During teeming from ladle to tundish, secondary metallurgy slag draggi
ng occurs when the level of metal is low in the ladle at the end of te
eming. This slag can then get into the continuous casting mold and be
trapped in steel during its solidification, causing a deterioration of
product inclusionary cleanliness during ladle changes (transient stat
es) and their possible downgrading. It is thus particularly important
to be able to detect the appearance of slag in,order to react rapidly
on the ladle closing system and thus to minimize the amount of slag go
ing into the tundish. This goal is realized with the AMEPA slag detect
ion system that was put into service at Sollac Dunkerque in November 1
991. The stakes linked with this slag detection are: reduction of the
amount of slag going from ladle to tundish; reproducibility at the lad
le end; tundish sequence lengthening for equal inclusionary cleanlines
s; at middle term, automating the casting stand since the personnel wi
ll no longer be present to detect slag passage visually. After solving
problems linked to the starting up (suitability of the ladle socket w
ith the steelmaking environment, separation of high potentials in the
connecting cable between central processing unit and control unit) and
regulation of the system (delay time , etc.), a detection rate of 95
% has been reached (number of ladles closed by AMEPA/total number of t
eemed ladles). The average service life of the detection loops is grea
ter than 10 months. As far as metallurgy is concerned, we must point o
ut: A decrease and a good reproducibility of the amount of slag going
from ladle into tundish. It went from 70 kg without AMEPA to 30 kg wit
h AMEPA. On the other hand, there is a clear improvement of the standa
rd deviation of ladle skull weight with the AMEPA system. Lengthening
of sequences for current steels. Inclusion cleanliness of slabs make i
t possible to go from 10 to 15 ladles by tundish with these grades ; f
or IFS, 6 ladles per sequence instead of 4. A study of the AMEPA signa
l was undertaken. Amplitude and slope of the signal characterizing the
slag appearance speed have been plotted on 1 127 heats as well as inn
er nozzle age, plates age, teeming time, ladle age, slag basicity inde
x and intermediary nozzle age. The result is that the amplitude greatl
y depends on the inner nozzle geometry (geometry of the runner entry)
and to a lesser degree on plates age (geometry of runner exit). On the
other hand, slope only depends on inner nozzle age. Otherwise, a stud
y is running to correlate the signal to the amount of slag going into
the tundish.