Multi-Objective Optimization Framework for Generation Expansion Planning Under Transmission Constraints

South Africa faces a severe electricity supply crisis driven by ageing coal infrastructure, declining plant reliability, and persistent generation shortfalls. While initiatives such as the Renewable Energy Independent Power Producer Procurement Programme have expanded renewable capacity, transmission bottlenecks and limited grid upgrades still hinder large-scale integration of variable renewable energy. At the regional level, the Western Cape Province remains heavily dependent on electricity imports from the national grid, making it particularly vulnerable to disruptions in external supply. Transmission congestion further limits the integration of local renewable generation. To address these challenges, the province aims to reduce its dependence on imported electricity by 5,700 MW by 2035 through increased local generation and storage capacity. Achieving this goal requires a generation expansion planning framework that optimizes the expansion of local generation capacity while minimizing total system costs and energy imports, subject to transmission constraints and limited grid upgrades. This study addresses the challenge by developing a multi-objective optimization framework for generation expansion planning. The framework is specifically designed for networks operating under limited grid upgrades. The optimization objectives are total system cost and energy imports, with total system cost minimized and energy imports minimized. The framework is developed by integrating the Python for Power System Analysis modeling tool with the Non-Dominated Sorting Genetic Algorithm II for multi-objective optimization. The developed framework is applied to the Western Cape electricity network as a case study to evaluate its performance and practicality in a real-world regional context. The PyPSA tool models generation, storage, and transmission assets, while k-means clustering is used to reduce temporal complexity and improve computational efficiency. The multi-objective optimization is executed using the Non-Dominated Sorting Genetic Algorithm II, where candidate system designs are iteratively refined through evolutionary operations until they converge to a Pareto set of optimal solutions. Scenario-based simulations are conducted to validate the framework, demonstrating its capability to generate technically feasible and computationally efficient capacity expansion plans. The successful validation indicates that the developed framework can be confidently applied to other power networks operating under similar transmission constraints and limited grid-upgrade conditions, supporting sustainable and regionally optimized energy planning beyond the Western Cape.

Grantor dc:publisher

Stellenbosch : Stellenbosch University

Year dc:date.issued

2026

Author dc:creator

• Ruzive, Grace

Advisors dc:contributor.advisor

• Garner, Karen S.
• Van Staden, Chantelle Y.

Language dc:language.iso

en

Repository record dc:identifier.uri

https://scholar.sun.ac.za/handle/10019.1/135887

OAI identifier oai:identifier

oai:scholar.sun.ac.za:10019.1/135887