You are on page 1of 8

Filterability test

There is no perfect method that can characterize sludge to predict membrane fouling propensity of the
sludge. The methods described below have their own strengths and limitations. The correlation between
the filterability test methods and actual membrane fouling rates must be checked case by case in order
to find the best method correlated with membrane fouling in a given location.
1. Free drainage test
Filterability of mixed liquor by membrane can be estimated by free drainage test as shown in Fig. 1. A
filter paper, e.g. Whatman 42, can be folded into a funnel. After pour 50 ml mixed liquor to the funnel,
filtrate volume is measured for a free determined time, e.g. 5 minutes, before the volume of filtrate is
measured. The test conditions must be kept constant including the volume of mixed liquor, filtration
time, etc. One downside is that maintaining surrounding temperature constant is tricky in this test
unless room temperature is controlled.
Although this method gives us an idea about the membrane fouling tendency by the mixed liquor used
in the experiment, following limitations are apparent.
Unlike actual filtration in membrane tank, there is only very slight driving force to filter mixed
liquor. Therefore cake compaction issue is much milder in the test than in the actual filtration.
No cross flow exists in the funnel unlike actual membrane filtration. Therefore, all particles and
large enough macromolecules in mixed liquor affect the filterability by forming a cake layer. In
crossflow filtration, large particles rarely form cake layer due to the back transport velocity from
membrane surface in the crossflow condition as discussed here.
Particle sizes dictate the filterability by affecting the permeability of cake layer. In actual MBR,
macromolecules such as SMP and EPS determine membrane fouling in general.
Filterability is affected by MLSS due to the lack of normalization against MLSS

Fig.1. Filterability test based on free drainage.

2. Modified free drainage test (Thiemig, 2011)


In order to eliminate or reduce the limitations of free drainage based filterability test, the mixed liquor
poured in a Buchner funnel is mixed using a bench mixer with a flat blade as illustrated in Fig. 2. In
addition, a larger amount of mixed liquor (500ml) is heated to 20 oC in water bath before it is poured
into funnel in this method in order to reduce the temperature effect. Finally the time constant obtained
is normalized by MLSS to reduce MLSS effect on filterability.
Experimental sequences are as follow
1. Take a 1 Liter mixed liquor shortly before the test
2. Insert a filter paper in the Buchner funnel (150 mm diameter, 0.6 micron pore size)
3. Place an empty 250 ml mass cylinder under the funnel
4. Clamp blade agitator at a height of 0.1 cm above the filter paper and mix at 40 rpm
5. Mix the mixed liquor sample and temper a 500 ml sample in a water bath to 20 oC
6. Quickly pour the 500 ml sample into the funnel
7. Activate a stopwatch when filtrate reaches 100 ml
8. Stop the stopwatch when filtrate reaches 150 ml
9. Measure MLSS as %
10. Calculate Sludge Filterability Index (SFI) as follow
SFI = delta_t (s) / MLSS (%)

Fig. 2. Setup and materials for measuring the sludge filtration index SFI (Thiemig, 2011)

3. Filterability test with pressure


Filterability of mixed liquor can be tested using a pressure vessel with stirrer as shown in Fig. 3. Relative
membrane fouling propensities of the mixed liquor can be estimated by measuring the time to obtain
certain volume of permeate under a constant operating conditions such as membrane area, pressure,
temperature, stirring rpm, etc. In the experiment performed by (Tarnacki, 2005), membranes with 38
cm2 was used at 1 bar, 20 oC, and 400 rpm.
Two indicators were defined. Filterability, L15, is estimated after 15 ml filtrate is obtained by dividing the
flux by pressure (equation 1). Fouling index, FI30, is calculated by dividing the flux at 30 minutes, J30, by
initial water flux, Jw,0 (equation 2).

------------------------------ (1)

------------------------------ (2)
One drawback of this method is that the effect of cake layer compaction is much more significant than in
MBR due to the high TMP (1bar) used in the test, which is 3-10 times higher than actual TMP in MBR.

Fig. 3. Fig. Filterability test cell (Tarnacki, 2005)

4. Time to filter, TTF (APHA, 1992)


TTF is determined by using an apparatus shown in Fig. 4 (Method 2710H). A 90-mm Buchner funnel is
used with Whatman #1, #2, or equivalent filter papers. The time required to filter 100 mL of a 200 mL
sample through the filter at the vacuum pressure of 51 kPa or 7.4 psi can be recorded as seconds. If
MLSS of the sample varies, TTF can be normalized by dividing by MLSS since TTF is affected by MLSS as
well as sludge quality.

Fig. 4. Apparatus to measure time to filter, TTF (APHA, 1992)

5. Delft filtration characterization method, DFCm


As shown in Fig. 5, DFCm relies on a single UF membrane (X-flow F5385, inside-out, ID=8 mm, nominal
pore size 0.03 m) under cross flow condition of 1 m/s to estimate the membrane fouling propensity of
mixed liquor. Flux is maintained constant at 80 LMH by vacuum using a peristaltic pump. Temperature,
pH, flux, dissolved oxygen (DO) concentration are recorded. The TMP at the feed, concentrate and
permeate are also recorded over the course of the experiment (Evenblij, 2006; Van den Broeck, 2011)
Subsequently, Rtotal (=Rm+Rc+Rf) is calculated using the resistance in series model, which is based on
Darcy's law, as shown in Fig. 6. The Rtotal at the permeate production of 20 L/m2 is denoted as R20 and is
used to compare membrane fouling propensities of different mixed liquors. Filterability is qualified as
poor when R20 is higher than 1x1012 m-1, moderate when 0.1 x1012 m-1 <R20< 1x1012 m-1 and good
when R20 values are lower than 0.1x1012 m-1.
This method provides closer imitation of real membrane filtration than the above methods. The flux
used, i.e. 80 LMH, however, is much higher than the flux commonly found in the MBR with immersed
membranes. The high flux might be inevitable to expedite the membrane fouling so that the test can be
completed within a reasonable time. As discussed here, the squeezing pressure on cake layer is directly
affected by the flux rather than TMP. Therefore, cake layer compaction becomes much more severe in
DFCm than in actual MBR. Under this circumstance, the effect of fine particles, EPS, and SMP can be

over expressed by filling up the void spaces of cake layer.

Fig. 5. Picture and flow diagram of the apparatus of DFCm (Geilvoet, 2010)

Fig. 6. Schematic representation of filtration characterization unit, DFCm (Krzeminski, 2010).

You might also like