Conditional equation is not working

[Christian Kral <dr.christian.kral@…>]

When simulating the example Modelica.Electrical.Machines.Examples.DCMachines.DCPM_withLosses machine dcpm1 does not consider brush loss and dcpm2 does consider brush loss. However, in case of dcpm2 the voltage drop across the brush model shall differ from zero during operation of the machine -- but it does not. Consequently, lossPower is zero, as well -- which shall not be the case.

This is a bug in OpenModelica 1.9.4~dev-101-g operated under Ubuntu 14.04, 32bit.

[Christian Kral <dr.christian.kral@…>]

dcpm2.brush.brushParameters.V=0.5

In this case the else branch should have been evaluated:

  if brushParameters.V <= 0 then
    v = 0;
  else
    v = smooth(1, if i > (+brushParameters.ILinear) then +brushParameters.V else if i < (-brushParameters.ILinear) then -brushParameters.V else brushParameters.V * i / brushParameters.ILinear);
  end if;
  lossPower = v * i;

[Christian Kral <dr.christian.kral@…>]

When investigating example Modelica.Electrical.Machines.Examples.DCMachines.DCPM_withLosses I noticed the same kind of issue with the friction model which is not working correctly in dcpm2. In this case the torque ŧau and lossPower shall be non-zero.

dcpm2.friction.frictionParameters.PRef=100

In this case also the else branch should have been evaluated:

  if frictionParameters.PRef <= 0 then
    tau = 0;
  else
    tau = -smooth(1, if w >= (+frictionParameters.wLinear) then +frictionParameters.tauRef * (+w / frictionParameters.wRef) ^ frictionParameters.power_w else if w <= (-frictionParameters.wLinear) then -frictionParameters.tauRef * (-w / frictionParameters.wRef) ^ frictionParameters.power_w else frictionParameters.tauLinear * (w / frictionParameters.wLinear));
  end if;
  lossPower = -tau * w;

[Per Östlund]

I'm not sure if it's actually a backend issue, but let's start there and see.

[Volker Waurich]

Using +d=dumpdaelow gives:

203/203 (1): dcpm2.friction.tau = 0.0    [dynamic]

So it seems that there is already something wrong in the frontend.

[Per Östlund]

Ok, I'll take a look at it then.

[Per Östlund]

Fixed in 10f5341.

[Christian Kral <dr.christian.kral@…>]

Thanks for fixing this issue; I just tested my models. Your solutions helps a lot in one of my projects.

[Martin Sjölund]

Milestone renamed

[Martin Sjölund]

Milestone renamed