Load factors when conducting an elastic-plastic FEA in accordance with EN 1993-1-5 1

[julian89]

Hi all,

I have begun to look into performing FEAs in accordance with EN 1993-1-5, and was a bit puzzled by Annex C of this standard which provides the rules. It allows the engineer to use any material model from linear elastic up to full elastic plastic, but does not mention any load factors associated with these models. My background is largely in the world of pressure vessel, and in that world there are distinct load factors associated with the material models one uses (e.g. ASME VIII Div. 2 etc).

Simply put, from what I can read from Cl. C.9 of EN 1993-1-5, we can use any material model, but the load factors and material factors have to be taken into consideration (here it is referred to EN 1990 Annex D which is for testing and has a strong emphasis on statistical approaches etc.).

JRC has released a document titled "COMMENTARY AND WORKED EXAMPLES TO EN 1993-1-5 'PLATED STRUCTURAL ELEMENTS'" (this is readily available on the web), but also here the guidelines are pretty diffuse.

Has anyone here had experience using elastic-plastic FEAs and EN 1993-1-5? What load factors do we use when performing an elastic-plastic analysis?

 

[desertfox]

Hi Julian

I am no expert in this field but reading the post you have presented it seems to me that the load factors have to be evaluated from actual physical tests however you don’t mention whether or not any information is available in this regard and I presume there isn’t any?
Also the EN specification points you to Annex D which you don't mention.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

 

[julian89]

Hi,@desertfox, you are correct. I'm also no expert as you can gather. I'm struggling a bit since from reading the standards here there are no 'rigid' guidelines.

Annex D as you point is based on testing (see image below) from EN 1990. The basic premise is that the partial factors and load factors should basically be the same as for 'conventional' design methods, but the guidelines in Annex D also have to take into account the statistical uncertainties with regards to the number of tests carried out. Thus, my conclusion from this is that a calculation can be carried out either as elastic or plastic (or any in between), but the same partial factors are employed. This is what seems counterintuitive, as the capacity will often increase dramatically when considering hardening of the material.

 

[desertfox]

Hi Julian

Capacity may well increase but from memory I have never known hardening (work hardening) to be a benefit, in other words if the component increases in its capacity due to hardening then that increase would not alter the load factors in an analysis.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

 

[julian89]

I find this confusing.. Maybe it's because my background is in pressure vessels..

But let's consider a plate in tension only. If the design tension (nominal tension load × load factor) is just slightly greater than the design stress (yield divided by partial factor), then the design is inadequate. But if we do the design with an FEA and include strain hardening (as above), then the design is considered to be OK by my interpretation, since the design load is the same and the collapse load much higher. What am I missing here?

 

[desertfox]

Hi Julian89

From my pressure vessel days and using the codes the yield stress divided by its partial factor gives the allowable stress. I which always perceived that to be an average stress over the vessel and that in certain small area’s of that vessel,the stresses locally maybe higher than the design stress but that was acceptable because it was confined to small area. It’s only when you use FEA that you can see area’s of the vessel that are stressed above the design limit.
I’m not sure that answers your question but I hope it might help

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

 

[Pieter1812]

You can find load factors in EN 1990 (usually permanent loads 1,35 and variable loads 1,50) and material factors in 1993-1-1 (usually 1,0 for steel).

 

[centondollar]

Does the design code you are applying allow the use of computational plasticity in determining the capacity of a pressure vessel? To put it bluntly, I cannot personally see the benefit of saving, say, 10 or 20% in materials by relying on computer modelling if the component in question is routinely subjected to design loading (such as internal pressure or a hydrostatic column) which can cause sudden collapse or explosion and associated shrapnel and possibly fatalities. There is no technical benefit to be had - pressure vessel weight is not a design problem in the real world, unlike e.g., the weight of an airplane, ship or armoured vehicle.

Unless you tune every aspect of the problem (including specifying realistic boundary conditions) and validate by experiments, fancy modelling of pressure vessels seems to me an unnecessary hurdle.

PS. If you want to evaluate a "real" safety factor (and I say "real" in quotation marks because the result is the output of an imperfect mathematical model), you could apply no safety factor on the load, a safety factor based on testing to yield strength and plasticity parameters, and ramp up the load until failure occurs. Make sure to involve all types of non-linearities (material, geometric, boundary conditions) and dynamics (inertia, damping). The results might seem optimistic, and you'd probably not feel comfortable in relying on the computer output to design a real-world pressure vessel, leading you back to the tried-and-tested code-based approaches.