Submersible Drone - Basic Concepts
Christopher Clayton
03/29/2023
I first started to think about drones on a grand scale of moving people through cities via flight, and in the context of the project management & business resources it would take to start such a project.
I am not an engineer, but I came around to returning to drone ideas from a practical design standpoint after receiving much additional tutoring in mathematics since then.
This iteration is an attempt at envisioning the requirements for a relatively small (camera payload) exploratory drone for undersea use. The simple ogive (bullet-shaped) nose as the initial conception for the front is to mimic the same contour as airplane noses, pending material choice and testing (smooth flow through a fluid with minimal drag). The nose in this context would need to be a transparent material for a spherical camera and that is where in practice joining different materials together (main body versus transparent front) becomes most critical to prevent failure under pressure. The rear is a parabolic side profile terminating in a logistc curve-shaped tail to mimick the tail of surface ships; an intent to maximally vent flow through the back at minimal drag as the vessel propels forward.
A curvature transition would be needed to put these two pieces (front and rear) together, which I haven't come up with. Ideally, the ogive-shaped front on the bottom would quickly transition to the cubic front profile of the rear to form the middle. Nacelles need to be added for the propeller housing. For the internals, I imagine the entire drone body to be hollow, with hollow stringers (beams) filled with ball bearings running at intervals, with each one fully looping across the width of the body, to resist the pressure of fluid from all sides (again pending testing to see if such a concept would work and to what depth).
The goal of such a design would be for deep exploration on the scale of full-size submarines (thousands of meters), whereas many consumer submersible drones that I have seen on the market from online searches are rated for a few hundred meters.
CAD modeling - 4/8/2024
Modeling this cone and tail out in ViaCAD, I started by laying out the wireframe using profile screenshots. This has been a big change to how I once created models using solid primitives in ZModeler 1 for video game modifications. In this case because of the complexity of the wireframes, I used surface covers. Because this model frequently involved curvatures in all three dimensions, and ViaCAD's surface cover algorithm attempts to account for them with multiple straight-edged, overlapping meshes, which then didn't always smoothly reconcile in the same way as Mathematica's formula-based generation, sometimes I did not utilize a particular spline. Then this mostly limited a complex shape, such as the tail, to a complex curvature in two of the three two-dimensional directions.
In some cases, I made a spline to edge surface cover in the third two-dimensional direction to create a smoother transition. Order of selection matters in these cover generations as to how the cover will initially flow, and then further how overlapping meshes will be generated (in cases of selecting more than two elements), whether spline to spline or spline to edge operations.
The parabolic nose concept I envisioned in Mathematica, terminating roughly in an ellipse, was eventually made to terminate into separate conics rather than one ellipse, corresponding to a matching number of splines at the tail. This allowed the surface cover generation for the middle of the drone to be relatively straight across the Y-axis (length), with a smooth curvature transition in the other directions.
Then I exported as STL format and repaired with Cotangent. I ended up doing an edge connection repair, followed by auto repair to entirely make sure the mesh had no holes, and finally conversion to a morphology in order for the complex curvatures to be converted into a solid properly (not as smooth of a curvature in some places as I expected). Then that allowed proper slicing for 3D printing gcode generation.
The idea after testing it in solid form is to make it hollow for the engines, stringers, and payload (i.e. a camera) to be installed.
For strength, the next test iteration in my concept is to make a series of molds for a given target drone size by inverting the model and dividing into pieces with a rectangular base for each mold modeled on the bottom of each inverted partial mesh. Then in this plan, I'd print the molds in a relatively low-strength plastic (size of prints limited by the 3D printer, which is why I envision it needing to be a multi-mold print). Then use strips of pre-produced carbon fiber or some other polymer that can be chemically cured to be set in the entire fastened-together mold. This would be repeated to represent the other half of the basic test object, followed by more strips and curative to adhere the halves together.
I envision this methodology to avoid needing a furnace or other high-heat method of production from a mold for small-scale, at-home samples.
Protype solid miniatures - May 2024
After finding a working setup with the Creality Ender 1 machine (dry glue on a glass bed), where the model processed properly in Creality's slicing software, I printed a test copy in PLA (polylactic acid). This turned out to be too top-heavy to properly float. The next iteration involved halving the maximum height to roughly match the maximum width of the body. This floated but not in the direction I expected. The third image is the orientation in which it floated in shallow water.
