Experiments in the study of transient aerodynamics typically require pressure measurements with a high spatial and temporal resolution. Existing commercial pressure transducers are expensive and they provide a spatial resolution only on the order of millimetres. The full bandwidth of commercial devices (which extends to around 200 kHz) can only be utilised by exposing the transducer to the flow environment with very little thermal or mechanical protection. If insufficient protection is provided, the expensive commercial devices are likely to be damaged. Inexpensive pressure sensors based on extrinsic Fabry-Perot fibre optic interferometry are capable of measurement with a high spatial and temporal resolution. Thermal protection or isolation for these sensors is still required, but they can be exposed directly to the flow if the sensors are disposable (low cost). Excessive thermal or mechanical protection is not required for these sensors because the damaging heat transfer and particle impacts that may occur in transient aerodynamic facilities generally occur after the useful test flow. In this dissertation, a variety of construction techniques for diaphragm-based Fabry-Perot fibre optic pressure sensors were investigated and the advantages and disadvantages of all techniques are compared. The results indicate that using a zirconia ferrule as the substrate, a liquid adhesive as the bonding layer, and a polished copper foil as the diaphragm provide the best results. It is demonstrated that a spatial resolution on the order of 0.1 mm and a bandwidth to more than 100 kHz can be achieved with such constructions. A variety of problems such as hysteresis, response irregularity, low visibility and sensor non-repeatability were observed. By using a thinner bonding layer, a larger bonding area, longer cavity length, increased calibration period, and applying load cycling to the diaphragm, the hysteresis was minimized. Sensor response irregularity was also minimized using a polished diaphragm. Visibility increased to about 90% using active control of the cavity length during the construction process. Non-repeatability was found to be a consequence of adhesive viscoelasticity and this effect was minimized using a thin layer of adhesive to bond the diaphragm to the substrate. Due to the effects of adhesive viscoelasticity, the pressure sensors indicate an error of up to 10% of mean value for the reflected shock pressure. This error could not be further reduced in the current sensors configuration. Some new configurations are proposed to decrease the effect of sensor non-repeatability. The effect of pretensioning the diaphragm was investigated analytically but the results do not indicate any considerable advantage for the levels of pretension likely to be achieved in practice. However, the results do indicate that pretension effects caused by an environmental temperature change can damage the sensor during storage. The effect of the initial diaphragm deflection on the sensor performance and temperature sensitivity was modelled and the results show that an initial diaphragm deflection can improve the sensor performance. The effect of the thermal isolation layer on the sensor performance was also investigated and the results show that for a shock tube diaphragm bursting pressure ratio up to 5.7, heat transfer does not contribute to sensor errors for the first millisecond after shock reflection. However, it was found that the use of a thin layer of low viscosity grease can protect the sensor for about 20 ms while only decreasing its natural frequency by typically 17%. The grease layer was also found to decrease the settling time of a low damping ratio sensor by 40%. The sensor was successfully employed to identify an acoustic disturbance in a shock tube.
展开▼
