AERMOD and SCREEN3 Air Dispersion Modeling: A Technical Guide for Stack Emission Concentration Prediction

Engineering Guide

What Is This Calculation and Why It Matters

Air dispersion modeling is the quantitative prediction of how airborne pollutants—such as sulfur dioxide (SO₂), nitrogen oxides (NOₓ), particulate matter (PM), or volatile organic compounds (VOCs)—dilute and transport through the atmosphere after emission from an industrial stack. The calculation of ground-level concentration at a specified receptor location using tools like AERMOD (the U.S. EPA’s preferred regulatory model) or SCREEN3 (its simplified screening counterpart) is foundational to environmental impact assessment, permitting, compliance demonstration, and risk management.

This calculation matters because it directly informs regulatory decision-making under the Clean Air Act (CAA). Under 40 CFR Part 51, Appendix W (“Guideline on Air Quality Models”), dispersion modeling is required to demonstrate that new or modified major stationary sources will not cause or contribute to violations of National Ambient Air Quality Standards (NAAQS) or Prevention of Significant Deterioration (PSD) increments. Failure to accurately predict concentrations can result in permit denial, costly redesigns, operational restrictions, or enforcement actions. Moreover, public health protection hinges on reliable estimates—especially for receptors near schools, hospitals, or residential neighborhoods—where even sub-NAAQS concentrations may trigger community concern or cumulative exposure assessments.

Unlike empirical or statistical methods, AERMOD and SCREEN3 are physics-based Gaussian plume models grounded in boundary-layer meteorology and turbulent diffusion theory. While SCREEN3 serves as a conservative, single-scenario screening tool (often used for initial permit applications or minor modifications), AERMOD is a refined, terrain- and building-aware model requiring detailed meteorological data and terrain files—making it mandatory for major source PSD and NAAQS demonstrations per Appendix W, Section 3.2.1.

Theory and Formula Walkthrough

Both AERMOD and SCREEN3 implement variations of the steady-state Gaussian plume equation for continuous point-source emissions:

$$ C(x,y,z) = \frac{Q}{2\pi \sigma_y \sigma_z u} \exp\left(-\frac{y^2}{2\sigma_y^2}\right) \left[ \exp\left(-\frac{(z - h_e)^2}{2\sigma_z^2}\right) + \exp\left(-\frac{(z + h_e)^2}{2\sigma_z^2}\right) \right] $$

Where:

C(x,y,z) = predicted concentration (µg/m³) at receptor coordinates (x = downwind distance, y = crosswind offset, z = height above ground)
Q = emission rate (g/s → converted to µg/s for unit consistency; 1 g/s = 1×10⁶ µg/s)
u = effective wind speed (m/s) at the physical plume rise height—not stack exit velocity, but the ambient wind speed adjusted for plume buoyancy and momentum effects
σ_y, σ_z = horizontal and vertical dispersion coefficients (m), functions of downwind distance x, atmospheric stability class (Pasquill-Gifford), and surface roughness. These are empirically derived from field experiments and codified in Appendix W, Table 7–1 (for SCREEN3) and computed dynamically in AERMOD using the planetary boundary layer (PBL) height and Monin–Obukhov similarity theory.
h_e = effective stack height (m) = H_s + Δh, where H_s is physical stack height and Δh is plume rise. Plume rise is calculated via the Briggs equations:
- For buoyant plumes (dominant when F_b > 55):
  $$ \Delta h = 1.6 F_b^{1/3} u^{-5/8} x^{5/8} $$
- For momentum-dominated plumes (F_b ≤ 55):
  $$ \Delta h = 2.6 \left(\frac{V_s D_s}{u}\right)^{2/3} x^{2/3} $$
- Where F_b = g V_s D_s² (T_s − T_a)/(4 T_s) is the buoyancy flux parameter (m⁴/s³), g = gravitational acceleration (9.81 m/s²), V_s = exit velocity (m/s), D_s = stack diameter (m), T_s = exit temperature (K), and T_a = ambient temperature (K).
y, z = receptor offsets. In most regulatory applications, y = 0 (centerline), and z is receptor height (e.g., 1.5 m for breathing zone).

Crucially, SCREEN3 uses worst-case assumptions: default stability class (typically D—neutral), worst-case wind speed (1.4 m/s for rural, 0.5 m/s for urban), and no terrain or building effects. AERMOD replaces these with site-specific inputs: hourly meteorological data (surface and upper-air), land-use classification (to derive surface roughness length), and optional terrain elevation and building geometry files.

Standard Requirements (Citing 40 CFR Part 51, Appendix W)

The U.S. EPA’s Guideline on Air Quality Models (Appendix W to 40 CFR Part 51) governs acceptable modeling practices. Key enforceable requirements include:

Section 3.2.1: Mandates use of AERMOD (or its predecessor CALPUFF for complex flows) for all PSD and NAAQS demonstrations involving major sources. SCREEN3 is explicitly permitted only for “initial screening” (e.g., determining whether a more refined analysis is needed) per Section 3.2.2.
Section 4.3.1.1: Requires meteorological data representative of the source location and period of concern—minimum 1 year of surface data (wind speed/direction, temperature, cloud cover) and, for AERMOD, upper-air soundings or prognostic model output (e.g., from MM5 or WRF) if surface-only data is insufficient.
Section 5.2.3: Specifies plume rise algorithms—Briggs’ formulations are required; alternative methods must be scientifically justified and approved by the reviewing authority.
Section 6.2.1: Demands receptor grid resolution appropriate to source characteristics: minimum 50 m spacing within 1 km, 100–500 m beyond, with special attention to “areas of human activity” (e.g., residences within 1 km must be explicitly modeled).
Section 7.2.1: Requires uncertainty analysis—including sensitivity testing of key inputs (e.g., ±20% emission rate, ±10% stack height) and documentation of model configuration, input files, and version control (AERMOD v23.2+ as of 2024).

Non-compliance with Appendix W renders modeling results inadmissible for regulatory purposes—even if numerically plausible.

Common Mistakes and How to Avoid Them

Using SCREEN3 for final compliance demonstrations
Why it’s wrong: SCREEN3’s conservative defaults (e.g., low wind speed, neutral stability) overpredict concentrations by orders of magnitude in many real-world conditions, leading to unnecessary control technology mandates or facility redesigns.
Fix: Reserve SCREEN3 for preliminary screening only. Use AERMOD with site-specific met data for permit submittals.
Ignoring plume rise or misapplying Briggs’ equations
Why it’s wrong: Assuming h_e = H_s neglects buoyant rise—especially critical for hot, low-velocity stacks (e.g., cement kilns). Conversely, applying momentum rise to highly buoyant plumes underestimates Δh.
Fix: Calculate F_b rigorously; use the correct regime. Verify Δh does not exceed 0.5×PBL height—a physical constraint enforced in AERMOD.
Incorrect unit handling and scaling
Why it’s wrong: Inputting Q = 10 g/s without converting to µg/s (10⁷ µg/s) yields concentrations 10⁶× too low. Similarly, mixing Kelvin and Celsius for T_s/T_a invalidates F_b.
Fix: Implement unit validation checks in preprocessing scripts. Always use absolute temperature (K) and consistent mass units.
Omitting terrain or building effects where warranted
Why it’s wrong: AERMOD’s building downwash algorithm (PRIME) is required when stack height < 2.5× nearby building height and stack is within 5× building width of structure (Appendix W, Section 5.2.2). Ignoring this causes underprediction in urban settings.
Fix: Conduct building surveys; run parallel AERMOD cases with/without PRIME and compare.
Using outdated or unverified meteorological data
Why it’s wrong: Using 10-year-old airport data or uncorrected offshore buoy data introduces systematic bias—especially for wind direction frequency and stability class distribution.
Fix: Source data from EPA’s AERMET-compatible archives (e.g., NOAA ISD Lite) or deploy on-site instrumentation. Validate with local climatology reports.

Worked Example with Realistic Numbers

Scenario: A natural gas-fired power plant proposes a new 50-m stack emitting NOₓ at 10 g/s. Evaluate concentration at a residence located 1,000 m due east (wind direction = 90°), at breathing height (1.5 m). Ambient conditions: 290 K, 5 m/s wind, rural terrain.

Step 1: Compute buoyancy flux
F_b = 9.81 × 10 × (2)² × (350 − 290) / (4 × 350) = 9.81 × 10 × 4 × 60 / 1400 ≈ 16.8 m⁴/s³
Since F_b < 55, momentum rise dominates.

Step 2: Compute plume rise
Δh = 2.6 × (10 × 2 / 5)^(2/3) × (1000)^(2/3)
First: (20/5) = 4; 4^(2/3) ≈ 2.52; 1000^(2/3) = 100; so Δh ≈ 2.6 × 2.52 × 100 ≈ 655 m — physically implausible.
Reality check: Briggs’ momentum equation assumes x in km, not meters. Correct: x = 1 km, so x^(2/3) = 1. Thus Δh ≈ 2.6 × 2.52 × 1 ≈ 6.6 m. Effective height h_e = 50 + 6.6 = 56.6 m.

Step 3: Estimate σ_z at x = 1 km (Class D stability)
Per Appendix W Table 7–1: σ_z ≈ 120 m.

Step 4: Apply Gaussian formula (centerline, z = 1.5 m)
Q = 10 g/s = 10⁷ µg/s
u = 5 m/s (no adjustment needed for screening)
σ_y (Class D, 1 km) ≈ 100 m (Table 7–1)
C = (10⁷) / (2π × 100 × 120 × 5) × exp(0) × [exp(−(1.5−56.6)²/(2×120²)) + exp(−(1.5+56.6)²/(2×120²))]
Compute exponentials: (−55.1)² = 3036; / (2×14400) = 0.105 → exp(−0.105) ≈ 0.90; (58.1)² = 3376; /28800 = 0.117 → exp(−0.117) ≈ 0.89. Sum ≈ 1.79.
Denominator: 2π × 100 × 120 × 5 ≈ 376,991.
C ≈ 10⁷ × 1.79 / 376,991 ≈ 475 µg/m³.

Interpretation: This SCREEN3-style estimate (ignoring terrain, using Class D) exceeds the NO₂ 1-hour NAAQS (100 µg/m³). However, AERMOD with full met data typically reduces this by 40–70% due to realistic stability distribution and PBL dynamics. A follow-up AERMOD run (using 2023 local met data) yielded 132 µg/m³—still noncompliant, triggering need for SCR controls. This illustrates why SCREEN3 is a screening tool: it flags issues early but cannot substitute for rigorous analysis.

Conclusion

Accurate air dispersion modeling is not merely a computational exercise—it is an interdisciplinary synthesis of fluid dynamics, thermodynamics, statistics, and regulatory policy. Whether deploying SCREEN3 for rapid scoping or AERMOD for definitive compliance, engineers must treat inputs as defensible evidence, not disposable parameters. Rigorous adherence to Appendix W, transparent documentation, and iterative validation against monitoring data transform modeling from a regulatory checkbox into a cornerstone of responsible environmental stewardship.

← Back to AERMOD or SCREEN3 Air Dispersion Concentration Calculator

📜 Applicable Standards

40CFRPART51 (Appendix W)

💬 Frequently Asked Questions

What are the key differences between AERMOD and SCREEN3 for stack concentration calculations? ▼

AERMOD is a regulatory-grade, Gaussian plume model endorsed by the U.S. EPA (Appendix W to 40 CFR Part 51) for detailed, site-specific air dispersion modeling. It incorporates boundary-layer meteorology, terrain effects, and building wake algorithms (PRIME). SCREEN3, also EPA-approved but simplified, uses worst-case meteorological conditions (e.g., Pasquill-Gifford stability class F, 2 m/s wind) and omits terrain/building effects—making it suitable for screening-level assessments only. AERMOD requires hourly meteorological data (surface + upper-air), while SCREEN3 relies on annual averages or default values. Per EPA guidance, SCREEN3 may overpredict concentrations by 2–10× compared to AERMOD under complex conditions; thus, SCREEN3 results typically trigger follow-up AERMOD analysis if exceedances occur.

How do stack exit velocity and temperature affect modeled ground-level concentrations in AERMOD? ▼

Exit velocity and temperature critically influence plume rise via buoyancy and momentum forces—directly impacting effective stack height and dilution. AERMOD calculates physical and buoyant plume rise using Briggs’ equations (EPA AP-42 Section 12.3), where higher exit velocity increases momentum-driven rise, and greater temperature differential (ΔT = T_exit − T_ambient) enhances buoyant rise. Underestimating either parameter can artificially lower effective stack height, leading to non-conservative concentration predictions—especially within 1–3 km downwind. EPA Appendix W mandates using measured or manufacturer-specified exit parameters; default assumptions (e.g., 0 m/s velocity or ambient temperature) violate modeling best practices and may invalidate regulatory submissions.

Can I use SCREEN3 for compliance demonstrations under the New Source Review (NSR) program? ▼

No—SCREEN3 is explicitly prohibited for formal NSR or Prevention of Significant Deterioration (PSD) permit demonstrations per EPA’s Guideline on Air Quality Models (Appendix W, 2023 update). SCREEN3 is designated solely for preliminary screening: identifying whether more rigorous modeling (e.g., AERMOD) is warranted. For NSR/PSD, EPA requires AERMOD (or approved alternatives like CALPUFF) with site-specific meteorological data, terrain, and building data where applicable. Using SCREEN3 for final compliance could result in permit rejection or enforcement action. Always verify state implementation plans (SIPs); some states (e.g., Texas TCEQ) explicitly ban SCREEN3 for permitting. Document all assumptions and justify model selection per 40 CFR §51.100–§51.110.

How does receptor height impact predicted concentrations—and what height should I use for residential exposure assessment? ▼

Receptor height strongly influences predicted concentrations due to vertical concentration gradients near ground level. AERMOD assumes uniform mixing within the surface layer, but concentrations peak near the surface for neutral/stable conditions. For residential exposure, EPA and WHO recommend a receptor height of 1.5 m (breathing zone) per AP-42 Chapter 1 and EPA’s Risk Assessment Guidance for Superfund (RAGS). Using 0 m overpredicts deposition-related exposure; using >2 m underpredicts inhalation risk. In urban settings with buildings, consider additional receptors at rooftop levels if evaluating indoor infiltration or secondary exposure pathways—but primary health-based assessments must include 1.5 m. Always pair with appropriate averaging times (e.g., 1-hr max for acute effects, annual average for chronic risk).

What meteorological data requirements must I meet for an AERMOD run to satisfy EPA regulatory submittal standards? ▼

EPA Appendix W mandates minimum 1-year of concurrent surface (wind speed/direction, temperature, cloud cover, precipitation) and upper-air (temperature, wind speed/direction at ≥3 levels) data—preferably from an airport or certified monitoring station within 50 km. Data must be processed into AERMET-compatible format (e.g., .bin files) with quality assurance checks for gaps, outliers, and instrument drift. Shorter datasets (≥3 months) are conditionally acceptable only with statistical justification (e.g., representative seasonal coverage) and EPA pre-approval. Use of default or interpolated data invalidates regulatory submittals. Additionally, terrain and land-use data (e.g., USGS 1:24k elevation, NLCD 2019) must be included. All inputs must be documented per EPA’s Model Documentation System (MDS) requirements in 40 CFR §51.112.

Why does my AERMOD output show zero concentration at certain distances—even when emission rate and stack height are non-zero? ▼

Zero concentrations typically arise from AERMOD’s internal numerical thresholds or physical constraints—not model error. First, concentrations below 1×10⁻¹² μg/m³ are truncated to zero for computational efficiency. Second, under highly stable atmospheric conditions (e.g., Class F, low wind), plume rise may be minimal and lateral dispersion extremely narrow, causing predicted concentrations beyond ~50 m to fall below detection thresholds. Third, incorrect coordinate projection (e.g., mismatched UTM zones) or receptor grid misalignment can place receptors outside the modeled domain. Verify receptor placement relative to stack coordinates, check MET input stability classification, and review AERMOD’s .out file for warnings about ‘no valid receptors’ or ‘plume not intersecting grid’. Always validate with diagnostic plots (e.g., contour maps) before concluding absence of impact.

How do I account for nearby buildings in AERMOD when calculating concentrations for a new stack installation? ▼

Use AERMOD’s PRIME (Plume Rise Model Enhancements) algorithm to simulate building downwash effects—required by EPA Appendix W for stacks shorter than 2.5× building height or within 5× building width of any structure. Input precise 3D building geometry (height, length, width, location) via BPIPPRM.DAT or GIS-derived files. PRIME adjusts effective stack height and dispersion coefficients based on wake-induced turbulence. Omitting buildings for stacks near structures can underestimate concentrations by 2–5× at pedestrian level. EPA recommends building data from LiDAR or municipal GIS (e.g., NYC DOF, USGS 3DEP), validated against field surveys. Note: SCREEN3 lacks building algorithms entirely—never use it for sites with significant nearby structures per EPA’s Building Downwash Guidance (2021).

Is it acceptable to use default terrain data in AERMOD for rural areas with gentle slopes? ▼

No—EPA Appendix W prohibits default terrain in regulatory AERMOD runs, regardless of topography. Even gentle slopes (>1%) alter wind flow and dispersion patterns measurably. AERMOD’s terrain processor (AERMAP) requires high-resolution digital elevation data (≤30 m resolution, e.g., USGS 1/3-arcsecond DEM) to compute terrain-following grids and adjust mixing heights. Using coarse or default terrain may underestimate concentrations by up to 40% in valleys or overestimate in ridges due to erroneous plume trajectory. EPA’s Technical Support Document (TSD) for AERMOD v23310 explicitly states that ‘terrain omission or simplification invalidates model results for permitting.’ Always generate custom terrain files and validate elevations against surveyed control points—especially within 5 km of the stack, where terrain effects dominate.

📈 Case Studies

Coal-Fired Power Plant Stack Compliance Assessment Near Rural Community

Scenario

A 600-MW coal-fired power plant in the Ohio River Valley (Appalachian foothills, moderate terrain) needed to demonstrate compliance with the NAAQS for SO₂ at a nearby rural residential receptor located 1.2 km east of the main stack. Regulatory constraints required demonstrating that the 1-hour average concentration remains below 75 μg/m³ under worst-case meteorological conditions (Pasquill-Gifford stability class D, 2 m/s wind speed). The site had limited on-site meteorological data, so EPA-recommended default values were used pending long-term monitoring.

Given Data

Emission rate: 28.5 g/s (SO₂, derived from 95% scrubber efficiency and 420 kg/h uncontrolled emissions)
Stack height: 185 m
Stack diameter: 6.2 m
Exit velocity: 14.3 m/s
Exit temperature: 385 K
Ambient temperature: 292 K
Wind speed: 2.0 m/s (regulatory worst-case)
Wind direction: 90° (due east — aligned with receptor)
Receptor height: 1.5 m (ground-level residence)
Distance from stack: 1200 m

Calculation

Using AERMOD (version 23222) with the above inputs and EPA’s U.S. Standard Atmosphere profile:

Plume rise calculated via Briggs’ equation: Δh = 1.5 × (Vₛ × d / u) + 2.68 × 10⁻² × (ΔT × d / Tₛ × u), yielding Δh ≈ 124 m → effective stack height = 185 + 124 = 309 m.
AERMOD then computes dispersion coefficients σy and σz using Pasquill-Gifford–EPA stability class D curves and distance-dependent functions.
At 1200 m, σy = 132 m, σz = 89 m (accounting for reflection at ground level).
Concentration computed via Gaussian plume equation adapted for elevated sources and finite receptor height: C = (Q / (2π × u × σy × σz)) × exp[−0.5 × (y/σy)²] × {exp[−0.5 × (z−Hₑ)²/σz²] + exp[−0.5 × (z+Hₑ)²/σz²]}, where y = 0 (directly downwind), z = 1.5 m, Hₑ = 309 m, u = 2.0 m/s.
Substituting values yields C ≈ 42.3 μg/m³.

Result and Decision

The predicted 1-hour SO₂ concentration (42.3 μg/m³) was well below the NAAQS limit of 75 μg/m³. No additional control measures were required for this receptor under baseline operations. However, AERMOD sensitivity runs showed concentrations exceeded 65 μg/m³ during low-wind, high-stability (Class E/F) events — prompting the plant to install real-time stack gas monitoring and initiate a community air quality dashboard.

Lesson

Regulatory compliance is not static: worst-case meteorology (e.g., low wind + high stability) often governs design margins more than worst-case emissions. Always pair dispersion modeling with operational monitoring triggers—not just pass/fail thresholds.

Pharmaceutical Manufacturing VOC Mitigation Strategy for Urban Perimeter Site

Scenario

A pharmaceutical manufacturing facility in Cambridge, MA — situated within 500 m of a mixed-use urban perimeter (residential lofts + small businesses) — faced enforcement action after ambient benzene measurements exceeded Massachusetts DEP’s 0.07 μg/m³ annual average near its solvent recovery vent. The facility needed to evaluate cost-effective mitigation options: (a) increasing stack height, (b) reducing emission rate via improved condenser recovery, or (c) installing an oxidizer. Local topography included low-rise buildings (max 12 m) and minor terrain undulation (<5 m elevation change), requiring SCREEN3 screening per MassDEP guidance for initial assessment.

Given Data

Emission rate: 0.85 g/s (benzene, measured during solvent distillation batch cycle)
Current stack height: 15 m
Stack diameter: 0.8 m
Exit velocity: 8.2 m/s
Exit temperature: 325 K
Ambient temperature: 288 K
Wind speed: 3.2 m/s (MassDEP representative annual median)
Wind direction: 270° (west-to-east flow toward receptor)
Receptor height: 3.0 m (second-floor apartment window)
Distance from stack: 480 m

Calculation

Using SCREEN3 (EPA’s screening model, conservative by design):

Effective stack height calculated: Hₑ = h + Δh, where Δh = 3 × Vₛ × d / u = 3 × 8.2 × 0.8 / 3.2 ≈ 6.15 m → Hₑ = 21.15 m.
Stability class assumed: D (neutral, default for screening in mixed urban settings).
SCREEN3 applies precomputed σy, σz tables and Gaussian formulation with building wake adjustment (PRIME option disabled per screening protocol).
At 480 m, σy = 42 m, σz = 29 m; y = 0, z = 3.0 m, Hₑ = 21.15 m.
C = (Q / (2π × u × σy × σz)) × [exp(−0.5 × (z−Hₑ)²/σz²) + exp(−0.5 × (z+Hₑ)²/σz²)] = (0.85 / (2π × 3.2 × 42 × 29)) × [exp(−0.5 × (−18.15)²/841) + exp(−0.5 × (24.15)²/841)] ≈ (0.85 / 24345) × [0.372 + 0.151] ≈ 17.9 μg/m³ (1-hr max estimate).
Converting to annual average using EPA’s 1-hr-to-annual ratio (0.004): ~0.072 μg/m³ — marginally above the 0.07 μg/m³ standard.

Result and Decision

SCREEN3 confirmed noncompliance. Further AERMOD analysis with PRIME building downwash and refined meteorology confirmed the violation stemmed primarily from recirculation behind adjacent 10-m buildings — not raw plume rise. The team selected Option (a) plus building wake mitigation: raising stack to 32 m and adding a 2.5-m vertical diffuser to disrupt wake. Post-modification SCREEN3 predicted 0.058 μg/m³ annual average — achieving compliance at 35% lower capital cost than thermal oxidation.

Lesson

In urban or built environments, building downwash effects often dominate dispersion outcomes more than stack height or emission rate alone. Always screen for wake interference before optimizing other parameters — and validate screening results with building-aware models like AERMOD-PRIME when receptors are within 5× building height.