oorja is a deep tech product engineering company that uses a hybrid approach (Physics + ML) to perform fast and accurate predictive analysis to solve complex engineering problems. oorja empowers customers to quickly solve degradation, thermal runaway and short range thereby enabling businesses to reduce the R&D cost and accelerate product to market.Book a Demo
60% of the cost of electric automobiles and energy storage companies is for the battery pack, and auto companies struggle to ensure that performance requirements are met and that the life of the battery pack is maximized.
With diverse operating conditions and high temperatures, the life and safety of a battery pack is significantly affected. There has been no easy way (until now) to predict cell temperature in a battery pack for various drive cycles, leading to “over engineering”, i.e. adding more cells to the pack, limiting acceleration and limiting fast charging speed, all of which increase costs and impact user experience negatively.
Traditional solution approaches are purely physics based, which are inaccurate since they are not based on real life data and are also rather complicated to use, requiring experts.very difficult to use since it requires experts to operate.
While we are fans of the Physics based model, we do not believe that it is the sole answer.
Getting Started With oorja
No installation, maintenance or computing power necessary. Log in, use it and move on
Electrochemistry , thermal and ML experts are just a phone call away
State of the art hybrid simulation technology for better pack design
So we put a PINN on it!!
At oorja we have created a simple, unique, hybrid approach for simulation (Patent under progress), using Physics Informed Neural Networks (PINN), which uses a combination of physics and machine learning.
First, an approximate physics-based model is used to get the pre-training data for the ML algorithm. Accuracy is then obtained by using minimal experimental data and using it in the pre-trained ML model. The requirement of only approximate physics-based models reduces complexity, while the minimal data required for subsequent ML reduces the need for data while not compromising on accuracy.
Extensive validation for both our thermal management application as well as the capacity fade application have been performed and our algorithms have been validated with experimental data with 90% accuracy.
oorja has made all of this available to auto companies, and other battery pack users and manufacturers, in an easy-to-use SaaS based tool which can help them predict pack life and thermal performance thereby optimizing pack performance greatly.
Property Data Base (DB) App is a central repository of all the properties of cells.
Design (DN) App creates module or pack 3D designs considering cell form factors and pack configuration and assigns appropriate material properties to them.