H2S Stripper Control

Regarding H2S stripper control in distillate strippers, we have seen similar concerns in our distillate stripper with regard to optimizing Flash Point. We have a fired reboiler heater configuration, and we do not have continuous analyzer measurement for the stripper bottoms product. Upon embarking on an APC application for Flash Pt. optimization, we built an inferential prediction for Flash Pt., but we are finding that this property prediction which is based on stripper bottoms tempertaure may not be sufficiently robust for realtime control, and we run the risk of having occasional "off-test" product if we attempt to control based on inferential prediction. We have Kerosene/Naphtha material in our feed which is rejected as a liquid overhead product. This net overhead liquid product seems to be a good indicator of when bottoms Flash Point might become too low.

So we are finding that the best control we might be able to do is to stabilize the tower with APC, but not necessarily optimize Flash Pt. closely. Since we have Kero/Naphtha is our stripper feed for sufficient control, should we be able to control Flash more tightly?, and if so what are toher key variable we should be looking for in our APC? One variable with have noted might be useful is reflux/(reflux + net overhead liquid), or R/D+R, sometimes refered to as Separation Index.

Thank You.