Are we becoming "sloppy" Engineers 12

[PSE]

One of the most (or perhaps the most significant) tools that has been placed at the engineer is the computer. This tool has become increasingly powerful over the years. I wonder though, if this tool is making sloppy engineers out of us. With ever more intricate algorithms, the capability of the engineer to independently assess the results of an analysis I feel is decreasing. At the same time, computer simulations are bringing products to market with fewer design iterations and even fewer prototypes. I thought to float the following questions for discussion.

Are we becoming overly reliant on a device and accompanying software that uses algorithms too complex to analyze, written by individuals who may or may not have any familiarity with the underlying engineering principles? How many bugs are still within such programs that could affect results? In recent memory, I cannot think of a software program or operating system that was bug free.

I haven't marked this post but will visit it regularly (I get enough e-mails). Interested in your general thoughts.

For a humorous close, from some of "Murphy's Laws"

Undetectable errors are infinite in variety in contrast to detectable errors which by definition are limited.

If builders built buildings the way programmers wrote programs, the first woodpecker to come along would destroy civilization.

There is always one more bug.

Never program and drink beer at the same time.

Regards,

 

[Tobalcane]

Computer software over the years has been growing increasingly useful and a standard at design reviews. I would say that we are in the generation of engineers that rely on analysis software to give us a better flavor of what “may” happen when we test. Now I say “may” because the software is not reality. An engineer may become sloppy if the person does not do hand calculations to make sure that the software is doing what it is suppose to do and does not do test to back up the analysis.

In my opinion analysis software is more like a tool to foresee problems that we may encounter not the definitive answer. I use Pro-Mech for my static and dynamic analysis and TAS for heat transfer analysis. During the analysis to make sure that the software is doing what it is suppose to do (garbage in garbage out), I spot check the results with hand calculations especially when it comes down to the final prototype design (when we are ready to cut metal). Now, the results of the analysis software are fantasy until it is correlated with test data; this is where the rubber meets the road. Once you have test data, the computer analysis can be updated accordingly.

I have been on a program where they went from computer to fabrication. That was a disaster. On top of that, it was a fixed price job so there were no extra funds to use (only overhead) to fix problems. At the end the project it cost more for the company than what we sold it for to the customer.

I don’t think the software is making the engineer sloppier, but making them even better. Back in my college days, one of the assignments for the class was to write code for the computer to solve problems. Back then I could not understand why would Mechanical Engineers program code to do problems, this is what computer sciences and electronic engineers did back in the early 90s. Now I have an appreciation of the professors that they had some foresight that we mechanicals would need an understanding of how analysis software worked.


Go Mechanical Engineering
Tobalcane

 

[BobPE]

Tobalcane hit a good point: <I don’t think the software is making the engineer sloppier, but making them even better.>

Software is making the engineer better, but at the same time, it is allowing the non-engineer to enter the field. This is the person that connot do the hand calculations to check the programming and is "engineering" without a thorough understanding of the garbage in/garbage out concept.

I have seen posts here before that talked about laypeople using our engineering tools. It is becoming more common place as evidenced by failures that make it to the news, and more frightning, failures that do not.

I really believe it is the people not qualified to be engineers that use the tools we created for ourselves that are contributing to the sloppiness (and downright danger)of engineering.

Good thought provoking post...

BobPE

 

[rerig]

If I had to do a 9x9 matrix analysis by hand, every time, to prove a composite repair required 1 additional repair ply, I would probably go crazy. Hence the "2 additional repair plys" rule given in most Structural Repair Manuals.
But, taking it back further...without the FEM analysis available by computer, there wouldn't be a Boeing 7E7.
If you have no idea what the results of the Program is going to be, then you shouldn't be designing with it. (the collective "you", not you personally)
The design is in my head...the fine tuning and proof comes from the computer.
So, is the " 2 additional repair plys" rule lazy or convenient.
If you go off and use a Computer Program without knowing what the results should be, the results won't make you question the input and program calculations.
I have been surprised by a program's results before and have spent many hours verifying the correctness of the program. It sure does bring back bad memories of doing calculations by hand, every time!
Without the present day computers and programs there would be no glut of Engineers...and Slide Rule sales would be on the increase. (but that is another thread)
Yes, there are bugs...there is also a Factor of Safety.
That is why we use Engineers.
Any IT department that purchases an Engineering Program, without the participation of the Engineering Dept. deserves to have the product fail. Any Engineering Dept. that uses a Program on Face Value of the Marketing provided on the box deserves to fall on their face. Someone must proof the program, but before I relied on the program I would do my own proof testing on known input - outputs.

"If an Airplane had to fly with the paperwork used to build it...it wouldn't get off the ground."

Denny Crain....I mean, Rerig.

 

[epoisses]

Computers are definitely a great help, but it increases the need for one very important thing: the engineer's intuition, or the ability to judge the outcome of a calc/simulation. Without redoing the calc, the engineer should be able to say "this looks funny" and reject the calc. I've met so-called brilliant people who were able to think only "behind the decimal point" and not in orders of magnitude. If a calc result were a factor 1000 off because of a kilo or a milli missed in the inputs, they wouldn't be able to tell. Before the computer, these people were simply dumb. Nowadays they are mass murderers...

 

[SlideRuleEra]

I think that there are two factors at work:

1. Computer software allows engineering students to muddle though and "produce" answers with little, or no understanding of problems. Before computers were common, those same students would have had no way solve these problems (using slide rules or basic calculators). The education system rapidly convinced these individuals to change to another major.

2. Unrelated to this is the increase in truly needed engineering fields that seem "softer" on number crunching - one example may be environmental engineering. This subtle change may have (incorrectly) created the impression that engineers (in general) do not have the "hard core" problem solving skills that were once part of the stereotype.

 

[TheTick]

Like any tool, a computer will serve to magnify the character of the engineer (or poser) using it.

 

[Tobalcane]

<<< Computer software allows engineering students to muddle though and "produce" answers with little, or no understanding of problems. Before computers were common, those same students would have had no way solve these problems (using slide rules or basic calculators). The education system rapidly convinced these individuals to change to another major.>>>

I beg to differ. Well I can only say for my time in college (early 90s) and I can safely say for today that the college students in mechanical engineering are not using the FEA in any of there classes to solve or learn engineering theories. Just in the classes where they are learning “how” to use the analysis software. In fact in my college I know that they are still making their students write code (Fortran) to solve mechanical engineering problems. Believe me, this is no easy effort. It is one thing to understand the theories and calculus, but it is another to program code to make the computer do the theories and calculus and come out with correct results. In fact, in order for them to prove their program works is to show the hand calculations on how they came to a particular or slue of answers. Hey this sounds familiar; this is what I do now!

In a way I do feel that the analysis software was not really meant for the engineers, but to show pretty pictures to management to influence there decision. It is one thing for an engineer to show all his equations for stress and shear stress to an audience, but man, when you throw up that slide with all the neat colors showing where all the stresses are, it can answer a thousand questions at once.


Go Mechanical Engineering
Tobalcane

 

[rerig]

By shear coincidence, today, I came across a fortran punch card and mylar punch tape for my collage 3 axis machine. Late 60's. (I really need to clean out my desk more)

In '97 I took the Boeing "Advanced Structural Repair for Engineer's" course. They wouldn't let us use the computers. Scientific Calculators only. I really needed my Slide Rule, but I probably forgot how to use it by now.
I didn't exactly feel Lazy....I just missed the ease of calculating on the computer. (and my home grown formulas)

 

[GregLocock]

In the auto industry I'd say computers have been a huge benefit, although I can't say I like the resulting cars as much as the old ones.

But... no significant model is uncorrelated. In my case the suspension is put together with a nasty mix of hand calculations, spreadsheets and sophisticated modelling programs. Every step of the way results are checked back against previous models, and sideways against prototypes or test parts, and the results from all three analytical methods are expected to agree to some extent.

The danger is that we are moving into a stage where the computer model has been 'right' for about 6 years - that is, we haven't learnt a great deal more from building the early physical prototypes, as they have performed pretty much as expected out of the box.

The temptation is to avoid building these rather expensive and troublesome cars, and design it all on the computer. It would cut about 3 months out of a program, and save millions of dollars.

The question then becomes how do we ensure the integrity of the models, even if they are as good as we think they are?

The answer is you have to build robust models, which is interesting. A robust design may not be as efficient as a less robust design, but will be more tolerant of errors in assumptions. That's quite funny in a way, it's almost going back to cookbook engineering.


Cheers

Greg Locock

 

[SlideRuleEra]

Tobalcane - I am sorry that I did not express myself clearly. There certainly continue to be numerous engineering students who learn and comprehend the full extent of science, math, and other disciplines that form the basis for engineering. What I did not make clear was my belief that before computers/software became readily available, OTHER college freshmen/sophomores (in engineering) quickly realized that they were in the wrong field - short of copying someone else's work, they were literally be unable to do homework, pass tests, etc.

For reference, my university (mid/late 1960's) had one computer available for student use. I was taught to program in Fortran II. Programming consisted of typing the input onto the now almost extinct IBM paper cards (about 25 of the mechanical punch-card machines were available for 15,000 students). You then turned in the cards at the Computer Center, left, and returned a few hours later to pick up your printed results. I remember it taking several days up to a couple of weeks to "get the bugs out” of the math and data entry. The key factor was that the computer was of no use at all in other classes – computer time was to valuable for routine use by undergraduates. All of this was typical for an engineering education at that time.

I majored in Mechanical Engineering, outside of these two Fortran classes, EVERYTHING else was done either on a slide rule, pencil/paper or hand drawn graphical solution – NO exceptions. If a student was “sloppy” in his setup of a problem, manipulation of a slide rule or drafting technique – you got the wrong answer. There was no other way.

My point is that things have changed for the better over the last 35 years, but SOME individuals who could not pass an engineering curriculum then, may do so now by using (but not understanding) the current technology.

 

[PSE]

It is gratifying to see that respondents talk about independent verification of model results. A couple of things had prompted me to start this thread. One was a posting that GregLocock had posted on finding errors within textbooks (errors can be there even after editorial reviews would give credence to the argurment that they are certainly possible in released software). The other was a discussion/argument that I had overheard about the differing result of a model run on two different software packages as to which one was right. Apparently no thought had been given to actually sit down and run through it by hand. The result of schedule pressures perhaps?

I too went through the engineering program in the "early days" of computer use. Mention FORTRAN these days and newer engineers or software folks look at you like you are a fossil! We always independently checked results. I wonder if the "new generation" of engineers feel similarly.

Regards,

 

[CajunCenturion]

From the perspective of a computer engineer, this is a quite interesting thread.

Let me be the first to say that I see a great deal of sloppiness invading the software engineering profession.

Good Luck
--------------
As a circle of light increases so does the circumference of darkness around it. - Albert Einstein

 

[chris9]

I used to be a mentor to design engineers who were learning FEA. They were astonished that just by changing the boundary restraints you can double the stress values. The real skill is not operating the software but making sense of the results. The main problem is that the people who run companies don’t appreciate this and employ technicians to operate software to do work that was previously done by professional engineers.

 

[Tobalcane]

Hi SlideRuleEra,
For students who find out that engineering is not the field that they would like to get into usally gets weeded out in the math, chemistry, and physics classes (remember the phrase by the professor “look to your left and look to your right, by the end of this class most of them will be gone”). By the time they get pass these classes; they usually know if they like doing the analysis needed for the engineering classes.

In a way I do believe that the current technology at the college level will actually help them learn the engineering theories. I mean software like the Excel spread sheets and Math CAD, even their calculators which are programmable, or programming in Fortran. These are software where the user has to know what they want before getting the software to do the task. Not going to the software to find answers. Because of this, the student will have to “think” while programming. It is almost a mental exercise.

I kind of see the software as another weeding out tool. The students have to know the engineering theories and know how to use the different software on top of that. Not to pick on the older engineers, but do you know how many times I hear the older engineers trying to do stuff in pro mechanica and scream “I can’t get this stupid box to work!”. They know what they want, but can’t get the software to do it.


Go Mechanical Engineering
Tobalcane

 

[CajunCenturion]

I kind of see the software as another weeding out tool. ...

That is shocking to me. How many very talented engineers are lost to the profession because they are unable to utilize a certain software utility?

For other than computer engineering professions, in my opinion, to deny a person "membership in the engineering profession" because they can't get Excel to work is sending the wrong message, and focusing on the wrong skill sets. I couldn't care less whether or not a civil engineer can use AutoCad. I would rather be confident that the engineer has properly evaluated the stress points. By all means continue to weed them out in the physics, chemistry, and math classes as appropriate.

Good Luck
--------------
As a circle of light increases so does the circumference of darkness around it. - Albert Einstein

 

[Tobalcane]

CajunCenturion,

It is not only the engineers. It other majors in college too. The last in my family that went to college was my youngest sister. When she started back in the late 90s, it was mandatory that all of her freshman class have a lap top. She was a Bio major and she had her own software master. If you did not realize it yet, we are in the computer industry. Actually I would be more scared if the person was not computer literate especially in Excel spread sheets.


Go Mechanical Engineering
Tobalcane

 

[CajunCenturion]

There is no question about the use of computer technology on the campus. As a software enginner, I am well aware of it.

But technology is and should be a tool, not a requirement.

Good Luck
--------------
As a circle of light increases so does the circumference of darkness around it. - Albert Einstein

 

[Tobalcane]

In today's day and age technology “is” a requirement. I do believe an engineer can not get a job if that person is not computer literate. At the most that person should know how to use Microsoft Office (i.e. Word and Excel and maybe Power Point), how to get on to the web and utilize it, things like that. If that person only used a pen and pad to communicate and actually mailed it to you and you can only mail your response by pen and paper back to the person that is not how we communicate today, or crunch numbers on graph paper and then present that to the customer as an influence tool, I can not see this person working in today’s cutting edge companies. I know for most mechanical engineering job requirements for the defense industries require that you know some sort of FEM, CAD, and Microsoft Office Suit.



Go Mechanical Engineering
Tobalcane

 

[GOTWW]

You people are lucky to even complain. The morons that have been breastfeed by the government virtual pearl harbor, idiocy, have outsourced all computer functions, leaving only the sacred Microsoft office. This is no joke. I am back to paper and pencil, but I can still do some plotting by using the HP calculator, and feel lucky to scrounge up log-log paper.

As far as canned software, the (black) buttonbox stuff does not doit for me, generally. Mathcad type mathematical program that you write yourself is a good way to verify the results. As for sim's, the high dollar FEA and nonlinear simulations are good.

Sincerely, screwedbynmci