The control structure selected has a strong effect on the performance
of a closed loop system with respect to its disturbance rejection capa
bility. A special capability-localizable disturbance rejection is stud
ied in this work. It is shown that this capability can be related to a
particular set of manipulated variables. A new measure, the input dis
turbance alignment (IDA) is proposed to identify these inputs. Some im
portant features of the new indicator and its relationship with other
well-known controllability measures are presented. The case studies in
cluded show that the IDA can be used for efficient selection of manipu
lated variables from a large number of candidate inputs. These then fo
rm the best control structure for localized disturbance rejection. (C)
1998 Elsevier Science Ltd. All rights reserved.