I'm a glutton for punishment...and crabby today, so for some reason I don't just want to drop it...

stress=My/GIp, where Ip is the torsional constant (for a round solid bar actually equal to the polar moment of inertia), M is the moment, y is the distance from center, and G is the shear modulus.

Ip is proportional to r^4, y is the same as r, so an r^3 ends up at the bottom of the fraction.

I would state that the change in stress concentration factor for a 0.5mm groove properly sanded and polished out is nil. The only thing that changes is r, so our stress increase is inversely proportional to ((r1^3)-(r2^3))/(r1^3). Assuming a 20mm bar, r1=10mm and r2=9.75mm. That yields a stress increase of 7.8%, which will not quite cut the life expectancy of the bar in 1/2. (10% stress increase halves the life, as a rule of thumb). Arguably the bar is now less durable. However, since fatigue testing regulary shows durability differences greater than a factor of 2, this change is not all that significant, ans it falls within the normal expected variation of the parts. In other words, it's worse but the world will not come to an end. Testing the bar may even yield a failure someplace other than the repair.

While I'm at it, high tensile strength steel is extremely notch sensitive. Very small nicks in the bar will increase local stresses far more than the theoretical 0.5mm diameter reduction, and will therefore affect durability to a greater extent.

The pre-stressing of the bar yields the outer section of the bar and gives it a n unloaded shear stress in the opposite direction of its intended loading. This zone is relatively deep (not shallow like shot peening), and while directionally incorrect the effect of removing 0.5mm of material is not going to be that significant.

Last, the empirical observation. The "how to buy a Porsche" books have lengthy laundry lists of common problems with these cars; anyone reading such a book would never buy a 911 if they were sane (I'm not, and boy does it hurt...). Torsion bars are not on the list. I'd bet 50% or more of all pre-89 911's in existence, if the A-arm bushings have not been replaced, have wear on the torsion bar. The bars must be pretty good, with a decent factor of safety...

OK, I'm done ranting.

I'd still replace the bar sooner than later.

:)

Oh, really?

(I did a quickie FEA model for ya)

http://forums.pelicanparts.com/uploa...1189891895.jpg

A couple hundred pounds twistin the splines (not much . . .read the numbers not the colors)

Simple scrape...
http://forums.pelicanparts.com/uploa...1189892011.jpg


Polished out (C2 curvature ...really smooth transitions)
http://forums.pelicanparts.com/uploa...1189892084.jpg


Again, read the numbers, not the colors. ...the colors 'go red' for the max value of each.

SO, is "nil" your Final answer? :cool:

The numbers were fascinating. I wouldn't have guessed that the jump in stress concentration would have been anywhere that high.

And the difference still doesn't include anything in the way of a surface treatment which is so important to fatigue resistance.

I say put them on a lathe, taper out the notch, and then sell them on ebay as progressive rate springs. ;)

.. . and have a hottie "holding" the bar for the ebay pic. ;)

Anyway, to be fair to burgermiester, he was saying a .5mm deep scrape, whereas what I modeled above is a 2mm deep 'scrape.' (on a 22mm bar) We don't know what the depth of the bar in question is. (it looks like more than .5mm. .l.less than 2mm) My only point with that FEA was to show that the effect of even a 'polished-out' dip, is much more than "nil." :)

I would estimate the wear at just over 0.5mm but less than a 1mm where the bar has rubbed.

If I get some time I well clean the worst bar and mic it up.

Thanks for all the input

Like previous post mine snapped with little evidence of surface damage. Can see where bushes have flowed though, so will need to replace these over winter. Anyone need to see another gratuitous broken bar picture?

yeh let's see it!

anybody have access to a sem?? it would be really cool to look at the grains along the strain in these things.

can "used" and new bars that look fine be plated to prevent rusting?
How do you all recommend to keep the bars in good shape till brake or are changed out?
I bought a set of saunders hollow bars and thay came w/ "zero" paint or rust preventative. So I sprayed them w/ a wurth spray lub that I cant remember that is very sticky and coats well. HOWEVER, I want to treat them 1x now and forget about them till they possibly break or are changed out for what ever reason.
Thanks!
Bob

trying picture load now.

http://i201.photobucket.com/albums/aa239/rs48635/911-torsion-end.jpg

http://forums.pelicanparts.com/uploa...1190075143.jpg

btw I followed advice here and wayne's book and applied grease all over my new bar. Then went out and did this

http://forums.pelicanparts.com/uploa...1190076621.jpg

I can use an SEM but doubt that it's worth sending the broken pieces across the pond...

No. The plating will crack before the steel, and then the crack will propagate into the steel. Plating is bad for fatigue life of steels. Even plastic parts fatigue sooner when chrome plated. Powder coat and grease is the way to go.

Island, I am curious:
Boundary conditions, element size, element type, code? I can talk shop in this regard, and I like to...happy to do it by email if you don't think we should bore the rest of the forum. Or here - I guess folks can just ignore it...
I am surprised at the amount of stress concentration caused by a 0.5mm dent. Hard to believe classical mechanics is that far out of whack. But, sometimes it surprises ya. If I have time I may try something similar...

A scratch is almost a singularity. Unless your element size approaches the grain size of the steel, you have not gotten the proper notch stress.....and I don't believe P-elements from Mechanica (should be written with that Metallica script IMHO :) )

Chuck Moreland plates his bars before sale. He can answer this for you definitively. Not sure if you can plate them once cold or what. But I think he buys from SRP or someone and then plates.

Island, first of all: Way cool to run an FEA. Thanks!

Aaahh, 2mm dent. Now it makes more sense.

My "nil" was for the notch factor - Kf - used specifically in a fatigue analysis.
Not for the basic stress of the reduced section, which is what the FEA shows.

Classical mechanics predicts a 25% stress increase for a 20mm section vs a 22mm section. Looking at your plot, 180MPa for the rest of the bar, that would be around 225 MPa. Your FEA shows 238MPa. Not all that far apart. Stress increase of 25% would cut life down by a factor of 6 or so - starts to be pretty significant. Even a 1mm scrape I'd consider too deep to fix. My bars have 1 to 1.5mm hollows - they're a gettin' replaced.

This is a cool thread IMO. But then I'm a geek...

Good night!

Hey Chuck Moreland, do you know if my used and new bars can be plated, or powder coated or what do you recommend so I can treat these and forget about them till a break or changed out to go up or down in size.

Thanks for your help!
Bob

no problem.:) --you raised the bar (so to speak) with all that geek-speak-- I had to get in on that too. ;)

I'll run one at .5mm when I get a chance.

red83911,
What kind of squeaking do you hear. Is it more like metal scraping or is it like "ticking"? I know squeaking is squeaking but I want to make sure.
Thanks.

mmmkay . .

did some at .75mm (~.030_in)

The reference stress was 1788psi

Wrong units (unless you want to do the math) on this one, but you can see the 'scrape"

http://forums.pelicanparts.com/uploa...1190176556.jpg


Here it is sectioned (w/ psi)
http://forums.pelicanparts.com/uploa...1190176617.jpg


Smoothed out
http://forums.pelicanparts.com/uploa...1190176944.jpg

Soooo . . . 2000 vs 1788psi.

Let's call it ~10% -- not "nil" . .. but not horrible either.

Again ... base stress is around +8% for the classical machanics approach. Kf would be a factor to multiply this stress by to arrive at the true notch stress. But there is no notch in the smoothed out bar - Kf is 1.0. There still is a stress increase - about 10% in your analysis. Ought to cut life at that location by 50%.


The bars in my 911 were greased, I don't know when. Does NOT appear that much maintenance was done on the car beyond regular oil changes and replacing things that made the car undriveable. Except where the bars rested on the A-arm, grease was intact and bar underneath was spotless.

http://forums.pelicanparts.com/uploa...1190195525.jpg


Not to be outdone... (Nastran (rocket science - ha!) - 1mm 2nd order tetras, 1E6 N*mm, 19mm bar, 0.5mm smooth depression and 0.5mm sharp scratch) :)

Island, what software are you using? Looks unfamiliar but cool.

Similar results (encouraging when different software has similar results! Happens less often that one might think...)
Base Stress: 1285MPa
Smooth Depression: 1394 (+8.5%)
Sharp Depression: 1630 (+27%)

Because of the element size, the stresses in the sharp depression are certainly underpredicted...unless it has a nice radius in the notch...