Going back to my undergrad days, yes, FEA can do just that and show you how the loads get distributed throughout the entire piece (provided you put everything in correctly and constrain it where it needs to be).

Finite element models...well, it does just what the name says. It breaks up a component into a finite number of smaller parts (as in, you can count them, as opposed to an infinite number of smaller parts), and calculates the loads over those smaller parts.

I may be a bit rusty on it, but it should give you an idea.

 

This is a bit of a 2 way street

Gears that have been installed correctly and thos who haven´t

Lets say that the gears are installed correctly but the diff blows up,that would be because of weak spider gears and springs,right?

Then why not make the spider gears and springs of better materials that has been treated to be tougher like shotpeend,magnafluxed,Cryo ect.. and same with the gears?

 

I think treated materials would be marginally better, but in the end, most will argue that it is the physical size of said parts that makes them weak (not the material selection). By the time you got done doing that, you would have also put a decent amount of money into it, which could have been spent on a physically larger rear. The argument just kinda goes on from there...

 

This could go on untill the end of time.. the bottom line is that everyone knows that the thing is junk... nobody is going to do anything about it because its simply not worth it....there are no rules saying we have to run a 7.5 in our cars so seek something more practical...Trying to move a 3700 pound car with a 7.5 inch ring gear is a bad idea...If there were no options avalible this would be something worth looking into.. if F bodys came with some sort of IRS or something or transaxle 3 of the best diffs ever made for performance vehicles are an easy bolt in for the average home mechanic.Call strange... Moser Currie and tell them you want some one off or limited run parts made so you can beef up 7.5 and they'll prob hang up the phone on you...I get a kick out of some people they'll spend 10 times the money and effort trying to jimmy rig what they got rather than buying something better suited like building a twin turbo motor thats going to make 1000 rwhp but they want to use the stock short block....and Magnafluxing has nothing to with adding strength....

 

Is there anyway to break the model down to say 100 Billion or more pieces (modeling every single molecule)?
Maybe something like infinite model analysis?

 

I'd say no unless you have access to a supercomputer of some sort. Not sure the results would warrant the hours involved...

 

Finite element models gain accuracy as you use more and more elements. Because each element has a limited capability of portraying stress and strain as it varies across each element, you use smaller and smaller elements to accurately model the variation in a part.

However, in years past, numerical precision in a computer was the limiting factor meaning that overly small elements would make a less accurate model. This issue is largely removed, but now we have a new problem.

Now, the problem is that metals and other crystalline structures have crystals of varying sizes, up to large fractions of a millimetre. At the crystal scale, material properties as we normally measure them, totally fall apart.

 

There is no need to. I'm going to give this a stab. The beauty of the finite element method (FEM), is that the solution can be determined as the convergence goes to zero. This is far outside the scope of the discussion though. Basically in FEM, there is something called the "mesh" and this is where the part (i.e. differential) is divided into a number of "elements". A tighter mesh results from smaller elements. Think of a 2 dimensional screen like you have on a window to keep the bugs out. The points where the fibers intersect are called "nodes" The nodes are where the model is "solved" based upon fundamental strength of materials equations (you can apply FEM to most physical problems, you just need to apply the correct equations) and it is only at these locations where a solution can be determined and not in between (there are exeptions to that) also the nodes are at the verticies of triangles not squares as I provided in the example. So the ideal case is you start with a course mesh, run the solver, check results. Then you refine the mesh to make is smaller and you resolve. This iterative process is run until the difference in subsequent solutions approaches zero (beware if you go too small in the mesh you solution can diverge as the infitesimally small elements require bit lengths longer than your computer can crunch and the solution goes to infinity (like dividing by zero).

So there you go, FEA in a bundle



You are going the wrong way with this. This isn't a "10 bolt vs. the world debate". You made your point and we all agree that there isn't a soul on this board that thinks a 10 bolt is a better rearend. I think I share the curiosity of many, when I asked about what exactly fails in this rearend. I always thought it was the housing being too weak but it seems I'm seeing more carrier failures and spider gear failures. Nobody would waste $1500 on a 10 bolt rear upgrade, but I'm probably one who would "invest" $200 for the sake of my curious mind as I ask myself, "What if?" questions. But then again we have not even begun to think about if there is a cheap solution to the weak areas of the 10. I would hope more people would describe there failures.

 

9.88 @ 3200 lbs with a 70 1/2 8.5" with stock axles and a spool only.

I was being cheap, I would bet axles & spool would handle 9's and last a long time.

 

Would this be similar to taking the limit of an equation as it approaches a given number?
And are you saying that it would not be beneficial (or maybe not possible at this point in computing power) to model each molecule, or in this case, each crystalline structure?
I guess where I am going with this would be using FEA to model, one molecule at a time, the flow inside a cylinder head, accounting for all the variables in a running motor.

 

No, it's more complicated than that. Basically, FEA can be broken into 2 catagories: 1. utilizing discrete elements 2. utilizing a continum of elements.
The first is used for determining displacements in a structural framework, much like a bridge truss (these calculations can easily be done with a pencil and paper. The second is what is used in commercial software like ANSYS, ALGOR, COSMOS, ABAQUS, etc. This method creates huge matrices which can easily be thousands by thousands in size. These matrices are the inputs and are what are solved by the computer). So in short, the software creates a huge system of equations based upon the mesh and solves each nodal point producing a solution.

Not possible using todays computers to solve for the flow in a head on molecular and time dependant basis. It's just to much to compute the movements of a partical at this level of fidelity. However using the properties of the molecules collectivley and treating them as a fluid, it easy to determine the flow of all the particals at a given location in the flow path,assuming you have access to some software. This is refered to as computational fluid dynamics.

 

The point he is making is valid. No matter how much money you put into the rear it will still be weak. Even if the case flex was limited (which I dont think is the main cause of failure), you would still be limited by physical size. You would have to brace the top and bottom of the case for hard launches, and the front of the case to distribute loads better, there isnt much you can do with the rear because of the cover. If the diff case and pinion were held perfectly in place there is only so much torque that the gears can transfer without failure. Even if the gears live there is only so much torque that the size limited, 2 spider differential case can take. Even if it had a full spool there is only so much abuse that the 28 spline axles can take.
You could bore out the housings to accept larger inner and outer pinion bearing to help reduce deflection, but then the housing is weaker. You could strengthen it back up I guess, but then you need to better hold the differential case in place. In the end if you have a better differential the ring and pinion would still break.
As far as the stock to aftermarket comparisons, the problem with aftermarket gears breaking is the fact that they are lower gears. Not only does a 4.10 have a physically smaller pinion gear than a 3.23 but it also multiplies 27% more torque, creating that much more stress.
In the end the weakest link in the 7.6 is its nearly 30 year old design for mid size GM vehicles (and monzas) that only made mid-100 hp. Everybody keeps talking about how they are S10 rears, but they go back further than that. The 305s, th350's and 200's(relatively tall first gear), and skinny *** 14" tires that the 78 mid size GM's came with stock put far less stress than a 4.3 5 speed S10 that may even carry a load. In the end for normal driving on a unmodified car the stock 7.6 is adequate. GM should have known how these cars would be treated and that many would not be left stock, but the bean counters looked at the most cost effective alternative.

 

Did you ever come up with anything hear guys . As i read on you all kinda lost interest in the whole 7.5 thing? I mean there has to be something that can be done.

 

I was wondering if someone was going to mention this before I got to the end. Almost the exact same thing I was going to say. GM did what they could by going from 7.5" to 7.625" ring gears and up to 28 spline axles, but the size of the unit is still to small for the loads our cars can place on it. Case in point- the restrictions they have in the hydraulics for clutch engagement to reduce the shock load on the drivetrain with stock power and tires.

Lets just say the differential is the point of failure, and you want to save the rear from self destruction, while at the same time upgrading to a more performance oriented gear ratio. A Zexel-Torsen HD diff will set you back $500, then $200 for gears, $160 for a cover that supports the carrier caps, and $125 for a deluxe install kit that includes pinion shims, carrier shims, crush sleeve, pinion seal, pinion nut, ring gear bolts, cover gasket, pinion bearings, carrier bearings, and wheel bearings and seals. Thats $985 in parts. I know it sounds like overkill, and it is, but you'll get all these components new in a bolt in 12 bolt or 9" rear. Add labor for setting all these up correctly, $150 -$200 as a ballpark, and you have a potential $1200 in a rear end that is still saddled with the same size primary gear components and stock axles. Granted, a 12 bolt for a 98+ car with ABS and Traction Control, and a posi (not a spool) is more than twice that at $2460 with 33 spline axles. But I don't think anyone would argue the 12 bolt would be much more durable in the end. Sell your complete 10 bolt while its in good shape and you close the gap a bit too.

Again, I realize that dollars and cents wasn't the discussion you wanted here, but that is much easier for everyone to digest than the mechanics of torque and shear in all of the components in the rear end.

 

Well im not a genious on rear diffs but this is what ive learned and have been told. I have a 98 transam with trailing arms and a eurethane tranny mount. My motor is stock except for free mods and my rear kept getting more and more loose until a loud clunk from the rear could be heard when shifting from reverse to drive. Later i totally blew apart the stock diff and destroyed the gears as a result. I got a auburn pro series lsd, richmond 3.73's and havent had a problem since. I had probably made 20 or so passes down the track and countless country drags before the stock unit finally failed. One thing ive heard about beefing the rear some is to weld the axle tubes to the center section because they are only pressed in. Does anyone have any thoughts on that?

 

I found this thread at camaroz28 to be a good one

http://web.camaross.com/forums/showt...+proof+10+bolt

 

Plain and simple GM gears arent meant for abuse. Ive nuked 3 pinions sets all factory, and my richmond set is still holding up, IMO the 7.5 is worth rebuilding with the right parts.

 

How do you check for carrier perload?
Also, It's not just the gears that fail. When mine detonated it was the carrier its self that had cracked.
Furthermore, a Torsen does not have spidergears. It has worm screw gears.

 

Actually in my car the right side gear actually chewed into the carrier which let the c clip fall out which then chewed up the gears. the only thing that was holding the axle in was the caliper!! kinda scarry concidering i was going in excess of 100 when i realized what had happened!!!

 

Great find! There are a lot of neat ideas in that post. Its not a bad thing to try and make your stuff last as long as possible...and have fun while you are doing it.