|
|
|

romBOX

Machine learning, reduced order models and multi-fidelity representation are at the heart of romBOX and will help you generate a homogenized view of your data and optimally combine your computational models.

What-If in real-time

With romBOX you can simulate the most complex What-If scenarios in real-time by exploiting state-of-the-art machine-learning algorithms. Responses can be obtained in fractions of a second instead of hours or days, leading to a democratization of engineering knowledge over all levels of business, from designers to managers, and helping in rationalizing and accelerating company-wide decision making.

Multi-fidelity

Different numerical tools and mathematical models are used at different stages of the design process. romBOX will bridge the gap among these tools to have a unique view over models and data. This will allow you to incorporate high-fidelity knowledge already in the preliminary phases of the design process.

Small data

Small data is often the rule rather than the exception in many engineering tasks. Our technology values even very limited, but relevant, datasets by combining machine learning with physical modelling.

Enablers for complex product design

Complex products are characterized by multi-disciplinary and multi-objective trader-offs. Minimizing these trade-offs requires concurrent engineering and interfacing different simulation pipelines. Encapsulating the different disciplines through romBOX will ease the communication among different teams and permit to update of digital models on-the-fly as soon as additional data is available.

Enablers for multi-scale problems

Expensive multi-level simulations are necessary when the physics is governed by phenomena occurring on different lengths and time scales romBOX can be used to derive a reduced-order-model of the smaller-scale simulations that can be coupled with the larger-scale model through its API, drastically reducing the run time without compromising the overall predictive capability.