Cycling pressure to vacuum

[FoxRox]

I have a new project for a vessel that cycles between 150 psi and full vacuum every 15 minutes during normal operation. This is new to me, and it's making alarms go off in my head about fatigue.

If using ASME BPVC Sec VIII Div 1, shall I just design it for the pressure and/or the vacuum (worst case)? Or are there additional calculations to be done for potential fatigue?

Of course, I'm doing my own digging and research on this, but I'd like any insights the fine members of this forum might have to offer. Thanks!

 

[SnTMan]

FoxRox, 1) you need to design for both pressure cases, and 2) if you are anticipating more that perhaps a few hundred cycles, yes definitely investigate fatigue. Another poster will be along with more on 2) shortly

Regards,

Mike

The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[TGS4]

I'm the other poster that SnTMan was expecting.

The cyclic condition is covered in UG-22 and must be included in the design, although there currently aren't any rules in VIII-1 to direct you.

As a first pass, I would recommend performing one of the fatigue screening assessments in VIII-2, Part 5, Section 5.5. If you pass Method A, then you're definitely ok. If your cycle time is 15 minutes continuously for 25 years, that's a total of ~875,000 cycles. That will certainly fail the Method A assessment. The Method B may pass, but only if the vessel is "over-designed". You will need to pay special attention to your design details, especially eschewing partial-penetration welds (fillet welds, etc).

I recommend getting expert-level help.

 

[FoxRox]

Thank you for the responses. That is exactly where I thought this would go, just didn't know the route.

 

[SnTMan]

TGS4, yes, you are

The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[FoxRox]

I found out a bit more about this vessel today.

The good: The vessel is to be rated for 150 psi pressure and 15 psi vacuum, but the operating conditions only cycle between 70 psi and -12 psi.

The bad: The previously given cycle time of 15 minutes is not correct. I was told the cycle could be as little 90 seconds. When asked, the customer said they did not think any fatigue analysis had been done on previous builds of this vessel, but they would have to check to know for sure. They've had some in service for at least 10 years.

My next question is this: Does endurance limit come into play? Based on some quick research, the endurance limit of steel is apparently higher than the ASME Sec II Part D allowable stress of SA516-70. That seems strange to me, but once again... new territory. If this is true, it seems cyclical loading is a non-issue for the shells and heads. Are the welds the main concern? Is that why it's important to go with full penetration?

 

[TGS4]

The endurance limit of typical pressure vessel steels is MUCH lower than the allowable stress. In your example, in 10 years of service the vessel would already have seen 3.5e6 cycles. Granted, the pressure range would be 82 psi compared to the design pressure of 150 psi, which will provide some relief.

However, 55% of the allowable stress (VIII-1) at ambient for a 516-70 material is still a stress range of 11 ksi, which translates to a stress amplitude of 5.5 ksi. So, for the plain shell (not including weld peaking effects, or weld effects for that matter), the stresses are below the endurance limit. However, weld peaking may add as much as 50% to the stress amplitude. And lack of inspection of welds (or just the higher stresses in the head knuckle) may add an additional factor as high as 3. That would take your stress amplitude up to about 24.75 ksi, which has a "life" of only ~3e4 cycles. Even factoring in a factor of 20 on mean life (the design margin built into the ASME curves), you could expect an average life of 6e6 cycles. Essentially, that vessel is on borrowed time.

You definitely need an expert to evaluate the fatigue life of this vessel. This is going to involve FEA. Depending on the prior inspections, you will definitely need a measure of longitudinal weld seam peaking. If it were me, I would likely prefer a fracture mechanics approach which would assume the presence of cracks in the welds.