Air Dispersion Modeling of Coke Battery Emissions and Off-Site Exposure Potential from an Industrial Steelmaking Facility
Project Overview
RoseyFields conducted a technical air-dispersion and emissions analysis for a coke battery and steelmaking-related industrial facility in southeastern Michigan, focusing on how emissions from coke ovens and associated sources could affect nearby residential and community receptors. The project integrated multiple years of facility emissions inventory data, source-term development for criteria pollutants and hazardous air pollutants, and AERMOD-based dispersion modeling to evaluate the magnitude and spatial distribution of off-site air concentrations. Rather than relying on a single annual inventory total or a single modeled source, the work considered both facility-wide annual emissions and coke-oven-specific emissions to assess how routine operations could contribute to chronic community exposure in the surrounding area. The uploaded modeling figures indicate that the analysis included a facility-wide annual concentration plot for “all sources” and a separate isopleth plot focused on coke oven emissions, each developed using EPA-style source terms and multi-year annual averaging.
The project appears to center on EES Coke Battery, LLC, a coke-making operation whose emissions inventories from 2012 through 2015 show substantial annual emissions of combustion-related and particulate pollutants, including carbon monoxide, nitrogen oxides, sulfur dioxide, PM10, PM2.5, volatile organic compounds, ammonia, benzo[a]pyrene, and polycyclic organic matter. The annual inventory reports also indicate that certain pollutant totals varied substantially from year to year, suggesting that source characterization and modeling needed to account for operational variability across the evaluated period. For example, sulfur dioxide emissions ranged from approximately 1,900.9 tons in 2012 to 2,451.4 tons in 2015, while PM10-primary emissions ranged from approximately 167.5 tons in 2014 to 406.4 tons in 2013. Benzo[a]pyrene emissions were also reported annually, and 2014 included a separate reported line item for “coke oven emissions” of 21.14112 tons.
From a forensic and exposure-assessment perspective, the key question was not simply how much the facility emitted in a given year, but how those emissions translated into off-site concentrations at nearby receptors. RoseyFields’ role in this type of matter is to bridge emissions inventories and community exposure by converting annual emissions into defensible model source terms, selecting appropriate averaging periods, and evaluating the spatial footprint of modeled impacts. The modeling graphics indicate that the work included thousands of receptors and produced concentration isopleths showing the distribution of annual average concentrations around the coke battery. The all-sources annual map reports a maximum annual average concentration of 2.46 µg/m³, while the coke-oven-emissions isopleth reports a maximum annual average concentration of 9.5E-03 µg/m³ for the modeled source group.
Chemicals of Concern
The emissions inventories identify a mix of criteria pollutants, particulate pollutants, and combustion-related hazardous air pollutants associated with coke battery operations. Across the 2012–2015 annual inventory reports, the principal reported criteria pollutants include carbon monoxide (CO), nitrogen oxides (NOx), sulfur dioxide (SO₂), PM10, PM2.5, volatile organic compounds (VOC), and ammonia (NH₃). These pollutants are relevant because coke-making and related combustion operations can produce both gaseous combustion products and condensable or filterable particulate matter that contribute to off-site inhalation exposure and regional air-quality burdens. The reported inventories show that sulfur dioxide and nitrogen oxides were among the largest annual emissions categories during the evaluated period, with sulfur dioxide exceeding 2,000 tons per year in 2013–2015 and NOx exceeding 1,000 tons per year in each year shown.
The hazardous and toxicologically significant pollutants reported in the inventories include benzo[a]pyrene and polycyclic organic matter (POM), both of which are especially important in coke oven and combustion-related exposure cases. Benzo[a]pyrene is a well-known carcinogenic polycyclic aromatic hydrocarbon and is often used as an indicator of PAH-related risk in air toxics assessments. The 2012–2014 inventories report annual benzo[a]pyrene emissions of approximately 1.287, 1.394, and 1.423 tons, respectively, while the 2015 inventory reports 0.105 tons, suggesting either a substantial operational change, revised emission factors, or a different reporting basis in that year. Polycyclic organic matter was also reported at substantial levels in 2012–2014, ranging from approximately 348.7 to 381.8 tons per year.
Because coke battery emissions can include both criteria pollutants and hazardous organic compounds, this type of project requires evaluating multiple pollutant classes with different regulatory and toxicological significance. Some pollutants are primarily relevant to criteria-air-quality attainment and long-range dispersion, while others are more relevant to chronic inhalation risk, community nuisance concerns, or source-apportionment questions. RoseyFields’ role in this kind of matter is to interpret those emissions in the context of exposure potential, spatial impact, and source-specific contribution rather than treating the annual inventory as a simple compliance ledger.
Methods
The first step in the project was assembling and reviewing the facility’s annual emissions inventory reports for multiple years of operation. The uploaded materials include annual emissions summaries for 2012, 2013, 2014, and 2015, each listing total reported emissions for criteria pollutants and selected hazardous or other pollutants. These reports provide the foundational data needed to identify the major emitted pollutants, assess year-to-year variability, and develop representative source terms for dispersion modeling or exposure assessment. For example, the inventories show that sulfur dioxide emissions were reported at 1,900.8885 tons in 2012, 2,180.4025 tons in 2013, 2,067.249 tons in 2014, and 2,451.4035 tons in 2015, while NOx emissions remained above 1,000 tons per year in all four years.
These inventory reports also help identify the pollutants most relevant to air-toxics or coke-oven exposure analysis. Benzo[a]pyrene was reported in each year, and polycyclic organic matter was reported in the 2012–2014 summaries. In addition, the 2014 inventory includes a specific line item for “COKE OVEN EMISSIONS (UNKNOWN)” totaling 21.14112 tons, suggesting that the project may have distinguished coke-oven-related emissions from broader facility-wide totals when developing source groups or selecting modeled pollutants. Reviewing these inventories across multiple years allows the analyst to identify whether a given year is representative, unusually high or low, or better suited to a worst-case or long-term average modeling scenario.
After reviewing the emissions data, RoseyFields would translate those annual totals into AERMOD-compatible source terms for the relevant emission units and source groups. In a coke battery case, this can include stacks, fugitives, oven doors, leaks, charging operations, pushing operations, quench emissions, and other battery-related release points, depending on the level of source detail available. The uploaded modeling figures suggest that at least two separate source-group analyses were prepared: one for all facility sources and another specifically for coke oven emissions. The all-sources map is labeled “DTE EES Coke Battery Annual Emissions – All Sources” and notes that EPA source terms were used when modeling for attainment, while the second map is labeled “DTE EES Coke Battery Air Emissions – All Sources” but includes comments specifically identifying coke oven emissions as the modeled basis.
Source-term development is a critical step because annual inventory totals alone do not determine off-site concentrations. The modeler must allocate emissions to source locations, release heights, stack parameters, or fugitive geometries; select an averaging basis; and decide whether the modeled scenario reflects annual average operations, attainment-style source assumptions, or another regulatory/forensic framework. In this matter, the figures indicate that the modeling used annual average values averaged across five years and that the results were plotted for a large receptor grid around the facility. That suggests RoseyFields’ type of work here would include converting annual emissions into a consistent multi-year modeling framework suitable for evaluating chronic exposure or long-term community impact rather than only short-term peak concentrations.
The core analytical tool in the project was AERMOD, EPA’s preferred near-field dispersion model for industrial air emissions. The uploaded figures show concentration isopleths generated for a dense receptor network around the facility. The facility-wide annual emissions plot indicates 15 sources, 13,500 receptors, and an output type of concentration, with a reported maximum annual average concentration of 2.46 µg/m³. The separate coke oven emissions isopleth shows 16 sources, 13,500 receptors, and a maximum annual average concentration of 9.5E-03 µg/m³. Both figures appear to be plotted in UTM coordinates and overlaid on a basemap showing the facility and surrounding neighborhoods, roads, and waterways.
The purpose of this modeling is to translate emissions into geographically resolved exposure potential. Instead of simply knowing that the facility emitted a certain number of tons per year, the model shows where the plume is predicted to travel, how annual average concentrations diminish with distance, and which nearby areas receive the highest chronic impacts. In community exposure and environmental justice matters, this type of spatial information is often as important as the emissions totals themselves, because it helps identify which residential neighborhoods, schools, parks, or other sensitive receptors lie within the highest-impact portions of the modeled plume.
A particularly useful feature of the uploaded materials is that they appear to distinguish between all-sources facility-wide impacts and a more focused coke-oven-emissions impact footprint. The all-sources annual concentration plot shows a broad plume with contour labels extending outward from the facility and a maximum concentration of 2.46 µg/m³, while the coke-oven-emissions isopleth shows a tighter, lower-concentration pattern centered on the facility with contours labeled in the 10⁻³ µg/m³ range and a reported maximum of 9.5E-03 µg/m³. This type of side-by-side modeling is valuable because it allows the analyst to distinguish between total facility burden and the specific contribution of a narrower source category such as coke oven emissions.
That distinction can matter in several ways. In a regulatory or litigation context, it can help answer whether a particular source category is a material contributor to community exposure, whether control measures should focus on a subset of operations, and whether toxic emissions associated with coke production are likely to drive a different risk picture than the facility’s broader criteria-pollutant inventory. RoseyFields frequently performs this kind of source-apportionment-by-modeling work to evaluate how much of the off-site impact is attributable to a specific operation rather than the site as a whole.
The final methodological component is interpreting how year-to-year emissions variability affects the defensibility of a chronic exposure analysis. The annual inventories show notable changes over time in several pollutant categories, including VOC, PM10-primary, carbon monoxide, and benzo[a]pyrene. For example, VOC emissions were reported at 119.8 tons in 2012, 200.1 tons in 2013, 61.1 tons in 2014, and 114.5 tons in 2015. PM10-primary likewise varied from 357.5 tons in 2012 to 406.4 tons in 2013, then down to 167.5 tons in 2014, before increasing again to 267.7 tons in 2015. These differences matter because the choice of modeling year—or the decision to average across multiple years—can materially change predicted long-term concentrations and therefore the characterization of community exposure.
A robust chronic exposure assessment therefore requires not only running the model, but also deciding whether the selected source terms are representative of typical operations, high-end operations, or a blended multi-year average. The uploaded figures suggest that the modeling used a five-year annual average framework, which is often appropriate for chronic exposure or attainment-style analysis because it reduces the influence of a single anomalous operating year. RoseyFields’ role in this type of case is to ensure that the modeled scenario aligns with the actual question being asked—whether that is typical community exposure, a health-protective upper-bound estimate, or a source-specific causation analysis.
Results
The first major finding is that the coke battery facility had a substantial annual emissions profile across multiple pollutant classes, including criteria pollutants, particulates, and combustion-related hazardous pollutants. Sulfur dioxide and nitrogen oxides were emitted at particularly high levels, each exceeding approximately 1,000 tons per year across the inventory years reviewed, while PM10-primary and VOC emissions were also substantial. The inventories further show annual reporting of benzo[a]pyrene and, in several years, polycyclic organic matter—pollutants that are especially relevant in coke oven and steelmaking-related health investigations because of their connection to combustion byproducts and chronic inhalation risk.
The second major finding is that air-dispersion modeling translated those annual emissions into a clear off-site impact footprint affecting the surrounding community. The uploaded AERMOD figures show annual average concentration contours extending beyond the immediate facility boundary and into nearby neighborhoods. The all-sources map reports a maximum annual average concentration of 2.46 µg/m³, while the coke-oven-specific isopleth reports a lower but still mappable maximum of 9.5E-03 µg/m³. The presence of thousands of modeled receptors and broad contour fields indicates that the analysis was designed to evaluate community-scale exposure potential rather than only on-site concentrations or fence-line impacts.
A third important finding is that source-specific modeling can materially improve interpretation of facility impacts. By distinguishing between all facility sources and a coke-oven-emissions source group, the analysis can separate the facility’s overall chronic burden from the contribution of a narrower set of coke-making operations. That is valuable in cases involving apportionment, source attribution, control effectiveness, or community risk because it helps identify whether a specific operation is likely to be driving the modeled plume or whether the observed impacts reflect the combined effect of multiple industrial sources across the site.
Finally, the inventory record shows that emissions were not static from year to year, which has direct implications for chronic exposure modeling and any conclusions drawn from a single year of data. Pollutants such as sulfur dioxide, VOCs, PM10-primary, and benzo[a]pyrene changed substantially across the 2012–2015 period. A technically sound long-term exposure assessment therefore needs to consider whether a single year is representative, whether multi-year averaging is more appropriate, and whether different pollutant classes behave differently across the facility’s operating history. The uploaded materials suggest that RoseyFields approached that issue by relying on a multi-year annual averaging framework rather than treating a single inventory year as determinative.
Conclusions
This project illustrates RoseyFields’ ability to conduct air-dispersion modeling and emissions-based exposure assessment for complex industrial facilities, including coke battery and steelmaking-related operations with multiple emission units and pollutant classes. The work required integrating annual emissions inventory data, pollutant-specific source characterization, and AERMOD modeling to evaluate how routine operations at a coke battery facility could contribute to off-site community exposure. By moving from inventory totals to receptor-based concentration fields, RoseyFields translated a large industrial emissions record into a practical assessment of where and to what degree surrounding neighborhoods may be affected by facility emissions.
The case also demonstrates the importance of pairing emissions analysis with source-specific modeling. Coke battery facilities can emit a combination of criteria pollutants, fine and coarse particulate matter, sulfur compounds, and hazardous combustion byproducts such as benzo[a]pyrene and other polycyclic organic compounds. Evaluating their impact requires more than reading a permit inventory; it requires understanding how those emissions are distributed across specific source groups, how they vary over time, and how they disperse into the surrounding community under long-term operating conditions. RoseyFields’ work in this type of matter is designed to answer those questions in a technically defensible way that can support litigation, regulatory review, or community exposure evaluation.
More broadly, this project reflects RoseyFields’ experience in industrial emissions analysis, air-dispersion modeling, and chronic inhalation exposure assessment for heavy industrial facilities. Whether the issue involves coke ovens, refineries, chemical plants, landfills, or combustion sources, RoseyFields can integrate emissions inventories, source-term development, atmospheric modeling, and exposure interpretation into a coherent technical framework for understanding off-site air impacts.
Dr. Paul Rosenfeld led the technical evaluation of emissions and air-dispersion impacts for this matter, applying industrial emissions analysis, source-term development, and atmospheric modeling principles to assess how coke battery and related facility emissions could affect nearby off-site receptors. His work focused on integrating multi-year emissions inventories, pollutant-specific source characterization, AERMOD concentration modeling, and source-group comparisons into a defensible assessment of chronic community exposure potential. The project reflects Dr. Rosenfeld’s broader expertise in air-dispersion modeling, industrial source apportionment, and forensic evaluation of inhalation exposures associated with large industrial facilities.
References
EES Coke Battery, LLC. (2012). 2012 Emissions inventory report: Emissions summary.
EES Coke Battery, LLC. (2013). 2013 Emissions inventory report: Emissions summary.
EES Coke Battery, LLC. (2014). 2014 Emissions inventory report: Emissions summary.
EES Coke Battery, LLC. (2015). 2015 Emissions inventory report: Emissions summary.
ZM Associates Environmental Corp. (2025, January 21). DTE EES Coke Battery annual emissions – all sources: Using US EPA source terms when modeling for attainment.
ZM Associates Environmental Corp. (2025, April 28). DTE EES Coke Battery coke oven emissions isopleth / annual values averaged across five years.