stastitical analysis of calculation of corrosion growth rates

[rudiawan .]

We would like to do Fitness for Services for the pipeline, but the client want to have the remaining life of the pipeline with method " stastitical analysis of calculation of corrosion growth rates ". is there any reference regarding this method ?

 

[fel3]

This is not my area of expertise, but I was curious so I typed "statistical analysis of calculation of corrosion growth rates" (i.e., with quotes and correcting the typo in "statistical") into the search bar in Microsoft Edge and got lots of hits. I cannot say if any of these will help you, but it appears that there are many references available.

 

[Galvatech]

The calculation of corrosion rates involves numerous variables, making precise predictions challenging. For example:

• Location of Corrosion: Is it occurring on the interior or exterior of the pipeline? Each environment introduces vastly different factors.
• Protection Measures: Are sacrificial anodes or cathodic protection systems in place? These can significantly influence corrosion rates.
• Environmental Conditions: Temperature, soil pH, moisture levels, and other environmental factors play a critical role. For instance, conditions can vary dramatically even within a short 10-foot span, leading to different rates of material degradation.

While I do have data for exterior exposure of unprotected steel and galvanized materials, what your client is asking for—an exact prediction under varied conditions—doesn’t exist in standard references.

The only feasible estimation approach is to evaluate the corrosion level at a known initial time (time 0) and compare it to the current state. Based on this data, and assuming environmental conditions remain constant, you could extrapolate that corrosion might double over double the time. However, this linear assumption doesn’t account for changes in variables like moisture, temperature, or physical damage, which could accelerate or decelerate the process.

For more accurate insights, an on-site assessment and periodic monitoring are recommended to refine predictions based on real-world conditions.

 

[EdStainless]

You end up with statistical distributions for rate of corrosion, changes in conditions, likelihood of defects, and so on.
Then all of these are combined.
I have seen this done using gaming theory.
You end up with very wide ranges of estimates, but at some confidence level you can establish an inspection interval.
As time goes on you use inspection data to refine the model.
If the client can hand you good inspection data for the last few decades, then you could get close.