"Engineering" of Automobiles

[Robert_A_Brady@rohmhaas.com (Robert A Brady)]

As a mechanical engineer, I find it amusing how much credit is given to 
automobile engineers.  The thread on modifications has included many 
postings from individuals who claim or suggest that the engineers of our 
Alfas went through painstaking analysis to determine the natural 
frequencies of components and the synergistic behaviors of the assemblies.  
Well, not to burst anyone's bubble, but... my impression and anecdotal 
evidence suggests that the "engineering" of automobiles has been (until 
recently) VERY empirical.  That is, rarely has rigorous analysis been done 
to components to optimize them to a given set of performance criteria.  Let 
me cite a couple of examples:

In the early 60's the US space program was having difficulties with their 
engine development for the Saturn V rocket.  Pressure pulses do to 
combustion/fluid dynamic interactions were causing the engines to fail.  So 
faced with this problem, did they whip out their slide rules and calculate 
the resonances and propose a solution?  Unfortunately, the reality was that 
mathematical models of the physics were not sufficiently developed to 
predict such resonances.  So, the engineers built and tested scores of 
baffle configurations until they found one that worked acceptably.  So much 
for "rocket science."

A friend of mine works at GM's Tech Center, on powertrain development.  A 
couple of years ago I asked him to find out what software GM uses to model 
their exhaust manifolds.  After he spent some time investigating this, he 
told me that they DON'T model their exhaust manifolds at all.  This friend 
also worked at the Proving Grounds in the mid 80's.  During that time, GM 
wanted to put the "cross-fire" fuel injection system on camaros.  So did 
they fire up their computers to model the system first?  No.  They pulled 
some FI systems from the production floor and plunked them into the 
camaros.  The result?  They struggled for weeks to understand why the 
systems wouldn't perform properly.  After weeks of tearing their hair out, 
they realized that the units they had pulled from production were 
defective.  So much for GM's engineering.

The engineering community's ability to model mechanical systems for stress, 
vibration, heat transfer, and flow has advanced by leaps in recent years.  
Automobile manufacturers probably DO model much of new cars.... it's 
certainly cheaper than prototypes.  Some phenomena still defy modeling, 
such as turbulent combustion.  So, was my '74 gtv's suspension components 
modeled to insure their performance to a given standard.  Well, considering 
it was developed in the 60's, I'd venture to guess not.  How about my 70's 
developed Milano Verde.  Probably not either.  What about the '98 Dodge 
Intrepid?  Probably, yes.  

My point?  Until very recently, I believe that automotive engineering was 
largely an effort of trial and error and extrapolation from experience.  To 
claim that cars developed in the 60's and 70's benefited from rigorous 
analysis of their systems and components is being naive.  I'm fairly 
certain that with my PC I can do more stress analysis on my GTV in a week, 
than it's developers ever did during it's lifetime. 

Can anyone substantiate or refute my views with REAL experience from the 
front lines?

I'm interested in the responses.

Bob Brady, P.E.
Yardley, PA
'74 gtv
'88 verde