For a more efficient design of dissolved air flotation (DAF) units, it is important to deepen our knowledge of the role that each constituent part plays with regard to overall performance. The velocity control of the flocculated water at the entrance (Ve) of the contact zone of DAF units is important because the degree of agitation in this region may affect the size of the flocs and the efficiency of the flotation process. This article shows the results of an experimental study conducted to verify the influence of Ve (and its associated average velocity gradient, Ge) on the performance of a DAF pilot plant (DAFPP) fed with synthetic water. Two different coagulation conditions of the same raw synthetic water were tested, and 11 assays with different Ve values for each condition were performed. The floc size distribution curves regarding two limit situations, the lowest and the highest Ve values, were obtained by using an appropriate image analysis method. The best results were obtained for Ve values ranging from 125 to180 m·h−1, and for Ve values outside this range, the decrease in DAFPP performance was not as substantial as expected.

INTRODUCTION

Currently, dissolved air flotation (DAF) can be considered to be a recognized and consolidated technique for efficient drinking water clarification, mainly for algae-laden or colored raw water. The success of the DAF process fundamentally depends on the following main sequential steps: (i) the raw water coagulation step, which is performed by adding the correct dosages of chemicals (e.g., coagulants, alkali and others if required); (ii) the flocculation step, where flocs of an appropriate size range are formed, aiming for separation by flotation (requiring floc size ranges lower than that required for the sedimentation technique); (iii) the micro-bubble generation step, which is performed by using high pressure saturation chambers (fed with pressurized recirculated water and compressed air) that are associated with special nozzles that are installed at the beginning of the contact zone (CZ), which are responsible for the sudden pressure decrease of the saturated water stream, thus permitting the generation of air micro-bubble clouds (bubble size usually ranging from 10 to 100 μm); and (iv) mixing the saturated water stream into the flocculated water to promote the collision and adherence of micro-bubbles to the floc structures, thus forming clusters, i.e., flocs containing several micro-bubbles attached to them (Leppinen & Dalziel 2004).

This last step takes place inside the CZ of DAF units and is initiated immediately after the water enters the CZ, passing through the opening located between the flocculation unit outlet and the entrance of the CZ; the mean water velocity is presented here as Ve. In rectangular DAF units, this passage is usually designed as one submersed rectangular opening at the bottom of the CZ or as several parallel openings (with circular or rectangular shapes), also at the bottom of the CZ. Either way, these openings are sized taking into account the Ve value, which must be correctly chosen to permit appropriate initial mixing between the flocculated water (with Ve) and the recirculated water (discharged by the nozzles near the entrance of the CZ and containing the air micro-bubbles) without causing excessive shear strength in the water flow, which can result in detrimental floc breakage. For a determined water temperature, the shear strength can be better characterized by the average velocity gradient (Ge) at the entrance of the CZ, which is directly dependent on the Ve value for a given entrance geometry. After the initial mixing of the flocculated water with the recirculated water, the flocs and the micro-bubbles continue to contact throughout the CZ, thus permitting efficient formation of clusters (floc plus micro-bubble agglomerates). When leaving the CZ, these clusters, presenting suitable rising velocities, enter the separation zone (SZ) of the DAF unit, where they float, whereas the water (free of clusters) is collected at the bottom of the SZ.

In the last three decades, knowledge of the DAF technique applied to water treatment has presented important advances in its basic aspects. Several theoretical models for the DAF process have been proposed, and the use of new analytical tools, such as computer fluid dynamics, ultrasonic equipment (Micro ADV) and special equipment for microscopic image capturing, has allowed for important studies to be performed as well as the improvement of the DAF technique (Fukushi et al. 1995; Lundh et al. 2002; Reali & Patrizzi 2007; Amato & Wicks 2009; Moruzzi & Reali 2010; Edzwald & Haarhoff 2011; Lakghomi et al. 2012, 2015).

However, research is still required to deepen our knowledge of the hydrodynamic characteristics in the CZ and their effects on the efficiency of floc separation. One aspect that is still not well understood regarding the CZ of rectangular DAF units is the influence of the entrance velocity (Ve), or more properly speaking, of the mean velocity gradient (Ge) on the resulting floc size distribution (after the opening) and the floc separation efficiency.

This paper shows the results of a study conducted using a 4.6 m3·h−1 rectangular dissolved air flotation pilot plant (DAFPP) that was equipped with a special apparatus for microscopic image capturing and aimed to verify the influence of Ve (or Ge) on the floc size distribution and the overall color and turbidity removal efficiencies of the DAFPP.

METHODS

The experimental work was divided into three phases. In the first phase, laboratory-scale batch flotation equipment (named Flotatest) was used to perform preliminary assays to determine good coagulation conditions of the study water (i.e., suitable pH and alum dosage values) for flotation. The study (synthetic) water was prepared in the same way for all three phases of the work. In the second and third phases of the work, a continuous-flow rectangular DAFPP (see Figure 1) fed with the same water as phase 1 was used to perform the sets of assays described below.
Figure 1

Schemes and pictures of the DAFPP: (a) superior view; (b) longitudinal section A-A; (c) picture of the upper part of the device that controls the height (he) of the rectangular CZ entrance and consequently the Ve value; and (d) general view of the DAFPP.

Figure 1

Schemes and pictures of the DAFPP: (a) superior view; (b) longitudinal section A-A; (c) picture of the upper part of the device that controls the height (he) of the rectangular CZ entrance and consequently the Ve value; and (d) general view of the DAFPP.

Characteristics of the synthetic raw water: in all of the sets of experiments with the Flotatest equipment (first phase) and with the pilot plant (second and third phases), the synthetic raw water was the same and was prepared by adding 1.0 mg·L−1 of humic acid (Aldrich 1,675–2) and 8.5 mg·L−1 of kaolin (Fluka 60609) to deep well water. The synthetic raw water was stored in four 15 m3 tanks, and it always presented a turbidity of approximately 7.0 NTU, with a true color of approximately 34 CU and alkalinity of approximately 33.0 mg CaCO3·L−1.

First phase of the work: the first-phase assays were performed by using batch flotation equipment (Flotatest) that contained four 2-L flocculation/flotation columns. Sixteen series of flotation tests were performed (by varying the alum dosage from 10 to 47.5 mg·L−1) to optimize the coagulation conditions. Each series of tests was performed with a fixed coagulant (Al2(SO4)3.14H2O) dosage and by varying the pH (six pH values were tested in each series). The following parameters were fixed during all of the Flotatest trials: a water temperature of 22 ± 1 °C, average velocity gradient (G) of approximately 900 s−1 and detention time (Td) of 15 s during the rapid mixing step, and G and Td during flocculation of 90 s−1 and 15 min, respectively.

Second phase of the work: the second-phase assays were conducted by using the 4.6 m3·h−1 rectangular DAFPP shown in Figure 1, which was fed with the same synthetic water as the first phase. Based on the coagulation tests performed in the first phase, two different good coagulation conditions were chosen for the second-phase assays. In this way, it was possible to verify the influence of Ve on flotation performance regarding two different coagulation conditions (producing flocs with different characteristics) applied to the study water.

The DAFPP was designed in a way that made it possible to vary the height of the rectangular entry of the CZ at the bottom of the DAF unit, thus varying the water velocity at that location (Ve). For this, a special submerged gate was installed at the CZ entry that had the same width as the CZ. Hence, it was possible to test 11 different Ve values while maintaining the other main fixed parameters (see Figures 1(b) and 1(c)). The raw synthetic water was stored in four agitated 15 m3 tanks. The following operational parameters were maintained as constant during all of the assays: (i) incoming flow in the DAFPP: 4.6 ± 0.1 m3·h−1 (automatically controlled); (ii) recycle flow: 0.46 ± 0.5 m3·h−1; (iii) gauge pressure in the saturation chamber: 450 ± 10 kPa; (iv) water temperature: 26 ± 1 °C (controlled by a heater); (v) G in the flocculation compartments: 80 s−1; (vi) flocculation time: 18.6 min; (vii) down-flow rate in the SZ: 7.7 ± 0.1 m·h−1; (viii) cross-flow velocity: 45 ± 1 m·h−1; and (ix) up-flow rate in the CZ: 180 ± 4 m·h−1 (contact time of 34 s). The Ve value was varied from 62 to 12,432 m·h−1. Figure 1(c) shows a picture of the device for adjusting the opening of the CZ entry of the DAFPP, thus allowing the desired values of Ve to be obtained.

Third phase of the work: the two trials of the third phase were performed with the DAFPP operating with the limit Ve values (i.e., 62 m·h−1 and 12,432 m·h−1). The other operational parameters were identical to those fixed in the second phase, differing only when concerning the saturation system, which was turned off. In this way, no micro-bubbles were generated by the recirculation flow to avoid interference in the optical parts of the equipment used to acquire the floc images from the final part of the CZ (i.e., the images used to obtain the floc size distribution curves regarding each tested situation). The floc size characterization was conducted using the image acquisition/analysis apparatus described by Pioltine & Reali (2011). For this characterization, a special high luminosity charge coupled device (CCD) camera (QImaging QICAM FAST MONO 12-bit) coupled to a colposcope (Olympus OCS-3) was used. To process the images and obtain the floc size distribution curves, a laptop with software for image acquisition (Image Pro Plus 7.01) was used.

RESULTS AND DISCUSSION

Results of the first phase of the experimental work

Based on the coagulation/flocculation/flotation assays performed using the Flotatest equipment, two coagulation conditions of the studied water were chosen for performing the second phase of the study: (i) the alum dosage of 22.5 mg·L−1 (with the pH approximately 6.4) presented better turbidity and color removal efficiencies than the other dosages, thus resulting in a residual apparent color of 1.0 CU (98.5% removal) and in a residual turbidity of 0.35 NTU (95.5% removal); (ii) another alum dosage that presented good turbidity and color removal results during the assays with the Flotatest equipment was 42.5 mg·L−1 (pH also ranging 6.4); under this condition, a residual turbidity of approximately 0.5 NTU (93% removal) and a residual color of approximately 2 CU (96% removal) were obtained. Then, for the second phase of the study, these two different coagulation conditions were chosen to form the assays using the DAFPP fed with the same synthetic water and varying the Ve value at the entrance of the CZ (Condition 1: 42.5 mg·L−1, pH of approximately 6.4 and positive zeta potential (ZP); Condition 2: 22.5 mg·L−1, pH of approximately 6.4 and near-zero ZP).

Results of the second phase of the experimental work

Table 1 shows the turbidity, color residual values, and respective removal efficiencies for both coagulation conditions when the Ve value (and consequently Ge) was varied. For each coagulation condition, 11 assays were performed, each with a different Ve value. Table 1 also shows the initial raw synthetic water color and turbidity values together with the ZP (after coagulation) during the assays.

Table 1

Color and turbidity residuals and removal efficiencies regarding each tested Ve value for both coagulation conditions

Assay n1234567891011
*he (mm) 200 150 100 69 50 20 10 
Ve (m·h−162 83 125 180 249 622 1,243 1,554 2,486 4,144 12,432 
Ge (s−10.6 1.0 2.2 4.7 9.2 63.6 280.6 453.5 1,248.9 3,764.1 40,538.4 
Coagulation condition 1: alum dosage of 42.5 mg·L−1 (positive zeta potential) 
Raw water Turbidity = 6.93 NTU Apparent color = 33 CU 
ZP (mV) 7.5 ± 0.5 8.2 ± 0.4 7.2 ± 0.3 8.1 ± 0.5 7.9 ± 0.8 7.1 ± 0.4 8.5 ± 0.8 7.0 ± 0.8 7.9 ± 0.1 8.9 ± 0.7 7.7 ± 0.5 
Residuals/Removal efficiencies (%) regarding each assay 
Turbidity (NTU) 0.73/ 89.5 0.71/ 89.8 0.68/ 90.2 0.66/ 90.5 0.76/ 89.0 0.76/ 89.0 0.78/ 88.7 0.81/ 88.3 0.82/ 88.2 0.85/ 87.7 0.87/ 87.4 
Color (CU) 4/ 87.9 3/ 90.9 2/ 93.9 2/ 93.9 2/ 93.9 3/ 90.9 3/ 90.9 4/ 87.9 4/ 87.9 4/87.9 4/ 87.9 
Coagulation condition 2: alum dosage of 22.5 mg·L−1 (near-zero zeta potential) 
Raw water Turbidity = 7.10 NTU Apparent color = 36 CU 
ZP (mV) 1.3 ± 1.3 0.8 ± 0.4 1.9 ± 0.9 1.1 ± 1.2 2.2 ± 0.8 0.7 ± 0.4 1.5 ± 0.8 1.1 ± 1.3 0.0 ± 0.8 0.5 ± 0.7 −0.8 ± 0.5 
Residuals/Removal efficiencies (%) regarding each assay 
Turbidity (NTU) 1.03/ 85.7 0.98/ 86.4 0.95/ 86.8 0.93/ 87.1 0.97/ 86.5 1.02/ 85.9 1.01/ 86.0 1.03/ 85.7 1.07/ 85.2 1.09/ 84.9 1.09/ 84.9 
Color (CU) 4/ 88.9 4/ 88.9 3/ 91.7 3/ 91.7 3/ 91.7 4/ 88.9 4/ 88.9 5/ 86.1 5/ 86.1 5/ 86.1 5/ 83.3 
Assay n1234567891011
*he (mm) 200 150 100 69 50 20 10 
Ve (m·h−162 83 125 180 249 622 1,243 1,554 2,486 4,144 12,432 
Ge (s−10.6 1.0 2.2 4.7 9.2 63.6 280.6 453.5 1,248.9 3,764.1 40,538.4 
Coagulation condition 1: alum dosage of 42.5 mg·L−1 (positive zeta potential) 
Raw water Turbidity = 6.93 NTU Apparent color = 33 CU 
ZP (mV) 7.5 ± 0.5 8.2 ± 0.4 7.2 ± 0.3 8.1 ± 0.5 7.9 ± 0.8 7.1 ± 0.4 8.5 ± 0.8 7.0 ± 0.8 7.9 ± 0.1 8.9 ± 0.7 7.7 ± 0.5 
Residuals/Removal efficiencies (%) regarding each assay 
Turbidity (NTU) 0.73/ 89.5 0.71/ 89.8 0.68/ 90.2 0.66/ 90.5 0.76/ 89.0 0.76/ 89.0 0.78/ 88.7 0.81/ 88.3 0.82/ 88.2 0.85/ 87.7 0.87/ 87.4 
Color (CU) 4/ 87.9 3/ 90.9 2/ 93.9 2/ 93.9 2/ 93.9 3/ 90.9 3/ 90.9 4/ 87.9 4/ 87.9 4/87.9 4/ 87.9 
Coagulation condition 2: alum dosage of 22.5 mg·L−1 (near-zero zeta potential) 
Raw water Turbidity = 7.10 NTU Apparent color = 36 CU 
ZP (mV) 1.3 ± 1.3 0.8 ± 0.4 1.9 ± 0.9 1.1 ± 1.2 2.2 ± 0.8 0.7 ± 0.4 1.5 ± 0.8 1.1 ± 1.3 0.0 ± 0.8 0.5 ± 0.7 −0.8 ± 0.5 
Residuals/Removal efficiencies (%) regarding each assay 
Turbidity (NTU) 1.03/ 85.7 0.98/ 86.4 0.95/ 86.8 0.93/ 87.1 0.97/ 86.5 1.02/ 85.9 1.01/ 86.0 1.03/ 85.7 1.07/ 85.2 1.09/ 84.9 1.09/ 84.9 
Color (CU) 4/ 88.9 4/ 88.9 3/ 91.7 3/ 91.7 3/ 91.7 4/ 88.9 4/ 88.9 5/ 86.1 5/ 86.1 5/ 86.1 5/ 83.3 

*he: height of the rectangular CZ entrance, which controls the Ve value; Ge: estimated mean velocity gradient at the entrance of the CZ.

Based on the data presented in Table 1 and to facilitate the visualization of the results, the graphics shown in Figure 2 were prepared, where the color and turbidity removal efficiencies were plotted as a function of the applied Ve value during each assay.
Figure 2

Apparent color and turbidity removal efficiencies of the DAF regarding each Ve value for two coagulation conditions: (a) with a coagulant dosage of 3.9 mgAl+3·L−1 and ZP of approximately 8.0 mV, and (b) with 2.1 mgAl+3·L−1 associated with a ZP of approximately 0.7 mV.

Figure 2

Apparent color and turbidity removal efficiencies of the DAF regarding each Ve value for two coagulation conditions: (a) with a coagulant dosage of 3.9 mgAl+3·L−1 and ZP of approximately 8.0 mV, and (b) with 2.1 mgAl+3·L−1 associated with a ZP of approximately 0.7 mV.

Analyzing the results presented in Table 1 and Figure 2, it can be observed that for condition 1 (alum dosage of 42.5 mg·L−1 and positive ZP), the turbidity and color residuals were slightly lower than the residuals obtained for condition 2 (alum dosage of 22.5 mg·L−1 and near-zero ZP), when considering the same applied Ve. This was not observed during the first experimental phase (with the Flotatest equipment), thus suggesting that the flocs formed by positive-ZP particles in the continuous flow pilot tests presented a slightly better tendency to adhere to the air micro-bubbles that present a negative ZP in the water (Edzwald & Haarhoff 2011; Han et al. 2004).

In this way, the aggregates, that is, ‘flocs + micro-bubbles’, could be more easily found and more strongly adhere to each other, thus resulting in more efficient turbidity and color removal by flotation. Another important factor that must be considered is that the higher alum dosage promoted the formation of higher concentrations of aluminum hydroxide, likely enhancing the entrapment of micro-bubbles in the larger floc structures and resulting in better performance than in condition 2. This factor is probably more important when flotation occurs in continuous flow pilot units than in (batch) Flotatest columns due to the quite different floc–bubble kinetic interactions that occur in the CZ of these DAF units.

Focusing on the influence of Ve on the DAFPP performance, the results presented in Table 1 and Figure 2 show that for both conditions, raising the Ve value from 62 to 12,432 m·h−1 did not cause expressive decreases in the turbidity and color removal efficiencies. The variations regarding turbidity removal ranged from 87.4% to 90.5% for condition 1 and 84.9% to 87.1% for condition 2. The variations regarding color removal ranged from 87.9% to 93.9% for condition 1 and from 83.3% to 91.7% for condition 2. These results indicate that the floc breakage caused by the higher shear strength (due to high Ve values such as 12,432 m·h−1) in the water at the entrance of the CZ was not harmful to the flotation process. In other words, the floc size distribution in the water after passing the CZ entrance in this critical situation (Ve of 12,432 m·h−1) still remained suitable for the flotation process in both studied coagulation conditions. These aspects will be further discussed when discussing the results obtained from the image analysis study (third phase).

From analyzing the results of Table 1 and Figure 2, it can be observed that the range of Ve values between 125 and 180 m·h−1 produced the best color and turbidity removal efficiencies for both coagulation conditions. Ve values lower or higher than this range caused a decrease in color and turbidity removal efficiencies. Therefore, for this type of rectangular DAFPP, it is recommended that Ve values be adopted in the range of 125 to 180 m·h−1.

Results of the third phase of the experimental work

Figure 3 presents the results of the third phase of the study, which was performed to investigate the floc size distributions that resulted when applying the lowest (62 m·h−1, associated with a Ge of 0.6s−1) and highest (12,432 m·h−1, associated with a Ge of 40,538 s−1) Ve values at the CZ entrance for coagulation condition 2. These results are presented in terms of the frequency number and cumulative number of particles (flocs) regarding each size interval (Feret mean diameter) for samples extracted from the final part of the CZ of the pilot plant during two different assays (one applying Ve of 62 m·h−1 and the other applying Ve of 12,432 m·h−1).
Figure 3

Floc size distribution curves with a Ve of 62 m·h−1 (associated with a Ge of 0.6 s−1 and he of 200 mm) and Ve of 12,432 m·h−1 (associated with a Ge of 40,538 s−1 and he of 1 mm) for coagulation condition 2 (alum dosage of 22.5 mg·L−1).

Figure 3

Floc size distribution curves with a Ve of 62 m·h−1 (associated with a Ge of 0.6 s−1 and he of 200 mm) and Ve of 12,432 m·h−1 (associated with a Ge of 40,538 s−1 and he of 1 mm) for coagulation condition 2 (alum dosage of 22.5 mg·L−1).

It can be seen that increasing the Ve from 62 m·h−1 to 12,432 m·h−1 caused a displacement of the relative floc size distribution curve to the lower size ranges. For 62 m·h−1, the frequency number presented a peak in the floc size range of 400 to 500 μm, and for 12,432 m·h−1, the peak moved to the floc size range of 200 to 300 μm. It must be emphasized that these curves regard the samples extracted in the final part of the CZ (just before the water enters the SZ). Due to the limitations of the apparatus used to capture the images, it was not possible to obtain data from the entrance of the CZ. This suggests that some uncertainty still remains in assuming that the floc size distribution at the entrance of the CZ is quite different from the distribution curve obtained at the final part of the CZ. Perhaps, such results can corroborate the concept that DAF is capable of successfully operating with a small floc size distribution at the inlet of the CZ (Edzwald & Haarhoff 2011). Additionally, it can be supposed that a regrowth of the flocs occurs during the floc–bubble aggregate (cluster) formation process in the CZ, hence resulting in a satisfactory (but not the best) particle removal efficiency in the subsequent SZ.

Nonetheless, the results showed that even when applying a very high Ve value – supposing that a severe floc breakage took place due to an estimated local Ge of approximately 40,500 s−1 – the CZ detention time of 35 s associated with a mean up-flow water velocity of 180 m·h−1 promoted the formation of floc–bubble aggregates with sizes and rising velocities that were not harmful to the flotation process, thus resulting in only a slight decrease in the DAF unit performance compared to the results obtained when applying the optimum range of Ve values (125 to 180 m·h−1).

At this point, it is worth mentioning that besides the influence of the water velocity at the CZ entrance on the floc size (the subject of this study), another aspect that deserves attention in the DAF system design is the water jet velocity near the outlet of the nozzles which are responsible for the mixing of the recirculation flow with the flocculated water. Improper design of such nozzles can produce jets with excessive velocity that are capable of causing undesirable fragmentation of the flocs. Thus, additional research (preferably using image analysis) is needed to investigate this important issue.

CONCLUSIONS

For both sets of experiments of each phase of the study, the best results were obtained when Ve values ranging from 125 to 180 m·h−1 were applied. When the Ve value was increased from 180 to 12,432 m·h−1, the color and turbidity removal efficiencies presented a decrease lower than expected (i.e., from 90.5% to 87.4% turbidity removal and from 93.9% to 87.9% color removal for phase 2). This may indicate that for the coagulation conditions (phase 1 and phase 2) of the water used in this study, the resulting floc size distribution after passing the CZ entrance of the DAF unit did not reach values that were harmful to the flotation process, even in the presence of high shear stress values in the water when applying very high Ve values. The same behavior was observed when the Ve values were lower than 125 m·h−1, thus indicating that there was an optimum Ve value range that promotes adequate floc and micro-bubble mixing conditions (125 to 180 m·h−1, with Ge ranging from 2.2 to 4.7 s−1). Such results corroborate the concept that DAFs are capable of successfully operating with a small floc size distribution at the CZ. Furthermore, a probable regrowth of the floc–bubble aggregates (clusters) could have occurred during the passage of the suspension through the CZ, thus resulting in a not optimum, but still satisfactory, cluster size distribution as well as rising velocities for particle removal in the subsequent SZ.

ACKOWLEDGEMENTS

The financial support for this research was provided by Fundação de Amparo à Pesquisa do Estado de São Paulo and Conselho Nacional de Desenvolvimento Científico e Tecnológico, Brazil.

REFERENCES

REFERENCES
Edzwald
J. K.
Haarhoff
J.
2011
Dissolved Air Flotation for Water Clarification
, 1st edn,
McGraw-Hill and American Water Works Association
,
Denver, Colorado
.
Fukushi
K.
Tambo
N.
Matsui
Y.
1995
A kinetic model for dissolved air flotation in water and wastewater treatment
.
Water Science and Technology
31
(
3–4
),
37
47
.
Han
M. Y.
Ahn
H. J.
Shin
M. S.
Kim
S. R.
2004
The effect of divalent metal ions on the zeta potential of bubbles
.
Water Science and Technology
50
(
8
),
49
56
.
Leppinen
D. M.
Dalziel
S. B.
2004
Bubble size distribution in dissolved air flotation tanks
.
Journal of Water Supply
53
(
8
),
531
543
.
Pioltine
A.
Reali
M. A. P.
2011
Influence of flocculated particles size on the efficiency of dissolved air flotation
. In:
8th IWA Leading-Edge Conference on Water and Wastewater Technologies, 2011, Amsterdam. Proceedings of the 8th IWA Leading-Edge Conference on Water and Wastewater Technologies
.
IWA
,
London
.
Reali
M. A. P.
Patrizzi
L. J.
2007
The influence of the contact zone configuration on the efficiency of a DAFPP
. In:
The 5th International Conference on Flotation in Water and Wastewater Systems, Seoul, Korea. Proceedings of the 5th International Conference on Flotation in Water and Wastewater Systems
.
IWA
,
London
.