Project goals:
  • Learn how to operate Ansys Fluent and setup complex problems.
  • A huge part of this project is to learn Siemens NX. Siemens NX is much more powerful than Solidworks in how explicit you have to be. I hope to be able to master it while I am still in school.
  • I am learning about turbo pumps in my free time so that I can hopefully work on them and design them.
  • Currently reading Modern Engineering for Design of Liquid-Propellant Rocket Engines to gain knowledge of turbo pumps and apply to my models.
Design Process:
  • I took inspiration for this design from Garrett Turbos. Although I don't have one to exactly measure so I am basing it off of features I see on their turbos.
  • I decided to do dimensions that I felt were roughly correct because I will be running the design through Ansys Fluent so I can iteratively refine the geometry.
  • The hardest part of running Turbo machinery through Fluent is how dense the mesh needs to be.
  • Early into this project I realized I couldn't run the model on my computer setup. So saved up my money and built out a computer that I knew would be perfect for simulating my turbo and future turbo pumps.
  • Computer specs: 5950x, 2080ti, 64 GB of Samsung B-die ram, Asus DarkHero mother board, and a full custom water cooling loop to ensure the computer stays cool while it runs simulations overnight.
  • I started with just the compressor wheel. I knew if could just make a nozzle to go over the blades that I knew that what I was doing was in the correct direction.
  • I even ended up 3D printing a version of the model at Launcher Space on the Formlabs 3L because my boss thought it was cool and he wanted to see what the new printer could do.
  • I started by creating a fluid domain in NX, and working around that. I then imported the model as a Step file into Ansys Spaceclaim. I sought out creating a mesh by tweaking the size of the tetrahedral elements on each surface. Final mesh was around 3.5 million nodes and 15 million elements. This took awhile because the mesh directly effects the convergence and the accuracy and I wanted the mesh to be as dense as I can get it. Next was setting up the Transient rotating mesh motion simulation using the k-epsilon model.
  • Success! Output. This simulation really doesn't say anything but I wanted to make sure that I could get anything before moving forward with more of my time.
  • So I spent more time creating a housing for the Compressor wheel! Then I ran the simulation using the same parameters and setting up boundary conditions and got output! I intend to backup what I am seeing with hand calculations and empirical data after I flush out the design some more.
  • From this data I saw that rotating the compressor wheel at only 15000 rpm wasn't enough to generate much pressure or mass flow. But at that rotational speed the air's max velocity is only around 100meters per second. Which means that I am only at a mach number of .29. Which furthermore means that I am not in compressible flow yet. So I will run it again at a higher speed when I don't have exams... TBC