Level 2.5 Fitness for Service Interpolator

Improve on API 579 Level 1 or 2 pressure capacity for local wall loss

The Fitness for Service Interpolator uses results from thousands of limit load cases on locally thin areas.

Industry cookbook methods give simplified and conservative assessments for local corrosion in pressure piping. But maybe you want to squeeze a bit more out of your system. Your next shutdown might be another two years away while corrosion continues unabated. Or it’s at the stage where you’re having to wind back your operating pressure. Maybe you’ve even got a tough call to make on whether to take a shut down.

pipe with corrosion
A local corrosion defect - prior to removal of loose rust.

To help you safely get a little closer to the mark, we set out to map the limit load failure pressures of a wide range of external corrosion defect size combinations and depths. We analysed various diameter to thickness ratios, compiling a map of the locally thin area ‘failure surface’ over four main variables (Length, Width, D/t, Remaining thickness ratio).

This map can be interpolated via curve fitting or machine learning functions to cover any combination of dimensions within its limits. The map grows in accuracy over time as more results are added and the known points become closer together. Currently we have over 1400 FEA results in our database (we use parabolic interpolation between results). Register below for access to the application.

A fifth variable, the material stress/strain curve, could be considered to arrive at more accurate material-specific failure maps. The current work relies only on allowable stress as the strength parameter. For pipeline codes, this is the yield stress times the applicable hoop stress factor.

Go to login page (register below for access):

Questions and Answers

The most comprehensive approach is to carry out finite element analysis. This typically requires an experienced analyst to conduct the work, who may not always be available at short notice. The analyst needs to then prepare a report, and have it checked. Consequently, the timeframe for getting a response may be longer than ideal.

For urgent situations (or if you’re drowning in defects), rather than waiting the defect can be quickly checked via our FFSI for immediate results. If desired, this can then be followed up with FEA to get the best accuracy possible. We offer a Streamlined FEA service with a simple One-Page FEA Report.

An interesting effect occurs in piping, aboveground pipelines and horizontal vessels – the longitudinal pressure serves to increase resistance to hoop stress. Restrained underground pipelines do not experience longitudinal pressure thrust so they do not benefit in this way. Neither does pipe with untied expansion joints. Tall column vessels or long vertical piping can also have differing longitudinal stresses due to the weight of steel above.

The reason for this benefit can be seen by inspection of the Von Mises failure locus as seen below. The greatest capacity for a material in tension occurs when one of the principal stresses is half the other. It just so happens in piping that longitudinal stress is around half the hoop stress. Therefore according to distortion energy theory, the capacity is increased by up to 15%. This effect becomes less significant as the defect area becomes thinner – the pressure thrust force is preferentially carried by the sound pipe, lessening the benefit to the defect.

Therefore these results should not be directly applied to restrained underground piping / pipelines.

equivalent stress graph

Forces and bending moments tend to have a secondary effect on pressure capacity by altering the Von mises equivalent stress. Thermal expansion also alters the longitudinal stress, and although this is usually thought of as secondary it may be non-conservative to ignore it. This is because deformation of the local thin area does not cause the load to relax significantly.

The FFSI calculator does not currently include the effect of forces and bending moments. If these are high, consider applying additional corrections, safety margin or FEA.

Tests are ongoing in which random defect sizes are checked using the FFSI. These geometries are then built as FEA models and checked by Level 3 limit load assessment. We normally find results to be within +/-5% of the FFSI figure (although results are skewed to the conservative side). Eventually we will give a statistical estimate of accuracy. For defects with less than 20% thickness remaining, the pressure capacity is very sensitive to the remaining thickness and accuracy starts to diminish. Here we find that results can be +/- 20% out.

We suggest you adjust results using a safety factor or allowable Remaining Strength Factor of your choosing. For instance, instead of the usual RSFa=0.9, you might choose to use 0.95. For defects where the margin of safety is not high enough for your comfort, carry out FEA checks or implement remedial measures.

Comparison to burst tests is not strictly required as this has already been done by standards bodies (API / ASME), and the limit load method is a permitted approach to assess local corrosion defects. That said, limit load assessments tend to show reducing capacity as the defect circumferential width increases, which is not so much the case with elastic /plastic analysis or API 579 Level 1/2. Our database shows increased capacity over Level 1/2 for defect widths which are about the same as the axial length, but may show lower capacity for wide defects.

Note: we limit the calculated allowable operating pressure to the API 579 MAWP for the uniform shell thickness with ASME B31.3 as the underlying code.

The main aspect of the FFSI method which is neither prohibited nor endorsed by FFS standards is the interpolation between results. Interpolation is a straight forward technique which is used by numerous engineers every day. However there are a number of interpolations required for this application. We use parabolic curve fitting between known points for greater accuracy. A parabola requires three points, therefore we create two interpolation curves – curve A with two points below the target and one above, and curve B with one point below and two above. We take the lowest of the two interpolated values.

parabolic interpolation chart

Note: this work is not associated with nor endorsed by API or ASME.

The FFSI is still in its infancy, so the more people who use or get exposed to it, the more feedback we receive and the more we can improve it.

Naturally we hope that if you find it useful, you may consider our services when you need FEA or stress analysis (these may be advertised within the web app). We offer the following services –

  • Streamlined FEA – get a fast and economical one-page FEA report for a local thin area defect
  • Full consulting service – experienced stress and FFS engineering as required
If you wish to support this work with a small donation, consider purchasing our Excel add-in, and receive something useful in return.
We reserve the right to charge for this product in future.

Third party licensing can be made available on commercial terms.

The method applies to cylindrical shells made from ductile material, where the wall loss is away from a gross structural discontinuity. In plain english that means carbon steel pipes where the defect is not close to or on welded supports, branches, caps or reducers. Defects on elbows or reducers could be accounted for by use of appropriate corrections and we may include this functionality at a later date.

The study work has been conducted on external defects in piping with internal pressure. Results for internal defects are ‘similar’ but we cannot give conclusive advice at this stage.

The application caters for any diameter cylinder by applying similarity – a 300mm OD shell with thickness 10mm behaves the same as a 600mm OD shell with thickness 20mm. Defect length and width are also normalised based on outer diameter. The database currently contains results for D/t ratios between 10.2 and 42.6. We plan to extend this to higher D/t ratios next.

The defect is modelled as rectangular, constant depth (bent into an arc to conform to pipe radius). The ‘width’ of the defect is its circumferential length at the outside surface. The ‘length’ refers to length along the longitudinal axis. Other forms of defect may be modelled in future (eg. ‘bathtub’ profile, oval shaped).

We may also add the ability to assess defects quantified on a pit depth-grid. Of course you can still do the up-front work to determine the equivalent wall loss yourself using the API 579 Level 2 method. Alternatively, start by using the deepest pit (minimum remaining wall thickness).

If you have multiple defects or a defect at an angle to the longitudinal, we suggest you follow API 579 recommendations for defect sizing prior to using the FFSI calculator.

Register to gain access