Does numerical simulation reproduce reality? A comparative assessment of indoor environmental quality in multifamily buildings

Accurately simulating indoor environmental quality is essential for designing energy-efficient and health-promoting residential buildings. As numerical simulation tools become increasingly central to evaluating ventilation performance, it is crucial to assess how well these tools reproduce actual indoor conditions. A performance-based methodology was developed, using TRNSYS and CONTAM coupling to simulate conditions in multifamily buildings equipped with mechanical extraction ventilation. The methodology relies on dynamic heat and mass transfer calculations in a multizone approach, considering airflow, moisture and pollutant transport (CO2, PM2.5, and formaldehyde). This study compares simulated outputs of one prototypical building with real-world data. The dataset includes national French indoor air quality campaigns (2003 - 2005) and involved data collection during one week in 567 homes from 74 locations. It provides weekly time series for CO2, temperature, and RH, and summary statistics (medians, percentiles, and maximum concentrations) for PM2.5 and formaldehyde by main rooms (bedroom and living room). Simulated outputs (CO2, formaldehyde, PM2.5 concentration and RH levels) are obtained varying boundary conditions, including airtightness, climate, and external pollution, reflecting the diversity of real-world variability and compared to the in-situ dataset to explore the reliability of the model, particularly regarding simulation parameters such as moisture and pollutant source strengths and schedules. Results show good agreement for temperature, relative humidity and formaldehyde in terms of medians and interquartile ranges. For PM2.5, the model reproduces typical levels reasonably well for living rooms but underestimates extreme episodes, consistent with highly variable short-term emissions, the absence of particle resuspension in the model, the exclusion of additional indoor sources (e.g., smoking and candle burning), and measurement protocols emphasizing evening/night and weekend activity periods. CO2 is underestimated and shows reduced variability, consistent with fixed generation rates/schedules and constant design ventilation in the model compared with variable occupancy and intermittent or reduced ventilation in real dwellings. Overall, the proposed approach supports performance-based assessment of indoor environmental conditions in mechanically ventilated dwellings and highlights key inputs needed to better reproduce exposure extremes.