ISTP-16, 005, PRAGUE 16 TH INTERNATIONAL SYMPOSIUM ON TRANSPORT PHENOMENA ADVANCED CO OCEAN SEQUESTRATION TECHNOLOGY FOR GLOBAL WARMING Shuichiro Hirai 1*, Shohji Tsushima 1*, Ryoichi Muraoka 1*, Hisashi Sanda 1* and Masahiko Ozaki ** * Research Center for Carbon Recycling and Energy, Tokyo Institute of Technology, ** Nagasaki R & D Center, Mitsubisi Heavy Industries, Ltd. Corresponding author: hirai@mes.titech.ac.jp, phone& fax +81-3-5734-3336 Abstract CO sequestration in the ocean would be a highly advanced countermeasure to global warming if biological impact induced by CO dissolution into ocean water is sufficiently small. Releasing liquid CO droplets from a pipe towed by a ship is effective because of dilution due to the moving release point. If liquid CO droplet sizes are controlled, dissolution into ocean water while rising up due to buoyancy, would take a long distance, which leads to vertical dilution. Dilution could be enhanced by a multi-nozzle array placed in the direction of ship width. Here we show that CO dilution could be effectively further promoted by dispersing liquid CO droplets horizontally by a turbulent wake formed by bodies placed at upper part of the multi-nozzle array. Turbulence formed at the downstream of the bodies could enlarge the width of rising liquid CO droplets by turbulent mixing between sea water and droplets. We made a measurement of flow and statistical turbulent properties needed for numerical simulation behind a body. Based on the measured data, CO enhanced dilution was simulated. Increase of partial CO pressure ( pco ) and acidification ( ph ) are drastically decreased as compared with the case without dilution. pco larger than 1000µ atm decrease to 56.1% to 10.% using the present dilution technology. 1 Introduction Modern civilization is being maintained by a large consumption of the fossil fuel, and it is being predicted that CO concentration, now at 365 ppm, would become several times larger in future, if fossil fuels are continued to burned at the present rate. Therefore, it is immediately required to take global warming measures to restrict the increase of atmospheric CO concentration. One of such measures is to increase the storage of CO in the ocean by direct artificial injection of CO removed from flue gas at fossil-fuel-fired power plants, which was first proposed by Marchetti [1]. For implementation of CO sequestration in the ocean, it is needed to suppress variation that is caused by injecting a large quantity of CO in the ocean artificially. It is acidification ( ph decrease ) and pco of increase of partial pressure of CO ( ) seawater caused by injecting CO that influences marine life in the ocean. The estimation was made by Caufield et al [] and Sato [3]. The extent of influence is strongly related to the releasing methods, which was also estimated by Caufield et al []. To release CO droplets from pipeline outlet directly introduced from power plant induces significant acidification since a large amount of CO is successively released from a fixed position of pipeline outlet. Biological impact induced by ph decrease and pco increase would be smaller by a technology to release CO droplets from a pipe of moving ship [4]. 1
Shuichiro Hirai, Shohji Tsushima, Ryoichi Muraoka, Hisashi Sanda, Masahiko Ozaki Released CO from towed pipe of moving ship, tip of pipe at the depth around 1500-000 m, dissolve in the process of rising upward due to buoyancy in the sea water. Dissolved CO in sea water is diluted due to both the motion of moving releasing point and the vertical distance that is needed for droplet dissolution. If could be further diluted using nozzle array expanded in ship-width direction. Following Sato [3], pco needs to be less than 1000µ atm considering biological impact. Dilution using three directions, 100kg/s of injected CO would cause pco that is larger than 1000µ atm to be 56% of CO dissolved seawater. Therefore, further dilution technology is now essential for implementation of CO ocean sequestration. The present paper presents a new advanced technology to further enhance dilution of CO in moving ship technology. Liquid CO droplets in rising motion due to buoyancy could be effectively dispersed in the horizontal direction (ship-width direction) by turbulent wake flow over a body placed at upper part of CO release nozzle array. We measured flow and turbulence statistics behind a body (cylinder),which would be useful covering wide various conditions, i.e., wide range of Reynolds number. The measured data are employed for a simulation of turbulent mixing between CO droplets and seawater. Decrease of ph ( ph ) and increase of pco ( pco) are quantitatively estimated to be sufficiently small. pco larger than1000µ atm reduces to 10.%. by dispersing liquid CO droplets horizontally by a turbulent wake formed by bodies placed at upper part of the multi-nozzle array. It is schematically show in Fig.1. In Fig.1, cylinders are employed for a typical shape of the bodies. CO droplets are released from nozzle array and rise upward. Turbulent wake is formed by cylinders at the upper part of nozzle array and the droplets encounter turbulent wake (A-1 and A-) where they are dispersed in horizontal (ship width) direction by an intensive turbulent mixing of wake. Because CO droplet rise speed is small, approximately 0.1m/s, and ship speed is large, 3m/s, distance between cylinders and droplets become large and wake effect becomes small rapidly, so next wake (B-1 and B-), cylinders to form wakes are positioned downstream than (A-1 and A-), to enhance dilution. Therefore, turbulence formed at the downstream of the bodies could successively enlarge the width of rising liquid CO droplets by turbulent mixing between seawater and droplets. In order to estimate the present dilution technology quantitatively and precisely, we made an experiment and numerical simulation. 3 Experimental Methods and Results Although, turbulent mixing behind cylinders plays a key role on horizontal dispersion of released CO, it is known that drag coefficient over the critical Reynolds number An Advanced Technology for CO Dilution We propose a new advanced technology to enhanced CO dilution of moving ship. In addition to two dilution effects, one in the ship direction by the moving release point and the other in the vertical direction by controlling releasing size of CO droplet to obtain large distance for dissolution, dilution could be enhanced by a multi-nozzle array placed in the direction of ship width. We propose that CO dilution could be effectively further promoted Figure 1. Schematic view of the present advanced technology to enhance CO dilution
ADVANCED CO OCEAN SEQUESTRATION TECHNOLOGY FOR GLOBAL WARMING (Re=350,000) sharply decreases. Reynolds number (Re) is defined as Re U d ship cyl = (1) ν sea When we use the ship speed U = 3 m/ s, diameter of cylinder d = 0.m, ship Re equals to critical Reynolds number. Turbulent flow over the cylinder might in the range, below and over the critical Reynolds number depending on the cylinder diameter. However, flow characteristics and turbulent properties that are vital for prediction of CO droplet dispersion are not fully investigated in detail. Therefore, it is needed to characterize turbulent flow fields below and over the critical Reynolds number. Re number, especially over 350,000 is a high Reynolds number flow and, it is by no means of ease to construct experimental apparatus because of its handling. We constructed laboratory-scaled experimental apparatus to carry out measurements under the condition of Reynolds number of which were 70,000 and 500,000, below and over critical Reynolds number 350,000, respectively. We measured statistical turbulent properties by applying time-resolved particle image velocimetry (PIV). Figure shows an experimental apparatus we constructed together with an applied measurement system. The apparatus consists of cyl a square reservoir tank and a receiver tank. In the test section with 0.5m square channel, a circular cylinder with 0.1m in diameter was assembled. Once a gate is opened, reserved 3 water with volume of about 3m starts to fall and is accelerated by gravity force. In the case of Re number equals to 70,000, we equipped a resistance at the outlet of the test section to control the flow rate. Visualization and PIV results of Re = 500,000 and 70,000 are depicted in Fig. 3 and Fig. 4, respectively. In the experiments, nylon particles, averaged diameter of which is about 30 particle was illuminated by laser light sheet emitted from an Ar-ion laser and motion of illuminated particle was captured by high-speed CCD camera with frame rate of 1000 (frames/sec) after amplifying scattered light intensity by an image intensifier with shuttering speed of 50 µ s. The visualization results ((Fig. (a) Visualized flow field (b) Instantaneous velocity vectors Figure 3. Visualization of turbulent flow over a cylinder and its PIV results (Re=500,000). Figure. Experimental apparatus and measuring system of turbulent wake over a cylinder. (a) Visualized flow field (b) Instantaneous velocity vectors Figure 4. Visualization of turbulent flow over a cylinder and its PIV results (Re=70,000). 3
Shuichiro Hirai, Shohji Tsushima, Ryoichi Muraoka, Hisashi Sanda, Masahiko Ozaki 3 (a), Fig. 4 (a)) are analyzed to get instantaneous velocity field ((Fig. 3 (b), Fig. 4 (b)). As shown in an instantaneous velocity field (Fig. 3 (b)) in the case of Re = 500,000, zigzag motion of turbulent flow associated with vortex shedding is packed into narrow region behind the cylinder. On the other hand, in the case of Re = 70,000, relatively large vortex and wide zigzag motion of the flow can be recognized (Fig. 4 (b)). This distinct difference observed between their flow patterns leads to difference on turbulent properties as the followings. To examine turbulent properties in detail, we measured profiles of mean velocity and relative turbulent intensities from time-resolved PIV results. Turbulent properties measured at the downstream of 1.5 times of the cylinder diameter (x/d=1.5) are shown in Fig. 5 for both Reynolds numbers. As shown in Fig. 5 (a), width of wake in case of Re=500,000 is narrow as much as the cylinder diameter. On the other hand, in case of Re=70,000, wake width is twice as much as the cylinder diameter. Furthermore, turbulent intensities normalized by surrounding flow velocity along main flow and transversal directions were also very distinctive as shown in Fig. 5 (b) and 5 (c). Relative turbulent intensities in case of Re=500,000 are weaker than the ones Re=70,000 in both direction. These differences between the two Reynolds number flows must be explained from the flow visualization results shown in Fig. 3 and Fig. 4. Compared the flow pattern with Re = 500,000 with 70,000, size of vortex and zigzag motion of the flow is distinctively different each other. In the case of Re = 500,000, vortex marching is limited to narrow region behind the cylinder and vortex size seems to be small compared to the one in the intermediate Reynolds number flow. The data presented in Fig.5 could be used for the turbulent mixing simulation covering below and over critical Reynolds number. (a) Mean velocity profiles (c) relative turbulent intensity of transversal component (b) relative turbulent intensity of main (d) Reynolds shear stress flow component Figure 5. Turbulent properties measured at x/d=1.5 4
ADVANCED CO OCEAN SEQUESTRATION TECHNOLOGY FOR GLOBAL WARMING 4 Calculation Method Calculation Method of CO dissolution for both cases, with and without dilution by wake is presented in this section. Firstly, dissolution behavior of a single droplet covered with hydrate is presented. The dissolution behavior of liquid CO droplets is described by the relation d ( ρco V) dt ( C ) = ka C () 0 The measured surface concentration C0 of hydrate-covered CO droplet, which prescribes and determines dissolution rate due to the existence of hydrate, was employed [5]. The mass transfer coefficient k in eq () for high Schmidt number flows satisfies the relation [6]. ( ) 1 0.47 3 Sh= 1+ Sc+ 1/ Re 0.75Re (3) Here, ρco is the liquid CO density, V and A are the volume and surface area of a liquid CO droplet, respectively, C 0 and C are, respectively, the surface concentration at the CO droplet and the concentration at infinity; the Sherwood number is Sh= kd / D, the Reynolds number is Re= ud / CO ν, the Schmidt number is Sc = ν / D,ν is the kinematic viscosity, u the velocity, dco the CO droplet diameter, and D the diffusion coefficient. Introducing the coordinate z that represents the vertical distance from the release point, we obtain from Eq. () ( CO) ( 0 ) ( ) d ρco CO dd kc C d = (4) dz u ρ dz 3ρ z CO CO The rise velocity u z was estimated using the following equation, u z = 4d ( ρ ρ ) CO sea CO 3C D ρ sea g CO D 0.7 410.15 ( )/ ( 1,000, ) 0.5 ( ) ( ) C = + Re Re Re < C = 0.55+ 4.8/ Re Re < 10,000. (5) D In the case of without dilution using turbulent mixing behind the cylinders, CO droplets rise upward keeping the same width set by the nozzle array. We obtained quantitative CO dissolution behavior by solving eq. (4) with boundary conditions, i.e., released droplet size. As for the dissolution behavior of a CO droplet that rise upward in the wake region it is described as follows. Number density of CO droplets C along with velocity fields are solved with the dissolution of a single droplet. The calculation in the wake region is based on the time-averaged conservation equations of mass, momentum and CO droplet number density in which boundary layer approximation is applied. The basic equations are U V ρ + = 0 x y ( ρ ) U U p ρ U + V = uv x y x y C C ρ U + V = ρvc x y y (6) (7) (8) where xy, are the coordinates in the flow, perpendicular to flow directions, respectively. UVand, uvare, the time-mean and fluctuating velocity components in the xy, directions, respectively. ρ is the fluid density, p the pressure. ( ) denotes the time average. The correlations of velocity fluctuations in equation (7), (8), are evaluated from the turbulence models. The turbulence models applied to the present calculations is the k ε two equation model. In the k ε two equation model, kinetic energy of turbulence k and its dissipation rate ε are calculated from the following transport equations[7]. 5
Shuichiro Hirai, Shohji Tsushima, Ryoichi Muraoka, Hisashi Sanda, Masahiko Ozaki k k µ k U U V x y y y y T ρ + = µ + + µ T σk 0.5 k ρε µ y ε ε µ T ε ρ U + V = µ + x y y σ ε y ε U C + µ ε Cε1 T k y k ρε (9) µµ T U + 0 (10) ρ y The turbulent viscosity µ t is determined from the relation, µ t = ρc f k µ µ / ε (11) { ( )} fµ = exp.5/1 + R t /50 (1) The empirical constants in equations (9)-(11) are adopted from Jones and Launder (1973), and are listed as follows. C µ = 0.09, C ε 1 = 1.45, C ε = 1.95, σ k = 1.0, σ = 1.3. The terms ρ uv and ρ vc in equation (7) and (8) are modeled by the turbulent viscosity as 5 Estimation of the Present Dilution Technology We simulated a case in which 100 kg/s of liquid CO is released from the pipe towed by a moving ship at the depth of 1500m. The ship is assumed to move at a speed of 6 knots (3.09m/s) which is now considered to be a maximum ship speed. The released CO droplets are assumed to disperse in 10m width at the releasing point by multi-nozzle array for both cases with and without dilution of the present method. Dissolved CO follows the reactions, between carbonic acid ( ) 3 HCO, bicarbonate ion ( HCO 3 ) and carbonate ion ( CO 3 ) with the hydrogen ions ( H + ) in seawater: ph and pco were calculated under these conditions. CO + HO HCO 3 (15) HCO H + HCO (16) + 3 3 HCO H + CO (17) + 3 3 Fig.6 shows the two-dimensional contour profiles of pco in cross section vertical to ship moving direction. CO droplets of 1.5cm diameter released at 1500m depth, could travel U ρ uv = µ t y (13) C ρ vc = µ t y (14) Ukε,, are given at the cross section 1.5 diameter downstream from the cylinder using the measured data presented in Fig.5. Initial (upstream) condition of C is given by the width of nozzle array. Because equations (7)-(10) are boundary-layer type, they are solved downstream. (a) without dilution (b) with dilution Figure 6. Two-dimensional contour profiles of pco 6
ADVANCED CO OCEAN SEQUESTRATION TECHNOLOGY FOR GLOBAL WARMING distance of 84m (top of the figure) in both cases. In the case of dilution, we simulated the case that 6 cylinders of 0.m in diameter and.5m in length are placed in 0 sections (Fig. 1 depict the case of cylinders placed in sections). 0.m is the largest diameter to use large turbulence intensities (below critical Reynolds number) depicted in Fig.5. Dilution induced by turbulent wake of multi-cylinders are noted as dilution zone in Fig.6. Because pco profiles are symmetric with respect to the vertical line on the center of the nozzle arrays, the area of blocked line in Fig.1 is depicted in the figure. 5m depicted in the figure is the position of half-width of nozzle array. For the case without active dilution, CO droplets rises upward straightly, so pco shows an stripe-like profile. Large pco at the deep region originates in the fact that CO droplet size is large that leads to large amount of CO dissolution. In the case of with active dilution, CO droplets would disperse in ship width direction, so pco becomes diffusive profiles. It can be seen from the figure that the active dilution effect is distinctive. Two-dimensional contour profiles of ph is depicted in Fig.7 for both cases with and without dilution. Acidification is also drastically reduced by the present dilution technology. Figs. 6 and 7 demonstrates that dilution is effective but could not directly make an quantitative estimation, since Sato s data (004) depicts biological impact as function of pco and exposure time. Figs. 8 and 9 show probability density function (p.d.f.) of pco and ph, respectively. We can see from these figure the extent of CO dissolution effect on pco and ph. For example, if we want to know volume rate that pco larger than 1000µ atm adding the plots those are larger than 1000µ atm leads to 56.1% for without dilution and 10.% for with dilution. Biological impact data (Sato, 004) show that biological impact would be negligibly small under the condition of pco less than several hundreds µ atm (less than 1000µ atm ). Therefore present advanced technology reduces volume rate of CO Figure 8. Probability density function (p.d.f.) of pco (a) without dilution (b) with dilution Figure 7. Two-dimensional contour profiles of ph Figure 9. Probability density function (p.d.f.) of ph 7
Shuichiro Hirai, Shohji Tsushima, Ryoichi Muraoka, Hisashi Sanda, Masahiko Ozaki dissolved water of less than 1000µ atm from 56.1% to 10.%. If we use a criteria of 500µ atm, present technology reduces volume rate from 84.0% to 43.%. 6 Concluding Remarks The present paper reports a new advanced technology of CO sequestration in the ocean that CO would be effectively diluted to drastically reduce biological impact. It is based on the moving ship technology to dilute CO by (1) moving release point, () vertical distance by controlling droplet size and (3) nozzle array system. Further dilution could be achieved by the present technology that applies turbulent wake active dispersion of CO droplet. Before making the simulation, experiment of flow over cylinder was made to obtain valuable data, i,e., turbulence statistics, to make the simulation correct. We used cylinders to form turbulent wake due to the reason that turbulent statistics are fundamental and simple to use (only Reynolds number is the parameter). But if we use other body, like a flat plate placed vertical to ship direction and its drag coefficient is larger than that of cylinder, turbulent mixing would be further enhanced and dilution promoted. Because CO droplets released from nozzle initially dispersed (near the nozzle, not at shallow depth), droplets rise upward keeping the dispersed profile. So length of the bodies to form wake need not be long so it would not be a drag to drive ship. The present paper shows a concept to further dilution of CO. Size, position and numbers of the bodies are not optimized sufficiently, so it is expected that further study could promote further dilution. References [1] Marchetti, C., 1977, Climate Change, Vol. 1: 59-68. [] Caulfield, J. A., Auerbach, D. I., Adams, E. E. and Herzog, H. J., 1997, Energy Convers. Mgmt, Vol. 38:343-48. [3] Sato, T. 004. J. Oceanography, in print. [4] Nakashiki, N., Ohsumi, T. and Katano, N., 1995, "Technical View on CO Transportation onto the Deep Ocean Floor and Dispersion at Intermediate Depths", Direct Ocean Disposal of Carbon Dioxide, N. Handa and T. Ohsumi, eds. Tokyo; Terrapub: 183-194. [5] Hirai, S., Okazaki, K., Tabe, Y., Hijikata, K. and Mori, Y., 1997, Energy Int. J., Vol., No. /3:85-93. [6] Clift, R., Grace, J. R. and Weber, M. E., 1978, "Bubbles, Drops and Particles", Academic Press, New York. [7] Jones, W. P. and Launder, B. E. 1973. Int. J. Heat Mass Transfer Vol.16:119-30. Acknowledgement The authors would like to express our thanks that a part of this work was supported by RITE (Research Institute of Innovative Technology for the Earth). 8