Physics

[ckuo@sprynet.com]

The torsen thread has been very interesting reading the past few months, but one 
of the things that has been bothering me is the use of the term 'physics'. 
Unfortunately, the debate has been using 'physics' to propose an absolute 
assertion, when in reality what I see being presented is a model. Models are 
based on physics. If we are talking about law of thermodynamics, conservation of 
energy, etc., that is physics.  We can generate models based on the 
understanding of physics. In this case, a model of an Audi with a torsen 
differential. Indeed, one of the common tools used in engineering is modeling 
and simulation of systems to predict behavior. What Scott is actually presenting 
is his model, not physics. Being an engineer, the use of the term of 'physics' 
has been very bothersome. I do, however, believe Scott has been making a 
noteworthy effort to try and generate a model.

Although I myself do not know the true answer to this spider debate, I do know 
what needs to be done is the development of a more accurate model of the 
torsen/chassis/drivetrain system. Unfortunately this is not easy, as an 
automobile is a non-linear real-time system, which is hard to predict using a 
linear static model that has been used to this point. For the behavior that we 
are trying to prove or disprove, it is extremely difficult (perhaps impossible) 
to use a simple Excel spreadsheet model to come up with an answer. 
Unfortunately, it does not appear if anyone has a good mathematical real-time 
model of the torsen (correct me if I'm wrong). Obtaining this model would be a 
key to the solution, as is accurately modeling the chassis, including variables 
such as slip angle. 

As an electrical engineer, I have used Excel (for quick and dirty first order 
approximations) and workstation simulation tools to simulate behavior of 
systems. Unfortunately, in some cases, even with a $120,000 Mentor workstation, 
simulations do not always fully represent real life. There can be so many 
variables that are unaccounted for, that the accuracy of the simulation can be 
affected. A good example- predicting weather! (Simulations can at least keep 
engineers from pursuing a dead-end path, however.) For an automobile, these 
variables include drivetrain losses, distortion of chassis and suspension 
members, etc. Often, the best way to define behavior is to perform real-life 
testing - although  there is debate in this area on this list as well.

Anyhow, sorry for this long-winded expression. I do think that the debate needs 
to remain objective from both sides, and from what I see, there is no solid 
answer as of yet.

Off my engineering soapbox....

Thanks, Chauncey Kuo
ckuo@sprynet.com

85 5KS Avant
84 Scirocco II
Rossignol Extreme monoski