Fluid Handling

Use the FRETT approach to prevent pipeline failures

Use two-part failure analyses to determine the cause of pipeline failures.

By Heinz P. Bloch P.E., Process Machinery Consulting

In brief:

  • This article will convey the essence of two-part failure analyses that have been used for a number of decades with great success.
  • These four basic agents or component failure mode sets are always force, reactive environment, time, and temperature (FRETT).
  • The first of the two failure-analysis steps is called the “Seven Root Cause Category Examination.”

In 2006, Plant Services published my article on using the FRETT approach to make problem pumps a thing of the past. With questions over pipeline failures, it may be time to revisit the topic. This article will convey the essence of two-part failure analyses that have been used for a number of decades with great success. Originally devised for machinery, the two-part analyses apply equally to pipeline issues. This will be confirmed when the dust settles or, to be more exact, when the flow slows to a trickle.

Failure analysis step no. 1: The “Seven Root Cause Category” approach

The first of the two failure-analysis steps is called the “Seven Root Cause Category Examination.” It accepts the premise that all equipment or hardware failures fall into one or more of only seven possible cause categories.

  1. design errors
  2. material defects
  3. fabrication and processing errors
  4. assembly and installation deficiencies
  5. maintenance-related or procedural errors
  6. unintended operating conditions
  7. operator error.

Using logical thought processes, we might ask ourselves which of these seven cause categories are influenced by the pipeline operator and which ones are under the full jurisdiction of the pipe manufacturer or pipe installers. The answer determines the cause categories where failure analysis efforts should be concentrated.

Suppose there was a case of a failed pipeline. It had been properly leak-tested when it was first commissioned years ago. At that time, all welds had been examined and their integrity verified by suitable non-destructive means.

If that pipeline had been designed by responsible engineers and was in service for a number of decades, we could rule out (1) design error. Next, assume it had been ascertained that no operator error occurred. Accordingly, category (7) is now being ruled out. Assume further that failure occurred in a buried portion of pipe — a segment unaffected by (5) pipeline maintenance. It would be reasonable to say there is thus no logical causal event falling into either the assembly and installation or maintenance and procedural error categories. Categories (4) and (5) can now be ruled out. One might focus next on items (2) material defects and (3) fabrication and processing errors. So, we might investigate these two.

Figure 1. Suppose one found no metallurgical evidence of flaws in the base material selected by the pipe manufacturer and the pipe had burst near a straight length buried in the same type of soil as all the undamaged sections of buried pipe. Blaming the calamity on a very old unseen pre-existing metallurgical defect would be like attributing the death of a 90-year-old to mistakes made by nurses present at the man’s birth.
Figure 1. Suppose one found no metallurgical evidence of flaws in the base material selected by the pipe manufacturer and the pipe had burst near a straight length buried in the same type of soil as all the undamaged sections of buried pipe. Blaming the calamity on a very old unseen pre-existing metallurgical defect would be like attributing the death of a 90-year-old to mistakes made by nurses present at the man’s birth.

But suppose one found no metallurgical evidence of flaws in the base material selected by the pipe manufacturer and the pipe had burst near a straight length buried in the same type of soil as all the undamaged sections of buried pipe (Figure 1). Blaming the calamity on a very old unseen pre-existing metallurgical defect would be like attributing the death of a 90-year-old to mistakes made by nurses present at the man’s birth.

Anyway, the reviewer would now be left with the cause category (6), operation under conditions not envisioned in the design. However, the analysis is still incomplete at this point in time. It will have to shift to Step 2, often abbreviated as the FRETT phase of an investigation.

{pb}

Failure analysis step no. 2: FRETT

It is universally recognized that all hardware component failures or hardware distress on the mechanical component level, are attributable to one or two of only four failure mode sets. These four basic agents or component failure mode sets are always force, reactive environment, time, and temperature. An easily remembered acronym, FRETT allows analysts to recall these four possible basic agents.

One would immediately rule out temperature if it had been possible to ascertain the pipe wall temperatures had always been well within the acceptable range. Whenever one might suspect that corrosion had something to do with pipe rupture, the RE component of FRETT would be involved. Depending on the severity of corrosion, closely measuring wall thickness, depth of pits, fissures, and other seemingly insignificant stress risers might be of great importance. In laymen’s terms, the force (F) acting on an effective cross-section of decade-old pipe wall might then reach magnitudes well beyond what the original designers had anticipated.

Heinz P. Bloch, P.E., is owner of Process Machinery Consulting (www.heinzbloch.com) in Westminster, Colorado, and the author of numerous articles and books, including “Improving Machinery Reliability” and “Pump Wisdom.” Contact him at heinzpbloch@gmail.com.

That would make the F in FRETT a second confirmable agent, meaning an existing failure mode set. A failure situation would be amplified because we can reason that two of the four failure mode sets (in this case F and RE) had combined. Both failure mode sets fit inside the description “cause categories of interest” which we earlier labeled (6), operating conditions not intended by the original designers. In this particular example problem, dealing with a pipeline, the analysts would view force as another logical failure mode set (in addition to RE) and would now advocate implementing Step 3, validation.

Step no. 3: Validation, or relating analytical models to field experience

Figure 2. Facts are difficult to obtain when pipelines fail. If the different parties concentrated on using adequate factors of safety and on retiring pipelines within a specified time, we would all profit from the experience.
Figure 2. Facts are difficult to obtain when pipelines fail. If the different parties concentrated on using adequate factors of safety and on retiring pipelines within a specified time, we would all profit from the experience. (Source: Photo courtesy of Dr. Anthony Sofronas, http://mechanicalengineeringhelp.com.)

Whenever a hardware component design is modeled for computer analysis, the analysts will generally make a number of assumptions and the validity of these assumptions must be ascertained. Common sense tells us that there are four possibilities and corresponding definite conclusions:

  1. Analysis says “OK,” but pipe fails:  Model or analysis technique flawed
  2. Analysis says “OK,” and pipe survives:  Model and analysis technique OK
  3. Analysis says “not OK,” pipe also fails:  Model and analysis technique OK
  4. Analysis says “not OK,” but pipe survives:  Model or analysis technique flawed.

As is so often the case, all the facts are difficult to obtain when pipelines fail. But if the different parties concentrated on using adequate factors of safety and on retiring pipelines within a specified time, we would all profit from the experience (Figure 2). We might liken it to the replacement of an automobile tire with very little tread left. Once the tire is bald, it might be serviceable for a horse-drawn vehicle, but not a passenger car that’s expected to keep up with conditions and criteria of modern traffic.

Reliability professionals should apply parallel reasoning and not stand idly by while the uninformed take inappropriate risks.


Reference

Bloch, Heinz P., and Fred K. Geitner, (4th Edition, 2012) “Machinery Failure Analysis and Troubleshooting,” Elsevier Publishing, Oxford, United Kingdom, and Waltham, Massachusetts 02451, ISBN 978-0-12-386045-3.