The 2012 - 2013 group was able to collect rough data, but there was an inexcusable amount of noise (red). After seeing how close that group came to acquiring good data, we planned to build off their work by making small modifications to their prototype.

Unfortunately, we quickly discovered that this would not be possible as the existing prototype was neither robust enough to handle the anticipated testing loads nor adjustable enough to accommodate required sensing equipment. This realization was a critical junction in the project’s progression and led us to completely redesign the prototype.

The primary driving objectives of the project were to collect accurate and repeatable experimental data, characterize the performance of the Walvisstaart system, and use our observations to validate or invalidate an existing computational fluid dynamics (CFD) model. To accomplish these objectives, we sought to measure thrust, side force, shaft torque, and spindle torque values for comparison with those predicted by the previously made CFD model.

In order to obtain these measurements, we determined that three key changes needed to be made to the prior prototype: 1.) Eliminate distortion of the drive shaft and foils 2.) Reduce noise by re-positioning the motor and 3.) Rebuild the frame out of more sturdy, reliable materials.

Our analysis of the prior prototype revealed that the drive shaft bent when relatively light loads were applied to the bottom of the foils. We were concerned about the bending because the foils must sustain high loads while running in the water. Therefore, we quantified how much a foil deflected as a function of applied force to the foil. The foil deflection was due to both the drive shaft bending and the individual foils bending.

To eliminate deflection of the drive shaft and foils, I performed finite element analysis (FEA) on the model and determined that increasing the diameter of the drive shaft from 0.5 inches to 2.5 inches would solve the problem. We modified the design of the prototype accordingly, and this modification alone enhanced the quality of the data significantly. 

To further optimize the data acquired by the system, we did the following:

1.  To reduce the noise in our data the motor was moved from the internal frame to the external frame and connected to the main drive shaft. A double U-joint was purchased and inserted, allowing for the desired freedom of movement.

2. The previous prototype was constructed using Unistrut, a metal framing system designed to eliminate welding and drilling. Testing found that using the Unistrut system led to a frame that suffered greatly from non-right angles and vibration due to a lack of proper bracing. Various alternatives were explored, including welding a frame from steel or aluminum. In the end, a frame constructed of 80/20 was settled upon for its ease and quickness of construction, availability, and versatility in case of later designs.

We fabricated the system completely in-house with the majority of the 71 machined parts made during the first five weeks. I personally was in responsible for creating the overall CAD model of the system, and the CAD models and part drawings for all of the individual parts.

Fabrication challenges existed for the large parts associated with the two inch drive shaft redesign as well as working with hardened steel. For example, to bore set screw holes, electrical discharge machining (EDM) was necessary instead of a conventional mill. Additionally, specialized collets were made to machine the large concentric cylinders.

For testing of the system, eleven measurements were made simultaneously using three LabQuests connected to a computer with Logger Pro. This schematic shows how all of the necessary measurements were made.

Once the machine was assembled and tested briefly in-house, we went to the Jere A. Chase Ocean Engineering Laboratory at the University of New Hampshire (UNH). During a full week of preparation and testing at UNH, we ran a total of 63 individual experiments.

From our testing at UNH, we found that our design was able to capture the necessary data. This data ultimately paved the way for future research and optimization of the system by substantiating claims of high efficiency and confirming the validity of the computational flow dynamics model.