SUGAR MILLING

G RESEARCH INSTITUTE
UNIVERSITY OF NATAL
KiNG GEORGE V AVENUE/FRANCOIS ROAD
WAD P.O. BOX 1525 - DURBAN - T E L E P H O N E 359192

TECHNICAL REPORT NO. 9 ^ 9

SUBJECT: OPERATION OF THIS FILTER-STATION AT UMZIMKULU MILL

PREPARED BY: MR. J.R. FITZGERALD & MR. J.P. LAMUSSE

PREPARED FOR: UMZIMKULU SUGAR COMPANY LIMITED.

DATE: 3 r d December 1 9 7 1 .

A request was received to investigate the general operating

conditions of the Filter-Station at Umzimkulu Mill and its possible
influence on the incidences of poor clarification. In general, it
was found that this station operates adequately and it is felt that
only under extremely adverse conditions, corresponding to an accumulation
of the various factors affecting filter performance, could the filtrate
return cause clarification malfunction. An investigation of the
various factors affecting filter performance was made and in some
cases it was found that more attention could be paid to these
parameters.

Description and Operation of the Filter Station

The filter station consists of three rotary vacuum filters:

1. 2 , 4 m dia x 3 , 7 m. Door Oliver filter.

2. 2 A m dia x 3 , 7 m. Standard Eimco filter.
3. 2 A m'di.i x 3 , 7 m. Converted Eimco belt filter.
2
giving a total area of 1 1 1 , 6 m for a crushing rate of approximately
160 m tons/hr. This ratio of screen area to crushing rate is satis­
factory.
The clarifier mud is pumped to a central head tank where a small
amount of flocculent may be added to improve filterability. Facilities
for the addition of dilution water to the mud before the mud pumps are
available should difficulty be experienced in pumping. The feed from
the head tank is split into 2 bagacillo mixing tanks before being fed
to the filters. The correct proportion of bagacillo to mud in each
tank is important in order to prevent the starvation 6 f filter-aid to
any one filter.

Overflow weirs are available on each filter trough to ensure the

correct pick-up level in the troughs, but it is general practice
at Umzimkulu not to utilize them due to the apparent danger' of
fermentation within the recyle loop. The recycling of filtrate is
similarly not utilized to accommodate fluctuations in feed-rate.
In view of this, special attention must be paid by the operators
to the levels in the troughs and the practice of running two troughs

2/...
 -2-
TECHNICAL REPORT NO. 9 ^ 9 3 r d December 1 9 7 1 .

full and allowing the third trough to fluctuate according to demand

should be avoided. The use of surface speed to control the levels
should be similarly avoided as an increase in speed will lead to a
thinner mat and a corresponding increase in the volume of cloudy
filtrate returned. In the majority of cases, it was found that the
filters were running at speeds in excess of the recommended 12 R.P.H,

The general operating conditions of the station appeared to be

sound; filter feed and wash-water temperatures were recorded at
An
order of 6 0 °C. The addition of wash-water however could be
v
improved. Four banks of water sprays are positioned within a 6 0 to
9 0 section of the circumference before the vertical so that the
cake is dry approximately 1 metre above the discharge. The close
proximity of the sprays tends to saturate the cake in this region
and the excess water runs back off the filter eroding the matt
formed. It is recommended that these sprays be repositioned over
a wider segment so as to obtain a better utilization of the water
and the filter area and that water be more evenly distributed among
the sprays.

A low pick-up vacuum of approximately 14" Hg and high vacuum

of 1 7 " Hg appear to be in use but their absolute values are in doubt
due to the questionable reliability of the gauges. The low vacuum
was however observed to fluctuate excessively due probably to changes
in the mud level in the troughs. Adequate vacuum capacity is
available from two high vacuum pumps, the low vacuum being obtained
by means of a manually operated air bleed into the smaller pump.
In place of the standard vacuum breakers on the filtrate separators,
a save-all is used to stop entrainment of filtrate. The save-all
tail-pipe seal overflow is returned to the overflow mud tank. This
is a potential source of fermentation and should be avoided at all
costs.

The removal of this source of fermentation to a separate drain

is of upmost importance and the complete re-engineering of the vacuum
system is recommended. The use of float-controlled vacuum breaker
valves and of an automatic valve to regulate low vacuum are also
recommended.

The pol in filter cake averaged lAk for the two months prior
to the investigation. This is above the industrial average of 1 , 3 3
and well above results obtained in some other mills. Since the
initiation of this investigation, the average has fallen to 1 , 3 2
while the implication of the recommendations made in this report
should enable filter losses to be reduced even further. Unfortunately,
no record of the weight of filter cake is available, therefore
the true magnitude of the loss can only be estimated.

3/...
TECHNICAL REPORT NO. 9 4 9 -3- 3 r d December 1 9 7 1 -

Analysis of Filter-Station Operating Parameters

The following analayses were obtained from the mill.

TABLE 1

17/11/71 18/11/71 19/11/71 22/11/71 24/11/7™

CLARIFIES MUD

% Solids V/V 41,0 38,0 35,0 35,0 41,0

% Dry Solids. WT/WT. 6,24 7,09 7,89 8,6 11,37
Brix supernatent 12,76 12,80 13,10 12,60 12,30
pH 7,00 7,25 7,25 7,20 7,20
CLOUDY FILTRATE
% Solids V/V. 5,0 6,0 5,0 6,0 7,0
% Dry Solids WT/WT. 0,28 0,80 0,85 0,87 0,94
Brix supernatent 8,56 10,20 8,40 8,94 7,97
pH 6,70 6,85 6,65 6,65 6,65
CLEAR FILTRATE

% Solids V/V 3*0 3,0 6,0

% Dry Solids See Note 0,39 0,39 0,85 0,76
Brix supernatent 10,10 8,50 8,94 7,97
pH 6,80 6,70 6,80 6,85
RETENTION %
CLOUDY FILTRATE V/V 81,82 80,19 77,72 75,84 73,65
93,31 85,84 83,20 85,74 87,24
CLEAR FILTRATE V/V See Note 89,99 86,79 75,84 77,79
W/W 93,03 92,38 86,07 89,68

Bagacillo Filter Feed 2,75 1 5K x n —, i ^

Bagacillo Ratio 71,1 39,9 42,6 36,0 15,7

Notf. Cloudy & clear filtrates mixed.

pH of wash-water - 7 , 4

(a) Bagacillo and Bagacillo Ratio

A dry seive analysis of a snatch sample yielded the following

size distribution for bagacillo;

Tyler Mesh Sjg» Ave Particle Diam (mm) Wt. Fraction %

±2 > 2.79 0,1

- 7 + 10 2,22 2,1
- 1 0 + 14 .1,41 7,2
-14 + 20 1,00 18,1
- 20 < 0,83 72,5
4/...
TECHNICAL REPORT NO. 9 ^ 9 3 r d December 1 9 7 1 -

This distribution is adequate and the high percentage of fines

should be conducive to good retention values.

The bagacillo ratio - bagacillo % dry solids in £lexifier mud

was found to vary from between 15 and 7 0 %. while it is generally
agreed that this ratio should be between 5 0 - 6 0 %, the fineness of
bagacillo and the relatively high dry solids concentration in the
clarifier mud should allow adequate performance at slightly lower
values of this ratio. However, neslec-ting; the sinsle value of
7 1 %, the average value is on the low side i.e. 3 4 % and should be
increased.

From a consideration of the size distribution, it would appear

that the amount of bagacillo may be effectively increased by both
increasing the screen mesh size as well as by increasing the screen
area. The recommended screen area per 1 0 0 m. ton of cane crushed
per hour is 4 , 1 m .
2

(t>) Retention

% Retention = 1 0 0 - Mud solids % Filtrate x Brix feed x 1 0 0

Mud solids % Feed Brix filtrate

As both cloudy and clear filtrate are returned independently to

the process, it is not possible to quote an overall retention for the
filter station without previous knowledge of their respective flows.
For this reason, retention has been quoted for both cloudy and clear
filtrate c.f. Table 1 .

Further, it is the practise at Umzimkulu to determine the retention

based on volumetric determinations of % solids rather than by weight
and similar values are quoted for comparison purposes. While the use
of the volumetric readings are slightly in error, it was noted that
no allowance has been made for the addition of wash-water in the
factory determination of this ratio, i.e. the brix of the supernatent
is not used to adjust the results. From the results obtained
however, retention based on volume and corrected for wash-water appears
to be about 5 - 1 0 % lower than the true value. This would tend to
indicate that 'volume' retention may be used as a rough and quick
guide to filter performance.

An analysis of the retention figures obtained, tends to indicate

good performance on the part of the filter-station although a
knowledge of the magnitude of the respective filtrate return flows
is required to confirm this conclusion. The dependence of retention
on bagacillo ratio is evident.

The pH drop from clarifier mud to filtrate return appears to be

slightly higher than values recorded at some other mills. Further
work is being done with regards to this aspect as it may be indicative
of fermentation.

5/...
 -5-
TECHNICAL REPORT NO. 9 ^ 9 3 r d December 1 9 7 1 .

Recommendations

In the light of this investigation, the following recommendation

are made:

1. The vacuum system should be completely redesigned. As a first

step, the tail-pipe overflow from the filtrate save-all must be
removed to a separate drain and an automatic control of low
vacuum be fitted. Low vacuum should be set at about 7" Hg.

2. The wash-water sprays should be repositioned to make fuller use

of the filter area available.

3. Ways of decreasing the fluctuations in feed level in the filter

troughs and decreasing the filter surface speed should be
investigated.

k. The bagacillo ratio should be increased by at least 1 0 to 2 0 %

5. Routine measurements of moisture content of filter cake should

be made.