A CFD simulation study of air-lift artificial upwelling based on Qiandao Lake experiment

A CFD simulation study of air-lift artificial upwelling based on Qiandao Lake experiment

Ocean Engineering 144 (2017) 257–265 Contents lists available at ScienceDirect Ocean Engineering journal homepage: www.elsevier.com/locate/oceaneng ...

2MB Sizes 4 Downloads 63 Views

Ocean Engineering 144 (2017) 257–265

Contents lists available at ScienceDirect

Ocean Engineering journal homepage: www.elsevier.com/locate/oceaneng

A CFD simulation study of air-lift artificial upwelling based on Qiandao Lake experiment

MARK



Haocai Huanga,b, Ji Wua, Zhao Yangc, Zhao Xuec, Wenke Gea, Yan Weia, , Wei Fana, Jiawang Chena, Ying Chena a b c

Ocean College, Zhejiang University, Zhoushan 316021, China Laboratory for Marine Geology, Qingdao National Laboratory for Marine Science and Technology, Qingdao 266061, China Development Center of Qingdao Nation Laboratory for Marine Science and Technology, Qingdao, Shandong 266237, China

A R T I C L E I N F O

A BS T RAC T

Keywords: Artificial upwelling CFD simulation Nozzle type Injection depth Volume of fluid method

The application of artificial upwelling technology could increase the primary marine productivity by bringing nutrient-rich subsurface water to surface. Although the artificial upwelling has been extensively studied, few numerical simulation studies on artificial upwelling device parameters, such as nozzle type, injection depth and injection hole diameter, have been conducted. This paper establishes a computational fluid dynamics (CFD) model to theoretically analyze how the upwelling efficiency were influenced by injection nozzle type, diameter of injection hole and injection depth. The comparison between the simulation results and the corresponding experimental data obtained in Qiandao Lake experiments firstly proves the accuracy of the predictions of CFD model in terms of upwelling efficiency via changing the parameters of the artificial upwelling device; secondly, it also indicates that the circular type of an air injection nozzle installed in deeper water depth with larger injection holes brings more deep water to the surface. It is believed that this study may shed some light on the improved design of an artificial air-lift upwelling device.

1. Introduction As resource on land has been consumed excessively, marine resource attracts more attention from human being. However due to the overfishing, it has been recently estimated that 75% of the world's commercial fish stocks get fished at or above mean sustainable levels (Kirke, 2003); moreover, caused by the environmental pollution and influenced by man's offshore economic activities, fishery resource has been declining significantly contrary to the increase of the world's population (Li, 2004; Sun, 2004). Unlike the advanced agricultural techniques, ocean fisheries still remain at the hunter-gatherer stage. Fortunately, as explained by Kirke, upwelling is the possible means to enrich the ocean water nutrients and thus, enhance the fish production in a sustainable way (Kirke, 2003). But natural upwelling only occurs in specific season and distributes unevenly in space and time. Upwelling regions are just a very small fraction of ocean (Polovina, 2001) and account for only 0.1% of the ocean surface but they yield over 40% of the world's fish catch (Roels et al., 1978). Therefore, an urgent need calls for artificial upwelling to bring to surface the Deep Ocean Water (DOW), which is the sea water of more than 200 m depth having very dense nutrient salts such as Nitrogen, Phosphorus, etc.



Correspondence author. E-mail address: [email protected] (Y. Wei).

http://dx.doi.org/10.1016/j.oceaneng.2017.07.048 Received 10 March 2017; Received in revised form 15 June 2017; Accepted 15 July 2017 0029-8018/ © 2017 Published by Elsevier Ltd.

To generate artificial upwelling, one method is to reshape the terrain under the ocean. That means to build seamounts to force current to become upwelling, example like placing fish-reefs (Nakayama, 2010). Another method, which is more achievable, is to resort to proper equipment to transport water and form upwelling. The Japanese researchers has proposed a scheme of large sea platform, called Ocean Nutrient Enhancer 'TAKUMI', which was equipped with upwelling pump to bring and discharge DOW into the euphotic layer to increase primary production of the sea and make a fishing ground (Ouchi, 2009, 2004). And in the University of Hawaii, a wave-driven artificial upwelling device was developed. When it heaves in the ambient waves, water inside the device keeps moving upward due to inertial forces, bringing the DOW to surface. Its performance with the consideration of the randomness of the incident waves has been analyzed by mathematical modeling (Liu and Jin, 1995). Utilizing the perpetual salt fountain (Stommel et al., 1956), Tsubaki research group has developed an apparatus consisting of pipe, buoys on the surface, measurement systems and a pump to capture the upwelling flow caused by the differences of the temperature and salinity (Tsubaki et al., 2007), but only under the condition of these differences existing does the method work.

Ocean Engineering 144 (2017) 257–265

H. Huang et al.

Apart from pumping water up directly, a kind of air lift pump for upwelling deep seawater was also proposed (Liang, 1996). Compared with the air flow rate, the seawater flow rate could become hundred times higher (Peng and Liang, 1999). The air-lift pump work principle is like this. The air is injected into a vertical pipe placed in water. Bubbles are rising inside the pipe and the density of the air-water mixture is less than that of the surrounding seawater. Therefore, the additional upward force acting on liquid is generated. Once the water level reaches up to the top of the pipe and the mixture flows out, the deeper seawater carrying the nutrient salt would be sucked continuously from the lower end (Liang, 1991; Liang and Peng, 2005). Based on Liang's theory, Fan's research group has evaluated the hydrodynamic performance of an air-lift artificial upwelling system, using experimental data and theoretical analysis. Their experiment was conducted in Qiandao Lake in Zhejiang Province, China (Fan et al., 2013). Although it is believed that experiments are able to provide reliable upwelling data, their costs cannot always be presumed acceptable. From this point, CFD numerical simulation may be considered as an alternative way to study artificial upwelling. Meng has taken a lake trial as the basis of simulation, in which a variety of time interval, number of grid and other initial parameters were set as simulation conditions, and then comparisons between the experimental results, the numerical simulations, and the theoretical predictions were made (Meng et al., 2013). However, with regard to how different types of air injectors influence the air volume flow rate, it is beyond the assumption of their model. In another simulation for a perpetual salt fountain, the k-ε and LES model were used to perform the turbulent flow computations. And the diffusion process was investigated to measure the effect of an artificial upwelling (Maruyama et al., 2013). Furthermore, Zhang has conducted the preliminary research on the influence of upwelling pipe's diameter and length on the lifting efficiency via simplified CFD model (Zhang et al., 2015). While in the study of heating for artificial upwelling, it applied the vertical velocity as a criteria to evaluate the efficiency of the artificial upwelling (Lv, 2014). In this paper, based on the previous studies and as the continuation of research on air-lift artificial upwelling system, we try to apply the CFD simulation method to investigate how different types of air injectors influence the air volume flow rate. The simulation results are verified by being compared with the experimental data. It is aimed to prove the accuracy of the simulation method in terms of predicting the tendency of the corresponding air volume flow rate when changing the parameters of air-lift artificial upwelling devices. Section 2 introduces the experiment setup in Qiandao Lake, in which the working principle of air-lift artificial upwelling system is also explained in details. Section 3 describes the computational approach, the governing equations are presented and the simulation boundary conditions are set. As a discussion part, Section 4 carries out the comparisons between the simulation results and the experimental results after the grid independence test and further analyzes how the performance of upwelling pipe is influenced by the injection nozzle types and injection depth. Section 5 wraps up the paper by the conclusion that the circular type of air injection nozzle installed in deeper water depth with larger injection holes brings more deep water to the surface.

Fig. 1. The schematic diagram of the experiment setup.

experiment, the upwelling pipe, made of reinforced PVC, with 0.4 m in diameter and 28.3 m in length was vertically deployed and completely submerged in water at depth 2.1 m. Its upper part was hoisted by the deck crane via two cables and a counter weight was hanging at its bottom to keep it in vertical position. During the test, the compressed air would be injected into water by a pump with an air injection nozzle. The nozzle, made of stainless steel tube, was placed at the center of the upwelling pipe so that the bubble instead of sticking to the pipe inner wall could blend with water freely. It divided the upwelling pipe into two parts, the air injection part (from O to B) which was about 7.5 m in length and the suction part (from B to C) about 20.8 m in length. Therefore the water suction part only had the water flowing within while the air injection part was filled with the two-phase water-air flow. The nozzle was designed with 4 different types, shown in Fig. 2. The (a) nozzle at the top left corner is a cross nozzle with 24 holes, 6 on each cross beam. And at the top right corner, (b) nozzle is the same but with 384 holes, 96 on each cross beam. The lower left (c) is a circular nozzle with 24 holes and the lower right (d) is the same but with 384 holes, on both of which were the holes distributed annularly and evenly. For 24 holes, each diameter was 2.0 mm; for 384 ones, each was 0.5 mm.

2. Experiment setup The research group traveled to Qiandao Lake in Chunan county, Zhejiang Province, to conduct the experiment. In this area, the average depth of water is around 34 m and little turbulent current was observed. The density of the lake water ranges from 997.79 kg/m3 to 999.40 kg/m3. The experiment location was chosen at 29°33′ 51′’ north, 119°11′ 9′’ east and a twin-hull craft was employed. On the deck of the twin-hull craft was arranged the experiment facilities which consisted of a power supply, a crane, deck working unit, data collection system, etc. The whole schematic diagram is illustrated in Fig. 1. In the

Fig. 2. The four types of the nozzle used in the experiment.

258

Ocean Engineering 144 (2017) 257–265

H. Huang et al.

Fig. 3. The hardware and software of the temperature monitoring system.

Therefore, for each nozzle type, its total area of the holes was 75.4 mm2. The inside diameter of all pipe tube was 8 mm. All the injection holes were drilled by plasma arc drilling. The configuration and specification of the nozzles can also be found in Fan and Meng (Fan et al., 2013; Meng et al., 2013). In the experiment, the air went through an air compressor (Model W-0.36/8), an on/off valve, a pressure control valve, a mass-flow meter (Model MF5619) and the nozzle and finally got injected. In the pressure valve, the pressure of compressed air was reduced to working pressure, about 1.2–3.2 bar. The mass-flow meter was used to control the air flow rate. Sensors were used to record the data, the. An electromagnetic flowmeter (MCG/KLL-magB) was deployed at the joint of the air injection part and the water suction part to measure the upwelling flow rate. And two sets of temperature sensor were placed at the inner upper end and the outer lower end respectively. Fig. 3 displays the hardware and the software of the temperature monitoring system. 3. Computational approach

Fig. 4. Boundary condition and the initial distribution of water.

In order to simulate the air-lift artificial upwelling process more accurately, a three-dimensional full scale hydrodynamics model was built. A CFD commercial software STAR-CCM+ was applied to solve the governing equations, namely the mass, momentum and energy conservation equations. The geometric parameters of the upwelling pipe and the computational domain, the governing equations, the boundary conditions and the settings for the commercial code would be explained in details in this section.

of 10% and the nozzle refined area with the size of 2% of the basic size. The reason to choose the inner pipe area and the around pipe area to be refined is that the water in these two areas is more involved in the artificial upwelling process compared with the whole water field. Both refined area are cylinders with the different diameters, 0.2 m for the inner refined area and 0.5 m for the other, but with the same length as the pipe. The bottom refined area and the nozzle refined area are significant too in the upwelling process. The former is the area where water begins to flow into the device and the later where air is injected and begins to mix with the water. Both areas are cylinders with the different diameters, 1 m for the bottom area and 0.2 m for the other, but with the same length 0.5 m, i.e. 0.25 above/below the bottom and 0.25 m above/below the nozzle respectively.

3.1. Geometrical parameters of the upwelling pipe and the computational domain In CFD simulation, the first step was to build a geometric model of the simulated objects. In this paper, the simulated objects were the upwelling pipe and the injection nozzle. Then a computational domain with the injection nozzle and the upwelling pipe contained would be set and its corresponding governing equations would be solved. As shown in Fig. 4, the computational domain is a cuboid with the dimension of 10 m×10 m×38.4 m and with the upwelling pipe in the center. The lower end of the upwelling pipe is deep water inlet and its upper end is deep water outlet. The distance between the deep water inlet and the bottom of the computational domain is 5 m. The initial distribution of water and air in the computational domain was set by a user-defined function. In the whole computational domain, the initial distribution of water is from the bottom to 2.1 m above the deep water outlet. It means that the water outlet was submerged beneath the water surface 2.1 m. The whole model is discretized into meshes, shown in Fig. 5. The basic size of gird is set as 0.5 m. In order to obtain more accurate simulation results, four refined grid areas are set, which are the inner pipe refined area with the size of 10% of the basic size, the around pipe refined area with the size of 50%, the bottom refined area with the size

3.2. Governing equations It is known that fluid in motion must satisfy the governing equations, namely the mass, momentum and energy conservation equations which need to be solved in the computational domain. Although the temperature of the Qiandao Lake water is not the same in depth direction, as the influence of temperature difference is trivial, we assume the flow in pipe is isothermal. This means the upwelling flow has no heat exchange in water and gas during the upwelling. The heat exchange does have some effect on mixing process of liquid and gas. Since the mechanism of air-lift artificial upwelling is buoyancy driven, the heat exchange has little influence on lifted water flow rate and thus it should be safe not to be considered here. So in the computational domain, only the mass and momentum conservation equations need to be solved. The differential forms of the mass conservation equation (the continuity equation) in Cartesian coordinate system is displayed as follows (Liu, 2002): 259

Ocean Engineering 144 (2017) 257–265

H. Huang et al.

Fig. 5. Simulation model and Demonstration of grid.

∂ρui ∂ρ + =0 ∂t ∂xi

μ ⎞ ∂ε ⎤ ∂ε ∂ε ∂ ⎡⎛ ε ε2 ⎢⎜μ + t ⎟ ⎥ + cε1 (Pk + cε3Gb ) − cε2 + ui = ∂t ∂xi ∂xi ⎢⎣⎝ σε ⎠ ∂xi ⎥⎦ k k

(1)

Where ρ is the density of the fluid; t is time; xi represents the direction of Cartesian coordinate e.g. direction of x, y and z; ui is the velocity component corresponding with xi . Since the flow speed in pipe in Qiandao Lake experiment was 0.318–0.955 m/s, it could be deduced that the Reynolds number was 1.27E5–3.82E5 and the flow in pipe was regarded as turbulence. Three basic approaches to modeling turbulence are available in STAR-CCM+, namely Reynolds-averaged Navier-Stokes (RANS) equations, large eddy simulation (LES), detached eddy simulation (DES). Due to the limit of computation resource, the first approach was chosen in simulations. The differential forms of the momentum conservation equation, RANS equation, in Cartesian coordinate system is displayed as follows (Liu, 2002):

⎡ ⎞⎤ ∂p ∂ ∂ ∂ ⎢ ⎛ ∂ui (ρui ) + ρ (uiuj ) = − + μ⎜⎜ − ρui′u′j ⎟⎟⎥ ∂t ∂xj ∂xi ∂xj ⎢⎣ ⎝ ∂xj ⎠⎥⎦

Here, σk is the Prandtl number often between 0.9 and 1.0.σk =1.0 was adopted in this model. Pk is the shear stress production term and Gb is the buoyancy production term; The shear stress production term Pk and the buoyancy production term Gb are:

⎞2 ⎛ Pk = μt ⎜ 2SijSij ⎟ ⎠ ⎝ Gb = − μt

(2)

cε3 = tanh |

k2 ε

V | U

(9)

3.3. Boundary conditions As showed in the Fig. 4, the bottom surface of the computational domain, used to model the lakebed, the surfaces of upwelling pipe and the injection nozzle, except the air injection holes on it, were set to be smooth wall boundaries. Side surfaces of the cuboid computational domain were set to be symmetry boundaries. Its top surface was set to be pressure outlet boundary with a standard atmosphere pressure. Since close to the wall boundary, turbulent flow has not fully developed; while a little distance away from the wall, the increased mean velocity gradient causes the turbulent kinetic energy to increase. Numerous experiments and researches have shown that the flow conditions in this two areas are apparently different. Therefore, the near-wall area were to be considered when wall boundary was applied in the computational domain. There are two ways to deal with the near-wall area, namely the wall function using empirical formula to solve the problem and the near-wall model with the means of enhancing the gird which indicates more computing resource and longer computing time. Therefore, the method of wall function was applied.

(3)

(4)

cμ is constant. cμ = 0.09. The additional transport equations of the k-ε turbulence models are as follows. μ ⎞ ∂k ⎤ ∂k ∂k ∂ ⎡⎛ ⎢⎜μ + t ⎟ ⎥ + Pk + Gb − ε + ui = ⎢ ∂t ∂xi ∂xi ⎣⎝ σk ⎠ ∂xi ⎥⎦

(8)

Where V is the component of the flow velocity parallel to the gravitational vector and U is the total component of the flow velocity perpendicular to the gravitational vector. Combining the Boussinesq approximation with the k-ε turbulence models, the governing equations could be closed. Next, the boundary conditions would be given to solve the governing equations.

μt is the turbulent viscosity; k is the turbulent kinetic energy. While some simpler models rely on the concept of mixing length to get the turbulent viscosity μt in terms of mean flow quantities, the k-ε turbulence models were applied to derive μt and additional transport equations of the k-ε turbulence models needed to be solved. μt = cμρ

g ∂ρ ρ ∂y

(7)

Where Sij is the mean rate of strain tensor; g is gravitational acceleration. ε is the turbulent energy dissipation rate. σε , cε1, cε2 , cε3 are coefficients. The coefficients adopted in this model are: σε = 1.3, cε1 = 1.44 , cε2 = 1.92 . The coefficient, cε3 by default, is computed as:

In turbulent flow, the fluid motion is decomposed into a timeaveraged mean flow component and a fluctuating component. Here, ui and uj are the velocity components in Cartesian coordinate. ui′ and u′j are the fluctuating components of ui and uj ; ui and uj are the timeaveraged mean components of ui and uj . xj is also the direction of Cartesian coordinate; μ is dynamic coefficient of viscosity and −ρui′u′j is the Reynolds stress tensor. In order to model the Reynolds stress tensor in terms of the mean flow quantities, the Boussinesq approximation is applied. The relationship between the Reynolds stress tensor and the averaged velocity gradient is established, which is:

⎛ ∂u ∂ uj ⎞ 2 ⎟⎟ − δijρk −ρui′u′j = μt ⎜⎜ i + ∂xi ⎠ 3 ⎝ ∂xj

(6)

(5) 260

Ocean Engineering 144 (2017) 257–265

H. Huang et al.

Table 1 The velocity for air injection holes. Air Flow Rate (m3/s)

1.26E−3

2.55E−3

3.81E−3

5.0E−3

At standard atmosphere pressure (m/ s) At 6.1 m depth (m/s) At 9.6 m depth (m/s) At 16.1 m depth (m/s)

16.71

33.81

50.5

66.31

10.53 8.69 6.52

21.30 17.58 13.19

31.82 26.26 19.70

41.78 34.48 25.86

For the air injection holes, the velocity inlet boundary was employed. The velocities are obtained as follows. According to the Qiandao Lake experiment data, four air injection flow rates (1.26E-3 m3/s, 2.55E-3 m3/s, 3.81E-3 m3/s, 5.0E-3 m3/s) were chosen in CFD simulations to get the corresponding deep water flow rates. The total area of the air injection holes is about 75.4 mm2. The velocity for each air flow rate at a standard atmosphere could be calculated thereafter, which are 16.71 m/s, 33.81 m/s, 50.5 m/s, 66.31 m/s respectively. Furthermore, with the consideration of air compressibility, the volume of air flow rate would reduce under hydrostatic pressure in water. Three different air injection depths, i.e. 6.1 m, 9.6 m, and 16.1 m, were chosen in CFD simulations. The reduction factors were obtained by using ideal gas state equation:

PV = nRT

Fig. 6. Grid independence test for water flow rate.

the other hand, to compute unlimited grids simulation also seems unlikely. Hence, it's necessary to prove the discretized grids, coarse or finer, capable of yielding the accurate results. Therefore, the grid independence test was performed. During the test, the injection nozzle was the cross shape with 24 air injection holes with 2 mm in diameter. The water depth for installing the injection nozzle was 9.6 m. Fig. 6 shows that along with the increasing air volume flow rate, the deep water volume flow rate climbed up steadily. These two curves obtained on different grid level, i.e. 1,050,000 and 1,530,000 mesh elements, agreed within a level of tolerance. Furthermore, in Fig. 7, two ratio curves also fitted well with one another, which demonstrated that the ratio between the water flow rate and the air flow rate was also independent of the grid, irrespective of its number. We could conclude that the CFD simulation is reliable and trustworthy.

(10)

Here, T is the temperature; n and R are the constants; P is the pressure; V is the volume at pressure P. Due to the assumption of the isothermal flow within the pipe, Eq. (10) could be rewritten as:

PV = c

(11)

Here, c is a constant. Thus, the reduction factors are obtained, namely 0.63, 0.52 and 0.39 in different air injection depths. Then the velocity values of velocity inlet boundary under different air injection depths could be calculated, shown in Table 1.

4.2. Comparisons between the simulation results and the experimental results

3.4. Settings before solving

Since deeper water volume flow rate implies more nutrient being brought to surface, it is regarded as an important parameter to evaluate the performance of air-lift artificial upwelling. During the Qiandao Lake experiment, it was detected by an electromagnetic flowmeter. In the test, the deep water flow rate data corresponding to different air injection types and different air injection flow rates were collected. We have compared the numerical simulation results with the experimental data thereafter. Fig. 8 to Fig. 11 show the comparison results, in which the curve marked with square represents the simulation result and the curve marked with dot the experimental data. In Figs. 8 and 10, the deep water volume flow rate is plotted versus the supplied air volume flow rate for

For CFD simulation, some settings need to prepare before solving. The computational domain needs to discretize into mesh elements. An unstructured polyhedral mesh method was applied and the computational domain was discretized into 1,530,000 mesh elements by STARCCM+, since it is relatively easy and more efficient to build and requires no more surface preparation than the equivalent tetrahedral mesh method. With regard to the deep water flow rate, at different water injection depths, it could be calculated by Eq. (12):

→ Vw = ∯→ v ⋅dS

(12) → Here Vw is the deep water flow rate, v is the velocity in vertical → direction and S is the area of observation plane. Three curves of flow rate would be drawn in the same diagram thereafter. Under the assumption that when the curves get close and finally coincide, it could be regarded that the air-lift upwelling is steady. The whole solving process continues until steady behavior of the lifted deep water flow rate is achieved. About the time step, it was set to be 5E-4 s initially. And along with the simulation model running, larger time step was set. The largest time step was about 0.005 s. Due to the different air injection flow rate, the time needed to allow the solution to become steady was also different and varied between 30 s and 100 s. 4. Results and discussions 4.1. Grid independence test It is generally acknowledged that the more grids are set in the simulation, the result is expected to get closer to the exact solution. On

Fig. 7. Grid independence test for ratio between water flow rate and air flow rate.

261

Ocean Engineering 144 (2017) 257–265

H. Huang et al.

Fig. 8. Water flow rate vs. air flow rate for cross injection nozzle.

rate; thus, the water flow rate differences are larger.

the cross injection nozzle and circular injection nozzle respectively. In Figs. 9 and 11, the ratio between water flow rate and air flow rate is plotted versus the air volume flow rate. It can be found that:

1. The tendency of the numerical simulation and experiment data is almost the same. The two curves in each figure agree well. Since in this paper, we are more concerned with the tendency of the curve, it would be considerably safe to draw the conclusion that even if there is only numerical simulation result curve, the tendency of deep water flow rate could still be predicted. From here, the geometric model and computation method proves valid for the time being. It is also understood that the numerical simulation model is not perfect and accurate and needs to be improved in future study. But at least to a certain degree, the existing experimental data has validated the CFD simulations and are able to provide strong support for the numerical analysis.

1. All curves with square are above the ones with dot, which means the simulation results are always higher than the corresponding experimental data. The possible reasons are presented here. First, the CFD model tends to be ideal. The wall is assumed to be smooth with less resistance and thus causes the higher deep water flow rate. Second, in actual upwelling pipe, there are some sensors and other equipment installed on the pipe wall. They would have hindering effects on lifting the deep water. However, these are not considered in the CFD model. Thirdly, the experimental data are inevitably prone to be less accurate due to the accuracy of equipment and researcher's operations.

4.3. Performance influenced by injection nozzle types and injection depth

1. The simulation results are getting closer to the corresponding experiment data as the air volume increases, i.e. the deviation between the two diminishes when the air volume flow going up and the water flow rate differences are larger when the air flow rate is lower. This may be explained by the phenomena that in the experiments, the increased air flow volume would weaken the hindering effects caused by the unsmooth wall and the sensors installed within the pipe; thus, the experimental data gets closer to the simulated results. While the air flow rate is lower, the hindering effects in the experiments become prominent and slow down the water flow rate significantly; on the other hand, they have not been considered in the simulation which would generate higher water flow

The performance of upwelling pipe has also been analyzed through varying the injection nozzle type and the injection depth. The simulation results are presented from Fig. 12 to Fig. 16 where curves marked with pentagon represent the circular injection nozzle and the ones marked with triangular the cross injection nozzle. It could be found that: 1. From Fig. 12 to Fig. 14, all curves with pentagon are above the ones with triangular. It means in terms of the deep water volume rate parameter, the circular injection nozzle showed better performance than the cross injection nozzle with different amount of holes in

Fig. 9. Ratio between water and air flow rate vs. air flow rate for cross injection nozzle.

262

Ocean Engineering 144 (2017) 257–265

H. Huang et al.

Fig. 10. Water flow rate vs. air flow rate for circular injection nozzle.

influenced by injection depth, in which the corresponding depths are 6.1 m, 9.6 m and 16.1 m. This indicates the performance of air-lift artificial upwelling is improved when the injection depth becomes deeper. This could be because the deeper injection depth means a longer air lifting path, which gives more time for buoyancy to lift the water. Thus it results high air and water volume flow rate. Nevertheless, the deeper injection depth would incur engineering difficulties, for example, the instruments are required to be more pressure sustainable and the pump needs to be high-powered. Finding the balance between the injection depth and the feasibility of engineering is also part of the future research.

different injection depth. The simulation has been repeated and still presented the same results. The reason could be that when using the cross injection nozzle, the injected air filled in the central area of the upwelling pipe. The air occupied most area of the transversal surface, leaving little space for water. Thus, only a small portion of water and air would be blended into mixture and the majority of lifted deep water would be extruded to the pipe wall resulting in the increased resistance near the wall. Consequently, it is comparatively more difficult for the deep water in cross nozzle to rise. But for the case of circular injection nozzle, the injected air occupied less central space of upwelling pipe and allowed more space for deep water to rise up. 1. Compare diagram (a) with diagram (b) in each figure, it is not difficult to find that under the condition of the same nozzle injection type and the same injection depth, the nozzle with 24 holes generated higher deep water volume flow rate than the nozzle with 384 holes did. This well indicates that the former performs better than the latter. Considering the effect of different amount holes, as the total area of the holes is the same, the larger quantity of the hole there are, the smaller a single hole is. When gave the same pump to provide air flow, small holes will cause more resistance, and this would severely affect the efficiency of water lifting. We can conclude that circular injection nozzle is better than cross injection nozzle and large holes are also better than small ones. Of course, the hole cannot be unlimitedly large. This parameter needs to be further studied in future work.

In this paper, we try to apply the CFD simulation tool to theoretically analyze the performance of an air-lift artificial upwelling system and further investigate the injection nozzle parameters that would possibly influence the upwelling efficiency. It is a full scale simulation and the size of the upwelling pipe and the air injection nozzles are the same as the experiment. Two types of injection nozzles, i.e. the cross shape and the circular shape, with 24 holes and 384 holes respectively were simulated in the injection depth condition of 6.1 m, 9.6 m and 16.1 m. After comparing the simulation results with the corresponding experiment, the following conclusion would be safely drawn:

1. With the air injection depth increasing, the deep water volume flow rate is climbing up too. Figs. 15 and 16 show the performance

1. The simulation results are always higher than the corresponding experimental data.

5. Conclusions

Fig. 11. Ratio between water and air flow rate vs. air flow rate for circular injection nozzle.

263

Ocean Engineering 144 (2017) 257–265

H. Huang et al.

Fig. 12. Water flow rate vs. air flow rate for injection depth of 6.1 m.

Fig. 13. Water flow rate vs. air flow rate for injection depth of 9.6 m.

Fig. 14. Water flow rate vs. air flow rate for injection depth of 16.1 m.

2. The figures also prove that the circular type nozzle performs better than the cross shape. 3. The larger diameter injection holes perform better than smaller ones. 4. The injection nozzle in deeper position performs better than in shallower one.

1. The simulation results are getting closer to the corresponding experiment data as the air volume increases. 1. The existing experimental data has validated the CFD simulation results and the tendency of deep water flow rate could be predicted by numerical analysis.

We believe this study may shed some light on the improvement of future design of an artificial air-lift upwelling device. For the further study, we would continue to investigate the optimal injection nozzle parameters based on CFD simulation, taking consideration the cost and the engineer-

1. All three device parameters, i.e. the nozzle type, the diameter of injection-holes and the injection depth, do have influence on the water-lifting efficiency.

264

Ocean Engineering 144 (2017) 257–265

H. Huang et al.

Fig. 15. Water flow rate vs. air flow rate for cross injection nozzle.

Fig. 16. Water flow rate vs. air flow rate for circular injection nozzle. Lv, M., Yan, X., Nie, X., Pan, H.C., Liu, H.Q., 2014. Experimental studies of differential heating for artificial upwelling. Comput. Model. New Technol. 18 (6), 299–303. Liu, C.C.K., Jin, Q., 1995. Artificial upwelling in regular and random waves. Ocean Eng. 22 (4), 337–350. Liu, G.Z., Miu, G.P., 2002. Advanced Fluid Mechanics. Jiao Tong University Press, Shanghai. Meng, Q., Wang, C., Chen, Y., Chen, J., 2013. A simplified CFD model for air-lift artificial upwelling. Ocean Eng. 72 (7), 267–276. Nakayama, A., Yagi, H., Fujii, Y., Sano, T., Takeda, M., Okano, T., 2010. Evaluation of effect of artificial upwelling producing structure on lower-trophic production using simulation., 66(1), pp. 1131–1135. Ouchi, K., Ohmura, H., 2004. Design concept and experiment of ocean nutrient enhancer "TAKUMI". Oceans 1, 322–327. Ouchi, K., 2009. Results from real sea experiment of ocean nutrient enhancer TAKUMI. In: Proceedings of the 28th International Conference on Ocean, Offshore and Arctic Engineering, Honolulu, HI, May 31-Jun. 05. Vol.4, 1473–1480. Peng, H.K., Liang, N.K., 1999. A study on air-lift artificial upwelling (continued). In: Proceedings of the 21st Conference on Ocean Engineering, Republic of China, 589– 596. Polovina, J.J., Howell, E., Kobayashi, D.R., Seki, M.P., 2001. The transition zone chlorophyll front, a dynamic global feature defining migration and forage habitat for marine resources. Prog. Oceanogr. 49 (1–4), 469–483. Roels, O.A., Laurence, M., Farmer, W., Van Hemehyck, L., 1978. Organic production potential of artificial upwelling marine environment. Process Biochem. 13 (2), 18–23. Stommel, H., Arons, A.B., Blanchard, D., 1956. An oceanographical curiosity: the perpetual salt fountain. Deep Sea Res. Part II Top. Stud. Oceanogr. 3 (3), 152–153. Sun, J., Liu, D.Y., Zhang, C., et al., 2004. The preliminary study in size-fractionated biomass of phytonplankton in the central Bohai Sea, the Bohai Straits and its adjacent area (in Chinese with English abstract). Acta Oceanol. Sin. 25, 103–112. Tsubaki, K., Maruyama, S., Komiya, A., Mitsugashira, H., 2007. Continuous measurement of an artificial upwelling of deep sea water induced by the perpetual salt fountain. Deep Sea Res. Part I Oceanogr. Res. Pap. 54 (1), 75–84. Zhang, Q., Xu, L., Yu, W., 2015. A simulation of air-lift artificial upwelling in vertical pipe. International Conference on Intelligent Systems Research and Mechatronics Engineering (ISRME 2015), 1347–1351.

ing feasibility. The corresponding new prototype of injection nozzles would be produced and new experiments would be carried out thereafter. Acknowledgements This work was supported by the NSFC Projects under Grants 41576031 and 51120195001 and the Scientific Research Foundation for the Returned Overseas Chinese Scholars, State Education Ministry with Grant [2014]1685. We would like to acknowledge Dr. HanGe, Miss ShanLin and Miss Jianying Leng who helped the preparation of the Qiandao Lake experiments and the crew members on the site. Additionally, the authors are particularly grateful to the anonymous reviewers for their inquiries and comments which not only make our manuscript more readable but shed light on the research itself. References Fan, W., Chen, J., Pan, Y., Huang, H., Chen, C.T.A., Chen, Y., 2013. Experimental study on the performance of an air-lift pump for artificial upwelling. Ocean Eng. 59 (2), 47–57. Kirke, B., 2003. Enhancing fish stocks with wave-powered artificial upwelling. Ocean Coast. Manag. 46 (9–10), 901–915. Li, G.C., Wang, B.S., Li, M., 2004. Research on artificial upwelling and fisheries in various waters of different depth. Ocean Sci. 28 (11), 49–56. Liang, N.K., 1991. Concept Design of Artificial Upwelling Induced by Natural Forces. Oceans '91.Ocean Technologies and Opportunities in the Pacific for the 90's. Proceedings, Vol.1, 391–393. Liang, N.K., 1996. A preliminary study on air-lift artificial upwelling system. Acta Oceanogr. Taiwan. 35 (2), 187–200. Liang, N.K., Peng, H.K., 2005. A study of air-lift artificial upwelling. Ocean Eng. 32, 731–745.

265