ORTIC: Online Real Time Parametric Internet Computing Via Webserver
Researchers have developed a computational web-server to fulfill the growing demand for numerical simulation analysis, that, in the last decades, are replacing the experimental measures.
An efficient mathematical tool is needed to overcome the growing demand for numerical simulations, especially in an industrial context. Reduced order modeling (ROM) is a computational reduction technique allowing an efficient exploitation of the numerical simulation, resulting in a real-time response framework for parametric problems. Thus, researchers have developed a computational web-server to fulfill the growing demand for ROM offline-online tools exploitable also by users lacking the strong technical background or the amount of time required to implement themselves the desired software. The key idea is to create a computational interface giving the opportunity to handle and analyze a parametric problem in a very intuitive way. Many tools will be available on the web platform, some with a general purpose and some for very specific problems: users can access directly through the web browser, using the available device, choose the application they need and finally use it to tackle the problem of interest.
Thanks to this technology, users have not to install any package by themselves, on their device. The final user has the possibility to run the applications with no efforts related to the installation/implementation phase. Moreover, by using the standard web protocol, any device having an internet connection can access to the web-server and run a web application. The real time response, due to the model reduction algorithms, combined with the easiness of use, matches the industrial demand of intuitive and immediate tools. An additional reason is that the offline-online computational paradigm fits well the client-server architecture. Moreover, having a centralized system to manage, it is possible to update all the systems and packages only once, guaranteeing the highest standard in terms of performance and security. Thanks to its capabilities, the web-server can enhance the design pipeline by handling in an efficient way the high-fidelity simulations and providing a user-friendly interface usable also to non-expert technical staff. Another vantage point is the multi-purpose applicability of the web-server. In conclusion, this computational infrastructure will lead to a reduced energy consumption for scientific computing.
Stage of development
A prototype version of the computational web-server, containing few demonstrative applications, has been already implemented, and it is available at web address argos.sissa.it. The applications already developed show a complex structural bifurcation test-case efficiently, analyzed using ROM, and a more general-purpose application to perform cluster analysis to an input dataset. The variety of these applications allows to understand the modularity and the multi-purpose applicability of the web-server: the offline-online paradigm, typical of the reduced order modeling, can be exploited also to develop application related to machine learning and artificial intelligence. In this context, the web-server can be exploited also to build applications for data science, or responsive digital twin.
Challenge and needs
In many industrial contexts, the entire analysis and design process can be very demanding due to the involvement of two (or more) offices/teams during different steps of the process, and due to the high computational cost for a single simulation. In parametric problems, a database of simulations can be treated using reduced order methods to exponentially reduce the computational cost of a single simulation. Especially in a many-queries context, these methods are able to make sustainable the study of complex phenomena, achieving a reduction in the cost per single simulation capable to reduce also the time-to-market. Moreover, the real-time response, the centrality of the web-server and the easiness of use help to simplify the interactions between heterogeneous offices, that in an industrial context are usually affected by diversified languages and tools, and by latency in computing the results, guaranteeing a further reduction in the global time-to-market.
The technology is not protected by a patent yet.
Potential markets and targets
The possible potential applications are many and varied: parametric problems are spread among several engineering fields, e.g. naval, nautical, aeronautical, mechanical, civil, wind, environmental, sport applications. In all these contexts, the computational interface can be easily applied to analyze, design and optimize the final products. Moreover, the wide applicability of the reduced order modeling method, as well as by a graphical interface, gives the possibility to disseminate the web-server potential also in many industrial workflows. In conclusion, researchers envision that ROM techniques could be adopted also in many other fields directly related to applied sciences, for example in medicine and environment.
Potential partners could be any SMEs interested in:
- real-time solution or multi-queries context (e.g. optimization, uncertainty quantification, parameter space analysis) for complex parametric problems;
- a software framework for real-time simulation via web;
- data analysis and modeling, visualization and assimilation.