Intelligent membrane management to enhance resource recovery opportunities Sandeep Sathyamoorthy IWA Aspire 2019, Hong Kong November 1, 2019

IOT & AI now commonly impact & benefit our daily lives

2

And is coming to a WRRF screen near you

Figure courtesy of Atonix Digital

Framework for today’s talk

Opportunities Solutions & Strategies

The Fine Print…

3

Opportunities to maximize the value of analytics, machine learning and connected analyses

4

0% 20% 40% 60%

PerformanceMonitoring

AssetManagement

TreatmentOptimization

2018 Strategic Directions: Black & Veatch Water Industry Report and Survey

Operational areas where data analytics and

automated monitoring will help improve

your organization?

Energy costs of membrane management in water reuse schemes

5

Non-potable Reuse Potable Reuse

Influent Pumping

5% Headworks1%

Biological Process

26%

MBR Management

39%

Disinfection10%

Odour Control15%

Misc.4%

Pumping6%

Headworks2% Biological

Process9%

MBR Management

13%

RO Management

41%

Advanced Oxidation

22%

Odour Control7%

Misc.0%

5% reduction in MBR management costs can result in 1-2% OPEX reduction

5% reduction in MBR & RO management costs can result in 2-3% OPEX reduction

Membrane replacement costs are directly influenced by operation and water quality

6

Water Quality (DOC, TOC,

MLSS, NHX-N,…)

Filterability &

Flux Decline

Short & long term

membrane performance

Membrane OPEX &

Replacement Costs

Membrane Operation

(flux, initial flux) Enhanced operation to increase membrane life results in 2%-5% reduction in 20-yr present worth

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20Years in Operation

Operational modifications to decrease membrane replacement frequency

Process Risk Management & Treatment Performance Optimization

7

Virus Giardia Cryptosporidium Regulatory Requirement

12 10 10

BNR-MBR RO UV-AOP Free-Cl2 Underground Retention

UF RO UV-AOP Free-Cl2 Underground Retention BNR

Process Risk Management & Treatment Process Optimization in Potable Reuse

8

Biological Treatment AWPF Supply Water Feed System

DPR Supply Water

Manage the quality/quantity of this (intermediate) raw material Other Uses

Critical Control Point

RAS

WAS

UF Membrane

Solutions & Strategies – Some Examples

9

Solutions and Strategies

Figures courtesy of Atonix Digital

Actively managing membrane energy requirements

10

In an MBR - energy costs driven by air scour

Current Approach(es) to MBR Air Scour

Continuous aeration (with high/low SPs)

Intermittent aeration (with high/low SPs)

=> Two basic control variables: Air Flow Time

• Results in (sometimes) OVER-scour & (sometimes) UNDER-scour

• Membrane fouling is a complex, multidimensional process impacted by a range of variables

• Need a scour tool capable of understanding, then managing scour based on fouling needs

Using machine learning & analytics tools for air-scour management

11

SRT

F/M

NHx-N

Flux

TMP, Permeability w/ history

Scour History

MLSS

Parameter X,Y,Z

.....

Neural Network Kohonen Bayesian

Deep Machine Learning & Scour Manager

Delphi knowledge (e.g., manual intervention - scour now)

Fouling model(s) Theoretical framework

Intelligent Adaptive MBR Scour Management System

Managing supply water quality in potable reuse

12

Measure Turbidity

No Action Required

Divert Supply Water Away from AWPF

Alarms Inform shift supervisors Initiate mitigation measures

Wait YY min

Alert Level Turbidity >

N

Y

Max. Level Turbidity >

N

Y

Inform Corrective Action Team shift supervisor, Sep’n process SME, AWPF lead operator, AWPF process SME

Initiate False Positive Verification

Automated Actions • Instrument self

verification Human Actions • Sampling system • Visual inspection

Wait XX min

Operational Verification & Recommended Actions

Operator System Reset

0

2

4

6

8

10

0 5 10 15 20Seco

ndar

y Eff

luen

t Tur

bidi

ty

(NTU

)

Secondary Effluent TSS (mg/L)

Using water’s fingerprint to manage attention points

13

0

2

4

6

8

10

0

5

10

15

20

Sec.

Eff.

Tur

b. (N

TU)

Sec.

Eff.

TSS

(mg/

L) S.E. SS S.E. Ntu

Turbidity/TSS

Fluorescence Spectrum

Early Warning Actions

KPIs

Attention Point Health

DECISION SUPPORT SYSTEM

• Big Data Analytics • Machine Learning

Using water’s fingerprint to manage attention points

14

0

2

4

6

8

10

0

5

10

15

20

Sec.

Eff.

Tur

b. (N

TU)

Sec.

Eff.

TSS

(mg/

L) S.E. SS S.E. Ntu

Turbidity/TSS

Fluorescence Spectrum

Early Warning Actions

KPIs

Attention Point Health

DECISION SUPPORT SYSTEM

• Big Data Analytics • Machine Learning

Figure courtesy of Atonix Digital

The Fine Print

15

things you need to really, really pay attention to if you want the system to actually work….

Xxxxx xxxx

Managing the data highway

16

What do you really want to

measure?

How much data is needed?

Accuracy of the data?

What do you want do with data?

Who needs the information?

• TSS vs. Turbidity • Ammonia-N in Oxic zone vs.

Supply Water Ammonia-N?

• Frequency

• Granularity – e.g., total particle counts vs. bins

• mg/L (ppm) vs. ug/L (ppb)

• Monitoring • Control • Reporting

Using the right tools for the job, in the right way

17

24-h Diurnal

2nd AE Zone 3rd AE Zone

Optimal Probe Range

Amm

onia

-N C

onc.

1st AE Zone

Ammonia-N concentration may be one of many predictor variables in a multivariate feature extraction algorithm or input to a neural net

Tools need to well looked after – enhancing the vernacular to include validation and data quality management

18

Process Function

Instrument Type Cleaning Validation Calibration Maintenance

Ammonia-N, Nitrate-N

Ion Selective Electrode (ISE)

1-2 times per week

2-3 times per week

1-2 times every 2 weeks

1-2 times per month

Dissolved Oxygen

optical (e.g., LDO)

1-2 times per week

2-3 times per week

1-2 times every 2 weeks

1-2 times per month

Turbidity Optical or Laser 2-3 times per week

1-2 times every 2 weeks

1-2 times per month

1-2 times per month

pH/ORP ISE 1-2 times per week

1-2 times per week

1-2 times per month

1-2 times per month

Managing instrument risks through self diagnostic tools

19

Key Take Aways

• IIOT/AI etc. is here – we need to “acclimate” and find the best applications

• Using AI is not likely to reduce the need for skilled operations professionals – BUT it will require new, enhanced skill sets

• GIGO – manage, maintain and validate instruments

20

Sandeep Sathyamoorthy +1.925.949.5913 SathyamoorthyS@bv.com