A Generation Expansion Planning model for integrating high shares of renewable energy: A Meta-Model Assisted Evolutionary Algorithm approach

This study presents a complementary model for Generation Expansion Planning (GEP). A GEP problem commonly determines optimal investment decisions in new power generation plants by minimizing total cost over a mid towards long planning horizon subjected by a set of constraints. The model aims to capture operational challenges arising when a transition towards higher shares of intermittent renewable generation is considered. It embeds a computationally expensive Operational Cost Simulation Model (OCSM), which may exhibit a high level of temporal and technical representation of the short-term operation of a power system to model the unit commitment. The emerging computationally expensive integer non-linear programming constrained optimization model is solved by a problem-customized Meta-model Assisted Evolutionary Algorithm (MAEA). The MAEA employs, off-line trained and on-line refined, approximation models to estimate the output of an OCSM to attain a near-optimal solution by utilizing a limited number of computationally expensive OCSM simulations. The approach is applied on an illustrative test case for a 15 year planning period considering the short-term operation of thermal, hydroelectric and storage units and generation from renewable energy sources. Moreover, the impact of technical resolution is examined through a simple comparative study. The results reveal the efficiency of the proposed problem-customized MAEA. Moreover, the trained approximation models exhibit a low relative error indicating that they may adequately approximate the true output of the OCSM. It is demonstrated that neglecting technical limitations of thermal units may underestimate the utilization of flexible units, i.e. thermal and non-thermal units, affecting the attained investment decisions.