Implementing the Oracle-Based Optimisation for an Integrated Assessment Model for Air Pollution: the Luxembourg Energy Air Quality model

In this paper, we present the implementation of an advanced convex optimisation method in the framework of integrated assessment modelling. The integrated assessment of optimal policies requires a flexible and modular optimisation framework to exchange information between models while providing a consistent response in a reasonable time. In the Luxembourg Energy Air Quality model, the NOx and VOC emissions are computed by an energy model and the dispersion of air pollutants, including produced ozone, are simulated by an air quality model. Here we implement an oraclebased optimisation technique to couple the energy model and the air quality model in cost-effectiveness mode. Each model is queried by a program, the so-called “oracle”, which supplies the model output and sensitivity information to a main program. The convergence of procedure is ensured given sufficient assumptions on the convexity by the cutting plane method Proximal-ACCPM. Encouraging results show the capability of the method to handle the non-convex behaviour of the ozone production, but also to find an optimal solution within a reasonable timeframe of a few hours.