Materials and Processes Research and the Information Highway: Summary Record of a Workshop (1996)

John Rumble

National Institute of Standards and Technology (NIST)

Rumble described four NIST experiences which provide lessons learned. One involved NIST's National Materials Advisory Property Data Network, which integrated multiple materials databases. The network failed because it lacked enough databases to attract large numbers of users. Metadata were handled well by the system, but some metadata were not available because industry was reluctant to provide it.

A second NIST effort, called A Computerized Tribology Information System (ACTIS), has experienced low use. He attributed the failure of this system to high expansion costs (a new architecture was required for each version) and lack of strong support from industry and academe.

The third was NIST's Personal Computer databases, with over 75 scientific and technological titles on disks. NIST itself markets these products because no large software vendor has shown an interest in doing so.

The fourth experience discussed by Rumble was NIST's International Standards Organization (ISO) standard for the Exchange of Product Data (STEP), which allows integration of engineering software. The limiting factor in its adoption has been that specific applications are labor-intensive, and implementation is not cheap.

Rumble then discussed the following questions:

• How can we develop coherency in a system of distributed databases?

• How do we integrate metadata from different databases?

• How can we design M&P databases when the tests producing the data may have as many as 150 or 200 variables?

• What can be done to accommodate error ranges?

• How can diverse databases be linked to user software?

Data must be integrated with the computer tools used in engineering. Although such things as clipboards, browsers, and search engines are coming into existence that simplify the movement of data, the problem is moving it automatically (i.e., without looking at the data). A handbook can be hardwired to perform computer tasks, but a standard format is needed. The automation of data transfer leaves unanswered the question of who will judge the quality of the information.

Computer programs are also being developed that will be capable of designing virtual materials. Tests of these virtual materials to determine their suitability, particularly in critical applications, will then become necessary. We need databases from the atomic to the macroscopic level to support materials design, as well as better models in these same regimes.

Standards are a necessity to enable the user to use multiple databases as if they were a single database. Without standards, the building and linking of diverse databases becomes overwhelming. The development of standards, however, will require the participation of the scientific community. The engineering community already understands the necessity of standards and is willing to participate in establishing them.

Rumble emphasized that information is not free. The quality of the information depends on designing for quality and supporting quality efforts. In the future, information-gatherers will be paid to improve the quality, cohesiveness, and consistency of data.

In summary, Rumble stated that the ideal system for M&P researchers would have the following characteristics:

• a single computer interface to provide access to diverse data sources

• data whose good quality is self-evident

• a full range of information

• ease of use

The chief obstacle to building such a system, he said, is lack of funding. The economic return for building and supporting this system is not apparent, so there is little motivation to do so.

Baglin stated that data come from a huge number of sources, and users therefore need information on quality. But, he added, it is not clear who should be responsible for validating and authenticating information. Rumble responded that the institutions that currently validate paper-based databases could perform the same function for electronic databases. These include, for example, ASM International, NIST, DOD, and the university community.

Kerschberg asked for a show of hands on how many participants had a home page. Most indicated that they did. He then asked how many of the participants shared data on the Internet. Only a few indicated that they did. This prompted Kerschberg to point out that researchers on the Human Genome project had found a way to share information and that perhaps the M&P community could develop similar techniques.

Rumble said that the Human Genome project has curators and is in any case only a subset of the field. He added that the Protein Data Bank and other

databases in molecular biology are not integrated. As a result, researchers spend a lot of time getting protein structures from PDB and comparing them with structures from other databases.

Laudise mentioned that optical fibers were just one example of a topic that could be aided by critically evaluated data made available on the Internet. Rumble, however, said that optical fibers were a specialized case, and that other materials information does not have to be as critically evaluated.

Losleben indicated that the issue on data economics is not black and white. Perhaps there is some middle ground on how data are treated on the information highway in terms of how much effort and cost are expended on validating it. A tradeoff should be made between cost and corresponding value, and the appropriate level selected.

Baglin stated that users need help to evaluate the quality of data. Goldie responded that new methods are not needed to validate data quality. Practices for data V&A could be transferred from paper to electronic media. However, providing such services on the Internet will not be free, and a mechanism to pay for validated data and information must be identified. Standards are the most important issue.

Thomas Rohmiller

Wright-Patterson Air Force Base

Rohmiller said there were unlimited opportunities for access to information on the Internet that has been provided by the development of better search tools. On the other hand, he noted the rise of access fees to sites containing databases of value. These are analogous to indexing and abstracting services in the print world, where one expects to pay a subscription price in return for knowledge about the location of published information.

The purpose of Rohmiller's demonstration was to illustrate the use of the Internet as a way of gaining access to traditional information sources, particularly indexes and abstracting services that cover materials science. Commercial electronic databases have existed for more than 20 years, many of them having been established as outgrowths of printed indexing and abstracting services.

To demonstrate, Rohmiller logged on to a server at Fort Belvoir through the Internet. He solicited a search topic from the audience—“heat-treating aluminum”—and then obtained a list of 640 papers on this subject. The GoldenGate software used in the demonstration illustrated the advantage of a common graphical interface to reach diverse databases.

Rohmiller classified the three basic information processes made possible by the Internet: Telnet, which provides remote interaction with a host computer; the

file transfer protocol that makes it possible to move files from one computer to another; and electronic-mail, for personal communications. Specialized software exists to facilitate use of these basic functions. Another example besides Golden Gate is UnCover Company's Personal UnCover Navigator (an on-line periodical search and delivery service). The uniqueness of such software is that it provides a way to search without detailed knowledge of search tools. The graphical interface on the monitor looks the same for any of the diverse databases. GoldenGate enhances successful searching by providing a common graphical interface for multiple commercial databases. GoldenGate has become a popular way to access abstracts; UnCover provides only information on periodicals.

These tools are used primarily by Wright-Patterson Air Force Base Laboratory researchers; access can also be given to recognized Air Force contractors.

The discussion focused on the cost of the service. The point was made that the costs are often hidden, and that costs may rise significantly in the future. In addition, pricing is complicated by multiple fees. However, Rohmiller said that costs can be reduced with additional human effort to focus searches. It was noted that the cost is low if titles only are retrieved. Retrieval of abstracts, on the other hand, may cost around $1.15 each.

Baglin indicated that the high costs of using electronic systems must be justified. Rohmiller suggested that speed of search often justifies the cost. Wiederhold said that it can take weeks to set up different accounts, and that it would be helpful if it were possible to do it once for a variety of services. The point was made that the example set by paper services has to be followed to make electronic service attractive: the user pays a one-time fee (say, annual) and is then covered for a period of one year. Baglin said that electronic access should be as simple as using a public library, where no borrower feels inhibited by price considerations. Woodall concluded the discussion by stating that paying for information is an important problem that needs to be addressed.

Frank W. Crossman

Advanced Technology Center, Lockheed Martin

Crossman discussed and displayed the information on materials-related Web sites that he had assembled into a homepage organized as an electronic notebook linked to the WWW. Internet links are categorized to provide ease in

accessing desired information in a timely manner. Categorization of information is largely according to personal preference. Information on materials science and engineering can be categorized in terms of performance, structure, properties, and synthesis.

Crossman used hypertext markup language (html) to create the list but noted that more user-friendly documentation methods would be needed for routine updates. Daily use of the Usenet to get information on materials is low. Crossman speculated that materials topics are hard to describe with text only, and that more robust descriptions involving graphics, charts, and photos were needed. He also noted the scarcity of on-line materials databases (either free or offered by subscription). Few journals on materials can be obtained on-line. Links to other lists of valued links is important in accessing worthwhile data.

Crossman also demonstrated how to search materials-related Web sites at Oak Ridge National Laboratories (ORNL), Sandia, and DARPA with local search tools (Harvest). The systems's ability to schedule manufacturing of MEMS chips remotely was demonstrated. Lists of nonprofit institutions that can provide materials-specific information were offered. Virtual materials databases and generic search engines were included. The still inconsistent or incomplete intranet linking of companies among large corporate organizations was noted.

Crossman observed that browsing the Internet can be time consuming, and that it is highly desirable to be able to access data to assess its value before being required to pay for the privilege of using it.

Rumble asked if the list generated was of value. Crossman responded that the link to ORNL had been used by him to prepare for a visit to ORNL during the following week.

LeClair noted that Microsoft Word and Excel products have been developed for easier generation of html documents. Crossman said that an understanding of underlying html tags saved time in updating existing documents.

Friday commented that indexing of information sites is valuable, that librarians do that well as a first cut, and that maintenance of electronic indices is needed for individual disciplines.

Crossman noted that the taxonomy of materials information was not universally agreed upon, and that the topic needed further study. Furthermore, lists of new developments in materials continue to evolve.

One participant asked how the quality of lists is assured. Crossman responded that the searcher must monitor quality or rely on others whose expertise can be trusted.

Laudise asked whether lists similar to phonebook lists should be generated—that is, with a name, address, and phone number associated to each entry.

Knowles then suggested that lists containing descriptions of uniform resource locator (URL) sites would be even more valuable, and that materials societies might become the creators of such lists.

Manzione stated it would be useful to have profiles of recently generated information made available to subscribers on a regular basis. As an example of this, Crossman noted that the San Jose (California) Mercury's Newshound service provides information from news services that is sent by e-mail to subscribers daily.

Rohmiller noted there will still be a delay between when data is obtained by a researcher and when it becomes available to the public in a refereed publication. So even though the Internet can be used to access this information in electronic journals quickly, it still won't be current.

Crossman noted the use of listservers (a program which manages and distributes information lists, electronic journals, discussion groups, etc.) on specialized topics where preliminary information is shared to prevent the delay that arises before information can be made available in printed journals.

Baskes said that some organizations were spending millions of dollars to create Web sites, and asked whether the organizations were benefiting from their efforts. The question remained unanswered.

Kevin Skinner

Technology Transition Office, Wright-Patterson Air Force Base

Skinner is the chief of TECH CONNECT at Wright-Patterson Air Force Base. TECH CONNECT is a clearinghouse for the distribution of Air Force information to DOD, academia, and industry. It is also the Air Force Materiel Command (AFMC) technology information base and coordinates all AFMC technology transfer activities. Since it began operating in mid-1993, TECH CONNECT has helped users search for helpful Air Force technologies, contact Air Force technology experts, and identify technology transfer opportunities. Users include the National Hot Rod Association, Caterpillar, Inc., University of Kansas, and Goodyear Tire and Rubber. So far, the clearinghouse has dealt with approximately 3,700 requests, many of them resulting from the organization 's advertisements in the journals of professional societies.

TECH CONNECT provides access to various databases covering the technical library at Wright-Patterson. It also utilizes UnCover (search for journal articles), other Air Force laboratories and TriNet. TriNet is a network that moves information among the Air Force, Army, and Navy. Electronic mail allows for rapid responses to technology questions. This network also offers a way to locate a specific technical expert and thus encourages combined efforts in solving common problems. Skinner emphasized the importance of the human-machine

interface in TECH CONNECT's operations. Analysts determine which tools and experts are most appropriate for a problem. Clearly, Internet use is substantial, but analyst assistance in searches is often required. Use of TECH CONNECT has been helpful in inventing a garbage truck that burns waste as it is dumped into the truck, identifying experts on Joint Standoff Weapons (the target need not be in sight—e.g., cruise missiles), evaluating the distance between aircraft in flight in upgrading flight simulators, and the use of ultraviolet lasers to cut textiles. Skinner also said that in the event of a war, TECH CONNECT's capabilities could be made available through classified electronic mail 24 hours a day. Although TECH CONNECT was initially developed to provide expert capabilities for Air Force laboratories, it is rapidly expanding to DOD, academia, and industry. The service is free, and could be available to foreign companies whose embassies request access.

Woodall remarked that TECH CONNECT's capabilities made it a much more valuable tool for research than an ordinary library, and moreover that it “takes care of the user-friendly problem by the use of analysts.” Pu asked if GoldenGate is also being used and was told that it was. Alexander referred again to the hidden costs problem, particularly the costs of human efforts, since the service is subsidized by the government.

John L. Mitchiner and Kim Walker Mahin

Sandia National Laboratories

Mitchiner (computer science) was the primary speaker, with Mahin (materials science) providing additional input on pertinent issues. Mitchiner described SmartWeld as an on-line, software-intensive, concurrent engineering system which integrates product design and processing decisions within an electronic desktop engineering environment. It is being developed to provide designers, process engineers, researchers, and manufacturing technologists with access to the right process information, process models, process experience, and process experts to produce a correct welded assembly on the first attempt. The impact of SmartWeld at Sandia National Laboratories has been to reduce the iteration times for redesigns of welded assemblies from weeks to hours. The goals

of SmartWeld are as follows:

• to develop the information infrastructure required to provide a generic foundation for concurrent engineering of products and processes

• to develop SMART access (automatic filtering of information to provide only what is most relevant) to databases, information sources, and computational tools

• to create a prototype of this concurrent engineering tool using the SmartWeld system of integrated design, analysis, and knowledge for arc, laser, and solid-state welding

• to translate current models of the welding heat flux (laser and arc) and materials response models into robust, user-friendly forms through the use of SMART Graphical User Interfaces(GUIs) and task models (to indicate model limitations)

• to develop DOE defense program (DP) product-specific knowledge bases to assist designers in choosing welding processes, materials joint geometries, and processing schedules

• to provide the foundation for rapid development of other SmartProcess systems for design and manufacturing

SmartWeld, Mitchiner said, integrates the science of welding in a variety of models, ranging from knowledge-based and empirically derived models to predictive mathematical models. It can also model the temperatures and stress distributions that will be generated during welding. The program enables engineers to utilize various software tools and databases to assess design changes by way of computer simulations rather than on the manufacturing floor. SmartWeld enables the user to predict how changes in early processing steps will influence later processing steps, and ultimately the properties of the final welded assembly. SmartWeld is also valuable in quickly identifying an optimum welding process for a particular set of user-input design and performance requirements.

Mahin noted that SmartWeld consists of several independent, modular components which are interfaced via the Internet. She suggested that a SmartWeld-type system (e.g., SmartProcess) could be used for many manufacturing processes by intelligently accessing remote knowledge bases and process models via the Internet. For example, a Design for Manufacturability Advisor would help designers improve the manufacturability of machined parts. This advisor, described as a “spelling checker ” for machinability problems, would incorporate manufacturability concerns into the design process. A Near Net-Shape Advisor would assist designers in selecting which near-net shape process to use prior to finish machining. This advisor considers such near-net processes as rough machining and casting, and will include forging in future versions.

Mitchiner and Mahin ran a demonstration of SmartWeld via the Internet using Netscape as one front-end and using Motif and x-windows as a second front-end. The Forging Advisor and the Welding Advisor were demonstrated using actual data, and they produced design recommendations in tabular form that were

relayed to the workshop via the Internet. Guidance for finite element analysis (FEA) simulations of welds can be obtained through an on-line, user-friendly workflow diagram that was designed by analysts, thus demonstrating improved access to information without requiring expertise in FEA.

At the conclusion of the demonstration, Mahin stated that SmartWeld allows design engineers to concentrate on their areas of expertise, without mandating that they learn several different computer codes. SmartWeld takes information that is input at a user-friendly front-end, and sends jobs to the appropriate encapsulated module to obtain design or process recommendations.

Kerschberg asked how long it took to run the finite element calculations that are utilized by SmartWeld. Mitchiner answered that calculation times depend on the complexity of the problem, but that problem set-up and the ability to view results were greatly enhanced by SmartWeld.

Friday asked if SmartWeld recognized features such as holes and pockets from computer aided design (CAD) models. Mitchiner responded that SmartWeld does not currently have feature recognition capabilities but that such capabilities are in the process of being integrated into the different system.

Wiederhold asked what the intermediate data exchange format was. Mitchiner answered that they are currently using neutral files but are converting everything to CORBA agents.

Kristen P. Constant

Iowa State University

Constant described the ongoing project at Iowa State University to develop multimedia courseware for education in materials processing, with an emphasis on industrial relevance. Constant underlined the project's emphasis on materials processing by quoting an unnamed government study which said that materials science is taught to budding scientists and engineers at the expense of materials processing.

Constant then noted another important issue in education in general —incompatibility in learning and teaching styles—for example, visual versus verbal. Multiple ways of presenting course content using multimedia courseware will thus assist students with different learning styles.

The industrial relevance of materials and materials processing is made clear to students by using an “inverted paradigm” relating industrially relevant materials

and processes to common products. The courseware describes 40 industrial products and their associated materials and processes, 12 of them in detail, including aerospace, industrial, transportation, and household products. The product used by Constant to demonstrate the courseware was a hand-held blender. Her discussion focused on the magnet inside the electric motor of the blender, describing the function of the magnet, the materials and design requirements of the magnet, and the processing steps used to make the magnet. Using the program, students are able to investigate different aspects of materials processing by selecting different processing variables to manufacture a magnet. Similarly, MathCad is used for mathematical modeling, so that students can see graphs change in real-time simulation as processing parameters are changed.

Constant said that the courseware is being used now in courses, and that demonstrations are available. A formal assessment and evaluation of the courseware is scheduled at Iowa State University in the fall of 1996.

In Constant's view, adaptation of the courseware for the Internet would allow for wider usage and program expandability. Obstacles to expansion, however, include questions about the protection of intellectual property, program maintenance, and slow real-time information transfer over the Internet.

Baglin asked about the publisher's role in updating the courseware. Constant replied that financial incentives are important, but that PWS (the publisher) is equipped to maintain the courseware in a format accessible through the Internet.

Kerschberg suggested that it may be possible to store the intellectual property on a CD-ROM, which could then be marketed by the publisher. In this scenario, the software user would be required to purchase the CD-ROM in order to access the program but would be able to access pertinent supplementary information from a designated site via the Internet.

Baskes asked if the course could be customized. Constant replied that the courseware can be edited and has inherent flexibility.

LeClair asked if there was a mechanism in place to improve the courseware and whether students can add information in an interactive way. Constant responded that project members are working with companies on an on-going basis to improve the product by implementing realistic materials processing and application scenarios into the courseware. The software is interactive in terms of allowing students to select the subject matter and depth of information they wish to study. However, the program is not interactive in an additive way (i.e., students cannot add information to the program).

A workshop participant stated that the information that was presented seemed static. Constant replied that the process simulation that was demonstrated was new and that she was in the process of getting additional information on injection molding.

Woodall asked what the role of the professor was in using the courseware program. Constant responded that although the program is self-guided, professors provide guidance as well. Woodall then asked whether elimination of the professor is avoided. Constant replied that there is a lack of resources for teaching, and that the courseware is designed as an educational aid, not as a replacement for professors.

Sonwalkar stated that he was trying to use computers in a similar manner at Massachusetts Institute of Technology, particularly in the liberal arts. He said that experience at MIT revealed little measurable progress by using such programs. Sonwalkar stated that evaluation is important. Constant responded that Iowa State was trying to conduct an assessment of the program in a systematic way. The program is only being used by materials science and engineering majors at Iowa State, and there is no control group. In addition, the course in which the courseware is being tested is being taught in a different way, and different material is being taught than was used in the past, making it difficult to measure the usefulness of the courseware.

Reifsnider said that Virginia Polytechnic Institute has an “electronic village project” and wants to develop paperless courses. He then asked what Constant 's experience was in this regard. Constant replied that students learn by different means.

Laudise suggested that students might not learn materials terminology very well if given access only to courseware. Constant said that a textbook which contains additional information is also used in the course. She then demonstrated the glossary of terms available from the software.

Baglin stated that the principle used in making the courseware could be used in other areas, such as presenting database information. There is a whole spectrum of choices (in user modeling).

Adams noted the ability of courseware to tailor the instruction to individual students. The Air Force has an “intelligent tutoring” computer program, which is used to give technical training to Air Force personnel. Ninth-grade calculus, for example, is taught on-line, and the computer keeps track of the progress of each student. In a lot of remedial courses, he said, students are disinclined to do the required work, but that putting the material on software helps to overcome that disinclination. Constant added that computerized software gives students more control of the learning process. Adams stated that this was helpful from a psychological point of view because it reduces competitive pressures among students.

Nishikant Sonwalker

Massachusetts Institute of Technology

Sonwalker presented a concept for distributed, networked, teaching hypermedia using html document-format teaching aids on the World Wide Web. The term “hypermedia” includes text, graphics, audio, video, animation, and simulation provided via an associative database and GUI. He then provided a brief description of the integrated framework called the Hypermedia Instruction and Teaching Environment (HITE), which was created for the design, development, and deployment of core courses. HITE architecture leverages the capabilities of the WWW and adds an outer software shell to facilitate pedagogical functionality. HITE architecture is used to create domain links between courses as a curricular integrative mechanism and as a pedagogical device to enhance guidance via text, audio, video, graphics, animation, and simulations of engineering concepts delivered by the Internet.

The course is designed for interaction with other courses. A trunk path linking the course to other courses is presented, and students are given a tour of the trunk path by the professor. For example, a fluid mechanics course will have links to courses in solid mechanics, heat transfer, and relevant mathematics.

Sonwalkar said that although hypermedia courses delivered through the WWW enhance the educational process, professors are still responsible for teaching the students. The design, development, and deployment of hypermedia courses is done through a hypermedia teaching facility (HTF). The HTF provides collection, conversion, and organizational tools to prepare courseware. Professors work with the HTF to create courseware from existing materials (e.g., textbooks, notes, demos, etc.).

There are HITE authoring tools, a WWW Server, and GUI browser elements for the local area network (LAN). HITE architecture integrates the tools and processes necessary for successful use of the material by students. Security elements make it possible to restrict the course to class participants only. Special links to simulation software are paid for on the basis of the amount of use of the software. Geographically dispersed classes have shown value in raising the level of class participation by all students, regardless of national origin, language proficiency, etc. Learn engines have been developed. CD-ROMs are used to archive the course and are used as take-home media.

Sonwalkar said that firms involved in commercial educational enterprises and publishing are happy with the focus on interdisciplinary course linking. Students are highly motivated. Trunk path architecture is a key contributor to teaching, and MIT plans to expand its remote site teaching to its graduates and to industrial firms that support programs at MIT. Fees are charges for the use of secondary software tools, so that the amount of usage by the students is an effective measure of tool value.

Mahin asked if the users are linked to the tools in a way that allows them to perform actual simulations or simply to obtain results. Sonwalker responded that, ideally, information should be presented in context. For example, he said, excellent movies showing mechanical phenomena have been made and are stored in libraries, but students rarely are inclined to watch them. If put on CD-ROM, such films are more accessible, and students are more likely to watch them.

Baskes asked why CD-ROM was used instead of putting the program online. Sonwalker answered that there is archival value to the information stored on them.

Goldie asked if the HTF has a complete facility, including staff to write software and provide information update. Sonwalker replied that the HTF has a complete facility.