With deregulation in the electric power industry, traditional approaches for minimizing production costs have become unfit for the present competitive environment. Owners of generation assets must now consider price uncertainty in solving unit commitment problems for scheduling and operating their power plants. Operation flexibility of the generating assets, such as fuel switching and overfire, becomes an important issue. Because in a competitive market with volatile electricity prices, these flexibility may add significant values. On the other hand, operational constraints, such as ramp and minimum uptime/downtime constraints, present physical limits for the generating assets to flexibly react to rapid price changes, which have a negative effect on the asset value. Both of the operational flexibility and operational constraints must be considered simultaneously so as to achieve optimal operation under uncertainty. This dissertation devotes to this very important subject.

Deregulation in the power industry allows new firms to freely enter the generation markets. As a result, capacity expansion is no longer the responsibility of local utility companies and has become a pure investment problem. Overestimating the value of a power may result in stranded capital for a long time period. Therefore, to ensure a successful investment a fair valuation method is essential. The generation asset valuation must fully account for market uncertainty, which results in not only risks but also opportunities. To minimize the risks, one must first have sound models for market uncertainties. In this research, we consider not only the uncertainties of electricity price and fuel price, but also environment temperature because some characteristics of power plants may be sensitive to the temperature. To fully capitalize on profitable opportunities arising in the marketplace due to price spreads of different commodities, such as fuel and electricity, a real options approach is considered, in which different options are exercised at different but `optimal' timings.

Overall, this research is expected to contribute a new methodology for fair generation valuation that accounts for multiple and interdependent uncertainties and complex physical constraints. The proposed approach can help operators achieving optimal operation and investors making appropriate investment decisions. In the long run, customers also benefit from the improved societal efficiency.