Tell us more about the columns and their respective models - what the specs are, number of stages, thermo methods used, feed condition, etc. Tell us also what measured information is available. The first thing that comes to mind is that the number of stages in the model is less than what the real column is providing, so that more traffic in the column is needed to achieve the desired separation. The next thing to consiser is the feed condition; if your model gives information that makes the feed liquid when in the real plant it is vapor, more energy will be needed at the reboiler. Discrepancies between measured and calculated numbers are usual in the course of fitting operating data to a model. One adjusts the model to obtain good agreement with operating data. It always is a good idea to have as much data on the operation as is possible, to permit cross-checking of measurements against each other. Flow measurements often are in disagreement with each other.

For the sake of brevity I am replying with data specific to the deethanizer (this is also the column with the bigger difference with respect to field data).

Column:  40 2-pass valve trays, total condenser (I have tried simulating with and without a vent), kettle reboiler

Measured values available:  feed flow & temp, product flows & temp, reflux flow & temp, product compositions (by GC), hot oil to reboiler temp in & out and flow, hot oil to heater (temp in and out and flow per pass).  Product flows are from accounting meters and heater and reboiler flows agree within 20 gpm (out of a total of about 600).  (total heat input approx 10 MMBTU/h).  My temperatures actually come out pretty close and of course flows match since I am specing out mole fractions in the feed and product.

Model:  40 stages (I have tried everthing from 15 to 80), Peng Robinson EOS, feed vap frac is 0.20 (based on measured temperatures, flows, & compositions and HYSYS' flash)

Comments:  Increasing number of stages does nothing (anything above 30 and the duties flatten right out).  HYSYS' duty is about 18 MMBTU/h (vs 10).  If I use a fully vaporized feed, I get a believable number for the duty (close to 10).  However the temperatures at operating pressure, and pressures at operating temperatures are not even close to believable.

Your comment about feed condition maybe an area for further investigation.