A statistical approach to model simplification and multiobjective analysis

Evolving national energy supply/demand distribution systems rely, at least in part, on quantifiable factors such as local and national environmental restrictions, resource availability (type, price, and quantity) and the associated transportation infrastructure, the amount and price of capital available to consumers and suppliers of energy, total annualized system cost, including the annualized cost of end-use devices, and the demands for energy and their price/supply responsiveness. The evolution also depends on nonquantifiable factors such as personal, regionally aggregated, or even national “utility functions” and institutional or social barriers. Many models have been formulated which attempt to simulate these complex interactions.This paper describes a systematic statistical methodology for capturing, both visually and quantitatively, the trade-offs between competing quantifiable, differentiable objective functions in a model of the national energy system (Brookhaven Energy System Optimization Model). The aim is to provide decision makers with a more easily understood tool and a more easily defensible methodology on which trade-offs between certain sensitive and competing energy issues can be based. The methodology has the additional advantage of providing insights into the inherent structural relationships of the model (model simplification).