Abstract:
The
present study aims at setting the basis for a assessment method in the three
principal areas of a power plant, that is its design and off design performance, operations
and maintenance
strategy and also economics decisions related to profitable operational
regimes and project investment.
The
principles of what is referred to as the novel technology of the two, i.e. the Solid
Oxide Fuel Cell are
presented, and a series of scenarios are examined from a technical
but also a degradation point of view, seeking to identify the relevance of the deterioration
mechanisms while at the same time
pointing out which components within the prime
mover
require more attention. In addition a presentation to the various methods that
faults in
prime movers can be identified is given leading to the isolation of Gas Path
Analysis as a good Engine Health Monitoring technique candidate. The work culminates
with the economic
analysis of a number of configurations developing a method to
calculate revenues and
applying an appraisal technique utilising a set of financial indices
that seek to evaluate the configurations on the basis of profitability.
Results showed than Fuel Cell Technology presents attractive characteristics on part load
operation, with average efficiency around 50%, an area proven to be weak for the
turbine. Further, due to the modular characteristics they can offer alternative modes of
operation like that of reduced stack operation. On a component level the anode electrode
seems to be the most
important one in the cell with the cathode following and the
electrolyte being the least important of the three. Finally with respect to the economics it
was
proven that cells couldn't compete as their production costs are still well above that
of Turbines with
prices as high as two times that of turbines, resulting in risky,
unattractive investments. With reference to the application of GPA on cells, what can be
considered as an initial
attempt is characterised as successful with a number of
approaches having been appraised and faults being detect with respectable accuracy.
Linear Root Mean
Square Errors were not far away from the critical value of 2 while
convergence for Non-Linear approaches was achieved with a small (~ 17) number of
iterations.