I seem to remember doing this a long time ago and I believe you use one half the weight of the big end of the rod in the calc.

regards

Chris_seven, dkirk, I see broken rods all the time on other engines. Calculations show that compressive forces are near double tensile in the example. Factor in boost, and it would theoretically go up.

When rods let go, what force is causing the failure? Compression or tensile (assuming oiling was adequate)?

I was always under the impression it was tensile?

Corky Bell once told me, "Boost doesn't kill rods, RPMs do." He was known to break ****, ol' Corky.

Maximum Boost

He also wrote that book when we weren't making much power.

Also, 10-15 lbs of boost was a lot back then. Now, folks run 50-60 lbs of boost on race gas and over 100 lbs on diesels.

It will always be a tensile stress or traction vectors that causes a component to break but we need to consider what we mean by failure.

The whole concept of 'Fatigue Prevention' is to eliminate catastrophic failure by understanding the 'safe' life.

If rods break then they are being used beyond their design limits.

Fatigue is responsible for around 95% of all engineering failures and it is insidious because it generally occurs at stresses much lower than the component will tolerate when loaded monotonically.

As I mentioned earlier it is conventional wisdom to consider that fatigue is a two stage process but it is equally valid to consider three stages.

The First Stage is the initiation of a fatigue crack which would typically be 90% of a components 'life'.

The stress range applied to the part is important and the compressive part of the loading spectrum makes a significant contribution as I discussed in an earlier post.

There is a well established convention for describing this stage of the process which is known as Miner's Rule which provides a basic understanding but a more modern approach would use a Weibull Statistical Model.

Miner’s Rule and Cumulative Damage Models

Once a crack has initiated the second stage of the process will involve crack growth and again reverse loading and crack closure effects will influence the crack growth rate (da/dN).

Eventually the crack will extend sufficiently that the 'remaining ligament' of the component will be insufficient to support the applied load and the part will fail catastrophically which would be the third and final stage.

The stress which causes this failure will always be tensile in nature.

I would always model a part such as con rod in terms of the Number of Cycles required to cause crack initiation and by using a Rainflow technique to establish the damaging part of the loading spectrum this could then translate to hours. That is to say that 'failure' should be considered to be crack initiation.

Some conservatism is needed in terms of either applied stress to ensure the failures never occur, which would be the case for production engines, or establishing a safe life for highly stressed racing parts.

If rod failures due to fatigue are commonplace this job isn't being done very well and would normally be associated with a poor understanding of the load spectrum.

We used to design parts with a concept known as a 'Factor of Safety' but when failures continually occur I think that the term 'Factor of Ignorance' would be more appropriate :D

Next, we can examine the 8000 rpm, 200 bhp, 103 psi BMEP condition in what would be a realistic shift point in a high-output, 3.2 liter, 911 engine. The PV curve appears as follows:

http://forums.pelicanparts.com/uploa...1489860228.jpg

Maximum cylinder firing pressure is 589 psi. The axial force applied to the con rod is now thus:

http://forums.pelicanparts.com/uploa...1489860289.jpg

Here is the important take-away on the con rod forces at the max engine speed point:

Force Max Compressive – 3776 lbf
Force Max Tensile – 6006 lbf
Load Cyclic Frequency – 66.7 Hz (@ 8000 rpm)

Tensile force now dominates and is at a maximum during the exhaust/inlet stroke. It is interesting to note that (just by coincidence) the peak values are numerically close to those obtained at 6100 rpm, just swapped in direction.

I am not considering the bending moment exerted on the con rod as this is typically not significant and is ignored (Ref. 4). This type of analysis is used for determining the forces and loads that the wrist pin and crankpin bearings support, and I have not incorporated the rod bending moment into the calculation but now have a good excuse to do so. The bending moment, even though of low magnitude, combined with a compressive force, can initiate a buckling failure. So for an improved understanding of the con rod loading, bending should be known. It may take some time as the equations are ugly.

So we now have a good approximation of worst case axial force conditions that are imposed on the con rod. Realistically though, these are only applied for brief moments during the operational life of the engine and are only useful for calculating the safety factor of the con rod, knowing the X-sectional area and material properties. For a realistic fatigue calculation, a weighted speed/load history on the engine for a sample time period would have to be acquired – this would take data recorders, many cars/owners, etc., to get a representative sample of how a typical 911 engine is used. My background is in the marine and aviation engine fields – here very accurate data has been accumulated on engine duty cycles over the useful life of the product. This is extremely useful for predicting L10 bearing life in an engine with say, a rolling element crankpin bearing. However, for an automobile, especially a Porsche 911, it is doubtful that such an engine duty cycle could be accurately defined due to the highly varied driving styles these cars are subjected to. Without this duty cycle info, accurate fatigue life calculations aren’t possible.

Chris – I hope this applied force info can be useful for predicting material stress, but I don’t know the con rod alloy nor the X-section area. Hoping you can use this to examine the stress levels at these two load conditions. Incidentally – I’ve incorporated your term “Factor of Ignorance” into my vocabulary – love that one!

These are the references I used and for those of you who are engine guys, I highly recommend all of these:

1) Gordon P. Blair, “Design and Simulation of Four-Stroke Engines”, Society of Automotive Engineers, Inc., 1999.

2) Wayne R, Dempsey, “How To Rebuild and Modify Porsche 911 Engines 1965 - 1989”, Motorbooks International, 2004.

3) Bruce Anderson, “Porsche 911 Performance Handbook”, Second Edition, Motorbooks International, 1996.

4) R. van Basshuysen and E. Schafer, “Internal Combustion Engine Handbook”, Society of Automotive Engineers, Inc., 2004.

Great stuff dkirk/chris_seven!!

Dkirk, what happens to the compressive from boost?

I'd argue that "cylinder filling" is not really understood and would be hard to figure true combustion pressure, no?

My example I always use is a large coffee can with a tiny hole and a 125 PSI air gun. If I hit the lever for a half-second, the coffee can does not get up to 125 PSI in that time. Not even close.

Same with a reciprocating engine.

The valves are only open in short durations. "Closer to cylinder filling" when talking about VE, to me, would be more accurate.

Point is, if you're using the same 930 engine, one with stock cylinder heads and one with Xtreme CNC'd heads with larger valves, bith using the same boost levels. The higher flowing engine would exhibit higher cylinder pressures - then resultant higher compressive loads on the conrods, no?

Love to hear from those in the know! :)

Tippy - You are correct, a turbocharged engine would exhibit less tensile and more compressive connecting rod loads running under boosted conditions, than a comparable naturally aspirated engine. The fact that cylinder pressure is above atmospheric on the exhaust stroke due to back-pressure imposed by the turbine, plus boost pressure on the inlet stroke being elevated as well, will impose a gas pressure load on the piston during the exhaust-inlet stroke. This will reduce the tensile force in the con rod during this portion of the cycle.

I chose not to look at turbo engines for this study as the N/A engine seemed to be the logical place to start. There is an upper limit on the BMEP range for a given N/A engine but for a turbocharged one, the sky's the limit.

Timing has a big effect on combustion compressive force. In these engines with single plug you have to fire earlier before TDC to get the max force where you want it after TDC (depends on the angle between the rod and crank pin which depends on the rod ratio). Earlier ignition moves that part of the curve in post 40 to the left. Too far to the left and the max force occurs too early past TDC (right after TDC in the graph) when there is too small of an angle between the rod and crank pin. That's when rods get bent, not in these engines but in others that have a larger rod ratio and are running boost on stock rods. Twin plug in these engines squeezes that part of the curve together so you can fire later and get the max force where you want it at the right point past TDC. The smaller the area under the combustion force curve before TDC the better because that force is trying to push the piston down when it is coming up. That robs power.

I will draw up a standard rod and see if we can carry out some basic FEA using the loads you have determined.

It will be quite straightforward to look at the minimum cross-section and look at stress amplitudes in this area.

Most after market steel rods in the UK are made from a 080M740 steel in a V Condition and this is very similar to 4340 heat treated to 170ksi (1150MPa)

I would think this type of steel would have an endurance limit of around 500Mpa - perhaps a little better.

Since this is talking about rods I have a set of 996 rods that are I believe Ti powder metal cracked rod technology.

The factory manual says DO NOT punch or stamp mark the rods.
A previous engine builder who was a total hack punch stamped the rod caps...
So are these rods now NFG?

from "rods" thread in turbo forum-

"The one now with Carrillo had stock rods with ARP studs. One bent on the dyno and caused the need of a crank, both case halves, oil pump, cylinder barrel, piston so a complete engine redo was needed ( well over $15K )."

So much for stock rods not bending..Good rods are CHEAP $2k vs. $15k in this case.

Question regarding rods
 

I am in the process of rebuilding my 3.2 SS. Based on the head's and cams, I anticipate spinning it to 8000 rpm. I am trying to fully understand the implications/engineering of replacing the stock rods.

From my research on this forum, and advice from experts from this forum in emails, I believe

Most rod failures are secondary to lubrication/bearing failure.
Stock Rod failures without oiling issues is actually rare
Cross drilling, case oiling passage mod and oil pump mod/update are critical for a high revving engine.
The stock rods with APR bolts are quite strong but heavy.
They handle 7500 rpm pretty well.
For engines revving over 7500 upgraded rods are advisable.

Are Pauter or Carrillo rods actually mechanically "stronger" than the stock rods? or do they survive at higher RPM better than the heavier stock rods simply because they are lighter which inherently creates an RPM safety margin because of the lesser forces associated with their lighter rotating mass at high RPM?

Look at post #9. You be the judge.

I posted a big block Chevy aftermarket rod that handles 8000 RPM.

I then posted a stock 3.2 rod.

How can the stock 3.2 NOT handle 8k with ease? It's very short and has lots of cross-sectional thickness.

I'm sure it can handle it but the question is what's the loss in rod bearing life doing it with such a heavy rod. There is no reason to have boat anchor rods if they aren't needed. I have a set of 997.1 3.6L Cup ti rods with 45 hours that I picked up at a swap meet for $100 because the owner thought they were "almost timed out". They have roughly the same cross sectional area as a stock 3.0L rod yet weigh 1/3 less. If one is having custom pistons made for a 3.2SS seems like it would be a shame to relegate them to paperweight duty unless there is sound engineering and real world results behind fatigue being a real problem. I only was around Cup car for 5 years but in that time I didn't see 1 rod failure. Maybe its because the guys that ran them 150hours + got lucky or maybe they are still beefy rods even in Ti.

You get some confidence in them when PMNA says the fast way in a cup is to do 10.5K RPM downshifts to lock the diff early so you can brake deeper.

The argument never changes with regard to these rods it is just Risk/Reward and there are no valid NDT techniques that will measure the accumulated fatigue damage.

There is no published data so when you run them for more then the recommended hours you risk a failure and just how big the risk is just a guess.

Real fatigue data is extremely expensive to generate in any meaningful way.

One technique used in the Automotive world involves a Weibull Model.

It is quite easy to show that fatigue life rarely follows a 'Normal distribution' and much more sophisticated analysis is needed.

"Weibull analysis" attempts to make predictions about the life of a component by fitting a statistical distribution to life data from a representative sample of units.

The parameterised distribution for the data set can then be used to estimate important life characteristics of the component such as reliability or probability of failure at a specific time, the mean life and the failure rate. Life data analysis requires that the following information is important:

Bearings for example are generally specified to a Weibull B10 Life and this is the time by which 10% of the bearing population would have failed. This can also be known as a L10 life.

It is possible to estimate an MTBF from a Weibull Life but again there will be an increased uncertainty.

I am confident that there is no data to support the life of these Ti rods one way or another and that that the good old 'Bulgarian Constant' has been applied to the design data.

It would be most likely that there will be an increasing probability of failure with increased duty and that at some time a rod will fail and the 'so far so good' strategy isn't too helpful. :D

I would consider using controlled shot-peening to ensure you 'zero-life' these rods as the cost is not too great.

When we 'peen' Ti Rods we make nylon coated plates to protect the Plasma/Thermal Moly Coating on the big ends and similar plates for the small ends.

We also use reduced head bolts for the caps to ensure that the radius can be correctly peened.

Dave - fascinating tour de force of engineering calculation (at least for the unwashed like me, who are perhaps easily impressed by this sort of stuff).
I have some information on a rod contemplated for 3.4 Caymans. I guess these water cooled motors need longer rods, as it is 145mm. At 608g (which I believe is stock) the calculated split is 443g rotating, and 165 reciprocating, which isn't quite the approximation you used. A shorter rod would have even a higher percentage of rotating weight? I'm probably missing how these two are calculated.

280 flywheel HP from a 3.2 is way above stock, but well below the 110 hp/L you can make with these motors in full race mode, but I suppose that doesn't matter for the purposes of what you are investigating.

And a motor which makes peak HP at 6,100 RPM is most likely going to be shifted at a much lower RPM than 8,000. Optimum shift RPMs are calculated from the torque curve and gear ratios, but I'd guess even shifting at 7,000 would be wasting potential acceleration. That would likely be as much as you'd stretch the motor for those few corners where it was better to waste potential to make up for upshift losses before braking.

I tried to calculate a cross sectional area, but could not find widths for the long stem of the rod (where cross section would be smallest)in either of the orientations needed on the drawing.

Again, not relevant to the questions addressed here.

Finishing up my homework problem on the connecting rod study –

I calculated the maximum bending moment in the plane of motion imparted to the con rod at the two engine speeds considered, these being 6100 and 8000 rpm. A bending moment is created due to the rod undergoing an angular “swing” (combination of translation and rotation), which it does during a revolution of the crankshaft. If one were to plot the path of motion of the CG (center of gravity) of the con rod, it would inscribe an ellipse. Energy is imparted to, and recovered by, the crankshaft in the force required to “swing” (i.e., impart angular acceleration) to the con rod about its CG. The force is resisted by the mass moment of inertia of the con rod with the result that a bending moment (torque) is exerted on the beam section of the rod. Most books on the subject state that this moment is so small in magnitude that it can be ignored. After going through the derivation and incorporating this calculation in my program, I certainly agree. Not having any information on the mass moment of inertia of the 3.2 L connecting rod (normally obtained from a CAD model or measured on a torsional pendulum) I used the lumped mass method, as we have previously made an educated guess as to the equivalent masses concentrated at the wrist pin and crank pin ends. Distances to the CG are by the parallel axis theorem. Results for the bending moment in the plane of motion are as follows:

Rpm = 6100: Max Rod Bending Moment = 11.57 in-lbf (1.3 Nm)

Rpm = 8000: Max Rod Bending Moment = 19.90 in-lbf (2.25 Nm)

The bending moment is applied and reverses once per every engine revolution. These bending loads are minor in the grand scope of things but should be known if we are to fully understand what the con rods are subject to in the running engine.

Tippy – Your last post is something I’ve wondered about too. Only thing I can come up with is that the Porsche rods were designed for turbo engines as well and need to be inherently stout. And to better answer an earlier question you posed – yes, we can calculate (with pretty good accuracy) the volumetric efficiency, peak pressure in the combustion chamber, and predict the power output knowing the bmep the engine can produce. Conversely, if we know the power output, displacement, and engine speed, we can calculate the bmep. Bmep is a relative pressure term, strictly calculated and not measured. It is the average cylinder pressure that would be required to make the brake horsepower at the crankshaft, at a given rpm. What is nice about bmep is it is a specific term, meaning we can compare various engines of totally different types. We know from looking at a bunch of 911 engines in various displacements, states of tune, and engine speeds, that a 180 psi bmep for a naturally-aspirated 911 engine is realistic and can be achieved on a well prepared engine. Knowing this, and picking a target rpm, we know the bhp, delivery ratio (volumetric efficiency) and can produce a theoretical cylinder pressure trace that is accurate within a few percent of the actual cylinder pressure. The delivery ratio (volumetric efficiency) required to produce this level of bmep is about 110%, made possible by harnessing the pressure waves in the induction and exhaust system to not only move air, but trap it efficiently.

Boosted79 brings up a good point about dual ignition and how this reduces negative work by speeding up the combustion process. Anytime the flame travel distance can be shortened, this is always of benefit, both in reducing the tendency to detonate and producing higher power output. The Porsche 2-valve engine is a true “hemi” and with the highly domed piston, the combustion chamber is divided and less than ideal. Dual ignition is the elegant solution that obviously works by effectively “halving” the rather tortuous flame travel distance. For my theoretical PV diagram, I used a single ignition source with the following inputs:

Ign Timing = 26 deg BTDC
Ign Delay = 13 deg
Total Heat Release Interval = 44 deg
Peak Pressure Occurs = 20 deg ATDC

The timing is right out of the specs on the 3.2 engine…the rest I had to make an educated guess for but are values suggested in Ref. 1 for engines of this type.

Concerning rods used in a turbo engine – engine destruction is totally controlled by a screw (the wastegate spring) and how brave the operator is. The con rods can easily be taxed well beyond their design point resulting in catastrophic failures as there is (in theory anyway) no limit to the BMEP that can be produced with a pressure-charged engine. That is why I wanted to limit the study to naturally aspirated engines.

Walt – thanks for the kind remarks and contributing info on the 3.4 L Cayman rod. From your numbers, this rod has a 27 / 72 % mass distribution, small end to big end. I used a 33 / 66 % distribution for the 3.2 L rod, just by guessing. This will have a slight effect on the results, but actually in the right direction as more recip mass will cause higher tensile forces in the rod. And yes, usually as the con rods decrease in length, the CG moves closer to the big end, thus more percentage of mass at the rotating end.

I chose the 280 bhp @ 6100 target as this seemed to be a good street engine that a prudent engine builder would not be concerned with using stock con rods in. I appreciate your comments on max engine revs – I couldn’t find any representative power or torque curves from Ref 3 that showed this kind of info, so I had to guess again.

Chris – will be interested to know what the rod stress is when you get the areas, plus your take on the combined bending and axial loads.

Great info! Enjoy the read.