Extensive application of advanced composite materials in rib-stiffened composite structures demands new fast and reliable simulation procedures and optimum design guidelines for efficient industrial implementation. Such procedures mostly involve a large amount of analyzed numerical samples that are unacceptably demanding in terms of solution time. Considering this, a more efficient strategy is applied, by arranging and adding new experimental points to an already existing design of experiments according to a space-filling criterion, thus achieving a good balance between the space filling quality in whole design space and quantitative improvement by added sample points. An advantage of the applied approach is the fine sampling quality even before all experiment runs are performed, which once elaborated could be made publicly available (www.rtu.lv/mmd/). Metamodeling approach was implemented for the analysis of composite stiffened shells subjected to buckling and loaded into the post buckling regime until collapse. Moreover, material degradation in terms of stiffness reduction in the skin stringer zone is carried out to assure design reliability. The acquired sequential metamodels demonstrated effectiveness in solution and hierarchical optimization of the direct and the inverse problem - establishing dependence of the load-shortening curve on material degradation, and determination of material degradation degree from the essential response parameters of the load shortening curve. The resulting design procedure provides an effective optimal design tool for the safe preliminary study of composite stiffened shells under axial compression.