A new approach for assessing nonlinear interaction effects and closed-
loop nonlinearity in multivariable processes is presented. It is based
on a differential geometric interpretation of the relative gain away
that leads naturally to systematic procedures for describing interacti
on effects of higher order and for assessing closed-loop nonlinearity
effects in nonlinear processes. Two types of nonlinear effects associa
ted with the behavior of a process are introduced. Between-channel non
linearity is associated with the nonlinear dependence of an output cha
nnel on other input-output pairings. Within channel nonlinearity is us
ed to identify the nonlinear effects that result from the inherent non
linearity of an individual output channel. A root-mean-squared measure
of nonlinearity is introduced and is used to evaluate the significanc
e of local nonlinear effects. Nonlinear interaction measures are deriv
ed that provide tools for assessing input-output pairings in a nonline
ar process. This new approach extends the standard techniques and prov
ides an estimate of the effect of nonlinearity on closed-loop interact
ions.