Download - Anaerobic Digestion: Effects on Particle Size and Dewaterability

Anaerobic Digestion: Effects on Particle Size and DewaterabilityAuthor(s): Desmond F. Lawler, Yoon Jin Chung, Shiaw-Jy Hwang and Barbara A. HullSource: Journal (Water Pollution Control Federation), Vol. 58, No. 12 (Dec., 1986), pp. 1107-1117Published by: Water Environment FederationStable URL: http://www.jstor.org/stable/25043143 .

Accessed: 14/07/2014 19:59

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Water Environment Federation is collaborating with JSTOR to digitize, preserve and extend access to Journal(Water Pollution Control Federation).

http://www.jstor.org

This content downloaded from 81.229.103.174 on Mon, 14 Jul 2014 19:59:13 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=wef

http://www.jstor.org/stable/25043143?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


FOCUS

Anaerobic digestion: Effects on particle size and dewaterability Desmond F. Lawler, Yoon Jin Chung, Shiaw-Jy Hwang, Barbara A. Hull

Anaerobic digestion is commonly used in wastewater treat

ment plants to stabilize sludge produced in primary and sec

ondary treatment, and to reduce the volume of solids for ultimate

disposal. Solid/liquid separation (secondary digestion, thickening or dewatering, or a combination of these) follows anaerobic

digestion to complete volume reduction. Nevertheless, no clear

picture of the effects of anaerobic digestion on subsequent de

watering (or other solid/liquid separation processes), or on the

particle characteristics that influence dewatering, emerges from

previously-published research.

The objectives of this research were to study the effects of anaerobic digestion (including design and operational variables of digesters) on particle characteristics and dewaterability of typ ical municipal wastewater sludges and to relate the particle char

acteristics to sludge dewaterability. These objectives were achieved through operation of carefully-controlled laboratory digesters under various conditions.

BACKGROUND

Digestion. Anaerobic digestion is currently thought of as a

three-stage process that involves hydrolysis of solid and complex soluble organics by fermentative bacteria, oxidation of volatile

acids and alcohols to acetate and hydrogen, and C02 and CH4

production by methanogenic bacteria.1'2 Design and operational variables that significantly influence digester performance include

temperature, solids loading rate, solids detention time, and in

fluent sludge concentration and composition (the relative mix

ture of primary and second sludges). High-rate digesters usually

are designed3 for 10 to 20 days detention time, influent solids

concentrations of 3 to 8%, volatile solids loading rates of 1.6 to

6.4 kg/m3 d, and operating temperatures of 30? to 40?C. Anaerobic digestion and dewaterability. Previous studies on

the effects of digestion on dewatering present a confusing picture.

Approximately equal numbers of studies show that digestion

improves dewaterability or makes it worse. Rudolphs and Heu

kelekian4 reported, on the basis of batch tests a well-digested

sludge had better drainability on a small sand filter than raw

sludge and that drainability improved with increased digestion time. Brooks et al.5 showed that digested sludge had better de

waterability (lower specific resistance) than activated sludge or

mixtures of primary and activated sludges. Haug et al.6 found

that without thermal pretreatment, digestion significantly im

proved dewaterability of primary sludge and a 1:1 mixture

(weight of volatile solids basis) of primary and activated sludges but had little effect on dewaterability of activated sludge alone. A U. S. Environmental Protection Agency (EPA) manual7 states that anaerobic digestion of primary sludge alone produces clearer

digester supernatant and a more easily dewaterable sludge than

when biological sludge is added. Kini and Nayak8 showed that

digested sludge had better dewaterability (measured in vacuum

filtration tests) than primary sludge and that dewaterability im

proved as digestion proceeded.

However, other authors reported that dewaterability deteri

orated with anaerobic digestion. Pearson and Buswell9 found that as digestion proceeded, sludge drainability was reduced.

Others10,11 concluded from plant experiences that digested ac

tivated sludge was more difficult to dewater than other sludges. Karr and Keinath12 reported that anaerobically digested sludge was more difficult to dewater (measured by capillary suction

time and specific resistance) than either primary or activated

sludge.

Particle characteristics and dewaterability. Many factors have

been reported to influence sludge dewatering characteristics. A

review of previous work concluded that particle size was clearly

the most important factor that influences dewaterability, and

that many other factors reported were important because of their

effect on particle size.12 In experimental work reported in the same paper, sludge samples were fractionated through a series

of steps that included sieving, flocculation and settling, and

membrane filtration at three pore sizes. After re-mixing sludges

using different proportions of their fractions, researchers found

that different types of sludge made in similar proportions had

equal dewaterability, and that the "supracolloidal" fraction (ap

proximately 1 to 100 /im) had the most significant effect on

dewaterability. Particle size distribution was not directly mea

sured. Instead the fractionation technique was used to estimate

size distributions.

Digesters should be designed and operated not only for

sludge stabilization, but also to improve

subsequent dewaterability.

Knocke and co-workers13,14 reported direct measurements of

particle size distributions and their effects on dewaterability for metal hydroxide and other water treatment plant sludges. A good correlation occurred between particle surface area determined

from size distributions and the Kozeny equation applied to spe cific resistance. Particle size distributions (1 to 80 /im) of two

anaerobically digested sludges showed that although most of the

particles, on a number basis, were in the smallest sizes measured,

the volume (or mass, assuming constant density with size) was

spread almost uniformly across a log diameter scale throughout

that range.15 No correlation between the size distributions and

thickening or dewatering measurements was attempted in that

study.

December 1986 1107



Lawler et al._

EXPERIMENTAL DESIGN

To define the interrelationships among anaerobic digestion, particle characteristics, and dewaterability, laboratory-scale high

rate anaerobic digesters were operated under various conditions.

Feed strength, feed composition, and solids detention time were chosen as primary control variables, and each was varied in

separate experimental sets while the others were held at a stan

dard or control value. Feed composition refers to the percent

by weight of activated sludge in the influent; the remainder was

primary sludge. The specific values used for each reactor in each

set are given in Table 1 ; exceptions to the control values were

necessary in some cases as indicated.

In Set 1, the detention time was 21 days for all reactors; this was expected to be the standard, but the resulting solids loading rates were too low. Ten days was chosen as the standard detention

time after Set 2; this change accounts for the apparent duplication between Sets 1 and 4 in varying the influent solids concentration. For the reactor in Set 3 with 100% activated sludge as the influent, it was not possible to achieve 3.5% solids concentration by gravity thickening; therefore the influent solids concentration for this reactor was 2.5%. Also in Set 3, one reactor was operated at 4?C

in an attempt to test the effect of mixing with little or no digestion. The two-stage operation in Set 5 separated the acid phase from the methane production phase in two reactors in series. These

reactors were operated with detention times of 2 and 8 days for the acid-phase and methane-phase reactors, respectively; the

short detention time for the acid-phase is consistent with the

findings of Eastman and Ferguson.16

Digester design. The laboratory digesters were acrylic cylinders (20-cm I.D.) with a liquid volume of 6.0 L and a gas space of

approximately 1.5 L. Plastic ball valves (1.27-cm I.D.) were used

as ports on the top and side for introduction and withdrawal of

sludge, respectively; the large size was used to minimize floe

breakup. Gas was collected in inverted bottles submerged in an acid brine solution (10% NaCl and 1.0 N H2S04) and measured

through a wet test meter. Continuous mixing was accomplished with a stirrer (20 rpm) that consisted of three submerged flat

impellers and a fourth impeller with a curved face protruding above the liquid surface to prevent scum layer development. Baffles (2 cm) were installed throughout the depth at 90-deg intervals.

Feed sludge. The entire sludge volume used in each set of

experiments was obtained at one time from a municipal treat ment plant in Austin, Tex.; influent sludge for the laboratory reactors was the same throughout each set but was different for

different sets. Primary and activated sludges were obtained sep

arately to control the mixture ratio. Sludge was passed through a screen ( 1.27-cm mesh) to remove large materials, mixed at the

desired ratio after measurement of the total solids content of

each, and stored at 4?C in full 1-L plastic bottles to minimize air contact before use.

Operation. For initial start-up of the laboratory digesters, active

digesting sludge was obtained from a local municipal treatment

plant. After each experimental set, the contents of all the digesters were combined, mixed, and divided again so that all digesters had identical conditions for the start of the next set.

Sludge was withdrawn from each digester just before each

feeding, with the amount (equivalent to the amount to be added) determined by the detention time. Feeding was done once per

day during Set 1 and twice daily thereafter. Batch feeding was

preferable to continuous feeding to avoid floe breakup in lab

oratory positive displacement pumps. All digesters were operated in constant-temperature rooms at the desired temperatures (35?C

and 4?C). To have the same average velocity gradient in the

4?C reactor as in those at 35?C, the mixing speed was increased to 33 rpm.

Monitoring. Digestion was monitored by measurements of

the quantity and composition of the gas produced and by analysis of sludge samples for pH, total solids, volatile solids, and chemical

oxygen demand (COD). Alkalinity and volatile acids are reported elsewhere.17 Gas production was monitored twice daily by re

leasing the collected gas through the wet test meter. The gas was

occasionally measured for carbon dioxide and methane content

using a gas Chromatograph. Twice each week pH was measured

immediately after sludge withdrawal; calibration was performed with buffers stored at the temperature of the sludges to be mea

sured. Total and volatile solids concentrations were determined

by Methods 209A and 209E, respectively, in "Standard Meth

ods."18 COD was determined by the dichromate reflux method

slightly modified from Method 508A in "Standard Methods."18 Instead of a sample volume, a weight of sludge (5 to 15 g) was

chosen and then diluted to 20 mL. Hence, COD data were in units of milligrams of COD per kilogram of sludge (mg/kg). Because the density of wet sludge is nearly that of water ( 1 kg/ L), the COD results are approximately equal to mg/L.

Measures of dewaterability. Two common measures of de

waterability, specific resistance and capillary suction time (CST), were used throughout this research. Despite their common use,

no standard methodology exists, especially for specific resistance.

The methods used in this research reflect the findings of Ka

vanaugh19 and Christensen and Dick.20,21 The CST device was

used with the small stainless steel reservoir ( 1.8-cm I.D. and 2.5

cm height). CST depends on the solids concentration, but one

Table 1?Research design.

Parameter varied*

Standard

value Variations

Set 1: Feed strength5, % total solids

Set 2: Detention time, days Set 3: Feed composition, weight % activated sludge Set 3: Temperature, ?C

Set 4: Feed strength, % total solids

Set 5: Staged operation

3.5

10

20

35

3.5

Single

2.0

7

0

4

2.0

2-Stageb

5.0

14

40

5.0

6.5

21

100b

6.5

1 All reactors were operated at the standard values for all parameters except for the one parameter varied in each set. 3 Reactors with exceptions to the standard values; see text.

1108 Journal WPCF, Volume 58, Number 12



_Focus on Dewatering

objective in this research was to compare dewaterability of dif ferent sludges with different concentrations. CST was measured both at the existing solids concentration of the sample and at a

standardized value, chosen as 2.1% total solids. This was chosen

because most digested sludge samples from the reactors fed with 3.5% raw sludge had this concentration in the early part of the research.

For specific resistance, the experimental apparatus consisted

of a B?chner funnel (7.6-cm I.D.), a 100-mL graduated cylinder with a vacuum port installed, a stainless steel screen (700-jum

openings), filter paper (Whatman no. 2, 9-cm diameter), and a

stopwatch. These details are provided because specific resistance

measurements depend somewhat on the equipment used. The

filter paper was positioned above the screen and sealed against the walls of the funnel by wetting with distilled water and ap

plying a vacuum. A sample of sludge (usually 50 mL) was poured into the funnel and measurements of the volume of filtrate (V) at different times (/) were taken until the rate of flow was very slow. All specific resistance tests were run at the original sample concentration.

Specific resistance (rs, in m/kg) is calculated by the following formula:

2PA2b r*

=- (1)

/JLW

Where:

P = the applied vacuum (Pa), A = the filter area, cm2,

fi = the viscosity, g/cm s, and

w = the mass of solids in the filter cake per unit volume of

filtrate, g/cm3.

Unit conversions are necessary to obtain the desired units. The

pressure and area values were constant throughout the experi

ments and the viscosity was taken as that of water at the measured

temperature. The values of b and w were calculated using in

formation from the plot of t/Vversus Fas shown in Figure 1. The slope of the straight portion of the graph is the value b (s/ cm6). The value of w was calculated as the product of the initial total solids (TS) concentration of the sludge and the ratio of the

0 10 20 30 40 50

FILTRATE VOLUME (cm3)

Figure 1?Plot used in determination of specific resistance. (Data shown

are from effluent of standard reactor in Set 3.)

initial volume of the sample to the volume of the filtrate at the end of the straight portion on the plot ( Vf on Figure 1 ). In this

work, the total solids concentrations were used rather than sus

pended solids because they were routinely measured in this re

search; the error is less than 5% and consistent for all samples. Preliminary experiments to test the validity of specific resis

tance measurements showed that the measurements were re

producible, independent of solids concentration (0.79 to 2.36%

sludge tested), and independent of the volume of sample used

(30 to 70 mL tested). Nevertheless, some difficulties were en countered in determining the b and Vf values for high concen tration sludges and activated sludges. For high concentration

sludges, the filter sometimes clogged, so that water no longer

flowed through the cake and accumulated above the cake. Such

ponding destroys the accuracy of the measurement because Vf is underestimated, so that specific resistance is also underesti

mated. To prevent this ponding, the applied sludge volume was

reduced to decrease the solids applied to the filter. A second

difficulty, encountered particularly when measuring the specific resistance of activated sludge, was previously defined by Chris tensen and Dick21 as "nonparabolic behavior." In such a test,

no well-defined straight line portion exists on the t/ V versus V

plot, so that neither the b nor Vf values are known well. No

solution was found for this problem during this research; mea

surements that were subject to these errors are noted.

Particle size distributions. The particle size distribution mea surements were made using an electrozone type instrument

equipped with a logarithmic amplifier and a 100-channel pulse height analyzer. Twelve and 30 pm apertures were used on all

samples, together with two ( 140 and 280 ^m) or three (100,200, and 560 pm) other aperture sizes. With this equipment, ex

tremely detailed measurements of the particle size distributions were obtained over a broad size range, with a lower limit of detection of 0.66 pm. Procedures for measurement are reported

elsewhere.22,23

RESULTS

Digester performance. Representative data for gas production,

volatile solids content, and total solids concentration are shown

in Figure 2. These data are from experimental Set 4 in which the influent solids concentration was varied between reactors.

Results from the reactor fed 6.5% influent sludge are not shown because a mechanical failure occurred after 20 days. Although that digester performed well before the failure, it never operated successfully after the repair, apparently because of the high load

ing rate. Results from the reactor fed 3.5% influent sludge are omitted from the solids measurements on the figure for the sake of clarity; in all cases, the results were between those shown for the 2% and 5% reactors and followed the same trends. The results for all three measures shown in Figure 2 indicate that reasonably steady conditions were achieved within 15 days of start-up. Gas

production was approximately proportional to the influent solids concentration. Influent sludges had total solids concentrations

slightly less than their nominal values. The results for both types of solids measurements show that significant digestion (destruc tion of solids) occurred.

The performance of all digesters in this research is summarized in Table 2. The averages reported were obtained using data after 30 days of operation, except in the case of the 6.5% reactor

which failed before that time. Several observations can be made

December 1986 1109



Lawler et al.

0 10 20 30 40 50 60 TIME (days)

Figure 2?Monitoring of digester performance in set 4 with varying in

fluent solids concentration. (Detention time 10 days, temperature 35?C, 20% by weight activated sludge in influent.)

from the data in this table. First, again with Set 4 as an example, the reductions in total solids, volatile solids, and COD for the

digesters fed 2.0, 3.5, and 5.0% total solids all reflect excellent

digestion, and show little variation between the reactors. The

data for the reactor fed 6.5% sludge, taken between 10 and 20

days, do not suggest that the reactor would have failed.

Second, within each set of experiments, the same trend of

little variation in any of the percent reductions can be seen. This

suggests that digester performance is quite insensitive to the

changes imposed. The principal exception to this is Set 3, in which the digester with 100% activated sludge as influent exhib ited notably poor performance and the digester that received

only primary sludge (0% activated sludge) performed better than the digesters that were fed mixtures. The small amount of diges

tion in the 4?C reactor in Set 3 and the first stage digester in Set 5 was expected; the similarity between these two reactors in all

measures taken in this research suggests that the 4?C reactor

behaved as a slow first-stage reactor.

Third, both the total solids and volatile solids reductions were less in Sets 1 and 2 than in the other sets. The result for the total

solids was expected because the volatile solids content was low

in the influent sludge for those two sets (62 and 55% of the total

influent solids in Sets 1 and 2, respectively, compared to 78 to

84% in the remaining three sets). The lower volatile solids re

duction in Sets 1 and 2 suggests that inorganic (nonvolatile) solids shield volatile solids and slow or prevent digestion of some

volatile solids.

Despite the exceptions and variations noted, the major con

clusions from the data in Table 2 are that laboratory digester performance was excellent and was relatively insensitive to vari

ations in operation.

Effects of digestion on dewaterability. Determining the effects

of anaerobic digestion on dewaterability was one objective of

this study. The data for specific resistance during Set 4 for three reactors are shown in Figure 3. For influent sludges, specific

resistance measurements at the different concentrations are ap

proximately equal, which confirms that this measure is inde

pendent of concentration. The results for reactors with influent

sludges at 2.0 and 3.5% total solids concentration are typical for

this experimental work. The most important observation is that

the effluent specific resistance was much less than that of the

influent; digestion significantly improved the dewaterability of these sludges.

The results shown in Figure 3 for the reactor with 5% total

solids in the influent are quite atypical, but nevertheless illustrate

two points. First, the variation for effluent samples was the worst

for any reactor in this research. These samples often exhibited

the ponding effect or nonparabolic behavior, or both, which are

the two significant causes of uncertainty in the measurement of

specific resistance. These errors lead to an underestimation of

the true specific resistance. Second, the average effluent specific

resistance (omitting the first two points because steady-state per

formance was not yet achieved) was worse than the influent for

this reactor. When considered with the results for the other two

reactors in Set 4, the confusion in the previously published lit

erature about the effects of digestion on dewaterability becomes

understandable. Performance of the three digesters, as measured

by the parameters listed in Table 2, was equally good, so that

digestion performance alone cannot explain the variation in de

waterability.

The average results for dewaterability for all of the experi

mental conditions are shown in Table 3; CST values reported are those from the standard solids concentration of 2.1%. Ratios

of the effluent to influent specific resistance and capillary suction

time are almost all less than 1 ; in those cases, anaerobic digestion

improved dewaterability. Of the 21 cases reported, there are six

exceptions to this rule, all from digesters that were considered

to be operated under stressed conditions. They are from the

digesters in Sets 1 and 4 with the highest influent solids concen

tration, the digesters in Set 3 with 40 and 100% activated sludge in the influent, the digester operated at 4?C, and the first stage

digester in Set 5. Performance of the latter three was significantly worse than the others; the first three achieved satisfactory diges

tion as measured by solids and COD reductions, but were near

the limits of operation of those digesters that did not perform well. It is clear that good digestion generally leads to improved

dewaterability, poor digestion leads to worse dewaterability, and

dewaterability is more sensitive to variations in digester design

and operation than digestion itself.

Trends in dewaterability within each set of experiments are

enlightening. Dewaterability of the digested sludge in Sets 1 and

4, worsened with increasing concentration. The differences be

tween the digesters fed 2.0 and 3.5% total solids are not statis




Focus on Dewatering

Table 2?Effects of operating variables on digester performance.

VS loading,

kg/m3d pH

%TS reduction

% VS reduction

% COD reduction

m3 CH4/kg VS added

Set 1: Total solids concentration, %

2.0

3.5

3.5

5.0

6.5

Set 2: Detention time, days 7

10

14

21

Set 3: % Activated sludge 0

20

40

100 Set 3: Temperature, ?C

4


2.0

35

5.0

6.5

Set 5: Staged operation Standard

1 st stage 2nd stage Overall 2-stage

0.58

0.98

0.98

1.41

1.91

2.67

1.87

1.34

0.89

2.86

2.62

2.57

2.02

2.73

1.59

2.75

3.94

5.30

2.73

13.7

2.84

2.73

7.4

7.4

7.4

7.5

7.5

7.2

7.2

7.2

7.3

7.2

7.2

7.4

7.2

7.1

7.3

7.4

7.1

7.3

5.3

7.3

34.8

34.3

30.7

30.3

34.8

28.6

27.7

29.8

30.1

56.1

48.5

48.4

25.0

12.2

49.6

54.6

52.5

53.9

50.2

15.4

36.7

46.5

52.9

50.4

49.1

50.1

52.5

47.0

48.5

50.8

52.6

63.4

56.8

54.9

31.9

6.9

58.3

60.3

59.4

59.6

58.8

16.6

45.4

54.5

60.4

59.8

58.7

58.6

58.5

53.8

56.3

56.9

59.8

64.0

63.1

58.6

33.8

2.2

61.6

61.9

60.3

55.9

61.5

4.3

52.6

54.7

0.43

0.45

0.47

0.46

0.48

0.42

0.43

0.44

0.45

0.52

0.47

0.46

0.23

0

0.52

0.48

0.50

0.44

0.003

0.48

0.51

I_I-1-1-1-1_I 0 10 20 30 40 50 60

TIME (days)

Figure 3?Dewaterability of raw (influent) and digested (effluent) sludges in set 4. (Detention time 10 days, temperature 35 ?C, 20% by weight

activated sludge in influent.)

tically significant but differences between the other reactors are

significant. The trend of decreasing specific resistance with in

creasing concentration for the feed sludges in Set 1 is thought to be caused by measurement error; the final procedures for

measurement of specific resistance were not adopted until the

end of that set. This trend is not found in the CST results for

Set 1 nor is it statistically significant in the specific resistance

results for Set 4. Within Set 2, the difference between the dewaterability of

feed sludge and digested sludge was statistically significant. However, the trend of increasing specific resistance with longer detention time in the digested sludges was not statistically sig nificant at the 0.05 level. This trend is not as obvious in the CST

measurements. It is important to note that both measures of

dewaterability for the influent sludge were low in this set in

comparison to the subsequent sets, apparently as a result of the

low volatile solids content.

Within Set 3, the results are quite complex. The results of

each reactor after digestion were different, and the best dewa

terability was achieved by the reactor that received the standard

mixture of 20% activated sludge and the remainder primary

sludge. As previously noted, in reactors that received 0% and

20% activated sludge digestion improved dewaterability, whereas

in the others dewaterability deteriorated. In both the first stage digester in Set 5 and the reactor at 4?C

in Set 3, dewaterability became significantly worse with digestion,

apparently because digestion was incomplete in both reactors.

The nearly identical response of these two reactors is further

indication that the digester operated at 4?C performed as a slow

December 1986 1111



Lawler et al.

Table 3?Sludge dewaterability: effects of digestion.

Specific resistance Capillary suction time

Influent,

1013

m/kg

Effluent,

1013

m/kg

Effluent:influent

ratio Influent, s Effluent, s

Effluent:influent

ratio


2.0

3.5

3.5

5.0

6.5


10

14

21


20

40


4


2.0

3.5

5.0


1 st stage 2nd stage overall 2-stage

12.38

10.77

10.77

9.42

8.40

8.38

13.34

16.67

15.00

15.24

16.67

23.30

22.80

19.55

16.06

16.06

32.83

16.06

6.81

9.81

6.76

7.52

8.50

2.96

3.33

4.85

6.28

10.56

3.15

19.36

26.37

30.46

4.89

5.76

26.83

1.01

32.83

0.41

0.41

0.55

0.91

0.63

0.80

1.01

0.35

0.40

0.58

0.75

0.79

0.19

1.29

1.73

1.83

0.21

0.25

1.37

0.06

2.04

0.01

0.03

162 186 186 188 189

125

197 242 210 262

242

139 155 188

145 145 283 145

41 93 44

151 218

32 35 35 45

115 54

266 371

484

45 60

296

27 283

16 16

0.26

0.50

0.24

0.80

1.15

0.25

0.28

0.28

0.36

0.64

0.22

1.27

1.42

2.00

0.32

0.38

1.57

0.18

1.96

0.06

0.11

first-stage reactor. Dewaterability after the second stage is slightly better than that of the standard reactor in the same set, although the difference is only statistically significant for the CST. The results of the two-stage operation suggest, as expected from the

ory, that the first stage broke down large particles into smaller

fragments, and that these fine particles were subsequently con

verted into soluble products and methane in the second stage.

Particle size distribution results, presented subsequently, support

this conclusion.

The ratios of both specific resistance and CST lead to the same conclusions and show the same trends with each variation

in operating conditions; nevertheless, there is not good numerical

agreement between the two measurements. This disagreement stems from fundamental differences in the two measurements.

Kavanagh19 mathematically demonstrated that, for samples

measured at their original concentrations, the CST (corrected

by subtracting the CST of pure water) should be proportional to the product of the specific resistance and w.19 These results

confirm that relationship; when plotted as shown in Figure 4, the slope was 1.09 which compares favorably with the predicted value of 1.0. The relationship also makes clear that CST is a

measure of dewaterability that focuses on the volume of water

removed per unit time, whereas specific resistance focuses on

the mass of solids that can be dewatered per unit time. Because

dewatering devices must be designed both with sufficient hy draulic loading capacity and sufficient solids handling capacity, both measures (and their trends in this work) are important.

o

160

15.5

ISO

145

140

135

130

n-r

Regression Analysis

Slope? 1.09 R2 - 0.88

n-r

05 1.0 1.5 2.0 25 3i0

LOG CST1

Figure 4?The relationship between sp?cifie resistance and capillary suction time. (CST in seconds, modified by subtracting the CST of water;

specific resistance in m/kg, influent and effluent sludges from Set 4

shown.)





Effects of digestion on particle size distributions. The detailed

particle size distributions measured in this research yield several

insights into the destruction of solids during digestion and the effects of digestion on dewaterability. The destruction of solids of different sizes is most clearly illustrated by the volume dis

tribution; a few examples are shown in Figure 5. The area under the curve between any two sizes (log diameter values, log dp) represents the volume concentration (/?m3/cm3) of particles in

that size range. Influent sludge for the standard reactor (20% by weight activated sludge) had a bi-modal distribution. The ma

jority of the particle volume (mass) was located in the range 10

/xm < dp < 100 ?mi ( 1.0 < log dp < 2.0), but a significant fraction in the measured size range less than 2.5 ?urn (log dp < 0.4). This

distribution was typical of all sludges studied with the standard mixture of primary and activated sludge. The 100% activated

sludge influent, also shown in Figure 5, was significantly different, with a much greater fraction of its particle volume in the large size range, as might be expected.

The destruction of particles in digestion is obvious in the re sults from the standard reactor shown in Figure 5; a significant reduction of the volume concentration occurred in both parts

of the bi-modal distribution. Overall, most of the destruction occurred in the large sizes because that is where most of the

particle volume was located. However, the fractional reduction

in the small particle size range was even greater. This overall

reduction throughout the entire particle size range was typical

of most reactors in this study.

As might be expected from the digestion and dewatering re

sults, the destruction of particles in the reactor fed 100% activated

sludge did not conform to the standard pattern, as indicated in

40

30h

E

o

I

?? tr

20 h

60

50

40

30

20

10

20% Activated Sludge Reactor

o Influent o Effluent

cb

*?

<?% ^? a a

?DD ?Q?

df?

' $>


o Influent

o Effluent

m T3?

-03 0.0 03 0.6 0.9 1.2

LOG OF PARTICLE DIAMETER

Figure 5?Effects of digestion on volume distribution. (Particle diameter,

dp, measured in ??m; data shown are from Set 3.)

120

KX) O ^

n ? O

8

0?<1

80

? 60h <

40

20 H

?bo n *

? a

O CD OP

OD

\

T ~V

Influent

Effluent Acid-Forming Stage Effluent' Methane-Forming Stage

0 -0. 0.0 03 Q6 0.9 12 15

LOG OF PARTICLE DIAMETER

1.8 2.1

Figure 6?Surface area distributions during 2-stage digestion. (Particle

diameter, dp, measured in ?im; all three sets of results essentially identical

for log dp > 0.75.)

Figure 5. A significant loss of large particles occurred as in the

typical cases, but the particle volume increased rather than de creased in the small size range. Apparently, some large floes

were broken in the digester but not destroyed, so that the loss of large particles represents both destruction of solids (digestion) and transfer from large to small particles.

On a number basis, virtually all particles were less than 2 pm in size (log dp

= 0.30) for both influent and effluent samples,

for all reactors. All digesters except the one fed 100% activated

sludge and the acid phase digester showed a decrease in particle number concentration. Complete results of the volume and

number distributions are not shown, because neither is consid

ered as important as surface area.

Surface area distributions. For both dewaterability and diges tion, the surface area distribution of sludge can provide more

important information than the other forms of the size distri bution. The surface area of the particles affects the dewaterability by providing frictional resistance to the withdrawal of water and a surface to which water can bind. The rate that particles can

be hydrolyzed to a size small enough for bacterial uptake also is related directly to the available surface area.

The surface area distributions of the three sludges from the

two-stage operation in Set 5 are shown in Figure 6. Analogous

to the volume distribution, the area under the curve between

two log diameter values on this plot represents the surface area

concentration (/?m2/cm3) of the particles between those two sizes.

For all samples, the surface area was dominated by particles less

than 2 pm in diameter (log dp < 0.3); 100% activated sludge samples had a noticeable but still small fraction of particles larger than 10 /urn (log dp > 1.0). As is obvious in the figure, the total

surface area increased slightly in the acid-forming (first) stage and then decreased substantially in the methane-forming (sec ond) stage. The overall change from the influent to the effluent

(second-stage digester) is typical of that seen in the single-stage reactors. The increase in surface area in the acid-forming stage

occurred in the very small particles (log dp < 0.04, or dp < 1.1

um). This is consistent with the view that anaerobic digestion breaks large particles into small particles and subsequently con

December 1986 1113



Lawler et al.

verts small particles into gaseous forms. The same presumably

occurs in single-stage digesters, but the separation of the two

steps in the two-stage process makes it clear.

Average values of the total surface area concentration for each

digester are shown in Table 4; ratios of the effluent to influent

values are included for comparison. Also shown are the specific surface area concentrations and ratios for this parameter. Results

in the latter columns of Table 4 are discussed subsequently. Several trends are obvious in the results for the surface area

concentrations. First, the ratios of the effluent to influent values

are almost all less than 1, which indicates that the surface area was decreased through digestion. Again the exceptions are the reactor fed 100% activated sludge (Set 3), the reactor operated at 4?C (Set 3), and the first-stage reactor (Set 5).

Second, the results from Sets 1 and 4 indicate that the reduc tion in surface area lessened as the influent solids concentration

increased. The results for the reactor fed 5% sludge in Set 4

suggest that this reactor was under stress; apart from the three reactors with ratios greater than 1, this reactor had the highest ratio of effluent to influent surface areas. The stressed condition of the reactor is clear from the particle size distribution mea

surements (and dewaterability data) even though it was not per

ceptible in the overall performance indicated by the measures

reported in Table 2.

Third, the data from Set 3 with varying sludge composition showed trends similar to the dewatering results: the standard reactor had the greatest destruction of surface area and the ac

tivated sludge reactor had no overall reduction. In the latter

case, some surface area reduction occurred in the particles larger

than 2 /?m, but the surface area increased in the smaller size

range. The small particles produced from the large particles could not be digested well under the operating conditions; perhaps a

longer detention time would lead to better destruction.

Fourth, a significant increase in the surface area occurred in

the reactor operated at 4?C. Operation of this digester was de

signed to test the effects of mixing alone on dewaterability and

particle size distributions, but the monitoring results suggest that this reactor operated as a slow first-stage (acid-forming) reactor;

therefore this study failed to isolate the effects of mixing. The relationship between particle size distributions and de

waterability. The similarity in trends of the dewaterability and surface area distribution results suggests that surface area is a

principal determinant of dewaterability. The Kozeny equation provides a basis for testing this relationship quantitatively. As shown in a previous study,24 the Kozeny equation can be written

in the form

KS0\\-e) ...

rs =-i- (2) ? Pp

Where

rs =

specific resistance, m/kg, =

fractional voidage of cake,

K = Kozeny constant,

Table 4?Particle surface area: effects of digestion.

Surface area concentration Specific surface area

Influent,

107 Mm2/cm3

Effluent,

107 Mm2/cm3

Effluentinfluent

ratio

Influent, Effluent,

1011 ?tm2/g 1011 /tim2/g

Effluentinfluent

ratio

Set 1 : Total solids concentration, %

2.0

3.5

3.5

5.0

6.5


10

14

21


20

40


4


2.0

3.5

5.0


1 st stage 2nd stage overall 2-stage

20.04

35.24

35.24

50.48

63.33

27.68

41.24

40.32

39.43

25.10

40.32

24.92

40.26

59.51

40.74

40.74

42.21

40.74

8.12

20.36

15.60

27.17

41.77

13.85

14.84

12.45

15.40

19.27

10.81

18.56

25.20

48.18

6.33

18.16

44.46

17.36

42.21

21.91

21.91

0.41

0.58

0.44

0.54

0.66

0.50

0.54

0.45

0.56

0.47

0.27

0.47

1.00

1.19

0.25

0.45

0.75

0.43

1.04

0.52

0.54

10.64

10.75

10.75

10.77

10.22

8.19

11.76

12.27

11.91

10.09

12.27

11.80

12.28

12.29

11.78

11.78

14.69

11.78

6.25

8.74

6.87

7.99

9.90

5.50

6.05

5.23

6.46

12.60

5.00

11.24

13.74

15.83

6.89

12.19

20.72

9.52

14.69

11.78

11.78

0.59

0.81

0.64

0.74

0.97

0.68

0.74

0.64

0.79

1.07

0.47

0.94

1.36

1.29

0.58

0.99

1.69

0.81

1.25

0.71

1.00




Focus on Dewatering

pp =

particle density, g/cm3, and

S0 =

surface area per unit volume of particles, pm2/pm3.

As shown in Equation 2, the specific resistance is related to the

particle density and, through S0, particle size and shape. No

attempt was made to determine the shape of the particles in this

study; in the analysis of measurements from the electrozone

particle counter, it is assumed that the particles are spheres. The

few particle density measurements made in this research were

considered unreliable. However, it is reasonable to assume that

neither the particle shape nor particle density varied as much

as the size distributions among the sludges used in this study. With that assumption, differences in specific resistance between

samples should be primarily related to differences in S0, ac

cording to the Kozeny equation.

S0 could be obtained as the ratio of the total surface area

concentration to the total volume concentration from the particle

size measurements. However, because of some unreliability in

the 560-/im aperture measurements (which affects the total vol

ume significantly but not the surface area because it measures

the largest particles), it was more reasonable to normalize the

surface area distributions by the measured solids concentrations.

The resulting value is termed the specific surface area (As), and

is related to S0 by Equation 3

S0 = Aspp (3)

This substitution requires the assumption that the particle density was constant across the size range.

Specific surface area distributions were plotted for all samples (Figure 7). The area under such curves represents the total surface

40

30

5 'o q: <n

Q o <60 ??

UJ

i 2?40 V)

^30 u. o ?J fe 20

10

U jp1*

-1-1-1-r


Influent o Effluent


a Influent o Effluent

cP

ft

-03 OJO 03 0? 03 12 15 IS 21 LOG OF PARTICLE DIAMETER

Figure 7?Effects of digestion on specific surface area. (Particle diameter,

dp, measured in pm; data shown are from Set 3.)

i-1-r

Regression Analysis

Slope- 1.82 o

R2- 0.85 * ,

a OMlttee hi R?er?Mi?i ? ?

/

? P

/o o

o/ Q)

oof o / / a

o /o

/?? / ?

J-oL_b-1-1 0 0.25 0.5 0.75 1.0 1.25 150

LOG OF SPECIFIC SURFACE AREA

Figure 8?Effects of particle size on dewaterability. (Specific surface

area, As, in 1011 Mm2/g; specific resistance, rs, in 1013 m/kg; results shown

are averages from each digester.)

area per unit mass of solids. The significant reduction in specific surface area in the standard reactor in Set 3 and the increase in this parameter in the reactor fed 100% activated sludge are ob vious in the figure. If the differences in solids concentration be tween two samples were caused simply by dilution or concen

tration, the two distributions would form a single line on these

plots. These plots were useful in comparing the influent to ef fluent particle size distributions, because the pattern of destruc tion of solids (particles) as a function of size is made clear. For almost all reactors, there was a net reduction in the specific sur

face area through digestion, as reflected in the ratios (less than

1) shown in the last column of Table 4. The interpretation is

that, in such cases, there was a selective destruction of small

particles, that is, a greater reduction in surface area (created

mostly by small particles) than mass or volume (distributed bi

modally in small and large particle sizes).

Substituting Equation 3 into Equation 2 yields

^izo (4) e

The relationship in Equation 4 suggests that a log-log plot of

specific resistance (rs) versus specific surface area (As) should

yield a straight line with a slope of 2. The average results for both influent and effluent of all reactors for specific resistance

(from Table 3) and specific surface area (from Table 4) yielded a straight line with a slope of 1.82, as shown in Figure 8. Three

points were considered outliers and were omitted in the regres

sion analysis. Two were from Set 5 (the standard reactor and

the second-stage reactor) and one from Set 4 (standard reactor); all had excellent dewaterability but no explanation could be

found for these exceptions. The close agreement between the

experimental results and the expectations from the Kozeny

i.i w

l? O z

?o CO ?J o:

o

o

o o

1.50 r

1.25

LOO

0.75

Q50

0.25

December 1986 H15



Lawler et al.

equation (including the assumptions about density in the analysis noted previously) confirms that the particle surface area is a

principal determinant of sludge dewaterability.

SUMMARY

Relationships among anaerobic digestion, dewaterability, and

particle size distributions were investigated and the relationships between pairs of the three factors have been presented. Figure 9 ties all three topics together. The form of the plot is suggested by the Kozeny equation and relates the square of the specific surface area to specific resistance; digestion is accounted for by using the ratios of the effluent to influent for these quantities.

Three lines are drawn on the figure in addition to the data

points that represent each reactor. The horizontal and vertical dotted lines represent no change through digestion in specific resistance and specific surface area, respectively. All but a few

reactors fall within the box formed by these two lines, which demonstrates that both specific resistance and specific surface area were reduced by digestion. If the changes in specific resis tance brought about by digestion were accounted for entirely by the changes in the specific surface area, the points would fall on the line with a slope of 1. Results for virtually all reactors fall

reasonably close to this line, especially when the level of uncer

tainty in the four measures that are necessary for each point and

the squaring of the abscissa values are considered. The only ex

ceptions are from the three digesters with excellent dewaterability omitted from the analysis of Figure 8 and from the digester fed 5% solids in Set 4; as noted earlier, there was considerable un

certainty in the specific resistance measurements for that digester.

The figure makes clear that digestion generally improves de

waterability and does so by the selective destruction of small

particles that contribute relatively more surface area than mass

in sludge samples. When digesters are stressed, they can still accomplish some

digestion but can create rather than destroy small particles. The four highest points on Figure 9 correspond to the reactors that

were under the most stress from a biological viewpoint, the first

Asz , digested/As2 , raw

Figure 9?Interrelationships among anaerobic digestion, dewaterability, and particle size characteristics.

stage reactor, the 4?C reactor, the reactor fed 100% activated

sludge, and the reactor fed 5.0% sludge in Set 4.

Implications for full-scale digesters. This research was carried out in small laboratory digesters under well controlled conditions, but some implications for design and operation of full-scale di

gesters are obvious. Most digesters are followed by dewatering devices, and dewatering is usually improved by polymer addition. It is reasonable to think that the optimal polymer doses are pro

portional to the surface area of particles because it is primarily polymer interactions with the surface of the particles that causes

effective flocculation. The effects, therefore, of digester operating variables on particle characteristics will affect polymer usage and costs.

Digesters should be designed and operated not only for sludge stabilization, but to improve subsequent dewaterability; the range of operating conditions that can lead to good digestion is wider than the range that can lead to good dewaterability. Designers should recognize that saving space by a reduction in detention time or by an increase in influent sludge concentration might ultimately be more costly; larger dewatering units and more

polymer are likely to be necessary to process the sludge after

digestion. Also, digestion of primary and activated sludges should not be separated.

Operators might find that dewatering problems are reduced

by better operation of digesters. The greater sensitivity of de

waterability than digester performance to operational variations

suggests that dewaterability measures should be added to the list of monitoring tools, and could yield earlier warning of stress in a digester.

CONCLUSIONS In carefully-controlled laboratory reactors, anaerobic digestion

generally improved sludge dewaterability. Operating variables

of digesters, including detention time, influent solids concentra

tion, and the mixture ratio of primary and activated sludges, influence digested sludge dewaterability. Dewaterability is more

sensitive to digester operating variables than is digestion itself.

Dewaterability is commonly measured by either CST or spe cific resistance, but there are fundamental differences in these

measures. CST is more aptly suited for plants whose dewatering

devices are limited by their ability to process the volume of

sludge; specific resistance is better in cases where the limitation

is in processing the mass of solids in the sludge. Anaerobic digestion changes the particle size distribution of

sludges. When digestion works well, particles of all sizes are de

stroyed, but there is a preferential removal of particles of small

sizes, a consequent loss of specific surface area, and therefore

an improvement in dewaterability. When digestion does not

work well, large particles are destroyed but small particles are

created with a consequent gain in specific surface area, and

therefore a worsening in dewaterability. Specific surface area

(jum2/g) is the principal determinant of sludge dewaterability, as

expected from the Kozeny equation.

ACKNOWLEDGMENTS Credits. The research reported in this paper was supported

by the U. S. Environmental Protection Agency, Office of Re search and Development, under grant R-810147-01-0. The con

tents of the paper do not necessarily reflect the views and policies of the Environmental Protection Agency. Donald F. Carey was

project officer.





Howard M. Liljestrand and Joseph F. Malina, Jr. suggested details of the measurement methodology and operation of lab

oratory digesters. James Stewart built the laboratory digesters and improved their design. Frank Husley solved many opera tional problems with the measurement instruments and digesters.

This paper was presented at the 57th Annual Conference of the Water Pollution Control Federation, New Orleans, Louisiana

(September 30 to October 4, 1984). Authors. Desmond F. Lawler is an associate professor in the

Department of Civil Engineering at the University of Texas, Austin. At the time the research was performed, Yoon Jin Chung,

Shiaw-Jy Hwang, and Barbara A. Hull were graduate students

in the Department of Civil Engineering at the University of

Texas, Austin. Dr. Chung is now an assistant professor in the

Department of Environmental Engineering at Ajou University in Suweon, Korea. Ms. Hwang is an engineer with the Los An

geles County Sanitation District, Los Angeles, Calif. Ms. Hull is an engineer with Turner, Collie, and Braden, Inc., Austin,

Tex. Correspondence should be addressed to Dr. Lawler at the

Department of Civil Engineering, ECJ 8.6; University of Texas at Austin; Austin, TX 78712.

REFERENCES 1. McCarty, P. L., "History and Overview of Anaerobic Digestion."

Second International Symposium on Anaerobic Digestion, Treve

munde, Germany (1981). 2. Speece, R. E., "Anaerobic Biotechnology for Industrial Wastewater

Treatment." Environ. Sei. TechnoL, 17, 9, 416A (1983). 3. Metcalf and Eddy, Inc., "Wastewater Engineering: Treatment, Dis

posal, and Reuse." Second Ed., McGraw-Hill, Inc. New York, N. Y. (1979).

4. Rudolfs, W., and Heukelekian, H., "Relation between Drainability of Sludge and Degree of Digestion." Sew. Works J., 6,6, 1073 (1934).

5. Brooks, R. B., "Heat Treatment of Sewage Sludge." Water Pollut.

Control (G. B.), 69, 1,92(1970). 6. Haug, R. T., et ai, "Effect of thermal pretreatment on digestibility

and dewaterability of organic sludges." J. Water Pollut. Control Fed.,

50, 1,73(1978). 7. U. S. Environ. Prot. Agency, "Process Design Manual: Sludge

Treatment and Disposal." EPA-625/1-79-011, U. S. EPA, Municipal Environmental Research Laboratory, Cincinnati, Ohio (1979).

8. Kini, A. D., and Nayak, S. L., "Optimizing Vacuum Filtration of

Sewage Sludge." Filtr. Sep., 17, 4, 313 (1980). 9. Pearson, E. L., and Buswell, A. M., "Sludge Ripeness Studies." Ind.

Eng. Chem., Analytical Edition, 3, 4, 359 (1931). 10. Morris, R. H., "Polymer Conditioned Sludge Filtration." Water

Works & Wastes Engineering, 2, 3, 68 (1965). 11. Bacon, V. W., and Dalton, E. F., "Chicago Metro Sanitary District

Makes No Little Plans." Public Works, 97, 11, 66 (1966). 12. Karr, P. R., and Keinath, T. M., "Influence of the particle size on

sludge dewaterability." /. Water Pollut. Control Fed., 50, 8, 1911

(1978). 13. Knocke, W. R., et al, "Vacuum Filtration of Metal Hydroxide

Sludges." J. Environ. Eng. Div., Proc. Am. Soc. Civ. Eng., 106, EE2,

363(1980). 14. Knocke, W. R., and Wakeland, D. L., "Fundamental Characteristics

of Water Treatment Plant Sludges." /. Am. Water Works Assoc,

75, 10,516(1983). 15. Faisst, W. K., "Characterization of Particles in Digested Sewage

Sludge." In "Particulates in Water." M. C. Kavanaugh, and J. O.

Leckie, (Eds.), Advances in Chemistry Series 189, American Chem

ical Society, Washington, D. C, 259 (1980). 16. Eastman, J. A., and Ferguson, J. F., "Solubilization of paniculate

organic carbon during the acid phase of anaerobic digestion." J.

Water Pollut. Control Fed., 53, 3, 352 (1981). 17. Chung, Y. C, "Interrelationships Among Anaerobic Digestion,

Sludge Dewaterability, and Particle Characteristics." Ph.D. Disser

tation, Dept. Civil Eng., Univ. of Texas at Austin (1985). 18. "Standard Methods For the Examination of Water and Wastewater."

15th Ed., Am. Public Health Assoc. (1981). 19. Kavanagh, B. A., "The Dewatering of Activated Sludge: Measure

ment of Specific Resistance to Filtration and Capillary Suction

Time." Water Pollut. Control (G. B.), 79, 3, 388 (1980). 20. Christensen, G. L., and Dick, R. I., "Specific Resistance Measure

ments: Methods and Procedures." J. Environ. Eng. Div., Proc. Am.

Soc. Civ. Eng, 111, 3, 258 (1985). 21. Christensen, G. L., and Dick, R. I., "Specific Resistance Measure

ments: Nonparabolic Data." J. Environ. Eng. Proc. Am. Soc. Civ.

Eng., 111,3,243(1985).

22. Lawler, D. F., et al, "Particle behavior in gravity thickening." J.

Water Pollut. Control Fed., 54, 10, 1388 (1982). 23. Lawler, D. F., et al., "Changes in Particle Size Distributions in Batch

Flocculation." J. Am. Water Works Assoc, 75, 12, 604 (1983). 24. Gale, R. S., "Filtration Theory With Special Reference to Sewage

Sludges." Water Pollut. Control (G. B.), 66, 6, 622 (1967).

December 1986 1117



Download - Anaerobic Digestion: Effects on Particle Size and Dewaterability

Top Related