Clark Punp Desalination Thesis

i

MASTER OF SCIENCE THESIS

AUTONOMOUS PHOTOVOLTAIC

POWERED SEAWATER REVERSE

OSMOSIS FOR REMOTE

COASTAL AREAS

E V A N G E L I A G K E R E D A K I

16th June 2011

AUTONOMOUS PHOTOVOLTAIC-POWERED REVERSE

OSMOSIS FOR REMOTE COASTAL AREAS

E V A N G E L I A G K E R E D A K I

For the degree of:

Master of Science in Sustainable Energy Technology

Date of defense: 16th June 2011

THESIS REVIEW COMMITTEE:

PROF. DR. IR. L.C. RIETVELD DELFT UNIVERSITY OF TECHNOLOGY SANITARY ENGINEERING SECTION

DR. IR. S.G.J. HEIJMAN DELFT UNIVERSITY OF TECHNOLOGY

SANITARY ENGINEERING SECTION PROF. DR. F.M. MULDER DELFT UNIVERSITY OF TECHNOLOGY

RRR/FUNDAMENTAL ASPECTS OF MATERIALS AND ENERGY

SANITARY ENGINEERING SECTION, DEPARTMENT OF WATER MANAGEMENT FACULTY OF CIVIL ENGINEERING AND GEOSCIENCES

DELFT UNIVERSITY OF TECHNOLOGY, DELFT

i

ABSTRACT

This master thesis is part of the venture ―Drinking with the sun‖ conceived by the section Sanitary

Engineering in CiTG. It is a part of the ongoing research in the section with regards to water

purification using renewable energy technologies.

The project ―Drinking with the Sun‖ involves system integration of solar photovoltaics with a reverse osmosis desalination system with the ultimate aim of producing safe but sustainable drinking water

from sea water. Such a technology may be ideally suited for application in remote coastal areas,

facing a shortage of fresh water, but endowed with ample insolation.

Solar Energy is, by nature, fluctuating, whereas existing reverse osmosis systems are designed for continuous operation. In most autonomous renewable energy powered systems, batteries are used

for energy storage. However, in the last few years, research interest has been towards elimination of

batteries due to the financial and often environmental costs involved with their maintenance and replacement. Making a batteryless system which will be economic and reliable at the same time is a

challenge for seawater PVRO desalination.

Another possibility for improving efficiency of PVRO units lies in utilizing some of the energy from the

waste high pressure stream of the RO unit. The present project involves testing a device called the ―Pearson Pump‖ manufactured by Spectra Watermakers, which is unique in the way that it

incorporates an innovative energy recovery mechanism in a small scale system.

The project work involved sizing the system components, building and testing the prototype in a real

scenario. For the purposes of the experimentation, the island of Crete (Greece) was chosen as the area of the PVRO installation and cooperation with the Technical University of Crete was established

for the successful implementation of the project and for the investigation of its further sustainability. The scientific approach followed in the course of the experimentation lies in the worst-case scenario

testing of the real prototype, in real insolation and feed water conditions.

The preliminary results of the system positively indicate the technical feasibility of the system, and

offer recommendations oriented towards improving the reliability of the system in practical field conditions.

ii

iii

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my supervisor dr.ir. S.G.J Heijman for giving me the

opportunity to carry out this project as my master thesis and for facilitating me to conduct my experiments in Greece. The flexibility, feedback and support I was given allowed me to explore my

capabilities, and develop myself. Secondly, I would like to thank prof.dr.ir. L.C. Rietveld for the

guidance and feedback review for this thesis as well as prof.dr.ir. F.M. Mulder for agreeing to be a member of my review committee. Special thanks to professor E.Diamantopoulos and the Laboratory

of Environmental Engineering and Management of Technical University of Crete, Greece for hosting me during my experiments and for all the generous help, space and equipment provision, which

ensured the smooth completion of the experimental phase of the project.

Last but not least, I would like to thank house no.5, permanent and guests, my family and friends for

giving comfort in moments of stress and hard times and providing me with distraction when I needed it. My sincerest gratitude and thanks to Yash, for all the unconditional and endless support and help

during my studies and for being always there for me no matter what.

Lastly, I would like to dedicate this thesis to the memory of my beloved grandmother, a woman of

great strength and courage who passed away just before the completion of this report.

v

CONTENTS

ABSTRACT ............................................................................................................................. I

ACKNOWLEDGEMENTS ...................................................................................................... III

1. WATER SCARCITY AND DESALINATION .......................................................................... 1

1.1. FRESH WATER SCARCITY AROUND THE WORLD ............................................................................... 1 1.2. WATER GUIDELINES AND HEALTH ISSUES RELATED TO WATER SCARCITY ................................................ 2 1.3. DESALINATION - THE SOLUTION TO FRESH WATER SCARCITY .............................................................. 2 1.4. DESALINATION TECHNOLOGIES ................................................................................................... 4 1.5. REVERSE OSMOSIS - THE LEADER IN DESALINATION ......................................................................... 5 1.6. DESALINATION SYSTEMS - SYNERGY WITH RENEWABLE ENERGY ......................................................... 6

2. SOLAR DESALINATION AND PV-RO SYSTEMS ............................................................... 11

2.1. INTRODUCTION .................................................................................................................... 11 2.2. PV-POWERED RO SYSTEM COMPONENT ANALYSIS ......................................................................... 12 2.3. SYSTEM REVIEWS ................................................................................................................. 22 2.4. ECONOMICS AND MARKET ISSUES ............................................................................................. 25

3. SIZING THE PV-RO SYSTEM COMPONENTS ................................................................... 29

3.1. INTRODUCTION .................................................................................................................... 29 3.2. SIZING APPROACH AND ASSUMPTIONS ........................................................................................ 29 3.3. CRITERIA FOR PRODUCTION OF SAFE WATER ............................................................................... 30 3.4. SIZING STRATEGY ................................................................................................................. 32 3.5. RESULTS ............................................................................................................................ 34 3.6. THEORETICAL COST ANALYSIS ................................................................................................. 35

4. EXPERIMENTATION ....................................................................................................... 39

4.1. SELECTION OF THE AREA OF STUDY AND PROJECT PARTNERS ............................................................ 39 4.2. LOCATION CHARACTERISTICS AND METEOROLOGICAL DATA .............................................................. 40 4.3. OBJECTIVES ........................................................................................................................ 41 4.4. EXPERIMENTAL SETUP............................................................................................................ 42 4.5. SYSTEM START UP ................................................................................................................ 47 4.6. SYSTEM OPERATION .............................................................................................................. 48 4.7. PROBLEMS FACED / TROUBLESHOOTING ...................................................................................... 49

5. RESULTS AND DISCUSSION ........................................................................................... 53

5.1. A SAMPLE DAY ..................................................................................................................... 53 5.2. GENERAL COMMENTS ON EXPERIMENTATION................................................................................ 56 5.3. ROLE OF LINEAR CURRENT BOOSTER (LCB) ................................................................................ 59 5.4. OBSERVATION OF SUDDEN QUALITY DETERIORATION ...................................................................... 65 5.5. LIMITING FACTORS................................................................................................................ 67 5.6. DIFFERENT MODES OF OPERATION ............................................................................................ 68 5.7. COMPARISON OF THEORETICAL AND PRACTICALLY OBTAINED WATER YIELDS ......................................... 69

6. CONCLUSIONS & RECOMMENDATIONS ......................................................................... 71

BIBLIOGRAPHY ..................................................................................................................... I

APPENDICES ........................................................................................................................ V

TABLE OF FIGURES

Figure 1: The alarming extent of water scarcity across the world is detailed in a map compiled by the

International Water Management Institute (IWMI) [6] .................................................................... 1 Figure 2: Total installed desalination capacity by country (2006)[17]................................................ 3 Figure 3: Global installed desalination capacity by feed water sources (in million m3/d)(2006) [17] .... 3 Figure 4: Main desalination processes [7] ....................................................................................... 4 Figure 5: Global installed desalting capacity by process (2006) [17] ................................................. 4 Figure 6: Principle Of Osmosis And Reverse Osmosis Process [26] ................................................... 5 Figure 7: Possible Uses Of Renewable Energy Sources With Desalination Systems [3] ....................... 7 Figure 8: Distribution of RE-powered desalination technologies [3] .................................................. 8 Figure 9: Average 3-year insolation from calibrated data collected by the NASA International Satellite

Cloud Climatology Project (ISCCP) (1991-1993) [29] ..................................................................... 11 Figure 10: Solar Peak Hours on an optimally titled surface during the worst months of the year [30] 11 Figure 11: Simplified general design of a PV-RO desalination plant [16] ......................................... 13 Figure 12: Solar Cell I-V curve with MPPT in varying sunlight [35] ................................................. 15 Figure 13: Spectra Clark Pump and Spectra Pearson Pump [48] ..................................................... 18 Figure 14: Design of the Pearson Pump [48] ................................................................................ 19 Figure 15: SEC for seawater and brackish water PV-RO systems [16] ............................................. 23 Figure 16: Indicative shares of total costs in conventional seawater desalination [3] ....................... 25 Figure 17: Photovoltaics industry from 1999 - 2008....................................................................... 25 Figure 18: PV Module price versus cumulative module production from 1979 - 2009 ....................... 26 Figure 19: Cost versus Conversion Efficiency in 2008 .................................................................... 26 Figure 20: Power Consumption versus Product Flow Rate and Water Quality [1] ............................. 31 Figure 21: Monthly Optimal Inclination Angles for Chania, Greece .................................................. 32 Figure 22: Irradiation during a typical day in Chania in December and its possible Exploitation using 20 PV Panels............................................................................................................................... 34 Figure 23: Variation of Annual Water Production Costs .................................................................. 36 Figure 24: Variation of Water Cost with Interest Rates .................................................................. 36 Figure 25: Contributions to the Total Cost of a PV-RO System ....................................................... 37 Figure 26: Location of the Technical University of Crete, Chania .................................................... 40 Figure 27: Meteorological Data for Chania: Annual Temperature, Precipitation, Wind Speed and Water

Temperature ............................................................................................................................... 40 Figure 28: Total solar energy (Monthly Values) over 2010 in Chania [Footnote 11] ......................... 41 Figure 29:The Experimental Setup in the WaterLab, CiTG, TU Delft [1] .......................................... 42 Figure 30: Roof on which the PV Panels were Installed.................................................................. 44 Figure 31: Solar PV Panels Installed on the Roof ........................................................................... 44 Figure 33: Electrical Connections of the Main Components of the System ....................................... 45 Figure 32: Linear Current Booster (LCB) ....................................................................................... 45 Figure 34: Instrumentation systems (from Left to Right): Analogue Pressure Gauges / FLowMeters, Digital Conductivity meters, pyranometer ..................................................................................... 46 Figure 35: RO unit cabinet on field and Desk Station on field ......................................................... 47 Figure 36: Simplified Process Diagram of PV-RO System ................................................................ 47 Figure 37: Relation between TDS and Electrical Conductivity of the feedwater used ........................ 48 Figure 38: Tank Configuration for Permeate Collection .................................................................. 50 Figure 39: Change in Feed Water Temperature during Daily Operation ........................................... 51 Figure 40: Final Process Flow Diagram and instrumentation ........................................................... 52 Figure 41: Solar Irradiation on a Sample Day (5/10/2010) ............................................................. 53 Figure 42: Variation of Operation Parameters in Time on Sample Day: 5/10/2010 ........................... 55 Figure 43: Relation between water quality, power consumption and flow on 5/10/2010 .................. 56 Figure 44: Worst-Case Solar Irradiation Patterns during experimentation period at chania ............... 57 Figure 45: Irradiation vs Feed Flow on 16/09/2010 ....................................................................... 57

Figure 46: Relation between Permeate Production and Power Consumption on a mostly clear day:

17/09/2010 ................................................................................................................................ 58 Figure 47: Power Consumption vs Irradiation on a mostly clear day: 17/09/2010 ............................ 58 Figure 48: Irradiation over time on 17/09/2010 and on 06/10/2010 ............................................... 60 Figure 49: Comparison of Variation in Motor Voltage for one clear day with the lcb and one clear day

without the lcb (Case Study 1) ..................................................................................................... 61 Figure 50: Comparison of Variation in Motor Current for one clear day with the lcb and one clear day without the lcb (Case Study 1) ..................................................................................................... 61 Figure 51: Comparison of Permeate Flow for one clear day with the lcb and one clear day without the lcb (Case Study 1) ....................................................................................................................... 62 Figure 52: Comparison of Variation in Permeate Quality for one clear day with the lcb and one clear day without the lcb (Case Study 1) .............................................................................................. 62 Figure 53: Daily Irradiation Values on 12/10/2010 ........................................................................ 63 Figure 54: Daily Irradiation Values on 11/10/2010 ........................................................................ 64 Figure 55: PV-RO Parameters as a function of Time (Quality Deterioration) .................................... 66 Figure 56: Layout of the RO membranes ...................................................................................... 66 Figure 57: Difference in Solar Irradiation for South and Southwest Facing Panels ........................... 68

List of abbreviations

MENA

SEMI RO

Middle East and North Africa

South European Mediterranean islands Reverse osmosis

RES Renewable energy source BW Brackish water

SW Seawater DW Demineralised water

PV Photovoltaics

PVRO SWPVRO

Photovoltaic reverse osmosis Seawater Photovoltaic Reverse osmosis

HPP High Pressure Pump LPP Low Pressure Pump

TDS Total dissolved solids

TMP Trans membrane pressure LCB Linear Current Booster

MPPT PLC

Maximum power point tracker Programmable Logic Controller

gpm η

ER

Gallons per minute Pump efficiency

Energy recovery

1

1. WATER SCARCITY AND DESALINATION

1.1. FRESH WATER SCARCITY AROUND THE WORLD

Fresh water scarcity poses a big problem in remote regions especially in the Middle East and North

African countries (MENA), the southern European Mediterranean Islands (SEMI) and isolated

communities in deserts [2]. Pollution and exploitation of groundwater aquifers and surface water

have contributed to the decrease in quantity and/or quality of available natural water resources in

those areas [3]. The availability of electricity networks in those areas is often as limited as the

availability of safe drinking water, even as technologies that are able to remove pathogens and

dissolved contaminants require substantial amounts of energy [4].

One of the most conspicuous phenomena of water-quality degradation, particularly in arid and semi-

arid zones, is salinization of water and soil resources. Salinization is a long-term phenomenon due to

which many aquifers and river basins have become unsuitable for human consumption owing to high

levels of salinity during the course of the last century. The salinity problem has numerous grave

economic, social, and political consequences, particularly in cross-boundary basins that are shared by

different communities [5].

FIGURE 1: THE ALARMING EXTENT OF WATER SCARCITY ACROSS THE WORLD IS DETAILED IN A MAP COMPILED

BY THE INTERNATIONAL WATER MANAGEMENT INSTITUTE (IWMI) [6]

For the Arab countries potable water is becoming as critical a commodity as electricity [7]. The per

capita share of total annual renewable water resources has dropped well below the UN threshold for

water poverty (1000m3/yr) with most of the Gulf Arab countries reaching below 200m3/yr per capita

[8]. The Aegean Archipelago islands in Greece have restricted water resources and for some of the

islands, salt water intrusion into aquifers has contributed to the deterioration of the quality of life of

the inhabitants. Especially during the summer, the majority of small and medium-sized Aegean

Archipelago islands have a significant clean water deficit and in several cases almost 50-80% of the

2

fresh water needed is transferred at a very high cost [9]. In July and August 2010, the cost for

transferring drinking water to 17 arid Greek islands rose to 12.5 €/m3, which is an extravagant cost as

compared to previous years. Transfer of water to arid areas in Greece has been happening since

1980 with the Greek government spending 7.5 million Euros for this year alone [10].

According to the World Water Council, today about three billion people around the world have no

access to clean drinking water, while, by 2020, the world will be about 17% short of the fresh water

needed to sustain the world population. Owing to the foreseen growth of the world‘s population

(especially in the developing countries), the problem is expected to become more and more critical

over the next two decades, bringing the lack of potable water to the top of the international agenda

[3, 6].

1.2. WATER GUIDELINES AND HEALTH ISSUES RELATED TO WATER

SCARCITY

Salinity in water is usually defined by the chloride (Cl-) content (mg/L or ppm) or total dissolved

solids content (TDS, mg/L or ppm), although the chloride comprises only a fraction of the total

dissolved salts in water. The Cl/TDS ratio varies from 0.1 in non-marine saline waters to 0.5 in

marine-associated saline waters. Water salinity is also defined by electrical conductivity (EC).

Based on salinity, water can be classified into three groups: Fresh Water - less than 1,000 mg/L,

Brackish Water -between 1,000 and 25,000 mg/L, Seawater - greater than 25,000 mg/L [11]. Most of

the water available on earth has salinity up to 10,000 ppm whereas seawater normally has salinity in

the range of 35,000-45,000 ppm in the form of total dissolved salts. According to World Health

Organization (WHO) latest guidelines (2008), water is considered to be good for consumption when

the TDS level is less than 600 ppm and for special cases goes up to 1000 ppm (based on taste

consideration). However, no health-based guideline value for TDS has been proposed due to the fact

that no reliable data on possible health effects associated with the ingestion of TDS in drinking water

were available [12].

Particularly for European countries, in the European directive 98/83/EC on the quality of drinking

water, TDS is no longer used as an indicator since it is very dependent on the time, season and water

source, and mostly affects the taste. Alternatively, Electrical Conductivity is given as a guideline and

the limit given for drinking water is 2500 μS/cm [13].

The consumption of brackish water has been linked to poor health, including diarrhea [11], kidney

and gastric disorders as well as possibly diabetes [14]. High levels of salinity are associated with high

concentrations of other inorganic pollutants (e.g., sodium, sulphate, boron, fluoride), and

bioaccumulated elements (e.g., selenium, and arsenic). In some parts of Africa, China, and India, for

example, high fluoride content is associated with saline groundwater and has been found to cause

severe dental and skeletal fluorosis [5].

1.3. DESALINATION - THE SOLUTION TO FRESH WATER SCARCITY

Under these circumstances, an increased trend towards use of desalination is observed around the

world as a means to reduce current or future water scarcity [3]. Although the costs for desalination

can be high because of its intensive use of energy, the cost to desalinate saline water is less than

other alternatives that may exist or be considered for the future in many arid areas of the world [7].

Effectively, this has led to the expansion of the desalination markets and the expectation to their

continuous expansion in the coming years particularly in the Mediterranean, Middle East and North

3

African (MENA) regions [15, 16]. Especially the use of desalination technologies to soften mildly

brackish water is increasing rapidly in various parts of the world [7]. A visual impression can be given

in the following map (Figure 2) where the distribution of installed desalination plants worldwide is

shown.

FIGURE 2: TOTAL INSTALLED DESALINATION CAPACITY BY COUNTRY (2006)[17]

The total capacity of desalination plants around the globe was 44.1 million m3/d by the end of 2006

according to the 20th GWI/IDA Worldwide Desalting Plant Inventory [17] whereas according to

the 22nd GWI/IDA Worldwide Desalting Plant Inventory the total installed capacity rose to

59.9 million m³/d in 2009 corresponding to more than 14,451 desalination plants worldwide [18]. It is

estimated that this will be more than double, increasing to about 107 million m3/day by 2016 [19].

Nearly half of the current global desalination capacity is located in the Middle East led by the Gulf

Cooperation Council (GCC) countries, with the remaining capacity distributed throughout North

America, Europe, Australia and Asia. New markets are opening in China, India and the USA [20]. Until

2009, desalinated water is used as a main source of municipal water supply in many areas of the

Caribbean, North Africa, and the Middle East, a fact which proves the dependence of these areas on

desalination as a highly reliable, non-conventional source of freshwater [16]. Figure 3 outlines the

installed global desalination capacity for various feed water sources. Seawater desalination is being

applied at 58% of installed capacity worldwide, followed by brackish water desalination accounting for

23% of installed capacity [7, 8, 21].

FIGURE 3: GLOBAL INSTALLED DESALINATION CAPACITY BY FEED WATER SOURCES (IN MILLION

M3/D)(2006) [17]

4

1.4. DESALINATION TECHNOLOGIES

Generally, desalination processes can be categorized into two major types: (1)phase-change

(thermal) and (2)membrane separation (concept of filtration). Some of the phase-change processes

include multi-stage flash (MSF), multiple effect distillation (MED), vapour compression (VC) which can

be thermal (TVC) or mechanic (MVC), freezing, humidification-dehumidification and solar stills.

Membrane based processes include reverse osmosis (RO), membrane distillation (MD) and

electrodialysis (ED)[7, 22]. The main desalination processes are shown in Figure 4.

FIGURE 4: MAIN DESALINATION PROCESSES [7]

FIGURE 5: GLOBAL INSTALLED DESALTING CAPACITY BY PROCESS (2006) [17]

Based on installed capacity for all source water types included, more than 80% of the world's

desalination capacity is provided by two technologies: Multi-stage flash (MSF), and reverse osmosis

(RO) (Figure 5). The MSF process represents more than 93% of the thermal process production while

RO process represents more than 88% of membrane processes production [3].

From Figure 5 for all sourcewater types, it can be seen that RO is the prevalent desalination process.

It accounts for sl ightly more than half (51% or 22.4Mm 3/day) of the globalcapacity. Forty percent or 17.7Mm3/day of the global production of desalinated water comes from

distillation plants, either using the MSF or the MED process. The picture changes if one distinguishes between the different source water types. Thermal desalination processses acocount for 61% (17.2

Mm3/d) of the production in all desalination plants that use seawater as raw water source, of which

50% is produced in MSF plants. Only 35% of the water comes from ro seawater desalination plants. On the contrary, RO accounts for 84% and 79% of the production in brachkish water and in

wastewater applications, respectively.

5

1.5. REVERSE OSMOSIS - THE LEADER IN DESALINATION

Among desalination technologies, reverse osmosis (RO) is rapidly overtaking thermal desalination in

terms of market shares and is forecast that this will continue with 59% of the total new built capacity

being membrane based (mentioned in literature of 2004) [3, 22].

One of the reasons for this is that reverse osmosis is commercially available in a range of sizes and is

one of the most efficient technologies having much lower specific energy consumption (SEC) (about

3—10 kWh of electric energy per m3 of fresh water produced from seawater) than the average of

desalination technologies (compared to MED and VC) [22, 23]. On top of that, the high share of

recovered product water (up to 55%), the modularity of the systems, the low unit investment costs

and the flexibility in site location, start-up and shut-down all add to the advantages of RO process

[23, 24], making it the best alternative, especially for applications in remote, often off-grid, areas

with small and medium local water demand, such as islands or isolated villages in coastal areas [9,

25].

During the past decade especially, two improvements have helped reduce the operating costs of RO

plants and thus the cost of water produced—the development of membranes that can operate

efficiently at lower pressures, and the use of energy recovery devices [7].

For all the above reasons, reverse osmosis is becoming the technology of choice with continued

advances being made to reduce the total energy consumption and lower the cost of water produced.

[3]

1.5.1. REVERSE OSMOSIS PRINCIPLE

FIGURE 6: PRINCIPLE OF OSMOSIS AND REVERSE OSMOSIS PROCESS [26]

Osmosis is a natural process of flow through a semi-permeable membrane. When pure water of the

same temperature is present on both sides of a membrane and the pressure on both sides is also

equal, no water will flow through the membrane. However, when the salt on one side is dissolved into

the water, a flow through the membrane from the pure water to the water containing salts will occur.

(Figure 6).Nature tries to equalize concentration differences [26].

Reverse Osmosis (RO) is a pressure-driven process that under high pressure forces salt water against

semi-permeable membranes so that water molecules can pass through membranes while the salts are

retained, when pressure is exerted on the side where the salts are added (Figure 6). The result is two

different flows one of freshwater permeate and one of concentrated brine. The system flow rate is

6

proportional to the difference between the applied - osmotic pressure differential between the brine

and the dilute sections [3]. The membrane itself represents a major pressure differential to the flow

of fresh water and as it is stated in the literature, commercially available RO membranes can retain

about 98-99.5% of the salt dissolved in the feed water [3]. The major energy requirement is for the

initial pressurization of the feed water [22].

Based on different literature and applications, operating pressures required for the desalination differ

with the salt concentration of the feed flow. Thus, for brackish water (BW) desalination the operating

pressures range from 15-30 bar [22, 24] (other literature mentions a range of 10-15 bar [3]), and for

seawater desalination (SW) from 55-70 bar [22] (while other literature gives a range from 55-80 bar

[24]). As desalinated water permeates across the membrane, the feed water becomes more and

more concentrated.

The amount of freshwater that can be recovered from the feed is limited by osmotic pressure and

scaling. Overall water recovery rates for seawater RO systems are typically 45-50% [3], whereas in

other references, recoveries are mentioned to be from 25 to 45% [2]. Brackish water RO plants have

recovery rates as high as 70% or even 90% if antiscalants are used [16, 22].

The disadvantages of the RO process are the sensitivity of the membranes to fouling, the high costs

of maintenance and repair, the risk of disruptions in supply, and the lower product water quality

(compared with thermal processes) [24].

Two major factors controlling the energy requirements of an RO system are membrane properties

and salinity of the feed water. Higher water salinity requires more energy to overcome the osmotic

pressure, whereas the RO system needs only mechanical power to raise the pressure of feed water

[22].

1.6. DESALINATION SYSTEMS - SYNERGY WITH RENEWABLE ENERGY

Despite the advances in desalination, the energy required to run these plants remains a drawback,

especially when it is supplied using conventional energy sources. However, the large ecological

footprint of fossil fuels and their fast on-going depletion in present times have led to a growing

interest in renewable energy sources. These technologies have certainly advanced technically over

the last quarter century to the point where they should now be considered clean-energy alternatives

to fossil fuels.

Many countries have already initiated the transition of their electricity supply schemes to higher

renewable energy shares, by supporting market introduction and expansion of those technologies.

The European Union set a goal to double its renewable energy share until 2010, and the

intergovernmental panel on climate change has recommended a worldwide reduction of 75% of

carbon emissions by the end of this century in order to avoid dangerous, uncontrolled effects on

climate and on the world's economy.

Fortuitously, renewable energy (RE) has unique synergies in regions where desalination is needed.

Many places all over the world which experience water scarcity like the Egyptian desert [35], rural

areas of Jordan [36], remote communities in Australia (e.g., [37]), Sicily [38], Ireland [86], and India

[87] have a high solar and/or wind energy potential [22]. It is expected that RE systems will flourish

in future as this synergy is exploited and play an important role in brackish and seawater desalination

in developing countries due to low operating and maintenance costs [3].

7

1.6.1. RE SYSTEMS IN DESALINATION

Renewable energies for use in desalination processes can include wind, solar thermal, photovoltaic

and geothermal. Among the several possible combinations of desalination and renewable energy

technologies (Figure 7), some are more (or less) promising than others in terms of economic and

technological feasibility.

Their applicability strongly depends on the local availability of renewable energy resources and the

salinity of feedwater, as well as on the remoteness of the region in consideration, accessibility to the

grid and technical infrastructure [22]. Moreover, the size of the plant plays significant role; some

combinations are better suited for large size plants, whereas some others are better suited for small-

scale applications.

FIGURE 7: POSSIBLE USES OF RENEWABLE ENERGY SOURCES WITH DESALINATION SYSTEMS [3]

In 2010, in the framework of ProDes Project financed by the Intelligent Energy for Europe

programme [27] an updated information collection of 131 RE-desalination systems installed from

1974-2009 was presented. Some of them were installed as pilot installations and have been dis-

mantled after some years of operation, but most of the installations are providing drinking water and

are used by the local populations.

Based on this collection, ProDes [27] made an overview of the most common or promising RE-

desalination technologies, including typical capacities, energy demand, estimated water generation

cost and the development stage (Table 1Error! Reference source not found.). As is claimed by

he authors, most technologies have already been tested extensively and the water generation costs

are estimated based on operational experience and real data.

RE-desalination systems are currently acknowledged as the most promising for remote regions in

most literature sources studied, where connection to the public electrical grid is either not cost

effective or feasible, and where water scarcity is severe.

In references of 2008 [22] it is mentioned, that RE-RO is most often chosen as it is one of the most

efficient in terms of energy consumption. Some RO plants are particularly suited for small

communities in remote locations, although others may find large-scale applications. Figure 8 shows

the distribution of renewable energy powered desalination technologies [3]. As can be seen from the

graph, photovoltaic-powered reverse osmosis (PV-RO) is the most popular design option and is

8

considered by many one of the more promising approaches, particularly for small systems where

other technologies are less competitive [24].

TABLE 1: OVERVIEW OF THE MOST COMMON RE - DESALINATION TECHNOLOGIES [27]

FIGURE 8: DISTRIBUTION OF RE-POWERED DESALINATION TECHNOLOGIES [3]

1.6.2. RE-DESALINATION DISADVANTAGES

As stated in [7], until 2002 desalination systems using renewable energy sources (RE) had been

scarce and of limited capacity representing about 0.02% of the total desalination capacity. Although a

very large amount of work has been conducted in this field because of significant financial - social

benefits of RE-desalination systems (including plant design and implementation, mathematical

models, and economic feasibility) - only a few are currently being used. Most of the desalination

plants are proposed for the purpose of providing drinking water to small communities, especially

remote ones.

WMVC

5% Solar MED

13%

Solar MSF

6%

Wind RO

19%Hybrid

4%PV-ED

6%

PV-RO

32%

Other

15%

Typical

capacity

(m3/d)

energy demand water production

cost (€/m3)

technical

development stage

Solar Still <0.1 solar passive 1-5 applications

solar MEH 1-100 thermal 2-5 applications/advanced

R&D

Solar MD 0.15-10 thermal 8-15 advanced R&D

Solar/CSP

MED >5000 thermal 1.8-2.2 advanced R&D

PV-RO <100 electrical BW: 5-7

SW: 9-12

applications/advanced

R&D

PV-EDR <100 electrical BW: 8-9 advanced R&D

Wind-RO 50-2000 electrical

Units < 100 m3/d:

BW: 3-5

SW: 5-7

Units of 1000 m3/d:

1.5-4

applications/advanced

R&D

Wind-MVC <100 electrical 4-6 basic research

Wave-RO 1000-3000 pressurised water 0.5-1 (prospective) basic research

9

The reasons for this are related to various, often correlated, aspects such as:

(i) Availability, where the geographical distribution of RE potential does not always comply with

the water demand intensity at a local level,

(ii) Costs, where the initial capital installation costs and various system components are still

expensive. Even though prices decrease continuously still in many cases they are still

prohibiting for commercialization,

(iii) Technologies involving the combination of energy conversion and the desalination systems

are still faced with the challenge of optimal design offering high efficiency at the required

volumes, thus decreasing costs,

(iv) Sustainability, where in most of the cases, the maturity of the associated technologies does

not match the low level of infrastructures which often characterizes places with severe water

stress. Experience has shown that several attempts to integrate advanced desalination

solutions in isolated areas failed due to lack of reliable technical support (adapted from [28]).

Conversion of renewable energies, including solar, requires high investment cost and though

the intensive R&D effort technology is not yet mature enough to be exploited through large-

scale applications [3, 28].

The real problem in RE-desalination technologies is the optimum economic design and evaluation of

the combined plants in order to be economically viable for remote or arid regions. The slow

implementation of renewable energy projects especially in the developing countries is mostly due to

the government subsidies of conventional fuels products and electricity. The economic analyses

carried out so far have not been able to provide a strong basis for comparing economic viability of

each desalination technology. The economic performances expressed in terms of cost of water

production have been based on different system capacity, system energy sources, system

component, and water source. These differences make it difficult, if not impossible, to assess the

economic performance of a particular technology and compare it with others. [3].

Nevertheless, the cost reduction of renewable energy systems has been significant during the last

decades. It is estimated according to literature that future reductions as well as rise in fossil fuel

prices will make seawater desalination driven by renewable energies more competitive in the years to

come [3].

10

11

2. SOLAR DESALINATION AND PV-RO SYSTEMS

2.1. INTRODUCTION

2.1.1. HIGH WATER SCARCITY IN AREAS WITH HIGH INSOLATION

As mentioned before, high water scarcity is a problem faced in remote regions in the Middle East and

North African countries (MENA), the southern European Mediterranean Islands (SEMI) and isolated

communities in deserts like in Australia. Fortunately, the climatic conditions in these areas are very

favourable for potential exploitation of solar energy. This fact can be seen from the following two

maps (Figure 9: Average 3-year insolation from calibrated data collected by the NASA International

Satellite Cloud Climatology Project (ISCCP) (1991-1993) [29] (Figure 9 and Figure 10) which depict

the spatially resolved solar irradiance in W/m2 and peak sun hours.

FIGURE 9: AVERAGE 3-YEAR INSOLATION FROM CALIBRATED DATA COLLECTED BY THE NASA

INTERNATIONAL SATELLITE CLOUD CLIMATOLOGY PROJECT (ISCCP) (1991-1993) [29]

FIGURE 10: SOLAR PEAK HOURS ON AN OPTIMALLY TITLED SURFACE DURING THE WORST MONTHS OF THE

YEAR [30]

12

Consequently, solar desalination systems and pilot plants have been installed around many of the

darker coloured places in the above maps like Saudi Arabia, Brazil, Australia, Florida, Gran Canaria,

Egypt, Gaza, Jordan, Eritrea, Sicily, Greece, USA, Mexico, Tunisia etc.

2.1.2. PV-POWERED RO SYSTEMS MOST COMMON IN RE-DESALINATION

The feasible solar energy-desalination technologies combinations have been described previously. The

combination of RO membranes and arrays of photovoltaic (PV) modules is the design option that has

been implemented most frequently in solar-driven RO desalination systems. As shown in Figure 8 PV-

RO systems account for 32% of the total installed capacity, while in PRoDes project [27] it is

estimated that PV-RO systems comprise 31% among combinations of 131 RE-desalination plants

reviewed in 2009.

However, the main problems nowadays are that RO presents a significant requirement for chemicals

and spare parts, and that PV panels still represent significant capital investment. Most importantly

however, PV-RO systems require some degree of technical skill to operate, involving understanding of

means to protect the membranes from fouling and to maintain pumps. For these reasons, even as the

feasibility of PV-powered RO systems as a valid means of desalination has been proven, they still

cannot compete favourably with fossil fuel-based desalination systems [31]. Until some years ago, the

cost for photovoltaics was a major constraint against PVRO systems. Yet, this situation has changed

with advance in PV technology.

Despite the above disadvantages, from a technical point of view today, PV as well as RO are mature

and commercially widely available technologies. RO is modular and compact and has proved to be the

lowest energy consuming technique, using nearly half the energy needed for thermal processes. This,

supplemented by the modular nature of PVs, their low environmental impact as well as the ease of

operation and maintenance are incentives for this combination of technologies to be used, especially

for application in remote areas [32].

The ADIRA project [31], funded by the European Union, examined the application of Autonomous

Desalination Systems (ADS) supplied by RE in several rural and other remote areas. Because of the

high cost of PVs, PVRO systems were only recommended for low quantities of output water either

from brackish or seawater feed.

The extent to which PV energy is competitive with conventional energy depends on the plant

capacity, on the distance to the electric grid and on the salt concentration of the feed [33]. Based on

that, many authors agree that photovoltaic powered reverse osmosis (PV-RO) systems may be the

only technically and economically competitive alternative for small and standalone applications with a

capacity up to 50 m3/day to provide drinking water in remote areas where access to fuel, electricity,

and technical expertise is not available [7, 24, 31, 33, 34]. The continuously decreasing capital cost of

PVs and RO units is also helping the feasibility of such systems.

2.2. PV-POWERED RO SYSTEM COMPONENT ANALYSIS

It has already been mentioned that the PV-RO system is regarded as the most promising approach

for desalination particularly for small systems in remote areas where other technologies are less

competitive.

Although research has been conducted concerning the PV-RO system design since the 1970s, a

standard design approach has so far not emerged. Design solutions may either include or omit energy

storage, inverters, energy recovery devices, etc. Despite the lack of standardization, however, several

components are common to all design approaches [16].

13

Generally, the major common components of PV-RO plants are [31]:

Photovoltaic modules

Pre-treatment units, which remove the dissolved and large suspended solids from the

feedwater prior to flowing through the membranes. This is done to protect the membranes

and to reduce salt deposits that can diminish the efficiency

high pressure pumps, which increase the feed water pressure on the membrane to the point

that it exceeds the osmotic pressure, thereby providing enough energy to move water across

the membrane

Reverse Osmosis Membranes

Post treatment units, in which the acidity of the water is neutralized and chlorine added to

disinfect it

A simplified, general design scheme for PV-RO desalination systems as given in literature [16] is

depicted in Figure 11. Blue lines show the water flow whereas the red line shows the recovery of the

concentrate stream. Dashed lines identify components and connections that may be absent.

FIGURE 11: SIMPLIFIED GENERAL DESIGN OF A PV-RO DESALINATION PLANT [16]

The most important components used by PV-RO plants will be discussed a bit more elaborately in the

following sections:

2.2.1. PV MODULES

In PV-RO desalination, the direct current (DC) electrical energy generated in the solar cells by silicon

or other semi-conductors is used—directly or after regulation—to power the pumps that generate the

pressure required for the feed water to permeate across the RO membranes.

Characteristics

PV panels today constitute the fastest growing renewable energy market. Present industrial

production focuses primarily on the production of (mono- and poly-) crystalline silicon and thin film

amorphous silicon cells. Over the course of several technological developments, the conversion

efficiencies of PV modules have reached 13-16% for poly- and mono- crystalline silicon cells and 6-

10% for thin film solar cells. Until recently, the high price for PV modules was also prohibitive in their

application in PV-RO systems. However, the prices of PV modules have been falling rapidly in recent

times year by year, a fact which means that the capital cost of PVs are no longer a major constraint.

PV modules need hardly any maintenance.

14

Until now, both mono-crystalline and multi-crystalline silicon modules have been used in PV-RO

experimental units but no thin film solar cells have yet been used. This fact has mostly to do with thin

film cells having; lower efficiency in normal light conditions (30% less efficient that single crystal)

and uncertain durability. Lower efficiency means that more space and mounting hardware are

required to produce the same power output, and this is a problem when space is a constraint. Also,

thin film materials tend to be less stable than crystalline ones, degrading over time.

2.2.2. INVERTER

Although PV panels supply DC current, many of the PV-RO systems use AC motors and require

inverters to convert DC current to AC. Reliability issues have been reported with inverters and power

conditioning equipment in grid connected and AC systems. Unscheduled site visits for maintenance

are required in such cases, in order to replace the complete inverter, or fuses, reset circuit breakers

etc [31].

2.2.3. TRACKING SYSTEM

Tracking and module orientation has been widely recognised as an important factor in determining

the electrical power output and thus the overall performance of the desalination plant. While modules

with fixed axes are tilted at a constant angle, modules with adjustable axes can be manually

repositioned based on seasonal changes, or, if a tracking system with controller and drive motor is

installed, the modules can automatically follow the sun's daily path in the sky. It has been estimated

by researchers that utilizing the seasonal tilt angle variation increases the yearly average permeate

flow of a PV-RO desalination plant in Saudi Arabia from 15 to 17 m3/d [16]. In Jordan gains in

electrical power output and permeate flow of 25% and 15% respectively have been reported when a

one-axis automatic tracking system was used rather than a fixed tilt plate [16]. In some applications

in Australia [4, 15] 30% more solar radiation was measured due to the single-axis tracker while in

other cases 60% higher permeate flow was produced because of tracking [16]. The model of the

complete PV-RO system developed in CREST research institute in Loughborough, UK [28] compared

the gains between a single-axis and a dual axis tracking system. The model showed that single-axis

tracking would increase the annual freshwater production by some 33% while dual-axis tracking

would provide a further increase of only 3%. Furthermore, it was concluded that only the single-axis

tracker is economically justified since it doesn‘t require automation like the dual-axis mechanism.

All in all, the advantages of the mounting of PV modules on trackers are proven. Nevertheless, the

high initial investment costs required to install tracking systems, have so far limited their use in PV-

RO desalination.

2.2.4. MPPT AND LCB

Maximum power point tracker (MPPT) circuits are basically DC-DC converters, generally used [4, 14,

28] to control the current drawn from the PV array and to maintain the system operation at a voltage

that achieves maximum power output (Figure 12) under varying conditions of irradiance and module

temperature [16, 21]. It is obvious from the graph that this voltage corresponds to a lower current

than the maximum current and the lower the irradiation, the lower the operating voltage. So, in case

the irradiance is very low, the DC/DC converter called (MPPT) will find, ideally, the optimal operation

point for that irradiation. This means that the motor will operate between certain voltage limits which

will correspond to the maximum power.

The MPPT algorithm has been used in many desalination pilots, and ensures efficiency under

fluctuating conditions. In 2002, A. Schafer and B. Richards [14] successfully tested a small batteryless

PV-RO prototype (power consumption 150 W and water production of just 500 l/d in a remote region

in Australia with the use of MPPT). It was the first time that an MPPT was used without use of

batteries but the authors did not specifically mention any information regarding its usage. At the

15

same year, CREST institute in Loughborough, UK [28] developed an MPPT algorithm specifically for

use with standard industrial inverters and without need of any additional sensors or calibration, in

Matlab/Simulink.

FIGURE 12: SOLAR CELL I-V CURVE WITH MPPT IN VARYING SUNLIGHT [35]

MPPT controllers for use in battery-less systems are not available commercially as of yet. MPPT

circuits exist mostly in the market as charge controllers which are connected with batteries. The only

available product for direct connection of MPPT between a PV panel and a pump is made from the

Australian energy research laboratories (AERL), however with very limited current output (16A),

which can be used for very small PV systems [36].

Linear current boosters (LCB) are also DC-DC converters mainly used in solar direct pumping

applications. They can achieve 30% increase in the water pumped than when the motor is connected

directly to the solar panels. An LCB basically exchanges voltage for current providing more current

produced by solar panels in order to use the maximum power for a particular time. LCB thus also

prevents stalling of the motor under less than full sun conditions. As the solar irradiation drops, the

output voltage of the LCB drops instead of the current, thus resulting in slowing down of the motor

(since the speed is proportional to the voltage), but continues to provide the same torque (the torque

being proportional to the current), thereby preventing stalling. By examining the effect of the

magnetic field in the motor, and realizing that magnetic flux is constant, we can arrive at the

following two equations relating the torque and speed output of the motor to the supplied current

and voltage:

v

m

E K

K I

Where: (in S.I. Units)

E = Motor Voltage

ω = Motor Speed

T = Torque developed by the motor

16

I = Motor Current

These are often known as the transducer equations for a motor, since a motor is really an electro-

mechanical transducer. The constants Kv and Km are dependent on the particular motor. Similarly, the

motor starts running much earlier in the morning at a lower speed (thus providing some useful

output) instead of staying stalled until full sun. This translates into more running time of the motor

where it spends a lot of time working instead of stalled doing no work [37].

Although its application has been rarely found in research publications, it is commonly used in almost

every private PV-direct water pumping application (particularly in America and Canada). LCB has been

noted to have been used by B. Richards and A. Schafer in 2002 in design considerations for a PV-

powered desalination system for remote communities in Australia [38]. Although they consider the

use of an LCB or MPPT essential for efficient operation when the PV array is connected directly to the

pump, there is no evidence in literature about the results of a field test with it. In contrast, 2 LCB

units from Solar Converters, Inc. (Canada) have been used in a PV-electrolyser system for hydrogen

production in order to improve the efficiency of the system [39].

2.2.5. PRE-TREATMENT UNIT

A major consideration in the design and operation of any RO system is the avoidance, or at least

management, of fouling and scaling of the membranes, since this determines the frequency of

required membrane cleaning and replacement, since the rate of membrane fouling and scaling is very

dependent upon feed-water quality and pre-treatment [40]. The use of an adapted pre-treatment

minimizes the fouling problems and can provide good protection of the membranes and a longer

lifetime [22]. Particularly, suspended solids and larger particulates are important to be removed as

they can damage the membranes when deposited on the surface of the membranes causing clogging

of spacers and increasing the resistances to the flow [14].

Depending on several parameters which influence the choice of the pre-treatment like dissolved

organic carbon, SDI, turbidity, algae content and their evolution during the seasons, and

temperature, the pre-treatment can comprise different technologies, such as conventional pre-

treatment (i.e. ballasted sedimentation, air flotation, dual-media filtration, mono-media filtration,

double stage filtration) or advanced technologies including membranes coupled with a conventional

process [22]. Specifically, the pre-treatment stage often consists of either 5 or 20 μm filters, or sand

filters. An interesting arrangement is the use of a beach well, which by using the sand of the beach,

can provide pre-filtered water and greatly reduce pre-treatment requirements [40].

Nevertheless, some references suggest an alternative solution for smaller systems, which involves

system operation at low recovery rates to prolong membrane viability and to reduce the costs [16,

24, 41]. It is argued that in systems using brackish feed waters, where scaling is a problem, fouling

can be greatly accelerated by use of excessive water recovery ratios, and this sets a limit on the

recovery ratio. In sea-water fed systems, on the other hand, the recovery ratio is limited primarily by

the osmotic pressure in the last element, where the feed water is most concentrated. Operating at a

lower recovery ratio will have little effect on the biological fouling, which tends to occur in the first

element [41].

UF pre-treatment has been used in some BW applications [4, 15, 25] in Australia. In Coober Pedy in

Australia, the experiments showed that an RO plant utilizing UF pre-treatment is a promising

alternative. Since the feedwater is disinfected physically using ultrafiltration (UF), the brine is free

from bacteria and most viruses and hence can be seen more as a reusable feed stream than a waste

stream with a disposal problem [4]. However, UF pretreatment involves higher investment costs than

conventional pretreatment, but because it removes significant numbers of microorganisms and

17

generally delivers higher qualify RO feed, eliminates the need for membrane disinfection. Also UF

pretreatment may reduce RO membrane cleaning and replacement costs [16].

2.2.6. HIGH PRESSURE PUMP

The high-pressure pump supplies the pressure needed to enable the water to pass through the

membrane and have the salts rejected. There are a vast number of possible configurations of motors

and pumps etc. that have been used in reverse osmosis systems coupled with photovoltaics around

the world. This wide range of choices is due to the different costs, efficiency and performances,

seawater compatibility and level of simplicity and other practicalities depending on the capacity and

on the year that each facility was installed.

Centrifugal pumps are regarded inappropriate to couple with PVRO systems. In order to achieve

optimum efficiency with a centrifugal pump, the rotor speed must be matched to the flow/pressure

operating point. However, this is not the case in PVRO systems where the fluctuations in irradiance

influence the flow and the pressure of the system. Due to their construction, they also cannot

manage the water recovery ratio with that injection system [41].

Due to the above mentioned reasons, positive displacement pumps are used as a rule because of

their higher energy efficiencies with respect to centrifugal pumps at the required flows and pressures

[16]. Based on the references which have been studied in the framework of this report, both rotary

positive displacement pumps (e.g., rotary vane [14, 42, 43] and progressive cavity pumps [4, 15, 25,

28, 41]) and reciprocating positive displacement pumps (e.g., piston [24, 28, 41, 44, 45]and

diaphragm pumps [46]) have been used.

Progressive cavity pumps are often used in PV-powered systems, however this is mostly in BW

systems (low salinity) since they are limited to a certain pressure (10-30 bar). Literature cases of BW

systems have reported this type of pumps to fully meet the targets set. Example of a progressive

cavity pump is the custom-designed pump for the projects of Schafer and Richards [4, 25] from

Mono-pumps manufacturer in Australia). However, in 2009 [15], the above mentioned authors

reported failure of the tested system to meet water quality standards when fed with high salinity

water, due to the pump pressure limitation. Another disadvantage reported in literature, is that this

kind of pumps can suffer from starting problems due to the static friction between the rubber stator

and the metal rotor - once turning, the water being pumped acts as a lubricant. However, it is

claimed that starting should not be a problem in a system employing a variable-frequency inverter

[41].

The amount of power, or more specifically current, required to drive the pumps is directly

proportional to the operating pressure. Higher the pressure, higher is the permeate flow. One of the

main operating characteristics or rotary vane high pressure pumps is that the permeate flow shows

very little dependence on the pressure which leads to essentially constant permeate flow. Dankoff

Solar Slow pumps are rotary vane pumps which have been used in PV-RO brackish water desalination

systems [42]. The pump was reported in literature [42] to work fine coupled directly off solar panels

with very few maintenance problems, and with motor brush replacements only every 3-5 years. The

small capacity (2.2 gpm and maximum pressure 17 bar) was mentioned to be one of the

disadvantages of the pump.

The Clark pump, is a positive displacement reciprocating pump (piston) that was specifically

developed for energy recovery in small desalination systems and was used in several seawater PVRO

applications since 2002 [28, 43, 45, 47]. Pearson pump, the successor of the Clark pump, is

nowadays the state-of-the-art pump technology in small reverse osmosis desalination systems (See

below energy recovery section).

18

FIGURE 13: SPECTRA CLARK PUMP AND SPECTRA PEARSON PUMP [48]

Most RO plants usually use AC for the pump, which means that DC/AC inverters are required [7].

However, there are systems that have been reported to have used DC pumps, to avoid energy losses

from the conversion DC-AC-DC [15, 24, 42, 43].

More recently, many PVRO plants chose to couple photovoltaics directly to high pressure pumps. In

general terms it is argued in most of the cases, that direct pump connection to the PVs eliminated the

need for inverters and batteries, resulting in low-cost, simplified and energy-efficient systems [4, 14,

15, 25, 28, 42].

2.2.7. ENERGY RECOVERY DEVICES

In conventional small Reverse Osmosis systems the high pressure needed in the membranes is

created and regulated by a back pressure regulating valve in the high pressure concentrate (brine)

discharge line. This pressurization of the saline water accounts for most of the energy consumed by

an RO system. All of the potential energy in the high pressure concentrate is lost in the back pressure

regulator, and the power required producing that energy is wasted. Consequently, the energy

efficiency of a seawater reverse-osmosis system is heavily dependent on recovering energy from the

highly pressurized brine stream.

In some cases, Danfoss axial piston motors have been reported to be used for energy recovery [41,

45]. In large reverse osmosis systems, pelton-wheel turbines, reverse running pumps and pressure or

work exchangers (mostly ERI type) have been used to recover energy [14, 24, 49]. According to the

company, ERI's PX Pressure Exchanger(R) device is a rotary positive displacement pump that

recovers energy from the high pressure reject stream of SWRO systems at up to 98% efficiency with

no downtime or scheduled maintenance [50]. Nevertheless, neither is available in small sizes.

The innovative Clark pump, developed by Spectra Watermakers [48] was the first pump for small

desalination systems which incorporated energy recovery and a pressure amplification innovation.

CREST institute (Loughborough) and Thomson tested the Spectra Clark pump with variable flow and

pressure conditions (using solar irradiation data) and concluded that the Clark Pump is very well

suited to a batteryless PV-powered system [41].

In 2009, Spectra Watermakers developed the Pearson Pump, which was a breakthrough in the

evolution and enhancement of pump design especially suited for small applications mainly because of

the incorporated energy recovery mechanism. It is a positive displacement three-cylinder

reciprocating high pressure pump with the same motors and crankcases used in conventional RO

system feed pumps. The Pearson Pump head delivers water to the membranes in the same way as

conventional feed pumps but is capable of recovering the energy in the concentrate stream. This is

done by returning the concentrate to the Pearson pump at high pressure, where it flows into the

pump cylinders on the undersides of the pistons, transferring its energy to the feed water entering

the membranes (Figure 14). The energy recovered from the concentrate leads to reduced load on the

pump motor, reducing the electrical consumption dramatically (to an impressive 2.64 kWh/m³) [51].

19

FIGURE 14: DESIGN OF THE PEARSON PUMP [48]

―Fixed recovery ratios‖ is another characteristic of the Spectra technology which differentiates it from

the conventional constant-pressure systems. The creation of the high pressure in a Spectra system is

not based on a back pressure regulating valve at the exit of the membrane, but on the innovative

pump design: A remarkably bigger size of the ceramic plunger occupies a significant part of the

volume in the underside of the piston thus reducing the space available for the returning concentrate

(Figure 14). Since feed water is forced out of the cylinder by the upper side of the piston, and only a

portion of that water can return to the underside, a ―Hydraulic Lock‖ is created, which induces high

pressure in the membrane. When the pressure rises high enough water is forced through the

membrane as permeate.

The proportion of the feed water discharged by the Pearson pump which permeates the membrane

and becomes product water is fixed by the percentage of the volume of the underside cylinder taken

up by the plunger. If the plunger takes up half the volume of the cylinder, then only 50% of the feed

water discharged by the Pearson Pump can return to the pump, and the other 50% must leave the

system as permeate. Thus, a Pearson Pump based system will be a fixed recovery ratio system, and

the operating pressure will vary with pump speed, feed water salinity, and feed water temperature

but the percentage of feed recovered as permeate will always be constant. Conventional systems are

constant pressure systems, where the back pressure regulator keeps the system at a fixed pressure,

and the recovery ratio will vary according to operating conditions [52].

In published literature however, use of the Spectra Pearson pump with PV-RO systems was not

found. However, the pump has been marketed in many solar and wind-powered land-based

desalination units for military, disaster relief, village level water supply, small eco-resort and remote

home applications [51]. In 2009, Brett Ibbotson used the Spectra Pearson pump to examine the

effects of fluctuating operation on reverse osmosis membranes, as part of his master thesis. After

more than 650 hours of fluctuating operation, he concluded that no deterioration of reverse osmosis

membrane performance was observed. From his experiments, SEC ranged between 2.66-2.86

kWh/m3 and TDS between 518-638 ppm [1].

2.2.8. MEMBRANE MODULE

The most common RO membranes used in desalination are spiral wound Thin Film Composites. They

consist of a flat sheet sealed like an envelope and wound in a spiral. In a reverse osmosis system, the

higher the pressure, the higher is the permeate flow and also the energy consumption. PV-RO

desalination systems are hence designed with generous membrane areas since for a fixed recovery

rate they can operate at lower pressures and thus with a lower energy consumption [41].

Nevertheless, although large membrane areas compensate for the reduced product flow with

decreasing operating pressure, they result in decreased water quality.

RO membranes are specified in terms of their rejection of NaCl at a specified pressure. Rejection

ranges vary from 95%-99.8% depending on the source of feed water. In seawater applications,

20

rejections values higher than 99.4% are used [53], while a membrane with rejection 95% is

acceptable for desalination of brackish water.

The most common RO configuration is single pass, in which the membranes are organised in series

within one or more pressure vessels [16].

After the introduction of battery-less PVRO systems, there have been several discussions about the

effects of fluctuating operation on seawater reverse osmosis membranes. Membrane manufacturers

suggest that an important factor for the lifetime expectancy of the membranes or RO plants is the

constant flow and pressure operation of the plant. For this reason, they have been reluctant to give

any warranties regarding their products under variable pressure-flow conditions. From various tests

conducted by different researches, the operation of PVSWRO systems under variable pressure flow

conditions indicated no serious problems [1, 41]. Schafer et al (2008) [25], evaluated the effect on

the performance of a PV-(BW)RO and concluded that the membranes tolerated large fluctuations in a

wide range of solar irradiance (500-1200W/m2), resulting in only small increases in the permeate

conductivity. One important point from the above research was the special care for the feedwater

pre-treatment to avoid fouling of the RO membranes.

2.2.9. BATTERIES

Solar power is intermittent and variable while RO plants are generally designed to run continuously

and at constant load. However, constant power supply can only be maintained through energy

storage or backup systems. Typically conventional lead-acid batteries have been used as energy

buffers during day time or for extending the hours of operation beyond sunset for photovoltaic-

desalination systems. There has been a lot of debate over the years about the suitability of batteries

as energy storage in PVRO systems. Many authors and membrane manufacturers often propose

batteries in order to address the problem of variable operating conditions (pressure and flow) that

follow solar fluctuations. On the other hand, others argue that lead-acid batteries (accumulators)

which have been used until now in PVRO systems are very troublesome and associate them with

various problems such as:

Increase environmental risk in case of spillage (because of accidents or improper disposal in

remote areas where recycling is limited)

Worse performance and faster degradation at higher temperatures, a fact which is relevant,

since most potential sites for PV-RO applicability do experience high temperatures. With

increasing operating temperature, the charge efficiency decreases, eventually reducing the

number of battery lifecycles [25].

Because of the relatively low number of discharge-cycles, lead-acid batteries need

replacement every 2 years (on an average) [28]. This offsets the overall sustainability

benefits from the use of PVs.

Conversion losses when charging in/charging out, reducing the system efficiency (efficiency

of Pb-acid battery is about 75-80%) [25]; this leads to a greater power requirements, thus a

bigger PV-array. In literature [24], it has been argued that when batteries are used, the PV

system has to be at least 4 times larger for a 24 hr/day operation.

Extra addition to the capital cost of the system

Significant maintenance which add in costs (absence of careful maintenance is typical in

remotely located systems [16])

Complicated system if all auxiliary components such as charge controller and wiring are

considered [16]

The main advantage of batteryless systems is the simplicity since there are no batteries and charge

controllers, capital and maintenance investment is lower and environmental hazards are less. Battery-

less PV-RO systems are based on the idea that water storage is often more efficient and cost-

21

effective than energy storage [15, 16]. The technical feasibility and short-term operation of variable

speed, batteryless PV-RO systems was tested in a series of studies [4, 15, 25, 28, 42, 47]. However,

long-term performance has not been monitored so far.

Thomson et al. from the CREST institute in Loughborough [28] in 2002 presented a model for a small

batteryless PVSWRO system, based on real data from the country of Eritrea and commercially

available components incorporating the innovative Spectra Clark pump and an MPPT algorithm. Based

on the model, the authors described the system as energy efficient and calculated the production cost

of drinking water to be ₤2/m3. In 2005, the same team [47], developed a prototype which was tested

in the laboratory in UK. However, results were inconclusive since permeate quality was beyond

guidelines and the reliability of the system was not proven. The authors remarked that system

reliability should be treated with equal importance as energy efficiency in PVRO systems.

In 2008, a comparative study between a batteryless and a battery PV-powered seawater system was

published by Mohamed et al [2]. The results obtained from this comparative experimental study

showed that although the battery system was more efficient, there was no real difference (less than

7%) in the SEC and water production between the battery and the batteryless system. The authors

concluded that economically, the extra costs for battery and charge controller were not justified.

The same year, Richards et al [25] in their research on the effect of energy fluctuations on the

performance of a direct coupled PV-RO system for brackish water, found that the system could

tolerate large fluctuations between 500-1200 W/m2).

In 2009, De Munari et al [15], tested a batteryless PVBWRO system in a remote area in Australia and

noticed that the power variations because of the lack of batteries do not affect the permeate quality

nor water production. However, it was indicated that results would not be the same for high salinity

water.

At last, the question of using batteries or not is very site and case specific. M. Thomson et al in 2008

argued that the underlying subject is identifying what is less troublesome and more cost-effective in

remote locations lacking skilled personnel where PVSWRO systems normally find their application,

replacing reverse osmosis membranes or replacing batteries [45].

Nevertheless, with the advance in battery technology recent years, new possibilities about energy

storage in batteries have evolved. Low-maintenance lead-acid rechargeable batteries called VRLA

batteries (valve-regulated lead-acid battery) with a lifespan of up to 10 years are commercially

available. Because of their construction, VRLA batteries do not require regular addition of water to the

cells. VRLA batteries are classified as absorbed glass mat batteries (AGM) and Gel batteries (gel cell).

These batteries are often colloquially called sealed lead-acid batteries, but they always include a

safety pressure relief valve. As opposed to vented (also called flooded) batteries, a VRLA cannot spill

its electrolyte if it is inverted. Because VRLA batteries use much less electrolyte (battery acid) than

traditional lead-acid batteries, they are also occasionally referred to as an "acid-starved" design.

However, cost is a constraint since they cost twice as the traditional lead-acid batteries while their

discharge cycle is less deep (around 50%).

In the market, hybrid Gel batteries, the 2nd generation VRLA batteries already exist. Manufacturers

claim that they have better service life, especially deep cycle life and high rates of discharge

compared to standard AGM batteries.

Currently, AGM and gel batteries are used in commercial PVRO systems of Spectra Watermakers [54].

22

2.2.10. FEED WATER INLET MEANS

Spectra systems require that feed water enters the system at a pressure of about 1 bar (gauge). For

the purpose of the trials, it would be easiest to have the feed water tank at the same elevation as the

pump, which leads to requirement of a feed pump. However, in a real system, the pressure head

required at the inlet may also be obtained by placing the feed tank at an elevation (greater than 10m,

correcting for the friction factor) higher than the pump inlet.

An extreme concept was proposed by Murakami [55], regarding a hydro-powered pumped water

storage system for co-generation of electricity and desalinated water. The proposed scheme involved

pumping sea water to an elevation of 589 m (above sea level) using hydroelectricity. Water from the

pumped storage reservoir would be used to regenerate electricity for 4 hours in a day in times of high

demand. For the remainder of the operational period (18-20 hours), the pressure head of the water

would be used to produce a desalinated stream of water by using reverse osmosis membranes. The

elevation can be potentially used to generate a head of ~ 58 bar (disregarding friction losses) which

can be used to overcome the osmotic pressure of sea water.

2.3. SYSTEM REVIEWS

Ever since the first technological breakthroughs on PV-powered RO desalination in the late 1970s,

vast experience has been gained with PV-RO systems from a large number of experimental units.

Although many authors in references agree that photovoltaic powered reverse osmosis (PV-RO)

systems are a proven technology nowadays, most of the units were lab demonstrations or pilot

systems which after their testing were abandoned [7]. The main reasons for this were the very high

capital cost and the high requirement for technical skills in order to protect the membranes and

maintain the pump [24].

The PV-RO technology has been implemented for the desalination of both brackish water (BW) and

seawater (SW), yet only in small and medium systems (less than 75 m3). It is interesting to note that

until 2010 (as far as the author understands), no PV-RO experimental study has been done on a

larger scale. Most systems have been developed in different parts of the world with intense fresh

water scarcity but in the same time with abundant solar energy potential (e.g Mediterranean region,

Australia, North Africa and Middle East).

The parameters determining the performance of a PV-RO system depends on the specific energy

consumption (SEC), on the total daily water production and on the salt rejection. SEC fundamentally

depends on the salinity of the water [22]; The higher the salinity, the higher is the SEC and the cost

of the water produced [4, 15]. In addition, small systems operate much less efficiently, which leads to

even higher SEC values, leading to an even higher cost of the system [24].

Because of cost considerations, brackish water solar desalination became mature and proved to be an

economically viable technology very early in contrast with seawater solar desalination [2]. Seawater

desalination is a very energy consuming process and has much higher SEC than brackish water

desalination. Until 2002 research was focusing mostly on PV-(BW)RO systems and demonstration

plants. In addition to this, until 2002, most plants were using batteries to balance the energy

fluctuations from the sun.

The situation changed after 2002 for seawater PV-powered desalination, with the invention of the

Spectra Clark pump, incorporating energy recovery mechanism designed especially for small plants.

Energy recovery reduced the SEC for seawater PVRO desalination. However, seawater desalination

powered by photovoltaics is (potentially) economically competitive only for remote areas with no

23

access to the grid areas (electricity / water) [16]. Indicatively, based on literature [4, 16, 24], the

following table shows values for SEC depending on the feedwater quality.

TABLE 2: INDICATIVE SEC VALUES FOR DIFFERENT FEEDWATER TYPES WITH AND WITHOUT ENERGY

RECOVERY

Feedwater Type Energy recovery SEC (KWh/m3)

Brackish No <10

Yes <2

Seawater No >10

Yes <5

FIGURE 15: SEC FOR SEAWATER AND BRACKISH WATER PV-RO SYSTEMS [16]

In Figure 15, different SEC values can be compared for brackish and seawater systems. Based on that

list, it is very interesting to see that the lowest seawater SEC value (for a real prototype and not a

theoretical estimate) is in Baja California Sur, Mexico. This case study incorporated the Clark pump by

Spectra Watermakers which tremendously reduced the SEC to 2.6 kWh/m3.

In the comparison made by Mohamed et al. in 2008 [18] between a seawater PV-RO system with and

without a battery, it was shown that the SEC for a battery system was around 4.3 kWh/m3 while for

the batteryless a bit higher (4.6 kWh/m3).

It is difficult to compare between different experimental studies with respect to water costs, since

each one is based on different assumptions (variable depreciation times, currency exchange rates,

frequency of replacement of components, feedwater quality, plant capacity, etc.) and is site and year

specific [3]. In the following table an overview of existing PVRO systems is shown, yet it is not

reliable for conclusions due to the aforementioned reasons. Very old systems for brackish PVRO

(before 1990) have been omitted.

24

TABLE 3: OVERVIEW OF PV-RO SYSTEMS (ADAPTED FROM [16])

Location and

country

Year Feed TDS

(mg/L)

PV capacity,

(kW)

Battery

storage

Pump Production

(m3/d)

SEC Cost,

(US$/m3

)

Brackish desalination systems

Coite-Pedreiras, BRA 2000 1200 1.1 yes DC/AC 6 12.8

Denver, ITN, USA 2003 1600 0.54 no DC 1.5 6.5

Lisbon, INETI, PRT 2000 2549 0.1 no DC 0.02 10.6

Hammam Lif, TUN 2003 2800 0.59 no DC 0.05 11.6

Concepcion del Oro,

MEX

1978 3000 2.5 yes DC 0.71 12.8

Mesquite, ITN, USA 2003 3480 0.54 no DC 1.28 3.6

Murdoch Univ., AUS 2003 3480 0.06 no DC 0.05 3.6

White Cliffs, AUS 2003 3500 0.26 no DC 0.06 8 9.0

Ksar Ghilene, TUN 2005 3500 10.5 yes AC 7.0 6.5

Aqaba, JOR 2005 4000 16.8 yes AC 58.0 9.8

Baja California Sur,

MEX

2005 4000 25 yes AC 11.5 9.8

Pine Hill, AUS 2008 5300 0.6 no DC 1.1 3.7

Sadous, Riyadh, SAU 1994 5700 10.08 yes AC 5.7 9.6

Lampedusa, ITA 1990 8000 100 yes AC 40 10.6

Coober Pedy, AUS 2008 BW 3 no DC 0.76 3.2

Central Australia,

AUS

2008 BW 3 no DC 1.1 2.3

Kuwait, KWT 2005 8000 0.3 yes DC 1.0 6.5

Seawater desalination systems

Agric. Univ., Athens,

GRC

2006 30000 0.85 no DC 0.35 9.8

CREST, GBR 2001 32800 1.54 no AC 1.45 3.0

Vancouver, CAN 1983 33000 0.48 no DC 0.86 9.0

Doha, QAT 1984 35000 11.2 no AC 5.7 3.0

Univ. of Bahrain,

BHR

1994 35000 0.11 yes DC 0.2 2.8

Pozo Izquierdo, ESP 2000 35500 4.8 yes AC 1.24 9.6

Jeddah, SAU 1981 42800 8 yes DC 3.22 6.5

Baja California, MEX 2007 SW no DC 19 2.6

CREST, GBR 2003 SW 2.4 no DC 3 3

Although the above comparative table is not reliable to draw conclusions, in general, we can observe

the following trends:

The older the system, greater is the cost of desalination (owing to systems with lesser SEC)

The lower the salinity, lower is the water cost (owing to lesser pressure requirements, and

hence a smaller SEC)

It is interesting to note that as far as seawater PV-RO is concerned there are few plants installed

worldwide and most of them have very high water cost. Reduction in costs of seawater solar

desalination appeared mainly after 2001 with the invention of the Spectra Clark pump incorporating

means of energy recovery, as mentioned in publications.

The cheapest water cost for a PV-(SW)RO system, was estimated by Thomson et al. from the CREST

institute in Loughborough [28]. In his model in 2003, he presented a small batteryless PV-(SW)RO

25

system which was incorporating energy recovery (Clark Pump) and MPPT. The water cost estimation

was ₤2/m3. However, this calculation was not confirmed by the lab scaled prototype which was

developed in 2005. Cost estimations by models are usually lower than in real experimental units.

Despite the technological advances in recent years in seawater PV-powered desalination, it is still not

competitive with grid-powered or fossil fuel powered desalination and mainly finds applications in

case of small plants in remote areas with no access to the grid or with expensive fuel supply chains.

2.4. ECONOMICS AND MARKET ISSUES

It is difficult to compare the costs of desalination installations since the actual costs are contingent on

situation-specific parameters. However, in general terms energy and capital costs are the main

driving forces in the desalination total cost (Figure 16).

FIGURE 16: INDICATIVE SHARES OF TOTAL COSTS IN CONVENTIONAL SEAWATER DESALINATION [3]

From Figure 16, it can be observed that for conventional seawater RO systems the largest cost

reduction potential lies in capital costs and energy. In case of PVRO seawater desalination, the capital

costs are much greater because of the high investment costs for the photovoltaics, whereas the

operational costs for electrical energy are non-existent. Feasibility in PV-RO systems can be reached

when the gain achieved by elimination of operating electricity costs outweigh the loss due to rise in

capital costs. PV technology is fast developing, and currently among the renewable energies, they

constitute the fastest growing market. The industry has almost doubled from 2007 until 2008 and

similar trends are expected to happen the following years (Figure 17):

FIGURE 17: PHOTOVOLTAICS INDUSTRY FROM 1999 - 2008

Capital

39%

Labor

4%

Maintenance

and parts

7%Consumables

3%

Electrical

Energy

47%

26

Extensive market penetration, led to the rapid reduction of the price of PV cells from 1979 until 2009

(Figure 18), and currently the prices stand at 3-4 $/Wp in the US and Australian markets. Similar

trends are also expected in the future, making photovoltaics more promising technology than how it

was viewed in the past.

FIGURE 18: PV MODULE PRICE VERSUS CUMULATIVE MODULE PRODUCTION FROM 1979 - 2009

FIGURE 19: COST VERSUS CONVERSION EFFICIENCY IN 2008

27

From another point of view, thin film solar panels are less expensive than the crystalline ones, in

terms of €/m2. However, the choice of the right technology lies strongly on the application and

purpose. PV experts generally agree that crystalline silicon will remain the "premium" technology for

critical applications in remote areas. Thin film will be strong in the "consumer" market where price is

a critical factor [56].

In addition to PV cost reduction, RO has seen also steep declines in prices; between 1990 and 2002

membrane costs have fallen 86% [7]. Steeply declining maintenance cost in combination with

relatively low capital cost, seems to offer hope that PV-powered seawater desalination will become

competitive with other water supply sources, in context of remote regions where freshwater deficits

are covered with transportation of water, or electricity from the grid is not an option.

The dilemma around batteries for PVRO systems is still unresolved. Other energy storage possibilities

like super-capacitors could also be investigated. Alternatively, it may be interesting to use a battery

array exclusively for the purpose of regulating the solar energy fluctuations in the power supplied to

the motor, instead of for extending operating hours.

However, following the present research trend in favour of batteryless systems for autonomous cost

effective solutions, the present project shall focus on implementation of a seawater PV-RO setup in

the real scenario with a view to minimizing costs, and understanding the potential obstacles to

implementation with respect to long term performance and reliability.

28

29

3. SIZING THE PV-RO SYSTEM COMPONENTS

3.1. INTRODUCTION

For the purposes of this project, an already existing RO-unit installed in the WaterLab, CiTG, TUDelft

will be coupled directly with photovoltaics. The RO-unit (SpectraTM LB 1800) is commercially available

from the American company Spectra Watermakers, Inc. and is designed for land-based applications.

The maximum capacity of the system (with continuous operation) is 7.9 m3/d and it comprises of1:

2 spiral wound membrane elements

The state-of-the-art Spectra Pearson Pump (HPP) which incorporates energy recovery.

A Low Pressure Pump (LPP)

A water storage tank of 1 m3

One microfiltration pre-treatment unit

From the literature search, it was found that post- 2002 research interest was driven towards

batteryless PVRO systems (mostly for brackish water) operating in fluctuating conditions of power

input, thus avoiding the use of the batteries due to the financial and often the environmental costs

involved with their maintenance and replacement. The investigation of batteryless PVRO systems

operating in fluctuating conditions has given positive indications in the past. In addition, previous

research on the current existing system gave encouraging results for the use of the system without

an energy buffer under fluctuating operation [1].

The ultimate effect of not including batteries is the system having to cope with variable operation

patterns of flow rates and pressure that follow instantaneous atmospheric conditions. Membrane and

pump manufacturers object to such operation regimes since RO systems are designed for constant

operation. Taking these facts into consideration, it was decided that the research results of this

project would be valuable and they would contribute to new knowledge in the field of seawater PVRO

for small systems, if the system did not incorporate batteries.

3.2. SIZING APPROACH AND ASSUMPTIONS

It is assumed that the system is operating under fluctuating conditions with no energy buffer. Sudden

start-stops can seriously cause problems in the pumps as well reduce the lifetime of the membranes.

For this reason, the challenge in the sizing of the components lies at a system design operating

without sudden start-stops but within the power range that the motor operates safely and gives safe

water.

The design of an autonomous desalination plant as well as the economic analysis is site specific and

cannot usually be generalized for applications in other situations. This sizing approach is site specific

to the city of Chania in Greece, with respect to tilt angles, radiation data, and the demand-driven

strategy of water production.

1 More information about the system configuration is given in chapter 4.

30

It has to be ensured that:

The given power to the system is within the operating range (670 W – 1200W) of the

motor2

The given power to the system will always be above a certain limit in order to produce safe

drinking water (<500 mg/l).

The following assumptions were made and used for the sizing calculations:

1. Radiation data is site-specific for Chania, Greece, and represents normalized values for a

typical day of each month3.

2. It should be noted that radiation data does not include stochastic variations (like passing

clouds, showers, etc.)

3. Optimized panel tilt angle for every month based on the latitude of the location.2

4. The water produced in a year is the mean of the projected annual water production on the

basis of June (summer) and December (winter).

5. The maximum rated production of the facility (7.9 m3/d) is taken into account.

6. No use of batteries (or any other energy storage device).

7. The constant efficiency of conversion of PV panels used in calculations is equal to its

maximum value.

8. The excess in power production by the PV panels is used to power the controllers; a power

requirement which is assumed to be adequately satisfied at all times.

3.3. CRITERIA FOR PRODUCTION OF SAFE WATER

3.3.1. MINIMUM AND MAXIMUM POWERS

Figure 20 correlates the power of the system (W) - the HPP including the LPP pump with the product

flow (l/min) and the water quality (TDS in mg/l). From the relation between the power consumption

and the water quality in TDS, it can be calculated that the minimum power required for the HPP-LPP

system to produce safe drinking water (<500 mg/l) is 793 W.

2 This operating range has been proved to be safe for the motors in previous experimental investigations on the same setup (By Brett Ibbotson). 3 http://re.jrc.ec.europa.eu/pvgis/apps/radday.php?lang=en&map=europe

31

FIGURE 20: POWER CONSUMPTION VERSUS PRODUCT FLOW RATE AND WATER QUALITY [1]

From the system specifications, the capacity of the system is 7.9 m3/d. Using the relation acquired

from the above figure for water production and power consumption, the maximum power which the

system can utilize can be calculated to be 1204 W.

3.3.2. OPTIMAL INCLINATION ANGLE

The angle of inclination of the solar panels is an important criterion affecting power production. For a

given location, the optimal angle for harnessing solar irradiation depends upon the time of the year.

Solar PV systems may hence be designed with a mechanism to tilt the plane of the panels to match

the optimal angle for a particular time of the year.

The optimal panel inclination angle is affected by several other minor factors, hence the following

location specific data about the location is essential:

Location: 35ο 30'26" North, 24ο 1'26" East Elevation: 20 m a.s.l

Nearest city: Khania, Greece (0 km away)

Land cover class: agro-forestry areas (CLC244)

Optimal inclination angle is: 28 degrees

Orientation (azimuth) of modules: -1.0o (optimum)

Annual irradiation deficit due to shadowing (horizontal): 0.0 %

The optimal inclination angles for every month (and a normalized yearly optimal angle) in Chania as

given by JRC are shown in Figure 21.

32

FIGURE 21: MONTHLY OPTIMAL INCLINATION ANGLES FOR CHANIA, GREECE

3.4. SIZING STRATEGY

The general procedure that is followed in the course of the sizing is as follows:

An area of the solar panels is assumed

Based on the area of the solar panels and their efficiency, the minimum required irradiance

W/m2 is calculated (where the power provided by the panels is more than the minimum

power required for safe water production=793 W)

The usable time frame within the month (June/December) is evaluated

All irradiation values leading to power greater than the maximum power (>1204W) which can

be consumed by the system are set to a fixed maximum.

The irradiation within the usable time frame is integrated over the usable time frame

Using the integral value in the relevant equation the total water production over a single day

is calculated

Assuming the relevant pay-back periods and costs of the system, the cost of water (per m3)

is calculated

The procedure is iterated to reach a value at which the cost of the system is as low as

possible while the energy wastage is minimized.

3.4.1. PV PANEL SELECTION AND SIZING

Many different PV panels were compared based on their capacity, efficiency and cost/Wp. The final

selected model of PV panel is the ―Energy Power Plus‖ which combines high conversion efficiency and

very low cost4:

TABLE 4: PV PANEL CHARACTERISTICS

Photovoltaic

type

Brand model capacity

(W)

Area

(m2)

Module

efficiency

Weight

(kg)

cost

(€)

cost

(€/W)

cost

(€/m2)

Polycrystalline

silicon

Solar Energy

Power Plus

SE230P-

20/Ac

230 1.62 14.0% 20 437 1.9 379

4 Referring to: http://www.solar-systems.gr/SPECIFICATIONS-POLY.htm

33

3.4.2. METHODOLOGY FOR THE CALCULATION OF THE DAILY WATER PRODUCTION :

From Figure 20 the relation between water production rate and power is:

y 0.004 x 0.6691 ... (1)

Where;

y= Water production rate (l/min)

x = Power (W)

Since the system is under fluctuating operation, the water production is not constant during the day.

Thus, the daily water production Y can be calculated by the integration:

stop stop

start start

t t

t t

Y y dt (0.004 x 0.6691) dt

... (2)

The Power x can be translated to the irradiance with the following relation:

px I A … (3)

Where: x = Power (W), I = Irradiance (W/m2), A= area covered by the panels (m2), ηp = max

conversion efficiency (%), tstart=start up time of the system, tstop=stop time of the system

From (2) and (3), the total water production over a single day (Y in liters) can be given from:

stop

start

t

t

Y 0.004 A I dt 0.6691 dt

… (4)

The water production was calculated for both a typical day in winter (December) and in summer

(June) using the irradiation data from JRC. The average of winter and summer production values

represents the estimation for the daily water production throughout the year.

In order to account for the minimum power to produce safe water (793 W) and for the maximum

power which the system can utilize (1204 W), the original irradiation data needs to be modified. In

order to do this, the irradiation (W/m2) corresponding to the maximum rate of absorption of power

from the solar panel array is first calculated (using equation 3). Following this, all values of irradiance

greater than this value are equated to the maximum value.

The second threshold is that the reverse osmosis plant can be in operation only when the power

produced by the panels is larger than the minimum power required by the plant to produce safe

water (793 W). The first instance of the day when this condition can be satisfied becomes the tstart

value, and the last possible time instance satisfying the condition becomes the tstop. The integral is

evaluated between the limits of tstart and tstop.

34

FIGURE 22: IRRADIATION DURING A TYPICAL DAY IN CHANIA IN DECEMBER AND ITS POSSIBLE

EXPLOITATION USING 20 PV PANELS

Taking as an example the use of 20 panels, Figure 22 was created. The area between the two lines in

Figure 22 (plotted for December) indicates the energy which needs to be disposed. A part of this

energy could possibly be used in powering the computer and other utilities. The energy which needs

to be disposed off in summer is even greater than what it will be in winter. The sizing is made on the

basis that the system provides adequate quantity of water at a low cost. If fewer panels are used in

the system, the quantity of water produced in winter decreases greatly. The loss of revenue on this

account may be greater than the savings due to a lesser PV Panel requirement (depending on the

cost of water in the winter months).

Since we have assumed that the system works at its maximum efficiency at all irradiation values, the

energy wastage is a little lesser in the real system. If the energy wasted in the summer was desired

to be reduced even further while maintaining the same cost of the system, there would be no

alternative but to use a battery.

3.5. RESULTS

Multiple iterations (trial and error) of different PV area values were conducted as described above.

The results of the daily water production and the final water cost within the possible time frame for

each area value were compared. It was seen that between 10 - 20 panels there is a balance between

energy wastage and cost of water. For the purposes of the experiments, it was decided to use 10

panels (area of 16.20 m2) taking into account the practical hindrances of temporary installation.

TABLE 5: RESULTS OF SIZING APPROACH

Area covered by the panels (m2) 16.20

Number of panels 10

Operation time in winter 7.25 hours

Daily water production (m3/d) in winter 1.16

Operation time in summer 11.25 hours

Daily water production (m3/d) in summer 3.12

Average annual water production (m3) 781

0

50

100

150

200

250

300

350

400

450

500

6.00 8.00 10.00 12.00 14.00 16.00 18.00

Time of the day (hr)

Irra

dia

tio

n (

W/m

2)

Solar Irradiation (W/m2) Absorbed Power

35

3.6. THEORETICAL COST ANALYSIS

The costs for the RO system, PV panels, PV installation and mounting structures as well as tanks and

the rest of the equipment is based on real market prices. The RO unit from Spectra Watermakers,

Inc. comes mainly pre-assembled so the assembly cost is minimal. For the purposes of this cost

analysis, apart from the LCB (Linear Current Booster), an PLC controller (to monitor and control the

system operational parameters and water quality in the real scenario) has been accounted for in the

automation costs. The different components of the system can be assumed to have different

depreciation periods. It is assumed that the value of the component at the end of the depreciation

period is zero.

TABLE 6: ANNUAL COST ANALYSIS

Particulars Cost Depreciation

Period (years)

Capital Costs

RO system+pretreatment+HPP € 13000 10

PV panels € 4347 25

PV installation and mounting structures € 1710 25

LCB Controller € 335 10

Feed Pump € 159 10

Tanks and other equipment € 300 10

Automation Costs € 1200 10

(Total capital cost) (€ 21051)

Annual Operational Costs

Membrane + Pre-filter Replacement5 € 900

Post-treatment6 € 19

Miscellaneous Operational Costs7 € 260

Maintenance cost8

Feedpump operational costs9

€ 335

€ 26.13

An important contributing factor to the cost of water production is the interest payment on the capital

investment. This is, however, a rather complex factor to account for in the cost calculations since it is

largely dependent on the economic and political scenario where the installation is to be placed. Due

to this variance, the total annual costs of water production (and hence the cost of the water) is

presented for different cases of real (inflation adjusted) interest rates.

5 Required every 6 months 6 Use of calcium hypochlorite tablet. Cost based on a chlorination cost of 2.4 €cents per m3, which is in turn based on a calcium hypochlorite cost of 5 USD per kg, and meeting the required dilution of 5 mg of chlorine equivalents per liter of water 7 Considered equal to 2% the cost of the RO unit 8 Considering the maintenance cost of 0.5% for all civil structures, 2% for Mechanical Elements and 4% for Electrical Components as recommended by de Moel [57] 9 Considering the feed flow of the system equal to 1.64 m3/h, the boost pressure required to feed the water to the HPP equal to 1 bar and a pump efficiency 70%,the hydraulic pump power is 0.657 kWh/over 10 hours of operation. As an example, the cost for the kWh in Greece is €0.11/kWh, which means the feedpump operation costs €0.07 per day.

36

FIGURE 23: VARIATION OF ANNUAL WATER PRODUCTION COSTS

Based on Figure 23, the cost of water based on an annual production of 772 m3 (based on the power

output of 10 PV modules10) is shown as a function of the interest rates in Figure 24, where we can

see a linear rise of the cost of water with the interest rates.

FIGURE 24: VARIATION OF WATER COST WITH INTEREST RATES

For a sample case, we consider an interest rate of 3% as advised in [57], with our desired configuration to understand the various contributions to the total cost of a PV-RO system. As we can

see from Figure 25, the PV panels account for a rather small percentage (6.7%) of the total system

10 The reason for choosing 10 solar PV modules with a total area of 16.2 m2 has to do with the maximum space availability on the roof of the testing facility

300031003200330034003500360037003800390040004100420043004400

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

An

nu

al

Co

st

of

Wa

ter

Pro

du

cti

on

(in

€)

Years

0% 3% 5% 0% Averaged 3% Averaged 5% Averaged

y = 15.909x + 4.2634

4

4.5

5

5.5

6

0% 2% 4% 6% 8% 10%

Co

st

of

Wa

ter

(€/m

3)

Interest Rate

37

costs in the given setup. At first glance, this is an indicator that the solar PV panels do not constitute

the dominant (or prohibitive) fraction of the total cost of a PV-RO system. It must be emphasized that the PV-RO system could not be optimized due to practical restrictions of roofing area at the test site.

FIGURE 25: CONTRIBUTIONS TO THE TOTAL COST OF A PV-RO SYSTEM

With increasing number of solar panels, it is possible to extend the operational period of the

batteryless PV-RO system. It is likely that our constrained limit of 10 panels is not optimized with

respect to the production capacity of the RO unit. In the following study, we try understand the

variation of the average water cost with increasing number of solar panels. It is assumed that the

real interest rate is 3%, and that the cost of the solar panels and fixtures scale linearly with the

installed capacity.

In Table 7, different number of panels and resulting values of water production and water cost are

shown for comparison:

TABLE 7: VARIATION IN WATER COST WITH NUMBER OF PV PANELS

Number

of panels

Area

(m2)

Operation

time in

winter (hr)

Operation

time in

summer (hr)

Water

production in

winter (m3/d)

Water

production in

summer (m3/d)

Average

Water cost

(€/m3)

10 16.2 4.25 9.25 1.16 3.12 4.74

15 24.3 6.75 10.75 2.10 3.54 4.17

20 32.4 7.25 11.25 2.40 3.74 3.61

25 40.5 7.75 11.75 2.59 3.91 3.55

We can see that as the number of panels increase, the average water costs decreases. Also seen is

that the reduction in water cost follows the law of diminishing returns. An innate assumption in the

above analysis is that the efficiency of the solar cell is constant at all values of irradiation. This in fact

is not true, and practically, there may be a much smaller gain in operational periods due to increasing

the number of panels than what this analysis bares out. Also, there are difficulties in obtaining larger

uni-directional roof surfaces for PV Panel installation, and beyond a point, it may be required to pay

rent for such an area, something that has not been considered in the present cost evaluation.

RO system, 35.9%

PV panels, 6.7%

Automation, 4.2%

Membrane Replacement, 24.9%

Other Maintenance, 9.3%

Other Operationa

l, 9.0%

Interest Amount, 10.1%

38

39

4. EXPERIMENTATION

The experimental part of the project was conducted in Greece in the campus of the Technical

University of Crete. The field study was carried out between the 3 September and 3 November 2010.

4.1. SELECTION OF THE AREA OF STUDY AND PROJECT PARTNERS

The location in Chania, Crete was selected for various reasons:

Firstly, Crete is one of the southernmost locations in Europe. The mean solar irradiation

values increase as we get closer to the equator which makes it the ideal European location to

experiment with solar PV systems. Furthermore, as it can be seen from graph 1, the

temperate Mediterranean climate in Chania ensures that temperatures do not rise to a point

where they seriously start affecting the performance of the solar cells.

Secondly, the presence of the Technical University of Crete in Chania facilitates the

availability of technological expertise, along with the technical support functions. In

particular, Prof. E. Diamantopoulos who heads the Laboratory of Environmental Engineering

and Management has been actively researching novel environmental engineering techniques

and the project ―Drinking with the sun‖ fits well in the context of the research profile of the

laboratory.

In the framework of this collaboration, the ―Drinking with the Sun‖ group was provided with the

following from the abovementioned laboratory:

A room for the facility, the pumps, electronic equipment and the storage tanks,

Suitable open roof for the installation of the solar panels

Electricity for the laptops, datalogger and feed pump

Assistance, guidance, and financial support for spare parts and pipes during the duration of

the experiments

Available workspace, internet and stationary

Lab facilities for the production of demineralised water and TDS analyses.

For experimental purposes, a reverse osmosis unit (SpectraTM LB 1800) originally placed in the

WaterLab, CiTG, TUDelft was transported and installed at the place of interest.

40

FIGURE 26: LOCATION OF THE TECHNICAL UNIVERSITY OF CRETE, CHANIA

4.2. LOCATION CHARACTERISTICS AND METEOROLOGICAL DATA

FIGURE 27: METEOROLOGICAL DATA FOR CHANIA: ANNUAL TEMPERATURE, PRECIPITATION, WIND SPEED11

AND WATER TEMPERATURE12

The altitude of the location is 137m above the sea level and with coordinates: Latitude: 35° 32' 00"

N and Longitude: 24° 04' 09. The location of the experiments is characterized with high insolation in

the summer period. In 2010, Chania had 2902 total accumulated sunshine hours, reaching peak in

July: 12.3 hrs daily sunshine and 383 sunshine hours in total were recorded13. The average value in

2010 for the daily sunshine hours in Chania was 8.0 hours.

11 The data are acquired from the meteorological station of the Laboratory of Atmospheric Aerosols at the Technical University of Crete. 12 http://www.zoover.co.uk/greece/crete/chania/weather 13 http://www.weatheronline.co.uk/weather/maps/city

41

For a notion of comparison, in Germany (Essen) 1402 sunshine hours were recorded in total in 2010,

while in Indonesia (Bali) 2405 sunshine hours were recorded in total. Furthermore, the average value

in 2010 for daily sunshine hours was 4.3 hours and 7 hours respectively for 2010.

FIGURE 28: TOTAL SOLAR ENERGY (MONTHLY VALUES) OVER 2010 IN CHANIA [FOOTNOTE 11]

A general observation from the Figure 27 and Figure 28 is that during the summer months (June until

September) the precipitation is negligible leading to water scarcity. However, the solar radiation

values reach its peak these months, fact which reinforces the synergy between photovoltaics and

reverse osmosis desalination systems.

4.3. OBJECTIVES

The main objective of the experiments was the testing of the batteryless PVRO prototype under

realistic conditions:

- At a place with high irradiance potential during fluctuating solar conditions

- Using real seawater with the highest salinity (39000 psu) in the World Ocean14.

Specifically, the experimentation results will assist in drawing conclusions about:

Test the integration of the PV panels + LCB controller + RO in one system. Does the system

achieve desalination of water in acceptable quality limits? What is the impact of fluctuating

operation in water quality?

Test the performance of the system with respect to stability and reliability issues. Which is

the response of the system in the absence of an energy buffer?

Test the efficiency of the system. Is it more efficient than similar existing products as

required? Comparing the Specific Energy Consumption of the system with other similar

systems will give an impression of the efficiency of the system.

LCB usability. Does the LCB affect the Specific Energy consumption and does it make a

difference in operational times and water quality in cloudy days?

14 Interconnected system of the Earth's oceanic (or marine) waters; comprises the bulk of the hydrosphere, covering almost 71% of the Earth's surface. http://upload.wikimedia.org/wikipedia/commons/2/28/EU-Glob_opta_presentation.png

42

What is the yield of water during experimentation? How can it be compared to the theoretical

values estimated in the sizing of the system?

Test response of membranes on power fluctuations. Is the lifecycle affected by intermittent

operation?

4.4. EXPERIMENTAL SETUP

In accordance with the component sizing as described in the previous chapter, the batteryless

experimental setup at Chania comprised of the following components:

The reverse osmosis unit with maximum capacity 7.9 m3/d including a high pressure and a

low pressure pump

10 Photovoltaic cells (2.3 kW total installed capacity)

A Linear Current Booster Controller (LCB)

A storage tank for the feed (1 m3) and a storage tank for the permeate (200 L).

4.4.1. THE REVERSE OSMOSIS UNIT

The reverse osmosis unit (SpectraTM LB 1800) which was used for the purposes of the project is

commercially available from the company Spectrawatermakers and designed for land-based

applications. The test rig used in these experiments was built in late 2009 by the MSc. student Brett

Ibbotson at the WaterLab in CiTG, TUDelft.

The facility comprises of:

One microfiltration pre-treatment unit (5 micron nominal filter cartridge, 4‘‘ x 20‘‘)

2 thin-film composite spiral-wound membrane elements for seawater connected in series,

(Filmtec SW30-4040) with salt rejection 99.4 %

The state-of-the-art Spectra Pearson Pump (HPP) which incorporates energy recovery.

(Leeson electric Model C4D18FK6, 2HP, 1800RPM, DC motor 24V, safe operating range: 450

W- 1200 W, rated current: 50 A)

The Spectra LB 1800 system used for the experimentation has a fixed water recovery equal to 20%,

which means that 20% of the feed flow is produced as permeate.

The use of a feed pump

The Pearson pump required the inlet flow to be supplied between a minimum of 0.7 bar and a

maximum of 1.4 bar. The pressurization of the inlet stream can be achieved by passive means such

FIGURE 29:THE EXPERIMENTAL SETUP IN THE WATERLAB, CITG,

TU DELFT [1]

43

as storing feed water in a tank located more than 12 m above the Pearson pump. However for the

purpose of experimenting in the given scenario (with limitations of space and infrastructure), active

means were used, involving a feed pump15 along with a pressure relief valve configured to open at

1.2 bar pressure. For the experimentation study, the system of the feed pump was considered

independent of the main system and was not powered by the PV source. This may be justified to an

extent, since the power consumption of the high pressure Pearson pump is much greater than that of

the feed pump.

System configuration

Since it was not practical to continuously consume sea water (due to logistics constraints of

transporting sea water to the site of experimentation), the feed water is recirculated. The

configuration of the experimental setup is such that all the water streams together form a closed

loop. The feed pump delivers water at around 1 bar to the high pressure pump via the 5 MF cartridge

filter. The water after passing through the RO membranes is separated into a concentrate and a

permeate stream. The concentrate stream after passing through the Pearson pump for energy

recovery is discharged back into the feed water reservoir. The permeate stream is also discharged

into the feed water reservoir. Since both the streams are directed back into the feed water reservoir,

the net salinity of the feed water remains unchanged. This operation is said to be the ―RUN‖ mode as

shown on the control panel of the LB1800.

Flushing operation

After a long cycle of normal operation or after an inactivity period of more than 12 hours, the

membranes need to be flushed with fresh water in order to be rinsed from the accumulated salts.

Although the LB 1800 is provided with a ―FLUSH‖ mode where water can enter the system via a

separate line, the system was mostly flushed using the normal ―RUN‖ mode (with manually altered

feed streams) in order to allow the fresh water to also pass through the prefilter.

In order to maintain the concentration of the artificial seawater at a constant level during flushing

with fresh water, the initial concentrated discharge is directed into a bucket to capture the salts being

flushed out. Once the conductivity of this discharged water approaches the conductivity of the feed

water, the remainder is directed to the gutter with the captured water being returned to the

reservoir. Similarly once the flush had been completed and the system is put back into a regular

cycle, the initially discharged water was discarded. Further adjustments to the level and conductivity

of the reservoir water were made as necessary to ensure constant characteristics.

4.4.2. PHOTOVOLTAICS ARRAY

Based on the sizing calculations and after a thorough market research was decided to use 10 ―Energy

Power Plus‖ panels with capacity 230 Wp each (rated at 30 V and 7.8 A), giving a total peak capacity

of 2.3 kW. Each panel covers an area of 1.62 m2, thus an area of 16.2 m2 in total was used.

The roof of the building where the collaborating laboratory was located was selected for the

installation of the solar panels. The location of the roof was ideal due to the following reasons:

- The building is located on a hill and there is no taller building in the vicinity which can cast a

shadow over any part of the panels during the peak hours of energy production

- The access to the roof is restricted by a gate which ensures the security of the array from

theft/damage.

15 The feed pump used was a Low Pressure Pump (Liquiflo™ 37R positive displacement magnetic-drive pump).

44

- The roof was exactly above the Laboratory and above the room where the desalination

system was installed, leading to short cable requirements and thus less power losses.

- The orientation of the roof was almost south, which is ideal for solar panels located in the

Northern hemisphere.

TABLE 8: ROOF CHARACTERISTICS

FIGURE 30: ROOF ON WHICH THE PV PANELS WERE INSTALLED

The roof inclination is coincidentally close to the optimal panel inclination angle proposed by Joint

Research Centre (JRC) of the European Commission for the area of Chania (28ο for the whole year)

hence no additional means were required to alter the angle of the panels. Furthermore, despite the

fact that tracking mechanisms may lead to increase in efficiency of the system, it was decided that it

would be unjustified with respect to the cost, since the optimal inclination angle during the period of

testing (September-October) is in any case close to the existing angle of the roof (33Ο for September

and 46ο for October) based on JRC data.

Special mounting structures were used for the panels in order to avoid drilling on the roof. The panels

were connected in parallel in order to keep the output Voltage at 30 V (rated) which was required by

the Pearson pump motor (rated at 24 V). Consequently, the maximum current of the solar array was

78 A. Further specifications for the PV panels can be found in the appendix.

TABLE 9: PV PANEL CHARACTERISTICS

4.4.3. LINEAR CURRENT BOOSTER (LCB) AND SYSTEM INTEGRATION

From the literature study, it was seen that a DC-DC converter was necessary between the

photovoltaics and the pump to optimize the integration of PVs with the pump motor. Since MPPT

circuits exist mostly as charge controllers, it was decided to integrate a Linear Current Booster (LCB)

in the PVRO system. As it was described in earlier chapters, the LCB is a converter which exchanges

voltage for current preventing the motor from stalling in times of low irradiation. The system

requirements were submitted to the Canadian company ―Solar Converters Inc‖ and the controller was

custom fabricated for the purposes of the project (Figure 32).

Roof characteristics

Length (m) 18

Inclination from

the horizontal (deg)

30ο

Orientation SSW (30ο from South)

PV characteristics (per panel)

Capacity (Wp) 230

Rated Voltage (V) 29.5

Rated Current (A) 7.8

Area (m2) 1.62

Size (mm) 1636 (L) x 992 (W) x 50 (H)

FIGURE 31: SOLAR PV PANELS INSTALLED ON THE ROOF

45

The LCB controller was supplied with incorporated various other user controls, LED displays and

protections. Most importantly, the device current limits its output to protect the unit and the load

from jammed motor/ low resistance operation. After the manufacturer‘s suggestion, fuses after the

LCB and before the motor were installed. Further specifications of the LCB can be found in the

appendix. For added security, fuses after the photovoltaics load and before the LCB were

incorporated. As an addition at a later stage, a switch was installed depending on the position of

which either the PV source or the AC/DC rectifier could be used to power the motor. This was very

useful in order to perform tests with constant power for reference and for troubleshooting needs.

The final electrical connections can be seen in Figure 33:

FIGURE 33: ELECTRICAL CONNECTIONS OF THE MAIN COMPONENTS OF THE SYSTEM

4.4.4. INSTRUMENTATION

For the requirements of the experimental part of the project, several measuring devices were used in

order to monitor and control important parameters of the process and acquire useful data for the

analysis of the experiments.

The following measuring instruments were used:

FIGURE 32: LINEAR CURRENT BOOSTER (LCB)

46

Pyranometer for the measurement of the instantaneous irradiation

Two electrical conductivity meters for the feed flow and the permeate flow (indirectly

measuring salinity)

Voltmeter and ammeter for measurement of voltage and current of the motor

Membrane pressure gauge

A flow meter gauge for the product flow

Two analogue pressure gauges for measuring the pressure of the feed at the inlet to the pre-

filter and at the exit of the membranes in bar

Analogue flow meters for the feed flow and the permeate flow

The analogue meters and gauges located on the Spectra control box were used to monitor the

changes in operating parameters during manual control of the Pearson motor speed. The

pyranometer used gives a linear voltage production until 2000 W/m2 and it has sensitivity 69.3

μV/W/m2. Temperature of the feed and permeate water flows could be observed on the display of the

two electrical conductivity meters. Since conductometric measurements are temperature dependent,

the meters use the function of temperature compensation to convert all measured values to a

reference temperature (25ο C).

FIGURE 34: INSTRUMENTATION SYSTEMS (FROM LEFT TO RIGHT): ANALOGUE PRESSURE GAUGES /

FLOWMETERS, DIGITAL CONDUCTIVITY METERS, PYRANOMETER

Most of the data variables from the measuring instruments were stored independently using a 16-

channel digital datalogger. The parameters could be monitored in real time from the datalogger

interface installed on a laptop, while in the same time were recorded. The calibrations of the

analogue data inputs are given in the appendix. Table 10 shows a list of parameters logged by the

instruments.

TABLE 10: PARAMETERS LOGGED BY THE INSTRUMENTS

Parameter Type of meter

Instantaneous irradiation at the flat surface (W/m2) Pyranometer

Motor Voltage (V) Voltmeter

Motor Current (I) Ammeter

Membrane pressure (bar) Pressure meter

Permeate flow (l/min) Flow meter

Electrical Conductivity of feed flow mS/cm Conductivity meter

Electrical Conductivity of permeate flow μS/cm Conductivity meter

47

FIGURE 35: RO UNIT CABINET ON FIELD AND DESK STATION ON FIELD

4.5. SYSTEM START UP

The reverse osmosis unit was transported from Netherlands to Chania, Greece together with the feed

pump, tubes and spare filters and membranes, packaged in a specially constructed wooden case. The

initial preparation activities involved setting up the location for the system and the work station,

obtain all necessary tanks and equipment, piping and modifications in valves and fittings, and

transferring water from the sea. Furthermore, the installations of a datalogger, electronic devices and

connections as well as calibrations were also done in preparation for the experimentation.

Since one of the objectives of the exercise was to approach reality as much as possible, it was

decided to use real sea water for the experimental purposes. Sea water was transferred from the

nearest organized beach (Agios Onoufrios), which had excellent water quality (zero tars, mineral oils,

seaweed or microbiological activity. The water was pumped from the sea with a submersible pump

provided by the laboratory group. The seawater was filtered with a 50 MF net to ensure that all

suspended solids are removed. A tank of 1 m3 was procured to store the feedwater at the location of

the RO system. A block diagram of the seawater PV-RO system installed at Chania is shown in Figure

36:

FIGURE 36: SIMPLIFIED PROCESS DIAGRAM OF PV-RO SYSTEM

48

4.5.1. DETERMINATION OF TDS AND ELECTRICAL CONDUCTIVITY RELATION

The TDS content and the electrical conductivity of a water sample are closely related. Conductivity

increases with increasing ion concentration, which means that in most cases it gives a good

approximation of the TDS measurement. As is found in literature16, there is no one single constant

linking the values of conductivity (in µS/cm) and TDS (in ppm). To determine the precise relation

between TDS and electrical conductivity for the sea water used for the experiments, TDS analysis was

carried out in the laboratory. This involved evaporating water from measured filtered solutions of

known dilutions and weighing the mass of the remaining solids. Figure 37 was plotted based on the

results obtained from the TDS analysis.

FIGURE 37: RELATION BETWEEN TDS AND ELECTRICAL CONDUCTIVITY OF THE FEEDWATER USED

The electrical conductivity value for the seawater used was measured to be 53.4 mS/cm (at 26ο C).

Based on the correlation factor obtained from the TDS analysis, it was calculated that the electrical

conductivity of the seawater used in the experiments (53.4 mS/cm) corresponds to TDS= 38353

ppm.

4.6. SYSTEM OPERATION

The motors of the two pumps (low and high pressure pumps) were fully controlled via a computer

when the system was placed in the WaterLab, in Delft. However, since the addition of an automated

controller increased the complexity of the system, it was decided to avoid any motor speed controllers

during the experimentation period in Crete. Since the Pearson pump is powered directly by

photovoltaic cells, the power available for the Pearson pump is varying, depending on the fluctuation

in the incident solar radiation. To prevent the pump from stalling in times of low power availability,

the pump speed needs to be controlled manually (using a knob on the control box). To exploit the

available solar energy to the maximum possible extent (thus maximizing production of fresh water), it

is necessary that the motor speed precisely follows the variations in solar irradiation.

16 http://www.appslabs.com.au/salinity.htm

y = 718.23x

R2 = 0.9958

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

11000

12000

13000

14000

15000

16000

17000

18000

19000

20000

21000

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

Conductivity (mS/cm)

TD

S (

mg

/l)

49

The speed of the feed pump motor was set constant at maximum value while the pressure relief

valve was used to prevent overpressurising the stream line. The system efficiency is not affected by

this apparent ‗loss‘ of feed water, since the feed pump is not powered by the photovoltaics.

The feed water was recirculated during the experiments the same way as in the WaterLab.

For the month of September, starting time operation was around 8:45 and ending time 18:00 for a

clear day while for the month of October, starting time operation 9:15 and ending time 17:30

respectively.

The activities to be done every day for a full cycle of operation can be summarized as follows:

Setting up the laptop and configuring a new logging session. For most of the experiments the

time step was set at 30 s.

Adjusting the configuration of the system and valves for the ―RUN‖ mode

Starting the PV system and the pumps

The manual intervention required during experimentation involved altering the speed of the

Pearson Pump depending on the solar fluctuations, and noting the values of certain system

parameters which were not being recorded by the data logger.

The system was flushed every evening after the sun had set with the 150 L permeate water

stored in a permeate barrel.

4.7. PROBLEMS FACED / TROUBLESHOOTING

After the first set up of the system, numerous problems or challenges were encountered which led to

various modifications to solve these problems. The main problems that required troubleshooting are

as follows:

Lack of unchlorinated tap water to cover flushing requirements of the system

One of the most important aspects in a reverse osmosis process is the fresh water flushing to reduce

biofouling. However, flushing with freshwater proved to be a great problem on site since the

municipal tap water in Chania is chlorinated for disinfection purposes. The sea water membranes

used (Filmtec SW30 40-40) have zero free chlorine tolerance and for this reason premature

membrane damage due to oxidation could occur in case of directly flushing the membranes with tap

water.

As a first solution, demineralised water was produced at the Laboratory facilities in order to flash the

system. However, soon it was noticed that this was neither a practical nor an economical solution,

since more than 200L of demi water was required to flash thoroughly the system.

As a final solution, permeate water was decided to be used for flushing to overcome this problem.

The following tank configuration was adopted to ensure that at the end of the day adequate quantity

of product water would be available for flushing:

A second tank (200 L) was placed above the 1 m3 tank, in order to store permeate water. The

intention was that during normal running of the system, the permeate flow would discharge in the

200 L barrel while the concentrate would return back to the 1 m3 tank. The 200L barrel has been

modified in such a way, that after filling in of 150 L, the permeate stream was overflowing back to

the 1 m3 tank.

50

FIGURE 38: TANK CONFIGURATION FOR PERMEATE COLLECTION

Increase in feedwater concentration because of new tank configuration

During the first ―running‖ of the system in the two-tank configuration, it was observed that the

salinity in the 1 m3 tank was increasing, since the permeate flow was discharged in the barrel and not

returning to the feedwater tank until the barrel started overflowing. At the point when the barrel

starts overflowing, steady state is reached, with a certain feed water concentration. Starting with an

initial salinity of the sea-water in Chania (38353 mg/l), it is found that at steady state, the salinity in

the feedwater tank had risen up to 43000 mg/l TDS. Hence, the salinity of the feed water which

would have been used in experimentation would have not been the same as the real salinity of the

seawater source at steady state.

To solve this problem, a part of the sea water contained in the 1 m3 tank was diluted with

demineralised water in order to achieve the desired feed water concentration at steady state

conditions (equal to the original feedwater source around 38353 ppm TDS). The required starting

feedwater concentration and the volume of demineralised water required to replace part of the

original seawater were found by solving simple salt mass balances:

1 1 2 20

0

V C V CC 32675 ppm

V 0

replace 0

1

CV V 1 148 L

C

Where: Vo = Total volume of the feedwater tank (1 m3)

Co = Starting feedwater concentration

V1 = Volume of the feedwater at steady state = (Vo-V2)

C1 = Desired feedwater concentration (at steady state) = 38535 ppm

V2 = Volume of the permeate water in the barrel (150 L)

C2 = Concentration of the permeate water in the barrel (500 ppm)

Increased water Temperature and overheating of the pump

Overheating of the Pearson Pump motor and of the feedwater in the tank was a main problem during

the experimentation period, especially in September when the ambient temperature was very high

(>30ο C). The problem was aggravated due to lack of ventilation within the room. A small table fan

and later a mobile air-conditioner unit were used in order to supplement the cooling effect of the

motor fan.

51

In addition to the overheating of the motor of the pump, the water temperature was rising on an

average from 27ο C to 32ο C during the course of the day. Various simple solutions were used to

tackle the problem, like insertion of icepacks in the tank and use of ice cubes around the water tubes.

However, these simple solutions did not manage to solve the problem effectively. Yet, temperatures

at the end of September decreased significantly and remained low for the rest of the experimentation

period and solutions like a heat exchanger in the tank or use of dry ice were avoided (Figure 39).

FIGURE 39: CHANGE IN FEED WATER TEMPERATURE DURING DAILY OPERATION

Damage of parts of instrumentation equipment and unavailability of spare parts

Very frequently accidents with the electronics and instrumentation equipment were occurring. One of

the most difficult to solve was the failure of the motor voltage measuring sensor which was basically

a voltage divider comprising of 2 resistors of very low tolerance. Availability of such spare parts was

very limited in Chania, and thus this influenced the data acquisition for about 3 days until the voltage

sensor was restored.

Figure 40 shows the final process and energy flows along with all the modifications made to the

system.

27.5

28.2

28.9

29.8

31.932

31.8

27.1

27.427.6

28.2

28.628.8

29.129.2

26.8

27.2

27.6

28

28.4

28.8

29.2

29.6

30

30.4

30.8

31.2

31.6

32

32.4

9:00 9:43 10:26 11:09 11:52 12:36 13:19 14:02 14:45 15:28 16:12 16:55 17:38 18:21

Te

mp

era

ture

(C

)

15/09/2010

19/10/2010

52

FIGURE 40: FINAL PROCESS FLOW DIAGRAM AND INSTRUMENTATION

53

5. RESULTS AND DISCUSSION

5.1. A SAMPLE DAY

An example of an experiment which combines clear and cloudy conditions is shown in Figure 41.

FIGURE 41: SOLAR IRRADIATION ON A SAMPLE DAY (5/10/2010)

Operation period for 05/10/2010 was a little bit less than 9 hours. Starting time was around 9:00 and

stopping time was around 18:00. An abrupt fall in irradiance occurred between 11:00-12:00. Similar

trends in operation time were observed during the other experimentation days.

The daily water production and the total energy consumption are calculated by integrating the

instantaneous flow rate and power consumption by the Pearson pump motor.

Total water production per day (m3):

The total water production (Y) can be calculated by integrating the rate of water produced (per

minute) (y) over the operation time period. Since the time step is constant, the total water production

can be calculated by the addition of each instant permeate flow (Qp) and by multiplying with the time

step (30sec).

stop

i

start

t 18:15:10

i p

t 9:00:00

l minY y dt y dt sum Q 30sec

min 60sec1.94 m3

Energy consumption (kWh):

Similarly, by integration of the power consumed, the total energy produced over the operation time

period can be calculated.

stop

start

t 18:15:10

i i

t 9:00:00

hrE Power dt Power dt sum Power W 30sec

60 60sec5381.44 Wh = 6.24 kWh

Specific energy consumption (kWh/m3):

E / Y = 3.22 kWh/m3

0

100

200

300

400

500

600

700

800

900

8:00 8:28 8:57 9:26 9:55 10:24 10:52 11:21 11:50 12:19 12:48 13:16 13:45 14:14 14:43 15:12 15:40 16:09 16:38 17:07 17:36 18:04 18:33 19:02

Irra

dia

nce (

W/m

2)

54

It is important to note that the instantaneous power delivered to the system is not only dependent on

the instantaneous irradiance value but also on the effect of light diffusion. For example, the power

produced in the morning with direct incident sunlight, with a specific irradiation value gives more

power than what is produced in the afternoon with diffused light having the same irradiance value.

This can be attributed to the characteristics of the photovoltaics.

In table 11 the accumulative results obtained for the above measurement are contained:

TABLE 11: PV-RO RESULTS FOR THE SAMPLE DAY (5/10/2010)

Date:

5/10/2010

LCB CONNECTED Parameters

Irradiation

(W/m2)

Voltage

(V)

Current

(A)

Power

(W)

Pressure

(bar)

Flow

(lpm)

Flux

(l/h*m2)

ECp

(μS/cm)

TDS

permeate

(mg/l)

max 841.11 30.30 43 1034.7

2

49.09 4.98 20.17 1761.20 1264.95

min 55.05 15.61 0 9.10 0.33 0.11 0.43 3.66 2.63

average 473.92 25.17 28 673.79 41.46 3.49 14.14 963.10 691.73

Daily water

Production

(m3)

1.94

Total Energy

consumption

(kWh)

6.24

SEC

(kWh/m3)

3.22

The feedwater temperature on 05/10/2010 ranged between 27 οC- 29.8 οC. The variations of the

above mentioned parameters during the day are shown in Figure 42.

55

Based on the above results and trends, the following observatins regarding general behaviour of the

system can be made:

In the beginning of the system operation, the permeate quality falls abruptly (conductivity

and TDS rises), due to the concentration of salts on the membrane surface. After some time,

conductivity falls rapidly due to the removal of the accumulated salts from the membranes.

With the increase in power (blue line), the cross-flow velocity rises and the permeate flow

(green line), flux (brown line) and pressure (pink line) follow the same increasing trend. The

reason why the permeate flow perfectly follows the motor power is due the fact that the

system has a fixed recovery ratio. The same can be said about the pressure (pink line), since

the reason for pressure rise in the membranes is also the fixed recovery ratio of the system.

The same does not apply however, for the electrical conductivity (purple line), which is also

affected by other factors such as cross-flow velocity, etc. Lower power leads to lower cross-

flow velocity and permeate flow leading to low dilution of the diffused salts on the permeate

side and potential increase in the concentration polarization.

In general, we can see that a sustained stable operation of the system, leads to a lower value

of electrical conductivity (which translates into better quality of the water output).

Rapid fluctuations in the solar irradiation do not directly affect the electrical conductivity as

they affect the flow and pressure. In contrast, electrical conductivity seems to be influenced

by longer trends in the solar irradiation. Sustained and sharp fluctuations which lead to

stopping of the system for long time (more than 5 min) significantly affects the quality of the

048

12162024283236404448525660646872768084889296

100104108112116120124128132136140144148152156160164168172176180

EC permeate/10 (μS/cm)

Power consumption/10 (W)

Pressure (bar)

Current (A)

Permeate Flux (l/h*m2)

Permeate flow (l/min)

FIGURE 42: VARIATION OF OPERATION PARAMETERS IN TIME ON SAMPLE DAY: 5/10/2010

56

water produced. After stopping and starting the system again, salts which are accumulated

on the membrane surface get diffused on the permeate side deteriorating the quality.

As it is seen from the solar irradiation profile, there was a significant cloudiness between 11

am and 12 pm leading to a large drop in solar PV power production. The abrupt drop in

radiation was to such a large extent (to <200 W/m2) that the system was stopped since it

could not continue running at the low torque. The conductivity and thus the water quality are

dramatically affected by the start-stop of the system. This can be attributed to the

accumulation of salts on the membrane surface due to the sudden stopping of the system in

response to the insufficient power.

At the following graph, the inverse relation between water quality and power (as well as permeate

flow) can be observed more closely:

When the motor runs in higher speed (higher power consumption), the flow through the pump is

increased and thus the permeate flow also increases, while the water quality improves significantly.

This can be attributed to the greater cross-flow velocity resulting from the increased feed flow.

5.2. GENERAL COMMENTS ON EXPERIMENTATION

From 14/09/2010 to 03/11/2010 several experiments were carried out on location in Chania, both

with and without the LCB connected. The experiments done during the first days were mostly to

observe how the system operates and do the necessary calibrations and modifications for optimizing

the process. For this reason, the experimental results during 14/09-16/09/2010 are not very

representative.

FIGURE 43: RELATION BETWEEN WATER QUALITY, POWER CONSUMPTION AND FLOW ON 5/10/2010

57

Furthermore, the experiments which were conducted during October represent the worst possible

scenario since the solar radiation pattern best resembles the winter season, with lots of passing

clouds. Two examples of the worst solar conditions in the month of October are shown in Figure 44,

during which operation of the pump was impossible.

FIGURE 44: WORST-CASE SOLAR IRRADIATION PATTERNS DURING EXPERIMENTATION PERIOD AT CHANIA

Various tests were carried out in the beginning in order to settle the multiple parameters calibrations

as well as to acquire trends among parameters and reach the optimum operational conditions. After

numerous measurements, a correlation between irradiation and feed flow was observed as shown in

Figure 45.

Based on these observations, it was discovered that above approximately 600 W/m2, the feed flow is

the maximum possible (6 gpm or 22.7 liters per minute) while for irradiation values lower than 200

W/m2 the feed flow is lower than 3 gpm (or 11.4 liters per minute). System start-up was attempted at

even lower values of incident irradiation; however, at these conditions the pump was clearly unable

to run normally and showed intermittent operation.

FIGURE 45: IRRADIATION VS FEED FLOW ON 16/09/2010

Based on the results of the experiments some general remarks can be made regarding the

relationship between power consumption, permeate flow and irradiation. As can be seen from Figure

46, the relationship between permeate flow and power consumption is nearly linear. One significant

observation is related with the cut-in power value (around 350W) under which the slope of the trend

58

changes (for permeate production less than 2 l/min). Another observation is that above 1000 W the

trend levels off at the maximum permeate flow achieved value of 5.2 l/min. However, it is important

to note that depending on the custom system configuration these boundaries may vary.

FIGURE 46: RELATION BETWEEN PERMEATE PRODUCTION AND POWER CONSUMPTION ON A MOSTLY CLEAR

DAY: 17/09/2010

In Figure 47, measured irradiation values have been plotted with the corresponding power

consumption values for the course of one mostly clear day. The main observation here is that a

certain power value can be achieved at lower irradiation values during the second half of the day than

in the first half of the day. This can be attributed to the light source (diffused or direct) and also to

the SSW orientation of the panels.

FIGURE 47: POWER CONSUMPTION VS IRRADIATION ON A MOSTLY CLEAR DAY: 17/09/2010

Evaluation of all full-day experiments showed the ranges for the important system parameters such

as total permeate production, total energy production and the specific energy consumption.

Second half of the day:

After-noon

First half of the day:

(Before noon)

59

TABLE 12: OBSERVED VARIATION IN PV-RO SYSTEM PARAMETERS

Depending on the weather conditions and the solar fluctuations, the total daily water production

varies. It can be observed that even in the worst case scenario, the SEC values do not increase

significantly. For comparison measured values during an 8-hour operation directly powered by the

grid can be found in Table 13.

TABLE 13: RO SYSTEM PERFORMANCE WHEN RUNNING ON GRID POWER

Experiment powered by the grid in

October

Value

Total Water production (m3) 2.2617

Total Energy consumption (kWh) 6.39

Specific Energy Consumption (kWh/m3) 2.83

5.3. ROLE OF LINEAR CURRENT BOOSTER (LCB)

In the last week of September (prior to LCB connection), the sky was heavily clouded (leading to a

(global) irradiation of <200 W/m2 for most time of the day for 4 days) and system operation was

difficult. The motor was abruptly stalling and restarting, leading to significant vibrations.

The motor was starting and stopping almost every 10 minutes. However, that practice is not safe for

the motor over extended time scales, and neither does it produce good results. The situation was

aggravated by the absence of a speed controller for the Pearson pump, the need for which is felt

especially in times when there are large fluctuations in the irradiation levels.

After the connection and operation with the LCB, the following were observed:

Since a greater torque is delivered to the motor at low voltages, the motor doesn‘t stall in

times of low irradiation

Solar fluctuations smoothened by the LCB. The controller avoids deep and steep changes of

motor speed.

By observing the LCB LEDs which measure the voltage level supplied every moment, it is easy

to predict when the motor speed has to be adjusted (in clear sky conditions).

Significant exchange of voltage with current occurs most of the time, in order to supply with

the maximum instant current possible. However, the limitation of the LCB is 40 A, and current

boosting above that limit is not possible. With the LCB connected, it was observed that the 40

17 The RO unit has a water production capacity of 7.9 m3/day which can be translated to 2.6 m3 for 8 hours. However, due to capacity limitations of the rectifier used, it was only possible to be produced 2.26 m3 in 8 hours.

Experiments powered by PV in

September

Range

Total Water production (m3) 1.65 - 2.55

Total Energy consumption (kWh) 1.05 – 8.58

Specific Energy Consumption (kWh/m3) 2.7 - 3.37

Experiments powered by PV in October Range

Total Water production (m3) 0.33 - 2.15

Total Energy consumption (kWh) 1.07 – 7.32

Specific Energy Consumption (kWh/m3) 2.73 - 3.50

60

A corresponded to feed flow of 6 gpm or higher, above which it is not possible to run the

system. Consequently, the limitation of the 40 A does not effectively limit the system‘s

capacity.

The LCB works better when there are no rapid fluctuations. When irradiation reduces

gradually, then there is better exchange of voltage with current.

The above points can be comprehended better in the following case-studies:

5.3.1. CASE STUDY 1: COMPARISON BETWEEN A CLEAR DAY IN SEPTEMBER

(WITHOUT AN LCB) AND AN ALMOST CLEAR DAY IN OCTOBER (WITH AN LCB)

Most of the days in October had significant fluctuations in irradiation with different length, depth,

frequency and intensity of cloud coverage. This made the comparison with each other difficult. By

comparison of two clear days it is easier to draw some conclusions about the effect of using the LCB.

A clear day in September, when LCB was not connected, was chosen to be compared with the only

clear day in October with the LCB connected. The two days were compared for a specific irradiance

range (500 W/m2 in the morning until 600 W/m2 in the afternoon) when no clouds existed to hinder

the irradiance in both days. Although the irradiation in October is lower than the irradiation in

September, useful conclusions can be deduced from the following comparisons:

17 September 2010 – NO USE OF LCB: Operating period: 9:58:22 (500W/m2) - 16:04:11 (600 W/m2)

06 October 2010 – USE OF LCB: Operating period: 10:20:29 (500W/m2) - 15:19:29 (600 W/m2)

The following observations were made regarding the daily irradiation values for the two days:

First observation is that in September irradiation reaches 500 W/m2 around 20 min earlier

than it reaches in October.

The maximum irradiation for the day in September was 837 W/m2 at 13:33, while in October

the maximum irradiation was 746 W/m2 at 13:00.

Finally, at the second half of the day, 600 W/m2 was reached at 16:04 in September, while in

October the same amount of irradiation was reached almost 40 minutes earlier, at 15:19.

FIGURE 48: IRRADIATION OVER TIME ON 17/09/2010 AND ON 06/10/2010

61

The following diagrams were created, to compare parameters for the same irradiation range in both

days:

FIGURE 49: COMPARISON OF VARIATION IN MOTOR VOLTAGE FOR ONE CLEAR DAY WITH THE LCB AND ONE

CLEAR DAY WITHOUT THE LCB (CASE STUDY 1)

The motor voltage in the day when LCB was connected was considerably lower than the Voltage in

September when LCB was not connected.

FIGURE 50: COMPARISON OF VARIATION IN MOTOR CURRENT FOR ONE CLEAR DAY WITH THE LCB AND ONE

CLEAR DAY WITHOUT THE LCB (CASE STUDY 1)

Despite the lower instant irradiation in October, the average current supply is much higher in October

due to the work of the LCB.

62

FIGURE 51: COMPARISON OF PERMEATE FLOW FOR ONE CLEAR DAY WITH THE LCB AND ONE CLEAR DAY

WITHOUT THE LCB (CASE STUDY 1)

Even as one would expect much lower permeate flow rates in October (as compared to September)

due to lower irradiation values and operational times, due to the use of the LCB, the average

permeate flow in October tends to approach the value of the permeate flow in September. (On

17/09/2010 without LCB: Qp=4.41 l/min while on 06/10/2010 with LCB: Qp=4.40 l/min).

The flux follows as expected the same trend as the permeate flow. The permeate flux in September is

generally higher than in October, due to the higher irradiation levels, and consequently the longer

operating period with high flowrates than in October. However, the higher the current, the higher the

permeate flux, and the use of LCB on 06/10/2010 compensates for the difference in irradiation and

operational time (On 17/09/2010 without LCB: average flux=17.87 l/h*m2 while on 06/10/2010 with

LCB: average Flux=17.84 l/h*m2).

FIGURE 52: COMPARISON OF VARIATION IN PERMEATE QUALITY FOR ONE CLEAR DAY WITH THE LCB AND

ONE CLEAR DAY WITHOUT THE LCB (CASE STUDY 1)

63

As far as the water quality is concerned, the following remarks can be made based on Figure 52:

The TDS obtained using an LCB is consistently lower than what is obtained without using an

LCB.

For higher current, permeate flow and flux are higher. And the higher the permeate flow and

flux, the better the water quality. This is the reason why in October although the irradiation

range is lower, the permeate quality manages to be even better than September.

Comparable TDS levels are achieved at much lower irradiation levels with the use of LCB. For

e.g. by observing the above graph, if the LCB was not used, to get a TDS level lower than

550 mg/l (say), we would need to wait until after mid-day! However, if an LCB is to be used,

that quality of water could be obtained since about 11 o‘clock in October (in September, this

would have been even earlier). Concluding, the starts up times are lower with the use of LCB

and thus operational times are much higher.

Although SEC on the day when LCB is used is lower than the SEC on the day when it is not

used, this does not apply to all other experimentation results in general, due to the varying

solar conditions.

5.3.2. CASE STUDY 2: COMPARISON BETWEEN TWO CONSECUTIVE SIMILAR DAYS IN

OCTOBER WITH AND WITHOUT THE USE OF LCB

The days of 12/10/2010 (without using an LCB) and 11/10/2010 (using an LCB) will be compared in

order to deduce the effects of using the LCB.

12 OCTOBER 2010 – NO USE OF LCB

Operating period: 9:25:10 - 18:24:10

FIGURE 53: DAILY IRRADIATION VALUES ON 12/10/2010

The day started sunny until noon when big clouds caused abrupt fluctuations until late in the

afternoon. The lack of LCB made the motor stall during these fluctuations and caused frequent starts

and stops. Accumulated data can be observed at the following table:

Irradiation over time 12/10/2010

0

100

200

300

400

500

600

700

800

900

1000

9:20 9:48 10:17 10:46 11:15 11:44 12:12 12:41 13:10 13:39 14:08 14:36 15:05 15:34 16:03 16:32 17:00 17:29 17:58 18:27

Irra

dia

nc

e (

W/m

2)

64

TABLE 14: PV-RO OPERATIONAL PARAMETERS (12/10/2010)

11 OCTOBER 2010 – USE OF LCB

Operating period: 8:55:09 – 16:49:30

FIGURE 54: DAILY IRRADIATION VALUES ON 11/10/2010

The day started partially sunny until noon when clouds caused intense fluctuations. After 14:00

serious cloud coverage occurred leading to abrupt fall in irradiance. The cloud coverage was just as

intense until system operation was stopped at around 17:00 since the motor speed was very low

leading to very low feed flows and bad water quality. In addition, operation of the motor in low

speeds is not a safe practice.

TABLE 15: PV-RO OPERATIONAL PARAMETERS (11/10/2010)

Irradiation over time 11/10/2010

0

100

200

300

400

500

600

700

800

900

1000

8:50 9:18 9:47 10:16 10:45 11:14 11:42 12:11 12:40 13:09 13:38 14:06 14:35 15:04 15:33 16:02 16:30 16:59 17:28 17:57 18:26

Irra

dia

nc

e (

W/m

2)

Date: 12/10/2010 Parameters

Irradiation

(W/m2)

Voltage

(V)

Current

(A)

Power

(W)

Pressure

(bar)

Flow

(l/min)

Flux

(l/h*m2)

ECp

(μS/cm)

TDS

permeate

(mg/l)

Max 857 33.61 38.31 982 51.08 5.25 21.27 1122.64 806.32

min 66 17.74 0.09 3 0.30 0.11 0.44 3.67 2.63

average 419 28.53 21.33 598 43.42 3.42 13.88 848.28 609.26

Total daily water production 1.85 (m3)

Total Energy consumption 5.38 (kWh)

Specific Energy Consumption

(SEC)

2.91 (kWh/m3)

Date: 11/10/2010 Parameters

Irradiation

(W/m2)

Voltage

(V)

Current

(A)

Power

(W)

Pressure

(bar)

Flow

(l/min)

Flux

(l/h*m2)

ECp

(μS/cm)

TDS

permeate

(mg/l)

max 901 30.33 37.96 920 50.81 5.02 20.37 1455.73 1045.55

min 57 11.13 0.00 0 0.29 0.11 0.43 616.68 442.92

average 442 24.29 27.28 654 45.60 3.80 15.42 774.04 555.94

Total daily water production 1.8 m3

Total Energy consumption 5.17 kWh

Specific Energy

consumption

(SEC)

2.87 kWh/m3

65

Some general comparative remarks for the above mentioned two consecutive days are the following:

Despite the intense solar fluctuations during these cloudy days, SEC is significantly low

(<2.91 kWh/m3). Although SEC on 11/10/2010 when LCB is used is lower than the SEC on

12/10/2010 when it is not used, this does not apply to all other experimentation results in

general.

Rapid and small fluctuations do not considerably influence the permeate quality and

conductivity.

In contrast, sustained and sharp fluctuations which lead to stopping of the system for long

time (more than 5 min), significantly affect the quality of the water produced. After stopping

and restarting the system, salts which are accumulated on the membrane surface get

diffused on the permeate side deteriorating the quality.

The difference in irradiation distribution, intensity and fluctuations in these two days makes it

very difficult to compare the two days with each other.

Nevertheless, during operation with LCB (11/10/2010) the current is higher than during

operation without the LCB (12/10/2010), the average permeate flow and flux is higher during

11/10/2010, which leads to average lower TDS (see tables with summarized results for these

two days in appendix).

5.4. OBSERVATION OF SUDDEN QUALITY DETERIORATION

After a long series of experiments in heavily clouded conditions during October, deterioration in the

water quality was observed at around the end of the experimentation period. Experimental sessions

powered by the grid at maximum possible feed flow rate (6 gpm or 23 L/min) were carried out to

observe the system‘s response under constant power conditions and constant feedwater salinity.

However, despite a steady rise of electrical conductivity, the other operational parameters

(membrane pressure, permeate flow) remained constant (Table 16). During 4 hours operation of the

system the total permeate conductivity rose from 830 μS/cm to 900 μS/cm (under constant power

regime) (Figure 55). Comparatively in the past before the occurrence of the sudden deterioration,

permeate conductivity was as low as 630 μS/cm (for the same max feed flow and power).

TABLE 16: OBSERVATION OF OPERATIONAL PARAMETERS (QUALITY DETERIORATION)

Operational parameters Value

Membrane Pressure (P) 50 bar

Feed Flow (Qf) 6 gpm (23 l/min)

Permeate flow (Qp) 1.2 gpm (5 l/min)

Water recovery (γ) 20 % (fixed)

Feed electrical Conductivity (ECf) 53.2 mS/cm

Max system salt rejection

Range of permeate conductivity (Ecp)

98.4 %

830 – 900 μS/cm

66

FIGURE 55: PV-RO PARAMETERS AS A FUNCTION OF TIME (QUALITY DETERIORATION)

Since the two membrane elements are in series, they can be represented as shown in Figure 56:

FIGURE 56: LAYOUT OF THE RO MEMBRANES

67

Simultaneous permeate samples from both membranes showed that the element 2 had unjustified

high electrical conductivity (ECp1= 453 μS/cm while ECp2= 1736 μS/cm). The water recovery ratio of

the first element was measured during operation: γ1=Qp1/Qf1=11.6% while the salt rejection was

calculated to be:

p1

1

f1

mS0.453C cmr 1 1 99.1%

mSC53.2

cm

Since the two membranes are in series, the concentrate of the first element is the feed for the second

element. For the calculation of the salt rejection of the second element, the mass balance for salt for

the element 1 was considered:

1f f1 c1 c1 p1 p1Q C Q C Q C

Assuming that the salt concentration in the permeate stream is negligible, and thatp1 1 f1Q Q , it

applies for the salt concentration of the concentrate from element 1: f1c1

1

CC

1=60 mS/cm.

Thus, the salt rejection for the 2nd element was calculated to be:

p2

2

c1

mS1.736C cmr 1 1 97.1%

mSC60.1

cm

It is obvious that the salt rejection of the 2nd membrane has considerable decrease in the salt

rejection (manufacture value is 99.4%). Possibility of fouling was eliminated, because of constant

pressure during operation. It can however be speculated that there is a pin-hole leakage in the

membrane. This could be likely attributed to the frequent start and stops of the system during the

preceding days. Since the deterioration happened during the last days of the planned period, due to

time constraints and lack of training, the case of the membrane was not opened to visually inspect

possible damage. Furthermore, in case of damage, replacement with a new membrane and storing it

with preservative until next spring was not considered to be a wise course of action.

5.5. LIMITING FACTORS

Southwest orientation of the roof

The orientation of the roof being SSW and not S which is optimum for the operation of photovoltaics

in the Northern Hemisphere is responsible for some losses in power production. However, by

comparison of the power production with the optimum south orientation and the available orientation

(SSW, 30ο), it was shown that the losses can be estimated at 0.191 kWh/d or 2.7% of the maximum

available energy.

68

FIGURE 57: DIFFERENCE IN SOLAR IRRADIATION FOR SOUTH AND SOUTHWEST FACING PANELS

Manual control of the high pressure pump

It was decided to keep the system as simplified as possible, for practical reasons. For this reason, no

automation mechanisms or controllers were used on site during experimentation in Crete, in order to

control the speed of the two pumps. Instead, the motor speed of the Pearson pump was regulated

manually by a knob on the control panel of the unit. However, the losses in power utilization and the

resulting drop in total daily water production have to be accounted for since due to the large

fluctuations in the solar radiation in the winter season (due to presence of clouds), it was very

difficult to follow the solar pattern accurately and to manually achieve the maximum flow rates at

every instance.

5.6. DIFFERENT MODES OF OPERATION

5.6.1. SEAWATER/BRACKISH WATER AS FEEDSOURCE CONSIDERATIONS

Photovoltaic powered seawater desalination has found applications in remote communities with no

access to the grid and when the option of brackish water as a feed source is not available. Brackish water desalination is definitely a much more viable process due to the much lower operation

pressures. With some simple assumptions, an impression can be given about the difference between the two processes.

A typical seawater-RO system without energy recovery ER (γ=20%, P=53 bar), and a typical brackish water – RO system without ER (γ=80%, P=15 bar) are assumed.

The ideal hydraulic power to drive a pump is dependent on the pressure difference and the feed flow (Hydraulic Pump Power=membrane pressure ΔP x Qf i) while the water recovery is: γ=Qp/Qf. To

simplify the calculations, constant efficiency and permeate flow Qp is assumed. It can be shown that brackish water can be 14 times less energy intensive than seawater:

hydr,SW

hydr,BW

53 barP 20% 14

15 barP

80%

. This factor proves that energy recovery is critical for SW but not for BW. Thus,

the question arising is how much difference the ER makes to the power consumption in a SWRO

0

100

200

300

400

500

600

700

800

900

1000

5:16 7:40 10:04 12:28 14:52 17:16Time (hrs)

Sola

r Ir

rad

iati

on

(W

/m2)

South facing

South-West facing

69

system. A SWRO system with ER is considered (a Spectra system similar to the one used in

experimentation: 80% ER, 2 hp motor, 20% fixed water recovery, 7.9 m3/d product flow). Assuming a pump efficiency of 70%, operating pressure of 53 bar and knowing that Qf=Qp/γ, it applies for the

hydraulic pump power:

35

hydr f

mP Q P 7.9 5 53 10 Pa 2423W

d.

The shaft power - the power required transferred from the motor to the shaft of the pump - depends on the efficiency of the pump and can be calculated as:

hydr

shaft

P 2423WP 3461W

70%.

However, the Spectra system considered incorporates 80% ER and the maximum power which it can

be run is at 1200W (to pressurise the water to 53 bar). Based on this, only 1200/3461=35% of the

power required by the unit comes from the motor and the rest comes from the concentrate stream.

Consequently, in a SWRO system with ER like the one used in this project, the power consumed is (14 times x 35%)=4.9 times more than a typical BWRO system (γ=80%, P=15 bar) without energy

recovery.

5.6.2. ON-GRID/OFF-GRID CONSIDERATIONS

Even as the present study involves ―autonomous‖ system operation, it is interesting to compare its

on-grid and off-grid applications.18 The advent of feed-in tariffs in Europe (and in other places of the

world) creates profitable opportunities for projects involving solar PVs. The current feed-in rate in

Greece for solar PV plants generating less than 100 kWp is set at 0.441 €/kWh19. On the other hand,

the average cost of electricity in Greece is approximately 0.11 €/kWh20.

Consequently, it would be beneficial in case of an on-grid system to sell all the power produced by

the PV directly to the grid whenever possible, and operate the RO facility using the electricity from the grid during times of little or no irradiance. Due to the policy currently in place, this would in fact

lead to generation of revenue by selling electricity in addition to the revenue earned by the sale of

drinking water. It would also enable a system operation under a non-fluctuating constant load, which would lead to better quality of drinking water, as well as increasing the lifetime of the membranes.

Of course, such a system would only be feasible in case there was access to a grid which could

absorb the power generated by the PV, and not to remote coastal areas in general.

5.7. COMPARISON OF THEORETICAL AND PRACTICALLY OBTAINED WATER

YIELDS

The water cost estimation in the sizing approach, was based on the average water yield of one day in

the winter and one day in the summer, using irradiation values for the area of experimentation. The

results of the sizing approach showed that for 10 panels and an average annual water production of

781 m3 (1.16m3/d in the winter and 3.12 m3/d in the summer), the water cost is estimated to be

around 4.74 €/m3. The true validation of the yield can only be done over much longer experimental

18 An on-grid system may well be considered autonomous if it returns to the grid the same quantity of energy as it consumes. 19 http://www.solarfeedintariff.net/greece.html

20 http://www.energy.eu/

70

time scales. However, we can compare the practically obtained water yields during experimentation

with the theoretical expectations for the months of September and October calculated as described in

the sizing (Chapter 3) . The water costs are obtained by dividing the daily cost of water production by

the water yield on a specific day.

TABLE 17: COMPARISON OF THEORETICAL AND PRACTICAL WATER YIELDS

Theoretical

Maximum

(in m3/day)

Theoretical

Cost of Water

(€/m3)

Obtained Range

(in m3/day)

Expected Cost of

Water (€/m3)

September 2010 2.61 3.8 1.65 – 2.55 6.01 – 3.89

October 2010 2.35 4.22 0.33 – 2.15 30.05 – 4.61

As can be seen, the range of yields obtained from the RO facility is expectedly lower than the

theoretical maximum. The naturally relates to having an inverse effects on the costs of water

production. The main reasons for this differences are:

1. The theoretical calculations fail to take into account the presence of clouds, which mainly

result in large drops in yields. The maximum of the ranges typically correspond to cloudless

days, which expectedly come close to the theoretical maximum

2. Another important factor not considered in the theoretical calculations is the drop in efficiency

of the PV panels with decreasing irradiation values. This results in an over-estimation of the

energy yield in the theoretical calculations especially during dawn / dusk, which reflect in

apparently high water yields.

3. It is assumed for the theoretical calculations that the panels face the south, whereas in reality

they face about 30° west of south. However, this should not result in a major deviation

(about 2-3%) from the theoretical estimations.

71

6. CONCLUSIONS & RECOMMENDATIONS

The purpose of this project, was to examine the technical feasibility and economical aspects of an

autonomous batteryless seawater PVRO system. The conclusions of the study, along with

recommendations are as follows:

Worst case scenario testing

The field study of the batteryless PV-RO system lasted two months in total (03/09/2010-03/11/2010)

and gave crucial conclusions regarding its operation. It can be noted that most of the results obtained

represent the worst case scenario due to the unexpectedly winter-like conditions prevailing for most

of the time. However, the ―worst case‖ situation in Crete does not resemble the circumstances in

remote areas with high insolation and high water scarcity e.g. Middle East, Australia, North African

countries. Thus, it can be expected that experimental results in such areas would have been better

with respect to water quality and water production.

For this reason, it is recommended that an extended experimentation during summer period in

Greece would approach more the conditions in remote places where a seawater PVRO system could

find its application.

Average solar irradiation is not the deterministic factor for the system‘s output

It is important to mention that all results from the previous comparisons are indications and not

deterministic quantities. For wind powered RO systems, the average wind speed is considered to be a

useful criterion for comparison of the output over two different days of operation, or for judging

whether a particular location is suitable for setting up a system. It was believed that similar to this

quantity, the average solar irradiation could be used to make such determinations for a PV powered

RO system. However, it was observed that the average solar irradiation is not the critical factor

deciding the performance of a PV-RO system. This is because factors such as extent of cloudiness,

type of cloudiness, time of the day when the cloudiness occurs has a greater role to play in

determining the system output.

Following the conclusions, it was proved to be difficult to compare two different days by quantitive

means such as average solar irradiation alone or by qualitive means such as ―clear day‖ or ―cloudy

day‖. For a complete study, a developed statistical quantity could help in comparing irradiation values

for two different days depending on the system in consideration and its desired output (for e.g.

electrical energy output for a PV solar farm or desalinated water for a PV-RO system).

Preferred configuration of the system in a real scenario

The preferred embodiment of an autonomous seawater PVRO system for remote coastal areas should

be as less complex as possible. Since for this project it was decided to follow the last years‘ research

interest towards elimination of batteries due to the financial and often environmental costs involved

with their maintenance and replacement, this approach is for a batteryless system.A high pressure

pump with incorporated energy recovery, like the Spectra Pearson pump, is suggested for such a very

energy intensive system. The associated feed pump which is used to pressurise the inlet stream to

1.2 bar can be replaced by a reservoir at 12 m high above the high pressure pump. In such a

72

configuration, a second pump is avoided and no automation/controller is needed to synchronise the

speed of the two pumps.

The feedwater coming from the reservoir should be pre-treated via 20MF and 5 MF filters before

passing from the RO membranes. The water after passing through the RO membranes is separated

into a concentrate and a permeate stream. The concentrate stream which is under high pressure,

before being discharged, passes from the high pressure pump for energy recovery while the

permeate stream is stored separately. In a remote area where there is lack of fresh water, an

adequate amount of permeate should be kept for flushing the system thoroughly after a completed

cycle of operation.

In order to achieve maximum power harvest, the PV-array should be installed on a tracker with the

optimum inclination angle per season and facing the nearest pole. 2-axis tracker, according to

literature is not justified economically. Since it is optimal to have a larger PV array to get long enough

operational times (table 7), there is high chance that a large amount of energy can be wasted during

summer season. This energy may be utilised in pumping the seawater into the 12 meter height

reservoir.

Motor Speed Controller

The power utilization from the PV panels is not maximized in the results of this experimentation due

to the limiting factors described in the previous chapter (SSW roof orientation and lack of motor

speed controller). The absence of the controller was felt stronger during cloudy days with intense

solar fluctuations (especially during October). It is suggested that for periods with sustained cloud

cover, a motor speed controller is used which would optimize the motor operation at regular intervals

with respect to solar fluctuations.

Water quality

The results obtained show that the fluctuating operation of an RO system has an impact on the water

quality of the produced permeate since very few average readings of permeate quality on very cloudy

days showed values under 600 mg/l, a value generally suggested by the World Health Organization.

However, it should be noted that no health-based guideline is proposed by WHO or by the European

directive (98/83/EC); Based on consumer taste considerations, it is suggested that drinking water can

get unpalatable at TDS levels greater than about 1000 mg/l. Thus, it can be suggested that water

quality levels similar to the ones obtained during the experimentation can be acceptable for

communities in remote arid areas.

Using a lower recovery ratio than the 20% which was used in this system could probably lead to

improvement of the quality of the water produced for systems under intermittent operation.

LCB issues

The LCB allows extended operating times for the PV-RO system, since it prevents stalling of the

system at low power, and is especially valuable on cloudy days. These extended hours of operation

lead to greater water production, which will ultimately lead to a decrease in the cost of water

production.

Despite the intense solar fluctuations, the Specific Energy Consumption (SEC) ranged around 2.73 -

3.50 kWh/m3 which can be considered considerably low for renewable energy powered seawater

desalination systems. For comparison, typical SEC values found in literature for similar seawater PVRO

systems are about 4-5.5 kWh/m3 for systems with ER and 6.3-17.9 kWh/m3 for systems without ER

73

(Table 2, Figure 15). Nevertheless, it was not possible to correlate the effect of the use of the LCB on

the SEC, due to the different daily solar patterns.

Although conclusions about the water production cannot be drawn based on the comparisons, the

LCB does influence the water quality for the better, due to increased cross flow velocity resulting from

the higher flow.

Energy Buffer requirement

The quality of the water, in fact, suffers significantly due to the frequent starting and stopping of the

system on cloudy days. The authors from literature [45] state that the decision of whether to use a

system with or without batteries depends on which option is less troublesome and more cost-

effective: replacing reverse osmosis membranes or replacing batteries. However, this is under the

assumption that utilizing either of the above options makes no difference to the water quality.

However, the experiments carried out at Crete confirm that intermittent operation (with a batteryless

system) does not affect only the life span of the membranes, but also the quality of the permeate

water. A significant rise in TDS was observed due to frequent start-stops of the system, especially on

days with passing clouds.

Hence, in case a batteryless system is to be implemented, it is recommended that an energy buffer is

used, which can smoothen the power curve to an extent that the number of start-stops of the system

is reduced, which can result in a better water quality. Options which could potentially be investigated

in this course are super-capacitors or a small battery arrays.

Water cost

The water cost estimation in the sizing approach, was based on the average water output between

one day in the winter and one day in the summer, using irradiation values for the area of

experimentation. The results of the sizing approach showed that for 10 panels and an average annual

water production of 781 m3 (1.16m3/d in the winter and 3.12 m3/d in the summer), the averaged

annual theoretical water cost is estimated to be around 4.74 €/m3.

From the experimental results it was found that the water yield, which ultimately determines the price

of water depends largely on climatic conditions. So it is difficult to obtain an accurate estimate of the

water yield without experimenting over long time scales. It is important to note that the numbers for

the cost of water are derived by assuming that the capital costs are to be repaid on a daily basis using

the quantity of water produced on that particular day itself and this cannot give an accurate estimation

of the net cost of water.Nevertheless, an impression of the water cost derived from the experimental

results for the months of September and October are given in Table 17.

Although these prices are much higher than the current mains delivered water e.g. in Netherlands

(0.5 €), it is important to note for comparison that in July and August 2010, the cost for transporting

drinking water to 17 arid Greek islands reached 12.5 €/m3. Furthermore, if we consider the cost of

the bottled water which is 0.5€/L or 500€/m3, we can easily understand that seawater desalination is

an alternative solution in places where fresh water distribution is not an option.

In conclusion, it should be noted that the viability of RE-desalination systems is very site specific.

Based on this research, it can be suggested that PV-RO seawater desalination is promising mainly for

remote arid regions with high insolation (where water scarcity is covered with transportation of water

and there is no access to the grid) and with no potential for exploiting brackish water resources e.g.

in places around the MENA region, deserts and remote islands.

74

I

BIBLIOGRAPHY

1. Ibbotson, B., The Effects of Fluctuating Operation on Reverse Osmosis Membranes. 2010, Delft University of Technology & University of South Australia: Delft.

2. Mohamed, E.S., et al., A direct coupled photovoltaic seawater reverse osmosis desalination system toward battery based systems - a technical and economical experimental comparative study. Desalination, 2008. 221(1-3): p. 17-22.

3. Eltawil, M.A., Z. Zhengming, and L.Q. Yuan, A review of renewable energy technologies integrated with desalination systems. Renewable & Sustainable Energy Reviews, 2009. 13(9):

p. 2245-2262.

4. Schafer, A.I., A. Broeckmann, and B.S. Richards, Renewable energy powered membrane technology. 1. Development and characterization of a photovoltaic hybrid membrane system. Environmental Science & Technology, 2007. 41(3): p. 998-1003.

5. Vengosh, A., Salinization and Saline Environments. Treatise on Geochemistry, 2003. 9.

6. Insights from the Comprehensive Assessment of Water Management in Agriculture, in World Water Week, C. International Water Mangement Institute (IWMI), Sri Lanka, Editor. 2006:

Stockholm.

7. Al-Karaghouli, A., D. Renne, and L.L. Kazmerski, Solar and wind opportunities for water desalination in the Arab regions. Renewable & Sustainable Energy Reviews, 2009. 13(9): p.

2397-2407. 8. El-Nashar, A.M., D. Al Gobaisi, and B. Makkawi, Solar energy for desalination in the Arab

world. Proceedings of Ises Solar World Congress 2007: Solar Energy and Human Settlement,

Vols I-V, 2007: p. 213-221. 9. Kaldellis, J.K., K.A. Kavadias, and E. Kondili, Renewable energy desalination plants for the

Greek islands - technical and economic considerations. Desalination, 2004. 170(2): p. 187-203.

10. Mygiakis, M., Έναρ ζηοςρ έξι ανθπώποςρ δεν έσει ππόζβαζη ζε νεπό (One in six people lack access to water), in Οικο-ηύπορ. 2010, Eleftheros-Typos: Greece. p. 1-2.

11. Salinity Levels (Official Website of the City of Phoenix, Arizona, USA). [8 November 2010];

Available from: www.phoenix.gov/waterservices/wrc/yourwater/salinity/index.html. 12. Guidelines for Drinking-water Quality (Third Edition). 2008, World Health Organization:

Geneva. 13. Council Directive on the quality of water intended for human consumption, in 98/83/EC.

1998: European Union.

14. Richards, B.S. and A.I. Schafer, Photovoltaic-powered desalination system for remote Australian communities. Renewable Energy, 2003. 28(13): p. 2013-2022.

15. De Munari, A., et al., Application of solar-powered desalination in a remote town in South Australia. Desalination, 2009. 248(1-3): p. 72-82.

16. Ghermandi, A. and R. Messalem, Solar-driven desalination with reverse osmosis: the state of the art. Desalination and Water Treatment, 2009. 7(1-3): p. 285-296.

17. Sabine Lattemann, I.E., Andrea Schafer, Chapter 2 Global Desalination Situation. Vol. 2.

2010: Elsevier. 18. Total world desalination capacity close to 60 million m³/d. 2009 [30/05/2011]; Available

from: http://www.desalination.biz/news/news_story.asp?id=5121. 19. Water Pumps for the Desalination Market. 2009 [30/05/2011]; Available from:

http://news.lubipumps-na.com/blog/2009/01/water-pumps-for-the-desalination-market/.

20. Total desalination capacity by country. 2006 [30/05/2011]; Available from: http://water.nationalacademies.org/basics_part_3.shtml.

21. Papapetrou, M., C. Epp, and E. Tzen, Autonomous desalination units based on renewable energy systems - A review of representative installations worldwide. Solar Desalination for the

21st Century, 2007: p. 343-353.

22. Charcosset, C., A review of membrane processes and renewable energies for desalination. Desalination, 2009. 245(1-3): p. 214-231.

II

23. Herold, D. and A. Neskakis, A small PV-driven reverse osmosis desalination plant on the island of Gran Canaria. Desalination, 2001. 137(1-3): p. 285-292.

24. Banat, F. and H. Qiblawey, Membrane desalination driven by solar energy. Solar Desalination

for the 21st Century, 2007: p. 271-291. 25. Richards, B.S., D.P.S. Capao, and A.I. Schafer, Renewable energy powered membrane

technology. 2. The effect of energy fluctuations on performance of a photovoltaic hybrid membrane system. Environmental Science & Technology, 2008. 42(12): p. 4563-4569.

26. van Dijk, J.C., Verberk, J.Q.J.C., Heijman S.G.J., Rietveld L.C., Drinking water treatment. 2009: Sanitary Engineering Department, Civil Engineering and Geosciences, TUDelft.

27. Papapetrou, M., M. Wieghaus, and C. Biercamp, Roadmap for the development of Desalination powered by Renewable Energy, PRODES.

28. Thomson, M. and D. Infield, A photovoltaic-powered seawater reverse-osmosis system without batteries. Desalination, 2003. 153(1-3): p. 1-8.

29. Loster, M. Total Primary Energy Supply - From Sunlight. 2010 [21 November 2010]; Available from: http://www.ez2c.de/ml/solar_land_area/.

30. World Solar Insolation Map. Available from: http://www.oynot.com/solar-insolation-map.html. 31. ADIRA Handbook: A Guide to Autonomous Desalination System Concepts: MEDA Water.

32. Ahmad, G.E. and J. Schmid, Feasibility study of brackish water desalination in the Egyptian deserts and rural regions using PV systems. Energy Conversion and Management, 2002. 43(18): p. 2641-2649.

33. Garcia-Rodriguez, L., Renewable energy applications in desalination: state of the art. Solar Energy, 2003. 75(5): p. 381-393.

34. Garcia-Rodriguez, L., Seawater desalination driven by renewable energies: a review. Desalination, 2002. 143(2): p. 103-113.

35. Solar Cell I-V curve with MPPT. 2011; Available from: http://en.wikipedia.org/wiki/File:Solar-

Cell-IV-curve-with-MPP.png. 36. Australian Energy Research Laboratories. [22 November 2010]; Available from:

http://www.aerl.com.au/. 37. Solar Converters - Product Line. [22 November 2010]; Available from:

http://www.solarconverters.com/product_frame.html.

38. Richards, B.S. and A.I. Schafer, Design considerations for a solar-powered desalination system for remote communities in Australia. Desalination, 2002. 144(1-3): p. 193-199.

39. Gibson, T.L. and N.A. Kelly, Optimization of solar powered hydrogen production using photovoltaic electrolysis devices. International Journal of Hydrogen Energy, 2008. 33(21): p.

5931-5940.

40. Thomson, A.M., Reverse-Osmosis Desalination of Seawater Powered by Photovoltaics Without Batteries. 2003, Loughborough University.

41. Thomson, M., M. Miranda, J. Gwillim, A. Rowbottom, I. Draisey Batteryless Photovoltaic Reverse-Osmosis Desalination System, in DTI Sustainable Energy Programmes. 2001.

42. Cheah, S.-F., Photovoltaic Reverse Osmosis Desalination System. 2004, ITN Energy Systems, Inc.: Littleton, CO, USA.

43. Mohamed, E.S., et al., The effect of hydraulic energy recovery in a small sea water reverse osmosis desalination system; experimental and economical evaluation. Desalination, 2005. 184(1-3): p. 241-246.

44. Desalination Units powered by Renewable Energy Systems: Opportunities and Challenges. in International ADU-RES Seminar. 2005. Hammamet, Tunisia: WIP & INGREF.

45. Bermudez-Contreras, A., M. Thomson, and D.G. Infield, Renewable energy powered desalination in Baja California Sur, Mexico. Desalination, 2008. 220(1-3): p. 431-440.

46. Joyce, A., et al., Small reverse osmosis units using PV systems for water purification in rural places. Desalination, 2001. 137(1-3): p. 39-44.

47. Thomson, M. and D. Infield, Laboratory demonstration of a photovoltaic-powered seawater reverse-osmosis system without batteries. Desalination, 2005. 183(1-3): p. 105-111.

48. Spectra Watermakers - Land Based Applications. [22 November 2010]; Available from:

http://www.spectrawatermakers.com/landbased/about.php.

49. Mohamed, E.S. and G. Papadakis, Design, simulation and economic analysis of a stand-alone reverse osmosis desalination unit powered by wind turbines and photovoltaics. Desalination,

2004. 164(1): p. 87-97.

III

50. ERI. ERI Technology Overview. 2011; Available from:

http://www.energyrecovery.com/index.cfm/0/0/33-Overview.html. 51. Wood, R. (2009) Small-scale desalination pump combines energy-recovery. Desalination &

Water Reuse. 52. SpectraWatermakers, Pearson High Pressure Pump Application Guide. 2009.

53. Reverse Osmosis Desalination Membranes. [22 November 2010]; Available from:

http://www.lenntech.com/products/membrane/sea/general/reverse-osmosis-desalination-membranes.htm.

54. Spectra - SSW 5800. [22 November 2010]; Available from: http://www.spectrawatermakers.com/landbased/products_ssw5800.php.

55. Murakami, M., Hydropowered Reverse-Osmosis (Ro) Desalination for Cogeneration - a Middle-East Case-Study. Desalination, 1994. 97(1-3): p. 301-311.

56. Dako Power - Three Photovoltaic Technologies. 2008 [23 November 2010]; Available from:

http://www.dako.co.za/photovoltaic_types.html. 57. de Moel, P.J., Verberk, J.Q.J.C., van Dijk, J.C., Drinking Water: Principles and Practices. 2007:

World Scientific Publishing Co. Pte. Ltd. .

IV

V

APPENDICES

A) Datalogger calibrations

Channel in Datalogger

Parameter recorded Range Calibration relation Gain

1 Solar irradiation (W/m2) (0-

0,05V) 69.3 μV/W/m2 8

2 Motor Voltage (V) (0-10V) [output V] x 4 no

3 Motor Current (A) (0-0.3V) [output V] / 0.0045 8

4 Membrane Pressure (P) (2-10V) P = 7.5 x [output V] - 15 no 5 Permeate Flow (l/min) (0-10V) Qp=0.98344x [output V]+0.053749 no

6 Feedwater Conductivity

(mS/cm) (0-2V)

ECf=94.663 x [output V] 4

7 Permeate water

Conductivity (μS/cm) (0-2V)

Ecp= 1002.36 x [output V] 4

B) Irradiation data used for the sizing approach (acquired from JRC)

VI

C) Chosen sizing approach excel sheet

VII

VIII

D) Summarized tables of most representative results

Date: 16/09/2010 NO LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

max 854.31 29.24 54 1133.60 57.90 5.31 21.52 1153.30 828.33

min 158.53 27.00 14 973.57 39.24 2.14 8.69 10.07 353.33

average 657.06 28.58 35 1031.95 47.23 4.88 19.78 700.83 505.06

Daily water Production (m3)

2.24

defect Voltmeter. Manual log of some Voltage data, but not enough to calculate energy and SEC.

Total Energy consumption (kWh)

-

SEC (kWh/m3) -

Date: 17/09/2010 NO LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

Max 836.70 33.79 42 1085.07 49.16 5.17 20.95 1939.37 1392.91

Min 114.50 16.99 0 1.03 0.32 0.11 0.43 747.91 537.17

average 604.09 29.15 31 891.42 45.65 4.41 17.87 864.64 621.03

Daily water Production (m3)

2.55

Total Energy consumption (kWh)

8.58

SEC (kWh/m3) 3.37

Date: 22/09/2010 NO LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

Max 947 33.21 39.13 1060 49.43 5.12 20.75 1443.50 1036.77

Min 123 29 5.63 163.25 33.25 0.73 2.95 667.72 479.58

average 556 29.73 24.25 717 43.64 3.60 14.60 911.12 654.39

Daily water Production (m3)

1.65

Total Energy consumption (kWh)

5.48

SEC (kWh/m3) 3.32

Date: 30/09/2010 LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

Max 1033 30.33 41.59 1028 49.11 5.06 20.51 1814.90 1303.52

Min 70 8.64 0.03 1 0.33 0.11 0.43 765.69 549.94

average 491 24.26 27.98 667 43.30 3.54 14.37 970.49 697.03

Daily water Production (m3)

1.81

Total Energy consumption (kWh)

5.66

SEC (kWh/m3) 3.14

IX

Date: 2/10/2010 LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

Max 1026 30.30 40.73 987 49.50 5.06 20.51 1770.00 1271.27

Min 22 11.64 0.00 0 0.30 0.11 0.44 901.00 647.13

average 320 25.60 18.55 448 28.38 2.37 9.61 1220.43 876.55

Daily water Production (m3)

0.83

Total Energy consumption (kWh)

2.62

SEC (kWh/m3) 3.15

Date: 5/10/2010 LCB Parameters

Irradiatio

n (W/m2)

Voltag

e (V)

Curren

t (A)

Power

(W)

Pressur

e (bar)

Flow

(l/min)

Flux

(l/h*m2)

ECp

(μS/cm)

TDS

permeate (mg/l)

max 841.11 30.30 43

1034.72 49.09

4.98 20.17 1761.20 1264.95

min 55.05 15.61 0 9.10 0.33 0.11 0.43 3.66 2.63

average 473.92 25.17 28 673.79 41.46 3.49 14.14 963.10 691.73

Daily water Production (m3)

1.94

Total Energy consumption (kWh)

6.24

SEC (kWh/m3) 3.22

Date: 6/10/2010 LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

max 746 30.32 44 1052 49.55 5.04 20.45 1280.21 919

min 99 11.78 0 0 0.31 0.11 0.43 720.69 518

average 566 23.96 36 869 46.53 4.40 17.84 815.95 586

Daily water Production (m3)

2.08

Total Energy consumption (kWh)

6.84

SEC (kWh/m3) 3.29

Date: 8/10/2010 LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

max 1077 30.33 38 963 50.17 5.15 20.88 1732.50 1244

min 44 9.71 0 5 9.58 0.11 0.43 227.97 164

average 307 25.87 18 437 37.32 2.48 10.05 1000.49 719

Daily water Production (m3)

1.13

Total Energy consumption (kWh)

3.33

SEC (kWh/m3) 2.94

X

Date: 11/10/2010 LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

max 901 30.33 37.96 920 50.81 5.02 20.37 1452.30 1043

min 57 11.13 0.00 0 0.29 0.11 0.43 615.23 442

average 442 24.29 27.28 654 45.60 3.80 15.42 772.21 555

Daily water Production (m3)

1.80

Total Energy consumption (kWh)

5.17

SEC (kWh/m3) 2.87

Date: 12/10/2010 NO LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

max 857 33.61 38.31 982 51.08 5.25 21.27 1120.00 804.42

min 66 17.74 0.09 3 0.30 0.11 0.44 654.00 469.72

average 419 28.53 21.33 598 43.42 3.42 13.88 855.52 614.46

Daily water Production (m3)

1.85

Total Energy consumption (kWh)

5.38

SEC (kWh/m3) 2.91

Date: 13/10/2010 NO LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

max 859 33.56 36.12 971 49.81 5.02 20.35 1120.00 804.42

min 68 17.93 0.12 2 0.32 0.11 0.43 654.00 469.72

average 247 29.37 11.28 314 32.52 1.63 6.63 810.14 581.87

Daily water Production (m3)

0.33

Total Energy consumption (kWh)

1.07

SEC (kWh/m3) 3.21

Date: 15/10/2010 NO LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

max 918 36.12 33.26 883 51.15 5.14 20.83 1536.60 1103.63

min 53 18.52 0.15 5 0.34 0.11 0.44 630.49 452.84

average 335 30.74 11.17 328 34.77 2.00 8.12 1014.18 728.42

Daily water Production (m3)

0.88

Total Energy consumption (kWh)

2.41

SEC (kWh/m3) 2.73

XI

Date: 17/10/2010 LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

max 903 30.31 41.62 1008 50.67 5.02 20.37 1531.10 1099.68

min 110 0.00 0.00 0 0.30 0.11 0.44 528.56 379.63

average 573 23.43 30.92 727 44.12 4.01 16.25 1108.30 804.55

Daily water Production (m3)

1.35

Total Energy consumption (kWh)

4.07

SEC (kWh/m3) 3.03

Date: 19/10/2010 LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

max 764 29.07 44.73 1117 49.65 5.16 20.93 1944.00 1396

min 53 0.01 0.06 0 0.32 0.12 0.47 8.85 6

average 438 23.82 31.60 750 45.47 4.03 16.32 1160.59 834

Daily water Production (m3)

2.15

Total Energy consumption (kWh)

6.68

SEC (kWh/m3) 3.11

Date: 25/10/2010 INVERTER Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

max 746 31.84 31.35 991 50.50 5.23 21.18 936.89 672.90

min 7 31.56 23.81 758 43.07 4.44 18.01 795.29 571.20

average 312 31.61 26.99 853 49.73 5.02 20.35 890.35 639.47

Daily water Production (m3)

2.26

Total Energy consumption (kWh)

6.39

SEC (kWh/m3) 2.83

Date: 03/11/2010 LCB Parameters

Irradiation (W/m2)

Voltage (V)

Current (A)

Power (W)

Pressure (bar)

Flow (l/min)

Flux (l/h*m2)

ECp (μS/cm)

TDS permeate (mg/l)

max 656 30.34 50.61 1216 51.50 5.17 20.97 987.24 709.07

min 57 18.72 0.03 1 0.27 0.11 0.44 639.65 459.42

average 503 24.39 41.84 1018 49.54 4.85 19.67 867.73 623.23

Daily water Production (m3)

2.09

Total Energy consumption (kWh)

7.32

SEC (kWh/m3) 3.50

XII

E) PV panels characteristics

XIII

F) 12/24 V @ 40 Amps Linear Current Booster/Pump Driver

Model: PPT 12/24-40 from Solar Coverters .Inc

Features:

- Voltage Limited Output to 15 / 30 V

- Greatly increases water flow, even under reduced sunlight conditions (over 40% likely) - Transient Protected on the input and output

- Float/Dry Switch input for on/off control

- Pump Protection Features - Well Dry, Low Sun, Current Limited, Temperature Limited - LED Display of pump performance

- Interface for Optional Digital meter MT-3 - Ultra High Efficiency >94%

- Rainproof NEMA 3R enclosure

i The relation can be derived from the main formula for hydraulic pump power: http://www.engineeringtoolbox.com/pumps-power-d_505.html

Nominal Voltage (V) 12 24

Nominal PV Input (V) 12 24

Maximum PV Input (V) 50 50

Maximum Output Voltage (V)** 15 30

Maximum Output (A) 40 40

Maximum Output (A) short term

- 10 sec

60 60

Efficiency >95% >95%