Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
Elevated Temperature material properties for SA516 3
- Thread starter McT178
- Start date
-
1
- #2
- Thread starter
- #4
Temperature is 1,000 deg F. I will look into Section II, part D and see if that will work. I am attempting to prove that the peak stresses of my FEA will relax and are not areas of concern. There is some debate as to where local primary stresses become peak stresses.
-
1
- #5
- Thread starter
- #6
Thanks, I will look into API-579; I think the desired life is known.
One more question on this same subject if you don't mind. I have read in the code that stresses a distance of SQRT(R*t) away from a peak should be considered local primary. If the peak stress are caused by a strap on a vessel, what distance should be used? Would R be half the width of the strap or would R be the radius of the vessel?
One more question on this same subject if you don't mind. I have read in the code that stresses a distance of SQRT(R*t) away from a peak should be considered local primary. If the peak stress are caused by a strap on a vessel, what distance should be used? Would R be half the width of the strap or would R be the radius of the vessel?
What you interpreted with respect to local primary and 1.0*sqrt(r*t) is incorrect. What the Code says is that, for a primary membrane stress to be considered local, the extent to which it exceeds 1.1*S must be less than 1.0*sqrt(r*t) in the meridional direction. It is possible that one could have a surface stress that exceeds the linearized membrane-plus-bending stress, and that stress would be considered "peak". Coincident to that, you could also have the membrane stress be either primary local or primary general. It's sort of an apple and oranges thing...
Note that if you are trying to determine if stresses will relax (creep), then you are doing a non-linear analysis. DO NOT linearize any results from a non-linear analysis...
Note that if you are trying to determine if stresses will relax (creep), then you are doing a non-linear analysis. DO NOT linearize any results from a non-linear analysis...
- Thread starter
- #8
I have gone back and read the code, but I still am not confident that I understand this. Is it true that a stress greater than 1.1S that is > sqrt(r*t) from a discontinuity should be considered a Pm? Maybe another way to say that is a stress that exceeds 1.1S and is greater than sqrt(r*t) from a discontinuity would exceed the allowable limit.
It is possible that one could have a surface stress that exceeds the linearized membrane-plus-bending stress, and that stress would be considered "peak"
Are the surface stresses (peaks) governed by 1.5S or 3S?
Is there a distance similar to sqrt(r*t) that defines a hotspot stress? For instance if the stress extends a certain distance it should be considered a local primary, or is this engineering judgement?
I really appreciate everyone's help on this. I believe I am getting close to understanding Part 5, but there are a few issues like this one I need to work out.
It is possible that one could have a surface stress that exceeds the linearized membrane-plus-bending stress, and that stress would be considered "peak"
Are the surface stresses (peaks) governed by 1.5S or 3S?
Is there a distance similar to sqrt(r*t) that defines a hotspot stress? For instance if the stress extends a certain distance it should be considered a local primary, or is this engineering judgement?
I really appreciate everyone's help on this. I believe I am getting close to understanding Part 5, but there are a few issues like this one I need to work out.
In short - maybe, but remember, we are ONLY talking about membrane stress, in this context.McT178 said:
No. A membrane stress may be classified as local if the extent to which it exceeds 1.1S in the meridional direction is greater than 1.0*sqrt(r*t).McT178 said:
No - they are governed by the fatigue rules.McT178 said:
No. Your membrane stresses will either be classified as general primary membrane (Pm), local primary membrane (Pl), or secondary. The through-thickness bending stresses will either be primary bending (Pb) if it is in a flat head, or secondary. Anything that exceeds the linear through-thickness stress distribution is peak.McT178 said:
Does this help?
- Thread starter
- #10
I probably should have mentioned earlier that the report I am looking at was created with plate elements. Is there a determined distance away from a hot-spot where primary local stresses are evaluated and then another distance where general primary stresses are evaluated?
Even in plate/shell elements you still have membrane and membrane+bending stresses. Evaluate each separately.
As far as how far away from discontinuities you should evaluate the stresses, I would suggest that you take the same approach that you would if you had 3D-solid elements: at the edge of the juncture ring. For a typical nozzle-type arrangement, that is usually 1/2t (to account for the thickness not being explicitly accounted for in the shell/plate elements) + fillet weld leg.
'
The good thing about plate/shell elements is that you can directly plot membrane stresses. Do that, with a contour of 1.1S. Then, you can directly measure the meridional extent to which the membrane stress exceeds 1.1S. If that extent is less than 1.0*sqrt(r*t), then the membrane stresses inside that region can be considered local and subject to the 1.5S limit. If the extent is larger that 1.1S, then the membrane stresses are not local and will fail the 1.0*S check.
As far as how far away from discontinuities you should evaluate the stresses, I would suggest that you take the same approach that you would if you had 3D-solid elements: at the edge of the juncture ring. For a typical nozzle-type arrangement, that is usually 1/2t (to account for the thickness not being explicitly accounted for in the shell/plate elements) + fillet weld leg.
'
The good thing about plate/shell elements is that you can directly plot membrane stresses. Do that, with a contour of 1.1S. Then, you can directly measure the meridional extent to which the membrane stress exceeds 1.1S. If that extent is less than 1.0*sqrt(r*t), then the membrane stresses inside that region can be considered local and subject to the 1.5S limit. If the extent is larger that 1.1S, then the membrane stresses are not local and will fail the 1.0*S check.
- Thread starter
- #12
My confusion is where a peak (3Sa) becomes a local (1.5Sa). For example, if I have a peak at a discontinuity, at what distance from that discontinuity would I stop evaluating the stress for 3Sa and start to evaluate at 1.5Sa. It seems like it might be a common area of confusion because I found another thread ( where similar questions as my were asked, but it did not seem to result in a definite answer. We commonly receive reports where a point a few inches from a peak are named PL and evaluated as so (1.5Sa). Then the area is evaluated with bending + membrane, and the peaks are evaluated as primary + secondary (3Sa).
Apples and oranges comparison.
The 1.5S check is for MEMBRANE (i.e. mid-surface) stresses. The 3S check is for MEMBRANE+BENDING (i.e. surface) stresses. Peak stresses (the ones that are handled in Article 5.5) are those surface that are beyond the M+B stresses.
The M+B stresses should be evaluated at the same locations that I indicated in my previous post.
And don't get to confused about primary vs. secondary. First you have to figure out whether it is membrane or M+B. That seems to be the problem to start...
The 1.5S check is for MEMBRANE (i.e. mid-surface) stresses. The 3S check is for MEMBRANE+BENDING (i.e. surface) stresses. Peak stresses (the ones that are handled in Article 5.5) are those surface that are beyond the M+B stresses.
The M+B stresses should be evaluated at the same locations that I indicated in my previous post.
And don't get to confused about primary vs. secondary. First you have to figure out whether it is membrane or M+B. That seems to be the problem to start...
- Thread starter
- #14
-
1
- #15
Yes. That is exactly what you need to do.McT178 said:
You won't find "peak" membrane stresses. Don't bother looking.McT178 said:
In fact, you won't find "peak" stresses (generally) in a shell-plate element analysis. The formulation of the shell-plate element is such that it ASSUMES a linear through-thickness gradient in stress - the exact definition of M+B. If you want actual "peak" stresses that you will evaluate in a fatigue analysis, you will either have to factor your M+B stresses by an SIF, or perform a 3D-Solid FEA.
- Thread starter
- #16
Ok, think I got this. A highly localized stress at a discontinuity (corner of strap welded to vessel for instance) should be evaluated as a primary local membrane (1.5S) when the FEA is set to membrane only, and then if set to membrane + bending, evaluated as a peak (3S). If that statement is correct then I believe I understand this.
- Thread starter
- #19
- Status
Similar threads
- Locked
- Question
- Locked
- Question
- Locked
- Question