3MEMS3: Topology Optimisation — Stabiliser pt2
As outlined in my previous robot blog on the topic of the testing and calculation I did, I must add an additional stabiliser to the robot. I like a challenge and as a mechanical design engineer making a simple structural device like this is right up my alley, so no bought-in solutions here!
When you design anything consideration must be made for the method of manufacturing you are using. As I am working on this project from home with no workshop space it was looking like the design was going to be way too simple for my interest. However with desktop 3D printers now common both in college and among some friends I decided my stabiliser design would be optimised for additive manufacturing.
The nature of additive manufacturing allows for the creation of highly complex geometries that would not be otherwise possible using traditional manufacturing. Additive manufacturing perfectly accompanies many algorithmic design optimisation tools as the raw most optimised result can be made and used without the need for tedious and difficult post-processing in CAD.
For my robot stabiliser, I decided to use a process called topology optimisation. Topology optimisation (top-op) is a discrete subtractive process that builds on the finite element method of structural analysis. It is used to achieve ideal strength to stiffness ratios based on applied boundary conditions.
For this project, I decided to use nTop Platform, a design and simulation tool developed by nTopology. The ease of use and end result from nTop has suited my work better than similar tools from Ansys or Solidworks.
To summarise top-op simply, the part is split up into evenly sized separate elements or chunks. These elements are then ranked based on the contribution they are making to the structure of the part. The lowest contributing elements are then deleted from the part.
Topology Optimisation Steps:
Step 1: Blank
First, you must make a blank part, usually very blocky in shape is a cad file that fills all of the available space which the final part could use. In my case, this is under the robot between and around the motors. There only needs to be detail in the connection regions (where the part will be fastened to the parts).
Step 2: FE Model
We must build a finite element (FE) model of this blank for the computer to work with. A material is assigned to the part and a mesh is applied. The mesh splits the working domain into a discrete number of cells. the quality of this mesh can have a great effect on the end result of the simulation. Cell size should be as near uniform as possible. There are many types of meshes that can be applied, some better than others but top-op is only compatible with a tetrahedral volume mesh.
Step 3: Boundary Conditions
To finish the FE Model the boundary condition of the part must be applied. these are the forces which the part is under as well as the regions that are fastened or supported. In this case, I added the force translated through the mount under max mass distribution on both y and z-axis as well as the connection regions which in this instance are fixed faces (many beginners often wrongfully use fixtures in FEM but they have applied correctly in this case).
Step 4: Topology Optimisation
Using the FE model for the blank with the topology optimisation tool we can remove material. For this part, I set the top-op condition so the end part would be 0.11 of the mass of the original. I started higher at 0.3 but it was clear more could be removed without sacrificing stiffness. the raw result shown has a very rough surface and lack some needed detail for connections that needs to be addressed.
Step 5: Post Processing
Rough surface lost details and sharp edges commonly occur on a raw top-op part. A smoothening tool was used in nTop to level out the surface, a subsection of this tool allows a minimum radius to be applied to the optimised geometry removing any locations of possible stress concentration.
The connection faces are marked as named selections from the original blank model. Material is added around these named selections by thickening the faces, this thick material is merged with the top-op’d part. A boolean intersect between 2 implicit bodies (the blank and the optimised part) is performed to restore the original geometry of the mounts. this allows for accurate hole size and flat faces for any bolts to sit properly.
Step 6: Validation
It is essential to run some form of structural analysis on the optimised part to ensure that the optimisation parameters were not too aggressive for the desired application. Stress and deflection are evaluated to ensure proper rigidity. I used multiple mesh element sizes to ensure mesh independence as best as possible.
All of the boundary conditions applied in both the top-op and the static analysis are very much worst-case scenario with a high factor of safety. the material properties of PLA plastic were uploaded to nTop however due to the layering of 3D printing the end part will not be isotropic. This has been accounted for however in the positioning of the part on the print bed, the layers will always be in compression.
The final optimised part is then re-meshed and exported as a CAD file. This allows me to add it to my robot assembly in Solidworks and very it all fits together before printing!
As expected the mount fits up perfectly and I think it looks great. Given more time I would have loved to redesign the packaging of the robot to make it a little more stable when self balancing but 3rd yr end of the semester is near and Formula Student manufacturing is ramping up so maybe over the summer I’ll start a new robotics project.
In my last blog post I calculated the mass transfer and load which the stabiliser will be under, do note I included and rechecked the mass and COG position of the robot with the stabiliser included.
So there it is my first designed parts properly unveiled. As this blog is posted the parts are finished on the 3d printer ready for collection!
Check my socials tomorrow for full images and installation.
