This is just a funny quirk that came up recently in some simulations that caused some consternation before the problem was nailed down. The root is one of those little details that you ignore until it turns into a big detail.
So, I have this big flowsheet in VMGSim, a fractionator with an associated co2 scrubber, everything is converged, everything is kosher, time to do the overall balance for completeness.
Well the mass balances within a very small margin of error, but the energy, well, wtf?
Here is where I reveal a little niggling detail: the entire flowsheet is not using the same property package. About half of the sheet is using APR, but there is a sub flowsheet that is running the amines package (the co2 scrubber). If we look at the stream that connects the two flowsheets we see...
The left window is the stream leaving the flowsheet using the APR property package, the right window is that same stream entering the flowsheet using the amines property package. Here we see a hint of the problem, the stream enthalpies calculated by the two property packages are totally different.
I'm guessing this is because the reference point for the enthalpy is different between the two property packages. Each side of the divide balances, but if I draw the boundary over the whole system (which is what the OBL unit does) then the energy balance fails (or appears to at least).