The COLLYEAR Report - Some personal comments

Philip Sargent
5 May 1985

I found this interesting if mildly depressing reading. Interesting because it had assembled in one place quite a lot of the currently 'hot' topics, and depressing because (a) it didn't speculate on some really specific uses that they might lead to, and (b) because with the picture of industry as painted, it doesn't really seem likely that industry will generate the required 'pull'. By which I mean that it doesn't look as if any amount of money spent on R&D in HEIs and government labs. will actually stir industry out of its reluctance to use NIMPs. The report states (5.1) quite clearly why companies look on NIMPs as a big gamble, but then concentrates too much on the research areas rather than on this single, majore and overwhelmingly important point. Section 5 is entirely devoted to this point, but it gets a bit lost amongst the rest of the report. Only section 5 is important, the rest is interesting speculation for those in materials research and those planning research programmes, but as far as government and policy goes it is too specifically technical.

In the introduction, (2.6) it says "this potential for technical change and competitive advantage will be realized only by using revolutionary changes in engineering design and production technology (my emphasis). This should be copared with (5.5): "The effective acceleration of the application of NIMP will result only if there is active and enthusuastic support by UK manufacturing industry... the real test will come when investment committment becomes an issue". This seems to be a very difficult thing to encourage, all before there is any return on the upheaval and investment involved. The education, training, new design codes and methods and longer term research (5.12 to 5.21, and 5.25) are all absolutely necessary, but the motivation is central. Perhaps it is a matter of confidence; that all this money and effort will actually produce marketable products whcih the industries can use before someone else does.

One point which came out extremely clearly int he Fellowship of Engineering's report on this subject was the absolutely crucial matter of a materials' supplier collaborating with an end-user when developing a new material. This is given a cursory mention in Collyear section 5.7, but it's not made clear in the report that NIMP is a single area, the new materials only appear as a result of new processes: the processes are in the hands ofthe suppliers and the specification is in the hands of the end users. I consider thsi to be a grave omission, particularly since I know only too well that ceramics (which take the starring role in 3 of the 7 recommended 'club' areas (5.6), and a subordinate role in 3) are desperately dependent on the precise production method for almost all their useful properties. I am assured by the other members of the research group that this is also true of fibre composites.

With the above reservations on the writing and presentation of the report, the substane of its recommendations (section 7) on the formation of R&D clubs seems useful (even if many British ceramics researchers have gone to the States, or left the field entirely). The only way to get the confidence that a leading edge has been obtained is if the clubs know for sure that they are up with the current state of play in the world in their development area, and are assured of steady support (money) until theyu can bring the product tomarket. British companies are mostly too small to keep up the continual research effort which is required even to be able to evaluate the current state of other companies developments.

There are two further points whcih are not made int he report. First, that the NIMP 'revolution' is not unique; the IT revolution is more immediate and both are concerned with companies adopting new methods. The problems of getting new developments underway in any company is really no different in NIMP than in anything else. Second, because time-scales are so long, many companies really can afford to ignore materials developments when they can't ignore IT . (By which I mean office automation and CAD in the short term, and computer aided manufacturing in the not quite so short term). This means that there is a shortage of active people in companies who are prepared and willing to organise 'change' in general, and they are all doing CADCAM, not NIMP . A  very large proportion of people who graduated with me in materials science, and perhaps an even larger proportion of those friends who did PhDs in materials at the same time as me, now work in software because there haven't been enough interesting and challenging jobs in materials over the past 5 or 6 years.

Materials Information Processing - a common thread in the report

The report makes a number of references to this field and seems very ill-informed about it. I thought that it would be useful to draw them all together because they have a lot in common and are potentially more important than the individual materials development ares that the report concentrates on. More important especially with reference to industry 'take-up' and to the important point of reducing time scales from the stultifying 10 to 20 years, whcih seems to be a recipe for government to do nothing urgently.

First, sections 3.2 and 3.3 describe why NIMP investment is inhibited, and section 3.5 describes three essential courses of action: reduction in time scales, awareness and expertise transfer, and collaboration and coordination. It seems to me that IT is vital to the latter two, and as far as the first is concerned, no recipe exists for encouraging it directlu; it can only be approached through the other two. Second, timescales are inherently shorter for implementing IT, so to get the programme of the ground in order to keep enthusiasm high (cf 5.5), some effort in materials and processes information handling seems to be indicated. Since IT penetrations in these areas has been small, it could achieve significant results quite cheaply. As an example, the current high temperature creep limited design methods can only be termed squalid from a materials engineering and information handling point of view.

The Collyear report mentions a few information technologies, a rather curious ragbag whcih illustrates that the committee was somewhat lacking expertise or interest in this area. They are: flexible robotic systems for tape-winding (complex structures from fibre composites) and laser welding (laser surface treatments), use of CADCAM for die and mould manufacture (and nothing else), database standardisation and verification, and a throwaway line about expert systems advice for the general user in manufacturing industry. These are all described in section 7.6 as 'proposals best implemented by the DTI through relevant sponsoring divisions and IREs' .

The robot systems are a very similar technology to those currently well established for painting and welding, and while new robotic methods have implications for some materials production methods, I would say that fibre lay-up systems are a very bad example for longer range studies: the USAF have an immense programme in this area for automated manufacture of carbon-fibre wing structures. There is an enormous difference between standard robots and the same with some visual ability. This is a large area, well covered by AI research and R&D companies such as IBM. Die and mould CADCAM packages are probably available almost off the shelf these days: in 1984 and 1985 CAD workstations underwent almost monthly revolutions because they are driven by the semiconductor industry's requirements. This is an area where the speed of development is not suited to the general pace of the DTI. The last three points on database are an entirely different order of problem, and it is very worrying that the Collyear committee does not seem to realise this.

An expert system for 'the general user in manufacturing engineering' for general materials selection or materials processing technique selection is so far away that it is dangerously misleading to talk about at the same time level as the other points. If Alvey launched a multi-million large scale demonstrator project on thsi specific problem, it would be optimistic to look for a prototype working system within four years. Collyear report section 5.20 describes, in its last sentence, a research programme larger than everything else int hereport put together: 'Provision should be made for integrated computer systems, encompassing materials information storage and retrieval, engineering design, materials processing and component manufacture, in programmes on advanced manufacturing technology being developed by the Department.' That is a worthy aim, and extremely important, but the danger is that too much will be attempted at once and nothing will be achieved. The database problem is a good place to start.

Materials Databases

The next worrying thing is that the report implies that the databases and design methods for metals are well-established and 'available'. This may be true in one sense, but false if you add the rider that that the data obtainable is also correct and useful. The Institute of Metals meeting on materials databases (May 12th, 1985) may have given to the outsider the impression that a lot is going on in this field, but there is no concrete effort in data validation, and no effort in storing physically meaningful data, rather than superficial lists of numbers. The problem is that there is money to be made selling bibliographic materials information, and in selling cute expert systems tailored to single, small areas, but not in producing standarsd, design methods and most importantly modelling procedures. It is also a well known fact that keeping any database validated takes more time, effort and software than reading it and updating it put together.

Materials data is not useful in isolation, it becomes useful when it is applied to (1) a design specification or (2) a mechanistic model whcih attempts to predict materials properties so that a new design can be made. Standards for materials properties are universally used and any database will have to take care of all the possible measurements that have been made. The problem here is that of two types of standard conversion: first, of using (say) elongation to failure data in the database when looking for a material with a high deep-drawing ratio, and second, of using DIN when looking for ISO standards. This is definitely an area for development whcih is eminently suitable for the DTI to fund because it aids all industry who have access to it in a non-competitive manner. However, there would be problems for materials whose properties are intimately tied up with their production processing, and some companies might wish to add confidential data to the uniform data structure whcih would then only be accessible by them. Some kinds of standars in data representation would also need to be developed. It might be relevant to point out here that the general 'problem' of constructing databases in general is a highly active and rapidly advancing area of computing science, and a materials database is a new variety which may well need original research in its development (Byte, April 1985, 10, 389-397).

All standard information processing is, almost without exception, completely useless for modelling. This fact needs to be fully appreciated before the limitations of hte database described above become apparent. The implication is that no automatic design aids can use this 'standards database' without including a very great deal of metallurgical 'knowledge' to interpret the data (which are just experiemntal measurements) into more fundamental information. This is not really a sensible approach, what is required is that modelling information is also included in the database (or in a separate database). This would include things like phase diagram information, thermal conductivity, diffusion coefficients, crystallographic information etc., as well as crack opening displacement (cod) and room temperature yield stress. There has been great progress in materials engineering in the past 15 years in understanding fracture toughness, fatigue crack propagation, creep deformation and stress relaxation, and too little of this body of knowledge is available to engineers or incorporated into design codes. With modelling methods utilising what is currently well known about materials and with a comprehensive materials properties database, very slightly different new materials and processes may be developed whcih nevertheless offer significantly improved properties and economics. Because the information which currently exists is put to better use, and because it will show how only slight differences to current plant can bring improvements, this seems to be a possible development route which is both rapid and cheap.

Dr. Philip Sargent
5 May 1985

Materials Group
Department of Engineering
University of Cambridge

Acronyms

DTI    Department of Trade and Industry
HEI    Higher Education Institute
IRE    Industrial Research Establishments
CADCAM    Computer Aided Design & Computer Aided Manufacturing
IT    Information Technology
NIMP    New and Improved Materials and Processes
DIN    Deutscher Industrie Norm
ISO    International Standards Organisation


The Collyear Report, influential in its time, has now largely vanished from memory and is only available in print form in official publications libraries.
"...the submission in 1985 to the Department of Trade and Industry of the Collyear report. the Collyear Committee proposed a 5-year, £120 million program for the wider application of new and improved materials and processes. " Materials Science and Engineering for the 1990s: Maintaining Competitiveness in the Age of Materials (1989) , 320 pages, NATIONAL ACADEMY PRESS Washington, D.C. 1989, ISBN 0-309-03928-2.