Least-squares polynomial quasi-Monte Carlo for short-term generation unit asset valuation

This paper presents an effective and efficient financial-structural model for short-term generating unit asset valuation. A novel forward-moving quasi-Monte Carlo (QMC) simulation technique in conjunction with backward-moving dynamic programming is introduced. At any time, when commitment decision can be made, it is relied upon the present realization of electricity and fuel prices. The decision whether to turn the unit “on” or “off” is equivalent in term of expected profit, when the electricity and fuel prices are on the “locus” or known as indifference locus (IL). The identification of the ILs is a crucial step in unit asset valuation method. A least-squares polynomial based technique to approximate the continuous path of ILs was used to further improve the simulation speed without compromising its accuracy. The quasi-Monte Carlo coupled with least-squares polynomial for constructing ILs represent a new approach to solve unit asset valuation problem. Testing and model verification were performed. They indicate that the proposed technique is significantly faster while producing accurate results. It reduces the computational burden by more than 3,000 times while results are of a 0.13% error level.