In this contribution we present an online scheduling algorithm for a real world multiproduct batch plant, which includes an explicit representation of uncertainties. An ideal online scheduler is approximated by a hierarchical approach with two-levels, on which optimisation problems are formulated as mathematical programs and solved by non-standard algorithms. The focus is on the hierarchical framework and the upper level planning problem where uncertainties are modelled explicitly. The key features of the planning model and of the solution algorithm are explained and numerical results are presented.