I've got a problem with ambiguous solutions within my models. The simple form of what I'm thinking about is like this:
model testquadratic
Real x;
Real y(start=0);
equation
x = time;
x = y^2;
end testquadratic;
If you set the "proposal-start-value" to start=0, my dymola6 plots the positive solution +sqrt(x). The same thing happens for start=+2 or any othe positive value. But when I say start=-2, then I get the negative solution -sqrt(x).
As it is only a proposal-start-value, y(time=0) starts at y=0 and not at the given proposal-start-value. Strange but okay.
########################################################################
How can I control such accidental / arbitrary behavior? Is there a "standard" way to deal with ambiguous solution trajectories?
########################################################################
The long form of this problem is the following: I'm still on hydraulics - currently on pumps which I model via a characteristic map like:
power_in = func1 ( pressure_head , flow ) (1)
Together with a hydraulic load being described by
pressure_head = func2 ( flow ) (2)
it's possible to compute the flow (iteratively) when defining a certain power_in.
In order to simplify things at the beginning, I thought of a ideal pump map like:
power_in = power_out = const * pressure_head * flow (3)
Now giving a value for power_in, theoretically, I may receive more than one solution for the combination of the to equations (2) and (3). One with negative flow and pressure_head and one with both being positive. Dymola plots only one of them.
Of course, it's the one I do not want 
- the negative curves - which correspond to a pump pumping into the contrary direction.###########################################################
Thank you for your help,
Hannes