Fatigue Prediction for Extended Riser Life and Improved Vessel-Response Analysis
This paper presents a fatigue-prediction methodology designed to extend the life of unbonded flexible risers and improve the accuracy of floating production, storage, and offloading vessel response analysis.
This paper presents a fatigue-prediction methodology designed to extend the life of unbonded flexible risers and improve the accuracy of floating production, storage, and offloading (FPSO) vessel response analysis. The methodology combines measured-motion-response, maritime-environment, and process data to improve traditional time-domain dynamic analysis models, along with machine-learning (ML) techniques to develop a heading model for the FPSO.
Life extension of unbonded flexible risers is a challenge because of uncertainties associated with key parameters driving deterioration, such as how the riser was designed, manufactured, and installed and how it is operated. Fatigue service life is calculated during the design phase using models based on field-specific environmental data derived for the area and loads for the required design conditions per project specifications and international standards.
Using real measured data from the riser motions and from environmental sources can reduce uncertainties significantly, improving life-extension analysis for fatigue and other failure drivers. These data, when combined with high-end engineering assessments and considerations on data reliability, provide better insight and contribute to continued operations with acceptable risk for the risers.
The complete paper presents a methodology that uses measured environmental data and FPSO and riser response data in an ML environment to build more-realistic riser-response and fatigue-prediction models. A case example is presented for the risers suspended from the FPSO Fluminense producing from the Bijupirá and Salema fields offshore Brazil. Because FPSO heading is important for vessel dynamics, especially roll, and the vessel dynamics are a key factor in the riser dynamics at this field, initial focus was on predicting vessel heading relative to swell. The heading model developed by ML showed good agreement and was used as a key tool in a traditional fatigue analysis. This analysis was based on historical sea states from the last 2 years. The results show that the fatigue analysis from the design phase is conservative and lifetime extension is achievable.
Because the fully instrumented measurement campaign ended after 4 months, the work focused on using all captured data to provide improved insight and develop both traditional simulation and ML models. For future fatigue predictions based on the developed fatigue counter, the objective is to maintain accuracy with less instrumentation.
In the present phase, FPSO and riser-response data from the 4-month campaign have been used to establish a correlation between riser behavior, environmental data, and FPSO heading and motion. Calibration of a traditional numerical model is performed using measurement data along with a direct waves-to-fatigue prediction based on modern ML techniques.
The ongoing work illustrates how real environmental data and response monitoring can reduce uncertainties and improve and extend flexible-riser life.
The improved armor-wire-fatigue methodology developed in this project uses accurate motion-response sensors directly installed on the riser and interfacing FPSO structures. The response data are combined with measured environmental data to build more-realistic fatigue models. The objective is to develop a model that represents the real-life responses of the FPSO in various swell- and wind-driven sea states, defining with high accuracy the long-term characteristics of the riser system.
Access to high-quality environmental data for tail-end production and life-extension projects of older assets may be a challenge. Efficient use of existing environmental data improves understanding of environmental loadings. Wind, wave, and surface-current data from different sources are assessed and verified for quality and potential weaknesses under varying conditions.
Actual operational data such as topside and subsea pressures and temperatures from the production information system are used instead of design limits to increase accuracy of fatigue development. Integrating the operator’s process-instrumentation network and riser-status databases enables use of actual exposure conditions. Measurement data, component data, and new insight from the improved analytical models are visualized by live contextualized dashboards that provide all stakeholders with the same easy-to-access, relevant, updated, and consistent information.
Environmental Data and Motion Monitoring
The complete paper discusses environmental data and motion monitoring. Because vessel roll is a key factor for the riser bending near the hang-off where the bend stiffener is located, models need a good representation of the wind, current, swell, and vessel heading to perform reliable fatigue analysis. When high-quality field measurement data are available at good resolution, better understanding of the dynamics and model improvement is achievable within a reasonable time. However, for periods without access to environmental and motion response data, good analytical models for both vessel response and for global and local riser analysis are necessary.
Swell is important for total wave conditions in Brazilian waters and is important for roll motion, especially for free-weathervaning FPSOs. In sea states dominated by swell, good insight in the heading of the FPSO relative to the incoming swell is important. During sea states with high wind, the FPSO will align with the wind direction and may expose the vessel to beam sea swell, possibly giving higher roll motions.
A key activity in this project was to develop a heading model for the FPSO, predicting the weathervaning response from input of wind, waves, and surface current. The authors’ preferred approach was to use field measurement data and ML. The more data are obtained, the more the model improves. Thus far, the model is trained on average from 3-hour sea states, so, for transient conditions, inaccuracies are inevitable. The model is trained as more field measurement data become available, and significant improvements are seen from the initial models to the current model. Several different ML approaches were tested to find the best possible approach for the data set of 4 months with 3-hour statistical data, giving approximately 600 samples after 3-hour periods with nearly no motions removed.
The ML processes used for developing the heading model were also used for the prediction of FPSO and riser-motion responses. The objective was to obtain a sufficiently accurate model enabling reliable prediction of dynamic responses in time periods where no response measurements are available.
The complete paper outlines the model-training process. The ML models proved to be on the same level or better than a time-simulation model in predicting motion responses when both models were fed with similar input from the environmental conditions.
As Fig. 1 illustrates, compared with the original design analysis, the accuracies of ML predictions based on real environmental data and models trained on measured field data enable a step toward reliable dynamics in flexible-riser fatigue.
Developing a reliable online fatigue counter for the flexible-riser armor wires began with validating the available environmental data for the FPSO site. A site with a comprehensive measurement program for waves, current, and wind with good time resolution would have been preferable. However, in this project, a robust and accurate alternative was found.
This approach will be relevant for the majority of brownfield assets. Newer installations in deeper waters likely will have more onsite measurements to support integrity-management analytics.
When combining quality weather data and observations; field measurements of FPSO, turret, and riser motion; and normalized fatigue/damage curves derived from validated models, the live fatigue counter can be established. The fatigue counter will account automatically for bore pressures different from planned operational pressure, as well as environmental conditions experienced in the field. Depending upon conservatism built into the design phase, possible life extension may be achieved and documented. By forecasting planned changes to operational conditions or possible changes in annulus environments, future scenarios may be exploited efficiently by assuming that wave, wind, and current in the coming years, on average, will be similar to those of recent years.
The complete paper includes a detailed discussion of the criteria and process used to develop the fatigue counter.
- A reliable correlation can be established between environmental information and FPSO response from motion sensors, followed by significant improvements in analysis models.
- The same methodology may be used for improved structural, mooring, and riser analysis, as well as development of ML and analytics for armor-wire-fatigue prediction.
- A live fatigue counter presented in an online dashboard provides stakeholders with necessary information to support extended operation of the investigated components.
- The information can be used to improve fatigue-calculation models for all types of dynamic risers, enhancing confidence in life extension and enabling a cost-effective decision-making process to mitigate risk.
This article, written by JPT Technology Editor Judy Feder, contains highlights of paper OTC 29531, “Flexible-Riser Fatigue Counter Developed From Field Measurements and Machine-Learning Techniques,” by Christoffer Nilsen-Aas, Jan Muren, and Håvard Skjerve, 4Subsea, et al., prepared for the 2019 Offshore Technology Conference, Houston, 6–9 May. The paper has not been peer reviewed. Copyright 2019 Offshore Technology Conference. Reproduced by permission.