Floc properties can be controlled only by selecting a coagulant type and adjusting mixing parameters. However, optimization of mixing is widely overlooked. Well-chosen mixing conditions increase floc separation efficiency, which decreases the operating costs of drinking water treatment plants (DWTPs). This paper presents guidelines for designing mixing parameters in water treatment. Special attention is paid to the determination of mixing intensity and the purpose of different mixing intensities with respect to the subsequent separation. For instance, single-stage separation by filtration must be preceded by homogenization and rapid flocculation mixing. Double-stage separation by sedimentation and filtration should be preceded by homogenization, rapid flocculation mixing and slow flocculation mixing.

  • Floc separation efficiency depends on floc properties controllable by mixing.

  • Mixing has different stages: homogenization and flocculation mixing (rapid and slow).

  • Homogenization and rapid flocculation mixing produce flocs suitable for filtration.

  • Homogenization and rapid and slow mixing should precede sedimentation and filtration.

  • Mixing intensity should be expressed by the global velocity gradient, not rpm.

Graphical Abstract

Graphical Abstract
Graphical Abstract

Floc separation depends on the physical properties of the floc, such as size, size distribution, density and settling velocity. These properties are determined by environmental factors (such as the composition and concentration of impurities) and by chemical (Gonzalez-Torres et al. 2017; Filipenska et al. 2019) and mixing conditions (Bache & Rasool 2001; Coufort et al. 2007; Xu et al. 2010) during flocculation. Therefore, the optimization of these conditions is necessary.

Chemical conditions are still the primary interest. Thus, optimization of the coagulant dose is common. The optimization of pH is less frequent; the importance of this step has been explained in Naceradska et al. (2019). However, the optimization of mixing conditions (i.e., mixing intensity and flocculation time) is usually neglected. Despite this, there are several cases of drinking water treatment plants (DWTPs) where mixing optimization resulted in more efficient separation and thus a more efficient overall treatment process within the DWTP modernization (e.g., Svetla nad Sazavou, Kutna Hora, Milence, CZ).

Therefore, we would like to show and emphasize in this paper that controlling floc properties by optimizing mixing leads to a high separation efficiency of flocs and is as essential as coagulant dose and pH optimization. We propose general guidelines for designing flocculation parameters in water treatment.

The properties of flocs determine their separation efficiency. The only variables that can be chosen and thus set to tailor flocs with suitable properties are coagulant type and mixing intensity. In particular, the role of mixing is often underestimated both in research and in DWTP operation. Therefore, optimization of mixing for a given separation technology is crucial and necessary. Another significant difficulty is inaccurate quantification of mixing conditions. The mixing intensity is very often characterized by the rotation frequency of a certain impeller in revolutions per minute (rpm). However, rpm units provide no information about hydrodynamics. It is, thus, always necessary to give the value of the global velocity gradient in research papers dealing with flocculation.

Mixing is a fundamental condition for efficient flocculation. The quantity used for characterization of the mixing intensity is historically a velocity gradient (shear rate), originally defined for laminar flow in terms of the power input from the mixer
formula
(1)
where P is the power input of the impeller, V is the volume of liquid, η is the dynamic viscosity and G is the velocity gradient.
For turbulent flow that exists in real conditions, the velocity gradient G from the previous Equation (1) is replaced by the global velocity gradient (Camp & Stein 1943)
formula
(2)
There are several possibilities for determining the power input of an impeller. The most precise method is to determine the dependence of the torque of the impeller on its rotation frequency. The power input is then calculated from the relationship
formula
(3)
where ω is the angular velocity, M is the torque and f is the frequency of impeller rotation.
If measuring the torque is not possible, the power input can be determined from the power number (Po) of the impeller usually given by the formula
formula
(4)
where P is the power input, dm is the impeller diameter, f is the impeller rotation frequency and ρf is the water density.
In the case of hydraulic mixing (e.g., in a fluidized bed or various static mixers), a pressure drop Δp (due to hydraulic resistance) arises that can be expressed from the Bernoulli equation as
formula
(5)
where ρf is the water density and hz is the head loss.
The power input P in Equation (2) equals the work (energy) W needed to raise a liquid with volume V and density ρf to a height corresponding to the head loss hz in time t
formula
(6)
By substitution in Equation (2), an alternative equation for calculating the global velocity gradient is obtained
formula
(7)

The velocity gradient is a fundamental quantity for mixing intensity characterization in various mixing devices (Equations (2)–(7)). In contrast, the frequently used units of rpm are insufficient for quantification of mixing intensity because they prevent comparison of different mixing devices.

Flocculation and separation reach the best results when mixing has a few different stages. Each mixing stage has its purpose. The first stage, homogenization mixing, is applied immediately after dosing the coagulant, and its primary purpose is to achieve efficient and rapid dispersion of the coagulant in the treated volume of water. Homogenization mixing is achieved either hydraulically (under plug flow), e.g., by using different static mixers, or mechanically (by back mixing), e.g., by using different impellers (Bratby 2006), and has high intensities of G > 300 s−1 (Morrow & Rausch 1974). In contrast, the purpose of subsequent flocculation mixing is to form flocs that are removable by commonly used separation processes, such as sedimentation, flotation and filtration. Thus, flocculation mixing allows collisions between the particulate pollutants and the coagulant and ensures the binding of these components into flocs. Since each of the separation processes requires different floc properties, different mixing intensities are needed, and flocculation mixing can be further divided into rapid and slow. However, this division is a rough representation, and the flocculation mixing intensity should be adapted solely to the required floc properties.

Rapid flocculation mixing

In practice, rapid flocculation mixing is often mistaken for homogenization mixing. However, the purposes of the two processes are different. Rapid flocculation mixing is a technological process in the initial phase of flocculation. The applied velocity gradients are in the range of approximately 100–400 s−1 (Polasek 2007). Mixing with higher intensities does not further affect floc properties (Mutl et al. 2006). The selected mixing intensity should be applied until the floc properties reach a steady state, i.e., the floc size and density do not change any further (Polasek 2007). Literature data on necessary flocculation times vary considerably due to differences in the definition of rapid mixing, inconsistencies in the way the velocity gradients are calculated, and different ways of evaluating the mixing efficiency. Rapid flocculation mixing produces small and dense flocs that are effectively separable by sand filtration (Ngo et al. 1995; Pivokonsky et al. 2011; Bubakova & Pivokonsky 2012).

Slow flocculation mixing

Slow flocculation mixing is characterized by velocity gradients ranging from 20 to 100 s−1 with flocculation times ranging from approximately 5 to 30 min (Polasek 2007). Slow flocculation mixing should follow rapid flocculation mixing if large and rapidly settling flocs are required. Nevertheless, the usual practice is the application of slow flocculation mixing immediately after homogenization mixing. Such mixing produces flocs with unsatisfactory properties; they are heterogeneous in size, irregular and very porous with a low settling velocity (Coufort et al. 2005; Pivokonsky et al. 2011; Bubakova et al. 2013; Filipenska et al. 2019). The technological process consisting of homogenization mixing and rapid and slow flocculation mixing produces large flocs with high settling velocities (Polasek 2011; Wang et al. 2011), which are effectively separable by sedimentation.

The efficiency of floc separation is influenced by floc properties (size, size distribution, shape, density and settling velocity) that can be controlled by adjusting mixing intensity. Based on the amount and properties of formed flocs, we can distinguish two types of separation: single-stage and double-stage separation.

Single-stage separation by sand filtration

Direct filtration requires flocs that penetrate the entire volume of the filter bed and exhibit sufficient adhesion ability onto the filter cartridge surface (sand). Small flocs, generally <50–60 μm with high density, have been proven to be ideal for direct filtration separation (Ngo et al. 1995; Bache & Gregory 2010; Pivokonsky et al. 2012). In contrast, large flocs (>100 μm) of irregular shape and relatively low density are not suitable for sand filtration separation because they are not able to penetrate deep into the filter bed. They clog the sand filter rapidly, which causes a rapid pressure drop, and filter cycles (filter run times) are shortened to a minimum (Pivokonsky et al. 2011; Bubakova & Pivokonsky 2012).

The pressure profiles in a filter bed during filtration of suspensions (flocs) prepared at different velocity gradients are illustrated in Figure 1. With increasing mixing intensity, the filtration efficiency increases. Large flocs prepared by slow flocculation mixing penetrate only the upper layers of the sand filter. The filter capacity is not fully utilized, and the filter cycles are short. When the mixing intensity increases, the floc properties change, and the flocs penetrate deeper into the filter medium. The sludge capacity of the filters increases significantly, and the filtration cycles are extended (Polasek & Mutl 2002).

Figure 1

Behaviour of pressure in the filtration bed of a pilot plant sand filter (adapted from Pivokonsky et al. (2011)).

Figure 1

Behaviour of pressure in the filtration bed of a pilot plant sand filter (adapted from Pivokonsky et al. (2011)).

Close modal

The velocity gradients and residence time required to form a suspension separable by filtration roughly correspond to the rapid flocculation mixing values (G = 100–400 s−1) (Pivokonsky et al. 2011; Polasek 2011).

Double-stage separation by sedimentation and sand filtration

When a suspension is heterogeneous in size, direct filtration is inappropriate. Therefore, floc separation has to proceed in two successive steps, sedimentation and filtration. Large flocs (>100 μm) with high density and resistance to tangential forces are ideal for separation by sedimentation (Edzwald 1995; Polasek & Mutl 2005). The remaining smaller flocs continue to the filtration step.

It has been suggested (Polasek 2011) and confirmed at several restored DWTPs that large flocs with high density can be prepared by applying a sequence of rapid and slow flocculation mixing (Mixing Purpose: Slow flocculation mixing). The formation of large flocs can be supported using a flocculation aid at the end of a rapid flocculation mixing phase. After homogenization with the flocculation aid, slow flocculation mixing follows.

Double-stage separation by flotation and sand filtration

During dissolved air flotation (DAF), flocs are trapped on the surface of air bubbles that lift them to the water surface. Therefore, small flocs of tens of μm, preferably 25–50 μm, with a low density close to the density of water are required (Edzwald 2010). Even large (>50 μm) low-density flocs formed by low-velocity gradients (G = 10 s−1) are removable by flotation with high efficiency (Vlaski et al. 1997). The reason might be that flocs are broken into sizes suitable for flotation when they enter the contact zone where velocity gradients are 10–100 times higher than those used for floc formation (Bache & Gregory 2010).

Since flotation requires only very small flocs, the residence time in flocculation is usually up to 10 min (Valade et al. 1996; Edzwald 2010). Thus, flocculation does not proceed to a steady state and usually continues in the flotation unit.

The separation of flocs by flotation is often compared with sedimentation. DAF is reported to be more effective than sedimentation in treating water containing cyanobacteria, algae, and humic substances or a low content of turbidity-forming particles (clay minerals and aluminosilicates). The reason for the high flotation efficiency of these pollutants is the low density of the flocs formed by their coagulation (Edzwald 2010). However, this assertion is based on the erroneous assumption that the separation of flocs formed under the same mixing conditions was compared (Khiadani et al. 2013).

Additional practical aspects

It is necessary to realize that the properties of flocs are also affected by their composition. Thus, applying the same velocity gradient to different components (impurities/coagulants) often results in flocs with different properties (Filipenska et al. 2019). This phenomenon can be seen in Figure 2. This fact must also be considered when choosing a separation method. For example, flocs created from a ferric coagulant and cellular organic matter (COM) from Microcystis aeruginosa are suitable only for sedimentation filtration due to their size. If an aluminium coagulant is used instead, even flotation filtration and direct filtration are possible when the mixing parameters are suitably optimized (Filipenska et al. 2019).

Figure 2

Effect of velocity gradient on the properties of flocs made from different coagulants and impurities, as published by the mentioned authors.

Figure 2

Effect of velocity gradient on the properties of flocs made from different coagulants and impurities, as published by the mentioned authors.

Close modal

In practice, the contaminant concentration also determines the separation method. In the case of raw water with low contamination requiring a small amount of coagulant, a small amount of suspension is produced. Thus, single-stage separation by sand filtration is possible. In the case of highly contaminated raw water requiring large coagulant doses, a large amount of suspension will be produced. Thus, double-stage separation by sedimentation and sand filtration is necessary. In conclusion, when selecting a separation method, it is necessary to take into account the type and concentration of impurities present in the raw water and the type of coagulants used.

Mixing conditions are a key parameter controlling floc properties. Mixing optimization can significantly improve floc separation efficiency, which leads to a decrease in DWTP operating costs. Mixing intensity is very often characterized by impeller rotation frequency (rpm), although rpm units provide no information about hydrodynamics. The usage of the global velocity gradient is more convenient and correct. The global velocity gradient can be calculated from the torque or power number of an impeller, or in the case of hydraulic mixing, from the pressure drop. With increasing velocity gradient, flocs are smaller and more compact. For single-stage separation by filtration, rapid flocculation mixing must be applied after homogenization to produce small and dense flocs that penetrate deep into the filter bed. Double-stage separation by sedimentation and filtration requires the application of homogenization, rapid flocculation mixing and slow flocculation mixing to produce large flocs with high sedimentation velocities. Double-stage separation by flotation (DAF) and filtration require small flocs with a low density. Such flocs are obtained either by rapid flocculation mixing after homogenization, shortening the flocculation time or breaking large flocs in the DAF contact zone. Nevertheless, floc properties are also influenced by the type and concentration of impurities and coagulants, which must also be taken into account when mixing is optimized for a given separation method. The ranges of velocity gradients G recommended for different mixing types and needed for the formation of flocs suitable for different separation steps are summarized in Table 1.

Table 1

Types and intensity of mixing for the formation of flocs suitable for different separation steps

Type of separationRecommended floc size for the mentioned type of separationMixing type, its purpose and recommended velocity gradients (G)*
Homogenization (applied after coagulant dosing) – purpose: achieving a dispersion of the coagulant in waterRapid flocculation (applied after homogenization) – purpose: producing small and dense flocsSlow flocculation (applied after rapid flocculation) – purpose: producing large flocs (dense for sedimentation, floating for flotation)
Filtration 50–60 μma,b (max 100 μm) G > 300 s−1c G = 100–400 s−1d,e – 
Sedimentation + filtration >100 μmf,g G > 300 s−1c G = 100–400 s−1d,e G = 20–100 s−1e 
Flotation + filtration 25–50 μmf G > 300 s−1c G = 50–100 s−1f (sometimes 150 s−1G = 50–100 s−1f (sometimes 30 s−1
Type of separationRecommended floc size for the mentioned type of separationMixing type, its purpose and recommended velocity gradients (G)*
Homogenization (applied after coagulant dosing) – purpose: achieving a dispersion of the coagulant in waterRapid flocculation (applied after homogenization) – purpose: producing small and dense flocsSlow flocculation (applied after rapid flocculation) – purpose: producing large flocs (dense for sedimentation, floating for flotation)
Filtration 50–60 μma,b (max 100 μm) G > 300 s−1c G = 100–400 s−1d,e – 
Sedimentation + filtration >100 μmf,g G > 300 s−1c G = 100–400 s−1d,e G = 20–100 s−1e 
Flotation + filtration 25–50 μmf G > 300 s−1c G = 50–100 s−1f (sometimes 150 s−1G = 50–100 s−1f (sometimes 30 s−1

*There are ranges of G due to the fact that floc properties (size, density, etc.) are influenced also by the type and concentration of coagulant and impurities (Filipenska et al. 2019). For example, flocs prepared with Fe coagulant are larger than Al coagulant, or AOM flocs are larger than kaolinite flocs; thus, lower G should be applied in the case of Fe coagulant or AOM flocs, respectively, to obtain flocs of similar size.

This work was supported by the Czech Science Foundation (project number 18-05007S) and the Czech Academy of Sciences (institutional support RVO 67985874).

All relevant data are included in the paper or its Supplementary Information.

Bache
D. H.
Gregory
R.
2010
Flocs and separation processes in drinking water treatment: a review
.
Journal of Water Supply: Research and Technology – AQUA
59
(
1
),
16
30
.
https://doi.org/10.2166/aqua.2010.028
.
Bache
D. H.
Rasool
E. R.
2001
Characteristics of alumino-humic flocs in relation to DAF performance
.
Water Science and Technology
43
(
8
),
203
208
.
https://doi.org/10.2166/wst.2001.0495
.
Bratby
J.
2006
Coagulation and Flocculation in Water and Wastewater Treatment
, 2nd edn.
IWA Publishing
,
London
.
Bubakova
P.
Pivokonsky
M.
2012
The influence of velocity gradient on properties and filterability of suspension formed during water treatment
.
Separation and Purification Technology
92
,
161
167
.
https://doi.org/10.1016/j.seppur.2011.09.031
.
Bubakova
P.
Pivokonsky
M.
Pivokonsky
R.
Filip
P.
2013
The effect of shear rate on aggregate size distribution and structure at steady state: a comparison between a Taylor-Couette reactor and a mixing tank
.
Journal of Water Supply: Research and Technology – AQUA
62
(
5
),
288
295
.
https://doi.org/10.2166/aqua.2013.005
.
Camp
T. R.
Stein
P. C.
1943
Velocity gradient and internal work in fluid motion
.
Journal of Boston Society of Civil Engineering
30
(
4
),
219
237
.
Coufort
C.
Bouyer
D.
Liné
A.
2005
Flocculation related to local hydrodynamics in a Taylor-Couette reactor and in a jar
.
Chemical Engineering Science
60
(
8–9
),
2179
2192
.
https://doi.org/10.1016/j.ces.2004.10.038
.
Coufort
C.
Bouyer
D.
Liné
A.
Haut
B.
2007
Modeling of flocculation using a population balance equation
.
Chemical Engineering and Processing
46
(
12
),
1264
1273
.
https://doi.org/10.1016/j.cep.2006.10.012
.
Edzwald
J. K.
1995
Principles and applications of dissolved air flotation
.
Water Science and Technology
31
(
3–4
),
1
23
.
https://doi.org/10.1016/0273-1223(95)00200-7
.
Edzwald
J. K.
2010
Dissolved air flotation and me
.
Water Research
44
,
2077
2106
.
https://doi.org/10.1016/j.watres.2009.12.040
.
Filipenska
M.
Vasatova
P.
Pivokonska
L.
Cermakova
L.
Gonzales-Torres
A.
Henderson
R. K.
Naceradska
J.
Pivokonsky
M.
2019
Influence of COM-peptides/proteins on the properties of flocs formed at different shear rates
.
Journal of Environmental Sciences
80
,
116
127
.
https://doi.org/10.1016/j.jes.2018.11.025
.
Gonzalez-Torres
A.
Rich
A. M.
Marjo
C. E.
Henderson
R. K.
2017
Evaluation of biochemical algal floc properties using reflectance Fourier-transform infrared imaging
.
Algal Research
27
,
345
355
.
https://doi.org/10.1016/j.algal.2017.09.017
.
Khiadani
M.
Kolivand
R.
Ahooghalandari
M.
Mohajer
M.
2013
Removal of turbidity from water by dissolved air flotation and conventional sedimentation systems using poly aluminium chloride as coagulant
.
Desalination and Water Treatment
52
(
4–6
),
985
989
.
https://doi.org/10.1080/19443994.2013.826339
.
Morrow
J. J.
Rausch
E. G.
1974
Colloid destabilization with cationic polyelectrolytes as affected by velocity gradients
.
Journal of American Water Works Association
66
(
11
),
646
653
.
Mutl
S.
Polasek
P.
Pivokonsky
M.
Kloucek
O.
2006
The influence of G and T on the course of aggregation at the treatment of medium polluted surface water
.
Water Science and Technology: Water Supply
6
,
39
48
.
https://doi.org/10.2166/ws.2006.006
.
Naceradska
J.
Pivokonska
L.
Pivokonsky
M.
2019
On the importance of pH value in coagulation
.
Journal of Water Supply: Research and Technology – AQUA
68
(
3
),
222
230
.
https://doi.org/10.2166/aqua.2019.155
.
Ngo
H. H.
Vigneswaran
P.
Dhamappa
H. B.
1995
Optimization of direct filtration: experiments and mathematical models
.
Environmental Technology
16
(
1
),
55
63
.
https://doi.org/10.1080/09593331608616245
.
Pivokonsky
M.
Bubakova
P.
Pivokonska
L.
Hnatukova
P.
2011
The effect of global velocity gradient on the character and filterability of aggregates formed during the coagulation/flocculation process
.
Environmental Technology
32
(
12
),
1355
1366
.
https://doi.org/10.1080/09593330.2010.536786
.
Pivokonsky
M.
Safarikova
J.
Bubakova
P.
Pivokonska
L.
2012
Coagulation of peptides and proteins produced by Microcystis aeruginosa: interaction mechanisms and the effect of Fe-peptide/protein complexes formation
.
Water Research
46
,
5583
5590
.
https://doi.org/10.1016/j.watres.2012.07.040
.
Polasek
P.
2007
Differentiation between different kinds of mixing in water purification – back to basics
.
Water SA
33
(
2
),
249
251
.
https://doi.org/10.4314/wsa.v33i2.49082
.
Polasek
P.
Mutl
S.
2002
Cationic polymers in water treatment: part I: treatability of water with cationic polymers
.
Water SA
28
(
1
),
69
82
.
https://doi.org/10.4314/wsa.v28i1.4870
.
Polasek
P.
Mutl
S.
2005
Optimisation of reaction conditions of particle aggregation in water purification – back to basics
.
Water SA
31
(
1
),
61
72
.
https://doi.org/10.4314/wsa.v31i1.5122
.
Valade
M. T.
Edzwald
J. K.
Tobiason
J. E.
Dahlquist
J.
Hedberg
T.
Amato
T.
1996
Particle removal by flotation and filtration: pretreatment effects
.
Journal of American Water Works Association
88
(
12
),
35
46
.
https://doi.org/10.1002/j.1551-8833.1996.tb06660.x
.
Vlaski
A.
van Breemen
A. N.
Alaerts
G. J.
1997
The role of particle size and density in dissolved air flotation and sedimentation
.
Water Science and Technology
36
(
4
),
177
189
.
https://doi.org/10.1016/S0273-1223(97)00438-1
.
Wang
D.
Wu
R.
Jiang
Y.
Chow
W. K. C.
2011
Characterization of floc structure and strength: role of changing shear rates under various coagulation mechanisms
.
Colloids and Surfaces A: Physicochemical and Engineering Aspects
379
(
1–3
),
36
42
.
https://doi.org/10.1016/j.colsurfa.2010.11.048
.
Xu
W.
Gao
B.
Yue
Q.
Wang
Y.
2010
Effect of shear force and solution pH on flocs breakage and re-growth formed by nano-Al13 polymer
.
Water Research
44
(
6
),
1893
1899
.
https://doi.org/10.1016/j.watres.2009.11.029
.