holds a doctorate in mechanical engineering from the same. Both researchers have over thirty years of experience.
Comparing the predicted concentrations of benzene at these facilities’ fence lines – based on the company’s reporting of emissions – to the levels of benzene actually measured at the fences, the experts concluded that the companies had underestimated their emissions by as much as 28 fold.
Moreover, the refineries’ reported annual emissions also did not capture the magnitude of large, short-term spikes in emissions that were identified by the fenceline monitors.
Two events, one at Chevron’s Pasadena Refinery along the Houston Ship Channel and the other at HollyFrontier’s Navajo Refinery in New Mexico, were particularly large.
“The model results show that during a peak emission ‘upset’ event that occurred at the Pasadena refinery during a two-week period in late October 2019, the actual benzene emissions were over 200 times the reported annual average emission rate,” Dr. Gray said. “The resulting maximum hourly average benzene concentration at the Magnolia Court Residences, located just over one kilometer from the refinery, was greater than 11 micrograms per cubic meter, which would likely result in significant health impacts.”
Air Alliance Houston, an organization committed to reducing the public health impacts of air pollution and who partnered with CAES and EIP to study the emissions at the Houston area refineries, expressed concern that this underreporting could contribute to underestimation of the cumulative health risks in nearby communities exposed to several industrial pollution sources.
“These troubling results show that downwind neighborhoods may be exposed to even higher levels of benzene than previously reported, beyond EPA’s lifetime safety limits,” said Dr. Inyang Uwak, Research and Policy Director for Air Alliance Houston. “This puts people at risk of major health conditions such as leukemia and other blood disorders. When industry downplays their pollution, communities suffer.”
In 2018, EPA began requiring oil refineries to measure benzene concentrations at their fence lines. This fenceline monitoring requirement was the result of a 2012 lawsuit that EIP, Air Alliance Houston and other allies filed against the EPA to better protect communities immediately downwind from refineries.
Benzene is known to cause leukemia and other illnesses at relatively low concentrations. The EPA benzene monitoring program requires that refineries report benzene concentrations at the fence lines of their facilities, providing critical information for evaluating health risks in adjacent communities, which are disproportionately low-income and communities of color. The program also requires refinery owners to investigate and address the source of the elevated benzene concentrations when the net annual average exceeds an action-level established by EPA (9 micrograms per cubic meter).
Two of the refineries analyzed by Dr. Gray and Dr. Sahu, LyondellBasell’s Houston Refining and Chevron’s Pasadena Refinery, are located along the Houston Ship Channel east of Houston. The area is densely populated – over 1,500 people live within one mile of each facility – and the communities are predominantly Latino. While there are multiple sources of air pollution along the heavily industrialized channel, Houston Refining and Pasadena Refinery are two of the larger sources of benzene near the cities of Galena Park, Pasadena, and other surrounding neighborhoods.
The third refinery, HollyFrontier’s Navajo Refinery, is in Artesia, New Mexico, where it is the only large point source of benzene. Artesia is a community of about 12,000 in southeastern New Mexico and approximately half of the residents are Latino. The refinery is on the eastern edge of the city and about one fifth of the city’s population lives within a mile of the refinery. These residents are disproportionately low-income.
At Houston Refining, the average of the annual benzene concentrations at the fenceline monitors was seven times greater than the average of the modeled benzene concentrations based on the companies’ reported emissions. (And this was even after benzene from nearby sources, like other industrial facilities or traffic, was subtracted out, leaving a net benzene level that came only from the refinery itself.)
At the Pasadena Refinery, the net observed concentrations of benzene at the fence line were 28 times greater than modeled concentrations based on the company’s reporting. Observed net concentrations at the Navajo Refinery’s fenceline were 27 times greater than the estimate based on reported emissions.
The predictions of fenceline concentrations by Dr. Gray and Dr. Sahu were made using software developed by EPA and the American Meteorological Society for this purpose, the American Meteorological Society/Environmental Protection Agency Regulatory Model (AERMOD). This same modeling software is used by EPA to assess health risks associated with air pollution, using National Emissions Inventory (NEI) data. The Air Toxics Screening Assessment (AirToxScreen) provides a rough estimate of the cumulative cancer risk associated with air pollution, which includes exposure to benzene and many other air pollutants not only from refineries, but also from other point and mobile sources. The NEI is compiled using the same emissions data that were used to for the models developed by Dr. Gray and Dr. Sahu.
EPA uses AirToxScreen to, among other things, target risk reduction activities, better understand risks from air toxics, and improve data in emissions inventories. Those purposes are undermined if AirToxScreen underestimates exposure and risk because the National Emissions Inventory underestimates emissions. The discrepancy between measured and modeled values found by Dr. Gray and Dr. Sahu demonstrates the importance of improving the data in EPA’s emissions inventories, because they do not currently line up with observed ambient benzene concentrations. EPA should use all the data at its disposal – community monitors, fenceline monitors, fenceline monitoring root cause analyses, and models – to understand where the National Emissions Inventory is failing and make improvements to it.
In addition to underestimated annual emissions, the analysis of the fenceline monitors identified at least one short-term release that exposed neighboring residents to benzene concentrations that exceeded one-hour health guidelines. A leaking tank at the Navajo Refinery in New Mexico released large quantities of benzene from late March through May of 2019. Based on the reported concentrations at nearby monitors, Dr. Gray estimated that over eight tons of benzene were released during this period, more than three times the total benzene emissions reported by the refinery for the year.
Because of the fenceline monitoring program, HollyFrontier was required to investigate the cause of the high concentrations at the monitors, leading to identification of the leak and removal of the tank from service. However, while the tank was in service, nearby residents were exposed to dangerously high benzene concentrations. Hourly concentrations of benzene at many homes, schools, and parks were likely greater than 27 micrograms per cubic meter, an exposure threshold established by the State of California. To prevent health impacts, this this exposure level should not be exceeded for more than one hour. While the leaking tank was in operation, the threshold was likely exceeded by more than ten times over dozens of one-hour periods at multiple community locations near the refinery.
The health effects of benzene are not simply theoretical. In 2010, a flaring event at a refinery in Texas City released roughly 8.5 tons of benzene (along with large quantities of other chemicals) over a period of 40 days. This release was associated with a range of toxic effects in local children including unsteady gait, memory loss, headaches, altered blood cell counts, and signs of liver toxicity.
This event demonstrates the critical importance of the fenceline monitoring program, not only for evaluating the accuracy of annual emissions estimates, but also for identifying and addressing dangerously high concentrations of pollutants. When fenceline monitors exceed the EPA action level, owners should immediately begin hourly monitoring at the fenceline location in question to ensure that the refinery is not contributing to acute health risks.
The event further demonstrates the importance of local air quality monitors in communities downwind of refineries. While there are several local monitors in the communities around the Houston Ship Channel that record long-and short-term concentrations of benzene and other pollutants, there are no local monitors in Artesia, New Mexico. A monitor there could have alerted the community to air pollution associated with the leaking tank and warned residents of exposure risks.
Lead photo: A Sept. 2, 2017, photo shows the refinery in Pasadena, Texas, formerly owned by Brazil's Petrobras. Chevron purchased the facility in 2019. (AP Photo/Frank Bajak)