A large-scale dynamic model of a board machine has been developed and implemented in Simons IDEAS. The model extends from the pulp storage towers to the finished sheet including the broke system and the long circulation of white water and predicts the fractions of four kinds of fibers, water, fines, filler, starch and COD. Additionally it can predict key paper quality properties such as density, E-modulus and bending stiffness. Three simulation examples are presented showing how the model is used to optimise production by focusing on grade changes, and to quantitatively explore the potential of improved, but costly, transducers. In closing, some reflections on why this technically successful simulation project failed to attract the interest and enthusiasm from industry that we thought it deserved.