PSI - Issue 17

Andra Gabriela Stancu et al. / Procedia Structural Integrity 17 (2019) 238–245 A. G. Stancu et al./ Structural Integrity Procedia 00 (2019) 000 – 000

243

6

Two types of signals can be distinguished in acoustic emission as shown in Figure 1, based on the energy released when a component undergoes stress, namely: burst type and continuous type. The first is more applicable to identifying defects such as cracks, corrosion, fibres breaking, which trigger the acquisition of the signal based on a pre-defined threshold. A continuous AE signal can be associated with yield rather than fracture mechanics. It captures variations in amplitude and frequency, being continuously recorded for a defined period of time, suitable for showing progressive damage and can be better correlated to dislocation avalanching at the leading edge of discontinuities, and discontinuity surfaces rubbing. For the present case study, the encountered damage mechanism can be described mainly by the lack of lubrication causing metal-to-metal contact of the shaft and inner bearing surface. Thus, the energy released in this case would be more sensitively captured in the form of continuous AE type of signals.

3.2. Analysis

The AE data adopted for validation analysis is sampled at 7.2 MHz rate continuously on a rotating machinery. The data analysis was performed as presented in Section 2. The convergence criterion implemented for baseline definition is shown in Figure 2 where the effect of the sample size reaches a standard deviation of 1.091% at approximately 300 measurements, corresponding to 30 minutes of continuous monitoring. Once the baseline condition is identified, the subsequent signatures are gathered, continuously updating the condition monitoring process using similarity analysis,

Figure 2. Baseline analysis, convergence at approximately 300 measurements; Mean (red line) and standard deviation (green line) of the baseline.

Figure 3. Similarity analysis results; Deviation from the baseline remains under 5%.

shown in Figure 3. In Figure 3, the measurements in percentage on the y-axis depict the deviations of the monitoring conditions (x axis) to the predetermined baseline, remaining steadily below 5%. Combining with the fact that no progressively increase trend is observed, it evidently suggests that the machinery being monitored is in healthy state. The percentage reference in this case can only be representative of how high the change is at the monitoring time. Finally, the operator establishes a limit of the deviation allowed, based on previous maintenance experience and observations of the machinery system’s working conditions, as well as the failure incidents history. This threshold is utilized to give out warnings once breached. A generally used model describing the likely failure rates of technologies and products, such as the bathtub curve is used to show the lifetime over a certain period of time. The monitoring process can be employed during any given phase if the three-part timeline presented in Figure 4, correlating any measured deviations from the normal operating conditions to the probability of failure characterized by the life stage of the machinery. It suggests that during useful lift period, the failure rate of the machinery is expected to remain a low and steady horizontal line. This is comparable to the results demonstrated in Figure 3, where the