Bending Moment

[sullivnc]

Hi everyone,

Wasn't really sure where to post this. I have a problem I'm having trouble wrapping my head around. I have a piece of glass that is clamped around the entire outer edge. The frame is then secured down to a surface. The glass is circular with a flat (OD of 7.95", 281 deg 18'30" arc with a flat 5.04" long). It is flat, 0.197" thick. The frame has a circular OD, with a bolt pattern OD of 8.375". There are several variables I am looking at. Two different tightening patterns (going around the pattern vs. across), different torques (the correct torque of 9 in*lbs vs an over-torque of 15 in*lbs), and a raised fastener pressing up on the bottom of the frame. The fastener is 1/16" raised and at -22.5 deg on the bolt pattern. Basically, to start with, I want to look at modeling the bending stress created by these different conditions. Can anyone give me a good starting point? Thanks for any help.

 

[KootK]

I see finite element modelling in your future. You'll be fine posting here. We've got a few glass guys that will show up any minute.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[desertfox]

Hi

Post a sketch please it speaks a 1000 words

 

[drawoh]

sullivnc,

Is this thing working in a constant temperature environment, or do you need to work out differential thermal expansion? The first indication you are doing anything wrong will be when the glass shatters.

When I clamp pieces of glass, I always include an elastic element to provide controlled force. Elastic elements like gaskets and O-rings also provide a seal, which you may need.

Read up on screw torques. I would not rely on screw torques to provide a precise compression force.

--
JHG

 

[sullivnc]

There are 2 gaskets, one protecting every surface of the glass the contacts the frame, and a flat gasket between the frame and surface. There is a thermal element as well, which I know I need to work out, but modeling the physical stress is what's throwing me right now.

 

[rb1957]

"The glass is circular " ... like a dome ?

OD = 7.95", with a flat 5.04" long (dia?) ... that`s a pretty big flat.

"281 deg 18'30" arc" ... ?

another day in paradise, or is paradise one day closer ?

 

[sullivnc]

Think a circle with the bottom chopped off. 281 deg 18' 30" of the circle are still intact. The flat along the bottom is 5.04" long. The uniform thickness is 0.197".

 

[racookpe1978]

There is no bending stress to cause failure if all of the edges are clamped evenly and all edges are uniformly "flat" unless the whole assembly (flat side + and around the arced side) is resisting pressure from inide the lens assembly to outside the lens assembly. .

 

[sullivnc]

As I said, all the edges are not flat, there is a 1/16" protrusion at -22.5 degrees around the bolt petter pressing up on the bottom of the frame.

 

[rb1957]

maybe one way to model it is as a flat plate, supported at each bolt, and loaded with an enforced displacement.

alternatively model the plate and the ground (with the protrusion) and have the plate resting on the ground, then apply bolt loads progressively.

if you have to live with this protrusion (obviously), i'd suggest either ...
1) add a spacer so that the glass is supported on a flat surface, or
2) add shims at each fastener, to gradually remove the protrusion height.

tell us more about what this glass is doing ... architectural detail ?, window in a bathysphere ?, window into a pressurised compartment ??


another day in paradise, or is paradise one day closer ?

 

[sullivnc]

It's a landing light filter. We don't have to live with the raised fastener, our customer needs to hire competent mechanics who wouldn't just leave a screw protruding after they shear the head off. But regardless, we have to prove that that's a problem. It's Borofloat glass.

 

[rb1957]

so you have to keep the OML ... spacers are out.

how about grinding the glass ?

how about reseating the protruding screw ?

how about deleting the protuding screw ?

sure you can model it, but living with this is a Bad way to resolve the problem. if somebody's pooched it, fix it.

another day in paradise, or is paradise one day closer ?

 

[sullivnc]

The problem has been fixed as far as their mechanics installing the assembly, but the Air Force still wants us to prove that that was the main culprit and not an inherent problem without design. We have proved it in practice, but they also want us to prove it theoretically. I'm looking for governing equations, basically. I'm attempting to write a MATLAB script that will do the FEM required for this, but it's been a while since I did a 3D FEM analysis.

 

[rb1957]

"The problem has been fixed" ... by installing over the protruding fastener head ?

it sounds like someone has gone and installed this light filter over the raised head and didn't stop to think. and now the customer has said "that looks like a dog's breakfast, show me it's good". personally, i think you're skrewed ... there's no way to show it good ... modelling the installed stresses will take a month (maybe a year) of sundays ...
1) how will you model the stiffness of the supports (not rigid, but maybe rigid is a conservative assumption) ? as i posted above, you can model the ground plane (with the protrusion), model the flat glass, then load the individual fasteners, forgot to mention you'll need to run non-linear. I've got no idea about writing a model from a MatLab script ... i`d use a pre- and post-processor like FeMap.
2) this'll get you the installed stresses (which are clearly ok as the filter hasn't cracked), but this is when the real questions start ...
a) what about thermal effects (cold at altitude, hot on the ground?) ?
b) fracture at altitude (cold generally reduces the fracture toughness) ?
c) thermal fatigue ?
...

personally, the customer has called your bluff, and you should fold ... "ok, we'll replace it; it was bad workmanship in the first place"

another day in paradise, or is paradise one day closer ?

 

[sullivnc]

Again, this is not our problem. Air Force mechanics, while installing our device on their planes, over-torqued screws past our given specifications and sheared the heads off the screws. Instead of removing the protruding screw, they left it there and continued the installation. Thermal fatigue is my next problem once I understand how the filter is being stressed by the bending moment created by the raised fastener.

 

[rb1957]

sorry, didn't understand the relationships. i'm surprised that they`re accepting the probelm (which sounds like bad installation workmanship). wonder how their quality bought it ?

like i said, don't how how MatLab scripts would create a model.

thinking about the model, if you model the plate sitting on the ground-plane (with the protusion), it is a little easier (i think) to apply a load at each fastener to see the deflection induced, to see when the plate (at the fastener) touches the ground. playing with this will give you different results ... next question, how did they actually tighten it down ? another scenario is applying a proportion of load at all fasteners, and seeing how it deforms.

another day in paradise, or is paradise one day closer ?

 

[sullivnc]

To show you the intelligence level of these mechanics, they are starting at the bottom of the pattern, and then just going around, rather than going across the pattern and tightening in increments like we told them to. The reasons the glass is cracking are obvious: a bending stress due to the raised fastener and improper tightening pattern, and over torque is deforming our gasket, allowing water to get in, and degrade the material and propagate any existing micro cracks. I'm trying to model all these scenarios and each combination.

 

[rb1957]

good grief !

i didn`t realise that when a part, obviously installed incorrectly, fails in service wwe need an FEA to tell us what we did wrong !

but at least that simplifies the modelling ... add fastener loads like they tightened the fasteners. you should be able to show that it's "naff" pretty easily ... much easier to show something down when it already is!

another day in paradise, or is paradise one day closer ?

 

[glass99]

Practically everything with structural analysis of glass requires FEA because if one tiny little piece is overstressed the whole panel breaks. You need to model the stiffness of the supports to get good results, which means modeling the gaskets at least as spring stiffnesses. It can be a fairly simple 1 day type analysis. I do similar things all the time.

Large bolt tensions will be fine, but you will need a hard surface between the bolt and the glass like a fiber gasket so the bolt does not detension. Is the glass laminated? If so, you need to model the interlayer, and you can expect the bolt to detension relatively quickly. Is the glass tempered or chemically strengthened? Do you have good load parameters?

 

[rb1957]

fastening glass plate over a step is common ?

another day in paradise, or is paradise one day closer ?