On 02.09.2011 22:49 meekerdb said the following:
On 9/2/2011 1:06 PM, Evgenii Rudnyi wrote:
On 02.09.2011 20:29 meekerdb said the following:
On 9/2/2011 5:17 AM, Evgenii Rudnyi wrote:
I have summarized my answers in respect to that the simulation
technology falls short of the simulation hypothesis at
http://blog.rudnyi.ru/2011/09/simulation-hypothesis-and-simulation-technology.html
It could be considered as some small empiric case study.
My practical experience with simulation has often been
disappointing. It works best when abstracting out a relatively
small number of relations and simple physics. But on the other
hand, what can be simulated has vastly expanded over the 50yrs of
my career. So when this or that ambitious project fails I don't
conclude that the trend is stopped.
Brent
Modern simulation software is actually not that bad. If one keeps
things simple, then the chances to get the right answer for the
first time are quite high even for a design engineer. I mean that
default settings and default meshing are working reasonably well.
This is one of the reasons that the simulation business grows
extraordinary well: design engineers can solve for example a linear
structural mechanics problems by themselves, the bachelor level
suffices.
The problem in the real world however is not just simulate at any
cost but rather to earn money. The IBM case is interesting exactly
from such a pragmatic viewpoint. If the business does not bet
anymore on monstrous supercomputers, then it is an interesting
sign.
When I talk to engineers working on electromobility, I mention that
theoretically one could think of simulating the whole hybrid
vehicle at once (structural mechanics, heat transfer, CFD,
electromagnetics in a single simulation) - they like it. Yet, they
do not bet on that, they are pragmatic.
Of course part of the reason they don't bet on that is that they
understand the vehicle pretty well and so they are confident that the
CFD won't interact with the instrument panel lights and the
suspension won't interact in some unforseen way the the engine.
However, if they were designing a robotic rover to investigate the
surface of Titan for example then they would be much more interested
more comprehensive simulations.
Just last year I wrote a heat transfer simulation for a vehicle and I
got different answers from the manufacturers simulation. It was
because he had neglected the fact that accelerations change the
internal convective heat transfer.
This is one of the reasons why CFD is popular nowadays. Yet, modeling of
boundary layers is still a tricky business. The default settings for
meshing are working but not all the time.
In general, industry is definitely interested in more comprehensive
simulations. The question is however how to reach such a goal and how
much it will cost. Say electronics now is a part of a product and it
must be simulated as electronics, then thermal, EMC, durability (it must
work 10 years), etc. A typical simulation scale (substitution level)
starts from nanometers on a chip and then it goes to decimeters at the
whole system. This is why engineers are not afraid of AI and simulation
hypothesis.
Evgenii
--
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To post to this group, send email to everything-list@googlegroups.com.
To unsubscribe from this group, send email to
everything-list+unsubscr...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/everything-list?hl=en.
