Lapas attēli
PDF
ePub

STATE OF THE ART IN JAPANESE

COMPUTER-AIDED DESIGN METHODOLOGIES FOR MECHANICAL PRODUCTS: REPORTS ON INDIVIDUAL VISITS TO COMPANIES AND UNIVERSITIES

The author spent 3 months in Japan as a temporary liaison scientist with
the Office of Naval Research Asian Office to survey Japanese use of
computers in design of mechanical products, to report on the state of
practice in Japanese companies, and to determine research needs and
trends in both industry and academia. A summary report was published in
the first issue of 1992; this is an appendix that contains detailed
information on the author's visits to companies and universities.

UNIVERSITY OF TOKYO: NEW
DEPARTMENTAL ALIGNMENT
IN MECHANICAL
ENGINEERING

5 June 1991

Background

Prof. H. Inoue is in the Mechanical Engineering Department and is currently head of the new MechanoInformatics Department. How this new department came into being is the subject of this report.

The University of Tokyo (Todai) is Japan's most prestigious. Its graduates go into the best government and university positions, including most of the new hires into Todai's own faculty. The buildings are quite old, solid reinforced concrete, and hard to modernize. Budgets are tight. Almost all the students are self-supporting, including graduate

by Daniel E. Whitney

students. Tuition is high but appar-
ently not nearly as high as at private
U.S. universities.

The typical course of study is 4 years
for a bachelor's degree, 2 more for a
master's, and 3 more for a Ph.D. The
undergraduate curriculum is almost all
classroom courses while graduate study
is mostly laboratory work and thesis
with only a few classes. This is impor-
tant to understand in view of the sub-
ject of this report, which depends heavily
on curriculum reform.

New Departments

Mechanical engineering used to be split into three subdepartments called Mechanical Engineering (ME), Mechanical Engineering for Production, and Marine Engineering. The latter came into being about 20 to 30 years ago as Japan became a prime shipbuilding country. Since Japan no longer

leads in shipbuilding, this department has been totally eliminated in the new structure. When it existed, it dealt primarily with engines and other ship machinery, not with ship structure or other traditional naval architecture. Elsewhere in Todai there is now a Department of Shipbuilding and Naval Architecture.

Three years ago the ME Department decided that it was losing students or would soon, with the defectors going into more modern technologies based on computers and information sciences. (Inoue said this twice during our talk.) The response was to "restructure" and modernize the curriculum.

The pressure to restructure came not only from trends visible in student registrations but also in general from the rush of technological change in society and industry. Japan identified information-intensive products as strategically important as early as 1970

with the launching of the PIPS (Pattern Information Processing Systems) 5-year national project and has pursued this area intensely since. Obviously mechatronic products will proliferate and engineers will be needed to design them. Industry is quite interested in this new restructuring.

Dept of Mechanical and Industrial Engineering (Broad ME)

Broad ME includes industrial engineering and production engineering. Both design and computing appear in

1. Production Systems, Manufactur- all three subdivisions. Students majoring Systems

[blocks in formation]

ing in any one of these three take courses from the various chairs, with 50% recommended from the home department and 25% each from the other two. There are no required subjects. The require

3. Systems Engineering, Security ments for what we call "humanities” Engineering

Inoue noted that the restructuring began 3 years ago and the Ministry of Education took until this January to give final approval. Todai is a national university subject to the Ministry's 4. Design Engineering governance. I do not know if there is an equivalent of ABET other than the Ministry, but I doubt it. He also notes that all the debate, curriculum creation, and course design occurred during this time, so the big fights are over and the new structure is fully in effect.

Department and Curricular Structure

The new department structure recognizes "traditional deep" ME, broad ME, and mechano-informatics (new ME).

5. Human Systems Engineering

subjects, basic science, and math are satisfied in the first 1.5 years when the students attend a different campus. This way of setting up the curriculum may have been adopted in order to reduce conflict between the advocates of the

6. Industrial Systems, Transportation new curriculum and those of the old Systems

7. Humanware Systems Engineering (donated chair by JR East Japan)

who usually ask in such debates what mechanical engineering really is. The new structure actually moots this question in a very realistic way, acknowledging the fact that the old curricular

Dept of Mechano-Informatics (New ME) and discipline boundaries have long

1. Electronics and Computer in Machinery (digital systems, microcomputer, interface, micro-machine)

Dept of Mechanical Engineering (Deep 2. Mechanism and Control (mecha2. Mechanism and Control (mechaME) tronics, control theory, mechanics and mechanisms)

[blocks in formation]

since been destroyed by external events and it is necessary to build new ones.

According to Inoue, the purpose of the Department of MechanoInformatics is to enhance the research and education of computer-intensive mechanical engineering. Primary research fields include:

⚫ creation of intelligent machinery such as robotics and mechatronics

⚫ computer-intensive design and analysis of mechanical systems

introducing new functions or approaches into machinery such as bionic functions, neuro science, and micro-machines

⚫ advanced human-machine interface, virtual reality, cognitive engineering, etc.

On subsequent visits I hope to delve deeper into such questions as the relation between university training and company training and whether the

university thinks any one student can really learn all the things that are offered. What should a competent design engineer know in a world of mechatronics? Since there are no required subjects, only "strongly recommended" ones, the department has not taken a rigid stand on these points. [This topic is addressed in the report on Hitachi Construction Machinery.]

I raised the question of the place of algorithms in this curriculum. It may seem odd to relate algorithms to mechanical design but Inoue agreed immediately that this is an essential ingredient. Many complex products are algorithmdriven by their embedded microprocessors. Many have complex user interfaces and multiple internal states, both mechanical and electronic. Thus a sense of algorithms is essential for a comprehensive design approach.

out before trickling it down to the undergraduate curriculum. This can take many years and lacks a departmentwide strategic approach. It also lacks a methodology for removing outdated material, leading to crowding in the undergraduate syllabus. At Todai the graduate curriculum has so few classes that this method may not be available.

The methodology at Todai is not totally clear to me, except that the pressure came from within the department, apparently, and not from the dean. The methodology for selecting elements of the new curriculum is also not clear, except that the chairs focus on areas that are related to their research. This creates expertise but does not guarantee that generic material will be taught or that the students will obtain a balanced education.

What is clear is that the change was quite radical and has defined “mechanical engineering" in a way that would be almost unrecognizable at many schools in the United States.

A related question is why algorithmaware students don't go into computer science (CS). The simple answer is that there is no CS department in Todai's engineering school! There is a CS department in the School of Science, Postscript however. I did not learn much about what it teaches. The electrical engineering (EE) department in the School of Engineering deals mostly with power and information systems, including signal processing and vision. Most U.S. universities have CS departments or CS divisions of EE departments. At Todai such competition does not exist, leaving a clear path to ME for such students who also have a mechanical bent.

Discussion

Many universities in the United States have trouble changing their curricula radically, in spite of obvious reasons to do so. At the Massachusetts Institute of Technology (MIT) I saw leading professors introduce new material at the graduate level and prove it

This report was distributed in draft form to many U.S. educators and drew an interesting response from Prof. Masayoshi Tomizuka, Vice Department Chairman of Mechanical Engineering at UC Berkeley. Tomizuka did his Ph.D. research with me at MIT in the early 1970s after getting his SB and SM from Keio University in Tokyo.

Tomizuka said (and Dr. Kozo Ono of Hitachi Construction Machinery Company and a Todai graduate confirmed) that Japanese undergraduate engineering education is broad and shallow. One is exposed to many fields shallow. One is exposed to many fields but learns almost no deep knowledge. Tomizuka said he was surprised by the depth of the MIT doctoral qualifying exams and had to study very hard to learn the material to pass them.

Tomizuka also said that curricular reforms like those described above actually happen fairly frequently in Japan and do not represent the revolution that is implied by my report.

The comments of Tomizuka and Ono raise a difficult question: if Japanese university education in engineering is so shallow, how come Japanese product engineering is so good? The answer apparently lies in the additional education the young engineers get on the job, plus such factors as lifetime employment, extensive use of past design data on new designs, and the length of time an engineer keeps the same job responsibilities.

An advantage of this kind of education is that it sets the pattern for “universal experience," meaning that an engineer more easily learns and practices many fields during his/her career. "Mechanical" engineering graduates do not feel a professional commitment to mechanical engineering but rather to their employer, who may alter their professional concentration as a result of assignments and training. These alterations apparently do not cause much discomfort. Cross-trained engineers perhaps can understand each other's design problems, making concurrent engineering easier to implement.

U.S. engineering students, on the other hand, devote a lot of their education to becoming "mechanical engineers," for example, and might feel their school time was wasted if their employer tried retraining them as EES. Similarly, U.S. companies expect new hires to function productively soon after being hired, just because of the focussed character of their education, and would not think of retraining them to a different field. Engineers thus rapidly become specialized and less able to communicate with engineers in other fields.

Therefore, education, career paths, and company training (or lack of it) are symbiotic in both countries.

PROF. FUJIMOTO, DEPT. OF ECONOMICS, UNIVERSITY OF TOKYO

7 June 1991

Background

Prof. Takahiro Fujimoto is a recent graduate of Harvard Business School, where he collaborated with Prof. Kim Clark. They have already written a book and are continuing their joint studies. The subject is management methods in the automobile business world-wide, with a prime focus on the product development cycle; specifically, the issues studied are lead time, development productivity, and total product quality. The questions they addressed

are:

1. How do auto companies in Europe, Japan, and the United States organize the product development process?

2. What are the main bottlenecks that cause the process to take long time?

3. What regional differences are there in the length of the process?

4. What managerial techniques and organizational practices account for the relative differences found between companies and regions?

Their past studies have focussed on body and chassis design of cars and the design of the manufacturing systems that make those parts. Although engines generally take longer to design than bodies, bodies change more frequently, so their development time, which is long as well, tends to dominate the design cycle for individual car models. Future studies will focus on engines, which have relatively more engineering analysis and less (though NOT no) esthetics behind them, and semiconductors, which have even more analysis

and no esthetics at all. (For more about engine esthetics, see below on designing the sound of a car.)

Research Methodology

Clark and Fujimoto pursue a style of research that is often called “determining best practice." It differs markedly from accepted academic research in most fields and at most business schools, where scholarly research is typically about economic models, financial analysis methods, or investment strategies. This method is practiced by conducting field studies, sending out questionnaires (typically 80 to 100 questions), and doing some statistical analyses on the results. Thus it is more like anthropology than management science. The results often make fascinating reading and are taken seriously by forward-thinking people in the subject industries. However, the studies often require a leap from "form" to "content" in the sense that the characteristics of a company that can be gleaned from questionnaires and interviews may not be the root causes of the differences between company achievements.

For example, a recent MIT study (Ref 1) revealed that car factories with more democratic management styles, more flexibility in job classifications, and more automation were able to make more cars in more model varieties with fewer defects than factories that lacked any one or two of the above characteristics. Yet it is not obvious that if one started up a new factory and included all these characteristics one would automatically obtain high quality, high model-mix production capability. The authors of the study show statistical significance in their survey results but this study method does not really lend itself to the typical methods of statistical analysis. For example, there is no way to create a control set; you cannot create a double blind environment; you cannot establish a controlled set of features whose effects are to be tested

in isolation since numerous other features are operating out of your control which may not be captured in the questionnaire.

Fujimoto says that companies are willing to participate in such studies because their own data are disguised and they are able to benchmark themselves against regional averages. Sometimes they can decode the data a little and discern information about single companies, but such opportunities are limited.

Importance of This Kind of Research

It is essential that different methods of managing the design and production process be identified and compared. It is well known that different companies in the same industry can differ in productivity by as much as a factor of two. Increasing attention is being given to the length of the product development cycle in particular because it gives so much competitive advantage to companies. They can follow the market as it changes, absorb new technology into their products sooner, and build experience in their technologies and in the design process itself at a higher rate. The ability to “climb the learning curve" of design and technology faster has been cited by Gomory and Schmitt (Ref 2) as a major factor in national and international competition based on national productivity and economic strength.

Two distinct approaches to speeding and improving the product development process have been identified. These may be called roughly the management-intensive approach and the technology-intensive approach. The former emphasizes management and organizational methods while the latter emphasizes use of computer-aided design and similar techniques. Companies using managerial methods employ design teams from different technical disciplines. Those pursuing

the technical track make use of computers for design or for internal communication. Others use modern computer-controlled manufacturing equipment to obtain better uniformity of output, hence higher quality. No company uses one technique alone. The differences are matters of degree only.

In some industries it is taken for granted that design technology is essential. Modern complex products require so much data to describe their design, so many calculations to determine performance, and so much attention to detail during fabrication and assembly

developed. Fujimoto believes that this is the right sequence for obtaining the best processes, at least in the auto industry. Others feel that the technical challenges of complex products are so great that without help from computers, no company can surmount them merely with management techniques.

Therefore the book is still open on what is the best approach or mix of approaches to improving product development methods. Clark and Fujimoto have learned a great deal about the auto industry, and a summary of their findings follows.

phase activities are finalized. Intense communication is used to minimize the effect of such risks, as discussed below (Ref 3).

Importance of Overlapping Design Functions and Interfunction Communication

The overlapping design process is often called "Concurrent Design" (CD) (Ref 4). Other synonyms are Concurrent Engineering and Simultaneous Engineering. All refer to a process during which product designers and production system designers exchange infor

that human capabilities are severely Fujimoto/Clark Major Findings mation in order to maximize the prochallenged. In the semiconductor industry, human capabilities in design and design checking were far surpassed over 15 years ago; only use of large scale computing permits modern microprocessors to be designed at all.

American companies in particular are known for using computers heavily in many aspects of design and production whereas Japanese companies are not so well known for this. In fact, some of the most productive Japanese companies, such as Toyota (the Just in Time (JIT) method for making cars) and Ishikawajima-Harima Heavy Industries (IHI) (the moduler method for making ships), achieved their famous production efficiencies without large scale use of computers for factory management. Since Japanese performance in these industries sets the standard for the world, it is somewhat ironic to find, at least based on current outside knowledge, that computers do not play a leading role.

The purpose of my own study in Japan has been to pursue the question of how computers are used in the design process, not the factory management process. It appears, however, that Japanese companies have spent the last 10 to 20 years refining both design and manufacturing processes independently of computers and are now applying computers to the methods already

Fujimoto and Clark approached their study as a management problem and asked management questions. They did not, in fact, investigate use of computers, either in extent or kind of use. The implications of this are discussed below. At the same time, Fujimoto knows that CAD/CAM has had a major effect on car development time, referring to Toyota's finding that wide introduction of CAD terminals reduced lead time half a year.

Their basic finding is that Japanese car companies on average take a year less to design a car than U.S. or European companies and that this shorter development time results from two factors. The largest one is overlapping of phases of the process that normally are accomplished sequentially. The second factor is that some of the phases are accomplished faster, although some of the time saved is used to perfect the design rather than to shorten the overall process further. The main method by which overlapping is achieved, according to Fujimoto/Clark, is via intense communication between product designers and production tooling designers. This communication allows the tooling designers to critique the design and to begin key phases of their own work. This process can be risky because second phase activities start before first

ducibility of the product. In more sophisticated CD environments, factors such as marketability and field repairability are also taken carefully into account. The contrasting situation, which occurs in the defense industry as well as in high performance companies such as Mercedes, is that product performance dominates design, with all other factors being secondary or not considered during design proper. They are considered later, during production system design, or not at all. This is called "throwing the design over the wall."

The trouble with allowing performance to dominate design is that the other factors cannot be served except by including them in the original design or by changing the design later. But changing the design is prohibitively expensive once it is complete. A design often represents a long chain of interdependent decisions and roads not taken. Changes that are more than cosmetic may threaten to unravel the entire chain. Producibility or repairability problems that are discovered after design are said to cost 10 times as much to fix as if they were discovered when the design was only drawings; if discovered during production the cost is 100 times as much. In addition, poor producibility and repairability lead to low quality and a bad image for the product in the marketplace.

« iepriekšējāTurpināt »