Investigation on the sensitivity of flexible foundation models of an offshore wind turbine under earthquake loadings

Highlights

• We developed a novel seismic analysis framework (SAF) for offshore wind turbines.

• We used target spectrum matching technique to adjust the ground motions.

• We examined three distinct flexible foundation concepts.

Abstract

This paper presents an investigation on the sensitivity of flexible foundation models of offshore wind turbines subjected to earthquake loadings. A novel seismic analysis framework (SAF) is developed and implemented in an open source aero-hydro-elastic analysis tool, “FAST”, for accurately modelling the effects of seismic loadings on offshore wind turbines. SAF has been validated through comparisons against experimentally validated numerical tools, GH Bladed and NREL Seismic. The behaviours of three flexible foundation models, namely, the apparent fixity (AF), coupled springs (CS) and distributed springs (DS) methods, subjected to earthquake loadings have been examined in relation to a fixed foundation. A total of 224 fully coupled nonlinear simulations of the foundation models are performed using a dataset of 28 earthquake records which are scaled using the target spectrum matching technique to represent the actual seismic effects of the selected sites. The results reveal that the AF model appropriately reflects realistic situations in comparison to the CS model. In addition, the amplitudes of vibration induced by the earthquake loadings are larger for flexible foundations compared to the rigid foundation. The main contribution to the out-of-plane bending moment of the support structure at the mudline comes from the wind loading for all the foundation models. This study has also found that the 2nd flap mode of blade is activated by the earthquake loadings for the AF and DS models but not for the rigid and CS models. As a result, the peak blade-root bending moment is found to be more sensitive to pseudo spectral acceleration (PSA) for the AF and DS models. Furthermore, the peak tower-top displacement and mudline bending moment increase linearly with PSA for all the examined models. This study contributes to the evaluation of the wind turbine responses subjected to earthquakes or combined multi-hazard loadings in the operational state.

Introduction

Wind energy is currently playing a leading role in the global production of cleaner energy as an alternative to fossil and non-cleaner fuels. The 2018 Global Wind Energy Council (GWEC) annual wind report states that 52 GW of newly installed wind capacity was added globally in 2017, and with 50% of the figure shared by China and the USA [1]. The southeast coastal areas of China and the west coast of the USA, located close to the Pacific seismic belts, are prone to earthquake. Wind turbines installed in these areas are susceptible to damage from the resulting earthquake loading coupled with the local wind loading. Similar circumstances exist for the wind farms located along the southern areas of Europe and New Zealand where there are rich offshore wind resources. Therefore, it is imperative to investigate the impacts of earthquake loading on wind turbines due to potential consequences on operation and supply of wind energy in these locations.

Environmental loads acting on wind turbines along with earthquake loadings have a significant influence on the accuracy of the seismic analysis of wind turbines. Dynamic behaviours of wind turbines under earthquake excitations have been studied over the past decades but with simplifications on the model geometries [2], [3], [4], [5], [6], [7], [8], [9]. In these studies, the rotor and nacelle were either completely ignored or simplified as a lumped mass. The unsteady wind loads are often treated as a rotor thrust, leading to inaccuracies in the prediction of aerodynamic loads acting on the blades. Generally, the aerodynamic loads increase exponentially with the rotor diameter for large-scale wind turbines. The resulting aerodynamic effects have been determined to be unneglectable from a comparative study on operational and parked states [10]. Therefore, over-simplification of aerodynamic loads is never precise, thereby undermining the accuracy of results in the seismic analysis of large-scale wind turbines. Therefore, in the seismic analysis of large-scale wind turbines, it is necessary to correctly take into account the coupled effect of wind and earthquake loadings.

One of the efficient approaches of improving the accuracy of coupled earthquake and wind loadings for wind turbines is by integrating an additional seismic module into an aeroelastic analysis tool. An early study on the coupled behaviour of earthquake and wind loadings was conducted by Witcher for a 2 MW wind turbine [11]. With the use of GH Bladed, Santangelo et al. [12] investigated the difference between the results from fully coupled and uncoupled time-domain simulations for a 5 MW wind turbine under the combined excitations of wind and earthquakes. Using FAST as a design basis, Asareh and Prowell [13], [14] developed a seismic module in order to examine the coupled effect of wind and earthquake. In the seismic module, the calculation of the earthquake loading is based on a specific ground motion, and the stiffness and damping properties of a damped actuator are located at the tower-base. Asareh et al. [15] used the improved FAST (also called NREL Seismic) to investigate the relationships between earthquake intensity and structural responses. Jin et al. [16] also used the NREL Seismic tool to predict the dynamic responses of a wind turbine under multiple hazards associated with earthquake and turbulent wind. Similarly, Yang et al. [8] proposed a numerical analysis framework coupled with FAST in order to obtain seismic responses of wind turbines. It is noted that the method of earthquake analysis proposed by Asareh and Prowell [13], [14] is different from the one applied to seismic analysis of buildings. The accuracy of predictions is significantly influenced by the stiffness and damping properties. The selection of the values of stiffness and damping depends on the experience of the involved analytical engineers.

However, it is noted that most of the aforementioned literatures focused heavily on earthquake effects for land-based wind turbines, which are significantly different from the offshore types. Since a large number of newly installed offshore wind turbines are located in earthquake-prone areas, it is necessary to investigate the seismic behaviour of offshore wind turbines in order to mitigate potential consequences of damage caused by earthquakes.

Offshore wind turbines have slender support structures resulting in large vibration amplitudes at the tower-top. In addition, the nature of the soils in the offshore environment often leads to more severe structural responses. The offshore soil is composed of detrital materials and sediments, implying that the wind turbine foundation is installed in a layer of less dense and less stiff soil [17], [18]. The soft soil condition is often associated with liquefaction in earthquake-prone offshore environments. This may affect the integrity or the serviceability of the foundation during its operational lifespan. As discussed by Wang et al. [19], the liquefaction is more easily caused by earthquakes leading to severe damage to the wind turbine under soft soil conditions. Some common foundation problems resulting from soil liquefaction include operational difficulty and loss of stability of the wind turbine. In addition, the cost of the foundation is approximately 30% of the total cost of a bottom-fixed offshore wind turbine and could reach up to 35% for the wind turbines installed in water depth of 30–40 m [20]. Hence, the deign of offshore wind turbine foundation subjected to earthquakes needs to be carefully handled due to its impact on the overall cost of wind turbines and the levelised cost of electricity (LCOE).

The soil structure interaction (SSI) model plays a key role in the design of a foundation as can be seen in Fig. 1. The accuracy of the results from foundation design analyses, including eigen-analysis and ultimate state analysis, is significantly influenced by the selection of the SSI model. This means that the selection of a SSI model determines the reliability of the foundation which costs over 30% of the whole wind turbine. Hence, the sensitivity analysis of SSI models is beneficial to the wind turbine industry for practical cost-reduction reasons when selecting the appropriate foundation concept during the design stage.

SSI can be modelled using three methods: apparent fixity (AF), coupled springs (CS) and distributed springs (DS) [21]. The CS model is the most widely used method in the dynamic analysis of offshore wind turbines and it is applicable to any type of offshore foundations due to its ease in obtaining results using typical theories [22], [23], [24], [25], [26], [27]. For the CS model, the foundation is modelled using a set of translational and rotational springs placed at the bottom of the structure to represent the SSI effect. Bhattacharya et al. [28], [29] investigated the SSI of a monopile wind turbine under different soil conditions by using scaled experiments and numerical analysis. It was found that the numerical models had first natural frequencies similar to those of the test models in most soil conditions including clay. In some foundation cases with saturated sands, however, over 20% discrepancies were observed between the numerical and experimental results. In another study conducted by Bhattacharya et al. [30], it was found that the stiffness of lateral springs could be reduced under cyclic loadings, which is a major contributor to fatigue. The study found that 30% change in the first natural frequency of the wind turbine system occurred after 10,000 cycles. This suggests that there is a limitation on the use of the CS model for SSI modelling of the dense soil condition.

The DS model is another widely used foundation modelling method for SSI [31], [32], [33], [34], [35], [36]. In this method, the SSI is represented by a set of lateral and vertical springs distributed along the embedded pile (usually, only the lateral springs are considered). The stiffness of the springs is obtained in accordance with p-y curves at different depths. Compared to the CS model, the DS model has an advantage that the responses of the embedded portion of the foundation can be investigated more specifically. The DS model is well suited for modelling pile in a multi-layered soil condition while the CS model could only model an overall effect of SSI at the seabed level.

The AF method is another modelling option which is an alternative to the CS and DS models. In this method, a fictive length is assumed to connect the bottom of the support structure and the foundation soils. The support structure is fixed and has the same mudline lateral deformation and rotation as the CS and DS models under external excitations. This approach is much easier to implement in any multi-body analysis tool for accounting the SSI effect. Damgaard et al. [37] investigated the dynamic responses of a monopile offshore wind turbine by considering the effect of SSI modelled using AF and CS models. The AF model results in similar fatigue damage compared to the CS model for two distinct types of soil. From the preceding literatures, it has been noted that the SSI effect for offshore wind turbines has been examined under multiple loadings with the exception of earthquakes.

Santangelo et al. [38] compared the structural responses of coupled and uncoupled time-domain simulations for an offshore wind turbine under earthquake loadings. Kim et al. [39] investigated the seismic fragility of a monopile offshore wind turbine by considering the SSI effect. They modelled the flexible foundation using a set of lateral springs distributed along the length of the support structure underneath the seabed. The stiffness of each spring at a corresponding depth was represented by a p-y curve. Mo et al. [40] also performed a seismic fragility analysis of an offshore wind turbine under different operating states by considering the effect of SSI. Wind loads were calculated using FAST and then applied to the FEM model for coupling with earthquake loadings in OpenSees. The probability of reaching damage states was discussed for different wind conditions and earthquake loadings. Alati et al. [41] studied the seismic responses of two bottom-fixed offshore wind turbines using GH Bladed in which the SSI model was represented by two transitional springs.

However, there are still some notable limitations in the above-mentioned literatures. First, the dynamic characteristic in the frequency domain of offshore wind turbines under multi-loadings, which is important in the control and mitigation of vibration induced by an earthquake [42], [43], has not been addressed. Secondly, although it has been widely accepted that the DS model offers the best approach for representing realistic foundation conditions, the difference between the three SSI modelling approaches (i.e. the AF, CS and DS) for seismic analysis of offshore wind turbines has not been thoroughly investigated. For offshore wind turbines located in earthquake-prone areas, the support structure suffers from high frequency and strong underlying excitations. This means that an investigation of the sensitivity of flexible foundation models becomes imperative in order to perform accurate seismic analysis for a reliable foundation design.

The purpose of this study, therefore, is to investigate the sensitivity of foundation models of offshore wind turbines under multi-hazards by including earthquake, wind and wave loadings. The structural responses of the wind turbine with distinct foundation models will be examined in both time domain and frequency domain. For this purpose, a seismic analysis framework (SAF) is developed to take into account the influences of earthquake loading and foundation flexibility by extending the capability of the FAST source code. One of the benefits of using SAF is that it is generic and can be applied to different types of wind turbine models compared with the NREL Seismic tool presented in [14], [15]. In addition, SAF offers capabilities for different SSI models to be examined as opposed to other tools that exclusively focus on the rigid foundation concept.