OKBNeftegazdiagnostika
Contact us ↗RU

International research

Increasing corrosion between surveys: Why not just subtract two depths

The convergence of signals, the stability of coordinates and the uncertainty of measurement determine the reliability of the corrosion rate assessment.

Technical analysis · OKB editorial team5 min read
An explanatory chart of the OKB. The experimental data are provided in the source.
An explanatory chart of the OKB. The experimental data are provided in the source.

Subject matter

The ROSEN 2016 report compares the approaches to corrosion rate assessment and discusses automated combination of MFL re-examination signals. The authors consider choosing an individual, segmental and conservative growth estimate for subsequent status analysis. The source is important because it links measurement processing to the final engineering solution.

Our review focuses on the logic of comparison. Even two quality inspections do not automatically produce reliable speed: one section, the same physical value and comparable registration conditions must be demonstrated.

One coordinate does not guarantee one defect

The odometer length may differ between runs due to slipping, change of trajectory and calibration. In addition, each report may break the corrosion area into elements by different rules. The line with close distance in the two tables is not yet confirmed by the pair.

The convergence usually requires supporting features, such as welds, and local clarification of signal structure. It is useful to keep a measure of conformity. If one broad area is visible in the first run and the other multiple ulcers are seen, then the geometry of the region should be compared, not the following lines should be forcibly chosen.

The difference in measurements contains two errors

Let the depth of the first survey be estimated with some error, and the second one. — The other one. Their differences include not only changes in metal but also errors in both estimates. With a short time interval, even a moderate depth error can cause large variations in the annual calculation speed.

A negative growth value does not mean the recovery of metal. It can reflect noise, change of size model or incorrect comparison. However, mechanical defrosting of all negative values can also shift statistics. Processing rules should be selected taking into account the error model and the purpose of the evaluation, while keeping the baseline results.

Signal and Size — Different levels of comparison

The comparison of the input signals helps to see a change that is lost after rounding up the dimensions in the report. But MFL amplitude is not only a matter of corrosion: magnetization, speed, gap and sensitivity of channels are essential. The signal must therefore be produced in a comparable manner without erasing the physical dynamics that may exist.

The dimensions calculated by the different versions of the software may also prove to be incomparable. One useful approach — Re-processing historical data with a harmonized methodology. However, the ability to process this depends on the preservation of the source channels and metadata, not just on the availability of a PDF output report.

Why the single speed for the whole line is disputed

Corrosion conditions may vary from site to site due to the environment, route profile, temperature, sediment and state of protection. The average speed of the whole line can hide a local active area. The maximum speed of one questionable measurement, on the contrary, may make the projection unduly conservative.

The break-up into segments should have a physical explanation. It is not enough to choose boundaries so that statistics look more comfortable. It is useful to compare measurements with operational data and verify the stability of the result when changing segment boundaries. The rare deep anomalies deserve separate consideration, especially if their mechanism is different from the main mass of damage.

The projection is not a continuation of the line

Linear extrapolation suggests that future speed is similar to the observed. This assumption may be broken after a change of mode, inhibition, cleaning or environmental composition. The projection should therefore contain scenarios and reasons for revision, not just the date of reaching the reference depth.

For the calculation of load capacity, it is important to consider a change in shape rather than one maximum point. Long growth and the association of adjacent ulcers can affect the result differently from even-handed depression. The choice of the model shall be consistent with the mechanism for damage and accessible detail of the reference measurements.

What should stay in the digital archive

For future comparisons, reference signals, channel identifiers, coordinate reference, instrument configuration information and processing versions are required. The masks of unreliable sites should also be kept. Otherwise, the absence of change can only mean that there is no suitable data in one run.

For each area compared, it is useful to store the reasons for the choice of pair, the conversion of coordinates and the results of a manual check. Such an archive makes re-examination a consistent observation of the object. Without it, each new inspection is actually restarted and expensive historical data are used only as a rough estimate.

OKB Engineering View

In developing the diagnostics software, we believe it is important to design data for reuse. The stored signal shall remain bound by the conditions for its receipt and the calculated size — with a specific version of the algorithm. This is a prerequisite for an audited time analyst.

The review does not set a universal corrosion rate and does not offer a safe life. It shows what measurement and information conditions are needed to make the calculation physically meaningful. The final assessment refers to the pipeline, its history and the adopted status analysis methodology.

Engineering output

Corrosion rate is the result of a comparison and a model of uncertainty, not just arithmetic of two reports.

Sources and further reading

The article was originally written in Russian. It summarises the cited sources. For the original text and illustrations, consult the publications listed below.