30.12.2014 Views

An assessment of the Silt Density Index based on RO membrane ...

An assessment of the Silt Density Index based on RO membrane ...

An assessment of the Silt Density Index based on RO membrane ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

DESALINATION<br />

ELSEVIER Desalinati<strong>on</strong> 15 l(2002) 229-238<br />

www.elsevieccom/locate/desal<br />

<str<strong>on</strong>g>An</str<strong>on</strong>g> <str<strong>on</strong>g>assessment</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> <str<strong>on</strong>g>Silt</str<strong>on</strong>g> <str<strong>on</strong>g>Density</str<strong>on</strong>g> <str<strong>on</strong>g>Index</str<strong>on</strong>g> <str<strong>on</strong>g>based</str<strong>on</strong>g> <strong>on</strong> <strong>RO</strong> <strong>membrane</strong><br />

colloidal fouling experiments with ir<strong>on</strong> oxide particles<br />

Stergios G. Yiantsios, <str<strong>on</strong>g>An</str<strong>on</strong>g>astasios J. Karabela.s*<br />

Chemical Process Engineering Research Institute and Department <str<strong>on</strong>g>of</str<strong>on</strong>g> Chemical Engineering,<br />

Aristotle University <str<strong>on</strong>g>of</str<strong>on</strong>g> Thessal<strong>on</strong>iki, PO Box 361, 570 01, Thermi, Thessal<strong>on</strong>iki, Greece<br />

Tel. +30 (310) 996201; Fax +30 (310) 996209; email: karabaj@cperi.certh.gr<br />

Received 9 July 2002; accepted 3 1 July 2002<br />

Abstract<br />

<strong>RO</strong> <strong>membrane</strong> colloidal fouling experiments were performed in <str<strong>on</strong>g>the</str<strong>on</strong>g> laboratory under well c<strong>on</strong>trolled and realistic<br />

c<strong>on</strong>diti<strong>on</strong>s. Ir<strong>on</strong> oxide was selected as atypical inorganic colloidal foulant, due to its importance, as evidenced from well<br />

known manufkcturer recommendati<strong>on</strong>s <strong>on</strong> ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>s in feed waters and from frequently encountered problems<br />

in <strong>membrane</strong> installati<strong>on</strong>s. A range <str<strong>on</strong>g>of</str<strong>on</strong>g> ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>s was identified where a linearrelati<strong>on</strong>ship existed between flux<br />

reducti<strong>on</strong> rate and c<strong>on</strong>centrati<strong>on</strong>. The performance <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> <str<strong>on</strong>g>Silt</str<strong>on</strong>g> <str<strong>on</strong>g>Density</str<strong>on</strong>g> <str<strong>on</strong>g>Index</str<strong>on</strong>g> (SDI) was tested <strong>on</strong> <str<strong>on</strong>g>the</str<strong>on</strong>g> basis <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> <strong>RO</strong><br />

fouling data obtained. The range <str<strong>on</strong>g>of</str<strong>on</strong>g> ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>s where measurable and meaningful SD1 values could be obtained<br />

was remarkably close to <strong>membrane</strong> manufacturer recommendati<strong>on</strong>s. A notable sensitivity <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> SD1 was also observed<br />

with particles for which retenti<strong>on</strong> is negligible. However, <strong>on</strong> <str<strong>on</strong>g>the</str<strong>on</strong>g> basis <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> <strong>RO</strong> fouling data obtained, it appears that<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> SD1 is not c<strong>on</strong>servative enough. Fur<str<strong>on</strong>g>the</str<strong>on</strong>g>rmore, since <str<strong>on</strong>g>the</str<strong>on</strong>g> SD1 cannot predict fouling rates, it cannot discriminate<br />

between different types <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>membrane</strong>s.<br />

Kqwor&:<br />

<strong>RO</strong> <strong>membrane</strong> colloidal fouling; Ir<strong>on</strong> oxide colloidal particles; <str<strong>on</strong>g>Silt</str<strong>on</strong>g> <str<strong>on</strong>g>Density</str<strong>on</strong>g> <str<strong>on</strong>g>Index</str<strong>on</strong>g><br />

1. Introducti<strong>on</strong><br />

One <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> most serious c<strong>on</strong>cerns regarding<br />

desalinati<strong>on</strong> <strong>membrane</strong> performance is <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

presence <str<strong>on</strong>g>of</str<strong>on</strong>g> colloidal particles in <str<strong>on</strong>g>the</str<strong>on</strong>g> feed waters<br />

and <str<strong>on</strong>g>the</str<strong>on</strong>g> associated fouling problems. The most<br />

comm<strong>on</strong>ly occurring inorganic colloids in natural<br />

waters are aluminum silicate clays, ranging in<br />

*Corresp<strong>on</strong>ding author.<br />

size between 0.3 to 1 pm, and colloids <str<strong>on</strong>g>of</str<strong>on</strong>g> ir<strong>on</strong>,<br />

aluminum and silica. Organic substances dissolved<br />

in natural waters, such as humic and fulvic<br />

acids, may be also characterized to a certain<br />

degree as colloidal foulants since many features<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g>ir behavior fall in <str<strong>on</strong>g>the</str<strong>on</strong>g> domain <str<strong>on</strong>g>of</str<strong>on</strong>g> colloid<br />

science and are comm<strong>on</strong> with inorganic particles.<br />

Organic deposits, toge<str<strong>on</strong>g>the</str<strong>on</strong>g>r with ir<strong>on</strong> oxide, silica<br />

and aluminum oxide, account for more than 70%<br />

00 1 l-91 64/02/$- See fr<strong>on</strong>t matter 0 2002 Elsevier Science B.V. All rights reserved<br />

PII:SOOll-9164(02)01015-9


230 S. G. Yiantsios, A.J. Karabelas /Desalinati<strong>on</strong> 151 (2002) 229-238<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> deposits detected in <strong>membrane</strong> autopsies<br />

throughout <str<strong>on</strong>g>the</str<strong>on</strong>g> world [l]. The issues, <str<strong>on</strong>g>the</str<strong>on</strong>g>refore,<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> predicti<strong>on</strong> and mitigati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> colloidal and<br />

organic fouling remain critical after many years<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> successful applicati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>membrane</strong> technologies<br />

to a variety <str<strong>on</strong>g>of</str<strong>on</strong>g> water sources including<br />

well and surface, brackish and seawater, as well<br />

as wastewater <str<strong>on</strong>g>of</str<strong>on</strong>g> various origins.<br />

The most comm<strong>on</strong> and widely accepted tool<br />

for predicti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> colloidal fouling is <str<strong>on</strong>g>the</str<strong>on</strong>g> <str<strong>on</strong>g>Silt</str<strong>on</strong>g><br />

<str<strong>on</strong>g>Density</str<strong>on</strong>g> <str<strong>on</strong>g>Index</str<strong>on</strong>g> (SDI). O<str<strong>on</strong>g>the</str<strong>on</strong>g>r proposed indices are<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> Modified Fouling <str<strong>on</strong>g>Index</str<strong>on</strong>g> (MFI), which was<br />

developed in <str<strong>on</strong>g>the</str<strong>on</strong>g> Ne<str<strong>on</strong>g>the</str<strong>on</strong>g>rlands [2] and gained<br />

acceptance mainly <str<strong>on</strong>g>the</str<strong>on</strong>g>re, and <str<strong>on</strong>g>the</str<strong>on</strong>g> Mini Plugging<br />

Factor <str<strong>on</strong>g>Index</str<strong>on</strong>g> (MPFI ). All three are <str<strong>on</strong>g>based</str<strong>on</strong>g> <strong>on</strong><br />

batch filtrati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> feed waters through a 0.45 pm<br />

Millipore micr<str<strong>on</strong>g>of</str<strong>on</strong>g>ilter. More recently, <str<strong>on</strong>g>the</str<strong>on</strong>g> UF-<br />

MFI was introduced which uses ultrafiltrati<strong>on</strong><br />

<strong>membrane</strong>s instead <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> Millipore micr<str<strong>on</strong>g>of</str<strong>on</strong>g>ilter<br />

[3,4]. O<str<strong>on</strong>g>the</str<strong>on</strong>g>r measurements such as turbidity,<br />

particle counts, and particle electrophoretic<br />

mobility are also used but are not c<strong>on</strong>sidered<br />

reliable criteria for <str<strong>on</strong>g>assessment</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> fouling trends.<br />

Membrane manufacturers recommend that SD1<br />

should not exceed 4 or 5 and set limits <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>membrane</strong><br />

productivity depending <strong>on</strong> <str<strong>on</strong>g>the</str<strong>on</strong>g> SD1 [5] (i.e.,<br />

8-14 gfd for SD1 2-4, 14-18 gfd for SD1 ~2 and<br />

20-30 gfd for SD1 cl). It should be emphasized<br />

that <str<strong>on</strong>g>the</str<strong>on</strong>g> SD1 test is, strictly speaking, an empirical<br />

<strong>on</strong>e and a very poor simulati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> actual <strong>RO</strong><br />

c<strong>on</strong>diti<strong>on</strong>s. The absence <str<strong>on</strong>g>of</str<strong>on</strong>g> actual <strong>membrane</strong>foulant<br />

interacti<strong>on</strong>s, <str<strong>on</strong>g>the</str<strong>on</strong>g> absence <str<strong>on</strong>g>of</str<strong>on</strong>g> shear (as<br />

opposed to <str<strong>on</strong>g>the</str<strong>on</strong>g> cross-flow operati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>membrane</strong><br />

modules), <str<strong>on</strong>g>the</str<strong>on</strong>g> significantly higher permeati<strong>on</strong><br />

rate (>l cm/s in <str<strong>on</strong>g>the</str<strong>on</strong>g> filter vs. c.001 cm/s in <strong>RO</strong><br />

<strong>membrane</strong>s), and <str<strong>on</strong>g>the</str<strong>on</strong>g> doubtful rejecti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

particles smaller than 0.45 pm (nominal pore size<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> filter) are <str<strong>on</strong>g>the</str<strong>on</strong>g> most important c<strong>on</strong>cerns.<br />

These poorly rejected particles are more likely to<br />

cause severe fouling to a <strong>membrane</strong> since <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

resistance <str<strong>on</strong>g>of</str<strong>on</strong>g> a cake layer is inversely proporti<strong>on</strong>al<br />

to <str<strong>on</strong>g>the</str<strong>on</strong>g> square <str<strong>on</strong>g>of</str<strong>on</strong>g> particle size. Schippers et<br />

al. [6] state that particles smaller than 0.05 pm<br />

are resp<strong>on</strong>sible for flux decline in <strong>RO</strong> mem-<br />

branes. <str<strong>on</strong>g>An</str<strong>on</strong>g>o<str<strong>on</strong>g>the</str<strong>on</strong>g>r criticism is that <str<strong>on</strong>g>the</str<strong>on</strong>g> index is not<br />

linearly related to colloid particle c<strong>on</strong>centrati<strong>on</strong><br />

and does not distinguish between fouling<br />

mechanisms during <str<strong>on</strong>g>the</str<strong>on</strong>g> test (pore blocking, cake<br />

filtrati<strong>on</strong>, cake c<strong>on</strong>solidati<strong>on</strong>).<br />

Despite <str<strong>on</strong>g>the</str<strong>on</strong>g> significance <str<strong>on</strong>g>of</str<strong>on</strong>g> colloidal fouling<br />

problems in desalinati<strong>on</strong>, <str<strong>on</strong>g>the</str<strong>on</strong>g>re are relatively few<br />

experimental studies at laboratory scale that<br />

represent typical water sources, realistic foulants<br />

and well c<strong>on</strong>trolled c<strong>on</strong>diti<strong>on</strong>s at <str<strong>on</strong>g>the</str<strong>on</strong>g> same time.<br />

Most studies use syn<str<strong>on</strong>g>the</str<strong>on</strong>g>tic waters <str<strong>on</strong>g>of</str<strong>on</strong>g> simple<br />

compositi<strong>on</strong> (e.g., distilled water to which NaCl<br />

is added) and model colloidal particles, which<br />

cannot be c<strong>on</strong>sidered typical <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> foulants<br />

encountered in practice [7-91. A sec<strong>on</strong>d related<br />

problem is <str<strong>on</strong>g>the</str<strong>on</strong>g> durati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> laboratory<br />

experimental studies. First <str<strong>on</strong>g>of</str<strong>on</strong>g> all, l<strong>on</strong>g times are<br />

required in order to obtain a stable <strong>membrane</strong><br />

flux under n<strong>on</strong>-fouling c<strong>on</strong>diti<strong>on</strong>s due to phenomena<br />

referred to as “<strong>membrane</strong> setting”. In<br />

additi<strong>on</strong>, due to <str<strong>on</strong>g>the</str<strong>on</strong>g>ir high resistance to water<br />

flow, <strong>RO</strong> <strong>membrane</strong>s are not very sensitive to<br />

colloidal fouling, in <str<strong>on</strong>g>the</str<strong>on</strong>g> sense that relatively thick<br />

deposits <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> typically employed particles<br />

(which usually exceed 100 nm in diameter) need<br />

to be developed in order to observe a detectable<br />

flux reducti<strong>on</strong>. This necessitates l<strong>on</strong>g experimental<br />

times and accelerated fouling c<strong>on</strong>diti<strong>on</strong>s<br />

with unrealistically high particle c<strong>on</strong>centrati<strong>on</strong>s<br />

(200 ppm or more). It may also be added that in<br />

a typically closed-loop laboratory system it is<br />

difficult to maintain c<strong>on</strong>stant c<strong>on</strong>diti<strong>on</strong>s due to<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> limited stock <str<strong>on</strong>g>of</str<strong>on</strong>g> fouling particles and to<br />

possible aggregati<strong>on</strong> phenomena.<br />

The present experimental efforts include two<br />

steps: first, to generate reliable <strong>RO</strong> fouling data<br />

under realistic, yet, well c<strong>on</strong>trolled c<strong>on</strong>diti<strong>on</strong>s;<br />

sec<strong>on</strong>d, <strong>on</strong> <str<strong>on</strong>g>the</str<strong>on</strong>g> basis <str<strong>on</strong>g>of</str<strong>on</strong>g> such results, to assess <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

performance <str<strong>on</strong>g>of</str<strong>on</strong>g> existing techniques for colloidal<br />

fouling predicti<strong>on</strong> (i.e., SDI, MFI). Ir<strong>on</strong> oxide<br />

was selected as a typical inorganic colloidal<br />

foulant due to its importance as evidenced by <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

frequent appearance in <strong>membrane</strong> autopsies [ 11,<br />

and <str<strong>on</strong>g>the</str<strong>on</strong>g> specific reference in manufacturers’


S. G. Yiantsios, A.J. Karabelas / Desalinati<strong>on</strong> I51 (2002) 229-238 231<br />

recommendati<strong>on</strong>s [ 10,111. Fouling data are obtained<br />

by c<strong>on</strong>tinuous introducti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> ferric i<strong>on</strong>s<br />

into <str<strong>on</strong>g>the</str<strong>on</strong>g> feed water and sp<strong>on</strong>taneous precipitati<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> colloidal ir<strong>on</strong> hydrous oxide. In <str<strong>on</strong>g>the</str<strong>on</strong>g> following<br />

secti<strong>on</strong> <str<strong>on</strong>g>the</str<strong>on</strong>g> experimental materials and methods<br />

are described. Then <str<strong>on</strong>g>the</str<strong>on</strong>g> <strong>RO</strong> fouling data obtained<br />

are presented and discussed. Finally, SD1<br />

measurements with <str<strong>on</strong>g>the</str<strong>on</strong>g> same feed water and<br />

various ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>s are presented and an<br />

<str<strong>on</strong>g>assessment</str<strong>on</strong>g> is made <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> performance <str<strong>on</strong>g>of</str<strong>on</strong>g> SD1 in<br />

light <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> fouling data obtained.<br />

2. Experimental materials and methods<br />

<strong>RO</strong> fouling experiments are performed in a<br />

laboratory unit with a narrow-channel type <str<strong>on</strong>g>of</str<strong>on</strong>g> test<br />

secti<strong>on</strong> using flat-sheet <strong>membrane</strong> pieces with an<br />

active filtrati<strong>on</strong> area <str<strong>on</strong>g>of</str<strong>on</strong>g> 130 cm*. A schematic<br />

diagram <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> experimental set-up is shown in<br />

Fig. 1. A 30-L vessel c<strong>on</strong>tains <str<strong>on</strong>g>the</str<strong>on</strong>g> feed soluti<strong>on</strong><br />

which is pressurized via a Grundfos CRNE 2<br />

pump, capable <str<strong>on</strong>g>of</str<strong>on</strong>g> reaching pressures up to<br />

300 psi. Pressure and cross-flow rate are<br />

m<strong>on</strong>itored through digital sensors and c<strong>on</strong>trolled<br />

through two needle valves at <str<strong>on</strong>g>the</str<strong>on</strong>g> entrance and<br />

exit <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> test secti<strong>on</strong>. Permeate flow is<br />

m<strong>on</strong>itored through a Hum<strong>on</strong>ics 1000 digital<br />

flowmeter c<strong>on</strong>nected to a PC for automatic data<br />

acquisiti<strong>on</strong>. Temperature is c<strong>on</strong>trolled in <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

system through a cooling coil c<strong>on</strong>nected to a<br />

water cooler. A c<strong>on</strong>troller and a flow c<strong>on</strong>trol<br />

valve maintain c<strong>on</strong>stant temperature in <str<strong>on</strong>g>the</str<strong>on</strong>g> feed<br />

vessel <str<strong>on</strong>g>of</str<strong>on</strong>g> 25rtO. l°C. The system can run virtually<br />

unattended for l<strong>on</strong>g periods <str<strong>on</strong>g>of</str<strong>on</strong>g> time. Experimental<br />

runs last typically for about 1 week.<br />

Initially, efforts were made to obtain fouling<br />

data similar to those presented by Cohen and<br />

Probstein [9] by employing m<strong>on</strong>odisperse ir<strong>on</strong><br />

oxide colloidal particles, 150 nm in diameter,<br />

prepared according to <str<strong>on</strong>g>the</str<strong>on</strong>g> procedures described in<br />

Scheiner and Matijevic [12]. However, several<br />

problems related to <str<strong>on</strong>g>the</str<strong>on</strong>g> large particle size and<br />

aggregati<strong>on</strong> effects as well as difficulties to<br />

maintain c<strong>on</strong>stant particle c<strong>on</strong>centrati<strong>on</strong> in <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

closed loop system were encountered. Thus, as in<br />

Cohen and Probstein, results could be obtained<br />

<strong>on</strong>ly with almost desalted water and insignificant<br />

fouling rates were observed. It is noted<br />

paren<str<strong>on</strong>g>the</str<strong>on</strong>g>tically that, in view <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> above, <str<strong>on</strong>g>the</str<strong>on</strong>g>ir<br />

“critical flux” observati<strong>on</strong>s must be viewed with<br />

skepticism. In <str<strong>on</strong>g>the</str<strong>on</strong>g> experiments presented herein<br />

a different route was followed. <str<strong>on</strong>g>An</str<strong>on</strong>g> open loop<br />

operati<strong>on</strong> was established by c<strong>on</strong>tinuous<br />

introducti<strong>on</strong> and withdrawal <str<strong>on</strong>g>of</str<strong>on</strong>g> water and foulant<br />

in <str<strong>on</strong>g>the</str<strong>on</strong>g> system. Instead <str<strong>on</strong>g>of</str<strong>on</strong>g> introducing particles<br />

A<br />

FeC$ tank Acid tank<br />

nrl<br />

Sand<br />

Filter<br />

t<br />

Cartridge<br />

Filter<br />

Fig. 1. Schematic diagram <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> experimental set-up used in <str<strong>on</strong>g>the</str<strong>on</strong>g> <strong>RO</strong> fouling experiments.<br />

Drain<br />

Reject


232 S.G. Yiantsios. A.J. Karabelas / Desalinati<strong>on</strong> 151 (2002) 229-238<br />

previously formed, a FeCl, soluti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> appropriate<br />

c<strong>on</strong>centrati<strong>on</strong> was injected into <str<strong>on</strong>g>the</str<strong>on</strong>g> feed<br />

stream at a rate <str<strong>on</strong>g>of</str<strong>on</strong>g> 10 mL/min and pH was<br />

adjusted to 7. The feed stream flow rate was<br />

1.8 Wmin. At a pH close to neutral, ir<strong>on</strong> oxide<br />

particles are sp<strong>on</strong>taneously formed and c<strong>on</strong>stant<br />

fouling c<strong>on</strong>diti<strong>on</strong>s are established regardless <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

colloid aggregati<strong>on</strong> phenomena. <str<strong>on</strong>g>An</str<strong>on</strong>g> important<br />

c<strong>on</strong>siderati<strong>on</strong> is to maintain ir<strong>on</strong> in <str<strong>on</strong>g>the</str<strong>on</strong>g> stock<br />

soluti<strong>on</strong> in dissolved form. According to Stumm<br />

and Morgan [ 131, <str<strong>on</strong>g>the</str<strong>on</strong>g> following reacti<strong>on</strong>s take<br />

place:<br />

Fe3’ + H 2<br />

0 f) FeOH’+ + H’ (14<br />

Fe3’+ 2H,O c) Fe(OH): + 2H’<br />

2Fe3’ + 2H,O f) Fe,(OH),4’ + 2H’<br />

Fe3++ 3H,O f) Fe(OH),(aq) + 3H’<br />

Fe3’ + 4H,O f) Fe(OH)i + 4H’<br />

3Fe3’+ 2H20 t, Fe,(OH),4’ + 2H’<br />

3Fe3’+ 4H,O f) Fe,(OH),S’ + 4H’<br />

Fe(OH),(s)+ 3H’ f) Fe3++ 3H,O<br />

(lb)<br />

(lc)<br />

(ld)<br />

(Ie)<br />

(lf)<br />

(lg)<br />

(lh)<br />

Using <str<strong>on</strong>g>the</str<strong>on</strong>g> equilibrium c<strong>on</strong>stants listed in [ 121,<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> pH necessary to avoid Fe(OH), precipitati<strong>on</strong><br />

in <str<strong>on</strong>g>the</str<strong>on</strong>g> FeCl, stock soluti<strong>on</strong> can be calculated. In<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> present experiments <str<strong>on</strong>g>the</str<strong>on</strong>g> pH <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> stock<br />

soluti<strong>on</strong> was maintained close to 2 by adding<br />

appropriate amounts <str<strong>on</strong>g>of</str<strong>on</strong>g> HCl.<br />

It is frequently menti<strong>on</strong>ed in <str<strong>on</strong>g>the</str<strong>on</strong>g> literature that<br />

a certain period <str<strong>on</strong>g>of</str<strong>on</strong>g> time is required for <strong>membrane</strong><br />

setting, which refers to <strong>membrane</strong> adjustment to<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> pressure applied and to <str<strong>on</strong>g>the</str<strong>on</strong>g> i<strong>on</strong>s present in <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

feed water. During this period, flux drops and<br />

rejecti<strong>on</strong> increases. <str<strong>on</strong>g>An</str<strong>on</strong>g> additi<strong>on</strong>al factor is <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

influence <str<strong>on</strong>g>of</str<strong>on</strong>g> preservatives to which <str<strong>on</strong>g>the</str<strong>on</strong>g> <strong>membrane</strong><br />

is soaked for storage. These may cause flux<br />

changes as <str<strong>on</strong>g>the</str<strong>on</strong>g>y leach, and <str<strong>on</strong>g>the</str<strong>on</strong>g> problem is more<br />

Table 1<br />

<str<strong>on</strong>g>An</str<strong>on</strong>g>alysis <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> feed water used in <str<strong>on</strong>g>the</str<strong>on</strong>g> <strong>RO</strong> fouling<br />

experiments<br />

Physicochemical<br />

PH<br />

C<strong>on</strong>ductivity, pS/cm<br />

P-alkalinity<br />

Total hardness, “F<br />

Carb<strong>on</strong>ate, “F<br />

M-alkalinity<br />

parameters<br />

Chemical parameters, mg/L<br />

Cati<strong>on</strong>s:<br />

Na’<br />

K’<br />

Ca*+<br />

MgZ<br />

Fe*+<br />

Zn’+<br />

Mn2+<br />

S13<br />

<str<strong>on</strong>g>An</str<strong>on</strong>g>i<strong>on</strong>s:<br />

Cl-<br />

HCO;<br />

so:-<br />

NO;<br />

NO;<br />

PO:-<br />

F-<br />

Neutral:<br />

co2<br />

SiO,<br />

NaOCl<br />

7.6<br />

680<br />

0<br />

22.8<br />

21.5<br />

4.3<br />

50.3<br />

1.2<br />

47.5<br />

26.5<br />

0.03<br />

0.19<br />

0.01<br />

0.47<br />

55.1<br />

262.3<br />

39.8<br />

7.5<br />

0<br />

0.01<br />

0.17<br />

15<br />

21.1<br />

0.1<br />

likely to occur in <str<strong>on</strong>g>the</str<strong>on</strong>g> typically closed laboratory<br />

systems. In <str<strong>on</strong>g>the</str<strong>on</strong>g> present experiments, all <strong>membrane</strong>s<br />

were soaked in distilled water before use<br />

and permeate and retentate were discarded.<br />

During <str<strong>on</strong>g>the</str<strong>on</strong>g> initial operating periods <str<strong>on</strong>g>of</str<strong>on</strong>g> several<br />

hours, flux changes were very severe and much<br />

larger than any anticipated effect <str<strong>on</strong>g>of</str<strong>on</strong>g> colloidal<br />

fouling. This behavior was found to be largely<br />

independent <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> type <str<strong>on</strong>g>of</str<strong>on</strong>g> feed water used (tap<br />

water, sand filtered tap water, de-i<strong>on</strong>ized water,<br />

Nanopure water, <strong>RO</strong> water). Typically, 48 h or


S.G. Yiantsios, A.J. Karabelas / Desalinati<strong>on</strong> 151 (2002) 229-238 233<br />

more were dedicated to <strong>membrane</strong> setting.<br />

Although flux drop was not completely eliminated,<br />

it was reduced to about 0.2% per hour or<br />

less, which is acceptable provided that higher<br />

fouling rates can be manifested after <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

introducti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> fouling species.<br />

The type <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>membrane</strong> employed was a flatsheet<br />

brackish-water <strong>RO</strong> <strong>membrane</strong> (Osm<strong>on</strong>ics<br />

AG). In all <str<strong>on</strong>g>the</str<strong>on</strong>g> experimental runs described in <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

next secti<strong>on</strong>, a c<strong>on</strong>stant cross-flow rate <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

0.5 L/min was maintained in <str<strong>on</strong>g>the</str<strong>on</strong>g> channel, which<br />

corresp<strong>on</strong>ds to an average velocity <str<strong>on</strong>g>of</str<strong>on</strong>g> 17 cm/s.<br />

This is in <str<strong>on</strong>g>the</str<strong>on</strong>g> range <str<strong>on</strong>g>of</str<strong>on</strong>g> typical cross-flow<br />

velocities encountered in spiral-wound <strong>RO</strong> <strong>membrane</strong><br />

elements. The feed water flow rate in <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

feed vessel was 1.8 L/min, and thus <str<strong>on</strong>g>the</str<strong>on</strong>g> mean<br />

residence time <str<strong>on</strong>g>of</str<strong>on</strong>g> water and fouling species in <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

system was about 15 min. Feed water passed<br />

through a sand filter and a cartridge filter to reach<br />

an SDI close to 2 and a turbidity typically less<br />

than 0.1 NTU, whereas raw water has an SD1<br />

close to 4 and a turbidity close to 1 NTU. <str<strong>on</strong>g>An</str<strong>on</strong>g><br />

analysis <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> feed water is shown in Table 1.<br />

2<br />

E<br />

3<br />

5<br />

a<br />

..j<br />

FeCl,addlthm : )<br />

0 5 10 15 20 25 30 35<br />

t Pf<br />

10 I I , I<br />

j<br />

3. Results and discussi<strong>on</strong><br />

3.1. <strong>RO</strong> fouling measurements<br />

Typical raw data obtained in an experimental<br />

sequence, which included pH adjustment, <strong>membrane</strong><br />

setting and fouling species introducti<strong>on</strong> are<br />

shown in Fig. 2.<br />

Membrane flux prior to fouling species<br />

introducti<strong>on</strong> was kept close to 8~10~~ cm/s (or<br />

28 LMH or 17 gfd), which is representative <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

<strong>membrane</strong> fluxes typically encountered in <strong>RO</strong><br />

applicati<strong>on</strong>s. As can be observed, a clearly<br />

detectable decline <str<strong>on</strong>g>of</str<strong>on</strong>g> permeati<strong>on</strong> rate, linear in<br />

time, is obtained in <str<strong>on</strong>g>the</str<strong>on</strong>g> ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong> range<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a few ppm. On <str<strong>on</strong>g>the</str<strong>on</strong>g> basis <str<strong>on</strong>g>of</str<strong>on</strong>g> such raw data,<br />

normalized flux decline data can be obtained<br />

prior to and after fouling species introducti<strong>on</strong>, as<br />

shown in Fig. 3. The difference in flux decline is<br />

attributed to <str<strong>on</strong>g>the</str<strong>on</strong>g> fouling effects <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> ir<strong>on</strong> oxide<br />

particles.<br />

4<br />

I I I<br />

0 10 20 30 40 50 60<br />

Fig. 2. Raw experimental data <str<strong>on</strong>g>of</str<strong>on</strong>g> permeate flow rate as a<br />

functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> time before and after fouling species introducti<strong>on</strong>.<br />

(a) Ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>: 5 ppm. (b) Ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>:<br />

3.5 ppm.<br />

Normalized flux reducti<strong>on</strong> data (in percentage<br />

reducti<strong>on</strong> per hour) vs. ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong> in <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

feed water are shown in Fig. 4. These data are<br />

c<strong>on</strong>sidered representative <str<strong>on</strong>g>of</str<strong>on</strong>g> fouling rate variati<strong>on</strong><br />

with ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>. As can be observed,<br />

a linear relati<strong>on</strong>ship between ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong><br />

and fouling rate is obtained with a satisfactory<br />

degree <str<strong>on</strong>g>of</str<strong>on</strong>g> c<strong>on</strong>fidence. Experiments performed at<br />

higher ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>s (10 and 20 ppm)<br />

showed fouling rates similar to those obtained at<br />

t W


234 S. G. Yiantsios. A.J. Karabelas /Desalinati<strong>on</strong> 151 (2002) 229-238<br />

-y = lOO- 0.56x R= 0,79<br />

y-100-3,84x R=O,99<br />

+ 0,61x R= 0.98<br />

1<br />

I<br />

,_,’<br />

_,’<br />

_,_’<br />

,.<br />

_,.’<br />

_,”<br />

,/<br />

,I<br />

70<br />

0 2 4 6 8 10 12 14<br />

105<br />

t(h)<br />

-y=lOO-0,22x<br />

R=O.89<br />

y= 100-2.02x R=0,99<br />

Fig. 3. Normalized <strong>membrane</strong> flux data as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

time before and after fouling species introducti<strong>on</strong>.<br />

(a) Ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>: 5 ppm. (b) Ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>:<br />

3.5 ppm.<br />

5 ppm, and thus <str<strong>on</strong>g>the</str<strong>on</strong>g> c<strong>on</strong>centrati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> 5 ppm is<br />

c<strong>on</strong>sidered an upper limit for linearity to hold<br />

under <str<strong>on</strong>g>the</str<strong>on</strong>g> present experimental c<strong>on</strong>diti<strong>on</strong>s (i.e.,<br />

water pH, i<strong>on</strong>ic strength and residence time in <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

system).<br />

It is clear that <str<strong>on</strong>g>the</str<strong>on</strong>g> present experimental results<br />

represent accelerated fouling c<strong>on</strong>diti<strong>on</strong>s since<br />

ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>s are far above <str<strong>on</strong>g>the</str<strong>on</strong>g> recommended<br />

values for <strong>membrane</strong> feed waters. Such<br />

[Fe”1(ppm)<br />

3 4<br />

Fig. 4. Normalized <strong>membrane</strong> flux reducti<strong>on</strong> rate as a<br />

functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong> in <str<strong>on</strong>g>the</str<strong>on</strong>g> feed water.<br />

accelerated c<strong>on</strong>diti<strong>on</strong>s are a necessity in laboratory<br />

investigati<strong>on</strong>s due to time and resource<br />

limitati<strong>on</strong>s. However, it is proposed that <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

linear relati<strong>on</strong>ship found between fouling rate<br />

and ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong> can be extrapolated to<br />

lower and more realistic c<strong>on</strong>centrati<strong>on</strong>s, and here<br />

lies <str<strong>on</strong>g>the</str<strong>on</strong>g> usefulness <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> data obtained. For this<br />

to be true two c<strong>on</strong>diti<strong>on</strong>s need to be met. First,<br />

ir<strong>on</strong> precipitati<strong>on</strong> at <str<strong>on</strong>g>the</str<strong>on</strong>g> chosen c<strong>on</strong>diti<strong>on</strong>s (i.e.,<br />

pH) must be complete down to lower c<strong>on</strong>centrati<strong>on</strong>s,<br />

and precipitati<strong>on</strong> kinetics must be much<br />

faster than <str<strong>on</strong>g>the</str<strong>on</strong>g> macroscopic characteristic times<br />

in <str<strong>on</strong>g>the</str<strong>on</strong>g> system, i.e., ir<strong>on</strong> residence time in <str<strong>on</strong>g>the</str<strong>on</strong>g> test<br />

loop. Sec<strong>on</strong>d, aggregati<strong>on</strong> phenomena <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

precipitated crystallites, due to Brownian moti<strong>on</strong><br />

or shear, must have a small effect <strong>on</strong> fouling,<br />

namely <strong>on</strong> depositi<strong>on</strong> rates <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> particles <strong>on</strong> <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

<strong>membrane</strong> and <strong>on</strong> deposit structure.<br />

Regarding precipitati<strong>on</strong> <str<strong>on</strong>g>the</str<strong>on</strong>g>rmodynamics,<br />

according to <str<strong>on</strong>g>the</str<strong>on</strong>g> reacti<strong>on</strong>s given in <str<strong>on</strong>g>the</str<strong>on</strong>g> previous<br />

secti<strong>on</strong> and <str<strong>on</strong>g>the</str<strong>on</strong>g> respective equilibrium c<strong>on</strong>stants<br />

[13], <str<strong>on</strong>g>the</str<strong>on</strong>g> solubility <str<strong>on</strong>g>of</str<strong>on</strong>g> ferric i<strong>on</strong>s at pH 7 is<br />

estimated to be 5.9~10-‘~ moles/L. Therefore,<br />

even at a total c<strong>on</strong>centrati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> 1 ppb, almost all<br />

ir<strong>on</strong> will be in precipitated form. Regarding<br />

kinetics, it may be recalled that <str<strong>on</strong>g>the</str<strong>on</strong>g> limiting step<br />

for ir<strong>on</strong> oxide particle formati<strong>on</strong> in <strong>membrane</strong><br />

5


S.G. Yiantsios, A.J. Karabelas / Desalinati<strong>on</strong> 151 (2002) 229-238 235<br />

feed waters is c<strong>on</strong>sidered to be <str<strong>on</strong>g>the</str<strong>on</strong>g> oxidati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> soluble ferrous i<strong>on</strong> to ferric. Thus, precipitati<strong>on</strong><br />

kinetics is assumed here to be fast relative<br />

to <str<strong>on</strong>g>the</str<strong>on</strong>g> ir<strong>on</strong> residence times in <str<strong>on</strong>g>the</str<strong>on</strong>g> present<br />

experiments.<br />

Regarding <str<strong>on</strong>g>the</str<strong>on</strong>g> effects <str<strong>on</strong>g>of</str<strong>on</strong>g> aggregati<strong>on</strong> <strong>on</strong><br />

depositi<strong>on</strong> rates and deposit structure, an indirect<br />

piece <str<strong>on</strong>g>of</str<strong>on</strong>g> evidence is presented below. C<strong>on</strong>sidering<br />

a cake filtrati<strong>on</strong> mechanism, an estimate <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> characteristic particle size, a, may be<br />

obtained as follows. Membrane flux as a functi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> time is given by<br />

J- -- A’<br />

where AP is <str<strong>on</strong>g>the</str<strong>on</strong>g> applied pressure, R, <str<strong>on</strong>g>the</str<strong>on</strong>g> <strong>membrane</strong><br />

resistance and R, <str<strong>on</strong>g>the</str<strong>on</strong>g> fouling cake resistance.<br />

The latter is given by<br />

R, = R’, H (W<br />

where<br />

R’ = 2 (pap<br />

c<br />

and<br />

2 a2(l-cp,Y<br />

t<br />

H = s<br />

J(p,lcp,dt<br />

0<br />

(3b)<br />

(3c)<br />

Here, R, is <str<strong>on</strong>g>the</str<strong>on</strong>g> specific cake resistance, H <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

cake thickness, qc <str<strong>on</strong>g>the</str<strong>on</strong>g> cake volume fracti<strong>on</strong>, (Pi<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> foulant bulk volume fracti<strong>on</strong>, and p <str<strong>on</strong>g>the</str<strong>on</strong>g> water<br />

viscosity. Combining <str<strong>on</strong>g>the</str<strong>on</strong>g> above equati<strong>on</strong>s and<br />

differentiating with respect to time, we find that<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> initial rate <str<strong>on</strong>g>of</str<strong>on</strong>g> flux decline is given by<br />

q/Jo)<br />

= _ R’c Jo % = _ %AJc~Jo<br />

dt t=O %% 2~2%(1-%~ ,.\<br />

(4)<br />

where Jo is <str<strong>on</strong>g>the</str<strong>on</strong>g> initial <strong>membrane</strong> flux. The<br />

parameters Jo and R,,, in <str<strong>on</strong>g>the</str<strong>on</strong>g> above equati<strong>on</strong> can<br />

be determined from <str<strong>on</strong>g>the</str<strong>on</strong>g> experimental measurements.<br />

The initial decline rate found in Fig. 4 is<br />

used. A typical value <str<strong>on</strong>g>of</str<strong>on</strong>g> 0.64 is used for cp=,<br />

whereas (Pi is estimated from <str<strong>on</strong>g>the</str<strong>on</strong>g> ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong><br />

assuming a density <str<strong>on</strong>g>of</str<strong>on</strong>g> 4 g/cm3 for <str<strong>on</strong>g>the</str<strong>on</strong>g> ir<strong>on</strong><br />

oxide particles. Then, Eq. (4) can be solved for a<br />

to obtain an estimate <str<strong>on</strong>g>of</str<strong>on</strong>g> 3 nm for <str<strong>on</strong>g>the</str<strong>on</strong>g> characteristic<br />

particle radius. Such a size may be<br />

expected to be characteristic <str<strong>on</strong>g>of</str<strong>on</strong>g> primary precipitated<br />

crystallites ra<str<strong>on</strong>g>the</str<strong>on</strong>g>r than aggregates since<br />

values <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> same order <str<strong>on</strong>g>of</str<strong>on</strong>g> magnitude have been<br />

obtained by low-angle x-ray scattering measurements<br />

[ 141. It must be emphasized that <str<strong>on</strong>g>the</str<strong>on</strong>g> above<br />

exercise is performed to provide <strong>on</strong>ly an order <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

magnitude size estimate since several assumpti<strong>on</strong>s<br />

have been made. In additi<strong>on</strong>, ir<strong>on</strong> oxide<br />

crystallites resemble el<strong>on</strong>gated ellipsoids ra<str<strong>on</strong>g>the</str<strong>on</strong>g>r<br />

than spheres c<strong>on</strong>sidered here. This size estimate<br />

for <str<strong>on</strong>g>the</str<strong>on</strong>g> particles forming <str<strong>on</strong>g>the</str<strong>on</strong>g> fouling deposit does<br />

not necessarily mean that aggregati<strong>on</strong> effects are<br />

absent. Indeed, it was observed by <str<strong>on</strong>g>the</str<strong>on</strong>g> naked eye<br />

that settleable floes were formed at an ir<strong>on</strong><br />

c<strong>on</strong>centrati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> 10 ppm within a period <str<strong>on</strong>g>of</str<strong>on</strong>g> 1 h.<br />

Ra<str<strong>on</strong>g>the</str<strong>on</strong>g>r, it suggests two things. First, that porosity<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> deposits does not dramatically change by<br />

aggregati<strong>on</strong> effects or that it is c<strong>on</strong>trolled by<br />

pressure and cake c<strong>on</strong>solidati<strong>on</strong>. Sec<strong>on</strong>d, that any<br />

effects <str<strong>on</strong>g>of</str<strong>on</strong>g> reduced depositi<strong>on</strong> efficiency <str<strong>on</strong>g>of</str<strong>on</strong>g> large<br />

size aggregates or any particle detachment effects<br />

are not significant.<br />

Efforts to obtain more precise informati<strong>on</strong> <strong>on</strong><br />

aggregati<strong>on</strong> kinetics by a dynamic light scattering<br />

technique were not successful due to <str<strong>on</strong>g>the</str<strong>on</strong>g> small<br />

scattering intensity <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> colloidal dispersi<strong>on</strong>s in<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> range <str<strong>on</strong>g>of</str<strong>on</strong>g> ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>s used in <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

fouling experiments. In general, such measurements<br />

are difficult with very small particles due<br />

to <str<strong>on</strong>g>the</str<strong>on</strong>g> c<strong>on</strong>flicting requirements <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> technique<br />

such as sufficient scattering and sufficiently slow<br />

aggregati<strong>on</strong> kinetics compared to <str<strong>on</strong>g>the</str<strong>on</strong>g> time<br />

required for a particle size distributi<strong>on</strong> measurement,<br />

which is roughly 1 min.


236 S.G. Yiantsios, A. J. Karabelas /Desalinati<strong>on</strong> 151 (2002) 229-238<br />

3.2. SDI measurements<br />

It is interesting to examine <str<strong>on</strong>g>the</str<strong>on</strong>g> predicti<strong>on</strong>s or<br />

recommendati<strong>on</strong>s that would be obtained by SD1<br />

measurements if <str<strong>on</strong>g>the</str<strong>on</strong>g> same water were to be used<br />

as <strong>RO</strong> feed water. For this task water was<br />

prepared in <str<strong>on</strong>g>the</str<strong>on</strong>g> same way as for <str<strong>on</strong>g>the</str<strong>on</strong>g> fouling<br />

experiments and ferric i<strong>on</strong>s were added at various<br />

c<strong>on</strong>centrati<strong>on</strong>s. SD1 measurements were performed<br />

according to <str<strong>on</strong>g>the</str<strong>on</strong>g> standard c<strong>on</strong>diti<strong>on</strong>s, that<br />

is, by use <str<strong>on</strong>g>of</str<strong>on</strong>g> 0.45 pm pore size Millipore filter<br />

disks at a pressure <str<strong>on</strong>g>of</str<strong>on</strong>g> 30 psi. At a c<strong>on</strong>centrati<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 1 ppm almost complete plugging <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> filter<br />

occurred within <str<strong>on</strong>g>the</str<strong>on</strong>g> first 2 min <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> test. The<br />

upper limit <str<strong>on</strong>g>of</str<strong>on</strong>g> c<strong>on</strong>centrati<strong>on</strong>s where measurable<br />

and meaningful SD1 values could be obtained is<br />

approximately 20 ppb, as can be seen from Fig. 5.<br />

Several interesting observati<strong>on</strong>s may be made<br />

by examinati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> results in Figs. 4 and 5.<br />

First <str<strong>on</strong>g>of</str<strong>on</strong>g> all, it may be recalled that manufacturers’<br />

recommendati<strong>on</strong>s for <strong>membrane</strong> feed waters<br />

under c<strong>on</strong>diti<strong>on</strong>s favoring oxidati<strong>on</strong> are that<br />

dissolved ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong> should be less than<br />

50 ppb. It may also be menti<strong>on</strong>ed that fouling<br />

problems due to ir<strong>on</strong> have been reported with NF<br />

<strong>membrane</strong>s at even lower c<strong>on</strong>centrati<strong>on</strong>s, and,<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g>refore, this limit could be c<strong>on</strong>sidered tentative<br />

[ 151. The upper limit <str<strong>on</strong>g>of</str<strong>on</strong>g> roughly 20 ppb for<br />

$3<br />

2<br />

1<br />

.;<br />

+<br />

b<br />

f .<br />

/<br />

c<br />

0 ~~.~_~.~~_..~~~~ _--t..“_-.-__._-.. _-.._I<br />

lFe3’~kW<br />

15 20<br />

Fig. 5. SD1 as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong> in <str<strong>on</strong>g>the</str<strong>on</strong>g> feed<br />

water.<br />

precipitated ir<strong>on</strong> obtained by <str<strong>on</strong>g>the</str<strong>on</strong>g> present SD1<br />

measurements is not <strong>on</strong>ly remarkably close to<br />

manufacturers recommendati<strong>on</strong>s, but also <strong>on</strong> <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

correct side. A sec<strong>on</strong>d observati<strong>on</strong> is that <str<strong>on</strong>g>the</str<strong>on</strong>g>re is<br />

a notable sensitivity to particles much smaller<br />

that <str<strong>on</strong>g>the</str<strong>on</strong>g> pore size <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> filter employed in <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

SD1 test. In this case, <str<strong>on</strong>g>the</str<strong>on</strong>g> observed filter fouling<br />

is more likely due to pore restricti<strong>on</strong> and<br />

plugging ra<str<strong>on</strong>g>the</str<strong>on</strong>g>r than to cake formati<strong>on</strong>. One<br />

might recall that a small change in pore diameter<br />

results in significant changes in pressure drop,<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> latter being proporti<strong>on</strong>al to pore diameter<br />

raised to <str<strong>on</strong>g>the</str<strong>on</strong>g> fourth power. Thus, despite <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

negligible particle retenti<strong>on</strong>, significant effects<br />

are observed due to pore restricti<strong>on</strong> and plugging.<br />

Finally, from <str<strong>on</strong>g>the</str<strong>on</strong>g> linear relati<strong>on</strong>ship obtained<br />

in Fig. 4 <strong>on</strong> <str<strong>on</strong>g>the</str<strong>on</strong>g> basis <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> present <strong>RO</strong> fouling<br />

data, it may be projected that a similar <strong>membrane</strong><br />

operating under similar c<strong>on</strong>diti<strong>on</strong>s would exhibit<br />

a 6% drop in productivity within less than 45 or<br />

30 days, if <str<strong>on</strong>g>the</str<strong>on</strong>g> feed water c<strong>on</strong>tained 10 ppb or<br />

15 ppb <str<strong>on</strong>g>of</str<strong>on</strong>g> ir<strong>on</strong>; <str<strong>on</strong>g>the</str<strong>on</strong>g>se c<strong>on</strong>centrati<strong>on</strong>s corresp<strong>on</strong>d<br />

to SD1 values close to 3 or 4, respectively, which<br />

are within <str<strong>on</strong>g>the</str<strong>on</strong>g> range <str<strong>on</strong>g>of</str<strong>on</strong>g> recommended values.<br />

However, <str<strong>on</strong>g>the</str<strong>on</strong>g> above rates <str<strong>on</strong>g>of</str<strong>on</strong>g> flux decline are<br />

ra<str<strong>on</strong>g>the</str<strong>on</strong>g>r severe, and hence, under <str<strong>on</strong>g>the</str<strong>on</strong>g> c<strong>on</strong>diti<strong>on</strong>s <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

<str<strong>on</strong>g>the</str<strong>on</strong>g> present tests, <str<strong>on</strong>g>the</str<strong>on</strong>g> SD1 does not appear to be<br />

c<strong>on</strong>servative enough. Moreover, <str<strong>on</strong>g>the</str<strong>on</strong>g> SD1 cannot<br />

predict fouling rates, and, hence, it cannot<br />

discriminate between different types <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>membrane</strong>s.<br />

For example, under <str<strong>on</strong>g>the</str<strong>on</strong>g> same c<strong>on</strong>diti<strong>on</strong>s,<br />

an NF <strong>membrane</strong> would be expected to foul more<br />

rapidly and a seawater <strong>membrane</strong> to be less<br />

affected since <str<strong>on</strong>g>the</str<strong>on</strong>g>ir resistances are different<br />

compared to that <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> fouling cake. It is,<br />

<str<strong>on</strong>g>the</str<strong>on</strong>g>refore, suggested that an improvement <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

SDI, in <str<strong>on</strong>g>the</str<strong>on</strong>g> directi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> predicting fouling rates<br />

and discriminating between various <strong>membrane</strong>s<br />

is needed.<br />

4. C<strong>on</strong>clusi<strong>on</strong>s<br />

In <str<strong>on</strong>g>the</str<strong>on</strong>g> present study ir<strong>on</strong> oxide was selected as<br />

a typical inorganic colloidal foulant due to its


S.G. Yiantsios. A.J. Karabelas / Desalinati<strong>on</strong> 151 (2002) 229-238 237<br />

importance, as evidenced from well known<br />

manufacturers’ recommendati<strong>on</strong>s <strong>on</strong> ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong><br />

in feed waters and from frequently<br />

encountered problems in <strong>membrane</strong> installati<strong>on</strong>s.<br />

Fouling data were obtained by c<strong>on</strong>tinuous<br />

introducti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> ferric i<strong>on</strong>s in <str<strong>on</strong>g>the</str<strong>on</strong>g> feed water and<br />

sp<strong>on</strong>taneous precipitati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> colloidal ir<strong>on</strong><br />

hydrous oxide. Thus, <str<strong>on</strong>g>the</str<strong>on</strong>g> experimental procedure<br />

followed simulates phenomena taking place in<br />

actual <strong>membrane</strong> installati<strong>on</strong>s and <strong>on</strong>ly skips <str<strong>on</strong>g>the</str<strong>on</strong>g><br />

oxidati<strong>on</strong> step <str<strong>on</strong>g>of</str<strong>on</strong>g> dissolved ferrous i<strong>on</strong>s. A range<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> ir<strong>on</strong> c<strong>on</strong>centrati<strong>on</strong>s was identified where a<br />

linear relati<strong>on</strong>ship existed between flux reducti<strong>on</strong><br />

rate and c<strong>on</strong>centrati<strong>on</strong> under <str<strong>on</strong>g>the</str<strong>on</strong>g> experimental<br />

c<strong>on</strong>diti<strong>on</strong>s employed. Although <str<strong>on</strong>g>the</str<strong>on</strong>g> results<br />

obtained represent accelerated fouling c<strong>on</strong>diti<strong>on</strong>s,<br />

it is proposed that <str<strong>on</strong>g>the</str<strong>on</strong>g> linear relati<strong>on</strong>ship can be<br />

extrapolated to lower and more realistic c<strong>on</strong>centrati<strong>on</strong>s.<br />

The upper limit <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> c<strong>on</strong>centrati<strong>on</strong><br />

range where measurable and meaningful SD1<br />

values could be obtained was roughly 20 ppb,<br />

which is remarkably close to <strong>membrane</strong><br />

manufacturers’ recommendati<strong>on</strong>s. A notable<br />

sensitivity <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> SD1 was also observed with<br />

particles for which retenti<strong>on</strong> is negligible, due<br />

mainly to filter pore restricti<strong>on</strong> and plugging.<br />

However, <strong>on</strong> <str<strong>on</strong>g>the</str<strong>on</strong>g> basis <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> <strong>RO</strong> fouling data<br />

obtained, it may be projected that <str<strong>on</strong>g>the</str<strong>on</strong>g> SD1 is not<br />

c<strong>on</strong>servative enough. Fur<str<strong>on</strong>g>the</str<strong>on</strong>g>rmore, since <str<strong>on</strong>g>the</str<strong>on</strong>g> SD1<br />

cannot predict fouling rates, it cannot discriminate<br />

between different types <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>membrane</strong>s (i.e.,<br />

NF vs. high pressure <strong>RO</strong>). It is, <str<strong>on</strong>g>the</str<strong>on</strong>g>refore,<br />

suggested that an improvement <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>the</str<strong>on</strong>g> SD1 is<br />

needed in <str<strong>on</strong>g>the</str<strong>on</strong>g> directi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> predicting fouling rates<br />

and discriminating between various <strong>membrane</strong><br />

types.<br />

Acknowledgments<br />

Financial support <str<strong>on</strong>g>of</str<strong>on</strong>g> this work by <str<strong>on</strong>g>the</str<strong>on</strong>g> Middle<br />

East Desalinati<strong>on</strong> Research Center (Project No<br />

98-BS-034) is gratefully acknowledged.<br />

References<br />

111 S.C.J.M. Van Ho<str<strong>on</strong>g>of</str<strong>on</strong>g>, J.G. Minnery and B. Mack,<br />

Performing a <strong>membrane</strong> autopsy, Desalinati<strong>on</strong> Water<br />

Reuse, ll(2002) 40-46.<br />

PI J.C. Schippers and J. Verdouw, The modified fouling<br />

index, a method <str<strong>on</strong>g>of</str<strong>on</strong>g> determining <str<strong>on</strong>g>the</str<strong>on</strong>g> fouling characteristics<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> water, Desalinati<strong>on</strong>, 32 (1980) 137-148.<br />

r31 S.F.E. Boerlage, M.D. Kennedy, M.R. Dicks<strong>on</strong>,<br />

D.E.Y. El-Hodali and J.C. Schippers, The modified<br />

fouling index using ultrafiltrati<strong>on</strong> <strong>membrane</strong>s (MFI-<br />

UF): characterizati<strong>on</strong>, filtrati<strong>on</strong> mechanisms and<br />

proposed reference <strong>membrane</strong>, J. Membr. Sci., 197<br />

(2002) l-2 1.<br />

[41 S.F.E. Boerlage, M.D. Kennedy, M.P. <str<strong>on</strong>g>An</str<strong>on</strong>g>iye, E.M.<br />

Abogrean, G. Galjaard and J.C. Schippers, M<strong>on</strong>itoring<br />

particulate fouling in <strong>membrane</strong> systems,<br />

Desalinati<strong>on</strong>, 118 (1997) 13 l-142.<br />

PI Hydranautics, <strong>RO</strong>DESIGN, Hydranautics <strong>RO</strong> System<br />

Design S<str<strong>on</strong>g>of</str<strong>on</strong>g>tware, VERSION 6.4 (c), Pretreatment and<br />

design limits secti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> help module, 1998.<br />

[61 J.C. Schippers, J.H. Hannemaayer, C.A. Smolders<br />

and A. Kostense, Predicting flux decline in reverse<br />

osmosis <strong>membrane</strong>s, Desalinati<strong>on</strong>, 38 (1981) 339-<br />

348.<br />

[71 E.M. Vrijenhoek, S. H<strong>on</strong>g and M. Elimelech,<br />

Influence <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>membrane</strong> surface properties <strong>on</strong> initial<br />

rate <str<strong>on</strong>g>of</str<strong>on</strong>g> colloidal fouling <str<strong>on</strong>g>of</str<strong>on</strong>g> reverse osmosis and<br />

nan<str<strong>on</strong>g>of</str<strong>on</strong>g>rltrati<strong>on</strong><strong>membrane</strong>s, J. Membr. Sci., 188 (2001)<br />

115-128.<br />

181 X. Zhu and M. Elimelech, Colloidal fouling <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

reverse osmosis <strong>membrane</strong>s: Measurements and fouling<br />

mechanisms, Envir<strong>on</strong>. Sci. Technol., 31 (1997)<br />

3654-3662.<br />

r91 R.D. Cohen and R.F. Probstein, Colloidal fouling <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

reverse osmosis <strong>membrane</strong>s, J. Coll. Interface Sci.,<br />

114 (1986) 194-207.<br />

[lOI Permasep Reverse Osmosis Product Bulletin 4010,<br />

Pretreatments for Permasep permeators, DuP<strong>on</strong>t,<br />

1994.<br />

r.111 J. Kucera, Properly apply reverse osmosis, Chem.<br />

Eng. Progress, February (1997) 54-61.<br />

PI E. Matijevic and P. Scheiner, Ferric hydrous oxide<br />

SOIS. III. Preparati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> uniform particles by hydrolysis<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> Fe(III)-chloride, -nitrate, and -perchlorate soluti<strong>on</strong>s,<br />

J. Coll. Interface Sci., 63 (1978) 509-524.


238 S. G. Yiantsios, A.J. Karabelas /Desalinati<strong>on</strong> 151 (2002) 229-238<br />

[13] W. Stumm and J.J. Morgan, Aquatic Chemistry.<br />

Chemical Equilibria and Rates in Natural Waters, 3rd<br />

ed., Wiley, New York, 1996.<br />

[14] W.R. Meyer, S.H. Pulcinelli, C.V. Santilli and A.F.<br />

Craievich, Formati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> colloidal particles <str<strong>on</strong>g>of</str<strong>on</strong>g>hydrous<br />

ir<strong>on</strong> oxide by forced hydrolysis, J. N<strong>on</strong>-Crystalline<br />

Solids, 273 (2000) 41-47.<br />

[15] J.A.M. Van Paasen, J.C. Kruith<str<strong>on</strong>g>of</str<strong>on</strong>g>, S.M. Bakker and<br />

F.S. Kegel, Integrated multiobjective <strong>membrane</strong><br />

systems for surface water treatment: pm-treatment <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

nan<str<strong>on</strong>g>of</str<strong>on</strong>g>iltrati<strong>on</strong> by riverbank filtrati<strong>on</strong> and c<strong>on</strong>venti<strong>on</strong>al<br />

groundwater treatment, Desalinati<strong>on</strong>, 118<br />

(1998) 239-248.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!