Lapas attēli
PDF
ePub

6. Better communication

100 times larger. Second, there are many
applications and engineers don't read
manuals, so good user interfaces are
needed. Third, technology and software
are always evolving and constant renewal
is needed. These descriptions do not
apply to business applications.

Historically these have been pursued
in the above order, starting in 1960
with data processing and 1965 for spec-
ifications and parts trees. CADI, Nissan's
first 3D CAD system, was installed in
1965 and used for making line draw-
ings of body panels. It was extended Future CAD/CAE Needs

into CADII, a surface modeler, by 1974
and used from that time to generate
NC data for die making. While CADII
was originally just used as a drawing
tool, it has now become a CAD/CAM
tool with many applications described

above.

By 1980 parts data trees, die designs, and test data were starting to be merged into one computer system. By 1985, three programs had emerged, one for test data, one for parts lists, and one for CAD data. There is still no common database for all of this, but that is their goal for the 1990s. In common with this is the push into concurrent engineering, which to them means doing more jobs simultaneously, including manufacturability analyses during early design, before any prototypes are built. Another goal is "production CAE," meaning computer engineering aids for production engineers.

The final goal is to put this common database and set of programs out on the network for all their worldwide offices to use. This was begun in the late 1980s and continues today. They say that when production CAE can proceed in parallel with car design, they will have achieved Simultaneous Engineering.

The main improvements they seek are global information sharing (design release and CAD data), global engineering system sharing (software, not just data), and global business information sharing (now done by fax). They apparently have had some trouble justifying some of this effort because they were at pains to contrast their work with typical business computer applications. First, CAD data volumes are

When I asked about this, there was laughter. Their answers were a strange mix of the mundane and the far out.

Katoh emphasized that you can't predict much in the car industry, in spite of the many years and many models under their belts. So much depends on experience and they mine this lode by keeping teams small and lines of communication open. Yet this practice cannot be continued forever, he says. Ways must be found to leverage this experience. Expert systems are not the answer. They are too limited in the kind of input they permit and the forms of knowledge they can store. What is needed is some easy way to put knowledge in and an easy way to get it out.

For example, existing design for assembly methods require too much information and too much detail about the parts before an analysis can begin. How can it be done at the concept phase when only the performance goals are available?? How indeed?

They want everyone to be linked on one network so they can easily send data to each other. This is merely a matter of buying it. However, they want any new program they buy to be instantly compatible in operating system and data format. This is not so easy. In general, Ono of Production Engineering would like tools that provide endto-end analysis of production problems and design/cost estimates of production and assembly steps. This requires Summary a long term effort at unifying software and databases, plus integrating engineering analysis and test data.

Another desire is to "standardize" design processes. This was a bit hard to interpret. Apparently they mean that they need a way to overcome the working style differences mentioned above. A suggestion is to make design tasks more orderly or procedural so everyone will know what to do and know what the other person is doing. It also means studying each design task, such as locating the washer tank in the engine compartment, and deciding in advance what the main issues and tradeoffs are so that they can be addressed systematically and in the same style each time the problem is faced, regardless of who is doing it and what country it is being done in. (At Hewlett-Packard they call this ensuring that the result is “a HewlettPackard product" adhering to the company's standards and capable of being made in any of its plants.)

Nissan is an important example in any study of use of computers in engineering design because, like other large Japanese companies, Nissan does not depend totally on the software vendor community to provide it with software that the community thinks is useful. Instead Nissan has great control over the "technology transfer process" that creates CAD software. Thus its priorities are worth noting in detail.

1. Continue the process of maturing the use of computers. This began as mere data processing in 1960 and has now become CAE. It is in the process of becoming CIE (computer integrated engineering), which means wide area sharing of software and design/test/manufacturing data files. In the next decade it will become CIM (computer integrated manufacturing), meaning direct links between marketing, design, manufacturing, and sales around the world.

2. Increase the ability to capture the experience of people who are currently not well supported by computers so that this expertise can be used by less experienced engineers. Much study of the design process is needed to identify efficient ways of performing many of the tasks, identifying tradeoffs early, getting quantitative models of them, and getting answers when there are still merely preliminary designs available.

3. Increase computer support for manufacturing and industrial engineering activities in order to optimize manufacturing operations and avoid discovering design mistakes after prototypes are built.

4. Find a way, if possible, to use com

puters to help spread the Japanese method of design to designers in other countries. The "fuzzy" method of job overlapping has resisted exportation so far.

HITACHI CONSTRUCTION MACHINERY CO.

11 July 1991

Background

Mrs. Whitney and I visited Hitachi Construction Machinery (HCM) Co. and had a short visit at the nearby Hitachi Mechanical Engineering Research Laboratory (MERL). Our host was Dr. Kozo Ono, whose main research activity for several years has been multi-axis force-torque sensors and robots that use such sensors for grinding and deburring. He collaborates with Prof. Hatamura of Todai and funds research there on similar topics. Ono recently became General Manager of the Technical Research Laboratory. This was my second visit there.

This company is part of the Hitachi group but is an independent company. A strong point of the Hitachi group,

says Ono, is that the member companies have close communication with Hitachi's research and development laboratories. They conduct joint research, and member companies receive research results which they convert into products. They also follow the parent company's lead in CAD/CAM, as I learned during this visit. MERL is part of Hitachi proper rather than a group affiliate.

The construction machine business is mature but not in the trouble it was a few years ago when I last visited. Yet HCM has been diversifying for at least 5 years into such far-flung areas as nondestructive testing equipment for the electronics industry, YAG laser products, force-controlled robots, piezoactuators for "nano-robots," and so on. Their main product lines include excavators and cranes, tunneling machines, digital-analog controllers for same, and several lines of small excavators that they make under the John Deere label.

The company has an image problem, namely, that it makes old-fashioned equipment. This hurts them in hiring new graduates and, apparently, in selling their main line of equipment. This latter impression can be gleaned from their introductory video, shown to the two of us alone on a super wide screen in an ultramodern 300-seat auditorium. It was called "Crossings," meaning technology combinations to create imaginative new products. The exclusively commercial (nonmilitary) use of these products, such as for environmental projects, was emphasized. The video showed extensive use of welding robots, lasers, ultrasonic testing of welds, computer simulation of crane performance, use of flexible manufacturing systems to make parts, and generally "adoption of new technology as soon as it becomes available." The video claimed that CAD and CAM are linked directly, but this is not so, according to Ono and based on answers to my questions.

Products and Design Methods

The company's main plant covers 430,000 m2 and a subassembly plant nearby covers 186,000 m2. Production capacity in excavators is 1,700/month. These are mainly tracked vehicles of medium size including front hoes, back hoes, and clamshell excavators. No large draglines are made as far as I know. A typical excavator has about 2,000 to 3,000 parts, but bought items like engines (mostly from Isuzu) count as one part.

There are about 400 designers, of whom about 150 work on excavators. Approximately 40% of the designers are university graduates. About 10 to 15 products are under design or redesign at any one time. They agree that this is a very heavy workload for a small group of designers. Of the 400 total, about 300 have access to CAD, but the penetration is about 1:2.9 with the objective being 1:2.3 by this time next year, at which point they will be "finished" facilitizing for CAD. The approximately 125 CAD terminals include 27 Apollo workstations with the DDM 3D modeler. The first four Apollo terminals were bought in 1986. More recently they have bought 109 seats comprising Sun 3s and a few SPARCS with “Advance CAD," a 2D CAD product of the CTC Company of Japan. CTC wrote a translator to convert DDM data. All of these workstations are on one or more LANS and there is a link through exactly one of these workstations to a Hitachi mainframe.

The product redesign cycle is about 4 years. However, everyone in the industry is exhausted by this pace and a "consensus" seems to be emerging that the cycle should be stretched. This "feeling" has emerged from informal discussions (at conferences and trade shows?) that designers and managers have with their counterparts at their main competitors Komatsu and Mitsubishi Heavy Industries.

Use of Computers in Design and Manufacturing

• Avariety of kinematics, noise, flow,
heat, vibration, and similar analysis
programs. Since most of these could
have been purchased, I asked why Other Comments
they were developed in-house
instead. Two answers emerged: some
were developed for special purposes
not available commercially, while
others were developed for sale.

the group uses, but HCM's workstations
are U.S. made, not Hitachi's own.

HCM is relatively new to CAD and has surprisingly little CAM. When I visited 4 years ago I was impressed by the robot gas metal arc welding of the main structures of cranes and excavators. The robots use simple sensors to find the weld gaps, which are beautifully flame cut and polished. In this Visit to HCM's CAE Laboratory way, an enormous amount of production welding is done almost unmanned. However, the parts are cut out by handprogrammed flame cutters and the robots are similarly hand-programmed. Since the parts in question are made of simple planar sections, it is surprising that there is no link between the simple 2D CAD and the CAM of cutter and robot programming. Mr. Moroshita, the director of CAD/CAM, said that until they install a data management system (a commercial item) there will be no such links. For this reason, the entire visit focussed on CAE, that is, analyses of mechanical performance of entire cranes or large subassemblies.

Visit to Hitachi MERL CAE
Laboratory

First we were shown a scrapbook of hard copy examples. This consisted almost entirely of output from SDRC's I-DEAS package versions 3.1 through 4, dated 1987-91. One example used ProEngineer Level 5. These consisted of pretty shaded images, analyses using IDEAS + ADAMS, several FEM studies of stresses in welded shovel buckets during digging, and so on. One interesting study was of the dynamics induced on the excavator and its tracks while swinging the arm rapidly. Another showed the excavator driving over a bump and experiencing oscillations in the arm and its hydraulics. Yet another showed what shape a bent hose would take depending on the internal pressure. I was also shown a demo of the robot simulation program Cim Station. The robot was "welding" a complex The robot was "welding" a complex item.

This laboratory has 40 programmers who work mostly in FORTRAN and C. Their "products" are mostly analysis programs for determining various per- Future Needs and Trends formance factors about cranes and excavators. Some products are developed for sale outside the company. We

saw:

• Geometric modeling, based on converting a 2D drawing to a 3D wireframe model

• Automatic mesh generation for FEM studies (written at the request of Hitachi's Software Laboratory and intended to be sold commercially)

HCM is clearly just starting down the road to CAD/CAE/CAM. I believe that this reflects the company's oldline background. On the other hand, in only a few years they have explored many meaningful areas in product performance analysis and will probably continue to do so.

In other respects, HCM is typical of the nonautomotive companies visited: they use the best U.S. workstations and software, and their engineers work until late in the evening. Much of the software is what the parent company of

Ono's English is very good and he is quite open, so some other topics of interest came up.

(1) Is it true that Japanese university engineering education is very broad but not very deep? I have been told that the reorganization of Todai's Mechanical Engineering Department looks like an upheaval, but in fact it just adds new electives while eliminating old ones, there being no required core.

His answer is that engineering education is indeed broad and shallow, but it exposes students to a lot of different technologies and leaves them malleable. When they enter a company, their initial training shapes them as the company wishes. Some standard training is given every new hire but in later years the manager decides what specific training each employee should get.

[I happened to read an article in a private circulation newsletter called Japan Design Today, covering graphic arts and product styling, that described Sony's training program for new designers (stylists to us). This is apparently a bit like boot camp in the sense that group bonding is a major aim of the 1-year course. The teachers are practicing designers who take time off from ongoing responsibilities that they must make up later in the evening. Design in this context includes carefully choosing exterior materials, colors, textures, "handfeel," "button-feel," and so on. The young designers are taught the "correct" view angle from which to draw a video camera. A student who chose his own view angle found that his model camera had been confiscated overnight and replaced by a Polaroid shot taken from the correct angle. In this way, Sony molds its new hires.]

Ono told of his company's difficulty hiring electronics and software engineers. To remedy this, they took the best of the mechanical engineers who showed up and promptly retrained them

ISHIKAWAJIMA-HARIMA
HEAVY INDUSTRIES (IHI),
AERO-ENGINES DIVISION,
TANASHI PLANT

in electronics and software. In the United 16 July 1991
States, such an act would cause the

student to regret his university educa- Background
tion and its typical commitment to a
discipline. Apparently this does not
happen in Japan. U.S. companies rankle
at having to supplement the education
of their new hires; Japanese companies
take advantage of the opportunity.

(2) Since I am told that the Japanese language is not very precise or quantitative, how come Japanese people are so good at science and engineering?

His answer is that most of the time Japanese people do not notice any disadvantage stemming from their language or methods of thinking. However, communication can sometimes be slow. Most Japanese sentences have no subject, and verbs do not distinguish number or gender. Also, he recalls translating one of his papers into English and discovering in the process that one of his Japanese sentences made no logical sense. He had been unaware of this when he first wrote it. He also added that logic is not the only component of practicing engineering. There is also "informal communication, group work, and a different way of thinking than the so-called logical way."

The shop is well equipped with modern equipment, mostly of Japanese manufacture, including several Toshiba five-axis NC machines of the type sold to Russia, some Huffman laser drilling machines, and the most modern Messer Griesheim e-beam welders. The FMS for aluminum parts comprises five Mitsui-Seki machines that run 24 h/ day in a temperature-controlled room.

My hosts were Mr. Nakajima (now
at the Mizuho plant) and Mr. Ochiai.
Both are managers of production engi- Software in Use
neering departments at their respec-
tive plants.

IHI's main business is shipbuilding
and heavy construction. Aircraft engines
are a small part. The aircraft engine
division has 3,600 employees and sales
of over $1 billion. IHI built Japan's first
jet engine during the war, making the
first test flight in a small fighter plane
in early August 1945. Since then IHI
has designed only three engines on its
own (plus many research and develop-
ment engines), relying mostly on man-
ufacturing licenses from General Elec-
tric, Rolls-Royce, and Pratt & Whitney
for most of its production. Products
include aircraft engines, stationary gas
turbines for generating electricity, gas
turbines for ships, and rocket engines
for space. Including several wholly owned
for space. Including several wholly owned
subsidiaries, IHI can produce all of the
main advanced technologies required
for such products, including intricate
investment castings and single crystal
or directed grain castings of turbine
blades, numerical controlled (NC)
machining, flexible manufacturing sys-
tem (FMS) operation, laser welding
and cutting, and some composite mate-
rials for nose cones that are transpar-
ent to infrared rays. They are especially
proud of their ability to laser weld
titanium. IHI is also a prominent par-
ticipant in Japan's proposed Space
Station module.

I was shown two examples of computer use, a shop floor scheduling system they wrote themselves and CATIA for CAD.

Shop Floor Scheduler. The shop grinds surfaces and dovetails of turbine blades and has about 100 workstations. There are at least 100 kinds of blades and about 300 blades in each batch. The shop handles more than 300 lots per month according to Mr. Shibata, who demonstrated the software. It takes about 4 months to process a lot. Each blade apparently requires dozens of work steps. We noted that some of the jobs currently in the shop are 50 to 75 days late and scheduled to stay that late or get even later. The usual response in such situations is to send some work outside, but IHI hopes the scheduler will permit them to improve efficiency and keep the work in-house.

The scheduling system was installed in April 1991 and they are still improving it, assessing how well it works and learning how to use it. A major feature is its excellent user interface, permitting the shop supervisor to view data about jobs pending or in progress from several viewpoints: by job, by batch, by job step, by machine, and so on. Clicking the mouse on a job reveals a window containing the details of the process. Thus the progress of the job or the scheduled job sequence planned for each machine can be viewed easily.

The objective of the software is to decide which machine should do which task step on which batch next. Different batches are waiting for machines, and different machines, capable of doing only some of the pending tasks, are waiting for work to be assigned to them, while other machines are still busy.

The task of the software is "resource selection and task assignment," a familiar task in management science algorithms. Typical criteria are to minimize the lateness of the jobs and/or to minimize the idle time of the machines. Occasionally a special high priority job must be wedged in.

The software was written using a commercial expert system shell to which IHI added its own rules. A clear explanation of how it worked was not available, but apparently it runs a number of simulations to verify its predictions. It first assigns jobs without regard to capacity limitations of the machines. Then it tries to shift work from overloaded machines to underloaded ones. Finally, it identifies jobs that cannot be assigned and puts them in a group to be assigned to outside contractors. This appears to be an imitation of how the human schedulers run the shop and does not utilize any of several algorithms in the scientific literature that could potentially solve this problem better.

CAD and CAE. Design is almost 100% paperless in the engine department. CATIA is used for 3D modeling, and CADAM and microCADAM are used for drafting. The usual CAE functions, such as mass and inertia properties, are done with CATIA. CATIA is also used for making nice pictures to show top management and customers. Piping design and interference analyses between pipes and engine structures are done in CATIA. Data are transferred to SDRC's solid modeler for thermal and stress analysis. CATIA is also used for preparing NC programs.

However, the data created by engineerdesigners are not sufficient for NC programming and must be augmented. But modifications are difficult because CATIANC output is not as easy to edit as APT, which has a simple line-by-line format. Also, I was told that CATIA's surface modeling capability is limited since surfaces tend to come out with low amplitude, low frequency waves.

Facilities consist of a Fujitsu supercomputer and a mainframe for supporting aerodynamics, structural and stress analyses, plus an IBM 3090 and 50 graphics terminals running CATIA and 100 PCs running microCADAM.

Today's applications for CAE are structural analysis and weight and moment analysis. Future applications are interface between engine and airframe, assembly procedures and instruction manuals, and a link to CAM.

Medium term they are rethinking the information flow in the design process, hoping to shorten the path. At the moment, parts such as blades are designed in 3D and then 2D drawings are made. In the shop, these drawings must be reconstituted into 3D data so that a workplan (NC programs) can be made.

Longer term they are hoping to create optimization methods for designing blades, disks, and shafts. (GE already does such things either routinely or in their design R&D laboratory.)

They wish 3D modeling could support dimensioning and tolerancing and have kept to 2D drawings because of their ability to represent such information. None of my hosts are aware in detail of the extensive CAD/CAE/CAM developed in the auto industry, although they have seen some things demonstrated at trade shows. They are afraid of in-house software development since they think it would be hard to maintain and improve the programs.

In addition, in spite of many years of cooperation between IHI and GE, the IHI people were unaware of the Defense

Advanced Research Projects Agency Initiative in Concurrent Engineering (DARPA/DICE) project. GE is a major participant in this project and design of turbine blades is an important demonstration of DICE capabilities.

Product Design Methodologies and Long Term Developments

Even though use of computers in the design process at IHI is a bit behind the state of the art, Nakajima is a careful thinker concerning organization of design processes. He shares Nissan's view that overlapping of tasks is essential to shortening the design cycle, and cycle shortening is a prime objective for IHI. Ochiai feels that having a common database without transcription errors is just as important if not

more so.

At present, they are using the development of the new HYPER 90 engine as the base project for design process improvements. This engine design is being funded by MITI and involves many Japanese and foreign companies. A hypersonic civilian transport plane is the eventual target. IHI wants to shorten the development time from a typical 30 months to 20 months. Note that this time is a very small part of the time needed for R&D of a new kind of engine which, together with manufacturing development and certification, can take up to 8 years.

IHI wants to use the methods and technologies of Concurrent Engineering to achieve this time reduction. The basic approach is, as stated above, to overlap the tasks of design, process planning, and manufacturing development. The most important action is to make sure designs are evaluated promptly for their impact on manufacturing, assembly, cost, and quality. Second, IHI must make a direct link between CAD and CAM. Third, they must improve overall data management, including learning how to manage

« iepriekšējāTurpināt »