Thanks LI. I agree. I think I was just looking for the forum to reinforce my own opinion (I was especially hoping for a response from yourself, BI, zdas04, or any of the other great contributors).
Now consider this very interesting real life example (follow link to video - below).
Burst Test in a Controlled Environment:
The N-Vision data logger is great to have in the screen shot since you can clearly track increasing operating pressure until failure.
Test Results:
Failure occurs at ~2,787psi.
Thought experiment (Part A):
Assume:
We found this leak in service at 2,700 psi.
The actual failure (rupture) of the line will take place at some unknown higher pressure (or possibly will experience a “
time dependent failure” if left at 2,700psi).
Let’s be conservative and assume the line is on the brink of failure at 2,700psi.
According to the empirical evidence developed by
Battelle and British Gas, we might decide to lower the discrete operating pressure at the leak location to no more than 80% of what it was when the leak was discovered (i.e. 2,700psi is a pressure that we know did not rupture the line so a safe operating pressure to investigate and repair the leak would be 80% of that).
Therefore, a safe pressure for inspection and repair would be
• 2,700psi x 0.8 = 2,160 psi
From the video, 2,160psi occurs at the ~1:08 minute mark in the video (as can be seen on the N-Vision data logger).
Compare the leak characteristics at ~1:08 to the rupture that occurs at the ~2:30 minute mark (at ~2,787psi).
Visually and audibly, the decision to lower discrete operating pressure at the leak location to no more than 80% of what it was when the leak was discovered appears to be valid.
And according to the empirical evidence developed by Battelle and British Gas, 2,160psi should be below the “
threshold for time dependent behavior” and therefore the leak should not experience a “
time dependent failure” at 2,160psi if left at that pressure indefinitely (neglecting any other component loads and setting erosion effects aside).
So here are the thought experiment questions for you:
Grade and WT for this test segment are unknown but you knew it was operating at 2,700psi without rupturing.
Assume that 2,700psi puts this test segment on the brink of failure (seems conservative).
By lowering the discrete operating pressure at the leak location to no more than 80% of what it was when the leak was discovered (i.e. lower the pressure to 2,160psi) it is predicted that the line is now safe to inspection and repair.
Question 1: Do you believe the test segment is in fact below the threshold for “
time dependent behavior” and therefore the leak should not experience a “
time dependent failure” at 2,160psi if left at that pressure indefinitely?
Question 2: Would you go have your lunch with an umbrella under that leak if you knew nothing about the grade and WT of the pipe? (Keep in mind the person tasked with the repair just might).
Empirical evidence developed by Battelle and British Gas suggests this would be safe (strickly from a material performance point of view, not from a safe work practice point of view). And the video appears to support their recommendation.
Note that the Battelle and British Gas results are based on material performance independent of knowledge of the material properties.
(write down all your answers before continuing…)
Remember, we do not know the nominal or discrete wall thickness at the defect location so we cannot calculate a pressure reduction that leads to a hoop stress equal to 30% SMYS.
Thought experiment (Part B)
The fact is that that burst test was completed using NPS12 x 0.219”mmWT X42 pipe with arc burns placed all over it.
The predicted ultimate bust pressure using minimum tensile strength is calculated as:
• P=2(60,000)(.219)/12.75 = 2,062psi
The pipe failed at ~2,787psi or ~726psi (~35%) higher than predicted ultimate burst pressure (and it had arc burns all over it).
If we had discovered the leak at 2,700psi and instead decided to proceed with a pressure reduction that leads to a hoop stress equal to 30% SMYS based on a hoop stress calculation using nominal WT we get:
P=[2(42,000)(.219)/12.75]*0.3 = 433psi
So for this example, a pressure reduction that leads to a hoop stress equal to 30% SMYS based on a hoop stress calculation using nominal is clearly far more conservative than a pressure reduction to no more than 80% of the operating pressure when the leak was found (i.e. 2,160psi versus 433psi respectively).
But again, from that video, you would never know the grade or wall thickness or the fact that it is covered in arc burns so you would not be able to calculate what pressure results in 30% SMYS…
So here again are the thought experiment questions for you:
Question 1: Do you believe we are in fact below the threshold for “
time dependent behavior” and therefore the leak should not experience a “
time dependent failure” at 2,160psi if left at that pressure indefinitely?
Question 2: Would you go have your lunch with an umbrella under that leak if you knew the grade and WT of the pipe was X42 x 0.219mmWT (i.e. you knew that you were having lunch beside a pipe operating at above predicted ultimate bust pressure)? (Keep in mind the person tasked with the repair just might).
Question 3: How about knowing it is covered with arc burns? They didn’t fail at 2,700psi either so in theory, if we are in fact below the threshold for “
time dependent behavior” at 2,106psi the other arc burns on the pipe should not experience a “
time dependent failure” either (related to ductile crack growth...unless you have brittle hard spots?). Have you changed your lunch plans now? Would you send a worker in to repair this leak at this pressure?
Empirical evidence developed by Battelle and British Gas suggests this would be safe. And the video appears to support their recommendation.
Note that the Battelle and British Gas results are based on material performance independent of knowledge of the material properties.
