Computing practices of new engineers

Software tools such as LabView and Matlab, which are heavily used in many engineering colleges, offer considerable potential for enhancing productivity in manufacturing, Tom Kurfess pointed out in this column in February 2004. Further evidence of this industrial-academic software connection is found in a recent study performed by the CACHE Corp.

By Thomas F. Edgar, University of Texas June 1, 2004

Software tools such as LabView and Matlab, which are heavily used in many engineering colleges, offer considerable potential for enhancing productivity in manufacturing, Tom Kurfess pointed out in this column in February 2004.

Further evidence of this industrial-academic software connection is found in a recent study performed by the CACHE Corp. (Computer Aids for Chemical Enginering). It focuses on the relationship between computing taught in the typical engineering curriculum and computing practices for new engineers in the workplace. Details on the survey of 300 chemical engineering graduates since 1998 (from Carnegie-Mellon University, Clarkson University, McMaster University, and The University of Texas at Austin) are available at www.che.utexas.edu/cache/survey/index_files/frame.html . Survey questions used were similar to a 1997 survey carried out by CACHE ( www.cache.org ), so some changes in the past six years could be noted. Most notable of these changes was the fact that 43% reported six hours or more per day are spent at the computer, compared to 19% six years earlier. Only 30% said less than four hours per day were involved with computer work.

New engineers were asked which software packages they used most frequently, how they are applied, and which vendor products are most popular. For spreadsheet applications, Microsoft Excel dominated. Industry clearly values the use of spreadsheets for a variety of applications (see graphic). The fraction of use for numerical analysis is higher than expected, certainly compared to typical use in engineering undergraduate courses. It is interesting that almost all students use spreadsheets, often with minimal formal instruction in their department.

Interesting trends also occur in other software applications; 27% of respondents use dedicated statistical software, such as JMP or SAS; 26% use numerical analysis software, such as Matlab or MathCAD; 66% use database management systems (Microsoft Access or SQL and Oracle software); and 10% use symbolic math software like Mathematica or Maple. It is noteworthy that even with continued improvements in packages such as Matlab and MathCAD, such software is still used much more heavily in academia than in industry. It is not clear why, although educational licenses are much cheaper than commercial ones. Only 6% use numerical methods libraries, which illustrates their general decline in popularity since the 1970s.

Respondents were asked how much initial training they needed to learn computer skills to carry out a job function; 60% reported less than one month; 20%, 1 to 3 months; and 15%, over 3 months. The primary source of training was: 71% self, 14% colleagues, 8% organization, and 6% training company/vendor. Recent graduates continue to rely on themselves or the help of colleagues to learn new computing skills. This supports the notion that universities should prepare their graduates to “learn how to learn.” The amount of formal training to use computing tools is fairly small.

When respondents were asked whether they wrote software in their job, 38% said yes, with Visual Basic (VB) being the most popular programming language. However, 78% said it was important for universities to teach one computing language to undergraduate engineers. The recommended languages were Visual Basic (33%) or C++ (21%), but 28% said the specific language did not matter. The comment that undergraduates should be exposed to some form of programming is not surprising, even though a minority of engineers write programs in the workplace.

What about the future? I believe that the types of software used will not change much, although in control-related functions, adoption of Matlab and LabView in industry will increase gradually.

Author Information
Thomas F. Edgar is a professor in the Department of Chemical Engineering at the University of Texas, Austin.