We present the analysis of the steady state backoff problem with state and
dynamic constraints of a non-linear chemical process described by almost 30
00 differential algebraic equations. The dynamic optimization is carried ou
t using a new approach based on an SQP algorithm for semi-infinite non-line
ar programming problems. The system equations are integrated with an implic
it Runge-Kutta method and 'reduced' gradients are evaluated by adjoint equa
tions. The high performance of the algorithm is analysed and compared to fu
lly non-linear programming proposals in which discretized system equations
are treated as general non-linear equality constraints.