ﴀﴀﴀ OGP Workshop on Lifetime extension – Pau – June 11 & 12, 2007 Wave data for fatigue analysis; Multi-modal wave & response spectra Valérie QUINIOU-RAMUS, TOTAL S.A. Michel OLAGNON, IFREMER

2 Many questions left unanswered… A proposal for a way forward… - ﴀﴀﴀ OGP Workshop on Lifetime extension – Pau – June 11 & 12, 2007

3 Background, from the metocean prospective A typical West African sea-state measured in Girassol, Angola B17 How to calculate the fatigue damage of a structure subject to the combined action of swell and wind sea?

4 - ﴀﴀﴀ OGP Workshop on Lifetime extension – Pau – June 11 & 12, 2007 Wave data - Case No.1: Typical of a W.A. shallow-water offshore platform Often scarce and old wave records with limited archived parameters  post-processing impossible to partition swell / wind sea Usual practice: use of hindcast model data (such as WANE/NOAA)  Hindcast is suitable for representing swell but not wind sea  If swell comes from hindcast (~10-20 years) and wind sea from “regional” records (~1 year), loss of time concomitance: how shall we recombine the two?

5 - ﴀﴀﴀ OGP Workshop on Lifetime extension – Pau – June 11 & 12, 2007 Wave data - Case No. 2: Nigeria OML130: 1 year of in-situ wave records Detailed in-situ data available but… Representative of a 20-year lifetime ? … Use of hindcast model data to extend the time period?  Same issue as for fixed platform (loss of time concomitance)

6 - ﴀﴀﴀ OGP Workshop on Lifetime extension – Pau – June 11 & 12, 2007 Wave data - Case No. 3: Angola B17: 3 ½ years of in-situ wave records Recent survey, relatively long duration Detailed data available Various ways to display the wave data to be used in design analyses: (assuming that 3½ years are representative of the lifetime environment, which is yet another debate…)  Individual wave count: zero-up/down crossing or cycle count with rainflow filter  Spectral parameters Hs-Tp-Dir representing the total energy  Extraction of wave systems (2 swells + 1 wind sea) and spectral parameters Hs_i-Tp_i- Dir_i for each partition :  Displayed as 3 independent sets of distributions + rules for recombining the systems together  Or: Display the 3 ½-year time series of the 3 simultaneous systems Which one is the best suited for fatigue design?

7 - ﴀﴀﴀ OGP Workshop on Lifetime extension – Pau – June 11 & 12, 2007 Various fatigue design methods, requiring different formats of wave data… Fixed Structures: deterministic (individual waves H-T) or spectral approach (sea-states Hs-Tp; swell + wind sea Hs i -Tp i )? Floating Structures (mooring systems): usually spectral approach, with Low and High/Wave Frequency components Flexibles Risers : deterministic only...?

8 - ﴀﴀﴀ OGP Workshop on Lifetime extension – Pau – June 11 & 12, 2007 Simple summation Simple summation is often used to calculate the damage to a system due to the combined action of swell and wind sea (or more generally LF and HF): The damage is calculated separately for each component and then added together.

9 - ﴀﴀﴀ OGP Workshop on Lifetime extension – Pau – June 11 & 12, 2007 WARNING: simple summation may not always be conservative! Two load histories shown in figure 1. Both are narrow-band, of the same amplitude. When considered individually, they contribute either 10 or 1 cycles of the same amplitude to the fatigue damage. Simple Summation would then assume that the total damage due to the sum of the two signals is that of 11 cycles of range 2. ___________________________________________________ In practice, the signal resulting from the summation of the two load signals of figure 1 is shown on figure 2. Rainflow counting gives 10 cycles, that one of those cycles joins the highest crest to the deepest trough, i.e. its range is almost 4, and the 9 other ones have range close to 2 (1.78 in average).

10 - ﴀﴀﴀ OGP Workshop on Lifetime extension – Pau – June 11 & 12, 2007 WARNING: simple summation may not always be conservative!

11 - ﴀﴀﴀ OGP Workshop on Lifetime extension – Pau – June 11 & 12, 2007 Alternate solutions for fatigue damage calculation Briefly speaking: rainflow = “exact” method should a response history be available, but time-consuming  Need to choose in the frequency domain a method that is guaranteed to be conservative (for instance Jiao-Moan is not always), but not too conservative (Dual Narrow-Band is much too conservative in some cases, same for combined Spectrum).  When the sea-state is bimodal (swell + wind sea), can we easily calculate the floater response from its response to each wave system separately?  Should the deterministic approach be selected, how physical and reliable is it to calculate load/stress cycles from the response to individual ocean wave counts?

12 - ﴀﴀﴀ OGP Workshop on Lifetime extension – Pau – June 11 & 12, 2007 Proposal for a way forward… IFREMER, based on their long experience in fatigue of fixed structures and metocean, have started comparing methods and proposing alternatives.  See Olagnon, M., Guede, Z., IFREMER, “Rainflow Counts for the Combination of High and Low-Frequency Loads”, submitted to Elsevier Marine Structures Journal  Various response spectra will be analysed (typical of deepwater moorings, risers, fixed structures etc.) Other study proposals have been discussed and drafted with:  PRINCIPIA for flexible riser response  STRUSIM for jacket response Partly funded by TOTAL, but could be open to a JIP in order to get a broader audience and ensure a transfer knowledge to the industry.