cfd for wastewater draft final - · pdf file• 3d fluent cfd • 1,100,000 hexahedral...

CFD for WastewaterPresentation to the Department of Civil

and Environmental EngineeringUniversity of California, Los Angeles

19 November 2015

Randal W. Samstag, MS, PE, BCEE

Civil and Sanitary Engineer

Bainbridge Island, WA US

CFD for Wastewater

• What is it?

• How is it done?

• What is it good for?

• Who is doing it?

“If we know what is happening within the vessel,then we are able to predict the behavior of thevessel as a reactor. Though fine in principle, theattendant complexities make it impractical to usethis approach.” – Octave Levenspiel (1972)

Computational fluid dynamics (CFD) changesthis picture. Using CFD, we can compute three-dimensional velocity fields and followinteractions of reactants and products througha tank. We can use this information to optimizetank geometry.

Types of Models

Black Box Models Grey Box Models

Glass Box Models

Black Box Models

Principle:

• Output = f (Input, Volume)

Examples:

• RTD Theory

• Box and Jenkins ARIMA

These can give approximate results because outputs usually are related toinputs and volume, but can’t tell us anything about the influence of theshape of the reactor on the outputs.

Grey Box Models

Principle:

• Use knowledge about thereactor plus inputs andvolume to predict outputs.

Examples:

• Tanks in Series Models

• Axial Dispersion Models

Grey box models use macroscopic analogs of diffusion over the tank topredict outputs. The IWA ASM models are often used in this type ofmodel.

Glass Box Models

Principle:

• Use three dimensionalvelocity field model as basefor prediction of outputs.

Examples:

• CFD Models

Glass box models predict the impact of geometry on process outputsbased on the three-dimensional velocity field. This realizes Levenspiel’s“impractical” goal.

CFD: What is it?CFD Solves the Reynolds Averaged Navier-Stokes

(RANS) Equations by Numerical Schemes

• Continuity Equation: Law ofMass Conservation

• Momentum Equations:Newton’s Second Law(Incompressible Flow )

0

i

i

x

u

t

ii

t

ij

ij

i Fx

u

x

p

x

uu

t

u

j

2

21

Breakdown of the MomentumEquation

ii

t

ij

ij

i Fx

u

x

p

x

uu

t

u

j

2

21

Unsteady Term

Advective TermPressure Term

Diffusion TermSource Term

The Momentum Equations are a SpecialCase of the Generalized Transport Equation

Sxx

uxt jj

j

j

)()()(

Advection Diffusion

Unsteady

Source

Discretization Techniques

• Finite difference

• Method of weighted residuals (finite element)

• Finite volume formulation

• Grid-less methods

All of these have been used in CFD for wastewater. Finitedifference was the first technique used. Finite element has beenused for clarifier modeling. Finite volume formulation is themost common commercial CFD software approach. Grid-lessmethods have not been much used, but may be promising.

How to eliminate pressure?

• Transform the governing equations:

– Vorticity / stream function method

• Use iteration and convergence

– SIMPLE

Both of these methods have been used in CFD for wastewater.

Turbulence Modeling• Simplest model: Prandtl’s Mixing Length

Hypothesis (Plane mixing layer, width δ)

• Two Equation Models: k – epsilon

y

ulmt

2

kMbk

jk

t

j

i

i

SYPPX

k

XkU

Xk

t

)()(

S

kCPCP

kC

XXU

Xtbk

jk

t

j

i

i

2

231 )()()(

2kCt

Rodi, W. (1980) Turbulence models and their application in hydraulics: A state-of-the art review.

07.0

ml

3D transport models can be coupled to the velocitycalculations to simulate sedimentation and mixing.

• Solids Transport

• Vesilind settling

• Density couple

kCos eVV

s

ww c

1

k

s

is

t

ii

i

x

CV

x

C

xx

Cu

t

C

Samstag, et al. (1992) Prospects for Transport Modelling for Process Tanks,WST, Vol. 26. No. 5-6. pp. 1401-1410.

3D transport models can be implemented forwastewater quality parameters as well.

Biokinetic Models

– ASM Models

– Advanced oxidationModels

– Disinfection models

Sobremisana, Ducoste, de los Reyes III (2011)

Multi-phase models can simulatemotion of water and air.

0)(

id

d vtt

0)(

ic

c vtt

idiiV

i

djid

j

id Tvux

pvv

xv

t,)()()(

iciiV

i

cjic

j

ic Tvux

pvv

xv

t,)()()(

Crowe, Sommerfield, and Tusji (1998) Multiphase Flows with Droplets and Particles.

How to Do CFD?Good Modelling Practice

Wicklein et al. (in press)

Good Modelling PracticeInitial Scoping

Wicklein et al. (in press)

Good Modelling PracticeBuild Geometry and Mesh

Wicklein et al. (in press)

Good Modelling PracticeSolve

Wicklein et al. (in press)

Good Modelling PracticePost Process

Wicklein et al. (in press)

How to do CFD?Software

• Hand Coded– Fortran– C++

• Commercial platforms (Examples)– ANSYS (Fluent and CFX)– Cd-adapco– Flow-3D– Comsol– PHOENICS

• Open source platforms– OpenFOAM

Hand Coded InterfaceTANKXZ

Fluent Interface

Standard OpenFOAM Interface

Visual CFD Interface for OpenFOAM

2015 OpenFOAM WorkshopSurfers Paradise

• Presentations by IWA WGmembers

• Teaching by NelsonMarques on OpenFOAM– Modeling– Meshing– Simulation– Post-Processing

• Example cases– Parshall Flume– Flow Splitter– Clarifier– UV Channel

OpenFOAM WorkshopCase Study: Splitter Box

OpenFOAM WorkshopCase Study: Activated Sludge Clarifier

What can be done with CFD?Wastewater Treatment Examples

Case Study: WWT Hydraulics

• Adverse Hydraulics

Vortices

Pre-rotation

Turbulence

Velocity Distribution

• Leading to

Decreased capacity

Poor flow distribution

Cavitation

Excessive wear

Provided by Jim Wicks, PhDThe Fluid Group

Case Study: WWT HydraulicsDrop Shafts

To transfer flood waterand sewage rapidlydown into a main sewerwithout blockage

Multiphase flow

Velocity dependent gritdeposition

Provided by Jim Wicks, PhD

Case Study: WWT HydraulicsFlow Splits

Equal transfer ofwastewater to multipledownstream units(ensuring solidscomponent accounted forcorrectly)

Velocity dependent gritdeposition

Multiphase flow

Provided by Jim Wicks, PhD

Case Study: BiokineticsEindhoven WWTP Facility

(The Netherlands)

From Rehman (2014)Provided by Ingmar Nopens, PhDGhent University

Configuration of the Bioreactor

Geometry

Diffuserssimplification

Vector Plots

• Bad mixing zones dueto recirculation

• Averaged out insystemic modeling

• Change in aeration leads to different flow patterns

Vertical cross section inthe aerated region

Dissolved oxygen and ammonia conc. (mg/L)

Oxygen concentration at 3.45m depth Ammonia concentration at 3.45mdepth

Case Study: Jet Mixing in a sequencingbatch reactor (SBR)

Case Study: MixingCFD of Jet mixing and aeration

415,350 mixed tetrahedralcells

2,108,308 nodes

Inlet flow into jet nozzles

Outlet flow to pump suction

Air added as second phase

Solids transport, settling, anddensity impact modeled byUDF

Samstag, Wicklein, et al. (2012)

Mesh projected onto modelsurfaces.

Case Study: MixingVelocity profiles for pumped mixing and aeration

Simulated Pumped Mixing Profile Simulated Aeration Profile

Case Study: MixingComparison of pumped mix velocity profiles for increasing jet velocities

Existing(2.5m/secJet)

3.0m/secJet

3.5m/sec Jet

4.0 m/secJet

Case Study: MixingComparison of solids profiles for increasing jet velocities

Existing(2.5m/secJet)

3.0m/secJet

3.5m/sec Jet

4.0 m/secJet

Case Study: MixingComparison of density-coupled and neutral density simulations

Density-coupled

Solids transport modelcalculates the local solidsconcentration based onflow regime.

The influence of the localsolids concentration on thelocal density is theniteratively calculated.

This approach was verified bythe field solids profile testdata.

Case Study: MixingComparison of density-coupled and neutral density simulations

Neutral Density

Solids transport model calculatesthe local solids concentrationbased on flow regime.

Influence of the local solidsconcentration on the localdensity was turned off.

This approach over-predictedmeasured solids mixing.

Case Study: Activated SludgeClarification

Case Study: Activated Sludge Clarifier

• Radial flow clarifier

• Questions:

– Optimum Depth?

– Optimum Inlet ?

– Optimum Feedwell?

– Optimum Effluent Zone?

Samstag and Wicklein(2010)

Radial Flow Clarifier CFD Model

• 3D Fluent CFD

• 1,100,000 hexahedralcells

• K-epsilon turbulencemodel

• User defined functions(UDF) to implement– Solids settling and

transport

– Density coupling

Case Study: Activated Sludge ClarifierCalibration CFD

Case Study: Activated Sludge ClarifierAlternative Inlet Configurations

Case Study: Activated Sludge ClarifierAlternative Velocity Vector Plans

Case Study: Activated Sludge ClarifierAlternative Velocity Plans

Case Study: Activated Sludge ClarifierAlternative Velocity Profiles

Case Study: Activated Sludge ClarifierComparison Solids Profiles

Case Studies: Disinfection

• Dye, disinfectant, andmicroorganismtransport in a chemicaldisinfection system

Provided by StevenSaunders, P.E.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5

no

rmali

zed

co

ncen

trati

on

t/TTDT

Residence Time Dstribution

T10

TTDT

T90

System Hydraulic Efficiency

The hydraulic efficiency of a contact tank systemis measured in CFD simulations with a tracer.Using a time dependent CFD model, tracer isintroduced at the point of disinfectant injection andits concentration is monitored at the model outlet.

Data gathered at the model outlet are used to plota residence time distribution (RTD) curve. Theslope and inflection points of the RTD curve areindicative of the hydraulic performance of thecontact system. Data points T10, TTDT and T90 areinputs used for accepted evaluation protocol.

T10: time at which tracer concentration hasreached 10% of the target value

TTDT: Theoretical Detention time ( Voltank/Qeffluent )

T90: time at which tracer concentration hasreached 90 % of the target value

BF: Baffle Factor – T10/TTDT Values near 1.0indicate good plug flow. Values lower than 0.3indicate some short circuiting is taking place.

MI: Morrill Index - T90/T10 Values greater than5.0 indicate some flow is getting hung up inrecirculation zones.

Streaklines colored by elapsed timenormalized against TTDT

Disinfectant Decay

The efficacy of the contact system is dependenton the time the active microorganisms in theeffluent are exposed to disinfectant at sufficientconcentration to neutralize them.

Disinfectants currently in use like sodiumhypochlorite or peracetic acid begin to decayimmediately upon introduction to the effluentstream. Their levels of concentration aremodeled by UDF based on published data.

Contours of disinfectant concentration normalized against targetconcentration level. Contour level 1.0 denotes the targetconcentration.

Log(N/N0) (bio. organisms)

Neutralization ofMicroorganisms

The rates at which microorganisms areneutralized are specific to the disinfectant in useand the target microorganism. For example,different coliforms will have differing sensitivitiesto the disinfectant in use. Their neutralization issimulated using UDF’s based on published data.

Contours of microorganism population. N is the localpopulation and N0 is the population in the effluent prior toexposure to the disinfectant.

Case Study: Dissolved Air FlotationDAF

Seawater DAF –desalination use

Prediction of absolutemaximum capacity(typically 10-20% aboveabove design max)

Comparison of nozzledesign and airintroduction

Provided by Jim Wicks,PhD

Case Study: DigestersDigesters

Non-Newtonian flow above3% DS

Rheology for all digestatesdifficult to measure.Typically we use of a libraryof values for various drysolids contents (and forvarious sludge types)

Useful for optimisingmethane gas production;reducing foaming; limitingsolids deposition andavoiding ‘dead’ volumes.

Provided by Jim Wicks,PhD

Case Study: DigestersDigesters

Structurally quite simpleand can compare differentmixers, baffles around theouter wall, distribution ofgas lances at the base etc

Dynamic modelling ispractical, with changes togas flow rate and locationover time to shift material

Velocity and (virtual)tracer measurementspractical (Volume withvelocity > 0.1 m/s used asa coarse measure inliterature)

Case Study: DigestersPitfalls

High levels of mixing donot equate linearly tomethane gas production

Density impact ofdigester solidsconcentration

Changes to dry solidscontent over time asdigestion processcontinues

Potential loss of activevolume around the basedue to solids deposition

Ongoing work

CFD plus “ADM1”

Who is Doing CFD?

• IWA CFD Working Group

“The WG intends to solve shortcomings arising fromlack of knowledge of CFD in the water andwastewater community in the short term byproducing papers and books as well as hands-ontraining for the IWA MIA members (and beyond).Furthermore, a book dedicated for training newpeople in the water/wastewater field will beproduced.”

IWA CFD Working GroupManagement Team

IWA CFD Working GroupWork Products

• Published Papers– Good Modelling Practice in Applying Computational Fluid Dynamics for WWTP Modelling, WEFTEC

2012– A protocol for the use of computational fluid dynamics as a supportive tool for wastewater treatment

plant modelling, WST– Computational Fluid Dynamics: an important modelling tool for the water sector, IWC Conference

• Submitted Papers– Good Modelling Practice in Applying Computational Fluid Dynamics for WWTP Modelling, WST– CFD for Wastewater: State of the Art, WST

• Workshops– WEFTEC 2012 (Fluent)– WWTMod 2012– Watermatex 2015 (OpenFOAM)– WWTMod 2016 (Proposed: Flow-3D)– IWA Biennium 2016 (Proposed)

• Book Plans– IWA Scientific and Technical Report– CFD for Water Book for Students and Practitioners

There is also a LinkedIn GroupCFD for Wastewater

Thank you!Questions?

Web: http://rsamstag.com/

Phone: +1 (206) 851-0094

Email: randal.Samstag@rsamstag.com

The Momentum Equations are a SpecialCase of the Generalized Transport Equation

Sxx

uxt jj

j

j

)()()(

)()()()(

w

zv

yu

xu

xj

j

)()()()(zzyyxxxx jj

Advection

Diffusion

z,w

y,v

x,u

The RANS Equations in RectangularCartesian Coordinates

0)()()(

w

zv

yu

x

)(1

2

2

2

2

2

2

z

u

y

u

x

u

x

p

z

uw

y

uv

x

uu

t

ut

)(1

2

2

2

2

2

2

z

v

y

v

x

v

y

p

z

vw

y

vv

x

vu

t

vt

z

w

wt g

z

w

y

w

x

w

z

p

z

ww

y

wv

x

wu

t

w

)()(

12

2

2

2

2

2

z,w

x,u

y,v

McCorquodaleComposite Settling Model

Settling Zones

1) Non-settleables (C<CN)

2) Discrete (CN<C<CD)

3) Transition (CD<C<CH)

4) Hindered (CH<C<CC)

5) Compressive (C>CC)

Formula

1) Vs = 0.0

2) Vsd = � � �

3) Vs = ζ �� +(1-ζ)Vsd

=

4) Vs = Vo�

5) Vs = VC

Bürger-DiehlComposite Settling Model

Settling Zones

1) Clarification (X<XTC)

2) Compression (X>XTC)

Formula

1) Vs = Vo�

2) Vs = VT(1-Dcomp )

VT = Vo�

Dcomp = �

� � � �

= ( �

� � � �

)bcomp

McCorquodale Rheology Model

1) X < 1 g/L

2) X >1 g/L

1) νb=νoe1.386 X

2) νb=2.9νoe0.322X

McCorquodale Flocculation Model

Shear induced

/ = KB*X*Gm- KA*X*G

Differential settling

1/ = −9/4 � � 1 2/ρ1ρ2(1+2 1/ 2) 1/ 1( � 2− � 1)