考量再生能源結算政策下為電力市場建立系統響應模 型之研究

Abstract

隨著臺灣能源部門朝向永續發展轉型，如何在高度集中化的電力市場中有效整合再生能源，成為一項兼具技術與制度層面的挑戰。本研究建立一個統一的建模框架，以分析臺灣電力系統營運者（台電公司）與再生能源憑證（T-REC）經紀人(broker)之間的市場均衡互動。此模型將台電與經紀人的行為分別建構為成本最小化問題，並透過市場清算條件連結兩者，使再生能源的發電與採購相一致。

所得的均衡問題以變分不等式（Variational Inequality, VI）表達，並採用兩種互補的求解方法：一為使用 PATH 求解軟體的混合互補問題（Mixed Complementarity Problem, MCP）形式，另一為自行實作的正則化額外梯度法（Regularized Extra Gradient, R-EG）。模型以臺灣實際的電力系統資料進行校準，包括發電結構、電力需求與成本資訊。

本研究數值模擬分析不同負載與再生能源採購情境下系統的行為。結果顯示，儘管三種解法對發電決策具有一致性，但再生能源的採購價格常因互補條件的結構問題而趨近於零，與實際情況不符。敏感度分析亦指出離峰時段的解法差異性，並突顯建立更合理經紀人定價機制的必要性。 本研究貢獻於建立一套應用於臺灣能源市場的均衡模型，驗證了商業解法與一階數值法於政策模擬上的可行性，並指出未來可在市場真實性、時間解析度與多代理人設定上進行延伸與優化。

As Taiwan's energy sector transitions toward greater sustainability, integrating renewable energy into its centralized electricity market presents both technical and institutional challenges. This thesis develops a unified modeling framework to analyze the equilibrium interaction between Taiwan's Energy System Operator (Taipower Company) and a renewable energy certificate (T-REC) broker. The model captures operational and market dynamics by formulating both the Taipower Company (ISO) and broker problems as cost minimization programs, linked through a market-clearing constraint that ensures consistency between renewable generation and procurement.

The resulting equilibrium problem is expressed as a Variational Inequality Problem solved using two complementary approaches: a Mixed Complementarity Problem formulation solved using the PATH solver, and an implementation of the Regularized Extra Gradient (R-EG) method. The model is calibrated with real-world data from Taiwan's power system, including generation profiles, demand, and cost structures.

Numerical experiments simulate system behavior under different load scenarios and renewable procurement targets. Results show that while generation decisions remain stable across solvers, procurement prices for renewable energy often collapse to zero, an unrealistic outcome driven by the structure of the complementarity conditions. Sensitivity analysis reveals solver-dependent behaviors in off-peak periods and highlights the need for more realistic broker pricing mechanisms.

This thesis contributes with an equilibrium model for the Taiwanese energy market, demonstrates the applicability of commercial and first-order method solvers in policy contexts, and identifies future directions for improving market realism and time resolution in renewable energy procurement modeling.