A3.2 - Towards a Medium-Independent Flow Sensor

SENSOR+TEST Conferences 2011
2011-06-07 - 2011-06-09
Proceedings SENSOR 2011
A3 - Flow Measurement
D. Reyes Romero, A. Cubukcu, G. Urban - Universität Freiburg (Germany)
73 - 77


We have successfully developed a low power 2D calorimetric gas flow sensor in order to measure the wind speed on a blimp. It features a central heater-thermistor pair surrounded by four additional germanium thermistors which allow a temperature sensing resolution under 1 mK. Flow measurements strongly depend on the fluid’s thermal conductivity (κ), specific heat (cp) and density (ρ). Since variations on these properties may result in a wrong flow rate measurement, recalibration is needed for different fluids. In this paper we discuss the effect of the thermal and physical properties of the fluid on the sensor’s response. With this, we aim to develop a selfcalibration method for the sensor. Simulations suggest that the frequency response of the temperatures at the thermistors exhibits a dominant second order low-pass behavior. Additionally, experimental results show that such frequency response does not seem to be affected by the fluid’s flow rate. The frequency response may be analytically predicted in terms zof the thermal conductivity and the heat diffusivity of the fluid. Consequently, a full spectra analysis and curve-fitting procedures would allow us to determine the fluid properties. However the flow rate measurement must be done at a single frequency where the flow rate sensitivity is expected to be optimal. This imposes two major requirements: to know the analytical frequency response a priori by means of either numerical or analytical approaches; and to perform concurrent signal treatment in order to fit the frequency response to a set of parametric curves. Under non-zero flow conditions, simulation results for many thermal conductivity values show that the phase shift between the middle thermistor and any other surrounding thermistor decreases as the thermal conductivity increases. Simulations also showed that, the phase shift has a stronger correlation to the thermal conductivity than to any other parameter. Thus, the complexity of the curve-fitting calculations may be decreased for
certain operational ranges.