Coming up next week, we have the last set of seminars for the semester, featuring another double-header of Nick Sirmas and Simon Freijer-Poulsen Cloutier. Simon will be giving a talk on “Condition Monitoring of Rotating Machinery via Harmonic Subset Analysis” and Nick will be giving a talk on “Shock Instability in Gases Characterized by Inelastic Collisions”. The talk abstracts follow below.
Date: Friday December 7th
Place: CBY D103 – NOTE ROOM CHANGE!
Condition Monitoring of Rotating Machinery via Harmonic Subset Analysis
Simon Freijer-Poulsen Cloutier
The analysis of system vibration is a popular industrial method for monitoring the health of rotating machinery, such as shaft bearings and gearboxes. Examining the overall energy of these systems with statistical measures, such as RMS or kurtosis, can be done quickly and has proved a popular solution, with the downside of often being vague in terms of the system’s actual condition. More intensive methods, such as neural networks or PSO, can provide more explicit information, but tend to require extended processing time (restricting real-time usage) and/or reference data from the system in a healthy state (difficult to obtain). Additionally, the physical structure of the system can potentially restrict how sensors can be placed and what kind of data can be collected. An automated, real-time solution is proposed which operates on the premise of Harmonic Subset Analysis, effectively using each individual frequency’s harmonic behaviours to filter the signal spectrum on a point-for-point basis. In this way, the algorithm achieves aggressive de-noising (above 99%) which highlights the regions of strong harmonic relevance denoting rotational events and greatly aids in automated detection. Through model fitting of the filtered spectrum, the system is then capable of rapidly identifying the target machine’s operating speed and any associated faults within range of each sensor.
The algorithm’s operational parameters are limited to fixed and known values, such that it can still be used in cases where the shaft speed is not known or where no healthy reference data is available. Its harmonic-based structure also implicitly retains low-amplitude harmonics while filtering out non-harmonic interference.
Practical testing, using laboratory data and without any tuning, produced the expected results: The algorithm was able to determine both the system’s shaft speed and fault state within a reasonable margin of error. Computation time in a non-optimised state was roughly five times the source signal’s length, still within the realm of continuous application.
Shock Instability in Gases Characterized by Inelastic Collisions
Supervisor: Dr. Radulescu
The current study addresses the stability of shock waves propagating through dissipative media, analogous to both granular media and molecular gases undergoing endothermic reactions. In order to investigate the stability, a simple molecular dynamics model was developed to observe shock waves and their structures with the inclusion of energy dissipation. For this, an Event Driven Molecular Dynamics model was implemented in a 2D environment, where a molecule is represented by a disk. The simulations addressed the formation of a shock wave in a gas by the sudden acceleration of a piston. Inelastic collisions were assumed to occur only if an impact velocity threshold is surpassed, representing the activation energy of the dissipative reactions.
Parametric studies were conducted for this molecular model, by varying the strength of the shock wave, the activation threshold and the degree of inelasticity in the collisions. The resulting simulations showed that a shock structure does indeed become unstable with the presence of dissipative collisions. This instability manifests itself in the form of distinctive high density non-uniformities behind the shock wave, which take the form of convective rolls. The spacing and size of this “nger-like” unstable pattern was shown to be dependent on the degree of inelasticity, the activation energy, and the strength of the driving piston.
The mechanism responsible for the instability was addressed by studying the time evolution of the material undergoing the shock wave compression and further relaxation. It is found that the gas develops the instability on the same time scales as the clustering instability in homogeneous gases, rst observed by Goldhirsch and Zanetti in granular gases. This conrmed that the clustering instability is the dominant mechanism.