Some people are good at certain things, others not so much. That’s as true in mechanical engineering as in any other field that requires specialized skills and knowledge. But what distinguishes an average proficient person from a true expert in, for example, the essential engineering skill of computer-aided design (CAD)? Is it simply a matter of experience, or are other factors such as an individual’s particular approach to a certain task involved? How can we speed up the process of turning the merely competent into CAD superstars?
Alison Olechowski, assistant professor in the University of Toronto’s Department of Mechanical and Industrial Engineering, decided to find out. Working with graduate students James Chen and Yuanzhe Deng, she designed a study to characterize the differences and similarities in CAD modeling approaches, patterns, and strategies between proficient and expert users. The work was published in the ASME Journal of Mechanical Design.
“My research group is interested in improving CAD collaboration and understanding how mechanical engineering designers can work together with CAD to design innovative new products,” Olechowski said. “But when we started digging into that problem, we realized that we’re lacking an understanding of how engineers design individually in CAD, and that’s an important first building block. A lot of the existing research on CAD collects data from student or novice users, but very little exists for what we call proficient or intermediate or expert users. So, we designed an experiment where we would have engineering designers join us on a cloud-based tool.”
Olechowski and her colleagues defined a set of descriptions of novice, proficient, and expert CAD users and then recruited a group of designers. “We invited folks to self-identify in those different levels of expertise. But what our experiment let us do is that we could then verify that,” Olechowski explained. “We also asked the expert users to report the number of years of experience that they had with CAD. Those who self-reported as expert had significantly more experience than those who were proficient.”
Participants were given engineering drawings from which to create a 3D parametric CAD model in four steps in a given amount of time.
“We had a number of metrics about the effectiveness, speed, and the quality of their design, and we could compare these things, so we could we look at the models that resulted from the experiments in and of themselves,” Olechowski said. “But we also looked at all the clicks, the design choices that were made by the users. Because there are many different ways to arrive at one final CAD model.”
The researchers found that years of experience definitely make a big difference. “We didn’t find one golden approach to building a CAD model, but we did notice some differences in user actions and how experts perform as differing from proficient users,” Olechowski said. “The experts were able to initially form a smart strategy and see the problem through from beginning to end, so that in the end their behavior had less iterations, they had to revise less.”
The merely proficient users, however, were found to use more of a trial-and-error process to arrive at their final goals.
“It’s kind of a chicken and an egg thing, but we noticed that the experts were much more effective at reading the engineering drawings that they were given and then translating that into the model,” Olechowski said. “It reminds me of studies that have been done in the past about chess experts. Apparently, if you ask a group of non-chess players and chess experts to memorize where the pieces are on a board that just has randomly placed pieces, they can remember them about as well. But if you put the pieces in an actual gameplay scenario, the experts can remember those much better than the novices can, because they have those connections made and that experience built.”
Olechowski and her colleagues believe that one explanation for the gap between the experts and the merely proficient is the difference between what psychologists call declarative vs. procedural knowledge. Declarative knowledge is basically all about following the instructions and knowing the specific commands and various procedures to do a task, while procedural knowledge goes beyond that to incorporate different strategies or choices.
“If we think about the trajectory from a novice to an expert, so much of the way that we teach CAD is through declarative knowledge,” she noted. “That’s the way most software programs are taught. ‘This is what the buttons do, these are the features and the capabilities.’ Whereas it’s hard to learn procedural expertise. It takes time.”
Ideally, she said, some kind of apprenticeship-style education would help, or being partnered with a great mentor, but such approaches aren’t always possible. “People are busy, and timelines are tight, so you might not get taught that procedural knowledge. And it’s not really written up in books either.”
It’s not yet clear how the results of this study might translate to CAD usage in other areas of engineering, or in architecture, in which problem solving might involve a higher degree of creativity. In this particular study, Olechowski said, “They have to figure out how to do the design, but they don’t have to be creative in what the design is. I would imagine that maybe an architecture audience would be more interested in more open-ended problems.”
She’s looking at such questions for further exploration. “Engineers are often trained to think about quality and manufacturability and cost from the very start,” she observed. “My understanding of architects is that they allow themselves to let go of some of those things, at least initially. So yeah, there are probably some differences there.”
Other angles to examine are the role of aptitudes such as a talent for spatial thinking, and the techniques of CAD users with more experience in newer additive manufacturing methods contrasted with those with more traditional subtractive manufacturing approaches—and how to find ways that designers with different talents and strategies, strengths and weaknesses, might work together better instead of possibly clashing over how to complete a project.
Clearly, there’s no magic formula to make any engineer a CAD expert, but it’s clear that experience is one part of it. This study is an important first step in understanding and facilitating the mysterious process of transforming an average user into a CAD Jedi Master.
Mark Wolverton is a technology writer in Narbeth, Pa.
