Screw failure causes

[gersen]

Hello,
recently I had the a problem with a bolted flange. After a repair some of the screw head pop out (see the picture below).
Is it possible that the failure is due to a too high torque during bolting-up?
All the damaged screw broke at the base of the head.
The screw are Socket Head Cap M24 class 12.9, they bolted them with 2 (don't know why) split-lock washer DIN 127 B.

 

Without seeing good macro pictures of the fracture surfaces, I would say there are only two realistic possibilities for the fracture mechanism:
(1) Fatigue; or
(2) Faulty manufacture.
The statistics, the time of life, and the circumstance of a change in supplier all point to faulty manufacture. In which case hydrogen cracking is the overwhelming favourite mechanism. But do not rule out the second favourite, significant forming defects - you never get just one in any batch.
A full work-up at a metallurgical laboratory is necessary to confirm the above.

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

 

[drawoh]

chicopee,

Cyclic loads on a hydraulic cylinder?

That implies to me that the end plate is lifting up a bit. Wouldn't we be seeing some hydraulic fluid?

--
JHG

 

[chicopee]

The OP did confirm after my first reply that there was cyclic loading perhaps more likely, in my mind, to be pulsating. My take on his problem is that the bolts have been removed and put back on the flange perhaps than a couple of times for maintenance reasons and the cap screws were stretched more than acceptable while being torqued and became too hardened for the application. Don't forget that in one of his replies, the OP stated that the cap screw heads were popping out suggesting pulsating loads on the flange.

 

[gersen]

To answer yout question:
the screws were new off-the-shelf, we remove the hex bolts and changed them with socket head cap screws. These weren't subjected to fatigue, after the bolting up, with a pneumatic tool, the cylinder received a simple test under pressure, one chamber and the other. During test we didn't see any damage, after some day we saw some of the heads popped, the cylinders were on the ground, not installed on the machine yet.
We proceded to remove all the screws, during the unbolting some of the head broke and some remained attached but with a visible crack (the pictures I attached were from different screws). The damaged head, still attached, pop out with a single hit with an hammer.
I noticed that were used screws from 2 different supplier and only one type were damaged.

 

[hydtools]

Now you need compensation from the supplier/manufacturer of the defective capscrews.

Ted

 

[drawoh]

chicopee,

I quoted a torque of 1200N.m, above. This is based on 75% of yield stress, and no lubrication. The screw should not be over-torqued. If I do 90% of yield, I get a torque of 1500N.m. If screws are being installed this way and then removed, there is a good chance some of them are being overstressed, due to the inaccuracy of the torque process.

Did all of the screws from the second manufacturer break?

--
JHG

 

[Screwman1]

Is there any sort of chemical environment that may be touching the heads of these fasteners? P.C. 12.9 fasteners are very susceptible to environmental stress corrosion failures that will exhibit themselves as delayed fractures under load. The other likely root cause if there is cyclic loading would be fatigue failures, and from the photos there 'may' be beach marks on some of the broken parts, but given the resolution it is impossible to be sure at all of the failure mode. If I was involved in this, I would send 3 or 4 samples to a good metallurgical lab for a failure analysis, anything else is just a bunch of folks on the internet tossing around ideas without any actual proof.

Dick

 

[enginesrus]

Heat treat, too hard in that area, and or not a small fillet, stress riser from cutting tool, combined with cyclic loading from the impact wrench or the application. In some cases the hex can be too deep too. A hardness test would be interesting.