Measurement Technique for Ideal Selection of Sensors and Accurate Force Recovery on Aerodynamic Models

Abstract

The accuracy of force measurement and prediction is of prime importance in the calibration of a force balance in aerodynamic models. Multi-point calibration has recently been introduced which has proven to be a more accurate prediction technique than the traditional single-point calibration technique. This paper describes the multi-point calibration of a double cone model using tri-axial and uniaxial accelerometers to obtain the maximum number of responses with accurate prediction using a minimum number of accelerometers. The multi-point calibration was performed by applying forces at nine locations on a double cone model considering two configurations. Firstly, accelerations were measured simultaneously using one tri-axial and two uniaxial accelerometers which were fixed at three locations of the model-balance assembly. Secondly, forces were measured from the responses obtained simultaneously using one uniaxial and one triaxial accelerometer fixed at two locations of the model-balance assembly. The prediction of the forces was performed using Adaptive Neuro Fuzzy Inference System (ANFIS). Genetic algorithm was used to obtain pure normal and pure axial forces from the components of applied forces at different calibration locations. The forces were predicted accurately from the responses measured using two accelerometers (one uniaxial and one triaxial accelerometer) as well as from the responses of three accelerometers (one tri-axial and two uniaxial accelerometers). Thus, tri-axial accelerometers can be used rather than using multiple numbers of uniaxial accelerometers. The use of tri-axial accelerometers leads to the ease of experimentation as mounting of multiple number of uniaxial accelerometers makes the calibration set-up complex and response measurement challenging. The attainment of responses in three directions at a single location is possible only using tri-axial accelerometer which helps in understanding the actual degrees of freedom experienced by the model. Thus, the use of tri-axial accelerometers for obtaining the responses in dynamic calibration is a better alternative than using uniaxial accelerometers.