Sumitomo
Sumitomo
Osaka Works
Solid-Liquid Separation Kunitoshi T AKAHASHI
Process & Production Technology Center
Susumu S HIOJIRI
Many manufacturing processes for active pharmaceutical ingredients (APIs) and their manufacturing inter-
mediates consist of reaction, post-processing, crystallization, filtration and drying operations. Particularly dur-
ing filtration operations, we sometimes have operational troubles which can cause product quality deterioration
and a large decrease in productivity and work efficiency. Therefore, it is important to stabilize filtration opera-
tion during API manufacturing processes.
In this paper, we would like to introduce several scale-up techniques from a small experimental scale to a
commercial manufacturing scale by focusing mainly on operational condition setting of key parameters in cen-
trifugal filtration processes.
This paper is translated from R&D Report, “SUMITOMO KAGAKU”, vol. 2007-II.
rpm
1000 p
Start 3 Stop
800
600 2
1
400
200
0
Filtrate
Slurry
ed
D Ri g
uo -1
D Rin 1
D uor -2
uo -2
of ape ing
m
e
g-
g-
r m ak
rin
iq se
iq se
liq ing
iu
fe
rin
rin
ng cr ap
c
ed
n
uo
ry
d
hi ns cr
iq
ur
S
el
Cake
Sl
te
el
el
D
fil
u
W ing
ov
as
m
Filter medium
Re
Fig. 1 Diagram of filtration, washing and deliquor- Fig. 2 Cross section drawing of filtration system
ing in centrifuge
Increasing the Scale of Slurry Feed and Rins- ence in pressure p on the slurry side and the filter
ing Operations medium side. Centrifugal, pressure and reduced pres-
sure filtering just have different operations that provide
We will describe a technique for increasing the scale p, and can use the same filtering theory.
from small-scale experimental data to actual equipment If the scale for centrifugal filtering is increased keep-
using mean specific filtration resistance αav, and the ing the centrifugal effect Z ( = ro 2/g, where is the
compression index n, which are basic physical proper- angular velocity of rotation and g is gravitational accel-
ties of filtration. eration) using the basket radius ro, for the centrifugal
separator, p (= 2 (ro – rL), where is the filter
L L
1. Concept of increasing the scale density and rL is in the radius of the liquid membrane
We will explain the basic theory of filtering using during filtering) increases from small centrifugal sepa-
Fig. 2. The driving force for the filtrate passing rators (6 inch, 15 inch, etc.) to actual equipment (48
through the filter medium and the cake is the differ- inch, etc.) as is shown in Fig. 3, and if the cake has
10000
Table 2 Evaluation examples of average specific filtra-
tion resistance and compressibility of filtration
cake
1000
mean specific < 5 × 108 very fast filtration/solid sedimentation
400G
filtration resistance 5 × 108 ~ 2 × 1010 best filterability
Z [–]
αav [m/kg]
1.0E+11
rL n = 0.25
n = 0.50
n = 1.00
1.0E+10
n = 1.25
rc
H 1.0E+09
0 50 100 150 200 250 300 350
1.0E+11 n = 1.00
high compressibility (cake with a large n), it is com-
n = 1.25
pressed. Because of this there are problems such as
1.0E+10
reduction in filtration rate, reduction in rinsing efficien-
cy and poor detachment of the filter bed.
1.0E+09
Therefore, it is necessary to increase the scale of 0 20 40 60 80 100
shows a cross-section of a centrifugal separator for cal- Fig. 6 Average specific filtration resistance plot-
culating p. ted against cake thickness
gets thinner, and that the thickness becomes one Table 3 Comparison of constant pressure filtration
where it is difficult to scrape the cake with an upright test and compression permeability test
centrifugal separator. Thus, we can see that any filter-
constant pressure compression
ing using an upright centrifugal separator a criterion of filter test permeability test
αav ≤ 1010 m/kg is preferable for αav. purpose of αav αp
measurement n n
accuracy of results fair very good
4. Measurement of mean specific filtration resis-
cost low high
tance equipment size small large
The constant pressure filtration test equipment and handling very easy easy
the compression permeability test equipment shown in measurement time short long
h0
Filter paper Compression permeability test
1.E+11
αav [m/kg]
1.E+12
Intermediate B er (compression becomes easier) if αav increases (fil-
tering gets poorer).
αav [m/kg]
p [kPa]
Increasing the Scale of Deliquoring Operations2)
Fig. 10 Average specific filtration resistance meas-
ured in constant pressure filtration test
To estimate the deliquoring time using a centrifugal
and calculated from results of compres-
sion permeability test (intermediate B) separator, the average liquid content of the cake after
deliquoring has conventionally been found using a
small centrifugal separator (6 inch) and a mid-sized
The measurement results for the constant pressure centrifugal separator (15 inch diameter, etc.). As a
filter test equipment and the measurement results for result of examining methods for testing with a smaller
the compression permeability test equipment are com- sample, it has become possible to find the liquid con-
pared in Figs. 9 and 10. tent simply using a table tops interviews (Fig. 13).
1.E+11
αav [m/kg]
1.E+10
1.E+09
1.E+08
Fig. 13 Table-top centrifuge
1.E+07
0 0.2 0.4 0.6 0.8 1
n
We will compare the measurement results for aver-
Fig. 11 Average specific filtration resistance meas-
age liquid content with the small and mid-sized cen-
ured in constant pressure filtration test plot-
ted against compressibility index of cake trifugal separators and the tabletop centrifuge and
introduce a test method for the tabletop centrifuge.
1.E+11
1. Average liquid content, average centrifugal
1.E+10
αav [m/kg]
Mother liquor
100%
Wet cake
50%
0%
0 2000 4000 6000 Electric weight
Z [–]
Centrifugal sedimentation tube
Calculation of cake liquid content
Fig. 14 Final liquid content of cake in centrifugal
deliquoring plotted against centrifugal Fig. 16 Procedure of deliqoring experiment by ta-
effect 2) ble-top centrifuge
Dry weight-based final liquid content of cake
200%
Experimetal centrifuge 6inch 3. Cake compressibility and average liquid content
Experimetal centrifuge 15inch Fig. 17 shows the relationship between uncom-
150% pressed cake and p’, and Fig. 18 shows the relation-
ship between the average liquid content of cake with
100%
compressibility and p’. The average liquid content for
the cake with compressibility and the uncompressed
cake agreed for the tabletop centrifuge and a small cen-
50%
trifugal separator used in the same manner. This
shows that measurement of the average liquid content
0%
0 200 400 600 800 1000
that has been carried out in small and mid-sized cen-
trifugal separators can be carried out using a tabletop
p’ [kPa]
centrifuge.
Fig. 15 Final liquid content of cake of low compres-
sibility in centrifugal deliquoring by experi-
mental centrifuges of different sizes plot-
Dry weight-based final liquid content of cake
using the tabletop centrifuge. Fig. 17 Final liquid content of cake of low compres-
Paper is put into a filtering tube, and the wet cake is sibility in centrifugal deliquoring plotted
inserted. Next, the filter tube is put into the centrifugal against apparent centrifugal deliqoring
pressure; comparison of results of experi-
sedimentation tube, and set into the tabletop cen-
mental centrifuges (open symbols) and
trifuge. The speed is set for a p’ close to the operating those of table-top centrifuge (closed sym-
conditions for the increased scale, and the deliquoring bols)2)
60%
Symbols Used
40%
A : filter area [m2]
g : gravitational acceleration [m/s 2]
20%
H : centrifugal separator basket height [m]
0% n : compression index [–]
0 200 400 600 800 1000
p : filtration pressure [Pa]
p’ [kPa]
p’ : apparent centrifugal deliquoring pressure [Pa]
Fig. 18 Final liquid content of cake of high com- ro : outside radius of the centrifugal separator bas-
pressibility in centrifugal deliquoring plot- ket [m]
ted against apparent centrifugal deliqoring
rc : radius of cake inside surface [m]
pressure; comparison of results of experi-
mental centrifuge (open symbols) and rL : radius of liquid membrane during filtering [m]
those of table-top centrifuge (closed sym- V : filtrate volume [m3]
bols)2) W : cake mass [kg]
Z : centrifugal effect [–]
αav : mean specific filtration resistance [m/kg]
αp : partial specific filtration resistance [m/kg]
Conclusion α 0 : αav = α 0 ( p)n [m/kg]
: filtration time [s]
In this paper, we have introduced the fact that it is L : filtrate density [kg/m3]
possible to estimate the filtering time when increasing L : filtrate viscosity [Pa · s]
the scale in slurry feed and rinsing operations by : angular velocity of rotation of centrifugal sepa-
obtaining αav and n using constant pressure filtration ration [rad/s]
test equipment as the technique for scaling up to pro-
duction from small-scale experiments and that it is References
possible to estimate the deliquoring time and average
liquid content by obtaining deliquoring data using a 1) K. Takahashi, Centrifugal Solid-Liquid Separation,
tabletop centrifuge. By making it possible to obtain the The Society of Chemical Engineers, Kansai
data for increasing the scale using small-scale experi- Branch, Frontiers of Separation Technology in API
ments, it is possible to reduce the centrifugal separa- Manufacturing (2004. 11)
tor tests at the middle scale for obtaining data for 2) S. Shiojiri, T. Okahashi, K. Takahashi, M. Iwata,
increasing the scale in solid-liquid separation opera- Scale-up Law in Spin Drying Operation of Centrifu-
tions using centrifugal separators and to increase effi- gal Filter, The Filtration Society Japan Association
ciency of testing. Filtration & Separation Symposium (2001. 11)
PROFILE