Analysis points the way to project savings

Design by analysis rather than the traditional formula approach is the way forward for pipeline operators and contractors looking to cut project costs and timescales, argues TRaC Global consultant Dr Peter Moir.

Analysis to validate the design of pipeline systems is normally based on simplified code of practice methods (‘Design by Formula'), an approach that demands that construction conforms to the assumptions on which the calculations are based. Build errors or mis-manufacture that introduces effects not accounted for in such methods will require remedial work to ensure the integrity of installations, resulting in both direct and project delay costs.

Given that there will be uncertainties associated with on-site construction etc, Design by Formula methods are necessarily conservative. Hence scope will always exist for the use of sophisticated ‘Design by Analysis' finite element methods to show whether operational stresses are actually within allowed limits. The attraction in using computer simulations is made the greater since the costs and timescales associated with such studies are typically a fraction of those that would result from on-site reconstruction work.

In the manufacture of large bore seamed pipe, which is widely used in pipelines, there are two main causes of local distortions that result in non-circular sections. The abutting plate edges may not be curved correctly during the forming process (‘roof-topping') or distortions may be caused by residual stresses in the longitudinal weld. Frequently this is minor, with pipe ends being trued up to ensure that no stress concentration effects (SCF) exist at butt welds between adjacent pipe sections or fittings. When pipes are cut to length on site, however, the section distortion at the point where the pipes are cut can be marked, so creating a hi-lo mismatch at joints that introduces an uncertain SCF.

There is no particular specification regarding the fit up of butt joints having noncircular hi-lo mismatch. However, ASME B31.8 sets limits on discrepancies in the radial alignment of butt welds between sections having unequal wall thicknesses, and this can be taken to apply to misalignment generally. For pipes close to the same wall thickness the limit on profile mismatch is nominally 2.4mm.

The implications for weld integrity of hi-lo mismatch beyond ASME limits was assessed by TRaC for a major client in the UK. The research investigated the impact of a 914mm bore gas line having hi-lo misalignments mainly in the range of 4-5mm in a system that comprises pipes connected to either other pipes, bends or tees. The client observed misalignment that extended circumferentially up to 300mm in extreme cases, ie up to 10% of the girth weld length. Wall thicknesses at the various butt joints varied from some 20mm for line pipe up to 40mm for tee fittings.

Using modern FEA tools, the analysis of the applied forces to assess durability against sustained, shakedown and fatigue criteria is straightforward. However, the number of geometry variations – pipe to pipe, pipe to bend and pipe to tee – compounded by the level, the shape detail and the orientation of the profile mismatch compared to the orientation of XY bending forces means that each joint should strictly be treated as an individual case. Since studying a large number of particular geometries would be lengthy and expensive, the risk was assessed by making geometric approximations and taking a ‘worst case' approach.

Three solid geometric models including weld preparation and profile geometry were used to represent the various welded joint configurations. Bends and tees were not modelled as such, each model being two sections of straight pipe with appropriate wall thicknesses either side of the weld junction. Whilst an approximation, this approach is technically acceptable since the detail of the bend or tee shape is secondary to the thickness of the fitting wall thickness limiting the ovalisation of the junction under pressure and bending. The FE model included a length of the line pipe equal to three times the pipe diameter either side of the weld so that the external loads could be applied at points remote from the local weld discontinuities. Provisions were made for creating stress classification lines for stress linearization, along which predicted stresses were compared to allowable values by creating a suitably positioned series of nodes on inner and outer surfaces around the critical circumferential lines. Figure 2illustrates the selected positions of the Stress Classification Lines. Note that in this Figure the highly refined mesh in the area of the weld that is made progressively more coarse further from the weld is omitted for clarity.

The finite element model of the pipe was rotated about the pipe axis so that the maximum resultant bending moment at the junction produced the maximum tensile axial stress at the point of maximum stress due to internal pressure alone. This ‘worst case' approach is judged to be conservative.

An analysis of the overall system, using 62 conventional load cases, was first carried out using CAESAR II, assuming a SCF of 1 at the misaligned weld joints. The global element forces and moments found in this model were extracted and applied directly to the end nodes of the pipe sections in the finite element model within Abaqus, and an elasto-plastic analysis was performed.

Typical results from the simulations are given in Figures 3 and 4, showing the overall deformed shape under internal pressure load and Tresca stress contours under sustained loading. The pipe to pipe junctions were found to be the most highly stressed, probably due to the greater reinforcement provided to the noncircular pipe by the thicker walls of the bend and tee fittings.

The analysis produced some interesting results, in particular the pressure case that will tend to cause distorted sections to become circular was found to introduce the highest stresses. This suggests that ovality, even in otherwise aligned joints, could be as much of an issue in causing a stress concentration effect as lateral misalignment.

Establishing values for allowable stresses can present some difficulties in that alternative codes of practice are not always completely consistent. For example if you compare the pressure vessel code PD5500 against the Design by Analysis method of Appendix 6 of the pipeline code IGE/TD/12 Edition 2, the general primary membrane allowable stress is lower for PD5500 than TD/12, whilst other components are the same. In addition, it should be noted that different mathematical techniques are used, with Design by Formula methods based on von Mises's stresses, whilst Design by Analysis uses Tresca stresses.

The analysis showed the weld joints to be satisfactory.

This meant that no remedial work was required, saving both project time and cost that would have been incurred if a conventional Design by Formula approach was used. Given the potential savings, we would expect an increasing number of pipelaying companies to adopt a Design by Analysis approach. OE

Dr Peter Moir, CEng FIMechE, worked as a senior engineer at the GEC Whetstone Research Laboratories and as a senior research fellow at Lanchester Polytechnic, working on wave energy. He later formed ETA Engineering Consultants with a colleague and wrote finite element software marketed worldwide by Hewlett Packard under its own brand name. Following the 1993 acquisition of ETA by TRaC Global he became technical director, serving in that capacity until his retirement in 2010 since when he has been retained in a consultancy role.