GOING THE DISTANCE: PRIMARY SLUDGE PUMPING

Transcription

1 GOING THE DISTANCE: PRIMARY SLUDGE PUMPING H Walmsley 1, 2, G Modulon 1, 2, W Schmiedte 1, 2, D Taylor 1, 2 1. GHD Pty Ltd, Sydney, NSW, Australia 2. 4Malabar Alliance, Sydney, NSW, Australia ABSTRACT The composition of primary wastewater sludge is determined by a number of factors which vary significantly between treatment plants. Composition in turn affects sludge rheological properties that govern flow behaviour. In the design of pumped sludge transfer systems, rheology testing should be carried out on relevant sludge samples, if available, to increase the confidence in the input parameters used for friction loss calculations. The accuracy of friction loss calculations to inform pump selection becomes more critical since a slight variation in calculated friction loss, and in turn required pump head, will be amplified as the transfer distance increases. During the design phase for the project, the variance between friction loss calculations using a number of theoretical models led to in-situ pressure testing. Pressure test results were used to validate the Herschel-Bulkley model as the most appropriate method for pipe friction loss analysis for primary sludge at Malabar Wastewater Treatment Plant. The Bingham plastic model was also found to be comparable. INTRODUCTION Malabar Wastewater Treatment Plant (Malabar WWTP) is located in Sydney s eastern suburbs, around 15 km south east of the city. Commissioned in the early 1970 s, it is Sydney Water s largest wastewater plant with a peak wet weather flow of 1,300 ML/d. Sydney Water is partnering with John Holland, United Group Limited and GHD to form the 4Malabar Alliance, to deliver the Malabar Process and Reliability/Renewals (PARR) Improvement Project. The project aims to: Improve the safety and working conditions at the WWTP Reduce operating costs by more efficiently operating and controlling the WWTP, and reducing manual intervention Ensure the plant complies with its environmental protection licence Improve biosolids product quality Improve the operational conditions, by providing appropriate process capability, reliability and operability of equipment. Malabar WWTP is a primary treatment plant, consisting of six coarse and fine screens, four aerated grit tanks and six primary sedimentation tanks. Primary effluent is discharged via a deep ocean outfall, 3.6 km offshore at a depth of approximately 80 m. At Malabar, each primary sedimentation tank has a dedicated primary sludge pump ( the pumps ) which transfers sludge via a single rising main to one of three digesters for stabilisation. The capacity of the existing pumps to satisfy operational requirements of the upgraded plant was reviewed as part of the PARR improvement project. The six existing primary sludge pumps are horizontal, end suction, belt-driven centrifugal open impeller type slurry pumps (Warman model TC 4/4 units) with a relatively low pumping efficiency of 20 to 30%. The nominal duty point of the units is 30 L/s at 40 m head. Primary sludge is transferred via a 505 m long mild steel rising main (consisting of 170 m of DN150 and 335 m of DN250) to the furthest digester. Scum from the primary sedimentation tanks is also collected and transferred into the primary sludge rising main intermittently at 30 L/s. The current system pumps sludge between 2.5% and 3.5% total solids residual (TSR) at approximately L/s. The behaviour of primary sludge is affected by the total solids residual (TSR) of the sludge. In turn, a change in TSR impacts on the operation of the sludge pump. As the TSR increases, the system pressure (for a given flow) also increases. The existing centrifugal primary sludge pumps see a significant reduction in flow as the system pressure increases, resulting in an insufficient volume of sludge being transferred from the primary sedimentation tanks to the digesters. This may lead to an increased sludge blanket depth within the primary sedimentation tanks, an increase in TSR and a further reduction in transfer capacity, creating an undesirable loop which continues until the point is reached where the centrifugal pumps can no longer deliver any flow to the digesters.

2 The PARR improvement project requires the reliable transfer of primary sludge, with up to a higher TSR of 5% at 30 L/s, to a new sludge screening facility complete with a storage tank for screened sludge, prior to being transferred to the digesters via another new pumping facility. The existing primary sludge rising main needs to be extended to the new facility, resulting in a total rising main length of 540 m. This change in parameters results in an altered duty point for the primary sludge pumps, determined by the required flow, static head, minor system losses and pipeline friction losses. There was concern that the existing centrifugal pumps would be unable to deliver the minimum requirement of 30 L/s at 5% TSR. Accurate friction loss calculations to inform pump selection become more critical as the transfer distances increase, since the variation in pump duty resulting from an underestimate/overestimate of friction losses may result in the selected pump operating outside its safe and efficient range. To inform the friction loss calculations, rheological testing of Malabar WWTP primary sludge was completed to provide relevant data for analysis using a range of theoretical models. The spread of calculated friction loss values was such that pressure testing of the existing primary sludge pumping system was carried out to enable correlation of theoretical calculations. This led to validation of the Herschel-Bulkley model and final pump head. Flow Case 3: an emergency pumping configuration whereby the new sludge screening facility is bypassed and the primary sludge is transferred at 30 L/s direct to the digesters. Transfer of sludge up to 5.0% at 30 L/s required friction loss calculations to be undertaken for the above flow cases over a TSR range of 3.0 to 6.0% to estimate the required pump operating duty. (The 6.0% case was estimated as an extreme event). Rheological Properties A fluid may be characterised by the relationship between the shear stress and shear strain of the fluid. A Newtonian, or ideal fluid, has a linear relationship between stress and strain, meaning the viscosity of the fluid is constant over all values of shear stress. Fluids not satisfying this criterion are broadly referred to as Non-Newtonian fluids. A Newtonian fluid (Figure 1) has the following shear stress-shear strain relationship: ( ) = (1) PROCESS This section outlines the process taken to determine the potential range of pump duty points for the upgraded primary sludge pumping system. This includes the establishment of design criteria and possible flow patterns, rheological testing to determine sludge properties for input to theoretical models and in-situ pressure testing to verify the calculated friction loss. Design Parameters The main pumping configurations that will be used following the PARR upgrade are: Figure 1: Newtonian Fluid A fluid characterised by a Power Law relationship no longer has a constant viscosity over all values of shear strain. A Power Law fluid (Figure 2) has the following shear stress-shear strain relationship: ( ) = (2) Flow Case 1: 30 L/s sludge transfer from the primary sedimentation tanks via a 540 m DN250 unlined steel duty sludge rising main to the new primary sludge screening facility, and then into the screened primary sludge storage tank. An additional 5 L/s of scum will be intermittently fed into the line at the existing scum tie-in point, 170 m downstream from the furthest primary sludge pump. Flow Case 2: as per flow case 1, however via a 540 m DN200 glass lined standby sludge rising main (repurposed disused pipework). Figure 2: Power Law

3 A Bingham Plastic fluid behaves somewhat like a Newtonian fluid; however the yield stress of the fluid is the additional parameter which distinguishes it from a Newtonian fluid. A Bingham Plastic fluid will not flow until the shear stress applied exceeds the yield stress of the fluid. This type of fluid may also be referred to as a visco-plastic or yield stress fluid, some examples include mayonnaise and toothpaste. Once the fluid is moving, the shear stress increases proportionally to the shear rate. A Bingham Plastic fluid (Figure 3) has the following shear stress-shear strain relationship: ( ) = + (3) Figure 3: Bingham Plastic A Herschel-Bulkley fluid combines the characteristics of a Power Law and Bingham Plastic relationship. A Herschel-Bulkley fluid is also a visco-plastic with the fluid behaviour at stresses above the yield stress being governed by an exponent n to characterise the shape of the curve. A Herschel-Bulkley fluid (Figure 4) has the following shear stress-shear strain relationship: ( ) = + (4) Figure 4: Herschel-Bulkley The Herschel-Bulkley model is particularly suited to assessing systems at low to moderate shear rates where there can be significant variability in the apparent viscosity (i.e. slope of the curve). At very high shear rates a Bingham model may be more suitable. Sewage sludge pumping systems generally operate in a low to moderate shear regime. Understanding sludge rheology is essential for predicting flow behaviour in sludge transfer systems. Using previously published rheological data to calculate system friction losses is considered undesirable for non-newtonian primary sludge due to the significant variation in sludge composition, and therefore rheological properties, between plants. At Malabar WWTP, even a slight variation in calculated friction losses within the system is amplified by the long distance the primary sludge is transferred. Therefore, samples of sludge were collected and rheological data derived from direct analysis of the plant s sludge was obtained to improve the accuracy of friction loss calculations to inform pump selection for the calculated duty. The rheology of Malabar WWTP primary sludge was initially characterised by a yield stress and a plastic viscosity, which both vary with %TSR (Figure 5). Malabar primary sludge was tested at concentrations of 2.64%, 3.14%, 3.7%, 4.63% and 4.96%. Rheology test results are summarised in Table 1 and a plot of shear stress against shear rate of Malabar WWTP primary sludge at varying concentrations is shown in Figure 5. Results of sitespecific rheology testing were compared against published data (Baroutian S, et. al, 2012; Frost RC, 1982; Novarino D, et al, 2010). There was a significant difference in these results, highlighting the importance of site-specific rheology testing. Determination of Friction Loss The characterisation of a fluid as Newtonian, Power Law, Bingham Plastic or Herschel-Bulkley is essential for accurate estimation of friction loss within a pumped system. The rheology test result, Figure 6, indicates the viscosity of the sludge sample decreases as the applied stress increases. As a result, Malabar primary sludge can be characterised as a non- Newtonian, yield pseudo-plastic fluid, with the Herschel-Bulkley model expected to be the most appropriate method for pipe friction loss analysis. Rheology test data for all concentrations were initially analysed in terms of the Bingham Plastic model. The yield stress and plastic viscosity at different %TSR values obtained from rheology tests were further analysed to determine the Herschel- Bulkley model parameters, the consistency index n and consistency coefficient K (Table 2). Sludge parameters for 5.5% and 6.0% TSR cases were extrapolated from the lower value test results as data above 4.96% was not able to be prepared and tested. The Herschel-Bulkley parameters were then used to estimate the friction loss within the system, generally under laminar or transitional flow regime conditions. If flow was found to be turbulent, an alternative Fanning friction factor method with

4 modified equations applicable to both laminar and turbulent flow were applied to estimate friction loss. The Metzner-Reed Reynolds number was used to determine laminar or turbulent flow conditions. Laminar flow occurs if the pipeline velocity is less than the upper critical velocity when the Metzner- Reed Reynolds number is equal to The transitional range occurs between 2100 and 3000; below 2100 is fully laminar (Metzner & Reed, 1955; Heywood, 1991). In some instances for the lower %TSR cases (i.e. < 3.7%), the flow in a run of pipe was deemed to be turbulent based on Reynolds number checks. In this case, an alternative predictive method for the friction factor f was used to calculate the loss for the primary sludge rising main. The Fanning friction factor method with modified equations applicable to both laminar and turbulent flow (Haldenwang, R et al, 2012). A Hazen-Williams calculation and a Bingham plastic model calculation were also undertaken for comparison of results. Calculations produced a wide range of friction loss results, and subsequently total dynamic pump head for the system. Results ranged from 33 m for flow case 1 at a low TSR of 3.14% to 157 m for flow case 2 at an extreme TSR of 6.0%. The significant length of the primary sludge rising largely impacts friction loss results. The variation of results led to the need for pressure testing to provide operational data for comparison with calculated results and to assist with the identification of a valid theoretical model for friction loss estimation to inform pump selection. In-situ Pressure Tests Field pressure readings of the existing primary sludge pumping system were taken in order to validate the application of the Herschel-Bulkley model. Three pressure tests of the existing primary sludge pumping system were completed on 04/08/14, 06/08/14 and 10/09/14. Results of the first pressure test were discarded due to low %TSR values, with majority of the test pumping between 1.0% and 2.0% sludge. These %TSR values were considered too low to represent future operating conditions. The second pressure test (Figure 7) was successful with %TSR values between 4.8% and 5.2% able to be generated. Primary sludge samples were taken every minute for the duration of the pressure test to provide another means of determining the sludge s %TSR. Laboratory %TSR results were slightly less than the %TSR recorded from the on-line instrument. The pressure during the pump test consistently measured around 4 Bar at the stabilised flow of L/s. The %TSR values are representative of future operating conditions and can be used to validate the Herschel-Bulkley model. The third pressure test transferred primary sludge between 2.35% and 3.8% TSR. The corresponding %TSR values measured in the laboratory varied between 0.9% and 2.55%. The on-line %TSR instrument is not reliable all the time, so the lab values were adopted in this instance. Pressure readings from the data logger were around 3 Bar at a stabilised flow of L/s, suggesting partial blockages at this relative low %TSR. Subsequent investigation found evidence of partial blockages and consequently this result was discarded. DISCUSSION The results of friction loss calculations completed using Hazen-Williams, Bingham Plastic and Herschel-Bulkley models for Malabar WWTP primary sludge transfer under flow case 1 only are shown in Table 3 for comparison. The Bingham Plastic method of calculation was comparable over the range of %TSR values with the Herschel- Bulkley Model. The Hazen-Williams results were comparable to Herschel-Bulkley for lower %TSR sludge where the sludge is more water-like, however significantly underestimated friction losses at higher %TSR values (i.e. > 3.7% TSR sludge) due to the absence of any yield stress input parameters. The rheology test results indicated the characteristics of the Malabar primary sludge would be best represented by the Herschel-Bulkley model, or the Bingham plastic model. The flow and %TSR values from the in-situ pressure test, completed on the 06/08/14, were used as inputs to validate the Herschel-Bulkley friction loss calculations. Using the previously completed rheology test results, the Herschel-Bulkley model adopted the following parameters (Table 5). Table 5: Herschel-Bulkley pressure test parameters Solids concentration (% w/v) 4.50% Pump flow (L/s) 15 Yield Stress (Pa) 25 Consistencey Index 'n' 0.38 Consistencey Coefficient 'K' 2.1 Using the above inputs, the Herschel-Bulkley model calculated an expected total dynamic pump head of 34.6 m.

5 For comparison, a Bingham Plastic model calculation was completed using the in-situ pressure test data. The Bingham Plastic model adopted the following parametes (Table 6). Table 6: Bingham Plastic pressure test parameters Solids concentration (% w/v) 4.50% Pump flow (L/s) 15 Yield stress of sludge (Pa) 25 Plastic viscosity of sludge (mpa.s) 75.6 Using the above inputs, the Bingham Plastic model calculated an expected total dynamic pump head of 34.2 m, which was comparable to the Herschel- Bulkley calculated value. The pressure measured during the field test was 35 m (corrected for instrument location) which is within 5% of the Herschel-Bulkley and Bingham estimates, indicating either model was an appropriate method in this instance for calculating system pressures for primary sludge pumping. The existing centrifugal primary sludge pumps could not be utilised to satisfy the required design criteria. The selection of a new centrifugal pump was considered unsuitable due to the range of calculated total dynamic pump head from 33 m to 157 m for a number of flow conditions and %TSR values. This variability led to the selection of a positive displacement pump for their reliability in delivering a constant flow across a range of required pump duty points, subject to sufficient motor power and casing design pressure. CONCLUSION The composition and properties of sludge vary significantly from plant to plant. Rheology testing should be done on the specific plant sludge to increase confidence in the parameters used to calculate friction losses within the transfer system. The yield stress values are particularly important as they have a significant effect on total pump head. The variance between friction loss calculations across a number of models led to in-situ pressure testing, which validated the use of the Herschel- Bulkley model as the most appropriate method for pipe friction loss analysis for this application. The Bingham plastic model was also found to be comparable. Accurate friction loss calculations become more critical as the transfer distance increases, since the variation in pump duty from an underestimate/overestimate of friction losses may result in the selected pump operating outside its safe and efficient range. ACKNOWLEDGEMENTS I would like to acknowledge the contribution of the co-authors Carlo Modulon, Wayne Schmiedte and Duncan Taylor for their technical contribution, review and guidance. Thank you to the 4Malabar team comprising of members from Sydney Water, John Holland, United Group Ltd. and GHD for your continued efforts over the development of the Malabar PARR project. NOMENCLATURE = ( ) = ( ) = = ( ) REFERENCES Abulnaga, B, 2002, Slurry Systems Handbook, McGraw Hill, USA Baroutian, S, Eshtiaghi, N, Gapes, DJ, 2012, Rheological Characteristics of a Mixture of Primary and Secondary Municipal Sludge, SCION, New Zealand. Frost, RC, 1982, Prediction of friction losses for the flow of sewage sludges in straight pipes, Technical Report RT175, Water Research Centre, Stevenage, England. Haldenwang, R., Sutherland, APN., Fester, VG., Holm, R., Chhabra, RP., 2012, Sludge pipe flow pressure drop prediction using composite power-law friction factor-reynolds number correlations based on different non-newtonian Reynolds numbers, Flow Process Research Centre, Cape Peninsula University of Technology, Cape Town, South Africa. Mays, LW, 1999, Hydraulic Design Handbook, McGraw-Hill. Modulon, G., Schmiedte, W., 2014, Primary Sludge Pumping Audit Report, 4Malabar, Sydney, NSW, Australia. Novarino, D, Santagata, E, Dalmazzo, D, Zanetti, M, 2010, Rheological Characterisation of Sludge Coming from a Wastewater Treatment Plant, American Journal of Environmental Sciences 6 (4): , ISSN X, Department of Hydraulics, Transport and Civil Infrastructures, Italy.

6 Qasim, SR, 1998, Wastewater treatment plants: planning, design, and operation, CRC Press, USA. Sanin, FD, Clarkson, WW, Vesilind, PA, 2011, Sludge Engineering: The Treatment and disposal of wastewater sludges, ISBN No , DEStech Publications, USA. Slurry Systems Engineering Pty Limited, 2014, Rheology Tests on Malabar WWTP Raw and Digested Slugde, Lochinvar Laboratory, NSW, Australia. Subramanian, RS, Non-Newtonian Flows, Department of Chemical and Biomolecular Engineering, Clarkson University

7 True Concentration (wt%) Table 1: Sludge Rheology Results: Raw Data (Slurry Systems) Measured Plastic Viscosity* (mpas) Test Temperature ( C) Plastic Viscosity at 20 C (mpas) Yield Stress (Pa)** *Yield Stress determined by rotating vane in beaker test results which were significantly greater than the viscometer derived values. **Viscosity results were determined using a viscometer. Table 2: Sludge Rheology Results: Herschel-Bulkley Parameters Parameter Minimum Average Typical Max. Abnormal Max. Worst Case % TSR * 6.0* Bulk Density (kg/m 3 ) Yield Stress (Pa) Consistency coefficient (K) Consistency index (n) *Rheology properties extrapolated highest TSR for which properties directly measured was 4.96%. Table 3: Comparison of Friction Loss Calculations for Flow Case 1 using Herschel-Bulkley, Bingham Plastic Model and Hazen-Williams Solids concentration (% w/v) 3.14% 3.70% 4.96% Pump flow (L/s) Hazen-Williams C factor Friction losses in straight pipe (m): Hazen-Williams Friction losses in straight pipe (m): Bingham Plastic Friction losses in straight pipe (m): Herschel-Bulkley Qasim, SR, 1998; Sanin, FD, et al, 2011 Figure 5: Sludge Rheology Results: Bingham Parameters (Slurry Systems)

8 Figure 6: Sludge Rheology Results: Raw Data (Slurry Systems) Figure 7: 06/08/14 Malabar WWTP Primary Sludge Rising Main Pressure Test Results