Currently a shift of focus towards 6-sigma quality and market responsive operation has been initiated in the chemical processing industries. The fast evolution of products and processes enforced by fierce global competition and by tightening legislation are major forces for new application development approaches and for new technologies. The need of high performance non-linear model based control, optimization, monitoring and soft sensing applications and the cost driven necessity of reuse of models and results of earlier engineering effort will be explained to be the drivers for the current and future industrial challenges in (hybrid) modelling and system identification. These market developments require more extensive application of (nonlinear) rigorous models extended with empirical model components to achieve the model accuracy requirements, the coverage of wide process operating ranges and minimization of engineering costs, which cannot be attained by application of pure black box modelling approaches. Besides the techniques applied for hybrid modelling and parameter estimation, the paper also discusses the techniques needed for model reduction, model tracking and state estimation to make the high performance model based applications work properly. An overview with some results is given of the techniques applied and tested in our current R&D industrial pilot projects.