**Pierluigi Leone**
*Dipartimento di Energetica. Politecnico di Torino Corso Duca degli Abruzzi 24 Torino, Italy*

**Massimo Santarelli**
*Politecnico di Torino, Dipartimento di Energetica Corso Duca degli Abruzzi 24 Torino, Italy*

### Abstract

The aim of this work is the description of a computer
experimental session, made on a simulation model of a tubular
SOFC CHP plant, whose aim has been the analysis (through a
statistical methodology: 2^{K} factorial experiments) of the effect
of the main operation variables on the electric and thermal
power provided by the stack, and on a particular performance
parameter, the steam-to-carbon ratio (S/C). The computer
experimental analysis has the aim to supply information
preliminary to the correct design of the forthcoming real
experimental tests on the generator, in order to outline
particular point of interest and operation variables dependence.
To perform the computer experimental analysis, a mathematical
model has been developed that simulates a natural gas fuelled
SOFC CHP system that generates and supplies net AC power to
the external grid and recovers heat from the SOFC exhaust to
provide hot water. The model has been calibrated for the CHP
100 kW_{e} Siemens tested for 20,400 hours in Holland by
EDB/ELSAM and in Germany by RWE. The statistical
methodology used for the computer experimental analysis is the
factorial design (Yates' Technique): using the *Anova* technique
the effects of the main independent variables (air utilisation
factor U_{ox}, fuel utilisation factor U_{F}, internal fuel and air
preheating and anodic recycling flow rate) have been
investigated in a rigorous manner. Each main and interaction
effect of the variables is analysed referring to the generated
electric power and the heat recovered, and subsequently
especially for the analysis of the steam-to-carbon ratio, whose values play a very important role on both performances and
degradation of a SOFC stack. The proposed methodology
seems to be powerful for system analysis; moreover, it allows the design of a set of experimental tests reducing the number of experiments and focusing the main expected results.