Dynamic optimization allows for the determination of the upper bound of achievable performance in the operation of continuous chemical processes in transient operation which often results from load or grade changes. Performance assessment can rely on the computation of optimal trajectories for selected scenarios which properly reflect the operational envelope. Realistic industrial problems, however, involve very large-scale dynamic process models and consequently require highly-efficient and robust optimization algorithms. In this work we demonstrate the feasibility of operability assessment by means of dynamic optimization in an industrial case study involving a large-scale process model comprising about 12,000 differential-algebraic model equations. The numerical strategy employed relies on a single shooting method combined with adaptive control grid refinement to minimize the complexity of the numerical problem to the extent possible. This algorithm proves to be the key to success and saves about 95% of computational complexity in comparison to a conventional equidistant discretization. (C) 2010 Elsevier Ltd. All rights reserved.