Simulation details

[EagleEye]

As I begin to unveil our kitchen...

Oh yes, it`s very interesting to read more about you`re "magics". Please continue!:) Maybe you could post some dev screens with your software...for better imagination?

 

http://kartsim.blogspot.com/ - A Karting sim that a good friend of mine is developing. They're doing much the same level of physics modelling for their engines & fuel etc.

Hmm, interesting. Added to my bookmarks.:smilewink:

[loche]

The systems is modelled using physical level of modelling and not a statechart. As we mentioned before the engines is modelled using thermodynamics model similar to the models used for engine design. Hydroulics and fuel system based on elements and equations of liquid flow.

This handling allows to have damage model parametrical and not scripted that is more flexible and convenient to implement.

We don't need to describe what must be with an engine if air dump valves stays open when the engine is working at TO power - just leave the valves open and the model will do the rest.

The engine icing is simulated throttling the compressor. Engine deterioration is simulated modifying some parameters of its parts.

 

There is no need to describe what will be if fuel pump 27 V bus is down. Just kill this bus and the system will do the rest... regarding its state and states of its valves, tanks and pipes.

Also we can drill a hole in a tank or a pipe, foul a filter...

Though it seems to be more complicated this handling is very fruitful allowing to use its advantages after you complete the first step of modelling.

Etc, etc...

 

Indeed, statechart-simulated system reacts adequately only within a frame of several predefined scenarios. However in some cases drawing statechart is much more faster and easier way to obtain desired object behavior rather than doing detailed modeling. Consider a simple valve. Having input voltage, it's slowly opening until full open. Without voltage it's slowly closing until fully closed. As you can see, observable behavior of valve is very simple. There is no need to model locking mechanics or it's electric motor or turbulence of liquid, flowing through it. And statechart diagram easy suites simulation needs of such kind of objects - I believe I can draw it for 15 minutes :smilewink:

 

The other case, where statecharts might be usefull, is complicated non-interactive processes, which could hardly be modelled directly. For example, current, voltage and power alternations during engine warm-up. The diagram of electric currents during such activity looks puzzly because it's affected by many different factors like chemical processes of accumulators discharge. In this case we just coding several "scenarios", each is characterised by specific outputs, and there is one control statechart which switch from one scenario to another.

Edited November 13, 2008 by loche

[Yo-Yo]

Indeed, statechart-simulated system reacts adequately only within a frame of several predefined scenarios. However in some cases drawing statechart is much more faster and easier way to obtain desired object behavior rather than doing detailed modeling. For instance we have simple valve. Having input voltage, it's slowly opening until full open. Without voltage it's slowly closing until fully closed. As you can see, observable behavior of valve is very simple. There is no need to model locking mechanics or it's electric motor or turbulence of liquid, flowing through it. And statechart diagram easy suites simulation needs of such kind of objects. I can draw it for 15 minutes.

 

The other case, where statecharts might be usefull, is complicated non-interactive processes, which could hardly be modelled directly. For example, current, voltage and power alternations during engine warm-up. The diagram of electric currents during such activity looks puzzly because it's affected by many different factors like chemical processes of accumulators discharge. In this case we just coding several "scenarios", each is characterised by specific outputs, and there is one control statechart which switch from one scenario to another.

 

Of course you are right. Finding the level of model detalisation is the art of simmaking. :)

I never get deeper if it is not useful to the gameplay or it does not make the modelling of functionality easier.