Reduction of fatigue computational time for offshore wind turbine jacket foundations

Abstract

Recognizing climate change as a result of more than 150 years of industrial activity, and due to the emission reduction protocols set internationally, wind energy has become the mainstream source of energy for many developed nations. Due to the prospect of large scale energy generation, wind energy has moved offshore. Offshore wind farms are relatively more complex to design and costlier to install. Since, substructure design of an offshore wind turbine (OWT) plays an important role and is one of the key drivers for capital cost, optimization of support structure design becomes imperative.
Driven by the universal goal of cost saving in offshore wind industry, the main objective is set to – “reducing computational time required for a typical fatigue design cycle of OWT support structure”. From the perspective of a foundation designer, following practical challenges are identified and investigated:
• Lumping methodology for correlated wind and wave scatter diagrams
• Reduction in number of fatigue design load cases in time domain simulations
• An alternative fatigue design method in frequency domain
This thesis focuses on the fatigue design of jacket type support structure. First, a conceptual 3-legged jacket with a light weight pyramid shaped transition piece is modeled in finite element software. Subsequently, an overview of loads, dynamic properties and fatigue analysis of support structure is presented.
In OWT support structure design, wind and wave loads are equally significant. For fatigue design, the combinations of loads which are likely to occur simultaneously during the design life of the structure are estimated. The correlation established between wind and wave is generally presented in the form of scatter diagrams. To reduce the computational time, common practice is to condense wind and wave scatter diagram and assess fatigue damage for the lumped states. Thus, in this thesis, an investigation is completed on commonly used lumping methodologies and comparison is established on the accuracy of fatigue results. In this study, existing lumping methods are validated for a jacket support structure and it is proven that dynamic characteristic of the structure plays an important role in condensing scatter diagrams.
Traditionally, the complete fatigue load case design of OWT support structure is based on time domain simulations, which is very time consuming. Hence, an important query raised by the Company (KCI) was, whether it is possible to reduce the number of design load cases while retaining a high level of precision in fatigue damage results. A novel method is presented in this thesis, which attempts to reduce the number of design load cases. For jacket support structure, linear-quasi-static assumptions are made to estimate approximate fatigue damages due to wind load alone. Using the estimated damage, critical wind seeds are identified and are sorted for reduction. Results from this case study show that with the reduction of load cases, the accuracy in fatigue damage is also compromised. This experimental method provides good insight into the load case reduction capabilities and is recommended for preliminary design phase.
Since, time domain fatigue design requires very large time records to accurately describe the random loading processes, it proves prohibitive for many finite element analyses. Thus, an alternate fatigue design approach in frequency domain is introduced in this thesis. Contrary to popular use of power spectral density functions, a different method is presented which successfully preserves the phase information of load time series. Frequency domain method offers many advantages like providing clear information regarding the environmental conditions and the dynamic properties of the structure. First, essentials of frequency domain method for a single degree of freedom system is described. Subsequently, this approach is exemplified on OWT support structure with interface wind loads. Based on the assessment of response from time domain and frequency domain analysis, it is proved that frequency domain method is an effective tool which can lead to vast savings in computational times while still producing accurate fatigue results.
To summarize, this report gives an overview of the work done to reduce the computational time and effort required for a typical fatigue design cycle of an OWT support structure. Three measures were presented which helps with the reduction of computational time and the accuracy of end results were checked with full time domain simulations. As a result, one can perform more informed design optimization leading to reduction in costs for the support structure.