Bending of beam under pure tension 1

[Nougatti]

Hi there!

Given a beam/profile like this

With a force applied normally to one end and given no degrees of freedom at the other, the beam is experiencing pure tensile stress.
It should be, anyway.

In reality, what happens is that the beam bends drastically outwards and experiences bending.

Can anyone share some insight into why this happens and how I can avoid it?

I have a gate where the back plate (not exposed to water pressure) is holding the front arced plate (exposed to water pressure) together.
I wanted to stiffen this large plate up with a few bars to avoid it being so flimsy, but it makes the whole back plate bend outwards.

With best regards
Daniel

---------------------------
I am Norwegian.
I design mechanicals for hydroelectric powerplants.
I use NX 8.0.3.4 and ANSYS 15
----------------------------
Tom you can't knit at these speeds! Nobody can! DON'T BE A GODDAMNED HERO TOM!

 

[Agent666]

Your tension load isnt applied through the centroid of the section, so an internal moment is generated within the section when the additional plate is attached to the side due to the eccentricity of the load with respect to the centroid of the section.


In your full model it could also be some moment is generated around the corner due to the continuity.


To avoid this either use a symetric beam/stiffner arrangement so you are loading it through the centroid, or load it through the centroid.

You cannot avoid it deflecting given the full arrangement, how much deflection is acceptable is really a matter of judgement. If the strength is ok, and it still performs the job (in that it wont leak if its retaining water) then what is the problem?

We dont know the scale of the item, but the 18mm deflection could be acceptable if its a few meters wide?

 

[desertfox]

Hi

Looking at your full model the values given are negative stress i.e.
compression, under that circumstance isn't the back plate under compression and starting to buckle?

 

[Nougatti]

desertfox: the full model at the bottom shows deflection.

Agent666: I've been thinking about putting a range of smaller bars on the opposite side of the stiffeners, to even it out, but I think the deflection shown here might be acceptable.


-Daniel

---------------------------
I am Norwegian.
I design mechanicals for hydroelectric powerplants.
I use NX 8.0.3.4 and ANSYS 15
----------------------------
Tom you can't knit at these speeds! Nobody can! DON'T BE A GODDAMNED HERO TOM!

 

[desertfox]

Oh sorry I thought they were stress.

 

[Agent666]

If you make it symmetrical by adding a similar plate on the other side do you see the same effect in the top model?

If you do then maybe its a modelling issue.

Instead of stiffening it with a flat plate you could use and angle, or channel section which would be stiffer.

 

[XR250]

Does you model take into account second order effects? As soon as it begins to deflect, the p-delta, will tend to bring it back inline

 

[Nougatti]

ExcelEngineering: Interesting. I have not taken into account any second order effects. Can you explain this further?

---------------------------
I am Norwegian.
I design mechanicals for hydroelectric powerplants.
I use NX 8.0.3.4 and ANSYS 15
----------------------------
Tom you can't knit at these speeds! Nobody can! DON'T BE A GODDAMNED HERO TOM!

 

[paddingtongreen]

If you could size the stiffener to bring the neutral axis to the surface of the plate, and also connect the load to that face, the problem would almost go away. "Almost" because the stiffener isn't full length; the plate would bend very slightly in the opposite direction.

Second order. Look at your diagram showing deformed shape, the load now has moved to the right of center of the fixed end, so it will cause a partial equalizing moment.

Michael.
"Science adjusts its views based on what's observed. Faith is the denial of observation so that belief can be preserved." ~ Tim Minchin

 

[rb1957]

why have a back plate ? if you're trying to stablise the front plate, won't discrete elements work better ?

the thin plate is bending due to inplane load; plate's don't like this. a question to ask might be "am i deflecting the plate too much ?" if your FEA deflection is bigger than the thickness of the plate you might need to consider membrane forces (simple plate bending assumes no axial load on the centroid, ie no membrane forces).

Quando Omni Flunkus Moritati

 

[Nougatti]

I've got a back plate because the alternative is tension bars (ideally with joints at each end), and I'd need to have a whole lot of them, or use extremely big ones to do the job.
The plate gives me a nice big area for the stress to be absorbed into (around 50% utilization) as well as the fact that it brings stiffness to the structure as a whole.
Oh, and the width of this whole thing is over 8 metres. The bottom image shows the lower half of the whole gate.

---------------------------
I am Norwegian.
I design mechanicals for hydroelectric powerplants.
I use NX 8.0.3.4 and ANSYS 15
----------------------------
Tom you can't knit at these speeds! Nobody can! DON'T BE A GODDAMNED HERO TOM!

 

[drawoh]

Nougatti,

Check the magnification factor in your FEA software. FEA grossly exaggerates strain in its displays. Often, it helps to turn this feature off.

--
JHG

 

[rb1957]

max deflection of the plate (1st pic) is 0.36 (m?, mm??)

i'd add stiffeners to the opposite side to balance the plate.

if weight is a concern, cut lightening holes in the plate (away from the stiffeners it's not that effective).

how is the plate attached to the curved wall ? is the curved wall putting moment into the plate ?

Quando Omni Flunkus Moritati

 

[JStephen]

When you actually build it, you'll have some weld distortion that may be more than the deflection you're discussing.

 

[CELinOttawa]

When you actually build it, you'll have some weld distortion that will be more than the deflection you're discussing.

Fixed it for you... Oh, and that weld distortion is likely to be in the opposite direction, and thus act as a pre-camber.

 

[CELinOttawa]

P.S. A star for you JStephen. Damn good point.

 

[Nougatti]

The first two pics of tha beam was just to explain the principle. I just applied a random load to randomly sized asymmetrical beam to show the problem. It isn't affiliated with the gate itself (third pic).

The deflection on back plate of the gate is about 15 mm. The image shows a magnification factor to the deflection of about 30.
The back plate is some 10-12 mm thick (don't remember at the moment, since I'm not at work).
The plate is connected to the curved wall as well as the curved wall's stiffeners; the latter which may lead to a small amount of moment being transferred.
It shouldn't be much compared to the tension forces though, but then again, they both induce bending in the same direction.

---------------------------
I am Norwegian.
I design mechanicals for hydroelectric powerplants.
I use NX 8.0.3.4 and ANSYS 15
----------------------------
Tom you can't knit at these speeds! Nobody can! DON'T BE A GODDAMNED HERO TOM!

 

[rb1957]

with deflections from plate theory something like the thickness, is typically considered beyond the limit of the theory.

support the plate symmetrically.

Quando Omni Flunkus Moritati

 

[Nougatti]

rb1957:

Will do.

---------------------------
I am Norwegian.
I design mechanicals for hydroelectric powerplants.
I use NX 8.0.3.4 and ANSYS 15
----------------------------
Tom you can't knit at these speeds! Nobody can! DON'T BE A GODDAMNED HERO TOM!

 

[Agent666]

If I am understanding this correctly only 15mm deflection over 8m with what would be a substantial water pressure seems quite small now we know the scale of the structure.

How much water head? Remembering back to fluids at university I can't remember if there are additional loads due to the fact that water is moving past the gate. I assume the water side is a more substantial structure than the same thin plate and stiffeners so it doesn't buckle if in compression. If it's 8m across and this is the bottom half the thing must be 20m tall, water pressures would be enormous and I'm having trouble believing relying on just the back plate works for strength?


Given the scale I would be providing far more substantial beams or something than a flat plate + stiffener.