Lagrangian decomposition for stochastic TIMES energy system optimization model

Abstract

Energy system optimization models play an essential role in current decision support on topics including energy security, sustainable development and environmental protection from industrial, regional, national and even global perspective. One of the key energy system optimization models applied in international energy policy is TIMES. The article establishes two basic deterministic TIMES models which cover an energy commodity (coal or gas), a three-step supply curve and an end-use energy service demand. Then we convert the deterministic TIMES models into a stochastic optimization problem with multiple scenarios, and implement the Lagrangian decomposition approach in solving the stochastic programming models. The numerical experiment shows the feasibility of the Lagrangian decomposition algorithm to solve stochastic TIMES models with a small amount of scenarios, and analyze several reasons for non-convergence cases including the choice of step length and initial values of Lagrangian multipliers.

Keywords

• the times model
• stochastic extension
• two-stage stochastic programs with recourse
• scenario decomposition
• lagrangian relaxation
• supergradient ascent method