Air Exchange Rate Estimation from CO2 Decay: Background-Bias Benchmarking, Time-Window Stability Mapping and Implications for Ventilation Heat-Loss Assessment

Reliable estimation of the air change rate from CO2 decay measurements is essential for ventilation assessment, but results may strongly depend on the assumed background concentration and the selected analysis window. This study compares the ISO 12569 two-point and multi-point methods with a study-specific nonlinear least-squares (NLSQ) exponential-fitting approach developed for systematic window-sweep analysis. A synthetic benchmark with known decay parameters was used to quantify the sensitivity of the estimated air change rate N to background-concentration error and time-window selection. The framework was then applied to CO2 decay measurements from four rooms and to an OpenFOAM-generated CFD decay case using dense N(tbeg,tend) stability maps and automatic plateau selection. The benchmark showed that admissible windows shrink as background uncertainty increases, while late, tail-dominated intervals are particularly prone to bias. The two ISO methods produced similar results, whereas the NLSQ approach generally preserved stable solutions over longer windows when fitting started sufficiently early. For three measured rooms, stable regions yielded effective N values of approximately 0.14, 0.75, and 1.24 1/h, while one room remained non-stationary. To assess energy implications, N was propagated into the ventilation heat-loss coefficient Hvent = 0.34 NV. Illustrative late-tail scenarios showed that background-sensitive window selection may translate into substantial over- or underestimation of ventilation-related heat-loss indicators.