Using recent data, obtained by Morgan, Poreh and colleagues, we produc
e correlations for the mass flow of a two-dimensional plume emerging n
ormal to the straight edge of a flat horizontal surface-the balcony-an
d rising up into a uniform atmosphere (the spill plume). A comparison
is made with an earlier correction of the same data by Poreh et al. wh
ich required values of the layer depth, D-B, in addition to those of t
he layer flow per unit length of line plume, M-B'. The treatment of Po
reh et al. followed others assuming the linear relationship typical of
far-field line plumes between the mass flow M' and the height z with
a correction Delta-the virtual source. This linearity is a theoretical
consequence of self-similarity (and a constant entrainment coefficien
t) in the velocity and temperature profiles across the plume, but rece
nt, as yet unpublished, studies including some by computational fluid
dynamics (CFD) cast doubt on the existence of self-similarity for-thes
e plumes at the low heights relevant in practice. However, a dimension
al analysis of the flow does not require the assumption of self-simila
rity and we have demonstrated the linearity as a conclusion and not an
assumption. The effective entrainment coefficient is, as found by Por
eh et al., less than the value 0.16 found by Lee and Emmons and used i
n early work by Morgan and Marshall. The lower figure of 0.11 is consi
stent with other recent work on line plumes. The experimental values o
f D-B, the layer depth reported by Poreh et al., are in reasonable agr
eement with theoretical values for small increases in temperatures onl
y. Experiments in model atria by Hansell, Morgan and Marshall which ar
e not fully two-dimensional are discussed. Our correlation of them can
be reconciled with that obtained by Law and subsequently used by the
Chartered Institution of Building Service Engineers (CIBSE). (C) 1998
Published by Elsevier Science Ltd. All rights reserved.