Structural Glass Canopy

[Contraflexure74]

Hi,

I need to run some numbers on a glass canopy design. Its supported on 4 corners by spiders but I have never done calcs like this before.

Has anyone got the critical formulas I need to run the numbers. I assume deflection and bending stress calculations are the most critical?

Any information would be appreciated.

Thanks.

 

[KootK]

It's design that you probably want to delegate to a specialty engineer working on behalf of the glass supplier or spider clip supplier. Deflection isn't terribly important for glass and the design will likely be governed by local bending stresses occurring near the spider clip connections. The plug and chug glass codes like ASTM 1300 usually only deal with edge supported glass and won't be terribly helpful other than to, perhaps, determine allowable stresses. If you were a glass specialty design engineer, you'd probably use MEPLA FEM software to tackle the design. Any FEM software package can be used similarly if a bit less elegantly for non-insulated glass panels like you'll be using here. Europe and Australia both have much better glass codes than what we have in Norrh America. I don't know them well but there must be provisions of value there.

Here's a starter doc: Link. If you're just confirming a glass thickness and panel size availability for your architect, I'd just run it by a local glazing contractor.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[Contraflexure74]

Thanks KootK. We have some FE software so I might try modelling it on that.

Any idea what allowable deflection I should allow for the supporting steel frame, (it's a cantilever).....?

 

[KootK]

WT cantilever or something like that? With the spider clips, the WT deflection translates into rigid body deformation of the glass which doesn't really affect the glass. You see pretty aggressive numbers like L/75 etc. You do want to ensure that drainage is not compromised but you usually get that from back slope.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[calvinandhobbes10]

I believe KootK is pointing out that you need to look at where the spider clips are located on the steel outrigger member. If they are at, say, the post connection and the free end, outrigger curvature doesn't really affect the glass plane, the spider clips just take a bit of rotation and the glass remains plane (except for its own C&C pressure). I would however, still impose the deflection limits on the steel frame as a measure of sizing the steel sufficiently to resist excessive movement. Plus, if you have multiple rows of glass panels along the cantilever, the sealed joint between the rows needs to take the rotation---deflection limits are written to help seal to hold.

For the steel outrigger (assuming that's what you have), I would apply vertical deflection per the IBC category "roof members supporting ceiling"---L/240 for wind/uplift (add in the 0.42 10-year conversion factor), L/180 for dead plus live. This is my interpretation of how the limit is intended; don't think it's spelled out in the Code.

 

[Contraflexure74]

Thanks guys. Very helpful.

 

[KootK]

You're most welcome. Calvin&Hobbes has translated my ramblings accurately. Because glass if brittle and scary, engineers often assume that the supporting structure needs to be designed to really tight deflection limits. That's rarely the case.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[NorthCivil]

KootK

After debating on it for a long while, my last workplace purchased Mepla, it was around $7,000. When it came to panels with holes supported by standoffs and fixtures, we found the models it turned out were overly conservative most of the time compared to what we had been doing over the history of the design consultancy. If we designed by it, it would have led to way overly conservative designs that no glazing contractor would have ever built.

The $7000 mepla software became shelf-ware.

From my experience, standard FEA packages (S-frame, solidworks) were adequate, and in many ways even more versatile than Mepla.

 

[KootK]

Thanks for your comment NorthCivil. It begs the question however: how do you know that the less conservative answers were the right ones?

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[firstoption]

I have also modeled glass structures in an FEA package.
definitely takes the stress out of calculating it by hand. (well, technically mathcad. still super time consuming)

 

[Contraflexure74]

I'm currently trying to model my panel on strap. Not too sure how to model the 3 different layers specified.

I have 10mm toughened glass on 1.5mm interlayer on 6mm toughened glass. Also what would permissible bending and shear stresses be for toughened glass? And permissible deflection limits?

Any help welcome.

 

[CANPRO]

I am not familiar with structural glass design, but it seems your questions would be best answered by the glass supplier (allowable stresses and deflection limits).

 

[Klitor]

When I did some glass design (years ago, and not much desing) there was a code table you could use to find effective thickness for your 10+6 composite... lets say it would come up to 12.1mm that you would than use for further stress analysys... your code/manual may have something similar

 

[RFreund]

When dealing with multiple laminated layers (in the US) we would refer to ASTM E1300, which provides an effective thickness for stress and one for deflection.
I would send some caution your way. There is (IMO) a lot of research that needs to go into understanding glass before heading into design. Not to say you can't do it or you may have a simple project to start with, but I've read numerous books and tech reports to help me better understanding things as there is no real "code" for glass.
Also you usually want an FEA package that can handle non-linearity as the deflections are usually larger than the thickness of the glass.

EIT

http://www.HowToEngineer.com

 

[KootK]

Also what would permissible bending and shear stresses be for toughened glass?

Transverse shear is generally a non-issue. Longitudinal shear stress between layers comes into play as RFreuend mentioned. Basically, there are two fundamental cases where your behavior will be less than 100% composite due to inter layer effects:

1) High temperatures.
2) Long duration loads (snow on skylights etc).


I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[NorthCivil]

KootK

regarding whether the less conservative models were right.

We would regularily run 3 different models in 3 different programs, and receive 3 different results. All 3 programs had their own limitations on creating a "true" form of the glass we were modelling, and all had limitations in accurately modelling the connection restraints.

using solidworks, we could draft a perfect version of the glass panel, there would be a variable density mesh near the connections (the main point of interest). however, connections were limited to fully fixed (possibly due to our limited version of the program).

using S-frame, we were free to select the type of FEA, define our own mesh density, define our connections as preferred. the panel was simplified. input glass properties and model in non linear fashion. Again, results must be interpreted.

The differences in inputs and restraints, and their outcomes for stress and deflection always seemed to make sense working between Sframe and Solidworks.

with mepla, stressed were kind of all over the place, depending on the sort of problem. but typically high. Deflections typically seemed unreasonably high. The biggest thing, MEPLA was a black box. I have never worked with FEA that simple to use. It was simple because the user was limited - you couldn't dive too deep into what sorts of functions the program was carrying out. Our conclusion - if we cant see what its doing, and its spitting out funny numbers, what reason do we have to trust it?

My guess? I think there must have been some safety factors built into it to exclude the developers from liability... call me a conspiracy theorist or paranoid, but things just didn't add up using it.

This was just my experience working for a specialty structural glass consultancy.

 

[KootK]

Thanks for sharing your experiences NorthCivil. I have access to both S-Frame and MEPLA so I may see if I can corroborate some of what you've described. Overall, how did you like working as a specialty glass consultant? From the outside, it appears to be a mix of extremes. 20% SUPER sexy / 80% pretty mundane.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[NorthCivil]

KootK

Glass was my first job, I left it to try out "traditional" structural, ie, wood, concrete, steel, masonry.

When you boil it down they are both pretty similar. Ive found sexy-ness and engineering appeal rarely line up . 50 story facades design packages were a laborous bore to grind through, as can be glass clad "architectural masterpieces". Some of my most memorable engineering was designing connection details for typical windows into questionable 80yr old hollow masonry structure.