結合電力系統模擬與生命週期評估之臺灣能源轉型路徑探討——以OSeMOSYS模型為例

Abstract

因應氣候變遷的嚴峻挑戰，實現電力部門能深度去碳化已成為全球淨零目標的重要策略。臺灣於2022 年電力部門之溫室氣體排放占全國的排放量的65%，顯示電力結構調整對我國2050 年淨零排放路徑具關鍵性的影響。過去能源規劃多著重於成本於排放效率分析，卻較少系統性的考量發電技術在全生命週期中所產生的環境衝擊。有鑒於此，本研究首次以開源能源系統工具OSeMOSYS (Open Source Energy Modeling System) 模型工具並整合生命週期評估 (Life Cycle Assessment, LCA) 方法，針對臺灣2018 至2050 年之電力系統進行模擬，評估不同情境下能源結構之轉變趨勢、環境衝擊與政策結果，並且以多目標分析環境與模型系統之取捨。 首先，本研究透過台灣電力公司 (Taipower) 2018–2023 年歷史發電數據模擬，確立OSeMOSYS 模型之準確性與調度邏輯，並且建立群組層級限制及再生能源優先調度、技術建置規劃等台灣的發電特徵。模型納入多種發電技術及未來情境（如技術進步、政策規劃）後，針對四種核心的情境模擬進行分析：以最小化成本為目標的「基本情境」 (Business as Usual, BAU)、以火力發展為主之情境、積極發展再生能源之情境以及以生命週期評估最小化環境衝擊為目標之情境。結果顯示，地熱發電為再生能源最具成本效益的選擇；若是以生命週期評估為目標優化，雖然能達成最低環境衝擊，但是系統成本負擔最大，主因在於大規模新建離岸風電。 本研究亦探討三種不同條件下的國家政策路徑。模擬結果指出，2030 年前透過燃氣取代燃煤之策略實現了顯著的溫室氣體減排及延緩再生能源建置之壓力、2030 年至2050 年需要加速再生能源的投資（高於預期裝設量1.3 倍），並於2050 年全面導入碳捕捉與封存（Carbon Capture And Storage, CCS）技術。透過多目標分析發現，雖然再生能源於營運間具備低碳排及空污之優勢，但若納入製造與建設階段之衝擊，其整體環境效益將因供應鏈與間歇性等問題被部分抵銷，導致燃氣發電於情境中仍具競爭性。 本研究的貢獻在於結合臺灣實際發電資料與生命週期評估，對臺灣能源轉型路徑進行了綜合性分析，考量了技術可行性、能源安全性、溫室氣體與空氣污染排放，並識別了不同發電技術的環境衝擊。這些結果對於未來優化臺灣電力結構模擬及綜合評估提供參考依據，亦對於LCA 與能源模型整合方法之理解與發展具有貢獻，為後續能源研究進行擴展及資料檢視。

In response to the pressing challenges of climate change, chieving deep decarbonization of the power sector has become a crucial strategy for global net-zero targets. Taiwan's power sector contributing 65% of the nation's total greenhouse gas emissions in 2022, underscoring the pivotal role of electricity structure transformation in the country’s 2050 net-zero pathway. While past energy planning has focused primarily on cost and emission efficiency analyses, the environmental impacts associated with different power generation technologies over their full life cycle have rarely been assessed in a systematic manner. This study is first to apply the Open Source Energy Modelling System (OSeMOSYS) integrated with Life Cycle Assessment (LCA) to Taiwan’s electricity system, simulating from 2018 to 2050. It evaluates trends in energy structure transitions, environmental impacts, and policy outcomes under different scenarios, and applies multi-objective analysis to assess trade-offs between environmental impacts and system performance. The study first validated the OSeMOSYS model using historical power generation data from Taipower (2018–2023) to verify its accuracy and applicability. It incorporates Taiwan-specific features such as group-level constraints, renewable energy prioritization, and technology deployment strategies. With various power generation technologies and future scenarios (e.g., echnological advancement, policy interventions) included, four core scenarios are analyzed: (1) a cost-minimizing Business-as-Usual (BAU) scenario, (2) a fossil-fuel dominant scenario, (3) a renewable energy dominant scenario, and (4) an LCA-optimized low-impact scenario. Results show that geothermal energy is the most cost-effective renewable option, yet even under aggressive deployment, it accounts for only 58.6% of electricity generation by 2050. The LCA-optimized scenario achieves the lowest environmental impact but incurs the highest system cost, mainly due to large-scale offshore wind development. Additionally, three national policy pathways under different assumptions are examined. The simulations indicate that substituting coal with natural gas before 2030 can lead to significant emission reductions and reduce short-term pressure on renewable deployment. However, from 2030 to 2050, accelerated renewable investment—1.3 times the anticipated capacity—is required, alongside the full adoption of Carbon Capture and Storage (CCS) by 2050. The multi-objective analysis reveals that while renewable energy offers low emissions and air pollution during operational stage, its overall environmental benefits are partially offset when upstream impacts from manufacturing, construction, and intermittency are considered, making natural gas remain competitive. This study contributes to the literature by combining empirical data from Taiwan’s power sector with life cycle-based environmental assessments. It provides a comprehensive analysis of Taiwan’s energy transition pathway by addressing technical feasibility, energy security, and emissions of both GHGs and air pollutants. The findings offer valuable insights for optimizing future electricity modeling frameworks in Taiwan and enhance the understanding of LCA-integrated energy system modeling methodologies for broader research applications.