Engineering design analysis utilizing a cloud platform. (January 2018)

• Record Type: Journal Article
• Authors: Suram, Sunil; MacCarty, Nordica A.; Bryden, Kenneth M.
• Is Part Of: Advances in engineering software. Volume 115(2018)
• Page Start: 374
• Page End: 385
• Publication Date: 2018-01
• Subjects: Computer-aided engineering -- Periodicals; Engineering -- Computer programs -- Periodicals; Engineering -- Software -- Periodicals; Periodicals; 620.0028553
• DOI: 10.1016/j.advengsoft.2017.10.004
• Languages: English
• ISSNs: 0965-9978

Abstract

Highlights: A novel schema for integrating models together is introduced. The component models are established as model-as-a-service components. The component models are stateless and web-enabled. System builders can access the models as API endpoints. This is demonstrated using a model for a small cookstove. Abstract: In this paper we present a novel methodology for modeling engineered and other systems based on integrating a set of component models that are accessible as "model-as-a-service" components within a cloud platform. These component models can be combined together to form a systems model. The component models are stateless and web-enabled. The advantage of being web-enabled is that developers can use the models as API endpoints as opposed to library components, hence making the models themselves language agnostic and less restrictive in their use. These ideas are presented within the context of a previously published engineering model for the thermal analysis and preliminary design of a small biomass cookstove. In this paper the monolithic biomass cookstove model is separated into six independent, stateless component models supported by a generic model application infrastructure. Interaction between the models is orchestrated by a federated model system. Finally, the results of the cookstove from the monolithic model were compared with the distributed systems model. It was found that there was no change in the results. However, the systems model increased the time-to-solution due to network latency. However, the ability to share models and data via API endpoints, will likely offset the overall wall-clock time for model integration, since model developers do not have to make code changes. In conclusion, it is advantageous to build web-enabled component models for their easy reuse across multiple systems models.