Perez Paricio 2

Chapter II: State of the Art
CHAPTER II
REVIEW –STATE OF THE ART
(...) «Y siempre reinterpretamos y recordamos el pasado desde los acontecimientos

que siguieron después, siempre coloreamos un incidente con el conocimiento que no teníamos
cuando estaba sucediendo.» (page 172)
ARIEL DORFMAN (from “Heading South, Looking North”. 1998)
This chapter contains a brief summary of the state of the art on clogging. It is not purported to be an
exhaustive review, but a compendium of the basic principles and, especially, the most interesting
models. There exist excellent reviews in the literature providing further information on physical
clogging (HERIZG ET AL., 1970; OSEI-BONSU, 1996; RYAN & ELIMELECH, 1996) and bioclogging
(BAVEYE ET AL., 1998; HOLM, 1999), apart from a few comprehensive reviews (PÉREZ-PARICIO &
CARRERA, 1999a). This chapter is based on the latter, even though a shorter version is presented in
PÉREZ-PARICIO & CARRERA (1998b).
The chapter focuses on conceptual issues (processes, parameters), prevention-redevelopment and

modelling. The main objective of this presentation consists in establishing the conceptual
framework of clogging and remarking the existing drawbacks, so as to implicitly motivate the need
for an integrated approach to the problem.
A remark has to be done before initiating the Chapter. From a conceptual point of view, there is no
difference between surface (basins) and deep (wells) recharge systems, because the physics of the
processes are equivalent in both types of systems. However, there are evident practical implications,
since the operational procedures, requirements on the quality of recharge water and objectives may
vary substantially. This will be taken into account when referring to technological aspects.
5
Integrated Modelling of Clogging Processes in Artificial Groundwater Recharge
2.1. CONCEPTUAL ASPECTS
The elementary clogging mechanisms have been listed in Chapter 1 (Introduction). This section
provides more details about such mechanisms, as well as enumerating some relevant parameters.
2.1.1. PHYSICAL CLOGGING
The term physical clogging refers to mechanisms that affect the movement, attachment (also
deposition, retention, colmatage) and detachment (also shear, unclogging, decolmatage) of inert
suspended particles in recharge water, whose consequence is a diminution of porosity. The reviews
by HERZIG ET AL. (1970) and MC DOWELL-BOYER ET AL. (1986) are especially recommended.
Depending on their size, suspended particles are usually classified under three categories: colloidal,
intermediate and large particles.
10-10 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2
M OLECULES
COLLOIDS
SUSPENDED PARTICLES
PARTICLES
BACTERIA ALGAE
VIRUS
M ICRO SIEVES
SIEVES
M EMBRANES
M OLECULAR SAND AC (GRAINS)
SIEVES FILTERS
DIATOMACEOUS
SILICA - GELS EARTHS
AC
m icro- pore openings
-pore s
Diameter (m)
Figure 2.1. Classification of particles and filters as a function of size: diameter and pores, respectively.
Delimitation of some categories is neither universal nor easy to establish. Dashed lines indicate diffuse separations
of especial cases. AC stands for activated carbon filters.
2.1.1.1. Colloidal particles
Their diameter is lower than approximately one micron (1 µm = 10-6 m), although 0.45 µm is
often used as threshold for classification purposes. In fact, Figure 2.1. assumes a limiting value
of 10 µm. Owing to their great specific surface (surface divided by volume), suspension of
these particles is controlled by physiochemical surface forces: London-van der Waals attractive
forces, attractive or repulsive electrostatic forces and Born repulsive forces. The former are
caused by electrical dipoles, while the latter are due to overlapping of electron orbitals (i.e.,
very short-range interaction, a few nanometers; 1 nm = 10-9 m). Several theories have been
developed to account for these interactions; one of the most famous ones is the DLVO
(Derjaguin-Landau-Verwey-Overbeek) theory for total potential energy of interaction between
surfaces. Brownian diffusion plays also a relevant role in the fate of colloidal particles in
porous media.
6
The role of colloids in transport of some species comes periodically under scrutiny (HOFMAN &
SCHÖTTLER, 1998). It has been shown that radionuclides and metals can get adsorbed to
colloids through ion exchange or surface complexation (CORAPCIOGLU & JIANG, 1996; RYAN &
ELIMELECH, 1996). This is very important to assess the mobility of contaminants in subsurface
environments (FRYCKLUND, 1999). That is, contaminants that would otherwise be immobile
may become mobile if attached to a colloid. On the other hand, the number and mobility of
colloids is usually low.
Field studies show that colloids may be formed by inorganic compounds such as minerals,
oxides and carbonates, or by humic substances and microbial exudates. Natural concentration
of colloids in groundwater is usually low, but colloids can be generated by means of:
1. Changes in the solution chemistry that cause repulsive interactions between colloids and
formation grains (KALLAY ET AL., 1987; MC DOWELL-BOYER, 1992; RYAN & GSCHWEND,
1994). Two factors must be considered: variations in ionic strength and in pH.
- On one hand, when recharge water has a lower ionic strength than groundwater the inter-
particle repulsion increases and induces the remobilisation of colloids (QUIRK & SCHOFIELD,
1955; MC NEAL & COLEMAN, 1966; ABU-SHARAR ET AL., 1987). Clay swelling and
dispersion is associated to this phenomenon. GOLDENBERG ET AL. (1983, 1984) have
reported great reductions in permeability due to a reduction in groundwater salinity. KHILAR
& FOGLER (1984) define a critical salt concentration value for particle dispersion in Berea
cores (with monovalent cations) as a function of temperature and cation type.
- On the other hand, an augment of pH tends to enhance clay dispersion and to reduce the
intrinsic permeability of the medium.
2. An increase in groundwater velocity due to rapid flow through fractures or to pumping, that
is, as a result of hydrodynamic changes (KAPLAN ET AL., 1993; BACKHUS ET AL., 1993;
RYAN ET AL., 1995). This is why sampling colloids –and colloids-sized microorganisms- in
aquifers requires special protocols (HOFMAN & ESCHWEILER, 1999)
2.1.1.2. Intermediate particles
Their diameter varies between 1 and 30 µm, so that they are subject to both surface and
volumetric forces. Surface forces can be neglected for diameters larger than 10 µm (SANFORD
& GATES, 1956), even though some authors state either that van der Waals forces are
significant (MINTZ, 1964; MACKRLE, 1960) or even that electrokinetic forces must be included
too (STANLEY, 1955; O’MELIA & CRAPPS, 1964; IVES, 1965).
For the same reason, sedimentation is also controversial. Some authors consider it (CAMP,
1937; FAIR, 1951; IVES, 1961), whilst others neglect it (STEIN, 1940; O’MELIA & CRAPPS,
1964; SMITH, 1967). Other researchers restrict the importance of sedimentation to aggregates
(DIAPER & IVES, 1965; ISON & IVES, 1969).
It is evident that discrepancies arise when dealing with particles that can not be clearly
associated to a specific group, that is, when both surface and volumetric forces can be relevant.
It seems though that both surface forces and sedimentation should not be neglected.
2.1.1.3. Large particles
Their diameter is larger than 30 µm and are only affected by volumetric forces. Therefore,
relevant processes include sedimentation, direct interception (straining) and hydrodynamic
effects. Large particles get retained in crevices, constriction sites and caverns. Some authors
state that spontaneous unclogging of large particles is improbable in natural conditions, even if
7
flow is increased by a factor of 3 or 4 (MAROUDAS, 1961; SAKTHIVADIVEL, 1966; HERZIG ET

AL.,1970).
It is also important to notice that some authors incorporate ‘mechanical jamming’, that is,
particle rearrangement, into another clogging category (SNIEGOCKI & BROWN, 1970; LEENHEER
ET AL., 1976; JOHNSON, 1981; HUISMAN & OLSTHOORN, 1983; RANDALL, 1995). PYNE (1989)
and JENNE ET AL. (1992) include clay expansion or dispersion. But, as commented for colloidal
particles, mechanical jamming and clay swelling plus dispersion can be viewed as particular
cases of physical clogging. Sometimes particulate organic matter (CUSTODIO, 1981; PAVELIC &
DILLON, 1996) can also provoke a specific flow reduction.
Figure 2.2. Mechanisms during water filtration.

At the beginning, blocking filtration (top) takes place when the particles’ diameter is larger than the pores. Then,
cake filtration (middle) proceeds as the previously retained particles lodge other mobile particles. This can occur
with or without compression of the cake itself (not explicitly shown). Deep bed filtration (bottom) is the fourth
mechanism, and occurs when the particles can travel a certain distance through the medium until they are deposited
by different causes. These mechanisms are not necessarily sequential.
8
When water flows through a permeable medium several types of accumulation of sediments can
take place (Figure 2.2.):
1. Blocking filtration occurs when a particle obstructs an opening or pore due to interception. This
originates an increase in head loss and a diminution in filtration rate.
2. Cake filtration without compression takes place when other particles are deposited on the
previously blocked particles. A cake or clogging layer develops on top of the soil filter. This is a
typical phenomenon in surface Artificial Recharge systems.
3. Cake filtration with compression occurs when the clogging layer is compressed during
accumulation of particles.
4. Deep bed filtration is characteristic of recharge wells, yet it can be relevant for surface systems
too. Particles smaller than the medium pores travel a certain distance across that medium, and
can be deposited by several capture processes during their passage (deep bed filtration).
How these mechanisms work depends on the type and characteristics of the medium, recharge water
quality, type (size and physicochemical properties) of the sediments and the infiltration extent. A
generic equation can be used to describe the first three mechanisms, whilst the latter will be
investigated in section 2.3.2. ORR (1977) proposed the following generic equation:
α2
d2V  dV 
= α1   [2.1.]
dt 2  dt 
where t [T] is time; V [L3] is the recharge volume; α1 [dimension depending on the mechanism] is a
coefficient used to define the mechanism and α2 [L0] is a constant that determines the actual
filtration mechanism.
• If α2=0, equation [2.1.] describes the mechanism of cake filtration. After integration, it yields:
t µ R µl I F
= l 0 + V [2.2.]
V ∆P A F 2 ∆P A 2F
where µl [ML-1T-1] is the dynamic viscosity of the liquid; R0 [L-1] is the initial resistivity of the
(clean) filter; P [ML-1T-2] is the applied pressure, ∆ denoting increment; AF [L2] is the membrane
filtering area; and IF [L-2] is a measure of the membrane fouling potential of water (SCHIPPERS &
VERDOUW, 1980). Equation [2.2.] shows a linear dependency of the inverse of flowrate, t/V with
respect to the filtered volume, V. The slope of this term is defined as the modified fouling index
(MFI), described in section (2.2.3.):
µl I F
MFI = [2.3.]
2 ∆P A 2F
This parameter –which equals α1 in equation [2.1.]– is very important for clogging purposes.
The MFI [TL-6] is commonly expressed in s·L-2. It is possible to use the same equation for cake
filtration with compression if the MFI is multiplied by a factor that incorporates the
compressibility of the solids:
β
MFI compr = MFI 1 + ∆P p  [2.4.]
 
where βp [L0] is a constant that contains the dependence on the compressibility of the particles.
9
• If α2=2, equation [2.1] reduces to the expression for blocking filtration and α1 equals:
∆P A F N p
α1 = [2.5.]
µl εF
where Np [L-3] is the number of particles per unit volume (concentration) and εF [L-1] is a
constant that depends upon the structure of the membrane filter.
• An intermediate value of α2 (α2=1) indicates that there is a partial pre-blocking of the pores
followed by cake filtration.
2.1.2. BIOLOGICAL GROWTH (BIOCLOGGING)
Although protozoa and viruses are present in groundwater and in fact are receiving a great deal of
attention because sanitary concerns, bacteria play the most relevant role regarding biological
clogging. CHAPPELLE (1992) and PAVELIC ET AL. (1996) provide further information about
microorganisms and groundwater, but many studies are available (REFSGAARD ET AL., 1991;
ALBRECHTSEN & WINDING, 1992; SCHOETTLER & SCHULTE-EBBERT, 1995).
In a detailed review on biological clogging, BAVEYE ET AL. (1998) identify four main mechanisms
of bacterial clogging. These are (1) the accumulation of cell bodies in the porous medium, (2) the
production of bacterial extracellular polymers, basically polysaccharides, (3) the entrapment of
gaseous products, especially carbon dioxide (respiration), nitrogen (denitrification) and methane
(methanogenesis), and (4) the microbially mediated accumulation of insoluble precipitates.
Many bacteria form a biofilm, which is composed by cells and extracellular material, basically
polysaccharides (ALLISON, 1947; SHAW ET AL., 1985). The biofilm is treated as a separated phase
by many researchers (TAYLOR & JAFFÉ, 1990a). The formation of a biofilm attached to the medium
reduces the porosity, thus causing clogging of Artificial Recharge devices.
Bacteria are also important because of their catalytic capacity with respect to redox reactions.
Bacteria need external electron acceptors for catabolism, and carbon (organic if heterotrophic and
inorganic if autotrophic) for synthesis. Oxygen is the most effective electron acceptor in terms of
energetic yield, so that its availability is crucial for bacterial development. Generation, transport and
consumption of certain basic species depend strongly on bacteria and aquifer conditions. Each year
new bacteria types are being identified in field studies (TAYLOR ET AL., 1997; ESCHWEILER ET AL.,
1998), which confirms the importance of this clogging mechanism.
Algae are one of the concerns for surface systems. Clogging by algal growth can be very severe, as
reported elsewhere. For instance, Synedra (size between 65 and 85 µm) and Nitzchia (150 µm) have
been detected by KONNO & SATO (1986) and ADIN & CHELMINSKY (1986), respectively. Algal
activity concentrates in a surface layer, mainly within the upper 7.5 cm (GOODRICH ET AL., 1990).
Photosynthesis sometimes originates precipitation of carbonates as well, because consumption of
carbon dioxide can lead to over-saturation of carbonaceous compounds -mainly calcite.
2.1.3. CHEMICAL REACTIONS
Chemical reactions that promote the precipitation of minerals can occur because of incompatibility
between recharge water and groundwater (LLURIA ET AL., 1991; HUISMAN & OLSTHOORN, 1983;
10
PETERS, 1994; and many other researchers). The precipitation of minerals provokes an immediate
reduction in porosity.
Chemical clogging of continuous recharge wells is unlikely, because groundwater is displaced by
recharge water and mixing takes place only at the recharged water front. Instead, pumping wells are
more susceptible. For instance, under reducing conditions, iron may be dissolved. Oxidation will
then cause the precipitation of iron compounds (STUYFZAND, 1993). This type of reactions is
usually catalysed by bacteria. Thus, it is difficult to uncouple chemical and biological processes in
many real situations.
Chemical composition of recharge water and groundwater, aquifer mineralogy and physical
variables (temperature and pressure) control precipitation. Reactive surface area of a mineral
dictates the rate of precipitation or dissolution. Surface area is essential, as explained in section 2.4.
Moreover, bacterial metabolism, addition or depletion of oxygen and carbon dioxide, and presence
of catalysts affect mineral precipitation and dissolution.
2.1.4. GENERATION OF GAS AND AIR
Gas can be originated by biochemical reactions in both surface and deep systems. Temperature
increases or pressure decreases can lead to releases of dissolved gases as well. On one hand,
temperature of recharge water should be kept above groundwater temperature because the solubility
of gases decreases with temperature. On the other hand, gas solubility is directly proportional to
pressure. “Bubbling” takes place when the water becomes over-saturated with gas and this separates
from the liquid in the form of bubbles. Negative (below atmospheric) pressures must be prevented
in pipelines carrying water to recharge wells in order to avoid bubbling of dissolved gases. These
phenomena are characteristic of pumping wells, where positive pressures must be preserved.
Two main sources of air exist in wells (OLSTHOORN, 1982): first, bubbling due to the free fall of
water inside a recharge well (cascading) and, second, air entrance into the well if negative (i.e.,
below atmospheric) pressures occur. Entrained air bubbles can move downward and get into the
aquifer formation, increasing the resistance to water flow and resulting in higher water levels inside
the well. Such a resistance build-up is usually quite rapid, although it levels off after some hours
have passed because bubbles redissolve into the water.
Release of dissolved gases is another problem, because gas binding can take place. Air entrainment
in wells can be avoided through an adequate design and operation of recharge wells. Prevention of
negative pressures is done either by maintaining positive pressures in the pump column prior to
discharge below the water level or by means of airtight systems.
In spite that air is generally a nuisance for recharge wells, it is also purported to be a problem in
surface systems. According to NIELSEN ET AL. (1986) “situations can occur where air is not free to
escape when it is displaced by water, thus leading to larger soil air pressures and increased
resistance to liquid flow. The most obvious situation is that of ponded infiltration over large areas,
especially in the presence of high (or perched) water tables or relatively impermeable soil layers”.
A third means of augmenting the extent of gaseous phase consists of bacterial metabolic products.
When biocatalysed reactions take place, an increase of certain species can occur. Gas will be
generated if oversaturation is achieved.
11
2.1.5. COMPACTION
This is a relatively rare case that only takes place in surface systems, mainly basins (BOUWER &
RICE, 1989; BOUWER, 1990). Compaction is due to an excess of water above ground level in surface
systems, which causes the clogging layer to compress by increasing the confining pressure.
In theory, increasing the water height has a direct incidence in hydraulic gradient and recharge rate
should augment too. But the clogging layer formed as a result of sieving of suspended particles and
algae offers a resistance to flow that is increased if compression occurs. The final effect, as
described below, is a reduction in porosity. In fact, the top sediment layer on sand filters of water
treatment plants can break down as a result of an unduly increase of the hydraulic gradient. This
causes particles to contaminate and spoil the filter.
IDEALIZATION: RESPONSE TO
DIFFERENT CLOGGING MECHANISMS
Physical
Biological
Biological (limited)
Gas
Figure 2.3. Clogging evolution as a function of the predominant process.

It is a conceptual distinction, but shows differences caused by different clogging sources. (Modified from PYNE,
1995). W represents the resistance (pressure build-up divided by flowrate times the viscosity), as introduced by
OLSTHOORN (1982), due to clogging.
2.1.6. INTER-DEPENDENCE OF CLOGGING MECHANISMS
Previous identification of individual processes is required in order to understand the basic clogging
mechanisms (Figure 2.3.). But these processes have a mutual influence in practical conditions
(Table 2.1.):
(i) Coupling particles-biology.

ERNISEE & ABBOTT (1975) and ROSOWSKI ET AL. (1986) mention the interaction between
microbial communities and colloidal mineral particles. AVNIMELECH & MENZEL (1983) report
the evidence of algal-clay flocculation.
(ii) Coupling particles-chemistry.
Colloids are subjected to surface forces that depend on chemical parameters such as ionic
strength. KHILAR & FOGLER (1987) detected clogging by formation and migration of fines,
colloids and precipitates. The origin of fines can be ‘chemical’, for instance coming from calcite
dissolution (PAVELIC & DILLON, 1996).
(iii) Biochemical coupling.
12
It is the most evident link attending to the catalysing effect of bacteria (JOHNSON, 1981). Specific
models have been proposed (BRUN ET AL., 1994; for instance).
Algal photosynthesis usually causes an increase in pH that, in turn, produces precipitation of
carbonates (DAVIS, 1989; BOUWER & RICE, 1989). This is especially important when pre-
treatment includes coagulation that requires pH correction. In that case, the surface water may
become calcite oversaturated or more saturated, prior to algal growth (STUYFZAND, 1986, 1989).
(iv) Biology-gas coupling.
Metabolic products of bacterial reactions can transfer to the gaseous phase if oversaturation
exists. For instance, denitrification gases during basin management have been found to cause a
remarkable clogging. (RICE, 1974; BOUWER & RICE, 1989). Algal photosynthesis is also an
important process that leads to oxygen production and carbon dioxide consumption.
(v)Chemistry-gas relations.
Supply of oxygen can accelerate iron precipitation (CUSTODIO, 1981; SHAVER & WUCETICH,
1995), while a carbon dioxide loss often conducts to carbonate precipitation (DEGALLIER, 1987).
A good representation of this phenomenon is given by THEISS & SINGER (1974), who studied the
relationship among iron, organic matter and oxygen .
(vi) Compression and algal growth.
An excess of water height above the soil of basins creates a compressive force which diminishes
infiltration rate, enhancing algae proliferation. Algal photosynthesis increases pH, so that calcite
precipitation is possible (BOUWER, 1990).
Table 2.1. Inter-dependence of clogging mechanisms.

Only connections cited in the text are included. The remaining relationships are not considered and
hyphens (-) are used.
PHYSICAL BIOLOGICAL CHEMICAL GAS COMPACTION
Coloid-Bacteria Surface Forces

- - PHYSICAL
Algae-Clay and Chemistry
Biocatalysts
Bacterial products - BIOLOGICAL
Photosynthesis
Addition of O2 Compression =>

CHEMICAL
Loss of CO2 precipitation
- GAS
COMPACTION
13
2.2. PREVENTION AND REDEVELOPMENT
Prevention is essential to minimise clogging. However, after a certain recharge period -whose
duration depends on design, water quality and environmental and hydrogeological conditions- it is
often necessary to apply a redevelopment method. Both topics are addressed in this section.
2.2.1. PREVENTION
The best preventive tactics consists of ideal design of systems together with high water quality and
optimum hydrogeological characteristics. Actual recharge facilities are not ideal, though. This is
why various measures must be adopted to maximise benefits. As shown in Table 2.2., some of these
are strictly necessary to obtain a satisfactory groundwater quality, preserving it from pollution.
Details about numerical constraints on certain parameters are presented in section 2.2.4.
Table 2.2. Summary of preventive measures depending on the dominant clogging processes.
Æ
PHYSICAL CLOGGING Coagulation-flocculation plus sedimentation prior to recharge
Æ
Prior filtration to eliminate particles sand filters, microfiltration
Æ
Deep systems Periodic backwashing to detach sediments
Surface systems Desiccation of basins (i.e. recharge cycles)
BIOCLOGGING Bacteria: - Disinfectants
- Reduction of nutrients
- Constraints on input concentration (TOC...)
Algae ⇔ - Prefiltration
(surface systems) : - Reduction of nutrients
- Chemicals
- Avoid direct light on the water surface
CHEMICAL CLOGGING Acidification
Daily pumping
Separate strata
Avoid oxygen
GAS GENERATION Adequate design (valves, positive pressures)
Recharge water temperature > Groundwater temperature
COMPACTION Limit water height (⇔ surface systems)
CORROSION Avoid: Work-hardened materials & Stressed parts
High temperature & recharge rate
Oxidising conditions
GENERIC RULES 1. Appropriate water treatment
2. Minimise aeration (wells)
3. Proper design
4. Install high-water alarms
Table 2.2. includes different possibilities that are not mutually exclusive. In general, it has been
observed that surface systems can maintain high yields if recharge cycles are applied, that is, if the
clogging material is allowed to dry and crack naturally. This tends to break clogging deposits, as
well as impairing algal growth and enhancing biodegradation within the non-saturated zone. As for
deep systems, the best management option consists of periodic backflushing (reversing flow
direction by pumping). In both cases, prefiltration can be very helpful. WEGELIN ET AL. (1991)
provide guidelines on horizontal roughing filters, which are simple, effective and inexpensive.
14
2.2.2. REDEVELOPMENT
Although prevention is a key aspect in the design and management of recharge plants, often it is not
enough to maintain the efficiency of the system at an adequate level. Prior to the occurrence of
notable clogging impacts, it is necessary to redevelop the recharge system. In fact, redevelopment
and prevention may not be separable in practical situations, because such distinction is quite
arbitrary. For instance, management alternatives such as desiccation or periodic backflushing can be
indistinctly considered as preventive or redeveloping measures. Table 2.3. shows redevelopment
techniques (OLSTHOORN, 1982; DRISCOLL, 1986) but, contrary to preventive measures, a clear
distinction is made between surface and deep systems. The distinction is due to technological
differences between them rather than to conceptual dissimilarities.
Table 2.3. Redevelopment techniques for surface and deep systems.

This distinction is motivated by the practical implications in real recharge plants, even though the processes
involved are identical. Recharge cycles are usually implemented in surface systems, which leads to the
classification in preliminary and during-operation techniques. As for deep systems, redevelopment methods
depend on the type of clogging processes.
Surface systems During runs: - Filtration: mats, filters

- Chemicals (calcium)
- Enhance natural factors: aeration, avoid light
- Others: underwater robot to scrape the clogging layer
Between runs: - Mechanical methods: tillage, scrapping
- Addition of chemicals (calcium)
- Natural methods: drying of basins, freezing
- Acid baths to remove carbonated deposits
Deep systems Generic methods - Pumping, since flow reversals promote detachment
- Dosing of specific chemicals: disinfectants, acids
- Methods based on the injection of fluids
- Adoption of recharge protocols
- Other, less frequent techniques
Physical clogging - Pumping
- Chemicals
- Fluid-based methods
- Protocols
Bioclogging - Pumping
- Chemicals
- Physically-based methods
- Combination of methods
Chemical clogging - Chemicals
- Hydro-mechanical methods
- TV surveys and scraping borehole walls by brushing
In summary, it can be stated that the following methods are preferable (PÉREZ-PARICIO &
CARRERA, 1999a) if combined with hydraulic methods:
• Polyphosphates for physical clogging because they reduce the stability of retained particles,
which enhances the efficiency of detachment.
• Addition of chemicals if bioclogging is the dominant process. Chemical oxidants, such as
chlorine (Cl2) or hydrogen peroxide (H2O2), promote the inactivation of bacteria.
• Addition of acids for chemical clogging, in order to remove encrustations of different minerals.
15
Hydraulic methods comprise methods such as pumping, surging or “juttering”. The latter consists of
intermittent pumping in the damaged well (OLSTHOORN, 1982).
2.2.3. USEFUL TOOLS
Given the significance of clogging, some authors have tried to find predictive tools. These include
specific parameters, relationships and generalised charts. Some of them are briefly described below.
Here the aim is to explain their applicability range and main drawbacks. Further information is
available in PÉREZ-PARICIO & CARRERA (1999a).
2.2.3.1. Specific parameters
Conventional parameters, such as turbidity and concentration of suspended solids (physical

clogging) or dissolved organic carbon (bioclogging), are useful, but they are less informative when
recharging high quality water. Alternative parameters have been investigated in the Netherlands:
(i) Modified fouling index, MFI (SCHIPPERS & VERDOUW, 1980). It is proportional to the
concentration of both colloidal and suspended matter, and is determined by membrane
measurements. Physically, it represents the slope of inverse filtered flow versus cumulative
filtered volume (see section 2.1.1.1.). It is used as a guideline in the Netherlands assuming a
maximum MFI value of 3 to 5 s·L-2 for well recharge in fine sand aquifers. However, some
limitations arise: it is only reliable when clogging is strictly physical, and threshold values can
only be established after doing field research, taking into account the aquifer characteristics.
(ii) Assimilable organic carbon, AOC (HIJNEN & VAN DER KOOIJ, 1992). It was introduced to
assess the biostability of drinking water (not subject to disinfection in the Netherlands). AOC
represents the fraction of organic carbon that can be rapidly assimilated by microorganisms,
which may be an oversimplification in field situations.
(iii) To overcome these and other limitations, another parameter is currently under study (VAN DER
KOOIJ ET AL., 1995): the biofilm formation rate, or BFR. This accounts for all biomass in the
water, providing an integrated quantification of bioclogging potential (HIJNEN ET AL., 1998).
Despite all limitations, both MFI and AOC (or, alternatively, the BFR) should be considered as
excellent clogging-indicators. Although prediction can not be adequately addressed (SCHIPPERS ET
AL., 1995), they can indicate whether clogging is to be relevant or not. Furthermore, MFI and AOC
measurements can be combined with parallel filters (HIJNEN ET AL., 1998), which refer to standard
filter media (i.e. core samples that can be considered to be representative of different lithologies)
that are flushed with samples of recharge water. Combining simple measurements with aquifer
replica may be a more realistic approach than standard laboratory estimates.
2.2.3.2. Preliminary assessment of clogging in recharge systems
(i) Normalised build-up (for recharge wells). OLSTHOORN (1982) proposed a standardised
equation to compare clogging effects in recharge wells under different circumstances. This
equation adopts several reference conditions: 20 oC water temperature, infiltration velocity v0 = 1
m/h, injection time t0 = 1 year:
  µ  v0  µ   v 0   v 0 t 0 
∆h =  h e  0    − h b  0      [2.6.]
  µe   Qe / A F   µb   Q b / A F   V / A F 
16
where subscripts 0, b and e refer, respectively, to the standard, initial and final values; h [L] is
the piezometric head, so that ∆h represents the build-up caused by clogging; v [LT-1] is the
infiltration velocity; AF [L2] is the well filtering area; and Q [L3T-1] is recharge flowrate.
(ii) Simple filtration formulae (for physical clogging). Darcy’s equation is applicable to clogging
when the temporal evolution of permeability (or resistivity, which is the inverse of permeability)
can be approximated or fitted to laboratory data. Darcy’s law can be directly used to express the
time evolution of filtration rate through a membrane (SCHIPPERS & VERDOUW, 1980; RICHAUD &
MILLISIC, 1991), as shown in section 2.1.1.1.
(iii) Parallel filters. Due to the problems associated to previous parameters, a new approach applies
simple semi-empirical equations to data recorded during laboratory experiments with parallel
filters. These are columns filled-in with aquifer material, dosing recharge water under field-like
hydraulic conditions (SCHIPPERS ET AL., 1995).
Parallel filters can provide very valuable estimates of clogging evolution, but it is difficult to
extrapolate laboratory conclusions to field conditions. This is especially questionable when
natural heterogeneity in the aquifer is high, what is very usual. Consequently, standard media are
being tested to enable comparing the clogging potential of water (HIJNEN ET AL., 1998).
R E L A T IO N S H IP C L O G G IN G -K -M F I
100.0 W ell A -6
0.167 W ell 97
N O R M A L I ZED C L O G G IN G R A T E
0.386
10.0 W ell 11A

0.083 Garf ield W ell
0.015
(ft/m o n th )
W ell C-14B
1.0 0.052
W ell B-44B
0.060
W ell C-26A
0.1 0.043
Metr o W ell
Oc . Rd. W ell
1.750
0.002
0.0
1 10 100 1,000 10,000
C O N D U C T IV IT Y (g p d / ft2 )
Figure 2.4. Normalised clogging rate versus estimated hydraulic conductivity (HUTCHINSON & RANDALL, 1995).
If the concentration of suspended solids in recharge water is also taken into account, it is possible to observe a
relationship among the 3 factors. This is a useful tool for physical clogging in recharge wells, but is not applicable
when biochemical processes play a role in the system. (Notice that units are non-metric in this example).
2.2.3.3. Special charts
Only the Hutchinson’s plot (HUTCHINSON & RANDALL 1995) is mentioned here, because it is
particularly important. It incorporates three different magnitudes, each one related to basic
properties of a recharge system: hydraulic conductivity of the aquifer, concentration of suspended
particles and clogging (normalised) build-up (Figure 2.4.). Suspended particles can be expressed by
17
means of conventional parameters such as the total concentration of suspended solids (TSS, usually
in mg·L-1) or other parameters, such as the MFI (in s·L-2).
Table 2.4. Recommended values for basic parameters.
CONCEPT LIMITATIONS COMMENTS

9 -1
[TSS] < 2 mg·L if K > 40 m·d-1
1,9
Deep [TSS] < 0.1 mg·L-1 if 4 m·d-1 < K < 40 m·d-1
2
PHYSICAL
Systems MFI < 3÷5 s·L-2
1,3
CLOGGING Tur < 1 NTU
Surface [TSS] < 10 mg.L-1
Systems Tur < 5 NTU
4
pH > 7.2 (iron bacteria)
4
Deep Avoid CO2
4
BIOLOGICAL Systems DOC < 2 mg·L-1 according to Swiss standards
4
CLOGGING Eh > 10 mV
Surface
TOC < 10 mg·L-1
Systems
6
[Fe2+]GW < 11.2 mg·L-1
7
[Saturation Index
8 More important for abstraction
Ryznar Index > 7
CHEMICAL 8 than for recharge wells
pH < 7.5
CLOGGING 8 Saturation indexes of minerals
Low [Ca2+] and [Mg2+]
8 are essential.
TDS <150 mg·L-1
8
[Cl-] < 500 mg·L-1
8
DO < 2 mg·L-1
8
Ryznar Index < 9
8
pH > 7
CORROSION 8 Applicable to wells and pipes
CO2 < 50 mg·L-1
8
[TSS] < 1,000 mg·L-1
8
Avoid H2S
1
Dutch standards (Peters, pers. comm.); 2Olsthoorn (1995); 3German standards (Schöttler, pers. comm.); 4Degalier
(1987); 5van der Kooij & Hijnen (1990); 6Lucas et al. (1995); 7Hills et al. (1989); 8Ford (1990); 9Pyne (1995); 10Hurni.
2.2.4. RECOMMENDATIONS / GUIDELINES
This section contains recommendations to minimise clogging. The suggested limiting values have
been obtained mostly from the literature and from commonly accepted operational practices, while
the remaining values were proposed by P ÉREZ-PARICIO & CARRERA (1999a).
Table 2.4. displays a set of threshold values, where differences between surface and deep systems
are explicitly remarked. Notice that some of these values may be difficult to meet, especially for
recharge wells. However, they can be relaxed if other conditions apply (limitations used in the
Netherlands are very tight, as there are many fine sand aquifers).
Table 2.5. shows simplified criteria to determine the extent of clogging based on a few simple
parameters. Such criteria are only indicative, as not all basic magnitudes are included (light,
temperature, other nutrients, hydraulic conductivity, recharge regime). However, suggested values
should be accounted for to improve management.
18
Table 2.5. Guidelines for clogging, based on simple parameters.

CLOGGING RECHARGE WATER REDEVELOPMENT
SURFACE SYSTEMS TSS < 10 mg·L-1

Natural drying and cracking
SLIGHT Tur < 5 NTU
Once a year: Mechanical
TOC < 10 mg·L-1
TSS = 10÷20 mg·L-1 Frequent drying and cracking

NOTABLE
TOC = 10÷25 mg·L-1 Twice a year: Mechanical
TSS > 20 mg·L-1

STRONG Prefiltration recommended
TOC > 25 mg·L-1
TSS < 10 mg·L-1

Frequent pumpimg
SLIGHT Tur < 5 NTU
DEEP SYSTEMS
Surging/Jetting: once a month

TOC < 10 mg·L-1
TSS = 10÷20 mg·L-1 Pumpimg once a day

NOTABLE
TOC = 10÷25 mg·L-1 Surging/Jetting: once a week
TSS > 20 mg·L-1 Daily pumpimg

SEVERE
TOC > 25 mg·L-1 Adapted protocol
19
Integrated Modelling of Clogging Processes in Artificial Recharge Systems
2.3. MODELLING
Quantitative approaches to clogging are basic to provide information about the effectively relevant
parameters and interactions. The increasing interest for alternative water management techniques
and –especially- for water production industry has motivated the development of physical clogging
models. Bioclogging models have received much attention too because of the concern about water
quality issues. The recent importance of reactive transport -particularly with respect to deep
repositories- has promoted research into reactive transport modelling. In general, all these studies
focus on one particular problem.
However, evidence that contaminants may adsorb on colloidal particles and that bacteria play a
crucial role in aquatic environments demonstrates the interconnectivity of different processes. But it
is not possible to find comprehensive models in relation to clogging of porous media, because
current models are concerned about specific individual processes. Most of the models do not
concentrate in the time evolution of porous medium properties. Porosity changes are hardly taken
into account, despite the great influence on the hydraulic behaviour of real systems.
This section makes a distinction between empirical (phenomenological) and theoretical models. The
former are only applicable to specific situations, given the site-dependence in terms of
hydrogeology, water quality and environmental properties. Yet, they can be very useful to obtain
preliminary assessments of clogging potential. The latter are more sophisticated and require a good
deal of data, although can give more insight into the clogging mechanisms and their relative
importance, thereby improving our understanding of the whole problem.
A detailed listing of models lies beyond the scope of this document. Further information on
modelling of physical clogging can be found in SPIELMAN (1977) and OSEI-BONSU (1996).
VANDEVIVERE ET AL. (1995), and HOLM (1999) give insight into fundamental issues of biological
clogging models. More comprehensive reviews on clogging modelling can be found in CUSTODIO
(1986), TAYLOR & JAFFÉ (1990a), BAVEYE ET AL. (1998) and PÉREZ-PARICIO & CARRERA (1999a).
2.3.1. EMPIRICAL MODELS
Empirical models establish a relationship between the time evolution of pressure build-up or
infiltration rate and some key parameters, such as the concentration of suspended solids and organic
carbon, for instance. The common feature of these models is that they are extremely site-dependent:
climatic conditions, aquifer properties, quality of recharge water and native groundwater, and
operational rules make them applicable only to the conditions they have been developed for.
Most of the relationships have been developed for surface systems, although it does not imply that
this kind of relationships can not be obtained for deep systems. In this document only a few models
are described.
2.3.1.1. Exponential decrease
A generalized equation that properly describes many clogging problems was proposed by BIANCHI
ET AL. (1978). This equation is formally identical to the equation introduced by HORTON (1933) to
quantify the evolution of the infiltration capacity of a soil with time in rain conditions. This
equation, applied to clogging, is as follows:
20
−ωH t
i = i ∞ + (i 0 − i ∞ )e [2.7.]
where i [LT-1] is the infiltration rate, defined as the actual recharged volume per unit area;
subscripts 0 and ∞ are referred to the initial and final values, respectively. An empirical or
adjustable parameter, ωH [T-1], controls the velocity of clogging. Similar equations, without the
asymptotic term, had already been used in the 50’s (BOUCHER, 1947) to describe the clogging
impact on surface systems. Figure 2.5. shows the typical evolution of infiltration rate.
i0
(i n i ti a l va l u e )
I N F I L T R A T I O N R A T E (i)
i00
(fi n a l v a l u e )
TIM E
Figure 2.5. Plot of HORTON-type equation with final asymptotic value.
Equation [2.7] has been successfully applied to biological clogging too (SCOTT ET AL., 1974). In
that case, the constant ωH (clogging constant) represents the biological growth constant rate when
there is no competition and the substrate (food) is unlimited. IVES (1975) stated that a linear relation
between pressure drop and filtration time should be used for initial times; after this time lag, surface
filtration starts and BOUCHER’s law should be used. There exist variations on Boucher’s law, such
as the use of power series (PHILIP, 1969) and potential decrease expressions, like KOSTIAKOV’s
equation (CHILDS, 1967)
2.3.1.2. Site-dependent equations
There are many site-specific relationships whose aim is to find a correlation between the time
evolution of key variables and a few measurable parameters. Only some of them are presented here.
One of the classical relationships consists of correlating the infiltration rate with the concentration
of suspended solids. For instance, this was done for the Peoria aquifer (CUSTODIO, 1986), where the
ratio of suspended matter reduction, χ [L0], was used as a parameter. The removal of suspended
matter during infiltration was assumed to be constant along time, as indicated by HARMESON ET AL
(1968):
−0.25
χ= 1.000 − 0.027B P (i ⋅ TSS) 0.33 d g [2.8.]
21
where Bp [L], in m, is the width of the filter material; TSS [ML-3], in mg·L-1, denotes the
concentration of suspended solids in recharge water, and dg [L], is the mean diameter of the filter
grains. With these units, the infiltration rate, i [LT-1], must be expressed in m·d-1.
SIEGRIST & BOYLE (1987) obtained a good correlation between infiltration rate at a given time t and
the cumulative loading of total biological oxygen demand, tBOD [ML-3]a −defined as the
carbonaceous plus the nitrogenous biochemical oxygen demand, BOD [ML-3]− plus the cumulative
density loading of total suspended solids, TSS [ML-3], for pilot infiltration cells:
241 exp[2.63 − 5.70( tBOD) + 41.08(TSS) − 0.048( tBOD ⋅ TSS)]

i= [2.9.]
1 + exp[2.63 − 5.70( tBOD) + 41.08(TSS) − 0.048( tBOD ⋅ TSS)]
They insisted that the deduced relation can not be used for other sites, because the differences in
specific soil, climatic conditions, infiltration surface (geometry and depth) properties, wastewater
quality and application regime.
TAYLOR & JAFFÉ (1990a) published a very interesting analytical bioclogging model accounting for
changes in the main parameters of the aquifer. But they previously attained an empirical correlation
(TAYLOR & JAFFÉ, 1990b) that quantifies the current value of intrinsic permeability, k [L2], relative
to the initial value, k0 [L2], as a function of bacterial organic carbon, BOC [ML-3]. This parameter is
defined as the total organic carbon, TOC [ML-3], minus the concentration of soluble organic carbon,
SOC b [ML-3].
k  exp { γ 1 [ BOC] + γ 2 [ BOC]2 } [ BOC] < 0.4 mg ⋅ cm −3

= [2.10.]
k0  γ3 [ BOC] > 0.4 mg ⋅ cm −3
where γ1 [M-1L3], γ2 [M-2L6] and γ3 [L0] are three adjustable parameters. Strictly, k and k0 are
tensors, but a scalar value was considered in this case (1-D vertical flow).
For recharge wells and physical clogging, OLSTHOORN (1982) obtained a semi-empirical equation
that is widely used for comparison purposes, because it considers standardised values that can be
easily extrapolated to particular cases:
2 TSS ⋅ µ l  2
where ω0 ≡   
i0
i= ,  i0 [2.11.]
1 + ω0 t  P  k 
HUTCHINSON & RANDALL (1995) developed a useful standard graph to assess the risk of physical
clogging in wells by using [1] the normalised clogging build-up as introduced by OLSTHOORN
(1982) and [2] type-curves for hydraulic conductivity and the modified fouling index. As mentioned
above, it is also possible to use the concentration of suspended solids instead of MFI.
a
Concentration can be expressed in several ways. The most frequent way consists in giving the mass of solute per unit
volume of solution [mg·L-1], but it is also frequent to do it in mmol·L-1, meq·L-1 or in relative terms [mg solute per mg
solvent]. This implies that proportionality factors must be used to transform between different formulations.
b
The soluble organic carbon, SOC, is obtained by filtration of recharge water through a 0.45 µm membrane.
22
2.3.2. THEORETICAL MODELS
There are numerous theoretical models in the literature too. None of them opts for an integrated
approach to the problem of clogging, but focuses instead on any of the individual mechanisms.
Other models deal with complicated reactive transport equations, but they ignore the changes in
porosity (and consequently in intrinsic permeability) caused by the physiochemical reactions. A
schematic list of models is presented below.
2.3.2.1. Physical Clogging Models
• Macroscopic model by Iwasaki
The first known and widely equation used was proposed by IWASAKI (1937). He introduced a
1-D macroscopic model to describe the behaviour of the concentration of suspended particles in
the liquid, clp [ML-3]c, as a function of penetration length in the medium, z [L]:
∂ c lp
= − λ 0 c lp [2.12.]
∂z
where λ0 [L-1] is the widely used filtration coefficient, which probably represents the most
important feature of this model.
• Macroscopic approach by Herzig et al
HERZIG ET AL. (1970) proposed a crucial model for the time evolution of concentration of
retained particles that is presented here in detail. A basic assumption is that interaction between
suspended particles is neglected, which can be a considerable oversimplification if a
coagulation/flocculation treatment is applied.
The starting point is the well-known advection-diffusion equation for the concentration of
suspended particles under steady state flow conditions:
∂ p p ∂ clp ∂ 2 c pl
(cs + φ cl ) + q −D =0 [2.13.]
∂t ∂z ∂ z2
where D [L2T-1] is the diffusion coefficient; φ [L0] is porosity; and q [LT-1] is Darcy’s velocity.
Concentrations in this model are expressed in volume of particles per unit of bulk volume. For
instance, csp is the volume of retained particles per unit of filter volume. In this model, the
porosity changes as a result of the attachment of particles in accordance to:
φ = φ 0 − β s c sp [2.14.]
where φ0 [L0] is the initial porosity of the filter and βs [L0] is the inverse of the compaction factor
of retained particles. In this model, βs was estimated at 2 or 3, and accounted for the effect of
imprisoning ‘dead liquid’ among the particles, which would affect the effective accumulated
volume of particles. However, βs is difficult to determine, since it depends on particles size and
c
Usually, the concentration of particles (p) in the liquid phase (l) is expressed in dimensionless form, i.e., in mass of
particles per unit mass of the liquid (Chapter III), wlp [L0].
23
shape and the texture of the porous medium (HERZIG ET AL., 1970). Further, βs cannot be
considered constant, since the blocking of a pore by one or more particles does not exclude the
deposition of more particles, which would modify the value of this parameter without
significantly affecting the flow distribution.
After establishing the mass balance for particles, they investigated three approximations:
3rd approximation: neglecting the diffusive term in equation [2.13.].
2nd approximation: taking a constant porosity value in equation [2.13.]
1st approximation: neglecting the contribution of suspended particles to the time derivative in
equation [2.13.], i.e. φ clp << csp.
However, the key aspect of this work was to conceptually define the form of the kinetic equation
that describes the mass transfer rate between mobile and retained particles. This was not done
theoretically because the mass transfer is a function of elementary processes, but was inferred
from experiments and also from the definition of attachment probabilities. The resulting kinetic
equation of clogging is:
∂ c sp
= λ ' c lp [2.15.]
∂t
where λ’ [T-1] is called the clogging rate constant.
It is very interesting to remark that Iwasaki’s equation is not exactly a kinetic equation, since
manipulating both equation [2.12.] and [2.13.] -with the third approximation (i.e. no diffusion)-
yields:
∂ (c sp + φ c pl )
= q λ 0 c lp [2.16.]
∂t
However, when the first approximation is justified then a kinetic equation is obtained through the
use of Iwasaki’s filtration coefficient,
∂ c sp
= q λ 0 c lp [2.17.]
∂t
i.e. the clogging rate can be approximated as the filtration coefficient times the flow velocity.
Finally, it is worth explaining that a global clogging rate constant can be obtained if both
attachment and detachment are lumped into a single parameter, whose definition is:
λ' d c sp
λ' = λ'a φ − [2.18.]
c pl
where λ’a [T-1] is the attachment rate and λ’d [T-1] is the detachment rate. However, only the
attachment term was considered in the work published by HERZIG ET AL. (1970).
• Microscopic attachment and detachment rates
More refined models were subsequently suggested to provide theoretical basis for the kinetics
coefficients (attachment and detachment parameters). A sophisticated approach was made by
YAO ET AL. (1971) in order to improve the description of the mass transfer rates by means of a
24
microscopic analysis. A demonstration of the power of microscopic analysis is contained in

RAJAGOPALAN & TIEN (1976), who separated the attachment rate into three additive terms:
Brownian diffusion, interception and gravity. By solving the trajectory equation for the motion
of a particle they obtained the following relationship:
D 1 17 ⋅ 10 − 3 1 / 2 − 2 / 5 
λ'a = B 1 / 8 15 / 8
 A s N LO N In N Pe + N Gr N In N Pe + A s N1Pe/ 3  [2.19.]
dg  4 20 
where DB [L2T-1] is the Brownian diffusion coefficient of the particles. As [L0] is a geometric
factor that depends on porosity; NLO, NIn, NPe and NGr [L0] are, respectively, the London,
interception, Peclet and gravity numbers. These five numbers are defined by next equation:
2 (1 − y 5 )
As =
(2 − 3y + 3y 5 − 2 y 6 )
4H dp
N LO = N In = [2.20.]
9πµ l d 2p q dg
d gq d 2p (ρ s − ρ l )g
N Pe = N Gr =
DB 18 µ l q
where y≡(1-φ)1/3. H [ML2T-2] is the Hamaker constant that controls the strength of London-van
der Waals attractive forces; dp [L] is the mean diameter size of the suspended particles; ρs and ρl
[ML-3] are, respectively, the density of particles and liquid; and g [LT-2] is the gravity
acceleration modulus. The model assumed that suspended particles and filter grains were
spherical. This, together with the use of attractive electrokinetic forces, was questioned by
several authors (TOBIASSON & O’MELIA, 1988; ELIMELICH & O’MELIA, 1990).
An important characteristic of clogging is that retained particles can act as collectors as well,
behaving like soil grains. This process is termed ‘ripening’, and was first included by O’MELIA
& ALI (1978) for a mono-disperse solution. They modified the removal efficiency to incorporate
the effects of ripening, that is, the new removal efficiency being equal to the sum of the former
efficiency plus the collected particles efficiency. This model also experienced several
modifications, like the one proposed by TARE & VENKOBACHAR (1985). In their approach,
surface coverage changes during the filtration process and there are important variations of the
concentration of retained particles with depth.
Other modifications focus on the effects of heterodisperse solutions (VIGNESWARAN & AIM,
1983; DARBBY ET AL., 1992) and on heterodisperse solutions plus breakthrough -that is,
detachment- (VIGNESWARAN & CHANG, 1986; VIGNESWARAN & TULACHAN ,1988). According
to CLEASBY (1990), filtration is a dynamic process consisting of a ripening stage, a period of
steady operation and a breakthrough stage. An interesting qualitative review, grouping different
analytical models looking at their main assumptions, is done by MORAN ET AL. (1993).
• Modifications of Herzig et al. model
There are various models that have been successfully used to interpret laboratory data. For
instance, RAJAGOPALAN & CHU (1982) used an analogous model -equations [2.13.], [2.15.] and
[2.18.]- but they: (1) explicitly included detachment and (2) related the attachment and
25
detachment rates to the physical clogging mechanism. The formal expression of this model is as
follows:
∂ c lp ∂ c lp ρs a s' (1 − φ) ∂ c sp
+ um + =0
∂t ∂z φ ∂t
[2.21.]
∂ c sp
= λa cl − λ d cs
' p ' p
∂t
where um [LT-1] is the water velocity and a’S [M-1L2] is area available for adsorption per unit
mass of grains. In equation [2.19.], clp is expressed as number of suspended particles in the liquid
per unit volume of suspension, whilst csp represents the number of retained particles per unit area
of grains. This is why the attachment and detachment rates, λ’a and λ’d, are not identical to the
ones appearing in equation [2.18.].
According to empirical observations, this model tended to underestimate attachment when

suspended solids and grains had the same charge, that is, when repulsive electric forces became
relevant. In fact, particles already adsorbed on grains begin to influence the flow very soon: they
act at very low coverage percentages (2% to 4%), due to geometrical effects and not to the
surface charge. The model proved to be quite accurate when suspended particles and grains had
opposite charges.
A similar 1-D macroscopic model was formulated by ADIN & REBHUN (1987), including
retention and re-mobilisation of suspended particles, by means of a simplified transport equation
without diffusion:
∂ cp ∂ cp
q l + s =0
∂x ∂t
[2.22.]
∂ c sp
= λ 0 q cl (F − cs ) − λ d cs J
p p ' p
∂t
where F [L0] is the filter capacity, which determines the maximum retention potential of the
medium; and J [L0] is the hydraulic gradient.
This model is equivalent to Herzig’s model without diffusion and using the first approximation,
i.e. φ clp << csp. The main difference lies in the form of the attachment and detachment terms.
The accumulation term is based on physical considerations, although the concept of the filter
capacity may be questionable (CLEASBY, 1984). The detachment term is also a subject of
controversy; some authors argue that it is not relevant at all (IVES, 1969, 1973), whilst it is
included by others (MINTZ, 1964; SHEKHTMAN, 1961). Shear forces increase as clogging
develops, due to a decrease in porosity and the corresponding augment in filtration velocity
(GHOSH ET AL., 1985). A linear dependence of time evolution of the concentration of retained
particles on the concentration of retained particles itself was suggested by (SAKTHIVADIVEL &
IRMAY (1966) and MINTZ (1964). Equation [2.22.] accounts for the influence of hydraulic
gradient on the detachment rate.
GOVINDARAJU & REDDI (1994) proposed a variation on the last model that did not use the
concentration of retained particles, csp. Instead, they worked with the concentration of suspended
particles alone, and introduced two new parameters: the detachability coefficient, ξ [L0], and the
critical shear stress, τcr [ML-1T-2]. The former influences the sink/source term, whilst the latter
26
establishes a limitation between attachment and detachment based on hydraulic conditions. In

sum, attachment and detachment depend on the hydraulic properties and status of the system.
• Purely Hydraulic Models
Various authors have searched for more elemental relationships based on hydraulic
considerations. Only three of these models are briefly described here.
IVES & PIENVICHITR (1965) proposed a model whose basic assumption was that the porous
medium could be represented by a bundle of tortuous, non-interconnected capillary tubes of
varying radii. Their predicted hydraulic conductivity reduction can be expressed as:
3 − 2 α IP 3 − 2 α IP
K  φ − φ0   β s c sp 
=  =   [2.23.]
K 0  φ0   φ0 
where K and K0 [LT-1] are the current and initial value of hydraulic conductivity, respectively
(strictly, K is a tensor); βs [L0] was defined in the model of HERZIG ET AL. (1970) and αIP [L0] is
a variable positive exponent that depends on how the diameter of the capillaries is assumed to
vary with time as a result of clogging. A value of αIP=1.25 (net exponent of 0.5) is equivalent to
assuming uniform coating of the tubes.
A generalisation of equation [2.23.] was implemented by TIEN ET AL. (1979):
−4 / 3
K  φ cp 
= (1 − c sp ) 3 1 + 0 s  [2.24.]
K0  1 − φ0 
OSEI-BONSU (1996) proposed a mathematical model for physical clogging based on experimental
results obtained under confined radial-flow conditions in a 90º laboratory-scale segment.
According to this model, the ratio of hydraulic conductivity reduction with time can be written as
a function of time and two parameters, αOB and ωOB (OSEI-BONSU & ARMSTRONG, in
preparation):
  K 1 / 3 
1 −    = (αOB1 t ) αOB 2 [2.25.]
  K 0  
where αOB1 [T-1] is a proportionality factor; and αOB2 [L0] is a parameter determined by the
dominant clogging mechanism. It is 0.5 for blocking filtration and 1.0 for deep bed filtration.
The model is based on the assumption that the changes in hydraulic conductivity caused by the
deposition of suspended particles can be assimilated to a series of steady-state changes of the
hydraulic gradient. One important finding of OSEI-BONSU was that, for the same mass flux, the
lower the concentrations of suspended solids the higher the rate of clogging.
2.3.2.2. Biological Clogging Models
Three types of bioclogging models exist: (1) macroscopic models, (2) models based on the concept
of micro-colonies, and (3) biofilm models. Although the conceptual framework of each group may
look different, BAVEYE & VALOCCHI (1989). have shown that macroscopic transport equations for
each of the groups are formally identical if the micro-colonies and the biofilms are fully penetrated.
27
The three approaches are described below by means of representative examples found in the
literature.
(i) Strictly macroscopic models
The best example of this type of models is provided by OKUBO & MATSUMOTO (1979). Their model
could be applied to other clogging sources, given the fact that only hydraulic considerations are
made. Unlike other biological models, they did not implement bacterial metabolism. Instead, they
considered the porous medium to be formed by a bundle of capillary tubes of equal initial diameter,
dt0 [L], embedded in the soil. According to the HAGEN-POUSEUILLE equation, the initial specific
discharge:
π d 4t 0 ρ l g
Q0 = −N t J [2.26.]
128 µ l
where Q0 [L3T-1] is the discharge; Nt [L-2] is the number of tubes per unit area and J [L0] is the
hydraulic gradient.
At time t, the discharge, Q, will be defined by an analogous equation but the diameter of the tubes
will be reduced due to bioclogging. The first assumption of the model was that the reduction in
diameter was uniform. Under the second assumption, the volume decrease per unit length of tube is
proportional to the cumulative discharge through an adjustable parameter:
∂ Vp π 2
= (d t 0 − d 2t ) = α OM 1V [2.27.]
∂z 4
where Vp [L3] is the volume of the void space (pores); dt [L] is the reduced diameter of the tubes;
αOM1 [L] is a fitting parameter; and V [L] is the cumulative discharge.
After some algebraic manipulations of the previous equations, it is possible to write a more compact
expression for the discharge affected by bioclogging:
∂V
Q= = Q 0 (1 − α OM 2 V ) [2.28.a]
∂t
where
4 α OM1
α OM 2 ≡ [2.28.b]
π d 2t 0
After integration, with the initial condition that V=0 at t=0, if is finally derived that:
Q0 t
V= [2.29.]
1 + α OM 2 Q 0 t
VIGNESWARAN & SUAZO (1987) used a modification of equation [2.29.] to predict bioclogging at
different infiltration rates. Predictions were based in laboratory experiments that served to obtain
the empirical coefficient, αOM1 (or αOM1). Results obtained showed a good agreement with
measured data. However, VANDEVIVERE ET AL. (1995) have demonstrated that the model defined by
OKUBO & MATSUMOTO is just a special case of the model proposed by IVES & PIENVICHITR (1965)
when αIP=0.5 in equation [2.23.].
28
(ii) Models based on Micro-colonies
Another approach consists in assuming that bacteria grow in discrete micro-colonies that are
attached to the porous medium. Micro-colonies are supposed to consist of isolated cylinders of
radius rc [L], height bc [L], mass mc [M], and density ρc [ML-3]. The volume of a micro-colony, Vc
[L3], can thus be written as Vc=πrc2 bc. The colony density is defined by NC [L-3] . Micro-colonies
may be separated from the water phase by a diffuse layer of thickness δc [L].
In accordance with previous investigations (RITTMAN & MCCARTY, 1980; BENEFIELD & MOLZ,
1985), MOLZ ET AL. (1986) developed a model for aerobic heterotrophic bacteria. This 1-D model is
described by the following set of equations:
 cs   co 
Y Rs = µ ρ c Vc  s c s   o c o  [2.30.a]
 g + c c   g + cc 
max
 co 
ro = f syn
o
Y R s + f eno µ dec m c  o c o  [2.30.b]
 g + cc 
 c s − c sc  2
D s  l  π rc = R s [2.30.c]
 δc 
 c o − c oc  2
D o  l  π rc = R o [2.30.d]
 δc 
∂
{φ c sl + c ss (a s − N c π rc2 )}= − ∂  φ u m c sl − φD s ∂c l  − D s  c l − c c
s s s
 N c π rc2 [2.30.e]
∂t ∂x  ∂x   δ 
∂
{φ c lo + c so (a s − N c π rc2 )}= − ∂  φ u m c lo − φ D o ∂c l  − D o  c l − c c N c π rc2
o o o
[2.30.f]
∂t ∂x  ∂x   δ 
• Equation [2.30.a] is a modification of MONOD kinetics for the utilisation of nutrients (substrate).
Y [L0] is the yield coefficient that (stoichiometrically) relates the growth of microorganisms to
the utilisation of substrate; Rs [MT-1] is the rate of utilisation of the substrate; µmax [T-1] is the
maximum growth rate MONOD’s kinetics; ccs and cco [ML-3] are the substrate and oxygen
concentrations within the colony, respectively; and gs and go [ML-3] are the half-saturation
constants for substrate and oxygen.
• Equation [2.30.b] incorporates PIRT kinetics for oxygen consumption by the micro-colonies. Ro
[MT-1] is the rate of utilisation of oxygen; fosyn and foen [L0] account for fraction of oxygen used
for synthesis and energy, respectively; µdec [T-1] is the microbial decay coefficient.
• Equations [2.30.c] and [2.30.d] refer to diffusion of substrate and oxygen, respectively, through
the diffuse layer. DS and DO [L2T-1] are diffusive parameters for the substrate and oxygen; cls and
clo [ML-3] are the substrate and oxygen concentration in water, respectively.
• Equations [2.30.e] and [2.30.f] account for the transport of substrate and oxygen within the
aquifer formation. css and cso [ML-2] are the mass of adsorbed substrate and oxygen per unit
area; as [L-1] is the specific reactive surface of the matrix.
A Freundlich isotherm is employed to relate the adsorbed mass of substrate or oxygen concentration
in water. After some considerations about the net biomass production rate for a colony and for the
total volume, they came to a final equation that, together with equation [2.30.a] constitutes the
proposed model:
29
∂ Nc
∂t
=
1
mc
(
Y R s − µ dec m c N c ) [2.31.]
To adopt equation [2.31.], a constant average colony size and constant mass was assumed.
(iii) Biofilm Models
The third approach includes a new phase, the biofilm, composed by bacteria and a matrix of extra-
cellular material, which forms a continuum along the porous medium. The biofilm is mainly
constituted by water, although interactions with the liquid phase must be quantified too. The basic
hypotheses of TAYLOR & JAFFÉ (1990a) open pore biofilm model, whose basic hypotheses are:
• Rigid and saturated porous medium that is chemically inert.
• There is a unique limiting substrate that cannot sorb onto the porous medium.
• MONOD kinetics apply, and bacterial decay follows a first order law.
• The aquifer parameters, such as porosity, permeability and dispersivity, change as a result of
biomass attachment and detachment; then, biofilm thickness controls the variation of those
parameters.
• Mass balance equations are established for biomass, waster and the substrate. Transport
equations include advection, diffusion, production and decay.
The model is defined by the following equations:
∂
(φ c bl ) = ∇(φ D∇c lb ) − ∇(q c lb ) + φ ρ l R bl + R she − R det [2.32.a]
∂t
∂
( w bb ρ b ) = w bb ρ b R bb + R det − R she [2.32.b]
∂t
∂
(φ c sl ) = ∇( φ D ∇c sl ) − ∇(q c sl ) + φ ρ l ( R slb ) [2.32.c]
∂t
• Equation [2.32.a] is the mass balance for the biomass in the liquid, whose concentration is
defined by clb [ML-3]. Rlb [T-1] the net rate of biomass production in the liquid phase, which is
described through MONOD’s equation and a first-order decay relationship; Rshe [T-1] is the shear
rate of biomass from the biofilm due to fluid mechanical forces; and Rdet [T-1] is the rate of
deposition of biomass from the liquid to the biofilm or the solid phase. Exact expressions for the
shear and deposition rates are based in previous developments of different authors. A classical
first-order dependency on the concentration of biomass in the liquid was adopted for R det (like
HERZIG ET AL., 1970), whilst the shear term, Rshe, was taken from SPEITEL & DIGIANO (1987).
• Equation [2.32.b] is the mass balance for the biofilm phase (the biomass occurs in a continuous
film, the biofilm). The volume fraction of biofilm phase is denoted by wbb [L0]. The net rate of
biomass production in the biofilm, Rbb [T-1], is expressed by integration of MONOD’s equation
over the biofilm thickness, Lb [L].
• Equation [2.32.c] represents the mass balance for substrate in the liquid. In analogy to the
previous equations, Rlbs [T-1] is sum of (1) the rate of substrate utilisation by the biomass in the
liquid phase and (2) the transfer of substrate from the liquid to the liquid phase by diffusion
across the boundary. Explicit relationships can be consulted in TAYLOR & JAFFÉ (1990a).
A key aspect of this model is the relationship between porous media parameters and biological
growth. This is a must for any clogging model, irrespective of the clogging mechanism. Within the
30
group of conduit flow models, two basic approaches can be used to describe changes in porous
media due to bioclogging:
1. Geometrical models do not consider the size distribution of pores in the medium. The KOZENY-
CARMAN model (KOZENY, 1927; CARMAN, 1937) exemplifies this approach:
3 2
k  φ   a s0 
=    [2.33.]
k 0  φ 0   as 
where as and as0 [L-1] are the current and initial values of the surface area of the porous medium
(internal surface area per unit bulk volume), respectively.
Equations that relate the changes in porous media parameters and biofilm coating of the grains
are based on DEB (1969). In case of cubic packing, the initial porosity and surface area of the
medium can be written as:
π
φ0 = 1 −
6
[2.34.]
π
as =
dg
Using these concepts, TAYLOR ET AL. (1990) arrived at the following mathematical expressions
for the porosity and surface area evolution of a porous medium subject to bioclogging –for cubic
packing-:
1  2 L  1  2 L  1  2 L  1
3 2
φ = 1 + π   b  +  b  −  b 
− 
   
 3  d g  4  d g  2  d g  6 
[2.35.]
  2L   2L  
2
a s = π 1 −  b 
− 2  b 

  g   g  
 d  d
whilst KOZENY-CARMAN equation fully describes the time evolution of hydraulic conductivity
due to bioclogging.
2. Statistical models consider the random nature of the interconnectedness of the pores. The cut-
and-random-rejoin-type model by Mualem (1976) proposes the following equations for
porosity, surface area and intrinsic permeability of the medium in the absence of clogging:
rmax
φ = ∫ f ( r ) dr
rmin
rmax
f (r )
a =2 ∫ dr [2.36.]
s rmin r
2
γ 
κ φ  rmax 
k= ∫ r f ( r ) dr
8 r 
 min 
where f(r) is the pore size distribution of the medium, which is defined by dθ≡f(r)dr, and θ(r) is
the unsaturated moisture content when all pores with radius smaller that r are filled; rmin and rmax
31
[L] are, respectively, the minimum and maximum pore radius; κ [L0] and γ [L0] are two
constants. MUALEM (1976) suggested that γ=0.5.
TAYLOR ET AL. (1990) particularised this formulation by using a power function to characterise
the pore size distribution. After applying the biofilm concept to MUALEM’s model, the
expressions for biofilm-affected porous media parameters can be obtained by modifying
equation [2.36.]. The exact expressions are available in TAYLOR ET AL. (1990).
(iv) Performance of Bioclogging Models
VANDEVIVERE ET AL. (1995) compared three types of models: (1) the hydraulic model developed by
IVES & PIENVICHITR (1965), (2) the KOZENY-CARMAN model of TAYLOR ET AL. (1990) model, and
(3) the MUALEM model of TAYLOR ET AL. (1990). Despite the different conceptual frameworks of
these models, all of them achieved reasonable results for coarse-textured materials. However, the
models have a tendency to underpredict the hydraulic conductivity reduction in fine-textured
materials by orders of magnitude.
Since the analysed models assumed a complete coverage of the grains (continuous biofilm),
VANDEVIVERE ET AL. (1995) argued that a simplistic model where biomass was distributed in plugs
would provide better predictions. This simple model is described by equation [2.37]:
K (φ ) 2
= 0 [2.37.]
K plug φ0 − φ
where Kplug [LT-1] is the hydraulic conductivity of the plugs, which accounts for all the head loss
within the capillaries.
This model has severe limitations (too simplistic geometry, ignoring the kinetics of plugs formation,
not incorporating bacterial mechanisms), but can give satisfactory results for fine-textured media. In
order to do so, it is necessary to propose a value for the ratio Kplug/K, i.e. to provide an estimate for
the reduction of conductivity. Values of 10-3 to 10-4 would be acceptable, in principle. However, it
is clear that some estimates have to be done before applying the model.
More recently, HOLM (1999) applied a statistical model to data from laboratory columns. The model
had difficulties in predicting the response to clogging and to obtain a consistent set of model
parameters (they had to be varied with varying input conditions). However, coupling empirical
relationships with detailed numerical modelling seems to be a promising approach to complicated
biological reactions and biofilm-induced clogging.
2.3.2.3.Chemical Clogging Models
Recent developments in reactive transport analytical and numerical (especially finite differences or
finite elements) models are outstanding. However, reactive transport models have not been applied
to clogging problems. This is probably due to the amount of data necessary to verify the conceptual
model and the difficulty to quantify changes in aquifer parameters and to relate them to pressure (or
head) build-up.
Instead of trying to describe those models, which can be found elsewhere (LASAGA, 1981; STEEFEL,
1992; SAALTINK ET AL., 1997), a specific model for bacteria-catalysed iron precipitation observed in
laboratory experiments (OMURA ET AL., 1991) is briefly commented. Their initial hypotheses are:
32
• Iron is the only limiting substrate. The oxygen concentration in infiltrating water is high enough
so as not to limit ferrous iron (Fe+2) oxidation in accordance to MONOD’s kinetics.
• Only the attached cells (bacteria fixed on the medium walls) are important as concerns oxidation
and growth. Mobile bacteria are not considered in the model.
• Cells detached from the support medium are instantly transferred to the liquid phase.
• The medium is assumed to be formed by three phases: the support medium (or matrix), the
stagnant liquid film, whose width is defined by δl [L], and the liquid.
The model performs mass balances for:

• Fe+2 in the support medium (matrix), which is formed by spherical grains of radius Rg [L].
Equation [2.38.a] describes this balance (0 < r ≤ Rg).
• Fe+2 in the stagnant liquid film (Rg ≤ r ≤ Rg+δl), Equation [2.38.b].
• Fe+2 in the bulk liquid, equation [2.38.c].
• Attached bacteria in the support medium, equation [2.38.d].
∂ c ss  ∂ 2 c ss 2 ∂ c ss  N sb µ max c ss
= D 2 + − [2.38.a]
∂t  ∂r r ∂ r  φ g s + c ss
∂ c sw  ∂ 2 c sw 2 ∂ c sw 
= D  +  [2.38.b]
∂t  ∂r
2
r ∂ r 
2
∂ c sl Q  s s 1− φ  (0.5 d g + δ w ) ∂ c sl
= c l − c − 3 
l   D [2.38.c]
∂t VR φ  in
 φ  d 3g / 8 ∂r
R g + δw
∂N b
µ c s
= N sb Y smax − (µ dec + R she ) N sb
s s
[2.38.d]
∂t g +c s
s
where css [ML-3] is the Fe+2 concentration in the support medium; cws [ML-3] is the Fe+2
concentration in the stagnant liquid film; cls [ML-3] is the Fe+2 concentration in the liquid, with
subscript in referring to the input concentration; Nsb [L-3] is the attached cell number in the support
medium; VR [L3] is the reactor volume; and Rshe [T-1] was defined in the model by TAYLOR &
JAFFÉ (1990a).
The problem resides in how to determine the three kinetics parameters: the yield coefficient, Y [L0],
the die-off rate, µdec [T-1] and the detachment rate, Rshe [T-1]. Apart from solving the system of
equations [2.38.a] through [2.38.d], OMURA ET AL. (1991) made a balance for mobile bacteria in the
column (reactor) and another balance for attached bacteria in the column, in steady state, with the
aim of finding a relationship among those parameters. The resulting equation is as follows:
Q N bl
=
[
c sl in − c sl ]
Q Y − µ dec [2.39.]
(1 − φ) VR N sb* (1 − φ) VR N sb*
where Nsb* [L-3] is the attached cell number in the support medium at steady state.
Equation [2.39.] can be used to determine the values of Y and µdec after substituting for the
empirical results, so that they are obtained from the slope and the intercept of a straight line,
respectively. Then, Rs can be deduced from the transient mass balance for mobile bacteria in the
column, which is not shown here.
33
OMURA ET AL. (1991) concluded that pH should be kept below 2 in order to avoid clogging by ferric
hydroxide. Also, they found that temperature variations did not prove to be very important. It is
evident that such a requirement is unfeasible from a practical perspective, but it helps in
understanding the clogging potential of chemical compounds by means of biotic (bacterially-
mediated) and abiotic reactions.
2.3.2.4. Compaction
Compaction is an uncommon situation, which takes place when water height is too big and causes
the clogging layer to compress. Therefore, compaction does not refer to the aquifer formation itself.
The work by PERSON ET AL. (1996) can be consulted for further information. Basically, deformation
of porous media is due to the application of external forces, F [ML-2T-2], which splits into:
(1) internal stress, given by the divergence of the tensor of porous medium stress, -∇σσ
(2) inertial forces, which are usually negligible for slow rates of deformation.
When small elastic strains are applicable to macroscopically continuous porous medium:
F ≈ −∇σ
σ [2.40.]
A constitutive equation is required to relate the strain tensor, ε [L], and the stress tensor, σ
[ML-1T-2]. A classical example is the relationship introduced by BIOT (1941):
ε=
1
E
[ ]
(1 + υ bi ) σ − υTr{σ }δ ij +
P
3H bi
δ ij [2.41.]
where υbi [L0] and Hbi [ML-2T-2] are Poisson’s modulus and a bulk modulus of the porous medium .
Finally, if the resulting displacement velocity, u, is small, then another equation which links
displacement and the strain rate tensor is obtained:
ε = ∇u [2.42.]
This displacement of the porous medium with regard to a fixed reference system must be included
when Darcy’s law is applied, because this fundamental law was originally derived for water flow
velocity referred to a fixed porous medium.
2.3.2.5. New Comprehensive Approaches
As a result of a 3-years period European Union Project on Artificial Recharge, two new clogging
codes were developed: CLOG (PÉREZ-PARICIO & CARRERA 1998a; PÉREZ-PARICIO ET AL., 1998;
PÉREZ-PARICIO & CARRERA, 1999b) and MIKE-SHE SC (VON CHRISTIERSON, 1999). The former is
the object of this thesis, so that will be exhaustively described in the following chapters. The
mathematical formulation of the latter is not presented here, but a qualitative description is
provided. Table 2.6. collects the essential features of both programs.
CLOG is a sphere-type model that can account for attachment and detachment of suspended
particles, bacterial growth on the porous matrix and precipitation/dissolution of minerals.
Multiphase flow, mechanical compaction and heat transport are also considered by the model.
CLOG was generated after linking two existing codes, CODE-BRIGHT (OLIVELLA &
34
GARCÍA-MOLINA, 1996) and RETRASO (SAALTINK ET AL., 1997), together with the
implementation of specific clogging modules (PÉREZ-PARICIO ET AL., 1998).
MIKE-SHE SC is a cut-and-random-rejoin-type (statistical) model that was implemented by adding

new modules to the MIKE-SHE system. MIKE-SHE SC is a soil-clogging model describing
microbial growth and transport of limiting substrates and biomass in porous media. The model is a
new implementation to the existing hydrological modelling system MIKE-SHE, developed at the
Danish Hydraulic Institute (Denmark). Aside from several utility programs and add-on modules
MIKE-SHE comprises two models: a flow model describing integrated water flow and a model for
integrated, multi- component, advective-dispersive conservative and reactive transport (VON
CHRISTIERSON, 1999).
Table 2.6. Some capabilities of the new codes CLOG and MIKE-SHE.
Both codes work with mean diameter sizes of particles and grains, but there are plans to improve this treatment in
the future.
CODE
Aspect \ CLOG MIKE-SHE
3-D ✓ ✓
Mesh Finite elements Finite differences
Physical ✓ ✓
Clogging
Biological ✓ ✓
Chemical ✓ -
Compaction ✓ -
Reactive transport ✓ ✓
Kinetics of bacterial growth Monod First order
Bacterial transport - ✓
Unsaturated flow ✓ Under completion
Temperature ✓ -
Direct substitution approach Sequential non-iterative
Numerical solution
(DSA) Approach (SNIA)
MIKE-SHE SC clogging model belongs to a family of so-called two-compartment models

developed in recent years by TAYLOR & JAFFÉ (1990a), ZYSSET ET AL. (1994), CLEMENT ET AL.
(1996) and HOLM (1999). The model basically consists of one compartment that provides a
description of the clogging processes: transport of bacteria and limiting nutrients, bacterial
growth/decay and detachment/attachment of biomass. Such processes are closely coupled to a
second compartment, which translates the biomass accumulation into soil permeability reductions.
At present, MIKE-SHE SC is limited to describing biogrowth and decay processes and

sedimentation/detachment of organic matter. Thus, the model does not include all biological,
physical and chemical processes that may cause soil clogging, such as mineral precipitation, air
bubble release or soil compaction of the clogging layer. Temperature variations have not been
included either in the model at this stage. The model largely follows the work of CLEMENT ET AL.
(1996), who derived an empirical formula that relates biomass concentrations to the hydraulic
conductivity. This model, which is based on a macroscopic approach and thus makes no
35
assumptions on the biomass distribution, is used in MIKE-SHE SC for relating biomass

concentrations to permeability reductions. The processes of growth/decay and
sedimentation/detachment are described by existing add-on modules in the MIKE-SHE modelling
system. Bacterial growth is solved by the MIKE-SHE BM module using a simple first-order
expression. Sedimentation, detachment and decay of biomass are solved by the sorption-
degradation module MIKE-SHE SD through kinetic first-order expressions.
The existing sorption formulation has been modified to include a fully kinetic isotherm for
describing sedimentation and detachment of biomass based on a model proposed by ADIN &
REBHUN (1987), which includes the effects of shear forces. The system of equations describing
clogging is solved for the saturated zone and partly for the unsaturated zone. The biological model
in the unsaturated zone was under development.
36

Perez Paricio 2

Uploaded by

Copyright:

Available Formats

Perez Paricio 2

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Perez Paricio 2

Uploaded by

Copyright:

Available Formats

Chapter II: State of the Art

(...) «Y siempre reinterpretamos y recordamos el pasado desde los acontecimientos

The chapter focuses on conceptual issues (processes, parameters), prevention-redevelopment and

2.1. CONCEPTUAL ASPECTS

2.1.1. PHYSICAL CLOGGING

10-10 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2

2.1.1.1. Colloidal particles

2.1.1.2. Intermediate particles

2.1.1.3. Large particles

flow is increased by a factor of 3 or 4 (MAROUDAS, 1961; SAKTHIVADIVEL, 1966; HERZIG ET

Figure 2.2. Mechanisms during water filtration.

2.1.2. BIOLOGICAL GROWTH (BIOCLOGGING)

2.1.3. CHEMICAL REACTIONS

2.1.4. GENERATION OF GAS AND AIR

Figure 2.3. Clogging evolution as a function of the predominant process.

2.1.6. INTER-DEPENDENCE OF CLOGGING MECHANISMS

(i) Coupling particles-biology.

Table 2.1. Inter-dependence of clogging mechanisms.

PHYSICAL BIOLOGICAL CHEMICAL GAS COMPACTION

Coloid-Bacteria Surface Forces

Addition of O2 Compression =>

2.2. PREVENTION AND REDEVELOPMENT

Table 2.3. Redevelopment techniques for surface and deep systems.

Surface systems During runs: - Filtration: mats, filters

2.2.3. USEFUL TOOLS

2.2.3.1. Specific parameters

Conventional parameters, such as turbidity and concentration of suspended solids (physical

2.2.3.2. Preliminary assessment of clogging in recharge systems

10.0 W ell 11A

2.2.3.3. Special charts

Table 2.4. Recommended values for basic parameters.

CONCEPT LIMITATIONS COMMENTS

2.2.4. RECOMMENDATIONS / GUIDELINES

Table 2.5. Guidelines for clogging, based on simple parameters.

SURFACE SYSTEMS TSS < 10 mg·L-1

TSS = 10÷20 mg·L-1 Frequent drying and cracking

TSS > 20 mg·L-1

TSS < 10 mg·L-1

Surging/Jetting: once a month

TSS = 10÷20 mg·L-1 Pumpimg once a day

TSS > 20 mg·L-1 Daily pumpimg

2.3.1. EMPIRICAL MODELS

2.3.1.1. Exponential decrease

Figure 2.5. Plot of HORTON-type equation with final asymptotic value.

2.3.1.2. Site-dependent equations

241 exp[2.63 − 5.70( tBOD) + 41.08(TSS) − 0.048( tBOD ⋅ TSS)]

k  exp { γ 1 [ BOC] + γ 2 [ BOC]2 } [ BOC] < 0.4 mg ⋅ cm −3

2.3.2. THEORETICAL MODELS

2.3.2.1. Physical Clogging Models

• Macroscopic model by Iwasaki

• Macroscopic approach by Herzig et al

where λ’ [T-1] is called the clogging rate constant.

• Microscopic attachment and detachment rates

microscopic analysis. A demonstration of the power of microscopic analysis is contained in

• Modifications of Herzig et al. model

According to empirical observations, this model tended to underestimate attachment when

establishes a limitation between attachment and detachment based on hydraulic conditions. In

• Purely Hydraulic Models

A generalisation of equation [2.23.] was implemented by TIEN ET AL. (1979):

2.3.2.2. Biological Clogging Models