`
`77
`
`Development of a new method of measuring bubble size
`
`M.Y. Han*, Y.H. Park* and T.J. Yu**
`* School of Civil, Urban & Geosystem Engineering, Seoul National University, San 56-1, Shilim-dong,
`Kwanak-gu, Seoul, Korea. (E-mail: myhan@gong.snu.ac.kr; yhpark@waterfirst.snu.ac.kr)
`** Department of Civil and Environmental Engineering, Kwangju University, 592-1, Jinwol-dong, Nam-gu,
`Kwangju, Korea. (E-mail: tjyu@hosim.kwangju.ac.kr)
`
`Abstract The use of bubbles in water and wastewater treatment, including dissolved air flotation (DAF) and
`electro-flotation (EF), is attracting much interest recently. These flotation processes are governed by
`characteristics of the bubbles as well as the particles, and therefore it is necessary to investigate the size
`distribution of the bubbles that are generated. In this research, a new method has been developed to
`measure the bubble size, using commercially available batch-type and on-line particle counters. The results
`are compared with the traditional image analysis method. Although there are some discrepancies, the results
`show that an on-line particle counter can produce reasonably accurate size distributions conveniently and
`efficiently. The bubble size measurement technique developed in this study will assist understanding and
`improvement of the DAF and EF processes, from both theoretical and practical points of view.
`Keywords Bubble size; dissolved air flotation (DAF); electro-flotation (EF); image analysis;
`particle counter
`
`Introduction
`The use of bubbles in water and wastewater treatment, including dissolved air flotation
`(DAF) and electro-flotation (EF), has attracted much interest recently. Although the funda-
`mental characteristics of the micro-bubble/particle/solution system should affect the
`removal efficiency of the process, the effect of each governing physical–chemical parame-
`ter has not been investigated fully, either experimentally or theoretically. According to
`recent modeling of the DAF process, the most important parameters that affect the removal
`efficiency are the size and zeta potential of both bubbles and particles (Han et al., 2001;
`Han, 2002).
`In DAF, bubbles are generated when air-saturated water is released into atmospheric
`pressure. The size of bubbles is mostly affected by pressure difference across the injection
`system and type of nozzle (AWWA, 1999). The size range is generally reported to be
`10–100 µm, with the average being approximately 40 µm, under a pressure of 4–6 atmos-
`pheres (Edzwald, 1995). In EF, hydrogen and oxygen bubbles are generated when current
`is applied to the solution through metal electrodes. The average size range is reported to be
`around 20–40 µm, which is a smaller range than that of DAF (Burns et al., 1997).
`Several methods have been developed to measure the size of bubbles. The most straight-
`forward method is image analysis. Because this method requires a complicated experi-
`mental setup and is time-consuming, it is not easy to produce enough data to generate size
`distributions under different conditions. Another method is to measure the rising velocity
`of the bubbles and to calculate the sizes by Stokes’ Law. However, because the sizes of
`bubbles are not uniform, and because the rising velocity of many bubbles is different from
`that of a single bubble, no general equations are available to predict the size distribution of
`bubbles from the rising velocities.
`In this study, a new method to measure the size of bubbles, using particle counters, was
`developed. The bubble counting results obtained from both image analysis and particle
`
`Downloaded from http://iwaponline.com/ws/article-pdf/2/2/77/408222/77.pdf
`by jbeitz@fredlaw.com
`on 05 October 2020
`
`Tennant Company Exhibit 1037
`
`
`
`counters are compared by measuring the bubbles that are generated under the same
`conditions in DAF and EF.
`
`Methods
`Bubble generation conditions
`Dissolved-air-flotation. Air was pressurized and dissolved into water under 6 atm. To
`reduce interference from particles of solids, distilled and deionized water was used.
`Although particles smaller than 10 µm can be detected by the particle counter, only those
`larger than 10 µm were regarded as bubbles. To avoid over-counting the larger bubbles
`formed by bubble coalescence inside the tubing, the observations were made directly after
`the valve. Only a small volume of bubbles was generated to avoid the possibility that a high
`concentration might decrease the accuracy of the particle counters and increase bubble
`coalescence.
`
`Electro-flotation. To generate bubbles by EF, distilled and deionized water (as above) was
`mixed with the same volume of tap water. Aluminium electrodes 5 cm square and of thick-
`ness 0.5 mm were used, and a DC voltage of 12 V was applied. The method relies on gener-
`ation of hydrogen bubbles from the cathode. When aluminium electrodes are used, Al3+
`ions and oxygen bubbles are generated from the anode. The aluminium ions hydrolyze in
`water, producing floc particles that interfere with the measurements of bubble size. To
`avoid this, the anode was wrapped with GF/C filters (pore size: 0.45 µm) to prevent floc
`particles and oxygen bubbles being introduced into the sampling tube.
`
`Bubble size measurement
`Image analysis. The image analysis system, which is illustrated in Figure 1, includes a
`measuring cell, a microscope, a CCD camera, and a computer for image processing.
`Bubbles were generated inside the measuring cell to prevent the change of bubble size
`when bubbles are introduced into the cell through a tube in both DAF and EF. Images of
`bubbles were taken using the CCD camera. Their sizes were measured using a micrometer.
`The upper part of the cell was kept open because the pressure difference inside the cell can
`affect the bubble size. The microscope was focused at the point directly after the valve in
`DAF and directly above the cathode in EF.
`
`Batch-type particle counters. A batch-type particle counter (Multisizer II, Coulter) was
`used to measure the sizes of bubbles. Figure 2 shows the configuration of the measuring cell
`
`M.Y. Han et al.
`
`78
`
`Figure 1 Image analysis system
`
`Downloaded from http://iwaponline.com/ws/article-pdf/2/2/77/408222/77.pdf
`by jbeitz@fredlaw.com
`on 05 October 2020
`
`
`
`M.Y. Han et al.
`
`79
`
`Figure 2 Schematic of batch-type particle counter
`
`of the Multisizer II. Bubbles generated inside the cell were introduced through the opening
`in the aperture. An aperture size of 200 µm was used in the experiment. In this method, a
`constant electric current passes between the two electrodes through the electrolyte. When
`particles (bubbles in this research) pass through the sensing zone, the electrolyte volume
`decreases, which increases the resistance to the electric current. The amount of resistance is
`exactly proportional to the volume of particles, and this volume is converted to the size of
`equivalent spherical particles or bubbles. Although the result might be considered accurate
`because of the narrowly divided channel (256 channels), the application is limited to
`laboratory experiments because sampling and measuring is quite difficult. Furthermore,
`bubbles generated by EF cannot be measured, because of the electrical disturbance to the
`measuring system.
`
`On-line particle counters. An on-line particle counter (Chemtrac Model PC2400 D, USA)
`was used to measure the sizes of bubbles. In this method, a laser light shines through the
`sensor onto the detector. When the sample passes through the sensor, the light is scattered
`and obscured by the particles. This scattering and obscuring of the light causes a decrease in
`the intensity of the light reaching the detector that is proportional to the particle size.
`According to the decrease, a voltage pulse is generated. Here, the number of pulses repre-
`sents the number of particles, and the height of the pulse the size of the particles. This
`instrument can measure over the size range of 2–400 µm in seven user-definable size
`ranges. In this research, two identical particle counters were used to record data for 14
`channels. To minimize possible bubble coalescence inside the tube, a straight tube, which
`was kept as short as possible, was used, and the sampling flow rate was kept at 100 ml/min
`which is recommended by the manufacturer.
`
`Downloaded from http://iwaponline.com/ws/article-pdf/2/2/77/408222/77.pdf
`by jbeitz@fredlaw.com
`on 05 October 2020
`
`
`
`M.Y. Han et al.
`
`Figure 3 Schematic of on-line particle counter and details of the sensor
`
`Results and discussions
`Comparison of each method
`Three different methods of measuring bubble sizes (image analysis, a batch-type particle
`counter, and an on-line particle counter) were tested, and the characteristics of each method
`are compared in Table 1. The batch-type particle counter is not suitable for bubbles gener-
`ated by EF. Continuous size measurement is not possible using the image analysis method
`and the batch-type particle counter.
`However, the results are much more accurate than those from the on-line particle count-
`er. The most useful feature of the on-line particle counters is the very rapid rate at which
`data can be acquired. The time needed to measure 2,000 bubbles by each method was 3,000,
`30, and 10 minutes, respectively.
`
`Bubble size in DAF
`The size and size distribution of bubbles generated from DAF were measured by image
`analysis, batch and online particle counters, respectively. It is important to keep the pres-
`sure constant (6 atm.) throughout the DAF experiments, because the size of the bubbles is
`dependent on the pressure. The results from each method are comparable, as illustrated in
`Figure 4 and listed in Table 2.
`The average bubble size and modal bubble size recorded by each method was similar,
`but the size range was not. The size range of bubbles from particle counting methods is
`wider than from image analysis, which means that a small number of larger bubbles was
`detected in the particle counting methods. One possible reason for this might be the coales-
`cence of bubbles during transport to the sensor. Another reason, applicable to the on-line
`particle counter, is possible overlapping of bubbles inside the sensor, which would result in
`counting fewer but larger bubbles. This is an inherent shortcoming of the instrument.
`
`Table 1 Comparison of characteristics of bubble size measurement methods
`
`Application
`DAF
`EF
`
`On-line
`measurement
`
`Size
`accuracy
`
`Measuring time
`(min)*
`
`O
`Image analysis
`Batch-type particle counter O
`On-line particle counter
`O
`
`O
`X
`O
`
`X
`X
`O
`
`Excellent
`Excellent
`Good
`
`3000
`30
`10
`
`80
`
`* For measurement of approximately 2,000 bubbles
`
`Downloaded from http://iwaponline.com/ws/article-pdf/2/2/77/408222/77.pdf
`by jbeitz@fredlaw.com
`on 05 October 2020
`
`
`
`M.Y. Han et al.
`
`Figure 4 Comparison of bubble size distribution of DAF
`
`Table 2 Size characteristics of bubbles measured by each method
`
`Size range
`(µm)
`
`Average size
`(µm)
`
`Modal size
`(µm)
`
`Image analysis
`Batch-type particle counter
`On-line particle counter
`
`14–56
`13–96
`15–85
`
`* Fraction of bubbles smaller than 55 µm
`
`32
`31
`28
`
`25
`25
`25
`
`d55*
`
`99.0%
`99.0%
`97.6%
`
`Nevertheless, the fraction of bubbles smaller than 55 µm (d55) in all measurements is more
`than 97%, so that the difference in size range is of little importance.
`From this comparison, the accuracy of measuring bubbles generated from DAF by a par-
`ticle counter is considered good enough to be used for process monitoring. The fast
`response of the on-line particle counter is an especially good feature.
`
`Bubble size in EF
`Both the image analysis method and the on-line particle counting method were used to
`measure the size of hydrogen bubbles generated from EF. The batch-type particle counting
`method cannot be used for EF, as described previously. The size distribution and cumula-
`tive distribution of bubbles are compared in Figure 5. The average size, modal size, and size
`range of bubbles are compared in Table 3.
`As the result of DAF experiments, a wider bubble size range was observed for the on-
`line particle counter compared with image analysis. The reason for this is expected to be the
`same as with DAF. The difference between the two methods for average bubble size and d35
`is slightly larger than observed for DAF. This is because of disturbance by the floc particles
`produced during electrolysis. Another error in the application of the new method in EF is
`
`Figure 5 Comparison of bubble size distribution of EF
`
`81
`
`Downloaded from http://iwaponline.com/ws/article-pdf/2/2/77/408222/77.pdf
`by jbeitz@fredlaw.com
`on 05 October 2020
`
`
`
`Table 3 Size characteristics of bubbles measured by each method
`
`Size range
`(µm)
`
`Average size
`(µm)
`
`Modal size
`(µm)
`
`Image analysis
`On-line particle counter
`
`5–40
`15–65
`
`18
`22
`
`15
`15
`
`* Fraction of bubbles smaller than 35 µm
`
`d35 *
`
`99.6%
`94.2%
`
`that bubbles smaller than 10 µm cannot be counted even though those bubbles are actually
`generated in EF. However, because the fraction of those bubbles is very small, 4% in this
`study, the new method is considered to be quite acceptable also in EF in spite of these
`problems.
`
`Bubble size comparison between DAF and EF
`Since it is found from above experiments that the on-line particle counting method can pro-
`duce data of reasonable accuracy, the sizes of bubbles generated from DAF and EF are
`compared as in Figure 6 and Table 4.
`The size of bubbles produced in DAF is in the range of 15–85 µm and the average size is
`around 28 µm, whereas the size produced by EF is in the range of 15–65 µm and the average
`is 22 µm. This result supports the generally known fact that DAF generates larger bubbles
`than EF does.
`The average size of the bubbles produced by DAF in this work is smaller than those sizes
`reported in the literature. The reason is that in this study the bubbles are measured immedi-
`ately after release from a 6 atm pressure vessel. Literature values are measured from a con-
`tact zone in an operating DAF plant in which the pressures are reduced by passage through
`piping, valve, and orifice. Lower pressures tend to increase the size of the bubbles. In addi-
`tion, because the chance of bubble coalescence increases with the time between generation
`and measurement, the size of bubbles will increase.
`In practice, it has been a generally accepted concept that smaller bubbles are preferred in
`order to achieve a larger bubble surface area and so to maximize mass transfer. However, if
`the collision and attachment mechanisms are considered in DAF and EF processes, the
`
`M.Y. Han et al.
`
`Figure 6 Comparison of DAF and EF bubble size distribution
`
`Table 4 Size characteristics of bubbles generated by DAF and EF
`
`Size range
`(µm)
`
`Average size
`(µm)
`
`Modal size
`(µm)
`
`DAF
`Electro-flotation
`
`15–85
`15–65
`
`28
`22
`
`25
`15
`
`82
`
`* Fraction of bubbles smaller than 35 µm
`
`d35 *
`
`87.6%
`94.2%
`
`Downloaded from http://iwaponline.com/ws/article-pdf/2/2/77/408222/77.pdf
`by jbeitz@fredlaw.com
`on 05 October 2020
`
`
`
`M.Y. Han et al.
`
`83
`
`optimum size of bubbles should be dependent on the size of particles to be removed. The
`effect of bubble size and particle size on the collision efficiency in DAF has been modeled
`by Han (2001).
`
`Conclusion
`In this research, a new method to measure bubble size distribution was developed by using
`commercially available batch-type and on-line particle counters. The results compare well
`with the traditional but laborious image analysis method. The batch-type counter is not
`suitable for measurement of the size of bubbles generated from EF because of disturbance
`by the EF electric current. Although there are some discrepancies, the on-line particle
`counter can produce reasonably accurate results in a very short time.
`The bubble counting method described in this paper will be helpful for research in DAF
`and EF processes, either theoretically or practically. The mechanism of bubble and particle
`collision and its effect on the removal efficiency can be described. An optimum operating
`condition of the bubble generation system and/or pretreatment system can be diagnosed by
`measuring the size of bubble at several places in the reactor and processes.
`
`Acknowledgement
`The authors wish to acknowledge J. Lee (Seoul National University) for the contribution
`and assistance during the experiment. This work is funded by MOCT under the project,
`2001-JAYU A09.
`
`References
`Water Quality and Treatment 5th Ed. (1999). American Water Works Association, McGraw Hill, USA.
`Burns, S.E., Yiacoumi, S. and Tsouris, C. (1997). Microbubble Generation for Environmental and
`Industrial Separations, Separation and Purification Technology 11, 221–232.
`Edzwald, J. (1995). Principles and Applications of Dissolved Air Flotation. Wat. Sci. & Tech., 31(3–4),
`1–23.
`Han, M.Y. (2002). Modeling of DAF: the effect of particle and bubble characteristics, Journal of Water
`Supply: Research and Technology – AQUA 51 27–34.
`Han, M.Y., Kim, W.T. and Dockko, S. (2001). Collision Efficiency Factor of Bubble and Particle (αbp) in
`DAF: Theory and Experimental Verification, Wat. Sci. & Tech., 43(8), 139–144.
`
`Downloaded from http://iwaponline.com/ws/article-pdf/2/2/77/408222/77.pdf
`by jbeitz@fredlaw.com
`on 05 October 2020
`
`
`
`Downloaded from http://iwaponline.com/ws/article-pdf/2/2/77/408222/77.pdf
`by jbeitz@fredlaw.com
`on 05 October 2020
`
`