CAT 988K Tilt Arm
The project was to create a bottom-up model of a product of our choice and perform shape optimization on the geometry for the MECH 465 class. For this project I selected a tilt link from a CAT 988k where the objective function was to minimize the volume which translates to the minimizing mass. The load applied was 130kN acting on the bucket end of the arm at 45 degrees. In reality this arm is usually encountering nonlinear loading at a range of angles however that is out of scope of this course.
Note that this project was done in ANSYS APDL and while it was a good to understand the backbone of ANSYS workbench, the use of APDL is obsolete and workbench should have been used.
Model Simplifications
Notably, the course requirement was to make this as a bottom up model, the model had to be simplified otherwise this project would not be completed in the course timeline. All fillets, chamfers and rounds where the hydraulic cylinders attach were removed.
Volume Mesh - Total 136 elements,
"Laptop" Mesh - 424 Elements
Optimization Mesh - 78 684 Elements
Mesh
Above you can see the volumes that were made of the model. It should be noted that this part is symmetric and was constrained for all nodes on the back face. The mesh that was operated on my laptop only used 424 elements however the mesh was parametrized such that it could be scaled to 78 684 elements which was was used for analysis and the optimization.
Results
As seen in Model Simplification, I started with a simplified CAD model and created 23 unique design vectors that would test iterate the optimization to determine the best result. It should be noted that each optimization run only generally ran 9 iterations before being caught up on local minimums even with a 1% tolerance.
von Misses Stress
Deflection
Errors
It should be noted an error was made and the position of the inner side of the lug relative to the symmetric face was not defined and thus the part got skinnier without consideration for the hydraulic cylinder that would normally sit there. This is an inaccuracy however due to the course structure and timeline this change was not possible before the course completion.
On the top face of the part, there is a stress riser as a result of a sharp incline. This error is a result of that point not being considered a design variable. In the future this can be elevated by making it a design variable or by adding a fillet for the final design.
Again, out of the scope of the project, a mesh convergence test would be beneficial in determining the ideal number of elements to reduce compute time.
Moving forward
On this project, overall I was able to achieve 95%. However it would be interesting to perform a topology study with the proper loading in the future.