The method is evolving. I have realized that by recording the cooling water temperatures, the method can be greatly simplified and also extended to more systems. In the figure below, the original energy envelope in blue shows the need for the estimation model to compute the boilup and the condensate. This requires both a good thermodynamic VLE model and a good model of the condenser. In the original example, only the solvent was volatile so the VLE was fairly well known. However, in cases where some of the reactants and products are volatile, the non-ideality of the solution and gas could create difficulty. Also the condenser model requires knowledge of the condensing and subcooling heat transfer coefficients and probably hydraulic modeling.
Changing the heat balance in the estimation model to the red envelope eliminates the need for VLE knowledge and a model of the the condenser. The additional temperatures around the condenser are used as input to the estimation model. The hydrogen flow rate remains as the data used for the minimization of the error.
The sensors shown in red are the minimum recommended for digital recording. It is assumed that the cooling water flow rate and the hydrogen feed temperature are held constant. If not, then they should also be recorded.