• No results found

~ P e r g a m o n P l h S 0 0 4 3 - 1 3 5 4 ( 9 6 ) 0 0 3 5 0 - 8


Academic year: 2022

Share "~ P e r g a m o n P l h S 0 0 4 3 - 1 3 5 4 ( 9 6 ) 0 0 3 5 0 - 8"


Loading.... (view fulltext now)

Full text


~ Pergamon

P l h S0043-1354(96)00350-8

Wat. Res. Vol. 31, No. 5, pp. 1126-1134. 1997 ,~~ 1997 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/97 $17.00 + 0.00


K O E N G R I J S P E E R D T ® and W I L L Y V E R S T R A E T E @*

Centre for Environmental Sanitation, Coupure L653, B-9000 Gent, Belgium (Received January 1996; accepted revised form October 1996)

Abstraet--A new method, using low magnification microscopy combined with image analysis to estimate settling properties of activated sludge, is presented. It is shown that the morphology of the activated sludge flocs correlates significantly with the settling properties of the sludge, as expressed by traditional settling tests. The morphology is expressed by two parameters: the mean equivalent circle diameter and the mean form factor of the sludge flocs. In contrast with traditional tests, the new method takes only between 10 and 20 min, is automated and can be incorporated in other sensors. In addition, a rapid and reliable estimate of the suspended solids concentration can be obtained by establishing a calibration between the mean field area of a series of images with the concentration of a certain type of activated sludge. The relevance of the novel method is clearly illustrated by its ability to monitor the rise and the subsequent cure of a severe case of bulking in an activated sludge pilot plant. It is shown that a more sensitive measurement of the sludge settleability is obtained by using image analysis. Consequently, the method presented may have good potential in being used as an early warning system. © 1997 Elsevier Science Ltd

Key words--activated sludge, image analysis, settleability, concentration measurement


a = the intercept of the linear concentration cali- bration curve

A = the percentage field area AR = 3-D aspect ratio

b = the slope of the linear concentration calibration curve (L3/M)

dSVI = diluted sludge volume index (L3/M) Do = equivalent circle diameter (M) FD = fractal dimension

FF = form factor RD = roundness

ZSV = zone settling velocity (M/T)


The activated sludge process is one of the most frequently used processes for the purification of wastewater (Metcalf a n d Eddie, 1992; Horan, 1991).

Usually, the effluent from the secondary elarifier is not treated any further. Since this effluent has to meet certain standards, the good operation of the settler is critical for the operation of the whole plant.

However, it is estimated that at least 70% of the activated sludge treatment plants experience difficulties with the settler at least once a year (Pujol a n d Canler, 1992). The problems are multifold and can be due to bad operating strategy or to poorly designed clarifiers, b u t the majority of settling failures can be attributed to bulking sludge (Wanner, 1994).

*Author to whom correspondence should be sent. [Fax:

+ 329 264 62 48; e-mail: Willy.verstraete@rug.ac.be].

In this case, there exists an imbalance between floc forming bacteria a n d filaments, preventing the formation of well settling sludge floes (Jenkins et al., 1993). Obviously this disturbance greatly affects the appearance of the sludge floes (Andreadakis, 1993).

Ideally, sludge floes should be firm and round, to allow a good compactation. Highly irregular sludge floes are characteristic in bulking sludge. Another cause of the malfunctioning of the settling stage is pin point sludge, where a large a m o u n t of sludge floes are so small that they do not settle, causing a very turbid effluent (Sezgin, 1982). Bulking sludge compacts very poorly, which affects process performance through the recycle flow and requires large settler volumes.

This is why activated sludge plants are commonly over designed a n d not optimally used.

The first step in the battle against bad settling sludge is the development of on-line systems to monitor the settling properties of the sludge (Sekine et al., 1987; Albertson, 1991; Vanrolleghem and Verstraete, 1993). The second step involves acquiring an adequate remedy. Image analysis is a promising technique that could be incorporated into sensoring techniques. It is widely used in all kinds of applications (Russ, 1990; Vecht-Lifshitz a n d Ison, 1992), n o t in the least because the price/quality ratio of image analysis systems has been decreasing exponentially during the last few years. Image analysis of activated sludge, at high magnifications, was used by W a t a n a b e et al. (1990) to determine the filamentous state o f the sludge. Li and Ganczarcz'yk 1126


Estimating settleability o f activated sludge (1990, 1991) e x a m i n e d t h e size d i s t r i b u t i o n a n d

t h e i n t e r n a l s t r u c t u r e o f a c t i v a t e d s l u d g e flocs b y u s i n g i m a g e analysis. T h i s w o r k f o c u s e s o n t h e use o f a u t o m a t e d i m a g e a n a l y s i s t o c h a r a c t e r i z e s l u d g e settling p r o p e r t i e s a n d also to o b t a i n s l u d g e c o n c e n t r a t i o n d a t a .


Image analysis system

The sludge samples were investigated using a stereo microscope (WILD M800, Switzerland). The maximum magnification of the microscope used is 50 × . In this magnification range 1he system focuses on the sludge macro flocs which comprise, mostly, o f the mass in an activated sludge system (more lhan 80%) (Li and Ganczarczyk, 1991).

On the one hand, it is possible to have an overview of several sludge macro flocs as needed for the determination of the size distribution o f the flocs. On the other hand, one can also monitor individual si[udge floes in more detail, in order to detect the floc morphology more accurately. The sample is visualized using a dark field lighting system (Caldwell et al., 1992). The image is captured by a colour CCD-camera (JVC, Japan) and digitized with a frame grabber (Digithurst, UK) in 1/25 s. The digitized image is processed on a PC, by means o f specifically developed software (in Microsoft Visual C + + ).

Imaging procedures

Before analyzing an image, a threshold has to be defined in order to distinguish objects from the background. For a monochrome range, this is simply a lower and upper limit o f intensity. Thanks to the dark field lighting floes appear as white objects on a black background, enabling us to use the same thresholds irrespective o f the source o f sludge. An example if a bitmapped image is presented in Fig. 1, Pixels have a value o f either one or zero, enabling the easy processing of the images with the PC.

Measuremen ts

Size. The size o f the sludge flocs is an important parameter with respect to the settling properties (Ganczarczyk, 1994). The measurement is restricted to macro flocs with a size larger than 10#m, since the magnification used for this measurement is 25 × . For ease

1127 o f interpretation, the size o f the flocs is expressed as the equivalent circle diameter D~, calculated from the real projected area:

D , = 2. A ~ a . (1)

Shape quantification. It is mentioned in the literature (Eriksson and H~irdin, 1984; Watanabe et al., 1990) that the shape o f sludge flocs is related to the settling properties.

Many shape quantifying parameters can be measured by means of image analysis (Russ, 1990; N~imer and Ganczarczyk, 1993; Pons et al., 1993). Four parameters were implemented:

• The form factor (FF) describes the deviation of an object from a circle. It is particularly sensitive to the

"roughness" o f the boundaries. A circle has a F F of one.

F ~ r = ~ : . 4.n-Area (2)

• The so-called three-dimensional aspect ratio, (AR), is sensitive to the extension of an object. The more elongated it is the larger the value o f this parameter. A circle has an A R o f one.

A R = 1.0 + 4 { Length )

n \ Width - 1.0 . (3)

• The roundness (RD) is mainly influenced by the elongation of an object. It varies between 0 and 1, a circle has an RD of one.

4. Area

RD - n.Length ~ (4)

• The fractal dimension (FD), of the perimeter o f an object, is a measure for the irregularity o f the perimeter (Peitgen et al., 1992). Logan and Wilkinson (1991) investigated the fractal dimension o f bioaggregates. Li and Ganczarczyk (1990) and Eriksson et al. (1992) reported on the fractal properties o f activated sludge flocs. The fractal dimension is determined in this work with the "mosaic amalgamation" algorithm as described by Russ (1990). The main disadvantage o f this method is that it is quite time consuming in comparison with the assessment of the other parameters ( + 5 min vs 20 s per image). A circle has an F D of one.

For these morphological parameters, and also for the determination o f the size o f the flocs, it was verified by Carrette (1995) that the measurements were reproducible and independent o f the dilution o f the sludge sample, provided the concentration remained between 0.5 and 4 g/I.

Table 1 shows the statistical analysis for 10 successive measurements on the same sludge sample. The coefficients o f variance ( = standard deviation/average*100) are very

Fig. 1. A binary image o f sludge flocs, pixels have a value o f either one or zero. The bar represents 100 pm. The

magnification is 40 x .

Table 1. Reproducibility of the quantification of the morphological measurements

Measurement FF AR

1 0.519 2.718

2 0.518 2.718

3 0.529 2.695

4 0.58 2.708

5 0.526 2.710

6 0.530 2.694

7 0.534 2.700

8 0.514 2.718

9 0.512 2.950

10 0.529 2.692

Average 0.529 2.730

Standard deviation 0.019 0.078 Coefficient of variance 3.59 2.86


1128 K. Grijspeerdt and W. Verstraete Table 2. Influence of the dilution: anova-analysis

Sum of Degrees of Weighted sum

Source squares freedom of squares F Ft,,.0.05

FF Treatment 0.01308 4 0.00327 1.867 3.0555

Error 0.02626 15 0.00175 9 - -

Total 0.03934 19 - - - - - -

AR Treatment 0.27722 4 0.06930 2.339 3.0555

Error 0.44431 15 0.02962 7 - -

Total 0.72153 19 - - - -

low, pointing to very good reproducibility. The influence of the dilution was examined by making a dilution series of the same sludge type and processing the results with an anova-analysis. Four replicates were done per dilution.

The results o f the anova-analysis are shown in Table 2.

Since the calculated F-value for each o f the morphological parameters is smaller than the critical F-value, it can be concluded that there is no significant difference between the series. In the same way, it was verified that the magnification showed a significant influence on the measurements (Carrette, 1995). Therefore, it was necessary to work with the same magnification for a specific type o f measurement.

Manual determination of sludge characteristics Image analysis can be subdivided in different stages:

• sample peparation and imaging;

• digitization;

• image analysis.

The last two stages are done through suitable hard and software and are easily automated. However, the first level requires delicate manipulation. The sludge sample must be sufficiently diluted to avoid saturation o f the image, and the liquid layer thickness must be small enough to be able to use relatively high magnifications, focusing on a thin plane. In the first phase, this was done manually (Grijspeerdt et al., 1994).

The procedure was as follows: 2 ml was taken from an activated sludge sample, diluted with effluent to 10ml and transferred to a glass petri dish. Plastic petri dishes could not be used due to the high hydrophobicity. The dish was put under the microscope. Figure 2 shows that the standard error on the measurements reaches a saturation value when the number o f objects is large enough.

Consecutive images were taken until the total number of detected objects was 200. This manual method is rather tedious as it takes between 1 and 2 h to examine one sample. Dr was not included in the measurements. Sludge



0.20 Q

"~1 0.1s




° o

" - . O o

" ° ' ° o - ° ° . o . . . .

0.00 i i i

0 50 100 150 200

N u m b e r o f d e t e c t e d s l u d g e f l o c s Fig. 2. The evolution o f the standard error o f the form factor (FF) and the aspect ratio (AR) as function o f the

number o f detected objects.

samples were collected from several treatment plants, treating domestic as well as industrial wastewater. The samples were stored in a fridge for no longer than a day before analysis.

Automatic determination of sludge characteristics

To have a practical measuring method, the sample preparation and imaging had to be automated. A flow through cell (Hellma, Belgium), designed for spectrophoto- metric applications, was used. This cell has a thickness o f 1 mm, which is sufficiently small for focusing. The sludge is sucked through the cell with a peristaltic pump at a flow rate of 200 ml/min. This pump is situated downstream o f the cell.

At certain intervals, the pump is stopped and an image is taken. N o significant influence on the measured parameters could be observed from the pumping (Carrette, 1995). This was verified by comparing the size and the morphology o f the sludge flocs before and after passing through a peristaltic pump. The results and the statistical analysis are summarized in Table 3. The variances were compared using a F-test, after which the averages were checked with a t-test (Press et al., 1992). The high values o f the probability levels show that the influence o f the pumping is statistically not significant. While the image is being processed, the sample is pumped through the cell again. This procedure is repeated until statistically relevant results are obtained. The control o f the pump, the lighting and the camera is done by the same PC that performs the image processing. After the first series o f experiments, the automatic method was adapted to allow a standardized flocculation before sampling. This was done by gently stirring (20rpm) the sludge sample in an erlenmeyer flask with a magnetic stirrer, which was stopped for a certain time t ~ ( = 60 s), to allow the sludge macro flocs to reflocculate. Then the same procedure as before was used. The experimental setup is presented in Fig. 3. Only 200 ml o f sludge is required and with this automated procedure it takes between 10 and 20 min to measure one sample.

Sludge samples from different sources were examined.

In addition, three types o f sludges were monitored over a longer period o f time. Sludge from a 150 1 pilot plant, a full scale industrial plant, and a domestic plant were used. These sludge types are denoted as type 1, 2 a n d 3, respectively.

Off-line determination of sludge settleability

The diluted sludge volume index (dSVI) and the corresponding zone settling velocity (ZSV) were determined for the sludge samples. These were measured in an Imhoffcone following the guidelines from Albertson (1992).

The dSVI is characteristic for the compression phase in the settling process, while the ZSV gives more information with respect to the hindered settling regime (Gregory and Zabel, 1990).

Statistical methods

On the basis o f the gathered data, two items were addressed: the relationship between the ~ttleability indexes (dSVI and ZSV) and the morphological parameters, and redundant parameters.

The first question can be addressed by performing a correlation analysis o f the data, while the second answer


Estimating settleability of activated sludge

Table 3. Comparison of the sludge flo¢ characteristics before and after passing through peristaltic pump

Average Variance F-test t-test

Before After Before After

D,(pm 2) 14.5-104 13.1.104 9077 9711 0.063 0.147 6835 8074 14.0.103 18.0.103 0.070 0.939

FF 0.671 0.624 0.059 0.063 0.108 0.435

0.668 0.701 0.067 0.070 0.155 0.961

AR 3.253 3.185 0.157 0.152 0.244 0.661

2.916 3.056 0.203 0.227 0.216 0.509


Edenmayeri flask F" ~

Magnel ic stirrer



]Image analysis[

r I system


Fig. 3. The setup used for automated measurements.

requires a multivariate analysis (Morrison, 1967). Both analyses were conducted with the StatGraphics software package (STSC, USA). A canonical analysis gave the best linear correlation between the two sets of data and attributes a weight coefficient to the different variables. One data set contained the off-line measurements, while the other comprised the variables obtained by image analysis.

It is also very important that the results are statistically relevant, i.e. enough images have to be analyzed. In general, at least 150 objects need to be processed before the standard error reaches a saturation value (Fig. 2), corresponding with _+ 10 images. This imposes a computational burden on the system, especially for the determination of the fractal dimension.


1 /

Intercept: .0.366:1:0.965 1' 30 - Slope: 10.726 + 0.545 [ ~ [

lO 5 o -

0.0 0,5 ~ 0 1.5 2 0 2.5 3 0 3 5


Fig. 4. The relation between the fields area (A) and the concentration of the pilot plant sludge (X). The field area represents the mean percentage detected area in a series of

images. The magnification used was 25 x .

Concentration measurement o f activated sludge

The image analysis system enables to measure fast and reliable the concentration of activated sludge. This is done by correlating the "field area" (the percentage area detected in an image, Russ (1990)) with the real concentration. To do this, a calibration curve has to be made of progressively diluted samples by measuring the corresponding field area A (Fig. 4). Statistically significant results are obtained by analyzing at least 10 images of each particular sample. This calibration is dependent on the sludge type monitored, so it must be done for each new sludge type. The magnification used for this measurement was 25 × .

Pilot plant measurements

For some studies, sludge was examined from a pilot plant of 1501 used for carbon- and nitrogen removal studies (Fig. 5). The plant was fed with synthetic wastewater with a concentration of 500rag/1 COD and 50mg/l total nitrogen. The sludge concentration in the bioreactor fluctuated around 2 g/1. For the pilot decanter design specifications, the reader is referred to Grijspeerdt et al.


RESULTS AND DISCUSSION M a n u a l m e t h o d

T h e m o r p h o l o g i c a l p a r a m e t e r s m e a s u r e d f r o m different sludge samples with the m a n u a l m e t h o d were statistically c o m p a r e d with off-line measure- m e n t s o f the settleability. T h e results o f the c o r r e l a t i o n analysis are s u m m a r i z e d in T a b l e 4. It c a n be seen t h a t the F F and, to a s o m e w h a t lesser degree, the A R are the best c o r r e l a t e d with the dSVI. This m e a n s t h a t the m o r e the f o r m o f the floc a p p r o a c h e s t h a t o f a sphere, the b e t t e r it settles. F i g u r e 6 shows WR 3 1 / 5 ~


1130 K. Grijspeerdt and W. Verstraete

Waste Effluent

Inlluent Selector Recirculation

I~ ~ ~ 301 301


601 ~ Decanter Anoxic Aero ,ic Recycle sludge

Chlorination ~ Hy[~chlorite

Fig. 5. Schematic overview of the 150 1 pilot plant used for carbon and nitrogen removal studies under laboratory conditions (20-22°C).

the relationship between these two morphological parameters and the dSVI. Table 4 indicates that the same two parameters also correlate very well with the ZSV, although not as pronounced as with the dSVI.

Since dSVI and ZSV are inversely related themselves, the correlation coefficients o f F F and A R logically change signs for these two parameters.

The canonical analysis (Table 5) revealed that for the dSVI the highest weight coefficient is attributed to the F F followed by the A R . These are the same two parameters that emerged from the correlation analysis. Such correspondence between correlation- and canonical analysis is not always the case (Morrison, 1967). F o r the case o f hindered settling, as expressed by the ZSV, A R is the most important factor. F r o m these first series o f experiments can be concluded that the F F and the A R seem to be the most suited morphological parameters to estimate the settleability o f activated sludge. The fractal dimen- sion is correlating to a lesser extent and its determination is quite time consuming ( + 3 h) and was hence abandoned.

It must be remarked u p o n that the noise level o f the measurements was not negligible. However, it must be kept in mind, that the anlaysis was done on sludge from very different sources. The noise can be expected to be less when focusing on the same sludge type for a time series o f samples.

Automatic method

In the first stage, the m e t h o d was used without the flocculation procedure. In contrast with the m a n u a l method, the noise on the measurements was unacceptably high, both for different sludge types and also for the pilot sludge, m o n i t o r e d for a longer period.

This divergence in the results could only be attributed to the a u t o m a t i o n procedure. F o r the manual m e t h o d a system is used where a certain mixing is possible (the petri dish), whereas for t h e automatic method no mixing was present at the level o f the flow through cell. D u e to the gently mixing a certain a m o u n t o f flocculation is possible, which means that the flocs observed generally are larger.

Table 4. The correlation matrix for the settling and morphological parameters as determined with the manual method.

The numbers between brackets are the significance of the correlation

Diluted sludge Zone settling Fractal

volume index velocity Form factor Roundness Aspect ratio dimension


dSVI 1 . 0 0 . . . . .


z s v (0.99) 1 . 0 0 . . . .

-0.82 0.56

FF (0.99) (0.99) 1.00 -- - - - -

-0.15 0.08 0.19

RD (0.65) (0.39) (0.78) 1.00 - - - -

0.81 -0.69 -0.76 --0.21

AR (0.99) (0.99) (0.99) (0.83) 1.00 - -

0.56 --0.39 --0.62 - 0 . 1 6 0.69

FD (0.99) (0.99) (0.99) (0.68) (0.99) 1.00


E s t i m a t i n g s e t t l e a b i l i t y o f a c t i v a t e d s l u d g e

4'5 I ~ ° ~ • •

4.0 olpO oo


e~ o ~ °o

3.o t o

2.5 • o •

t . 0


~ o .8

f " " . ' . : l - • e

2 . 0 . . . .


1.5 .5

1.0 ' ' ' , , J .4


100 200 300

0 100 200


dSVI (ml/g) clSVI (ml/g)

Fig. 6. The relationship between the aspect ratio (AR), the form factor (FF) and the dSVI for the manual method. As a matter of reference, for a circle the AR and FF are equal to one.


When no gentle mixing is present, it is not possible for the micro flocs to reach each other and aggregate to form macro flocs ( > 10/~m), the category for which the system is designed for. Due to transport in the sampling tub,: and the peristaltic pump, the macro floes probzbly suffer from shear forces and break down. The gentle mixing helps to restore the macro flocs.

It is inherent to image analysis systems that smaller objects produce larger measurement errors.

An object is approximated by square pixels, resulting in a less accurate measurement of smaller objects.

This problem could be solved by imposing a lower limit on the number of pixels in an object. Russ (1990) noted that objects comprising a number of pixels smaller than 0.1% of the total number of pixels of an image should not be taken into account. The system used for this study has a total number of pixels of 350,000, resulting in a minimum number of pixels of 350.

The aggregation phenomenon occurring during the manual method was simulated by allowing the sludge to flocculate in an erlenmeyer flask before sampling (Fig. 3). Apparently, because the influence of the size of the sludge flocs is important, the equivalent diameter D, was also included in the next series of measurements.

The results obtained with respect to the dSVI for different sludge samples with this adapted automatic method are summarized in Fig. 7. A trend can be observed for the ]?F, for the De this is much less pronounced. Similar results are obtained for the AR, but with a lower correlation. Corresponding to the manual method, the correlation for the ZSV shows

the same trend, but the noise is higher. Table 6 summarizes the multiple correlations for the different cases, and also the correlations for the individual parameters. These numbers show that De has less influence on the dSVI, but is, nevertheless, significant.

It is also remarkable that this correlation points to a (small) inverse relation with De.

Apart from these sludge samples, three other types were monitored over a longer period of time.

For these three types, mentioned in Table 6, the settleability did not change drastically during the time they were monitored. The correlations show that for individual types of sludge the noise is less, as could be expected. The results also indicate that the sludges do not behave in entirely the same way with regard to the relationship between the measurements and their settleability. Probably this is due to other factors influencing the settleability not accessible through image analysis (Urbain et al., 1993).

This indicates that this procedure should be used with a certain precaution and must not be considered as an absolute measurement. Ideally, a calibration should be done for every single sludge type.

Especially the evolution of the morphological state of the sludge floes with time sould be considered as an indication about changes in the settling state of the sludge.

Practical application

The image analysis procedure outlined above was used to monitor the settleability of the sludge of the pilot plant. In normal conditions, the dSVI varies between 150 and 200 ml/g, at sludge concen- trations around 2 g/1. At a certain time, the sludge

Table 5. The canonical weight co¢tilcients for the different image analysis parameters as determined with the manual method


FF RD AR FD correlation Significance

dSVI - 0 . 5 7 -0.001 0.53 - 0 . 0 9 0.88 0.99

ZSV 0.24 - 0 . 2 0 - 1.01 0.25 0.72 0.99


1132 K. Grijspeerdt and W. Verstraete

055 450 900

/ " Start chlorination +

/ 0 50 • Installation selector "~"----~. s00

900" / 045 - , ° 700

/ ~ 350

800 / / 0 4 0 " : ~ 600

6oo. / / oso 400

50o / 025 ~/ . 2so

, 300

400 / / 020 200 2o0

8oo' / / ols I I I I to 150 100

200 ~ ~ , ~ 300 50 100 150 200 250

250 D a y s

10o ~ ~ Fig. 8. The evolution of the form factor (FF), the equivalent

o ~ ~-~ ~ ~ o ~ diameter (De) and the diluted sludge volume index (dSVI)

0.4 /~, )0 5 5 ~ 100 lO0 during a bulking period.

Fig. 7. The relationship between the form factor (FF) and higher sensitivity. This behaviour can be used in an the equivalent diameter (Do) as measured by the automated

method and the diluted sludge volume index (dSVI) for early warning system for bulking onset, allowing different sludge sources. The arrow indicates that the dSVI more time to investigate possible causes o f the settling for activated sludge flocs tends to a minimum when they are problem.

quasi-circular and of the order of 50 #m.

started to bulk very seriously (due to a proliferation o f N o s t o c o i d a l i m i c o l a ) , causing dSVIs higher than 500 ml/g. Since this restricted the working o f the plant, due to an overloading o f the decanter, remedial actions were taken. A selector was installed to bring the recycle sludge in contact with a high concen- tration of the influent (Patoczka and Eckenfelder, 1990). Because the effect o f a selector has to be situated at a longer term (several weeks), chlorination was also applied in order to try to reduce the bulking in a shorter period (Fig. 5). The chlorine was applied as hypochlorite following the guidelines described by Jenkins e t al. (1993). Sludge was continuously taken from the aeration basin and brought into contact with hypochlorite in a 5 1 vessel. The concentration o f the hypochlorite was 170mg/1, the contact time 1 5 m i n and all the sludge was exposed to the hypochlorite twice a day. The evolution o f the dSVI, the F F and the De are summarized in Fig. 8. The bulking problem was solved, although the dSVI did not return completely to the original level, remaining between 200 and 250 ml/g. However, this value was sufficient to operate the plant under normal conditions. The image analysis parameters appeared to be m o r e sensitive to changes in settleability than the dSVI, since both the F F and the D0 started to change at least 5 d earlier. The F F seems to have most potential for monitoring purposes, because of the

Table 6. Correlation study for different sludge types

Sludge Multiple Number of

t y p e correlation WeightFF WeightD, samples

1 0.86 0.91 0.09 10

2 0.84 0.89 0. I 1 16

3 0.91 0.96 0.04 30

Overall 0.81 0.85 0.15 63

C o n c e n t r a t i o n m e a s u r e m e n t s

F o r 10 types o f sludge, the concentration calibration was carried out (Table 7). In almost all cases, a very good linear relation was obtained.

However, for two cases the calibration gave unusual results, which could not be explained. By doing a

Table 7. Concentration calibration (A = a + b.X) for different sludge types

Sludge a b Correlation

1 - 0 . 3 6 6 10.73 0.99

2 0.056 7.59 0.98

3 0.248 8.14 0.94

4 - 0.174 6.54 0.59

5 0.210 6.89 0.97

6 0.109 9.21 0.95

7 0.054 5.72 0.98

8 - 0 . 1 0 5 7.19 0.94

9 - 0.284 8.04 0.42

10 0.305 10.18 0.96


~ 6.5

.~ s.o


~ 5.5 U 0 C 0

(J 0.5


Time (d)

Fig. 9. The evolution of the calibration coefficients of the linear relation A = a + b ' X over a period of 3 weeks. A is

the field area and X is the concentration.


Estimating settleability of activated sludge 1133 linearity test on the calibration, the validity of the

measurement is e~Lsily tested. If the sludge is too concentrated ( > 4 g/I), the image gets saturated and a linear calibration is no longer valid. Therefore, a suitable dilution of the sludge sample under study had to be carried out. From the results it is also clear that the calibration is sludge dependent.

The concentration measurement based on the image analysis system was tested by determining the calibration coelficients for the pilot sludge during a longer period of time. The results are shown in Fig. 9, where it can be seen that the calibration remains sufficiently constant for at least 3 weeks.

It is advisable, however, to do a test analysis once a week and compare this value with the predicted one.

The main advantage of this measurement is the independency of the colour of the sludge. Reports have mentioned the failure of suspended solids probes when the colour of the sludge changes (Sikow and Pursiainen, 1995). It is also clear that the calibration is dependent on the density of the sludge floes. Further research is needed to see whether this dependency can yield useful information (Dammel and Schroeder, 1991).


In this paper, a method to estimate the settle- ability of activated sludge using image analysis is described. The results show that it is possible to relate the morphology of the sludge floes to traditional settling indexes. There is a clear statistical correlation between the mean form factor of sludge floes and the diluted sludge volume index. Although an idea about the settling state of the sludge can be obtained in le:gs than 30 min, it is not meant to be an absolute measurement. The value of this technique has above all to be situated at the level of automatic on-line monitoring of activated sludge plants. Especially changes in the measured par- ameters can indicate an upcoming problem with the settleability of the sludge. The application in a practical case strongly indicates that the measured parameters are more sensitive to changes in settleability than the conventional methods. This shows that the method could be useful for application in control schemes incorporating the settleability of the sludge, because it can act as an early warning system. The possibility of estimating the concen- tration of activated sludge adds to the attractiveness of the technique.

Acknowledgements--Fhis research was financially sup- ported by the Flemish Institute for Encouragement of Scientific Technological Research in Industry (IWT).


Albertson O. E. (1991) Bulking sludge control--progress, practice and probl,ems. Wat. Sci. Technol. 23, 835--846.

Albertson O. E. (1992) Clarifier design. In Water Quality Management Library: Design and Retrofit o f Wastewater Treatment Plants for Biological Nutrient Removal. (Edited by Randall C. W., Barnard J. L. and Stensel H. D.), pp. 185-254. Technomic Publishing Company, Lancaster, PA.

Andreadakis A. D. (1993) Physical and chemical properties of activated sludge floe. Wat. Res. 27, 1707-1714.

Bouwer E. J. (1989) Transformation of xenobioties in biofilms. In Structure and Function of Biofilms (Edited by Characklis W. G. and Wilderer P. A.), pp. 251-267.

Wiley, New York.

Caldwell D. E., Korber D. R. and Lawrence J. R.

(1993) Analysis of biofilm formation using 2D vs 3D digital imaging. J. appl. Bacteriol. Syrup. Suppl. 74, 52S-66S.

Carrette R. (1995) Image analysis and nutrient control in relation to sludge quality. Engineers Thesis, Faculty of Applied Biological Sciences, University of Gent, p. 121 (in Dutch).

Dammel E. E. and Schroeder E. D. (1991) Density of activated sludge solids. Wat. Res. 25, 841-846.

Eriksson L. and H/irdin A.-M. (1984) Settling properties of activated sludge related to floc structure. Wat. Sci.

Technol. 16(10-11), 55-68.

Eriksson L., Steen I. and Tendaj M. (1992) Evaluation of sludge properties at an activated sludge plant. Wat. Sci.

Technol. 25(6), 251-265.

Ganczarczyk J. J. (1994) Microbial aggregates in wastewater treatment. Wat. Sci. Technol. 30(8), 87-95.

Gregory R. and Zabel T. F. (1990) Sedimentation and flotation. In Water Quality and Treatment, 4th edn (Edited by Symons J. M.). McGraw-Hill, New York.

Grijspeerdt K., Vanrolleghem P. and Verstraete W. (1994) Measuring settleability of activated sludge using image analysis. Med. Fac. Landbouww. Univ. Gent. 59/4a, 2037-2045.

Grijspeerdt K., Bogaert H. and Verstraete W. (1996) Design and verification of a small scale secondary clarifier (submitted).

Horan N. J. (1991) Biological Wastewater Treatment Systems: Theory and Operation. Wiley, Chichester.

Jenkins D., Richard M. G. and Daigger G. T. (1993) Manual on the Causes and Control o f Activated Sludge Bulking and Foaming, 2nd edn. Lewis, London.

Li D.-H. and Ganczarczyk J. (1990) Structure of activated sludge floes. Biotechnol. Bioengng 35, 57-65.

Li D.-H. and Ganczarczyk J. (1991) Size distribution of activated sludge floes. J. Wat. Pollut. Control Fed. 63, 806-814.

Logan B. E. and Wilkinson D. B. (1991 ) Fraetal dimensions and porosities of Zoogloea ramigera and Saccharomyces cerevisae aggregates. Biotechnol. Bioengng 38, 389-396.

Metcalf and Eddy (1992) Wastewater Engineering: Treat- ment, Disposal, Reuse, 3rd edn. McGraw-Hill, New York.

Morrison D. F. (1967) Multivariate Statistical Methods.

McGraw-Hill, New York.

N~imer J. and Ganczarczyk J. J. (1993) Settling properties of digested sludge particle aggregates. Wat. Res. 27, 1285-1294.

Patoczka J. and Eckenfelder W. W. (1990) Performance and design of a selector for bulking control. J. Wat. Pollut.

Control Fed. 62, 151-159.

Peitgen H.-O., Jiirgens H. and Saupe D. (1992) Fractals for the classroom. Part One: Introduction to Fractals and Chaos. Springer, New York.

Pons M. N., Vivier H., R6my J. F. and Dodds J. A. (1993) Morphological characterization of yeast by image analysis. Biotechnol. Bioengng 42, 1352-1359.

Press W. H., Teukolsky S. A., Vetterling W. T. and Flannery B. P. (1992) Numerical Recipes in C: the Art o f Scientific Computing, 2nd edn. Cambridge University Press, London.


1134 K. Grijspeerdt and W. Verstraete Pujol R. and Canler J. P. (1992) Biosorption and dynamics

of bacterial populations in activated sludge. Wat. Res. 26, 209-212.

Russ J. C. (1990) Computer-Assisted Microscopy: the Measurement and Analysis o f Images. Plenum Press, New York.

Sekine T., Tsugura H., Urushibara S., Furuya N., Fujimoto E. and Matsui S. (1989) Evaluation of settleability of activated sludge using a sludge settling analyser. Wat.

Res. 23, 361-367.

Sezgin M. (1982) Variation of sludge volume index with activated sludge characteristics. Wat. Res. 16, 83-88.

Sikow M. and Pursiainen J. (1995) Use and maintenance of on-line measurements at suomenoja wastewater plant. In Proc. IA WQ Specialized Conf. on Sensors in Wastewater Technology, 25-27 Oct, Copenhagen, Denmark.

Urbain V., Block J. C. and Manem J. (1993) Bioflocculation

in activated sludge: an analytic approach. Wat. Res. 27, 829-838.

Vanrolleghem P. A. and Verstraete W. (1993) On-line monitoring equipment for wastewater treatment pro- cesses: state of the art. In Proc. TI-KV1V Studiedag Optimalisatie van Waterzuiveringsinstallaties door Pro- ceskontrole en- sturing, Gent, Belgium, pp. 1-22.

Vecht-Lifshitz S. E. and Ison A. P. (1992) Review:

biotechnological applications of image analysis: present and future prospects. J. Biotechnol. 23, 1-18.

Wanner J. (1994) Activated Sludge Bulking and Foaming Control. Technomic, Basel, Switzerland.

Watanabe S., Baba K., Matuzaki H., Hiraoka M. and Tsumura K. (1990) Operation condition diagnosis system for activated sludge processes based on microorganism image information. In Advances in Water Pollution Control (Edited by Briggs R.), pp. 683-690. Pergamon Press, London.


Related documents

We recommend to the annual meeting of shareholders that the income statements and balance sheets of the parent company and the group be adopted, that the loss of the parent company

Konsumtionsutvecklingen utav trävaror i Europa har alltsedan 2003 utvecklats på ett positivt sätt där konsumtionen, enligt uppgifter från ECE Timber Committee, under 2005 ökade med

Aktieägarnas inflytande i Bergs Timber utövas via årsstämman som är Bolagets högsta beslutande organ. På årsstämman behandlas bolagets utveckling och beslut tas i centrala

Tillgångar och skulder i koncernen har värderats till anskaffningsvärden med avdrag för ackumulerade avskrivningar och eventuella nedskrivningar. Goodwill utgörs av det belopp

Hence the fi nal price of the gold sold (the sales value) can diff er markedly from the acquisition value of the gold produced. In the same way as the gold price can fl uctu-

The part of Meda’s operation that is concerned with product development, clinical trials, production, marketing, and sales of the company’s products entails a risk of

The consolidated accounts include the Parent Company and all companies in which ReadSoft controls more than 50 percent of the votes or otherwise has a controlling infl uence. If

RR 32 innebär att moderbolaget i årsredovisningen för den juridiska personen tillämpar samtliga av EU godkända IFRS enligt koncernens tillämpning av dessa principer och uttalanden