"Virtually Edison" Feature Article, October 2004

virtually Edison

Analysis software makes it easy to experiment early and often. It doesn't eliminate the hard work, but it can get more out of less.

by Jean Thilmany, Associate Editor

Wasn't it Thomas Edison who made the quip about success? That it's 10 percent inspiration and 90 percent perspiration.

It makes you wonder what Edison might have explored if he had access to analysis software. Computers don't do away with perspiration, but virtual testing lets engineers extract more information today from much less physical prototyping than they could in the past.

Modern analysis technology allows engineers to answer what-if questions that would have been impractical, or perhaps impossible, to manage in Edison's day, according to Stefan Thomke, an associate professor of business administration at the Harvard Business School.

What if the earth quaked near a dam you designed? How would the dam be affected? How could you design around that possibility?

You could model the soil in a computer and analyze its ability to remain stable under a variety of conditions.

"Today's new technologies let experiments happen quickly and easily," Thomke said. "Many new possibilities are created through the what-if experimentation these technologies allow for. When testing is very expensive, we don't try things out. These technologies let us do that."

For large designs like a dam, engineers rely mainly on analysis software.

Thomke is the author of Experimentation Matters: Unlocking the Potential of New Technologies for Innovation (Harvard Business School Press, 2003), which looks at how new technologies affect experimentation at businesses. He says many businesses aren't properly set up to take full advantage of new product development tools.

"Engineers can innovate, but as they come up with these new ideas, they need them to be tested," he said. "They wait for feedback from analysts, but there's a wall between analysts and engineers. Analysts think of engineers as customers sometimes, and they have a long queue of projects they need to look at. An engineer's project could get put on a list, and he'd get a report back three months later."

That waiting game does not drive innovation. Because technology has moved so quickly, companies have a hard time figuring out how to best incorporate new analysis technologies into their established testing and analysis structure.

Edison's was a classic way to innovate. He pushed his engineers to experiment early, often, and always; he frequently claimed his measure of success was the number of experiments that he could fit into 24 hours, Thomke said.

But in Edison's day, experimentation took place quite differently. In the inventor's laboratory, experimenter and machinist sat across from each other. The engineer handed the machinist a part, who made it. Then they tried it out. The engineer immediately redesigned the part based on that trial, then handed the new design back to the machinist, who made the new part. Then they tried it out again. You get the idea. That back-and-forth continued until the experimenter was happy with the result.

Edison pushed engineers to test their designs as thoroughly as they could. No one is about to suggest that physical prototyping and real-world testing will ever become a thing of the past, but building prototypes has become more expensive over the years as products have become more sophisticated. Successful companies today allow for experimentation that strikes the right balance between computerized simulations and Edison's hands-on prototyping method.

A TEST RUN (LITERALLY)

Engineers at LifeFitness in Franklin Park, Ill., follow Edison's testing dictum, although they've upgraded his design-test-redesign method for modern times by merging simulated analysis with real-world testing. Managers follow Thomke's advice by making sure they keep the lines of communication between engineering and analysis departments open and up to speed.

"We all work together from a design and analysis aspect," said John Rogus, a mechanical engineer at the company and part of the newly formed mechanical engineering analysis group. His department works closely with the design group to analyze parts as they're designed.

The company makes workout machines like treadmills and stair-steppers for both cardiovascular and weight-bearing workouts at home or at the gym. Of course, the company field-tests its equipment at the places where it's used the most.

"We have arrangements with different clubs for field testing," Rogus said. "We do in-house testing where different people of different weights, sizes, and conditioning use the machine, then we measure different aspects of the machines with different people on them in different conditions."

But the company can't get by on field-testing alone. It also relies on a full complement of analysis and reliability tests. For analysis, the gym equipment maker uses Cosmos software from Structural Research and Analysis Corp. of Los Angeles and a seat of analysis software from Ansys of Canonsburg, Pa.

"We use analysis to determine and predict where the equipment is going to fail and for comparative analysis—to say this geometry is better than that geometry in terms of stresses and things of that nature," Rogus said.

His company doesn't run FEA studies on every part. "That would take forever," he said.

"But if a part fails during testing, we go back and look at it with FEA to see why it failed and try to understand it better," he said.

Engineers and analysts work closely with the reliability engineers, who are also in the loop. The latter group ensures that the machines are as reliable as possible. They're the engineers who perform in-house accelerated and component testing, working mainly with already-built machines.

TOO BIG TO BUILD TWICE

Sometimes a project is so big or so complex it can't be prototyped for real-world study and Edison's back-and-forth method just won't apply. When engineers at the FEA technology maker Abaqus of Pawtucket, R.I., were asked to advise on how best to build an earthen dam, they couldn't run out and construct a series of dams for study.

A South African company associated with Abaqus, Finite Element Analysis Services, passed on information from a customer who had a question about an earthen dam. Because Abaqus engineers knew certain problems applied across the board to all dams made from earth, they decided to make a study of the topic and prepare a paper from their findings, said Deepak Datye, an Abaqus engineer.

An earthen dam is made by compacting layers of soil. Thus, designers have to take into account two events that may cause their dams to fail: an earthquake or a rapid emptying of the reservoir that the dam embanks.

With analysis technology, engineers can ask many what-if questions. Engineers at LifeFitness of Franklin Park, Ill., use simulated analysis and real-world testing to find the most workable design.

Engineers in Rhode Island couldn't travel to South Africa to study the particular dam problem close up. And they certainly couldn't build a large prototype dam for study and real-world analysis.

Nor in fact did they need to. They were analyzing problems designers might face in constructing earthen dams in general. Local conditions were of secondary importance. The Abaqus engineers relied mainly on their FEA technology to describe how earthen dams are best designed to avoid the earthquake or rapid emptying problems.

"We analyzed building a dam and filling it with water to find out what the distribution of water pressure does to the dam," Datye said.

Experimentation, analysis, and prototyping—both physical and virtual—can't be sidestepped. But engineers need to strike the right balance between physical and virtual experimentation appropriate to their companies. And they need to ensure open lines of communication between analyst and designer.

"New technologies are revolutionizing how products are developed today," Thomke said. "But, like anything new, they have their challenges."