A Numerical Method for Optimizing Oxyfuel Power Plants

1.Introduction

Choosing between near-term oxyfuel combustion technologies and novel next-generation technologies is difficult and requires a means of making comparisons across vastly different technologies. This paper proposes a computational methodology to evaluate different technologies and to identify optimal technologies based on a user supplied set of evaluation criteria. The tool makes it possible to compare different modules of oxyfuel power plants, different oxyfuel combustion power plant designs, and different technology pathways leading from today's power plants to the power plants of the future.

2.Different oxyfuel combustion technologies

Oxyfuel combustion technologies involve individual modules or components (for example air separation units, oxyfuel boilers, flue gas processing units, etc). The fossil fuel, usually coal, is combusted in pure oxygen that may be diluted with combustion products. The output is a nearly pure stream of carbon dioxide (Croiset, 2000). Oxyfuel combustion technologies are among the most promising low-emission power plant generation technologies.

The need for a fundamental redesign of today's power plant is driven by the rapidly growing consensus that excess carbon dioxide will cause a significant change in climate and will have repercussions on a wide variety of human activities. It is also driven by the prospect of increasingly stringent environmental regulations, which will eventually demand zero emission power plants that do not emit any gaseous effluent to the atmosphere (Lackner 2001). Oxyfuel combustion technologies, among all low-emission power generation technologies, offer a unique opportunity for improved environmental control performance. The new combustion environment without nitrogen makes it possible to design power plants where the entire off-gas stream is disposed of and kept out of the atmosphere.

A spectrum of competing oxyfuel combustion power plant designs already exists. On one end of the spectrum is a class of plant designs that simply modify existing pulverized coal plants. Such designs are suitable for a retrofit and they consist of adding a conventional air separation unit for producing oxygen and sealing the boiler system so as to avoid the accidental introduction of nitrogen rich outside air (Wilkinson et al., 2003). On the other end of the spectrum are plants that are based on entirely novel technologies. This includes new oxygen production methods, for example by separation with high temperature ion transport membranes (Sinner and Kilner 2003; Bouwmeester and Van Laar, 2002; Dyer et al,. 2000; Bredesen et al., 2004). One might even consider as advanced oxyfuel designs chemical looping systems that oxidize a metal in contact with air and carry the oxygen to the fuel bound to the metal (Zafar et al., 2005).

Choosing between different oxyfuel combustion technologies has a profound impact not only on the cost-efficiencies of an individual power plant, but more importantly on the pathway that will lead from today's power plant technology to future technologies. In order to choose from a wide range of technologies, and more importantly from different technological pathways, one needs to make comparisons between different plants and different pathways.

This paper aims to create conceptual plant designs and conduct engineering assessments of the component modules of an oxyfuel power plant; evaluate different power plant designs under various energetic, economic, environmental and infrastructural constraints; and perform optimization not just of the individual power plants, but also of pathways connecting current power generation technologies to future technologies.

3.Novel evaluation criteria based on a penalty model

Power plant development leading to a zero emission plant design could move through a set of new plants, each designed to the best available knowledge at the time and with little regard to the long term goal, or to the basic knowledge that is embedded in previous designs. In such a strategy one may introduce technologies even though it is clear from the outset that they do not lend themselves to further advances. For example, post combustion technology, may well be in this category. Any R&D investment into flue gas scrubbing is most likely made obsolete by the next generation of power plants. Alternatively, the goal could be achieved by a set of incremental improvements that are introduced in each new plant or in each upgraded plant, where changes are designed to build upon each other. Oxyfuel combustion designs are likely to fit into this category.

A consideration of the intermediate plant designs can reduce the long-term cost of power plant designs. However, a rational implementation of such an approach requires the means of making comparisons across different technologies and across different times. We propose a methodology by which we can make such an assessment. The method introduces a penalty function that can be applied to modules, plants, and sequences of plant designs. In optimizing the design, one varies design parameters so as to minimize the penalty function. The penalty function is zero for some perfect state of the system which is typically not attainable, and the penalty function is optimized by varying all the available design parameters. The penalty can be thought of as a sum of penalties for specific aspects of the plant, for example its efficiency, its cost or its environmental impact. Individual modules may have component penalties. Some aspect of the penalty will depend on properties that can only be defined for the entire plant, or even for a sequence of plants.

The relative weights of these penalties can be chosen appropriately by a user, who has specific goals. For examples, penalties one may associate with having to build a new plant on a new site may vary for users in different countries. For example, building new plants in China is likely to introduce a relatively small penalty for greenfield plants. The same decision in the West is likely to introduce a much bigger weight, because the political difficulties of opening up new sites are much larger. The different weights may result in alternative development pathways.

It is also recognized that the availability and maturity of novel technologies, as well as environmental thresholds for existing and potential criteria pollutants are likely to change over time, thus the weights for these penalties should not be considered as static. The dynamic nature of these penalty specifications, allow users to choose, instead of the best possible plant at a specific time ti, the best possible pathway connecting technologies from time t1 to time tn. Pursuing the optimal pathways on the basis of the minimum total pathway penalties helps users lower the cost of achieving the specific goals, even if it results in seemingly sub-optimal outcomes for individual plants.

Much of the work we present is in defining an appropriate set of penalties on which the optimization rests. The underlying algorithms have been studied for other applications. The approach outlined here has taken its inspiration from the typesetting software TeX, in which Donald Knuth demonstrated the power of these penalty based optimization algorithms for trading off between very different and very difficult to quantify properties of text. We will show how very similar algorithms can be used to select appropriate modules, and power plants to arrive at a sequence of plant designs that provides an advantageous technology pathway from today's power plant designs to a future design that has far higher efficiency, avoids all emissions to the air, and provides the CO2 produced in a concentrated stream ready for disposal.