Solidworks' FEA - first section modelled

[boost monkey 0]

So I got a copy of SW2008 Student Designer Kit free from uni today, and between doing real coursework on it i've been playing around with the FEA utilities.

 

I'm hoping to be able to mock up a 16v intake manifold and then model airflow through it in a rudimentary fashion. :cheers:

 

I know some of the guys on here have done similar things for big projects / dissertations, but anyone else just been casual tinkering with it?

 

Are there better FEA systems which are easily accessible/available?

 

J-unit.

[vr6-phil 0]

In my experience SolidWorks is a very nice and fairly straight forward programme to use.

 

It has been a little while since using it and I didn't have to model car parts...I think the last thing I did was a shopping trolley wheel assembly...very interesting. Just keep creating layers and it's easy to plonk parts where you want them!

 

I don't think SW was FEA software when I used it actually...has it developed since 2002?

[boost monkey 0]

I think it must have Phil, the FEA side of things is done by a utility/add-on called CosmosWorks although I haven't got it to work properly yet so will keep tinkering.

[vr6-phil 0]

Can't say I have heard of CosmosWorks so it must be something new since I used it.

 

Good luck with the tinkering then!

 

Post up some images when you have modelled the manifold (or whatever you create) as I would be interested in seeing it.

[Tempest 0]

In my days as a student, my mechanical engineering colleagues (not one myself) used Ansys for FEA. That's all I can remember. Mate of mine modelled a tennis racket using Ansys and the stresses of the strings and racket material (shaft, head, etc.) when playing tennis.

 

Tempest

[delfinis38 1]

Cosmos has been around for a long time in various forms. it's just an add on to the standard SW. I've used it in the past and it's very good.. better than Ansys and Pro Mechanica in my view

[boost monkey 0]

I worked through the tutorial in SW last night just to get used to it.

 

You take a sample-drawn ball valve and model the flow of water through it by setting volume flow rate and a few other variables. Just like with modelling forces, the particles change colour with respect to velocity so you can see where they are moving quickly or slowly: this is theoretically proportional to cross sectional area of the pipe.

 

It can then export all the data into Word; apparently highest particle velocity was 7.5m/s :grin:

 

Shouldn't be too hard to mock up 2 boxes with 4 inlet tracts running between them, and then tweak the dimensions, materials, surface texture as and when I can get the proper data. I have a spare 42mm manifold which I could take some measurements from hopefully.

 

Jon.

 

p.s. I'll try to stick up the ball valve screenshot tonight.

[skinnyman9000 0]

Ive used Cosmosxpress, but not the full cosmosworks. Although i do have SW2008 professional, purchased of course

[boost monkey 0]

I r in ur valvez, finitely analyzing ur elemants! :nuts:

 

[boost monkey 0]

Ok, been a while since I updated this but I developed the first bastardised upper intake today.

 

It's bastardised because:

 

other than the 42mm inlet tracts, none of the dimensions are accurate for a 16v intake,

material it's modelled in is translucent plastic for sake of seeing particle flow - think the manifold is actually magnesium?

no radiused/chamfered/filleted edges = lots of evil 90deg corners for air flow to literally get trapped in.

 

Would be nice to find some actual dimensions for the intake without cutting up my spare if I can help it!

 

I modelled the wall thickness as 5mm, negligible surface friction (i.e. not textured as the real intake is) and the program demanded I had an inlet air pressure higher than the outlet, so I modelled 2bar inlet (i.e. 1 bar boost) and outlet at 1bar which is not particularly accurate but for a first run it will do.

 

Unfortunately didn't work out how to use multiple outlets so I had to each tract seperately which is time-consuming as the modelling takes about 3-5 mins on my AMD 2.4 with 1Gb.

 

Here are the pics:

 

 

Internal flow volume of my model. This is what the FEA uses when working out where each particle will go. I modelled 50 particles.

 

 

Manifold changed into translucent plastic so that flow could be observed. Arrows show inlet at the end of the plenum and outlet at the end of tract 1. You have to "close" the system by plugging up the holes for the analysis to work.

 

 

Inlet Tract Flowrates

 

Cylinder 1 - Max velocity 297.352 m/s

 

 

 

Cylinder 2 - Max velocity 306.398 m/s

 

 

 

Cylinder 3 - Max velocity 309.715 m/s

 

 

 

Cylinder 4 - Max velocity 297.933 m/s

 

 

 

Obviously, flow data could change when the manifold is correctly scaled (might even be able to model the lower part too) but it's a good base to continue on from I think.

 

Any comments / criticisms welcome :salute:

[boost monkey 0]

Air intake temp was 293K or 20degrees Celcius.

[Dec 1]

Looking good! Never used that package myself, but I've been working on CFD / Aero stuff like that for the last 6yrs'ish now.

 

You could do some quite interesting things with modelling the four outlet pipes! In theory you could apply an outlet pressure for each runner that varies with time, to approximately model the effects of the valves opening + closing and then run a time dependent simulation to see how the flow varies through one complet engine cycle.

 

3D CFD, like modelling an inlet is notoriously processor hungry though, especially to get accurate results. As an example, I modelled a rectangular box with a small inlet jet flowing into one side of it (quite similar to the top box of the inlet mani you have there). In order to get 100% accurate results it took 8 days running on 2048 processors of a large parallel supercomputer, and that was even with a super efficient code developed just for modelling this jet!!

[boost monkey 0]

:shock: :shock: :shock: :shock: woah! that's insane.

 

variable outlet pressure with respect to time would be AMAZING! definitely final year project material I think :salute: Man i'd love to get a first...

 

Really need to scale this stupid thing to make it look less like a 2nd grade science experiment and more Uni-grade I think! I swear i made something like this out of a cornflakes packet and loo rolls when I was a kid :scratch: Without an accurate scale it's just a bunch of pretty lines really. Working out the air pressure at the valves would be useful too: Now I think about it, I'm pretty sure intake pressure will go UP and not DOWN especially at higher rpms.

 

yes you're right this is actuall computational flow dynamics and not FEA at all! I must be losing it :nuts:

[skinnyman9000 0]

I'll attempt to get some screen prints tomorrow of some of the stuff i knocked up last year, modelled and assembled a full moving single cylinder engine.

[boost monkey 0]

great news skinnyman, get it up on here! :grin:

[boost monkey 0]

Does anyone know where I can get technical drawings from which I can 3D model this manifold, or am I consigned to having to cut one up? :pale:

[KipVR 1]

Just read this thread, good stuff!

 

I just asked a colleague of mine who really knows his sh1t, (Jeremy Bliss- Thrust SSC Aerodynamacist/Systems Engineer) he recomended Flowmaster as a good CFD package. Although I've shown hime the full version of the SW2009 CFD and he seems relitavely impressed for the simpler stuff. SW is a really good all round package really buddy :D , Catia is the mainstream for Automotive though (and pays better :( )

[boost monkey 0]

Thanks for that Kip, very useful! There seem to be a few packages around at the mo: SW, AutoCAD, Catia and Pro-E spring to mind. SW is quite easy to understand, but I guess it's subsequently limited by what it can do because of the simple user interface. At the mo it does everything I need and it's free, but it'd be good to start using something else aswell.

 

I'm thinking I would like to do this kind of thing for my placement year but am unsure where I could apply to :scratch: