the computer assist

In less than 40 years, a novelty has grown into a mainstay of engineering practice.

As ASME celebrates its 125th anniversary this year,
Mechanical Engineering will run articles each month highlighting key influences in the Society's development. This, the 10th in our series, examines the emergence of computers in engineering in the 1970s.

Harry Armen, who would later serve as president of ASME, worked as a structural engineer at Grumman and was one of the engineers at the meeting. Armen told his boss that the engineering staff was responding to pressure from the U.S. Navy and other customers to upgrade the computational capabilities of both hardware and software in an effort to speed design processes and automate analytical techniques, with the aim of reducing developmental costs.

"Our customer had become very sophisticated, and its higher performance requirements exceeded what we were able to do with standard tools on the drawing boards," Armen recalls. "The company had no choice but to comply."

Pushed in many cases by systems users, particularly military and other government customers, the computer began arriving at corporate engineering workstations in the mid-1970s. Before then, the new technology was not widely used in the engineering department. Despite the promising demonstrations of Nastran, the powerful structural analysis system developed at NASA, and the presentations of the computer pioneer John H. Argyris at a series of conferences organized by the U.S. Air Force beginning in 1965, computer-aided engineering was not standard practice heading into the decade of the 1970s. The only computers at the disposal of engineers were mainframes such as the IBM 360 Series and pocket devices like the HP-35, which let practitioners perform routine calculations, store sales data, and process inventories.

Only a few forward-looking technology companies invested in computers, primarily mainframe systems. While bringing the benefits of data management and real-time processing to engineering, the mainframes were also a headache. Engineers spent countless hours correcting functional problems and writing programs.

The programs, particularly large-scale ones involving difficult computations, were executed in batch processing mode, meaning that the engineer had only one attempt each day to run the programs. In the days of the mainframes, "developing programs to execute the algorithms was a laborious task," said Karl S. Pister, former dean in the College of Engineering at the University of California, Berkeley.

Finite Element Analysis

UC-Berkeley in the early '70s was a focal point for the development of computer-based structural analysis. There, Edward Wilson, Ray Clough, Klaus-Jurgen Bathe, and other pioneers worked on finite element methods, techniques that geometrically render a structure into a series of elements having prescribed properties.

Finite element analysis, or FEA, methods proved to be extremely useful in engineering. When the discrete elements were assembled to form a structure, or continuum, what resulted were sets of defined equations used to represent the behavior of a system under load. FEA tools involved the development of algorithms that helped engineers analyze such physical phenomena as deflections and stresses in complex structures. The pioneers of FEA carried out numerous applications, including analysis of aircraft components such as wings.

Engineers in 1965 test processors for the IBM System/360 Model 40 computer. A mainframe, it was one of few models available in the early days of CAE.

In 1973, Bathe developed the structural analysis program SAP IV, which was released at no cost to interested users. "The SAP IV source code offered numerical algorithms that engineers could apply to their own applications," said Bathe, who is now a professor at the Massachusetts Institute of Technology. "The code was distributed worldwide."

While military contractors and large corporations like General Electric and Hitachi could not ignore the role of finite element analysis in their technology programs, during most of the 1970s the general engineering community did not capitalize on the full potential of the computer. One main problem was lack of access. The prohibitive cost of the early mainframes and glitches in the software deterred many companies from investing in computer technology; those firms that installed systems often did not give priority to engineering departments. Another problem was education: In the 1970s, computers were relegated to the back burner as engineering schools instead earmarked funds for the construction of ultramodern laboratories and design centers.

Trends started to shift toward the latter part of the decade. As computer speed and memory increased, finite element analysis began to spread beyond research centers and military labs to diverse industrial applications. Before long, midsize firms and even small engineering consultancies embraced this powerful tool that brought speed and precision to systems analysis. By 1980, the computer revolution finally had taken hold in engineering, spurred by FEA.

The Emergence of CAD

The ASME Computers in Engineering Division was established in 1980, and by then computer-assisted engineering was gaining strongly in engineering practice. As engineers became increasingly adept at using finite element methods for simulation, a group of software companies was poised to take computer technology to the next level. These firms introduced computer-aided design and computer-aided manufacturing systems, enabling engineers to perform 2-D and 3-D modeling on computer screens. Then, the vendors augmented the CAD/CAM programs with powerful parametric systems to automate tasks associated with the generation of geometry, signaling the beginning of knowledge-based engineering.

"We see the development of computers in engineering starting with finite difference and finite element methods, which support simulation and analysis, and progressing to CAD/CAM, which supports design and manufacturing, and then to knowledge-based systems," Harry Armen said.

Knowledge-based systems allowed engineers to implement rules, which are entered in the computer's database to be used as constraints during the design of products and systems. Such rules could be used for optimizing a design against factors like cost, weight, manufacturing constraints, and standards and regulations. Technology firms that invested the necessary time and resources in the development of knowledge-based engineering systems were able to integrate design, testing, and manufacturing, in the process automating tasks, saving time, and reducing costs.

Two ages of CAD: A screen from AutoCAD 9, issued in 1987 (left), puts fewer options and features at the designer's fingertips than does the screen of AutoCAD 2002 on the right.

ASME members embraced knowledge-based engineering, believing that there was great potential in knowledge-based methodologies to support modeling, design, and manufacturability. Members conducted extensive research in the field, and presented their findings at conferences as well as in technical journals.

The 1988 ASME Computers in Engineering Conference had 23 sessions on knowledge-based engineering. Also, the Society's flagship publication, Mechanical Engineering magazine, developed "Computers in Mechanical Engineering," a regular feature offering updates on knowledge-based systems. Indeed, ASME played a major role in advancing the field.

Computer capability in engineering has grown exponentially since the 1970s. Engineers solve hundreds of thousands of equations in minutes on laptop computers. Still, according to experts, engineers have not realized the full potential of computers.

"A future challenge will be to integrate computer-aided design and computer-based analysis and to render the two technologies fully compatible with models embodying analysis and geometry," said Farrokh Mistree, associate chair of the George Woodruff School of Mechanical Engineering at Georgia Institute of Technology in Savannah.

According to Armen: "The engineering community must advance computer technology to the level where engineers can validate a structure completely using computational tools, without having to develop physical models and prototypes. The next step is cognitive information processing using the computer to actually mimic the attributes of the human brain."

John Varrasi is a senior writer in the Public Information Department of ASME in New York.

home | features | breaking news | marketplace | departments | about ME back issues | ASME | site search

© 2005 by The American Society of Mechanical Engineers