## Fatigue analysis: Introduction |

The OrcaFlex fatigue analysis tool is a post-processor which calculates fatigue damage using a variety of methods. Fatigue damage is collated and summed for specified load cases and then presented in graphs or tables. The fatigue analysis tool is essentially a self-contained sub-program within OrcaFlex, with its own menus, data and results. The tool is accessed from the Results | Fatigue analysis menu item.

Fatigue damage can be calculated in a number of ways, depending on the application:

- Various S-N curve methods recover stress using
**homogeneous pipe stress**(for metal risers),**stress factors**(for umbilicals and flexibles), or**externally calculated stress**(suitable for bespoke applications), then calculate damage based on the specified S-N curve. **Mooring**line fatigue calculates damage from effective tension ranges using T-N curves.

The damage, having been calculated, is summed by one of the following analysis types:

- deterministic
**regular**wave fatigue analysis - deterministic irregular wave fatigue analysis using the
**rainflow**cycle counting method - stochastic irregular wave fatigue analysis using
**spectral**methods

Another calculation method is **SHEAR7**, under which damage is calculated by SHEAR7 rather than by OrcaFlex. OrcaFlex then collates the damage from a number of SHEAR7 load cases.

The **histograms** method does not involve damage at all. Instead, this method performs rainflow cycle counting and collates these cycles into histograms, to be used as inputs to subsequent post-processing – for instance to perform an engineering criticality assessment (ECA).

The steps involved in performing a fatigue analysis are:

- With the normal OrcaFlex facilities, set up and run simulations modelling the various load cases the line will experience. Alternatively, for a SHEAR7 analysis, create a set of SHEAR7 .plt output files to represent your VIV load cases.
- Open the fatigue analysis tool and set up the fatigue analysis data. These data are held separately from other OrcaFlex data, and saved in a separate file.
- Check the data for errors.
- Calculate and collate the damage.

Notes: | The calculation stage of a fatigue analysis can take a long time, especially a rainflow analysis with a lot of load cases. To help you manage this, you can estimate the calculation time, and fatigue analyses can be run in batch mode. All available processor cores are used, to process load cases concurrently. |

The overall speed of the fatigue calculation is often limited by disk access – consequently it is important that the disk access speed is as fast as possible. This usually means that the simulation files are best stored on a local disk of the machine performing the fatigue calculation. |

Before the fatigue analysis can be performed you must first prepare a set of OrcaFlex simulation files that model the same system but under the various load conditions that the system will experience in its lifetime. The approach is to divide the range of sea states which the system will experience into a number of **wave classes**; typically this is done with a wave scatter table.

For both regular and rainflow analysis, each wave class is represented by a distinct OrcaFlex dynamic simulation. For regular analysis, the simulation should use a regular wave representative of the wave class; for rainflow analysis, the simulation should use a representative irregular wave.

For regular analysis, time domain dynamics must be used. For rainflow analysis, either time domain or frequency domain dynamics can be used. If frequency domain dynamics are used, then the rainflow calculation is based on synthesised time histories.

In a frequency domain spectral analysis, each wave class should be represented by a distinct OrcaFlex frequency domain simulation that uses an irregular wave representative of the wave class.

For response RAO-based spectral analysis, multiple wave classes with similar $\Hs$ values may be represented by a single simulation. This must be either a time domain response calculation simulation, or a frequency domain simulation. In either case, OrcaFlex derives response RAOs and combines them with wave spectra to produce response spectra, for which damage is calculated.

SHEAR7 fatigue analysis requires that the load cases are specified by a set of SHEAR7 .plt output files. These are most easily generated using the OrcaFlex SHEAR7 interface together with the standard OrcaFlex automation facilities – the .plt files are automatically exported if you run the direct SHEAR7 interface in batch mode.

Each load case is assigned an exposure level. For regular load cases, this is the total number of occurrences of waves within the wave class; otherwise, the exposure level is the total time exposed to waves within the wave class.

You must choose (other than for histogram and SHEAR7 calculations) one of the three analysis types for your fatigue analysis: regular, rainflow or spectral. To help guide your choice, the characteristics of these types are as follows.

**Rainflow** fatigue is the most accurate of the methods, but also the most demanding of computer time and disk storage. These requirements can be somewhat alleviated by careful selection of load cases. The other factor which can be adjusted is the duration of the irregular wave load case simulations: reducing this duration will ease the time and storage requirements, at the cost of making the analysis less accurate.

**Regular** wave fatigue analysis is much faster and requires much less disk storage than rainflow fatigue. The wave scatter conversion facility provides an efficient and productive way to generate a regular wave scatter table from a random sea scatter table. Provided that the regular wave bin discretisation is performed well, the results of a regular wave fatigue analysis will generally be in good agreement with those of an equivalent rainflow analysis.

The **spectral** methods are also faster than rainflow but, again, can be less accurate. If frequency domain analysis is applicable and provides sufficiently accurate results then it is to be preferred over the response calculation alternative. However, if your model is not amenable to frequency domain analysis then response calculation may be a good choice. The quality of the results obtained using the response RAO-based spectral method depends critically on the accuracy of the response RAOs. You should take care to check that these RAOs are reasonable.