A Design of Experiments (DOE) approach to optimise temperature measurement accuracy in Solid Oxide Fuel Cell (SOFC)

32nd UIT (Italian Union of Thermo–fluid-dynamics) Heat Transfer Conference

物理学;力学

Barari, F.^1,2 ; Morgan, R.^1 ; Barnard, P.^2

University of Brighton, Cockcroft Building, Lewes Road, Brighton

BN2 4GJ, United Kingdom^1

Ceres Power Limited, Viking House, Foundry Lane, Horsham

RH13 5PX, United Kingdom^2

关键词: Full factorial;    Gas temperature;    Hot gas temperatures;    Interaction effect;    Statistical design of experiments;    Temperature measurement systems;    Temperature range;    Wall temperatures;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/547/1/012004/pdf
DOI  :  10.1088/1742-6596/547/1/012004

学科分类：力学，机械学

来源: IOP

【 摘 要 】

In SOFC, accurately measuring the hot-gas temperature is challenging due to low gas velocity, high wall temperature, complex flow geometries and relatively small pipe diameter. Improper use of low cost thermometry system such as standard Type K thermocouples (TC) may introduce large measurement error. The error could have a negative effect on the thermal management of the SOFC systems and consequential reduction in efficiency. In order to study the factors affecting the accuracy of the temperature measurement system, a mathematical model of a TC inside a pipe was defined and numerically solved. The model calculated the difference between the actual and the measured gas temperature inside the pipe. A statistical Design of Experiment (DOE) approach was applied to the modelling data to compute the interaction effect between variables and investigate the significance of each variable on the measurement errors. In this study a full factorial DOE design with six variables (wall temperature, gas temperature, TC length, TC diameter and TC emissivity) at two levels was carried out. Four different scenarios, two sets of TC length (6-10.5 mm and 17-22 mm) and two different sets of temperature range (550-650 °C and 750-850 °C), were proposed. DOE analysis was done for each scenario and results were compared to identify key parameters affecting the accuracy of a particular temperature reading.