The wave tank located in the Aquaculture Center at the University of Maine is where all testing was carried out.  The wave tank was used to create a laminar flow through a scaled model of a tidal turbine by attaching the Turbine Test Bed to a carriage, which traveled down the length of the wave tank at a controllable speed.  From these short time intervals, data was produced which showed turbine characteristics. 

    Multiple data runs were performed for speeds ranging from 1 to 2 meters per second; possible flow rates for different tidal sites along Maine's coast. The speed of the carriage was recorded using two stopwatches and dividing the recorded time by the distance the carriage covered once steady state speed had been reached

    Data was gathered at three different test speeds of 3.00, 4.53, and 6.00 feet per second. for each speed, six runs were conducted at preloaded torsional resistances. The pancake motor created the pre-load using a variable resistor connected to a power supply, which varied voltage and therefore torsional resistance. The torsional resistance was varied to develop the optimum torque in the shaft in order to achieve the maximum power produced by the turbine.

        During each data run shaft torque, thrust, and rpm were measured several times a second using a reaction-torque load-cell, load-cell on shaft axis, and encoder respectively.  An encoder based carriage velocity sensor was designed into the test bed system however, was not used for this study.  A lab-view program then stored the data collected by the various measuring instruments.  The data from each data run was then converted into a useful data point by taking an average of the steady state data from the middle of the run.  Non-dimensional analysis was then performed to create operating characteristic charts for the tested turbine

CALIBRATION

The calibration of both load cells was conducted on the same day testing was preformed. Note the importance of calibration for each test session due to the harsh environment the tow tank is subjected too.

Calibration of the frontal load cell was preformed by running thin rope attached to one end of the load cell over a low friction sleeve covering a bracket to a vertical weight hanger. The gimble load cell was calibrated by hanging a thin rope perpendicular to the motor from the same radius the load cell was connected to the motor. The calibration range for both load cells was zero to five pounds. The weights were calibrated on a digital scale for accuracy. 

RESULTS

Raw Data 

The figure to the right shows the raw data from a single tow tank run, showing the RPM and torques signals. The RPM data was used to determine steady state speed. The data over this steady region was averaged to get the mean rotational speed and torque for each data run.

Dimensional Analysis

The figure on the right shows the dimensional performance data curves for each of the three test speeds. The mean torque, rpm and carriage velocity data for all runs are depicted.

Non-Dimensional Analysis

The turbine performance data was non-dimensionalized using the following relations:

Tip Speed Ratio: 

 Torque Coefficient:      

Thrust Coefficient: 

The figure below shows the non-dimensional performance of the turbine at the three test speeds. Note that the curves collapse to a single curve for the middle and higher speeds, indicating that the test apparatus was properly measuring the turbine performance. The lower speed data may have suffered from the lower measurement forces due to the low speed, but the number of data points collected was insufficient to make any concrete conclusions.  The data showed that the loading resolution of the device at higher preloaded torques needs to be improved to capture the lower speed, higher load region of the power curve.

Discussion

A test bed for the testing of scale model tidal turbines was developed and implemented in the University of Maine tow tank.  A sample turbine was tested which successfully demonstrated the test beds measurement capabilities.  Data was collected over a range of three speeds and it was shown that the non-dimensionalized power curves collapsed to a single performance curve as expected. The uncertainty in the power, power coefficient, and tip speed ratio is 3.01%, +/- 3.8, and 3.8% respectively. The power coefficient had an uncertainty of plus or minus 3.8 directly.  No percentage of actual value need be found.  This percentage represents a direct value of turbine efficiency. These values were used to validate the performance of the test bed measurement system.