AixLib contains models with experiment annotation that have parameters without bindings and without explicitly set start values.

[casella]

Dear AixLib developers,

please check the latest OpenModelica regression test report. For the latest released version of AixLib (AixLib) and for the development version of AixLib (AixLib_dev) there are about 35 models that fail since we fixed ticket OpenModelica/OpenModelica#10386.

According to the latest version of Modelica Specification, i.e., 3.6, when you simulate a model all parameters with fixed = true should have a binding equation (either provided as a default in the components, or set by modifiers when instantiating them). The only exception is parameters that have an explicitly set start attribute - in that case, if a binding equation is lacking, the start attribute is used to set the parameter value and a warning should be issued.

The first failing case is AixLib.Electrical.Machines.PVInverterRMS, which is actually a component, not a model that should be simulated. In this case it is OK not to have a binding for some parameters, as they will be provided when instantiating the component, but you should remove the experiment annotation, since it makes no sense to simulate this component by itself.

In all other cases, some parameter bindings are missing. You can

• add a default binding to the parameters in the component definitions, or
• add an explicit binding when you instantiate the components or,
• add an explicit start = 0 attribute to the parameters in the component definition, which will be used as a fallback value, with a warning being issued.

This will make those models compliant with the Modelica Specification in force, and get them running again in OpenModelica.

Thanks!

[FWuellhorst]

@casella Thanks for raising this issue, we will fix it asap!