Engineering with Computers 8, 243-252 (1992)
Engineering C6i puters 9 Springer-Verlag New York Inc. 1992
Engineering Materials Information Systems P.M. Sargent Materials Group and Engineering Design Center, Cambridge University Engineering Department, Cambridge CB2 1PZ, UK
Abstract. Materials properties information systems are poorly understood. Many databases of materials properties and designations have been produced but, except in the most modest of cases, they have been less successful than their creators had hoped. Knowledge based systems (KBS) are subject to exactly the same problems as data based materials information systems and it is important to realize what these are before the special character of KBSs can be used to alleviate them. This paper surveys the unusual and difficult aspects of engineering materials information that must be handled by any organizing methodology, whether manual or computerized, data based or knowledge based, handling information which is stored or inferred.
1
Introduction
Materials information has a n u m b e r of unusual characteristics c o m p a r e d with information from other engineering domains insofar as its representation in c o m p u t e r i z e d s y s t e m s is concerned. T h e y are p e r h a p s not fundamental differences, but while for m a n y engineering p r o b l e m s some simplifications are possible which m a k e data modeling m u c h easier, for materials these approximations often m e a n that the e s s e n c e of the matter has been omitted [1].
2
Materials Information Characteristics
The following five u n c o m m o n matters are characteristics of materials information: o p e n endedness, state and process, abstraction levels, default information and capricious properties. The two points which follow these are also characteristic of materials information s y s t e m s but are also c o m m o n l y found in other engineering domains: data quality and decision support.
Offprint requests: P.M. Sargent, Materials Group and Engineering Design Center, Cambridge University Engineering Dept., Trumpington St., Cambridge CB2 1PZ, UK.
2.1
An Open Ended Domain
Materials information is a thin thread that runs through m a n y different engineering activities: design, manufacture, maintenance, lifetime assessment, disposal, etc. Thus m a n y different communities require materials information, and while it m a y be represented in different w a y s for these different groups, it is often nevertheless the s a m e information in m o s t cases [2,3]. In addition to the n u m b e r of different types of people involved in both using and generating information, there are also a great n u m b e r of different materials. An estimated 80,000 structural engineering materials are in use [4,5], and for m a n y purposes information on obsolete materials m u s t also be retained, whereas for others information regarding materials yet to be fully developed must be estim a t e d [6]. It is also found that information needs are very widely spread, that one material or standard will be used globally in conjunction with other materials d e v e l o p e d independently by quite different organizations. This m e a n s that no industry segment, no country, no continent even, can isolate itself from the worldwide m a i n s t r e a m of materials information [7-14]. T h e s e three factors: people, materials, and geography, each and together imply that any materials information s y s t e m will never be complete: there will always be relevant information that it does not contain b e c a u s e global coordination is just not good enough to k e e p e v e r y o n e up to date. T h e r e will always be new materials, new m e a s u r e m e n t techniques, new materials properties that the original designers cannot anticipate. It therefore means that any realistic materials information system should f r o m the beginning take as a fundamental assumption that to be truly useful, it will h a v e to interwork with other materials informations systems developed b y other groups, and also that the style of that interaction cannot be prespecified b y any individual s y s t e m designer. (Too often, developers construct
244
P.M. Sargent
grandiose schemes which will work for all people everywhere if they only adopt a particular way of dealing with information that the scheme designer thinks everyone should use. This is unworkable and inappropriate.) Materials information is thus characteristically open ended: the closed world assumption is not valid. This means that N U L L values and three-valued logic (true, false, maybe) are essential attributes of any broadly based materials database or knowledge base [15].
2.2
Properties: Measurement, State, and Process
Material properties are less simple than might be imagined. The difficulties are based in the relationships between "similar" properties and in the likelihood that "similar" materials will have "similar" values for the same property.
2.2.1 Measurement distinctions The " s a m e " property can, if measured by two different methods, give quite validly different results because the two methods do not quite test the same things. This does not usually matter at a crude level but for detailed work it must be taken into account. The classic example is the measurement of density; not a property about which there is usually thought to be much ambiguity. There are three ways of measuring density: by external measurement and weighing, by liquid displacement (Archimedes' method), and by x-rays. The last of these measures the interatomic spacing of the solid so that the density of the "perfect" crystal can be calculated, but his ignores the effect of dislocations and vacancies (which are always present) as well as all types of porosity. Liquid displacement includes all microstructural features including closed porosity. External measurement includes open porosity as well. 2.2.2 State and process properties Material properties can be roughly divided into two types, those that are characteristic of a state of the material (like density) and those which are characteristic of a process going on inside the material (like time-dependent creep or fatigue). The effect of this distinction is that the absolute value of state properties is expected to reflect underlying physics and so similar materials should have similar values for state properties. Conversely, the results of a process can be highly dependent on slight differences in the controlling parameters and so similar
materials can have widely varying varying process properties. There is a further issue in that the distinction between state and process depends on the definition of the defining parameters: the state variables and the input parameters to the process. This is not just a matter of detail but of essence. In engineering we must limit ourselves to considering observable parameters and variables, and not just parameters that are observable in principle, but those that can be measured in a practical manner. Thus some properties which might be imagined to be simple, state variables, such as yield stress, must instead be more properly considered as process variables. The yield stress of an alloy depends on its microstructural state, but this state is unobservable ~ and is determined by the processes of microstructural development that occur during heating, cooling, and deformation. This should be contrasted with corrosion and fatigue properties where the property itself is inherently a dynamic process.
2.3
Abstraction and Detail
There are two important matters concerning the levels of abstraction that must be represented for materials designation: relevance and synthesis.
2.3.1 Definition by relevance Materials are defined in practice not by their definition in some standard but by a set of properties considered to be relevant to the current task. Thus so far as organizing the recycling of soft drink containers is concerned, the only distinction of note is that between steel and aluminum. Practical situations are always limited in this sense (by a set of properties), so to be properly useful materials information systems must also present their information at suitable levels o f abstraction determined by the appropriate situation. 2.3.2 Synthesis from measurement Materials properties are all, eventually, derived from some experimental measurements made on individual specimens. The problem is how to relate these sets of numerical measurements to the more abstract concepts of "material" and "property." We have already seen that the " s a m e " property can be measured in several different ways which give slightly different answers. (In practical engineering the most important effect of this type is the
All states are unobservable, measurement is a process itself.
Materials Information Systems
difference between the different methods for measuring fracture toughness.) Also for many purposes we want to aggregate information from many "similar" materials since for the purpose at hand they can all be considered to be equivalent. This is a matter of definition by relevance referred to above and an example might be the typical value of hardness for a set of aluminum alloys which are to be recycled and crushed if the attribute of importance is to ensure that the crushing rollers are not damaged.
2.4
Incomplete, Default, and Estimated Information
Because of the open ended nature of materials information, any realistic database of materials properties, whether implemented as a conventional database or as a frame structure within a knowledge based system, will have missing values for some properties [51,52]. Incomplete databases severely limit the types of inference which can be made from them because many operations which produce definite, useful answers for complete databases produce undefined results for incomplete ones, or at least, results which require further qualification. A good example is the concept of average value: for incomplete databases the average is no longer the sum of all property values divided by the number of materials in the database. Numerical models fail completely if some of their input parameters, which may be materials properties, are missing. Absent values create such serious problems that it is generally advisable to replace them by default values (it is presumed that sufficient data modeling has been done to ensure that all N U L L values do represent missing, rather than inappropriate, data) [15,5]. Thus techniques for estimating defaults assume a central importance [16,17]. This is itself an interesting problem since materials cover such a wide range of behavior that estimation techniques must use local information, for "similar" materials and "similar" properties, and thus must "understand" the matters of abstraction and detail referred to above. Using these ideas we can identify which material properties are easily predictable and which are capricious.
2.5
Capricious Properties
When estimating default values for properties, it is also necessary to have some idea of the accuracy of the estimate and also the confidence with which that
245
P4
xXX ~ xO ~
3r parameter
parameter
Fig. 1. Clustering and nonclustering properties for two materials.
accuracy is assigned. Some types of properties are easier to estimate than others: process properties are always much more variable for "similar" materials than are state properties.
2.5.1 Similar materials Those properties which, when plotted on property charts, show local clusters of similar materials appear so because they depend largely on the strength of the atomic and molecular bonds, on the identity of the atoms involved, and to a lesser degree on the crystallographic structure of the material. Other properties do not display such close clustering of similar materials. The corrosion behavior of an aluminum alloy can be widely different from that of pure aluminum. Figure 1 shows that two classes of material (denoted by circles and crosses) might cluster on one property but not on another. This does not necessarily mean that the second property appears to have widely varying values with respect to a more fundamental parameter (although it usually does); in the figure this is illustrated by showing a correlation between property 2 and the varying parameter. The nonclustering properties are "capricious" and can be identified with processes rather than states in the materials. Thus corrosion behavior is not a simple ranking based on the electrochemical series, but a complex result of perhaps a dozen different competing processes. Because it is the balance between competing processes that is changed by external factors (temperature for instance, or oxygen partial pressure), the resulting behavior is sensitive to fundamentally small causes. Things become clearer if we ask precisely what the process-properties are capricious with respect to? The answer is with respect to what we commonly think of as sets of related materials, which are materials which are related by having similar compositions. Compositions imply atoms, and proportions of different atoms directly imply bonds and crystallography.
246
o
P.M. Sargent
D
\
E ~.
,< []
..-.-~
D/
MECHANISMS
Fig. 2. A property determined by many parameters.
2.5.2 Predictability There are degrees to this capriciousness and they can be estimated by considering what software models one would have to write to predict the behavior. Without knowing what the precise relationships and coefficients are, several decades of materials science theory often enable us to identify the number and identity of processes involved, and the number and identity of the more fundamental parameters that go into these models. (The interpretation of the distinction in terms of levels of predictability with respect to the type of models required to simulate behavior has not previously been made [18].) Not surprisingly perhaps, the better theories use some parameters which are closely tied to the underlying physics, such as activation energies which in turn depend on the identity of the atoms and the stiffness of the bonds. The more numerous the parameters, the more variable the property. However, most of the apparent unpredictability comes not from just rearranging these fundamental properties in different ways, but from the explosion of state variables required to represent the materials' microstructure.
2.5.3 Microstructural combinations The microstructure of a material is everything that is larger than atomic size but small enough that a mechanical engineer can imagine that it is a continuum. It includes dislocations, stacking faults, grain boundaries, phase boundaries, and all of their threedimensional arrangement. It is this geometric and topological freedom which produces a combinatorial explosion of different behaviors (see Fig. 2). Any measured properties determined by the microstructure are describable by a lumped-parameter model, and since the internal (lumped) parameters are individually not observable, the result appears to have little coherent relationship to the observable inputs. As has been mentioned above, there is a deep difference between properties which are characteristics of a material's state and those which are char-
acteristic of some process that the material undergoes. Distinguishing between them, however, is a matter for the observer to decide, depending on how many parameters the observer wishes to use to totally define a material's state. The position and orientation of every grain in a specimen is an unworkably large list of parameters but it would permit precise prediction of some properties. The average grain size and standard deviation is a short list (two parameters), but some behavior (such as "earing") will be indescribable.
2.5.4
Concurrent engineering
Most manufacturing processes rely on capricious, process type properties. This implies (and is found in practice) that competitive manufacturing processes are always in a state of semi-ignorance. There is always more that could be discovered about the process but that effort is better spent in empirical tuning of an existing procedure in order to optimize performance. All process models are accurate only to some level of detail; to get better performance than the accuracy of model, empirical tuning is always necessary.
2.5.5
Structure sensitive properties
The distinction developed here is similar, but not identical to, the distinction between "structure sensitive" and "structure insensitive" properties proposed in old metallurgical textbooks. There the classification was made purely to bring some order into discussing properties for teaching purposes and the distinction is one between crystallographically determined properties, properties of the perfect crystal, and microstructuralIy determined properties which result from crystal defects. The stable and capricious properties proposed here are better distinguished as representing " s t a t e " and "process", irrespective of microstructure because it is up to the observer to decide which defining parameters are appropriate.
2.5.6 Ill-structured problems There is an inexact parallel between stable materials properties and well-defined design problems, and between process-properties and ill-defined design problems. Systematic methods can be used for the first class, usually after a prolonged phase of analysis which for basic properties would be the task of accumulating the data for all candidate materials and arranging them in a form that enables merit indices to be calculated [19,4,20,21]. The second class is best tackled by looking first for a single, feasible solution (a material that will function cor-
Materials Information Systems
rectly, if not well) and then exploring potentially better solutions incrementally from that starting point. This is explicitly the approach taken by the SPLINTER system which takes the view that materials selection is inherently an open ended ill-structured problem [23,22]. Ill-defined design problems are never understood properly without relating them to a potential solution, whereas process-property problems might well be understood conceptually, but the lack of either numeric data or a reliable predictive model nevertheless implies an exploratory type of search. An image might help: the comparison is between moves in chess where each piece can confidently move to any feasible position and the moves made by an explorer attempting to traverse a swamp where he does not know whether a clump of grass will support his weight until it is stepped on. Cast as an optimization problem, this is the same distinction as between solving for the solution directly and having to use a step-wise search algorithm.
2.5.7 Less capricious properties There is one type of process-property which is less capricious than others and that occurs when all the competing processes have something in common. Thus the yield strength of plastically deforming materials depends on many different dislocation-dislocation interaction mechanisms, but these are all directly proportional to the stiffness of the material and related to the density of dislocations [17]. That is why strength and fracture toughness, despite depending on numerous internal mechanisms, are well-behaved enough to appear on Ashby's charts [4,5], though a detailed examination will show rather wider variation for them than for the other properties.
2.7 Information Quality Quality in materials information systems can in principle be assured [1] in regard to 1. 2. 3. 4. 5. 6.
individual data items collected sets of data organizations which measure and evaluate data organizations which deliver data database systems for delivering data organizations which maintain database systems
and these are not independent. The quality of individual data items depends on the quality of the collection of which it is a part, and the quality of a data collection depends on the quality of the individual items and on the organization which collected it.
247
Even individual data items have several different types of quality depending on whether the item is a standard, a specification, or a measurement. Work must proceed in producing guidelines and standards in each of these areas, but each development must also be cognizant of the important interrelationships. The quality of materials information in databases has long been a matter of concern but now at least some of the operations guidelines produced by the Commission of the European Community (CEC) as part of the Materials Databank Demonstrator Project (1984-89) have been adopted as a standard by ASTM as standard E 1407-1991. Much further work is still required to ensure a consistent view of quality of the information in databases however (as the list above indicates), but fortunately most of the principles involved: audit control, recorded evaluation techniques etc., are directly applicable to information in knowledge based systems [18].
2.8 Decision Support Knowledge based systems almost invariably attempt to supply explicit support to users in making decisions in addition to supplying information. Existing systems, both knowledge based and data based, have much to learn in this regard from conventional decision theory as practiced by schools of management science [24-25]. Several techniques are directly applicable to materials selection or decision problems and this can be expected to be an area of growth in the future [26,18,27].
3 Creating Useful Systems The size and scale aspects of materials information inevitably mean that any useful information system will have to be able to communicate with other information systems and probably also with other types of software.
3.1 Materials Information Interchange Materials information is hard to transfer between independently developed databases and knowledge bases for the same reasons: different data modeling decisions will have been taken, different relevant properties will have been used to define the materials designations in the context of the data stored (even if standard international designations are used for identification), and different levels of abstraction will been used with respect to experimental raw
248
P.M. Sargent
data as opposed to evaluated design data. Even where all those aspects are identical, standards of quality may vary. The only solution is not to legislate that certain types of information be represented in a standard manner 2, but to provide standard ways of describing these aspects of any data or knowledge collection so that the receiver can interpret it correctly-as much as will fit with the receiver's framework and abstractions. 3.2
Representing Schema Information
The problems are these: hierarchical representations are easy at first, but become difficult and arbitrary at detailed levels (different people produce different trees). But pure relational database approaches fail to capture many hierarchical associations explicitly leaving many functional dependencies only implicit in their tabular structure. The answer is to use a tabular system but then to supply extra, explicit catalog tables which explicitly describe the functional dependencies. A standardization of such a system is proceeding (see IRDS, below). A data dictionary is a complete documentation of every field of every table comprising the database, of every program that uses the data, and why it uses it. It usually is stored in an entirely distinct database with a unique structure. Conversely a data catalog is the formal description of a particular database schema, stored as tables in the same database management system as the database it describes. Since the database schema documented in the database's catalog assumes such importance, it would be useful to communicate the catalog in some common medium for expression. Pending the development of an object-oriented standard at the right level of abstraction 3 to be communicated along with the data, the only clear candidate for such a medium is a set of tables which could be communicated by any tabular data interchange protocol. The International Standards Organization (ISO) has been working on defining the level of abstraction above data dictionaries IRDS [28,29], and at this level all the different existing materials databases can find agreement [30].
2 That would solve the problem, and will always help so far as it is feasible, but the problem is too big for it to be a complete solution. 3 T h e E x p r e s s language [31]. is suitable, but currently requires extraordinary expertise for effective use.
1
domains, associations
2
data-dictionary schema +-- fundamental data
3 4
database schema +- data-dictionary data database data
Fig. 3. Information resource dictionary (IRD) levels.
Figure 3 shows the four levels of ISO's Information Resource Dictionary System (IRDS), though this scheme did not originate with ISO [32]. At the bottom level is the data, e.g., "Copper", "1356", "Kelvins" At the level above this is the data dictionary 4 which names the fields of the database. For the values just listed these would be "material name," "melting point," and "units" respectively. This level also contains the database schema which defines the fields: "material name" and "units" are text, "melting point" is a positive numeric, and all represent data fields. At the next higher level the valid value domains for the database, e.g., text, numerics, or integers, are defined together with the valid constraints on the data and access permissions for users. These are all held in the data dictionary schema. The database schema is, from the point of view of the data dictionary, just data. Similarly the structure of the data dictionary is just data described formally in the data dictionary schema. (The database catalog described earlier would be just a subset of this data dictionary). Also defined at this level are the list of permitted concepts: the concepts of "fields", "domains", and "constraints" are defined in terms of "fundamental concepts", the data model (top level). In brief, each level contains data in a format which given meaning by the level above [28]. Communication using a shared data dictionary puts the complexity burden on the receiver rather than on the sender. The sender has to ensure that the information transferred is a complete description; effectively it "publishes" its own capabilities and descriptive methods together with the data [32,29]. The receiver has to read the data and description and has to ensure that appropriate translations are made for nonidentical but similar concepts. 4 IRDS u s e s a m u c h m o r e specific definition for the term data
dictionary than is m o r e generally understood.
Materials InformationSystems
249
The self-describing database approach to data interchange requires that all the collaborating databases do share the same global view of how materials data should be organized, i.e., the same data model even if they are represented differently at the data dictionary and database schema levels [54].
4 Big Materials KBS Projects Deep causal modeling systems have implementation problems as the following case study indicates. 4.1
USER
~
~
USER
DB Fig. 4. Subdatabasesusedby knowledgesystemsand database intelligentinterfaces. [38,29]. A monolithic knowledge based system unfortunately implies an all-or-nothing approach.
Case Study: ALADIN
The ALADIN aluminum alloy design assistant illustrates many important aspects of developing systems which represent deep physical reasoning about materials, physics, microstructures, and properties [33]. ALADIN, a technical success [34], was developed at Alcoa Technical Center (Pennsylvania) at a cost of probably over $1 million. It is currently unused. Other large corrosion knowledge based systems include PRIME [35] and ACHILLES [36]. The ALADIN project never made the transition from the artificial intelligence department to the alloy design department since the latter could not afford the high maintenance costs associated with large knowledge based systems. Such knowledge bases require continual upgrading as new information becomes available, as new alloys are developed, and as new knowledge about existing alloys is discovered. Also as any large system is used, unexpected interactions between items of information must be documented or changed. Alloy design is an important, expert task but too few people do it to support such heavy maintenance costs. It is such a specialized task involving so few staff that a company can generally afford to hire the best people. Since it is their expertise which would be embodied in the system, there is the possibility that it would be less useful than if it were to be used by lessskilled staff. (This is not always true for expert systems: the PAL adhesives system sometimes surprises its creators with unexpected but good selections [37].) The system probably attempted to much. Some small, separate software tools (perhaps knowledge based, perhaps not) for manipulating databases of alloys, for running numerical predictions of specific alloying element interactions, or for managing default parameters based on alloy classification could have been adopted and might have been used
5 Databases and Knowledge Systems 5.1
Combined Knowledge~Databases
Many knowledge based systems include databases, and some databases have knowledge based user interfaces [39,29]. Neither approach represents a true merging of the technologies (see Fig. 4). Nearly all large, modern knowledge based projects in engineering use a frame based, object oriented structure to hold unchanging knowledge together with a system of rules and inference/ deduction engines which represent how to solve problems dealing with the frame based knowledge [40,41,34,38]. Some also use frames to store some strategic information representing how to go about applying the rule systems in particular circumstances. Engineering and CAD/CAM information presents some unusual problems for both data and knowledge representation [42,27]. A true merging of knowledge and database technologies will occur when databases begin to cope with the rich, complex functional dependencies in frame based knowledge stores [43]. If these can be represented in standard ways, perhaps using an inheritance-extended version of IRDS [32,28,29], then they can be made independent of specific research projects (see Figs. 4 and 5) and will also be able to be communicated between systems, (assuming that appropriate concept ontologies have previously been built [44,45]). Only then will deep causal models of materials processes become successful, tradeable products of independent value [40]. 5.2
Reusable Concept Hierarchies
Databases are easily developed as a common resource to be used by many different programs, but the construction of knowledge bases (stored as
250
P.M. Sargent
modules. The representation schemes used were implicit in the CRL and not formally defined but even so some knowledge was " m i n e d " from it to be reused in the CORDIAL system [50]. urrent
6 Fig. 5. Eventual explicit and independent information representation.
frame systems) to be similarly used by several different types of programs is currently a research issue. The best hope for reuse is that knowledge becomes represented in concept hierarchies using such explicit tools as CODE, CRL, a data thesaurus, or more implicitly in SPLINTER or Express [33,30,46,23,22,31]. These would form the basis for independently developed information systems which nevertheless would share a common semantic heritage thus making later integration much easier. Such concept modelling is now recommended as thefirst step of developing any complex information system, to be carried out at the same time as the initial systems analysis [45,25]. Although framelike expressions would be most appropriate for materials concept structures [44], the universal problems that always accompany data interchange would indicate that a tabular representation might be more convenient as a common medium of expression [47-49, 18].
5.3
Re-use o f A L A D I N
ALADIN was created with facts represented in the frame based language CRL, rules in OPS5 and other functions in LISP because its complexity and depth required specific features not then found in commercial packages. This makes it difficult, but not impossible, to extract useful modules for use in other knowledge based projects. Such things as a list based representation of the periodic table, small databases of alloys, classification schemes of alloy classes, and manufacturing processes etc. are embedded in ALADIN and one of the justifications of the project was that the cost of such knowledge acquisition and representation would not have to be repeated for other projects. Unfortunately since ALADIN was not really designed for such reuse and because of intense time pressure on the development programmers, the system was not built in properly distinct extractable
Component Lifetime Assessment
Most research emphasis on information use in engineering concentrates on design and manufacturing but materials information has another crucial role in the assessment of the remaining lifetime of existing structures. In most engineering this involves much the same information as required during design but materials change with time: cracks grow s, high temperature leads to creep deformation and recrystallization, surfaces and crevices corrode, and even sunlight affects many structural polymers. This extra information is far more voluminous and often far less controlled than that used in design. Component assessment from a materials engineering point of view inevitably requires a great deal of detailed data, so primary roles for knowledge based systems must be: 1. to help locate relevant data, and 2. to help interpret the relevance of accessible data to the problem under study. The former is much easier and is a prerequisite for the latter. However, it is not so easy as might be imagined. The following are precisely alternative ways of looking at the same issues which show that even the first role has its creative aspects: 1. to advance hypotheses about the problem at hand, and 2. to assess the validity of hypotheses. These problems are clearly in the class of cuttingedge knowledge based systems, yet they lie just beneath the surface of the primary roles just described. A hypothesis might be whether stress corrosion cracking could be relevant, in which case assessing it would involve examining data to see whether it could be ruled out. This is the way in which CORDIAL works.
Conclusions Individual, stand-alone knowledge based systems have intrinsic difficulties dealing with the required 5 They are always present in every structure.
Materials Information Systems
breadth of materials knowledge except in relatively narrow domains. For databases this is alleviated by using several databases together but the technology to allow information and knowledge sharing between independently developed databases and KGSs is not yet available (or at least, is not yet in use).
Difficulties with default information, N U L L values and property estimation mean that the systems have to be substantially more complex than is feasible with expert system shells or business oriented database packages, even for simple problems. In addition, all KBSs require large associated databases of detailed materials properties if they are to offer sensible aid in any field of engineering design, manufacture, or assessment.
References 1. Sargent, P.M. (1991) Data quality in materials information systems, Engineering Design Centre Technical Report CUED/C-EDC/TR.2 April, Cambridge University Engineering Department UK 2. Kaufman J.G. (1988) Sources and standards for computerized materials property data and intelligent knowledge systems, Engng. with Computers 4 (1/2) 75-86 (SpringerVerlag) 3. Glazman J.S.; Rumble J.R. Editors (1989) Computerization and networking of materials databases, Proc. 1st Intl. Symp., Philadelphia Nov. 2-3 1987. ASTM STP 1017 4. Ashby M.F. (1988) Materials selection in engineering design, Materials Science and Technology, June 5, 517-525 5. Ashby M.F. (1989) Overview No. 80: On the engineering properties of materials, Acta Metall. Mater. 37 (5) 1273-1293 6. Bamkin R.J. and Butler C.E. (1989) CAE - The integration with materials data and information, ASTM 2nd Int. Syrup. on Computerization of Materials Property Databases, Orlando Dec. 1 7. Westbrook J.H., Behrens H., Dathe G. and Iwata S. (1986) Editors, Material Data Systems for Engineering, Proc. CODATA Workshop, Schluchsee, FRG, 1985, pp 78-80, 110. ISBN 3-88127-100-7 8. Westbrook J.H. (1987) Designation, identification and characterization of metals and alloys, in Proc. 1st Intl. Syrup. on Computerization and Networking of Materials Property Databases, Philadelphia Nov. 2-3 1987. ASTM STP 1017, pp 23-42 9. Kaufman J.G. (1987) Standards for computerized material property data, in Proc. 1st Intl. Syrup. on Computerization and Networking of Materials Property Databases, Philadelphia Nov. 2-3 1987. ASTM STP 1017, pp 7-22 10. H. Krockel; J.H. Westbrook (1987) Computerized materials information systems, Phil. Trans. R. Soc. Lond. A322, 373391 11. H. Krockel; K. Reynard; J. Rumble, Editors (1987) Factual material databanks: The need for standards, Report of VAMAS Technical Working Area 10, available from J. Rumble, National Institute for Science and Technology, A323 Building, Gaithersburg, MD 20899
251
12. Plastics and Rubber Institute (1989) Standardisation of the presentation of data for plastics, Int. Conf. on Polymer properties for CAD/CAM, 13-14 Dec. 1989, London. Publ. Plastics and Rubber Institute, ISBN 1-871571-06-5 13. Eriguchi K.; Shimura K. (1990) Factual databases for materials design and manufacturing, ISIJ Intl. 30 (6) 409-416 14. Waterman N.A. (1990) Final Report on the Evaluation of the Materials Data-Bank Demonstrator Programme of the Commission of the European Communities, Quo-Tec Ltd., August 1990, Amersham HP6 5AE, UK 15. Date C.J. (1990) An Introduction to Database Systems, Fifth Edition, Vols I (1990) and II (1988), Addison-Wesley Publ. Co., Reading, MA., ISBN 0-201-51381-1 16. Sargent P.M. (1987) Model-based procedures to aid estimation and validation of materials property data, presented at the Institute of Metals Conf. Mathematical Models for Metals and Materials Applications, Sutton Coldfield, UK, Oct. 1987 17. Frost H.J. and Ashby M.F. (1982) Deformation Mechanism Maps, Pergamon Press, Oxford, UK 18. Sargent P.M. (1991) Materials Information for CAD/CAM, Butterworth-Heinemann Publ., August, 170p, ISBN 0-75060277-5 19. Dieter G.E. (1983) Engineering Design: A Materials and Processing Approach, McGraw-Hill publ. ISBN 0-07-016896-2 20. Ashby M.F. (1991) Overview No. 82: Materials and shape, Acta Metall. Mater. 39 (6) 1025-1039 21. Cebon D.; Ashby M.F. (1991) Materials selection for mechanical design, in Computerization and Networking of Materials Databases: Third Volume, ASTM STP 1140, Thomas I. Barry; Keith W. Reynard, editors, American Society for Testing and Materials, Philadelphia, 1992 22. Zucker J.; Demaid A. (1989) A software machine designed for selection, Knowledge-Based Systems J., Butterworths, 2 (3) 178-184 23. Demaid A.; Zucker J. (1988) A Conceptual Model of Materials Selection, Metals and Materials, May 291-297 24. Dawes R.M. (1988) Rational Choice in an Uncertain World, Harcourt Brace Jovanovich Inc. ISBN 0-15-575215-4 25. de Nenfville R. (1990) Applied Systems Analysis: Engineering Planning and Technology Management, McGraw-HiU 0-07-016372-3 26. Sargent, P.M. (1991) Decision techniques for materials selection, Engineering Design Centre Technical Report (CED/ C-EDC/TR.4 August, Cambridge University Engineering Department UK 27. Hopgood A. (1992) Knowledge-Based Systems in Engineering, CRC Press, Times Mirror books, in press 28. Gradwell D.J.L., Ed. Information resource dictionary system: Framework, ISO/IEC JTCI/SC21 N2642, and "Data Modelling Facilities" ISO/IEC JTC1/SC21/WG3 N634. Data Dictionary Systems Ltd., Camberley, UK 29. Beynon-Davis P. (1991) Expert Database Systems: A Gentle Introduction, McGraw-Hill ISBN 0-07-707240-5 30. McCarthy J.L., "Information systems design for materials property data" in Computerization and Networking of Materials Data Bases. ASTM STP 1017, J.S. Glazman and J.R. Rumble Jr., editors, American Society for Testing and Materials, Philadelphia, 1989, pp 135-150 31. "Information Modelling Language: Express" Annexes A, B and C of ISO Draft Proposal for an international standard DP-10303, previously issued as N287, N280 and N281 32. Mark L.; Roussopoulos N. (1986) Metadata management, IEEE Computer Dec. 1986, 26-36
252
33. Hulthage I.; Przystupa M.; Farinacci M.; Rychener M.D. (1987) The metallurgical database of ALADIN - An alloy design system, in [41] 34. Hayes-Roth F. (1989) Towards benchmarks for knowledge systems and their implications for data engineering, in [40] 35. Jovanovic A.S.; Kussmanul K.F.; Lucia A.C.; Bonissone P.P. Editors. (1989) Expert systems in structural safety assessment, Lecture Notes in Engineering 53, 375-389, Springer-Verlag 36. Balkwill P. (1990) Achilles, An expert system on corrosion and corrosion control, Proc. 1st Intl. Conf. on Computer Applications to Materials Science and Engineering: "Computer Aided Innovation of New Materials", Aug. 28-31, 1990, Ikebukuro, Tokyo, Japan 37. Lees W.A. (1991) "The PAL program: Permabond adhesive locator" in Computerization and Networking of Materials Databases: Third Volume, ASTM STP 1140, Thomas I. Barry and Keith W. Reynard, editors, American Society for Testing and Materials, Philadelphia, 1992 38. Payne E.C. ; McArthur R.C. (1990) Developing Expert Systems: A Knowledge Engineer's Handbook for Rules and Objects, John Wiley and Sons, ISBN 0-471-51413-6 39. Kaiser G.E., et al. (1988) Database support for knowledgebased engineering environments, IEEE Expert, Summer 1988 pp 18-32 40. Kellog C. (1986) From Data Management to Knowledge Management, pp 75-84 Special issue on data engineering, IEEE Computer, January 1986 41. Harrison R.J.; Roth L.D., Editors (1987) Artificial intelligence applications in materials science, Proc. Syrup. held in Orlando, Pla., Oct. 8 1986. Publ. The Metallurgical Soc. Inc. (AIME) ISBN 0-87339-067-9 42. Hartzband D.J.; Maryanski F.J. (1985) Enhancing knowledge representation in engineering databases, IEEE Computer Sept. 1985, 39-48 43. R. Hull; R. King (1987) Semantic database modelling survey, applications and research issues ACM Computing Surveys 19 (3) 201-260 44. Sargent, P.M.; Subrahmanian, E.; Downs, M.; Greene, R; Rishel, D., Materials information and conceptual data modeling in Computerization and Networking of Materials Databases: Third Volume, ASTM STP 1140, Thomas I. Barry
P.M. Sargent
45.
46.
47.
48.
49.
50.
51. 52.
53.
54.
and Keith W. Reynard, editors, American Society for Testing and Materials, Philadelphia, 1992 Skuce D; Meyer I. (1990) Concept analysis and terminology: A knowledge based approach to documentation, 13th Intl. Conf. on Computational Linguistics, COLING 90, 20-25 August, Helsinki, Finland McCarthy J.L. (1988) "The automated data thesaurus: A new tool for scientific information", 1lth CODATA Conf. Karlsruhe, Sept. 26-29 Sargent P.M. (1989) A Survey of technologies for materials data interchange, Technical Report CUED/E-MANUF/ TR.1 February Cambridge University Engineering Department, UK Sargent P.M. (1989) Use of abstraction in creating data dictionaries for materials databanks, ASTM 2nd Int. Symp. on Computerization of Materials Property Databases, Orlando Dec. 1 Sargent P.M. (1989) Associativity in materials property data and a tabular materials data interchange format CUED/CMATS/Tr. 162, and TR. 163 August 1989, Cambridge University Engineering Dept. Technical Reports, UK Boag W.A. Jr.; Reiser D.B.; Sprowls D.O; Rychener M.D. (1987) CORDIAL - A knowledge based system for the diagnosis of stress corrosion behavior in high strength aluminium alloys, in [41] Sargent P.M. (1985) Expert systems in metallurgy and materials engineering, Metals and Materials 1 540-545 Sargent P.M. (1990) Materials data interchange for component manufacture, Engineering with Computers 6 (4) 237247 Ramamoorthy C.V; Wah B.W. (1989) Knowledge and data engineering, Special first issue, IEEE Trans. Knowledge and Data Engineering 1 (1) March 1989 Rumble J.R; Smith F.J. (1990) Database Systems in Science and Engineering, Adam Hilger Publ. ISBN 0-7503-0048-5
Cambridge University Engineering Department technical reports are available from the author, CUED, Trumpington Street, Cambridge CB2 1PZ, UK, for a nominal fee of s