Periodic gas temperature measurement as an inverse heat transfer problem, utilization of the Particle Filter algorithm and mechanistic models

Abstract

In many engineering applications one can find periodically variable gas temperatures. Possible examples include the temperature inside the cylinders of Stirling Engines and other reciprocating machines. New methods are proposed and evaluated for the measurement of periodically varying temperature by thermocouples. A Computational Fluid Dynamics model is created to assess the method viability and efficacy in a realistic setting while avoiding measurement errors. The first method is based on matrix analysis, the two others on the particle filter algorithm used to estimate either the heat transfer coefficients with the additional use of other flow information, or to estimate the thermocouple constants without said information. It is shown that the matrix method is sufficient for cases with constant velocities. For variable velocities, the PF algorithm shows promising results as based on measurements generated from the numerical simulation.