Final conclusion, Dec2, 2005 - Given all of the information below, and the fact that folks have been flying Corvairs for years with no crankshaft problems up until now, I've reached the following conclusions. Cranks have only broken in KR based Corvair installations. We tend to turn them faster than most, climb and turn quicker than most, and generally hotdog a lot. And the guys who've broken cranks also FLY a lot. As mentioned on William Wynne's Crankshaft 101 web page, there seem to be extenuating circumstances for the other cases (very long prop hub, insufficient journal radius, etc), but mine's not so clear cut (although WW seems to blame a "blob of Loctite" on mine, which we both know is BS). I do have a slightly longer prop hub, which is 1.25" longer than WW's front starter setup (3" + .75" adapter puck = 4.25" compared to my 5" hub). That doesn't sound like much, but it IS 33% longer, and therefore multiplies bending forces created by the prop by 33% when they get to the crankshaft. I also have to mention that my engine was a 3100cc, which by definition means the crank is seeing 15% higher loads from combustion alone, so now we're up to 48%. That's a pretty big number, when you start thinking about safety margins.

After my crank problem, John Kearney decided to build an engine vibration test machine based on a laptop, using two tri-axial sensors, one located on the engine and another on the airframe (in order to difference out the airframe vibrations). We have tested several engines to date, and I've tested both my old prop and new prop with this setup, as well as with different indexing. Although John hasn't synthesized the data yet, his initial comment after reviewing a few slices of data was "The propeller forces that I reviewed for about three different RPM ranges were 10 to 15 times the magnitude of the engine power pulses making the prop forces the dominant input to the engine". This also supports the thought that the prop is a huge factor with regards to the forces that the crank sees.

Comments from experts in the field are that torsional forces were not an issue with my crank breakage...it was pure bending. I also have to mention that my engine was a 3100cc, which by definition means the crank is seeing 15% higher loads from combustion alone.

Nitriding is a significant amount of insurance against fatigue fracture of the type that broke my crank (see comments down below). Whether or not you'll ever need it depends on a lot of factors, many of which you have no control over and probably are not even aware of. I'm now building another 3100cc engine, with nitrided crank, and perfectly radiused journals. I'll also shorten the prob hub down to 3" (I have one of WW's hubs), because my cowling would be just as streamlined even if it was 2" shorter. If a shorter prop hub works, there's no reason to go with anything longer.

What else am I doing differently? The prop is indexed so that it is vertical when number 6 piston is at TDC. What this does is takes the extra forces due to P-factor (descending blade getting a better bite) and distributes them to the number 6 rod journal 90 degrees out of phase from when the rod bending is highest, so now it's perpendicular and not additive. My "new" 2700cc engine is as smooth as I ever imagined it could be, and I fly with a lot more confidence than I did with my first engine. I now have over 200 flying hours on this new crankshaft (as of June 2006), with no problem. I place most of the credit on the nitrided crankshaft, but prop indexing is probably also a factor.

Read the comments by experts below, and you'll come away with the appreciation for nitriding and carefull attention to journal radiusing that I now have. The crank in my current 2700cc engine was nitrided to a depth of about .015" at Advanced Heat Treat Corporation in Monroe, Michigan, (http://www.ahtweb.com/) at a cost of $150 (they have since informed me that for aircraft use, it's more than that). That's still some cheap insurance! My new (next) 3100cc engine crank was nitrided at Nitron, Inc.,25 Wellman Street, Lowell, MA 01851, phone 978-458-3030 (ask for Pramod). Price was fifty bucks, but turnaround time was slow.

Earlier....July 5th, 2005
I had 50 hours on my 3100cc Corvair engine when the crankshaft broke (at 5000'). Well, really there's no telling how many hours it had on it before I stuck it in my airplane. After all...it IS 40 years old. This webpage will provide a description of the problem, and eventually (I hope), detail a solution. Sorry about the large photos, but I figured you'd want to be able to get a good look at the fracture. Please send comments to N56ML@hiwaay.net. Any idea as to possible causes or the mode of failure would be welcome, as well as clues as to what to check or change to prevent this from happening again!

Here's the big picture of what the crank looks like, with prop hub on the right end, harmonic balancer (and flexplate) on the left end. The break is right next to the front main bearing, and originated from the foreward-most rod journal, #6.

Here's the break area. The crack extends from the #6 rod journal to the first main journal, fillet to fillet.

Here's a closer look at the area. The "beach mark" striations point to the origin of the crack, like where a rock created ripples in a pond. The fillet radius is where the fracture originated. There are at least two large crack areas that met up at the oil hole. The oil hole is not the origin, but the localized stress concentration is where the cracks met and joined, forming the "cliffs" that you see. A real closeup is next.

A closeup of the rod side, showing where the crack originated at the fillet radius. Nice radial cracks very close to initiation site are called "ratchet marks". The fact that there are so many of these marks is an indication that the crank failed in several places at once, indicating an overall overstressing of the material, not just one imperfection that weakened the crank. This crank was magnafluxed by my local crank grinder before installation. Fuzz is from the paper towel that I wiped it off with.

This is the prop hub end, showing the main journal, along with a main bearing and a rod bearing. They are quite trashed by contamination, and of course the front one was wobbling all over the place after breakage, so it's not going to look very good after that abuse. I can see how this could happen pretty quickly at 3100 rpm, maybe in the 12 seconds that I went from "fine" to having to switch the engine off! Oil pressure was 39 psi right up until I throttled back. There was a quarter inch of aluminum shavings in the oil pan, most of it looking like it came off the cam gear in the last few seconds of operation, and a lot of finer stuff shaved off the engine case as the crank throw tried to machine its way to freedom. Oil temps were fine, and there were no other indications of problems (according the the Engine Information System which records EGTs, CHTs, and a lot of other info). No burned looking bearings, but they were all scored by aluminum chips.

A closeup of the photo above. Shiny polished places are where the crank rubbed against itself before I shut it down.

Here's a side view of the break at the #6 rod journal. Obviously almost a 45 degree angle where the beach marks are. You can also see the fillet radius at the bottom of the photo. Late Corvair "stroker" cranks are made of 5140 steel, and are forged. This one was ground .010" under, as are almost all Corvair cranks used in aircraft. Standard cranks are hard to find. It was not nitrided (and neither are most flying Corvairs), but I've come to realize that it should have been!

Here's a picture showing the radius on the #6 journal (left) compared with the radius on one of William Wynne's "aircraft" cranks. There appears to be no appreciable difference, so I believe incorrect fillet radius can be ruled out.

Update, as of July 3, 2005 - Preliminary "report"

I have heard from five experts in fracture analysis who've seen the photos above.

The first was introduced to me as "the world expert on aircraft crankshaft dynamics and failure modes". First thing he asked was "was this crank nitrided, and the prop indexed properly?". His analysis is that "combustion forces are the primary source of the bending fatigue fracture", and there was "no torsional component". "The fracture is multi origin bending fatigue from the forward fillet of the #6 crankpin. The position of the origin is approximately coincident with the occurrence of peak combustion pressure from #6 cylinder, which is what I would expect in the case of a primary bending fracture. I see no obvious defects in the fracture origin area, including no unusual geometric stress raisers in the affected fillet. The multi origin nature of the fracture indicates that the part was overstressed by a moderate amount. You need a stronger crank (nitriding) or lower combustion forces. "

When I asked for details, he mentioned that one possibility was that I've bored it out to 3100cc (from 2700) and the rod is now transmitting 15% higher forces to the crank than it was previously. While you could argue that the same crank handles 180 hp in the turbo version, the actual total number of cycles at a load high enough to exceed the fatigue strength are rarely attained in the automotive mode, yet we call on it regularly to takeoff and climb in aircraft use. Also, 180 hp Corvair cranks are nitrided and mine was not! I should know even more after he gets a better look at it, and will report his findings. The answer to both nitriding and prop indexing questions was "no", unfortunately, but that makes it awfully easy to increase my safety margin next time around. When asked, he said there was no way detonation could break a properly designed crank.

Regarding nitriding and prop indexing, he wrote: "Your nitriding depth of .015 sounds acceptable. If the prior stress relief was above the nitriding temperature, there should be little warpage. For your two blade propeller, just index it 60 to 90 degrees from the #6 crankpin. On GA aircraft the prop is indexed in order to minimize blade bending stresses. The prop does not significantly affect crank bending stresses [on GA aircraft]. However, you will have noted that the front main bearings in a GA engine have a much greater effective length than in the Corvair crankcase. The latter is therefore more susceptible to crank/prop interaction."

Another expert is the metallurgist for one of the world's largest aircraft engine manufacturers. Upon seeing the photos, his first question was "are you sure that crank was nitrided?" He also said "you may be fighting a duty cycle problem". Keep in mind that "bending" is not just something that could come from the prop and hub, but is also what the #6 connecting rod does to that journal every time the cylinder fires on the power stroke, and returns the piston after the exhaust stroke. He also pointed out the sometimes when dealing with generously radiused journals, the edges of the bearings can "cut" into the fillet, creating a stress raiser.

We have a metallurgist on staff that spent years analyzing fractures in machine parts. She said "classic fatigue failure due to bending", and to prove her point, showed me a picture from a fractograph book that looked exactly like it. That still doesn't answer the question of exactly how, but I'm hoping to find that out from expert #1, who now has the parts in hand for a better look.

Another guy (pilot and homebuilder) who does this for a living for a large diesel engine manufacturer in Detroit said "textbook case of a bending fatigue failure". No signs of anything torsional. He said the break is normally closer to the source of the torsional problem, so that's another clue that the flexplate is not the problem. One of his co-worker experts also said it was highly unlikely that the extended length prop hub caused it, but I'm not so sure either of us go for that . And he suggested I seek employment as a fractograph photographer, since my pictures were so good!

Here are some nitriding comments by a metallurgist for a "major US helicopter manufacturer":
Nitriding does indeed improve dramatically the fatigue life in all failure modes, i.e. bending, torsion etc. But it only improve the cycles to initiation. The cycles to propagation quickly pick up once the fatigue progression sets in. After 0.015" of fatigue progression in a crank with 0.015 nitride depth = no benefit from nitriding. To get even better fatigue benefit, the heavy duty market, i.e. Cat, Cummins Diesel etc. will induction harden the throws and many times will also incorporate a rolled (compression) radius coming off the throw. Typical induction hardened journals will be in the neighborhood of 0.050- 0.075" deep and a rolled fillet can impart a compressive zone upwards of 0.100 deep. Always remember cracks ONLY open under tension. Any residual compression buys countering insurance. Nitriding, Carburizing, Induction Hardening, Rolling and shot peening all belong in the powerful chapter of Metallurgy dealing with fatigue improvement. Carburizing and induction hardening on a unit stress basis probably imparts the best fatigue improvement for cranks. The gyroscopic forces imparted from a spinning prop are to taken very seriously and treated with the up most respect. I spent a lot of years working for a big helicopter mfg. - they use REALLY BIG props.

He also wrote: To do it right, the sequencing on your special processing needs to be correct. I would #1 Stress relieve, #2 Nitride, then #3 shot peen. The nitriding process is a heat treating process at around 980-1000 deg. F (depends if it is a 2 stage process to minimize the white layer) I would definitely request that the hard brittle white layer left from the nitriding process be removed by citric acid or abrasive blasting. If left on, it in itself can cause fatigue problems in the future. The beneficial effects of shot peening go away at around 600 deg. F sooooooooo........ must do nitriding first, then shot peen last.

From New Zealand: I was shown your "lovely photos" by a co-worker who is building a Corvair engine for his aeroplane. They are amongst the best photos of a classic crankshaft fatigue failure that I have ever seen. I have 26 years (mostly Military) fulltime NDT experience in 5 methods and have seen quite a few broken crankshafts and many with cracks. I have read your excellent article and thought I would add my 5 cents worth. You are on the right track with the Nitriding, etc and it is essential that there is a LARGE radius at the corner of the journal, the bigger the better, but it depends on the amount of room you have. "Good" machine shops will actually work this out and "dress up" a grinding wheel to the correct radius for your job.

The two points of origin of the initial cracks indicate to me that there was either existing cracking in the radius missed by Magnafluxing, or areas in the radius of "localised grinding burns" caused from when the crank was ground undersize. It is rare for a good Magnaflux operator to miss crankshaft cracks as long as the equipment and processes are "up to specification". Grinding burns can only be reliably detected by etching the journal with a special weak acid to show up any "decarburised" areas. These areas are very brittle Martinsite and cracks WILL start from there, even if the item is just in "normal" service. There is a Mil Spec available for the "Temper Etching Inspection" , its MIL-STD-867A (USAF), however most books on metallurgy will cover what/how to do it. Grinding burns are not limited to crankshafts and can happen any time there is a grinding operation carried out. We do a lot of grinding of aeronautical parts and all critical parts are Magnafluxed and Temper Etch inspected, irregardless of who was running the grinder. Most grinder operators are aware of the causes of grinding burn, but it can also "just happen" and not be evident to the eye. Well, good luck with it all. This has probably just given you something more to consider as you seem to have most of the other bases covered in your article.

What this is starting to look like is that there is no smoking gun pointing to anything "easy" like an existing imperfection, improper radiusing, oil starvation, foreign object scoring of the crank, or anything else other than simply exceeding the duty cycle fatigue strength for the crank. It simply saw too many cycles at a level that was above the fatigue limit, and failed, just like cranks do that have been overloaded for too long. This could be a combination of any or all of the following: