EVALUATION OF FATIGUE LIFE MODELS AND ASSESSMENT PRACTICE FOR TENSION LEG PLATFORMS (PHASE 1: TENDON SYSTEM FATIGUE)

OBJECTIVE: This project will seek to evaluate the state of practice in fatigue assessment in the offshore industry today. Of interest is an understanding of how fatigue life calculations are undertaken and how both load and resistance side uncertainties are employed in design for fatigue. The project will review relevant API Recommended Practices (e.g., API RP-2T, “Recommended Practice for Planning, Designing, and Constructing Tension Leg Platforms,” 2nd Edition, 1997, as well as drafts of a proposed 3rd Edition) and implied safety factors that are inherent to them, as well as compare these with safety factors in other standards (such as DNV’s Offshore Standards and Classification Notes—e.g., DNV-OS-C105, “Structural Design of TLPs (LRFD Method),” October 2005).

BACKGROUND: Tendon systems in tension leg platforms (TLPs) are comprised of various components connected in series. As such, in fatigue design, the predicted fatigue life of a single component must be larger than that of the tendon. Significant uncertainty is associated with assessing fatigue life and the use of reliability theory and statistical methods is unavoidable. Uncertainties arise from both the loading (stress) side and the resistance side. Tendon fatigue safety factors may be expected to depend on the intended service life, nature of in-service inspection, availability and quality of S-N data (resistance side), availability and quality of environmental loads data, and analytical assessment methods employed.

Questions have arisen recently regarding fatigue assessment procedures that relate primarily to whether or not fatigue safety factors employed in the design of offshore structures are adequate or, indeed, the degree of confidence with which these factors can be assessed. Often simplified procedures have been employed for which it is difficult to clearly quantify the achieved factor of safety. There is therefore some interest in better understanding how different assessment procedures compare with each other it terms of bias and uncertainty in the factor of safety.

APPROACH: This first phase of research will focus on fatigue of TLP tendon systems and will seek to provide:

• An overview and evaluation of fatigue life models and fatigue assessment practice currently used for tendon systems.
• A review of the relevant API Recommended Practice (particularly, RP-2T, 2nd Edition, 1997 as well as committee drafts of the 3rd Edition) as it pertains to fatigue assessment models, including deterministic as well as stochastic (spectral) models, and with a focus on rainflow-counting approaches, correction factors sometimes used in damage calculations, etc. A comparison of API RPs with other standards (e.g., DNV) will also be undertaken.
• A report of the differences between different assessment models will be developed especially concerning safety factors and assumptions concerning safety. Ultimately, the goal here is to assess how different practices influence uncertainty in fatigue damage and life assessment and what are the safety factors that result from their use.

While understanding what safety factors are implemented in today’s tendon system fatigue designs is the focus of this study, consideration for manufacturing and fabrication techniques involved, quality control, inspectability, consequences of failure will be addressed while summarizing the state of practice and the safety factors. On the “loading” side of the fatigue limit state, some of the factors that are important especially for probabilistic evaluation are the selection of appropriate fatigue seastates and wave scatter diagrams; the use of spectral versus time-domain (e.g., rainflow-cycle counting) analyses in the time domain; the sensitivity to mean tensile stress levels; cycle rates of loading; etc. On the “resistance” side, likewise, consideration for stress concentration; material selection; S-N curves versus fracture mechanics (and associated inspectability issues) is also important and will be addressed. Finally, in the reliability analyses, system redundancy and inherent uncertainties in modeling will be considered.

Representative TLP configurations will be selected for purposes of evaluating the different tendon component load/resistance models and the fatigue life assessment models. The selected TLPs will include one conventional 4-column TLP and one mini TLP. In order to have a range of contrasting characteristics associated with frequency content, extent of nonlinearity, etc., tendon tension data from model tests in fatigue and extreme (storm) seastates will be used. While having these model test data sets is vital to the project, they will be made available under confidential release; extensive details related to the specific platform will not be provided. To supplement the model test data, we will pursue getting access to some field measurements as well.

The scope of the current project will involve fatigue assessment of various components in a TLP tendon system including the tendon pipe, top and bottom tendon interfaces, intermediate connectors, and flex bearings. Fatigue design is carried out based on fatigue test (S-N) data and cumulative damage analysis and/or methods based on fracture mechanics. Regarding loading, first-order wave loads are most important but tendon loads depend on the platform response which is influenced by wind, waves, current, tides, etc. Also, second-order difference-frequency forces and more importantly second-order sum-frequency (springing) forces can be important. For fatigue, wind and wave spectra and directionality data are important. Tendon load analysis must account for static and dynamic components; the dynamic component of greater importance in fatigue analysis may be studied in either the frequency or time domain. Frequency domain analysis is often used for fatigue assessment since transfer functions can be applied directly with spectra representing different seastates to yield tendon load statistics.

Prediction of fatigue life or damage after fatigue load and resistance models have been defined requires statistical analysis. Probabilistic approaches of varying complexity have been employed in the industry. These include methods based on (i) discrete wave heights and periods (using wave scatter diagrams along with transfer functions); (ii) frequency-domain computation (wave spectra along with Response Amplitude Operators generated from regular or random wave analyses); and (iii) time-domain computation (using stochastic simulations for each seastate in the wave scatter diagram and derived response/load statistics). So-called single-event fatigue damage accumulation associated with low-cycle fatigue from rare extreme events such as 100-year storms are often assessed separately.

DEPLOYMENT OF RESULTS: Results obtained from this study will be communicated to TLP designers, owners, and regulatory agencies through conference presentations and publications, as well as by dissemination of project reports and student theses through normal OTRC and MMS channels. It is anticipated that the results will provide insights into the fatigue design process for TLPs in the Gulf of Mexico as well as help to assess current procedures used in the industry.

PROJECT PLAN FOR 2007 - 2008: The project will be carried out in 7 separate tasks as described below:

Task 1: Literature Review
A detailed review of the literature will be carried out to assess what is the state of practice today with regard to fatigue design of TLP tendon systems. Industry guidelines and recommended practices (API RP-2T, DNV-OS-C105, for example) as well as conference proceedings from OTC, OMAE, ISOPE, etc. will be the focus. The literature review will attempt to establish the most significant sources of uncertainty involved in the fatigue assessment process and to identify the numerous assumptions and gaps that will be evaluated in studies to be performed in the remaining tasks of this project.

Task 2: Definition of Illustrative TLPs
For the purpose of making specific any findings related to the implied levels of conservatism in today’s fatigue design of tendon systems, two illustrative TLP configurations will be selected for this study. Tendon tension time series records available from model tests will be used in the analyses in this task. In order to have a range of contrasting characteristics associated with frequency content, extent of nonlinearity in the loading, etc., the selected TLPs will include one conventional 4-column TLP and a one mini TLP. Records from model tests in fatigue and extreme (storm) seastates will be used. In addition, we will pursue getting access to field measurements.

Task 3: Alternative Fatigue Resistance Models
For the different tendon system components that will be evaluated, it is likely that alternative fatigue resistance models can be employed. This affects the “resistance” side of the fatigue limit state equation and, hence, the implied reliability or derived fatigue life. Industry-accepted fatigue S-N curves as well as others will be considered—for steel components, data on tensile and yield strength, fatigue strength, toughness, and ductility and associated uncertainty will be taken into consideration. Thickness and diameter transitions in tendon pipe, girth weld S-N or fracture mechanics analysis data, and other specific considerations for the various components in a tendon system will be studied in the resistance models.

Task 4: Alternative Fatigue Load Models
As discussed above, there are various approaches with different degrees of complexity that can be employed to assess fatigue loads. These include approaches that use (i) discrete wave heights and periods (with wave scatter diagrams and accompanying transfer functions); (ii) frequency-domain analyses (with wave spectra and accompanying Response Amplitude Operators generated from regular or random wave analyses); and (iii) time-domain analyses (using stochastic simulations for each seastate in the wave scatter diagram and empirically derived tendon load probability distributions). These various load models will be studied.

Task 5: Assessment of Fatigue Life
Based on the resistance and load models described in Tasks 3 and 4, probabilistic approaches will be employed in reliability analyses for fatigue limit states. A suite of different studies will be carried out with models or assumptions on the load and resistance side varied systematically and exhaustively to evaluate the probability of attaining a target reliability level or target service life. Equivalently, this will lead to an assessment of the derived safety factor for fatigue of individual TLP tendon system components. System-level reliability or system factors of safety will also be derived. In this manner, it will be possible to assess the various models employed for resistance and load (from Tasks 3 and 4) for the different TLP configurations selected (in Task 2).

Task 6: Critical Review and Interpretation of Results
Tendon load analysis (Task 4) and fatigue life assessment (Task 5) are expected to account for the most substantial effort in this project. Results from these analyses, especially with regard to safety factors, will be compared for the TLP configurations studied and the load and resistance models employed. On the basis of the various analyses carried out, an attempt will be made to answer the following questions:

• Which environmental conditions or seastates induce largest fatigue damage on tendon components?
• To what degree does resistance model uncertainty influence derived safety factors?
• To what degree does load model uncertainty influence derived safety factors?
• Which components in a tendon system dominate fatigue failure?
• How do fatigue life assessments for different tendon load analysis models compare?
• How is tendon system fatigue reliability influenced by individual component’s reliabilities?
• How different are assessment procedures in the different industry standards (e.g., API versus DNV)?
• To what extent do extreme seastates associated with rare 100-year events (as in hurricanes) account for fatigue damage accumulation?

This task will address all of these questions. Numerical calculations supporting the answers will be provided in the final report. Additionally, these results will be presented in meetings with MMS and industry representatives.

Task 7: Final Report Preparation
A final report will be prepared that documents all of the work carried out in the various tasks. Specifically, details will be presented regarding the various TLP configurations analyzed, the loading conditions and seastates considered, the fatigue load and resistance models used in the life assessment techniques, the assumptions regarding uncertainties at all stages, and the derived safety factors based on current conventional and alternative procedures.

PRINCIPAL INVESTIGATOR(S) & OTHERS INVOLVED IN PROJECT:

Principal Investigator: Dr. Lance Manuel, University of Texas, Austin

Support Staff: One graduate student to be determined