How can a country minimize long-term power-system costs when future electricity demand is uncertain? Yuang Chen (PhD ISyE Georgia Tech and now assistant professor at the Chinese University of Hong Kong at Shenzhen), Beste Basciftci (PhD ISyE Georgia Tech and now assistant professor at the University of Iowa) and Georgia Tech’s Valerie M. Thomas (an Energy Policy & Innovation Center affiliate) tackle this by creating a planning model that lets planners meet most —but not necessarily every last kilowatt-hour—of demand in the early years, while still guaranteeing full coverage at the end of a 15 year energy system expansion plan.
Applied to a case study of Rwanda using real-world data, the model lowers the expected 15-year system-wide cost from US $3.42 B to $3.37 B (-1.46 %), and in the highest-demand scenario from $4.01 B to $3.82 B (-4.74 %). Investment and fuel expenses fall because planners can delay capacity that would otherwise hedge against low-probability peaks. Under an extreme set of scenarios (up to 20 % annual electricity demand growth), savings rise to -1.83 % system-wide and nearly -5 % in the top-demand scenario.
A key feature of the model developed by the authors is flexibility and allowance for the possibility that demand will exceed supply in some years. Allowing a modest shortfall (up to 20 % of demand going unmet in a few mid-decade years) captures most of the benefit; relaxing standards further gives only small extra gains.
Because the model presented by Valerie and her coauthors builds fewer diesel and peat units and more solar plus extra high-voltage lines, it cuts fuel use, keeps emissions on track, and protects the budget. The chance-based flexibility “helps reduce the overall expected cost and also mitigates the costs associated with unexpectedly high demand scenarios” without compromising the end electrification goal.
This summary was written with the assistance of Microsoft Copilot on July 1st, 2025. Its content was edited and verified by EPIcenter staff and affiliates.
Read the full paper: https://www.sciencedirect.com/science/article/pii/S0142061523005562?via%3Dihub