Documentos de Académico
Documentos de Profesional
Documentos de Cultura
D. J. Burt*, BEng, MSc, CEng, MIMechE J. Ganeshalingam+, BSc, MSc, PhD, AMIChemE Presented at the CIWEM / Aqua Enviro joint conference Design and Operation of Activated Sludge Plants 19th April 2005.
ABSTRACT
A Computational Fluid Dynamics (CFD) prediction procedure for computing the internal hydrodynamic behaviour of final clarifiers is presented. Calculations are carried out and presented for comparison with experimental measurements of point velocities in a shallow circular clarifier in operation at a UK waste water treatment works. The calculations compare favourably with the data and the model has subsequently been used for many design studies. A separate study is described where CFD predictions are used to investigate the influence of introducing energy dissipating influents (EDI), varying the stilling well diameter, adding Stamford baffles at the side wall or placing an influent floor baffle (McKinney baffle) below the stilling well. From the study it is possible to determine the sizes and combinations of internal baffling likely to give the best clarification performance across a range of operating conditions. The success of the optimum design is judged by the depth of the settling sludge bed and the magnitude of the effluent suspended solids (ESS) for the state points considered. Key words: Computational Fluid Dynamics (CFD), Energy Dissipation Influent (EDI), Stirred Sludge Volume Index (SSVI), McKinney-baffle, Stamford-baffle. *Senior Engineer, MMI Engineering, Bristol, UK. and Dept of Mechanical Engineering, Queens Building, University of Bristol. + Project Engineer, MMI Engineering, Bristol, UK.
INTRODUCTION
According to the CIWEM handbook(1), Activated sludge plants (ASPs) are responsible for the treatment of about 50% of all sewage treated by biological oxidation in the UK. These plants are able to produce effluents in compliance with the current legislative requirements for suspended solids (SS), biochemical oxygen demand (BOD) and nitrate content (ammoniacal N). However, European legislation is demanding improvements in effluent quality and the requirement to treat larger volumes in our expanding towns and cities, is putting considerable pressure on the existing sewage treatment infrastructure. Consequently, there is much interest in the enhancement and improvement of the performance of the final stages of the activated sludge process. For, it is in the final clarifier that the sludge settles and it is the separation efficiency of the clarifier that largely determines the effluent quality of the ASP. It is well known that the settling efficiency of the clarifier is greatly affected by the hydrodynamic flow paths within the tank. In 1940, Anderson(2) published a paper showing how the internal flows within the tank led to a billowing of suspended solids immediately below the effluent weir. Figure 1 is taken from the work of Anderson and shows a section through a typical circular clarifier of 40m diameter and 3.5m side wall depth. The influent has an axial riser delivering feed into the tank through a central radial diffuser. Flow rates for this tank are in excess of 1000 m3/hr giving inlet velocities of order 0.1 m/s. A conventional stilling well is used to attenuate the turbulent inlet flow; the diameter and depth of this well are significant design parameters. The measurements show the typical density driven current observed in all circular clarifiers. The
dense feed with a concentration of mixed liquor suspended solids (MLSS) typically ranging from 2000 to 6000 mg/l, falls from the influent and generates a radial underflow. In order to balance this momentum, the underflow is matched by a return flow at higher depths in the tank. The resulting re-circulation is one of the reasons why the sediment blanket appears to lift at the side wall below the effluent weir. This early work clearly shows that the flow in a circular clarifier is far from one dimensional but can be considered as close to two dimensional with axis symmetry. The standard technique for designing a secondary clarifier is mass flux theory, sometimes referred to as state point analysis. This method uses a one-dimensional settling model that cannot take account of the complex internal flow patterns flow present in the clarifier. Shortcircuiting of the influent flow, scouring of solids and re-entrainment of solids into the recirculating flow pattern all contribute to the effluent suspended solids (ESS). Excess influent momentum, perhaps from an undersized stilling well, can invoke a complete failure with overflow of the sludge blanket. Even when the clarification surface area is over sized, relative to mass flux theory, the tank may still fail, or perform badly in practice because of these internal flow features. Clearly there is a need to use a new design method for clarifiers that is capable of overcoming the limitations of mass flux theory. In this work, a CFD modelling technique was developed, verified and validated to determine the internal hydrodynamic performance of secondary clarifiers. The model is based on an adaptation of the IAWQ drift flux model part of which is discussed in the Scientific and Technical report No 6 by Ekama et al(3). Results from the model were used to demonstrate several characteristic internal flow features that are thought to be causes of poor clarification performance. A validation study is presented which compares local velocity data with experiment for the Rye Meads clarifier, measured extensively by Richardson et al(4)(5). Also a single design study is presented for a typical UK clarifier where a number of internal modifications have been investigated in order to understand the influence of introducing energy dissipating influents (EDI), varying the stilling well diameter, adding Stamford baffles or placing an influent floor baffle (McKinney) below the stilling well. In the design study, the flow and settling behaviours were calculated for a base case configuration and then for alternative designs. The aim of the work was to determine the internal design likely to give best clarification performance across a range of operating state points. The success of a design was judged by the height of the settled sludge bed and the magnitude of the ESS for all state points.
the mass flux curve as there are significant differences in these. Figure 2 shows how correlations can differ greatly when used in a practical selection application.
CFD MODELLING
Most of the CFD modelling work for secondary clarifiers reported in the literature uses a form of the algebraic slip (ASM) or drift flux model of Wallis(9) to represent the two-phase mixture of water and activated sludge. Zhou and McCorquodale(10) describe how the density variations and settling velocity relationships may be modelled. Lakehal and Krebs (11) have extended this model to include variations in fluid mixture rheology. Two recent PhD theses discuses the model in greater detail and include extended rheology functions, see Armbruster(12) and deClercq(13). In this implementation, a modified version of the CFX code was used as the modelling tool with an adaptation of the IAWQ drift flux model(3). The simulations were performed in twodimensional, axis-symmetric co-ordinates with models for sludge mixture density, viscosity, following Bokil and Bewtra(14), or Dahl(15), Dick and Ewing(16), and the double exponential settling function, following Takcs(17). A low Reynolds number k - turbulence model(18) was used in deference to the large variation in mixing time scales present in a clarifier. The model uses a multiphase method where the sludge is able to move independently with respect to the water but only in the direction of a slip vector; where, in this implementation the slip only acts in the direction of gravity.
Boundary Conditions
Even with its limitations, there is still value in using mass flux theory to calibrate the boundary conditions for subsequent CFD analysis. In the process of developing the generic CFD studies reported here it was discovered that, without applying any factors of safety, tanks with sufficient surface area to satisfy mass flux theory always fail before the mass flux limit when modelled with CFD, in other word mass flux theory is the absolute upper bound of performance. Mass flux theory also provides a useful indicator of the likely RAS solids concentration.
inclined at 8, they are equipped with a central feed well, perimeter weir and bridge driven scrapers. The conventional centre well is a 9m diameter, open bottomed, circular baffle arrangement, 1.5m deep (1.3m submerged). The tanks were analysed experimentally by Scriven and Richardson(5) at flow rates representative of the bounds of operation for the tanks. Comparisons between CFD and experiment are presented here for a forward flow of 360 m3/hr, mixed liquors at influent of 2825 mg/l, a RAS ratio of 1.0 and SSVI of 80 ml/g. The experimental work provided information on effluent quality, fluoride tracer break through curves, approximate blanket depth and Acoustic Doppler Velocimetry (ADV) measurements. The ADV data was found to be particularly useful for validating the CFD model results.
DESIGN STUDY
The design study is for a typical UK clarifier 55 (16.76 m) diameter and 8 (2.438 m) side wall depth. The tank has a 7.5 sloping floor with bridge scrapers and a pumped central hopper RAS extraction. As built, the centre well is only 2m (12%D) diameter and 1.5m deep. Modifications based on a larger stilling well were considered either with a central EDI, similar to that shown in Figure 4a or by mounting a baffle plate, Figure 5c, directly below the enlarged stilling well. This second design, known as a McKinney floor baffle is described in Ekama(3), it is located at a depth close to the side water depth with a gap sized for densimetric Froude number, Frd 0.7, at the highest flow rate. The key design parameter for the McKinney baffle is the slot gap between the bottom of the stilling well and the baffle. This is sized to keep 0.5 < Frd < 0.7 such that re-entrainment into the stilling well is prevented(12),
Frd =
U w gh w
where U is the average velocity through the slot, h is the height of the slot, is the mixture density at influent and w is the density of water, g is the gravitational constant. Alternative designs were analysed at various state points representative of the bounds of operation. Comparisons are presented here for a forward flow of 156.3 m3/hr, with mixed liquors at influent of 3700 mg/l, RAS ratio 0.51 or 1.0 and SSVI of 80 ml/g.
CFD Models As Built EDI McKinney Flow ratio R 0.51 a) 1 0.51 b) 1 0.51 c) 1 ESS Depth below TWL (mg/l) (m) 312 0.245 9 1.450 328 0.255 12 1.925 24 0.825 10 2.170 Status Fail Pass Fail Pass Pass Pass
Table 1: Summary of bed depth and ESS for the model variations. Figure 5 and Table 1 summarise the results for the study. Two state points are considered for three geometries allowing the performance of each design to be compared. The first thing to note from table 1 is that a critically loaded clarifier can be brought back into compliance by simply increasing the RAS rate. Figure 5a shows how the smaller stilling well gives rise to a higher sludge bed but, where the influent flow is now directed through the bed, the ESS is seen to be the lowest of the three configurations. Figure 5c shows how the McKinney baffle acts to separate the stilling zone from the settling zone and this design has the ability to carry the greatest volume flow rate, the results in Table 1 show that, in this case, only the McKinney design is compliant at both of the RAS rates investigated. Comparisons for the EDI design with and without a Stamford baffle showed no difference in the effluent quality, see Figure 6.
Although there may be merit in deep designs (this is still a topic of investigation), the modeling techniques applied here have largely been used to obtain maximum performance from flat bottom or shallow angle clarifiers with various forms of RAS removal system. From these studies a number of observations have emerged that have provided guidance on internal baffling which can improve effluent quality and allow the tank performance to approach the mass flux limit.
a way that the effluent flow effectively by passes the location of the Stamford. In flat bottomed deep tanks, with low beds, the shear layer tends to persist all the way to the side wall and the Stamford can have a much more significant influence on the effluent quality.
CONCLUSIONS
An integrated CFD and mass flux modelling approach has been developed that may be used to optimise the internal designs of final clarifiers. During the course of the development work a number of key conclusions have arisen both with reference to current clarifier design methodology and to current design practice. Flows in secondary clarifiers are characterised by a strong density current arising from the influent that drives a radial shear layer above the settling sludge bed. Several flow feature exist that must be designed out to optimise the tank performance. The flow is far from one dimensional but it can be considered as close to two dimensional and in a circular clarifier there is axis-symmetry. Tanks with sufficient surface area to satisfy mass flux theory will always fail before the mass flux limit when modelled with CFD, in other words mass flux theory is the absolute upper bound of performance. The stilling well dimensions are important, too large a diameter or too shallow, can serve to enhance the density current momentum through re-entrainment. Two options are currently favoured as a means of breaking the density current. An EDI or a McKinney baffle. A central EDI helps to diffuse the density current by spreading the load uniformly near the top of the stilling well and reducing the density gradients in the stilling pond. A McKinney baffle cuts the density current and, if designed with 0.5 < Frd < 0.7 it distinctly separates the stilling zone from the settling zone introducing the flow as a direct radial jet into the settling zone. The McKinney design tends to favour a limited operating range as it provides maximum benefit only when the baffle sits slightly above the settling sludge bed. In its best operating range it will tend to out perform an EDI. Sidewall baffling has only a limited influence in shallow clarifiers but can be beneficial in flat bottomed clarifiers with deep side walls. It is now possible to check clarifier retrofit and final design options with CFD modelling prior to executing a civil engineering project.
ACKNOWLEDGEMENTS
The author would like to thank Dr Pete Pearce of Thames Water for his help, advice and continued support. Thanks also to other colleagues at Thames Water, United Utilities, Montgomery Watson Harza, Severn Trent and Yorkshire Water who have all supported projects contributing to the overall understanding of internal clarifier flows.
REFERENCES
[1] Chartered Institution of Water and Environmental Management (CIWEM), Activated Sludge Treatment, Handbooks of UK Wastewater practice, London, 1997. [2] Anderson, N.E., Design of Settling Tanks for Activated Sludge, Sewage Works J., 17(1), 50-63, 1945. [3] Ekama, G.A., Barnard, J.L., Gunthert,F.W., Krebs,P., McCorquadale, J.A., Parker, D.S. and Wahlberg, E.J., Secondary Settling Tanks, Theory, Modelling, Design and Operation, International Association of Water Quality, Scientific and Technical Report No 6, 1997. [4] Richardson, D.S., Hydraulic Considerations of Final Settlement Tank Design, Cranfield University School of Water Sciences, M.Sc. Thesis, 1998. [5] Scriven, R. and Richardson, D.S. Rye Meads STW Stage 1 Final Settlement Tanks Process Investigation, Thames Water Technical Report, R19808, August 1998. [6] Pitman, A.R. Settling Properties of Extended Aeration Sludge, J. Wat. Pollut. Control Fed. 52(3), 524-536, 1980. [7] White, M.J.D., Settling of Activated Sludge, Technical Report TR11, Water Research Centre, Stevenage, UK, 1975. [8] Wahlberg, E.J. and Keinath, T.M. Development of settling flux curves using SVI J. Wat. Pollut. Control Fed. 60 (12), pp2095-2100, 1988. [9] Wallis, G.B., One-Dimensional Two Phase Flow, McGraw-Hill, 1st Ed, 1969. [10] Zhou, S. and McCorquodale, J.A., Modelling of Rectangular Settling Tanks, J. Hydr. Eng., ASCE, 118(10), October 1992. [11] Lakehal, D., Krebs, P., Krijgsman, J. and Rodi, W. Computing Shear Flow and Sludge Blanket in Secondary Clarifiers, J. Hydr. Eng., ASCE, 125(3), 1999. [12] Armbruster, M., Untersuchung der mglichen Leistugssteigerung von Nachklrbeken mit Hilfe numerischer Rechungen, PhD thesis, University of Karlesruhe, August 2003. [13] De Clerq, B. Computational Fluid Dynamics of Settling Tanks: Development of Experiments and Rheological, Settling and Scraper Sub Models, PhD Thesis, Dept of Applied Math, Biometrics and Process Control (BIOMATH), University of Ghent, Belgium, 2003. [14] Bokil, S.D. and Bewtra, J.K., Influence of Mechanical Blending on Aerobic Digestion of Waste Activated Sludge, Proc., 6th Int. IAWPRC Conf. on Water Pollution Res., Int. Assoc. on Water Pollution and Control, London, 421-438, 1972. [15] Dahl, C.P., Larsen, T. and Peterson, O., Numerical Modelling and Measurement in a Test Secondary Settling Tank, Water Sci. and Technology., 30(2), 219-228, 1994. [16] Dick, R.I. and Ewing, B., The Rheology of Activated Sludge, J. Water. Pollution Control Fed., 39(4), 543-560, 1967. [17] Takcs, I., Patry, G.G., and Nolasco, D. A Dynamic Model of the Clarification Thickening Process., Water Res, 25(10), 1991
[18] CFX International, CFX-4.4 Solver Manual, Vol 3, AEA Technology, Harwell, 2001. [19] Burt D.J. and Ganeshaligam, J. Validation study for the Witney Clarifier, MMI Engineering Report, MMU035, February 2005. [20] Vesilind, P.A. Theoretical considerations: Design of prototype thickeners from batch settling tests, Water and Sewage Works, 115 (July), 302-307, 1968. [21] Burt, D.J. and Gilbertson, M.A. Flocculation Frameworks and CFD Modelling for Activated Sludge Clarifiers, 5th Particle Technology Forum, University of Sheffield, July 2003. [22] Biggs, C. A. Activated Sludge Flocculation: Investigating the Effect of Shear Rate and Cation Concentration on Flocculation Dynamics, PhD Thesis, Dept of Chem Eng, University of Queensland, Australia, 2000.
Figure 1: Taken from Anderson(1) shows concentration gradients through a circular clarifier and velocity vectors mapped at discrete points.
RAS = RQ 360.00 m3/h Recycle 360.00 Tank Diameter SST Area (m2) Forward Flow (Q, m3/h) Design Overflow Rate (m/h) SSVI (ml/g) 3 Influent MLSS ( kg/m ) RAS ratio 3 RAS_surplus (m /h) Solids Loading (l/m2h) Estimated RAS Conc. ( kg/m3) 28 615.75 360.00 0.58 100 2.825 1.00 0 220 5.6500 m /h
3
SSVI = 100 [Pitman and White (1980, 1984)] SSVI = 100 [Wahlberg and Keinath (1998)] Overflow Line Underflow Line Influent Conc.
6
3
10
Figure 2: Mass flux graph comparing Pitman(6) and White (7) with Wahlberg and Keinath(8) for the Rye Meads clarifer at average flow and 100 SSVI.
a)
Density Waterfall 0.35 0.5 0.64 0.78 0.96
b)
R / Rmax =0.35
R / Rmax =0.50 0.0 0.1 0.2 0.3 0.4 H / Hmax[-] H / Hmax[-] 0.5 0.6 0.7 0.8 0.9 1.0 -0.2 0.0 0.1 0.2 0.3 0.4
R / Rmax =0.64 0.0 0.1 0.2 0.3 0.4 H / Hmax[-] 0.5 0.6 0.7 0.8 0.9
R / Rmax =0.78 0.0 0.1 0.2 0.3 0.4 H / Hmax[-] 0.5 0.6 0.7 0.8 0.9
R / Rmax =0.96
0
V / Uin [-]
0.2
0 V / Uin [-]
0.2
-0.05
0.05
0.15
1.0 -0.15
-0.05
0.05
0.15
1.0 -0.05
0 V / Uin [-]
0.05
Expt T4
T3 T5
Expt T4
T3 T5
T3 T5
c)
0.0 0.1 0.2 0.3 0.4 R / Rmax =0.35
R / Rmax =0.50
R / Rmax =0.64
0.0 0.1 0.2 0.3 0.4 H / Hmax[-] 0.5 0.6 0.7 0.8 0.9 1.0
R / Rmax =0.78
0.0 0.1 0.2 0.3 0.4 H / Hmax[-] 0.5 0.6 0.7 0.8 0.9 1.0
R / Rmax =0.96
H / Hmax[-]
Figure 3: CFD analysis for Rye Meads, T3, T4(14) and T5(11) represent alternative rheological models compared with the experimental data of Richardson et al(4)(5). The results show good agreement for radial velocity profiles and bed height.
CROSB Y
VARIATION
PLAIN
McKI NNEY
CANTILEVERED TROUGH
a) Typical flocculating centre well arrangement with central EDI, swirling ports and large, 30% D, stilling well.
b) Various alternatives for baffling the side wall. The Crosby is also sometimes called a Stamford.
Figure 4: a) A central EDI, spreads the load out uniformly near the top of the stilling well reducing the density variation. b) The return current at the side wall can be disrupted by various baffle options.
b) Revised design with flocculating stilling well at 20%D and enclosing an EDI. The EDI is a variation on Figure 4a.
c) A McKinney baffle design with stilling well at 20%D and gap sized for densimetric Froude number, Frd 1, at the highest flow rate.
Figure 5: A typical UK Clarifier diameter 16.76m. Comparisons between designs are presented here for a forward flow of 156.3 m3/hr, mixed liquors at influent of 3700 mg/l, a RAS ratio of 1.0 and SSVI around 80 ml/g. A logarithmic scale is used to show the solids distribution 1 mg/l to 10,000 mg/l. The lightest shading at 1000 mg/l is approximately at the transition into the settled bed.
Figure 6: A Stamford baffle was included for the EDI design shown in Figure 5b. The flow patterns near the effluent are redirecated but there is no significant difference in the effluent quality.