Structural Performance and Integrity
The Structural Performance and Integrity research team is led by Professor David Nowell, who has over 30 years’ research experience in the field. Professor Nowell leads the overall Dynamics Group, and is Director of the Rolls-Royce University Technology Centre (UTC) in Vibration. From 1999 to 2009 he was Director of the UTC in Solid Mechanics at Oxford.
Professor Nowell’s interests focus on the manner in which structural performance affects mechanical integrity and vice-versa. Current research projects are listed below, and fall into the following general areas:
- Behaviour of frictional interfaces
- Fretting fatigue
- Fatigue response, including multiaxial fatigue, short cracks and life prediction methods
- Experimental techniques in solid mechanics, including digital image correlation (DIC) and FIB-DIC for residual stress measurement
- Applications of machine learning in solid mechanics
In addition to his research work, Professor Nowell undertakes a wide range of consultancy with industry. Current and recent customers include Rolls-Royce plc; London Underground; Voith Hydro; Mercedes Formula 1; Deeside Power; and Equinor. In 2013, he acted as an Assessor, assisting the judge interpret technical evidence in the AS$ 2 billion Kilmore East-Kinglake Bushfire case.
Professor Nowell is Editor of the Journal of Strain Analysis and is a Trustee of the Institution of Mechanical Engineers. He is a visiting professor at the University of Oxford. He joined Imperial College in 2017 and his inaugural lecture may be viewed on YouTube.
Lu Yin’s PhD project aims to combine two aspects of the complex problem of brake squeal. Often work in this area is undertaken either from a dynamics perspective, using a simple model for friction, or from a tribological perspective with little consideration of system dynamics. By combining these two aspects, Lu’s work hopes to shed new light on a capricious and complex phenomenon, which affects many frictional braking systems.
- A pin on disk rig has been designed and measurements of squeal have been made.
- A finite element model has been developed and predictions of squeal frequencies have been made using complex eigenvalue analysis.
- Further work is taking place to understand the complex nature of the frictional force between the pin and the rotating disc, and its variation with temperature and with wear.
David Nowell has been involved in research in fretting fatigue for many years. This is a complex phenomenon which can occur in engineering systems when oscillatory forces are transmitted between components. The resulting high stress and surface damage can combine to substantially reduce fatigue performance. Most of David’s fretting work was conducted at Oxford, but a new rig is currently under construction which will enable a series of fretting fatigue experiments to be conducted at Imperial for an industrial customer.
- A neural network approach has recently been used to analyse historical fretting fatigue data [link to Machine Learning project above].
- A fretting fatigue rig is being manufactured for use on the biaxial fatigue machine at Imperial College.
- David Nowell has been collaborating with the University of Sheffield (Prof. Luca Susmel and Cedric Kouanga) in carrying out a series of fretting fatigue tests for Cummins UK.
Machine Learning Applications in Solid Mechanics
Bemin Sheen; Peter Nowell
Machine learning is starting to become a useful tool in a wide range of engineering applications. However, its use in solid mechanics is relatively novel. A number of preliminary investigations have been carried out which illustrate how neural networks, in particular, may be useful in interpreting experimental data and in predicting stress distributions in different geometries of components. More substantial research projects in this area are now under consideration.
- An investigation has been carried out concerning the use of machine learning to predict fretting fatigue life. The results have been published in Tribology International (Nowell, D. and Nowell, P.W., ‘A machine learning approach to the prediction of fretting fatigue life’, Tribology International, 141, 105913, 2020).
- Bemin Sheen’s final year undergraduate project has successfully demonstrated that a neural network machine learning approach may be applied to the prediction of stress distributions in a number of simple situations. The technique is particularly useful for ‘what-if?’ design assessments.
Measurement of micro-scale residual stress
Rodolfo Fleury (University of Hanover); Jacob Schneider-Martin (Medtronic)
Many engineering structures and systems are affected by significant levels of residual stress, present when the component is unloaded. These may affect the performance and life of the structure when it is in operation, and there is a need to evaluate these in order to take them into account. The FIB-DIC technique is a promising new method, combining Focused Ion Beam milling with Digital Image Correlation. The approach enables measurement of residual stress in very rapidly-varying stress fields, since the diameter of the ring machined in the surface is of the order of microns. The challenge that follows is to interpret the information that is obtained, since the measurement is a combination of macroscopic and microscopic residual strains.
- Rodolfo Fleury’s work has demonstrated the capability of the FIB-DIC process to measure successfully residual stress states in a practical situation (small dents induced by handling damage). Fleury, R.M.N., Salvati E., Nowell, D., Korsunsky A.M., Silva F., and Tai, Y.H., ‘The effect of surface damage and residual stresses on the fatigue life of nickel superalloys at high temperature’, International Journal of Fatigue, 119, 34-42.
- Jacob Schneider-Martin’s MSc project investigated the use of FIB-DIC to measure residual stress in a small medical device (a stent) manufactured from Nitinol. This is a shape memory alloy which can sustain significant plastic strain. The results were reported at the BioMedEng19 Conference in September 2019 (p 181of the conference proceedings).
João Sahadi Cavalheiro; Bemin Sheen
Work is ongoing in the area of multiaxial fatigue. Components in service are quite likely to be loaded in two (or potentially three) principal directions. The complexity of simulating these stress states in the laboratory means that materials data is usually provided only for uniaxial loading. It is therefore necessary to estimate multiaxial fatigue performance from uniaxial data. This can be challenging, particularly where the loads do not vary in proportion or where the principal directions change during the loading cycle. Our research work aims to better understand these phenomena to enable more effective fatigue life prediction under multiaxial loading.
- João Sahadi has recently completed his thesis. This addresses a number of topics, including evaluation of traditional criteria, validation with cruciform and tension-torsion experiments, and development of a crystal plasticity model for multiaxial fatigue.
- Funding has now been obtained for a PhD project which will specifically examine long crack growth under a changing multiaxial stress state. The work will include development of a novel test geometry as well as formulation and validation of an appropriate crack growth criterion.
Rigid body dynamics with frictional interfaces
Underplatform dampers are a key component in reducing vibration amplitudes in gas turbines. Energy is dissipated by friction as the blades vibrate, but since this introduces a non-linearity and the level of friction is potentially quite variable, the performance of the damper may vary from blade to blade. Hence, the ‘best’ design may not be the one which gives the most damping, but rather one which gives a reasonable level of damping and which is robust to variations due to manufacturing tolerances, frictional force, and wear. Myles Kelly’s MSc project produced a simple ‘rigid body’ solver which could be used to investigate damper motion under different combinations of geometry, friction, and load. This provided a useful design tool which can predict behaviour and highlight possible designs worthy of more detailed investigation.
- Myles Kelly has completed his MSc project. The results are being written up as a journal paper.