Two approaches are traditionally used in the design of a battery.
With the first, the behavior of a battery is simulated through the use of CAE (Computer Aided Engineering) technologies, which allow the product to be represented using numerical models focused on physics modeling and analysis. But, for an accurate prediction, it is necessary to introduce multi-physics analysis, i.e. to introduce further parameters, the electrophysical ones characteristic of batteries for example, which make the actual physical model more complex. To this we must add the increase in product complexity, determined by the introduction of new technologies and new materials, and by attention to environmental sustainability. And it is this complexity that has brought out the limits in adopting this approach: from the increasing calculation times, to the extensive skills that are required (on the software to be used, on the types of analyzes to be carried out, on the characteristics and behaviors of the new materials, to name the main ones).

The second approach, more recent, is instead linked to automatic learning models (Machine Learning). In this case, datasets are used to design a new battery, consisting of a large number of data collected experimentally.
But how reliable is this data when designing a new product? The design and performance of a battery depend on many factors: the limit in using this approach is the availability of data that is actually useful in the early design phases.

Oorja fits into this scenario: the strength of this platform lies in the adoption of a hybrid approach, which exploits the advantages of the two methodologies described above, and overcomes their limitations.

Oorja uses an approach based on simple and fast physical models, which will form the basis for the machine learning algorithm, thus reducing the number of data needed for the initial dataset.

Oorja simulates, predicts and optimizes the behavior of batteries, analyzing different performances, such as, for example, the quantity of current produced, the "capacity/power fade", overheating during use, fast charging protocols and related aspects to the guarantee.

At the basis of the methodology there is a workflow made up of 7 modules:

• Material: to define the material
• Data: for using Machine Learning
• Design: to create or import the "pack geometry" to optimize
• Range: to predict performance related to battery current
• Volt: to optimize performance (e.g. BoL: Begin of Life performance)
• Fade: for risk analysis, analyze aspects related to the guarantee, analyze the "capacity/power fade"
• Heat: thermal analysis, design optimization to define the distances between cells or cooling

To underline the extremely "user friendly" graphic interface of Oorja, which makes the use of the complex methodology extremely simple: it is based on the use of a "wizard", i.e. an automatic system, which guides the user step by step in workflow generation.