Gains:
- Ability to use AQI (AQI), PM2.5/PM10, NOx and emission factor concepts correctly
- Ability to accelerate emission calculation, distribution interpretation and report drafting with AI
- Ability to verify the emission factors and limit values given by AI with official sources
You work in the peripheral unit of a natural gas cycle power plant. You need to prepare an annual emission inventory and you are asked to calculate and report the NOx load coming out of the chimney. You are also expected to interpret PM2.5 and NO2 data from the air quality station in the district where the facility is located. You tell the AI "what is the NOx emission factor of this plant and calculate the annual load"; The model immediately gives an emission factor number and an agency guide name as the source. Here is the critical moment: is this emission factor true or made up? This unit teaches how to use concepts such as AQI, PM, NOx correctly, speed up emission calculation and distribution interpretation with AI, but most importantly, verify each emission factor and limit value given by AI with the official source.
Basic concepts: AQI, PM, NOx, SO2
The language of air quality management is based on several basic indicators:
concept
expansion
unit
Importance
AKI (AQI)
Air Quality Index
Index (0–500)
Expression of pollution in public health language
PM2.5
≤2.5 µm particulate matter
µg/m³
Reaches deep lung, most risky
PM10
≤10 µm particulate matter
µg/m³
Respirable coarse particle
NOx
Nitrogen oxides (NO + NO₂)
µg/m³ or kg
Combustion-derived, ozone precursor
SO2
sulfur dioxide
µg/m³
Sulfur fuel combustion, acid rain
AQI is a conversion that converts raw concentration (µg/m³) into health categories (good, fair, unhealthy for sensitive groups, unhealthy...). Important point: AQI thresholds and spreadsheets differ by country and authority. If you ask AI "AQI threshold for PM2.5", do not use it without knowing which system (national index, EPA, EU) gives the threshold.
Tip: Ask the AI to clearly state which index system and which averaging period (1-hour, 24-hour, or annual) it uses when interpreting air quality data. "50 µg/m³" has a different meaning for a 24-hour value and a completely different meaning for an annual average.
Emission factor approach
The basic equation of the emissions inventory is simple:
Emission = Emission Factor
The emission factor is the most sensitive and hallucinogenic component of this calculation. Factor; varies depending on fuel type, combustion technology, control equipment (e.g. low-NOx burner, SCR) and plant age. AI may give you a number that seems reasonable, but that number may not match your facility's actual configuration.
Example calculation: annual NOx load
Let's do a calculation with realistic units and see each step:
# Inputs fuel_flow = 45000.0 # Nm3/hour natural gasoperating_hours = 7500.0 # hours/year# Emission factor: THIS VALUE MUST BE VERIFIED FROM THE LEGISLATION/GUIDELINES!EF_NOx = 0.15 # g NOx / Nm3 (EXAMPLE - source confirmation required)# Annual fuel consumptionannual_fuel = fuel_flow * operating_hours # Nm3/yearprint("Annual fuel:", annual_fuel, "Nm3/year") # 337,500,000# Annual NOx loadNOx_g = annual_fuel * EF_NOx # grams/annualNOx_ton = NOx_g / 1_000_000.0 # ton/yearprint("NOx load:", round(NOx_ton, 1), "ton/year") # 50.6 tons/year
The result is approximately 50.6 tonnes/year NOx. But this whole number is based on the assumption of EF_NOx = 0.15. If the actual factor is 0.20, your load will be 33% higher. Therefore, the emission factor must be verified from the official source before entering into the decision.
Caution: Emission factor hallucination is one of the most expensive mistakes in environmental engineering. An incorrect factor affects the entire annual inventory, emissions permit and possible penalties. Be sure to compare the factor given by the AI with the "EMEP/EEA guideline", the national emissions guideline or the facility's continuous measurement (CEMS) data.
Weak prompt / Strong prompt
WEAK PROMPT: "What is the NOx emission factor of a natural gas boiler? Calculate the annual load."Problem: Forces the model to fit a single factor; combustion technology, control equipment, source not specified. If the resulting number goes into the decision, the entire inventory will be wrong. STRONG PROMPT: "Explain which variables (combustion technology, low-NOx burner, SCR/SNCR, load) depend on the NOx emission factor for a natural gas-fired industrial boiler. DO NOT make up an exact number; instead, from which official sources the factor is obtained (e.g. EMEP/EEA Guidebook, national emission guide) and explain step by step which category to choose. Set up the account structure (Emission = EF x activity), but write 'to be verified_from_source' instead of EF. "Difference: The powerful prompt sets up the Skeleton of the account, but leaves the critical number to the legislation.
Distribution interpretation
The amount of emissions alone is not enough; How the pollutant disperses in the atmosphere, what concentration it reaches at ground level — this is the subject of dispersion modelling. Chimney height, outlet velocity, temperature, wind speed and atmospheric stability determine the distribution.
AI can help explain distribution concepts and establish the structure of a modeling study: what inputs are required, which meteorological scenarios are critical, how to interpret the result. But:
- AI does not run a true distribution model (like AERMOD, CALPUFF).
- Those tools calculate the ground level concentration estimate with your input.
- "Worst case" and limit value comparison must be based on legislation and actual meteorology.
Quest
The role of AI
verification
Explain the concept of distribution
tutorial
Textbook/guide
List modeling inputs
checklist
Expert + field
Ground level concentration calculation
doesn't
AERMOD/CALPUFF
Comparison with limit value
draft
Official legislation
mini case
An engineer asks the AI for the SO2 emission factor of a thermal heat treatment furnace for a quick quote. The model gives a confident factor and a guide name that looks real as the source. The engineer records this in the inventory. During the audit, it is noticed that the given factor does not belong to the facility using low sulfur fuel; The true factor is much lower and the plant has declared excess emissions. A mistake in the opposite direction was also possible: underreporting and risking penalties. Lesson: emission factor never enters inventory without verification by fuel analysis and official guidance. The AI factor was an initial hypothesis, not evidence.
Common mistakes
- Taking the emission factor from the AI and not confirming its source (EMEP/EEA, national guidance).
- Using AQI thresholds without knowing which index system they belong to.
- Mixing average duration (1-hour / 24-hour / yearly).
- Mistaking the emission amount for distribution and mistaking it for ground level concentration.
- Treating AI as a real distribution model and expecting an AERMOD/CALPUFF result.
- Using a single "average" factor, ignoring combustion technology and control equipment (SCR, low-NOx burner).
In summary
- Air quality management requires the use of indicators such as AQI, PM2.5/PM10, NOx and SO2 with the correct unit and averaging time.
- The basis of the emission calculation is the equation "Emission = factor × activity"; The most sensitive component is the emission factor.
- AI speeds up the calculation structure and concept explanation, but cannot reliably provide the emission factor and limit value.
- The emission factor is not included in the inventory until it is verified with official sources (EMEP/EEA, national guideline) or with the continuous measurement data of the facility.
- Explains the AI concept for distribution interpretation; Tools such as AERMOD/CALPUFF calculate the actual ground level concentration.
- AI is like a senior intern: they prepare the draft, but the numerical and legal accuracy of the emissions inventory is the engineer's responsibility.
Application task
Establish an annual NOx or SO2 load calculation from your own plant or from a hypothetical combustion source (boiler, furnace, generator): determine fuel type, consumption/hours of operation and assume an emission factor, calculate the annual load in tonnes/year with the equation "Emission = factor × activity". Then explain in a paragraph from which official source (EMEP/EEA Guidebook, national emission guide or SEÖS data) and according to which category you will verify this emission factor. Finally, using the “strong prompt” pattern above, ask the AI to skeletonize the account, but deliberately leave the emission factor as “verified at source”; Create a checklist by marking each numerical claim and each source citation in the output as to be verified/verified.