ORALS

SESSION: BatteryTuePM2-R6	Kumar international Symposium (8th Intl. Symp. on Sustainable Secondary Battery Manufacturing & Recycling)
Tue. 28 Nov. 2023 / Room: Orchid
Session Chairs: Claudio Capiglia; Satheesh Krishnamurthy; Session Monitor: TBA

16:00: [BatteryTuePM209] OS Keynote
APPLICATION OF AI/ML IN THE BATTERY INDUSTRY
Himanshu Paliwal¹ ; Alhad Parashtekar¹ ; Nithin Reddy¹ ; Pawan Kumar¹ ; Astitva Mishra¹ ; Claudio Capiglia² ; ¹Reliance, Mumbai, India; ²Reliance Industries Limited, Mumbai, India;
Paper Id: 397 [Abstract]

Current process simulators such as ASPEN, PRO-II, gPROMS etc. do not have thermo-physical and transport property database of the materials used or produced in the battery industry. The process simulators lack physics-based models for modelling unit operations such as mixing, wet milling, coating, drying, calendaring. It is thus hard for process simulators to capture the effect of process variables in materials manufacturing or in cell manufacturing units on cell performance. Therefore, the process development and control in the battery industry is mostly done using a trial-and-error approach which makes it time and cost intensive. The recent advances in machine learning, artificial intelligence and sensing technologies can be leveraged to reduce the cost, accelerate the development and retuning of plants in the battery industry [1, 2]. The data from the pilot lines or from the manufacturing lines could be used to develop data-based models. There are only a few academic groups who have worked on data-based models for cell manufacturing units [2]. However, there is no work on developing data-based models for battery materials manufacturing unit. The biggest bottleneck in developing data-based model for cell manufacturing and battery materials manufacturing units is the availability of feature rich and diverse dataset.The scarcity of data could be mitigated by coupling multiscale modelling with data driven approach. Multiscale modelling gets computationally intensive when models are built to scale. Plant data still is orders of magnitude expensive and hence exists in much smaller volumes. One possible solution is to build multiscale models with the help of plant data and run these simulations for different scenarios creating a diverse enough dataset [3,4]. The simulated data along with plant data could then be expanded using generative AI to a much larger data set enabling higher accuracies of the data-based model. However, this concept still needs to be tested and validated.

References:
[1] Sinha, A.; Radhakrishnan, V.; Vadari, R.; Capiglia, C. Scaling up Li-ion cell production: Building a Gigafatory. Sustainable Industrial Processing Summit and Exhibition 2022, 14.
[2] Bockwinkel, K.; Nowak, C.; Thiede, B.; Nöske, M.; Dietrich, F.; Thiede, S.; Haselrieder, W.; Dröder, K.; Kwade, A.; Herrmann, C. Enhanced Processing and Testing Concepts for New Active Materials for lithium‐Ion Batteries. Energy Technology 2019, 8 (2).
[3] Duquesnoy, M.; Liu, C.; Dominguez, D. Z.; Kumar, V.; Ayerbe, E.; Franco, A. A. Machine Learning-Assisted Multi-Objective Optimization of Battery Manufacturing from Synthetic Data Generated by Physics-Based Simulations. Energy Storage Materials 2023, 56, 50–61.
[4] Turetskyy, A.; Thiede, S.; Thomitzek, M.; von Drachenfels, N.; Pape, T.; Herrmann, C. Toward Data‐driven Applications in lithium‐Ion Battery Cell Manufacturing. Energy Technology 2019, 8 (2).