2026 Volume 21 Issue 2 Pages JFST0013
Indoor airflow monitoring typically relies on a few fixed sensors and provides only ambient information. The flow field can instead be reconstructed using computational fluid dynamics simulations and corrected with sparse measurements via the ensemble Kalman filter, thereby estimating the velocity even at sensor-free locations. This study, under common conditions, simultaneously compared fixed sensors that measured airflow at several locations and a mobile sensor that visited each location individually, in terms of threshold-crossing time and the number of assimilated point-time records. Twin experiments were performed in a two-dimensional laminar partitioned cavity with periodic wall motion for single- and two-sided cases. Observation schemes were evaluated using the threshold-crossing time, defined as the first time at which the full-field root-mean-square error of the velocity magnitude decreased below a prescribed threshold. In both cases, the best fixed-sensor scheme reached the threshold earliest. The best mobile-sensor schemes used only 12 point-time records over the common evaluation window and reached the threshold later than the best fixed-sensor schemes; however, they did so earlier than the no-assimilation baseline. Intermittent fixed-sensor controls showed that sparse updating at a well-placed single point can also reduce the error. At the primary threshold, however, visiting multiple locations yielded a shorter threshold-crossing time. Increasing the number of simultaneously observed points did not consistently shorten the threshold-crossing time under the study’s settings. These results demonstrate a tradeoff between the threshold-crossing time and the number of assimilated point-time records. Because the comparison is based on a prescribed candidate set and a small ensemble, the rankings should be interpreted as screening results for the tested conditions.
