Blogs

FEA Used to Verify Fillet Welded Patch Plate Repairs for Corrosion Under Pipe Supports

by VITALY DRAMARETSKIY CENG

Introduction
Corrosion under pipe supports (CUPS) is a variation of touch point corrosion, which is a damage mechanism that many industrial facilities have to continuously fight with as they age. Contact points between pipes and their supports create areas for dissimilar metal and oxygen concentration cell corrosion, leading to a rapid and highly local­ized external wall loss (see Figure 1), which ultimately poses an inherent integrity threat as this type of corro­sion is difficult to avoid, identify, measure, and remediate.

Facilities having long runs of off-plot piping, such as tank farms, have to spend millions of dollars on inspection campaigns, aiming to timely detect and quantify the damage before it is too late. As a result, it is often found that numerous lines are no longer fit for service and have to be repaired or replaced. Off-plot pipes typically run for hundreds of meters without a single isolation valve. Taking them out of service for a partial or complete replacement is a major exercise, which leads to significant financial losses. On the other hand, a permanent repair on the run, such as restoring the wall by a local weld buildup, is rarely possible due to the surface roughness, corrosion scale, lack of remaining wall thickness or uncertainties in estimating the extent of metal loss.

Out of all possible permanent repair options, one economically attractive solution is fillet welded patch plates (see figure 2), described in ASME PCC-2.

Fig. 2 Filet welded patch plate repair at CUPS location.

Article 212 provides a conservative method to design such repairs; however, its analyt­ical design rules are based purely on hoop and longitudinal loads resulting from pressure. While the code requires an additional consid­eration to be given to external loads, it provides no specific guidance on how it shall be approached. For CUPS locations, a reaction force from the underlying support is an additional load that must be taken into account as its magnitude may have a significant effect, especially in large diameter lines. 

Elastic Stress Distribution Analysis

As a first step in the study, a stress distribution analysis was carried out based on a 508 mm (20 in.) carbon steel (C ≤ 0.3%) pipe in its standard schedule, operating at 20 barg (290.075 psig) / 50 °C (122 °F), with a fully welded 60° wide / 400 mm (15.75 in) long / 10 mm (0.4 in) thick patch plate (see Figure 3 on page 20). Pipe support was initially excluded from the analysis to assess the effect of its presence on the stress distribution at a later stage.

Also, as it is virtually impossible to monitor the remaining pipe’s wall thickness underneath the patch plate after it is installed, it was conservatively assumed that the wall is completely corroded away up to the edges of the patch, ultimately making it part of the pressure boundary. (See Figure 4.)

The following observations were made in this part of the study:

• Maximum principal and equivalent stresses were highest at the axial midplane on the longitudinal sides of the plate and the peaks were at the weld’s toe on the pipe’s side;
• Hoop stress peaked at the corners of the plate at the weld’s toe and the linearized average membrane stresses were slightly higher than at the plate’s center, while the computed membrane + bending invariants were significantly lower;
• ASME PCC-2 method significantly over predicted the maximum membrane + bending stress at the weld compared to the Finite Element Analysis (FEA) results.

Pipe Support Effect

·    To model the effect of a pipe support, a fixed 25 mm (0.98 in) steel rod representing a typical sleeper was brought in contact with the pipe underneath the patch plate. Shear forces and bending moments were applied to the pipe’s ends to represent the free hanging spans up to the next supports six meters (20 feet) away. The immediate finding was that the maximum equivalent stress raised to an extreme value of >1000 MPa (145038 psi) at the contact point, despite the fact that the contact pressure was spread over a number of nodes. The stress field; however, was highly localized and did not interact with the plate’s welds. (See Figure 5.)

·        The second major finding was that both maximum principal and transformed hoop stresses coincided at the midplane, hence aligning with the axial position of the support.

To further assess how the pipe with a patch plate would fail, an elastic-plastic analysis was carried out assuming isotropic hardening. The main conclusions were as follows:

• No structural instability was observed even for the highest load factor. No local failure was identified and the maximum equivalent plastic strain capped at 6.4%; 
• Redistribution of stresses at the contact point due to local yielding did not affect the overall load transfer and stress levels at the patch plate’s welds; 
• Maximum equivalent plastic strain was observed at the axial midplane at the weld’s toe on the pipe’s side, pointing towards a potential failure initiation at this location in case of overload; 
• Pressure is the failure-driving load, while the support‘s reaction / bending moment led to the concentration of the prevailing stresses at the midplane. (See Figure 6.)

Parametric Analysis

In order to study the effects of changing geometrical features, several additional models were run. An increased plate thickness / weld leg, longer / wider patch plates and effects of a pinhole / undamaged pipe wall on various pipe sizes / schedules were analyzed. The find­ings can be summarized as follows:

• Presence of the pipe wall under­neath the patch plate majorly reduces the bending stresses at the weld; however, that effect is significantly diminished when a pinhole is made in the wall;
• Increasing the patch plate thickness / weld leg reduces the bending stresses at the weld, although it may lead to the shift of stress peaks to the weld’s root;* 
• Increasing the length / width of the patch plate had no effect on stress levels.* 
  * Pipe support was omitted from these models. 

Bulk Verification

After analyzing the findings, a bulk verification using the elastic stress analysis method was done for various pipe sizes and schedules. The following stress allowables were used:

• Average membrane stress – 138 MPa (20015.2 psi); 
• Membrane + bending stress – 207 MPa (30022.8 psi);
• Triaxial stress limit – 552 MPa (80060.8 psi). 

The above limits were set assuming ASME B31.3 as the design code and ASTM A106-B steel as the base material. Design conditions were 20 barg (290.075 psig) / 150 °C (302 °F), and the span length was six meters (20 feet). It shall be noted that the allowables did not include a weld efficiency factor. 

During the analysis, stresses at the midplane were linearized at the weld’s toes, both through the pipe and the plate. Patch plate dimen­sions that satisfied the criteria are tabulated in Table 1.

All patch plates analyzed had the circumferential width of 60° and were centrally positioned at the low point of the pipe. Forming strains are given for reference, it shall be noted that above 5% it is required to carry out appropriate postforming stress relief prior to installation, as per ASME PCC-2.

During the analysis, a number of patch plates also failed the assess­ment (see Table 2).

More than 90% of these failed due to the exceedance of the membrane + bending stress levels while the few remaining models showed exceedance of the average membrane and triaxial stress limits. In all cases, the highest stresses were located at the weld’s toe on the pipe’s outer surface.

All weld leg sizes were equal to the plate thickness minus 2 mm (0.08 in) to avoid melting of plate’s edges and creating a stress riser. Further increasing the thickness of the plates that failed the assess­ment was not seen appropriate as it would end up with a weld throat thicker than the pipe’s wall. For large diameter lines, it was also discovered that while increasing the thickness reduces the bending component, it also causes the average membrane stress to rise due to the shift of the neutral line relative to the pipe’s axis, which is caused by the increased plate thickness.

Implementation Risks

The patch plates described here were intended to be welded to live equip­ment, which is a high risk activity on its own and a thorough assessment shall always be done to confirm that it can be executed safely (e.g. see API RP 2201) and no burn through will occur during the process.

Apart from that, welding a plate under a support implies that the pipe will need to be lifted, which also requires an engineering assessment, especially in view of the reduced wall thickness due to corrosion.

The other issue is the level of non-destructive examination that one can do at the touch point location itself and the weld around the patch. Even a tiny leak of a flammable substance during welding may have disastrous consequences. Thus, one needs to gain certainty around the amount of wall thickness left at the corrosion patch.

On the other hand, the lack of possibility to execute volumetric inspection on the welds and the fact that the calculations assumed a geometrically perfect weld without a weld efficiency factor, calls for a careful approach to welding quality and NDE. Remedies such as multi­-layer MT/PT and job-specific qualifi­cation of welders are some options that may help mitigate the risks.

Last, but not least, installing a patch plate may create an internal crevice corrosion risk, so an appropriate inspection and monitoring strategy may be needed for such cases.

Conclusion

Viability of patch plate repairs for CUPS locations in off-plot piping has been demonstrated and the results of this study have already been successfully used to carry out numerous repairs, saving millions of dollars in replacement costs. However, engineers shall always assess the limitations of applying this repair method and the risks it carries versus advantages of other permanent repair types and/or replacing the corroded lines, either partially or completely.

This article was originally published in MTI CONNECT 2023, Issue 2. See the full article for all figures and tables.