Below is a paper I wrote to compile the methods and progress of one of my independent research projects. I've had to solve a lot of unique problems and adapt to some pretty interesting workflows, so take a look if you're interested- otherwise, farther down the page is a more concise summary of the actual workflow itself without in depth explanations of the underlying techniques.
Like all good things, many of my flying designs begin in various aerodynamic simulation software. The two that I've found to be the most accessible are OpenVSP and XFR5. Both are open source, and both provide parametric aircraft design and aerodynamic approximation. This results in a great jumping off point for me where I can vet projects and quickly go through various design iterations based on readily accessible data. XFLR5 is particularly nice because you can work pretty closely with your aerofoils and hone in on chord point and therefore get a better idea of weight distribution and structure, which brings me to the next phase of the workflow.
The first and primary step of converting aerodynamically designed geometry is taking the dimensions and remeshing. There's a fundamental difference in the way models are generated in prior mentioned aero software vs. typical CAD: in NURBS style editing, i.e. AutoDesk Inventor, geometry is calculated via mainly booleans and extrusions, and therefore point and triangle based meshes don't work. Hence the first step: replicating all of the dimensions used in the aero simulation, and then slicing the model up into digestible bits that can be individually focused on.
I need to slice up my models into parts because of the constraints of my 3D printer build volume (220mm in the xy, 240mm in the z). Once done, I assemble everything back into a single file and begin making sure the whole thing actually fits together. This includes wing spar locations, wire holes, mechanical aspects like control surfaces, and most importantly for my modular wing design, structurally sound screw in mounts for every single part. Most (albeit limited sample size) 3D printed airplanes throw away a huge advantage of 3D printers (versatility, ability to quickly modify parts) by making every component glue together. I've tried to avoid this in as many places as possible.
Once I've completely 3D modelled the structure and sliced the vehicle into properly sized parts, I can begin optimizing the GCODE. This includes 3D printing infill, width, speed, and a whole slew of other settings that I tweak strategically for the best results. In order to make my models compatible with as many GCODE slicers as possible, my planes typically rely heavily on this phase, because the internal structure of the wing is generated here. I use a cubic subdivision mesh, because my tests have revealed that it offers the best strength in the lateral directions I'm worried about.
After slicing up the GCODE, the printing process can begin. With prints properly scheduled, the printers can be running 24 hours a day and output a significant amount of parts. Seen here is the printing process of two mirror base wing sections. These hold the linkages that join the wing and the body as well as the modular attachment points that join the outer sections.
Once all parts are printed, assembly can begin, and all electronics can be wired up. The plane on the left is an example of the result of the cumulative work outlined in this design process. In this image the plane is mounted on the pneumatic launcher I designed, however, that's a whole separate topic.