Degree Type


Date of Award


Degree Name

Master of Science


Agricultural and Biosystems Engineering

First Advisor

Brian L. Steward


A change in a fluid’s dielectric properties can be investigated using dielectric spectroscopy to gain valuable insight into the changing condition of the fluid. A dielectric spectroscopic sensor was developed using a cylindrical capacitive sensing unit with the fluid as the dielectric media. The sensor was used to estimate or detect contaminants in a hydraulic fluid and a compressed air stream. Tests were performed with a hydraulic fluid in which the dielectric sensor’s performance was evaluated in detecting iron powder and ISO medium test dust particles as contaminants in the fluid. Using iron powder as contaminants, two tests were performed with central electrodes of diameters 6.35 mm and 17.7 mm inch placed inside the capacitive dielectric sensor. The results from partial least squares (PLS) regression showed that the root mean square error of calibration (RMSEC) and the root mean square error of cross-validation (RMSECV) for a 6.35 mm (0.25-inch) diameter central electrode were 1.1 and 1.39 of adjusted ISO cleanliness code respectively. For a 17.7 mm (0.70-inch) diameter central electrode, the RMSEC and RMSECV values were 0.62 and 0.83 of adjusted ISO cleanliness code, respectively. Similarly, a test was performed using ISO test dust particles as contaminants with a central electrode of 17.7 mm diameter. The RMSEC and RMSECV values from the model for ISO test dust were 1.29 and 1.48 of adjusted ISO cleanliness code, respectively. Tests were also conducted to investigate the efficacy of dielectric spectroscopy in detecting water and oil droplets in a compressed air stream. Spray nozzles were used to produce fine droplets of deionized water and light lubricant oil. Multivariate statistical techniques, principal component analysis (PCA) and linear discriminant analysis (LDA), were used to develop statistical classifiers, which determined the performance of dielectric spectroscopic sensor in differentiating the dry compressed air from an air stream with entrained liquid droplets. Through model calibration and cross-validation, the classifiers were able to separate the two cases without any errors, validating the dielectric sensor’s ability to detect of liquid droplets in an air stream.


Copyright Owner

Safal Kshetri



File Format


File Size

84 pages