Vibration analysis for condition monitoring of mechanical presses

Abstract: In a world where the concept of just-in-time has been lifted outside of the manu-facturing world and onto the market, the increasing demand of luxury wares like cars is accompanied only by tougher industrial conditions and no room for error. Loss of productivity and decrease in efficiency can be devastating to any manufac-turing company in today’s market climate. One of the largest causes of these is-sues is production downtime due to machine breakdown or unexpected failure. To prevent such events from happening, Volvo Cars has decided to investigate the possibility of using vibration analysis for predictive maintenance and condition monitoring of presses used to manufacture sheet-metal car body components. Although the concept of using vibration analysis for predictive maintenance is not new and the positive effects have been established, uses in complex machinery with intermittent cycles as mechanical presses have at the time of this thesis not been widely studied. This thesis aims to research if vibration analysis is a useable tool for predictive maintenance in mechanical presses and what information it might provide. Vibration analysis is a wide field of technology and there are endless tools and methods to perform analysis. This thesis mainly focuses on the possibilities in using Power Spectral Density (PSD), Fast Fourier Transform (FFT) and statistical parameters like Root Mean Square (RMS) and Kurtosis. These are used in many applications and there are many companies that are familiar with implementing these tools, making them suitable to research. After a thorough investigation of the use of PSD, FFT, RMS, and Kurtosis on the main components of the press crown and driving system, it is possible to conclude that vibration analysis is useable when using predictive maintenance on a press. Some configuration of PSD, FFT, and RMS gives the most benefits in detecting and analysing faults and issues. However, more data must be collected to establish when alarms must be issued to the maintenance crew and to build thorough models that provide enough information for making informed diagnosis and decisions based on vibration analysis.