A1704
Title: Application of receding horizon optimal control to DICE integrated assessment models
Authors: Timm Faulwasser - Karlsruhe Institute of Technology (Germany) [presenting]
Lars Gruene - University of Bayreuth (Germany)
Christopher M Kellett - University of Newcastle (Australia)
Steven Weller - University of Newcastle (Australia)
Abstract: To quantify the damages from anthropogenic emissions of heat-trapping greenhouse gases, specifically carbon dioxide (CO2), integrated assessment models are used to describe the dynamics of climate-economy interactions. We consider the computation of the Social Cost of Carbon (SCC) via the Dynamic Integrated model of Climate and the Economy (DICE). Typically, any SCC computation is based on the solution of a single (long-horizon) optimal control problem. We show that receding-horizon strategies can also be used to compute the SCC. In receding-horizon optimal control, also known as model predictive control in the field of systems and control, one solves a sequence of short-horizon optimal control problems, of which only the first part of the optimal solution is used, instead of tackling the long-horizon optimal control problem directly. We demonstrate that tools developed in a systems and control context can be used to analyze the receding-horizon approximation of the SCC. Furthermore, we show that the receding-horizon strategy facilitates feedback, introducing an element of robustness. Additionally, we comment on different strategies of computing the SCC and a recently published open-source MATLAB implementation of DICE.
Journal