I've seen this when you formally have a degree of freedom but the margin is very low because you are close to a limit .
Take for example a dearator: water liquid + water vapor + dissolved oxygen.
Components: 2
Phases: 2
F = C-P+2 = 2. For a given composition, you should be able to set pressure and temperature, but at ppm oxygen concentrations, the system behaves almost as pure water. So once you set one variable (P or T), the margin f on the other is very low. If you set the pressure, the margin on temperature (between the gas phase disappearing and the liquid phase disappearing) is very small. The simulator will let you set an arbitrary set of pressure and temperature because the degree of freedom checks, but the solution in the L-V equilibrium zone is not feasible.
Same happens when you are close to an azeotrope. The exact azeotropic point reduces a degree of freedom (you cannot specify pressure, temperature, and being at the azeotrope simultanously) , but when you are close to it, while formally you have the degree of freedom, the margin is low and you may get into an unfeasible condition for L-V equilibrium with just a few tenths of a degree.
I your case you are basically trying to set pressure and temperature at the top of the column in a region near the azeotrope.
So, try with actual variables like distillate rate and reflux ratio. Then use internal specs or external vary to get to the desired ouput (ppm water in the bottom, máximum EDC at the top).