Fastener Tightening Strategies Help Requested 7

[iqsy59]

As a Quality Engineer for an automotive assembly plant, I have placed myself in the middle of a heated debate over fastening strategies. We currently employ a torque controlled/angle monitoring method that appears prone to inaccurate results with varying degrees of criticality.

It occurs to me that employing an angle controlled/torque monitoring strategy would greatly reduce these errors. However, the Manufacturing Engineering department is resisting my proposal with all of their resources, even though they have not presented valid justification for it. The Design group is on the fence, but is leaning towards supporting the fastener experts in the ME group.

I have mathematically applied the angle control method to 100 torque curves that were incorrectly passed or failed using the torque control method, and every result would have been accurate using angle control... at least on paper.

What are the true benefits and risks of each method? Surely there must be a disadvantage to employing angle control, but I cannot find it! I found one statement that this method is best used with fixtured tooling, but I cannot identify any significant error that would be introduced by handheld tooling.

Are there any experts out there who can provide insight?

 

[Metalguy]

In order for angular control to work right, the fasteners must be used with only a very slight (if any) difference in rigidity of the connections and being able to get a consistent starting point. In other words, it only works if the angle of rotation develops bolt stretch and isn't lost on flattening the mating surfaces, etc.

It really *can be* far more accurate than torque, which is known to cause large variation in bolt stress.

But the best way is to either measure bolt stretch (directly of ultrasonically) or actually pull a stud with a hyd. tensioner, assuming you can for your app.

Read up on J. Bickford.

 

[unclesyd]

I have to echo metguy's statement that you have to have a starting point for this method to be effective.

There are several papers out on turn of the nut(angular) method bolt tightening bolts. The most common use of this method is by the structural steel people.

There are guidelines and methods published by the Research Council on Structural Connections (RCSC) for tightening by turn of the nut methods.

You might want to check out this site as this instrument definitely would compare the results.

http://www.surebolt.com

 

[iqsy59]

Thanks for the tips so far! What I haven't heard yet is a reason for me to stop advocating the angle-controlled method. I'm just certain that I'm overlooking something. I just can't figure out what it is!

What I don't have is the advantage of an angle-controlled specification. I am given only a torque range. Does that preclude the use of an angle-controlled strategy? In our current situation, we are applying an arbitrary seating torque and then applying the specification nominal torque and measuring the angle between for a sanity check.

The problem is that there is no way for the tool to analyze what happened between the two values... and I have seen classic strip-outs passed by the tooling. The angle window is simply "opened up" when too many tightenings are rejected by the system.

My logic tells me that if I have known good torque curves, then I can use that linear relationship to achieve specification using angle, rather than torque, and then verify at the end that torque is within the value specified by the print. I have applied the theory on paper and it seems to work.

The designers are skeptical because I would no longer be targeting nominal, but to me if the torque curves are consistent, my Cpk should be excellent. I guess what I'm looking for is a reason NOT to advocate the angle-controlled method on a torque-controlled specification. I know the current torque-controlled system rejects good fastenings about 8% of the time and I have examples of classic strip-outs AND cross-threads being accepted by the current system.

Regards,
Michael

 

[Metalguy]

Before we spend too much time guessing at your bolting arrangement, can you provide meaningful drawings/photos? There are way too many variables involved.

 

[MintJulep]

Iqus,

If you are trying to develop an angular method of bolt tightening you need to stop playing around with torque to do it.

The goal of tightening a fastener is to develop a desired tension in the fastener. The relationship between torque and tension is really pretty tenuous.

 

[iqsy59]

I know that I am looking for an easy answer to a complex problem. To answer your question, Metalguy, My thought was to employ the following method to all of our automotive assembly Atlas Copco tools (approximately 100), and then dispatch the few with inconsistent joints (ie crushing or rubber) that did not work:

1) rotate to seating torque
2) rotate a predetermined angular rotation based on known good torque curves
3) verify final angle is within specification

Currently, we do it backwards:

1) rotate to 50% of nominal torque
2) rotate to nominal torque
3) verify angle is within predetermined value based on torque curves.

Our current method has too much error, in my opinion. For one thing, it gives me zero torque tolerance because the tool will continue to attempt to reach nominal in every case. Is there a reason why I wouldn't want to implement method #1 across the board?

Regards,
Michael

 

[mloew]

Michael,

Perhaps this will be helpful to you:

2001-01-0978 A New Tightening Method for Bolted Joints By the Simultaneous Application of Torque and Compressive Force

By:
Kazuya Mori - Kumamoto Univ.
Shinji Hashimura - Kurume National College of Technology
Yukitaka Murakami - Kyushu Univ.
Kunihiko Mineki - Central Motor Wheel Co., Ltd.

http://www.sae.org/servlets/productDetail?PROD_TYP=PAPER&PROD_CD=2001-01-0978

Best regards,

Matthew Ian Loew
"Luck is the residue of design."
Branch Rickey

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

 

[Metalguy]

Let me try to explain it this way: *IF* you are working with a very rigid connection, with flat machined surfaces such as installing a flywheel on a crank, the angular "turn-of-the-nut" method is FAR better than torque. But torque is better for non-rigid connections.

I don't know what kind you have.

 

[yates]

Mint Julep is right, the main reason for the better performance of the angle method (on rigid connections only)is the fact that torque tension curves are so inconsistent. It is useless to try to impose a tightening angle which corresponds to a certain torque, when the whole object of the exercise is to obtain a certain joint tension, not a torque.
Provided your connection is rigid (and not several layers of sheet steel for example - which makes finding the 'clamping point' very difficult to determine) the best method is to experimentally apply different angles to the nut while measuring tension, then apply the required angle during production.

 

[TVP]

FYI, RS Technologies has a couple of good technical articles available on their website related to this subject:

http://www.rstechltd.com/Technical_articles.htm

 

[griffengm]

iqsy59,
I will play devil's advocate for a moment.
First, it is likely that tools, procedures, and training have all been centered on torque as the tightening method of choice. What you propose may entail a huge expense to fix something that your ME dept regards as "not broke."
Secondly, if this connection must be duplicated in the field, the same issues are compounded.
I have little doubt that angle of rotation works but you need to either define the expected benefits of AOR better( as in $$$$) or show excessive costs to continued use of the torque method.

Griffy

Technology is great but why buy a laser when a match will do.

 

[drawoh]

iqsy59,

I have looked into angle control for some of my bolts too, and I have found the idea to not work. Consider a grade five bolt, 1/4-20UNC, with a clamped thickness of one inch. I will use a proof stress of 85ksi. Steel has an elastic modulus of 29e6psi.

Strain = proof stress / elastic modulus * Length
= 85e3/29e6*1in = 2.90e-3in

Turn of bolt = 2.90in * 20 threads/in = 0.06 turns.

That means that when I go from a snug bolt to one tightened to the proof stress, I have stretched it less than three thou. My structure and my washers each could be out of flat by this much. I cannot control this strain to within the tolerances I require.

All of the numbers above can be fudged, at least in theory. Use a longer bolt with a higher proof stress and a lower elastic modulus (titanium?). Make the clamped section as accurate as possible and purchase or fabricate precision washers. Use finer threads to increase the rotation of the screw. Hopefully, you are not clamping something relatively flexible, like a printed circuit board or a gasket.

I cannot work around any of this. Perhaps you can.

JHG

 

[CoryPad]

The contributions above have been good, but there are some missing details.

The causes for varying tension levels from a given input torque are 1) joint members not in solid contact which requires extra tension force in the fastener with no compression force through the joint; 2) friction variation at the thread interfaces and screw/nut head interface; and 3) prevailing torque, i.e. torque that does not produce joint compression, such as the use of deformed thread forms for anti-rotation or thread forming.

I have extensive experience with automotive fastening applications. The majority of critical joints (let us say thread size greater than M6) should be rigid/seated, so point number 1 is not common. Point number 2 requires friction control of fastener coatings (using specifications, calculations, and lab testing), and point number 3 requires lab testing.

In my opinion, there is vast opportunity to improve your operations using angle control strategies. The key item to consider is lab testing - you need relevant testing results (actual production parts and power tools) to determine the appropriate parameters (angle target, low & high angle limits, low & high torque limits). The answer to your question about the disadvantages of angle tightening strategies is that it requires information, testing, calculation, and effort - not technical challenges, but human challenges that may be difficult to overcome. Good luck.

Regards,

Cory

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

 

[CoryPad]

drawoh,

For your example, rotation angle theta = 21 degrees. Angle encoders for power tools are accurate to less than 1 degree, and typical tolerance would be 5 to 10 degrees. Let us say that you wanted to limit screw plastic strain to 5% (typical heat treated screws have plastic strain capacities of > 8%). For your 1 inch grip length, that translates into 360 degrees rotation. Say you chose a seating torque = 5 N m (3 ft-lb) and a tightening angle target = 40 degrees (which should allow for the part variation you stated). You still have 320 degrees more until your screw is nearing its limit. This strategy gives you enormous flexibility to accomodate varying friction levels, and by setting the appropriate windows (low/high angle, low/high torque), you can identify part quality problems, power tool problems, etc.



Regards,

Cory

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

 

[zepp]

With regard to the original question, If you are going to advocate angles after torque, one thing so far has not been mensioned, and that is using a bolt load stress meter.
Usualy in the construction industry, a batch of bolts from any manufacturer would have so many bolts tested, the test bolts would be screwed into the bolt load test meter using the specified torque, then the angle would be applied, this is the critical part, and I do understand why angles would not be prefered, without such a test on every batch of bolts that come in, you may be taking your bolt up past its yield point, and its this extra work that puts most people off, In the automotive trade, I dont know of any fasteners that are used in conjunction with angles, most bolting applications are now carried out these days with multip spindle nut runners or torque guns, the latter being prefered , with certain manufacturers of toque guns, angle encoders can be included, and with this unit, there is also a built in torque transducer that will feed back information to a desk top computer, it all sounds like way to much science, but once installed, its very easy to run.
but to be honest, If you have engineers that are against inclusion of angles to your bolting process, no matter what you try to explain, they wont go for it, some engineers like to work with angles, some dont, and the arguments for and against will never be fully satisfied.
I hope the info given above helps you in a way that you may be able to convince the engineering dept of both methods being used together and controlled from the same tool.

 

[CoryPad]

In the automotive trade, I dont know of any fasteners that are used in conjunction with angles

Your knowledge of automotive assembly is incomplete. Tightening specifications using angle are used widely. Many manufactureres use angle specifications for performance critical joints like connecting rod caps, main bearings, and head to block joints. They also are used for safety critical joints like steering rod to upright.

Regards,

Cory

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

 

[zepp]

thanks for the pointer. I havent seen angles quoted before, I have always had to deal with specific torque values where the automotive industry is concerned.
I will bear it in mind in future.

 

[vonlueke]

The approach in the drawoh post, above, looks incorrect because it omits the stiffness of the clamped plates. I think the formula for the angle-controlled (turn-of-the-nut) tightening method prior to yielding is:

theta = (sigma/E)(L)(360 deg)/[(1-C)*p],

where sigma = stress on bolt tensile stress area,
E = bolt tensile modulus of elasticity,
L = bolt grip length,
p = thread pitch (bolt longitudinal distance per thread),
C = k_b/(k_b+k_p) = bolt stiffness relative to total joint stiffness, where k_b = bolt stiffness, and k_p = stiffness of plates in bolt grip region. A formula for k_p might be, e.g., k_m in thread725-22449.

As an example, let's say we have an M6 x 1 bolt, property class 10.9, grip L = 30 mm, nut, washers, and plates stronger than the bolt, and let's say C = 0.307. Therefore, if the bolt is snugged to the zero angle position, the bolt yield strength (940 MPa) would supposedly be reached at theta = [(940 MPa)/(200000 MPa)](30 mm)(360 deg/thread)/[(1-0.307)(1.0 mm/thread)] = 73.2 deg.

Or, assuming C = 0.307 for drawoh's example gives theta = 30.5 deg.