The industrial benefits of batch reactor optimisation and optimising control come both during plant operation as well as during the initial product design and scale-up stages. At the pilot and production stages, batch and semi-batch reactor optimisation and optimising control can contribute significantly to the achievement of consistency of production and the minimisation of batch production times. On-line estimation of reactor and reactant properties, impurities, fouling, process disturbances, etc., as well as the process states, leads to the updating of the process model and optimal control strategies thereby providing real-time adaptation and on-line real-time optimisation of time varying process behaviour. The development of dynamic phenomenological (mechanistic) process models is time consuming and difficult especially in highly responsive multi-product and agile manufacturing. Robust empirical and hybrid models (the conjunction of reduced complexity mechanistic behaviour with empirical representations) provides an alternative approach to the fast development of process representations for use in on-line optimising control. The issues of batch reactor optimisation, the building of robust empirical and hybrid models and fault detection and isolation were investigated.