Professor John Durodola

Ultrasonic de-agglomeration and dispersion of oxides is important for a range of applications. In particular, in liquid metal, this is one of the ways to produce metal-matrix composites reinforced with micron and nano sized particles. The associated mechanism through which the de-agglomeration occurs has, however, only been conceptualized theoretically and not yet been validated with experimental observations. In this paper, the influence of ultrasonic cavitation on SiO2 and MgO agglomerates (commonly found in lightweight alloys as reinforcements) with individual particle sizes ranging between 0.5 and 10 μm was observed for the first time in-situ using high-speed imaging. Owing to the opacity of liquid metals, a de-agglomeration imaging experiment was carried out in de-ionised water with sequences captured at frame rates up to 50 kfps. In-situ observations were further accompanied by synchronised acoustic measurements using an advanced calibrated cavitometer, to reveal the effect of pressure amplitude arising from oscillating microbubbles on oxide de-agglomeration. Results showed that ultrasound-induced microbubble clusters pulsating chaotically, were predominantly responsible for the breakage and dispersion of oxide agglomerates. Such oscillating cavitation clusters were seen to capture the floating agglomerates resulting in their immediate disintegration. De-agglomeration of oxides occurred from both the surface and within the bulk of the aggregate. Microbubble clusters oscillating with associated emission frequencies at the subharmonic, 1st harmonic and low order ultra-harmonics of the driving frequency were deemed responsible for the breakage of the agglomerates.

Machine learning is now applied in virtually every sphere of life for data analysis and interpretation. The main strengths of the method lie in the relative ease of the construction of its structures and its ability to model complex non-linear relationships and behaviours. While application of existing materials have enabled significant technological advancement there are still needs for novel materials that will enable even greater achievement at lower cost and higher effectiveness. The physics underlining the phenomena involved in materials processing and behaviour however still pose considerable challenge and yet require solving. Machine learning can facilitate the achievement of these new aspirations and desires by learning from existing knowledge and data to fill in gaps that have so far been intractable for various reasons including cost and time. This paper reviews the applications of machine learning to various aspects of materials design, processing, characterisation, and some aspects of fabrication and environmental impact evaluation.

The objective of this paper is to model and analyse the dynamic response of an automotive lamp assembly. Modern automotive lamp assemblies have complex geometry and are composed of different parts made of polymer materials. As part of design verification, automotive lamp assemblies are subjected to accelerated random vibration tests to assess their integrity over a lifetime exposure to mechanical vibration loading. Understanding the dynamic behaviour of the lamp is crucial in the numerical evaluation of the fatigue life. Dynamic analysis involves characterising the modal and harmonic behaviours. In this work, numerical modal properties and harmonic responses were validated using experimental testing. The numerical analysis was carried out using the ANSYS finite element analysis (FEA) software. Experimental modal properties including mode shapes and corresponding frequencies were determined using Polytec PSV-500 Xtra laser scanning head at a frequency range of 10 to 1000Hz. The experimental harmonic transmissibility responses of all the components of the lamp assembly were determined using a vibration shaker. The experimental and numerical mode shapes and responding frequencies obtained in the analyses compared well thus validating the numerical modal model. Furthermore, the mode shapes showed that the lamp assembly was mostly vibrating in bending, therefore subsequent analysis should take this into account. Harmonic response validation showed that the first few numerical resonant frequencies, that dominate the response, compared well with experimental results.

One of the main applications of ultrasonic melt treatment is the grain refinement of aluminium alloys. Among several suggested mechanisms, the fragmentation of primary intermetallics by acoustic cavitation is regarded as very efficient. However, the physical process causing this fragmentation has received little attention and is not yet well understood. In this study, we evaluate the mechanical properties of primary Al3Zr intermetallics by nano-indentation experiments and correlate those with in-situ high-speed imaging (of up to 1 Mfps) of their fragmentation process by laser-induced cavitation (single bubble) and by acoustic cavitation (cloud of bubbles) in water. Intermetallic crystals were chemically extracted from an Al-3 wt% Zr alloy matrix. Mechanical properties such as hardness, elastic modulus and fracture toughness of the extracted intermetallics were determined using a geometrically fixed Berkovich nano-diamond and cube corner indenter, under ambient temperature conditions. The studied crystals were then exposed to the two cavitation conditions mentioned. Results demonstrated for the first time that the governing fragmentation mechanism of the studied intermetallics was due to the emitted shock waves from the collapsing bubbles. The fragmentation caused by a single bubble collapse was found to be almost instantaneous. On the other hand, sono-fragmentation studies revealed that the intermetallic crystal initially underwent low cycle fatigue loading, followed by catastrophic brittle failure due to propagating shock waves. The observed fragmentation mechanism was supported by fracture mechanics and pressure measurements using a calibrated fibre optic hydrophone. Results showed that the acoustic pressures produced from shock wave emissions in the case of a single bubble collapse, and responsible for instantaneous fragmentation of the intermetallics, were in the range of 20–40 MPa. Whereas, the shock pressure generated from the acoustic cavitation cloud collapses surged up to 1.6 MPa inducing fatigue stresses within the crystal leading to eventual fragmentation.

This study aims to determine the most suitable flow metering device for the characterisation of heavy duty diesel injector behaviour. The study focuses on three commercially available metering devices and the main principles they employ. An experiment was carried out to benchmark the performance of each device’s measurement repeatability in the characterisation of fuel injector behaviour. This study then compares the capabilities and suitability of each for use in a production environment. The comparison was carried out for Delphi Technologies using the new DFi21 heavy duty diesel injector which uses the miniaturised hydraulic three way control valve technology.

There has been a lot of work done on the analysis of Gaussian loading analysis perhaps because its occurrence is more common than non-Gaussian loading problems. It is nevertheless known that non-Gaussian load occurs in many instances especially in various forms of transport, land, sea and space. Part of the challenge with non-Gaussian loading analysis is the increased number of variables that are needed to model the loading adequately. Artificial neural network approach provides a versatile means to develop models that may require many input variables in order to achieve applicable predictive generalisation capabilities. Artificial neural network has been shown to perform much better than existing frequency domain methods for random fatigue loading under stationary Gaussian load forms especially when mean stress effects are included. This paper presents an artificial neural network model with greater predictive capability than existing frequency domain methods for both Gaussian and non-Gaussian loading analysis. Both platykurtic and leptokurtic non-Gaussian loading cases were considered to demonstrate the scope of application. The model was also validated with available SAE experimental data, even though the skewness and kurtosis of the signal in this case were mild.

The use of artificial intelligence especially based on artificial neural networks (ANN) is now prevalent in many fields of data analysis and interpretation. There have been a number of papers published in the literature on the use of ANN for fatigue characterisation. Most of these have however been developed for rather focussed application with limited capability for fatigue life prediction for a broad scope of material and loading conditions. The authors recently presented a uniquely generalised ANN model that is capable of making fatigue life prediction for a broad range of material fatigue properties and loading spectral forms. The model was developed using simulated data albeit subject to conceivable constraints between possible materials properties and load forms. This paper presents a validation of the ANN model using a Society of Automotive Engineers (SAE) random fatigue loading experimental test data. The capabilities and potentials of the model are demonstrated by comparison with the SAE random load fatigue test results and with results obtained from other predictive methods. The performance of the ANN is highly encouraging as a general tool for random loading fatigue analysis.

The exposure of pistons to extreme mechanical and thermal loads in modern combustion engines has necessitated the use of efficient and detailed analysis methods to facilitate their design. The finite element analysis has become a standard design optimisation tool for this purpose. In literature two different approaches have been suggested for reducing the geometry of the cylinder and crank slider mechanism, to idealise piston finite element analysis load models,whilst trying to maintain realistic boundaries to obtain accurate results. The most widely used geometry is the combination of piston and gudgeon pin while the second geometry includes some portion of the connecting rod’s small end and cylinder in addition to the piston and gudgeon pin.No clear analyses have been made in literature about the relative effectiveness of the two approaches in terms of model accuracy. In this work both approaches have been carried out and analysed with respect to a racing piston. The results suggest that the latter approach is more representative of the load conditions that the piston is subjected to in reality.

A novel test rig for bending fatigue test that based on specimen resonant behaviour has been developed. Determining bending fatigue properties of polymer materials with the standard test systems is challenging, and in some cases results are unattainable. This is particularly true of polymers that exhibit a high level of non-linearity and large deflection. This novel test setup is similar to that of four point bending arrangement resulting in a simple support. The loading is achieved by inertial effect of small masses mounted on the test specimen. A vibration shaker is used to base excite the specimen at the first resonance frequency until it breaks. The proposed test setup reduces the time taken to obtain Stress v/s number of cycles (S/N) curves, typically 1/10th of the universal testing machine based approach. The effect of nonlinearities can be reduced by application of larger loads at higher frequencies using large acceleration and smaller deflection combination. The results based on the proposed approach are in good agreement with tensile fatigue results. It has been successfully used to determine the bending fatigue properties of Polycarbonate (PC) of which determining the tensile fatigue properties were difficult to obtain. The significance of this novel test rig is that it accelerates the fatigue testing and allows the determination of the fatigue properties of some materials that cannot be obtained with existing systems.

The objective of this paper was to model random vibration response of components of an automotive lamp made of Polycarbonate/Acrylonitrile Butadiene Styrene (PC-ABS), Polymethyl methacrylate (PMMA) and Polypropylene 40% Talc filled (PPT40) materials using a nonlinear hyperelastic model. Traditionally, the Rayleigh damping matrix used in the dynamic response analysis is constructed considering linear elastic behaviour based on either initial stiffness or secant stiffness. The performance of linear stiffness matrices is compared in this work with that based on the nonlinear hyperelastic, Mooney-Rivlin model, specifically addressing Rayleigh damping matrix construction. The random vibration responses of 10 samples of each material are measured. The mean square error of acceleration response was used to assess the effectiveness. Considering three materials of study, the hyperelastic model resulted in the reduction of the least square error at best by 11.8 times and at worst by 2.6 times. The Mooney-Rivlin material model based Raleigh damping matrix was more accurate in modelling the dynamic behaviour of components of nonlinear materials and it also represented the manufacturing variabilities more reliably. 

many structural and machine components. The analysis of these types of problems is often

carried out using either time domain or frequency domain methods. Time domain rainflow

counting together with Miner’s linear damage accumulation assumption is widely accepted as

a method of rationalising stress amplitude and mean stress from random fatigue loading and

the damage caused to the component. Frequency domain methods provide a faster alternative

for the analysis of the same problem but the results are generally conservative compared to

those obtained using time domain methods. This paper presents an artificial neural network

(ANN) machine learning approach for the prediction of damage caused by random fatigue

loading. The results obtained for ergodic Gaussian stationary stochastic loading is very

encouraging. The method embodies rapid analysis as well as better agreement with rainflow

counting method than existing frequency domain methods.

The objective of this article is to introduce and assess a new plate buckling analysis procedure which can be used for quick, approximate analysis of buckling loads in preliminary design. The method is applied to a range of plate edge support condition combinations including many where results are not readily available. The results obtained using the new procedure were compared against theoretical formulae available in the literature and by finite element analyses with good agreement

A series of carbon fibre laminate bearing versus bypass load tests were carried out investigating the effect of hole clearance, laminate lay-up, washer contact size and clamping force value. Explanations of the underlying effects that influence the results are discussed. The results compare well with literature but hole clearance was shown to increase joint strength when bypass loads are dominant, which seems counter intuitive, and has not been reported in literature

It is shown that the reduced order model decreases the simulation time by at least one order of magnitude, while maintaining the desired accuracy of the state variables under investigation. The method could prove as a valuable tool to analyze complex coupled structural–acoustic systems, and their subsequent optimization or sensitivity analysis, where, in addition to fast analysis, a fine frequency resolution is often required.

Adhesively bonded lap shear joints have been investigated widely and several ideas have been proposed for improving joint strength by reducing bondline stress concentrations. These include application of adhesive fillets at the overlap ends and use of adhesive with graded properties in the overlap area. Another, less common, approach is to deform the substrates in the overlap area in order to obtain a more desirable bondline stress distribution. Previous work carried out by the authors on a number of different substrate materials indicated that a reverse-bent joint geometry is useful for increasing joint strength. Results from static stress analysis and experimental testing demonstrated that significant improvements could be achieved. This paper presents results of further work carried out to assess the fatigue performance of reverse-bent joints. Substrates with different yield and plastic deformation characteristics were used and the effects of different overlap lengths on strength were examined. The results of this research show that the improvements obtained under static tests conditions translate to even higher benefits in fatigue. The paper also explains the failure mechanism of the joints under fatigue loading.

Adhesive bonding is used increasingly by the automotive industry to join structural components of metallic and composite materials. The most common joint configuration is the lap-shear joint, which has been investigated widely and several ideas have been proposed to improve its performance. For example, the introduction of fillets at the overlap ends or tapering of the substrates can reduce the peel stresses at the overlap end. Other approaches include the use of"reverse-bent" substrates and"wavy joints" to reduce peel stresses. This paper compares the stress distribution of the"reverse-bent" and the"wavy joint" , with the stresses of the traditional lap-shear joint, using finite element analysis (FEA). A parametric study was carried out showing trends influencing stresses in the adhesive layer. Experimental tests were conducted to evaluate the assumptions used in, and the findings of, the FEA. The joint strength of"reverse-bent" joints was found to be up to 40% higher compared to flat joints using various substrate materials, adhesives and overlap lengths.

An inverse artificial neural network (ANN) assessment for locating defects in bars with or without notches is presented in the paper. Postulated void defects of 1mm x 1mm were introduced into bars that were impacted with an impulse step load; the resultant elastic waves propagate impinging on the defects. The resultant transient strain field was analyzed using the finite element method. Transient strain data was collected at nodal points or sensors locations on the boundary of the bars and used to train and assess ANNs. The paper demonstrates quantitatively, the effects of features such as the design of ANN, sensing parameters such as number of data collection points, and the effect of geometric features such as notches in the bars.

A generalisation of McLean's (Directionally Solidified Materials for High Temperature Service, 1983, The Metals Society, London) uniaxial creep model which accounts for the effects of inherent residual stresses, matrix primary creep and the treatment of the effect of possible instantaneous matrix yield upon loading is presented. The model and numerical micromechanics approach are used to analyse the uniaxial creep of Ti-6AI-4V/SiC metal matrix composites. Excellent agreement is obtained between the analytical and numerical models and predicted results are in reasonable agreement with published experimental results.