1st ICAI 2020

International Conference on Automotive Industry 2020

Mladá Boleslav, Czech Republic

The burst is excited 10 000 times per second and wave itself oscillates with frequency 350 kHz. The sensor, placed in offset position from transducer senses whole wave. In the controller, however, only two parameters of selected peak of the wave are recorded. Particularly, it is an amplitude of selected peak in volts and duration of its propagation from a moment of excitation to its arrival to the receiver in nanoseconds. The tracking of selected peak during operation is performed by internal algorithm. From the two characteristics, the presented research focused only on the analysis of propagation time. The reason for it was that the propagation time characteristic is less sensitive to quality of the contact between the ring and PZT elements than the wave amplitude. During current experiment, the sensor and transducer were both placed in the load zone of the bearing. Mutual distance of PZT elements was 20°. Spacing of rolling elements in the bearing was 30°. This configuration therefore, allowed sensing the pass of individual rolling elements in-between. Figure 2: a) Attachment of sensors on the bearing, b) Detail of sensed SAW impulse with selected peak of which characteristics are recorded and analysed.

Source: Own elaboration (2020) 2.4 Design of experiments

Experimental runs were divided into two groups. The first followed a method Design of Experiment (DOE) – two-level full factorial experiment. Levels of parameters used in this group are listed in the Table 1. The objective of this group was to provide indication of signal responses to various operating condition defined by: speed, load and viscosity. The other group was based on speed ramp-ups from 0 to 3000 rpm with constant acceleration 150 rpm/s lasting 20 s each. Such short duratio allowed keeping the bearing temperature within a band of 1.5°C from initial temperature.This procedure created virtually isothermal conditions during experiment, well corresponding to inputs and assumptions of mathematical models. As a result, measured data could be directly compared with predictions made by models. All runs were repeated for oils with viscosities 22 cSt, 46 cSt and 100 cSt to achieve transition of bearing into EHL model at different speeds. For last group also a control cases with dry bearing were performed to directly show how signal for dry bearing differs from others.

135