StudiesonFactorsInfluencingHydrodynamicCharacteris

J. Ocean Univ. China(Oceanic and Coastal Sea Research)https:///10.1007/s11802-019-3706-zISSN 1672-5182, 2019 18 (1): 193-202/xbywb/E-mail:xbywb@Studies on Factors Influencing Hydrodynamic Characteristicsof Plates Used in Artificial ReefsTANG Yanli1), YANG Wenzhao1), SUN Liyuan2), ZHAO Fenfang1), *, LONG Xiangyu1),and WANG Gang1)1) College of Fishery, Ocean University of China, Qingdao 266003, China2) Shandong Hydrobios Resources Conservation and Management Center, Yantai 264003, China(Received October 25, 2017; revised January 29, 2018; accepted September 20, 2018)© Ocean University of China, Science Press and Springer-Verlag GmbH Germany 2019Abstract As a simplified model of artificial reefs, a series of plate models punched with square or circular openings are de-signed to investigate the effects of openings on the hydrodynamic characteristics of artificial reefs. The models are grouped by various opening numbers and opening-area ratios. They are physically tested in a water flume or used in the numerical simulation to obtain the drag force in the uniform flow with different speeds. The simulation results are found in good agreement with the experimental measurements. By the non-dimensional analysis, the drag coefficient specified to each model is achieved and the effects of openings are examined. It is found that the key factor affecting the drag coefficient is the open-area ratio. Generally, the drag coefficient is a linear function of the open area ratio with a minus slope. The empirical formulae for the square and circular openings respectively are deduced by means of the multiple regression analysis based on the measured and numerical data. They will be good references for the design of new artificial reefs. As a result of numerical simulation, the vorticity contours and pres-sure distribution are also presented in this work to better understand the hydrodynamic characteristics of different models.Key words artificial reefs; drag coefficient; hydrodynamic characteristics; model experiment1 IntroductionArtificial reefs are man-made structures placed in the natural water area, either in the bays or along the coastal line, to restore and optimize the ecological environment. It has been reported that the artificial reefs can affect the bay’s natural cleaning capacity to improve the water quality (Antsulevich, 1994). The artificial reefs play a role of protecting coastal environment for nursery, micro- habitat and food supply against the illegal otter trawling, which increase the biodiversity and biomass of benthos and fishes, and provide facilities for ecotourism as well (Relini et al., 2007). The main function of artificial reefs is to disturb the flow field and form upwelling or com-plex wake regions, which can promote the circulation of the water column (Kim et al., 2014; Jiang et al., 2016). Once the reef is installed to the sea bottom, it is sub-jected to the action of waves and sea current, namely, the hydrodynamic force. As an obstacle fixed in the water flow, the term of drag force is borrowed to designate this force here. The reef will be displaced when the drag force outweighs the maximum static friction between the reef and the seabed. Meanwhile, when the overturning * Corresponding author. Tel: 0086-532-82032511E-mail: zhaoff@ moment of drag force acting on the reef is larger than the restoring moment caused by the weight of the reef body, the reef will be turned over. In both cases, it might de-crease the desired ecological effect of reefs or reduce their service life. Therefore, it is important to consider the structure stability of artificial reefs. However, most of the researches on the artificial reefs have been focused on their effects on the ecological environment. In order to ensure the stability of artificial reefs, it is necessary to investigate the surrounding flow field and the hydrody-namic forces acting on the reefs (Liu et al., 2011; Liu et al., 2014).There are mainly three methods to study the flow field around the artificial reefs to obtain the hydrodynamic forces, i.e., field investigation, experiments in flume or wind tunnel, and numerical simulation. The first method, i.e., field investigations, is very costly both in money and time due to the complexity of sea environment. Thanks to the development in modern computational mechanics and computer technology, satisfactory accuracy can be achieved by numerical calculation with reasonable efforts (Woo et al., 2014). The model tests in flumes or wind tunnel give the direct and intuitive results. Carrying out both physical experiments and numerical calculation can ob-tain reliable results (Ross et al., 2004; Nisugi et al., 2004; Su et al., 2008; Cui et al., 2009; Liu and Su, 2013; Taka-hashi et al., 2015; Mellibovsky et al., 2015; Xu et al.,TANG et al . / J . Ocean Univ . China (Oceanic and Coastal Sea Research ) 2019 18: 193-202194 2017).The artificial reefs are usually made of block bodies, opened with several holes to stimulate flows. Their hy-drodynamic characteristics are mainly determined by their shapes including the opening rate of those holes. In the flume experiments of porous cubic reefs, a concept of open area ratio is proposed by Kageyama and his col-leagues (1981). It is defined as the ratio between the pro-jected area of the opening part, i .e ., the holes, and the full projection area of the reef surface.10/A A γ=, (1)where γ is the open area ratio (or briefly opening ratio); A 1 is the projected area of openings, and A 0 is the total projected area of the reef. As pointed out by Kageyamaet al. (1981), this parameter plays an important role to theflow around the artificial reefs.In order to investigate how the hydrodynamic charac-teristics of artificial reef are affected by its opening area, including the shape of reef holes, their numbers and openarea ratio, a series of model tests are performed in thepresent work. The models are square plates punched withsquare or circular holes. The models are grouped accord-ing to their opening modes. In one group, the reefs havethe same number of holes but with various opening ratios. In the other group, the reefs have various numbers ofholes while with the same opening ratio. These modelsare tested in a water flume at different flow speeds tomeasure the drag force acting on the model. Meanwhile,numerical simulation is made with the same reef modeland at the same flow speed. The results of these two ap-proaches are compared to check the accuracy. Based onboth the experimental and numerical results, the rela-tionship between the hydrodynamic characteristics of artificial reefs and their various opening shapes are de-tected.2 Experiments in the FlumeThe experiment is carried out in the hydrodynamic flume of Tokyo University of Marine Science and Tech-nology. The main scale of water flume is 22.00 m (length) × 8.35 m (width) × 3.05 m (height). The flow speed can be adjusted from 0.10 to 2.0 m s −1 with an accuracy of 2%. The testing model is fixed in the middle part of the flume experimental section, where an observation window is installed with a scale of 9.00 m (length) × 2.20 m (width) × 1.60 m (water depth).To avoid the complexity of reef shapes and focus on the effects of reef openings, the models used in the pre-sent work are square-shaped plates made of acrylic mate-rial. They take the same size, i .e ., with a side length a = 0.2 m and a thickness of t = 0.005 m. The plates are punched with square or circular holes neatly. The dimen-sions of all the models are listed in Table 1. The opening area of a models is given by A 1 = nL 2 for square openings and A 1 = n πR 2 for circular openings, where n is the num-ber of openings, L is the side length of square openings,and R is the radius of circular openings as shown in Table 1. According to the definition made in Eq. (1), the open-ing ratio of the models is calculated as γ = nL 2 /a 2 in the case of square openings or γ = n πR 2/a 2 for those with cir-cular openings.The models are divided into three groups. In the first two groups, the opening ratio varies between 0.1 and 0.6 but the number of openings is kept unchanged, one open-ing for the first group and four for the second one. In the third group, the opening ratio is kept to γ = 0.3 but the number of openings varies.Table 1 The dimensions of all 30 models usedin the experimentOpening shape Circle Square Number of holes n Opening ratio γRadius R (m) Side length L (m)0.1 0.0357 0.0632 0.2 0.0505 0.0894 0.3 0.0618 0.1095 0.4 0.0714 0.12650.5 0.0798 0.141410.6 0.0874 0.15490.1 0.0178 0.0316 0.2 0.0252 0.0447 0.3 0.0309 0.0548 0.4 0.0357 0.0632 0.5 0.0399 0.0707 40.6 0.0437 0.0775 2 0.3 0.0437 0.0775 6 0.3 0.0252 0.0447 8 0.3 0.0219 0.0387Fig.1 The sketch of the circulative hydrodynamic flume and the arrangement of model testing devices.In the experiments, the water depth is set to 1.60m. The tests are conducted in the middle part of the flume experimental section. The model is fixed to a strut which is connected to the rotating table of the flume frame through a load cell (see Fig.1). The center of the model is located at a depth of 1.25 m beneath the water surface. By adjusting the rotating frame of the flume, the model is fixed to a position that its plate face is perpendicular to the incoming flow. The speed of testing flow is set from V = 0.1 m s −1 to 1.0 m s −1 at a step of 0.1 m s −1. The flow speed is very slow so that the effect of free surface is considered to be negligible. The force in the direction of the flow, i .e ., the drag force F x is measured by the load cell during the tests. The drag force acting on the strut is measured at each flow speed in advance of the modelTANG et al . / J . Ocean Univ . China (Oceanic and Coastal Sea Research ) 2019 18: 193-202195installation, which will be subtracted from the former measurement for model tests. The interference of the strut is thus corrected.3 Numerical SimulationThe numerical simulation is performed by means of the software CFX in the present work. The software CFX is developed under the assumptions that the fluid is an in-compressible and Newtonian viscous fluid. The flow has reached at a steady state. Based on these hypotheses, the governing equations, i .e ., the continuity equation and the Reynolds-averaged Navier–Stokes equation in the three- dimensional Cartesian coordinate system, are expressed as follows:()0i iu t x ρρ∂∂+=∂∂, (2) ''()()()()i i i j i j j i j j ju p u u u u u t x x x x x μ∂∂∂∂∂∂+=-++-∂∂∂∂∂∂,(3)where u i and u j represent the average velocity (i = 1, 2, 3 represent components x , y , and z in the Cartesian coordi-nate system), p is the static pressure, ρ is the density of the fluid, μ is the dynamic viscosity of the fluid, u i ' and u j ' represent the fluctuating velocity, and ''i j u u - are the Reynolds stresses.The standard k-ε turbulence model is accepted in the present numerical analysis, which is described by the following equations (Liu et al ., 2013; Li et al ., 2017):()()()i t k i jk j ku k k G t x x x ρμρμρεσ⎡⎤∂∂∂∂+=++-⎢⎥∂∂∂∂⎢⎥⎣⎦, (4)(i t i j j u t x x x ερεμρεεμσ⎡⎤∂()∂()∂∂+=++⎢⎥∂∂∂∂⎢⎥⎣⎦ 21k C G C kkεεεερ2-. (5)Here, G k represents the generation of turbulence ki-netic energy owing to the mean velocity gradients; C μ, C 1ε, and C 2ε are constants; and σk and σε are the turbulent Prandtl numbers for k and ε, respectively. These parame-ters were calculated as follows:()j i ik t j i j u u u G x x x μ∂∂∂=+∂∂∂, 2,0.09t k C C μμμρε= =,121.44, 1.92, 1.0, 1.3.C C εεκεσσ= = = =In the above equations, the summation convention hasbeen adopted, i.e., any term in the equation that containsthe same index twice should be summed over that index. In this simulation, the element-based finite volume method (EbFVM) (Profito et al ., 2015; Kim et al ., 2016) in the CFX was used to discretize the governing equa-tions. High resolution advection scheme, which is robust in the solving advection term, was used in this research. A coupled solver, which solved the hydrodynamic equa-tions as a single system, was adopted in the procedure of obtaining the numerical solutions of discretized govern-ing equations. For steady-state problems, the time-step was to guide the approximate solutions in a physically based manner to a steady-state solution. This reduced the number of iterations required for convergence to a steady state (ANSYS-Inc, 2013). Furthermore, the solution was assumed convergent, whereas all residuals including the continuity, velocities (x , y , z components), pressure, tur-bulent energy and dissipation rate fell below 10−4, and a maximum of 1000 time steps in the steady state simula-tions was used to ensure the final results were stable and converged.The boundary conditions are the same as those used in the verification tests when studying the effect of spacing on the flow field. Fig.2 shows the boundary conditions. It is assumed a boundary condition of trivial pressure gra-dient at the downstream surface. The boundary at both sides and the bottom of the fluid domain are rigid sta-tionary no-slip walls. Since the effect of free surface is neglected in the flume tests, it is assumed a free stream over there in the numerical simulation. The center of the model is set at the same depth as in the flume tests. In the numerical simulation, the upstream is a uniform flow of constant speed.Fig.2 The computational domain and boundary condi-tions used in the numerical simulation (L is reef length, Wis reef width, and H is reef height).The computational domain can be narrowed properly in order to reduce computation, and the dimension of the computational domain is set according to the size of the reef monomer (Cui et al ., 2011; Zheng et al ., 2015). The length of computational domain is 601 times of the thick- ness of the reef (3.005 m). The width is 10 times of the reef width (2 m). The height is 10 times of the reef height (2 m). The reef models are located at the center of the computational domain. The grid system is shown in Fig.3, the total number of nodes and elements were approxi-mately 4.50×105 and 2.49×106 in each case after meshing respectively.TANG et al . / J . Ocean Univ . China (Oceanic and Coastal Sea Research ) 2019 18: 193-202196Fig.3 Grid system in computational domain.In order to verify the accuracy of the numerical simu-lation, the models with different opening numbers (n ) and an open area ratio 0.3 are taken as examples to be evaluated, and the results are compared to the experiment measurements (see Table 2).Table 2 The comparison between experimental results andsimulated values of C d (γ = 0.3)Drag coefficient C d nOpening shape Experimental results Simulated values Relativedifference2 Square 0.982 0.981 0.10% 2 Circular 0.977 1.013 3.68%4 Square 0.989 1.029 4.04% 4 Circular 1.008 1.020 1.19% 8 Square 0.997 0.921 7.62% 8 Circular0.9730.9572.36%Then, a group of twelve models with nine opening holes (see Table 3) are used in the numerical simulation. The open area ratio varies from 0.1 to 0.6 as those used inthe experiments. The calculated results are sump up to the empirical formula to further extend the experimental re-sults.Table 3 The parameters of all twelve models in thenumerical simulationOpening shape Circle Square n γR (m) L (m) 0.1 0.0119 0.02110.2 0.0168 0.0298 0.3 0.0206 0.03659 0.4 0.0238 0.04220.5 0.0266 0.04710.6 0.0291 0.05164 Results and Discussion The drag force measured from model tests are plottedin Fig.4 against the flow speed. In the legend of Fig.4, regardless to the opening shape, it can be observed that the drag force increases with the decrease of opening ratio at the same flow speed when the opening number is kept unchanged (see Figs.4(a)–(d)). However, if the open- ing ratio is fixed, the change of opening number will not make significant change in the drag force (see Figs.4(e, f)). The drag force of panels with no openings in all cases are completely larger than the others. Generally, the drag force is roughly a quadratic function of the flow speed. To express the results in dimensionless expressions, the drag coefficient and the Reynolds number are adopted in the present work. The drag coefficient is defined as:22/(0.5)d x C F V a ρ=, (6)while the Reynolds number is given by/Re V a v =⋅, (7)here, V is flow speed (unit is m s −1); v is the kinematicviscosity of the water (m 2 s −1); a (m) is the characteristic length, namely the side length of the plate models.Expressed as the drag coefficient, the experimental data are presented together with the Reynolds number inFig.5. The conventions used in the legend of Fig.4 are also applied here and the plots in Fig.5 are arranged in a similar way as those in Fig.4.Similar to the results in Fig.4, the drag coefficient de-creases as the opening ratio increases for the same open-ing number (Figs.5(a)–(e)), when the Reynolds number isgreater than 0.8×105. In the same region of Reynolds number, the influence of opening number is not obvious for fixed opening ratio (see Figs.5(e) and (f)). Further-more, when the Reynolds number exceeds 0.8×105, the drag coefficient for a certain model remains basically unchanged even though the Reynolds number increases. It is said that the automatic model region has been reached. An average value of the drag coefficients in the automatic model region are evaluated by the SOLVER function built in the Excel and this value is referred asthe drag coefficient specifying to that model later on.To examine the relative importance of the three factors, i.e ., the opening-area ratio, opening shape and opening numbers, to the drag coefficient C d , the correlation ana- lysis for three-factor variances is made by means of a program written in R language by calculating the F valueTANG et al . / J . Ocean Univ . China (Oceanic and Coastal Sea Research ) 2019 18: 193-202197that can judge the difference. The results are shown in Ta-ble 4. It can be concluded that the influence of the opening- area ratio on the drag coefficient is dominant (P < 0.001) while the opening shape and opening numbers have rela-tively less influence on the drag coefficient (P < 0.05).Table 4 The results of correlation analysis amongthree factorsF valuePr (>F ) Correlation analysisOpening shape 5.429 < 0.05 P < 0.05Open area ratio 115.799 < 0.001 P < 0.001 Opening numbers5.754 < 0.05P < 0.05To examine the effects of opening shapes, the drag co-efficient C d for the models of group 1 (or group 2), i.e .,with one (or four) opening(s), is plotted in Fig.6(a) (or Fig.6(b)) verse the opening ratio. Meanwhile, results in Fig.6(c) is the drag coefficients for a group of models with nine openings, as a result of numerical simulation. It can be seen that the influence of opening shape is not so significant especially when the opening number is in-creased. Generally, the drag coefficient C d decreases as the opening ratio increases for fixed opening number. A regression analysis is made to the above data to obtainempirical formulae for models with the same opening number and opening shape. The results are presented in Table 5 together with the coefficient of determination R 2, which is an indicator to show the fitness of the regression equation to the used data. It can be observed that the drag coefficient C d is basically a linear function of the opening ratio γ.To compare the drag coefficient C d for models with different opening numbers, the results are reorganized so that they are presented in the same illustration for the models with the same opening shape in Fig.7(a) for square opening and (b) for circular opening. It can be observed from the diagram in Fig.7 that the drag coefficient C d for models with nine openings is obviously greater than the other ones. In addition, the opening number has more significant influence on the drag coefficient than the opening shape.To obtain a more general empirical formula for the drag coefficient C d as a function of both the parameters, i.e ., the opening-area ratio γ and the number of openings n , a multiple regression analysis is made based on the experiment and numerical data. The results are given by the following two equations.Fig.4 The drag force measured from flume tests as a function of flow velocity. (a) one square opening; (b) one circular opening; (c) four square openings; (d) four circular openings. The opening ratio is fixed at γ = 0.3 with varying opening number in (e) for square openings and (f) for circular openings. S means square and C means circle, which both indi-cate the shape of openings. Panel represents the model with no openings.TANG et al. / J. Ocean Univ. China (Oceanic and Coastal Sea Research) 2019 18: 193-202198Fig.5 The drag coefficient vs. Reynolds number of experimental data: (a) one square opening; (b) one circular opening;(c) four square openings; (d) four circular openings. The opening ratio is fixed at γ=0.3 with varying opening numbers in (e) for square openings and (f) for circular openings. S means square and C means circle, which both indicate theshape of openings. Panel represents the model with no openings.Fig.6 The relationship between the opening shape and drag coefficient C d for models with (a) one opening; (b) four openings; (c) nine openings (results of numerical calculation).TANG et al . / J . Ocean Univ . China (Oceanic and Coastal Sea Research ) 2019 18: 193-202199Table 5 Regression equations of the drag coefficient Cd as a function of the opening ratio γ formodels with the same opening shape and numberSquare openingsCircle openingsOpening numberRegression equationR 2Regression equationR 21 C d = −1.0062γ + 1.2467 0.99669 C d = −1.2416γ + 1.3136 0.99832 4 C d = −1.0725γ + 1.2889 0.98579 C d = −1.0267γ + 1.2666 0.95379 9 C d = −0.9007γ + 1.2759 0.95098 C d = −0.9350γ + 1.2875 0.88409Note: R denotes the coefficient of determination.Fig.7 The relationship between opening numbers and drag coefficient C d for models having (a) square openings (b) circular openings.For square openings (with R 2 = 0.975):1.260.0070 1.05d C n γ=+-. (8)For circular openings (with R 2 = 0.932):1.280.0068 1.14d C n γ=+-. (9)It can be observed from the regression equations that the coefficient of the opening ratio is much greater than the coefficient of the opening number. Comparing Eqs.(8) and (9), the difference between the corresponding coefficients is not so significant. Therefore, the opening- area ratio plays an important role to influence the drag coefficient C d among these three factors. The drag coef- ficient is generally linearly decreases as the opening ratio increases. The confidence interval of the regression equa-tion is 95%, it can be observed that the regression equa-tions have a significant regression effect at the sig- nifi-cance level of α =0.05.Fig.8 The vorticity contours along the center cross-section of the fluid domain for 4 models with opening ratio γ = 0.3. (a) S1-0.3, one square opening; (b) C1-0.3, one circular opening; (c) S4-0.3, 4 square openings; (d) C4-0.3, four circu-lar openings.TANG et al . / J . Ocean Univ . China (Oceanic and Coastal Sea Research ) 2019 18: 193-202200 The numerical simulation is also performed in the pre-sent work. As a validation of the numerical results, the models with four openings, square or circle, and the opening ratio 0.3 are taken as examples to be calculated by the software CFX. The results of numerical calcula-tion are compared with the data measured in the flume tests and the comparison is shown in Table 2. The rela-tive error is less than 8%.Then, a group of twelve models with nine openings and various opening ratios (Table 3) is used in the nu-merical simulation to further extend the experimental results. The calculated drag coefficient for these models are demonstrated in Fig.6(c) as a supplement of meas-ured data.For a better understand of the mechanism how the opening mode affects the hydrodynamic characteristics of artificial reefs, the fluid field around the plate models is numerically simulated by means of ANSYS CFX. The models used in the simulation have one or four square (or circular) opening(s) with opening ratio γ = 0.3. The in-coming flow speed is set to V = 1.0 m s −1 in the numerical simulations.The simulated flow field is illustrated in the diagrams of velocity vectors (Fig.8). It can be observed that there exists an eddy wake behind the model. Compared to the flow behind one-opening model, the vortexes in the wake of four-opening model are greater and the flow becomes more unstable. It might cause a bigger pressure differ-ence between the front and back surfaces of the modeland lead to a larger value in drag coefficient. On the other hand, disturbed flow field around the artificial reefs is just the purpose of their construction to improve the surrounding environment and attract more fishes. So it would be a trade-off between their functions and stability in the design stage of the artificial reefs. The wake vol-ume size of different simulated models was compared (Fig.9). The wake volume has a negative correlation with the open area ratio, and the volume reaches the maximum when the open area ratio is 0.1. In the actual production, we need to increase the wake volume and reduce the drag coefficient at the same time. The panel with 4 openings (γ = 0.3) is the ideal design on account of its stable hydro-dynamic performance and larger wake volume.Fig.9 The relationship between wake volume and open area ratio.Fig.10 The diagrams of pressure distribution on the front surface of 4 models with opening ratio γ = 0.3. (a) S1-0.3, one square opening; (b) C1-0.3, one circular opening; (c) S4-0.3, 4 square openings; (d) C4-0.3, four circular openings.The pressure distribution is evaluated in the simulation as well. As an example, the pressure distribution on thefront surface of four models is demonstrated in Fig.10. It has a tendency that the pressure is more likely to concen-TANG et al . / J . Ocean Univ . China (Oceanic and Coastal Sea Research ) 2019 18: 193-202201trate to the area near the openings of the model especially at the sharp corners of openings. The model with less opening number has a smaller drag coefficient under the same opening ratio due to fewer sharp corners. Taking the models with square openings as an example, the length of the boundary line for 4 openings is two times of that for a single opening. We can find the drag force of plate with 4 openings is bigger than that with 1 single opening when the flow velocity is 1 m s −1 (Fig.4). The reason might be the length of the boundary is in propor- tion to the frictional force, and the drag force increases with longer boundary line. It can be concluded that the drag coefficient may be determined by both the number of sharp corners and the length of the boundary line.5 ConclusionsIn the present work, a series of models are designed to investigate the effects of openings to the hydrodynamic characteristics of the artificial reefs. To simplify the problem, the models are basically two-dimensional and square-shaped plates punched with various numbers of square or circular openings. Various opening ratios are obtained by adjusting the size of openings. Both flume experiments and numerical simulations are carried out with these models in a uniform flow. The drag forces acting on the models are measured in the tests or evalu-ated in the simulation.It is observed that the drag force is generally a quad-ratic function of the flow speed. Expressed in dimen-sionless terms, it is found that the automatic model re-gion has been reached as the Reynolds number exceeds 0.8×105. The drag coefficient is approximately a constant regardless of the change of Reynolds number in that re-gion. Then the average value is obtained by the regres-sion analysis, which is referred as the drag coefficient C d specifying to that model.The dependence of this drag coefficient C d on the opening modes, i.e ., the opening shape, opening number and opening ratio, is investigated. It can be observed that the drag coefficient C d is mostly affected by the opening ratio especially in the case of circular openings. It de-creases linearly as the opening ratio increases. Increasing opening number may increase the value of drag coeffi-cient C d , but the contribution is much less than the open-ing ratio. Comparing the coefficients in these two equa-tions, it can be observed that the opening shape will not give much difference to the drag coefficient C d , particu-larly as the opening number increases. The empirical formulae offer a reference for the stability evaluation in the design stage. For the practical application in the fish-ing engineering, more complicated models including the three dimensional effects should be investigated in the future work. At the same time, the problems emerged in the actual production should be noticed. The effects of sessile organisms attached on the artificial reefs should be considered, considering the sessile organisms can in-crease the surface friction to influence the drag force and the fluid field around the reefs.The numerical simulation is made in the present work. The numerical results are in good agreement with the measured data, and the accuracy of numerical calculation is acceptable. In addition, the numerical simulation pro-vides pictures of the fluid field and the pressure distribu-tion to make a better understand of the hydrodynamic characteristics of artificial reefs. It will be a useful and powerful method in the development of new type of arti-ficial reefs.AcknowledgementsThe study is supported by Shandong Provincial Oceanic and Fishery Department (Ecological Simulation Test of the Offshore Area in Shandong Peninsula) and the Pri-mary Research and Development Program of Shandong Province (Nos. 2016CYJS04A01 and 2017CXGC0107).ReferencesANSYS-Inc, 2013. ANSYS Fluent Theory Guide, Release 15.0. Antsulevich, A. E., 1994. Artificial reefs project for improve- ment of water quality and environmental enhancement of Neva Bay (St. Petersburg County region). Bulletin of Marine Science , 55 (2-3): 1189-1192.Cui, Y ., Guan, C. T., Wan, R., Li, J., and Huang, B., 2009. Re- search and numerical simulation on features of flow field around artificial reefs. Journal of System Simulation , 21 (23): 7393-7396, DOI: 10.16182/ ki.joss.2009.23.007 (in Chinese with English abstract)Cui, Y ., Guan, C. T., Wan, R., Li, J., and Huang, B., 2011. Nu- merical simulation on influence of disposal space on effects of flow field around artificial reefs. Transactions of Oceanology and Limnology , 22 (2): 59-65 (in Chinese with English ab- stract).Jiang, Z. Y ., Liang, Z. L., Zhu, L. X., and Liu, Y ., 2016. Nu- merical simulation of effect of guide plate on flow field of artificial reef. Ocean Engineering , 116: 236-241, DOI: 10. 1016/ j.oceaneng. 2016.03.005.Kageyama, Y ., Osaka, H., Yamada, H., Hasegawa, H., and Naka- kura, Y ., 1981. Water tunnel visualization around the per- forated cube: 1st report, flow visualization in a center plane. Memoirs of the Faculty of Engineering Yamaguchi University , 31: 351-359 (in Japanese).Kim, D., Woo, J., Yoon, H. S., and Na, W. B., 2014. Wake lengths and structural responses of Korean general artificial reefs. Ocean Engineering , 92: 83-91, DOI: 10.1016/j.oceaneng. 2014.09.040.Kim, D., Woo, J., Yoon, H. S., and Na, W. B., 2016. Efficiency, tranquillity and stability indices to evaluate performance in the artificial reef wake region. Ocean Engineering , 122: 253-261, DOI: 10.1016/j.oceaneng.2016.06.030.Li, J., Zheng, Y . X., Gong, P. H., and Guan, C. T., 2017. Nu- merical simulation and PIV experimental study of the effect of fl ow fields around tube artificial reefs. Ocean Engineering , 134: 96-104, DOI: 10.1016/j.oceaneng.2017.02.016.Liu, J., Huang, H. L., Chen, S., Wu, Y ., Li, L. Z., and Rao, X., 2014. Model test of the hydrodynamic characteristics of two vertical cambered V type otter boards. Chinese Journal of Hydrodynamics , 29 (2): 183-188, DOI: 10.3969/j.issn1000- 4874.2014.02.007 (in Chinese with English abstract).Liu, J., Xu, L. X., Zhang, S., and Huang, H. L., 2011. Research。