Abstract
On the whole, scientific software community has focused in developing faster algorithms, setting aside flexibility, reuse, maintainability and scalability. The common approach used to develop software is the functional decomposition. However, the development of new applications is currently more complex than the former software development. In this work, we present a pattern object oriented architecture (POO) to approximate solutions of initial value problems (IVPs) with different time scales. We propose the use of design patterns' techniques in the developing process of the software architecture. These techniques allow to easily add new numerical methods and use them to create different multirate configurations. Besides, we give evidence that design patterns' techniques are suitable in the analysis, design and development of scientific software.