INFORMAnON TO USERS

This manuscript has been reproduœd from the microfilm mater. UMI films the
text directly tram the original or copy submitled. Thus. seme lhesis and
dissertation copies are in typewriter face. while oIhers rnay be from 8ny type of
computer printer.

The qu.lity of th. reproduction .. dependent upon the quallty of the copy
submitted. Broken or indistinct prim. coIor8d or poor quality illustrations and
photographs, print bleedlhrough, substandaRt margins. and improper alignment
can adversely affect reproduction.

ln the unlikely event that the author did not send UMI a complete manuscript and
there are missing pages. these will be noted. Also. if unauthorized copyright
material had to be removed, a note will indicate the deIeIïon.

Oversize materials (e.g.• maps, drawings. charts) are reproduced by sectioning

the original. begiMing st the upper Ieft-hand corner and continuing from Ieft to
right in equal sections with small overtaps.

Photographs incIuded in the original manuscript have been reproduced

xerographically in this copy. Higher quality 6- x 9" btac:k and white photographie
ponts are available for any phoIographs or illustratiOnS appearing in this copy for
an additional charge. Contact UMI directly to 0Rter.

Bell & HoweIllnfarmation and Leaming

300 Nor1h Z8eb Road. Ann Arbor. MI 48108-1346 USA
e

UMI 800-521-0600
• Biological and Physico-Chemical Removal

of Iron from Potable Waters

Redox Potential as an Indicator of Treatment Effectiveness

By

Catherine V. M. Tremblay

McGill University, Montreal

Department of Civil Engineering and Applied Mechanics

November 1997

A thesis submitted to the Faculty ofGraduate Studies and Research in

partial fulfillment of the requirements of the degree of

Master of Engineering.

o Catherine V. M. Tremblay, 1997

•
 Bibliothèque nationale
1+1 National Ubrary
of Canada du Canada

Acquisitions and Acquisitions et

Bibliographie Services services bibliographiques
395 Wellington Street 395. rue Wellington
Ottawa ON K1A 0N4 OIIawa ON K1A 0N4
Canada canada
Your.Vœ.~

The author bas granted a non- L'auteur a accordé une licence non
exclusive licence allowing the exclusive permettant à la
National Library of Canada to Bibliothèque nationale du Canada de
reproduce, lo~ distribute or sell reproduire, prêter, distribuer ou
copies of this thesis in microform, vendre des copies de cette thèse sous
paper or electronic formats. la forme de microfiche/film, de
reproduction sur papier ou sur format
électronique.

The author retains ownership of the L'auteur conserve la propriété du

copyright in this thesis. Neither the droit d'auteur qui protège cette thèse.
thesis nor substantial extracts from it Ni la thèse ni des extraits substantiels
may he printed or otherwise de celle-ci ne doivent être imprimés
reproduced without the author' s ou autrement reproduits sans son
pernnsslon. autorisation.

0-61244045-1

Canadl
 ABSTRACT

• The tirst objective of tms research was to evaluate oxidation-reduction potential (ORP) as

an indicator of effective iron removal in a biological process and to detennine its

relationship to dissolved oxygen (DO) and residual iron in the filtered water. Biological

removal of iron to produce drinking water was established on one full-scale and two

pilot-scale plants al two sites in France. Results show that below a minimum DO

concentration of approximately 0.3 mgIL~ residual iron concentration was related to ORP

in the filtered water. Above the minimum DO requirement. ORP on the order of

approximately 500 mV consistently reflected effective iron treatment and residual iron

concentrations less than the French norm for potable water of 0.2 mgIL.

Secondly, two catalytic materials of filtration known as Ferrolite MC2 and

Purolite were investigated on pilot-seale for their capacity to oxidize iron and manganese.

Results showed that both materials eliminated dissolved iron to below detection levels

regardless of DO or filtration rates up to 15 m/h (Ferrolite MC2) and 20 mIh (Purolite).

During the trial period where no tilter regeneration was carried out, the elimination of

dissolved iron and dissolved manganese, respectively, were 100% and 77% for Purolite

and 100% and 65% for Ferrolite MC2 for a raw water with 3.47 mg/L of dissolved iron

and 0.317 ofdissolved manganese. Purolile filtered a total volume of 1127 Lor 137

L/kg of catalytic material and Ferrolite MC2 filtered a total of 1457 L or 217 L/kg.

• 11
•
RÉsUMÉ

Le premier objectif de ce travail était d~évaluer I~emploi du potentiel d~oxydo-réduction

(potentiel redox) comme indicateur de déferrisation par un procédé de déferrisation

biologique et d'établir un rapport entre le potentiel redox et l'oxygène dissous (00) ainsi

qu'avec la concentration de fer résiduel dans l'eau filtrée. Un procédé industriel de

production d'eau potable et deux procédés pilotes ont été utilisés à deux endroits

différents en France. Les résultats indiquent que~ lorsque l'On est inférieure à 0,3 mgIL,

la concentration en fer résiduel dans l'eau filtrée est liée au potentiel redox de l'eau. Aux

concentrations d'oxygène dissous supérieures à cette valeur seuit le potentiel redox

demeurait constant à plus de 500 m V et indiquait une élimination satisfaisante du fer et

des concentrations en fer résiduel inférieures à la norme française de 0,2 mg/L.

Deuxièmemen~ deux matériaux filtrants catalytiques, la Ferrolite MC2 et la

Purolite, ont été étudiés en pilote pour établir leur capacité d'oxydation du fer et du

manganèse. Aux concentrations en oxygène dissous étudiées et à des vitesses de filtration

jusqu'à 15 m/h (sur Ferrolite MC2) et 20 mIh (sur Purolite), la concentration de fer

dissous résiduel est demeurée inférieure au seuil de détection pour les deux matériaux.

Durant toute la durée des essais et sans aucune régénération des filtres, les rendements

d'élimination du fer et du manganèse ont été respectivement de 100% et de 77% pour la

PuroUte et de 100% et 65% pour la Fe"olite MC2, pour une eau brute contenant 3.47

mgIL de fer dissous et 0.317 mgIL de manganèse dissous. Les volumes d'eau filtrée ont

été respectivement de 1127 L ou de 137 L par kg de matériau catalytique pour la Purotite

et de 1457 Lou 217 L par kg pour la Ferro/ite MC2.

• 111
•
ACKNOWLEDGMENTS

l would like to express thanks toward my supervisors~ Dr. James A. Nicell and Dr. André

Beaubien~ whose guidance and encouragement helped me greatly throughout the research

and writing.

l am also gratefuJ toward the Research Center for the Lyonnaise des Eaux

(C.I.R.S.E.E.) in France~ who funded the project through Degrémont Infilco of Montreal.

My gratitude is expressed toward my supervisor in France~ Philippe Charles~ for his

patience and perseverance, and to my colleague~ Sébastien Defois, for technical

assistance and support.

Special thanks are extended to everyone at the local Lyonnaise des Eaux offices in

Hossegor and Soissons~ especially to Mr. Luc Rives of Hossegor whose collaboration and

support were invaluable.

• IV
•
TABLE OF CONTENTS

ABSTRACT ii

RÉsuMÉ iii

ACKNOWLEDGMENTS iv

TABLE OF CONTENTS v

LIST Of TABLES ix

LIST Of FIGURES x

LIST Of ACRONYMS xii

1. INTRODUCTION _ 1

2. LITERATURE REVIEW _ 4

2.1 CHEMICAL CHARACTERISTICS OF IRON AND MANGANESE•.•••••••••••.••.•.••••.••.•••••..••••.•• .4

2.1 . 1 Iron 4

2. 1.2 Manganese _ 7

2.2 PHYSICQ-CHEMICAL TREATMENT ••.•••.•••..••...........•..•.•..•••.•.•••••••.•.•••••••.•...•.••••.••.••••.••• 8

2.2.1 Aeration 9

2.2.2 Chemical oxidation _ 10

2.2.3 Filtration 11

2.2.4 Filtration on catalytic materials 12

2.2.4.1 Purolite 1-1

2.2.4.2 Ferrolire MC2 1-1

• v
•
2.3 BIOLOGICAL TREATMENT ••.••••••••.•••••••.••••.•••.••••.•.•.•••••..••.••••......••••....••.•.......•..••....... 15

2.3.1 Iron bacteria 16

2.3.2 Mechanism for bacterial iron uptake 20

2.3.3 Growth conditions of iron bacteria 22

2.3.4 Full-scale application of biological iron removal .24

2.4 OXIDATIaN-REDuCTION POTENTIAL .•.••.••...............................•..•............................ .27

3. METHODS AND MATERIALS 30

3 .1 PRESENTATION OF THE SITES .....•....•......•..•................•.............•....•........................... 30

3.1.1 Seignosse 30

3.1.2 Septmonts 32

3.2 MATERIALS 33

3.2.1 Installation on site at Seignosse 33

3.2.2 Biological pilot plant at Seignosse 34

3.2.3 Biological pilot plant at Septmonts 36

3.2.4 Catalytic pilot plants at Septmonts 36

3.3 ANALYTICAL METHODS 38

3.3.1 Electrochemical measurements 38

3.3.2 Sampling 38

3.3.3 Chemical analyses 40

3.4 EXPERIMENTAL METHODS ....•...•..•..•....•..••••..••...•...•...........••.•••.•.....•......................... 40

3.4.1 Normal operation of the facilities in Seignosse .40

3.4.2 Aeration-ORP trials 41

• vi
•
3.4.3 CataJytic triaIs 42

4. RESULTS AND DISCUSSION FOR SEIGNOSSE .43

4.1 INITIAL STUDY ON FULL SCALE 43

4.1.1 Normal operation 43

4.1.2 Aeration-ORP trials 46

4.2 BIOLOGICAL PILOT STUDY.•.••••••..••••..•••..•••••••....••.•••..•.....•.•••.••••••.•...•..••..••..•..••........53

4.2.1 Startup of pilot tilter 53

4.2.2 Pilot tilter backwash 54

4.2.3 Aeration-ORP trials 56

4.3 SECOND STUDY ON FULL SCALE••..••.••••••..••...•.••••.•••.•.........•••..........•.•...•••..•......•......59

4.3.1 Primary tilter startup without pH adjustment 59

4.3.2 Biological treatment hypothesis 61

4.3.3 Summary on primary tilter operation 62

5. RESULTS AND DISCUSSION FOR SEPTMONTS 66

5.1 B IOLOGICAL PILOT STUDY 66

5.1.1 Startup of pilot tilter 66

5.1.2 Aeration-ORP trials 67

5.1.3 Summary on the biological pilot study at Septmonts 69

5.2 CATALYTIC PILOT STUDY ••..•...••••..••..••••...•...••...•.•..••••..•.....•.•••...•••..••...•...........••......71

5.2.1 Ferrolite Me2 71

5.2.2 Purolite 73

5.2.3 Summary on the catalytic pilot study 75

• vii
•
6. CONCLUSIONS 76

REFERENCES 79

APPENDIX 1: RAW W ATER ANAL YSES AT SEIGNOSSE JUL y 1996 1-1

APPENDIX 2: DÉFERRISATION BIOLOGIQUE DE SEIGNOSSE: RAPPORT DE LA

VISITE EFFECTUÉE LES 21 ET 22 MARS 1991 2-1

• viii
•
LIST OF TABLES

Table 1. Sorne iron bacteria 16

Table 2. Organization of the study at two sites 30

Table 3. Seignosse mean raw water parameters ± standard deviation.,lune 1996 32

Table 4. Septmonts mean raw water parameters ± standard deviation.. November 1996..33

Table 5. Raw water quality evolution after startup .43

Table 6. Primary filter intennittent operation, July 1996 .46

Table 7. Water quality before and after primary tilter, June and JuIy 1996 .50

Table 8. Summary of previous and current operating conditions on the primary tilter 63

Table 9. Results from trials on biological pilots at Seignosse and Septmonts 70

Table 10. Startup of the Ferrolite MC2 pilot: Rawand filtered water parameters 72

Table Il. Iron and manganese oxidation on Ferro/ite MC2 73

Table 12. Iron and manganese oxidation on Purolite 74

• ix
•
LIST OF FIGURES

Figure 1. The states of iron (after Degrémoo4 1991) 5

Figure 2. pH-potential diagram for iron (from Hem, 1961 ) 6

Figure 3. pH-potential diagram for manganese (from Pisarczyk.. 1995) 8

Figure 4. Adsorption of metal ions on hydrous manganese (IV) dioxide (after Pisarczyk..

1995) 13

Figure 5. Leptothrix ochracea (from Mulder and Deioema.. 1981) 18

Figure 6. Gallionella ferruginea (from Hanert, 1981 a) 18

Figure 7. Siderocapsa treubii slide ongrowth in aquarium-natural water culture (from

Hanert.. l 981 b).............................•............................................................................... 19

Figure 8. Metallogenium with buds and daughter colonies from lakes 19

Figure 9. pH-Eh diagram of iron bacteria activity (after Mouchet, 1992) 25

Figure 10. Drinking water treatment facilities in Seignosse 31

Figure Il. BiologicaI pilot plant facilities in Seignosse .35

Figure 12. CataIytic pilot plant facilities in Septmonts 37

Figure 13. Evolution of raw water quaIity with plant startup after 14 hour downtime 44

Figure 14. Evolution of residual iroD after ovemight stoppage and during tilter

maturation " .." 44

Figure 15. Evolution of DO and ORP during the aeration-ORP trials on the primary tilter

in June 1996 47

Figure 16. Evolution of total and dissolved residual iron during the aeration-ORP trials.47

• x
•
Figure 17. Relationship between DO and both ORP and DR! during the aeration-ORP

trials ' 49

Figure 18. Relationship between ORP and DR! during the aeration-ORP trials .49

Figure 19. pH-Eh diagram indicating water conditions before and after primary tilter

(after Mouchet~ 1992) 51

Figure 20. Startup of biological iron elimination on pilot scale 54

Figure 21. Results from the aeration-ORP trials on pilot-seale in Oetober 1996 .57

Figure 22. Correlation between ORP and TRI during start-up and trials on pilot-scale 57

Figure 23. Primary tilter start-up without pH adjustment in October 1996 60

Figure 24. Primary tiIter maturation after 3 different backwashes, October 1996 65

Figure 25. Iron and manganese content of raw water at Seignosse during study 65

Figure 26. Results from the aeration-ORP trials on pilot scale, November 1996 68

Figure 27. Relationship between ORP and TRI during aeration-ORP trials 68

• xi
•
LIST OF ACRONYMS

DO = dissolved oxygen

DR! = dissolved residual iron

ORM dissolved residual manganese

ORP = oxidation reduction potential

TRI total residual iron

TRM = total residual manganese

• xii
 1

•
1. INTRODUCTION

The presence of reduced foons of iron and manganese in groundwater is a naturaI

phenomenon, particularly in acidic waters. Concentrations generally found in naturaI

waters do not have toxicologica1 significance (Lehr et al.~ 1980). However~ when these

groundwaters are tapped for potable water, reduced ions come into contact with oxygen

and are precipitated as red or black insoluble oxides. These precipitates can lead to

corrosion and obstructions in underground water canals, and problems for the consumer,

such as tinted, metallic-tasting drinking water (Lehr et al., 1980). Dissolved iron and

manganese ions are a1so a nuisance for industries which rely on clear water, such as the

food industry and the pulp and paper industry (Lehr et al., 1980).

Therefore, in groundwater that is destined for drinking water production~ the level

of ITon is regulated based on aesthetic and taste considerations, rather than concem over

physiological efIects (Lehr et al., 1980). French regulation indicates maximum

admissible concentrations of 0.2 mg/L and 0.05 mgIL for iron and manganese,

respectively (Degrémont, 1991). Canadian water quality guidelines indicate maximum

allowable concentrations of 0.3 mg/L for iron and 0.05 mgIL for manganese (Health

Canad~ 1996).

Treatment of iron and manganese cao he accomplished by either physico-

chemical means~ which includes the use of catalytic materials, or by biological means. [n

the case of a conventional physico-chemical process~ the precipitation is induced by an

intensive aeration, usually followed by the addition of a chemical oxidant (in the case of

manganese), and then filtration on a sand filter. In the catalytic process, the filtration

•
 2

•
material is, for example, a specially coated sand or ceramic material which catalyzes the

oxidation. Biological treatment makes use of SPecific micro-organisms which instigate

the oxidation of reduced forms of iron and manganese in raw water, and also act to retain

the precipitates in the biological sand tilter.

The primary objective of this research is to investigate oxidation-reduction

potential (ORP or redox potential) as an indicator of biological iron treatment

effectiveness. This is accomplished by evaluating the impact of dissolved oxygen

concentration (DO) on the iron removal process and then detennining the relationship

between ûRP and DO and between ORP and the residual iron concentration in the

filtered water. ORP has been widely studied for application in control of biological

wastewater treatment systems (Charpentier et al., 1987; Wareham et al., (993).

Biological drinking water treatment is similar in that specifie bacteria Mediate the

presence of reduced and oxidized species (i.e. iton) which in turn impact upon the ûRP.

The second objective of the research is to evaluate iron and manganese oxidation

on two granular catalytic materials known as Purolite and Ferrolite MC2. Each matenal

is studied on pilot-scale at different velocities of filtration and at various DO Levels. The

performance of each is evaluated by the volume of raw water filtered per unit mass of

material for given residual concentrations of iron and manganese. These two materials

were selected after they were studied as part of previous unpublished research (Thill,

1995) through the International Center for Research on Water and the Environment

(Centre International de Recherche Sur rEau et l'Environnement or C.I.R.S.E.E.) which

evaluated severa! prevalent catalytic materials for their effectiveness with respect to iron

• and manganese treatment. Ferrolite MC2 is selected for the current study as it was found
 3

•
to achieve the hest overall perfonnance of the materials tested. Purolite is reconsidered

as it was shown to he the material most widely used in the full-scale treatment of iron and

manganese.

Although the research may demonstrate a promising application of ORP as an

indicator in biological treatlnent of iron, it should he noted that raw water conditions vary

greatly from installation to installatio~ making the operating conditions at different

treatment facilities quite diverse. PiIot-scaIe studies are required to determine the

importance of ORP for any speeific application. Similarly, full-seale use of a catalytic

materia! of filtration sueh as Purolite or Ferrolite Me2 must first be verified on a pilot-

seale under site-specifie conditions. Therefore, the current work simply provides an

example as to how ORP can he evaluated as an indicator in biologieal treatment and the

aforementioned catalytic materials cao he evaluated for iron and/or manganese oxidation

under a particular set of conditions.

•
 4

•
2. LITERATURE REVIEW

2.1 CHEMICAL CHARACfERISTICS OF IRON AND MANGANESE

Iron is the fourth most abundant element on the earth's surface (Stolzenberg, (995), while

manganese is twelfth (pisarczyk, (995). Surface water usually contains ooly precipitated

forms of ieon and manganese and they are easily removed by simple clarification. The

presence of reduced iron and manganese is then generally reserved for deep groundwater

sources, where the water table is deprived of oxygen and is in contact with minerais

which contain iron and manganese (Lehr et al., 1980). These two elements are frequently

found together in groundwater sources, but there have been cases where only one or the

other is present in high concentration (Degrémont, 1991).

To properly specify a method oftreatment for the selected groundwater source, it

is important to identify severa! parameters of the raw water, in addition to concentrations

of total and dissolved iron and manganese. The measure of pH, alkalinity~ hardness,

redox potential, coloue, organic matter, silicates, phosphates, chlorides, sulfates, turbidity,

ammonia, nitrates, dissolved oxygen, carbon dioxide and hyclrogen sulfide are analyses of

significant importance (Degrémont, (991). Many parameters need to be identified on site

ta avoid alteration of chemical species upon contact of the groundwater with air.

2.1.1 Iron

Elementary iron exists in either of two reduced states: fenic iron (trivalent) which is

represented by Fe3+ and ferrous iron (divalent), represented by Fe 2+ (Stolzenberg, (995).

Figure 1 presents the forms ofiron found in naturaI water sources.

•
 5

• Iron (II)
Total Iron

Iron (Ill)

--.
free chelated free

1
minerai organic
precipitated dissolved precipitated
complexes complexes
1 1 1
1 1 1
Rerained onfi/rer Not retained onjilter Retained on jifler

Figure 1. The states of iron (after Degrémont~ 1991).

The presence of chelated iron can complicate removal, but only when

conventional physico-chemical methods are employed (Mouchet, 1992). Hence, when

considering this treatment it is important to he able to differentiate between the forms of

iron not retained on a filter. A method to determine the amount of iron not retained

which is bound with complexing agents is weil documented (Degrémont, 1991 ).

Examples of chelating agents which have been found to associate with ferrous and ferric

iron are minerais (silicates, phosphates, polyphosphates, cyanates, or sulfides) and

organics (humic, fulvic or tannic acids) (Degrémont 1991; Mouchet, 1992).

Figure 2 illustrates the importance of pH and redox potential with respect to the

states of iron in aqueous solution. It should be noted that this figure applies for aqueous

systems in which no interference is presented by chelating agents. ft can be easily

verified from Figure 2 that in a mildly acidic, reducing environment (absence of oxygen),

•
 6

•
iron is generally stable, in dissolved ferrous ion form (Stolzenberg, 1995). AIso, Figure 2

shows that ferrie iron exists in dissolved fonn only for very Low values of pH.

1.0
" ~'"
F.···
...FcCOH)r·

~ _.- Oaldiled

""
rIOH··
0.8

~
0.6

o.•
'" \J Ï'..

'" ""~
f\
r.lOHh
ISoIidl
rc··
..
~ 0.2

'\
t
rot
:
i ~
0
i\,
~ , l "
i\",
~ 1_
-02

"~ I~d~~reco~ :
lé' .-
1 ,

-0.- :
"'R -.: Il
:" .. 1
'~
-06 W~l~ Reaucea ;! "" ~I
... • FeIOHI,_
...... ~
1
! (So1î41
; 1
i

-0.8 0
2
!
6
1
8 10
~
12 1.

Figure 2. pH-potential diagram for iron (trom Hem, 1961).

•
 7

•
Within a normal aquatic pH range (slightly acidie to slightly basic), a slow

oxidation of Fe2+ to Fe3+ takes place with increasing redox potential (dissolution of

oxygen). The oxidation rate increases with pH and precipitation of ferrie hydroxide

(Fe(OH)3) results (Droste, 1997).

The lower right band corner of Figure 2 indicates other important compounds

which can arise in the iron-aquatie system. Ferric carbonate or ferrie hydroxide species

will oecur in the absence of oxygen and at high pH and their appearance is dependent on

the carbonate content of the water (Stolzenberg, 1995). Carbonate speeies activity

equivalent ta either 100 ppm or 1000 ppm as bicarbonate are represented by the solid

boundaries or dashed boundaries, respectively. As the pH is inereased further, the ferrous

ions fonn stable hydroxide precipitates (Hem, 1961).

2.1.2 Manganese

The element manganese exists in oxidation states ranging from -3 to +7 (Pisarezyk,

1995). The ion with an oxidation state of +7 is chemieally produeed and is very

important eommercially in the formation orthe permanganate ion, Mn04 -, an extremely

powerful oxidizing agent. As will he shown later, potassium permanganate is often used

in the ehemical oxidation of iron and manganese in physico-chemical treatment plants.

2
In deep ground water sources, the divalent fonn, manganous ion Mn +, is found ta

be stable in solution up to a pH of about 8 or 9. This behaviour is represented in Figure

3, a diagrarn for manganese which is analogous to Figure 2 for iron.

Figure 3 shows that Mn(OH)2 is produced in the absence of oxygen when the pH

•
rises above about 8.3 (pisarezyk, 1995). Trivalent and mixed valent manganese
 8

•
compounds (Mn20 3 and Mn304~ respectively) will result under mildly oxidizing

conditions. When the pH is further increased into the range 9 - lO and redox potential

between 500 and 600 mV~ Mn2+ and Mn3+ are oxidized to manganese (IV) which readily

precipitates as Mn02 •

Manganese (IV) dioxide is of special interest for its surface characteristics, which

make it capable of adsorbing charged species. This characteristic is relevant to catalytic

materials used to eliminate iron and manganese, as will he discussed in Section 2.2.4.

2.2,..----....----...,.----......,.....-----.----,

1.4

, MnOz (aq)
0.6
, "
"~""11~"à"
-
>
~ ,,~(aq)
....
"
""
"
'
' .....
~-
",
--------
_ 11J 0 C (ilq)
- - - _- MnO;- •
-0.2

Mn

Mn(OH)z
-1.0

Mn

-1.8
a 4 8 12 16
pH

Figure 3. pH-potentiai diagram for manganese (from Pisarczyk, 1995).

2.2 PHYSICO-CHEMICAL TREATMENT

The classical physico-chemical approach to iron and manganese removal is known as

conventional treatment (Mouchet, 1992). The more recent developments in the realm of

•
physico-chemical treatments include the use of catalytic materials of filtration.
 9

•
Basic iron removal from water with the conventional approach cao be achieved

through aeration followed by filtration. Figure 2 illustrates the behaviour of iron species

in such a system. Treatment becomes more complicated when complexed forms of irou

or iron concentrations above 5 mg/L are detected, and when manganese is present in high

concentrations (Degrémont, 1991).

2.2.1 Aeration

[n the removal of iron from ground water, it is always advantageous to carry out an

aeration as the frrst stage, even if a chemical oxidant is required further on. The aeration

step allows for savings on costly chemical oxidants because the easily oxidized species

are already precipitated. In addition, efficient aeration permits an adequate concentration

of dissolved oxygen for the final product (Degrémont, 1991).

Aeration is carried out either under pressure in an oxidation contact tower or at

atmospheric pressure. Atmospheric installations include trickling over cascades or over

contact media (with or without air circulation), open air spraying, diffused aeration or

mechanical aeration with turbines (Oroste, 1997).

The pressurized installations are advaotageous when the treatment plant operates

under the back pressure in the water distribution network. But open aeration is also

useful for the elimination of dissolved carbon dioxide and hydrogen sulfide gases. If the

carbon dioxide concentration is high and cannot be removed in open aeration~ costly

neutralization May be necessary.

•
 ID

•
The oxidation of ferrous iron by elemental oxygen is expressed by the following

equation (Oroste, 1997):

(1)

[t has been experimentally determined that a pH of at least 7 is required to achieve

effective oxidation of iron by elemental oxygen (Degrémont, 1991; Mouchet, (992).

Since Many raw ground waters are mildly acidic, a pH adjustment by addition of lime or

caustic soda is sometimes necessary (Degrémont, 1991).

2.2.2 Chemical oxidation

Ordinarily, if manganese is present in high concentrations along with iron in raw water, it

will not precipitate following simple aeration (Degrémont, 1991). As shown in Figure 3,

a much higher pH is required for maoganese to he oxidized with oxygen than for iron.

Thus the addition of a chemicaI oxidant becomes a necessary step to achieve efficient

manganese removal.

High concentrations of iron and the presence of complexed forms of iron aIso

require the use of chemical oxidizing agents. Oxidants such as chIorine dioxide (CI02J

and potassium permanganate (KMn04) are used to provoke the oxidation of iron and

manganese. Ozone is aIso used in the treatment of manganese. These chemical

oxidations are illustrated by the following reactions (Oroste, 1997; Oegrémont, (991):

2Fe 2 + +C/2 +6H2 0-+ 2Fe(0H)3 +6H+ +2Cr (2)

3Fe 2+ + MnO; +50/1 +2H2 0-+ 3Fe(0H)) + Mn02 (3)

Mn 2+ +2C/02 +2H2 0 ...... Mn02 +202 +2Cr +4H+ (4)

• Mn 2 ++0) +H2 0-+ Mn02 +02 +2H+- (5)

 LI

•
(6)

With the use of chemical oxidants, a contact basin is often employed to improve

oxidation efficiency. In addition~ coagulants and floccuJants are sometimes added to

facilitate removal of fine precipitates. Finally, with high levels of iron or manganese to

be removed and/or the presence of complexed forms of iron or manganese, a decantation

stage May be necessary (Degrémont, (991).

As an alternative to chemicai precipitation of iron and manganese compounds..

sequestering agents such as sodium silicates, phosphates or polyphosphates are added to

water. These bind with iton and manganese in polymerie colloidal complexes and inhibit

them from forming colour or turbidity in the water distribution network (Oroste, (997).

2.2.3 Filtration

Operating conditions for filtration are likely to vary significantly from one installation to

another, due to different precipitates which arise from the various previous stages

(Mouchet, 1992).

[n the case of classic sand filters, the effective size of the sand particles ranges

from 0.5 to 1 mm and the rate of filtration between 5 and 15 m/h. The quantity of iron

retained on the filter is typically between 0.2 and 1.2 kg iron per m 2 (cross seetional area

of the filter). Head loss is generally on the order of 60 kPa. Crushed anthracite (hard

coais) has aIso been used as fl1ter material and in combination with sand in bi-layer filters

(Degrémont, 1991).

At the end of the iron removal process, disinfeetion is necessary (generally with

•
chlonne), to eliminate any bacteria., including the iton bacteria., which are often present in
 12

•
the raw water. These latter micro-organisms are known to obstruct pipelines and cause

corrosion (Droste,. 1997; Degrémon~ 1991).

A backwash using treated or raw water and air scour is the primary form of

maintenance for physico-chemical filters.

In brief summary, conventional physico-chemical treatments are generally reliable

methods of elimination for iron and/or manganese. After many years of use,. their most

favorable conditions are weil known,. which assures their continued successful application

(Mouchet, 1992).

2.2.4 Filtration on catalytic materials

From within the realm of physico-chemical treatments of iron and manganese has

emerged the use of specially prepared materials of filtration. According to Cole (1986),.

iron and manganese removal by catalytic materials is a known reliable method of

treatment in North America.

These filters generally consist of a material which naturally contains manganese

(IV) dioxide, or a material artificially coated with it. As discussed earlier,. Mn02 has been

studied for its amphoteric nature and ability to partake in surface - solution exchanges of

hydrogen and hydroxide ions (Pisarczyk, 1995). According to Stumm and Morgan

(1996), the surface charge of the hydrous Mn02 particle in a packed tilter is negative

above a pH of about 5. Thus it bas an outer layer of exposed OH- groups capable of

adsorbing charged species such as H+ and metal ions. Figure 4 illustrates this

phenomenon.

•
 13

•
(a) 1 (b) 1 (c) 1
0 0 o
1 1 /H 1
Mn Mn-~ Mn-OH- +n H+
1 1 H 1
0 +nH20 " 0 o
1
Mn
" 1
Mn-~
/H 1

~
1 1 H
(d) 1 \ /
0 0
1 1
~ lM
Mn-"
1 H
o
1 /H
Mn-O +20H+
1 \
~ lM,
Mn-O
l 'H
o
1

Figure 4. Adsorption of Metal ions on hydrous manganese (IV) dioxide (after Pisarczyk,

1995). Ca) catalytic materia! (b) hydration ofMn02 (c) surface-solution exchanges with

H+ (d) surface-solution exchanges with Metal cations, represented by 'M'.

Following adsorption, catalytic oxidation of the Metal ions takes place while the

local surface manganese dioxide is reduced. The sorption/oxidation mechanism for iron

and manganese can be expressed by the following equations:

(7)

(8)

After a certain time of operation, the tilter becomes saturated and needs to be

regenerated by a solution of permanganate, as shown by the following reaction:

•
(9)
 14

Several different catalytic materials exist on the market, but only Purolite and

• Ferrolite Me2 are considered in this study. The selection of these two materials was

based primarily on previous related research described in an unpublished report (Thill,

1995). Thill (1995) used severa! pilot columns which were assembled in parallel, each

capable of operating independently of the others. Gravitational filtration on each material

was studied without the use of reagents or an air injection. The cross sectional area and

bed depth of each pilot was 0.002 m 2 and 1 ~ respectively. Average raw water

conditions included a pH of 7.1, Eh of 175 mV, DO content of 0.8 mgIL, total iron

content of 1.65 mgIL and total manganese content of 0.115 mgIL.

2.2.4./ Purolite

Purolite is a trade name for a greensand which has been artificially coated with a layer of

manganese (IV) dioxide. Greensand is a granular, naturally occurring zeolite, with the

mineral composition (K,Na,Ca)I.2_Z -(Fe3+ ,Al, Fe, Mg)4 Si7_7.6AlI_1.402(OH)4 . nH 2 0

(Oroste, 1997). Purolite is generally used alone as the material of filtration (not in

combination with sand). Thill's (1995) results for the pilot plants described above show

that Purolite was effective al removing 100% of the iron and 95% of the manganese for

slow rates of filtration in the range of 1 - 3 m/h. For elevated rates of filtration (7 (n/h),

iron removal was still nearly 100% but only up to 30% of manganese was eliminated.

2.2.4.2 Ferro/ite MC2

Ferrolite MC2 consists of a granular ceramic medium which, like Purolite, is coated with

Mn02. Thill (1995) showed Ferrolite MC2 to behave as weil as Purolite with respect to

• iron elimination, and al higher velocities of filtration (13 and 24 m/h). [t was observed
 15

•
that for iron removal to below the detection limi4 ORP was consistently greater than 500

mV. Ferrolite MC2 a1so performed weil for manganese removal at the lower flow rate

and removed up to 75% of the manganese at the higher flow rates. ln terms of

manganese eliminatio~ Thill (1995) showed that Ferrolite MC2 filtered approximately

280 L or 42 Llkg before requiring regeneration.

Thus these two materials seem to be promising alternatives to conventional

physico-chemical treatment. However, their reliability must he evaluated for a specific

application~ which forms the second part of this study.

2.3 BIOLOGICAL TREATMENT

For many years physico-chemical treatment was considered to be the only reliable

method to effectively cid drinking water of high irOD and manganese concentrations

(Mouchet~ 1992). In the United States, investigations of conventional treatment plants

revealed that up to 40% of installations were Dot eliminating enough iron (Robinson et

al.~ 1987; Andersen et al., 1973). Similarly in France in the 1970s, it was found tbat only

two-thirds of conventional plants were achieving sufficient removal of iron and

manganese (Mouchet, 1992). On the other band, certain installations were performing

adequately even if their operating conditions were not favourable to effective physico-

chemical elimination (Mouchet, 1992). In these latter cases, massive developments of

iron and manganese bacteria were discovered in the sludge from the backwash and

determined to be the cause of the satisfactory removal. Thus biological treatment,

previously thought to be ineffective, came to the forefront of new studies in the

elimination of iron and manganese (Hettler, 1982; Viswanathan et al., 1991).

•
 16

•
2.3.1 Iron bacteria

These microorganisms are ubiquitous in the environmen~ being round not only in ground

water, water treatment facilities and raw and treated water canals, but aIso in spring

water, marshes, ponds, lake sedimen~ mines and various soils (Degrémont, 1991).

Although many iron bacteria are also associated with manganese, the present work is

concemed only with iron removal by biological means. Thus, the bacterial characteristies

discussed will foeus primarily on iron bacteria. For simplieity, the designation "iron

bacteria" encompasses the bacteria whieh MaY elsewhere be differentiated as either iron-

oxidizing or iron-aeeumulating baeteria (Ehrlich, 1990).

Table 1 indieates the bacteria of interest, which are sessile (flXed growth)

microorganisms. They include filamentous ('sheathed') baeteria (i.e. Leptothrix), stalked

bacteria (Le. Ga/lionel/a) and eneapsulated bacteria (Le. Sidercapsa).

Table 1. Some iron bacteria.

Order Fa ID ily Genus Observations

Chlamydobacteriaceae Leplolhrix. Sphaerolilus Filamentous; FelMn
Crenothricaceae Crenothrix. Clonothrix deposits; heterotrophic or
Chlamydobacteriales facultatively alltotrophic
Siderocapsaceae Siderocapsa. Siderobacter. Encapsu1ated;
Sideromonas. Naumanniella. Fe/Mn deposits;
Ochrobium. Sidercoccus heterotrophic
Caulobacteriales· GaIlioneIIaceae Gallionella Stalked; Fe deposits;
autotrophic

•
 17

In general~ iron bacteria can he distinguished by their morphology (Degrémon~

• 1991). While the filamentous and staIked bacteria are easily identified onder microscope~

the Siderocapsaceae family of bacteria are difficult to detect upon direct examination due

ta their rads or cocci ceUs which become covered in iron precipitates (Mouchet~ 1992).

Figures 5 and 6 show the characteristic forms of species Leptothrix ochracea and

Gallionellaferruginea~respectively. Figure 7 presents the typical Sidercapsa treubii.

Leptothrix spp. take the forro of small rods in chain-like formation (a trichome).

enclosed in a sheath in which iron precipitates are deposited (Mulder and Deinem~

1981). Gallionella consists of a vibrioid (bean-shaped) cell from which a fibrillar stalk is

excreted on the concave side. The long and twisted stalk becomes heavily encrusted with

ferric hydroxide and is the most distinct feature in Figure 6 (Hane~ 1981a).

The genus Metallogenium, of most significance in the oxidation and accumulation

of manganese~ is also capable of accumulating iron oxides, which make it relevant in the

present study (Zav~ 1981). Sorne authors have linked Metallogenium with the

Siderocapsaceae family (Mouchet, (992) and others with the appendaged (stalked) and

budding bacteria (Staley et al., (981). At a certain stage Metallogenium is a coccoid~

from which cocci sprout by radial filaments, giving the organism a characteristic "star"

shape~ as shown in Figure 8. Deposits of manganese over rime lead ta encrustation of the

colony~ which cao appear as an amorphous lump (Zavarin, 1981). Like Siderocapsaceae

microorganisms~ Metallogenium spp. May resemble sorne precipitates~ making them

difficult to clearly identify (Ehrlich, 1990). Metaf/ogenium is reportedly heterotrophic,

but may aIso be facultatively autotrophic (Ehrlich, 1990).

•
 18

(a) (b)

Figure 5. LeplOlhrix ochracea (a) cells moving out ofsheaths and lorming nc\v sheaths

(b) broken old sheaths covered and impregnated with ferric hydroxide in slowly running

iron (ll}-containing soil extract (fronl ivlulder and Deinema. 1981). Bar = 10 ~lm.

Figure 6. Gallionella.!e,.,.uginea (a) cxtrenlely pure deposit in drainage line near the

groundwater table (b) heavy iron-coated stalks (from Hancrt. 1981 a).

•
 r9

Figure 7. Sidercapsa treuhii slide ongrowth in aquarium-natural water culture (a) single

cells at the initial phase of capsule formation (b) n'lo ceUs coated with slime and iron

particles Cc) slime capsule completely coating cell (from HanerL 1981b).

Figure 8. .Hetallogenium with buds and daughter colonies from lakcs (arrow) (x 10 0(0)

sho\'ling excretions of iron and manganese hydroxides lfrom Rhcinhcimcr. 19(2).

•
 20

• Nearly all the species described in Table 1 are found to accumulate both iron and

manganese (Han~

accumulate iron alone

1981b; Mulder and Deinema, 1981). Only Gaflione/la is thought to

(Hane~ 198Ia). The mechanisms of oxidation and/or

accumulation are discussed in the following section.

2.3.2 Me~haDism for bacterial iron uptake

Ali ifon bacteria are aerobic or microaerophilic.. meaning that they require oxygen; and

chemotrophs~ meaning they obtain requ>ed energy through enzymatic oxidation of

organic or inorganic matter (chemosynthesis) in their environment. The energy is

subsequently used to assimilate their source of carbon. which cao either come ftom

organic (heterotrophs) or inorganic (autotrophs) materia! (Droste~ (997).

The concept of autotrophy was first proposed by Winogradsky in 1888 (Starkey,

1945).. following studies on sulfur bacteria. Winogradsky concluded that iron bacteria

had similar physiology, using iron Ç)xidation as a source of energy. He proposed the

following mechanism: the iron bacterium oxidizes ferrous iron inside its ceU, ferric

hydrate is precipitated outside of the cell, and energy is used to assimilate the carbon in

carbon dioxide (Starkey, 1945).

As referenced in Viswanathan and Boettcher (1991), Temple and Colmer (1951)

accounted for the efficieney of carbon assimilation and the faet tbat iron oxidation does

not release much energy in their estimation that obligate aerobes require 577 moles of

ferrous iron per mole of carbon. This fmding may explain why biological iron removal

facilities tend to fimction well at high ftow rates (Mouche~ 1992).

•
 21

•
The microorganism Gallionella is thought to he the only strict autotroph (Wolfe~

1958; Rheinheimer7 1992)7 as mentioned in Table 1. Although enzymatic oxidation or

primary intracellular oxidation (Mouche~ 1992) of iron by Gallionella has not been

proven this hypothesis is supported by the fact that the iron bacterium has been
7

developed in ferrous-containing minerai media and is sensitive to some organic matter 7

except at very low concentrations (Starkey, 1945; Wolfe 1958).

7

The microorganisms of the Sphaerotilus-Leptothrix group, listed as heterotrophs

or facultative autotrophs, are also thought to oxidize reduced forms of iron (Mulder and

Deinem~ 1981). However, the oxidation is not universally considered enzymatic

because most of these iron bacteria develop in the pH range (6 - 8) in which

autooxidation of iron occurs (Mulder and Deinem~ 1981). It has been suggested that

oxidation and accumulation of minerai iron by autotrophs in this group are achieved with

extracellular polymers such as polysaccharides proteins or POlysaccharide-protein-lipid

7

complexes (Rheinheimer, 1992). The genus Gallionella is aIso thought to demonstrate

this mechanism, termed secondary extracellular oxidation by Mouchet (1992), where the

extracellular materiaI is produced in the fonn of the stalk which becomes encrusted with

iron precipitates.

In the presence of iron-organic complexes, the heterotrophs in the Sphaerotilus-

Leptothrix group and of the genera Sidercapsa, Naumanniella, Ochrobium and

Sidercoccus are capable of obtaining the bound organic matter, tbereby liberating the

minerai iron to autooxidation (Ehrlich, 1990; Mouchet 1992). As with the above case,
7

the mineraI iron is then accumulated through the use of extracellular polymers.

•
 22

A fourth possible explanation for iron precipitation in the biological environment

• has been put forth to he metabolically-produced hydrogen peroxide (H20:0 which acts as

a chemical oxidizing agent (Ehrlich, 1990). Two bacteria which have exhibited tms

mechanism in iron and/or manganese oxidation are: Leptothrix pseudo-ochracea and

Metallogenium (Zavarin, 1981; Ehrlich, 1990).

For all bacterial uptake ofiron, the precipitates are slightly hydrated iron oxides in

the forms of lepidocrocite, y-FeOOR, and goethite, a-FeOOH (Mouchet, 1992). These

crystalline compounds are much more compact than the precipitates formed during

chemical oxidation (ferrihydrite, 5F~03·9H20, and iron hydroxide, Fe(OH)3), leading to

the formation of a denser sludge in the biological process than in the physico-chemical

process (Mouchet, 1992). In the transition between the biological and the physico-

chemical environment, lepidocrocite is replaced by ferrihydrite (Vuorinen et al., 1988).

2.3.3 Growth conditions of iron bacteria

Numerous species of iron bacteria considered significant for drinking water treatment

have been identified in their naturaI habitats. These include the following fi 1amentous

species: Leptothrix lopholea, L. ochracea, L. pseudo-ochracea, L. discophora, L.

cholodnii, Sphaerotilus natans, Crenothrix polyspora, Clonothrix fusca, and Lieskeella

bifida (Mulder and Deinem~ 1981). Leptothrix ochracea and the sole stalked species

thought to exist, Gallionella férruginea, are considered the most prevalent iron bacteria

(Hanert, 1981a; Starkey, 1945).

The natural habitat of G. ferruginea is very pure~ iron-bearing water which

contains only trace organics (Hanert, 1981 a). From laboratory studies, a favourable

•
 23

•
environment has a slightly acidic pH (6.0 - 7.6) and an oxidation-reduction potential

between +200 and +320 mV (Wolfe" 1964; Hanert" 1981a). These conditions~ when

located on Figure 2" correspond with the zone of ferrous iron stability. Other parameters

for cultivation of G. jërruginea are dissolved oxygen content hetween 0.1 and 1 mgIL"

temperature between 8 and 16°C, ferrous iron content hetween 5 and 25 mgIL, dissolved

carbon dioxide content ~ 20 mgIL, and low organic matter content (Hanert" 1981a).

Similar to the Gallionella species" Leptothrix ochracea occurs in sIowly running,

iron-containing water which is poor in readily decomposable organic material (Mulder

and Deinema" 1981). As previously mentioned, the favourable pH range is between 6

and 8 for accumulation of iron. L. ochracea bas been round in temperatures ranging from

1 to 25°C, where the optimum is reported to he 15°C (Starkey, 1945).

The better known species of the Siderocapsaceae family include: Sidercapsa

treubii, S. major, and Naumanniella neustonica. Severa! Siderocapsaceae species have

been identified in deep wells and water pipes, such as Sidercapsa botryoides,

Naumanniella pygmaea. N. elliptica. N. minor, LV. catenata, Siderobacter graci/is, S.

brevis. and S. lalus (Hanert, 1981 b). Favourable conditions reported for these organisms

include low temperature (10 - 20°C), low dissolved oxygen concentration (0.1 - 1 mg/L)

and the presence of organic matter (Starkey, 1945; Hanert, 198Ib). Although no

optimum pH level is given for the deep weIl species, an optimum range of neutral to pH

8.7 is reported for organisms in soil and lake habitats and an optimum range of pH 6 to 7

is reported for Siderocapsaceae species in swamps, ditches, and stagnant pools (Hanert,

1981b).

•
 24

•
Two species of the genus Meta/logenium have been identified: M personatum and

M symbioticum (Zavarin~ 1981). These are reportedly found in places containing

manganese~ including lakes, sediments, and soils (Zav~ 1981). Typical parameters for

development include a pH range from 4.1 ta 8.9, Eh from +200 to +650 mV 9 manganese

concentration ~ 50 mgIL and oxygen concentration from 0.01 to 10 mg/L (Zavarin..

1981 ).

2.3.4 Full-sale application of biologieal iron removal

In current water treatment practice, bacterial germination on a sand tilter is accomplished

by artificial or natural means. Artificial germination introduces sludge from the

backwash of an operating biological tilter into the new tiller. A natura! germination is

carried out by adjusting the chemical parameters of the raw water to suit the needs of the

desired iron bacteria. If the raw water is suitable for biological iron removal, this practice

should take betweeo 2 to 5 days to complete.

Suitable raw water should cootain 00 material which is considered to be taxie ta

the sensitive bacterial speeies (Le., Gallionella). This can include hydrogen sulfide

(H2S)9 heavy metals or organic pollutants. Baas-Becking and colleagues have studied the

area of iron bacteria activity in terms ofaquatic pH and Eh conditions, as shown in Figure

9 (Wolfe 9 1964; Mouche~ 1992). The lower lintit in Figure 9 corresponds to rH = 149

where rH represents the negative logarithm of the hydrogen ion concentration producing

the existing redox conditions (Rheinheimer~ 1992). This cao be expressed by the

following equation:

•
 2S

•
Eh
rH=-+(2·pH) (10)
c

where Eh is expressed in millivolts (mV) and c is a constant equal to 0.029 mV- 1

(Degrémont, 1991).

900
/
Physico-chemical
800 oxidation of iron
Theoreticallimit /
Fe2+ IFe(OH)J
700

_ 500
>E
600
1 " OXI atlon

--
~ 400
" ofiron

""
Competition between
physico-chem icaI and

"" , biological oxidation

300
" "-
200 " ", ,
100
" ":,
0
5 5.5 6 6.5 7 7.5
" 8
pH

Figure 9. pH-Eh diagram ofiron bacteria activity (after Mouchet~ 1992).

•
 26

•
The upper limit is also indicated but seems ooly to be of importance when the pH

exceeds approximately 7.2, at which point biological action competes with physico-

chemical oxidation (Mouchet, 1992). It should he noted that the biological region

straddles the theoretical boundary between the region of ferrous ion stability and ferric

hydroxide precipitation. This is related to the fact that iron bacteria which rely on iron

oxidation as a source of energy are gradient organisms, meaning that they are not likely to

develop under strongly reducing or strongly oxidizing conditions, but rather at the point

where they have a source ofboth ferrous ions and air (Wolfe, 1964).

In general, a pH range from 6 to 8 is recommended and the corresponding Eh

range cao be detennined from Figure 9. Dissolved oxygen is a variable which is adjusted

per installation, according to th.e pH (increasing dissolved oxygen increases Eh) and the

needs of the aerobic or microaerophilic bacteria (Mouchet, 1992).

When the raw water is suitable for the development of iron bacteri~ biological

elimination of iron is the method of choice over physico-chemical treatment for several

reasons. Chemicai oxidants, coagulants, and flocculaots are not required and aeration can

be achieved with in-line air injection (Degrémont, 1991). A higher flow rate is possible

(20-40 m/h superficial velocity), meaning more throughput can he achieved (Degrémont,

1991). Because of the bacterial action in accumulating the iron oxides, sand with a larger

effective size (1.18 to 2.05 mm) cao be employed (Mouchet, 1992), which facilitates the

higher-flow operation. AIso, the overall space requirements for the plant are reduced

since oxidation and filtration stages are combined into a single reactor (Mouchet, 1992).

And finally, a higher removal capacity is realized. generally on the order of 2 to 5 kg per

• m 2 of tilter cross sectional area, (Mouchet, 1992).

 27

processes~

•
As with conventional filtered water is disinfected (usually with

chIorine) before discharge to the water distribution network and a backwash using raw

water and air scour is required periodically to regenerate the tilter.

2.4 OXIDATION-REDUCfION POTENTIAL

The concept of oxidation-reduction potential (ORP) cornes from the exchange of

electrons among chemical species in a given medium. These species can exist in either

reduced or oxidized form and are transformed from one to the other by either the

acquisition of electrons (reduction) or the loss of electrons (oxidation) in what are called

redox (oxidation-reduction) reactions. These reactions depend upon chemical conditions

of the system and the bacteria which mediate redox reactions.

Redox reactions occur in pairs such that the species beiog oxidized is the reduciog

agent and the species beiog reduced is the oxidizing agent. For example, the redox

reactions for the oxidation of iron and the reduction of oxygen are expressed as follows:

(1l)

(12)

Adding the above equations, balancing and canceling the electrons gives the

following reaction, a slightly modified version of reaction (1).

(13)

The capacity of a system to donate or receive electrons is expressed with ORP,

which is determined by the Nernst equation (Charpentier et al., 1987):

o RT [ox.] (14)
E,} =E +-Iog-[-]
lI
nF red.

•
 28

•
E,~ = redox potential measured by the electrode

EOu = standard potential of the system at equilibrium

R = ideal gas constant (8.314 J/K/mole)

T = absolute temperature (K)

n = number of electrons

F = Faraday's constant (96500 coulombs/mole)

[ox.J = concentration ofoxidized matter (mgIL)

[red.] = concentration of reduced matter (mgIL)

The measurement of redox potential requires a standard reference which is., by

convention, the universai hydrogen electrode. Because the hydrogen electrode is not

practical to use in most applications., another reference is used as the internai electrode

reference to the measuring electrode. However., the redox potential is always expressed

in tenns of the universaI reference, according to the following expression:

EH = E" +E rTf (15)

EH = potentiaI in terms of the universal hydrogen electrode

E" - potential in terms ofthe internai electrode (measured)

En! standard potential of the internai electrode

In biological wastewater treatment, the ORP-time profile has been correlated with

biological activity related to redox reactions (Wareham et al.., 1993). The relationship of

ORP to dissolved oxygen (DO) bas aIso been investigated as a way ta optimize aeration

of activated sludge reactors and energy savings (Charpentier et al., 1987; Heduit and

•
 29

•
Thevenot, (989). ORP has the advantage over the traditional dissolved oxygen indicator

in that ORP is still significant when DO levels approach zero (Charpentier et al., (987).

Clearly~ the above features of using ORP to monitor a water treatment system

may be applied in drinking water treatment. Also, ORP in conjunction with pH is crucial

in determining whether or not the requirements for bacterial development are met., and if

physico-chemical oxidation of iron will compete with bacterial oxidation~ as indicated in

Figure 9. Thus~ the application of ORP as a control tool in drinking water treatment

systems represents a promising area of study.

•
 30

•
3. METROnS AND MATERIALS

3.1 PRESENTATION OF THE SITES

Two similar sites of full-scale scale drinking water treatment facilities were selected for

the study. Both plants are equipped with two biological filters which operate in series to

carry out biological iron and manganese removal. The study was organized into severa!

phases at each site~ as depicted in the following table. Each site and its facilities will be

described initially~ followed by a discussion of the materials installed at each site and

methods used.

Table 2. Organization ofthe study at two sites.

1. Seignosse U. Septmonts

Pbase Date Phase Date

A. Primary filter June - July 1996 A. Biological pilot November 1996
B. Biological pilot Aug. - Oct. 1996 B. Catalytic pilots December 1996
C. Primary filter October 1996 - -

3.1.1 Seignosse

Seignosse is a seaside town in the southwest corner of the France where the plant

operates during the summer months on an intermittent basis, producing drinking water at

a rate of about 1000 ml Per day. The primary tilter bed measures 1 ID in depth~ has a

cross sectional area of 3.14 m 2 and consists of sand with an effective size of 1.18 mm. In

total~ the reactor stands 3 m in height. The second filter has similar dimensions to the

•
 31

•
primary tilter with the exception of a slightly larger cross sectional area which measures

4.52 m 2 • Figure 10 shows a schematic of the full-seale plant.

Caustic soda Caustic soda

injection injection

Polyphosp ales C lorine

Air injection
injection injection

Figure 10. Drinking water treatment facilities in Seignosse.

The raw water was analyzed in June and July to give the average parameters and

standard deviations which are summarized in Table 3. A complete set of biological and

chemical analyses were carried out on a regular basis by the local laboratory. A copy of

analyses perfonned in July 1996 on the raw water is contained in Appendix 1.

Before entering the primary filter~ as shown in Figure 10, the raw water was

injected with caustic soda solution and air to increase pH to within the neutral range and

to increase the DO content. The superficial velocity of filtration was approximately 30

m/h which corresponded to a flowrate of70 m 3/h. Average pressure drop across the filter

during a filtration cycle was approximately 20 kPa. Gauge pressure measured at the head

of the tilter varied between 470 and 500 kPa during the filtration cycle. A backwash of

•
 32

•
the primary filter usÏDg raw water and air scour was carried out after every 500 ml

(approximately 7 hours) offiltration.

Table 3. Seignosse mean raw water parameters ± standard deviation, June 1996.

Total iron (mg/L) 3.80 ± 0.19

Dissolved iron (mg/L) 3.61 ± 0.14
Total manganese (mg/L) 0.042 ± 0.002
Dissolved manganese (mg/L) 0.042 ± 0.001
pH 5.73 ± 0.03
Eh(mV) 288±7
Dissolved oxygen (mg/L) O.OO±O.OO

Air was injected a second time to the primary filtered water before it was

delivered to the second tilter. Pressure fluctuation during a cycle of the second filter was

similar to the first, though it filtered 2 500 ml per cycle. Thus a backwash of the second

tilter using raw water and air scour was carried out with every fifth regeneration of the

primary tilter.

At the exit of the second tilter, caustic soda was injected a second time, the water

was chlorinated and polyphosphates were added to inhibit corrosion. Finally, the treated

water was sent to the water tower (chateau d'eau) for distribution.

3.1.2 Septmonts

The water treatment plant in Septmonts functions on an intermittent basis (10 - 13 hours

per day) nearly year round to produce drinking water for the local community. The

• treatment plant in Septmonts aIso consists of two biological filters in series.

 33

•
Schematically the plant is similar to the facilities in Seignosse as represented in Figure

10. The research at this site was concemed essentially with studying biological and

catalytic treatment of the raw water on pilot scale. Therefore further details of the plant
t

itself are not presented. Raw water parameters are compiled in Table 4. The raw water

also contained 0.4 mgIL of ammonium and a low concentration ofhydrogen sulfide (0.01

mg/L).

Table 4. Septmonts mean raw water parameters ± standard deviatio~ November 1996.

Total iron (mgIL) 3.82 ± 0.82

Dissolved iron (mg/L) 3.47 ± 0.74
Total manganese (mgIL) 0.317 ± 0.019
pH 6.70 ± 0.06
Eh(mV) 174±27
Dissolved oxygen (mgIL) 0.16±0.11

3.2 MATERIALS

3.2.1 InstaDation on site at Seignosse

In arder to enable sampling of filtered water and measurement of chemical parameters, a

valve at the exit (base) of the primary tiller was tapped and used to continuaIly supply a

22 L open air tank: which overflowed to the drain. An automatic valve was used ta hait

flow ta the tank whenever the plant was not in operation.

The tank aIso served as a probe bath to measure parameters such as pH. Eh. and

•
DO. Polymetron pH and Eh probes were furnished by Zellweger Analytical. The ORP
 34

•
probe had a platinum ring electrode and contained an internai reference solution of

AgiAgCI. The internaI reference potential (E~f) of 198 mV was used to caIculate the

standard potential measurements. The oxygen probe utilized was the Oxi 323-8 by

WTW (Germany). Electrochemical measurements were related to a data acquisition

system through a central processor equipped with a 10 channel input/output (1/0)

multiplexer cardo

3.2..2 Biological pilot plant at Seignosse

Figure Il shows the schematic of the pilot constructed in August 1996. Sand of 1.35 mm

effective size filled the polyethylene column to a 1 m depth. In total, the pilot measured 3

m in height, with an inside bed cross sectional area of 0.005 m2 • Ali valves and nozzles

were ofPVC construction.

A sampling port on the raw water inlet line to the plant was tapped to supply the

pilot. An automatic valve was used to halt flow to and from the pilot when the plant was

not in operation. Compressed air from the plant was used for În-line air injection to the

pilot. Air and water flow for bvth normal operation and backwashing were controlled

using rotameters.

A 22 L open air tank was placed at the exit of the pilot to facilitate sampling and

measurement of pH, E~ and DO concentration. Continuous measurement of

electrochemical parameters was achieved with the same equipment described in the last

section.

•
 35

•
c
__- D . . c J o - - - -.. 9

A
I-Plant raw water tapped
2-Process water to pilot
6 3-Backwash water to pilot
4-Air injection to process
2 3 S-Air injection to backwash
6-Biological pilot plant
7-pH. Eh, DO probe bath
To 8-To data acquisition
plant 9-To drain

8
~

;11
1 1 1
9
Raw
water Process air

Figure Il. Biological pilot plant facilities in Seignosse.

Initial gauge pressure at the head of the tilter was regulated to 50 kPa using the

valve on the exit line of the tiller. Pressure drop across the tilter was calculated as the

difference in water level between the inlet of the column and the outlet (open air tank). A

difference in height of approximately 2.3 m gave an initial pressure drop of 23 kPa.

Superficial velocity was established at 45 m/h (225 Lib) and air injection equivalent to

0.6 mgIL DO in the filtered water. Raw water and air scour were used to perform a

backwash (valves A and 0 closed; valves B and C open) when the gauge pressure

increased above approximately 100 kPa or filter saturation was perceived.

•
 36

•
3.2.3 Biologieal pilot plant at SeptalOD"

An identica1 biologica1 pilot plant to the one in Seignosse was constructed in Septmonts

at the end of October 1996. However, the raw water was supplied by a separate pump

which was lowered 14 m (dynamic hydraulic level) ioto the well where the raw water for

the full-sca1e operation was being drawn. Compressed air was obtained from the plant

and was cootrolled using a separate rotameter. The same materials were used as in the

last two sections to continually measure pH, Eh, and DO concentration.

Initial gauge pressure was regulated to 50 kPa, resulting in a pressure drop of

approximately 26 kPa. Superficial velocity was set to 14 m/h (70 LIb) and air injection

equivalent to approximately 1.00 mg/L DO in the filtered water. Raw water tlow to the

pilot was controlled by adjusting the variable frequency on the pump. Raw water and air

scour were used to regenerate the tilter when the gauge pressure increased above

approximately 100 kPa or tilter saturation was detected.

3.2.4 Catalytic pilot plants at Septmonts

After having completed the study of biologica1 treatment of iron on pilot scale, the pilot

was dismantled, the sand removed and the interior rinsed. The polyethylene column was

then charged with Ferrolite Me2 to a 1.2 m depth. A second identical polyethylene

column was assembled and filled with Purolite to a 1.2 m depth. Both products were

black granular material having smaller effective sizes than the sand used previously in the

biological filter. Thill (1995) determined the effective size to be 0.52 mm and 0.27 mm

and the bulk density to he 1120 kg/m3 and 1370 kg/m3 for Ferrolite Me2 and Purolite~

• respectively.
 37

•
It was not possible for the two columns to operate simultaneously under pressure

as in the case of the biological pilots. However, it was desirable to he able to inject air ta

the raw water, and since the columns were open to the air al the top, it was necessary to

reverse the tlow with respect the previous pilot studies. Thus filtration took place with

raw water entering at the base of the column and flowing out through a valve near the top.

A schematic of the two catalytic pilots is shown in Figure 12.

....,-~:..-.. 5

2 I-Ferrolite MC2
Process air 2-Purolite
3-pH, Eh, DO probe bath
4-To data acquisition
S-To drain

Process
water
4
~
/1
/1 1

Figure 12. Catalytic pilot plant facilities in Septmonts.

Flow to both columns was modified using the adjustable frequency on the pump

and total flow was indicated by a single rotameter, as was the air injection to the raw

•
water. If bath pilots were operating simultaneously, the flowrate at the exit of one
 38

•
column would he measured with a graduated cylinder. Often one column or the other

would operate alone, since only one set of measuring probes were available during

experimentation. The hydraulic depth of each tilter bed was 1.08 m and 1.12 m for the

Purolile and Ferrolile MCl, respectively.

3.3 ANALYTICALMETHOOS

Throughout the various phases at each site, analytical methods and maintenance protocol

were consistent.

3.3.1 Electrocbemical measuremeDts

The pH probe was calibrated frequently using buffer solutions of pH 4.01 and pH 7.00.

The DO probe was equipped with its own calibration capsule which simulated an

atmosphere of air saturated with water. A comparison of the DO reading with literature

values for air at 100% relative humidity and the given temperature were then used to

determine the accuracy of the reading. If the DO measurement was not within ± 0.01

mg/L of the standard references, the probe required membrane or electrolyte replacement.

The ORP probe was verified periodically with 230 mV buffer solution and the

platinum ring was polished daily during the trials. Instrument literature indicated a

measuring precision of ± 10 mV.

3.3.2 SampliDg

Collection of water samples for chemical analysis was achieved with wide-mouthed

plastic bottles. AIl samples for chemical analysis had to be obtained quickly, but without

agitation (no mixing with air), and immediately analyzed to ensure sample integrity. This

•
 39

•
was especially true of raw water samples which were effectively free of oxygen. Thus~

upon mixing with air, there was a possibility ofaltering the iron (manganese) species due

to dissolution of oxygen and evacuation ofcarbon dioxide.

For the full-scale plant at Seignosse, raw water was collected from a valve on the

inlet to the primary tilter (before caustic and air addition). Similarly on the pilot plants,

raw water samples were taken from a valve on the inlet to the pilot before air injection.

The valve was connected to a hose or tubing which was placed al the bottom of the

sampling bottle. Sampling was complete once the volume was overtumed severa! times.

Filtered water samples were collected from the open-air tank. The sampling

bottle was tirst rinsed then filled with the desired sample.

Samples of raw water after the addition of caustic and air were taken from the

full-scale facilities al Seignosse in a sunilar fashion to collection of raw water samples.

Aerated raw water samples were also taken from the biological pilots through sampling

ports located above the tilter bed of the column.

During the study of the full-scale facilities in Seignosse, sorne samples of

secondary filtered water were also analyzed. These were taken from the laboratory sink

faucet during ofI-season (June and October 1996), as it was directly supplied with water

from the second tiUer. From JuIy to September, the tap water was chlorinated which

would have affected composition, thus secondary filtered water was obtained from an

upstream sampling port.

•
 40

•
3.3.3 Chemieal analyses

Analysis of total and dissolved iron and manganese were performed with the DR2000

spectrophotometer from RACH. Ail methods used were documented in the

accompanying instrument manual (HACH, 1992). AlI reageots were fumished by

HACH. The differentiation between total and dissolved species was achieved by filtering

samples on individual disposable 0.45 JlIIl membranes by Millipore. Samples with

concentrations exceeding the upper concentration limit for the iron determination method

were diluted 1:10 with demineralized water.

The HACH method 260 for analysis of iron gave readings with precision of ±

0.0027 mg/L and the method 290 for analysis of manganese indicated values with

precision of ± 0.0049 mgIL.

3.4 EXPERIMENTAL METBODS

3.4.1 Normal operation of the faeilities in Seignosse

The tirst sets of experiments were carried out between late May and early June on the

full-scale scale facilities in Seignosse. During this time, it was possible to manipulate the

start-up of plant operation in order to evaluate the effect of long periods of downtime on

the concentration of iron in the raw water. The evolution of residual iron in the filtered

water after the backwash was aIso studied to detennine "maturation time" or time until

the lowest residual iron concentration was achieved in the filtered water.

For the first of these objectives, the full-scale plant was stopped overnight then

put back into operation in the moming. During the ensuing cycle, both the raw water and

•
 41

•
filtered water were analyzed periodically for concentrations of total and dissolved

residual iron, and occasionally for manganese. Corresponding electrochemical

measurements (pH, Eh and 00 concentration) were also recorded.

To evaluate the maturation time, an automatic timer was used to start the plant up

during the night. Thus a backwash was effectuated automatically during the day and the

evolution of total and dissolved residual iTon over the length of the following cycle was

monitored while the corresponding electrochemical measurements of pH.. Eh and DO

concentration were ootOO.

A second set of experiments, a1so conceming normal operation of the full-scale

facilities, were carried out in July. At this lime the plant was operating intermittently,

based on a signal indicating the level in the reservoÎr. During this peak period, the plant

was found to he operating up to 14 hours over a 24 hour period or !wo hours at a time

with approximately one hour of downtime between times of operation. Residual iron

concentration and electrochemical measurements were recorded during times of

operation.

3.4.2 AeratioD-ORP trials

The aeration-ORP trials involved severa! aims, the tirst of which was to establish the

influence of dissolved oxygen on biological treatment of iron. Secondly, the relationships

between ORP and both DO concentration and residual iron concentration in the filtered

water were investigated. Lastly, ORP was evaluated as an iodicator of biological iron

removal from drinking water.

•
 42

•
Over the course of severa! days, air injection to the raw water was varied. After

each modification, the pH, Eh, DO concentration, total residual iron and dissolved

residual iron of the filtered water were systematically recorded.

These trials were carried out on the primary tilter of the full-scale facilities at

Seignosse in June. The air injection was adjusted one hour after the backwash had been

completed. Sampling and analysis were then performed every fifteen minutes for the first

two hours and every hour following up until 3 hours into the next cycle (approximate

length of one cycle was 7 hours). During each run, periodic samples of raw water were

also collected and analyzed for iron.

Aeration-ORP trials were perfonned on the biological pilots in Seignosse

(October 1996) and Septmonts (November 1996). Because the pilots tended to operate

for longer periods oftinte, sampling and recording of electrochemical measurements with

each adjustment of air were only perfonned until parameters and residual iron were

stable.

3.4.3 Catalytie trials

Initial operation of the catalytic pilot columns was aimed to remove the fme particulate

matter of the tilter bed, or "fines". This process was quite lengthy because of the reverse

flow configuration, and some fmes still escaped with the exiting tiltered water during

experimentation. Thus nearly all samples analyzed for iron and manganese were first

filtered to focus on evaluating the oxidation capacity of the catalytic materials. Each

material was tested at different velocities and various levels of DO concentration in the

•
fil tered water.
 43

• 4.1
4. RESULTS AND DISCUSSION FOR SEIGNOSSE

INmAL STUDY ON FULL SCALE

4.1.1 Normal operation

Raw water iron content and composition may he atTected by long periods of downtime.

One contributing factor is a low groundwater velocity which increases the contact time

bet\veen the \\-Clter and minerais containing iron. Also, when the pump is in operation, the

water table descends, Conning a vortex around the pump. The wetted minerais above the

vortex are exposed to air, resulting in the formation of precipitates. When pumping stops,

the water level retums to Donnai and precipitates are re-entrained in the groundwater.

These precipitates are taken up by the pump once operation resumes. Thus ovemight

stoppage is studied to evaluate the impact of these factors on the iron removal process.

After startup of the pump, a certain period of time is needed for the raw water

quality to become more or less stable. This is observed in Figure 13, showing the

evolution of raw water quality over a six hour period on two separate days after startup.

The mean values for the entire cycle of each run in Figure 13 are summarized in Table 5

with their standard deviations. The Mean values are a1so indicated for each nID not

including the frrst 100 minutes, where the iron content appears to he the most variable.

Table 5. Raw water quality evolution after startup.

Date Total irea co"tent (mWL)

Entire cycle Cycle les5 fint 100 miDutes

• 29/05/96
04/06/96
3.S8 ± 0.16
3.50 ± 0.09
3.S0 ± 0.03
3.52 ± 0.05
 44

• 4

-
~ 3.9
-+-29.05.96 ~04.06.96

-=
E
oS
3.8

-=...=
41
CJ
3.7

Q
CJ
= 3.6
=
Q
.: 3.5

-
ëii
Q
E-
3.4

3.3
0 50 100 150 200 250 300 350 400

Time (miD.)

Figure 13. Evolution of raw water quality with plant startup after 14 hour downtime.

~E 1.2

-
.s=
~29.05.96 -.-30.05.96 31.05.96 ~04.06.96

-
...
=
=
CJ
=
o
0.8
G.I

CJ
0.6
=
o
.!::
0.4

0.2

O+---+---+---+-~I---+---+---+---+----t--+---+---+-----+---+---t'-----i
o 50 100 150 200 250 300 350 400
Time(min.)

• Figure 14. Evolution of residual iron after ovemight stoppage and during tilter maturation.
 45

• The results from the filter maturation study are illustrated by the evolution oftotal

residual iron (TRI) in the tiltered water over the length of two filtration cycles (7 hours)

on two different dates (30/05/96 an~ 31/05/96)., shawn in Figure 14. 1be evolution of

TRI on 29/05/96 and 04/06/96 corresponding to the raw water analyses ftom Figure 13

(after ovemight stoppage) are also shown on the same plot. Figure 14 shows tbat the

evolution of TRI appears ta accur overall in the same manner; that is, after an initial

decrease. TRI increases steadily towards the end of the cycle for the four sets of data.

This observation was later thought to he attributed to physico-chemical oxidation which

was taking place in the primary tilter. This is further discussed in the next section.

The maturation lime is determined as the time required for TRI to attain a

minimum value. From Figure 14 the average maturation time was about 100 minutes.

ft should he noted that for the tilter maturation study., the total iron in the raw

water was found to he much more variable on 30/05/96 (total iron = 3.74 ± 0.19 mgIL)
than on 31/05/96 (total iron = 3.65 ± 0.06 mgIL). Despite the greater variation in raw

water quality on 30/05/96. the evolution of filtered water quality was essentially the same

for bath cycles. Thus it seems that th.: tre::;"~ent of iron was not affected by the observed

variations in raw water iton content.

Normal operation of the full-scale facilities was a1so studied in July to evaluate

the effect of intermittent operation on iron removal. The primary tilter was in operation

14 hours per 24 hour period with a low air injection (DO = 0.2 mgIL). After each brief

downtime., it was noticed that treatment appeared to recover immediately and tilter

•
maturation required approximately 30 minutes. This is illustrated in Table 6.
 46

• Time
Table 6. Primary filtèr inteimittent operatio~ July 1996.

(mia utes)
Residual iro. CODceDlntioD UpoD IUrtup (IIIIIL)
Alter iDteralitteDt iDterruptioD Alter b.ckwult
0 0.182 1.385
30 0.220 0.276
60 0.217 0.247
90 0.223 0.234
120 - - 0.235

4.1.2 Aeration-ORP trials

The aeration-ORP trials on the primary tilter evaluated dissolved oxygen levels varying

between approximately 0 and 7 mgIL.

Figure 15 shows the evolution of Eh and DO concentration for the trials carried

out on the primary tiller in June. With each modification of air injectio~ the ORP

measurement seemed to suggest a proPOrtional relationship throughout the cycle with the

DO concentration in the tiltered water.

Figure 16 presents the changes in total and dissolved residual iron concentrations

during the aeration-ORP trials~ where the nms shown correspond to the nIDS indicated in

Figure 15. By comparing the correSPQnding runs on Figure 15 and 16~ it is observed that

as DO increased in runs 1 through 5. the dissolved residual iron (DRI) content in the

fi 1tered water decreased nearly to zero.

A comparison of the two plots also shows that the TRI attained a maximum value

at the end of the tirst filtration cycle in each run (recall that each trial consisted of

• monitoring conditions from the point al which the air injection was modified to the end of
 47

• 600
Run 1 2 J 4 5 7
-'-Eh -0-00
8 9
7.00

6.00

-
500

5.00 ~
-
>
400
4.00
-•
E
ID
-S300 ~
.&: 0
~ 3.00
200

100
2.00

l.00
1
0 0.00
4-100-96 7-Iun-96 lo-Iun-96 IJ-lun-96 16-1un-96 19-Iun-96 22-100-96 2S-Iun-96
Date

Figure IS. Evolution ofDO and ORP during the aeration-QRP trials
on the primary tilter in June 1996.

Run 1 2 8 9

~E 1.20 -A-TRI~DRI

-
1:
.~ LOO
- .
ii
=
..
~
ii 0.80

lO.60
i
.fl
~ 0.40
..
~
1:
ii 0.20
~
r- 0.00+---
4-Iun-96 7-Iun-96 IO-Iun-96 I3-Iun-96 16-Iun-96 19-Iun-96 22-Iun-96 2S-Iun-96
Date

• Figure 16. Evolution oftotal and dissolved residual iron during the aeration-ORP trials.
 48

•
that cycle and up to 3 hours into the following cycle; or a total of approximately 9 hours

of analyses). These results correspond to previous observations of an increase in TRI

towards the end of the filtration cycle (Figure 14). The level of this maximum TRI
-
anained seems to he proportiona1 ta DO levels in each run. Thus, it appears that

increasingly poor retention was occ:urring as DO injection was increased.

It is postulated that tilter operation was relying on a combination of physico-

chemical and biological oxidation. Tbus increasing the DO injection increased the

proportion of physico-chemical oxidation taking place. And since the sand size in the

tilter (ES 1.18 mm) was suited ta biological as opPOsed to physical retention, poor

retention of precipitates was the observed result.

ft was possible to establish a relationship between DO and E~ as shown on Figure

17. A natura! logarithm correlation was detennined using Excel S.O software and

rendered the following relationship. with a correlation coefficient (R2) of 0.706.

Eh = 59.7·10&:[DO] + 414.7 (16)

It is notable that the above relationship resembles the form of the correlation established

between DO and Eh by Heduit and Thevenot (1989) in activated sludge reactors. Reduit

and Thevenot (1989) found a slope between 55 and 65 mVand an intercept of 410 mV

for sludge aerated for severa! hours without feeding (pH between 7 and 7.6).

The evolution of DR! with increasing DO is also shown in Figure 17. ft appears

that DR! tends to decrease ",ith increasing DO. an observation which supports the eartier

postulation of physico-chemical oxidation. [t should be noted that linear correlations

were investigated between DR! and DO as weil as between TRI and DO but gave poor

• ~

R- values. due to the spread in data throughout the filtration cycle.

 49

• 600
6DRI .Eh
0.160
6
500 • 0.140
-
~

400
0.120 ~

0.100
-
1:::
~
;;:-
-
~
e 300 6
O.OSO
S
oS
.!
}66~6~
~ A

-
0.060 ~
200 ~
>
A 0.040 i
100
• •
6
~t~A
6 A 6
A
A ~6
~Q
6
0.020 Q
'"

0 0.000
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00
DiDolved O~ICD (mgIL)

Figure 17. Relationship between 00 and bath ORP and DRI during the aeration-ORP trials

600

6 A 6
500
6 6
" r A Jt.6 A
400 66666 6 6
6
;;:- 6

- e
..c::
~
300
6 6
6

200

100

0
0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160

Dissolved raidual iron (mglL)

• Figure 18. Relationship between ORP and DR! during the aeration-ORP trials.
 50

• Finally~ a plot of Eh against DR! is presented in Figure 18,. where it is suggested

the ORP measurement decreases with increasing DR!. It should he noted tbat a linear

correlation between Eh and DRI yielded a poor correlation coefficient (R2) of 0.365. A

similar relationship investigated between Eh and TRI yielded no identifiable trends.

Generatly, the aeration-oRP trials revealed that the pH. Eh,. DO, TRI and DR!

rapidly achieved stable values after modification of the air injection. Overall,. the plant

was producing water which conformed to French regulation but the trials suggested that

physico-chemical oxidation was a major factor in iron removal. Also,. analyses of the

backwash sludge did not reveal the expected iron bacteria (i.e.,. Gallionella or Leptothrix).

Further analyses were canied out in July on the raw water after the addition of

caustic an~ air before it entered the primary filter (pre-filtered primary water). These

analyses are summarized in Table 7 along with Mean values for raw and primary filtered

water. The results support the presumption of physico-chemical oxidation by indicating

that 95 • 100% of total iron was in dissolved fonn in the ra'W' water and only 25 - 28%

\vas dissolved after the addition of caustic and air (before filtration).

Table 7. Water quality before a'1d after primary filter~ June and July 1996.

Water quality Mean values * Std. Dev.

parameters Raw Pre-raltered primary Prima." filtered
Total iron (mgIL) 3.80 ± 0.19 4.82 ± 0.38 0.339 ± 0.277
Dissolved iron (mgIL) 3.26 ± 0.14 1.26 ± 0.04 0.130 ± 0.078
pH 5.73 ± 0.03 7.30 ± 0.12 6.91 ± 0.06
Eh (mV) 288±7 157 ± 23 349 ±60

• Dissolved oxygen (mgIL) 0.00 ± 0.00 1.06 ± 0.09 0.40 ± 0.84

 51

•
The above conditions of pH and Eh are placed on the pH-Eh diagram showing

iron bacteria activity (Figure 9) and presented in Figure 19 to demonstrate that the raw

water parameters were clearly in the region of iron bacteria activity. Points A, B, and C

correspond to the raw water, the pre-fiitered primary water and the primary filtered water,

respectively.

900
/
Physico-chemical
800 oxidation of iron
Theoreticallimit ,/
Fe2+lFe(OH))
700

600
1...... ,
Biological
oxidation
-500
>
-e
..c
,
,ofiron
Competition between
'-l 400
"" , c
physico-chemical and
biologicaloxidation

300 "
"" , ,
200

". ,
100

o~_

5
..........--+-..........--'-~~""""--"'~+----~~~~--'-----+~~~
5.5 6 6.S 7 7.S
" 8
pH

• Figure 19. pH-Eh diagram indicating water conditions before

and after primary tilter (after Mouchet, 1992).
 52

•
Thus it was thought tbat iron bacteria which are difficult to detect upon direct

examination (Hanert, 1981b) were perhaps present~ such as those from the

Siderocapsaceae family. Therefore, the plant was operating in the transition area between

biological and physico-chemical regimes. Because the operating conditions of the

facilities could not he a1tered during the peak. summer months, the pilot study was

undertaken 00 site to verify biological treatment feasibility and if successful, to repeat the

aeration-ORP trials on pilot scale.

From Figure 19, it is noted that as the pH increased from 5.7 to 7.3 with the

addition of caustic soda, the Eh decreased frOID 290 to 160 ID V. This is expected as pH

and Eh have been shown to exhibit an inverse relationship (Heduit and Thevenot, 1989).

Evidently, the change in iron concentration of the water was significant eoough aftec

passing through the tilter that this ultimately produced an increase in Eh to approximately

350 mV.

Theoretically, the pH should not change across the tiller. It is possible that the

sampling point was too close to the point of caustic injection. Nevertheless, the aeration-

ORP trials suggested that at least sorne physico-chemical oxidation was taking place and

thus, the results could not he interpreted for evaluating ORP as an indicator in biological

iron treatment.

•
 53

•
4.2 BIOLOGICAL PILOT STUDY

4.2.1 Startup of pDot tilter

After 16 days of operation onder seemingly favourable conditions for a natural

germination, the pilot was still not eliminating any iron. Because the raw water pH was

slightly lower than recommended (Mouchet, 1992), a second trial to instigate growth was

attempted by temporarily supplying the pilot with pre-filtered water from the plant (see

Table 7 for chemical parameters). After filtering the higher pH water for approximately

20 hours, the pilot visibly contained a considerable amount of red precipitate, and the

gauge pressure at the head of the column had increased to approximately 80 kPa (increase

in pressure drop of 30 kPa). The tilter was regenerated and the supply line changed back

to the raw water.

The pilot was put back ioto operation with a reduced flowrate of 150 Lib, or a

superficial velocity of 30 mIh, and column pressure regulated to 50 kPa. The evolution of

TRI and ORP in the filtered water from the pilot, shown in Figure 20, took place over 122

hours of operation from September 9 to October 7, 1996. It appears from Figure 20 that a

correlation exists between ORP and TRI and this is further discussed with results from

the aeration-ORP trials below. Total residual iron was used, instead of dissolved residual

iron, because it was observed that total iron was essentially in dissolved form at the exit

of the filter and there was more data available of TRI than DR!. It should be noted that

the pH of the filtered water was virtually unchanged from that of the raw water (pH 5.7).

•
 54

• 600 A
A
.Eh A TRI 3

a
500
•·i • •• • 2.5 :8
-
•••
Z!

-
>5
-
.c
400

300
• •
A

.1
'If-. 2
=
I:l~
1.5 = u
.: E
I:l
u
CI
I:l
CI _

~
1;-
200 66 1 =
~

100
6
..
.;;
u
6,. 0.5 i
A
66
=
E-
0 e
0
7-Sep-96 14-Sep-96 21-Sep-96 28-Sep-96 S-Oct-96 12-0ct-96
Date

Figure 20. Startup ofbiological iron elimination on pilot scaIe.

4.2.2 Pilot tilter ba<=kwash

Essentially ~ the backwash is expected to preserve the biomass while removing the sludge

which progressively clogs the filter. During baclewashing of the pilot column, this was

gauged by visual inspection of the backwash water which should not become clear but

rather retain a reddish-orange tint.

Two tilter regenerations on pilot scale were carried out in September L996 during

startup~ generally for the purposes of collecting samples of baclewash sludge. The

backwash at this time consisted of both water and air sent countercurrently (wash)

through the column, followed by water aIone (rinse). In October it was found that the

•
 55

•
rinse portion of the backwash was canied out cocurrently on the full-scale filters and so

the protocol for the pilot tiller was changed ta correspond to the full-scale facilities.

The flowrate and duration of the pilot backwash were also modified when it

appeared that the tilter required a lengthy maturation lime (2 - 3 hours) before achieving

relatively consistent residual iron concentration.

Samples of sludge collected in September and Oetober from the pilot did not

reveal the MOst common iron bacteri~ Gal/ioneHa or Leptothrix, nor iron bacteria of the

genus Sidercapsa. It should he noted that the Siderocapsaceae family of iron bacteria

encompasses many genera which are said to he diffieult 10 identify on direct examination.

These baeteria are generally said to develop in environments where there is a source of

organie carbon and a pH near neutral, which was not the case for the raw water at

Seignosse. However, sorne Siderocapsaceae species have been observed in lower pH

environments and exhibiting facultative autotrophy, which would malee them potentiaI

candidates for the present system. The development conditions of the genus

Metallogenium aIso malee it a possibility. In any case, biological treatment cao be

strongly postulated as severa! performance factors attested to biological and not physico-

chemical oxidation.

Firstly, the pH and Eh of the raw and filtered water were clearly in the region of

iron bacteria activity when placed on Figure 9 or 19 (pH 5.7 was unchanged, Eh increased

from 290 mV to 510 m V). Secondly, the velocity of filtration was typical for biological

iron removal (30 m/h), whereas physico-chemical processes are usually restricted to a

maximum rate of 15 mIh (Mouchet~ 1992). AIso, the effective size of sand in the pilot

• was 1.35 mm whereas physico-chemical oxidation requires sand between 0.5 and 1 mm
 56

•
in size (Degrémont, 1991). And lastly~ the residual iron was essentially ail in dissolved

fonn while previously there was a discrepancy between the total and dissolved forms of

residual iron during the aeration-ORP trials on the primary tilter.

4.2.3 Aeration-ORP trials

Once desirable conditions and satisfactory iron removal were established in the pilot~ it

was possible to carry out the aeration trials in early October. Because the gauge pressure

in the column was ooly slightly elevated (70 kPa)~ the trials were initiated without first

regenerating the filter. Figure 21 shows the evolution of DO~ TRI and Eh over the 7 runs.

Despite the large variations in DO, the removal of iron was practically constant,

as was the Eh, with the exception of run 2 where air injection was completely halted. At

the very end of the trials, TRI increased suddenly from 0.005 to 0.330 mgIL,

accompanied by a decrease in ORP from 500 to 390 OlV. The tilter had achieved

saturation and a backwash was initiated. Thus the dramatic drop in ORP signaled the

need for filter regeneration.

The trials revealed that an effective removal of iron can he achieved with a low

level of dissolved oxygen (0.2 - 0.6 mgIL). A comparison of runs 1 and 2 shows the

value of using ORP in contrast to DO to evaluate process reliability. In both cases, the

DO levels are similar~ 0.20 and 0.16 mgIL, respectively. However, the ORP is much

higher in run l, corresponding to excellent removal of iron. [t should be noted that

analyses indicated that TRI in the filtered water was essentially in dissolved fonn.

•
 57

• 600
Run 1 2
___ Eh - . - TRI
3
~

-r----~.----_:IJt_--w---------r---.....,._----
4
DO
5 6
4

3.5
500

38~
400
2.51 E
.-
.
$'
.s 300
-=
riIii1
2
.- .. -
1:
0·-
g
1.5= ~
200 "CI ~

100
1 !S
0.5
Oi--__,.-.;~_,....-....:;a. ...........,.--.........-.,...___.~. .~_._-.........._+O
7-Oct-96 8-Oct-96 9-OCt-96 IO-Oct-96 II-OCt-96 12-Oct-96
Date

Figure 21. Results from the aeration-ORP trials on pilot-scale in October 1996.

600
o Startup data
• Trials data

500

$' D

-E 400
..c
CiIi:l •
300
0 • • •
200
0 0.5 1.5 2 2.5 3 3.5 4
Total raidual iroD conceDtntion (lDgIL)

• Figure 22. Correlation between ORP and TRI during start-up and trials on pilot-scale.
 58

Severa! authors attest to the fact that physico-ehemical oxidation of iron cannot

• take place at an appreciable rate when the pH is below neutral (Ehrlich, 1990; Mouchet,

1992). The implication of this statement for the present research is twofold. Firstly, it

supports the presumption of biological action. Secondly, because the pH of the raw water

studied was low, it would he theoretica1ly possible to further increase the air injection

without greatly influencing physico-ehemical oxidation. Thus, the dissolved oxygen

content could he increased to saturation level and there would he little to no competition

for ferrous ions hetween biological and physico-ehemical oxidation. Conversely, in

biological processes where the raw water has a pH in the neutral range, increasing the DO

level would possibly provoke sorne physico-chemical oxidatio~ such as was seen during

the aeration-ORP trials on the primary tilter in June. Thus, evaluating the effect of DO

on biological iron treatment and controlling the air injection with ORP May he of greater

importance for such facilities, in order to favour biological action.

As with the pilot start-up shown in Figure 20, it was possible to see a correlation

between ORP and TRI concentration for the aeration-ORP trials shown in Figure 21. The

data from Figure 21 was combined with the data from stan-up of the pilot (Figure 20) to

demonstrate the relationship between ORP and TRI as shown in Figure 22.

A non-linear regression between ORP and TRI was detennined using Sigmaplot

4.0 software. The software rendered a correlation coefficient (R2) of 0.8848. The

relationship was modified to solve for total residual iron and is shown as follows.

Eh-a
when 300 < Eh < 470 mV (17)
[nu] = b - c(Eh)

•
 59

where [TRI] = total residual iron concentration in tiltered water (mgIL)

• Eh = standard oxidation-reduction potential of filtered water

(mV)

a = constant, 525.603 m Y

b = constant, 574.893 mY·L/mg

c = constant, 2.157 L!mg.

The regression model May he applied within upper and lower Iimits to predict TRI

concentration for a given ORP measurement. The correlation appears to become

ineffective at ORP levels greater than approximately 470 mV, where experimental results

showed iron concentrations to be consistently below regulation limits. ln addition, for

values of ORP below approximately 300 mV, total iron levels were found to be

consistently greater than or equal to 3 mgIL. Thus, for the present application of

biological elimination of iron from drinking water, equation 17 can be used to predict

total residual iron concentration for ORP levels between 300 and 470 mV.

4.3 SECOND STUDY ON FULL SCALE

4.3.1 Primary filter startup witbout pH adjustment

Once the plant was no longer producing water for the water distribution network, it was

possible to stop the injection of caustic soda (October 1996) in an attempt to develop the

same perfonnance as on the pilot. As soon as the tilter began to operate without the pH

adjustment, the residual iron content in the tiltered water increased to approximately 1.10

mg/L and remained on the order of 1 mgIL for several cycles. During this time, it was

• noticed that pressure drop across the tilter did not increase by more than 10 kPa
 60

•
throughout a cycle~ 50 it was proposed that the length of the cycle he increased from 500

m J to 1000 mJ.

In addition~ visual inspection Îndicated that the backwash was not preserving

enough of the bacterial mass to accomplish sufficient removal (backwash water was

becoming clear). Thus, the duration of the air and water portion of the backwash was

shortened from 5 minutes to 3 minutes.

These changes are shown in Figure 23 which encompasses approximately 169

hours of tilter oPeration (3 cycles of 500 mJ and 10 cycles of 1000 mJ) with a DO

concentration of 1.0 mgIL. Towards the end ofthis period~ primary filtered water had a

consistent TRI concentration of 0.025 ± 0.007 mgIL and an Eh of 508 ± 9 mV.

oEh 6 TRI First backwash

lit" after 1000 ml 1.6
600

500
•e
6
•
•• Backwash
IfJ EJ~
1.4.2

1.2
-..•
1:

ëal

t~ .\~ II" shortened

-
~
~

- 400 1:
1 el
~
>E
- ft a I:~
e 'ail
- -E
300 0.8
.c -
~
200 • 0.6 ;
"'C

100
•• 6
• 6 • •
0.4

0.2
.=
'E
E-
.~ 0
0
5-0ct-96 12-Oct-96 19-0ct-96 26-0ct-96
28-Sep-96
Date

Figure 23. Primary tilter start-up without pH adjustment in October 1996.

•
 61

4.3.2 Biologiea) treatllleDt hypothesis

• Sïnce the pilot column and the primary tilter were operating under approximately the

same conditions in October, it was assumed that the same treatment mechanism was

taking place in each. As discussed earlier, analysis of pilot sludge did not reveal the iron

bacteria responsible for iron elimination though severa! factors supported biological

activity. On-going study of the site is attempting to identify the micro-organisms present

in the primary filter.

Another factor which supports the postulation of biological activity is the filter

retention capacity. The iron retained during one filtration cycle is calculated with the

following equation:

[11] - [nu] x V = CR (18)

SA

where [TI] = total iron concentration in the raw water (g1m3)

[TRI] = total residual iron concentration in the filtered water (g/m3 )

SA = filter cross sectional area (m2)

V = volume of water filtered in one cycle (m3)

CR = tilter retention capacity (glm2 ).

Given that SA is 3.14 m 2 and V is 1000 ml; and assuming a raw water iron

concentration of 3.8 mgIL and a filtered water total residual iron concentration of 0.03

mg/L, CR is calculated to be 1.20 kglm2 • This value corresponds closely to the typical

value of 1.25 kg/m 2 reported by Mouchet (1992) for biological filters having a superficial

velocity of30 mIh and containing sand of 1.18 mm effective size.

•
 62

•
4.3.3 Summary OD primary tilter operatioD

Upon initial startup in 19907 the full-scale facilities were not immediately successful at

iron removal. In fac~ after severa! months of operation with ooly air injection, the

primary tilter was essentially not removing any ïron. Finally, following a study on the

site by P. Mouchet of Degrémont in March 1991 the pH adjustment was put into place

(Mouchet, 1991). These results were communicated in an unpublished internaI report

which is contained in Appendix 2. Because the tilter began to remove iron aImost

immediately, the pH adjustment by caustic soda was deemed necessary for reliable

operation. Analyses of the backwash did not revea1 the typical bacteria such as

Gallionella or Leptothrix at the time, and Mr. Mouchet suggested that bacteria from the

Siderocapsaceae family were possibly presen~ though this was never confirmed.

It should he recalled that analyses of sludge from the primary filter in June did not

reveal any typical iron bacteria. However, the trials in June suggested that both

biological and physico-chemical oxidation were taking place. Thus it is possible that

sorne bacteria (i.e., from the Siderocapsaceae family) were present at the time, given the

higher pH environment suited to both their development and the auto-oxidation of iron.

Furthermore, poor fllter operation (residual iran concentration above the French nOIm of

0.2 mgIL) could have been due to the shorter filtration cycles and longer backwashes

which hindered bacterial development.

A comparison of previous and current operating conditions on the primary tilter

and iron treatment effected are shown in Table 8. As shown from the initial study on the

full-scale plant in June 1996, the primary tilter rarely eliminated more than 90 - 95% of

•
 63

•
raw water total iron and tiltered water usually contained greater than 0.20 mgIL of iron

(French nonn in drinking water). It is evident that the current operating conditions

without the pH adjustment have led to more effective iron removal (98 - 99%) and

savings in backwash water. A pH adjustment is still necessary in the rmal product to

increase pH to approximately 7.6, the standard for potable water.

Table 8. Summary of previous and current operating conditions on the primary tiIter.

Parameter June 1996 October 1996

Volume tiltered per cycle 500m"' 1000 m"'
Length ofhackwash (air+water) 5 min. 3 min.
Filtered water pH 6.9 5.7
Filtered water Eh 349mV 500mV
Raw water total iron 3.80 mg/L 3.80 mgIL
Total residual iron 0.339 mg/L 0.028 mgIL
Dissolved residual iron 0.130 mgIL 0.016 mg/L
% Total Iron Removed 91.1 % 99.3 %

The time of maturation after a backwash was evaluated a final time for the

primary filter. Figure 24 shows the TRI to he weil beneath the regulation standard for

iron (0.20 mg/L) at the beginning of each of 3 cycles. Maturation tinte was observed to

be approximately [5 minutes in each case. When compared with the maturation time of

100 minutes observed on the full-scale facilities in June 1996, it is evident that a more

effective backwash was taking place in October. Furthennore, it was observed that the

TRI was essentially constant at approximately 0.3 mgIL up to tilter regeneration. Thus

•
 64

no increase in TRI was seen towards the end of the cycle as compared to the primary

• filter results from June (recall Figure 14) where TRI attained up to L.3 mg/L near the end

of the cycle.

(t should he noted that the secondary tilter, intended for manganese removal, was

not removing any manganese throughout the duration of the study. This was oot of

concern at the beginning of the summer, for the concentration of manganese in the raw

water (Table 3) was consistently around 0.04 mgIL (below the French norm of 0.05

mg/L). However, over the course of the research, changes were noticed in raw water

quality, notably in the manganese content.

Figure 25 indicates the manganese content increased significantly to 0.055 mg/L

in September 1996. Still, at this lime, the secondary fllter was not eliminating any

manganese. The raw water quality retumed to 0.04 mg/L shortly afterwards but the

problem remained ofhow to eliminate manganese in the case of future fluctuations in raw

water quality. The operating modifications effected on the primary filter in October have

enabled a satisfactory elimination of iron to take place, thus making the secondary tiller

available for changes in operating conditions to induce manganese removaL

•
 65

• ~ ~21.10.96

-
~
0.18
0.16
18.10.96 -6-22.10.96

--
E
.s=
0.14

--.=.
= 0.1
41
Col
0.12

= 0.08
Q
Col

.::=
Q 0.06
~ 0.04
= 0.02
"'0
";j
41
~ 0
0 15 30 45 60 75 90
Time after backwasb (miD.)

Figure 24. Primary tilter maturation after 3 different backwashes, October 1996.

~Fe~Mn
4.5 0.07
_ 4

r 3 .5
1:::3

-- -
0.0l§
e
o.oi
.~
-=
~
3
=_
0
Col

l:: 2.5
0.01 ~
~

~
Q
l;J

a 1.5
2
0.03=
.-
c:
M
E

.:: e
-
-=
Q
1 ...
o.o~
~ 0.5

o +-----~I__----_+_-----_+_----___l~----_+ 0.01
35187 35224 35261 35298 35335 35372
Date

• Figure 25. Iron and manganese content ofraw water at Seignosse du ring study.
 66

•
5. RESULTS AND DISCUSSION FOR SEPTMONTS

5.1 BIOLOGICAL PILOT STUDY

5.1.1 Startup of pilot tilter

Upon arrivai at the site of the biological pilot study in Septmonts, the pilot was

operationai and appeared to he functioning under biological conditions. Previous

analysis of the sludge generated in the primary filter of the full·scaie plant had revealed

the common iron bacteria Gallionella. Sïnce the pilot was essentially oPerating onder the

same conditions (air injection but no pH adjustment of raw water), it followed that the

same activity was present in the pilot.

However, it was soon noticed that the column was not operating on a continuous

basis, which led ta variations in DO, TRI and DR! levels in the filtered water. The use of

a separate pump to supply the eolumn was the source of the problem, as each time the

full-seale faeilities began to operate, the water table level would descend from 9 m (statie

level) to 14 m (dynamie level) 50 as to result in no flow to the pilot. This occurred

despite the fact that the pump supplying the pilot was situated at a depth of 14 m.

The use of a separate pump to supply the pilot meant that the pilot received raw

water with variable total and dissolved iron content at the point when the full-scaIe

facilities shut down. At the moment of plant downtime, the total iron content in the raw

water (delivered ta the pilot) inereased ta, for instance, 5.00 mgIL on November 20,

1996, while the dissolved iron content increased to 4.65 mg/L. Analyses performed one

hour after plant shutdown on November 28th indicated raw water total iron concentration

•
 67

•
had decreased ta about J .24 mgIL. Manganese content of the raw water was always in

dissoIved form.

The supply pump for the pilot column was lowered from its initial depth of about

14 m to 23 m in an attempt ta rectify the problem. When this action failed to have an

effect on pilot operation, a level indicator was instaUed on the open air tank at the exit of

the pilot column in an effort to regulate Oow from the column ta the tank. This, tao, did

not keep the column from emptying during times of plant operation.

The remaining solution was ta adjust the pump speed with each startup and

shutdown of plant operation, in order to maintain Oowrate to the column during

experimentation. At the end of daily experimentation, it was also necessary to shut off

the pump and close all valves exiting the column ta prevent variations upon the following

startup. Thus, it was possible ta carry out the aeration-ORP trials on the biological pilot

column.

5.1.2 Aeration-ORP trials

The aeration-ORP trials were initiated on the pilot during the month of November, once it

was decided ta regulate pilot operation manually. Flowrate was maintained at

approximateIy 70 Lib or 14 m/h, which corresponded with plant operation. The ensemble

of analyses performed are presented in Figure 26, showing the variation in Eh, DO and

TRI in the filtered water for 7 runs.

•
 68

• Run 1 2 3
___ Eh -e- DO -&- TRI
4 5 6 7
600 6

500 5_
~
400 4!.
~ =
41
~
!..300 3 ~
-=
~
=
200 2]
= lI'J

1=
lI'J

100

0 0
25-Nov-96 27-Nov-96 29-Nov-96 I-Oec-96 3-Dec-96 5-Dec-96
Date

Figure 26. Results from the aeratïon-ORP trials on pilot scale, November 1996.

600

500 ••
• •
~ • •
--=
E 400
~

300

200 -I----l~-_+--_+_-~I__-~--_+_--+__-_+--_+_-__I

o 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Total residual iron eoaeentration (mgIL)

• Figure 27. Relationship between ORP and TRI during aeration-ORP trials.
 69

•
Each adjustment of pilot operating conditions (with plant startup or shutdown)

still resulted in a minor upset to filtered water conditions. With the exception of the data

from the final run where the air injection was halted~ the concentration of TRI was

consistently low (0.013 ± 0.016 mgIL) for the range of DO levels observed; and the Mean

Eh value recorded was 542 ± 33 mV. In the last run where air injection was zero? the

Mean TRI was 3.58 ± 0.08 mg/L, as reflected by a lower value of Eh (202 ± 16 mY). It

was noticed that a similar relationship between Eh and TRI, as seen previously in Figure

22, likely also existed for the data at the present site. The resulting Eh and TRI values are

plotted in Figure 27. Unfortunately, there was a significant lack ofreliable Eh-TRI data

points from the startup of the biological pilot (for Eh between 200 and 400 MY). Thus a

good correlation between the two parameters could not he obtained.

It can he concluded from the above trials that Eh is a good indicator of iron

removal. An ORP reading of 440 mV or higher indicates the TRI concentration is less

than 0.1 mg/L. A1so~ a low level of dissolved oxygen concentration, on the order of 0.3

mgIL, is sufficient for effective removal ofirone The elimination of iron does not seem to

be hindered nor improved by increasing the DO level.

5.1.3 Summary on the biologieal pilot study at Septmonts

Biological treatment of Ïron at Septmonts bas revealed Many of the same trends observed

at the previous site. These include: a low level of dissolved oxygen is required to

accomplish effective iron removal, the measurement of DRP appears to be a reHable

indicator of iron concentration in the tiltered water~ and increases in DO level do not

•
improve nor hinder tilter operation.
 70

•
In view of the similar level of DO that was required at both Seignosse and

Septmonts, it is interesting to compare the two sites. A low level of DO (0.3 mg/L) was

sufficient for effective biological iron removal in two raw waters: one with a pH of 5.7

and an Eh of 290 mV; and one with apH of6.7 and an Eh of 174 mV. Given the pH ofa

system, a certain no concentration is required to increase the Eh to the level required for

bacterial development. Thus, based on the fact that in the lower pH environment

(Seignosse) the Eh was significantly higher, the same level of DO (0.3 mg/L) produced

essentially the same effect as in the higher pHllower Eh environment.

It can he recalled that while typical iron bacteria of the genus Gallionella were

confinned at Septmonts, the iron bacteria at Seignosse were not identified. And with the

exception of the pH and Eh of the raw water and rates of filtration, the results from the

two sites were quite similar, as shown in Table 9.

Table 9. Results from trials on biological pilots at Seignosse and Septmonts.

Parameters Seignosse Septmonts

Filtered water pH 5.7 6.7
Filtered water Eh 522mV 538mV
Raw water total iron 3.80 mgIL 3.82 mgIL
Total residual iron 0.017mg/L 0.014 mg/L
Dissolved residual iron 0.015 mgIL 0.014 mg/L
% Total iron removed 99.6% 99.6%
Average flow rate 150 L/h 70 Lib

Total time of filtration 34 hours 36 hours

FiLter retention capacity 3.9 kglm~ 1.9 kg/m~

•
 71

•
The results from the aeration-ORP study at each site (not including runs when air

injection was zero) are used to compare the two sites, since regular analyses were carried

out for the trials and reliable average values could he obtained. It is observed that the

retention capacity (recall equation 18) was twice as high at Seignosse than at Septmonts.

However, this appears to he due to the difference in flow rates at the two sites, which are

aIse related by a factor of about two. Other parameters relevant to equation 18 are

approximately equivalent: pilot cross-sectionai area (0.005 m2), total time of filtration,

and average difference between inlet and outlet concentrations of iron ([TI] - [TRI]). The

retention capacities shown in Table 9 aIso represent values typical to bioLogical iron

treatment [2 to 5 kg/m2 (Mouchet, 1992)].

5.2 CATALYTIC PILOT STUDY

As previously mentioned, elimjnation of fines on the two materials was the tirst required

step before experimentation. This was a lengthy process in the case of Purolite and

analyses began even though Îme black particles were still escaping with the filtered

water.

5.2.1 Ferrolite Me2

After two hours of fines removal, normal operation commenced on the Ferrolite Me2

column at a medium flowrate (12 m1h) and low air injection. The tirst run shown in

Table 10, along with raw water conditions, indicates excellent removal of both total and

dissolved iron and sorne removal of manganese. After 5 hours of operation at the initial

conditions, the manganese concentration increased and soon exceeded the raw water

• content. Thus the catalytic material seemed to be interfering with the manganese content
 72

•
of the filtered water. Elimination of iron continued to he satisfactory and appeared to he

indicated by a high ORP (510 m V).

Table 10. Startup of the Ferrolite MC2 pilot: Rawand filtered water parameters.

Parameten MeaD values :1: Std.Dev.

Rawwater Filtered water
pH 6.70 ± 0.06 6.60 ± 0.05
Eh(mV) 174±27 510 ± 24
Dissolved oxygen (mgIL) 0.16 ± 0.11 0.26 ± 0.04
Total iron (mg/L) 3.82 ± 0.82 0.010 ± 0.005
Dissolved iron (mgIL) 3.47 ± 0.74 0.004 ± 0.003
Total manganese (mgIL) 0.317 ± 0.019 0.132 ± 0.018

The ensuing experiments involved analyses at various combinations of high and

low air injections, coupled with high and low velocities of filtration. The results are

summarized in Table Il. Dissolved residual iron (DRI) and dissolved residual

manganese (DRM) are indicated to evaluate the oxidation capacity of the material and

these values cao he compared with raw water concentrations shown in Table 10. It

should be noted that total manganese in the raw water was completely in dissolved fonn.

Even though Fe"olite MC2 seemed to eliminate total iron quite weil in the first

run of Table Il (which corresponds ta the startup conditions shown in Table 10), as saon

as the air injection was increased (run 3), air bubbles significantly disturbed the filter bed,

causing a surge of fine particles with the flow exiting the column. Therefore, the

dissolved ieon (or manganese) gave the best indications of oxidation occurring. An OR?

•
 73

reading above 500 mV was a good indication of iron oxidation taking place al nearly

• 100% efficiency. In total, the Ferro/ile MC2 pilot filtered 1457 L ofraw water.

Table Il. Iron and manganese oxidation on Ferro/ite MC2.

RUB Velocity Dissolved oxygen DRI DRM Eh

(mIh) (mgIL) (mgIL) (mgIL) (mV)

1 medium 12 low 0.26± 0.04 0.004 ± 0.003 0.112 ± 0.010 510 ± 24

2 high 15 none 0.11 ± 0.11 0.002 ± 0.001 > 0.770 503 ± 13
3 high 15 high 4.22± 0.11 0.000 > 0.770 594±5
4 low 5 low 0.52 ± 0.16 0.000 > 0.770 519± 8
5 low 5 high 6.69 ± 0.07 0.000 > 0.770 543 ± 26

Ferrolite Me2 appears to fonction weil in the oxidation of iron at bath lowand

high velocities and equally weIl for high and low levels of DO. Oxidation of manganese

at approximately 65% efficiency occurred in the tirst 5 hours of operation (300 L) after

which point the tilter was producing water with more manganese than the raw water.

Thus, it would be necessary to regenerate with potassium permanganate either on a

frequent or continuai basis for this materia! to function continuaIly.

5.2.2 Purolite

After severa! hours of rmes removal, analysis finally began on the water filtered by the

second catalytic matenal, Purolite. At a medium flowrate (12 mIh) and low air injection

(1.07 ± 0.59 mgIL) the column was found to eliminate nearly 100% ofboth the total and

dissolved irone Sorne manganese (total and dissolved) was also removed.

•
 74

• Table 12 summarizes the oxidation of iron and manganese by Purolite at various

levels of DO and velocities of filtration. Initial conditions in the raw water can be

referred to in Table 10.

Table 12. Iron and manganese oxidation on Purolite.

RUB Veloeity Dislolved osygen DRI DRM Eh

(mIh) (mgIL) (mgIL) (mgIL) (mV)

1 medium 10 low 1.07 ± 0.59 0.000 0.101 ± 0.015 548 ± 12

2 medium 10 high 3.87 ± 0.22 0.000 0.093 ± 0.007 580±9
3 low 4 low 1.68 ± 0.20 0.000 0.106 ± 0.026 590±9
4 low 5 none O.09±0.OS 0.000 0.085 ± 0.008 498 ± 13
5 high 20 none 0.12 ± 0.04 0.000 0.089 ± 0.020 512 ± 16
6 low 5 high 4.56 ± 1.09 0.000 0.080 ± 0.003 570 ± 16

Oxidation of iron was exceptional on the catalytic material Purolite. Furthermore~

66 - 75% of manganese was oxidized up to a velocity of 20 m/h and without the tilter

requiring regeneration. This performance was observed throughout experimentation on

Purolite~ during which time the tilter processed approximately 1127 L of water. As with

the Ferrolite MC2~ a redox potential greater than approximately 500 rnV indicated that

100% of iron was being oxidized.

•
 75

•
S.2.3 Summary on the catalytic pilot study

The catalytic materials Ferrolite Me2 and Purolite were investigated in pilot columns

with an upward flow configuration which resulted in particulate matter escaping with the

filtered water. Therefore~ the apparent removal capacity of the two materials tested was

not determined. However~ the oxidation capacity of each was shown to be excellent for

iron, which seemed to be reflected by an ORP greater than 500 mV in both cases.

The performance of each material is further evaluated for the total volume of

water filtered per mass of catalytic material when oxidation of iron (manganese) was

consistent. The total mass of Ferro/ite Me2 is calculated from the bulk density (1120

kg/m3) and the dry volume occupied in the column (1.2 m x 0.005 m2 ) ta be 6.72 kg.

Thus Ferrolite MC2 filtered 1457 L or 217 Llkg and consistently oxidized iron al 100%.

In tenns of manganese oxidation al 65%, Ferrolite MC2 filtered 300 L or 45 L/kg before

reaching saturation.

Similarly, Purolite, (bulk density 1370 kg/m3) with a total mass of 8.22 kg~

filtered 1127 L or 137 L/kg while consistently oxidizing iron at 100% and manganese at

75%.

Based on the results from the pilot study al Septmonts~ the catalytic material

Purolite gave a superior performance to Ferrolite Me2.

•
 76

6. CONCLUSIONS

• Biological iron removal for potable water production was investigated on pilot-scaIe and

full-scale facilities. From the investigation of iron treatment on two pilot columns at two

sites (Seignosse and Septmonts), it was observed that the measure of oxidation-reduction

potential (ORP) appears to be a good indicator of changes in ITon concentration in the

filtered water of a biological process. A system-specific minimum value of ORP can be

determined which corresponds ta a maximum acceptable level of iron in the filtered

product. Thus, any measure of ORP which faIls below the system-specifie minimum

value would indicate perturbations in iron treatment which could be due to, for instance,

tilter saturation.

The etIect of varying the dissolved oxygen (DO) content of the raw water was

aIso explored for its impact upon the biological iron removal process and the ORP

measurement. It was perceived that for the raw waters examined (pH 5.7 and pH 6.7), an

increase in DO content above the minimum requirement did not improve nor hinder the

biological treatment of ïron. This observation is consistent with literature which indicates

that for a pH below neutral, physico-chemical oxidation is not significant and does not

compete with biological oxidation, regardless of the DO concentration (Ehrlich, 1990;

Mouchet, 1992). In addition, the variations in DO did not produce a quantifiable

response in ORP above the minimum required DO concentration.

Based on the results from the pilot at Seignosse, operating conditions of the full-

scale primary tilter at the site were altered (no pH adjustment of the raw water, decreased

frequency and length of backwash) to produce improved treatment of iron and savings in

•
 77

•
backwash water. The operating modifications were significant since iron treatment was

made possible in a biological tilter operating at a pH of 5.7, which was previously

considered infeasible.

Samples of sludge collected in September and October from the pilot plant at

Seignosse did not reveal the most common iron bacteri~ Gallionella or Leptothrix, nor

iron bacteria of the genus Siderocapsa. On-going study of the site is attempting to

identify the micro-organisms present in the primary tilter. Biologjcal treatment cao he

postulated as several performance factors attested to biological and not physico-chemical

oxidation. These included: pH and Eh of the raw water (5.7, 290 m V), velocity of

filtration (30 m/h), effective size of sand in the pilot and primary filter (1.35 and 1.18

mm), residual iron concentrations in dissolved fonn, and filter retention capacities on the

pilot and primary tilter of3.9 and 1.2 kglm2 , respectively.

In general, it was found that the frequency and duration of the backwash are

significant factors which affect the biological tilter. Thus, the measure of ORP cao he

very useful as ao indicator of the point at which tilter regeneration is required. In future

related studies ORP could he investigated as an indicator in determining the optimum

times of the backwash and rinse of the tilter (recall that the backwash consists of raw

water and air sent countercurrently through the tilter whereas the rinse portion is raw

water alone sent cocurrently). Thus, for various tinite backwash sequences, the ORP

could be measured in the tiltered rinse water to establish the minimum required rinse

duration corresponding to the backwash. Such an investigation would permit the

minimizing of filter maturation time, while maximizing the filtration cycle with respect to

• filter regeneration.
 78

•
The oxidation capacities of two catalytic materials, Ferrolite MCZ and Purolite.

were evaluated under site-specific conditions. The raw water tested contained

approximately 3.47 mgIL of dissolved iron and 0.317 mg/L of dissolved manganese.

Both materials oXÎdized iron at 100% efficiency throughout the entire time of the trials

regardless of DO concentration and for velocities up to 15 mIh on Ferro/ite MCZ and 20

m/h on Purolite. ORP measurements which were invariably greater than 500 mV during

the trials seemed to refIect the consistent oxidation of iron observed. Ferrolite MCZ

filtered a total of 1457 L or 217 L per kg of catalytic material while consistently

oxidizing iron at 100%. With respect ta manganese oxidation al 65%, Ferrolite MCZ

filtered 300 L or 45 Llkg. Purolite filtered a total of 1127 Lor 137 L per kg of catalytic

material while consistently oxidizing iron al 100% and manganese al up ta 77%.

The two catalytic materials were investigated in pilot columns with an upward

flow configuration. Thus it was not possible to determine the filtration retention capacity

of Ferrolite MCZ and Purolite. Nevertheless, initial analyses of filtered water showed

that both materials eliminated both total and dissolved species trom the outset of the

investigation. It was ooly when the air injection to the raw water was introduced that the

filter bed was significantly disturbed. Therefore, it is recommended that a gravitational

flow configuration he adopted for any future studies with these materials if air injection is

to be used. In addition, because Ferrolite Mez achieved saturation with respect to

manganese oxidation after filtering 45 Llkg, it appears that it wouId require regeneration

with potassium permanganate on a continuai or frequent intermittent basis.

•
 79

•
REFERENCES

Andersen~ D.R., Row, 0.0., and Sindelar, G.E. (1973) hon and manganese studies of

Nebraska water supplies. J.AWWA. 65 (10) pp. 635-641.

Charpentier, J., Florentz, M., and David, G. (1987) Oxidation-reduction potential (ORP)

regulation: A way to optimize pollution removal and energy savings in the Low

10ad activated sludge process. WaJ. Sei. Teck 19 (3-4) pp. 645-655.

Cole, L. (1986) Update on common water treatment methods. Water Weil Journal. 40

(11) pp. 58-60.

Degrémont Infilco Ltée. (1991) Water Treatment Randbook 6th ed. vol.l and 2,

Lavoisier Publishing, Paris.

Droste, R.L. (1997) Theory and Practice of Waret and Wastewatet Treatrnent. John

Wiley and Sons, Inc., Toronto.

Ehrlich, H.L. (1990) GeomicrobiolQQY. Znd ed. Marcel Dekker Inc., New York.

HACH Company Ltd. (1992) HACR DR20QQ Procedures Maoual. French ed. Loveland,

USA.

RaneIt, H.H. (1981 a) The genus Ga/lione/la. The Pmkacyotes; A Handbook Qn habitats.

iSQlation. and identification of bacteria vol. 1, Springer-Verlag, New York,

pp. 509-515.

Hanert, H.H. (1981b) The genus Sidercapsa (and other ïron- or manganese- oxidizing

eubacteria). The Prokaryotes: A HaodboQk on habitats. isolatiQn. and

identification of bacteria. vol. 1, Springer-Verlag, New York, pp. 1049-1059.

•
 80

Health Canada (1996) Gujdelioes fQr Capadiao priplçjni Water QualitY. 6th ed. Ministry

• of Health, Ottawa.

Heduit, A. and Thevenot, D.R. (1989) Relation between redQx poteotial and oxygen

levels in activated sludge reactors. Wal, Sci. Teck, 21 (8-9) pp. 947-956.

Hem, J.O. (1961) Stability field diagrams as aids in ironchemistry studies. J. AWWA.

53 (2) pp. 211-228.

Hettler, J.-P. (1982) Une station de déferrisation biologique. T.S.M L'Eau. 77 (10)

pp. 481-484.

Lehr, J.H. et al. (1980) OQmestic WaterTreatrnent. MeGraw-Hill, Ine., New York.

Mouehe4 P. (1991) Déferrisation biologique de Seignosse: Rapport de la visite effectuée

les 21 et 22 mars 1991. Dégremont, internaI report.

MQuchet, P. (1992) From eonventional tQ biological removal of iran and manganese in

France. J. AWWA. 84 (4) pp. 158-167.

Mulder, E.G. and Deinema, M.H. (1981) The Sheathed Baeteria The ProkEUyQtes: A

Haodbook Qn habitats. isolatiQn and identificatiQo Qfbacteria. vol. 1, Springer-

Verlag, New Yorlc, pp. 425-440.

Pisarczyk, K. (1995) Manganese Compounds. Encyclopedja Qf Cbemical TecboQloiY·

4th ed. VQl. 15, John Wiley and Sons, New York, pp. 991-1055.

Rheinheimer, G. (1992) AQuatic MjcrobjQloiY. 4th ed. John Wiley & Sons, Inc.

Chichester.

Robinson, R.B. et al. (1987) Committee report: Research needs for the treatment of

iron and manganese. J. AWWA. 79 (9) pp. 119-122.

•
 81

• Staley, J.T., Hirsch, P. and Schmidt, l.M. (1911) Introduction to the Budding and/or

Appendaged Bacteria. The Prolguyoles: A Handbook 00 habitats, isolation, and

identification pfbacteria. vol:I,

"-
Sp~ger-Verlag,New Yo~ pp.
-
451-455.

Starkey, R.L. (1945) Transfonnations ofiron by bacteria in water. J. AWWA. 37 (10)

pp. 963-984.

Stolzenberg, A.M. (1995). Iron Compounds. EOCèyclopedia of Cbemjeal TechooIQ&Y, 4th

~ vol. 14, John Wileyand Sons, New Yo~ pp. 873-902.

Stumm. W. and Morgan. J.J. (1996). Aquatie Cbemjstry. 3rd ed. John Wiley &. Sons,

[ne., New Yorle

Temple. K.L. and Colmer, A.R. (1951). The autotrophic oxidatioo ofiron by a new

bacterium: ThiobacillusJérrooxidans. J. Bacteriology 62 pp. 605 - 611.

Thill, A. (1995). Traitement CatalytjQue et BjolQiigue du fer et du ManiMèse,

C.I.R.S.E.E., internal report.

Viswanathan. M.N. and Boettcher. B. (1991). Biological removal ofiron from

groundwater. JVal. Sci. Tech. 23 (7) pp. 1437-1446.

Vuorinen. A. et al. (1988) Chemical. lninerLllogical and microbiological factors

affecting the precipitation of Fe and Mn from ground water. Wal. Sci. Tech.

20 (3) pp. 249.

Wareham. D.G.. Hall. K.J., Mavinic, D.S. (1993) Real-time control ofwastewater

treatment systems using ORP. Wal. Sci. Tech. 28 (11) pp. 273-282.

Wolfe, R.S. (1958) Cultivation, morphology and classification of the Îron baeteria J.

•
,
AWWA. 50 (9) pp. 1241-1249.
 82

•
Wolfe. R.S. (1964) Iron and Manganese Bacteria. In: Princip/es and Applications in

Aquatic Microhi%gy, H. Heukelekian and N.C. Dondero (eds.)., John Wiley &

Sons.. [nc... New Yor~ pp. 82-97.

Zavarin, G.A. (1981) The genus Meta//ogenium. lbe Pmkatyotes; A Handbook on

habitats. isolatjon. and identificatjon of bacteria. vol. 1, Springer-Verlag, New

York.. pp. 524-528.

•
 1-1

• APPENDIX 1:

RAW WATER ANALYSES AT SEIGNOSSE, ruL Y 1996.

•
 1-2
~\EçU Ue LYONNAISE DES EAUX

•
LABORATOIRE DE LA D. R .
AV DE CAMBO 64600 ANGLET

•
, 5 JUIL 1996
Rép:-----------

ANALYSE DE CONTROLE N° 719

Prélèvement effectué le : 01/07/96 par AGENT DU SECtBUR

Commune de : SEIGNOSSE
Lieu de prélèvement : SEIGNOSSE BOURG USINE PaRAGE E4 EAU BRUTE
Nature du prélèvement : RESSOURCE
Résultat transmis le : 10/07/96 Type d' analys'e PA)

RESULTATS UNITES

Paramètres organoleptiques :
ASPECT CLAIR
COULEUR <2.5 MG/L PT/CO
TURBIDITE 0.34 N.T.U
ODEUR o TX DILU.
ODEUR DETECTEE /
Paramètres Physico-chimiques :
TEMPERATURE 13.8 DEG.C
PH SUR PLACE 6.10
IS A LA TEMPERATURE DU PRELEVEMENT -4.64
CONDUCTIVITE A 20 0 C 142 p..S/CM
CHLORURES (TITRIMETR.IE) 29.5 MG/L CL
SULFATES (NEPHELOMBTR.IE) 4.5 MG/L S04
CALCIUM (COMPLEXOMBTR.IB) 4.5 MG/L CA
MAGNESIUM (COMPLEXOMETR.IE) 2.8 MG/L MG
SODIUM (AA FLAMME) 17.7 MG/L NA
POTASSIUM (ICP) 1.1 MG/L K
ALUMINIUM (COLORIMETRIE) 0.07 MG/L AL
DURETE TOTALE <COMPLBXOMETR.IE) 2.3 DEG.F.
TITRE ALCALIMETRIQUE COMPLET (TITRIMETRIE) 1.6 DEG.F.
CARBONATES (CALCUL) o MG/L
HYDROGENOCARBONATBS (CALCUL) 19.50 MG/L
ANHYDRIDE CARBONIQUE LIBRE 28.60 MG/L C02

Substances Indési.rables :
NITRATES (TECHNICON) <0.5 MG/L N03
NITRITES (TECHNICON) <0.05 MG/L N02
AMMONIUM (COLORIMETRIE) o MG/L NH4

•
 1-3
LYONNAISE DES EAUX

•
LABORATOIRE DE LA O.R .
AV DE CAMBO 64600 ANGLET

• ANALYSE DE CONTROLE ND 726

Prélèvement effectué le : 01/07/96 par AGBNT DU SECTEUR.

Commune de •••••.•.•.•.•• : SEIGNOSSE

Lieu de prélèvement ••••• : SEIGNOSSE BOURG US INB FORAGE 84 BAU BRtrI'E

Nature du prélèvement .•• : RESSOURCE

Résultat transmis le •••• : ~0/07/96 Type d' anaJ.yJe B2A

RESULTATS UNITES

Paramètres Microbiologiques .••.••••••••.•.•.••••• :

COLIFORMES TOTAUX (MF) o /l.OOML

COLIFORMES THERMOTOLBRANTS o /l.OOML
STREPTOCOQUES FECAUX (MF) o /l.OOML
DENOMBREMENT DES GERMES TOTAUX 37 DEG.C 13 /ML
DENOMBREMENT OES GERMES TOTAUX 22 DBG. C 82 /ML

CONCLUSION: LES RESULTATS DES PARAMETRES ANALYSES SONT CONFORMES AUX

LIMITES DE QUALITE DES BAUX BRUTES POUR LA PRODUCTION D' EAU
DESTINEE A LA CONSOMMATION HUMAINE.

Le 10/07/96, LE RESPONSABLE DU LABORATOIRE

•
 1-4
ANALYSB DE CONTROLE N° 719

• Commune de
•
Prélèvement effectué le : 01/07/96
Lieu de prélèvement
: SEIGNOSSE
: SEIGNOSSE BOURG OSINE PaRAGE E4 EAU BRUTE

RESULTATS .UNITES

OXYDABILlTE ICMN04 A CHAUD MILIEU ACmE 0.47 MG/L 02

FER (COLORIMBTllIB) 3500 MICROG/L FE
FER DISSOUS 2700 MICROG/L
MANGANESE (COLORIMETRIE) 60 MICROG/L MN
MANGANESE DISSOUS 60 MICROG/L MN

CONCLUSION LES RESULTATS DES PARAMBTRBS ANALYSES SONT COHFORlfES AUX

LIMITES DE QUALITE DES BAUX BROTBS POUR LA PRODuctION D' BAU
DESTINEE A LA CONSOMMATION HUMJUHB. BAU AGRESSVIE-

Le 10/07/96, .LE RESPONSABLE DU LABORATOIRE

•
 2-1

• APPENDIX2:

DEFERRISATION BIOLOGIQUE DE SEIGNOSSE: RAPPORT DE LA VISIT

EFFECTUtE LES 21 ET 22 MARS 1991.

•
 2-2

•
DEPAUIHIIIr DES LAllDES

000

DGAO'OBOZE

000

DDaPTSAnOR IIOLOGIqUZ DE SUGROSSB

000

!APPORT DE LA. VISXD U'iECiDU LBS Z1 et 2Z KARS 1"1

Nous avons ausculté cette .tation, avec K. DUPUIS et J.L. PICCIlILLI.

La station était l l'arrlt 1 notre .r~i"ée. Hous l'avons d'abord remise en .er-
vice dans le. condition. d'e••ais antérieurs, c' •• t-~-dire l dêbit r'duit et
sans contre-pre••ion (6vacuation de l'.au fi1tr'. ~ l"sout). L•• par... t~e. de
fonctionnement 'taient alors 1

Q eau - 43,S m3/h (rappel 1 Q nominal - 7S m3/h)

Pression amont - 0,6 b

Q air - 2,5 Hm3/h (.Q atr - 5,75 1)

q eau

•
 2-3

• .."

Résultats après 1 heure de fODCtiODD-..nt

.
;
;
Eau Brute Bau Filtr'.
1

Temp4irature C-C) 14 3.4

1
pH S,7 5,7
1
Eh (mV/HZ) + 60 aV
(.tabili••tion
-
difficile

(DII/1) abaence 1,4

°2
lfH. (mail) O,OS 0,0.5

HOZ (mail) 0,03 0,02

Fer (111/ 1 ) 3,3 2,8"

Manganèse (1IIS/1) 0,02 0,02

"rH 13,S -
TAC (-P) 2.2 2.2

L'augmentation du d4ibit d'air ~ 3,8 Hm3/h ou 8,75 % du d6bit d'eau, 4ilêve LE

concentration en Oz di.soul l 2,8 me/l mai. ne chanle pa. la teneur en fe~
résiduel dans l'eau filtr.e.

Premiêres conclalfou.

les seuls problèmes pos'. par cette .au sont le fer et l'agressivité carboni
que; en revanche. pas de probl"e de manlantse. RB•• H02.
les premières analyses, effectuêe~ par le Laboratoire D'parte.ental d~
Mont-de-MarIan en 1988, mentioanaient un pH de 6,8 (5.8.88) ~ 5,9 (2.11.88) •

•
 24

•
A la mise en service Cêtê 90), 'le pB (...urê lur place) .ta~t de 6,3, puil l'elt
abaissé progressivement pour .e fixer . .intenant ~ S.6/S.7. c· •• t aDe valeur de
pH que nous n'avons pa. encore.conltat' en dêferr~.ation biologique et qui
dlailleurs ne peut pa. corre.poDdre l une precipitation totale du fer, comme
nous le verrons plus 10~.

***********

Trois heures après la remi.e en .ervice, nou. a.OD' mil la .tation 'OUI prel-
SiOD. A une valeur proche de la fu~ur. pr.llion d. l.rY~ce lor.qu'elle refoulera
dans le réseau ; cette mil. IOUI pre••ioD a êt' obt~u. en fer..amt partiellement
la vanne de 'rinçage- (utili••e pour la matouraton du filtre apr.. la.aa.) qui
est placée sur la tuyauterie de vidaDse par o~ .'êcoule actuellement l'e.~ fil-
trée. Les paramètre. de fODct~ODD"'Dt lont dev.nu. r

Q eau • 36 m3/h
Pression amont - 4,6 b
Q air - 4.5 Hm3/h ( 0 air • 12,S I)
Q .au

Résultats après 3 heures de fonctioan8ment

Eau Brute Bau Piltree

pB 5.65 S,6S
.
Eh (IIlV/Hz) + 80 raV + 335 aV
Oz (mail) .ablence . 3,2

Fer (l1li/1) 3,3 2,6

•
 2-5

• -."

COllclusiODS

- A ce pB, malgré un rapport air/eau théoriquement fayorable ~ ane bonne

oxygénation de l'eau. le pourc.ntale de saturatioD en Oz dans l'eau filtrée
dépa•• e ~ peine 30 1.

- Le pB et le potentiel Redoz auzquels on parvient dans l·.au filtré. ne

corre.pondent pa. encore l aD. pr.c~p~tatiOD total. du fer (yoir fil. 1. o~ lE
point 1 est repréaentatif de l'.au brut. et le point 2 de l'eau aérée .t
filtrée) •

*******

Le débit d'air a été ensuite porte ~ 5,9 Nm3lh. Le rapport Q air/Q eau était
alors de 16,8S % (Q eau • 35 m3/h).

R.êsu1tats 2 heures .pr" 1& ..ac:U.fi.C.Ù.OIl

.Jau Brute Eau Filtrée

1
pH 5,65 S,65

Eh (IlY/HZ) + 80 aY + 340 DIV

Oz (l1li11) ab••nce 3,75

Fer (aII/l> 3,5 2,7

Sept heures après la modification. 1•• résultat. re.taient inchangés.

e·
• Conc1.usiOll

A un tel pH, la dêferri.a~ion bio~olique .emble impos.ible.

*•••****

Nous avons alors mi. en .ervice la pompe de soude, qui venait 4'.~re ln.~all'e
pour relever le pH de l'eau bru~e.

nouveau cycle de filtration: l'ouverture de la vanDe placte .ur

Avec un pB réglé l 6,9, nous avons voul.u laver le filtre pour COIIIIle1lcer un
la tuyauterie
d'évacuation de. eaux de lavale a _lheureu._nt provoqut un coup de ~lier qui.
a cass' cette tuyauterie, en PVC. Il n'êtait donc plu. que.~iOll de laver le
filtre .ous peine d'inonder la .~~ian. e~ l· •••ai l conttDu6 avec un filtre qui
alcalines, et i l reste donc une certaiDe incertitude
être expos', dan. ce qui .uit.
.ur
avai~ .ucces.ivement fonctionné dan. de. cOlld~~ion. trIs acide., puis ~r' •.
le. r6.ultat. qui vont

La bODJle va1.eur de pB a '~ê ob~eDue 1.e 23 mars en début cl'aprt.-llidi, dan. le.
conditions de fonc~iaDD...nt .uivant.. z

Q eau : 35 I113/h

Pression amont • 4,6 b

Q air - 7,1 Nm3/h (Q air - 20 1)
Q .eau

Eau Brute Bau Filtrée

pB 5,6 6,9

Eh (aV./RZ) + 70 aV + 350 aV

Oz (JIIg/l) absence .
10,2 IIg/1
(- 100 % .aturation)

tAC COF) 2,1 5,4

e·
•
2-7

•
L'augmentation de tAC corr~lpond l un taux de traitement en loude, avant
filtration, de 26,5 11m3 (en HaOB pare), loit .ncore 2,3 l/h de leslive de .oud
l 400 aIl pour 35 ~/h d'e.u bru~e.

Evolution des teneurs en fer (en . . Il)

Beure Bau Brut. Eau Piltrêe ;

i
1
15 h 3,2 1,6 f
t

16 h 30 3,0 1,1

17 h 30 3,1 0,3

COIlclusions

-.
La déferrisa tian biolosique se.bl. donc dtmarrer spontanément lorlque l'.au est
mise dans des condi~ions de pB le.trement inferieures l 7, ce qui la place dan.
des conditions de precipitation totale du fer (concretis6.s par le point 3 .ur
la figure 1). Il .ubsist. toutefoia de nombreuse. incertitudes 1
- confir.mation du bon rêsultat qui a 6tê obtenu 1. 23 mar. en fin de journée,
lor.que le filtre aura 'te reaia en I.rvic. aprt. lava.e ;
vérification que cette êvolution aboutit bien l une eau traitee totalement
exempte de fer (il n'. pas ett po.s~ble d'attendre c. relultat le 23 . .r.)
- si la voie biologique ••t confi~e. 1'.lit-il de bacterie. iDdiltne. ou de
ge~es subsiltant de. precedent. en.~Dc...nt. 7.J'ai cherché, au microlcope,
d'éventuelle. ferrobactêrie. d.n. le. tcumes qui se for.ment .pontan6ment l la
surface de l'eau dans 1•• tr.nch•••· d'infiltration qui reçoiv.nt 1• • •aux de
lavage des filtr.s (et, actuellement. qui reçoivent au••i l'eau filtree tant

e· qu'elle n'est pas refoule. sur 1. re.eau) : il en existe de tr.s rares

 •
2-8

e
exemplair•• (fila.ant.u••• ou GalliaDel1a). en quantité in.uffi.ante pour en
tirer une conclusion po.itive : .n reYaDcbe GD peut 1 voir de. trI. nombreu.es
bactéries libre•• et en outre 1. floc ferrique ob.erv' dan. ce••ca-e.
présente l'a.pect d'ua precipite foaDf par YOi. bialoliqu. ; il pourrait donc
très bien y avoir pr.dnefn8 ace d·autr••••plc•• d. ferrobact'ri••• camm. le
lenre Si.derocap.. qui ne fo~ IÛ 'Ihe lU p6dancal.. :
- enfin, et surtout ••i. tout ce qui. pr'clde est po.i.tif. v'rific.tion de la
faisabilité du proce•• au debit n~l de 1. Itation. c'.lt-&-dire 75 m3/h
(soit v - 24 mth).

*****'*'*'*
Conclu.tcm.. gdD.êrales

1. Sur cette eau, on ne peut espérer obtenir un bon ré.ultat qu'en élevant le pH
l une valeur approxJ-.tiveaene comprise entre 6,6 et 7. Cette contrainte
n'implique que la d'pen.e cl • une pompe clo.eu.e ,uPP~"'ntair., mai. n'aupente
pas la consommation Ilchale .n réactif puisque le taux de traitement final,
pour le relêvement du pB de l'eau trai.t6e, le trouvera dtmtnu6 de la quantité
déjà introduite dans l'eau brute.

2. Nous envisageona de poursuivre la mise au point de la .tation de la façon

s\U.vante :
- éventuellement, ensemencement du .able avec les écumes collectées dans les
trancbées de réception de. eaux d~·lav.le,
mdse en place d'un d6bibnêtre d'air sradué de 0 & 10 Hm3/h 'OUS 5,5 b
- réparation de la tuyauterie d'.vlcuation de• •aux d. llva.e que nouS
remplacerons par une tuyauterie en ac~er :
- mise en place d'un circuit définitif d'injection de la loude dans l'eau
brute <en remplacement du tuyau provisoire actuellement en place>
- après lavase du filtre, red.marrale dana le. conditions de fonctionnement
du ~3 mars après-midi (P • 4,5 b ; Q - 35 ~ 40 m3/h ; pH - 6,g : Oz voil1n
de 100 % .at.) ;

e·
 2-9

•
- si les résultat••Oht bon., affinas_ de. par. .ttre. de fonctionnement 1

* proportion minimale air/eau pour obtenir 90 l 100 % de la .aturat~on

en 02 di.sou. sous la pre••ion de service ;
* définition de la zone d. pB dans laquelle 1. d'ferri.ation biololique
atteint .a pleine efficacité et du pB mini l re.p.cter ~p'ratiYement

si on a pu obtenir un cycl. .atisfai.ant l demi-d'bit, pas.age ~ plein

débit après lavale du filtre (pa. d. d'colmatal_ ; 45 .ec. .oufflase; ~
min. de rinçage, l auamenter 'ventuellemene plu. tard,.i le. ré.ultats
.ont bons, ju.qu'au temp. D.c••••ire pour renouye1er la tranche d'eau .ale
au-des8us du .able, c. qui devrait.repr'••nter environ 2 minutes) ;
en ca. de résultats in.uffisants, (au ca. oà cette eau, dont le caractère
·inhabituel pour \Dle d'ferri••Uoll biolol1que Il'a't'
r'y'l' qu'. la lliae en
service, ne contiendrait pa•••••z de le~. de bact'rie. du fer dont le
domaine d'azistellce ne coue.pond que p~ur.a lJm.1te la plu. extrtme, aux
caractéristique. c~qu•• acta.llemant •••ur'e.) l demi-d'bit oa l pl.ta
débit, procéder l nouv.au l un .n....nc...nt avec 1•• bou.. d'um. autre
staUon.

P. MOUCHET

e·
 2-10
i • Ir • . If

OIAGQ~MME pOTEnTIEL - pH DU FEO ~

• EH (volr)~ (DANS L'EAU .Di5TillEE)

O-k-------~~+-------____t

-02 ~

-0 , 1

-06 ~

10 11 1'2 13 14 pH
e·