Chronicles are temporal patterns well suited to capture dynamic process thanks to an event abstraction of the information of interest. Designing chronicles from a journal log is not a trivial task considering the huge amount of data generated by highly-advanced systems. Chronicle discovery is a mean to help expert design chronicles that are representative of a system behavior from direct observations. In this paper, a clustering approach to the chronicle discovery problem is considered. To improve the discovered chronicle quality, an order in the design of interesting pattern is introduced. This allows a better robustness to small perturbations in the input journal log. The efficiency of the ordered chronicle discovery algorithm is evaluated on a real dataset.