Università di PalermoDipartimento di Ingegneria Civile, Ambientale, Aerospaziale, dei Materiali (DICAM)

Athens

14-16 September 2016

Interlinkages between operational

conditions and direct and indirect

greenhouse gas emissions in a moving bed

membrane biofilm reactor

G. Mannina, M. Capodici, A. Cosenza, D. Di Trapani

Introduction

Wastewater treatment entails:

• direct emissions of greenhouse gases (GHGs), such as

nitrous oxide (N2O)

• indirect emissions resulting from power requirements

N2O Unwanted even at small levels due to the high global

warming potential 310 higher than CO2

Introduction

reduction of NO2- as terminal

electron acceptor to N2O (AOB

denitrification)

incomplete oxidation of

hydroxylamine (NH2OH) to NO2

intermediate of the incomplete

heterotrophic denitrification

N2O Production Pathways

Nitrification Denitrification

Introduction

Process operations aimed at the reduction of N2O could

conflict with the effluent quality and increase the operational

costs

Challenge

Operationalcosts GHG

emissionEffluentquality

To identify GHG mitigation strategies as trade-off between operational costs

and effluent quality index is a very ambitious challenge

Aim

Performing a multivariate analysis

+

University Cape Town (UCT) moving bed (MB) membrane

bioreactor (MBR) pilot plant.

Simple model for interlinkage among operational conditions/influent

features/effluent quality and emitted N2O.

Methods

Pilot plant

Anaerobic Tank Anoxic Tank Aerobic Tank

MBR Tank

Clean In Place Tank

ODR

Qin

Qout

QR2

QR1

QRAS

Gas

Funnel

Gas

Funnel

Gas

Funnel

Gas

FunnelSuspended

Carriers

QIN = 20 L h-1

QR1 = 20 L h-1

QRAS = 80 L h-1

150 days of experimentation

Mixture of real and synthetic

wastewater!

Three experimental phases:

Phase I: SRT = ∞

Phase II: SRT = 30 days

Phase III: SRT = 15 days

QR2 = 100 L h-1

QOUT = 20 L h-1

Pilot plant

PURON 3 bundle ultrafiltration module (pore size

0.03 μm, surface 1.4 m2)

AMITECH carriers in anoxic and aerobic

reactors with a 15 and 40% filling fraction

respectively

TSS, VSS, CODTOT, CODSOL, N-NH4,N-NO3, N-NO2,

TN, TP, P-PO4, DO, pH, T,

N-N2O as gas and dissolved

Two time per week in each tank

Measured data

Indirect emissions

Pw [kWh m-3] energy required for the aerationPeff [kWh m-3] energy required for permeate extractiongpower,GHG geconversion factors, 0.7 gCO2eq and 0.806 € kWh-1

EF [€m-3] cost of the effluent fine including N2O

The Operational Costs (OCs) were evaluated using conversionfactors (Mannina and Cosenza, 2015 ):

Indirect emissions

QIN and QOUT are the influent and effluent flow, respectively;Daj is the slope of the curve EF versus Cj

EFFwhen CjEFF< CL,j (in this

case, the function Heaviside =0);Dbj represents the slope of the curve EF versus Cj

EFFwhen CjEFF> CL,j

(in this case, the function Heaviside =1);b0,j are the increment of the fines for the latter case.

The effluent fine (EF) was evaluated using:

EF =1

t2 - t1×

1

QIN

×QOUT ×Da j ×Cj

EFF + QOUT( )×

b0, j + Cj

EFF -CL, j( )×Db j - Da j( )××

××

×

×

××

×

×

×××Heaviside×Cj

EFF -CL, j( )( )j=1

n

××

×

××

×

×

××

×

×

××

×

×

××

t1

t2

× ×dt

Indirect emissions

bCOD, bTN, bPO, bN2Ogas and bN2O,L are the weighting factors of theeffluent CODTOT, TN, PO, liquid N2O in the permeate and gaseousN2O.

The effluent quality index (EQI) was evaluated using:

EQI =1

T×1000×

bCOD ×CODTOT + bTN ×TN + bPO×PO+

bN2Ogas×N2Ogas + bN2O,L ×N2OL

×

××

×

×××QOUT ×dt

to

t1

×

Multiregression analysis

Performed to point out general relationships for the N-N2O

and the plant operation conditions or the available measured

data

Two type of analysis

Complex regressionsSimple linear regression

Simple linear regression

Y = dependent variable; X1 = independent variable; c1, c2 regression coefficients

N2O-N fluxANAER (N2O-N flux emitted from the anaerobic tank)

N2O-N fluxANOX (N2O-N flux emitted from the anoxic tank)

N2O-N fluxAER (N2O-N flux emitted from the aerobic tank)

N2O-N fluxMBR (N2O-N flux emitted from the MBR tank)

N2O-N dissolvedOUT (N2O-N permeate dissolved concentration)

Dependent variables

Complex regressions

Y = dependent variable; X1,…,Xm= independent variable; c1,…,cn regression coefficients

∑N2O-N flux (sum of the N2O-N flux emitted from each tank)

N2O-N dissolvedOUT (N2O-N permeate dissolved concentration)

Dependent variables

Multiple linear (LINm)

Multiple exponential (EXP)

Sum of exponential

(SumEXP)

Independent variables Influent concentration

Effluent concentration

Intermediate concentration

N-NO2_AER, N-NO2_ANOX, DOAER, DOANOX, pHAER, pHANOX, DOMBR

CODTOT,OUT, BOD5,OUT, N-NH4,OUT, N-NO3,OUT, NO2-N,OUT, P-PO4,OUT

CODTOT, IN, N-NH4,IN, PTOT,IN, P-PO4,IN, C/N

hCOD,BIO, hCOD,TOT, hNITR, hDENIT, hNTOT, hP

Performance indicators

Operational conditions

TSS*, SRT, Biofilm*

Numerical settings

10,000 Monte Carlo simulations varying coefficients

Evaluation of Nash and Sutcliffe efficiency for each simulation

Ymeas,i = measured value of the ith dependent state variable; Ysim,i = simulated

value of the ith dependent state variable; Yaver,meas,i = average of the measured

values of the ith dependent state variable

Results

Simple linear regression analysisMaximum efficiency

Dependent variables

N2O-N

fluxANAER

N2O-N

fluxANOX

N2O-N

fluxAER N2O-N fluxMBR N2O-N dissolvedOUT

Phase

I

Independent

variableTSS NO2-NANOX NO2-NANOX NH4-NIN NO3-NOUT

Efficiency 0.11 0.52 0.52 0.2 0.1

II

Independent

variableNO2-NANOX NO2-NANOX DOAER hNITR CODOUT

Efficiency 0.35 0.6 0.5 0.26 0.72

III

Independent

variablepHAER Biofilm Biofilm PO4-POUT NO2-NAER

Efficiency 0.12 0.36 0.67 0.52 0.94

Varying the SRT different variables can be adopted to

predict the N2O

N2O dissolved in the permeate

depend on CODOUT

NO2 accumulation influence the N2O

production

Simple linear regression analysis

0 0.002 0.004 0.006 0.008Parameter- c2

N2O-N dissolvedOUT

(f)

-19 -18.5 -18 -17.5 -17Parameter- c1

N2O-N flux AER

(d)

-6.5 -6 -5.5 -5 -4.5Parameter- c1

N2O-N flux ANOX

(b)

0.0

0.1

0.2

0.3

0.4

9 9.5 10 10.5 11

Effic

iency [

-]

Parameter- c2

N2O-N flux ANOX

(a)

0.4

0.5

0.6

0.7

34 34.5 35 35.5 36

Effic

iency [

-]

Parameter- c2

N2O-N flux AER

(c)

0.0

0.2

0.4

0.6

0.8

1.0

0 0.02 0.04 0.06

Effic

iency [

-]

Parameter- c1

N2O-N dissolvedOUT

(e)

Scatter plots Phase III

High dependence Poor dependence

Combining effect

Complex multiregression analysisLINm - Maximum efficiency

∑N2O-N flux N2O-N dissolvedOUT

Efficiency Efficiency

Independent

variable0.015 0.244

C/N

N-NH4,IN

TSS

Biofilm

SRT

DOAER

N-NO2_AER

pHAER

DOANOX

N-NO2_ANOX

pHANOX

LINm poorly reproduces the measured

data for ∑N2O-N flux (efficiency 0.015).

Efficiency obtained for the N2O-N

dissolvedOUT is slightly higher than for

∑N2O-N flux (equal to 0.244)

Complex multiregression analysisEXP and SumEXP - Maximum efficiency

EXP SumEXP

∑N2O-N flux N2O-N dissolvedOUT ∑N2O-N flux N2O-N dissolvedOUT

Efficiency Efficiency Efficiency EfficiencyIndependent

variable 0.125 0.164 0.198 0.178C/N

C/N

N-NH4,IN

N-NH4,IN

TSS

TSS

Biofilm

Biofilm

SRT

SRT

DOAER

DOAER

N-NO2_AER

N-NO2_AER

pHAER

pHAER

DOANOX

DOANOX

N-NO2_ANOX

N-NO2_ANOX

pHANOX

pHANOX

Poor efficiency values obtained

for both the investigated

dependent variables

Conclusions

Reasonable agreements for simple regression equations

Dependency of N2O flux with SRT and plant sections

SRT of Phase III makes the conditions of N2O production

more sharped

None of the investigated equations for complex multivariate

analysis is able to provide satisfactory efficiencies

Message to take home!

The interactions among the key factors affecting the

N2O make difficult to establish an unique equation valid

for different operational conditions for predicting N2O

Università di PalermoDipartimento di Ingegneria Civile, Ambientale, Aerospaziale, dei Materiali (DICAM)

Athens

14-16 September 2016

Thank you for your attention

Giorgio Manninagiorgio.mannina@unipa.it

Acknowledgements

This research was funded by Italian Ministry of Education, University and Research (MIUR) through the

Research project of national interest PRIN2012 (D.M. 28 dicembre 2012 n. 957/Ric − Prot. 2012PTZAMC)

entitled “Energy consumption and GreenHouse Gas (GHG) emissions in the wastewater treatment

plants: a decision support system for planning and management − http://ghgfromwwtp.unipa.it”

PRIN Research project of

national interest

2012

Green House Gases emissions from Wastewater

Treatment Plants

Università di PalermoDipartimento di Ingegneria Civile, Ambientale, Aerospaziale, dei Materiali (DICAM)

Athens

14-16 September 2016

FICWTMOD2017 - Frontiers International Conference on wastewater treatment and modelling

1st International Conference

21 – 24 May 2017, Palermo, Italy Supported by

www.ficwtmod2017.it