Failure Analysis Criteria

[sean001]

I am analyzing an aluminum enclosure subjected to an interior pressure. The FEA package I'm using allows me to view several stress results such as Von Mises, max and min principle, etc...

I want to know whether or not the enclosure will rupture with the presence of an internal pressure. Which stress (Von Mises, principal, ???) do I compare against the yield and ultimate strengths of the aluminum?

 

[rb1957]

there shouldn't be much difference between the two (vM or principal, or tresca) but for material allowable i'd use fcy if the sections were heavy (thick, stable in compression). if you've "optimised" your design for minimum weight your thickness might be small enough to cripple (localised buckling)

 

[GBor]

If you go beyond yield, you will need a non-linear analysis. Assuming you realize this, rb is correct that there shouldn't be much difference so long as a particular stress field dominates (tension along a specific direction). For enclosures, I generally use VM since it is a combined stress theory and the stress fields of most pressurized enclosures are a little complex, however, a metallurgist will generally use Max Principle because that is what his formulas are built around (I think).

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.
Lower Alabama SolidWorks Users Group

 

[sean001]

Thanks, rb1957 and GBor.

Sean

 

[vonlueke]

sean001: von Mises is applicable to ductile materials. Maximum normal (principal) stress theory is more applicable to brittle materials and can be unsafe for ductile materials.

 

[TGS4]

vonlueke - best not tell that to the large number of people who design their pressure vessels to ASME - all of the Sec VIII, Div 1 vessels use the max normal stress theory.

sean001 - Like GBor said, it isn't really going to matter. However, your are going to find locations where the local stress exceeds yield - and that _may_ be OK. I would again recommend that you check out ASME Section VIII, Division 2 - particularly Appendix 4.

Are you doing an actual failure analysis of a failed part, or are you designing a part to avoid failure?

 

[cbrn]

Hi,
as pointed out by TGS4, I'd first of all check if you have to conform to some Norm or not. There are only very few cases where you can design a pressure vessel without refering to some Norm, be it ASME or european PED or others. Then, your design and calculations will have to carefully respect the appliable Norm.
For example, ASME will "oblige" you to consider Trescà-Guest, while PED is built upon VonMises.
If you are free to choose (lucky...), I'd suggest european PED instead of ASME because it's far more recent and has made treasure of the latest improvements in material science. Its "direct route", which needs non-linear solution by default, is computationally "heavy" but can lead to exceptional results if you are designing a new component (less useful if you are verifying a failed component, though...).

Regards

 

[corus]

If you're designing a pressure vessel then you should use a pressure vessel design code such as the british standard or eurocode equivalent, or the ASME code, and follow their rules. It's not as simple as comparing a stress against yield.

corus

 

[sean001]

TGS4,

The enclosure is not a pressure vessel (per se) or boiler. It is an explosion proof electronic enclosure. The design is guided by UL 1203 and ATEX Zone 1 specs. I reveiwed the specs and cannot see anything about failure analysis.

This is a new design. I'm trying to determine if the design will contain an explosion if the vapors inside ignite. I am not investigating a failed component.

The software we use does not have non-linear analysis, though it can be added on. Based on GBor's earlier comment, if the stress stays below yield that should not present a problem.

Thanks everyone.

Sean

 

[TGS4]

Given what you describe, I would agree that keeping all of the stresses below yield (with a suitable design margin, of course) should suffice.

 

[rb1957]

i read the OP as tho' there was external pressure trying to crush the pressure box.

obviously if the box is trying to contain an explosion, then it'll be loaded in tension.

how do you calculate the pressure of an explosion ? i guess chemistry allows to you to calculate the volume of the products, how about the time release of these ??

think too about a "leak before burst" type of design ... maybe it's enough for the pressure box to absorb some of the energy of the explosion rather than all of it.