Friday, July 16, 2010

Faulty NTSB Conclusions N9348S

I have to stand up for the mechanic as this NTSB  report's  "probable cause" is just plain stupid.


NTSB Identification: CEN09LA209
Accident occurred Sunday, March 15, 2009 in Bellefontaine, OH
Aircraft: BEECH B24R, registration: N9348S
Injuries: 2 Uninjured.
During cruise flight the pilot noticed abnormal engine noises and a partial loss of engine oil pressure. He immediately diverted to the nearest airport, but during the turn to base leg the engine oil pressure dropped to zero pounds per square inch and the engine seized. The airplane was not in a position to reach the runway threshold or to clear the airport perimeter fence. During the landing rollout the airplane impacted the airport perimeter fence, damaging both wings and the nose landing gear. An engine teardown examination revealed that the Number 3 cylinder connecting rod assembly had separated from its corresponding crankshaft journal. The journal surface was blue in color, consistent with exposure to excessive heat and lack of lubricant. The oil suction screen was obstructed with bearing material. The Number 3 cylinder connecting rod cap was found jammed beneath the counterbalance weight. One of the two connecting rod stretch bolts remained intact. The corresponding nut was found finger tight. The measured torque for the Number 2 cylinder connecting rod bolts were significantly less than the manufacturer's specification. The engine had accumulated a total of 3,799 hours since new and 492.7 hours since its last overhaul in 1999. The engine was last inspected 23.9 hours before the accident occurred.
The National Transportation Safety Board determines the probable cause(s) of this accident as follows:
The inadequate torque of the Number 3 cylinder rod bolts by maintenance personnel, which resulted in a failure of the connecting rod and a total loss of engine power.
---------------------------------------------



" The fallacy here is the idea that measured bolt torque today is the same as what was originally applied some time in the past? The ASTM Specification for Structural  Joints sums up the problem: "The uncertainty of the effect of passage of time and exposure in the installed condition." I would add especially after the engine broke into pieces possibly stretching and bending the bolts that remained intact!

When using a torque wrench to break-loose a nut the torque wrench is measuring the amount of torque required to overcome friction.
1. A little bit of off-torque applied to the wrench might be just enough overcome friction and get the nut to turn if the twisting force was locked into the bolt.

2. Friction might have changed due to time or the forces of engine failure on the bolt. The NTSB took a friction measurement that has an unknown relationship to torque applied some time in the past.

3. The two "intact" bolts most certainly suffered from some unusual loading as the engine destructed. If they were yielded and stretched then of course the nuts would be loose.

4. Fretting of the faying surfaces is a classic example of how torque is lost due to surface wear. So the answer is no for 4 reasons.

Torque is a twisting force that tightens the joint not only by stretching the bolt along its axis, but also by twisting it. We stretch AND twist the bolt into a coiled spring.  The stretch is necessary as it keeps our surfaces clamped together, but what is the twisting force doing? It might have gone into loosening our nut slightly the moment we released tension on our wrench; or it might still be locked into the joint in which case all of the wrench force you used to turn the nut may still be there trying to loosen the nut. It all depends on the friction between the nut face and the faying surface (an important lubrication consideration).

What happens if we use a torque wrench to “check the torque” on a bolt? If we torque the nut until we reach the “breakaway torque” we have to apply enough torque to overcome the friction between the nut face and joint surface. Until this happens none of our torque is felt by the bolt. Once the nut face is released our applied torque starts to twist the bolt until we apply even more torque until it overcomes the friction between the male and female threads. Only now does the nut move relative to the bolt and we detect the “breakaway” torque. Notice that the only thing we are measuring is friction. Everything depends on friction. If the friction is more or less than when the bolt was originally tightened then our “breakaway” torque will be more or less than the torque that the assembler applied to the nut. That is the quandary.  Did the assembler improperly torque the nut or did the friction change? We have no way of knowing.

“The nuts were loose.” Can we check assembly torque by loosening the nut?  Same problem but only worse because if the twisting force is locked into our bolt then it might take only the smallest amount of torque to get the nut moving (our breakaway torque) as the bolt tries to unwind.

But what if the nut really is loose? Then take your pick; all or some of the reasons listed below:

1.Embedment relaxation of the faying surfaces
2. Bolt stretch from metal creep (especially at high temperature)
3. Nut backed-off  due to vibration loosening
4. Wasn’t tightened properly to begin with
5. Elastic interactions between multiple bolts in a flange has reduced preload (crosstalk).

Checking the degree of tightness in a joint by using a torque wrench to measure the breakaway torque is not accurate and leads to incorrect conclusions.  Beware of inspectors carrying torque wrenches.

" When bolt pretension is arbitrated using torque wrenches after pretensioning, such arbitration is subject to all of the uncertainties of torque-controlled calibrated wrench installation that are discussed in the Commentary to Section 8.2.2. Additionally, the reliability of after-the-fact torque wrench arbitration is reduced by the absence of many of the controls that are necessary to minimize the variability of the torque-to-pretension relationship, such as:
(1) The use of hardened washers;
(2) Careful attention to lubrication; and,
(3) The uncertainty of the effect of passage of time and exposure in the installed
condition." quote from Specification-for-Structural-Joints Using ASTM A325 or A490 Bolts


No comments:

Post a Comment