A Protocol for Optimization of Activated Sludge Mixing · A Protocol for Optimization of Activated Sludge Mixing Randal W. Samstag1 and Edward A ... Carollo Engineers also conducted

A Protocol for Optimization of Activated Sludge Mixing

Randal W. Samstag1 and Edward A. Wicklein2

1PO Box 10129 Bainbridge Island, WA 98110 (Email: [email protected])2Carollo Engineers, 1218 Third Avenue, Suite 1600, Seattlle, WA 98110 (Email: [email protected])

AbstractThe paper outlines a protocol for comprehensive evaluation of mixing devices for activatedsludge tanks using field testing for calibration of computational fluid dynamic (CFD)modeling to compare the impacts of tank and mixing device geometry on mixing and energyefficiency. The protocol includes a CFD model for activated sludge solids settling andtransport which captures the influence of solids concentration gradients on fluid motion. Thiselement of the protocol is unique in that almost all analyses completed to date for activatedsludge biokinetic modeling or mixing have assumed either 1) complete mixing or 2) neutraldensity CFD. The protocol has been applied to date to several types of mixing devicesincluding jet aeration and mixing, horizontal shaft propeller mixers, and diffused aeration.Field testing of several other types of mixing devices has also been accomplished. Theprotocol is recommended to optimize design and application of mixing devices for activatedsludge service in aerated, anoxic, and anaerobic tanks as part of biological treatmentprocesses. The approach can be extended to incorporate biokinetic models that moreaccurately predict the impact of tank geometry and mixer configuration on treatmentefficiency than can be achieved assuming complete mixing or neutral density CFD.

KeywordsMixing, activated sludge, modeling; CFD, energy efficiency

INTRODUCTIONWith increasing recognition of the importance of nitrogen and phosphorus removal fromwastewater discharges and recognizing the proven economy of biological treatment processesfor nutrient removal, the wastewater treatment industry has seen an increase in the use of un-aerated tanks for anoxic uptake of nitrate and anaerobic tanks to facilitate phosphorus removalfrom wastewater effluents. Anaerobic tanks are also increasingly used to improve activatedsludge settleability. In order to facilitate these treatment goals, biological treatment tanks withsignificant concentrations of suspended solids must be mechanically mixed. With this increasingimportance of mixing in standard biological treatment processes, optimization of tank geometryand mixer configuration becomes more important. We want the most efficient mixing in twosenses: 1) we want near uniform distribution of suspended solids across our treatment tanks and2) we want to use tank geometries and mixer configurations that minimize consumption ofenergy. It is towards this optimization of mixing and energy efficiency that the proposedprotocol aims. This paper is an extended version of a poster and paper prepared for the JointWEF and IWA, WWTMod Conference (Samstag and Wicklein, 2014.)

PROTOCOL APPROACHThe elements of the proposed protocol include the following:

Field testing for comparison with CFD results Development of CFD models for the conditions of the field test Calibration of the CFD models Development of CFD models for alternate basin geometry and mixer type and

configuration and comparison of CFD results from alternative geometries and mixingdevices in terms of mixing and energy efficiency

Field TestingField tests used by the authors to date have included: 1) solids profiles, 2) velocity profiles, and 3)residence time distribution tests. In the current work each of these test techniques will be discussed.

Solids Profiles

Solids profile tests are arguably the most important data for evaluation of mixers. In suspendedgrowth wastewater applications a uniform distribution of solids concentrations at the lowest powerlevel is the primary goal. Solids profile tests can be implemented in the field by a number of means.Solids samples can be withdrawn by Kemmerer samplers or by a series of sample pumps using thetechniques developed by Robert Crosby (Crosby and Bender, 1980) or by solids probes. In theCrosby technique a grid of approximately 25 samples are withdrawn across the tank width anddepth at locations chosen to illustrate mixer influence. The samples are then analyzed for totalsuspended solids content (Method 2540D, APHA et al., 2013)

The authors have used the Crosby technique for sedimentation tank tests. In recent tests for tankmixers, the authors have used an optical solids probe for direct measurement of suspended solidsconcentration. For measurements discussed below we used an Insite Instrumentation Group Model3150 probe. This is a model frequently used by wastewater treatment plant (WWTP) operators forroutine sampling of mixed liquor suspended solids (MLSS) concentrations. Initially, three grabsamples are taken and total suspended solids (TSS) concentrations measured in the laboratory usingthe Standard Methods procedure referenced above. Probe measurements are then calibrated bycomparison with grab samples and clear water prior to each test. Detailed protocols for the TSStests are available from the authors. The general approach is to take three to five sets of samplesfrom one or more cross sections of the mixed tank at three to five depths. This, of course, is highlydependent upon available access.

Detailed TSS measurements for four vertical shaft mixers were conducted by Carollo Engineers forthe Orange County Utilities’ (OCU) South Water Reclamation Facility (SWRF) in Orlando, Florida(Carollo Engineers, 2013.) A cross sectional grid of 25 data points were measured for each mixer.Three hydrofoil mixers from different manufacturers were compared to a hyperboloid mixer.Hydrofoil I had three flat impellers each with a downward bending trailing edge. This impeller wasmounted 0.6 meters off the tank floor. Hydrofoil II had three curving impeller blades of relativelylarge diameter. The impeller was mounted with the shaft connection approximately 2 meters off thetank floor. Each blade projected downwards approximately one meter. Hydrofoil III wasconstructed of a single flat plate with three downward folding projections. This impeller wasmounted approximately one meter off the tank floor. These impellers produce a downward pumpingaction through the impeller. The Hyperboloid I mixer was mounted approximately 0.5 meters offthe tank floor and produced a downward pumping action using a series of upwardly projectedshallow ridges of hyperboloid shape.

These mixers were all arranged in anoxic zones of a plug flow (converted racetrack) reactor tankwithout internal recycle pumping. Measurements of TSS concentrations were taken at five differentlocations in a cross-section approximately 3 meters downstream of the mixers, perpendicular to themain flow through the tank and at five different depths. The measurements were taken over thecourse of a one-hour period on different days for each mixer. The deviation of the measuredconcentration from the average concentration across the entire section was then calculated. Thisdeviation may be considered the coefficient of variation (CoV) of the sample data for each mixer.

Contour plots of the measured concentration data are presented in Figure 1. The contour plots forHydrofoil I and Hydrofoil II indicate an asymmetry across the tank. This asymmetry is thought tohave been caused by upstream conditions in the tank for these mixers. The three hydrofoils were allarranged on their vertical shafts at approximately one third depth in the tanks. The one hyperboloidmixer was nearer to the bottom of the reactor tank. The contour plot for this mixer shows relativelyuniform concentrations across the tank, but significantly higher concentrations in the tank bottom.

Table 1 presents a summary comparison of the data. The smallest CoV among the four mixers wasfor Hydrofoil 1. But this mixer also had the highest measured power draw. Power measurementswere conducted throughout the test period using a current meter. The table shows the measuredpower draw, the power draw per unit of mixer reactor volume, the CoV, and the equivalent powerrequirement for an extrapolated CoV of 10 percent, a common requirement for specifications formixers for activated sludge reactor mixers. The equivalent power was estimated based on theassumption of the direct (linear) proportionality of mixing efficiency with power draw. The tableindicates that while Hydrofoil I had the lowest measured CoV, and Hydrofoil II the highest, whencorrected for unit power, Hydrofoil II had the lowest predicted power requirement for a 10 percentCoV. The variation in equivalent power requirement for a 10 percent CoV between the mostefficient to the least efficient mixer was over three to one.

Carollo Engineers also conducted limited field tests of jet mixing and aeration of an operatingsequencing batch reactor at the Blacks Ford Regional Water Reclamation Facility (BFRWRF) ofthe JEA Utility in Jacksonville, Florida. The tests were conducted to establish solids concentrationprofiles under normal operating conditions for use in calibrating CFD modeling. Solidsconcentration measurements were taken at multiple depths at two locations at the edge of theoperating sequencing batch reactor (SBR) using a calibrated optical solids measurement probe.Sample locations were: 1) near to the wall at one location adjacent to the main platform between theeffluent decanters and 2) adjacent to the access ladder for one auxiliary header submersible pump.Sample locations are shown schematically in Figure 2.

Mixed liquor suspended solids (MLSS) measurements were made during mixed cycles with both airand pumping operational and during pumping-only mix cycles. The measured MLSS concentrationfor the day of the test was approximately 2,400 mg/L and the sludge volume index (SVI) wasapproximately 100 mL/g.

Figure 3 presents results of sampling during a period when both the pumped mix system and theaeration blower were operational. The data indicate a fully mixed condition with a slightaccumulation of floatable solids. Solids concentrations were in the range of 2,320 to 2,380 mg/L forthe four measurement locations below the water surface and 2,550 mg/L for the measurement at thewater surface.

Figure 4 presents results from sampling at the location of the main platform during a period with themixing pump in operation and with the aeration blower off. A series of four measurements weremade at the main platform location at different times following shutdown of tank mixing pumps: 1)immediately after shutdown, 2) 25 minutes after shutdown, 3) 66 minutes after shutdown, and 4) 83minutes after shutdown. The data indicate that at this location, concentrations less than 1,000 mg/lwere present at the top 1.5 m level of the tank after 83 minutes of mixing and reached a maximumof 2,950 mg/L at the tank bottom after 25 minutes of mixing. At later times, the measured bottom

concentrations were approximately 2,600 mg/L.

Figure 5 shows results of measurements adjacent to the location of the auxiliary header submersiblepump access ladder. The access location in the test tank was on the opposite side of the tank fromthat shown in the figure. The figure is based on the tank model, which was based on manufacturer’sinstallation drawings. After 42 minutes of mixing, the solids concentration at the 1.5 m level layerwas 1,380 mg/L and the concentration at the bottom was measured at 2,200 mg/L. After 75 minutesof pumped mixing, the top-level concentration was 650 mg/L and the bottom concentration was2,450 mg/L. These results have been previously reported in Samstag, et al. (2012).

Figure 1. Solids profiles for four vertical shaft mixers.

Table 1. Comparison of mixer power levels.

Impeller TypeMeasured

PowerDraw (kW)

Unit Power(W/m3)

CoV (%)Equivalent

Power for CoV10% (W/m3)

Hydrofoil I 5.0 7.85 3.7% 2.9

Hydrofoil II 0.9 1.38 9.2% 1.3

Hydrofoil III 4.8 7.54 5.5% 4.1

Hyperboloid I 2.2 5.31 7.5% 4.0

Figure 2. Field Test Sample Locations.

Figure 3. Field Test Results with Air On.

Figure 4. Main Platform Field Pump Mix Test Results – Solids Concentration.

Depth (m) Air On - Main Platform Legend

0.0 2550 Concentration (mg/L)

1.5 2380 < 1000 650

3.0 2350 1000 < C < 1500 1250

4.6 2370 1500 < C < 2000 1800

6.1 2320 2000 < C < 2500 2450

2500 < C < 3000 2650

> 3000 3400

Depth (m) Main Platform - 0 minutes Legend

1.5 2230 Concentration (mg/L)

3.0 2210 < 1000 650

4.6 2100 1000 < C < 1500 1250

6.1 2070 1500 < C < 2000 1800

6.4 2290 2000 < C < 2500 2450

2500 < C < 3000 2650

Depth (m) Main Platform - 25 minutes > 3000 3400

1.5 2150

3.0 2270 ``

4.6 2390

6.1 2950

Depth (m) Main Platform - 66 minutes

1.5 2450

3.0 2350

4.6 2550

6.1 2640

Depth (m) Main Platform - 83 minutes

0.0 650

1.5 2510

3.0 2590

4.6 2690

6.1 2540

Figure 5. Pump Ladder Field Pump Mix Test Results – Solids Concentration.

Velocity Profile Tests

For the same plug flow racetrack mixers discussed above the authors made measurements ofvelocities in the mixer zones using a Hach Model FH950 velocity meter. Velocity measurements inthe horizontal and vertical direction were taken at the approximate locations of the TSS profilemeasurements for each mixer. Figure 7 shows velocity magnitude (indepent of direction) for the foreach mixer in a cross section facing the mixer from the mixer walkway at the opposite side of thewalkway. The velocity magnitude is the square root of the sum of the squares of individualhorizontal or vertical velocity measurements at each location. As such, the magnitude is non-directional; the velocity magnitude is the absolute value of the average velocity at each samplepoint. The mixers all showed strong downward velocity in the center of the tank cross section withupward velocities on either side near the tank walls.

The average velocity magnitude for each mixer cross section is shown in Table 2. The Hydrofoil Imixer had the highest average velocity magnitude at 0.22 meters per second (m/sec). The otherthree mixers showed average velocity magnitude in the range of 0.11 to 0.15 ft/sec. An averagevelocity magnitude of approximately 0.15 m/s was measured for the Hydrofoil II mixer, whichachieved a CoV in solids profile tests of approximately 10 percent. This velocity magnitude valuewas therefore taken as the magnitude required for a 10 percent CoV. Assuming a linear relationshipbetween velocity magnitude and solids profile CoV, the Hydrofoil II mixer showed the lowest unitpower requirement to achieve the 0.15 m/s average velocity magnitude.

Depth (m) Pump Ladder - 42 minutes Legend

1.5 1380 Concentration (mg/L)

3.0 1920 < 1000 650

4.6 2180 1000 < C < 1500 1250

6.1 2200 1500 < C < 2000 1800

2000 < C < 2500 2450

Depth (m) Pump Ladder - 75 minutes 2500 < C < 3000 2650

0.0 650 > 3000 3400

1.5 1590 ``

3.0 2450

4.6 2610

6.1 2450

Figure 7. Velocity profiles for four vertical shaft mixers.

Table 2. Test Results from Velocity Profile Tests.

Mixer TypeMeasured Power

Draw (kW)Unit Power

(W/m3)

VelocityMagnitude

(m/sec)

EquivalentPower for 0.15m/sec (W/m3)

Hydrofoil I 5.0 7.9 0.22 5.4Hydrofoil II 0.9 1.4 0.15 1.4Hydrofoil III 4.8 7.5 0.11 10.2Hyperboloid I 2.2 5.3 0.11 7.3

RTD Tests

Residence time distribution (RTD) tests can be used as an indicator of the degree of mixing in aflowthrough tank. In the RTD test a slug of non-reacting tracer is added to the upstream end of the tankand the concentration of the tracer measured at a downstream point over time. RTD theory wasdeveloped by Danckwerts (1953) and is described in detail by Levenspiel (1972). RTD theoryrecognizes that “elements of fluid taking different routes through the reactor may require different

lengths of time to pass through the vessel. The distribution of these times for the stream of fluidleaving the vessel is called the exit age distribution, E, or the residence time distribution RTD of thefluid (Levenspiel (1972) p. 255).” The exit age distribution can be thought of as the ratio of themeasured dye concentration from the tank divided by the concentration representing theinstantaneous mixing of the tracer slug into the tank volume. The exit age distribution, E, istypically plotted against the relative residence time, θ, to illustrate the RTD of the tank. The relative residence time is the expired time for a given dye sample divided by the theoretical hydraulicresidence time (HRT) for the tank volume. The HRT is the tank volume divided by the flowthrough the tank.

The aim of the RTD test is to determine the overall hydraulic character of a mixer tank zone basedon an equivalent number of tanks in series for each mixer tank zone between the extremes of 1.0 fora completely mixed tank to infinity for a plug flow tank. Using this number as an index of mixing;the mixer zone with the closest approximation to a 1.0 tanks in series (TIS) RTD fit would be themost effectively mixed. The RTD for a completely mixed tank, which would be modeled as a singleTIS, would characteristically show an exit age curve that started at the flash mix concentration ofthe dye slug at time zero and gradually decayed over time. An ideal plug flow tank would produce aRTD plot that had an infinite effluent dye concentration exactly at the mean HRT of the tank.

Figure 8 presents a graph of an example RTD test in the mixer tank with the Hydrofoil II mixer asdescribed above. The RTD test was conducted using Rhodamine WT fluorescent dye and a TurnerDesigns AquaFluor® handheld fluorometer. A slug of Rhodamine WT was added to each mixerzone at a location upstream of the mixer and samples were withdrawn at intervals from a locationdownstream of the mixer. The resulting graph of E versus θ is compared to a least-squared best fit to a TIS model of the mixer zone (See Levenspiel, 1972.) Similar RTD tests were conducted foreach of the four mixers described above. Table 3 presents results from the RTD tests for eachmixer. The table shows the best fit of the equivalent number of TIS for each mixer, the square of theerror between the RTD data and the TIS model and, for comparison, the CoV for the solids profiletest conducted for each of the mixers. The RTD test results would rate the mixers in approximatelythe same order as the solids profile CoV test. The disadvantage of the RTD test compared to thesolids profile test is that it cannot be plausibly normalized to rank mixers on the basis of equvalentpower input.

Figure 8. RTD Test for the Hydrofoil II Mixer.

Table 3. Test Results from RTD Tests.

Impeller TypeEquivalentNumber of

TISSquared Error

SolidsCoV (%)

Hydrofoil I 1.1 .137 3.7%

Hydrofoil II 2.3 .049 9.2%

Hydrofoil III 1.7 .101 5.5%

Hyperboloid I 1.7 .04 7.5%

CFD modelsThe first step in development of a calibrated CFD model is to produce a geometric andcomputational mesh of the field-tested reactor. Modern tools for CFD allow construction ofextremely detailed models of complex geometries. The first step in this process is creation of thetank model. Tools for this purpose include GAMBIT and DesignModeler, proprietary softwareowned and licensed by ANSYS, Inc. GAMBIT provides for both geometric modeling and meshing.DesignModeler is used for creation of geometric models. ANSYS Meshing is the proprietarymeshing software to create numerical grids or meshes for models created by DesignModeler andother modeling software. The authors have used ANSYS GAMBIT, Version 2.4.6 for bothmodeling and meshing for the case studies referenced here. We have used GAMBIT on acomputing platform of 64-bit workstations with multiple CPU cores running a 64-bit Windows XPoperating system.

Figure 9 presents an illustration of the three-dimensional geometric model and a projection of thecomputational mesh prepared for the BFRWRF SBR tanks for CFD analysis. The model shows thethree jet headers, the main header pump intake, the auxiliary header pumps and intakes, and theeffluent decanters. The effluent decanters were not required for the flow simulation, but wereincluded to simulate the fluid environment. The polyhedral computational mesh of approximatelyone million cells is shown, projected onto model surfaces.

Figure 10 presents the model and mesh constructed for another project simulating mixing in aracetrack reactor tank converted to anoxic and aerobic plug flow configuration (Wicklein et al.,2013). This tank was modeled with a polyhedral mesh of over two million cells. Figure 11 presentsa schematic detail of a submersible mixer modeled in the anoxic reactor using a rotating frame ofreference model.

Figure 9. SBR Tank Geometric Model and Computational Mesh.

Figure 10. Geometric Model and Computational Mesh for an Oxidation Ditch Simulation.

Figure 11. Detail of a submersible mixer modeled using a rotating frame of reference.

Calibration of the CFD modelsModern CFD models can be used without calibration. The physics of CFD have been verifiedwithin the tolerance of most field measurements many times. A possible exception is in the area ofturbulence estimation. It is widely believed that the k-epsilon turbulence model is appropriate forthe types of recirculating flows commonly seen in activated sludge sedimentation tanks, althoughother turbulence models can develop superior calibrations, depending on the problem geometry.Calibration is more important where the CFD is used as a base for solids transport or biokineticmodels, since many of the empirical parameters used in these models are much less well establishedthan the physical parameters applicable to fluid flow. Calibration techniques have included solidssettling rate testing, solids profile matching, and clarifier sludge blanket matching during dynamicflows. (See Griborio et al. (2008), Samstag and Griborio (2010), Wicklein and Samstag (2009), andSamstag et al. (2010).) For this protocol we demonstrate use of one of the most powerfultechniques, solids profile matching.

Figure 12 presents contour plots of CFD simulations of dynamic solids profiles for the BFRWRFunder conditions of normal operation, the facility where field solids profile measurements shownabove were obtained. The figure shows resulting solids profiles for simulation of the operatingcondition at subsequent times after aeration has been turned off and the tank is mixed solely by thejet mixing system. The three figures show predicted solids profiles at three different times afterturning off tank aeration. While far from a perfect fit, comparison with field test data shown inFigures 4 and 5 above confirms the development of relatively clear water in the top of the tank assedimentation takes place in the inadequately mixed tank. Figure 13 shows the simulated conditionfor the tank with aeration turned on. This well matches the fully mixed condition shown in the fieldtests (see Figure 3 above.)

Figure 12. Contour plots of CFD simulations of dynamic solids profiles for the BFRWRF.

Figure 13. Estimated Concentration Profiles with Full Aeration.

These simulations were conducted using UDF for solids settling and transport with coupling of theinfluence of solids gradients on the density profile and fluid flow. Hindered settling velocities werecalculated based on value for the sludge volume index (SVI) of 150 mL/g, using the revised Daiggerequation (Daigger, 1995). Further details are presented in Samstag et al. (2012).

The work for the BFRWRF did not include velocity profiling of the SBR tanks. This providesanother opportunity for calibration. Calibration of velocity fields has been accomplished insedimentation tanks. See the early work of Larsen (1977) which is compared to CFD work inWicklein and Samstag (2009), and the comparison of drogue velocity measurements to CFD modelpredictions in Samstag et al (2010). Figure 14 presents the predicted velocity profile for the SBRtanks at the BFWRF from the CFD model. The figures show the condition in a tank with aerationturned off in which has very high velocities at the exit from the mixing jets, but which dissipaterapidly further out from the jet exits. With the aeration on, higher velocities penetrate to the upperreaches of the tank resulting in complete solids mixing.

The importance of including density coupling in the CFD simulation is illustrated in Figure 15. Thefigures compares the results from simulation of 25 minutes of pumped mixing in the SBR tanksafter turning off aeration with the density couple active compared to a neutral density simulationwhere the effect of concentration gradients on the density field was turned off. The result is thatwithout including the density couple, the CFD simulation predicts relatively complete mixing forthe pumped mix condition. This result is unrealistic based on the field tests. Since neutral densityCFD simulation of mixing is common in the industry, this is a significant finding.

Figure 14. Velocity profiles for the pumped mixing (above) and air mixing (below) for BFRWRF.

Figure 15. Comparison of solids profiles from density-coupled and neutral density models

CFD Model AlternativesAfter calibration of the CFD model to conditions of operation in the field, the CFD model can beused to consider alternative configurations to improve operation. For the BFRWRF project a seriesof alternatives were considered in which the jet velocity was increased in an effort to improve solidsmixing during the pumped mix cycle. Figure 16 shows comparative predictions of solids profilesunder four different jet velocities from 2.5 m/sec to 4.0 m/sec. The simulations indicate thatincreasing the velocity to 3.0 m/sec would reduce solids deposition in the tank after 25 minutes ofpumped mixing after aeration is turned off, but that increasing the jet velocity to 4.0 m/sec would berequired to achieve a CoV for solids concentration less than 10 percent. This is illustrated in Table4. Power requirements for this increase in mixing intensity would triple the installed power for thisfacility. These results show that the original design was significantly inadequate to meet aspecification of CoV less than 10 percent.

Figure 16. Solids concentration profiles with increasing jet velocity.

Table 4. Average TSS concentrations by layer from CFD modeling (150 mL/g SVI).

Layer

Average TSS Concentration (mg/L)

2.5 m/sec 3.0 m/sec 3.5 m/sec 4.0 m/sec

Top 1,208 1,404 2,102 2,155

2 2,385 2,331 2,280 2,285

3 2,519 2,374 2,322 2,308

4 2,538 2,422 2,448 2,387

5 2,554 2,518 2,526 2,443

6 2,604 2,620 2,511 2,456

Bottom 3,008 2,806 2,559 2,500

Average 2,402 2,353 2,392 2,362

CoV (%) 50% 40% 12% 9%

In another project Carollo investigated the efficiency of solids mixing in the plug flow racetrackconfiguration previously shown in Figures 9 and 10. Two different mixers were investigated usingan un-calibrated CFD model incorporating density-coupled solids settling and transport. The projectalso included a two-fluid model (water and air) of the aerated zone in a side-sloped reactor. Figure17 presents the resulting solids concentration profiles from a simulation of an initial configurationusing two 15 hp (11 kW) horizontal propeller mixers with submersible motors. Figure 18 presentsthe results of an optimized solution which also included the effects of increased mixed liquorrecycle (MLR) flow through the reactor zones. Figure 19 shows simulated velocity and solidsprofiles for vertical hyperboloid mixers in the same reactor.

SUMMARY RESULTS

Table 5 presents summary data for mixing efficiency derived from the field tests and calibrated andun-calibrated CFD simulations discussed above for:

Pumped jet mixing Vertical hydrofoil mixers Vertical hyperboloid mixers Horizontal propeller mixers

The data indicate that the hydrofoil, hyperboloid, and horizontal propeller mixers have the potentialfor much greater power efficiency than pumped mixing. The equivalent power required for thesemixers varied in the range from 1.3 to 4.5 W/m3. This unit power range to achieve a 10 percentCoV of solids mixing is less by an order of magnitude than that of pumped mixing. Within thegroup of hydrofoil, hyperboloid, and horizontal propeller mixers, the most efficient, based on thisunit power criteria (operating without the influence of mixed liquor recycle) was the Hydrofoil IImixer, followed by the Hydrofoil I mixer, and lastly the hyperboloid and horizontal propeller

mixers, which were approximately equal in mixing efficiency. Based on the testing work, theHydrofoil II mixer was approximately three times more power efficient than the hyperboloid andhorizontal propeller mixers to achieve a 10 percent CoV for mixing of activated sludge solids.

It should be noted that this result is dependent on the assumption of direct proportionality of powerwith mixing intensity, since CoV values from mixers operating with different power levels werenormalized to a CoV of 10 percent by this assumption. The assumed scaling relationship was asfollows:

CoV10% * PL10% = CoVx%*PLx%

� � � � % =� � � % � � � � %

� � � � � %Where,

� � � � � % = Coefficient of variation of 10 Percent� � � � % = Power level at 10 percent CoV� � � � % = Coefficient of variation of test (x) percent� � � % = Test power level at x percent CoV

The linear relationship is suggested by analogy to the formula for pumping power, which is theproduct of system head and flow (velocity). A squared relationship would be suggested by the

formula for energy ( ��

2 � � ).The authors have compared this linear relationship to alternative

relationships in which the CoV at 10 percent was set equal to the square root or the square of theratios of the test CoVs times the power level at the test CoV. Neither of these alternativessignificantly changed the order of comparison of the different types of mixers, although absolutevalues of the predicted power level varied widely with these different assumptions for scaling. Thelinear relationship has the virtue of simplicity and that it produces a power level for the Hydrofoil IImixer at 10 percent CoV appropriately close to the measured value at 9.2% CoV. The authorsconsider this an area where future research could resolve the most appropriate scaling relation bycomparing measured CoV values (by field test or CFD) for different power levels with the samemixer in the same geometry.

Figure 17. Initial simulation of horizontal propeller mixers.

Figure 18. Optimized simulation of horizontal propeller mixers with increased MLR flow.

Figure 19. Simulation of vertical hyperboloid mixers with increased MLR flow.

Table 5. Comparison of required volumetric power input for different mixer types.

Type ofMixing Reference Basis of Test CoV

Power Level(W/m3)

Equivalent 10%CoV Power Level

(W/m3)

Pumped jet Samstag et al. (2012) 2.5 m/sec jet CFD 50.0% 7.7 38.5

Pumped jet Samstag et al. (2012) 3.0 m/sec jet CFD 40.0% 13.0 52.0

Pumped jet Samstag et al. (2012) 3.5 m/sec jet CFD 12.0% 20.7 37.0

Pumped jet Samstag et al. (2012) 4.0 m/sec jet CFD 9.0% 30.8 27.7

VerticalHydrofoil I Carollo Engineers (2013) Field test 3.7% 7.9 2.9

VerticalHydrofoil II Carollo Engineers (2013) Field test 9.2% 1.4 1.3

VerticalHydrofoil III Carollo Engineers (2013) Field test 5.5% 7.5 4.1

VerticalHyperboloid I Carollo Engineers (2013) Field test 7.5% 5.3 4.0

VerticalHyperboloid Oton et al. (2009) Field test 11.0% 4.0 4.4

VerticalHyperboloid w/MLR Wicklein et al. (2013) CFD Simulation 2.1% 6.8 1.4Horizontalpropeller(initial) Wicklein et al. (2013) CFD Simulation 10.3% 8.3 8.5

Horizontalpropeller (final) Wicklein et al. (2013) CFD Simulation 5.4% 8.3 4.5Horizontalpropeller w/MLR Wicklein et al. (2013) CFD Simulation 1.9% 8.3 1.6

CONCLUSIONS

Three different calibration techniques have been discussed for calibration of CFD models foractivated sludge mixing: solids profile tests, velocity profile tests, and RTD tests. Of these, thesolids profile tests is recommended, since it measures most directly the objective of activated sludgemixing; solids distribution, the measure of which, as CoV, is commonly used for specification ofactivated sludge mixing devices.

A protocol has been developed for comparisons of the efficiency of activated sludge mixingsystems using field-calibrated CFD models. A crucial difference in this protocol from previouswork is to incorporate the effects of solids settling and transport on fluid motion. Density couplingin the CFD model was found to be necessary to adequately represent mixing efficiency. Elements ofthe protocol have been applied to jet aeration and mixing, vertical shaft hyperboloid mixers,horizontal propeller mixers, and diffused aeration. We propose this protocol as a comprehensiveapproach to optimizing activated sludge mixing and urge its adoption in future studies. We suggestthat future work be done to further identify the appropriate scaling relationship between CoV andthe unit power ratio. The current protocol uses a linear relationship, but other scaling approachesshould be investigated.

ACKNOWLEDGEMENTThis work has been accomplished as part of Carollo Engineers’ CFD modeling practice for severalclients. Support for this work by these clients and by Carollo Engineers is gratefully acknowledged.Field testing for this work was assisted by staff from Carollo Engineers Orlando office, includingRoderick Reardon, Sudhan Paranjape, and Erica Stone, whose help was greatly appreciated.

REFERENCESAmerican Public Health Association (APHA), the American Water Works Association (AWWA),

and the Water Environment Federation (WEF) (2013) Standard Methods for theExamination of Water and Wastewater, Online Edition(http://www.standardmethods.org/store/ProductView.cfm?ProductID=63).

Bosma, A.J. and Reitsma, B.A. (2007), Hydraulic Design of Activated sludge Tanks with CFD.Proceedings of the COMSOL Users Conference, Grenoble.

Carollo Engineers (May 2013) Orange County Utilities, Mechanical Mixing Equipment Evaluation,Technical Memorandum, Final.

Crosby. R. M. and Bender, J.H. (1980) Hydraulic Considerations That Affect Secondary ClarifierPerformance, EPA Technology Transfer, United States Environmental Protection Agency,905N80001.

Daigger, G.T. (1995) Development of Refined Clarifier Operating Diagrams Using UpdatedSettling Characteristics Database. Water Environment Research, 67, 95.

Danckwerts, P. V. (1953) Continuous flow systems. Chem. Eng. Sci. 2, 1-13.

Griborio, Alonso; Rohrbacher, Joe; Taylor, Ronald; Pitt, Paul; Latimer, Ron (2008) “Evaluation ofWet Weather Strategies and Clarifier Optimization Using State-of-the-Art Tools,”Proceedings of the Water Environment Federation, WEFTEC 2008, Session 61 throughSession 70, pp. 4957-4968 (12).

Larsen, P. (1977) On the Hydraulics of Rectangular Settling Basins, Experimental and TheoreticalStudies. Department of Water Resources Engineering. Lund Institute of Technology /University of Lund, Report No. 1001, Lund, Sweden.

Levenspiel, O. (1972) Chemical Reaction Engineering, John Wiley & Sons.

Oton, S. et al. (2009) The Fine Line Between Thorough Mixing and Energy Consumption. WEFNutrient Removal Conference Proceedings, 2009.

Samstag, R.W. and Wicklein, E. A. (2014), A Protocol for Optimization of Activated SludgeMixing. Proceedings of the 4th IWA/WEF Wastewater treatment modelling seminar, Spa,Belgium.

Samstag, R.W., Wicklein, E.A., Reardon, R. D., Leetch, R. J., Parks, R. M., Groff, C. D. (2012),Field and CFD Analysis of Jet Aeration and Mixing. Proceedings of the Water EnvironmentFederation, WEFTEC Conference.

Samstag R. and Griborio, A. “Calibration and validation of CFD Models – Case Study: CFDModeling of Secondary Clarifiers.” Second International Water Association/WaterEnvironment Federation Wastewater Treatment Modeling Seminar, Mont-Sainte-Anne,Quebec, Canada. March 28-30, 2010.

Samstag, R., Zhou, S., Chan, R., Royer, C., and Brown, K. Comprehensive Evaluation ofSecondary Sedimentation Performance. Proceedings of the Water Environment Federation82nd Annual Technical Conference and Exposition, New Orleans, LA. 2010.

Edward A. Wicklein and Randal W. Samstag (2009) Comparing Commercial and Transport CFDModels for Secondary Sedimentation. Proceedings of the 81st Annual WEFTECConference; Orlando, Florida.