Dynamics of rising CO 2 bubble plumes in the QICS field experiment. Part 1 - The experiment

The dynamic characteristics of CO 2 bubbles in Scottish seawater are investigated through observational data obtained from the QICS project. Images of the leaked CO 2 bubble plume rising in the seawater were captured. This observation made it possible to discuss the dynamics of the CO 2 bubbles in plumes leaked in seawater from the sediments. Utilising ImageJ, an image processing program, the underwater recorded videos were analysed to measure the size and velocity of the CO 2 bubbles individually. It was found that most of the bubbles deform to non-spherical bubbles and the measured equivalent diameters of the CO 2 bubbles observed near the sea bed are to be between 2 and 12mm. The data processed from the videos showed that the velocities of 75% of the leaked CO 2 bubbles in the plume are in the interval 25–40cm/s with Reynolds numbers (Re) 500–3500, which are relatively higher than those of an individual bubble in quiescent water. The drag coefﬁcient C d is compared with numerous laboratory investigations, where agreement was found between the laboratory and the QICS experimental results withvariationsmainlyduetotheplumeinducedverticalvelocitycomponentoftheseawatercurrentand the interactions between the CO 2 bubbles (breakup and coalescence). The breakup of the CO 2 bubbles has been characterised and deﬁned by Eötvös number, Eo , and Re . © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
In addition to the carbon emission trading (TradeXchange, 2013), another potential solution to mitigate greenhouse gases release to the atmosphere and to meet the obligations specified by the Kyoto Protocol, is the disposal of carbon dioxide (CO 2 ) in the deep geoformations or the ocean (Freund and Ormerod, 1997), which was first proposed by Marchetti (1977).This process is known as carbon capture and storage (CCS) by which CO 2 is captured from power plants and industrial sources, before it is emitted into the atmosphere, and then injected into deep sub-seabed reservoirs/geological structures for permanent storage (Han et al., 2012).
The greatest concern on performing CCS in the engineering scale, is the risk of leakage from storage sites, it is therefore necessary to investigate the leakage possibility and the impacts of any potential CO 2 leakage on the environment, especially on the marine life for under seabed storage (Noble et al., 2012).Dispersion and dissolution of CO 2 bubbles in seawater are of special interest from the biological point of view, because of its importance in the changes of water quality.
In order to study the effects of a potential leak from a carbon under seabed storage on the UK marine environment, the Quantifying and Monitoring Potential Ecosystem Impacts of Geological Carbon Storage (QICS) project, was launched in 2010 (Blackford et al., 2014).QICS is a scientific research project involving a field experiment of injection of CO 2 into shallow marine sediments.One of the main objectives of the project, in addition to the development of monitoring and observation methods, is to generate experimental data to calibrate and develop models for predicting the change in pH or pCO 2 of the seawater in and above the sediments from leaked CO 2 .The changes in pH (or pCO 2 ) are vital data for the biogeochemical and ecological models in order to predict the impact of CO 2 leakages on the marine biological system in a variety of situations.Sufficient understanding of CO 2 bubble rising and dissolution characteristics in a plume are necessary and fundamental to the development of two-phase plume models for simulation of the changes in the seawater.In addition to the small-scale ocean turbulent model and bubble dissolution model, the twophase plume model requires two key parameters which are: (i) the initial CO 2 bubble size and leakage flux to be used as input and (ii) a correlation of the drag coefficient as function of the CO 2 bubble Reynolds number as it dominates the rise velocity and distance of the bubbles.By precisely measuring the size of the CO 2 bubbles and the rising velocities, these two parameters can be established.
The prediction of the dispersion of CO 2 bubbles has been attempted by numerical models (Chen et al., 2005(Chen et al., , 2009;;Dewar et al., 2013).Numerous experimental studies aimed at determining the drag coefficient correlation of gas bubble or liquid droplet as function of Reynolds numbers have been carried out, which had been summarised by Clift et al. (1978).Bozzano and Dente (2001) proposed a different correlation of drag coefficient as function of a group of dimensionless parameters of Reynolds number (Re), Morton numbers (Mo) and Eötvös number (Eo) of a single gas bubble or liquid droplet rising into a liquid phase (the details of the definitions of those dimensionless parameters can be found in Section 3, Eqs.(3)-( 5)).Bozzano and Dente (2001) described the comparison between their correlation and experimental data as satisfactory.Rodrigue (2001) developed a generalised correlation for the motion of single gas bubble in high viscosity Newtonian fluids.The correlation was compared with the previous experimental literature data for a broad range of Morton numbers.It was found that the correlation showed a good agreement for highly viscous fluids (Mo > 10 −8 ).Kelbaliyev (2011) presented an extensive review of twelve developed correlations of drag coefficient for a wide Re range up to 10 6 .All of the correlations were studied for single bubble rising in Newtonian fluids and developed for various ranges of a group of dimensionless parameters.The dynamics of the CO 2 bubble with the existence of other CO 2 bubbles in the near environment (CO 2 plume), however, differs notably from the one of an isolated bubble due to bubble interactions and the interactions between bubbles and water in the plume.In this work we report the results from observing the dynamics of CO 2 bubbles rising freely in a plume, created in the QICS field experiment (Blackford et al., 2014).It should be noted that, in addition to the free rising bubbles and plumes, the dynamics of bubble flows in pipes have been extensively investigated and the results obtained from those studies, can be the references of the studies of CO 2 plumes developed in the ocean.Details of the predictions of void fraction and heat transfer characteristics in two-phase systems can be found from recently published paper by Brooks et al. (2012).
The development of the drag coefficient correlation requires experimental data of the velocity and correspondent sizes and geometries of the rising gas bubbles in the fluids.Most of the previous studies have used cameras to gather these data; the motion of a single bubble was investigated experimentally using a camera to follow the rising bubbles (Zhang et al., 2008).Bando and Takemura (2006) used CCD (charge-coupled device) cameras to study the behaviour of carbon dioxide sphere rising in fluids, Takemura and Yabe (1999) used CCD cameras to study the rising speed and dissolution rate of CO 2 bubbles in slightly contaminated water, Zhang et al. (2012) used high speed cameras to study the multi-bubble behaviour in carbon capture system with ionic liquid, Brewer et al. (2002) used HDTV (High Definition Television) cameras and ROV (Remotely Operated Vehicles) in the field to measure the rise rate and dissolution rate of freely released CO 2 droplets in deep ocean, Luke and Cheng (2006) used high speed cameras to study the bubble formation with pool boiling, Zaruba et al. (2005) used CCD cameras to study the bubble motion in a rectangular bubble column, and many others (Wang and Dong, 2008;Luther et al., 2004;Bian et al., 2011;Wang and Dong, 2009;Hongyi and Feng, 2009) have used video cameras to study the velocity and size of a bubble rise in a liquid column.
In this paper, the dynamics of leaked CO 2 bubbles in a plume in the Scottish seawater are studied experimentally using video recordings.The experiment was carried in the ocean by creating a leakage scenario 12 m beneath the sediment and capturing video sequences of rising CO 2 bubbles reaching the seawater.

Experimental method
The QICS experiment is novel, in that for the first time CO 2 was injected into a natural system in a way that would closely mimic CO 2 leakage.The CO 2 release experiment was carried out under the Scottish sea at Ardmucknish Bay (56 29.55 N, 05 25.71 W) by drilling a borehole 12 m deep underlying the sandy mud sediments as illustrated in Fig. 1.The release of the CO 2 was controlled and monitored from a mobile laboratory at a nearby site after seeking the permission from the government and local authorities.The experiment was conducted by the Scottish Associations for Marine Science (SAMS) laboratory, near Oban in Western Scotland.After the migration of the CO 2 released in the sediment where part of it was dispersed and dissolved, the CO 2 reaches the seabed in gas bubble phase at 9-12 m water depth.
The motions of the CO 2 bubbles were captured using highdefinition (HD) video system to investigate the rise of CO 2 bubbles in seawater.A HDTV Canon EOS 5D Mark II video camera was used to track the rising CO 2 bubbles with a frame rate 30 fps (frames per second) producing digital HD images with resolution 1080 pixels vertically and 1920 pixels horizontally per frame.The elevation of the camera is approximately 20 cm above the sea floor.A ruler has been aligned with the CO 2 bubble plume as a referential dimension.The rising CO 2 bubbles were exposed to a green light during the capturing of the video as illustrated in Fig. 2.
The physical properties at ambient temperature of the seawater (10.8 • C) and CO 2 bubbles are reported in Table 1.

Rising CO 2 bubbles video processing
There is a direct analytical relationship between the drag coefficient (C d ) and the CO 2 bubble velocity (Dewar et al., 2013).To find this relationship, the experimental sizes and shapes of the CO 2 bubbles and the correspondent velocity are acquired.A method was developed that allowed the measurement of the CO 2 bubble sizes, velocities and their distributions.The bubble trajectories and the interactions among the CO 2 bubbles were investigated by analysing the videos recorded using the underwater digital video camera; the videos were processed using image processing software ImageJ (Schneider et al., 2012).The locations and the edge of the CO 2 bubble surfaces were determined as a result of manual image processing of each frame of the video separately.An edge contour was sketched around each bubble studied with the help of ImageJ to detect the cross-sectional area of the CO 2 bubble.A sample of the image of the bubble and the results are illustrated in Fig. 3.
For small CO 2 bubbles, their sizes can be more accurately measured since the bubble is almost spherical in shape, except where they are less than 2 mm where the resolution is too poor to conduct measurements.When the CO 2 bubbles are large, they take the shape of a perfect ellipsoidal or wobbling ellipsoidal shape; the evaluation of the size is characterised by an equivalent diameter (d e ).The CO 2 bubbles equivalent diameter is defined as the diameter of an equivalent sphere with the same bubble area of the equivalent CO 2 bubbles: where A is the measured area of the CO 2 bubble.It has been noted that the images taken from the video are two-dimensional vertical sections of the bubble, therefore, the measured area, A, is the vertical cross-sectional area of the bubble.The geometry of the large CO 2 bubbles studied is characterised by two dimensions: the major axis dimension (D mj ) and the minor axis diameter (D mi ) of the vertical cross section.In this study the unit of bubble size, including d e , D mi and D mj , are generally in metre if not specified.
The vertical movement of CO 2 bubbles were measured.Consider two successive frames taken from the record video, the coordinates of the CO 2 bubbles in the first frame and second frame are y 1 (m) and y 2 (m) respectively.The time interval between the two fames is t (=1/30 s).Then the velocity vertical velocity, V (m/s), of the CO 2 bubble is calculated by It has been noted that the velocity measured by Eq. ( 2) is the 'gross' velocity, rather than the relative velocity of the CO 2 bubble to that of seawater.Such a 'gross' velocity measured from the field observations can be defined as the velocity integrated with the seawater background velocity (current and tides) and plume velocities generated by the interactions between seawater and bubbles, and among the bubbles themselves.The effects of these factors on the bubble rising velocity and drag coefficient will be discussed in Section 4.2, in comparison with that of a single rising bubble in quiescent water.The dimensionless numbers used in this study to characterise the motion and shape of the CO 2 bubbles are: Reynolds number (Re), Eötvös number (Eo), and Morton number (Mo), which are defined as,

Re =
w Vd e w (3) where w (Pa s) is the dynamic viscosity of the seawater and w (kg/m 3 ) is the seawater density.where g (m/s 2 ) is gravitational acceleration and (N/m) is the interfacial tension between the water and the CO 2 gas. (5) where CO 2 (kg/m 3 ) is the density of the CO 2 .The data of physical properties of seawater and CO 2 in the field conditions can be found from Table 1.
The data obtained from the directly observations, such as the 'gross' velocity and the bubble size are referred to as the raw data in this study, in order to distinguish the data obtained with corrections.

Results and discussion
The CO 2 bubbles were studied up to 30 cm from sea floor at their initial states when they leak from the sediments into the seawater.The velocities (V) and sizes (d e ) of the bubbles have been recorded to generate the results to be discussed in this section.
It was found from the data obtained by the experiment, as shown in Fig. 4, that the size of the leaked CO 2 bubbles varies between 0.2 and 1.2 cm with a correspondent velocity varying between 20 cm/s and 45 cm/s.

Distributions of bubble size and velocity
The size of the CO 2 bubbles is the key parameter for the dynamics of free rising bubbles, including the dispersion and dissolution.The larger the bubble, the further it will travel in the seawater and the longer it will take to dissolve.For this reason, the distribution of the initial bubble size is vital data for the model prediction of the height travelled by the CO 2 bubbles in the water column before dissolving, as well as the changes in pH and pCO 2 of the water to be created, for which the details have been modelled and discussed in the second part of the paper (Dewar et al., 2014).It was found from QICS field experiment that more than 50% of the leaked CO 2 bubbles have a d e varying between 0.65 cm and 0.9 cm, with only  a low presence (<1.5%) of both the small bubbles (d e < 0.4 cm) and large bubbles (d e > 1.1 cm), respectively.The bubbles, after migrating from the sediment to the seawater, have the distribution of the different sizes as illustrated in Fig. 5 which has been used in the bubble plume modelling to set the initial bubble size (Dewar et al., 2014).
Another important parameter of the free rising CO 2 bubbles is the rising velocity.It was found, as illustrated in Fig. 6, the distribution of the leaked CO 2 bubble velocities, that most of the CO 2 bubbles (>75%) rise with a velocity between 25 cm/s and 40 cm/s, again we would like to highlight that these are the "gross" rising velocities of bubbles in the plume, which are detected directly by the video images, rather than the relative rising velocities of bubbles to the seawater.

The drag coefficient of leaked CO 2 bubbles
The drag coefficients of the leaked CO 2 bubbles are calculated by assuming the rising velocity measured being the terminal velocity.For bubbles rising freely in the seawater, the forces acting on the bubbles are mainly buoyancy and drag force.Taking into   (Dewar et al., 2013) and the data from QICS experiment directly by Eq. ( 7) (blue markers) and the data from QICS experiment by Eqs. ( 8)-( 10) (dark red markers) and with the corrected velocities of plume induced velocity (light red markers).consideration the equilibrium of the buoyancy and drag force, the equation of motion for the CO 2 bubble can be written as The density of the CO 2 gas can be negligible in comparison with the density of the seawater; therefore w − CO 2 ≈ w .The incorporation of this assumption into Eq.( 6) gives the drag coefficient C d of the CO 2 bubbles as, Using Eqs. ( 7) and ( 3), the drag coefficient C d and the Reynolds number Re were calculated for the studied CO 2 bubbles by using the observed velocity (V).The relationship between the C d and the Re obtained from the QICS experiment by the raw data is shown in Fig. 7 in blue markers, along with the drag coefficient results of methane gas bubbles obtained from the experimental study carried out in laboratory (Dewar et al., 2013) shown in black solid markers.It was found that only a few of the CO 2 bubbles studied within the QICS experiment match with the laboratory results for the Re range between 500 and 3500.Meanwhile, the data from QICS field experiment are broadly diverse and the majority of the QICS CO 2 bubbles had a variation of the C d between 0.4 and 2.3 for the same Re number which is relative smaller than those of an individual bubble.The difference can be explained due to three major factors.The first factor considered in the QICS experiment results in the variation of the C d is due to the CO 2 bubbles studied rising in a plume of bubbles compared to individual bubbles studied in the laboratory experiments.In the QICS experiment, the velocity of the CO 2 bubbles measured is the absolute ('gross') velocity of the bubbles carried by the water plume, for which, the effects vary depending on the location of the CO 2 bubbles in the plume, where the larger the velocity approaching to the centre of the plume.In general, the absolute velocity of bubbles in the plume is larger than the relative velocity of bubble to the water, which leads to a relatively smaller C d .Also providing the variation in the drag coefficient, as the same size CO 2 bubbles can have different velocities depending on its position in the plume (Domingos and Cardoso, 2013).The second factor that should be considered is the effects from the interactions among the CO 2 bubbles studied in the QICS experiment.It is detected from the observation data that the larger bubbles breakup as rising, meanwhile, coalescence of two or more CO 2 bubbles also occurs.The interactions change the velocity of the CO 2 bubbles due to the exchanges in momentum and the difference in sizes from the attraction or collision of the CO 2 bubbles (Liao and Lucas, 2009).
The third factor is the presence of the vertical component of the seawater tidal velocity.
In order to examine the effects from the seawater plume on the individual bubble dynamics, two-phase plume model simulations were carried out in part 2 of this study (Dewar et al., 2014) and it was found that the seawater in the plume is rising at velocities ranging from 3.5 cm/s to 5.0 cm/s in the centre of the plume, which results from the interactions between the momentum transferred by the rising bubbles and that of pealing down due to the increasing of density of CO 2 solution.This plume background seawater velocity data (V s = 3.5-5.0cm/s) is used to estimate the relative velocities of observed bubbles (V r ), V r = V − V s , where V is the bubble rising velocity observed from field experiment and shown in Fig. 4.
The effect of bubble deformation is then investigated by employing the correlations proposed by Bozzano and Dente (2001).In general, the drag coefficient C d of a spherical bubble can be expressed by the Reynolds numbers (Sommerfeld et al., 2010) alone, for which the correlations were proposed (Kelbaliyev, 2011).For large and deformable bubbles, however, the effect from flows generated by changes in shape of bubbles on the drag coefficient must be taken into account.In practice, additional dimensionless parameters, Morton number, Mo, and Eötvös number, Eo, are used for construction of the correlations (Bozzano and Dente, 2001), which is represented with the solid line in Fig. 7.The correlation is defined as: The correlation is tested by using the data of bubbles equivalent diameters d e and the corresponding velocities from the QICS experiment for calculation of the dimensionless parameters, Re, Eo, and Mo.
As shown in Fig. 7, in comparison with the raw data (blue markers), the effect of bubble deformation on the drag coefficient of plume bubbles is predicted (dark red markers) estimated from the correlation Eqs. ( 8)-( 10), using experimental data of d e and V.The deformation of bubbles generates a larger drag force, for which cannot directly detected by the raw data using Eq. ( 7).The light red markers in Fig. 7 are data representing the effects of induced seawater velocity on the estimation of the drag coefficient, calculated from correlation Eqs. ( 8)-( 10) and data of d e and the corrected rising velocity V r .It seems that the effects of bubble deformation are dominate in comparison with that of the induced seawater velocity, which is relatively smaller than the velocity of the bubbles observed in the field experiment.
Although the data are still divergent with a difference observed, in general, with that predicted by the correlation without using the QICS data (solid line).The divergent data are due to the results of the variable velocities at a given bubble equivalent diameter (refer to Fig. 5) as discussed in the first part of this section.It can be found from the results that in comparison with the results using Eq. ( 7), the corrected data are more approaching to those from laboratory experiments and the correlation (Bozzano and Dente, 2001) from the individual bubble rising in the quiescent water.The effects from the plume on the individual bubble rising dynamics are demonstrated, meanwhile, it is also demonstrated that the correlation proposed by Bozzano and Dente (2001) is applicable to the relative dilute bubbly plume simulations.There is another factor that should be taken into account when the comparisons with the laboratory experimental results (Dewar et al., 2013), which is the effects from dissolution.The available data (Dewar et al., 2013) are from methane bubbles, which are almost insoluble in the water, whereas the field experimental results are from CO 2 bubbles with relative large solubility.This effect should be investigated individually by well-designed lab.experiments or numerical simulations.

CO 2 bubbles shape characterisation
Considering the important effects of the shapes on the drag coefficient and further the dissolution rate, the data obtained from QICS experiment are analysed for investigation of the geometric characterisation of the different CO 2 bubbles in the open seawater.
Different CO 2 bubbles shapes were observed in the seawater during the QICS experiment.Fig. 8 shows photos captured of the six typical shapes of the bubbles, two of them characterising the breakup moment when the bubbles are about to divide (Fig. 8(e)) and breaking up in two (Fig. 8(f)).The CO 2 bubble shapes can be categorised into types of spherical (small size), cap and ellipsoidal shape, and ellipsoidal wobbling shape (large size).
The characteristics of bubbles deformation can be discussed by using Eötvös and Reynolds numbers.It can be found from the data retreated from recorded videos of CO 2 bubbles leaked in the QICS experiment in the Eo-Re diagram, shown in Fig. 9, that the different bubble shapes can be characterised.When Eo < 2, the CO 2 bubbles are small and have spherical shapes; 2 < Eo < 7, the CO 2 Fig. 9. Characterisation of the CO2 bubble shapes observed from QICS experiment.bubbles have ellipsoidal shapes; and Eo > 7 the CO 2 bubbles have ellipsoidal wobbling shapes.
Experiment data shows that the wobbling CO 2 bubbles were potentially going towards two possible shape situations: breakup into two smaller bubbles or becoming stable into a perfect ellipsoidal shape after losing part of its volume due to dissolution in seawater.
The diameter considered in the calculation of the Reynolds numbers is the equivalent diameter (d e ), however, as can be seen in Fig. 8, the CO 2 bubbles with the same d e have different widths expressed by the major axis dimension (D mj ).As have been discussed by Brooks et al. (2012), the major axis dimension (D mj ) is a parameter that characterises the bubble breakup.From QICS experiment, the data identify clearly when d e > 0.5 cm, as in Fig. 10, a  where D mj and d e both are in cm.This relation is suggested to represent the parameter for characterising the breakup in the numerical modelling as the CO 2 bubbles stretch horizontally before breaking up.The CO 2 bubble shape can be characterised, in another way, by the aspect ratio as defined as, Since the Eo number is the ratio of buoyancy force to interfacial tension, the parameter in the determination of bubble shape, the aspect ratio is examined against the Eo as illustrated in Fig. 11.It shows, that the aspect ratio decreases with Eo increases, the larger buoyancy force enhances the deformation of the bubbles to create a large width.In comparison with lab.experiment data, it shows a good agreement of the QICS results (circles in Fig. 11) with the experimental results and simulation presented by Bozzano and Dente (2001), thus validating the use of Eo to characterise the shapes of the CO 2 bubbles.

Interaction of CO 2 bubbles in seawater
From the processing of the videos filmed during the QICS experiment, interactions between the CO 2 bubbles have been observed.The interaction occurs as breakup of some of the large CO 2 bubbles which reduce their size and therefore velocity or coalescence between two CO 2 bubbles to give birth to a larger CO 2 bubble with higher velocity.
The example of the CO 2 bubble breakup, as shown in Fig. 12, observed in the QICS experiment captured videos.It can be seen  that the CO 2 bubble in the first photo (Fig. 12(a)) circled with red reached a large D mj before breaking into two CO 2 bubbles as showing in the two following photos (Fig. 12(b) and (c)).
The clearest breakups of the CO 2 bubbles have been selected and illustrated with the green triangles in Fig. 13 represented in the Eo-Re diagram.From the data in the Eo-Re diagram, it seems that the bubbles that experience break-up (the blue triangle) are hard to be differentiated from the other rising bubbles as the data shows they all fit a smooth trains.In order to indicate the breaking bubbles, it was proposed that the Eo and Re should be defined by using the major dimension, D mj , instead of the equivalent diameter, d e .By this definition, Eötvös number as Eo b and Reynolds number as Re b , as shown by the red circles in Fig. 13, the breaking bubbles are clearly identified.It can be concluded that the breakup occurs for the CO 2 bubbles when Eo b > 20 and Re b > 3500.For modelling of the breakup in the two-phase plume modelling in part 2 of this study (Dewar et al., 2014), Eq. ( 12) can be used to calculate the D mj of CO 2 bubbles for estimating the breakup Eötvös number and Reynolds number.
In addition to the CO 2 bubbles breakup, coalescence between bubbles has been observed at a frequency of 2.5 coalescences every second at the first 30 cm high above the sea floor.When the CO 2 bubbles coalesce, they form a larger bubble that will take longer to dissolute in the seawater.The data of bubbles coalescence frequency of 2.5 (Hz) observed from QICS experiment can be the reference value to develop the bubbles coalescence models for plume simulations in part 2 of this study (Dewar et al., 2014).

Conclusion
The dynamics of rising CO 2 bubbles in the Scottish seawater was investigated experimentally within the QICS project.Using video footage of the CO 2 bubble plume, the dimensionless numbers, such Re and Eo, have been used for data analysis and identifying the characteristics of leaked CO 2 bubbles.The results obtained from QICS experiment were compared with results published by studying the motion of a single CO 2 gas bubble in laboratory conditions.The agreement shows with a certain variation for the drag coefficient range mainly due to the difference between the experimental conditions: laboratory and open field experiments.
The bubbles leaked from QICS experiment are the bubbles with size ranging from 2.0 to 12.0 mm in equivalent diameter and velocity from 20 cm/s to 45 cm/s, which give the Reynolds number varying from 500 to 3500, respectively.The leaked bubbles experience the break-up and coalescences.The critical break-up Eötvös number is found to be E ob > 20, which should be characterised by the major dimension, D mj , rather than the equivalent diameter d e .
Some observation errors generated by setting the monitoring system, such as the location of the rulers, focal blur, the plume effect and the tidal effects along with the lack of observation in three dimensions due to a single video camera was used in the experiment, can be improved by redesign the system and measuring the seawater velocity simultaneously to detect the bubble relative velocity for identifying bubble dynamics more reasonably.Additionally, Greene and Wilson (2012) suggest that an improvement on gaining the initial bubble distribution would be through an acoustic method, proved through investigating initial bubble sizes with greater accuracy than imaging methods.
The interaction between the CO 2 bubbles is a very important phenomenon to characterise analytically.The experiments with larger leakage rate would generate a plume with strong bubble interactions, from which more suitable data can be obtained for development of the suitable correlations for plume model.Future experimental work on observing the bubble interactions in different water conditions (bubble size, bubble shape, directional velocity of water and bubble, temperature, salinity) are suggested to be carried extensively in the laboratory as well, in order to develop a statistical relation of coalescence or breakup of CO 2 bubbles.

Fig. 2 .
Fig. 2. Experimental set-up: (a) The observation field of multi-bubble-plumes.(b) Schematic view of the observation system.

Fig. 3 .
Fig. 3.A sample of the image of the bubble and measurement of CO2 bubble size.

Fig. 4 .
Fig. 4. The distribution of sizes and velocity of leaked CO2 bubbles measured directly from videos.
good liner relation is shown between the equivalent diameter of the CO 2 bubbles and their major axis, with a gradient of 1.82, D mj = d e if d e ≤ 0.5 cm 1.82d e − 0.4 ifd e > 0.5 cm (11)

Fig. 10 .
Fig. 10.The relation between major axis of the bubbles (D mj ) and the equivalent diameter (de) from QICS experiment (the symbols) and the liner equation (solid line).

Table 1
Physical properties of the selected fluids at 10 • C.