Biological aerated filters (BAFs) are submerged three-phase fixed-medi
a reactors for wastewater treatment. A major characteristic of BAF rea
ctors is the use of granular media which allows solids separation as w
ell as secondary or tertiary biological treatment in one unit. The aim
of this work was to design a simple empirical model relating influent
soluble chemical oxygen demand (sCOD) to effluent sCOD and reactor he
ight, verify experimentally the suitability of this model and ascertai
n the relationship between model constants and reactor performance. Th
e theoretical model was based on that designed for trickling filters a
s both fixed-film processes show a similar plug flow pattern. Two reac
tors were set up to run parallel treating settled domestic sewage usin
g media identical in size and shape, except one was less dense than wa
ter (relative density 0.92) and the other was denser than water (relat
ive density 1.05). The reactors were run upflow with liquid flowrates
of 0.29-0.58 m(3) d(-1) (0.2-0.5 litres min(-1)) and an air:liquid rat
io of 10:1. After 4 weeks from start-up, steady-state was reached. Fro
m this point, samples were taken at different heights along the reacto
rs at timed intervals and profiles of sCOD removal against reactor hei
ght were produced. This analysis was repeated for a number of differen
t flowrates and organic loadings (0.57-1.40 kg sCOD m(-3) d(-1)). The
resulting data was then used with the empirical model, which was based
on a first-order reaction, to calculate the values of k (overall pro
cess constant) and n (media factor). A higher value of k was found fo
r the floating media (55) compared with the value found for the sunken
media (33). This indicated the greater efficiency of sCOD removal in
the floating media. The values of the media constants were similar, sh
owing the similarity in the media shape and size. (C) 1997 Elsevier Sc
ience Ltd.