A new two-element thermal sensor for a single-point measurement of wall shear stress fluctuations (τ') in turbulent flows has been developed to overcome the deficiencies of the conventional single-element hot-film.
This thesis has three objectives; first, to establish, for available single-element hotfilm sensors, the extent of loss of resolution due to probe size as the Reynolds number (Re) is increased; next, to establish that decreasing sensor size alone still leaves us with significant errors in resolving ability and in the measured spectrum of τ': third, to present a concept for new sensors to overcome these errors and show how they have been made using microfabrication (MEMS) techniques. Experiments with the new sensors are the subject of a separate investigation and will not be covered in this thesis.
This study has involved the partial construction of a high Reynolds number water pipe flow facility (0.3 m diameter) in which we measured τ' over the Re range 5×10⁵ to 1.1×10⁶, using a 750/μm x 200μm single-element sensor. We found that values of the moments of r decrease continuously with increasing Re, beyond the Re value at which probe width exceeds ten wall units. For the spectrum of τ', substrate conduction causes a spurious amplification at low frequencies, while fluid thermal boundary layer inertia causes a spurious attenuation at high frequencies. Our analysis and numerical simulations of governing equations for the fluid and substrate show that the onset frequencies for these spectral deviations are close enough to distort the spectrum over its entire range. The variation of Iength-scale resolving ability with fluctuation strength has also been quantified.
The sensor design conceived to reduce these problems has an encompassing guard heater run at the sensor temperature using a separate thermal anemometry circuit. Sensor sizes of 12×60, 24×96, 72×288 and 250x1000 (μm) were proposed to resolve τ' over our range of Re. A microfabrication process with six levels of photolithography and suitable packaging has allowed this concept to be realised. Tests with a packaged sensor in a wind tunnel boundary layer showed a stable turbulent signal, responding to changes in overheat.