The process needs to change 4

[mloew]

I thought you might enjoy my rant in reply to an article in Machine Design Magazine:

Changes in the Engineering Profession Over 80 Years

I'm compelled to offer an opinion I have about the engineering process. Many years ago, before CAD became commonplace, the engineering process began with engineers developing the products, starting at the system level, with heavy use of models and calculations based on engineering fundamentals to guide the design. It was only much later in the process when 'draftsmen' became involved to render the product so it could be manufactured.

Today, most product is created in CAD as 3D solid models and this geometry is used to simulate the performance with FEA, CFD, etc. and the design is iterated until management is happy with the performance. Final drawings are created off of the CAD model (a real time saver).

Although CAD and the other modeling and simulations tools like FEA have allowed for more rapid and robust performance simulation, it is an iterative process that allows for designs to have been based on created geometry rather than sound, engineering-based design concepts. I believe however there is a significant rush to start modeling too early and often by those without the specific expertise in the functional areas for the system. I'm not saying that CAD should be abandoned, but conceptual development of the system should be considered and developed before significant solid modeling resources are committed to the design realization process. I think it is a great shame that many engineering organizations believe it is acceptable to replace engineering fundamentals and systems expertise with an iterative approach. We have moved from engineer-develop-design (drawings)-build-test-produce to a system of render-simulate-iterate-design (draw)-build-test-produce. The result in many cases is that simulate-iterate approach is the engineering process. It is not. Let's not get lazy with engineering just because we can make good looking designs in CAD early and can quickly make colorful stress plots in FEA.

I'd like to see a marriage of the traditional approach that harnesses the best of engineering expertise and the utilization of engineering fundamentals with the power of the tools we have at our disposal.

Thoughts?


Best regards,

Matthew Ian Loew

http://www.daxcon.com

 

[GregLocock]

Yes. I am fortunate in that my drafties are very experienced and know roughly what sizes are required, but then our designs are largely package limited and the manufacturing process defines a lot of the features in the design.

The way I have worked in the past on novel designs is to do the solid modelling myself, including all FEA and functional checks, and then throw it over the wall to an experienced CAD guy who would then 'productionise' it.

Most of the problem is down to the usual one with engineers, an unwillingness to communicate.

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[MikeHalloran]

The operating theory appears to be that if you put a trained monkey in front of a CAD station running some sofware with the word 'engineer' in the title, that you don't have to employ, or pay, any actual engineers.

I didn't say it was _my_ theory, or that I support it in any way, but I have seen involuntarily departed engineers' salaries redirected to pay for the CAD software, at more than one outfit. No doubt an MBA came up with the idea.

I certainly don't think the world is a better place for it.

Apologies to actual trained monkeys.
Apologies to actual product designers.
No apologies to actual MBAs.



Mike Halloran
Pembroke Pines, FL, USA

 

[IRstuff]

Seems a bit reactionary, "In the good ole days..."

The reality is that many systems being designed today are too complex for simple, or even complex hand calculations. Can't imagine anyone advocating using hand calculations to analyze Taipei101, any of the newer structures that have no symmetry and have complex shapes.

And, while you might find trained monkeys, you'll also find that they can't iterate terribly well without some level of talent.

I see similar things in optical design. No one, but no one, uses hand calculations to design optics today. However, a master designer can iterate a design in hours, while someone less talented or experienced would take days or weeks to iterate the exact same design.

TTFN

FAQ731-376

 

[mloew]

IRstuff,

The whole point was that this is not a "in the good old days we did X". However, much of what we have now is an automation of an old design-build-break-redesign... loop. There is still room to (re)introduce some engineering fundamentals and discipline before unleashing the hoards of designers to crank out CAD parts that stream through simulations.

Architecture is probably better, actually, because there is a lot of oversight on what the lead architect wants. The subsequent effort there is likely far more controlled and reviewed by the experts. I think this is a very different situation from automotive and most other industries.

I just fear that we have substituted tools for much actual process. There is a better balance to be made.

Best regards,

Matthew Ian Loew

http://www.daxcon.com

 

[TheTick]

I put trained monkeys at my workstations to design a new lift mechanism. Stupid monkeys just wrote Shakespeare.

 

[mloew]

Don't feel too bad, Congress tried to get Shakespeare and ended up with the "stimulus" spending bill.

Best regards,

Matthew Ian Loew

http://www.daxcon.com

 

[vc66]

Tick-

How many did it take?

V

 

[TheTick]

Just one. I discovered later he was actually an unemployed journalism major who learned CAD at night school. Later I got a real monkey and noticed a marked improvement in work output and hygiene.

 

[drawoh]

mloew,

As far as I am concerned, 3D[ ]CAD is a design tool, not a drafting tool. If the engineer sits at another desk and visits the drafter once or twice a day, either some other tool is being used to do design, or you have a grossly inefficient process, or both. 3D[ ]CAD is a powerful tool for working out component layout in limited space as well as for working out mechanisms. If I were designing a fluid system such as a chiller, I would do my first analysis by hand, then I would produce a functional schematic. I would move as quickly as possible to the 3D.

If the engineer does his analysis by pencil and then hands a sketch to the drafter, the drafter might as well use a cheap 2D[ ]CAD package. The engineer will have worked around most of of the useful features of 3D.

Back in the days of drafting boards, the drafter had to be good with linework and lettering. The ability to drop the drawing view on the correct part of the drawing page saved redrawing and cutting and scotch taping, later. The mechanical skills required to do drafting have gone way down with CAD. This creates the opportunity for the drafter to provide more knowledge and skill for the engineering.

With 3D[ ]CAD, there are usually several ways to model anything. Your modeling technique affects your ability to analyze your design and make changes. 3D[ ]modeling must be done in context of the design and its requirements. It must be done by a skilled designer.

An unskilled drafter (or monkey) will find one of the many ways to do it wrong. The engineer will be left either scribbling and sketching on a piece of 1/4"[ ]grid paper, or waiting for pieces of metal to come in so that they will have something to hack with.

JHG

 

[KENAT]

Not sure if it's because I started out on CAD or if it's just me but I tend to think 'geometry' first.

However, I've come to appreciate the 'calculations first' approach, having been shown up on at least one occasion because of it, even if it's just in getting a better appreciatio of the loads etc.

There's probably time & place for both, or perhaps more often some reasonable combination.

I do agree there sometimes seems over reliance on CAE in its various flavors.

And as for CAD Monkies...

KENAT,

Have you reminded yourself of faq731-376 recently, or taken a look at posting policies:

http://eng-tips.com/market.cfm?

What is Engineering anyway: faq1088-1484

 

[GregLocock]

Since I work in a mature industry CAE is often all we've got - that is, the program manager's ideal of zero prototypes is slowly becoming a reality - certainly the days of building 40 or 100 integrated engineering prototypes is long dead.

So we use the CAE because it is cheaper and quicker, and we trust it because we spend a lot of time correlating the models, and developing the links between what the customer wants and what we can measure.

When the design is done we build confirmation prototypes off tool and then check that their performance matches the CAE.

Obviously there are still enormous gaps in what CAE can do, so we build attribute prototypes rather than integrated prototypes - that is the car is modified to test a subset of the overall performance, rather than an integrated EP which was a fully functional vehicle with all the new systems in.

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[mloew]

I think in some areas, like vehicle dynamics, the automotive industry gets it right. Most of the time, the dynamics simulation group is making models that are not dependant on CAD geometry and simulating the performance of the vehicle against the functional requirements and the resulting architecture is used to drive the physical design. Many of the basic engineering requirements (including packaging, assembly process and access, etc.) can still be solved before there are detailed CAD models created.

In this process, the right level of engineering expertise is used to drive the design and let CAD models be a resultant of the engineering being done up-front. This is often called Simulation Driven Engineering.

Best regards,

Matthew Ian Loew

http://www.daxcon.com

 

[drawoh]

mloew,

I find 3D[ ]models to be very useful for working out and confirming assembly procedures. If you plan to be competitive in mass manufacturing, your product design and your manufacturing process have to be worked out in parallel, using the same database.

Most of the stuff I design involves packing stuff into the smallest space possible. The sooner I get into scale drawings and models, the better. In SolidWorks, I usually create an imaginary blob, representing the final size of the system. I insert the main components. I insert the covers, and I work out assembly procedure and access. Then, I design the structure. A 3D[ ]model is useful throughout this process.

Once the structure is worked out, I usually delete the blob. Before CAD, I made a lot of use out of 5H[ ]pencils. If your analysis does not require scale information, you can do it before drafting and modeling. If you need scale information, the scale[ ]model is a tool.

JHG

 

[KENAT]

When I was talking about over reliance on CAE I'm talking about managers asking for FEA of a simple right angle bracket with a simple loading case or the like.

Or when the first thought of anyone when concerned about the strength of a component being "just run an FEA on it using the CAD package's inbuilt stress analysis tool even though you have no background in stress analysis or training on the analysis software". The part in italics goes unsaid of course, probably not even considered.

KENAT,

Have you reminded yourself of faq731-376 recently, or taken a look at posting policies:

http://eng-tips.com/market.cfm?

What is Engineering anyway: faq1088-1484

 

[mloew]

Drawoh,

What you are describing is using CAD for conceptual or functional geometry. This is absolutely fine in my process. Plus, it is being developed by an expert in the subject matter. I strongly encourage this and have presented on this at conferences.

My big gripe is the substitution of these methods by assuming because someone is using CAD that a product is being developed. There is simply way too much CAD geometry being developed too early at an unnecessary level of detail by those not properly qualified to do engineering.

Best regards,

Matthew Ian Loew

http://www.daxcon.com

 

[civilperson]

Designers are NOT engineers and companies which turn designers loose on the preliminary design get low quality products when the only decision from the engineers is strength and size of members.

 

[mloew]

civilperson: I could not agree more.

Best regards,

Matthew Ian Loew

http://www.daxcon.com

 

[drawoh]

civilperson,

Perhaps this is more of a civil issue than a mechanical one, but I am not sure where you are coming from. I have been claiming that much of the non-functionality of CAD comes from the use of CAD[ ]operators, rather than designers. Designers should have substantial knowledge of the product, its requirements, of engineering, and of the processes required to build or manufacture the product. There is a continuum of ability ranging from professional engineers, down through people with engineering degrees, technologists with three and two year diplomas and then to industrial designers. Then, there is practical experience, the more, the better. The designer has to know stuff, and they have have know their limitations. At some point, you have to ask for help.

CAD software is user friendly. Anyone with the qualifications to do design, can learn CAD. Most people without the qualifications to do design can learn CAD too, but this does not make them useful.

JHG

 

[KENAT]

I agree with drawoh here, while it may for good reason be different in structural, in many areas of mechanical work much of the design work can be done by decent designers.

While I have concerns about unqualified/suited people using CAE and getting in over their head but not realizing it, arbitrarily excluding everyone with out a Bachelors in Engineering (or perhaps some think Masters and/or PE or equivalent) from taking part in anything more than detailing, in every situation, seems excessive to the point of stupidity.

Of course, getting a working definition of 'decent designer' could be almost as difficult as deciding who gets to be an 'Engineer';-).

KENAT,

Have you reminded yourself of faq731-376 recently, or taken a look at posting policies:

http://eng-tips.com/market.cfm?

What is Engineering anyway: faq1088-1484