Workspace: A workflow platform for supporting development and deployment of modelling and simulation

Abstract

In modelling and simulation a user performs complex sequences of tasks that are intrinsically linked in the form of networks. Tasks include simpler steps such as copying and transforming files to using mesh generation or visualisation software to create inputs or to analyse outputs. Typically, these sequences involve manual steps which are time expensive, often involve use of multiple, incompatible software tools of various levels of usability, which are usually not well documented, and so create significant challenges for provenance, reproducibility and transfer to other users. These semi-manual workflows also inhibit sharing of capabilities and commercialisation. This paper explores how the adoption of the scientific workflow system Workspace can significantly improve such research and engineering activities by streamlining and automating the workflows involved. Benefits ensue in terms of improved collaboration through the sharing of workflow components, higher productivity, reduced costs and shorter development cycles through the reuse of software components and workflows and a reduction in risk due to improved reproducibility, documentation and provenance. The use of the Workspace platform allows such workflows to be executed via an interactive GUI, in batch mode via the command prompt, or embedded within self-contained applications. These applications can take the form of closed source binaries, web services or ‘software as a service’ applications. This supports significantly improved translation of research prototype software to end users and creates low cost pathways for commercialisation.