Vous êtes sur la page 1sur 19

See discussions, stats, and author profiles for this publication at: https://www.researchgate.


The Prediction of Filter Belt Press Dewatering Efficiency for Activated Sludge
By Experimentation on Filtration Compression Cells

Article  in  Environmental Technology · January 2005

DOI: 10.1080/09593332508618474 · Source: PubMed


13 2,386

2 authors, including:

Jérémy Olivier
Université de Pau et des Pays de l'Adour


Some of the authors of this publication are also working on these related projects:

Limit dryness View project

Gravity drainage index View project

All content following this page was uploaded by Jérémy Olivier on 31 March 2016.

The user has requested enhancement of the downloaded file.



1- Laboratoire de Thermique, Energétique et Procédés, ENSGTI

Rue Jules Ferry, BP 7511, 64075 Pau cedex, France
(*tel : 33 (0)5 59 40 78 11, fax : 33 (0)5 59 40 78 01,  : jean.vaxelaire@univ-pau.fr)

2- E.M.O. s.a.
40 rue du Bignon, Immeuble le "Ponant ", B.P. 17, 35574 Chantepie Cedex France

The filter belt press is commonly used to dewater activated sludge. However, little research has
been done on this process and the prediction of its efficiency. Experimentation have been
carried out in a filtration compression cell (FCC) and in a pilot scale filter belt press. It offers a
way of determining filter belt press efficiency thanks to simple laboratory research. The
pressure distribution around the pressing roller was measured inside the pilot scale filter belt
press. It showed progressive increase (up to a certain maximum value : plateau), which was
followed by a rapid decrease. The impact of the progressive increase of applied pressure onto
the dry solid sludge content was observed in FCC. Similar dry solid contents were obtained
from both the above laboratory devices when the application of the pressure is comparable (in
time and increasing rate).

Keywords: Wastewater treatment, filter belt press, dewatering, sewage sludge.


The enforcement of current European regulation on wastewater treatment has led to a

significant increase in the amount of municipal sludge being produced. Consequently,
management and treatment of these sludges have became more important. Currently, two
major methods are available for sludge disposal; incineration and landspreading. For both, a
preliminary step of dewatering is usually required. Due to their relatively low energy cost,
mechanical devices, such as filtration or centrifugation, are currently used to achieve this
dewatering. Amongst the common devices, filter belt presses are widely used in municipal
wastewater treatment plants.
In spite of its widespread usage, the design of this type of press remains essentially empirical
and the operating parameters are generally fixed according to field testing [1,2]. However tests
on field industrial systems are not always easy to set up and some alternative methods of
characterisation may be preferred. Only a few published studies specifically deal with the
prediction of the efficiency of the belt filter press. Some authors [3] have established, from a
study carried out on a laboratory scale pilot belt press (containing 8 rollers), different equations
for predicting the efficiency of the belt filter press. These equations allow one to estimate the
solid content of the final cake, the total filtrate flow rate, the cake’s width and the solids’
recovery efficiency according to the input sludge’s flow rate, the initial solid content of sludge
and belt speed (Table 1).
Two years later, from a compilation of performance data obtained from a survey of over 100
American installations, some general equations to predict dry solid content in pressed cakes, as
compared to dry solid content when entering the device, were proposed [4] (Table2).

These different equations do not correlate well with the majority of the experimental data from
current literature. Consequently the interest of such global relationships remains very limited.
To establish more practical equations, filter belt press efficiency must be evaluated with regard
to some measurable physicochemical characteristics of sludges. In this way, a study [5] tried to
define the major factors (among 30) that affect the dewatering characteristics of sewage sludge
in belt filter presses. They showed that only the viscosity of raw sludge could be correlated to
the moisture content of dewatered cake for mixed and anaerobically digested sludges. The
viscosity was affected by the composition in terms of colloids, fibres, volatile suspended solids
and ash.

Due to the current problem of anticipating the behaviour of such an elusive material to study, it
seems useful to develop some laboratory tests capable of easily predicting the efficiency of the
industrial devices. Several techniques, classically used to characterise solid-liquid separation,
have been tested; such as gravity drainage methods [6,7], capillary suction time measurement
(CST) [6,8,9], measurement of the specific resistance of filtration (SRF) [6,9,10] in filtration
compression cells (FCC) or on funnel-type vacuum units [8], laboratory centrifugation [5,11], or
the measurement of infinite cake solid content [12].
The inability of all these tests to accurately predict the efficiency of industrial belt filters are due
to several aspects, such as differences in dewatering times [6,13] and in the value of applied
pressures. To progress on these aspects other methods of investigation were proposed. A
classical FCC was modified with a rotary piston to take into account the shearing observed in
belt filters [10]. Nevertheless, the results obtained from this modified device cannot be directly
applied to the design of industrial devices. Other authors [14, 15] have used a permeable piston
and consequently modified the FCC to a double-sided piston filter press. This modification
enables better prediction of industrial effectiveness, but does not consider the pressure
distribution around the roller. Some specific laboratory devices dedicated to the study of the
belt filter press have also been developed. A device constituted of a belt (length = 1.8 m, width =
0.2 m) tensed by two rollers (diameter = 0.25 m) was used [16]. The belt speed (1-17 m.min-1)
was set by an electric motor and a pneumatic jacket allowed to fix the belt tension. For
experimentation, a sample of drained sludge was laid on the belt and trapped with a piece of
filter cloth (length = 0.6 m) which was on the sludge and clipped at the ends. This stage of
installation of the sludge is delicate and can have some effects on the experimental results,
which could explain why this apparatus is used rarely. Another laboratory device called the
"Crown Press " was also developed [17]. It is a PVC pipe cut in half lengthwise (the crown) with
slots to allow filtrate drainage [18-21]. A curved surface was used to better simulate the
pressure distribution around pressing rollers. A piece of filter cloth was fixed on the crown and
a sample of sludge was laid on it. Then, another piece of cloth was fixed on the crown and put
over the sludge to squeeze it. The bottom of this second piece of cloth was tensed by a hook
attached to a pinion system. This specific device has given some interesting results but it is not
often used. Equipment suppliers generally prefer simpler and more standard tests such as FCC.

The aim of this paper is to propose a procedure to carry out FCC tests according to belt filter
specificity. This work should improve and facilitate the design of industrial belt filter presses
for biological sludge dewatering.

Experiments were carried out on activated sludge with two different devices: a pilot belt filter
press and a filtration compression cell (FCC).

The laboratory scale belt filter press

The pilot device was a small scale belt press (0.5 m * 2 m * 1.5 m). This is where the sludge
dewatering was studied, only around one pressing roller, the roller 3 (Fig 1). Different
diameters of the pressing roller can be adapted to this device, the one chosen for the present
work was of 0.27 m. All the other rollers were necessary to ensure the position, the tension and
movement of the two belts. More specifically, the rollers 1 and 11 are fixed to pneumatic jacks to
set the belt tension. An electric motor equipped with a regulator allowed us to set the belt
speed. The roller 9 is covered by a rubber liner and transmits the movement. The direction of
the moving belt can be reversed to study dewatering under several series of pressing cycles in
order to come closer to industrial devices (which operate with a series of pressing rollers). The
belt was a 16-6 herringbone twilled wave belt manufactured by Rai-Tillères.
The filtrate was collected and weighed for every pressing cycle around the designated roller.
The dewatered cake was removed and weighed at the end of the experiment. The dry solid
content in the final cake was measured after being dried at 105°C. It was also calculated after
each cycle (across the pressing roller), from a mass balance based on the mass of collected
Four replicated tests were carried out to evaluate the reliability of the experiments in terms of

dry solid content in the final cake. The relative error ( Cmax Cmin ) was estimated to be under 3%.
Moreover, some tests were carried out to compare dewatering efficiency between the laboratory
device and an industrial belt filter press operating in the Pau-Lescar municipal wastewater
treatment plant (France). These experiments were performed on the same sludge in similar
operating conditions, in terms of belt tension, belt speed, sludge loading, and the number of
cycles (across pressing rollers). The results of this comparison have shown accurate agreement
between the laboratory device and the industrial one [22].

The filtration compression cell (FCC)

The filtration compression cell was built according to French standards (the AFNOR T 97-001).
It is a 0.15 m deep cylindrical stainless steel chamber with an internal diameter of 0.07 m. A
perforated disk was located at the bottom of the cylinder in order to support the filter medium.
The filter cloth was similar to the belt used on the pilot device. The pressure on the piston was
applied and controlled by pressurised air. The pressure range was 0.5 to 15 bar. The mass of
collected filtrate was recorded over time on a personal computer using a software program
developed in the laboratory.
The reliability of the experiment was evaluated on four replicated tests. The curves describing
the kinetic of dewatering were accurate and the relative error of the dry solid content in the
dewatered cake was estimated to be under 6% [22].

The sludge
The activated sludge extracted from the thickener at the Pau-Idron municipal wastewater
treatment plant (France). It was conditioned by a high molecular weight and high charge
density (80%) cationic polymer (SNF Floerger, France, ref. EM 840 TRM). The conditioner dose
was fixed from preliminary experiments to 7 g kgDS-1.
The polymer solution (2.15 g l-1) was added with a syringe while the sludge was gently shaken
in a stirred vessel for 30 s.
In order to simulate industrial practices the sludge was drained during five minutes on a
gravity drainage system before the dewatering experiments began (on the pilot device and in


Pressure measurements
To evaluate belt filter press effectiveness it is important to know the operating parameters,
especially the pressure applied around the pressing rollers. Usually this pressure is determined
by the following equation [19,23] (not including rolling friction and belt elasticity) :

Proller  2Tbelt (i)

According to this equation, the pressure is constant around the roller. However, it was shown
(from direct measurements of pressure in a belt filter press) that equation (i) considerably
under-predicts the true applied pressure [24]. Besides, the pressure was not constant around the
roller. To clarify this aspect, we have set up some direct pressure measurements in the pilot
filter belt press. A miniature pressure transducer (Kyowa, Japan, ref. PS-5KA) was placed
between the two belts and carried around the pressing roller while in rotation. These
measurements have shown, similar to the results of the literature [24], that the pressure was not
constant, with a progressive increase, a plateau and steep decrease (Fig 2). The shape of the
pressure distribution around the roller is not really affected by the tension applied on the belts
(Fig. 2a) but depends on the roller diameter (Fig. 2b). Moreover, the maximal pressure reached
(Pmax) was much higher than the one calculated from equation (i). In this study (0.3 m belt width
and a 0.27 m roller diameter) it can be calculated from the following empirical equation :
Proller  1.06 10 3 Tbelt  0.6186 (ii)

with Tbelt in N and Proller in bar.

It is this "true" pressure which should be used for tests in FCC.

The impact of the applied pressure

Another important aspect to specify, was the influence of the applied pressure on sludge
dewatering by compression. On this particular point, the data reported in current literature
generally concerns initial liquid sludge (not drained) and leads to quite divergent results.
Indeed, for undrained sludge, few authors [25,26] have observed during the filtration step, the
significant influence of the applied pressure, whereas others [27,28] have shown that for a
highly compressible material such as activated sludge, pressure has no effect on filtrate flow
rate. During the compression phase (a phase particularly important in the belt filter press
process), it was shown [12,29] that a variation of pressure (after several minutes of compression)
had no effect, either on the filtrate flow rate, or on the dry solid content in the final cake.
Due to this lack of agreement and of data on drained sludge, some experiments were carried
out on FCC at different pressures. The results, obtained when the whole pressure was applied
instantaneously, do not show a significant impact of the pressure on the dry solid content of the
sludge (the observed variations are within the limits of accurate reliability). However, when the
pressure was increased progressively over the first two minutes of the experiment, a significant
evolution in the flow rate of the filtrate is observed (Fig 3). The dry solid content in the
dewatered cake is also very different under these operating conditions. When the increase of
pressure was delayed after a time lapse of two minutes, no impact was observed in the
dewatering. This is similar to the results reported in literature [12,29].
The impact of the pressure in the initial phase of compression is an important aspect which
should be considered to analyse belt filter presses because in these devices the sludge is
progressively pressurised when it reaches the first roller.

Laboratory procedure to predict industrial efficiency

To predict filter belt press dewatering efficiency, it seems interesting, according to the previous
results of this work, to carry out some tests in FCC with a progressive pressurisation. Thus,
comparative experiments were performed simultaneously in both devices: the pilot belt filter
press and the FCC, on identical sludge. For these experiments the quantity of sludge
introduced into the FCC was adjusted to obtain a thickness of the final cake close to 1 cm. This
value was approximately the thickness of the dewatered cake on the belt filter press. The
pressure was progressively increased in the FCC according to pressure measurements carried
out around pressing rollers, and finally fixed to a constant value which corresponded to P max
(Fig 4). The time of pressurisation into the FCC was set according to the number of pressing
cycles and the belt speed in the filter. This procedure correlates with our previous results which
have shown that after an initial period of a few minutes, the increase of pressure did not affect
the compression anymore.
However, the time pressurisation modified significantly the dry solid content of the final cake.
To take into account this aspect, the duration of pressurisation in the FCC was equal to the real
time of pressing in the belt filter press. This time was calculated by :

tn (iii)

One example of these comparative tests is reported on figure 5. It shows good agreement
between both experiments with a maximal deviation of 1%. The result of one complementary
experiment, carried out in FCC at a constant pressure (Pmax), indicates the importance of the
procedure of progressive pressurisation (Fig 5). The data presented on figure 5 also indicates
that the number of pressing rollers seems to have an impact only on the pressing time. Other
similar tests performed under different operating conditions led to the same results and
confirmed the interest of measurements in FCC to predict filter belt press dewatering efficiency.
This work has contributed to the application of a classical laboratory test to the design of filter
belt presses. It has presented a procedure to predict, from experiments carried out in the
filtration compression cell (FCC), the effectiveness of industrial devices. This study has shown
the necessity to be very aware of the distribution of pressure around the pressing rollers,
especially the first one. The addition of other pressing rollers essentially contributes to increase
the time of pressing. When these two aspects (progressive pressurisation and time of pressing)
are correctly assessed for experiments in FCC, the dry solid content of the sludge, dewatered by
the filter belt press, can be accurately estimated.

This study was carried out on behalf of E.M.O. Ltd.

C : Final dry solid content of filter belt press (gDS gcake-1)
Cmax : Maximal dry solid content measured during experiments of reliability (gDS gcake-1)
Cmin : Minimal dry solid content measured during experiments of reliability (gDS gcake-1)
Caverage : Average dry solid content obtained during experiments of reliability (gDS gcake-1)
C0 : Initial dry solid content of sludge (gDS gcake-1)
C’0 : Initial dry solid concentration of sludge (kgDS m-3)
Droller : Diameter of belt filter press roller (m)
E : Solid recovery efficiency (-)
l : Cake width (m)
lbelt: : Belt width (m)
Lc : Length of the pressing zone around a given roller (m)
n : Number of pressing rollers
Proller : Pressure drop around the cake around belt filter press rollers (Pa)
Q : Filtrate flow rate (m3 h-1)
Q0 : Input sludge flow rate (m3 h-1)
s : Belt speed (m s-1)
Tbelt : Belt tension (N)
X : Position in the pressing zone around a roller (m)

1. Haworth P.J. and Roberts K., Development and application of filterbelt presses. Filt. Sep.,
12, 572-576 (1975).
2. Johnson G., Buchanan G.G. and Newkirk D.D., Optimizing belt filter press dewatering at
the Skiner filtration plant. J. Am. Water Works Association, 84, 47-52 (1992).
3. Lotito V., Spinoza L. and Santori M., Analysis of sewage sludge dewatering operation by
beltpress. In: Proc. of the 4th World Filtr. Congress, Ostende, Belgium, organised by the
Royal Flemish Society of Engineers, 4.43-4.47 (1986).
4. ASCE (Task Committee on Belt Filter Presses), Belt filter press dewatering of wastewater
sludge. J. Environ. Eng., 114, 991-1006 (1988).
5. Hashimoto M. and Hiraoka M., Characteristics of sewage sludge affecting dewatering by
belt press filter. Water Sci. Tech., 22,143-152 (1990).
6. Baskerville R.C., Bruce A.M. and Day M.C., Laboratory techniques for predicting and
evaluating the performance of a filterbelt press. Filt. Sep., 15, 445-454 (1978).
7. Poduska R.A. and Collins B.H., A simple laboratory technique for determining belt-filter
press operation. Environ. Tech. Letters, 1, 547-556 (1980).
8. Karr P.R. and Keinath T.M., Influence of particle size on sludge dewatering. J. Ferm.
Bioeng., 50, 1911-1930 (1978).
9. Pan J.R., Huang C., Cherng M., Li K-C. and Lin C-F., Correlation between dewatering
index and dewatering efficiency of three mechanical dewatering devices. Adv. in Environ.
Res., 7, 599-602 (2003).
10. Halde R.E., Filterbelt pressing of sludge – a laboratory simulation. J. Water Pollut. Control
Fed., 52, 310-316 (1980).
11. Bullard C.M. and Barber J., A factor in belt filter press performance. Water Environ. Tech., 8,
67-70 (1996).
12. Gazbar S., Evaluation et amélioration des performances des procédés de déshydratation
mécanique des boues résiduaires. Thèse, Institut Polytechnique de Lorraine, Nancy, France
13. Rhemat T. Branion R., Duff S. and Groves M., A laboratory sludge press for characterizing
sludge dewatering. Water Sci. Tech., 35, 189-196 (1997).
14. Novak J., Knocke W., Burgos W. and Schuler P., Predicting the dewatering performance of
belt filter presses. Water Sci. Tech., 28, 11-19 (1993).
15. Miller S.A., Development of industrially relevant test methods for belt-press filtration. In:
Proc. of the 9th World Filtr. Congress, New Orleans, Louisiana, organised by the American
Filtration and Separation Society, 224-238 (2004).
16. Reitz D.D., Municipal sludge dewatering using a belt filter press. M.Sc. Thesis, Virginia
Polytechnic Institute and State University, Blacksburg, Department of Environmental
Engineering (1988).
17. Severin B.F. and Collins B.H., Advances in predicting belt press performance from lab
data. In: Proc. of the Water Environ. Fed., 65th Annual Conf. and Expo., New Orleans,
Louisiana, (1992).
18. Galla C.A., Freedman D.L., Severin B.F. and Kim B.J., Laboratory prediction of belt filter
press dewatering dynamics. In: Proc. of the Water Environ. Fed., 69th Annual Conf. and
Expo., Dallas, Texas, (1996).
19. Emery B.P., Predicting belt filter press performance using laboratory techniques. M.Sc.
Thesis, University of Illinois, Urbana, Department of Civil/ Environmental Engineering
20. Galla C.A., Laboratory prediction of belt filter press dewatering dynamics. M.Sc. Thesis,
University of Illinois, Urbana, Department of Civil/ Environmental Engineering (1996).
21. Galla C.A., Freedman D.L., Severin B.F. and Kim B.J., Pressing solids. Water Environ. Lab.
Solut., 4, 8-10 (1997).
22. Olivier J., Etudes des filtres à bandes pour la déshydratation mécanique des boues
résiduaires urbaines. PhD Thesis, Université de Pau et des Pays de l’Adour, France (2003).
23. Tokunaga K., Fujinami S., Ishimi T., Okahashi H. and Nakano I., High pressure filtration
and/or squeezing of sewage sludge. Filt. Sep., 20, 450-456 (1983).
24. Badgujar M.N. and Chiang S.-H., An analysis of belt filter press dewatering mechanism.
Filt. Sep, 26, 364-367 (1989).
25. La Heij E.J., An analysis of sludge filtration and expression. PhDThesis, Technische
Universiteit Eindhoven, Netherland (1994).
26. Léonard A., Etude du séchage convectif des boues de stations d’épurations – suivi de la
texture par micrographie à rayon X. PhD Thesis, Université de Liège, Belgique (2002).
27. Novak J.T., Agerbaek M.L., Sorensen B.L. and Hansen J.Aa., Conditioning, filtering and
expressing waste activated sludge. Water Res., 34, 1-20 (1999).
28. Lee D.J., Ju S.P., Kwon J.H. and Tiller F.M., Filtration of highly compactible filter cake:
variable internal flow rate. AIChE J., 46, 110-118 (2000).
29. Baudez J-C., Rhéologie et physico-chimie des boues résiduaires pâteuses pour l’étude du
stockage et de l’épandage. PhD Thesis, Ecole Nationale du Génie Rural, des Eaux et Fôrets,
Centre de Paris, France (2001).
Table1. Estimation of belt filter press efficiency on municipal sludges by Lotito et al. [3].
(for 1 < Q0 < 3 m3 h-1, 3 < sb < 12 m min-1 and 13 < C’0 >34 kg m-3)

Parameter Equation Correlation coefficient

Cake width l 0.706 0.381 0.912
 88.945Q0 s

Filtrate flowrate Q  0.007Q0

1.135 0.108
s C' 0
0.646 0.975

Solid content C  9.241Q0

 s 0.006  C' 0
0.279 0.880

Solid recovery efficiency E  788.61Q0

 s 0.298  C' 0
0.414 0.915
Table 2. Cake solid content versus solid content at the filter entrance (ASCE [4]).

Primary Secondary Equation Number Correlation Standard

sludge sludge (%) of points coefficients (R) error

0 - 10 90 – 100 0.044  0.0426 C0 30 0.89 4
10 – 40 60 - 90 0.0297  0.0402 C0 7 0.90 3

40 - 60 40 – 60 0.059  0.0307 C0 35 0.84 3
60 – 80 20 - 40 0.062  0.0306 C0 17 0.82 3
80 - 100 0 - 20 0.071  0.0266 C0 12 0.87 3
Figure 1. Laboratory filter belt press.

Pneumatic jack
driven by

the electric 9

Gravity drainage zone 8


Wedge zone

1 2 4 5

Filtrate tank D Belt

High pressure zone 3 R

(pressing roller)

I 6

Filtrate tanks A

Figure 2. Applied pressure versus a dimensionless length: ratio of any length over the whole
length (in the roller pressing zone). 2a. Effect of belt tension on pressure distribution around a
( : 206 N, : 382 N, : 971 N, :1383 N, : 2354 N, : 2943 N, : 3531 N)
2b. Effect of roller diameter on pressure distribution around a roller ( : Droller= 0.27 m, : Droller=
0.17 m). Belt tension: 197 N.

4.5 1.6
a b
4.0 1.4
Pressure ( bar)
Pressure (bar)

1.0 0.4

0.5 0.2

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

X / Lc X / Lc
Figure 3. Effect of the pressurisation procedure ( : 1.5 bar applied instantaneously, :
1.5 bar applied progressively over the first two minutes)
Figure 4. Pressure distribution: in FCC ( ______ ), and in the pilot filter belt press for a experiment
carried out with a series of seven pressing rollers (diameter 0.27 m) and a belt speed of 1.5 m
min-1 (------).
Figure 5. Comparison of FCC ( ) and the pilot filter belt press ( ) results in terms of cake
solid content.

Dry solid content (%)

0 20 40 60 80 100
Time (s)

View publication stats

Centres d'intérêt liés