Department of Ocean and Mechanical Engineering

Wall-bounded turbulent flows are pervasive in numerous physics and engineering applications. Such flows tend to have a strong impact on the design of ships, airplanes and rockets, industrial chemical mixing, wind and hydrokinetic energy, utility infrastructure and innumerable other fields. Understanding and controlling wall bounded turbulence has been a long-pursued endeavor yielding plentiful scientific and engineering discoveries, but there is much that remains unexplained from a fundamental viewpoint. One unexplained phenomenon is the formation and impact of coherent structures like the ejections of slow near-wall fluid into faster moving ow which have been shown to correlate with increases in friction drag. This thesis focuses on recognizing and regulating organized structures within wall-bounded turbulent flows using a variety of machine learning techniques to overcome the nonlinear nature of this phenomenon.
Deep Learning has provided new avenues of analyzing large amounts of data by applying techniques modeled after biological neurons. These techniques allow for the discovery of nonlinear relationships in massive, complex systems like the data found frequently in fluid dynamics simulation. Using a neural network architecture called Convolutional Neural Networks that specializes in uncovering spatial relationships, a network was trained to estimate the relative intensity of ejection structures within turbulent flow simulation without any a priori knowledge of the underlying flow dynamics. To explore the underlying physics that the trained network might reveal, an interpretation technique called Gradient-based Class Activation Mapping was modified to identify salient regions in the flow field which most influenced the trained network to make an accurate estimation of these organized structures. Using various statistical techniques, these salient regions were found to have a high correlation to ejection structures, and to high positive kinetic energy production, low negative production, and low energy dissipation regions within the flow. Additionally, these techniques present a general framework for identifying nonlinear causal structures in general three-dimensional data in any scientific domain where the underlying physics may be unknown.

Navigation of unmanned underwater vehicles in coastal zones, tight spaces and close to structures such as ports, ship hulls and pipelines remains a difficult challenge. Currently Autonomous Underwater Vehicles (AUVs) use a variety of techniques for motion control, including single thrusters with diving planes or hydrofoils, robotic wrists, or a moving mass. However, these techniques provide limited maneuverability. The objective of this work was to understand the mechanics of elongated fin propulsion for swimming and motion control of underwater vehicles. This bio-inspired propulsion is used by several fishes that swim by undulating a thin and elongated median fin that allow them to perform forward and directional maneuvers. In the first chapter we present the literature review as well as the mathematical formulation using thrust vectoring approach to achieve forward and turning maneuvers. In the second chapter, we used a robotic vessel with elongated fin propulsion to determine the thrust scaling and efficiency. Using precise force and swimming kinematics measurements with the robotic vessel, the thrust generated by the undulating fin was found to scale with the square of the relative velocity between the free streaming flow and the wave speed. In addition, a hydrodynamic efficiency is presented based on propulsive force measurements and a model on the power required to oscillate the fin laterally.

The detection of rebar corrosion in reinforced concrete is important due to the high costs of corrosion related damages to infrastructure. One such method of rebar corrosion lies in the use of non-destructive ultrasonic testing. To date, acoustic methods require either the training of an artificial neural network or a theory of acoustic wave propagation. Using a more complete acoustic model such as the Biot-Stoll model avoids algorithm training requirements by directly modeling the acoustic environment. A problem with this method lies in the complexity of the model and the selection of free parameters. The problem of parameter selection is addressed by a series of targeted measurements using ultrasonic transducers on a set of existing reinforced concrete samples placed in a saltwater solution. This data can then be analyzed by a non-linear least squares solver to produce a better fit for the acoustic signal.

The objective of this thesis is to study the proper placement and denoising of Total Field Magnetometers (TFM) installed on an Autonomous Underwater Vehicle (AUV), in support of a long-term goal to perform geophysical navigation based on total field magnetic sensing. This new form of navigation works by using the magnetic field of the Earth as a source of reference to find the desired heading. The primary tools used in this experiment are a REMUS 100 AUV, a QuSpin scalar magnetometer, and a TwinLeaf vector magnetometer. The Earth’s magnetic field was measured over periods of several hours to determine the range of values it provides under natural conditions. Digital filters were created to digitally reduce fluctuations caused by sources of external interference and sources of internal interference. To mitigate the issue of platform based interference, two methods were examined. These methods involved the use of the Tolles-Lawson model and Wavelet Multiresolution Analysis. The Tolles-Lawson model is used to determine the compensation coefficients from a calibration mission to mitigate the effects from the permanently detected magnetic field, the induced magnetic field, eddy currents. and the geomagnetic field. Wavelet multiresolution analysis follows the same basic steps as Fourier transformations and is used to analyze time series with power sources in motion over a frequency spectrum. Several acquisitions were run with the QuSpin in various locations around and along REMUS, and it was concluded that placing the sensor at the very front of the vessel which is approximately 1.8 [m] from the DC motor, with assistance from wavelet analysis was acceptable for the project.

Turbulent flow is a complex three dimensional system of velocity and pressure fluctuations in a fluid that creates vorticity, eddies and other flow structures. In this study we are specifically concerned with the surface pressure fluctuations below a turbulent boundary layer which is one of the primary sources of panel vibration on aircraft fuselages and ship hulls as well a major issue in ship hydrodynamics. The most accepted analytical approaches to describe the surface pressure fluctuations are the Chase model [1] for the surface pressure wavenumber spectrum and Goody’s model [2] for the pressure spectrum at a point. The most accurate numerical approach to use is Direct Numerical Simulations (DNS) [3]. In this study we compared Chase and Goody’s models against DNS of a turbulent channel flow in the space–time and wavenumber-frequency domains and estimated regions of convergence between the analytical models and the DNS data.

The aim of this thesis project was to design, develop, and test, a continuously variable transmission (CVT)-based power take off (PTO) sub-system, and its controller, for a small scale marine hydrokinetic turbine (MHK) developed for low-speed tidal currents. In this thesis, a CVT based PTO and controller was developed for a predefined MHK and validated through simulations. A testing platform was subsequently developed including an emulation system to replicate the MHK for testing of the coupled MHK/PTO system. Laboratory testing of the emulation system, PTO component efficiencies, and full system with controls was then conducted. The results showed the mechanical PTO design to be a valid solution and the control methods to be marginally stable with adequate power conversion at low-speed current conditions. The results also identified future work in continued controller development, alternate PTO component testing, and continued testing in parallel with that being done on the MHK prototype.

This experiment used different methodologies and comparisons that helped to determine the direction of future research on water-based perception systems for unmanned surface vehicles (USV) platforms. This would be using a stereo-vison based system. Presented in this work is object color and shape classification in the real-time maritime environment. This was coupled with HSV color space that allowed for different thresholds to be identified and detected. The algorithm was then calibrated and executed to configure the depth, color and shape accuracies. The approach entails the characterization of a stereo-vision camera and mount that was designed with 8.5° horizontal viewing increments and mounted on the WAMV.
This characterization has depth, color and shape object detection and its classification. Different shapes and buoys were used to complete the testing with assorted colors and shapes. The main program used was OpenCV which entails Gaussian blurring, Morphological operators and Canny edge detection libraries with a ROS integration. The code focuses on the area size and the number of contours detected on the shape for successes. A summary of what this thesis entails is the installation and characterization of the stereovision system on the WAMV-USV by obtaining specific inputs to the high-level controller.

This thesis presents the development a sliding mode controller and vehicle allocation to control a surface vessel platform within a high degree of accuracy. This is part of ongoing development on the WAMV platform at Florida Atlantic University to improve autonomy in marine systems. By developing models for the untested thrusters currently used, the efficacy of a Sliding Mode Controller is evaluated, and a new control allocation developed based on the gradient descent optimization method is developed to manage the thrusters’ constrained angles of thrust generation. The official simulation for the WAMV platform was then modified to include these aspects and the system was tested under wind conditions and was successful in achieving control to waypoints. The gradient descent optimization used for the control allocation did manage to increase the accuracy of both heading and position of the system at convergence. The sliding mode controller navigated to the desired waypoint however maintained oscillations of cross track that were less then 2m and heading error less 20 degrees.

Hand amputation is a devastating feeling for amputees, and it is lifestyle changing since it is challenging to perform the basic life activities with amputation. Hand amputation means interrupting the closed loop between sensory feedback and motor control. The absence of sensory feedback requires a significant cognitive effort from the amputee to perform basic daily activities with prosthetic hand. Loss of tactile sensations is a major roadblock preventing amputees from multitasking or using the full dexterity of their prosthetic hands. One of the most significant features lacking from commercial prosthetic hands is sensory feedback, according to amputees. Many amputees abandoned their prosthetic devices due to the lack of tactile feedback. In the field of prosthetics, restoring sensory feedback is the most challenging task due to the complexity of integration between the prosthetic and the peripheral nervous system. A prosthetic hand with sensory feedback that imitates the intact hand would improve the lives of millions of amputees worldwide by inducing the prosthetic hand to be a part of the body image and significant impact the control of the prosthetic. To restore the sensory feedback and improve the dexterity for upper limb amputee, multiple components needed to be integrated together to provide the sensory feedback. Tactile sensors are the first components that needed to be integrated into the sensorimotor loop. In this research two tactile sensors were integrated in the sensory feedback loop. The first tactile sensor is BioTac which is a commercially available sensor. The first novel contribution with BioTac is the development of an ANN classifier to detect the direction a grasped object slips in a dexterous robotic hand in real time, and the second novel aspect of this study is the use of slip direction detection for adaptive robotic grasp reflexes. The second tactile sensor is the liquid metal sensor (LMS), this sensor was developed entirely in our lab (BioRobotics lab). The novel contribution for LMS is to detect and prevent slip in real time application, and to recognize different surface features and different sliding speeds.

This research consists of the numerical model development and simulation of two prototype Wave Energy Convertor designs (WECs) across three simulation types. The first design is an oscillating body WEC called the Platypus designed to capture wave energy as three paddle arms actuate over the surface of the waves. The second design is an overtopping type WEC called the ROOWaC which captures and drains entrained water to generate power. Modeling of these systems was conducted using two techniques: the Morison load approach implemented using hydrodynamic response coefficients used to model the Platypus and a boundary element method (BEM) frequency-domain approach to model both WEC designs in the time domain. The BEM models included the development of hydrodynamic response coefficients using a discretized panel mesh of the system for calculation of added mass, excitation, and radiation forces. These three model families provided both performance predictions and power output information to WEC developers that supply important data for future full-scale designs. These models were used to predict power generation estimates for both WECs as follows: the Platypus WEC was predicted to have a maximum efficiency range between 14.5-35% and the ROOWaC WEC was predicted to generate a maximum peak average power of 19 W upon preliminary results.