Unbounded limiter in MSL 3.2.3 causes negative phi

[Mathadon]

Since MSL 3.2.3 the model Modelica.Blocks.Nonlinear.Limiter has the parameter homotopyType, which sets a homotopy option. By default linear is used, which neglects the min/max bounds. When updating IDEAS to Dymola 2020 this caused the PI controller output bounds to be ignored at initialization, leading to a PI outlet control signal of -10, which then fed into a valve and caused the resulting algebraic loop not to converge due to the negative control signal. I propose to disable homotopy by default in IBPSA.Controls.Continuous.LimPID to avoid this.

[Mathadon]

@damienpicard this can explain convergence problems in your model too

[mwetter]

Disabling homotopy makes sense in this situation.

[Mathadon]

many models seem to be bugging out though.. I'll look into it.

edit: @mwetter does IBPSA already use Dymola2020 for the regression tests? It seems to have a newer version of MSL 3.2.3. The old version does not yet have this parameter and may be causing the models to fail for that reason.

[Mathadon]

@mwetter does IBPSA already use Dymola2020 for the regression tests? It seems to have a newer version of MSL 3.2.3. The old version does not yet have this parameter and may be causing the models to fail for that reason.

[mwetter]

@Mathadon : IBPSA is using Dymola 2020 for the regression tests.