The motivation for the development of the spillway turbine comes from the idea of having a turbine that can be placed in a stream without extensive civil construction and installation work and without significant disruption of the local ecosystem. The spillway turbine is intended to be an impulse turbine with the design inspired by the shape of Savonius hydrokinetic turbine shown in Figure 1 and on a Pelton wheel bucket, the flow velocity components of which are presented in Figure 2.

If considering frictional flows, the influences of Coriolis and centrifugal forces can be neglected comparing to the contributions of the frictional forces and the relative velocity is majorly affected by flow friction [13]. Similarly, the resulting impact force is significantly greater than Coriolis and centrifugal forces. The impact force appears due to the flow experiencing a sudden change in flow direction due to bucket curvature [14]. Hence, bucket design has a major influence on the performance of the turbine as positive work from forces should be maximised whilst minimising losses due to friction [15].

The spillway turbine is intended for use in manmade steeply-sloped channels such as chutes, spillway, etc. A spillway is a channel that ensures a safe transition of water from a reservoir to a stream [16] and the flow over a spillway can be classified as a super critical shallow flow. Examples of existing spillways include the Alqueva Dam spillway in Portugal [17] and Llyn Brianne spillway in Wales [18], both shown in Figure 3. Spillways are a common occurrence in reservoir and dam schemes.

If the spillway is large in length, it would be possible to place two or more turbines in series. This would maximise the usage of flow power and therefore result in higher energy extraction. The design scaling and adjustments would be recommended for each application site individually.

The runner was designed in such a way that different design variations can be tested with minimal structural changes. The basic runner design with no inserts is shown in Figure 4.

The runner’s shaft is a stainless steel, 0.02 m diameter rod. The acrylic disks, 0.01 m thick were placed on both sides of the runner to protect it from side friction and to deliver structural stability to the system, while allowing to observe from the outside how the water interacts with the blades. The disks placed on the runner have a diameter of 0.2 m and are attached to the runner with 24 socket head cap screws on each side. The runner has 6 blades is 0.3 m wide with a diameter of 0.2 m. The basic runner design is then modified by implanting 2 cup or 4 cup inserts into the runner. Detailed CAD drawings of both inserts are shown in Figure 5 and Figure 6.

The runner and the cup inserts are made of polyamide powder with a 3D printing technique called selective laser sintering (SLS). This production technique is based on a laser bringing powder material close to its boiling point and the heating of the powder particles results in formation of a solid shape. Synthetic laser sintering allows for quick and cheap production of the runner and the “cup inserts” and therefore it is chosen as a production method. Alternatively, the runner could have been manufactured from CNC machining of a solid block of material (e.g. aluminium or high-performance plastic), most likely the method of choice for real-world installations of the system.

The purpose of the accelerator channel(s) is to direct high-speed flow towards the runner. The accelerator channel consists of a 0.015 m thick stainless steel insert and acrylic sides. After analysing lower and higher thicknesses for the steel section in the initial stages of testing, it was concluded that the thickness of 0.015 m should be adopted and taken forward. Three different wedge designs were tested with the same thickness stainless steel inserts and are given below. The lengths of all wedges were adjusted to fit the laboratory spillway but they can be further elongated to fit specific sites. With the goal to test the influence of the accelerator channel width on the turbine’s performance, the runner with 2 cup inserts was tested with a 0.15 m wide accelerator channel (A) shown in Figure 7 as well as a 0.126 m wide accelerator channel (B) shown in Figure 8. Channels A and B direct the flow towards the blades in the same manner as shown in Figure 7. The only notable difference between these two channels is the width. Whilst channel A’s width-to-runner-ratio is 1:2 and channel B’s width-to-runner-ratio is 1:2.4.

The runner with 4 cup inserts is tested with the channel (C) shown in Figure 9. The 4 cup insert diameter is 0.075 m, same as the width of channel C. The runner with no cup inserts is also tested with the accelerator channel A.

A fully contracted rectangular weir is placed 10 m downstream from the runner. It is used to measure the discharge of the flow. The weir is designed [22] with dimensions in millimetres shown in Figure 11. The discharge can be calculated using Equation (1) as follows [22]:

Q = 1.84 ( L − 0.2 h w e i r ) ( h w e i r ) 3 / 2 (1)

where h_weir represents the height of water above the weir opening in meters, Q (m/s³) is discharge and L (m) is the width of the weir, which is 0.5 m.

Data is logged to calculate the power of flow going into the system and the power produced by the turbine. The coefficient of power, or turbine efficiency is calculated by using Equation (2), which reads:

C p = P o u t / P i n (2)

where P_in is the power of flow and P_out is the power produced by the turbine. P_in is obtained from:

P i n = ρ g H Q (3)

with density of water (ρ) is ρ = 1000 m³/kg, gravitational constant (g) is 9.81 m²/s, H is the head measured in meters and Q is discharge obtained from the weir equation.

Torque (T) s measured in Nm and rotational velocity (ω) in rad/s and hence P_out can be calculated from:

P o u t = T ω (4)

The specific speed N_s is calculated using Equation (5):

N s = N P o u t / H 5 / 4 (5)

where N is rotational speed in rev/min, P_out is power produced by the turbine in kW and H is effective head in m. The speed factor φ is calculated using Equation (6) where D is the diameter of the runner in m and N and H are the same as in Equation (5).

φ = D N / 84.6 H 0.5 (6)

For the power out (4) calculation, torque and the rotational velocity are measured with the torque transducer Futek TRS605.

The turbine laboratory set up consists of 4 main components which are shown in Figure 12 and are summarised below. The bridge, the drive train, the generator and the supporting structure components are kept the same throughout all tests. A movable bridge is used to support the drive train and the generator. The belt is tensioned by moving the bridge horizontally away from the runner in order to avoid belt slipping.

The drive train consists of 5 components as shown in Figure 13. The role of the drivetrain and generator are to transfer and convert mechanical rotational energy at the runner to electrical energy.

For the runner, a 20 mm stainless steel shaft is used with a 30 teeth pulley (4) placed at the end. A similar pulley is mounted at the end of the drive shaft (3). Both pulleys are connected with a Continental HTD 1500-5M timing belt (5) which transfers the rotational mechanical energy from the runner to the drive train. The torque transducer (2) is placed on the main shaft and is used for torque and rotational velocity measurements. At the very end of the main drive train shaft, a low speed AC generator (1) is mounted which converts the turbine system’s mechanical energy into electrical energy. The entire process from the spinning of the runner to electricity generationis depicted in Figure 14.

Runner with 2 cup inserts (Runner 2C) is tested with accelerator channels A and B. The results in terms of power coefficient as a function of torque produced of Runner 2C with channel A are shown in Figure 16 and Figure 17, and these show how changing the “x” and “y” geometric parameters affect turbine performance. Maximum efficiencies for power curves shown in Figure 18 are summarised in Table 2. The channel bed distance was increased to 0.385 m for results shown in Figure 19 and maximum efficiencies achieved for this setup are summarised in Table 3.

The runner performs best when it is placed 0.02 m from the wedge and 0.375 m from the channel bed achieving efficiency of 43.4%. The efficiencies drop when the runner is moved away from the wedge in x direction and the channel bed in y direction, as it can be observed from Table 2 and Table 3.

This observation is also made when the runner with 2 semi-circular cup inserts is tested with channel B, which is narrower than channel A. Channel A and B test results are plotted in Figure 18. The runner is placed 0.02 m from the wedge in x direction and 0.375 m from the channel bed in y direction for both cases, as this variable set up resulted in best turbine performance.

The maximum efficiency achieved with the 0.15 m wide channel (A) is 43.4% and the maximum efficiency attained with the 0.126 m wide channel (B) is 34.9%. Therefore, it can be noted that runner 2C performs better with a wider channel than for a narrower channel, suggesting a channel-width-to-runner-width-ratio of 1:2 for the final design of the turbine.

Runner 4C with the accelerator channel C is tested with different “x” and “y” geometric parameters shown in Figure 15. A photo of this setup is shown in Figure 19.

The measured power curves for the 4C runner with different “x” and “y” parameters are presented in Figure 20 and Figure 21. The power curves are shown in Figure 20, with a channel bed distance of 0.375 m and maximum power coefficients are given in Table 4. The channel bed distance is then increased to 0.385 m and the power curves are presented in Figure 21 and maximum efficiencies achieved for this setup are summarised in Table 5.

The runner with 4 cup semi-circular inserts performed best when it was placed 0.02 m from the wedge and 0.375 m from the channel bed achieving a maximum power coefficient (or efficiency) of 34.36%. The efficiencies dropped when the runner was moved away from the wedge in “x” direction and the channel bed in “y” direction, as it can be observed from Table 4 and Table 5.

Finally, the runner with no cup inserts is tested with the accelerator channel A in order to quantify the effect of the cup inserts on turbine performance.The peak power coefficient achieved with different “x” and “y” distances are summarized in Table 6.

From previously presented results it is noted that all runners performed best when placed 0.02 m horizontally from the wedge and 0.375 m from the channel bed. Hence, these parameters are used in the performance comparison plot of runner 2C, runner 4C and runner with no inserts which is presented in Figure 22.

It can be noted that the runner with 2 cup semi-circular inserts and 0.15 m wide channel A showed best performance from all setups tested. It is noteworthy that the turbine’s performance is similar to recently investigated horizontal [23] [24] and vertical [19] [20] [21] axis hydrokinetic turbines and hence offers an opportunity to harness very shallow high-speed flows which were previously deemed unsuitable for hydrokinetic energy exploitation.

The specific speed and speed factors are calculated using Equations (5) and (6) respectively. The calculated values for peak power coefficients are given in Table 7. The speed factors against the power coefficient for all three runners are depicted in Figure 23.

From Figure 23 and Table 7 it can be noted that the runners with cup inserts have higher specific speeds and speed factors than the runner with no cup inserts. Hence, the idea of achieving similar water blade interaction as it occurs in Pelton blades shown in Figure 2 and Figure 3 is considered validated. As the speed factors fall into 0.43 - 0.48 range of impulse turbine speed factors [11], it can be concluded that the spillway turbine with 2 cup inserts can be classified as an impulse turbine. The specific speed [25] range shown in Table 7 also confirms that the spillway turbine can be defined as an impulse turbine.