C. Sujatha

Using neural networks for the diagnosis of localized defects in ball bearings

Two neural network based approaches, a multilayered feed forward neural network trained with supervised Error Back Propagation technique and an unsupervised Adaptive Resonance Theory-2 (ART2) based neural network were used for automatic detection/diagnosis of localized defects in ball bearings. Vibration acceleration signals were collected from a normal bearing and two different defective bearings under various load and speed conditions. The signals were processed to obtain various statistical parameters, which are good indicators of bearing condition, and these inputs were used to train the neural network and the output represented the ball bearing states. The trained neural networks were used for the recognition of ball bearing states. The results showed that the trained neural networks were able to distinguish a normal bearing from defective bearings with 100% reliability. Moreover, the networks were able to classify the ball bearings into different states with success rates better than those achieved with the best among the state-of-the-art techniques.

Static and dynamic analysis of spur and bevel gears using FEM

This paper presents the findings of three-dimensional stress analysis of spur and bevel gear teeth by finite element method using cyclic symmetry concept. The displacement of a tooth is computed for each Fourier harmonic component of the contact line load and all the components are added to obtain the total displacement. This displacement is used in the calculation of static stress in the teeth. The natural frequencies and mode shapes are obtained using the submatrices elimination scheme.

Disc flexibility effects in rotor bearing systems

Abstract The importance of the inclusion of disk flexibility on the natural frequencies in a rotor dynamic analysis has attracted the attention of many researchers in the recent past. Such an analysis helps to predict the dynamic response of rotors more accurately from the natural frequencies and mode shapes calculated hence. In the present analysis a semi-analytical conical shell finite element is used for the modelling of the rotor with the inclusion of the support bearing flexibility. Such an element could include the flexibility of the disks, shaft and the bearings. A parametric study was also performed for looking into the frequency characteristics of the system. The study brings out the additional disk modes a well as the behaviour of the system at higher circumferential modes.

Modelling of automotive disc brake squeal and its reduction using rotor design modifications

This paper proposes a numerical approach to reducing disc brake squeal through rotor design modifications using the finite element method (FEM). The finite element (FE) model is validated based on the measured data extracted from experimental modal analysis for individual brake components in free-free boundary condition and brake assembly under applied pressure. The FE model is used to investigate brake squeal through two numerical approaches namely complex eigenvalue analysis and dynamic transient analysis. Experimental squeal tests are performed using a brake test rig for verification of the predicted results. It is observed that the results of both the complex eigenvalue analysis and dynamic transient analysis agree well with experimental squeal frequencies. In order to reduce brake squeal, a number of structural modifications on the disc are evaluated. The predicted results show that the squeal noise of disc brake is influenced by the natural frequency of the brake rotor and its mode shape. It is also found that a good choice of rotor geometry in the pre-design stage could help in reducing squeal noise.

Surface Contact Fatigue Failure Assessment in Spur Gears Using Lubricant Film Thickness and Vibration Signal Analysis

Gears are very common machine elements that are used to transmit power. During power transmission, the gear teeth are subjected to high contact pressure, leading to wear and tear, even though adequate lubrication is used. When gears operate near their maximum load capacity, very high contact pressure occurs at the mesh interference. This may lead to partial breakdown of the lubricant film at the gear teeth contact surface, thereby triggering distributed faults on gear teeth, viz. scuffing, scoring, spalling, pitting, and mild wear. Currently there are three different approaches to the detection of faults in geared systems: vibration analysis, oil/wear particle analysis, and acoustic signal analysis. Generally, in an industrial environment, vibration analysis and oil/ wear particle analyses are conducted independently, but of late there are requirements in the diagnostic world for more reliable and effective diagnostic information regarding fault growth in machinery components. In the present work, experimental investigations are carried out to assess surface fatigue wear on spur gear teeth of back-to-back power recirculation-type spur gearbox; in particular, vibration signal analysis using continuous wavelet transform is used in conjunction with specific film thickness studies. Specific film thickness analysis provided details of lubrication regimes in gear teeth contacts and corresponding wear propagation on gear teeth, and the vibration signals acquired from a gear pair are postprocessed using Morlet wavelet transform. Results obtained from the integration of specific film thickness analysis and wavelet transform provide better diagnostic information than the lubricant film thickness estimation alone for detection of gear faults.

Track Modulus Analysis of Railway Track System Using Finite Element Model

With the development of ever faster trains, the problem of excessive ground vibrations has increased. Dealing with ground vibration from surface and underground trains is a challenging issue for the railway industry. In the present work, the analysis of a typical Indian railway track system has been done with special emphasis on “track modulus” for Prestressed Concrete (PSC) sleepers and Wooden (WOOD) sleepers. The track consists of two rails of standard length, rail pads and sleepers with constant sleeper spacing, ballast and subgrade covering the length of the rails as per the Indian Railways standards. Finite Element models have been developed for computer simulation of the dynamic behaviour of the railway track system for 52PSC, 60PSC, 52WOOD and 60WOOD track. In this model, subgrade, ballast and rail pad parameters have been considered for parametric analysis of the track modulus. Finally the track system is excited harmonically over a range of frequencies to predict the dynamic variation in the track modulus.

Dynamic response of railroad vehicles: a frequency domain approach

A very elaborate Finite Element (FE) model and a rigid body model of a typical electrical multiple unit trailer coach are described. These models were used to find the dynamic response to track irregularities in the frequency domain. The Power Spectral Density (PSD) of track irregularities was used as input to the system. The influence of different track irregularities on dynamic response and coupling between vertical and lateral dynamics was investigated. Extensive experiments were carried out, and analytical results were compared with the measured response.

Response of a half-car model with optimal magnetorheological damper parameters

In this paper, control of the stationary response of a half-car model with a magnetorheological (MR) damper moving over a random road is considered. The MR damper is characterized using Bingham and modified Bouc–Wen models whose parameters are determined optimally using a multi-objective optimization technique and nondominated sorting genetic algorithm II. The multi-objective optimization problem is solved by minimizing the difference between the root mean square sprung mass acceleration, the sum of the front and rear suspension strokes and the sum of the front and rear road holding response of the half-car model with the MR damper, and those of an active suspension system based on linear quadratic regulator (LQR) control. The control force of the MR damper suspension is constrained to lie within ±5% of the control force corresponding to the active suspension system based on LQR control. It is observed that the MR damper suspension systems with optimal parameters perform an order of magnitude better than the passive suspension and perform as well as active suspensions with limited state feedback, and closer to the performance of fully active suspensions. In the case of MR damper suspensions, the vehicle response statistics are obtained using the equivalent linearization method and verified by Monte Carlo simulation.

Vibration and ride comfort studies on a tracked vehicle

This paper describes a field test conducted on a military tracked vehicle. The accelerations experienced at selected locations on the vehicle for four different types of terrain were recorded and analysed in spectral and statistical modes. The first few natural frequencies of the vehicle were obtained using an in-plane N+2 degree-of-freedom linear dynamic model of the vehicle with two degrees of freedom corresponding to bounce and pitch modes of the hull and N degrees of freedom corresponding to the bounce modes of the N road wheel assemblies. The analytically obtained natural frequencies match with peaks in the experimentally obtained transfer function and acceleration power spectral density curves. From the point of view of crew comfort, acceleration levels encountered in the crew cabin have been compared with those specified in ISO 2631 standard. Ride quality assessment has also been made on the basis of absorbed power.

Modal and Vibration / Stress Analysis of a Passenger Vehicle by FEM

Finite element method has been used for modal and vibration / stress analysis of a passenger vehicle. The study vehicle has been discretised taking into account the chassis elements, axles, suspension and tyres. The Lanczos method has been successfully used for the modal analysis. The power spectral density (p.s.d.) of acceleration of a track measured by using three height sensors has been fed as input to the tyres and the dynamic response of the vehicle in terms of acceleration and strain has been computed at all the nodes using finite element modeling and the random vibration concept. The experiments were carried out using piezo-electric accelerometers and strain gauges to measure the vibration and strain levels at critical points of the vehicle. The vibration and strain levels calculated through f.e.m. match well with the experimental values. Parametric study has also been carried out with a view to find the optimum suspension / tyre characteristics for maximum ride comfort in the vehicle.