Why M96 986/996 Heads Crack

Click here to see video of a  996 3.4 M96 head being pressure tested, revealing a microscopic crack that was not visible to the naked eye. This particular head was a prime example that a head can not be judged to be crack free without being pressure tested. I know where to look, but I didn’t see this crack til I subjected it to 100PSI air pressure. 

Anytime an engine overheats it’s no big surprise when it leads to a cracked head. 

But in many cases overheating was not reported. Indeed I often hear, “My engine did not overheat, so why did I end up with a cracked head?” 

I can not give a definitive answer, but I have some ideas.

–I have flushed water pump impeller blade(s) from many -but not all- of the cracked heads I’ve repaired.  Even if the pump has shed just one blade, if that blade makes its way to one of the very narrow internal head passages that feeds coolant to a specific area, that area will see a severe spike in surface temp.  And if that area is one of those discussed in the Crack Repair article as having a thin cross section a crack is very likely, if not certain to appear. In this case the overall coolant temp would be within the normal operating range, but localized overheating will occur. A guage would be of no help in warning of this sort of overheating. 

If the head failure occured after the replacement of a water pump that shed one or more blades, I believe there is a high likelyhood that we can pin it on a blade lodging in a critical passage.

It’s tough to explain why pulling the heads after a water pump failure when the engine did not overheat might be . But these pieces floating through my engine would worry me. Seems like something of a crap-shoot.     

Here’s a picture of a typical collection removed from one head. I have extracted these from both cracked and un-cracked heads.  



I would expect that the water pump was replaced because of some sort of overheating scenario.  I believe it is pretty likely that the blades from these pumps don’t all break off at the same time; meaning at least one has probably been lodged somewhere in a head for some time. Even a brief, minor overheating event that was caught quickly by the driver can be harmful if one or more passages in a head is already clogged from pump blades. Crack propagation can occur over time and may begin with one major event, like a pump failure, blown coolant hose, etc, but take some period of time to spread from the water jacket to the surface.   

On engines that did not suffer a pump failure or overheating event, but still managed to crack a head a potential explanation could be that the thin areas that typically crack are gathering points for casting stress.  With the head mounted to a precision fixture on a mill, dial indicator mapping of the machined surface upon which the follower housing sits reveals that this surface warps, much like the deck of a head will.  Every crack I’ve seen walked across this surface.  On the M96 3.4 heads – which comprise over 90% of the cracked M96 heads I’ve seen- I routinely measure .002″-.005″ variance from twisting and sagging of the surfaces. These variances are evidence of stress. The act of cracking serves as a form of stress relief for any casting.  

These stresses are very much like earthquake fault lines. When the stress reaches threshold a crack will appear.  Of course thermal shock increases the stress on these areas.  The sort of stress caused by rapid warm-up of a cold engine could be enough to push the stress level past the threshold point.  Heavily loading a cold engine greatly accelerates the rate of heating- and therefore expansion- of the cylinder heads.  That significantly elevates the stress throughout the casting and could certainly lead to or hasten the formation of a crack. This condition would be magnified in winter months. And I have seen plenty of cracked heads in the winter months.