The licensors process design called for 154ºF overhead drum temperature with the reflux as a saturated liquid. However, the drum actually operated at between 100ºF and 113ºF with the reflux as a sub-cooled liquid. The plant was running at the design reflux rate to the column. The plants engineers wondered if there was a problem with this operation and if they should make control changes to get a saturated liquid reflux. To the plant staff, it looked like the condenser was over-designed.With the control configuration shown, it is extremely difficult to operate the reflux drum to make a bubble-point liquid product. The configuration is a variation of a hot-vapor bypass. The reflux drum does not operate at equilibrium conditions. The vapor sent through the bypass to the drum has to be condensed in some way to prevent a net vapor product. Condensation occurs where the vapor contacts the liquid in the drum. To get the vapor to condense, the liquid must be sub-cooled.The vapor rate to the drum to maintain pressure is balanced against the condensation in the drum for steady-state operation. The heat transfer depends upon the surface area of the liquid, the amount of sub-cooling and the vapor and liquid turbulence in the drum. While simulation models and theory say that the drum can be perfectly balanced, in practice the liquid to the drum must have a sub-cooling margin to allow for stable control. All the variations of hot-vapor bypass for pressure control must have sub-cooled liquid product to work. While the condenser appears over-designed compared to the required process conditions for ideal operation, the extra cooling capability is vital for the control system to work.Typically, sub-cooled reflux to a tower has several results. First, the cold liquid back to the tower condenses rising vapor. So, the final liquid rate inside the tower exceeds the ideal liquid rate based on the bubble-point reflux. Second, the heat transfer required to raise the reflux to the bubble point reduces the fractionation effectiveness near the top of the tower. The actual tray (or packing) efficiency will be lower than expected. Third, because the duty removed at constant reflux rate increases with the colder return temperature, the reboiler duty requirement will be higher and thus, the vapor rates inside the tower will be, too.Most new units include enough design margin so that operation at higher internal reflux rates is not a significant problem unless reboiler heat is limited or expensive. The simplest method to deal with any problems here is to reduce the external reflux rate to adjust column performance.This case provides an extremely important lesson, even though the consequences turned out to be innocuous. Just because you can simulate an operation does not mean that the specific operating point can be maintained with real as distinct from virtual equipment. Never forget the requirements of control and start-up and the consequences of over-design or under-design of upstream and downstream equipment.
