"Clamshell Effect" bolting together parts with flat surfaces 7

[kidvb]

We a reproducung a part for a customer that is two pieces bolted together as an assembly. The mating surface of each part is flat w/in .0003". The surface opposite both of the mating surfaces is parallel w/in .0008".

When the parts are bolted together, we use four bolts torqued to 185 in/lbs. The problem is that the overall parallelism of the assembled and bolted parts goes to .0035-.0040"!

The Design Engineer from the customer mentioned the "clamshell effect". This makes logical sense when you look at the parts as the four bolts are grouped near each other and poorly placed for uniform compression. In addition to this, the mating surfaces don't make a complete surface across the entire part and the bolting is all on one side.

When the bolts are at 50 in/lbs, the overall parallelism is .0005". At 100 in/lbs, the parallelism is .0012". At 150 in/lbs the parallelism is .0035". By compressing one side of the part, the other side spreads apart.

The attached jpg shows the approx location of the four bolts (blue) and the mating surface area (red).

Has anyone heard of this before? Is there any information on this?

Thank you,

kidvb

 

[rb1957]

have you ever tightened a cyclinder head onto a cyclinder block ? there is a sequence for tightening the bolts.

i think you have a good description of your problem ... the bolts don't offer a uniform pressure onto the contact surface. since the bolts are not uniformly spread about the contact surface, i'd suggest varying the preload (or torque) applied. you might tighten in steps, say 25% of the final torque, to see how much torque you need on the other bolts to maintain a parallel contact face.

 

[MikeHalloran]

I haven't heard of "The Clamshell Effect" before.

I have seen it, however; cinch up 3 adjacent bolts of an 8 bolt flange, and the flange leaks. Duh.

In this case, the part approximates an 8 bolt flange that only _has_ 3 adjacent bolts; it will never _not_ leak.

Well, maybe with a bit of Loctite on the faces. ... or if the assembly is a pump sort of thingy, not clear from the provided information, maybe the hose clamps on the nipples will even out the stress a little, so you get an assembly whose performance is highly dependent on assembly technique.

I suppose it could serve as a bad example in some future design book.


Mike Halloran
Pembroke Pines, FL, USA

 

[Tmoose]

What is this made of? Non-metallic, I'm guessing.
How thick are the flanges? What medium needs to be contained?
How much pressure?
If 185 in/lbs makes the flange bow, and the parts are metallic, I'm guessing the faces are not really vary flat.

If the bolt pattern can not be improved, and the material and flange geometry are rugged enough to support it, I'd think about grinding/bending a profile into one mating face so the bolts nearly at 180 would be at the most open spots.

http://rspa.royalsocietypublishing.org/content/461/2064/3955/F2.large.jpg

depending, it might be easier to use an o-ring or gasket

 

[MiketheEngineer]

Sounds dumb - but maybe your bolt holes are "too" tight not allowing the bolts to re-align as they wish??

And staggering the tightening is a given - even done on car tires.

 

[1gibson]

If you are stuck with a strange geometry (sounds like you are) have you considered washers or custom shaped plates underneath each bolt head to spread some of the clamping force?

Has this design ever been succesful, or did they change suppliers instead of correcting the design?

 

[kidvb]

People,

Thank you all for the input!

Here's a little more information.

The two parts are Titanium, 6AL-4V.

I did try several techniques on the tightening of the bolts. Mostly bringing them up in 50 in/lbs increments uniformly. Once I get to 150 in/lbs, the distortion takes place (out of parallel condition).

I was wondering about the compressive strength of titanium. It stands to reason that if the titanium was compressing .0001-.0002/in, over a 4 1/2" distance we could see .00045-.00090 compression in the area of the bolts. In the other areas, where the compression didn't take place, the material could act as a fulcrum and increase the distance to an out of parallel condition.

Still speculation, but it makes sense.

kidvb

 

[rb1957]

where are you measuring your "out of parallelism" ? your reading could be very sensitive to some changes at the interface.

how consistent are the interface flanges ? if there's an out-of-plane stiffener that exaggerate small changes at the interface.

what's the problem with this result ?

how well is the joint sealing ??

 

[MintJulep]

Well the sketch is just confusing.

But it sounds like a stiffness problem to me.

Stiffness is a combination of material properties and geometry.

 

[MiketheEngineer]

You may have answered your own question -

Does it need to be so tight??

 

[kidvb]

Thank you all again!

I agree, the sketch is very vague. I was hoping that it wasn't so vague as to be useless as I cannot post the exact configuration.

This isn't a pressure vessel. In simplest terms it is a two piece pillow block to hold bearings. The placement of the bolts was dictated by a range of motion requirement.

I think that in conclusion, I have to attribute this problem to the stiffness (compression) of the material and the bolt placement. It was at the very least a stunning example the strength (or lack there of) of the design in terms of bolt placement.

Thank you again for all the help.

kidvb

 

[hydtools]

Will three fasteners do the job?
Eliminate the left fastener of the two that are closest together. This would leave three that are about equally spaced.

Ted

 

[iainuts]

A fastener and the joint are two parts of a set and they work together in a sense. As the bolt is tightened, it produces more load and stretches. The joint is defined as the parts surrounding the bolt. There is a volume of material beneath the head of the bolt that will compress. These two parts acting together (bolt and joint) act as two springs. The more you tighten and stretch the bolt, the more you compress the joint.

The shape of the bolt is pretty obvious, but the shape of the 'joint' is not simply the shape of the two parts being bolted together. For a bolt passing through the center of two circular plates, the joint is generally envisioned as a conical shape with the tip of the cone being under the head or nut as shown in this picture:

http://www.roymech.co.uk/images16/screw_17.gif

That's not the best way to visualize it, but it gives you the idea. In actuality, the compressive stress is highest in the joint material closest to the bolt and gradually drops to zero the farther away you go. The cone shape is intended to help visualize lines of equal compressive stress. The compressive stress isn't strictly a function of distance from the bolt centerline, it is also a function of distance from the points of contact under the bolt head and nut, so these lines of equal compressive stress fan out under the bolt head in a cone shape.

You might imagine that with the conical volume being compressed to some small degree, and the material farther away from the bolt having been compresed less, the two plates will have to dish slightly. That's hard to imagine for two flat, circular plates, but consider that the two plates in this conical area are now thinner than they were prior to the bolt being installed. Remember, the bolt stretches and the joint compresses. The joint actually gets thinner where the bolt is, so the two plates are forced apart by the surrounding material that isn't compressed as much.

You can actually SEE this affect if you take two rectangular rubber erasers and press them together in the center. You should be able to see the edges of the erasers move away from each other. You can also try putting one rectangular eraser down on your desk and press in the center with a blunt instrument such the blunt/flat end of a pen. In this case, the edge of the eraser bows upwards away from your desk.

What you're seeing is the same effect. Under the bolts, the joint material is compressing. In your case, the bolts are right on the edge of the part, which means that you won't have a nice symetrical bowing out of material, and the affect will be worse. There is no material to counter one side coming up because there's no material on the other side. You're compresing just one edge, so the opposite edge comes up farther due to the moment created. Try and visualize what the compressive stress looks like in the joint.

Again, you can see this affect with a rubber eraser. Put the eraser on the desk and press down on one end or edge with the flat end of a pen. Notice how much more it comes up at the opposite end compared to pushing in the center.

 

[kidvb]

Home Run!

iainuts, you have described superbly what I have been able to confirm with a secondary experiment.

I took one of the details ant bolted it to a 1" thick steel plate. I applied torque to the bolts and measured the parallelism as follows:

50 in/lbs, .0004"
100 in/lbs, .0010"
150 in/lbs, .0025"

These measurements are better when bolted to a steel plate than when both details are bolted together. A nifty experiment that displays your theory. Lucky for me I did the experiment w/in the last hour!

The coolest measurements were that the detail is shortest near the bolts and ranges out of parallel to the worst condition farthest away from the bolts.

Titanium, as tough as it is, certainly compresses at a much higher rate than steel. I'm not sure exactly what property directly correlates to this phenominom (tensile, elasticity, yield), but it is very obvious as the torque values are increased.

I regret that I cannot share the exact configuration of the part as it is proprietary to our customer. This was an invaluable lesson in the transfer of energy for me.

Again, Thank you All for your help!, especially you iainuts who explained the theory so sweetly.

kidvb

 

[JustinMEEE]

You could lay some fuji paper between your parts to get an understanding of how the contact pressure is distributed. By selecting the appropriate pressure range, you may be able to detect how your parts are deforming do to the compressive stress at the bolts.

Larger bolt flanges will reduce your compressive stress.

 

[Tmoose]

Is the goal is to eliminate the problem of joint opening using the existing design, or is it sufficient to show the existing design is creating a problem?

what size are your fasteners?
what is the flange and mating component thickness?
Is the mating face machined flat at the end of manufacturing?

I still think the manufactured geometry may not be as flat as you expect. Clamping adjacent to a high spot would hoist the far end into the air more easily than compressing titanium with a few thousand pounds of clamp load.

I'd gently rub the flange face on a sheet of 600 grit wet/dry sandpaper on a surface plate.