Effect of decoupling techniques in multivariable systems
Abstract
This paper presents a performance analysis of decoupling methods
considering its effect in the global system and the control system performance.
The system performance was studied using static gain (DGE),
approximation dead time (DTM) and feedforward (DAA) decoupling
methods. As study case a multivariable (2x2) first order plus dead
time model of a Wood and Berry distillation column was used. The
performance of the decoupling methods were analyzed for modelling
errors in the parameters: static gain (K), time constant (?) and dead time
(t0). Decoupling techniques were evaluated using a PID control scheme
strategy, testing the control system robustness against modelling errors
and set point changes. It was proved that for static gain (K) and time
constant (?) parameter changes, the feedforward decoupling (DAA)
presents better performance using the IAE criteria. The effects for dead
time changes are not significative, showing IAE variations of 0,05% for
distillation column set point changes.