Multi-parameter detection in fluid flows

A micromechanically fabricated thermal flow sensor is presented. The sensor incorporates multiple, 100 μm spaced, resistive sensing elements on a glass substrate in a silicon flow channel. This sensor works on the principle of a travelling heat pulse through the fluid. The response to this heat pulse at different positions upstream and downstream from a heater is used to determine not only flow velocities but also fluid properties. Theoretical analysis of the sensor response shows that the sensor is more able to discriminate between flow velocities and fluid properties when certain combinations of sensing element signals are used. It is shown that the sensor can also measure mass flow as long as the 'time of flight' of a heat pulse can be measured at equal distances from the heater upstream and downstream. Combination of the 'time of flight' at two different positions downstream can be used to determine the diffusivity of the fluid. The sensor can be made sensitive to flow velocity by taking the heat pulse response at two different locations downstream at an instant in time when the signal amplitudes are equal. The 'time of flight' measured at one position downstream is only accurate when the velocities are high enough since the diffusive effect can be neglected. The ability of an artificial neural network to learn to discriminate between the flow velocity and fluid properties is analyzed.