914RS4 Tube Chassis 996 Suspension Audi 2.7TT Stage 3, ADDED RECAROS Options

[Curbandgutter]

Well the dream begins. I'm embarking on the building of a 914 project. Here is the candidate. Bought a roller from LA area. Looks like its not going to be too much rust. Firsts thing will be to strip it. Taking front fenders and rear quarters off. Then sending to Cal Blast for media blasting. I look forward to the knowledge that I can gain from fellow enthusiasts. Plan on taking a lot of pics Stay tuned......


DISCLAIMER: Do not attempt to duplicate anything that is mentioned or illustrated in the entirety of this thread. I do not make any warranties of any kind. If you try to build what I am doing, you take full risk. Do not try this. It may be dangerous to your health and may get you killed

[jd74914]

Sweet! You don't see too many people doing FEA iterations on chassis. (IMG:style_emoticons/default/smilie_pokal.gif)

What software package is that? Almost looks like Ansys APDL or maybe more like Grape? You're modeling with beam elements or trusses? Can you glue a shell element floor on to better match reality? In one of the racecars I worked on laminating a thin carbon shear panel on the floors increased stiffness by 25-30% (experimentally verified via twist test too).

I think you're seeing artificially high loads in N43A because of the huge open box in the cockpit. If you add triangulation near the doors and closer the floor I bet you'll move some of that load path (if my visual FEA makes any sense haha).

Your loading seems a little weird. Why load the lower arms so highly in the vertical direction? They shouldn't see all that much vertical force since it's really all desisted by the spring/damper unit. Why not model the suspension in as an infinitely stiff member and then load with bump, lateral, and longitudinal forces? Then you wouldn't have to draw a free body diagram to figure out point loads.

With actual loading N36B might be more heavily loaded than you think (and in some combined weird bending/buckling mode so it probably needs a big safety factor). If you're doing frame stiffness twisting it (via moment about the front suspension nodes with the back fixed) seems more widely accepted than loading one side. There is a good SAE paper on it, I might be able to find it somewhere.

Sorry if you know all that stuff already. I'm not a mechanics guy by any means but I've spent a bunch of time doing chassis design/analysis for some tube frame racecars and a Lotus.

[Curbandgutter]

Sweet! You don't see too many people doing FEA iterations on chassis. (IMG:style_emoticons/default/smilie_pokal.gif)

What software package is that? Almost looks like Ansys APDL or maybe more like Grape? You're modeling with beam elements or trusses? Can you glue a shell element floor on to better match reality? In one of the racecars I worked on laminating a thin carbon shear panel on the floors increased stiffness by 25-30% (experimentally verified via twist test too).

I think you're seeing artificially high loads in N43A because of the huge open box in the cockpit. If you add triangulation near the doors and closer the floor I bet you'll move some of that load path (if my visual FEA makes any sense haha).

Your loading seems a little weird. Why load the lower arms so highly in the vertical direction? They shouldn't see all that much vertical force since it's really all desisted by the spring/damper unit. Why not model the suspension in as an infinitely stiff member and then load with bump, lateral, and longitudinal forces? Then you wouldn't have to draw a free body diagram to figure out point loads.

With actual loading N36B might be more heavily loaded than you think (and in some combined weird bending/buckling mode so it probably needs a big safety factor). If you're doing frame stiffness twisting it (via moment about the front suspension nodes with the back fixed) seems more widely accepted than loading one side. There is a good SAE paper on it, I might be able to find it somewhere.

Sorry if you know all that stuff already. I'm not a mechanics guy by any means but I've spent a bunch of time doing chassis design/analysis for some tube frame racecars and a Lotus.

Now that's what I'm talking about! (IMG:style_emoticons/default/aktion035.gif) (IMG:style_emoticons/default/aktion035.gif) (IMG:style_emoticons/default/aktion035.gif) Love this kind of input. Now to answer some of your questions. The lower A arms were not loaded, what you are seeing is the 3 points where the suspension cradle is bolted to the chassis. The 900 lb vertical load represents a 5g load on the wheel. The next step will be to simultaneously load a 900 lb load in a downward fashion on the opposing suspension cradle support points to create a couple, or rather twisting of the frame as you mentioned. I will be running the same scenario at the rear and then run another scenario to simulate bending forces in the frame. I will go ahead and model the floor and both firewalls with plate elements and see what happens. Might as well model the "longs" as well. this will give a better representation of the behavior.

I'm modeling with beam elements with fixed joints in all directions being that the joints will be notched and welded.

I would love to get my hands on that SAE paper. I'm sure that I will learn a couple of things.

[jd74914]

Now that's what I'm talking about! (IMG:style_emoticons/default/aktion035.gif) (IMG:style_emoticons/default/aktion035.gif) (IMG:style_emoticons/default/aktion035.gif) Love this kind of input. Now to answer some of your questions. The lower A arms were not loaded, what you are seeing is the 3 points where the suspension cradle is bolted to the chassis. The 900 lb vertical load represents a 5g load on the wheel. The next step will be to simultaneously load a 900 lb load in a downward fashion on the opposing suspension cradle support points to create a couple, or rather twisting of the frame as you mentioned. I will be running the same scenario at the rear and then run another scenario to simulate bending forces in the frame. I will go ahead and model the floor and both firewalls with plate elements and see what happens. Might as well model the "longs" as well. this will give a better representation of the behavior.

I'm modeling with beam elements with fixed joints in all directions being that the joints will be notched and welded.

I would love to get my hands on that SAE paper. I'm sure that I will learn a couple of things.

(IMG:style_emoticons/default/beerchug.gif) I get worried that people will take offense to comments like those sometimes.

Gotcha, I just looked at your pictures again; didn't realize the 996 stuff was all on a subframe. Now it all makes sense.

Looking at that model again it might get stiffer if you switch from the bent front windshield frame halo-style bar rearward facing bars like found in a non-halo cage. Then you'd spread the longitudinal bars (N44 and N43 maybe) out towards the edges moving them further from your head and building a better node at the windshield corners. I just get scared seeing cross-bracing put hoops in bending. It might also be worth switching around some of the triangulation (ie: in front of and behind the door) to meeting at the same places to form some "super nodes." I noticed on the full tube chassis that this seemed to help stiffness without any weight penalty. The x-bracing on top and bottom of the rear might make maintenance very difficult. I did this over a chain drive differential in the name of stiffness and really hated myself for it after the fact. You could probably get most of the stiffness with a bolted shear panel. Hopefully that makes some sense; I can draw it tomorrow really quickly too.

The 5g load is pretty conservative; I've always designed around 3g bump, 2g lateral, and 2g longitudinal loading (though not all at the same time since tire friction circles limit the combined grip) and haven't had many problems. Your analysis plan sounds good to me! Unfortunately I'm at a conference right now and having trouble remoting into my regular computer to look but I'll check for the paper as soon as I get home.

Totally unrelated to the design stuff, but when you notch everything be sure to drill small holes in all of the receiving tubes at the joints. Being able to back purge while welding makes the whole process much better. You have less problems with oils, etc. running out and end up with much higher quality welds. (IMG:style_emoticons/default/smile.gif)