Session: Poster Session

Paper Number: 161992

Acoustic Cavitation Detection in a Centrifugal Pump Outlet 

Our goal is to develop a detection system that can alert the user to the occurrence of cavitation in real time in order to prevent the damaging effects of cavitation in the outflow of a centrifugal pump without cautiously running equipment at prohibitively low pressures. To achieve this, we use acoustics as a detection method by recording the noise produced by cavitation with an ICP microphone (130D20, PCB Piezotronics) and analyzing the frequency domain of the results. Data acquisition is performed using an NI DAQ system and LabVIEW software on a PC. Our theory on the functionality of this equipment is that we will use a large magnitude of noise at a particular frequency as a signal for the presence of cavitation. The subject of the study is a two-inch PVC pipe that carries water from the outlet of a centrifugal pump to a large tank used by a heat pump. In order to collect data, we initially hold the microphone by hand, but later use a 3D-printed fixture that is bolted together around the pipe to hold the microphone steady. We collect multiple 5-second sound recordings of the pump at pressure differences ranging from 5 psi to 50 psi, representing the full operating range of the pump. The test pressure is controlled via a butterfly valve built into the pipe and an attached pressure gauge.

The recorded data is then analyzed using a Python program to perform deconvolution, comparing a sample recorded under normal pressure conditions to samples recorded at various pressure deviations. For this experiment, we consider 5 psi to be the normal conditions, as that is the lowest operating pressure of the pump, and cavitation should not occur there. The analysis demonstrates that cavitation can be detected acoustically, but not in the way we initially predict. The FFTs of the deconvolved samples reveal distinct peaks at higher pressure deviations, indicating cavitation. We then use this knowledge to develop a Python program that can identify when cavitation is occurring in the pipe by collecting several-second-long acoustic recordings and searching for repeated distinctive peaks in the deconvolved frequency domain plots. We do not identify one specific frequency where cavitation noise occurs, but we are able to identify cavitation qualitatively and quantitatively by looking at the strong peaks across multiple frequencies in the deconvolved frequency domain plots. Thus, we are able to successfully create a cavitation detection system for this centrifugal pump to prevent internal damage during use and to serve as an educational tool.

Presenting Author: Anthony Martin Valparaiso University

Presenting Author Biography: I am an undergraduate student in mechanical engineering at Valparaiso University, and I am also minoring in bioengineering with a focus on prosthetics. I am a member of Tau Beta Pi and I work in the University as an engineering tutor while also swimming competitively in the University's D1 program. Some of my special interests in engineering are 3D printing and CAD design. I'm originally from Bartlett, Illinois.

Authors:

Anthony Martin Valparaiso University
Julia Radcliffe Valparaiso University
Shahin Nudehi Valparaiso University