基於可靠度設計最佳化之離岸風場運維策略評估

Abstract

隨著全球可再生能源的需求迅速成長，離岸風力發電被視為相當具有前景的能源產生來源之一。然而，離岸風力發電場相對一般的陸上風場面臨更多挑戰與不確定性，這使得有效的運維（O&M）策略對於提升其發電效率和成本效益變得至關重要。本研究針對這些挑戰，開發符合台灣環境條件的運維（O&M）模型，並進一步尋找最佳解決策略：在最小化運維（O&M）成本情況下，同時確保充足的電力供給，本研究中以發電量不足天數計算。 利用台灣的海能離岸風力發電場（Formosa 2 OWF）作為案例研究，本研究分別蒐集海象觀測站與中央氣象局資料，整合台灣的風場特性和極端颱風事件，開發原型模型（Prototype Model）與進階模型（Proposed Model）。運維行動分為三種：失效更換、預防性更換以及預防性維修。設計變數共有三個：兩愈值及運維區間。其中設定兩閾值決定預防性維護進行之時機，運維區間則決定多久派遣一次海上運維團隊。原型模型主要參考既有文獻框架，以元件隨時間依照特定機率分布退化結合台灣氣候條件特性；而進階模型基於原型模型之上，結合一種新的元件使用壽命更新工具，稱為貝氏韋伯第一階可靠度方法（Bayesian Weibull First Order Reliability Method, BW-FORM）。進階模型結合了原型模型與BW-FORM，將一般元件僅依特定機率分布之退化，改為經偵測當下元件健康狀態後回傳數值、更新元件使用壽命的工具，以提供更即時和準確的元件健康狀態評估。針對原型和進階模型的表現在指定策略和情境下進行評估，並對差異進行了詳盡的比較和分析。 確認進階模型表現後，本研究運用基於可靠度設計最佳化（Reliability-Based Design Optimization, RBDO）來尋找最佳化的運維（O&M）策略。基於可靠度設計最佳化不同於一般的定然性最佳化在於，限制條件考慮了失效機率的概念，而非定值。故本RBDO結合蒙特卡羅模擬計算失效機率、以內點法進行最佳化搜索，找到了一些最佳解決策略。由結果可發現，這些最佳解決策略並非收斂於單一數值，而是分布在一特定範圍內：其在滿足限制條件，也就是限制發電量不足天數的前提下，產出最小化目標值。此方法能有效幫助研究在最小化運維成本和確保電力供給滿足需求之間取得平衡。最佳化結果在進行驗證過程中，同時也考慮了風速帶來的不確定性。

With the rapid expansion of renewable energy all around the world, offshore wind power has become one of the most promising source of power generation. However, offshore wind farms encounter unique challenges and uncertainties, making the development of effective Operation and Maintenance (O&M) strategies crucial for improving their efficiency and cost-effectiveness. This study addresses these challenges by developing an O&M model to match the specific environmental conditions of Taiwan, and further finding the optimal solution to minimize the O&M costs while ensuring reliable electric power generation. Using Taiwan's Formosa 2 OWF as a case study, this research integrates local wind characteristics and extreme typhoon events to develop both a prototype and an advanced proposed model. Three types of maintenance actions are considered, which are failure replacement, preventive replacement, and preventive repair. The decision variables include three elements: maximum threshold Amax, minimum threshold Amin and dispatch interval of the O&M team Tint. Amax and Amin are regarded as the criterion that influence the type of maintenance actions; and the dispatch interval determines how frequently the O&M team is deployed, impacting the overall maintenance strategy and effectiveness. The proposed model applied a novel technique for updating the useful life (UL) of components, known as the Bayesian Weibull First Order Reliability Method (BW-FORM). This advanced model combines probabilistic failure degradation with BW-FORM to provide a more immediate and accurate assessment of component health. The performance of both the prototype and the proposed model is evaluated across various designated strategies and scenarios, with a thorough comparison and analysis of the differences. Further, the study employs Reliability-Based Design Optimization (RBDO) to formulate an optimized O&M strategy. By utilizing Monte Carlo simulation to compute the failure probability, and the interior point method for optimization, some optimal solutions are found. The solutions identified do not converge to a single value but rather fall within a certain range, while enable to balance cost minimization with the requirement for reliable power generation. Validation and uncertainty from wind speed are accounted as well.