2003 Residuals and Biosolids

Baltimore, Maryland USA

Mick Dawson

, Martin Tillotson*

Monsal,

BHR Group Limited,
*
Yorkshire Water
Tel: ++ 44 (0)1623 429500; Fax: ++ 44 (0)1623 429505; email:
aidancumiskey@monsal.co.uk; Web: www.monsal.com
ABSTRACT
Advances in the design of anaerobic digesters have seen a trend towards digestion of thicker
sludges in the US, UK and elsewhere. This includes a number of new facilities using bolt on
technologies to mesophilic anaerobic digestion such as Cambi Thermal Hydrolysis, Pre-
pasteurisation and Biological Hydrolysis. Cambi Thermal hydrolysis utilises thickened
sludge and the digestion process is fed at high DS%, typically 12-14%DS thereby
intensifying the digestion process and reducing reactor volumes.
There is limited experience in the design of digestion systems for such applications including
heating, mixing and pumping. As the trend continues towards advanced digestion
technology, a greater emphasis will be placed on these basic unit process operations. Monsal
have conducted fundamental research into digester heating and mixing technology in
collaboration with BHR and Yorkshire water. A number of plants have been designed using
design tools generated from the research. This includes the digester mixing systems for a
major sludge processing facility in Scotland, which is an advanced digestion facility
employing Cambi Thermal Hydrolysis followed by mesophilic anaerobic digestion. This
advanced facility treats a design EP of 562,000, processing 16,000 tds/year of hydrolysed
sludge utilising two 4000 m
3
digesters.
This paper looks at the design of the mixing technology for thick feed sludges for the full
scale plant. The background research is examined and reviewed. The conceptual design is
examined and reported. The full scale design and testing is also reported and examined being
compared with the design stage. The paper looks at the use of fundamental sludge rheology
to design effective systems for thick sludge digestion and reports on the effects of thermal
hydrolysis pre-treatment.
KEY WORDS
Thick sludges, Rheology, Biosolids, Monsal, Digestion technology, Mixing, Hydrolysis,
Advanced Digestion
INTRODUCTION
Digester mixing is a key unit process operation that has come under scrutiny in recent years
as the nature of the anaerobic digestion process changes to meet emerging industry trends.
Gas mixing is still the most widely accepted form of mixing of digesters in the UK. There are
wide variations in the design of gas mixing systems, with both confined and unconfined
systems in operation. However, it has generally been accepted that traditional forms of gas
mixing systems can be energy intensive. One of the key benefits of gas mixing systems has
always been the absence of moving parts inside the digester.
In recent years there has been a proliferation of large blade impeller based systems installed
in new digesters along with other forms of mechanical mixers such as the draft tube type.
One of the drivers for mechanical based mixing systems has been improved energy efficiency
compared with traditional gas mixing designs. With the improvements in sludge screenings
on the newer sludge plants, it is anticipated that previous operational problems such as
ragging will not occur, however it is still early days in the UK to get reliable, long term
experience.
Design of many of the gas systems were based on empirical correlations or rules of thumb
using:
Power input per unit volume (Wm
-3
)
Gas flowrate per unit volume of sludge (m
3
h
-1
m
-3
)
Gas flowrate per unit area of digester (m
3
h
-1
m
-2
)
This design methodology did not take into account the sludge characteristics (DS%,
rheology) or the dimensioning of the digester (aspect ratio, floor slope). In particular there
was little information available on mixing systems for thicker sludges, a trend which had
primarily started from 1995 onwards as part of the second UK water industry investment
cycle with a move from traditional sludge feed thickness to digesters of 4.0 DS% towards
6.0% and beyond.
In addition to thicker sludges becoming the norm, a number of proprietary processes became
available on the market. These include Cambi Thermal hydrolysis, pre-pasteurisation and
biological hydrolysis systems. Many of these processes seek to optimise the digestion process
and utilise thicker feed sludges, normally above 6.0% DS. The nature of processes also
results in significant changes to sludge rheology and these must be taken into account to
ensure correct design and operation of the digestion process. These greater demands on the
mixing technology have largely been un-quantified and fundamental work is required to
ensure that the implementation new processes are not constrained by the prevailing unit
processes of pumping, heating and mixing.
A collaborative research project was initiated between BHR Group Limited, Yorkshire Water
Services and Monsal aimed to characterise digested sludge rheology and investigate the effect
of sludge rheology, digester geometry and mixing system design on mixing performance at
both laboratory and pilot scales. A primary driver of this work was to build knowledge on the
rheology of thick sludges now becoming more prevalent in the UK and elsewhere. A number
of different mixing systems were investigated which included impeller mixing, jet mixing
and continuous unconfined gas mixing and sequential unconfined gas mixing systems. Some
of the findings from the rheology survey and laboratory scale work have been published
1,2
,
but several of the key findings and developments are reviewed for the purposes of this paper.
On the basis of this research work, a novel method of digester gas mixing was developed
the Monsal SGM system which is particularly suited to thick sludge digestion.
The Monsal SGM system differs from conventional gas mixing systems. It injects gas
through individual nozzles to provide high localised mixing energies. The sequencing order
and duration take into account the digester parameters including the feed positions, feed
duration, sludge recirculation position and outlet positions. This integrated approach to gas
mixing design now means that superior gas mixing systems can compete in terms of energy
efficiency with the mechanical based systems.
REVIEW OF RESEARCH
There were a number of key findings from the research:
Thick Sludge Rheology
The effect of digesting high dry solids sludges was investigated in this work. The objectives
of this work was to develop a better understanding of the rheological properties of sewage
sludge and provide good quality data which could be used for further work. The surveyed
digested sludges ranged from 2.5 to 5% DS. To provide thicker sludge samples a 10%DS
sludge was formed by evaporation in an oven at 35C. Three sludges (2.5% DS, 5% DS and
10% DS) were chosen as representative of the surveyed range of sludge rheologies and are
shown in Figure 1.
It can be seen that the increase in apparent viscosity between 2.5% and 5% DS is an order of
magnitude with a further order of magnitude increase between 5% and 10%. When designing
effective mixers, the ability to predict the influence of viscosity becomes more important in
the thicker sludge bands (5-10 DS%) because of the more arduous mixing duty.
Predicting Sludge Viscosity
Understanding the rheology of sludge to be mixed is essential and a key goal of the research
was to develop better predictive methods for estimating sludge viscosity.
The accepted UK industry standard prediction method is described in the Water Research
Centre (WRc) TR185 report
4
. Figure 2 shows an example of a comparison between the
measured rheology of a digested sludge and that predicted from TR185, in this case the
agreement between measured and predicted viscosity is poor. Generally, the TR185
predictive method is not very accurate for thicker digested sludges and a different model was
adopted in this work. Rheological properties were fitted using the Herschel-Bulkley model
1
and a modified predictive correlation arrived at from digested sludges.
As can be seen, the actual data for the sludge is considerably different to that predicted by
TR185 whereas the Herschel-Bulkley model closely matches the experimental data.
Effect of Gas Mixing System Configuration
The performance of two unconfined gas mixing methods (sequential and simultaneous), two
liquid jet orientations, and an impeller system were investigated and compared in terms of
blend time and active volume. The overall results are presented elsewhere
1,2
; only the gas
mixing systems will be discussed here. Two gas mixing configurations were investigated as
shown in Figure 3 below.
Sparger A provides a central core of diffusers which operate continuously. This configuration
is similar to the standard adopted by a number of the UK water companies. The continuous
gas flows simultaneously through all the nozzles generating an axial flow mixing pattern.
Sparger B provides two cores, an outer and inner. However the gas flow is injected
individually through each nozzle for a period of time then sequenced in a pre-determined
way.
The results from research were as follows:
At equal net power inputs, with well spaced diffusers, sequential gas mixing achieves
greater active volume and shorter blend times than a continuous system.
The superiority of sequential gas mixing increases as the sludge DS% and hence
viscosity increases.
Nozzle sequence, gassing duration per nozzle and sequence integration with feed
cycle are important.
Case Study : Thick Sludge Digestion
The plant is an advanced digestion facility employing Cambi Thermal Hydrolysis followed
by mesophilic anaerobic digestion. This advanced facility treats a design PE of 562,000,
processing 16,000 tds/year of hydrolysed sludge utilising two 4000 m
3
digesters. The project
is a Private Finance Initiative (PFI) and will be operated by a consortium for 25 years. Energy
costs therefore become more significant.
The sludge centre is a regional plant processing a number of sludges from satellite works,
imported sludges and the indigenous sludge produced at the new works on site (table 1).
The contractual mixing system performance criteria were:
1. Feed sludge dispersal within 120 minutes
2. Active volume > 90%
3. Less than 5% feed volume short circuiting
The design feed sludge is thermally hydrolysed and fed to the digesters at 37-40
o
C with 10-
12DS%.
Thermal hydrolysis is a pre-treatment process heating raw sludge to approximately 165
o
C
and 6 bar pressure prior to digestion. The high temperatures and pressure result in cell lysis.
The resultant hydrolysed sludge is fed to the digester at much higher DS%, typically 12%
resultant in high organic loading (see table 2). The digesters on this plant are of a modern
design with a good aspect ratio (1: 1) for mixing (see figure 8). This combination of high feed
solids coupled with a need to minimise power use led to the client selecting the Monsal
sequential gas mixing system which could be designed using the available rheological
information from similar plants.
Hydrolysis changes the properties of sewage sludge. Rheology data provided by the main
contractor is shown in Figure 5. These parameters were input into the mixing model derived
in the stage one research. In general the hydrolysis process changes the rheological properties
of sludge, the resultant effect of hydrolysis is to effectively make the sludge behave as if it
were thinner than its equivalent non-hydrolysed counterpart. This is shown in Figure 5 where
a 12% hydrolysed sludge is compared with two digested sludge samples at 7-8%DS. It can be
seen that the rheological properties are similar. This approach has allowed an optimised
solution to be designed to meet the clients needs on this site, in particular with achieving the
desired active volume (>90%) at low mixing energies.
During the design, as in the other case studies, an integrated energy input approach was
adopted with particular attention paid to dispersing this thick feed sludge. Hydrolysed sludge
enters the digester at 40C and there is considerable danger of an 'inactive zone' developing
as the hot feed accumulates on the surface. This problem also occurs in conventional
digesters and has been shown to result in a high acid concentration zone in the digester and
serious foaming and other operability problems including untreated sludge short-circuiting to
the digester outlet.
The design information for the mixing system is presented in table 3
DESIGN
Application of the research has been used in the following way:
Selection as sequential mixing as preferred option for thicker sludges
Use of BHR Groups Herschel-Buckley model to more accurately predict sludge
rheology
Use of predictive models for active volume and blend time
Use of flow visualisation techniques to improve nozzle positioning
The approach that the partners have taken is to produce a mixer-sizing model that
incorporates rheological data, either taken from measurement or predicted using the yield
stress prediction model produced within this project. Both approaches have been used in the
case studies that follow.
The design of the mixing systems used an integrated approach taking into account the feed
positions, outlet positions and sludge re-circulation lines.
Sequential Gas Mixing is carried out using high efficiency sliding vane compressors and gas
is injected into the digester via the use of suitable gas solenoids. Control of sequencing is via
a PLC and both cycle times and duration of opening can be altered. This allows more
flexibility to fine tune the system during plant commissioning.
TESTING
Tracer studies have been used in the water industry to determine the actively mixed volumes
and flow patterns within operational sewage sludge digesters. In addition, tracer tests are
used to demonstrate effective performance of newly commissioned digester mixing systems.
The mixing performance of the digester was tested during November 2001.
The digester tested is cylindrical in shape and constructed with a sloping base. The installed
working volume of the digester is reported to be 4000m
3
. The digester feeding system is
pump-in, overflow-out, with a single raw sludge entry point near the base and a single
digested sludge over flow pipe located within a concrete chamber close to the digester roof.
The digester feed pump is reported to run in an automatic sequence of 15 minutes ON 15
minutes OFF.
The main objectives of the work were as follows:
! To conduct a tracer test on one of the two live digesters at Nigg WWTP, under normal
operating conditions.
! To determine the blend time within the digester
! To determine the Residence Time Distribution (RTD) within the digester so that the
active volume, minimum retention time and short-circuiting can be derived.
RTD and Flow Pattern Characterisation
Mixing vessels usually exhibit a flow pattern that lies somewhere between a completely
back-mixed flow pattern and a plug flow pattern. A completely back-mixed system would
exhibit an RTD curve as shown in Figure 6 featuring exponential decay of the tracer pulse
concentration at the outlet. An ideal plug flow system is one where each element of the
vessel contents would have the same residence time. The RTD curve for a plug flow system
with some dispersion is also shown in Figure 6. Sewage sludge digesters typically exhibit a
fully back-mixed RTD, ensuring contact between feed sludge and the active biomass.
Results
The results are summarised in table 4.
Based on the initial 12-hour mixing test period the results showed that the actively mixed
volume of the digester was completely mixed within 75 minutes. The tracer signal remained
relatively flat after the initial mixing and did not indicate any pockets of lithium solution that
were subsequently reintroduced back into the active zone, while the mixing test was running.
Based on the data gained from the remaining 36-day washout curve, the actively mixed zone
for the NIGG Digester consisted of approximately 93% of the total digester capacity. Hence,
the percentage dead volume in the digester was approximately 7%.
The calculated degree of short-circuiting was 5.8% of the average feed volume.
The nominal retention time of the digester, at an average feed volume of 168 m
3
/day was 23.8
days. The calculated mean retention time from the linear regression of the washout curve
was 23.5 days.
Mixing Test
The results of the initial 12-hour mixing test period can be seen in Figure 7. The test showed
the following:
Fully mixed conditions achieved at: 75 minutes
Mean concentration of lithium between 75 and 720 minutes: 3.02 mg/l
The fully mixed blend time of 75 minutes was based on the time after which all the corrected
data was within the limits of +/- 5% of the mean concentration.
Sludge Feed Rate Data
The sludge feed profile can be seen in Figure 7 over the 36-day test period (15th November
21st December 2001).
In order to calculate the hydraulic performance of the digester in terms of active volume and
short-circuiting, Levenspiels model (Reference 7) was applied to the experimental data.
Levenspiels model assumes a perfectly mixed zone, with a dead zone attached and a bypass.
One of the requirements of the model is that a constant or mean feed rate is used to achieve
meaningful results.
Due to the nature of commissioning there was some variability in the feed rate over the 36-
day test period.
The mean sludge feed rate over the whole of the test period was calculated to be 168m
3
/d
which, is superimposed onto the chart in Figure 8.
The results for the 36-day washout curve are presented in Figure 9. Linear regression
conducted on the results can be seen in Figure 10. It can be seen that the best-fit line to the
analytical data is very good with and R
2
value (correlation coefficient) of 0.9383 (the closer
this value is to 1 the better the fit is). Figure 10 gives the values of the y intercept and the
gradient of the curve. This data can then be applied to the Levenspiel model.
The performance criteria for the mixing test states that the mean sludge retention exceeds
twelve days, that the active volume is a minimum of 90% and that short-circuiting is less than
5%. The results showed that 92.8% of the digester volume was active and the mean sludge
retention time was 23.5 days. The short-circuit volume was calculated to be 5.8% of the
average daily feed volume. This exceeded the limit by 0.8 %.
Discussion of results
The full scale system as designed meets the performance tests required under the contract.
The use of fundamental rheology has allowed for a more accurate design of a the mixing
system for thick sludges.This has been validated by full scale performance testing.
A review of the mixing system design is presented in table 3. This demonstrates very low
specific mixing energies of 1.49 W/ m
3
with a daily power consumption of 90.2 Wh/ m
3
/day.
This is very favourable compared with the stated power consumption of other mechanical
systems
5
; draft tube 2 3.5 W/ m
3
and impellers about 2 W/ m
3
.
The system is currently run 24 hours a day. Now that the dispersion information is available
taking the more conservative dispersion time of 90 minutes, it is possible to optimise the run
times further to reduce mixing energies.
CONCLUSION
With increasing pressures on the sludge digestion Energy efficient gas mixing systems can
now be designed which can compete with the best mechanical systems in terms of energy
consumption. The work to date by the research partners has shown this and they have been
employed at a number of sites across the UK.
Fundamental research has improved our understanding of sewage sludge rheology
Sequential gas mixing is more effective at mixing thick sludges that a continuous
systems at the same net energy inputs
Design guidelines have been developed for mixing thick sludges in digesters
Effective gas mixing systems can be designed with specific mixing energies of 1- 2
W/ m
3
.
The Monsal SGM system compares favourably with mechanical systems with specific
mixing energies in the 40 90 Wh/ m
3
/day.
REFERENCES
1. DAWSON M., CHRISTODOULIDES, J., FAWCETT, N. AND BRADE, C. A
Comparison of Mixing Systems in a Model Anaerobic Digester 5th European
Biosolids and Organic Residuals Conference, Aqua-Enviro, Wakefield, UK. 19-22
Nov. 2000
2. BARKER, J., DAWSON, M. Digester Mixing: Theory and Practice 3rd European
Biosolids and Organic Residuals Conference, Aqua-Enviro, Wakefield, UK. 16-18
Nov. 1998
3. WHORLOW, R.W., Rheological Techniques, Ellis Horwood Ltd., 1980
4. MAY, P. Renewal of Mogden STW digester heating and mixing systems Sludge
Digester Mixing and Heating systems, Cranfield, UK. 22 May 2001
5. LAHNER E, WERNER U. Draft tube sludge mixer a unique solution for the
process of digestion Sludge Digester Mixing and Heating systems, Cranfield, UK.
22 May 2001
6. FROST, R. C. How to Design Sewage Sludge Pumping Systems Technical Report
TR185, Water Research Centre, 1983
7. Levenspiel, O., Chemical Reaction Engineering, Wiley International Ed., 1972.
ACKNOWLEDGEMENTS
The authors would like to thank EarthTech Engineering, in particular Martin Jolly and Steve
Wright for their help in preparing this paper.
TABLES AND FIGURES
Table 1 Raw sludge make up to sludge processing plant
Catchment Load (TDS per day) Sludge Type
Nigg STW 18.9 50% high rate PSTs (lamella)
50% BAFF (non- nitrifying)
Persley STW 2.75 Non nitrifying ASP without
primary settlement
Fraserburgh STW 2.75 Primary + ASP from fish
processing (non- nitrifying)
Peterhead STW 6.55 Primary + ASP from fish
processing (non- nitrifying)
Imported Cake 1.37 Municipal
Imported Liquid 6.03 Municipal
3
rd
party imports 5.5 Industrial from fish
processing
TOTALS 43.85
Table 2 Design conditions for digesters at Aberdeen
Sludge Type Primary + SAS
Digester temperature 35
o
C
Raw sludge feed 365 m
3
/day
Digester Design HRT 18 days @10% feed
TS raw sludge 10-12 %
VS raw sludge 75 %
Digester organic loading 4.1 kgVS/m
3
.day
TS digested 7-8%
Table 3 - Operational data of Monsal SGM System
Aberdeen
Digester volume 4000 m
3
Immersion depth 17 m
Compressor rating 216 m
3
/h
Adiabatic Compression
power
6.97 kW
Isothermal expansion power 5.96 kW
Power consumption during
gas mixing
1.49 W/ m
3
Absorbed power on
Compressor Motor
15.1 kW
Energy efficiency of mixing
system
39%
Operational time of gas
mixing
24 h
Daily power consumption 362.4 kWh/d
Daily power consumption per
digester volume
90.6 Wh/ m
3
/day
Table 4 - Summary of mixing test results
Days (1-36)
value of C
0
used c
0
= 3.09
average V, m3 4000
average Q, m3/day 168
M (from line fitted to wsht data) -0.042606
k (from line fitted to wsht data) -0.045813
R
2
0.9383
V
A
m3 3712.8
Q1 m3/day 158.2
% flow short-circuited 5.8
% available volume used 92.8
MRT
1
= V
A
/q
1
= -1/m 23.5
MRT
(ideal)
= V/Q 23.8
Figure 1: Range of apparent sludge viscosities found from site survey
Figure 2: Log-log plot showing comparison of apparent viscosity - TR185 predicted and
measured for a 4% DS digested sludge
0.01
0.1
1
10
100
0.01 0.1 1 10 100
Shear Rate (s-1)
A
p
p
a
r
e
n
t

V
i
s
c
o
s
i
t
y

(
P
a
.
s
)
2.5% DS
5% DS
10% DS
Increasing DS
0.01
0.1
1
V
i
s
c
o
s
i
t
y

(
N
/
m
2
)
10 100
Shear rate (s-1)
Measured
TR185 median
TR185 upper
TR185 lower
Herschel-Bulkley
Figure 3: Different sparger arrangements used at pilot scale
Figure 4: Section of the digesters at Aberdeen
x
x
x
1
2
3
4
5
6 7
8
(b) Sparger B
9
10
11
12
x
(a) Sparger A
Inlet position
Inlet position
Figure 5: Rheology data for a 12%DS hydrolysed sludge compared with non hydrolysed
sludge
Figure 6: RTD curve for a fully back mixed system
Rheogram to compare Hydrolysed sludge with digested sludges of 7 to 8 %DS
0.1
1
10
100
0.01 0.1 1 10 100
Shear rate (s-1)
A
p
p
a
r
e
n
t

v
i
s
c
o
s
i
t
y

(
P
a
s
)
Hydrolysed
8% DS digested sludge
7.6% DS digested sludge
Figure 7: Nigg Digester: corrected data for the initial 12 hour period
NIGG WWTP - 15/16 Nov 01
Digester Lithium tracer test
0
1
2
3
4
5
6
0 100 200 300 400 500 600 700
Time (minutes)
L
i
t
h
i
u
m

C
o
n
c
e
n
t
r
a
t
i
o
n

(
m
g
/
l
)
Test Samples Mean Concentration
Figure 8: Sludge feed rate over 36-day test period with mean sludge feed rate
Figure 9: Lithium tracer washout curve for the Nigg Digester over 36-day test
0
50
100
150
200
250
0 5 10 15 20 25 30 35
TIME (DAYS)
S
L
U
D
G
E

F
E
E
D

R
A
T
E

(
m
3
/
d
)
ACTUAL
MEAN
168 m
3
/d
y = 2.9538e
-0.0426x
R
2
= 0.9383
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0 5 10 15 20 25 30 35 40
Time (days)
C
o
n
c
e
n
t
r
a
t
i
o
n

(
m
g
/
l
)
Figure 10: Washout curve Linear regression on Nigg Digester
y = -0.0426x - 0.0412
R
2
= 0.9383
-1.80
-1.60
-1.40
-1.20
-1.00
-0.80
-0.60
-0.40
-0.20
0.00
0 5 10 15 20 25 30 35 40
Time (days)
l
n

(
C
/
C
o
)
ln C/Co
Linear (ln C/Co)