13th September 1999
At a meeting in Sheffield organised by Prof. John Beynon in October 1998, concern was expressed about the health of materials process modelling in the UK, in particular the state of education and training of researchers in the techniques and application of modelling. A working group was formed to decide what should be done to address the problem, and to consider an EPSRC Network in this area. Members of the working group were:
Prof. John Beynon (Sheffield)
Dr. Esteban Busso (Imperial)
Dr. Paul Buckley (Oxford)
Prof. Alan Cocks (Leicester)
Dr. Hugh Shercliff (Cambridge)
Members of this group had met three times, to maintain the debate and to organise this discussion workshop. It was recognised that "Materials Modelling" was very broad – hence the focus on "Constitutive Modelling for Materials Processing", being a sub-area within the overall field for which a need has been clearly identified, and which commands direct relevance for UK industry.
Dr Hugh Shercliff (University of Cambridge) opened the workshop with a review of the status of UK materials modelling in the context of this meeting, in order to explain the purpose of the workshop and pose a number of points for discussion.
Defining the "UK Materials Modelling Group"
The breadth of "materials modelling" can be seen by considering the many different modelling communities:
It was noted that modelling communities should include more than modellers:
In addition it was noted that UK academic and industrial modellers collaborate widely in Europe and worldwide.
A major purpose of the day was therefore to define the remit of a possible Network in "Constitutive Materials Modelling" and to set this within the wider UK materials modelling community.
Status of UK materials modelling
A recent review "Modelling of Materials and Processes" provided a number of starting points for this workshop. The review, published in June 1997, was a response to Technology Foresight for the OST & Institute of Materials (available via the Web:
http://www-materials.eng.cam.ac.uk/hrs/publications/modelling-report.html).
Key points from this review were summarised under three headings:
(1) People and Training
There is a continuous shortage of well-trained materials modellers, in both academia and industry.
Modelling is a key area to develop in degree courses at undergraduate and postgraduate level.
A major factor in training modellers is establishing the right attitudes:
(2) Model building and validation
Computer power is absorbed far too readily in added complexity, rather than in thorough testing and application of existing models.
FE methods are becoming routine tools at continuum level. Major opportunities are now:
Software engineering developments will be needed alongside materials modelling development.
The data needs of any modelling activity are paramount.
(3) Technology transfer
In the UK, industrial awareness/takeup of modelling is very non-uniform:
Academic modelling too rarely leads to usable software for industry, but software development should largely be the role of intermediate research organisations or spin-off companies.
Centres of excellence are needed to offer impartial advice and training in modelling, and for benchmarking software.
Modelling centres should be closely integrated with practitioners and experimentalists (test facilities, pilot-scale plant, real control systems etc).
Programme for the workshop
The invited talks for the workshop were intended to illustrate the following underlying types of activity:
To explore the possible synergy across different material sectors, the talks also fell into 3 areas which were considered to provide current and future challenges:
Professor Norman Fleck, University of Cambridge
Yielding of metal powders and metal foams
Professor Fleck described the flavour of his work as the combination of experiments and micromechanical models to produce practical constitutive laws. The types of model employed include: finite element, discrete element, bounds and estimates with analytical approximations. All are forms of homogenisation from the local behaviour to the macro-level response. Professor Fleck gave examples in: (i) powder compaction, where he noted that some issues required different sub-models for various regimes, with checks and smooth handing over from one sub-model to another (an issue connected to the appropriate level of simplification); and (ii) metallic foams, where the micromechanical model was insufficient to cover the whole deformation range, and the modelling shifted to a continuum mechanics approach with a fitting parameter. A micromechanical model based on Voronoi meshing provided a reasonable representation as well as insight, particularly on the effect of flaws in the structure.
Professor Howard Chandler, University of Aberdeen
Cyclic deformation of stiff ceramic pastes
This work arose out of practical problems, for example, the difficulty of extruding ceramic pastes, such as alumina and water mixes. Although additives can facilitate such extrusion, the mechanism by which they facilitate this is not understood. Two views prevail, based on chemical or mechanical explanations. Professor Chandler proposed that one way of dealing with such differences is to attempt to model extraordinary behaviour in a simple manner, rather than an expensive, detailed, systematic approach. The example he chose was the cyclic deformation of stiff ceramic pastes whereby a mechanical explanation could explain the response to small strain reversals, using arguments similar to those used in soil mechanics. Professor Chandler did consider the possibility of this modelling being applicable to the thixotropic forming of metal slurries, although one difference is that Professor Chandler’s work is largely rate insensitive: he deals with strain and small deformations, not typical of viscometry, which imposes large deformations.
Dr Paul Buckley, University of Oxford
Constitutive modelling of polymers in solid-state forming processes
Dr Buckley considered wholly non-crystalline polymers whereby forming is in the amorphous state. The range of behaviour spans viscous, elasto-viscous, hyperelastic, viscoelastic and elastic-plastic. Since structure only appears at a molecular scale, Dr Buckley does not use a micromechanical approach, rather he considers the free energy of the material. Dr Buckley stressed that such modelling is still unresolved with different approaches in use, with issues such as the onset of crystallisation still being resistant to accurate modelling. Further issues subject to future research are the effects of strain path change and strain rate change tests.
Dr Helen Atkinson, University of Sheffield
Thixotropic forming
Dr Atkinson began by demonstrating everyday examples of thixotropic behaviour with mascara and ketchup. She stressed some distinctions with the paste behaviour described earlier by Professor Chandler, namely that in thixotropic metals there is some bonding between the solid particles, and that the particles coarsen as they are held at high temperature. Successful thixotropic forming depends on achieving laminar flow throughout the moulding. At rest, metals in the slurry state have a viscosity greater than 106 Pas, but this drops very rapidly (within about 0.1s) to 103 Pas upon deformation. This value of 103 Pas is still well above the steady state value which would be measured by standard rheometry. Mechanisms occurring at the microstructural scale within thixotropically formed metal have been identified but are still not fully modelled. It is difficult to measure such viscosities and the normal tests for these materials, namely compression between flat platens, itself requires modelling so that material properties can be extracted from the test data. Dr Atkinson is working with a computational fluid dynamicist to develop a model for thixotropic forming, one of many examples during the day of a multi-disciplinary approach to this area of work.
Dr Daniel Read, University of Leeds
Finite element simulation of the mechanical responses of spinodal structures
Phase separation in some polymers (i.e. co-polymers) gives a structural scale which is larger than the molecule, and is more akin to crystalline microstructure scales, though without sharp boundaries between the phases. Dr Read showed a modelling technique based on assigning different material behaviour to the two phases and modelling the composite’s behaviour. Currently this work has been limited to the elastic state, but it is early days.
Dr John Sweeney, University of Bradford
Modelling solid phase deformation processing of polymers
The application here involves large deformations with controlled necking, which need to be modelled. The approach taken by Dr Sweeney is to use the network model by Ball et al. implemented in commercial finite element code. He stressed the need to account for the rate sensitivity and demonstrated that the modelling is good for predicting the final shape, though still less satisfactory for predicting the intermediate deformations. His example was the drawing of a grid to provide a civil-structural mesh.
Dr Fionn Dunne, University of Oxford
Modelling superplastic deformation for process simulation
Dr Dunne began with a description of the superplastic forming process for the titanium alloy Ti-6Al-4V, undertaken around 9000C with a maximum strain rate limitation to ensure superplasticity throughout the forming operation. Dr Dunne then explained the microstructural mechanisms involved, including the sensitivity of superplasticity to grain size. This led into the model description based on distributions of state variables and following the evolution of grain size. The results of the calculations showed a good fit with experimental results, including where grain size evolved during deformation. The next challenge is to develop models to capture microstructural inhomogeneity which can lead to necking. Also needed is a better description of friction between the workpiece and the tools.
Dr Esteban Busso, Imperial College
Continuum mechanics based modelling approaches for coupled multi-physics processes in solids
Dr Busso suggested that the terminology "multi-physics" can be defined differently by different communities. He also considered the range of length scales addressed, from defect mechanics (e.g. dislocations) through micromechanics and mechanics, to continuum mechanics. He described the construction of a model as involving: identifying the physical phenomena, developing an analytical formula, obtaining relevant data, and developing a predictive model. He showed examples for residual stresses in microchip manufacture, nucleation of cavities on stressed interfaces, and degradation of ceramic-metal interfaces for thermal barrier coatings. The physically based, internal state variable models developed by Dr Busso have been implemented into commercial finite element code, i.e., a continuum mechanics approach. Dr Busso concluded that continuum mechanics offers a powerful framework to model/predict the micromechanisms of deformation and fracture of materials, and interfaces involving multi-physics processes. Crucial to this is the understanding of the physical phenomena, development of multi-scale links, coupling (explicitly) of the multi-physics processes, development of robust implicit algorithms, and formulation of user-friendly predictive models for process optimisation and life prediction.
Professor Tom McLeish, University of Leeds
Multiscale constitutive modelling for branched polymer processing
Professor McLeish stressed that to analyse polymer melts, considerable simplification is needed to move from molecular rheology up to production scale. He also stressed that a multidisciplinary approach was essential and proceeded to demonstrate this by explaining the broad collaboration across many universities of which he is part. In a similar manner to Professor Chandler’s presentation, Professor McLeish highlighted an example of peculiar behaviour which current models have been unable to explain, proposing it as a good test case for the development of an improved model. The example was the very different behaviour of high-density and low-density polyethylene in extrusion, despite very similar mechanical and physical properties. He explained that although mean field theory can be used because of the large number of interactions of each polymer chain with its neighbours, linear and long-branched polymers ("star" polymers) behave in very different ways because of the constraints on their behaviour. The explanation proposed by Professor McLeish was based on the "tube model" for reptation. Professor McLeish stressed the value of model materials in bridging between theory and industrial processing and praised the efforts of collaborators in achieving highly controlled polymer melts. This had led to the development of a molecularly derived constitutive equation, "the Pom Pom model", based on the idea that a minimum of two branch points are needed per molecule for behaviour which is significantly different to a linear polymer. Such a model has been able to explain previously puzzling viscosity observations, such as differences under shear and stretch loading conditions. He concluded his presentation with video and computer graphic demonstrations of the model predictions.
From his overview, Dr Shercliff summarised a number of discussion points under 4 headings:
Defining the "UK Materials Modelling Group"
What "sectors" of materials modelling could be brought together within a network, or series of networks?
Can we bridge effectively between materials and between process sectors?
Is there enough interaction between the UK and Europe, and worldwide (especially USA)?
People and Training
Do current courses deliver the type of inter-disciplinary skills needed for academic and industrial modelling?
Are students and post-docs exposed to different modelling strategies and given hands-on opportunities with different types of software?
How could education in modelling be enhanced by networks?
Model building and validation
Do too many modellers tackle the next big challenge, without thoroughly evaluating their current capabilities?
Where are the best opportunities for integrating FE analysis with microstructural and micromechanical modelling ?
Are academic modellers sufficiently aware of the demands of industrial data transfer, software engineering and the development of standards?
In what areas do modellers feel the greatest lack of data?
Technology transfer
Is the current level of interaction between academia and industry sufficient in materials modelling research?
Are we doing enough to demonstrate the importance of modelling and to increase its uptake in all industrial sectors?
Centres are likely to be specialised - casting, polymer moulding etc. Could networks serve to cross-fertilise between these centres?
A range of comments was made, in response to the prompts above.
Theme-based groupings were recommended for a network – several generic problem areas had come up during the day, which crossed several materials. The researchers involved had not previously been aware of the other work, but were excited about forming links across the conventional material boundaries. Successful instances were cited of such "lateral thinking" – e.g. between soil mechanics and powder processing.
Networking within the UK should be the initial focus, but links to Europe and further afield should be developed once a UK group was established. Alternative sources of funding could be sought to develop this dimension.
Many "materials modelling groups" are known to exist within the UK. Care will be needed to define the remit (and title) of a network clearly, and to encourage synergy with related modelling activities (such as the existing polymer network, and a proposed "materials microstructure" activity).
People exchange was widely supported as the most important activity, particularly at RA level. Concern was also expressed at the lack of motivated PhD students coming through the system to do research in materials generally, and at the difficulties caused by funding agencies not providing for the cost of programmers.
Coupling modelling to FE was considered to be an essential area for a network to address, in various aspects: how to condense complex behaviour into a form suitable for embedding in a reliable UMAT, running and testing such UMATs, and linking FE experts to microstructural and micromechanical modellers.
Industrial collaboration was strongly encouraged. It was noted that industrial backup is very fluid (due to large company takeovers) so the academic community had a very important stabilising role in maintaining long-term development.
Interaction with SMEs was considered to be a role for centres in given process sectors. Support for SMEs appears to be shifting to regional development coordinated by the DTI, but it was noted that it is difficult working with SMEs when many are simply concentrating on survival.
Suggested themes are currently:
Further comments on the Workshop and this summary may be addressed to:
Dr Hugh Shercliff (University of Cambridge): hrs@eng.cam.ac.uk
Suggestions and expressions of interest in the Network may be addressed to:
Prof Alan Cocks (University of Leicester): acfc1@leicester.ac.uk
| Name | Affiliation |
| Prof. Howard Chandler | University of Aberdeen |
| Dr Alfred Akisanya | University of Aberdeen |
| Mr Gary Menary | Queen’s University, Belfast |
| Dr Peter Martin | Queen’s University, Belfast |
| Prof Cecil Armstrong | Queen’s University, Belfast |
| Dr John Sweeney | University of Bradford |
| Prof John Whiteman | Brunel University |
| Dr Hugh Shercliff | University of Cambridge |
| Prof Norman Fleck | University of Cambridge |
| Prof Mike Ashby | University of Cambridge |
| Dr Malcolm Bolton | University of Cambridge |
| Dr Michael Sutcliffe | University of Cambridge |
| Dr Mark Roberts | University of Cambridge |
| Dr. David Knowles | University of Cambridge |
| Dr Esteban Busso | Imperial College |
| Dr. Noel O’Dowd | Imperial College |
| Dr. Pat Leevers | Imperial College |
| Dr Alojz Ivankovic | Imperial College |
| Prof. V. Gambin | Imperial College |
| Prof. Tom McLeish | University of Leeds |
| Dr Alan Duckett | University of Leeds |
| Dr Paul Unwin | University of Leeds |
| Dr Daniel Read | University of Leeds |
| Dr Anne Whitehouse | University of Leicester |
| Dr. Bob Wood | University of Loughborough |
| Dr Paul Buckley | University of Oxford |
| Dr Fionn Dunne | University of Oxford |
| Prof PhilipWithers | University of Manchester |
| Prof. David Hayhurst | UMIST |
| Prof. T. Hyde | University of Nottingham |
| Prof John Beynon | University of Sheffield |
| Dr Ian Howard | University of Sheffield |
| Prof. Geoff Greenwood | University of Sheffield |
| Dr Helen Atkinson | University of Sheffield |
| Dr J. Pan | University of Surrey |
| Dr George Peric | University of Wales, Swansea |
| Dr Peter Smith | EPSRC |
| Dr Peter Morris | British Steel |
| Bruce Adderley | British Steel |
| Dr Keith Waterson | Alcan International |
| Dr Steve Rogers | Alcan International |
| Dr Jean Savoie | Alcan International |
| Dr George Durrant | Rolls-Royce |