Indirect composition control used in many industrial distillation columns is freouently practical because of difficulties associated with on-line composition measurements. Brosilow has proposed the inferential control using selected secondary measurements in a linear combination to estimate and control the effect of unmeasurable input disturbances on the product composition.
In this study utilization of the inferential control scheme for control of top product composition of a binary distillation column has been investigated. Experiments were run on a packed distillation column, 10.16 cm in diameter and 253 cm in height, separating a methanol/water mixture at atmospheric pressure. All instrument was pneumatically controlled. The column was interfaced with HP-2647A intelligence terminal, HP-2240A measurement and control processor which allowed extensive data acquisition and provided the means of implementing the inferential control algorithm. Result from simulation studies and experimental evaluation of the inferential control on a pilot scale column are compared with the performance achieved using conventional single temperature feedback control with proportional plus integral action.
Both digital simulation and experimental verification showed that the performance of inferential control was superior to that achieved with conventional single temperature feedback control using proportional plus integral action.