Measurement of ventilation air flows.

The common way to determine air infiltration, exfiltration and interzonal flows from tracer gas measurements in multizoned buildings is to rely upon the standard single or multizone model, Vc(t) = Qc(t)+p(t) . Here c, p are zonal tracer concentrations and injections, t is time and V, Q are the sought volumes and flows. This model may work well provided that all zones are sufficiently well mixed and all flows really are constant during the measurements. The latter can be doubtful in naturally ventilated buildings, especially as the measurements may require several hours. This paper discusses one single question: What could be gained if the flows of the model are allowed to vary in time?, i.e. we will consider a (bilinear) model: V c(t) = Q(t) c(t) + p(t) . The variation of Q(t) will, however, be restricted to be piecewise linear. This extension will complicate the identification procedure, i.e. the computations, somewhat. On the other hand it does not alter the tracer gas measurements needed. The idea is rather to pick out more information from the standard tracer gas measurements. To test the extended model, single-tracer gas measurements have been carried out in a natural ventilated row house. Some basic concepts are introduced with one and two zone models and the house is modelled with a nine zones model. As intended, the extended model matches the measurements better. However, as demonstrated in the paper, this does not guarantee that the model found actually is better.