the performance and metallurgical efficiency of a snif sheer ... - Pyrotek

Abstract
THE PERFORMANCE AND METALLURGICAL EFFICIENCY
OF A SNIF SHEER P-30HB DEGASSING UNIT
Presented at the International Melt Quality Workshop,
Madrid, Spain, 25-26 October 2001
Neil J Keegan 1 , Robert A Frank 2 , Nicholas Towsey 3 , Angela Hardman 4 ,
1 Pyrotek Engineering Materials Ltd., Dudley, West Midlands, UK.
2 Pyrotek Inc., SNIF Systems, 765 Old Saw Mill River Road, Tarrytown, NY, USA
3 VAW aluminium AG, Research and Development, Bonn, Germany.
4 London & Scandinavian Metallurgical Co. Limited, Rotherham, South Yorkshire, UK.
The test facility at VAW’s Rheinwerk Centre 40 offers a
unique opportunity to quantify and baseline the
metallurgical efficiency (i.e. hydrogen and inclusion
removal efficiency) of a SNIF degassing system under
controlled conditions. Started early in 2001, this
programme of work is the 3 rd phase in the joint partnership
on metal treatment of wrought aluminium alloys,
augmenting the earlier filtration programme carried out
between 1995 and 2000.
This paper describes the major steps proposed in this 3 rd
phase of work, along with some preliminary results. These
include the work carried out to date to assess the best
technique to evaluate inclusion removal. This primarily
focuses on the LiMCA method’s difficulty in measuring
accurately inclusion concentration after a degasser.
PoDFA and Prefil were used to supplement this study.
This paper also presents the preliminary results of the
initial baseline evaluation of the SNIF system using argon
over a range of rotor speeds. Once this baseline study has
been completed the work will move on to study the
interaction with grain refiners and CFF’s and the impact of
Cl2 on the inclusion removal efficiency.
Superimposed on the parameter variations would be the
longer term influences of real conditions such as
component wear, dross build up, etc. It is hoped that the
effects of each parameter can be isolated from these
influences.
1.0 Introduction
In this research work we have historically adopted a “layer
on layer” approach so that we can develop and build upon
the results obtained in earlier phases. The first phase of
our investigation looked at ceramic foam filters (CFF’s) in
isolation of other in-line treatments to give us a
fundamental understanding on the filtration process and a
performance baseline. In the second phase of this research
we introduced grain refiner into the system that we had
carefully baselined. This phase was intended to be closer
to typical production conditions existing in many DC cast
houses. In this latest work we start to consider the impact
of a degasser on downstream metal quality. Having
established its performance on its own we will then move
on in the future to look at the combined effects of a
degasser and a CFF on the quality of the metal processed.
Naturally a most important area to investigate is how the
performance of the SNIF varies with its life and what
impact any build up on the walls or the degradation of the
rotor etc., has on the overall performance of the system.
2.0 Experimental Procedure
The trials in this study were conducted using the specially
dedicated, production scale R&D unit at the VAW
Rheinwerk plant. AA1050 alloy, batched using reduction
line metal and cast into ingots by the direct chill process at
a flow rate of 10 tonne/hr, was used throughout. Metal
quality measurements were made in the launder using
LiMCA and PoDFA techniques. The inclusion content of
the AA1050 metal flowing from the furnace, was varied
by a stirring or settling practice. To markedly raise the
inclusion content, the melt was air-stirred for five minutes
prior to casting.
Key to the success of this work is the ability to be able to
carry out meaningful inclusion measurements downstream
of a degasser. The options available were LiMCA run in
normal mode using extension tubes as recommended by
Bomem, LiMCA pressure mode measurements as
developed by VAW R+D department 1 , Prefil flow rate
curves and PoDFA analysis of the solidified Prefil sample.
The LiMCA has a problem in measuring inclusion
concentration downstream of a degasser because of the
presence of gas bubbles in the metal. These are counted
along with inclusions. During LiMCA testing a small
volume of liquid aluminium is drawn through a 300µm
orifice in a glass tube. The LiMCA has electrodes inside

and outside the tube. The electrical conductivity of the
metal is changed as non-conducting inclusion particles are
drawn into the tube along with the liquid metal. The
particles can, therefore, be counted and their size
estimated in a way similar to that of a Coulter counter.
The normal way that these measurements are done is on
the up cycle known as the normal or vacuum mode. In
order to try to discount the effect of gas bubbles a method
was developed by VAW R+D 1 whereby the the counting is
done on the down cycle as the metal is ejected through this
orifice. This is referred to as the pressure mode.
Recently Comalco 2 reported some work which looked at
the pressure method and the use of extension tubes
independently of our testing and whilst they felt it had
potential they raised concerns that inclusion losses were
occurring during these measurements.
In our work our first aim was to verify that the use of 2
and 3 LiMCA’s at different locations with normal mode of
operation and extension tubes gave us repeatable results.
We then wanted to establish what errors we had by using
the pressure mode and extension tubes. To do this we
carried out trials running the LiMCA’s side by side in our
casting line without gas bubbles in the system – i.e. before
the installation of the degasser. We then carried out
similar measurements downstream of a SNIF degasser on
a production line in the Rheinwerk plant to make
comparisons of the results in the presence of bubbles. As a
result of this we felt that we were in a position to progress
on to install the SNIF P30 on the experimental casting
line. We also concluded that N20 and possibly higher
would now be used as our baseline as N15 was considered
too sensitive to gas bubbles.
Figure 1 shows the SNIF SHEER P30-HB as installed on
the experimental casting line during its commissioning
cast. This is a sealed single rotor SNIF system. Figure 2
shows a photograph of the experimental casting unit and
launder runs taken from the roof of the SNIF P30. Metal
enters the SNIF from the 20 tonne cylindrical casting
furnace on the left before exiting via the launder on the
bottom right hand side of the photograph. This launder
continues fully 9m before reaching the casting launder.
The experimental casting line was set up like this in order
to try to limit the influence of gas bubbles on our LiMCA
measurements. Two LiMCA’s can also be seen in this
photograph as used during the experimental programme.
One is just in front of the SNIF P30 on the bottom left
hand side and the second is at the top centre of the
photograph showing measurements being conducted
downstream of the SNIF unit just ahead of the casting
launder.
Figure 1: SNIF SHEER P30-HB on Rheinwerk casting line.
Figure 2: Experimental Casting Pit Layout.
3.0 Results and Discussion
3.1 LiMCA testing using normal and extension tubes in
normal and pressure mode without argon gas bubbles in
the system.
Figure 3 shows the LiMCA traces for 3 LiMCA’s ran
side by side with extension tubes on our casting line
without bubbles i.e. without a degasser in the line. We can
see that the three units are quite repeatable over the cast as
a whole. There is a higher degree of scatter which is quite
normal when running with extension tubes. The areas of
the traces outlined with the red circles show where one of
the units was run in pressure mode. The first and last is
where the LiMCA with the black line is run in pressure
mode. In the first instance the pressure mode seems to
have made very little difference. In the latter case as with
the LiMCA shown by the pink line there may be an
indication that the readings are slightly lower in this mode
than in normal mode. In contrast, however, other casts
showed the pressure mode sometimes gave higher results

than vacuum. This was probably, given general
experiences with extension tubes, due to the scatter that
occurs whilst running these tubes.
Figure 3: 3 LiMCA’s run in normal mode with extension
tubes - no argon gas bubbles – i.e. without degasser.
Figure 4 shows a trial where all three units were run
initially in normal mode with normal tubes for the without
bubbles case i.e. without degasser. The repeatability is
again very good. After 35 minutes an extension tube was
put on the second LiMCA, the one represented by the blue
line. It can be seen that this has led to a lower level of
inclusions being counted.
Figure 4: 3 LiMCA units run in normal mode, 2 with
normal tubes, 1 with extension tube – no argon gas bubbles.
Figure 5 shows clearly the offset that occurs when using
extension tubes compared with normal tubes. This is again
for the without bubbles case. The green line is our
reference line and shows the trace for normal mode with
normal tubes. The blue and pink lines of the second and
third LiMCA are run using extension tubes. It can be seen
that there is a clear offset between the counts obtained
with extension tubes as compared with those gained using
the normal tube. After approximately 25 minutes the
second LiMCA shown with the blue line was run in
pressure mode. In this case, admittedly at low inclusion
levels, the pink and blue lines show little difference
between the pressure and normal mode.
Figure 5: 3 LiMCA units, 1run in normal mode with a
normal tube, 2 with extension tubes – no argon gas bubbles.
3.2 LiMCA testing using normal and extension tubes in
normal and pressure mode with argon gas bubbles in the
system.
Having carried out some validation trials using the
LiMCA without bubbles, we then went on to evaluate
some production metal downstream of a SNIF degassing
unit which represented the ‘with argon bubbles’ case.
Figure 6: LiMCA results before and after a SNIF R140 on
Rheinewerk production unit, with extension tubes.
Figure 6 shows a typical trial result. The pink line is from
the LiMCA downstream and close to the degassing unit.
It can be seen that the first measurements made in normal
mode using extension tubes show that the LiMCA counts
are much higher after the degasser than before. This is

elieved to be as a result of counting bubbles. When we
switched the LiMCA to pressure mode we can see that for
these conditions we appear to be eliminating most of the
gas bubbles from our counts. We do not know the loss of
inclusions due to the pressure mode and extension tube
combined. When we switched back to normal mode the
line immediately returned to a level similar to that
detected before. Going back to pressure mode again gave
us a value similar or slightly lower than that of the
incoming metal counts.
Figure 7: LiMCA results before and after the SNIF
SHEER P30-HB with extension tubes in pressure and
vacuum mode.
Figure 7 shows one of the first LiMCA trials run with the
SNIF P30 for the “with gas bubbles” scenario. In this trial
the SNIF was set to run at 500 rpm and with 3Nm 3 /hr
argon. We were now varying the SNIF parameters and
along with the PoDFA trying to establish the measurement
errors for each setting due to the different inclusion and
bubble losses. It must be assumed that the inclusion and
gas bubble size distribution is also changing accordingly.
The before LiMCA curve is in pink and it can be seen that
the incoming metal was very clean. This was verified by
the PoDFA counts taken to supplement the LiMCA
measurements. During this trial consecutive points were
taken in normal and pressure mode using extension tubes.
The pressure mode has lead to lower counts than was the
case for the normal mode but the downstream LiMCA
values can be seen to be still higher than those of the
upstream LiMCA . The PoDFA showed otherwise being
very clean at less than 0.01mm 2 /kg. It appeared, therefore,
that despite having the LiMCA over 9m away we were
still counting very small gas bubbles after the degasser.
The marked reduction in counts detected using the
pressure mode as opposed to the normal mode is shown
more clearly in Figure 8, the histogram for the previous
trial.
Figure 8: LiMCA histogram results for after the SNIF
SHEER P30-HB with extension tubes in pressure and
vacuum mode.
The SNIF P30 operating parameters were then changed to
check our measurement technique over a wider range of
settings. In Figure 9 we see that for these parameters the
LiMCA’s perform differently to before. The incoming
metal is dirtier than for the previous trial and is shown
with the blue line. This was again found to agree with the
subsequent PoDFA measurements. The downstream
LiMCA SNIF values are again shown in pink. This time
when we switch to pressure mode as represented by the
red points, we seem to be excluding most of the gas
bubbles generated for these SNIF settings. The PoDFA’s
were again as clean as for the last example which
supported this conclusion.
Figure 9: LiMCA results for before and after the SNIF
SHEER P30-HB operating at 400rpm with 2Nm 3 /hr argon
run with extension tubes in pressure and vacuum mode

Figure 10: Typical PoDFA fields for before and after the
SNIF SHEER P30-HB on the experimental casting line for
cast 020401-2
Figure 10 shows some typical fields from the
metallurgical evaluation of the Prefil samples. The left
hand side shows the before SNIF sample with a large
aluminium carbide layer adjacent to the top of the sample.
This has a PoDFA count of 0.6 mm2/kg. The after SNIF
sample can be seen to be very much cleaner with much
less of a layer being formed in the sample. This had a
PoDFA count of 0.2mm2/kg.
Figure 11: PoDFA total inclusion contents for the SNIF
SHEER P30-HB on the experimental casting line.
Figure 11 shows the results of the PoDFA counts for the
before and after SNIF P30 samples taken during our trials.
It can be seen that for the clean SNIF unit a good inclusion
removal is consistently being obtained. This shows that
the LiMCA is indeed measuring bubbles after the SNIF
rather than inclusions.
It appeared, therefore, that we still needed to adapt our
LiMCA measurement technique if we were to be able to
use them successfully downstream of a degassing unit.
With this in mind we have come up with a special
approach. We have called this the “background method”.
The logic is that we collect data upstream and downstream
of the degasser for a period until the readings are
considered to be stable. Figure 12 shows a trial where we
tested this new approach. The downstream SNIF values
are again in pink and can be considered to be our
background due to gas bubbles which should not vary for
these particular SNIF settings. We then stirred the furnace
during the cast. It is assumed our background for
semiquantitative purposes will not alter too much as a
result of the varying error in inclusion counts, whilst the
incoming inclusion loading should naturally go up. We
can, therefore, judge semi-quantitatively how the SNIF
has performed by the response of the downstream line. If
as in this case the line remains fairly constant despite a
large increase in loading upstream of the SNIF it can be
assumed that the SNIF is doing a good job of cleaning the
metal.
Figure 12: LiMCA results during a trial using the
“Background method” for the SNIF SHEER P30-HB on
the experimental casting line.
If this is the case as is shown again in Figure 13 we have
what we will call the best case scenario. The downstream
line has remained constant so we have what we will grade
as an “A” performance.
Figure 13: LiMCA results during cast 020501-1 using the
“Background method” for the SNIF SHEER P30-HB.

In Figure 14 where we changed the SNIF P30 parameters
to a more unsuitable setting we see the downstream line
has increased along with the upstream line after stirring.
This is what we rank as a “B” performance. Similarly if
the downstream line appears the same or worse than the
before SNIF curve then we shall have what we consider to
be a “C” performance.
Figure 14: LiMCA results during cast 030501-2 using the
“Background method” for the SNIF SHEER P30-HB.
These rankings will be validated over a large number of
casts and in conjunction with PoDFA analysis. With this
methodology we aim to make sound assessments of the
inclusion removal capability of the SNIF and how this
changes with varying operational parameters. Due to the
shortcoming of the LiMCA in the presence of gas bubbles
we cannot produce quantitative results. However, with full
PoDFA back up we hope to get a semi-quantitative
classification.
Figure 15: PoDFA total inclusion contents for the SNIF
SHEER P30-HB on the experimental casting line.
Finally, just to confirm the inclusion removal that we
anticipate to have occurred during our background method
trials, Figure 15 details the accompanying PoDFA counts
for our latest series of trials. It can be seen that as was the
case for the first trials we are indeed getting good and
verifiable inclusion removal using the clean SNIF.
Figure 16 shows the Alscan hydrogen removal
efficiencies for the same trials. This shows that the SNIF
is working fundamentally as expected and appears in line
with the positive inclusion removal data obtained. This
obviously needs to be verified over a much larger number
of trials before a proper comparison with the inclusion
removal data can be made.
Figure 16: Alscan hydrogen removal efficiencies for the
SNIF SHEER P30-HB on the experimental casting line.
4.0 Conclusions
(1) The “Background” method appears promising but
needs be confirmed.
(2) This and PoDFA indicates good inclusion
removal with argon and a clean SNIF.
(3) The pressure mode alone is not accurate enough
to compare parameters.
(4) Prefil curves alone appear invalid so it is
necessary to rely on PoDFA metallographic
analysis.
(5) Alscan will be required to supplement inclusion
removal measurements.
5.0 References
(1) H-P. Krug, & W.Schneider,.
"A contribution to inclusion measurement after in—line
degassers with PoDFA and LiMCA", Light Metals 1998,
pp.863-870.
(2) M.Cooksey, T.Ware & M.J.Couper
”Effect of pressure cycle and extension probe on LiMCA
measurement of inclusions”, Light Metals 2001, pp 965-971