In this work, a novel method of viscosity measurement is proposed using a device comprising a compliant mechanism, a vibration source, and a piezoelectric sensor. The vibration source creates linear harmonic vibrations in the compliant mechanism suspended in the liquid, and the acceleration response of the mechanism is measured using the piezoelectric sensor. The vibration source is located in the central mass of the compliant mechanism, which is designed to have the necessary directional stiffness. As the mechanism vibrates, the links in the mechanism undergo damping due to the shearing action of the fluid because of its viscosity. A series of viscosity measurements are carried out with the use of water–glycerol solutions such that the acceleration of the mass is influenced by the fluid’s viscosity. During the working of the device, the mechanism is immersed in the liquid whose viscosity is to be measured. The acceleration response of the mass is recorded as time domain data using NI Lab View hardware and software, which are used to train a machine learning model. Later, a regression-based machine learning model is used for the estimation of dynamic viscosity for the given acceleration input. Experiments are performed with the prototype device using the water–glycerol solution within a viscosity ranging from 10 cP to 60 cP. The proposed sensor can be used for in-line measurements or used as a handheld instrument for quick measurements. The machine learning model achieved a high level of accuracy, evidenced by an R-squared value of 0.99, indicating that it explains 99% of the variance in the data.
Read full abstract