- Email: [email protected]
The effect of inlet design on the flow within a combined waves and current flumes, test tank and basins
Coastal Engineering 95 (2015) 117–129
Contents lists available at ScienceDirect
Coastal Engineering journal homepage: www.elsevier.com/locate/coastaleng
The effect of inlet design on the flow within a combined waves and current flumes, test tank and basins Adam Robinson ⁎, David Ingram, Ian Bryden, Tom Bruce Institute for Energy Systems, School of Engineering, The University of Edinburgh, Kings Buildings, Mayfield Rd, Edinburgh EH9 3JL, United Kingdom
a r t i c l e
i n f o
Article history: Received 30 April 2014 Received in revised form 15 October 2014 Accepted 16 October 2014 Available online xxxx Keywords: Current Wave Flume Test Tank Basin
a b s t r a c t The motion of the sea, through waves and currents, represents a large source of clean and safe energy. However, any structure built to operate in the sea will experience large varying forces and a difficult environment. It is therefore crucial to develop realistic and repeatable sea-like conditions in a laboratory in order to lower the cost and risk of developing off-shore structures. Building on previous efforts, an experimentally validated numerical model is used to predict the current-only flow in flumes capable of combining waves and current. This model is then used to simulate the flows within common flume configurations and within a new concept known as the “isolating inlet flume”. The results of these simulations are then analysed to assess the performance of each flume type and to understand the fluid dynamics that govern each type. Flume performance is found to be largely determined by the creation and dissipation of shear layers. The tests proved that a flume using the isolating inlet requires significantly less downstream length to achieve a developed flow and acceptable turbulence levels than the previous flume configurations. The isolating inlet has the additional benefit of creating a still zone where a conventional wave-maker might be used. Further simulations are used to investigate the design of the isolating inlet flume and demonstrate how it works. This paper should be of use to scientists and engineers seeking to design flumes, test tanks and basins that create sea-like test conditions, thus improving the scope and range of laboratory testing. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Flumes and test tanks that combine waves and currents to produce realistic two and three-dimensional seas provide an opportunity to test scaled off-shore devices and structures, reducing the cost and risk of development. To provide this testing capability several configurations of Combined Wave and Current (CWC) test tanks and flumes have been developed. The current is usually driven either by a pump or by an impeller and then guided into the test tank. The waves are made using an oscillating paddle or piston. The wave-making paddles also have the task of reabsorbing the waves to remove unwanted reflections from the tank (Salter, 1981). To date the production of the current compromises the wave-field and vice versa once a certain flow speed is reached. This means that it has not been possible to create sea-like test conditions representative of important potential tidal power generation sites like the Pentland Firth in terms of wave-field, current speed and turbulence level. Here a numerical model based on the Reynolds Averaged Navier– Stokes (RANS) equations is developed to aid flume design and assess current-only performance of CWC flumes, tanks and basins. The method is then used to investigate the flow within various configurations of ⁎ Corresponding author. E-mail address: [email protected] (A. Robinson).
http://dx.doi.org/10.1016/j.coastaleng.2014.10.004 0378-3839/© 2014 Elsevier B.V. All rights reserved.
flume that have been designed to include wave-making and current creation simultaneously. Here the current is assessed in isolation of the waves with the effect of waves on performance not assessed. All of the flume designs tested are based around a re-circulation principle where flow moves across the test section before it is drawn below a floor, then accelerated with an impeller or pump to be reinjected at the start of the test section (Fig. 5). Test tanks exist where current is provided by either jets supported on frames or by arrays of propellers in the test areas of the flume or tank. These configurations are useful in some cases but cannot achieve test conditions as searepresentative as a recirculating tank. Here four configurations of re-circulating CWC flume are tested at the same scale and operating velocity to allow comparison. The first is up-welling paired with a conventional flap wave-maker where the current enters the tank vertically directly in front of the wave-maker (Fig. 1). The second method of combining currents and waves could be described as undershot where the current is inlet horizontally below the wave-maker (Fig. 2). The third method of combining currents and waves works by using a shaped current guiding wave-maker to produce a racetrack-shaped flume (Fig. 3). The final method tested is the ‘isolating inlet flume’. This new inlet design introduces current into the tank in a way that creates a turbulent
118
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 1. Up-well type flume sliced along the centre line.
mixing layer which divides the area in front of the wave-maker and the fast moving flow entering the flume, minimising energy transfer, as shown in Fig. 4. Once the waves reach the dividing shear layer they pass through this boundary and are imposed on the current stream. The wavelength should then elongate in a consistent way dependent on the current velocity and direction. Fig. 5 is a diagram showing this inlet system implemented on a 3D CWC test tank. In the test region of a 3D tank (Fig. 5) where arrays of devices may be tested it is critical that the flow has developed a consistent velocity profile. Only changes in the velocity profile due to the upstream device, not flow development, should be experienced by a downstream device. This also determines how much of a 3D tank can be used for array testing and how large the tank needs to be to provide a given test area. The distance the flow takes to develop for a given flume configuration is a critical performance criterion and will be assessed here using a validated numerical model. The geometry of the isolating inlet is critical to its performance, in particular the inlet angle, the effect of which will be investigated here through simulations. This investigation aims to improve the physical understanding of what determines flume, tank and basin performance. The numerical model and functional explanations will provide a means of creating a test tank where more realistic representations of the tidal channels in terms of current speed and wave-field can be produced. Thus lowering the cost and risk of developing off-shore devices and structures. This paper will begin with a review of existing CWC tanks and flumes (Section 2). Following this will be a description of a numerical model based on the RANS method useful for flume design (Section 3.1). A description of the experiment used to validate the experiment is available in Section 3.2. In Section 4.1 these experimental results will then be used to validate this method. The numerical model is used to assess the
performance of different common flume configurations in Section 4.2. Section 5 will summarise the findings of this paper. 2. Background 2.1. Wave/current interaction in nature In nature waves and currents combine in areas of the sea such as tidal channels and costal zones as the tides come in and out. To an engineer the combination of waves and current is of interest because it leads to increased stresses and wear on off-shore structures (Salter, 2003; Wolf and Prandle, 1999). Current affects waves in the following ways: • Wavelength increases when the current and waves are in the same direction and decreases when they move in opposing directions. (Jonsson, 1990; Wolf and Prandle, 1999). • Currents reduce or increase wind shear on the surface affecting wave shape (Wolf and Prandle, 1999). • Waves turn to the current direction over time (Wolf and Prandle, 1999). Waves affect currents in the following ways: • The presence of waves can increase turbulent intensity of the bulk flow (Kemp and Simons, 1982). • Increase wind drag on surface due to waves alters the velocity profile (Janssen, 1989). • Waves increase friction on the sea bottom (Wolf and Prandle, 1999). Most of the effects described above do not influence the behaviour in a combined wave and current test tank. If the waves and currents combine in a similar way in a flume as they do in a tidal channel the wavelength should elongate and contract in a predictable way.
Fig. 2. Undershot type flume sliced along the centre line.
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
119
Fig. 3. Shaped wave-maker type flume.
2.2. Turbulent mixing layers Turbulent mixing layers are central to the new method of combining waves and current described in this paper. They are a Kelvin–Helmholtz instability that develops after the point where the flows are combined resulting in a 2D span-wise series of vortical structures known as rollers (Loucks and Wallace, 2012): see Fig. 6. More details on turbulent mixing layers relevant to combined current and wave tanks can be found in a related paper (Robinson et al., Submitted for publication). 2.3. Wave and current test tank design Although many tanks exist where waves and currents can combine none have the capability to produce 3D waves with 3D currents. The use of curved and round wave-tanks with wave-makers around the perimeter is well-developed and is described. (Minoura et al., 2009; Naito, 2006; Taylor et al., 2003). The work on 3D currents is limited to one
source (Salter, 2001). The design that could combine 3D waves and currents was only implemented on a scale model only capable of creating 3D currents (Salter, 2001, 2003; Taylor et al., 2003). This method generates a 3D current by rotating aerofoils around the perimeter of a tank and works on the same principal as a Voith–Schneider Propeller. Testing the scaled model proved that the design generally provides a robust means of producing a 3D current (Salter, 2003). Wave-tanks exist where current is provided inside the test tank by ether jets supported on frames in front of the wave-makers or by arrays of propellers (Kirkegaard, 2013). Although useful in some cases it is difficult to achieve sea-like velocity profiles and turbulence levels within a reasonable downstream distance. It is possible to remove turbulence and condition the flow in a flume or tank using screens and honeycomb (Miller, 1990) although placement is difficult where wave-makers are present. From surveying existing CWC tanks and flume designs it is apparent that there are three existing methods that could be used to make 3D currents in the presence of a 3D wave-field. The first is up-welling
Fig. 4. Method of isolating the generation of wave and current.
Fig. 5. A 3D CWC test tank sliced along the centre line.
120
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 8. Example mesh (Each line is approximately representative of 10 in the mesh used).
Fig. 6. Turbulent mixing layer diagram (Loucks and Wallace, 2012).
where the current enters the tank vertically directly in front of a conventional flap wave-maker (Fig. 1). With an up-well inlet the flow has to turn from vertical to horizontal before moving down the flume. This lifts the free-surface and may limit the flow speed for a given depth. The up-welling inlet is the easiest to implement in an existing wave-only flume and therefore many examples exist (Yao and Wu, 2005). The second method of combining currents and waves could be described as undershot where the current is inlet below the wave-maker (Fig. 2). Examples of this type of flume in use can be found from many sources (Ifemer, 2010; Kemp and Simons, 1982). The undershot configuration is liable to produce turbulent eddies after the slitter plate above a certain flow speed (Robinson et al., Submitted for publication). This configuration is the easiest to implement in an existing current-only flume. The final method of combining currents and waves uses a shaped current-guiding wave-maker (Fig. 3). This method was proposed but has never been implemented in a CWC tank suitable for testing (Salter, 2001, 2003). The shape of the flume known as a racetrack is proven to not add any turbulence to the flow and is used in many applications where a high speed/low turbulence flow is required (Ifemer, 2006). When the wave-maker is used on a flume of this configuration it will affect and be loaded by the current complicating generation and absorption. 2.4. Simulations combining waves and current Many useful CWC tank simulations have been created (Buchmann et al., 1998; Ortloff and Krafft, 1997) These models were produced to test ocean structures and do not include direct modelling of wave and current generation. Waves and currents are more easily included in simulations by prescribing boundary conditions. Numerical CWC tanks have been successfully created using a RANS-based method (Ortloff and Krafft, 1997). Only one example of absorbing wave-makers being directly modelled exists (Maguire, 2011). This method uses the flow3D multi-phase RANS code (Hirt, 1988) to simulate the behaviour of a wave-maker both producing and absorbing waves. A RANS-based method has also been proven for simulating the turbulent mixing layers usually present in CWC tanks (Robinson et al., Submitted for
Fig. 7. Contours of volume fraction for a 2D slice through a 2 m up-welling flume simulation with a free-surface present (black represents water and white is air).
publication). These works combined demonstrate that all the capabilities required to directly simulate a 3D combined current and wave tank are feasible within a RANS code.
2.5. Numerical model A computational model was created using the CFD code FLUENT (Fluent, 2006) to investigate the performance of various flume configurations. The model was first validated against experimental data in Section 4.1 then used to investigate the performance of various flume types in Section 4.2. The computational methodology was developed and validated specifically for accurate representation of flows containing turbulent mixing layers (Robinson et al., Submitted for publication). Although waves and moving wave-maker are not included in this simulation the representation of moving geometry is available in RANS-based codes (Hirt, 1988) and therefore could be included in the model described here. The approach is based on solving the RANS equations in an Eulerian fixed grid implemented within FLUENT. The k − ε model RNG turbulence model (Yakhot et al., 1992) is used for closure. Further detail of the method used can be obtained from a previous paper (Robinson et al., 2014b). The flow is simulated in 2D with the boundary conditions providing the complexity of the flow. The geometry is a rectangle of a given flume height that stretches in the flow direction far enough to ensure sufficient flow development and to ensure that the outlet boundary condition has a negligible effect on the flow in the area of interest. The validity of the 2D assumption is tested by measuring velocity profiles across the flume at various points. The velocity at the centre of the flume was found to be within 1% of the average velocity of a profile measured from the centre to the wall of experimental flume (Section 3.2) in the y direction.
Table 1 Water properties. Density (kg/m3)
Viscosity (kg/m-s)
Temperature (°C)
998.2
1.003 × 10−4
20 °C
Table 2 Mesh independence comparison between maximum velocity present in the velocity profile at various x locations for MI1 and MI2. X position (m)
Max × velocity (m/s) MI1
Max × velocity (m/s) MI2
Difference %
3 5 7.5 10 15 20 25 30 35 40
1.853 1.513 1.197 0.976 0.863 0.846 0.843 0.844 0.847 0.851
1.849 1.511 1.208 0.992 0.872 0.851 0.846 0.846 0.849 0.852
0.216 0.201 0.096 0.984 1.654 0.967 0.574 0.390 0.250 0.154
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
121
Fig. 9. A 2D stream-wise slice of the ECWC flume. The flume has a uniform cross-section in the y direction (400 mm wide). The datum used throughout is at the intersection of the wavemaker face at flume floor.
Fig. 10. Vector and contour map on the xz plane from x = −100 to 800 and z = 10 to 360 mm compiled with experimental data for Case 1A. Velocity is proportional to vector length and contour colour (Key given in figure).
Fig. 11. Vector and contour map on the xz plane from x = −100 to 800 and z = 0 to 400 mm predicted using a numerical model for Case 1A. Velocity is proportional to vector length and contour colour (Key given in figure).
122
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 12. Mean velocity versus distance from the bottom wall for various X positions for Case 1B. The dashed line is the simulated result, black is the experimental measurements.
Fig. 13. Mean velocity versus distance from the bottom wall for various X positions for Case 1C. The dashed line is the simulated result, black is the experimental measurements.
The flume inlet is described by a boundary condition and is set by prescribing node values for u and v velocities, turbulent kinetic energy, k and turbulent dissipation rates, and ε to match those recorded at the corresponding location in the experiment. The bottom boundary is set as a wall with a roughness matching the experiment (roughness height 0.0015 mm, roughness coefficient 0.5). The downstream boundary condition is set as a Neumann pressure boundary. The boundary condition configuration is shown in Fig. 8. The simulation of the free-surface in the flume represents a large increase in the computational cost of using this model and may not be necessary to accurately predict the downstream velocity profile development. To investigate this the method was extended to include the VOF multiphase model (Hirt and Nichols, 1981). The implementation used here is detailed in an earlier paper (Robinson et al., 2010a,b). The flume configuration that would have the most effect on the freesurface is the up-welling type (Fig. 1), where the inlet flow is directed vertically. To test whether it is necessary to model the free surface a simulation was run of an up-welling flume with a 2 m water depth and an average flume velocity of 0.8 m/s matching Case 2A. The result is shown in Fig. 7. In the simulation the free-surface was only lifted by 0.11 m with no significant change to velocity profile development. When an isolating inlet flume was tested (25° inlet angle) with a free-surface, no deformation was detected.
Fig. 14. Experimental image in front of a fixed wave-maker for the flume described in Section 3.1 in up-welling mode (0.6 m/s depth averaged flume velocity, 0.4 m water depth).
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
123
Fig. 15. Up-welling simulation configuration.
Having found that modelling the free-surface made little difference to the result and added greatly to the computational cost it was substituted with a free-slip wall boundary condition in all other tests. The physical properties of the water are set within the models and are based on conditions recorded in the experiments. The values used in the computations were found by using the average experimental temperatures and tables from the literature (White, 1994) and are given in Table 1. The mesh created for the computational model is a structured hexahedral mesh with a higher concentration of cells at the fluid inlet and boundary layer. The basic mesh structure is shown in Fig. 8. In the stream-wise direction the mesh expands at a rate of 1.05 from the inlet boundary to the outlet boundary. Vertically the mesh expands at a rate of 1.05 to the upper boundary from the lower boundary to properly resolve the boundary layer on the bottom wall. This mesh strategy is used throughout. A detailed mesh independence study of relevance to this case was in a previous paper (Robinson et al., 2014b), and the strategy and densities found were transferred to the simulations done here. To ensure that the solutions were independent of the number and location of the cells two meshes were created: MI1 and MI2 with identical geometry and identical boundary conditions. MI2 had four times the number of cells as MI1. For each mesh the downstream velocity profiles were recorded and analysed. Table 2 shows a comparison between the maximum u velocity within the velocity profile at various x locations downstream of the wave-maker face (x = 0). When the results from MI1 and MI2 were compared the average difference was always less than 2%, proving MI1 is sufficiently meshing independent. The compute time for the case with 29,204 cells (Case 2E) to develop from a state where the velocity is equal to zero in every cell to steady state flow seen in Fig. 23 was 3300 s. This simulation was carried out on a four-core 2.93 GHz Intel i7 870 processor with equation residuals converged to 1 × 10−7. The experiment used to validate the numerical model is described in the following section.
Fig. 17. Velocity profiles at various positions downstream of the wave-maker face (x = 0 m) for an upwelling flume simulation (Case 2A). Origin location for x and z positions is given in Fig. 9.
2.6. Validation experiment In order to validate the numerical model a flume with a water depth of 0.4 m and a length of 8 m (wave-maker to wave-maker) was used. The experiment is an identical scaled representation and 2D slice of what would be used in a 3D test tank (Fig. 5). The resulting flume is shown in Fig. 9. The flume has absorbing wave-makers at both ends however they are locked in the vertical position throughout these tests. The water is typically driven in the direction shown in Fig. 9 to allow the maximum distance for the propeller wake to develop and turbulence to dissipate. The flume walls are constructed with glass to allow optical access. 2.6.1. Measurement setup Point measurements of velocity in components u, v and w were taken using a Vectrino + ADV (Nortek-AS, n./d.). These measurements were used to calculate mean velocity and turbulent intensity. The flume is designed to maintain a steady state flow once it has accelerated from rest and therefore mean velocity measurements taken from large numbers of samples are suitable for this characterisation. Checks are built into the measurement processing to ensure that no transient effects caused by flume operation, start-up or large turbulent structures are present in the velocity measurements taken. Initially a sample of 360,000 measurements over 30 min (velocity setting 1C, Table 4) was taken to detect flume transience. By sub-dividing this into minute blocks, then comparing average velocities of each, it was found that 8 min (96,000 samples) were required to ensure the error in average velocity due to transience that was less than 1%. A further test was built into the measurement processing that compared the average velocity
Fig. 16. Vector and contour map around the inlet for an up-welling flume simulation (Case 2A). Velocity is proportional to vector length and contour colour (Key given in figure).
124
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 18. Undershot flume simulation configuration.
of the first 48,000 samples of every point measurement against the second 48,000 samples. This test ensured that the difference in average velocity between each half never exceeded 1% for the measurements presented in this paper. This high number of velocity samples is statistically sufficient to allow turbulence characterisation by Reynolds stresses or turbulent kinetic energy (Chanson et al., 2007). Along with velocity, the ADV derives two other quantities which relate to the quality of each measurement being taken. For every sample signal to noise ratio (SNR) and correlation are given. Signal to noise ratio was maintained above fifteen throughout this experiment in line with previous work (Cea et al., 2007; Lohrmann et al., 1994; Rusello et al., 2006). Correlation was used to filter the data from the ADV throughout (Martin et al., 2002; Rusello et al., 2006). Previous investigators have recommended 70% as a minimum threshold value; however 75% was used here (Martin et al., 2002; Rusello et al., 2006). The correlation value decreases as the turbulence increases (Robinson et al., 2014a). By extensive testing of ADV settings it was possible to achieve average correlation values exceeding 95% for all measurements in the working section of the flume (Section 4.1). In the more turbulent inlet and still zones (Sections 4.2 and 4.3) average correlation was always above 85%. These high correlation values ensured that at least 95,000 of 96,000 samples remained after filtering for working section and 75,000 of 96,000 for the inlet zone. This ensured accurate average velocity values and turbulence characterisation. To achieve this great care was taken to find ADV settings that would lead to the highest possible correlation values and smooth consistent velocity profiles. Typically a sampling rate of 50 Hz is used (Cea et al., 2007; Chanson et al., 2007; Martin et al., 2002; Robinson et al., 2014a; Rusello et al., 2006); however better results could be achieved at 200 Hz. The ADV measures an approximately 6 mm diameter cylindrical volume with a variable length. For the measurements in the working section of the flume (Section 4.1) a large measurement volume (9.2 mm) and a 2.4 mm acoustic pulse length were found to give the highest correlation values and give the smoothest and most consistent velocity profiles. In the inlet zone where the flow was more turbulent the acceptable correlation levels could only be achieved by minimising the size of the measurement volume (3 mm) and the length of the acoustic pulses (0.3 mm). The measurements in the inlet zone were at the limit of the ADV capability and therefore an improved ADV or alternative technique would be
Fig. 20. Velocity profiles at various positions downstream of the wave-maker face (x = 0 m) for an undershot flume simulation (Case 2B). Origin locations for x and z positions are given in Fig. 9.
required to achieve accurate measurements beyond an average flume velocity of 0.6 m/s (Table 4). The power level was set at the highest level possible to maximise the signal to noise ratio and correlation values. The acoustic streaming effect exacerbated by high power levels should be negligible for the velocities measured in this flume (Poindexter et al., 2011). Seeding was provided by neutrally buoyant hollow glass spheres with a mean diameter of 11.7 μm. By adding seeding until the correlation value stabilised the correct density was found. A good general overview of Acoustic Doppler Velocimetry is provided by Lohrmann et al. (Lohrmann et al., 1994) with further information on the specific ADV used here available from Nortek (Nortek-AS, n./d.).
2.6.2. Experimental uncertainty The geometrical tolerance of the experimental rig was less than ±1 mm with the measurement setup manufactured to ±0.1 mm. The x and z positions given here were measured with tapes and therefore have an accuracy of at least ± 1 mm. The y position is measured by the ADV and is accurate to 0.1 mm (Nortek-AS, n./d.). Immediately before the experiments were reported here the ADV was calibrated in a towing tank and was found to have an accuracy exceeding ±0.5% for velocity measurements. Using the error level found during the instrument calibration (±0.5%) and the potential errors in position measurement it is possible to assess maximum total error in various areas of the flume. The maximum combined position error in the x and z directions is ± 2 mm. By multiplying this by the maximum velocity change per mm recorded by the ADV whilst characterising the flume the maximum error can be found. The highest velocity gradients are found in the inlet area and reduce the further downstream as the water travels. The maximum error in velocity for the experiments for various flume areas is given in Table 3.
Fig. 19. Vector and contour map around the inlet for an undershot flume simulation (Case 2B). Velocity is proportional to vector length and contour colour (Key given in figure).
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
125
Fig. 21. Isolating inlet flume simulation configuration.
Table 3 the first column, maximum velocity gradient, this is found by analysing the ADV data. The second column, maximum velocity error due to position error %, is calculated by multiplying the maximum gradient by the maximum position error (2 mm). The fourth column combines the error in velocity due to positional inaccuracy with velocity measurement error of the ADV found during calibration. Temperature was measured using a thermocouple mounted inside the ADV with a quoted accuracy of ±0.1 °C (Nortek-AS, n./d.). The temperature of the flume was stable during the measurement campaign with the minimum temperature measured during the tests of 19.1 °C and a maximum of 21.2 °C. 3. The effect of inlet design on the flow within a flume To validate and check the numerical model and then investigate different flume configurations a series of tests were created. The details of these tests are described in Table 4. 3.1. Model validation To enable validation a series of measurements were taken in an isolating inlet flume (Fig. 9) at three different total flow rates. The numerical model described in Section 3 is set up to match the experiment described in Section 3.2. The initial test for the numerical model is its ability to accurately predict bulk flow patterns. In an isolating inlet flume the critical area for establishing how the water will move is the zone containing the wave-maker face and the inlet. The first assessment of the model will be its ability to predict how the fluid behaves in this area. The second assessment will be to compare how the flow develops as it moves downstream. In Fig. 10 a map of the flow in front of the wave-maker made using experimental measurements (ADV) is given. An equivalent flow map predicted using the numerical method described in Section 3.1 is shown in Fig. 11. By comparing the vector and contour maps for the numerical (Fig. 11) and experimental (Fig. 10) results it is clear the numerical model has accurately predicted the behaviour of the bulk flow. Most importantly the simulation has predicted the still zone in front of the
Fig. 23. Velocity profiles at various positions downstream of the wave-maker face (x = 0 m) for a 45° isolating inlet flume simulation (Case 2C). Origin location for x and z positions is given in Fig. 9.
wave-maker (x = −120 mm, Z = 0 mm to 400 mm). The low speed rotations above and below the inlet stream are present in the CFD but were undetected in the experiment, likely due to the limitations of the measurement technique. In the transition layer (0.2–0.5 m/s) thickness and expansion are comparable between the CFD and experimental result. The transition consists of a mixing layer made of complex turbulent structures (Robinson et al., Submitted for publication). The k − ε model used here to model turbulent behaviour does not directly model these structures and therefore there may have been inaccuracy in the simulation of the effects of the turbulent mixing layer. This result indicates the model's suitability. To ensure the method accurately predicts the development length in flumes velocity profiles were taken at various positions downstream of the inlet. A comparison of these velocity profiles for both numerical and experimental results for Cases 1B and 1C is shown in Fig. 12 and Fig. 13. The simulation predicts the shape of the velocity profile well, especially close to the inlet. The growing difference is likely due to the differences in behaviour between the outlet in the flume and the CFD boundary condition. Table 5 shows the average error in the prediction of point velocities in each velocity profile taken between the CFD and the experiment. With the numerical model able to predict the flow development in a flume to a reasonable degree of accuracy the model will now be used to predict and analyse the flow in common flume types. 3.2. Development length of various flume types Having developed and validated a numerical model for predicting the flow development in a flume the method will be used to investigate other flume types. The flumes will be simulated with a two metre water depth representative of the Flowave Test Basin, Fig. 5. The configurations tested are up-welling (Fig. 1), undershot (Fig. 2) and the isolation
Fig. 22. Vector and contour map around the inlet for a 25° isolating inlet flume simulation (Case 2E). Velocity is proportional to vector length and contour colour (Key given in figure).
126
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Table 3 Velocity measurement error due to position error (locations marked on Fig. 9). Area
Inlet (x = 0–460, Z = 15–360) x = 910 (z = 10–340 mm) x = 1910 (z = 10–340 mm) x = 4910 (z = 10–340 mm)
Maximum velocity gradient (m/s per mm)
Maximum velocity error due to position error %
Total maximum error %
0.020 0.015 0.005 0.001
6.7 2.1 0.8 0.3
7.2 2.6 1.3 0.8
type (Fig. 4). The shaped wave-maker type (Fig. 3) is not tested here as it represents the ideal flume in the current-only model and inlets fluid in a way that has the minimum effect on development length. For the isolation-type flume simulation several configurations are tested to assess the effect of inlet angle on development distance. The situations in this section are set up identically to the validation case described in Section 4.1 with the different flume configurations simulated by adjusting the positions and node values of the inlet boundary. Turbulent intensity at the inlet is set at 10% for all flumes with characteristic turbulent length scale set at 0.1 m due to the likely inlet guide vein spacing. 3.2.1. Up-welling flume The up-welling configuration shown in Fig. 1 directs the flow directly up at the free-surface in front of the wave-maker. This configuration results in significant energy loss due to free-surface lifting before the water is turned and flows down the channel. This free-surface disruption is significant and often unsteady and effectively leads to a wave-maker hinge depth that varies with current velocity (Fig. 14). Both these features may lead to difficulties in wave-maker control when absorbing and creating waves. In the up-scaled flume simulated here (0.8 m/s depth averaged flume velocity, 2 m water depth) the extra water depth suppresses the surface lifting to a negligible level (Section 3.1). A simulation of a 2 m depth up-welling flume was created using the method described in Section 3.1; the model setup is shown in Fig. 15. The resulting flow pattern around the inlet for the up-welling simulation is shown in Fig. 16. From the vector map above it is clear that the up-welling inlet results in a strong recirculation above the bottom wall which will act to increase the development distance. Velocity profiles at various positions down the length of the simulated flume are shown in Fig. 17. 3.2.2. Undershot flume The undershot configuration shown in Fig. 2 directs the flow horizontally underneath the wave-maker. A simulation of a 2 m depth upwelling flume was set up using the method described in Section 3.1; the flume model configuration is shown in Fig. 18. Water is inlet through a one metre vertical inlet at 1.6 m/s once the flow mixes
Table 5 Average % error in velocity profile for Case 1B and 1C. Case
X position (mm) 600
975
2000
5000
1B 1C
8.03% 5.43%
16.45% 13.02%
8.65% 9.61%
16.31% 16.72%
downstream. A 0.8 m/s depth averaged flume velocity should result in the 2 m deep flume. By passing a flowing stream under still fluid the conditions for a strong turbulent mixing layer are set up, resulting in the creation turbulent structures and a large recirculation flow above the inlet (Fig. 19). Velocity profiles at various positions down the length of the simulated flume are shown in Fig. 17. 3.2.3. Isolating inlet flume The isolating inlet flume aims to create a still zone in front of the wave-maker by introducing the fluid at an angle as shown in Fig. 4. The flume simulation configuration is shown in Fig. 21. Water is inlet through a two metre horizontal inlet at a speed that results in a 0.8 m/s depth averaged flume velocity in the two metre deep flume. Introducing the water at an angle results in the creation of a turbulent shear layer much like the undershot method; however the layer is limited by contact with the free surface, reducing turbulence production (Fig. 22). To assess the effect of inlet angle on development distance the simulation was set at three different angles: 45° (Fig. 23); 25° (Fig. 24); and 10° (Fig. 25). 3.2.4. Flume type discussion In this section the performance of each of the flume types is assessed and compared in the context of a current-only flume with a wavemaker present but not moving. Following this an effort will be made to understand why each flume type performs the way it does. To aid a discussion on the flow development of various flume configurations, a graph (Fig. 26) has been produced. An indication of how developed the velocity profile is at a particular point downstream can be found by dividing maximum velocity in a profile at the stated x location by the maximum velocity in the developed velocity profile. This calculated variable is labelled “% developed” in Fig. 26 (100% is fully developed). In Fig. 26 the flow development histories of five flume configurations are compared: • • • • •
Upwelling. Undershot. Isolating inlet set with vein angles set at 10°. Isolating inlet set with vein angles set at 25°. Isolating inlet set with vein angles set at 45°.
Table 4 Numerical results table for all the different flume configurations. Case
Flume type
Average velocity (m/s)
Mesh size (cells)
Note
1A 1B 1C 2A 2B 2C 2D 2E 2F MI1 MI2
Isolating inlet 25–37° Isolating inlet 25–37° Isolating inlet 25–37° Upwell Undershot Isolating inlet 45° Isolating inlet 10° Isolating inlet 25° Isolating inlet 25–37° Isolating inlet 25° Isolating inlet 25°
0.31 0.54 0.80 0.80 0.80 0.80 0.80 0.80 0.80 0.80 0.80
29,204 29,204 29,204 29,204 29,204 29,204 29,204 29,204 29,204 29,204 116,816
Validation case to match experiments Validation case to match experiments Validation case to match experiments
Mesh independence case (same as 2E) Mesh independence case
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 24. Velocity profiles at various positions downstream of the wave-maker face (x = 0 m) for a 25° isolating inlet flume simulation (Case 2E). Origin location for x and z positions is given in Fig. 9.
From the results it is clear that flume configuration has a significant effect on development distance. The development distance determines how long a flume would need to be, or how large the diameter of a circular tank would need to be, to have a useful test area. A likely deployment area for tidal turbine arrays is the Pentland Firth (Shields et al., 2009). If the conditions of the Pentland Firth were scaled to 1/20 in a test tank/flume the water depth would be two metres and the average velocity would be 0.8 m/s. If the upwelling or undershot configurations were used to create these conditions a flume would have to be 25 m long before the test zone could start. A round tank using an upwelling or undershot configuration would need to be 60 m in diameter to have a 10 m test zone at the centre of the tank, and to maintain the ability to change the flow direction without moving the useful test zone. The isolating inlet allows the test zone to start at around 10 m into a flume and reduces the diameter required for a round tank to 25–30 m. By reducing the diameter of a round tank to 25–30 m the construction costs of such a facility are considerably lowered making it a practical proposition. Turbulence is also critical in test tanks, firstly because it has an effect on how the flow develops, and secondly the turbulence level in the test zone of the tank must remain sea representative (10–15% turbulent intensity (Giles et al., 2011)). The configuration of the flume has a significant effect on how much energy is put into turbulence and how quickly it dissipates. Fig. 27 illustrates how turbulence is generated and dissipated by different flume configurations. Fig. 27 is a plot of the maximum turbulent intensity on a span-wise line at the given x position along the length of the flume. The isolating inlet type flumes dissipate turbulence more quickly than the upwelling
Fig. 25. Velocity profiles at various positions downstream of the wave-maker face (x = 0 m) for a 10° isolating inlet flume simulation (Case 2D). Origin location for x and z positions is given in Fig. 9.
127
Fig. 26. A comparison of flow development for various flume types.
and undershot types. From this analysis it is clear that only an isolation-type flume with the correct inlet angle could produce a searepresentative level of turbulence within a distance that would be economically achievable in a round tank. The inlet turbulent intensity might be further improved by reducing turbulence level before the inlet using flow conditioning. The rates at which the velocity profile develops and turbulence dissipates are related to shear layers. The difference in velocity profile development length for each flume configuration can be explained by the number and size of the shear layers they create. The two flume configurations that perform the worst are the upwelling and undershot: the reason for this can be seen when examining the X = 3 m velocity profiles in Figs. 17 and 20. A shear layer is set up in the flume where about half the fluid is moving at a high speed and the other half below it with either zero or negative velocity. For the flow to recover the energy in the top stream it has to equalise with the lower stream. The upwelling and undershot configurations maximise the amount of energy that has to be transferred before the flows equalise, increasing development distance. One means of reducing the energy transfer is to reduce the amount of fluid that has to be accelerated for the flows to equalise. The isolating inlet flume reduces the imbalance by angling the inlet flow so that the area containing low speed flow below the inlet stream is reduced (Fig. 28). This angling creates an extra shear layer above the inlet flow which intersects the free-surface and creates a closed pocket, reducing the length of the upper shear layer. (See Fig. 29.) By further angling the shear layer the energy imbalance between the inlet flow and fluid below it can be reduced, improving flow development (Fig. 24). Beyond around a 25° inlet angle the development distance begins to increase again (Fig. 25) as the upper shear layer stops being interrupted by the free-surface and a large energy imbalance forms between the fast flow at the bottom of the flume and the slow flow above it.
Fig. 27. A comparison of turbulent dissipation for various flume types.
128
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 28. Contours of velocity magnitude (m/s) on an xz plane on the centre line for an isolating inlet flume with vein angle set at 45° (Case 2C).
Fig. 29. Contours of velocity magnitude (m/s) on an xz plane on the centre line for an isolating inlet flume set at 10° (Case 2D).
The best design for a current-only flume would be one where no shear layers were created; a design that would do this is shown in Fig. 3. Although this design is ideal for a current-only flume and has been implemented before (Ifemer, 2006) there are difficulties implementing wave-makers in combination with it. 4. Conclusions The paper began by describing a numerical method to simulate the flow in flume in the absence of waves. Once validated with experimental results the method was used to assess the performance of common flume types. The tests proved that a flume using the isolating inlet requires significantly less downstream length to achieve a developed flow and acceptable turbulence level than previous flume configurations. The isolating inlet has the additional benefit of creating a still zone where a conventional wave-maker might be used. Analysis led to the conclusion that the downstream length required to develop the flow was related to the restoration of balance between shear layers. A flume that minimises the energy difference on either side of the shear layer will develop more quickly. It was found that the energy balance was minimised when the inlet angle was set at 25°. The results and analysis presented in this paper should be of use to scientists and engineers seeking to design flumes, test tank and basins that create sea-like test conditions, thus improving the scope and range of laboratory testing. The numerical model described here should become a useful design tool for basins, tanks and flumes. Acknowledgements The authors would like to thank the Engineering and Physical Sciences Research Council for funding this research [EP/H012745/1].
References Buchmann, B., Skourup, J., Fai Cheung, K., 1998. Run-up on a structure due to second-order waves and a current in a numerical wave tank. Appl. Ocean Res. 20 (5), 297–308. Cea, L., Puertas, J., Pena, L., 2007. Velocity measurements on highly turbulent free surface flow using ADV. Exp. Fluids 42 (3), 333–348. Chanson, H., Trevethan, M., Koch, C., 2007. Discussion of turbulence measurements with acoustic Doppler velocimeters by Carlos M. García, Mariano I. Cantero, Yarko Niño, and Marcelo H. García. ASCE 133, 1283–1286. Fluent, 2006. Fluent 6.3 User's Guide, 10 Cavendish Court, Lebanon, NH 03766, U.S.A.. Giles, J., Myers, L., Bahaj, A., Shelmerdine, B., 2011. The Downstream Wake Response of Marine Current Energy Converters Operating in Shallow Tidal Flows. Hirt, C.W., 1988. Flow-3d Users Manual. Flow Science Inc. Hirt, C.W., Nichols, B.D., 1981. Volume of fluid (VOF) method for the dynamics of free boundaries. J. Comput. Phys. 39, 201–225. Ifemer, 2006. Hydrodynamics Facilities Newsletter Issue 2. Ifemer, 2010. Hydrodynamics Facilities Newsletter Issue 5. Janssen, P., 1989. Wave-induced stress and the drag of air flow over sea waves. J. Phys. Oceanogr. 19 (6), 745–754. Jonsson, I.G., 1990. In: Le Méhauté, B. (Ed.), Wave–current interactions. The sea Wiley Interscience, pp. 65–120. Kemp, P.H., Simons, R.R., 1982. The interaction between waves and a turbulent current: waves propagating with the current. J. Fluid Mech. 116 (1), 227–250. Kirkegaard, J., 2013. Hydrolab4 dhi offshore wave basin, http://www.hydralab.eu/ facilities_view.asp?id=3. Lohrmann, A., Cabrera, R., Kraus, N.C., 1994. Acoustic-Doppler Velocimeter (ADV) for Laboratory Use. Proc. Conf. on Fundamentals and Advancements in Hydraulic Measurements and Experimentation, Buffalo, NY, pp. 351–365. Loucks, R.B., Wallace, J.M., 2012. Velocity and velocity gradient based properties of a turbulent plane mixing layer. J. Fluid Mech. 699, 280–319. Maguire, A.E., 2011. Hydrodynamics, Control and Numerical Modelling of Absorbing Wavemakers. The University of Edinburgh. Martin, V., Fisher, T.S.R., Millar, R.G., Quick, M.C., 2002. Adv data analysis for turbulent flows: low correlation problem. ASCE 113, 1–10. Miller, D.S., 1990. Internal flow systems. BHRA (Information Services), Bedford, UK. Minoura, M., Takahashi, R., Okuyama, E., Naito, S., 2009. Generation of Extreme Wave Composed of Ring Waves in a Circular Basin. Proc 19th Int Offshore and Polar Eng Conf, Osaka, ISOPE, pp. 389–396. Naito, S., 2006. Wave Generation and Absorption-theory and Application. In: Chung, J.S., Kashiwagi, M., Ferrant, J., Mizutani, N., Chien, L.K. (Eds.), Proceedings of the Sixteenth International Offshore and Polar Engineering Conference, pp. 1–9.
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129 Nortek-AS, Vectrino data sheet. Ortloff, C.R., Krafft, M., 1997. Numerical Test Tank: Simulation of Ocean Engineering Problems by Computational Fluid Dynamics. Offshore Technology Conference (p.^pp.). Poindexter, C.M., Rusello, P.J., Variano, E.A., 2011. Acoustic Doppler velocimeter-induced acoustic streaming and its implications for measurement. Exp. Fluids 50 (5), 1429–1442. Robinson, A.J., Morvan, H.P., Eastwick, C.N., 2010a. Computational investigations into draining in an axisymmetric vessel. J. Fluids Eng. 132 (12), 121104. Robinson, A., Eastwick, C., Morvan, H.P., 2010b. Further Computational Investigations Into Aero-engine Bearing Chamber Off-take Flows. Gas Turbine Technical Congress and Exposition, Glasgow, UK (p.^pp.). Robinson, A., Bryden, I., Ingram, D., Bruce, T., 2014a. The use of conditioned axial flow impellers to generate a current in test tanks. Ocean Eng. 75, 37–45. Robinson, A., Richon, J.-B., Bryden, I., Bruce, T., Ingram, D., 2014b. Vertical mixing layer development. Eur. J. Mech. - B/Fluids 43, 76–84. Robinson, A., Ingram, D., Bryden, I., Bruce, T., 2014n. The generation of 3d flows in a combined current and wave tank. J. Ocean Eng. Submitted for publication. Rusello, P.J., Lohrmann, A., Siegel, E., Maddux, T., 2006. Improvements in Acoustic Doppler Velocimetery. Proceedings of the Seventh International Conference on Hydroscience and Engineering, Philadelphia, PA (p.^pp). Salter, S.H., 1981. Absorbing Wave-makers and Wide tanks. Directional Wave Spectra Applications, Berkeley,pp. 185–202.
129
Salter, S.H., 2001. Proposals for a Combined Wave and Current Tank with Independent 360° Capability. Proceedings Marec 2001, 2-day conference on marine renewable energies, Newcastle, UK (p.^pp.). Salter, S.H., 2003. Design and Construction of a 360-degree Flow Table with Control of Velocity Gradient, IGR Report to EPSRC GR/R20694/01. Shields, M.A., Dillon, L.J., Woolf, D.K., Ford, A.T., 2009. Strategic priorities for assessing ecological impacts of marine renewable energy devices in the Pentland Firth (Scotland, UK). Mar. Policy 33 (4), 635–642. Taylor, J.R.M., Rea, M., D.J., R., 2003. The Edinburgh Curved Tank. 5th European Wave Energy Conference, Cork, Ireland, pp. 307–314. White, F.M., 1994. Fluid Mechanics. McGraw-hill. Wolf, J., Prandle, D., 1999. Some observations of wave–current interaction. Coast. Eng. 37 (3), 471–485. Yakhot, V., Orszag, S.A., Thangam, S., Gatski, T.B., C.G., S., 1992. Development of turbulence models for shear flows by a double expansion technique. Phys. Fluids 4 (7), 1510–1520. Yao, A., Wu, C.H., 2005. An automated image-based technique for tracking sequential surface wave profiles. Ocean Eng. 32 (2), 157–173.
Contents lists available at ScienceDirect
Coastal Engineering journal homepage: www.elsevier.com/locate/coastaleng
The effect of inlet design on the flow within a combined waves and current flumes, test tank and basins Adam Robinson ⁎, David Ingram, Ian Bryden, Tom Bruce Institute for Energy Systems, School of Engineering, The University of Edinburgh, Kings Buildings, Mayfield Rd, Edinburgh EH9 3JL, United Kingdom
a r t i c l e
i n f o
Article history: Received 30 April 2014 Received in revised form 15 October 2014 Accepted 16 October 2014 Available online xxxx Keywords: Current Wave Flume Test Tank Basin
a b s t r a c t The motion of the sea, through waves and currents, represents a large source of clean and safe energy. However, any structure built to operate in the sea will experience large varying forces and a difficult environment. It is therefore crucial to develop realistic and repeatable sea-like conditions in a laboratory in order to lower the cost and risk of developing off-shore structures. Building on previous efforts, an experimentally validated numerical model is used to predict the current-only flow in flumes capable of combining waves and current. This model is then used to simulate the flows within common flume configurations and within a new concept known as the “isolating inlet flume”. The results of these simulations are then analysed to assess the performance of each flume type and to understand the fluid dynamics that govern each type. Flume performance is found to be largely determined by the creation and dissipation of shear layers. The tests proved that a flume using the isolating inlet requires significantly less downstream length to achieve a developed flow and acceptable turbulence levels than the previous flume configurations. The isolating inlet has the additional benefit of creating a still zone where a conventional wave-maker might be used. Further simulations are used to investigate the design of the isolating inlet flume and demonstrate how it works. This paper should be of use to scientists and engineers seeking to design flumes, test tanks and basins that create sea-like test conditions, thus improving the scope and range of laboratory testing. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Flumes and test tanks that combine waves and currents to produce realistic two and three-dimensional seas provide an opportunity to test scaled off-shore devices and structures, reducing the cost and risk of development. To provide this testing capability several configurations of Combined Wave and Current (CWC) test tanks and flumes have been developed. The current is usually driven either by a pump or by an impeller and then guided into the test tank. The waves are made using an oscillating paddle or piston. The wave-making paddles also have the task of reabsorbing the waves to remove unwanted reflections from the tank (Salter, 1981). To date the production of the current compromises the wave-field and vice versa once a certain flow speed is reached. This means that it has not been possible to create sea-like test conditions representative of important potential tidal power generation sites like the Pentland Firth in terms of wave-field, current speed and turbulence level. Here a numerical model based on the Reynolds Averaged Navier– Stokes (RANS) equations is developed to aid flume design and assess current-only performance of CWC flumes, tanks and basins. The method is then used to investigate the flow within various configurations of ⁎ Corresponding author. E-mail address: [email protected] (A. Robinson).
http://dx.doi.org/10.1016/j.coastaleng.2014.10.004 0378-3839/© 2014 Elsevier B.V. All rights reserved.
flume that have been designed to include wave-making and current creation simultaneously. Here the current is assessed in isolation of the waves with the effect of waves on performance not assessed. All of the flume designs tested are based around a re-circulation principle where flow moves across the test section before it is drawn below a floor, then accelerated with an impeller or pump to be reinjected at the start of the test section (Fig. 5). Test tanks exist where current is provided by either jets supported on frames or by arrays of propellers in the test areas of the flume or tank. These configurations are useful in some cases but cannot achieve test conditions as searepresentative as a recirculating tank. Here four configurations of re-circulating CWC flume are tested at the same scale and operating velocity to allow comparison. The first is up-welling paired with a conventional flap wave-maker where the current enters the tank vertically directly in front of the wave-maker (Fig. 1). The second method of combining currents and waves could be described as undershot where the current is inlet horizontally below the wave-maker (Fig. 2). The third method of combining currents and waves works by using a shaped current guiding wave-maker to produce a racetrack-shaped flume (Fig. 3). The final method tested is the ‘isolating inlet flume’. This new inlet design introduces current into the tank in a way that creates a turbulent
118
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 1. Up-well type flume sliced along the centre line.
mixing layer which divides the area in front of the wave-maker and the fast moving flow entering the flume, minimising energy transfer, as shown in Fig. 4. Once the waves reach the dividing shear layer they pass through this boundary and are imposed on the current stream. The wavelength should then elongate in a consistent way dependent on the current velocity and direction. Fig. 5 is a diagram showing this inlet system implemented on a 3D CWC test tank. In the test region of a 3D tank (Fig. 5) where arrays of devices may be tested it is critical that the flow has developed a consistent velocity profile. Only changes in the velocity profile due to the upstream device, not flow development, should be experienced by a downstream device. This also determines how much of a 3D tank can be used for array testing and how large the tank needs to be to provide a given test area. The distance the flow takes to develop for a given flume configuration is a critical performance criterion and will be assessed here using a validated numerical model. The geometry of the isolating inlet is critical to its performance, in particular the inlet angle, the effect of which will be investigated here through simulations. This investigation aims to improve the physical understanding of what determines flume, tank and basin performance. The numerical model and functional explanations will provide a means of creating a test tank where more realistic representations of the tidal channels in terms of current speed and wave-field can be produced. Thus lowering the cost and risk of developing off-shore devices and structures. This paper will begin with a review of existing CWC tanks and flumes (Section 2). Following this will be a description of a numerical model based on the RANS method useful for flume design (Section 3.1). A description of the experiment used to validate the experiment is available in Section 3.2. In Section 4.1 these experimental results will then be used to validate this method. The numerical model is used to assess the
performance of different common flume configurations in Section 4.2. Section 5 will summarise the findings of this paper. 2. Background 2.1. Wave/current interaction in nature In nature waves and currents combine in areas of the sea such as tidal channels and costal zones as the tides come in and out. To an engineer the combination of waves and current is of interest because it leads to increased stresses and wear on off-shore structures (Salter, 2003; Wolf and Prandle, 1999). Current affects waves in the following ways: • Wavelength increases when the current and waves are in the same direction and decreases when they move in opposing directions. (Jonsson, 1990; Wolf and Prandle, 1999). • Currents reduce or increase wind shear on the surface affecting wave shape (Wolf and Prandle, 1999). • Waves turn to the current direction over time (Wolf and Prandle, 1999). Waves affect currents in the following ways: • The presence of waves can increase turbulent intensity of the bulk flow (Kemp and Simons, 1982). • Increase wind drag on surface due to waves alters the velocity profile (Janssen, 1989). • Waves increase friction on the sea bottom (Wolf and Prandle, 1999). Most of the effects described above do not influence the behaviour in a combined wave and current test tank. If the waves and currents combine in a similar way in a flume as they do in a tidal channel the wavelength should elongate and contract in a predictable way.
Fig. 2. Undershot type flume sliced along the centre line.
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
119
Fig. 3. Shaped wave-maker type flume.
2.2. Turbulent mixing layers Turbulent mixing layers are central to the new method of combining waves and current described in this paper. They are a Kelvin–Helmholtz instability that develops after the point where the flows are combined resulting in a 2D span-wise series of vortical structures known as rollers (Loucks and Wallace, 2012): see Fig. 6. More details on turbulent mixing layers relevant to combined current and wave tanks can be found in a related paper (Robinson et al., Submitted for publication). 2.3. Wave and current test tank design Although many tanks exist where waves and currents can combine none have the capability to produce 3D waves with 3D currents. The use of curved and round wave-tanks with wave-makers around the perimeter is well-developed and is described. (Minoura et al., 2009; Naito, 2006; Taylor et al., 2003). The work on 3D currents is limited to one
source (Salter, 2001). The design that could combine 3D waves and currents was only implemented on a scale model only capable of creating 3D currents (Salter, 2001, 2003; Taylor et al., 2003). This method generates a 3D current by rotating aerofoils around the perimeter of a tank and works on the same principal as a Voith–Schneider Propeller. Testing the scaled model proved that the design generally provides a robust means of producing a 3D current (Salter, 2003). Wave-tanks exist where current is provided inside the test tank by ether jets supported on frames in front of the wave-makers or by arrays of propellers (Kirkegaard, 2013). Although useful in some cases it is difficult to achieve sea-like velocity profiles and turbulence levels within a reasonable downstream distance. It is possible to remove turbulence and condition the flow in a flume or tank using screens and honeycomb (Miller, 1990) although placement is difficult where wave-makers are present. From surveying existing CWC tanks and flume designs it is apparent that there are three existing methods that could be used to make 3D currents in the presence of a 3D wave-field. The first is up-welling
Fig. 4. Method of isolating the generation of wave and current.
Fig. 5. A 3D CWC test tank sliced along the centre line.
120
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 8. Example mesh (Each line is approximately representative of 10 in the mesh used).
Fig. 6. Turbulent mixing layer diagram (Loucks and Wallace, 2012).
where the current enters the tank vertically directly in front of a conventional flap wave-maker (Fig. 1). With an up-well inlet the flow has to turn from vertical to horizontal before moving down the flume. This lifts the free-surface and may limit the flow speed for a given depth. The up-welling inlet is the easiest to implement in an existing wave-only flume and therefore many examples exist (Yao and Wu, 2005). The second method of combining currents and waves could be described as undershot where the current is inlet below the wave-maker (Fig. 2). Examples of this type of flume in use can be found from many sources (Ifemer, 2010; Kemp and Simons, 1982). The undershot configuration is liable to produce turbulent eddies after the slitter plate above a certain flow speed (Robinson et al., Submitted for publication). This configuration is the easiest to implement in an existing current-only flume. The final method of combining currents and waves uses a shaped current-guiding wave-maker (Fig. 3). This method was proposed but has never been implemented in a CWC tank suitable for testing (Salter, 2001, 2003). The shape of the flume known as a racetrack is proven to not add any turbulence to the flow and is used in many applications where a high speed/low turbulence flow is required (Ifemer, 2006). When the wave-maker is used on a flume of this configuration it will affect and be loaded by the current complicating generation and absorption. 2.4. Simulations combining waves and current Many useful CWC tank simulations have been created (Buchmann et al., 1998; Ortloff and Krafft, 1997) These models were produced to test ocean structures and do not include direct modelling of wave and current generation. Waves and currents are more easily included in simulations by prescribing boundary conditions. Numerical CWC tanks have been successfully created using a RANS-based method (Ortloff and Krafft, 1997). Only one example of absorbing wave-makers being directly modelled exists (Maguire, 2011). This method uses the flow3D multi-phase RANS code (Hirt, 1988) to simulate the behaviour of a wave-maker both producing and absorbing waves. A RANS-based method has also been proven for simulating the turbulent mixing layers usually present in CWC tanks (Robinson et al., Submitted for
Fig. 7. Contours of volume fraction for a 2D slice through a 2 m up-welling flume simulation with a free-surface present (black represents water and white is air).
publication). These works combined demonstrate that all the capabilities required to directly simulate a 3D combined current and wave tank are feasible within a RANS code.
2.5. Numerical model A computational model was created using the CFD code FLUENT (Fluent, 2006) to investigate the performance of various flume configurations. The model was first validated against experimental data in Section 4.1 then used to investigate the performance of various flume types in Section 4.2. The computational methodology was developed and validated specifically for accurate representation of flows containing turbulent mixing layers (Robinson et al., Submitted for publication). Although waves and moving wave-maker are not included in this simulation the representation of moving geometry is available in RANS-based codes (Hirt, 1988) and therefore could be included in the model described here. The approach is based on solving the RANS equations in an Eulerian fixed grid implemented within FLUENT. The k − ε model RNG turbulence model (Yakhot et al., 1992) is used for closure. Further detail of the method used can be obtained from a previous paper (Robinson et al., 2014b). The flow is simulated in 2D with the boundary conditions providing the complexity of the flow. The geometry is a rectangle of a given flume height that stretches in the flow direction far enough to ensure sufficient flow development and to ensure that the outlet boundary condition has a negligible effect on the flow in the area of interest. The validity of the 2D assumption is tested by measuring velocity profiles across the flume at various points. The velocity at the centre of the flume was found to be within 1% of the average velocity of a profile measured from the centre to the wall of experimental flume (Section 3.2) in the y direction.
Table 1 Water properties. Density (kg/m3)
Viscosity (kg/m-s)
Temperature (°C)
998.2
1.003 × 10−4
20 °C
Table 2 Mesh independence comparison between maximum velocity present in the velocity profile at various x locations for MI1 and MI2. X position (m)
Max × velocity (m/s) MI1
Max × velocity (m/s) MI2
Difference %
3 5 7.5 10 15 20 25 30 35 40
1.853 1.513 1.197 0.976 0.863 0.846 0.843 0.844 0.847 0.851
1.849 1.511 1.208 0.992 0.872 0.851 0.846 0.846 0.849 0.852
0.216 0.201 0.096 0.984 1.654 0.967 0.574 0.390 0.250 0.154
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
121
Fig. 9. A 2D stream-wise slice of the ECWC flume. The flume has a uniform cross-section in the y direction (400 mm wide). The datum used throughout is at the intersection of the wavemaker face at flume floor.
Fig. 10. Vector and contour map on the xz plane from x = −100 to 800 and z = 10 to 360 mm compiled with experimental data for Case 1A. Velocity is proportional to vector length and contour colour (Key given in figure).
Fig. 11. Vector and contour map on the xz plane from x = −100 to 800 and z = 0 to 400 mm predicted using a numerical model for Case 1A. Velocity is proportional to vector length and contour colour (Key given in figure).
122
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 12. Mean velocity versus distance from the bottom wall for various X positions for Case 1B. The dashed line is the simulated result, black is the experimental measurements.
Fig. 13. Mean velocity versus distance from the bottom wall for various X positions for Case 1C. The dashed line is the simulated result, black is the experimental measurements.
The flume inlet is described by a boundary condition and is set by prescribing node values for u and v velocities, turbulent kinetic energy, k and turbulent dissipation rates, and ε to match those recorded at the corresponding location in the experiment. The bottom boundary is set as a wall with a roughness matching the experiment (roughness height 0.0015 mm, roughness coefficient 0.5). The downstream boundary condition is set as a Neumann pressure boundary. The boundary condition configuration is shown in Fig. 8. The simulation of the free-surface in the flume represents a large increase in the computational cost of using this model and may not be necessary to accurately predict the downstream velocity profile development. To investigate this the method was extended to include the VOF multiphase model (Hirt and Nichols, 1981). The implementation used here is detailed in an earlier paper (Robinson et al., 2010a,b). The flume configuration that would have the most effect on the freesurface is the up-welling type (Fig. 1), where the inlet flow is directed vertically. To test whether it is necessary to model the free surface a simulation was run of an up-welling flume with a 2 m water depth and an average flume velocity of 0.8 m/s matching Case 2A. The result is shown in Fig. 7. In the simulation the free-surface was only lifted by 0.11 m with no significant change to velocity profile development. When an isolating inlet flume was tested (25° inlet angle) with a free-surface, no deformation was detected.
Fig. 14. Experimental image in front of a fixed wave-maker for the flume described in Section 3.1 in up-welling mode (0.6 m/s depth averaged flume velocity, 0.4 m water depth).
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
123
Fig. 15. Up-welling simulation configuration.
Having found that modelling the free-surface made little difference to the result and added greatly to the computational cost it was substituted with a free-slip wall boundary condition in all other tests. The physical properties of the water are set within the models and are based on conditions recorded in the experiments. The values used in the computations were found by using the average experimental temperatures and tables from the literature (White, 1994) and are given in Table 1. The mesh created for the computational model is a structured hexahedral mesh with a higher concentration of cells at the fluid inlet and boundary layer. The basic mesh structure is shown in Fig. 8. In the stream-wise direction the mesh expands at a rate of 1.05 from the inlet boundary to the outlet boundary. Vertically the mesh expands at a rate of 1.05 to the upper boundary from the lower boundary to properly resolve the boundary layer on the bottom wall. This mesh strategy is used throughout. A detailed mesh independence study of relevance to this case was in a previous paper (Robinson et al., 2014b), and the strategy and densities found were transferred to the simulations done here. To ensure that the solutions were independent of the number and location of the cells two meshes were created: MI1 and MI2 with identical geometry and identical boundary conditions. MI2 had four times the number of cells as MI1. For each mesh the downstream velocity profiles were recorded and analysed. Table 2 shows a comparison between the maximum u velocity within the velocity profile at various x locations downstream of the wave-maker face (x = 0). When the results from MI1 and MI2 were compared the average difference was always less than 2%, proving MI1 is sufficiently meshing independent. The compute time for the case with 29,204 cells (Case 2E) to develop from a state where the velocity is equal to zero in every cell to steady state flow seen in Fig. 23 was 3300 s. This simulation was carried out on a four-core 2.93 GHz Intel i7 870 processor with equation residuals converged to 1 × 10−7. The experiment used to validate the numerical model is described in the following section.
Fig. 17. Velocity profiles at various positions downstream of the wave-maker face (x = 0 m) for an upwelling flume simulation (Case 2A). Origin location for x and z positions is given in Fig. 9.
2.6. Validation experiment In order to validate the numerical model a flume with a water depth of 0.4 m and a length of 8 m (wave-maker to wave-maker) was used. The experiment is an identical scaled representation and 2D slice of what would be used in a 3D test tank (Fig. 5). The resulting flume is shown in Fig. 9. The flume has absorbing wave-makers at both ends however they are locked in the vertical position throughout these tests. The water is typically driven in the direction shown in Fig. 9 to allow the maximum distance for the propeller wake to develop and turbulence to dissipate. The flume walls are constructed with glass to allow optical access. 2.6.1. Measurement setup Point measurements of velocity in components u, v and w were taken using a Vectrino + ADV (Nortek-AS, n./d.). These measurements were used to calculate mean velocity and turbulent intensity. The flume is designed to maintain a steady state flow once it has accelerated from rest and therefore mean velocity measurements taken from large numbers of samples are suitable for this characterisation. Checks are built into the measurement processing to ensure that no transient effects caused by flume operation, start-up or large turbulent structures are present in the velocity measurements taken. Initially a sample of 360,000 measurements over 30 min (velocity setting 1C, Table 4) was taken to detect flume transience. By sub-dividing this into minute blocks, then comparing average velocities of each, it was found that 8 min (96,000 samples) were required to ensure the error in average velocity due to transience that was less than 1%. A further test was built into the measurement processing that compared the average velocity
Fig. 16. Vector and contour map around the inlet for an up-welling flume simulation (Case 2A). Velocity is proportional to vector length and contour colour (Key given in figure).
124
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 18. Undershot flume simulation configuration.
of the first 48,000 samples of every point measurement against the second 48,000 samples. This test ensured that the difference in average velocity between each half never exceeded 1% for the measurements presented in this paper. This high number of velocity samples is statistically sufficient to allow turbulence characterisation by Reynolds stresses or turbulent kinetic energy (Chanson et al., 2007). Along with velocity, the ADV derives two other quantities which relate to the quality of each measurement being taken. For every sample signal to noise ratio (SNR) and correlation are given. Signal to noise ratio was maintained above fifteen throughout this experiment in line with previous work (Cea et al., 2007; Lohrmann et al., 1994; Rusello et al., 2006). Correlation was used to filter the data from the ADV throughout (Martin et al., 2002; Rusello et al., 2006). Previous investigators have recommended 70% as a minimum threshold value; however 75% was used here (Martin et al., 2002; Rusello et al., 2006). The correlation value decreases as the turbulence increases (Robinson et al., 2014a). By extensive testing of ADV settings it was possible to achieve average correlation values exceeding 95% for all measurements in the working section of the flume (Section 4.1). In the more turbulent inlet and still zones (Sections 4.2 and 4.3) average correlation was always above 85%. These high correlation values ensured that at least 95,000 of 96,000 samples remained after filtering for working section and 75,000 of 96,000 for the inlet zone. This ensured accurate average velocity values and turbulence characterisation. To achieve this great care was taken to find ADV settings that would lead to the highest possible correlation values and smooth consistent velocity profiles. Typically a sampling rate of 50 Hz is used (Cea et al., 2007; Chanson et al., 2007; Martin et al., 2002; Robinson et al., 2014a; Rusello et al., 2006); however better results could be achieved at 200 Hz. The ADV measures an approximately 6 mm diameter cylindrical volume with a variable length. For the measurements in the working section of the flume (Section 4.1) a large measurement volume (9.2 mm) and a 2.4 mm acoustic pulse length were found to give the highest correlation values and give the smoothest and most consistent velocity profiles. In the inlet zone where the flow was more turbulent the acceptable correlation levels could only be achieved by minimising the size of the measurement volume (3 mm) and the length of the acoustic pulses (0.3 mm). The measurements in the inlet zone were at the limit of the ADV capability and therefore an improved ADV or alternative technique would be
Fig. 20. Velocity profiles at various positions downstream of the wave-maker face (x = 0 m) for an undershot flume simulation (Case 2B). Origin locations for x and z positions are given in Fig. 9.
required to achieve accurate measurements beyond an average flume velocity of 0.6 m/s (Table 4). The power level was set at the highest level possible to maximise the signal to noise ratio and correlation values. The acoustic streaming effect exacerbated by high power levels should be negligible for the velocities measured in this flume (Poindexter et al., 2011). Seeding was provided by neutrally buoyant hollow glass spheres with a mean diameter of 11.7 μm. By adding seeding until the correlation value stabilised the correct density was found. A good general overview of Acoustic Doppler Velocimetry is provided by Lohrmann et al. (Lohrmann et al., 1994) with further information on the specific ADV used here available from Nortek (Nortek-AS, n./d.).
2.6.2. Experimental uncertainty The geometrical tolerance of the experimental rig was less than ±1 mm with the measurement setup manufactured to ±0.1 mm. The x and z positions given here were measured with tapes and therefore have an accuracy of at least ± 1 mm. The y position is measured by the ADV and is accurate to 0.1 mm (Nortek-AS, n./d.). Immediately before the experiments were reported here the ADV was calibrated in a towing tank and was found to have an accuracy exceeding ±0.5% for velocity measurements. Using the error level found during the instrument calibration (±0.5%) and the potential errors in position measurement it is possible to assess maximum total error in various areas of the flume. The maximum combined position error in the x and z directions is ± 2 mm. By multiplying this by the maximum velocity change per mm recorded by the ADV whilst characterising the flume the maximum error can be found. The highest velocity gradients are found in the inlet area and reduce the further downstream as the water travels. The maximum error in velocity for the experiments for various flume areas is given in Table 3.
Fig. 19. Vector and contour map around the inlet for an undershot flume simulation (Case 2B). Velocity is proportional to vector length and contour colour (Key given in figure).
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
125
Fig. 21. Isolating inlet flume simulation configuration.
Table 3 the first column, maximum velocity gradient, this is found by analysing the ADV data. The second column, maximum velocity error due to position error %, is calculated by multiplying the maximum gradient by the maximum position error (2 mm). The fourth column combines the error in velocity due to positional inaccuracy with velocity measurement error of the ADV found during calibration. Temperature was measured using a thermocouple mounted inside the ADV with a quoted accuracy of ±0.1 °C (Nortek-AS, n./d.). The temperature of the flume was stable during the measurement campaign with the minimum temperature measured during the tests of 19.1 °C and a maximum of 21.2 °C. 3. The effect of inlet design on the flow within a flume To validate and check the numerical model and then investigate different flume configurations a series of tests were created. The details of these tests are described in Table 4. 3.1. Model validation To enable validation a series of measurements were taken in an isolating inlet flume (Fig. 9) at three different total flow rates. The numerical model described in Section 3 is set up to match the experiment described in Section 3.2. The initial test for the numerical model is its ability to accurately predict bulk flow patterns. In an isolating inlet flume the critical area for establishing how the water will move is the zone containing the wave-maker face and the inlet. The first assessment of the model will be its ability to predict how the fluid behaves in this area. The second assessment will be to compare how the flow develops as it moves downstream. In Fig. 10 a map of the flow in front of the wave-maker made using experimental measurements (ADV) is given. An equivalent flow map predicted using the numerical method described in Section 3.1 is shown in Fig. 11. By comparing the vector and contour maps for the numerical (Fig. 11) and experimental (Fig. 10) results it is clear the numerical model has accurately predicted the behaviour of the bulk flow. Most importantly the simulation has predicted the still zone in front of the
Fig. 23. Velocity profiles at various positions downstream of the wave-maker face (x = 0 m) for a 45° isolating inlet flume simulation (Case 2C). Origin location for x and z positions is given in Fig. 9.
wave-maker (x = −120 mm, Z = 0 mm to 400 mm). The low speed rotations above and below the inlet stream are present in the CFD but were undetected in the experiment, likely due to the limitations of the measurement technique. In the transition layer (0.2–0.5 m/s) thickness and expansion are comparable between the CFD and experimental result. The transition consists of a mixing layer made of complex turbulent structures (Robinson et al., Submitted for publication). The k − ε model used here to model turbulent behaviour does not directly model these structures and therefore there may have been inaccuracy in the simulation of the effects of the turbulent mixing layer. This result indicates the model's suitability. To ensure the method accurately predicts the development length in flumes velocity profiles were taken at various positions downstream of the inlet. A comparison of these velocity profiles for both numerical and experimental results for Cases 1B and 1C is shown in Fig. 12 and Fig. 13. The simulation predicts the shape of the velocity profile well, especially close to the inlet. The growing difference is likely due to the differences in behaviour between the outlet in the flume and the CFD boundary condition. Table 5 shows the average error in the prediction of point velocities in each velocity profile taken between the CFD and the experiment. With the numerical model able to predict the flow development in a flume to a reasonable degree of accuracy the model will now be used to predict and analyse the flow in common flume types. 3.2. Development length of various flume types Having developed and validated a numerical model for predicting the flow development in a flume the method will be used to investigate other flume types. The flumes will be simulated with a two metre water depth representative of the Flowave Test Basin, Fig. 5. The configurations tested are up-welling (Fig. 1), undershot (Fig. 2) and the isolation
Fig. 22. Vector and contour map around the inlet for a 25° isolating inlet flume simulation (Case 2E). Velocity is proportional to vector length and contour colour (Key given in figure).
126
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Table 3 Velocity measurement error due to position error (locations marked on Fig. 9). Area
Inlet (x = 0–460, Z = 15–360) x = 910 (z = 10–340 mm) x = 1910 (z = 10–340 mm) x = 4910 (z = 10–340 mm)
Maximum velocity gradient (m/s per mm)
Maximum velocity error due to position error %
Total maximum error %
0.020 0.015 0.005 0.001
6.7 2.1 0.8 0.3
7.2 2.6 1.3 0.8
type (Fig. 4). The shaped wave-maker type (Fig. 3) is not tested here as it represents the ideal flume in the current-only model and inlets fluid in a way that has the minimum effect on development length. For the isolation-type flume simulation several configurations are tested to assess the effect of inlet angle on development distance. The situations in this section are set up identically to the validation case described in Section 4.1 with the different flume configurations simulated by adjusting the positions and node values of the inlet boundary. Turbulent intensity at the inlet is set at 10% for all flumes with characteristic turbulent length scale set at 0.1 m due to the likely inlet guide vein spacing. 3.2.1. Up-welling flume The up-welling configuration shown in Fig. 1 directs the flow directly up at the free-surface in front of the wave-maker. This configuration results in significant energy loss due to free-surface lifting before the water is turned and flows down the channel. This free-surface disruption is significant and often unsteady and effectively leads to a wave-maker hinge depth that varies with current velocity (Fig. 14). Both these features may lead to difficulties in wave-maker control when absorbing and creating waves. In the up-scaled flume simulated here (0.8 m/s depth averaged flume velocity, 2 m water depth) the extra water depth suppresses the surface lifting to a negligible level (Section 3.1). A simulation of a 2 m depth up-welling flume was created using the method described in Section 3.1; the model setup is shown in Fig. 15. The resulting flow pattern around the inlet for the up-welling simulation is shown in Fig. 16. From the vector map above it is clear that the up-welling inlet results in a strong recirculation above the bottom wall which will act to increase the development distance. Velocity profiles at various positions down the length of the simulated flume are shown in Fig. 17. 3.2.2. Undershot flume The undershot configuration shown in Fig. 2 directs the flow horizontally underneath the wave-maker. A simulation of a 2 m depth upwelling flume was set up using the method described in Section 3.1; the flume model configuration is shown in Fig. 18. Water is inlet through a one metre vertical inlet at 1.6 m/s once the flow mixes
Table 5 Average % error in velocity profile for Case 1B and 1C. Case
X position (mm) 600
975
2000
5000
1B 1C
8.03% 5.43%
16.45% 13.02%
8.65% 9.61%
16.31% 16.72%
downstream. A 0.8 m/s depth averaged flume velocity should result in the 2 m deep flume. By passing a flowing stream under still fluid the conditions for a strong turbulent mixing layer are set up, resulting in the creation turbulent structures and a large recirculation flow above the inlet (Fig. 19). Velocity profiles at various positions down the length of the simulated flume are shown in Fig. 17. 3.2.3. Isolating inlet flume The isolating inlet flume aims to create a still zone in front of the wave-maker by introducing the fluid at an angle as shown in Fig. 4. The flume simulation configuration is shown in Fig. 21. Water is inlet through a two metre horizontal inlet at a speed that results in a 0.8 m/s depth averaged flume velocity in the two metre deep flume. Introducing the water at an angle results in the creation of a turbulent shear layer much like the undershot method; however the layer is limited by contact with the free surface, reducing turbulence production (Fig. 22). To assess the effect of inlet angle on development distance the simulation was set at three different angles: 45° (Fig. 23); 25° (Fig. 24); and 10° (Fig. 25). 3.2.4. Flume type discussion In this section the performance of each of the flume types is assessed and compared in the context of a current-only flume with a wavemaker present but not moving. Following this an effort will be made to understand why each flume type performs the way it does. To aid a discussion on the flow development of various flume configurations, a graph (Fig. 26) has been produced. An indication of how developed the velocity profile is at a particular point downstream can be found by dividing maximum velocity in a profile at the stated x location by the maximum velocity in the developed velocity profile. This calculated variable is labelled “% developed” in Fig. 26 (100% is fully developed). In Fig. 26 the flow development histories of five flume configurations are compared: • • • • •
Upwelling. Undershot. Isolating inlet set with vein angles set at 10°. Isolating inlet set with vein angles set at 25°. Isolating inlet set with vein angles set at 45°.
Table 4 Numerical results table for all the different flume configurations. Case
Flume type
Average velocity (m/s)
Mesh size (cells)
Note
1A 1B 1C 2A 2B 2C 2D 2E 2F MI1 MI2
Isolating inlet 25–37° Isolating inlet 25–37° Isolating inlet 25–37° Upwell Undershot Isolating inlet 45° Isolating inlet 10° Isolating inlet 25° Isolating inlet 25–37° Isolating inlet 25° Isolating inlet 25°
0.31 0.54 0.80 0.80 0.80 0.80 0.80 0.80 0.80 0.80 0.80
29,204 29,204 29,204 29,204 29,204 29,204 29,204 29,204 29,204 29,204 116,816
Validation case to match experiments Validation case to match experiments Validation case to match experiments
Mesh independence case (same as 2E) Mesh independence case
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 24. Velocity profiles at various positions downstream of the wave-maker face (x = 0 m) for a 25° isolating inlet flume simulation (Case 2E). Origin location for x and z positions is given in Fig. 9.
From the results it is clear that flume configuration has a significant effect on development distance. The development distance determines how long a flume would need to be, or how large the diameter of a circular tank would need to be, to have a useful test area. A likely deployment area for tidal turbine arrays is the Pentland Firth (Shields et al., 2009). If the conditions of the Pentland Firth were scaled to 1/20 in a test tank/flume the water depth would be two metres and the average velocity would be 0.8 m/s. If the upwelling or undershot configurations were used to create these conditions a flume would have to be 25 m long before the test zone could start. A round tank using an upwelling or undershot configuration would need to be 60 m in diameter to have a 10 m test zone at the centre of the tank, and to maintain the ability to change the flow direction without moving the useful test zone. The isolating inlet allows the test zone to start at around 10 m into a flume and reduces the diameter required for a round tank to 25–30 m. By reducing the diameter of a round tank to 25–30 m the construction costs of such a facility are considerably lowered making it a practical proposition. Turbulence is also critical in test tanks, firstly because it has an effect on how the flow develops, and secondly the turbulence level in the test zone of the tank must remain sea representative (10–15% turbulent intensity (Giles et al., 2011)). The configuration of the flume has a significant effect on how much energy is put into turbulence and how quickly it dissipates. Fig. 27 illustrates how turbulence is generated and dissipated by different flume configurations. Fig. 27 is a plot of the maximum turbulent intensity on a span-wise line at the given x position along the length of the flume. The isolating inlet type flumes dissipate turbulence more quickly than the upwelling
Fig. 25. Velocity profiles at various positions downstream of the wave-maker face (x = 0 m) for a 10° isolating inlet flume simulation (Case 2D). Origin location for x and z positions is given in Fig. 9.
127
Fig. 26. A comparison of flow development for various flume types.
and undershot types. From this analysis it is clear that only an isolation-type flume with the correct inlet angle could produce a searepresentative level of turbulence within a distance that would be economically achievable in a round tank. The inlet turbulent intensity might be further improved by reducing turbulence level before the inlet using flow conditioning. The rates at which the velocity profile develops and turbulence dissipates are related to shear layers. The difference in velocity profile development length for each flume configuration can be explained by the number and size of the shear layers they create. The two flume configurations that perform the worst are the upwelling and undershot: the reason for this can be seen when examining the X = 3 m velocity profiles in Figs. 17 and 20. A shear layer is set up in the flume where about half the fluid is moving at a high speed and the other half below it with either zero or negative velocity. For the flow to recover the energy in the top stream it has to equalise with the lower stream. The upwelling and undershot configurations maximise the amount of energy that has to be transferred before the flows equalise, increasing development distance. One means of reducing the energy transfer is to reduce the amount of fluid that has to be accelerated for the flows to equalise. The isolating inlet flume reduces the imbalance by angling the inlet flow so that the area containing low speed flow below the inlet stream is reduced (Fig. 28). This angling creates an extra shear layer above the inlet flow which intersects the free-surface and creates a closed pocket, reducing the length of the upper shear layer. (See Fig. 29.) By further angling the shear layer the energy imbalance between the inlet flow and fluid below it can be reduced, improving flow development (Fig. 24). Beyond around a 25° inlet angle the development distance begins to increase again (Fig. 25) as the upper shear layer stops being interrupted by the free-surface and a large energy imbalance forms between the fast flow at the bottom of the flume and the slow flow above it.
Fig. 27. A comparison of turbulent dissipation for various flume types.
128
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129
Fig. 28. Contours of velocity magnitude (m/s) on an xz plane on the centre line for an isolating inlet flume with vein angle set at 45° (Case 2C).
Fig. 29. Contours of velocity magnitude (m/s) on an xz plane on the centre line for an isolating inlet flume set at 10° (Case 2D).
The best design for a current-only flume would be one where no shear layers were created; a design that would do this is shown in Fig. 3. Although this design is ideal for a current-only flume and has been implemented before (Ifemer, 2006) there are difficulties implementing wave-makers in combination with it. 4. Conclusions The paper began by describing a numerical method to simulate the flow in flume in the absence of waves. Once validated with experimental results the method was used to assess the performance of common flume types. The tests proved that a flume using the isolating inlet requires significantly less downstream length to achieve a developed flow and acceptable turbulence level than previous flume configurations. The isolating inlet has the additional benefit of creating a still zone where a conventional wave-maker might be used. Analysis led to the conclusion that the downstream length required to develop the flow was related to the restoration of balance between shear layers. A flume that minimises the energy difference on either side of the shear layer will develop more quickly. It was found that the energy balance was minimised when the inlet angle was set at 25°. The results and analysis presented in this paper should be of use to scientists and engineers seeking to design flumes, test tank and basins that create sea-like test conditions, thus improving the scope and range of laboratory testing. The numerical model described here should become a useful design tool for basins, tanks and flumes. Acknowledgements The authors would like to thank the Engineering and Physical Sciences Research Council for funding this research [EP/H012745/1].
References Buchmann, B., Skourup, J., Fai Cheung, K., 1998. Run-up on a structure due to second-order waves and a current in a numerical wave tank. Appl. Ocean Res. 20 (5), 297–308. Cea, L., Puertas, J., Pena, L., 2007. Velocity measurements on highly turbulent free surface flow using ADV. Exp. Fluids 42 (3), 333–348. Chanson, H., Trevethan, M., Koch, C., 2007. Discussion of turbulence measurements with acoustic Doppler velocimeters by Carlos M. García, Mariano I. Cantero, Yarko Niño, and Marcelo H. García. ASCE 133, 1283–1286. Fluent, 2006. Fluent 6.3 User's Guide, 10 Cavendish Court, Lebanon, NH 03766, U.S.A.. Giles, J., Myers, L., Bahaj, A., Shelmerdine, B., 2011. The Downstream Wake Response of Marine Current Energy Converters Operating in Shallow Tidal Flows. Hirt, C.W., 1988. Flow-3d Users Manual. Flow Science Inc. Hirt, C.W., Nichols, B.D., 1981. Volume of fluid (VOF) method for the dynamics of free boundaries. J. Comput. Phys. 39, 201–225. Ifemer, 2006. Hydrodynamics Facilities Newsletter Issue 2. Ifemer, 2010. Hydrodynamics Facilities Newsletter Issue 5. Janssen, P., 1989. Wave-induced stress and the drag of air flow over sea waves. J. Phys. Oceanogr. 19 (6), 745–754. Jonsson, I.G., 1990. In: Le Méhauté, B. (Ed.), Wave–current interactions. The sea Wiley Interscience, pp. 65–120. Kemp, P.H., Simons, R.R., 1982. The interaction between waves and a turbulent current: waves propagating with the current. J. Fluid Mech. 116 (1), 227–250. Kirkegaard, J., 2013. Hydrolab4 dhi offshore wave basin, http://www.hydralab.eu/ facilities_view.asp?id=3. Lohrmann, A., Cabrera, R., Kraus, N.C., 1994. Acoustic-Doppler Velocimeter (ADV) for Laboratory Use. Proc. Conf. on Fundamentals and Advancements in Hydraulic Measurements and Experimentation, Buffalo, NY, pp. 351–365. Loucks, R.B., Wallace, J.M., 2012. Velocity and velocity gradient based properties of a turbulent plane mixing layer. J. Fluid Mech. 699, 280–319. Maguire, A.E., 2011. Hydrodynamics, Control and Numerical Modelling of Absorbing Wavemakers. The University of Edinburgh. Martin, V., Fisher, T.S.R., Millar, R.G., Quick, M.C., 2002. Adv data analysis for turbulent flows: low correlation problem. ASCE 113, 1–10. Miller, D.S., 1990. Internal flow systems. BHRA (Information Services), Bedford, UK. Minoura, M., Takahashi, R., Okuyama, E., Naito, S., 2009. Generation of Extreme Wave Composed of Ring Waves in a Circular Basin. Proc 19th Int Offshore and Polar Eng Conf, Osaka, ISOPE, pp. 389–396. Naito, S., 2006. Wave Generation and Absorption-theory and Application. In: Chung, J.S., Kashiwagi, M., Ferrant, J., Mizutani, N., Chien, L.K. (Eds.), Proceedings of the Sixteenth International Offshore and Polar Engineering Conference, pp. 1–9.
A. Robinson et al. / Coastal Engineering 95 (2015) 117–129 Nortek-AS, Vectrino data sheet. Ortloff, C.R., Krafft, M., 1997. Numerical Test Tank: Simulation of Ocean Engineering Problems by Computational Fluid Dynamics. Offshore Technology Conference (p.^pp.). Poindexter, C.M., Rusello, P.J., Variano, E.A., 2011. Acoustic Doppler velocimeter-induced acoustic streaming and its implications for measurement. Exp. Fluids 50 (5), 1429–1442. Robinson, A.J., Morvan, H.P., Eastwick, C.N., 2010a. Computational investigations into draining in an axisymmetric vessel. J. Fluids Eng. 132 (12), 121104. Robinson, A., Eastwick, C., Morvan, H.P., 2010b. Further Computational Investigations Into Aero-engine Bearing Chamber Off-take Flows. Gas Turbine Technical Congress and Exposition, Glasgow, UK (p.^pp.). Robinson, A., Bryden, I., Ingram, D., Bruce, T., 2014a. The use of conditioned axial flow impellers to generate a current in test tanks. Ocean Eng. 75, 37–45. Robinson, A., Richon, J.-B., Bryden, I., Bruce, T., Ingram, D., 2014b. Vertical mixing layer development. Eur. J. Mech. - B/Fluids 43, 76–84. Robinson, A., Ingram, D., Bryden, I., Bruce, T., 2014n. The generation of 3d flows in a combined current and wave tank. J. Ocean Eng. Submitted for publication. Rusello, P.J., Lohrmann, A., Siegel, E., Maddux, T., 2006. Improvements in Acoustic Doppler Velocimetery. Proceedings of the Seventh International Conference on Hydroscience and Engineering, Philadelphia, PA (p.^pp). Salter, S.H., 1981. Absorbing Wave-makers and Wide tanks. Directional Wave Spectra Applications, Berkeley,pp. 185–202.
129
Salter, S.H., 2001. Proposals for a Combined Wave and Current Tank with Independent 360° Capability. Proceedings Marec 2001, 2-day conference on marine renewable energies, Newcastle, UK (p.^pp.). Salter, S.H., 2003. Design and Construction of a 360-degree Flow Table with Control of Velocity Gradient, IGR Report to EPSRC GR/R20694/01. Shields, M.A., Dillon, L.J., Woolf, D.K., Ford, A.T., 2009. Strategic priorities for assessing ecological impacts of marine renewable energy devices in the Pentland Firth (Scotland, UK). Mar. Policy 33 (4), 635–642. Taylor, J.R.M., Rea, M., D.J., R., 2003. The Edinburgh Curved Tank. 5th European Wave Energy Conference, Cork, Ireland, pp. 307–314. White, F.M., 1994. Fluid Mechanics. McGraw-hill. Wolf, J., Prandle, D., 1999. Some observations of wave–current interaction. Coast. Eng. 37 (3), 471–485. Yakhot, V., Orszag, S.A., Thangam, S., Gatski, T.B., C.G., S., 1992. Development of turbulence models for shear flows by a double expansion technique. Phys. Fluids 4 (7), 1510–1520. Yao, A., Wu, C.H., 2005. An automated image-based technique for tracking sequential surface wave profiles. Ocean Eng. 32 (2), 157–173.