Experiments on microfiltration of raw cane remelts with tubular mineral mem
branes show that rigorous methodology is required if laboratory data are to
be extrapolated to an industrial setting. We studied two main variables of
ultrafiltration: sugar dilution and the control mode (constant pressure or
constant permeate flux). Optimal sugar dilution, on the basis of the highe
st dry substance Bur criteria, was in the range 28-33 g DS/100 g. Compariso
n between the two control modes, with the remelt at 30 g DS/100 g and 80 de
grees C and tubular mineral membranes (kerasep 15 kDa, Orelis, France), sho
wed that use of a constant permeate flux reduced fouling of the mineral mem
brane without altering selectivity. The optimal flux setpoint of 100 l h(-1
) m(-2) corresponded to the maximal productivity of 2.5 m(3) m(-2) day(-1)
(permeate volume collected per membrane area and per day, including time lo
st to cleaning) and the lowest cleaning frequency. The permeate flux contro
l mode was favourable in ultrafiltration, but was less influential than in
microfiltration. Ultrafiltration of remelt through spiral organic membrane
(S4 HFK VSF 131, 5 kDa, Koch, USA) showed that the control mode was also le
ss important in this case. Our results confirmed that the control mode is o
f importance in the particular case of microfiltration through tubular mine
ral membranes. Few relevant studies have been published, as these membranes
are not widely used. (C) 2000 Elsevier Science Ltd. All rights reserved.