Control strategy for autonomous remediation of marine oil spills
Becker, Christopher J.
MetadataShow full item record
This thesis presents a novel formulation of a gradient-type controller to address the problem of cleaning up marine oil spills. Little work appears to have been done in developing autonomous oil spill clean-up devices, with most research efforts directed toward developing improved oil collection strategies. It does not appear that previous work in this field has included development of control algorithms specific to addressing the problem of deployment strategies for multiple clean-up devices. This thesis provides a framework for deployment of multiple clean-up agents and makes the following contributions to the field. We first develop a mathematical representation for the effect of a clean-up agent as a line-sink and introduce this term into an existing oil spill spreading model. The augmented oil spill spreading model is simulated for a finite volume of oil released within a region Q' which contains multiple clean-up agents. Second, we use the augmented oil spreading model to develop a cost function and derive a gradient controller that seeks to maximize the oil removal rate for a system of N clean-up agents. Several key properties of the controller are presented. Finally, we demonstrate the effectiveness of our controller through a MATLAB simulation. The performance of the controlled agents, measured by the total volume of oil removed over the simulation, is compared to the performance of static and randomly moving clean-up agents. The results from MATLAB simulations presented in this thesis demonstrate that the proposed control strategy is more effective at removing oil than static or randomly moving agents. The formulation of the control law directs clean-up devices toward areas in Q' experiencing the greatest volumetric change in oil, thereby maximizing the volume of oil that is removed by each agent. The controller presented in this thesis is adaptable to a range of clean-up devices and we present several future research avenues that could be pursued to further develop this concept.
Thesis (M.S.)--Boston University