Nowadays thermal comfort and energy saving issues are getting more and more an important role and are actually involved at the very early design stage. In residential building projects it’s possible to predict the position of cooling sources in order to have a uniform temperature distribution saving energy. CFD can be used as a virtual tool to quickly predict the temperature distribution in presence of heating/cooling facilities. Compared to other fields of application this kind of modelling is very lightweight and fast using HPC facilities. For these reasons in CINECA we designed a cloud-based service built on top of OpenStack technology coupled with state-of-the-art open-source CFD solutions able to manage in an automatic fashion the whole workflow.
The advantage of using a cloud service relies to the strong flexibility that can be issued on request, reducing severely the cost of maintaining such a service. For sake of completeness we compared the performances of the cloud implementation (computational resources virtualized) with the standard one built on the HPC cluster showing that the two solutions are comparable for this kind of application.