Hydrodynamic Study on Energy Capturing Performance of Horizontal Axis Blades Under Sub-Low Speed Tidal Current

The research on the hydrodynamics of blades is mainly focused on sea areas with high-speed current. However, the average velocity in most territorial waters of China is smaller than 1 m/s, and the lift type of airfoil blades has limited application in most of these conditions. Therefore, it is of great significance to study the tidal current energy capture of blades in sub-low speed sea areas. The effect of flow impact resistance on the blade at sub-low current speed is considered and a new type of thin-walled blade based on the lift type of blade is proposed, and then the lift-impact combined hydrodynamic model of horizontal axis blade is established. Based on this model, and considering the characteristics of tidal current and velocity in the sea area of Yushan Islands, simulation and optimization of blade design are carried out. Additionally, the horizontal axis thin-walled blade and the NACA airfoil contrast blade under the same conditions are developed. By using a synthetical experimental test system, the power, torque, rotational speed and load characteristics of these two blades are tested. The performance of the thin-walled blade and the design theory are verified. It shows that this type of blade has much better energy capture efficiency in the sub-low speed sea area. This research will promote the study and development of turbines that can be used in low- speed current sea areas in the future.


Introduction
The blade is the key component of tidal current energy capture device, and its design is directly related to the efficiency of energy capture. China has a vast territorial water. However, according to the results of FVCOM (Qi et al., 2009) simulating model of tidal current in China's sea area during a tidal period, 70% of China's sea areas belong to the low velocity tidal current area, which shows that the average velocity is smaller than 1.5 m/s. The velocity distribution is shown in Fig. 1 (Song et al., 2006;Ashkenazy and Gildor., 2011). Among them, the low-speed tidal current area is divided into sub-low speed tidal current area and slow-flow area. The sub-low speed tidal current area is defined as the area where the maximum velocity of neap tide is smaller than 1 m/s, the maximum velocity of spring tide is smaller than 1.5 m/s, or the average velocity is between 0.75 and 1 m/s. At present, the research on highspeed tidal current power generation device has been carried out for many years (Chen et al., 2018), and the underlying theory has been gradually improved. There are MWlevel tidal power stations in foreign countries that have started demonstration operation, and many large-capacity development projects are also under way. However, the research on tidal power generation in China is still at its infancy. There are few studies on tidal power development in low-speed sea areas, and the types of blades are mainly vertical-axis Savonius blades (Derakhshan et al., 2017;Ma et al., 2016). In view of China's special conditions, there is a need to conduct research on tidal current energy capturing of blades in sub-low speed sea area. The reserve of tidal current energy at sub-low speed is huge (Uihlein and Magagna, 2016) in China. Thus, it is necessary to carry out studies on tidal current power generation devices in China.
Among horizontal axis blades, Chen et al. (2013) designed a horizontal axis turbine based on BEMT theory (Guo et al., 2015;Belloni et al., 2017). The cross-section airfoil of the blade is NACA 63415 airfoil, which has good lift-drag characteristics in the high velocity region. The UK MCT developed SeaGen (Thiringer et al., 2011.), which is the first tidal current energy capture device with yaw system, with pitch-controlled turbine capable of pitching at 180°. The design is for flow velocity being 2.25 m/s, and the energy utilization power factor is as high as 0.45. GEM of Naples University in Italy (Coiro et al., 2017) plans to use suspension impeller with proposed design velocity of 2.7 m/s and estimated power of 100 kW. It was tested near Venice in 2011. The average flow velocity at the test site was about 1.5 m/s, and its output power reached 20 kW. OpenHydro, an Irish company, has developed Open-Center turbines using the outer-loop diversion function to increase the incoming flow velocity (Baker et al., 2014), and successfully connected them to the grid on Orkney Island. Fuglsang and Bak (2004) of Denmark optimized the blades with multi-variables, aiming at reducing the cost of output energy for blade power generation. Green Energy Corporation of the United States has designed a tidal current power generation device for automatic convection system. When the tidal current velocity reaches 2.2 m/s, the device generates high power, up to 35.9 kW. The device can automatically rotate at the optimum angle to capture more tidal currents according to the change of the direction of tidal current, which improve the efficiency of energy capture. The Lunar Energy Ltd of UK has designed a blade with a diameter of 19.5 m and a diversion tube with a diameter of 25 m. The device adopts a symmetrical double diversion hood design, which can collect a large amount of tidal energy. The expected power of the device is 2 MW. The re-commended flow velocity can reach 3 m/s due to the energy gathering effect of the diversion hood. China has also participated in some related research, such as LI Wei of Zhejiang University, who has designed a 25 kW independent horizontal axis tidal current energy generation system (Liu et al., 2016). The diameter of impeller and hub are 4.4 m and 0.44 m, respectively. The advised flow velocity is 2 m/s and the design tip speed ratio is 5:1. Wang used flexible blades to realize optimal airfoil transformation at variable velocity. Chen et al. (2010) and Xu et al. (2015) used CFD software to simulate the geometry of the blade, and obtained lift and drag coefficients of different airfoils, which provides reference for the selection of airfoil. Shun (2010) and Chen (2012) proposed the method of servo variable pitch to improve the efficiency of tidal current energy capture and start-up performance.
The research conducted by domestic and abroad scholars has laid a good foundation for the development and utilization of tidal current energy (Xu et al., 2015). However, most of the research focuses on high-speed tidal current capturing blades. Usually, the maximum tidal current velocity (at least 2 m/s) is chosen as the design velocity, and the lift design method (such as Glauert design method and Wilson design method) is adopted, while the impact force is ignored. In the case of time-varying flow velocity, the tidal current velocity is lower than its design velocity at most of the time, and the total energy capture capacity of lift type blades in this case is lower. The quantity is very low, and the actual applicable sea area is rather unclear. On the other hand, turbines driven by impact resistance are almost Savonius vertical axis turbine. Therefore, there is a need to optimize the design of horizontal axis turbine blades in sublow speed tidal current region (Goundar et al., 2012), which is the focus of this paper. In Section 2, the lift-impact combined hydrodynamic model of horizontal axis blade under sub-low speed tidal current is established, and a new type of the thin-walled blade is proposed. In Section 3, the blade is designed, optimized and simulated according to specific characteristics of Yushan Islands. And in Section 4, a con- GAO Ru-jun et al. China Ocean Eng., 2020, Vol. 34, No. 3, P. 374-386 375 trast lifting blade is designed, and the power, torque, rotational speed and load characteristics of these two blades are tested and analyzed.
2 Hydrodynamic analysis of blades used to capture tidal current energy at sub-low speed 2.1 Theory of designing sub-low speed blades 2.1.1 Deficiencies of existing blade design theories According to Bernoulli's principle (Lam and Long, 2014), as shown in Fig. 2a, the force generated by different velocities and pressures on the upper and lower surfaces is called lift of differential pressure around the wing. The common design methods are based on differential pressure around the wing. On the other hand, when the water impacts the blade, the velocity direction should change, and a direct impact force will be produced on the blade surface (Zhang et al., 2019), as shown in Fig. 2b. The lifting force and impact resistance around the blade together provide the torque that makes the impeller rotate.

C p
The power coefficient of the impeller is can be expressed as: (1)

C M
And the torque coefficient of the impeller is can be expressed as: (2) where, T is torque of turbine, D is diameter of turbine, b is turbine extension, V is the inflow speed, R is the turbine radius. Eq. (1) shows the relationship between the power coefficient , the moment coefficient and the circumferential velocity ratio of the impeller.
The main idea of Glauert design method (Sherry et al., 2013) and Wilson design method (Wu et al., 2008) is based on high flow velocity and tip speed ratio to ensure the max-imum lift coefficient at the angle of attack (Wei et al., 2015). According to Eq. (1), the maximum capture efficiency can be achieved if the maximum moment coefficient at a large tip speed ratio Benini and Toffolo, 2002) is guaranteed. But two other factors should not and cannot be neglected in the realistic blade design especially under the conditions of low-speed tidal current. The first is that power is consumed by impeller rotation and rear stage loss, and the second is that low-speed blades usually fail to achieve a high spike-to-speed ratio. Thus, the power coefficient expression obtained by the latter stage can be rewritten as follows: is the loss coefficient and is the loss power.
In the case of low Reynolds number, the proportion of impact force in the total force increases. In addition, the impact force is less sensitive to angle of attack than the pressure differential force around the wing. The blade designed mainly by impact force has a higher energy capture power coefficient in a wide velocity range, and the impact force is characterized by heavy load and low speed. Such method is ideal for the design of blades with low flow velocity and tidal current energy. Therefore, we propose that it is necessary to introduce impact force in the design of blades. We should model it, substitute it into the equation of axial force and torque, and couple the total balance equation to establish the theoretical hydrodynamic model (Guo et al., 2018;Batten et al., 2008) for the sub-low speed blade and improve the theory behind the design of low-flow blade.

Thin-walled airfoil
Thin-walled airfoil is a type of airfoil suitable for combining lift and impact force at sub-low speed. As shown in Fig. 3, because the lower wing surface is the same as the upper wing surface and concave, it is easy to stall at a large Reynolds number. But at a low Reynolds number, the lower wing surface bears much impact force and has a certain lift around the wing. It is suitable for hydro-turbines designed for medium-to-low-speed to have the characteristics of low  starting torque, wide energy trapping velocity range, and heavy load at low flow velocities.

x f
The main parameter of the thin-walled airfoil is the bending distribution. As shown in Fig. 3, indicates the chord position of the bending distribution, f is the arc height and c is the chord length. f/c is the relative curvature, among which the maximum relative curvature and the position of the maximum curvature have the greatest impact on the performance of the blade, which need to be analyzed and discussed in the design.
The low-speed and small-radius blades are aimed at the sub-low speed tidal current area. The thin-walled airfoil is the best choice for this kind of application. We analyze the force and hydrodynamic characteristics of the blade, establish force balance equation by introducing the impact force, and consider the effect of the velocity on the blade design in order to maximize the efficiency of the whole energy capture cycle in tidal current (Batten et al., 2006).

Establishment of blade impact model
According to the theory of Finnemore, as shown in Fig. 4, for the incompressible steady flow, the impact force produced by the flow on each micro-segment of the blade is as follows: is the outlet velocity, is the inlet velocity, P is the fluid density, and Q is the flow of the swept area of the impeller. The impact force is decomposed into the axial and tangential directions, and for each micro-segment of the blade, the torque received is as follows: Assuming that the axial inducing factor is a and the tangential inducing factor is b, the axial force can be approximately simplified as follows: ϕ where, is an inclination angle.
Torque can be reduced to the following formula:

Establishment of pressure differential model of blade
According to Bernoulli equation, if there is no interference between each blade element, the axial force and torque on each blade element can be calculated when the differential pressure force acts on the blade. The stress analysis of blade element is shown in Fig. 5.
According to the classical design theory (Huang et al., 2016), the axial force generated by the differential pressure of the flow around the wing can be expressed as follows: The tangential component of the differential pressure around the wing is as follows: The torque generated by the differential pressure around the wing is as follows: In the formula, is the coefficient of axial force and is the coefficient of tangential force, which can be calculated from lift coefficient and drag coefficient . (13)

Establishment of blade equilibrium equation
Considering the fluid impact force and pressure difference around the wing, the total equilibrium equation is established according to the torque and axial force functions respectively.
The torque balance equation considering impact force is as follows: The impact coefficient f is defined as the ratio of blade impact force to total force, which is determined by tidal current velocity and can be expressed as the following formula.
The coefficient of impact force reflects the consideration of impact force in blade design. When the tidal current  GAO Ru-jun et al. China Ocean Eng., 2020, Vol. 34, No. 3, P. 374-386 377 velocity is larger than 2 m/s, the pressure difference between impact force and winding wing is very small, and f is 0. When the flow velocity is smaller than 1 m/s, the impact force is the main source of the blade rotating torque, so the impact force should be considered to design the blade in low velocity sea area. If the design velocity is 1 m/s, f can be 0.6. According to BEM theory, the expression of total torque is obtained as follows: Eqs. (9), (12), (15), and (17) are combined to obtain the following formulas: The equation of axial force balance considering impact force can be written as follows: (19) According to BEM theory, the expression of total axial force is obtained as follows: Eqs. (9), (10), (19), and (20) are combined to obtain the following formulas: The total equilibrium equation obtained by the simultaneous Eqs. (18) and (21) is as follows: ( After introducing impact force, Eq. (22) is the relationship between the axial and tangential inducing factors of the blade, which means that we establish the restrictive condition function for transforming the axial velocity of the blade into the tangential velocity. Taking the equilibrium equation as the constraint condition of non-linear equality, the expressions of the axial and tangential inducing factors of the energy capture power coefficient are the objective functions. Through numerical iteration, the values of the axial and tangential inducing factors of the maximum energy capture power coefficient can be obtained, and then the twist angle and chord length of the blade can be determined

Design conditions of blades
The tidal current characteristics and the performance of power generation devices determine how much tidal current energy is captured. Thus, it is necessary to study the tidal current characteristics before designing blades to capture tidal current energy. As shown in Fig. 6, the experimental base for capturing tidal current energy is Yushan Islands. Starting from Shipu, Xiangshan, Zhejiang Province, about 25 nautical miles southeast to Yushan Islands, the southeast most island of Xiangshan. It shows that the tidal current period is about 12.42 hours after testing the tidal current characteristics of the sea area with profiled velocimeter. The average velocity is between 0.6 and 0.8 m/s, and the maximum velocity is 1.2 m/s, which belongs to the standard semidiurnal tide. It is measured values of the tidal current velocities of a high tide are shown in Fig. 7.
According to the measured tidal current velocity, the function of velocity and time is obtained as follows: V m where is the maximum tidal current velocity, with a value of 1.2 m/s, and T is the tidal period, which is about 12.42 hours. Eq. (24) gives the fitting curve of tidal current velocity in one day. However, due to the periodicity of tidal movement, the maximum tidal current velocity of different  days per month also varies. The velocity curves in the halfmonth cycle are drawn, as shown in Fig. 8. V m According to the daily variation of maximum tidal current velocity, as long as the coefficients are added before , the functional relationship between the current velocity and time at any time can be obtained.
is the monthly cycle of tidal current, which is 30 days.
is the number of days per month. K=0.21 is obtained by fitting the measured tidal current velocities in the waters of Yushan Islands. Eq. (25) gives the variation function of flow velocity at any time. The blades studied are mainly used in sub-low velocity sea areas. The design velocity is usually of the maximum velocity. Since the average daily maximum tidal current velocity in the waters of Yushan Islands is about 1.2 m/s, and 1.0 m/s is chosen as the designed velocity. According to GB/T 13981-1992, the corresponding relationship between the number of blades and the tip speed ratio of the horizontal axis turbine can be obtained. In this case, the number of blades is 6. Studying the hydrodynamic performance of 200 W small blades, the blade radius can be calculated by the following formula: If we consider that the energy capture power coefficient is 0.4, the design velocity is 1.0 m/s, and the density of sea water is 1.039 g/cm 3 , we can calculate the impeller radius. When the rated power is 200 W, the impeller diameter is about 1.1 m.
In summary, the basic design parameters of the blade are shown in Table 1.

Determination of key blade parameters
(1) Choose airfoil For airfoil to capture more impact force, it is necessary to make the airfoil have larger leading edge radius and trail-ing edge radius. However, the larger radius of the leading and trailing edges could also make the blade stall easily and the lift of differential pressure around the wing would decrease sharply. In order to ensure a certain lift coefficient, the radius of the leading and trailing edges should be taken as an intermediate value. To sum up, the designed airfoil is shown in Fig. 9.
(2) Choose the impact force coefficient to determine the equilibrium equation From Fig. 10, the impact coefficient corresponding to the design velocity of 1 m/s is 0.6. The impact coefficient is substituted into the equilibrium equation and simplified as follows: Calculate the best tip velocity ratio λ 0 = 2 The optimal spike ratio can be obtained by the optimum calculation method.
(4) Calculate the twist angle and chord length of blade by dividing blades into equal parts After removing hub blades, the total extension is 0.5 m. It is divided into 10 parts and has 11 sections. The distance between adjacent sections is 0.05 m.
According to the equilibrium Eq. (27), the genetic algorithm toolbox of MATLAB is used to solve the optimal   solution problem under the condition of non-linear constraints. The optimal axial inducing factor 'a' and tangen-tial inducing factor 'b' for each section are obtained, as shown in Table 2.
The formula of section inclination is as follows: The inclination diagrams of each section are shown in Fig. 11.
The chord length of each section can be calculated by the following formula: The distribution of the modified chord length is obtained as shown in Fig. 12. (5) Establish a three-dimensional model According to the distribution of the chord length and the dip angle obtained above, a three-dimensional model can be established as shown in Fig. 13.

Pressure analysis of CFD
A common solver Fluent is used. According to the turbine environment, the blade is designed as the hollow wall. The grid in the light blue area is the inlet of incoming flow velocity and is set to the velocity inlet. The grid in the red region is the outlet of the flow, set to the out flow, the rest of the boundary is the wall, and the outer surface of the grid in the rotating region is the interface (Lee et al., 2015;Edmunds et al., 2017).
The meshes as shown in Fig. 14 are imported into Fluent as airfoil.msh file, and the turbulence model is selected as the SSK-e model suitable for the flow field of hydraulic turbine. The steady-state and transient solver based on pressure form is adopted. The medium of the basin is water, the inlet velocity is 1 m/s, and the angular velocity of the rotating domain is 1.88 rad/s. The pressure distribution of the simulation results is shown in Fig. 15.     It can be seen from the figure that in the process of rotating the turbine blade faces a great positive pressure. The pressure of the leading edge is larger than that of the trailing edge. From the root to the tip, the pressure increases and the pressure range increases gradually. The negative pressure on the backflow surface is huge, and the negative pressure on the trailing edge is larger than that on the leading edge. The negative pressure mainly concentrates on the middle and root positions.

Experiments and results of blades capturing tidal current energy
Based on extensive research on the blade capturing tidal current energy and its corresponding experimental test system, a synthetic experimental test system for 2-kW blade capturing tidal current energy is designed. The experimental system can be used to simulate the tidal current motion below 2 m/s. The characteristics of energy capture power coefficient, torque, speed and load of the blades under different flow fields can be tested experimentally.
4.1 Design of experimental system and comparative experiment 4.1.1 Synthetic experimental platform The experimental system consists of experimental pool, transmission mechanism, measurement and control device. The overall scheme is shown in Fig. 16. The working principle is that the experimental pool produces a simulated tidal current at a certain speed under the control of the measuring and controlling device. The tidal current generates a moment through the blade to drive its rotation, and the kinetic energy is transmitted to the generator shaft through the transmission device. Generator creates voltage and current to drive the rear load. At the same time, the measurement and control system records various motion parameters of the transmission system and tests the hydrodynamic performance of the blade comprehensively. The whole experimental system is based on the Lshaped tank which is 70 m long and 4 m wide. The detailed design scheme is shown in Fig. 17.

Design of contrast blades
In order to verify the design theory of sub-low velocity and thin-walled blades, a horizontal-axis lift-type blade capturing tidal current energy is designed according to Glauert design method under the same conditions. The experimental results are compared with those of a new type of thin-   GAO Ru-jun et al. China Ocean Eng., 2020, Vol. 34, No. 3, P. 374-386 381 b walled blade designed in Section 2 of this paper. Considering that the lifting blade needs a larger lift coefficient, NACA6412 is chosen as the airfoil. The tip ratio is 2.0 based on the design power and flow velocity, and the number of blades is 6 based on the tip ratio. According to the total equilibrium equation , the optimal axial and tangential inducing factors of each section are calculated. The final inclination distribution of the blade is shown in Fig. 18 and the chord length distribution is shown in Fig. 19.
According to the chord length and torsion angle distribution of the blade calculated above, a three-dimensional model is built and the processing method is considered. The lift type blade airfoil is more complex and needs to be machined by milling machines, and the high strength nylon is selected as the material. The thickness of lift-impact blade airfoil is constant. The strength requirement and airfoil requirement can be met by using common steel plate. The final impeller can be assembled as shown in Fig. 20.
In this paper, the different hydrodynamic characteristics (Ouro et al., 2017) of the impeller blade are analyzed by measuring the energy capture power, speed and torque of the impeller blade under different flow velocities and loads. The total energy of the impeller blade in the whole energy capture period is calculated by discrete integral of the tidal current velocity time function in the waters of Yushan Islands, and the hydrodynamic performances of the two kinds of impellers are comprehensively compared and analyzed. The installation, construction and debugging of the whole system as well as the physical drawings are shown in Fig. 21.
The experimental design is as follows: (1) To measure the capture power by changing the tidal current velocity from 0.6 m/s to 1.5 m/s at 0.1 m/s interval; (2) To measure the torque of impeller by changing the current velocity from 0.6 m/s to 1.5 m/s at 0.1 m/s interval, and (3) To measure the impeller speed by changing the tidal current velocity from 0.6 m/s to 1.5 m/s at 0.1 m/s interval.
(4) To measure the impeller's energy capture power, torque and rotational speed by changing the load at intervals of 3 Ω from 1 Ω to 24 Ω.     Fig. 22a is a curve of the energy capture power coefficiency, which varies with the flow velocity. According to the measured results in the figure, the maximum energy capture power coefficients of the two blades are at the flow velocity of 1 m/s, that is to say, the maximum energy capture power coefficients are obtained when the inflow velocity is at the design flow velocity. The maximum energy capture power coefficient of the lift type blade is about 0.31, while that of the combined lift-impact blade designed in this paper is about 0.35, which is slightly better than the former. In addition, the energy capture power coefficient of the combined lift-impact blade is insensitive to the angle of attack. In wide flow velocity range, the energy capture power coefficient is higher, which indicates the high performance of the combined lift-impact blade in the low flow velocity range. Although the maximum energy capture power coefficient of lifting blade is not bad, the curve of energy capture power coefficient changing with the inflow velocity is steep. The average energy capture power coefficient of lifting blade in the whole low velocity range is not high, which suggests that this design is suitable for high-speed constant velocity energy capture, but not for time-varying low velocity energy capture. This is because when the velocity changes, the angle of attack varies greatly, which leads to the reduction of lift coefficient and the low efficiency of energy capture. Fig. 22b is a curve of energy capture power varying with inlet velocity. The measured energy capture power increases geometrically with the inlet velocity before reaching the design velocity, and the energy capture power of the lift-impact combined blade is 243 W when the design velocity is 1 m/s. And the maximum energy capture power is 476 W when the inflow velocity is 1.3 m/s. In comparison, the overall performance of lift type blades at low flow velocity is not good, and the capture power is 163 W at design speed (1 m/s), which is much lower than that of liftimpact combined blades, and the maximum energy capture power is 255 W when the inflow velocity is 1.3 m/s. This is because the power generation efficiency decreases sharply due to the blade arrest. The summary shows that the average power generation coefficient and power generation of the combined lift-impact blade in the flow velocity range of 0.6−1.5 m/s are larger than those of the lift type blade.

Torque characteristics
When we keep the load resistance at 20 Ω and change the velocity of incoming flow, the curve of impeller torque with the velocity of incoming flow is shown in Fig. 23. The impeller torque increases linearly with the inflow velocity when the inflow velocity of both blades does not reach the design velocity; after reaching the design velocity, the impeller torque gradually stabilizes. The maximum impeller torque of the lift-impact combined blade is 54 N·m when the inflow velocity is 1 m/s. When the velocity of lift type blade is 1.5 m/s, the maximum moment is 41 N·m. The torque of the combined lift and impact blade is much larger than that of the lift type blade, because the impact coefficient is introduced into the combined lift and impulse blade, and the impact force is taken into account. The designed blade can capture more impact force, and the chord length, density and tangential force are bigger, so the torque is bigger. In order to ensure good hydrodynamic performance (Bahaj et al., 2007), lift type blades usually have smaller chord length. The tangential force is smaller, so the torque is smaller. When the incoming flow velocity reaches the design flow velocity, the tangential force on the blade increases with the increase of the flow velocity, but the angle between the tangential direction of the blade and the circumferential direction of the rotation axis increases continuously, and most of the absorbed energy is converted into resistance. Therefore, the torque of the blade tends to be stable with the increase of the flow velocity.

Speed characteristics
When the load resistance is kept at 20 Ω but the incoming flow velocity changes, the impeller speed varies with the incoming flow velocity as shown in Fig. 24. The impeller speed of the lift-impact combined blade increases with the GAO Ru-jun et al. China Ocean Eng., 2020, Vol. 34, No. 3, P. 374-386 383 increase of the incoming flow velocity, and the increasing speed slows down when the velocity reaches 1.3 m/s. This is because before reaching the design flow velocity, the blade capture power increases gradually, and the square of the load loop voltage is proportional to the capture power, and the blade speed is proportional to the load loop voltage. Therefore, when the incoming flow velocity approaches the design flow velocity, the impeller speed increases geometrically with the incoming flow velocity. When the design velocity is reached, although the total energy flowing through the blade increases, the blade's energy trapping ability gradually decreases, and the impeller's torque tends to stabilize and then decreases. The front end of the speed curve of the lifting blade is similar to that of the lift-impact combined blade, but the lifting blade is more sensitive to the flow velocity. When the design flow velocity is not reached, the speed of the lifting blade rises rapidly with the inflow velocity and decreases sharply after exceeding the design flow velocity.

Load characteristics
When keeping the incoming flow velocity as the design flow velocity but changing the load resistance, the curve of energy capture power coefficient with load resistance is shown in Fig. 25. The load resistance corresponding to the maximum capturing power coefficient of the lift-impact combined blade and the lift type blade is 19 Ω. When the lift-impact combined blade is in non-optimal resistance, the energy capture power coefficient is slightly lower than that of the blade in optimal resistance, with little difference. However, the energy capture power coefficient of lift type blade under non-optimal resistance is much lower than that under optimal load resistance, which is consistent with CFD simulation. This is because the load resistance is directly related to the blade angle of attack. The optimal angle of attack is generally 6.10°. The closer the angle of attack is to this value, the higher the capture power coefficient is. Therefore, fixed pitch turbines need to control the load resistance to make the blades be operated in the best performance.
When we keep the incoming flow velocity at the designed flow velocity but change the load resistance, the impeller torque curve can be obtained as shown in Fig. 26. The impeller torque decreases gradually with the increase of load resistance. When the load resistance is 4 Ω, the impeller torque of the lift-impact combined blade is the greatest, which is 57 N·m. When the resistance is 24 Ω, the minimal torque is 48 N·m. The maximal impeller torque of lifting blade is 29 N·m when resistance is 1 Ω, and the minimal is 25 N·m when resistance is 24 Ω. These data suggest that the general trend is that the load resistance increases, the load acting on the impeller decreases, and thus the impeller's torque decreases.
When we keep the incoming flow velocity at the de-    signed flow velocity but change the load resistance, the obtained impeller speed curve is shown in Fig. 27. When the load resistance is 1 Ω, the speed of the lift-impact combined blade is the slowest, which is 0.45 r/s. When the load resistance is 24 Ω, the speed is the fastest, which is 0.82 r/s. When the load resistance is 1 Ω, the speed of lifting blade is the slowest, which is 0.89 r/s. When the load resistance is 24 Ω, the speed of lifting blade is the fastest, which is 1.60 r/s. Thus, the general trend is that the impeller speed increases with the increase of load resistance. This is because the force of flow on the blade remains unchanged, but the load decreases with the increase of the resistance of the rear stage, the resistance of the latter stage acting on the impeller decreases, and therefore the blade rotates faster.
In conclusion, we analyzed and discussed the power characteristics, torque characteristics, speed characteristics and load characteristics of the blade in detail through the experimental results. Generally speaking, the lift-impact combined blade has larger torque, slower speed, and larger energy capture power coefficient in a wide low velocity range, which is suitable for the capture of tidal current energy at low velocity. However, the lift type blade has smaller torque and faster speed. Only in the small range of the design velocity, the power factor of energy capture is higher, which is suitable for the capture of tidal current energy at constant velocity or high velocity. The later stage load has a great influence on the performance of these different blades. The optimal performance of the blades can only be achieved when the appropriate resistance value is obtained. The electronic control load box should be designed so that the optimal value of the load can be obtained continuously with the change of the flow velocity.

Conclusions
The design of small radius thin-walled blades under sublow speed tidal current is mainly studied. The reasons for the poor performance of common design theory in low speed area are analyzed, and a new theoretical method to design blades is proposed, which combines pressure differ-ence and impact force around wings. The hydrodynamic model and theoretical model of designing small radius horizontal blades under sub-low speed tidal current are also established. On the basis of this theoretical model, the horizontal axis thin-walled blade for capturing tidal current energy in the waters of Yushan Islands was optimized. With the combination of theory and practice, this paper discusses the actual hydrodynamic performance of the blade by using the dual evaluation method of CFD simulation and physical experimentation, which verifies the effectiveness of the design theory. In conclusion, we propose that: (1) The torque of the blade comes from two aspects: the pressure difference and the impact force around the wing. The pressure difference around the wing should be taken into consideration when designing the blade at high flow speed. The impact force should be considered when designing the blade at low flow speed, and the impact force coefficient should be introduced to represent the impact force, which provides the theoretical model to design blades under sub-low speed current.
(2) At low velocity, the maximal energy capture power coefficient of the lift-impact combined blade is similar to that of the lift type blade, but the average energy capture power coefficient in wide range of velocity is much higher than that of the lift type blade.
(3) The characteristics of lift-impact combined blade are low-speed and heavy-load, while the characteristics of lift type blade are high-speed and light-load.