Condition monitoring for membrane systems W - … monitoring for membrane systems ... While for a 20-year life cycle cost calculation, ... ending with reverse osmosis to remove particles

Water scarcity

Increasinglevel of waterscarcity

Suffi cientwater resources

Water surplus

Water & Wastewater Asia • December / January 2011 35

PERFORMANCEMONITORING

Water stress is of increasing relevance on a global scale. More and more regions are affected by it. As of up to now, it is estimated that 700 million people in

43 countries are already facing water supply issues. Amongst others, water stress provides for negative impacts on health (reoccurrence of specifi c diseases for example), security (for example, risk of war on access to water sources) and economic issues (prohibition of growth due to lacking water suppliers). Available projections show a dramatic increase of countries affected by water stress in the next 10 to 15 years. Impacts of water stress will get more and more observable, for example, in the South Western part of the United States, the Mediterranean region, the Arabian Peninsula up to China or in Australia (see Figure 1).

Two thirds of the world’s population are expected to live in water-stressed countries by 2025 – from originally expected one third of the global population. The revision of this estimate demonstrates that it is important to keep a stronger focus on the topic of water stress, taking also into consideration that the total world’s water consumption is expected to further increase by 40% in 2025.

DesalinationTo overcome water stress and the related impacts, water desalination is one of the measures to support future water supplies security. Desalination is used for desalinating brackish water and seawater. Due to improvements in design and materials, capital and operational expenditures have been optimised over recent years – desalination is now a mature technology being increasingly attractive from an economic point of view as a water treatment technology. For desalination, different process technologies can be applied:

• T h e r m a l d e s a l i n a t i o n technologies such as multistage fl ash (MSF), multi-effect distillation

Condition monitoring for membrane systems

This paper outlines a new solution developed by ABB for advanced operation of membrane processes, specifi cally for reverse osmosis and nanofi ltration

Figure 1: Water Stress Map 2025

(MED), or vapour compression (VC) using the effect of evaporation/distillation.

• Membrane desalination technology such as reverse osmosis (RO) or nanofi ltration using the effect of physical separation based upon pressurising feed water and passing it through a semi-permeable membrane.

The most dominant process technology in recent years in terms of number of installations and additionally contracted capacity is reverse osmosis and this trend is expected to be continued as reverse osmosis provides for a much higher fl exibility in terms of operation. Thermal desalination technology is strongly used throughout the Arabian Peninsula and is expected to play an important role in this region in the future.

Energy consumptionA critical factor in desalination plant operations is energy consumption. While for a 20-year life cycle cost calculation, energy consumption is about 50% for thermal desalination (mainly steam energy is required), energy consumption represents 30% to 50% of life cycle costs for membrane processes with electrical energy mainly required for

36 Water & Wastewater Asia • December / January 2011

PERFORMANCEMONITORING

Figure 2: Reverse osmosis cost breakdown.pressurising feed water. In the following estimate for reverse osmosis (Figure 2), electrical energy makes up 44% of the total life cycle costs.

Comparing different membrane systems, the energy use in reverse osmosis and nanofiltration is higher as higher levels of pressure need to be applied following required process conditions. In the case of reverse osmosis, the pressure levels are required to overcome the osmotic pressure of the solute as well as to cause transport of the solvent from feed side to permeate side (product water).

Ineffi ciencies in plant operations cause higher energy consumption and thus negatively contribute to climate change which impacts the water supply situation as climate change is a main cause for water stress. Therefore, it is of utmost importance to further improve energy effi ciency in treatment processes in general and desalination in specifi c.

Better management of operationsFor desalination, the need for better management of operations moves on the agenda in order to reduce energy consumption, to minimise costs and to optimise productivity.

Required are energy-effi cient, highly reliable products and solutions, the consideration of the life cycle cost related aspects right from the conceptual stage as well as solutions for process optimisations integrated into the operational environment to provide for highest benefi ts.

This paper outlines a new solution developed by ABB for advanced operation of membrane processes, specifi cally for reverse osmosis and nanofi ltration.

Membrane systems in general are applied in different treatment applications including the desalination of brackish and seawater, the water and wastewater treatment as well as water reuse. The objective is to remove unwanted particles and to produce a product water stream fulfi lling quantitative and qualitative requirements. Different technologies are applied, starting with media fi ltration for removal of macro particles ending with reverse osmosis to remove particles belonging to molecular range size. Unwanted particles cover include organic (for example, bacteria) or inorganic (for example, salts) contaminants.

While for water and wastewater treatment, pressure levels are moderate ranging roughly from 1 to 5 bar (for example, media fi ltration application), pressure levels for desalination range from 5 to 80 bars causing high levels of energy consumption. The challenging aspect is now to get an indication of the optimal operation set-points to achieve most energy-effi cient and productive operation as these change over time with a changing operational characteristic of the membranes.

As the solute passes through the semi-permeable membrane, particles accumulate on the membrane surface of the pressurised feed water side leading to so-called

Figure 3: Impact of decreasing membrane performance.

“concentration polarization”. This accumulation causes fouling and thus blockage of membranes which ends up in reduced productivity (lower permeate fl ux). The decreasing membrane performance goes along with an increased differential pressure between feed water and reject water side and a decreasing permeate fl ux (see Figure 3).

The phenomenon of fouling is outlined in Figure 4 on page 37:

As fouling is a highly dynamic, non-linear process (see Figure 5) – depending on operational (feed pressure, feed fl ow) as well as environmental conditions (for example, feed water temperature or salt concentration) – an online analysis of the root cause is more or less impossible.

Taking proper maintenance actionsIn order to take proper maintenance actions, it is required to be aware of the condition of the membranes, taking into consideration the fouling and its dynamic nature. Increased fouling leads to energy-ineffi cient operation as the specifi c energy required to produce 1 m³ of permeate increases. Typical maintenance measures to overcome fouling are:

• Backwashing• Chemical cleaning• Partial membrane replacement• Total membrane replacementIn addition to the above given options, also operational

4% 7%5%

3%

44%

37%Capital costsElectrical energyChemicalsMembraneLabour and capacity chargesMaintenance

Permeatefl ux

Differentialpressure

Mem

bran

e pe

rfor

man

ce

Time

Separationlayer

Supportlayer

High pressure

Productwater

Water & Wastewater Asia • December / January 2011 37

PERFORMANCEMONITORING

Figure 4: Particles accumulation on membrane surface.

Figure 5: Dynamic nature of fouling.

conditions (set-points) can be optimised to minimise fouling and to achieve higher productivity. This needs to be addressed in a way considering the actual level of fouling (membrane condition) and to run a prediction based upon a model-based approach considering real-time and historical data.

Different approaches are available to determine the optimal point in time to take a maintenance measure and various R&D initiatives are going on addressing this specifi c operational aspect. A typical approach is to follow the recommendation as given by the membrane supplier, typically following an indication of predetermined time periods. To demonstrate the drawbacks of this approach, the following points should be taken into consideration:

• If condition of membranes does not require maintenance, additional costs (for example, for chemicals) and production losses might be the result.

• If membranes are in a condition that cleaning is overdue, membranes might have already been damaged up to a point where even chemical cleaning does not give required improvements in terms of restoring productivity.

Even approaches that are based on differential pressure of feed on reject side do not provide for capturing the dynamic nature of fouling.

Thus, a new approach has been developed to overcome drawbacks of existing approaches such as the aforementioned ones and to provide for online, real-time conditional assessment.

The developed solution consists of two modules, one to cover the functional scope of performance monitoring, the second to cover operation optimisation.

Performance monitoringThe function of the performance monitoring module is to calculate selected key performance indicators (KPI) that refl ect the dynamic nature of fouling and provide for proper monitoring of the membrane fouling condition. The calculation of the KPIs is done using a fi rst principle model as basis. For the calculation, nominal process data, such as feed temperature, feed pressure, differential pressure or reject fl ow rate, are required – no additional measurements are required and thus the solution can be added to new or existing installations,

The calculation is done on a train-by-train basis and the trains are described by models. As it can be seen from

Figure 6: Key performance parameters refl ecting dynamic nature of fouling

Inorganic andorganic particles

Results in membrane fouling

Semi-permeablemembrane

Reject

Feed waterTime

Fou

ling

rate

an

d Fo

ulin

g th

ickn

ess

Fouling Rate

Fouling Thickness

1st F

lush

ing

2nd F

lush

ing

3rd F

lush

ing

1st C

hem

ical

c

lean

ing

Part

ial

repl

acem

ent

Part

ial

repl

acem

ent

Part

ial

repl

acem

ent

Tota

lre

plac

emen

t

Monitoring withcleaning andfl ushing

Monitoring withoutcleaning andfl ushing

1st Cleaning 2nd Cleaning 3rd Cleaning 4th CleaningInitial state

Time

Per

form

ance

In

dica

tor

FP 1

FP 2

38 Water & Wastewater Asia • December / January 2011

PERFORMANCEMONITORING

Figure 7: Examples for alarm limit confi guration.

Figure 8: Results visualisation giving membrane condition and estimated due date per train.

– membrane requires cleaning, for example, < 60 days

Red

Yellow

Green

– membrane requires cleaning, for example, < 15 days

– membrane requires cleaning, for example, < 15 days > 60 days

Last Calculation10:30am, Monday, 14 Sept 2009

Current Time:9:30PM, Monday, 14 Sept 2009

Train 1 67.1 bar 67.2bar 67.2bar 680m3/hr 682m3/h 682m3/h 26 Sep 2009Train 2 66.4 bar 66.3bar 66.3bar 671m3/hr 672m3/h 672m3/h 22 Oct 2009Train 3 66.1 bar 66.0bar 66.0bar 669m3/hr 667m3/h 667m3/h 21 Nov 2009Train 4 65.2 bar 65.3bar 65.3bar 690m3/hr 691m3/h 691m3/h 26 Jan 2010Train 5 65.9 bar 65.8bar 65.8bar 685m3/hr 684m3/h 684m3/h 10 Mar 2010

Figure 6, the two main KPIs show an adverse effect with occurring fouling: While the one KPI increases (FP2), the other one decreases (FP1).

A combined analysis of both allows getting an insight to the actual condition. The calculation is done over time on a regular basis, for example, every three hours. Once required, for example, in case membranes have been chemically cleaned, the model is tuned to refl ect the real plant behaviour and characteristics.

A prediction based on actual as well as historical data using the tuned model provides for an estimated due date for taking chemical cleaning measures (advisory function). Based upon pre-confi gured limits, the status of the train is indicated using colour coding, for example, if cleaning due date is estimated to be within the next 15 days period, the train condition is colour-coded in red on the operator screen (see example as shown in Figure 7). The alarm limits can be fl exibly defi ned.

Thus, an intuitive and easy way of working and applying the solution is provided. The performance monitoring is applicable for different membrane confi gurations such as hollow fi bre and it does not require additional sensors required. It considers the hydrodynamics of membrane fouling at the membrane surface and it addresses the complete membrane life cycle except for partial replacement.

The performance monitoring function also allows an assessment of the quality of taken maintenance measures by comparing the condition before and after

Trains Actual FeedPressure

Actual Feedpoint

Optimal FeedPressure

Actual FeedFlow Actual set Point Optimal Feed

fl owDue Date For cleaning

Unit 1 Unit 2 Unit 3 Unit 4 Unit 5 Unit 6 Unit 7 Unit 8

Figure 9: System architecture.

OPTIMAX® MembranePerformance (Client)

PGIMSever

Online MembranePerformance Monitoring

Fouling status Optimal operatingcondition

Online MembraneProcess

Optimisation

UpdatedParameter

DCS Interfaces DCS Interfaces

RO unit 1ABB DCS

RO unit 23rd party

DCSDCS (ABB or 3rd party)

Water & Wastewater Asia • December / January 2011 39

PERFORMANCEMONITORING

Figure 10 – Exemplary Operator Screen used to present membrane performance monitoring and optimisation

the maintenance measure using an analysis of the main two key performance indicators.

Operation optimisationThe second module addressing operation optimisation uses results from performance monitoring (performance prediction) as the basis. Optimal operation conditions are calculated considering the operational and physical constrains. Depending on whether variable frequency drives (VFD) are used to drive the pump motors or not, either feed pressure and feed fl ow (VFDs used) or feed pressure or feed fl ow (operation without VSDs) set-points can be calculated. As aforementioned, the fouling rate dynamics depend on the operational set-points (feed fl ow, feed pressure) – this is considered for the calculation of the optimal set-points as the calculation not only aims at increasing productivity levels but also to optimise the fouling rate. The optimisation can be run regularly and might be implemented for open open-loop or closed-loop operation. The optimal set-points are suggested by the system follow the operational and physical constraints.

In order to have high fl exibility in terms of application, the modules are based on an ABB information management system which is used for data handling, storage and information management. This is outlined in Figure 9.

The information management system is capable of consolidating data from various process control systems being the source of required process data. In addition, the system provides for visualisation of results, for example, using trends of lists and even more extensive features such as alarm management (see Figure 10).

Results from the performance monitoring and optimisation solution are stored in the information management system

and can – if required – also be transferred to the process control system using standard interfaces for visualisation, for example, in alarm lists. Extensive reporting function is available with the information management system. Reports can be created in Microsoft Offi ce and can be deployed as html fi les on the web server of the information management system – the reports are accessible and viewable using thin client technology.

Successful implementationThe implementation of the performance monitoring and optimisation solution has successfully been implemented using afore-described architecture. With the pilot it was possible to demonstrate that the solution is capable of capturing the dynamics of fouling in real-time and to well give an insight to the membrane condition. Applying the optimisation function, it is possible to reduce the gap between actual and optimal set-points by gradually applying optimal set-points and

thereby to increase productivity. By gradually implementing optimal set-points, it was possible to achieve a 2% productivity increase during the pilot and to optimise the fouling rate. Optimal set-points have not been fully applied, so additional improvement potential by further implementing the suggested optimal set-points.

Benefi tsIn terms of benefi ts, the solution maximises the productivity by allowing to get higher product fl ow rates. In addition, operation and maintenance costs are minimised by improving the energy effi ciency and lowering the amount of chemicals required for cleaning as cleaning measures follow the condition of the membrane system. The membrane lifetime is increased as the risk of membrane damage is minimised following condition-based membrane maintenance measures. Unbudgeted membrane replacement can be avoided. Besides all this, the plant availability is increased by lowering cleaning and replacement activities and thus reducing plant downtimes. The solution is applicable to different membrane confi gurations such as hollow fi bre and can be used with existing or new installations without requiring additional measurements.

Overall, with this newly developed approach, the maintenance process for membrane systems can be changed from reactive, preventive to a predictive, condition-based way of operation. WWA

This paper is written by Mr Markus Gauder (business unit power generation, ABB AG, Mannheim, Germany), Mr Senthilmurugan S (ABB Corporate Research, Bangalore, India) and Mr Marc Antoine, (ABB Switzerland, business unit power generation, Baden, Switzerland).

RO Unit 1 RO Unit 2 RO Unit 3 RO Unit 4Train FTrain H

Close

Lest Calculation Time10.12.2009 05:99:00.000Current Time10.12.2009 11:26.54

268.16 0.00 299.11

Train H: RO Monitoring and Optimisation Solution

Optimal Values

Permeability

Actual OptimalSet PointProduct fl ow rate (m3/rr)