AIRBORNE INFECTION RISK CALCULATOR
Assess the airborne infection risk in your indoor environment.
How would you like to specify the ventilation rate?
Assess the airborne infection risk in your indoor environment.
How would you like to specify the ventilation rate?
Derive the air-exchange rate from CO₂, then apply it as the AER above. CO₂ tracks outdoor-air ventilation only — not in-room filtration.
Ambient CO₂ and per-person generation rate below are used to derive indoor CO₂ from AER — whichever way AER is set. Per-person CO₂ generation follows the susceptible activity chosen in “Exposure & occupants”.
BMR via Schofield (1985), adults 30–60; M (met) per ASHRAE 55; V̇CO₂ = BMR·M·0.000484 at RQ 0.85 (Persily & de Jonge 2017).
Steady-state estimate from room volume, AER above, occupants (I+S), and the ambient CO₂ / generation assumptions set in the “Ventilation from CO₂” panel below. Feeds the viral-inactivation term λ only.
| Exposure | Infection risk | Max occupancy for Revent < 1 |
|---|
Freeze every input at its current value above, vary one parameter across a range, and see how infection risk and Revent respond. Example: keep everything as-is and sweep AER from 5 to 10 h⁻¹.
A well-mixed (single-zone) box model. With a steady quanta emission ERq from I infectious occupants in volume V, the airborne quanta concentration relaxes toward steady state at the infectious-virus removal rate IVRR:
The time-averaged concentration over exposure t, the inhaled dose, and the risk to one susceptible (exponential dose–response) are:
ERq is not a single number: log₁₀(ERq) is normally distributed (mean μ, st.dev σ) per pathogen × expiratory activity. The reported risk integrates the per-ERq risk over that distribution by quadrature on a 199-point percentile grid (0.005→0.995) — identical to the original workbook:
The μ, σ values for I = 1…10 infectious occupants are the pre-fitted summed-lognormal parameters tabulated in the source tool (not re-fit here). The Fixed ERq mode evaluates a single deterministic emission rate; Custom mode takes your own μ, σ. Max occupancy for Revent < 1 is ⌊1/R⌋, with R capped at 0.99.
Secondary transmission. The event reproduction number is R_event = S · R for S susceptibles. But the mean hides the overdispersion. Rather than treating C = R·S as a continuous value, each susceptible is an independent Bernoulli trial at the per-emission probability P(ERq), so conditional on an emission the case count is Binomial(S, P(ERq)). Integrating that over the ERq distribution (same 199-point grid) gives the full discrete distribution:
This Bernoulli/binomial treatment (per Arpino et al. 2022, Indoor Air) replaces the earlier continuous R·S binning and yields the correct probability of zero cases and of a superspreading tail — the hallmark of airborne overdispersion.
Caveats: assumes instantaneous, perfect mixing and steady emission; no near-field/proximity effect, no masking or filtration terms in this build, no transient occupant arrival/departure (that is the TEC model). Estimates are relative-risk guidance, not exact predictions.
CO₂ ventilation front-layer (v3.2). The air-exchange rate can be derived from CO₂ rather than entered directly, then applied as the AER above. CO₂ is an inert gas, so its removal rate is the outdoor-air exchange rate — particle deposition k is unaffected, and in-room recirculating/HEPA filtration is not captured. Viral inactivation λ, however, now also depends on the CO₂ level for SARS-CoV-2 (see below). A well-mixed single-zone mass balance gives:
Steady state. A measured plateau C_ss inverts to airflow, with ACH = Q/V:
An equilibrium guard flags readings taken before the room nears steady state (~3 time constants, τ = V/Q).
Decay. During the unoccupied tail (n = 0) the concentration relaxes to ambient and the decay constant is the air-change rate directly — no G required:
The tail is fit by least squares in log space (linear in k for a given Cₐ), with Cₐ found by a deterministic grid search; R² is reported. Pairing a decay (gives k) with a steady plateau (gives Q) yields an effective volume V = Q/k.
CO₂ generation G. Per-person generation follows the susceptible-occupant activity level:
with BMR [MJ/day] from Schofield (1985) by body mass and sex, and metabolic rate M (met) per ASHRAE 55. Methods: steady-state and decay CO₂ mass-balance inversion; Persily & de Jonge (2017) for G; Schofield (1985) for BMR; ASHRAE 55 for M.
CO₂-dependent viral inactivation (v3.2). For SARS-CoV-2 and SARS-CoV-2 (v) only, the temperature/humidity/UVB decay term above is combined with an additional CO₂-dependent inactivation term, evaluated at the estimated indoor CO₂ concentration:
For any other pathogen, this SARS-CoV-2-specific decay model does not apply; instead λ is a flat, user-editable input parameter (default 0.63 h⁻¹).
EN 16798-1 combines a person-based term (Pp × occupants n) with an area-based term (Ra × room area A). ASHRAE 62.1 (Table 6.2.2.1, full category/subcategory list) combines a person-based term (Rp × occupants) with an area-based term (Ra × room area). Both standards share the same A (Environment panel) and occupants n = I + S (Exposure & occupants panel) — change those there and both bars update. Uses A = 200 m², n = 2. Coefficients are typical reference defaults — editable.
Defaults are typical reference figures for illustration — verify exact Rp/Ra values against the current edition of each standard before use in design. The “Current” bar is not a requirement — it converts the AER you entered above (Environment panel) into an equivalent outdoor airflow (AER × V), so you can see at a glance whether your actual ventilation meets either standard.
Infectious occupants I and susceptible occupants S are read directly from the AIRC Calculator's "Exposure & occupants" panel, there is nothing to set here. Change I, S, AER, room area/ceiling height, or the pathogen on the AIRC Calculator (tab.2) and this view updates automatically. The red mist and larger droplets drifting from each infectious occupant are illustrative of ongoing aerosol emission. The exhaust fans/streamlines spin and flow faster or slower as the air exchange rate (AER) changes; none of this adds to the underlying risk math.
Infection probability and max occupancy are the headline far-field risk and ⌊1/R⌋ at exposure time t₁ on the Calculator tab; expected new infections is Revent = S × R at the exposure time selected in that tab's "Secondary transmission & overdispersion" panel. These are the exact same figures shown there — this view adds no independent calculation of its own.
A worked University classroom example featuring 1 infected student during a 2-hour lecture with 29 susceptible students and 1 instructor.
| Parameter | Symbol | Input Value | Unit | Annotation / Justification |
|---|---|---|---|---|
| Duration of Exposure | t | 2.0 | hours | Standard class period length. |
| Room Volume | V | 240 | m³ | Length × Width × Height. |
| Ventilation Rate | Q | 2.0 | ACH | Air Changes per Hour supplied by the HVAC system. |
| Temperature | T | 25 | ℃ | Assumed room Temperature. |
| Relative Humidity | RH | 50 | % | Assumed relative humidity. |
| Infectious Persons | I | 1 | persons | The index case present in the room. |
| Susceptible Persons | S | 30 | persons | Remaining uninfected occupants. |
The results below report the individual infection risk, the classroom reproduction number, and the maximum occupancy for the two scenarios.
| Scenario | Ventilation | Individual infection risk | Reproduction Number | Maximum occupancy |
|---|---|---|---|---|
| Baseline | 2 ACH | 4.22% | 1.3 | 23 people |
| Mitigation | 4 ACH | 3.13% | 0.95 | 31 people |
This walks through the same model implemented in Tab 2, in the order the calculator actually evaluates it.
On the intro screen you pick either “Provide Air Exchange Rate (AER)” enter a known h⁻¹ value directly or “Provide CO₂”, which derives the AER for you from a steady-state CO₂ plateau or a CO₂ decay curve (see step 8). Either path feeds the same downstream model.
Enter room volume V, the number of infectious occupants I, and susceptible occupants S. These set the denominator of the well-mixed box and the population over which secondary cases are later counted.
Rather than asking you to type a quanta emission rate directly, AIRC asks what the infectious person is physically doing, resting, standing, light/heavy exercise, breathing, speaking, loudly speaking, or singing, for a chosen pathogen (SARS-CoV-2, its variants, influenza, rhinovirus, measles, TB). Each combination maps to a pre-fitted log-normal distribution of quanta emission rate (ERq), summed appropriately for I infectious occupants. This is the feature the comparative literature specifically calls out as distinguishing AIRC from calculators that require a manual quanta input (see validation, below).
A separate activity selector for the susceptible occupants (resting, standing, light/heavy exercise, or a custom value) sets the inhalation rate IR in m³/h, which multiplies exposure to give inhaled dose.
Three independent loss terms are added into a single infectious-virus removal rate: the air-exchange rate (AER, from step 1), gravitational/particle deposition (k, default 0.24 h⁻¹), and biological decay (λ). For SARS-CoV-2 specifically, λ is further adjusted upward as indoor CO₂ rises, on top of its baseline temperature/humidity/UVB dependence; for every other pathogen λ is a flat, user-editable input.
For a given ERq, the quanta concentration n(t) rises from zero toward a steady value at rate IVRR, and AIRC works with its time-average over the exposure window t - this is what lets a short exposure in a poorly-ventilated room be told apart from a long one in a well-ventilated room, rather than only using the eventual steady-state value.
The exponential dose–response relationship converts inhaled quanta into a probability of infection for one susceptible person. Because ERq is a distribution rather than one number, AIRC doesn't just plug in a mean, it numerically integrates the per-ERq risk over the full log-normal distribution (a 199-point quadrature), so the reported risk already reflects biological variability in how much a given infectious person emits.
For S susceptibles, each is treated as an independent trial at the per-emission probability, so the case count conditional on a given emission rate follows a Binomial(S, P) distribution. Integrating this over the ERq distribution produces a full probability distribution over the number of secondary cases C, capturing not just the expected event reproduction number but the chance of zero transmission and the chance of a superspreading tail.
If you chose the CO₂ path in step 1, AIRC fits either a steady-state plateau or an unoccupied decay tail to back out the air-change rate, feeding it into step 5 as if it had been typed in directly. Deposition and decay terms are unaffected by this CO₂ constrains outdoor-air exchange only.
Tab 3 takes the same room area and occupancy and checks the airflow implied by your AER against EN 16798-1 and ASHRAE 62.1 minimum outdoor-air requirements, so you can see whether the ventilation used in the risk model also satisfies conventional design guidance.
See the room come alive in 3D (drag to orbit, scroll to zoom, hover any occupant): infected occupants emit red aerosol particles whose density and reach scale with real exhalation rate and ventilation, while susceptible occupants shift color from green to red based on their calculated infection risk. It's a direct visual read-out of the same math driving the Calculator tab, not a separate model.
Walk through a real university classroom scenario step-by-step, from entering pathogen and room parameters to interpreting infection risk, expected new cases, and safe occupancy limits. Ideal for first-time users who want to see how each part of AIRC connects before applying it to their own space.
The comparison below draws on a 2025 peer-reviewed review that evaluated AIRC alongside eleven other airborne infection risk calculators using identical, standardized inputs.
The study identifies AIRC as the “Airborne Infection Risk Calculator” (Mikszewski, Stabile, Buonanno & Morawska, 2021) and places it in the quanta-emission-based group of tools, alongside the REHVA Calculator and the COVID-19 Aerosol Transmission Estimator. All twelve reviewed tools were screened for a transparent, published mathematical basis and open web access before inclusion, AIRC met both criteria.
The authors confirm that AIRC relies on the same exponential infection-risk model and the same first-order mass-balance solution for quanta concentration as the other two quanta-based tools the identical structure implemented in Tab 2 above (steps 6–7). They highlight one distinguishing feature: unlike REHVA and the COVID-19 Aerosol Transmission Estimator, which take quanta emission rate as a direct manual input, AIRC instead asks the user for metabolic and respiratory activity and derives the corresponding quanta itself, the same design used in step 3, which the review describes as allowing a more customized and precise assessment.
Run on an identical classroom scenario at three quanta-emission levels (3.8, 38, and 380 quanta/h) and two ventilation rates (0.5 and 5.0 ACH), AIRC's output tracked REHVA and the COVID-19 Aerosol Transmission Estimator closely the relative percentage difference between AIRC and the other two tools stayed within about 0–5% across every scenario tested, which the authors describe as "almost negligible" compared with the much larger spread seen among viral-load-based calculators. Increasing ventilation ten-fold (0.5→5.0 ACH) cut infection risk by roughly the same amount in all three quanta-based tools about 57% at low-to-moderate quanta emission, easing to the low-50s percent at the highest emission rate indicating AIRC's ventilation sensitivity behaves the same way as its peers.
AIRC was one of only four tools the authors could find retrospectively tested against a real, well-documented event: the March 2020 Skagit Valley Chorale rehearsal, where 53 of 61 attendees (87%) were confirmed or strongly suspected to have been infected. Using AIRC, the developers found that a quanta emission rate of about 341 quanta/h over 2.5 hours reproduces the observed 53% attack rate and that this required emission rate falls at roughly the 92nd–93rd percentile of the independently-measured quanta emission distribution for singing (Buonanno et al., 2020), which is the physiological activity the choir was actually performing. That the number needed to match the outbreak lands in a plausible, non-extreme part of an independently derived distribution is a point in favor of AIRC's underlying emission-rate model.
The review is explicit that this should not be read as independent validation. Retrospective fits like the one above were carried out by the tool's own developers, and aside from room volume and occupant counts most other inputs (ventilation rate, exposure time details, emission rate itself) were assumed in order to make the model match the known outcome, rather than measured independently beforehand. The authors note more broadly that no public health body other than the WHO is reported to use any of these twelve tools prospectively or retrospectively, and that validating any of them against real outbreaks remains difficult because outbreak reports rarely capture the building-science details (ventilation, filtration, room volume) these models need.
The authors' overall recommendation is that viral-load-based calculators be favored for detailed, tailored case studies, since they expose more adjustable inputs (inhalation rate, viral emission, mask efficiency). Quanta-based tools such as AIRC are described as more generalized, with fewer knobs to turn, but also as showing the tightest cross-tool agreement of any group reviewed, and no methodological red flags relative to REHVA or the COVID-19 Aerosol Transmission Estimator. In short: the published comparison does not find fault with AIRC's mathematics, but it does mean AIRC's outputs are best used for general guidance and relative comparisons (e.g. "how much does raising ACH from 0.5 to 5 help?") rather than as a precise, standalone prediction for any single real-world event.