Modelica: Equation balance in hierarchical model

I have the following simple hierarchical model comprising a reaction inside a chamber. The reaction model simply sets the mass rate to the mass, and it's connected to the chamber that encloses it.

connector Conn
  Real mass;
  flow Real massRate;
end Conn;

model Reaction
  Conn conn;
equation
  conn.massRate = conn.mass;
end Reaction;

model Chamber
  Real mass(start = 1);
  Reaction reaction;
  Conn conn;
equation
  conn.mass = mass;
  der(mass) = conn.massRate;
  connect(reaction.conn, conn);
end Chamber;

Now to my mind, there are in total five variables: mass, reaction.conn.mass, reaction.conn.massRate, conn.mass, and conn.massRate. And there are five equations: two in the equation section, viz.

conn.mass = mass
der(mass) = conn.massRate

one from the reaction

reaction.conn.massRate = reaction.conn.mass

and two from the connect equation:

reaction.conn.mass = conn.mass
reaction.conn.massRate = conn.massRate

So we have five equations in five variables. The solver logic can simply combine these equations like so

der(mass) = conn.massRate = reaction.conn.massRate = reaction.conn.mass
= conn.mass = mass

and since the mass is given a start value, Bob's your uncle. The mass should increase over time. But my simulation environment (Wolfram|One) says

Error: Simulation model is not globally balanced, having 1 variable and 2 equations.

Can someone please explain what is going on? Note that if I take the reactor outside the chamber (reversing the sign in the reaction to give conn.massRate = -conn.mass), it works as intended.

Solution

This is described in section "4.7 Balanced Models" in the Modelica specification.

The problem is that chamber has a top-level connector conn and there is also one equation related to that; e.g. conn.massRate=0; corresponding to a default-connected connector in a hierarchical model. Not only is it one equation too many - it is also wrong.

In addition if you used the Chamber as a sub-model, and wanted to add an external flow into/out of the Chamber you could connect to conn - but the equations wouldn't make sense.

When you move the reactor out of the chamber you avoid that dangling connector. Another alternative, that might be clearer would be to have two models such as:

  model Chamber
    Reaction reaction;
    MassPart source;
  equation 
    connect(reaction.conn, source.conn);
  end Chamber;

  model MassPart
    Conn conn;
    Real mass(start=1);
  equation 
    conn.mass = mass;
    der(mass) = -conn.massRate;
  end MassPart;

(assuming I got the sign-part right). Here you can see that the MassPart and Reaction in themselves both have the same number of equations and variables, when you add one equation for the flow-variable of the connector.