Stress Corrosion/Hydrogen Embrittlement/low Clampload help

[preload]

This is the new problem I am working on, on one of the mounting bolts. This time I will try my level best to give you guys as much of info as possible with pictures. This is our major issue (Customer satisfaction)

Problem where find: In the field, somewhere between 5hr to 440hr of engine run in water (no problem in assembly). Problem is seen only on high horsepower engines.

Joint Description: the whole engine will be supported by a structure, the engine is mounted on to the structure with 2 upper mount bolts and 4 lower mount bolts. So in service the whole load is carried by those 6 bolts and the problem is with the 2 upper mount bolts which are breaking underhead. Warranty issue. Some of them breaking at underhead of bolt and some of them come loose.

Fasteners used: ½-13 UNRC-2A,17.4 strength bolt (grade 8 high strength stainless steel) cadmium coated, fastener driven into blind hole (no nut), split lock washer under head and loctite 272 on the threads.
Depending on the heat treatement of the bolt, the proof load of the bolt can vary from 14,687lbs min to 17500lbs max. our supplier can get a mixed fasteners so we really want to keep the 14,687lbfs as proof load for margin of safety.But our tensile lab did some tests on 4 production bolts and they found the proofload to be 18000lbs.

Clampload we are shooting to: 9000 lbs average on each bolt.

Service loads on the joint : don’t know, design engineer dosent know either,but he said he gonna try to get the number.

Engineering dept thought: our eng dept thinks that , the failure/fracture of bolt head is due to hydrogen embrittlement or stress corrosion failure. But we see some of the bolts failed in 5 hrs of the engine run, so how can it be a corrosion failure if u have the engine in water for just 5 hrs? some of them failed in 440 hr (this I can agree it may be of corrosion failure)

What my guess is 9000 lbs of clamp load is like 50-60% of the bolt proof load. I think the bolts are failing because of the low clampload as cyclic loading/vibrations.services loads acting on the joint.

What do u guys think?

 

[preload]

Screwman,

Yes our material lab confirms the same thing. But corrosion is something which u expect to see after the engine runs for some hrs in the water. But if u see the pic below,shows the no of hours the engine run in service and the failure mode.

http://img337.imageshack.us/img337/5465/hrscc9.png

u can see some of them broke, or fractured in 0 hr or 5 hrs. In this case how can we attribute the failure to corrosion?

Volpe,
Depends on the customer use, some of them will be in salt water (florida) and some of them will be in clear water (in lakes)

 

[CoryPad]

It is possible that the difference is due only to the different Loctite products. 242 is a medium viscosity product, while 271 is a low viscosity product. Perhaps 271 is squeezed out of the joint more than 242, allowing more metal-to-metal contact, and hence, more friction.

Regards,

Cory

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[Screwman]

One thing that you should do right away is to eliminate the helical "lock washer". It is contributing nothing to keeping the joint intact and may be creating a stress riser on the underhead surface that is contributing to the failures. 242 Locktite should be more than enough to keep that bolt from loosening.

 

[TVP]

I completely agree with Screwman-- get rid of the lockwasher. The Loctite plus preload is sufficient for resistance to loosening.

 

[preload]

Unclesyd,

Here is the info you asked for,

Fastener material - Stainless Steel: Group 7 per ASTM F593 Except to age condition to H-1100 (HRC 30-38 REF) with cadmium plating. I have the the treatement process for the fastener with me. Please let me know what you want.It is a 16 page process , so I dint upload it. I can do that if u want.

Washer: split lock stainless steel washer. Fastener under head contact surface.

Tapped hole material: Investment Cast SS CF-8 per ASTM A743-81A. Fastener thread contact surface.

Please let me know if u need more info

 

[preload]

TVP & Screwman,

Yes, we are planning to get rid of that lock washer for the next year model.

 

[volpe2]

You can test your inventory for hydrogen embrittlement. See the ASTM F519-06e1 Standard Test Method for Mechanical Hydrogen Embrittlement Evaluation of Plating/Coating Processes and Service Environments for details. It is a fairly quick (4 days max, often times 2-3 hours) and dirty way to check your inventory.

Essentially you load the bolts up to the proof load on a fixture and wait 96 hours. Remove them and if they haven't completely failed check for cracks.

If you have lot control you can sample a few from each lot. If not, take a larger sample from your inventory.

It is a destructive test, so even if they pass do not use the test specimens.

Also, stainless steel can be effected by localized galvanic corosion if exposed to salt water. The Cad plating provides a sacrifical layer, so you may want to keep it or something similiar (zinc).

good luck

 

[MikeHalloran]

Odd. Loctite 271 is normally used for sealing porosity in castings, or for application _after_ wrenching a threaded connection. It's plenty strong, essentially permanent, but it wicks into every available crevice, which may not be a desired behavior in all circumstances. Its low viscosity suggests that it may not be a particularly good thread lubricant.

The 242 is a better choice if you ever want to get the joint apart.







Mike Halloran
Pembroke Pines, FL, USA

 

[preload]

I just did a comparison study using skid more and ultrasonics in the lab. I see a real big difference in the clampload in production line joint and in the lab

In production line, I gave the machined bolts to the operator and asked him to run them at 66 ft-lb torque and used ultrasonics to see the clampload and I got an average of 14500 lbs .

I replicated the same test in the lab using skidmore and also using ultrasonics (used both to see if there is any big difference between ultrasncs and skidmore).
On skidmore I always got less than 10000lbs for the same 66 ft-lb torque. Please see the attached results.

http://img255.imageshack.us/img255/3080/skidmoreil4.png

What is the reason of getting higher clamploads ( around 5000 lbs) for the same torque on production line than in the lab.

FYI: clutch tool we are using on line is doing really good. I cannot attribute the reason to the gun.

And also could you please help me understand why the torques are different to achieve the same load in different bolts?

For example, if you see the link, to achieve 10,000 lbs
Bolt 1 took 94ft-lbs
Bolt 2 took 77ft-lbs
Bolt 3 took 83ft-lbs

 

[preload]

For those who dont know,
minimax is the ultrasonic instrument.

Initially I thought if skidmore is doing something wrong, but I get the same value of load when I cross checked with ultrasonics minimax. so skidmore proved that the results are good.

For some reason, the same minimax is hsowing very hight loads,with the same torque on production line. I dont understand why?

Thanks in advance for your ideas

 

[CoryPad]

Friction variation and geometry variation can explain differences of 10-15%, which is what you are seeing.

Difference between lab and line results can be due to tool speed variation, which affects friction.

Did you know that the most highly engineered fastener coatings have friction variation of [±] 25%? For example, Magni 565 with [μ] = 0.13 [±] 0.03.

Regards,

Cory

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[preload]

Cory,

Does friction due to tool speed and goemetry variation has such a big effect (around 5000 lbs difference)?

friction variation gives me the answer to clamp load variation, but how can I attribute the friction variation to consistent low tension in skidmore and consistent high tensin on line?

Thanks a lot.

 

[CoryPad]

Usually not a 34% difference (5000 out of 14,500), more like 5% to 10%. Do you know the rotation speed for each test? If the Skidmore test was 10 rpm and the power tool was 10,000 rpm, then maybe it can account for this. Your materials and coatings are not ideal to minimize variation.

Regards,

Cory

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[preload]

Cory,

If I have a grip length change from lab to production joint,will there be a difference in clampload for the same torque and same environment?

 

[Screwman]

When doing the comparative tests make sure that you duplicate both the production line nut member and bearing surface. Each of those will contribute nearly 45% of the total friction in the joint. Also use all new (virgin) components for each test. Once the parts have been used you change the friction characteristics significantly.
One thing that mey be contributing some amount of difference between the production line and the lab could be the comparative differences in joint stiffness between the two systems. This will only account for maybe 10% difference though. But with the "lockwasher" in the joint you may be getting quite a bit of emmbedment when power tool tightening.

 

[preload]

Screwman,

If there is an embedmenet due to lockwasher and power tool on the production line, then I should see less clampload right? But I am seeing more clampload on the production line.

The differences in the test are

On the skidmore, our technician used hand wrench , in production line its clutch tool (will find out the speed)

Tech dint use any loctite on lab joint (on production line we use loc 272)

Tech used washer plates to simulate the joint and grip length(geometry variation),In production its actual production joints

 

[TVP]

Did the technician use any lubricant at all? If not, then this is the only answer you need. As mentioned previously, Loctite 272 will act as a lubricant during tightening.

 

[Screwman]

The locktite and also the use of washer plates if they were what the lock washer bore up against, along with the power tool vs hand tool all added to the variation.
You can not believe how critical lubrication consistency has to be on some joints. We had a case where head bolts were breaking at random and it turned out that a leaky cylinder was occaisionally dripping oil into the tapped hole and that was enough to cause the bolts to tensile during installation. Process control is critical on really fussy joints; everything has to be kept spot on for the units to turn out right.

 

[unclesyd]

Make sure the washers that you are using on the test stand are ground and better yet hardened. Stacking washers can act like a short spring.

The mention of speed of tightening can play an big part in the actual applied torque. We had an incident where the mechanics were running in a 3/8-16 H11 SHCS with a speed handle and then walking around the fixture with a torque wrench. The torque wrench clicked at the 45 ft lb set point but the actual torque was above 55 ft lb. We lost fasteners when we went to dismantle the fixture.



Here is a nice bolt stress calculator that I ran across while looking for an article on the values of K and thread friction factors. Click on the high high level of detail buttons to get all the results.

http://www.turula.com/bolt.html

 

[preload]

Unclesyd,

Thanks for the ideas.

In the link u provided, how can I get the value of
"Bolt Torque Factor = q = ratio of torque in the stressed bolt cross-section to the torque applied to the bolt"