Gas turbine power stations are being added to the electricity grid, primarily to provide capacity to supply peak loads. Usually these are intended for future conversion to combined-cycle base load operation, but to date only a few installations have made this transition. Gas turbines provide operational flexibility and, in the combined-cycle configuration, significantly reduce carbon emissions.
Most Australian pipelines are designed for transmission capacity, and do not provide sufficient storage to deliver gas for long periods of time at the rate required by a large gas turbine power station – for example, 6.5 TJ per hour for a 600 MW unit. Because peak load power stations are operated infrequently, the gas shipper is required to compensate the pipeline operator for the loss of capacity resulting from the supply pipeline being used to provide a storage service. This has resulted in the development of a number of gas pipelines specifically intended to be operated for gas storage, and it seems probable that this demand will increase in the future.
Storage pipelines are designed to operate over a large pressure (hoop stress) range. The maximum pressures between 10 and 15 MPa, and minimum pressures of 3-4 MPa are used. While this may appear to be the primary driver for fatigue failure, temperature cycles associated with the pressure cycle and intensified at geometric features such as bends, thickness changes and welds, may significantly reduce the number of stress cycles to failure, compared with the average transmission pipeline.
Storage pipeline design
Three types of pipeline are used for gas storage:
1. A large diameter, long pipeline with spare capacity – for example the DN 850 mm, 1,300 km Moomba to Sydney Pipeline. This pipeline type can usually tolerate a peak load demand without a significant pressure (and associated temperature) cycle. This type of pipeline is represented in Figure 1 as Pipeline 1: a 1,000 km, 800 mm diameter pipeline with a pressure of 3.8-3 MPa.
2. A modest diameter, modest length pipeline – for example the DN 400 mm, 100 km Condamine to Braemar Gas Pipeline. This pipeline type experiences a modest temperature change above soil ambient at the gas receipt point during pressurisation, and a modest temperature change below soil ambient during depressurisation. In Figure 1 this is Pipeline 2: 100 km, 400 mm diameter pipeline with a pressure of 9-3 MPa.
3. A large diameter short high pressure pipeline – for example the
DN 1050 mm, 9 km Colongra Lateral. This pipeline type experiences significant pressure change over the whole of the pipeline, with temperatures above ambient during pipeline compression, and temperatures below ambient during depressurisation. In Figure 1 this is Pipeline 3: 8 km, 1,050 mm diameter pipeline with a pressure of 13-3 MPa.
The effect on fatigue life of stress ranges produced by the pressure load and thermal load cycles are significantly different for each pipeline type, and are critical inputs to a subsequent fatigue analysis of the pipeline.
For the first and second pipeline types the temperature changes through each operating cycle can be predicted with a reasonable level of confidence, using typical transient hydraulic computer programs whose heat transfer calculations have been reasonably calibrated with the operation of a similar pipeline. The third pipeline type requires detailed thermal analysis of the pressure-temperature cycle in the proposed pipeline, and it is essential that the computer tools used for this work are demonstrated as being capable of modelling the pipeline performance, and that the designer has knowledge of the soil properties and annual temperature ranges at pipeline depth.
Fatigue affects the pipe base metal, but failure occurs in the welds first due to geometric features such as misalignment and weld shape that intensify the stress, and also because of the metallurgy and structure of the weld, which may include features such as voids, cracking, cast-like microstructures and undercut.
Fatigue analysis methods
There are two types of analysis that can be applied to pipe welds.
1. SN curves (stress range versus number of stress cycles), which assess the possibility of cracking or failure in a nominally defect-free weld; and,
2. Fracture mechanics analysis (fatigue crack growth), which calculates crack growth in defect of a defined size.
SN curve method
SN curves for welds are derived from testing defect-free welds with normal small manufacturing effects such as undercut. AS2885.1 Appendix N provides a conservative assessment basis, using a maximum stress range of 165 MPa for pressure stresses only – not for thermally induced stresses. Where these guidelines are not met, due to a higher stress range or significant thermal stresses, the standard recommends a detailed fracture mechanics calculation be undertaken by BS7910, a guide on methods for assessing the acceptability of flaws in metallic structures, or similar.
The investigations undertaken for a number of cases showed:
1. The fatigue life of the longitudinal seam weld of submerged arc welded (SAW) line pipe is approximately one order of magnitude greater than the typical design life. Electric resistance welded (ERW) and seamless pipes are expected to resist more stress cycles. From this it could be concluded that the pipe cylinder life is unlikely to be limited by fatigue failure.
2. The fatigue life of girth welds under cyclic pressure loading only is around three orders of magnitude greater than the typical design life. From this it could be concluded that the life of pipelines similar to type 1 (with essentially pressure loading only) is unlikely to be limited by fatigue failure.
3. The fatigue life of girth welds under a combination of cyclic thermal and pressure loading may be one order of magnitude less than the typical design life. From this it must be concluded that detailed fatigue studies are required for pipelines similar to type 3, and are probably required for pipelines similar to type 2.
However this simple assessment does not necessarily tell the full story.
Fatigue crack growth calculations
In a pipeline that passed the AS2885.1 Appendix N fatigue criterion, a crack could still grow to failure by cycling. Hence the need for a fatigue crack growth calculation when the cyclic loading from process temperature changes is considered.
For example, in a 1,000 mm outer diameter X70 pipeline with a wall thickness of 25 mm, which is operating with a 30 per cent specified minimum yield strength (SMYS), the hoop stress pressure range is designed for 4,000 cycles. By Appendix N the stress range of 145 MPa gives a life of 9,360 cycles. Fatigue crack growth in a seam weld was analysed with a 3 x 25 mm surface breaking internal crack. Two cases were analysed to show the effects of variables:
* Case A has a pressure range of 14.49 to 7.245 MPa (60 to 30 per cent SMYS), an offset of 3 mm, and peaking of 3 mm, with ovality of 0.75 per cent, using the API definition of ovality = (Dmax-Dnom)/Dnom. The seam weld was not post weld heat treated (PWHT) and the pipeline was hydrotested to 75 per cent SMYS.
* Case B was analysed with a pressure range of 8.245 to 1.0 MPa (34 to 4 per cent SMYS), an offset of 1 mm, and peaking of 1 mm, with ovality of 0.5 per cent. The seam weld was PWHT and the pipeline was hydrotested to 100 per cent SMYS.
The results are shown in Figure 2 and Figure 3. Case A has a predicted life of 150 cycles and Case B has a predicted life of 69,000 cycles. The large difference is due to weld geometry, relaxation of stress intensification from the weld geometry due to hydrotesting, and residual stress (which is affected by both hydrotesting and PWHT).
Stress analysis
Stress ranges to use in crack growth analysis are obtained through a number of steps.
Temperatures and pressures for operation need to be identified. The temperatures come from gas flow calculations. An important observation, which links with research being undertaken in APIA’s Low Temperature Excursions project, is that designers and pipeline operators have little experience in modelling the thermal cycles associated with the cyclic operation of the storage pipeline, particularly heat transfer between the gas and the soil. Until a better understanding of this modelling occurs, conservative methods and assumptions should be used. It should also be noted that full depressurisations may be a significant cause of fatigue damage.
These temperatures and pressures are used in a piping stress analysis package to calculate the global stresses at various points on the pipeline for each operating condition.
The global stresses from the piping stress analysis package are intensified by welds through features such as thickness change, axial offset, angular misalignment, and weld toe angle. The actual stresses can be calculated by analytical methods or by finite element analysis.
Fatigue crack growth is calculated using the actual stress ranges at welds with an assumed crack size.
Residual stress has little effect on crack growth rates but affects the final critical crack size. The residual stresses can be estimated from BS7910.
Design and construction
A high stress range may cause rapid crack growth and lead to very small acceptable defects and frequent inspections. In some projects flaw sizes allowed by AS2885.2 tier 1 may not be acceptable.
The geometric anomalies associated with weld need to be controlled:
* Tight limits on axial (Hi-Lo) misalignment < 5 per cent, are recommended.
* Tight limits on angular misalignment (based on cut end tolerance in the line pipe specification and the use of internal fit-up clamps) are recommended, preferably less than 0.2 degrees.
* Tight limits on internal weld bead angle, < 30 degrees, are required.
* For seam welds misalignment and peaking must be controlled.
Hydrostatic testing or cold expansion can significantly reduce the effects of these geometric features in seam welds (and may need to be specified during construction), but hydrostatic testing has little effect on the fatigue life of girth welds.
Recommendations for developers and designers of gas storage pipelines
* Gas storage pipelines require much more detailed analysis than a regular pipeline.
* Long, modest diameter pipes are less likely to have their life limited by fatigue than are short, large diameter pipes.
* Temperature and stress modelling is recommended before pipe procurement.
* Hydraulic models used to predict the pressure and temperature relationships through operating cycles may require calibration if the pipeline pressurisation and depressurisation rates are sufficient to introduce significant temperature changes
* Failure by crack growth at seam and girth welds is a possibility within the operating life of a storage pipeline, even in pipelines that pass the AS2885.1 Appendix N fatigue stress criterion.
* This effect is enhanced in short, large diameter storage pipelines.
* The designer may need to define very small acceptable defects, tight tolerances on weld geometric features, and significantly increased inspection.
* If a new pipeline may be used for cyclic loads in the future, it should be constructed with tight weld geometry and weld acceptance criteria.