Alcastek2002_filtrat.. - Pyrotek

An Evaluation of Industrial Filtration Systems
Neil J Keegan 1 and Steve F Ray 2 ,
1 Pyrotek Engineering Materials Ltd., Dudley, West Midlands, UK.
2
Pyrotek SA, Ile Falcon, 3960 Sierre, Switzerland
Abstract
This paper overviews the performance of a range of ceramic foam filters and
compares their efficiency to other industrially available in-line filtration systems. The
performance and consistency of these ceramic foam filters is shown to be clearly
related to their respective cell size. Results from an in-depth experimental
programme have been compared with examples taken from industrial trials thus
demonstrating the excellent efficiencies obtained using this type of filter media.
Introduction
An extensive programme of work has been carried out since 1995 in order to build up
knowledge on the performance of key individual in-line treatment systems used in the
casthouse 1,2 . This work was conducted on an experimental casting line in a step-wise
manner so as to endeavour to build on the knowledge that was gained at each stage
of the modular study. The initial aim of this research work was to generate an
understanding of what impact each of the in-line components eg. ceramic foam filters
(CFF’s), degassers, grain refiners etc had on the quality of the liquid aluminium. This
work sought initially to understand the performance of each component in isolation,
before continuing to investigate for any interaction effects that might occur when
these systems were run in combination, as is the case under normal production
conditions.
Filtration Efficiency Studies
The first phase of the programme evaluated the performance of a wide range of
ceramic foam filters before going on to compare these with other common industrial
filtration systems. This initial filtration study was carried out in the absence of grain
refiner.
Typical LiMCA II evaluation data for a 50 grade SIVEX CFF is shown in Figure 1. It
can be seen that excellent inclusion removal was observed. The inclusion size
removal data accompanying this cast is shown in Figure 2. A remarkably good
removal efficiency was achieved across the whole of the limited inclusion size range
which was present within the potroom metal that was used for these trials.

6
LiMCA-Evaluation 17" 50 grade CFF
No GR
Impurity Level N 20 [k/kg]
5
4
3
2
before CFF
after CFF
1
0
0 10 20 30 40 50 60 70 80
Casting Time [min]
Figure 1 : LiMCA II Evaluation data for a 50 grade SIVEX CFF
100
90
80
70
% Efficiency
60
50
40
30
20
10
0
N20 20-
25
25-
30
30-
35
35-
40
40-
45
45-
50
50-
55
55-
60
60-
70
70-
80
80-
90
90-
100
100-
120
120-
150
150-
300
Particle size range [µm]
Figure 2 : LiMCA II inclusion size removal data for the 50 grade SIVEX CFF shown
in Figure 1.
Figure 3 shows a comparison of the filtration performance, based on LimCA II data,
of a variety of grades of SIVEX CFF that were tested extensively in this programme.
In this figure the range of efficiencies that was measured during the trials for each
grade of CFF is shown as a solid block. The range of efficiencies measured during
these trials is included in brackets. The mean efficiency of each grade of SIVEX CFF
is shown by the solid line.

100
80
(33)
(27)
(12)
60
(41)
(35)
Filtration Efficiency %
40
20
(72)
0
15" Grade 30 15" Grade 50 17" Grade 30 17" Grade 50 17" Grade 65 17" Grade 80
Figure 3 : Comparison of Filtration Efficiency Ranges for the Standard and Fine
Pore Ceramic Foam Filters.
Figure 3 shows that all of the SIVEX CFF’s are capable of producing high levels of
filtration efficiency but that the range of the measured filtration efficiencies decreases
quite significantly as we move to finer porosity of ceramic foam filter grade. In other
words: the reliability of the filtration system increases as we move to finer pore filters.
It was also found that the finer the filter the greater its ability to remove smaller sized
inclusions. This becomes more important if the alloy being cast is destined for critical
products such as foil or litho sheet.
The reason for the greater reliability and improved performance of the finer pore
CFF’s is best understood by relating these liquid metal results with the physical
characterisation of the CFF’s that were made prior to their use in the programme 3 .
Figure 4 shows a summary of this data. The lines indicate pore size distribution and
the bars show the range of inclusion removal efficiencies recorded for each filter.
Green represents grade 80, blue grade 50 and red grade 30. Both pore size and
filtration efficiency vary less with finer filters. The Grade 30 filter has a pore size
range (within a single filter) of around 1400 microns. The Grade 80 filter has a range
of less than 300 microns. The probability of inclusion capture varies with pore size,
so a larger variation in pore size can result in unpredictable and, therefore
inconsistent, filtration performance.

Figure 4 : Pore size and inclusion removal efficiency distribution for SIVEX CFF
grades 30, 50 and 80
Figure 5 shows the in-line systems performance comparison for the complete trial
programme. It can be seen that the performance of the SIVEX CFF’s compare very
favourably with the more complicated tube and deep bed systems 2,4 . The results for
the single SIVEX CFF’s were also found to be better than those determined for the
17” 30/50 sandwich filter which was run under identical conditions.
Figure 5 : In – line filtration systems performance comparison

50 ppi, NO rod, stirred during cast
18
16
sampling before fpf extremely difficult, changed probe twice
14
Impurity Level N15 [k/kg]
12
10
8
6
STIR
STIR
After filter
Before filter
4
2
0
0 10 20 30 40 50 60 70 80 90
Casting Time (min.)
Figure 6 : Stability of 50 grade SIVEX under severe metal level disturbances 3 .
Another important result for the SIVEX CFF’s measured in the earliest phases of this
programme (without grain refiner) was their stability under severe metal level
disturbances, as well as increased, and sustained, high inclusion loadings 4,5 .
Figure 6 shows the LiMCA traces for a trial where the loading on the CFF was kept
at a high level by stirring the furnace continuously throughout the cast. Grain refiner
was omitted from this trial to check if high inclusion loading alone resulted in release
effects. It can be seen that, despite a very high incoming inclusion loading (>16k/kg),
no release effects at all are evident. In fact, despite there being severe metal level
disturbances due to the vigour of the stirring, the 50ppi CFF displayed a very high
efficiency and a stable and consistently low post-filter LiMCA value (0.25k/kg) 5 .
6
50ppi, CONTROL, LSM 3:1 Tibor in at start, stirred
5
Impurity Level N20 [k/kg]
4
3
2
rod in
After fpf
before fpf
1
0
0 10 20 30 40 50 60 70 80
Casting Time (min.)
Figure 7: LiMCA II Evaluation data for a 50 grade SIVEX CFF with a 3%Ti:1%B rod
fed ahead of the CFF - “high inclusion loading”.

The final phase of the filtration programme investigated the impact of adding a grain
refiner upstream of the SIVEX CFF 5,6 . The extremely high filtration efficiencies
measured for the 50 and 80 grade SIVEX CFF’s in the absence of grain refiner were
reduced when an Al-3%Ti-1%B rod was introduced ahead of the CFF. This was only
found to be significant at high inclusion loadings 4,5,6.
As an example of this, Figure 7 shows the LiMCA II evaluation for a 50 grade SIVEX
run in conjunction with LSM, Al-3%Ti-1%B rod. It can be seen that the use of a CFF
is still very beneficial even though the previously extremely low LiMCA levels
detected after the filter (Figure 1) have increased from typically 0.25K counts/kg to
~1K counts/kg.
The N20 inclusion size removal data is shown in Figure 8. This shows a good
inclusion removal performance across the size range of inclusions found within the
metal that was tested.
100.00
90.00
80.00
70.00
Removal Efficiency %
60.00
50.00
40.00
30.00
20.00
10.00
0.00
N20 20-
25
25-
30
30-
35
35-
40
40-
45
45-
50
50-
55
55-
60
60-
70
70-
80
80-
90
90-
100
100-
120
120-
150
150-
300
Particle Size Range (Microns)
Figure 8 : LiMCA II inclusion size removal data for the 50 grade SIVEX CFF shown
in Figure 7 - “high inclusion loading”.
Figure 9 shows the effect of the introduction of this grain refiner on a settled, ‘clean’
cast. It can be seen that the grain refiner has had little impact on the downstream
metal quality in this case.

50ppi, settled,1kg/T LSM 3:1 TiB rod in at start of cast - CONTROL type cast
6.0
Impurity level N20 k/kg
4.0
2.0
rod in
Before CFF
After CFF
0.0
0 10 20 30 40 50 60 70 80
Casting Time (min.)
Figure 9 : LiMCA II Evaluation data for a 50 grade SIVEX CFF with a 3%Ti:1%B rod
fed ahead of the CFF – “low inclusion loading”.
Figure 10 : Effect of grain refiner on average CFF efficiencies
Figure 10 shows the average filtration efficiency based on the N20 LiMCA II values
for casts where grain refiner was introduced at the start of casting and the inclusion
loading was high, compared with the average performance without any grain refiner.
It can be seen that the N20 removal efficiency has been reduced to approximately
70%.

The high filtration efficiencies determined for the CFF’s in the absence of grain refiner
is thought to be related to the build up of ‘bridges’ of carbides and oxides within the
upper regions of the filter. These bridges were seen in 50 grade filters and above but
were particularly prevalent in the 80 grade CFF’s. Specifically, the second phase of
the filtration evaluation programme conducted jointly between VAW, LSM and
Pyrotek, showed that these bridges did not appear to be present within drained
CFF’s examined for those trials that were conducted with a grain refiner ahead of the
CFF. This work showed and reported that it was the TiB 2 component within the grain
refiner rod that appears to be responsible for this effect 5,6 .
It would appear that the choice of grain refiner composition can have an effect on the
magnitude of its impact on the CFF performance 4 . 5%Ti:1%B, 3%Ti:0.2%B and
1.2%Ti:0.4%B are reported to have a less severe impact on the CFF performance
than that demonstrated here using a 3%Ti:1%B rod 4 .
Cast #
Before SIVEX
(mm2/kg)
After SIVEX
(mm2/kg)
Filtration
Efficiency %
0.12 0.024 91.8
1 0.069 0.052 24.6
0.724 0.06 91.7
0.142 0.07 50.7
2 0.172 0.055 68
0.724 0.042 94.2
3 0.756 0.029 96.2
0.212 0.133 37.3
4 0.138 0.004 97.1
0.022 0.008 63.6
0.025 0.01 60
0.109 0.038 65.1
5 0.187 0.019 89.8
0.36 0.058 83.9
0.234 0.046 80.3
6 0.116 0.043 62.9
0.454 0.026 94.3
7 0.208 0.092 55.8
0.058 0.038 34.5
0.05 0.019 62
8 0.046 0.038 17.4
0.124 0.02 83.9
0.225 0.045 80
9 0.121 0.014 88.4
0.039 0.022 43.6
0.819 0.051 93.8
10 0.138 0.008 94.2
0.094 0.048 48.9
Mean Efficiency 69.8%
TABLE 1 : Filtration Efficiency data for a 23” 40 grade SIVEX , 6XXX series
production trial, with a 5%Ti:1%B rod ahead of the CFF.

It is interesting to now balance these experimental results with examples taken from
industrial trials with grain refiner introduced ahead of the CFF, where the filtration
performance of the SIVEX CFF has been measured using ‘PoDFA’ assessment.
Table 1 shows the results of a 10 trial evaluation of a 6XXX series alloy that was run
using a 23” 40 Grade SIVEX CFF. A low flow rate of 5mm/sec within the CFF was in
operation for these trials. The grain refiner added ahead of the CFF was 5% Ti:1% B.
Six Prefil samples were taken during the duration of the cast (3 before and 3 after the
CFF) for 8 casts and two pairs of samples for two additional casts, giving a total of 56
samples in all. It can be seen that the filtration efficiencies recorded are very good.
The range of filter efficiencies determined is not dissimilar to those found in our
experimental programme for the coarser SIVEX CFF’s. It is also interesting to note
that the mean filtration efficiency determined for all the sample pairs taken in these
10 casts is approximately 70% (compare with Figure 10).
Further evidence of the very good efficiencies that have been obtained during
production trials in the presence of grain refiner are shown in Figure 11. Three of
these trials were conducted with 5% Ti:1% B and one with 3% Ti:1% B rod fed ahead
of the CFF and two with grain refiner fed after the CFF (and downstream of the ‘after
CFF’ samples) . It can be seen clearly that significantly improved levels of cleanliness
have been achieved using a variety of grades of SIVEX CFF despite the presence of
grain refiner ahead of the CFF.
Inclusion Counts (mm2/kg)
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
Customer D
Customer C
Customer B
Customer A
a
0.1
0
before after before after before after before after before after before after
15 inch Grade 40
with Grain Refiner
5%Ti:1%B
1xxx alloy
17 inch Grade 30
with Grain Refiner
3%Ti:1%B
6xxx alloy
23 inch Grade 40
with Grain Refiner
5%Ti:1%B
6xxx alloy
23 inch Grade 40
with Grain Refiner
5%Ti:1%B
3xxx alloy
23 inch Grade 65
no Grain Refiner
3xxx alloy
23 inch Grade 80
no Grain Refiner
3xxx alloy
16 samples
2 casts
18 samples
3 casts
56 samples
10 casts
20 samples
4 casts
12 samples
2 casts
14 samples
2 casts
Figure 11 : Cleanliness data for production trials using a variety of SIVEX CFF’s.

Conclusions
1. SIVEX ceramic foam filters have the capacity for high filtration efficiencies in
the absence of grain refiner, even under severe disturbance conditions and
with sustained high inclusion loading throughout the cast.
2. The variability of the measured filtration efficiencies reduces quite significantly
as we move to finer porosity SIVEX ceramic foam filter grades. In other
words: the reliability of the filtration system increases as we move to finer
pore filters. It was also found that the finer the filter, the greater its ability to
remove smaller sized inclusions.
3. The introduction of grain refiner upstream of the SIVEX CFF can lead to a
reduced filtration efficiency. This only appeared significant at high inclusion
loadings. The choice of grain refiner type can effect the magnitude of this
impact on CFF performance 4 . 5%Ti:1%B, 3%Ti:0.2%B and 1.2%Ti:0.4%B are
all reported to have a less severe impact on CFF performance than that
demonstrated here using a 3%Ti:1%B rod 4 .
4. Results of production trials in the presence of grain refiner agree well with the
findings of the filtration evaluation programme conducted jointly between
VAW, LSM and Pyrotek. These production trials also confirm that significantly
improved levels of cleanliness have been achieved using a variety of grades
of SIVEX CFF despite the presence of a grain refiner ahead of the CFF.
References
1 N.J.Keegan, W Schneider, H_P Krug, “Evaluation of the efficiency of
ceramic foam and bonded particle cartridge filter systems”, Light Metals
1997, pp 973-982
2 N.J.Keegan, W.Schneider, H.P.Krug, “Evaluation of the Efficiency of Fine
Pore Ceramic Foam Filters”, Light Metals 1999, pp 1031-1041.
3 S.F.Ray, N.J.Keegan, “Measurement of Cell and Window Size in Ceramic
Foam Filter Manufacture”, Light Metals 1998, pp 885-892
4 N.Towsey, W Schneider, H_P Krug, “A comprehensive study of ceramic
foam filtration”, 7 th Australian Asian Pacific Conference, Aluminium
Casthouse Technology, TMS (The Minerals, Metals and Materials
Society), 2001, pp125-137
5 N.Towsey, W Schneider, H_P Krug, A Hardman, N.J.Keegan, “Impact of
grain refiner addition on ceramic foam filter performance”, Continuous
Casting Conference, Wiley/DGM, 2000, pp26-32
6 N.Towsey, W Schneider, H_P Krug, A Hardman, N.J.Keegan, “The
influence of grain refiners on the efficiency of ceramic foam filters”, Light
Metals 2001, pp 973-977.