[jjhritz]

My boss wants to test some turbine designs using OpenFOAM to better refine them before fabricating prototypes. He wants me to do it, but I don't have any familiarity with CFD. My background is computer systems engineering and the closest I've ever been to CFD is University Physics 1 and Ordinary Differential Equations six years ago when I was working on my degree. It's not that I don't want to learn -- I find it rather fascinating and this is one of the few concrete skill-building opportunities this job has given me -- but I'm in way over my head here. I need guidance to get started.

Compounding this matter is the fact that my boss doesn't want to spend any money. He doesn't want to pay to get me a training course, he doesn't want to pay for CFD software, and he doesn't want to pay an outside consultant to do this for us. He also doesn't want me spending a month or more teaching myself with online tutorials, since he wants to start getting prototypes into field tests in the next couple weeks.

Is there some kind of crash-course out there that will teach me how to mesh an STL file, run it through an OpenFOAM simulation with a rotating/mixed frame of reference, then get useful information for calculating power output and efficiency in post-processing, all for the budget of zero dollars and in less than two weeks?

[HPE]

Personally, I have spent a decade of my lifetime on CFD theory and applications, and I still feel like I don't know anything, and everyday I have been evolving to a more ignorant engineer with more time dedicated.

The links below may help (and I hope they may help) , but I would give your boss my middle finger, which would be more informative for his demands.

[jjhritz]

Yeah, one of the first things I told my boss is that CFD is an engineering discipline unto itself with its own specific Master's degrees. My friend just finished his degree in Aerospace Engineering (literally rocket science), so I have passing familiarity with the complexity of it that I tried to impress upon my boss. Like everything else involving reality, it just slid off him like water off a duck.

I'll read through the links and see if I can get something useful out of it. I'm a quick study once I get my feet under me; let's hope that'll happen here.

Believe me, I wish I could flip my boss off -- this is not the first time he's had me reenacting the Charge of the Light Brigade. I've wanted to quit for months, but I'm still a novice engineer and I don't have serious industry contacts, so I haven't been able to find someone else willing to hire me.

[geth03]

hi there,

you get what you pay for. so your boss will get nothing! you need proper education. when i started my first job in january my boss told me he assumes that i will need approx. 2 years to get familiar with openFOAM.

well, i didn't need more than 4 monthes, i am able to modify solver code and also code my own ideas. but you need to know that i could really code stuff with C++ before, i worked in several small but complex projects in my uni times. i also wrote postProcessing libs, which speeds up our analysis. my firm did it with exporting stuff to excel and doing the postProcessing manually, which takes too much time.

it took me about 3 monthes to understand the derivation of the Navier-Stokes-Equations and the other kind of equations (turbulence models, population balance modeling, multiphase flow regarded force interactions ...)

my next goal is to dig down and understand the equation solving procedure with openFOAM, i want to know what openFOAM does behind the scenes.
it will take me several monthes i guess.

oh btw, it took me about 2-4 weeks to produce quality meshes for basic machines with ansys fluent (thats the software we create our meshes with).
meshing is an art on its own.

you could basically use freeCAD (opensource CAD software) to generate your geometry and STL-files and create your mesh with snappyHexMesh (openFOAM mesh utility). but that alone would take 1 week at least.

if you are a novice engineer like me, you should not quit at this stage of your career. corona fucked up the economy, it could be hard to find another job. tell your boss thats not possible to do that in that short amount of time and you need to spend time learning CFD.

[GerhardHolzinger]

There are plenty of attempts on "a beginner's guide for OpenFOAM" around.
So, if you're an absolute beginner, you are the perfect test case.

Somewhat related: XKCD on manuals


As for your problem: This sound's more like a line from marketing material for commercial software, with the exception of the zero dollars part.

to mesh an STL file, run it through an OpenFOAM simulation with a rotating/mixed frame of reference, then get useful information for calculating power output and efficiency in post-processing, all for the budget of zero dollars and in less than two weeks


Just out of curiosity, why did your boss bring up on OpenFOAM? Was it one suggestion among many, or did your boss specifically suggest OpenFOAM?


There's software, which promises to make CFD easily available, e.g. as an add-on module to CAD software. In those cases, a finite element method is used for the CFD simulation. This is un-usuall but not unheard of - part 3 of Zienkiewicz and Taylor's The Finite Element Method deals with fluid dynamics.

However, even with the structural mechanics FE add-ons more widely available with CAD software, you have very little control, and even less of an idea over what happens in the simulation.

[jjhritz]

Originally Posted by GerhardHolzinger

Just out of curiosity, why did your boss bring up on OpenFOAM? Was it one suggestion among many, or did your boss specifically suggest OpenFOAM?

He specifically suggested it, I suspect because it's the one he's familiar with that doesn't have any paid features. For example, SimFlow (a GUI frontend for OpenFOAM) limits the free version to 200k node meshes and 2 CPUs for parallel mesh generation and simulation. However, the line you picked out as sounding like a marketing blurb is really all he wants (with the added step of being able to deform the mesh for fine-tuning the design) and it doesn't sound like he's super picky about how he gets it as long as he doesn't have to pay for it.

Originally Posted by GerhardHolzinger

Somewhat related: XKCD on manuals

That is actually where I usually start learning new software, and about when I got to OpenFOAM's file structure it became less intuitive than I could manage. However, since it obviously had to make sense to somebody at some point -- lots of somebodies, given the project's contributor list -- I started doing researching into what things like "Gauss linear orthogonal Laplacian scheme" are to get a better understanding of what all the dials in the files mean and do and which might be useful. Unfortunately, I have a daily Zoom call with my boss to report on my progress and when I tell him I'm doing research into what the dials do he insists I "don't waste time learning things I don't need to know for this" and "just stick to learning how to make and test a mesh." The fundamental problem with that is I don't know enough to know the difference between what I do need to know and don't need to know.

[geth03]

how to create 3D Geometries and export it to STL-files to create meshes with openfoam just with opensource tools:

freeCAD:
https://wiki.openfoam.com/Collection_by_authors
Stefan Radl has one tutorial on how to export your geometry to STL-files.

snappyHexMesh:
https://wiki.openfoam.com/Meshing
here are good tutorial on how to create meshes,
i think Stefan Radl also has one tutorial on how to mesh the exported
geometry, maybe its the one in which he shows how to export it.

that was how i did learn it.

i think there are other options, free tools, like Salome, which are more
convenient:
https://www.salome-platform.org/

i am not familiar with it, but i think there are tutorials on youtube i guess:
https://www.youtube.com/watch?v=Iw7lqV26tGs

after creating your mesh you need to select a solver which fits your needs.

[jjhritz]

Alright, I'll take a look at those. I'd heard about Salome elsewhere and have been digging into it to figure it out.

Any particular advice for using snappyHexMesh to get a good result on something with sharp edges like a turbine/fan blade? I tried a few different refinement settings in SimFlow just to get an idea of how things changed in a graphical environment and the result always looked like someone had taken chunks out of the turbine blades with rocks.

[geth03]

http://www.wolfdynamics.com/wiki/meshing_OF_SHM.pdf

here is a decent document with snappyhexmesh related stuff.
you can use layers at those edges so no blocks are cut out from
your domain.

[FaridElson]

Originally Posted by jjhritz

My boss wants to test some turbine designs using OpenFOAM to better refine them before fabricating prototypes. He wants me to do it, but I don't have any familiarity with CFD. My background is computer systems engineering and the closest I've ever been to CFD is University Physics 1 and Ordinary Differential Equations six years ago when I was working on my degree. It's not that I don't want to learn -- I find it rather fascinating and this is one of the few concrete skill-building opportunities this job has given me -- but I'm in way over my head here. I need guidance to get started.

Compounding this matter is the fact that my boss doesn't want to spend any money. He doesn't want to pay to get me a training course, he doesn't want to pay for CFD software, and he doesn't want to pay an outside consultant to do this for us. He also doesn't want me spending a month or more teaching myself with online tutorials, since he wants to start getting prototypes into field tests in the next couple weeks. So I even started to think about loans and I even took one from here https://triceloans.com/700-dollar-loan/. Those 700 dollars are not enough for my course but I will try to earn more money to add to that value and buy courses of my dreams. Money is really important if you want to learn something new in this cruel world.

Is there some kind of crash-course out there that will teach me how to mesh an STL file, run it through an OpenFOAM simulation with a rotating/mixed frame of reference, then get useful information for calculating power output and efficiency in post-processing, all for the budget of zero dollars and in less than two weeks?

It is really sad that your boss don't want to pay for your study because It will boost your productivity to the top!