Predicting indoor infection risk with airflow modelling
- Babak Baghaei
- Sep 17
- 2 min read
Addressing airborne infection risk in indoor environments remains one of the most important challenges in building design, healthcare safety and workplace resilience. Although recent years have produced significant advances in ventilation research, many findings have still not been translated into practical, actionable guidance for engineers and facility managers.
A new peer-reviewed study authored by Mohammad Elsarraj, with co-authors Dr Yasser Mahmoudi and Dr Amir Keshmiri (Founder and Technical Director of Mansim), provides one of the most comprehensive frameworks to date for linking airflow behaviour, ventilation effectiveness (VE), infection probability and CO₂-based sensing.
Understanding airflow to understand risk
Airborne pathogens move according to the physics of indoor airflow, shaped by ventilation strategy, room geometry, heat sources and occupant positions. The authors apply a Probability of Infection (POI) metric inside a detailed office environment to explore how:
airflow patterns
ventilation rates
diffuser configurations
VE and “age of air”
influence both infection risk and CO₂ concentration distribution.
Ventilation Effectiveness: the critical parameter
A major contribution of the study is quantifying how Ventilation Effectiveness (VE) governs infection risk.
Key findings include:
Dead zones and recirculations increase air age and dramatically reduce VE.
Low VE sharply increases infection risk even at higher ventilation rates.
Systems with higher VE require far smaller flow-rate increases to achieve safe POI thresholds.
VE depends heavily on ventilation layout, diffuser placement and airflow pathways.
This demonstrates that quality of ventilation matters more than quantity.
Worst-case infector positioning
The study also introduces a method to determine the worst-case infector position, a vital insight for space planning and risk mitigation.
In these worst-case scenarios:
infectious particles travel through the maximum number of occupants
airflow patterns delay particle removal
CO₂ tends to correlate with elevated POI
This method provides designers with an evidence-based approach for identifying and addressing high-risk layouts.
Estimating the number of infectious individuals
Using local epidemiological data, the authors propose a practical method for estimating the likely number of infectious occupants within a space and identifying an acceptable POI threshold. The study adopts a threshold of 7.5%, intended to support ventilation-design decisions aimed at reducing airborne transmission.
From CFD to real-time prediction
Building on the team’s prior work, the paper outlines a framework for:
generating representative CFD cases
extracting key features from airflow and contaminant fields
training Random Forest (RF) machine-learning models
using CO₂ sensors as a real-time proxy for infection-risk prediction
This provides a pathway toward automated indoor-air mitigation systems, bridging high-fidelity modelling and operational building management.
Why this matters for industry
The findings offer actionable guidance for multiple sectors, healthcare, education, offices, housing, transport, including:
how to identify and remove stale-air pockets
how VE influences required ventilation upgrades
where to position diffusers for safer airflow
when CO₂ sensors can act as reliable indicators
how ML can support adaptive ventilation control
As respiratory-virus risks continue globally, this research provides a robust, physics-based toolkit that moves beyond static ventilation guidelines.
It also reflects Mansim’s commitment to using CFD and data-driven modelling to inform safer, healthier and more resilient built environments.
Comments