Methodology for technology evaluation under uncertainty and its application in advanced coal gasification processes

Abstract

Integrated gasification combined cycle (IGCC) technology has attracted interest as a cleaner alternative to conventional coal-fired power generation processes. While a number of pilot projects have been launched to experimentally test IGCC technologies, mathematical simulation remains a central part of the ongoing research efforts. A major challenge in modeling an IGCC power plant is the lack of real experience and reliable data. It is critical to properly understand the state of knowledge and evaluate the impact of uncertainty in every phase of the R&D process. A rigorous investigation of the effect of uncertainty on IGCC system requires accurate quantification of input uncertainty and efficient propagation of uncertainty through system models. This thesis proposes several uncertainty quantification methods which expand the sources of information that can be used for parameter estimation. Key features of these methods include the use of entropy maximization to translate subjective opinions to probability distribution functions, and a more flexible probability model that easily captures anomaly associated with small sample data. In addition, Bayesian estimation is extended to dynamic models. Aided by a computationally efficient algorithm, termed sequential Monte Carlo method, the Bayesian approach is shown to be an effective way to estimate time-variant parameters. Uncertainty propagation is performed using the deterministic equivalent modeling method (DEMM) which is based on polynomial chaos representation of random variables and probabilistic collocation algorithm. One major issue often overlooked in the analysis of IGCC models is to represent correlation in the input parameters. This thesis proposes the use of principal component analysis (PCA) to represent correlated random variables. The resulting formulation is the same as the truncated Karhunen-Lodve expansions. Explicit incorporation of correlation not only improves accuracy of the approximation but also reduces the overall computational time. A comprehensive study of the MIT-BP IGCC model is carried out to determine uncertainties of the key measures of performance and cost, including energy output, thermal efficiency, CO 2 emission, plant capital cost, and cost of electricity. Whenever possible, the probability distributions of input parameters are estimated based on realistic data. Experts' judgments are solicited if data acquisition is infeasible. Uncertainty analysis is conducted in a three-step approach. First, technology-related input parameters are taken into account to determine uncertainties of plant performance. Second, cost uncertainties are determined with only economic inputs in order to identify important economic parameters. Finally, the plant model is integrated with cost model and they are evaluated with the key technical and economic inputs identified in the previous steps. Our study indicates the property of coal feed has a substantial impact on the energy production of the IGCC plant, and subsequently on the cost of electricity. Immature technologies such as gasification and gas turbine have important bearing on model performance hence need to be addressed in future research.