Towards a Standard Approach for Optimization in Science and Engineering

Optimization plays a fundamental role in engineering design and in many other fields in applied science. An optimization process allows obtaining the best designs which maximize and/or minimize a number of objectives, satisfying at the same time certain constraints. Nowadays, design activities require a large use of computational models to simulate experiments, which are usually automated through the execution of the so-called scientific workflows. Even if there is a general agreement in both academy and industry on the use of scientific workflows for the representation of optimization processes, no single standard has arisen as a valid model to fully characterize it. A standard will facilitate collaboration between scientists and industrial designers, interaction between different fields and a common vocabulary in scientific and engineering publications. This paper proposes the use of BPMN 2.0, a well-defined standard from the area of business processes, as a formal representation for both the abstract and execution models for scientific workflows in the context of process optimization. Aspects like semantic expressiveness, representation efficiency and extensibility, as required by optimization in industrial applications, have been carefully considered in this research. Practical results of the implementation of an industrial-quality optimization workflow engine defined in terms of the BPMN 2.0 standard are also presented in the paper.