Masters of Engineering Research Project
From January 2014 to May 2016 I worked as a researcher under Professor Charles Williamson and PhD student Riley Schutt in Cornell's Fluid Dynamics Research Laboratory, on a project focused on understanding the fluid dynamics associated with Olympic level sailing. In high level sailing events, in small boats, sailors use their bodyweight to induce several types of dynamic motions in their sails, creating fanning or heaving motions that are known to improve performance. Understanding the fluid dynamics behind these effective motions is the focus of this research project. I was involved in several different aspects of this research, including generating CAD models of existing experimental equipment, like the towing tank shown in Figure 2, conducting on the water experiments (Figure 1), processing experimental data, and designing and building experimental equipment. I also presented a talk on the research at the 2015 meeting of The American Physical Society, Division of Fluid Mechanics. In my 5th year at Cornell I completed my 8 credit Masters of Engineering project designing and building a rotational control system (Figure 3) for the FRRL xyz-towing tank (Figure 2). The original towing tank could drag a testing body through water, following a precise motion profile in x, y, and z, while measuring forces on the body, or using particle image velocimetry to analyze flow characteristics. This system was used by the project to conduct controlled laboratory experiments using approximated 2D motions to compliment data captured in on-the-water testing, but was unable to properly model one of the motions we were interested in, specifically the roll tack (Figure 6). The following section gives an overview of my MEng. Project design and execution.
Vortex Dynamics of Olympic Sailing
Figure 1: Laser Sailboat used for on the water testing. This laser was sailed by top level collegiate sailors, and equipped with an array of sensors including two inertial measurement and GPS units, five cameras for recording the sail shape, and an anemometer. The array of dots on the sail was used along with the array of cameras to analyze the curvature of the sail.
Figure 7: The Rotational Stage Installed in the Real Tank under the linear variable displacement transducer in the actual tank. The protractor and indicator are clearly visible (printed in a white plastic) at the bottom of the picture.
Figure 5: Labeled Rotational Stage, showing all major components.
Figure 4: Towing Tank with Labeled Components
1. Steel Tank Frame
2. X-Axis control motor and block assembly
3. Force Sensor - Linear Variable Displacement Transducer (LVDT)
4. "Carriage" - rolls on rails in the x-axis direction
5. Y-Axis Control actuator
6. Z-Axis Control actuator
Figure 6: Roll Tack. Starting from top left (1.) to bottom right (4.), this sequence shows the major steps of a roll tack. Step 1, the boat is sailing at the minimum possible angle with respect to the wind and her sailors decide to initiate a tack. Step 2, the sailors steer the boat through the "no-go zone" created by the wind. Step 3, the boat has passed out of the "no-go zone" and is now at a minimum sailable angle on the other side of the wind, where the sailors use there bodyweight to lean the boat over aggressively. Step 4: the sailors flatten the boat quickly and accelerate away on their new angle. Note, in this scene the wind is coming directly toward the viewer, and that the boat is a "Club 420," which is a training boat for the Olympic class the 470.
FIgure 2: The FDRL Towing Tank was used to conduct laboratory experiments based on motions captured in on-the-water experiments using the setup displayed in Figure 1. This image is a color coded rendering made with a CAD model based on the actual tank that I made with another student in 2014. The original tank allows the user to move a testing body (in this case a model sail) through the water over a precisely controlled profile in z, y and y, while measuring forces on the body with an LVDT (Linear Variable Displacement Transducer) and/or using flow visualization techniques to record flow characteristics. The original system did not allow for movements involving rotation, which were required to understand the motions we were interested in.
Figure 3: The Towing Tank Rotational Control Stage, was designed to be 3D printed and installed in the towing tank shown in Figure 2. This retrofit allowed the towing tank to measure the effects of rotational motions of the testing body. This rendering shows the different components in different (unrealistic) colors to make distinguishing the elements easier. The design uses a high precision digital servo connected with a linkage to hydrofoil, which is, in this case, a NACA 0012 profile, used for baseline testing. In other tests we used a foil based on the sail shape captured in on-the-water testing. Figure 5 shows the different components labeled.
Rotational Control Design
I set out to design the system so that it could be built inexpensively and would fit into the footprint of the existing LVDT (linear variable displacement transducer) base. The group settled on using a high performance digital servo designed for model airplanes, for its performance characteristics and small size. Coupling a sail shape directly to the servo was a problem, so, instead, the design offsets the shaft and uses a small linkage to connect the servo to the sail. I went through several design iterations before arriving at the design shown in Figure 3. Figure 5 shows the same design with labeled parts. This system is designed to fit into the LVDT, which is shown in yellow in Figure 4. The design includes an angle indicator for a human operator, which was useful for debugging problems, and a torsion spring on the shaft, which was used to reduce the effect of backlash in the actuator. Most of the components are designed to be 3D printed, including the LVDT base, the linkage components, and the hydrofoil, so there was no need to design them to be easy to machine. Designing the system to be 3D printed in a strong plastic also allowed us to keep weight low, which is important because the LVDT is essentially a spring mass system. This design iteration met all of the design criteria and was successfully used in several experiments. Figure 7 shows the system installed in the actual tank.
Design Motivation and Specifications
The project had previously used the towing tank shown in Figure 2 to run experiments on several different sailing motions that did not require a rotational motion, but in order to study roll tack maneuvers rotation is required. In sailing, it is impossible to sail directly into the wind, or in into the range of angles approximately 45 degrees on either side. This zone is typically called the "no-go zone" by sailors. In order to travel in this direction sailors will sail at an angle as close as possible to the wind for some time and then "tack" turning the boat to the closest possible heading on the other side of the no-go zone. To put it another way, tacking is a turning maneuver in which the boat moves from a heading around 45 degrees to one side of the wind vector, to a heading about 45 degrees on the other side of the wind vector. Tacking is typically slow, so racing sailors use a maneuver called a roll tack, where they aggressively pitch and flatten the boat during the turn to maintain or temporarily increase speed. See Figure 6 for a visual demonstration of a roll tack. We conducted several sets of on-ther-water experiments to capture a motion profile for a roll tack, but the motions are impossible to study in a laboratory setting without rotating the model sail. In order to achieve this rotation we decided to augment the existing towing tank. This rotational system had several requirements. First, it needed to fit into the existing system, which imposed a tight size requirement. Second, the rotational system needed to be controlled with the same lab view control program which coordinates the motion of the rest of the tank. Third, the rotational system needed to be powerful and accurate enough to turn a hydrofoil in water and accurately represent the motions we were commanding to within 0.5 degrees. Finally, the system had a total budget of $500, meaning that we would have to build the system without purchasing any custom components.
In order to control the rotational actuator accurately I set up an external servo control board, which accepts serial commands and converts them to the 1-2ms pulse format used by the hobby style servos. I heavily modified the original LabView program to generate and send serial commands with precise timing. Accuracy tests revealed that the movements of the rotational actuator and the tanks linear actuators were synchronized to within 0.02 seconds, which was accurate enough, considering that tacking maneuvers typically occur slowly, over several seconds.