BOUNDARY

Q: How is the nub-valley contact accounted for? I can’t see the contact pressures in XPOST when opening the xx_bflex2010.raf file. Shall it be activated in the VISRES or similar card?

A: You will see the contact pressures directly by looking at the Normal stress (sigma-zz) result type for the boundary group

Q: In BFLEX, which friction coefficient is used for the nub-valley contact in BOUNDARY?

A: The friction coefficient will follow the value provided in the FLEXCROSS input for the corresponding layer.

Q: Nub-valley friction induced flexlok section transverse stresses (boundary), can they be visualised?

A: Look at Transverse stress (sigma-yy) and Shear stress (sigma-yz) result type for the boundary group

Q: INOD, would there be results differences depending on which INOD is selected?

A: The user can select the position of the boundary model. This is to be able to choose a critical position along for example a bend-stiffener. So yes, changing INOD may mean different curvature histories and hence difference in the boundary and pflex stresses.

Q: The Axial stress results in the -boundary model looks strange? Axial stress should be uniform across the cross section, and it is for a single zeta profile, but after running BOUNDARY the six windings show quite different axial stresses.

A: There are two things to check.

First, the Axial stress, Normal curvature stress and Transverse curvature stress results are populated by PFLEX, even though the results can be seen in the -boundary model part in Xpost. You should hence perform the PFLEX analysis before inspecting these results. Note that for the Transverse curvature stress results, the -boundary model results only contain the dynamic, bending induced part of the stresses from ovalization.

Second, a variation between the six windings is expected, and is due to end effects arising from the BOUNDARY model being truncated. The model contains only six spirals, and the outermost are supported by a neighbouring profile on one side only. That asymmetry lets the profile rotate slightly about its own axis: one side moves to a slightly larger radius relative to the pipe centre and picks up higher tensile stress, while the opposite side moves to a slightly smaller radius and lower stress. In the real pipe every winding has neighbours on both sides.

Use the results from one of the centre spirals, where the contact conditions are symmetric and therefore representative. The PFLEX model is not affected by the end effects in the same way.

Q: I set the friction and bending steps for BOUNDARY according to my Bflex2010 load history, but BOUNDARY does not seem to use them?

A: BOUNDARY can only operate at the steps that were stored to the result database, so check the DTVI parameter in the TIMECO card. For example:

Here the analysis step is DT = 0.1, but DTVI = 1.0 stores results only every 1.0 second, so BOUNDARY is executed only at steps 1, 10, 20, 30, 40 and so on. Setting DTVI = 0.1 stores every step and makes them all available to BOUNDARY.

Note also that the first result step is not time 0 but the first stored step, which may already carry some internal pressure. That is usually why stresses appear at what looks like the initial step.

On the choice of steps: friction onset should be slightly after the onset of bending, and the start of bending in BOUNDARY should be at a step where there is already a small bending in the analysis model. This helps convergence.

Q: During tension and pressure loading my boundary model of the zeta profile moves a lot in the transverse direction?

A: Check your transverse curvature parameter trcurp in the *.boi file. If this is zero or very small, an increased value will provide a small transverse stiffness.

Q: I observe a longitudinal stress pattern which is present from the beginning of the analysis, with horizontal stripes of stress variation?

A: For a non-linear material in a pressure spiral, the initial stress-free configuration is straight. (This differs from elastic material, where the initial helical shape is assumed to be stress-free.) Before the analysis load steps are applied, the production-induced stresses are accounted for. Consequently, the pressure spiral is pre-curved to a level that yields a net-zero bending moment for the model’s helix radius. This implies a spring-back from a larger initial curvature, resulting in the stress pattern observed.

Q: The Axial stress and Normal curvature stress are observed to be identical for the -boundary and -pflex model part in Xpost. However the Transverse curvature stress differs. Why is this?

A: The Axial stress, Normal curvature stress and Transverse curvature stress are populated by PFLEX both for the -boundary and -pflex part of the model. For Transverse curvature results the -boundary model results only contain the dynamic bending induced part of the stresses, excluding the axi-symmetric part.

Q: The Total longitudinal stress differs from the -boundary to the -pflex model part in Xpost. Why is this?

A: The BOUNDARY and PFLEX procedures are different analysis models with different underlying assumptions. The obtained stress components may hence differ. PFLEX is responsible for populating the Axial stress sigma_ax, Normal curvature stress sigma_nx and Transverse curvature stress sigma_tx result types both for the -boundary and -pflex part of the model. For the -pflex model results sigma_ax + sigma_nx + sigma_tx = sigma_xx, Total longitudinal stress. However BOUNDARY also calculates its own axial and transverse stresses in the analysis step before PFLEX has been executed. The Total longitudinal stress reported in the -boundary part of the model is based on these results. It includes the axial stress, and the transverse curvature stress arising from tilting of the components. It will normally differ slightly from the one reported in the -pflex part of the model. As PFLEX has populated the separate stress components, sigma_ax + sigma_nx + sigma_tx is not equal sigma_xx for the -boundary results.

Q: Which stress components are relevant to include in the fatigue assessment?

A: For fatigue assessment the -pflex model results stress results can be applied for transverse cracking of the wires.

For longitudinal cracking, a combination of the results from BOUNDARY and PFLEX must be applied. As the BOUNDARY procedure does not include stresses due to ovalization, PFLEX will calculate and report this. PFLEX will also report the local bending stresses due to global bending, both into the -boundary model. These normal and transverse curvature stresses must be added to the Total longitudinal stress from the -boundary model. The Normal, Transverse and Shear stress from the -boundary model must also be included in the von Mises stress calculation.

To obtain consistent results with LIFETIME the following approach must be followed: