Session: 03-09-01: Data-Driven Innovation in Smart Product Design and Manufacturing

Paper Number: 159148

An Integration of Support Vector Machine for Predicting Mechanical Properties in IIoT Enabled Investment Casting 

Investment casting is very complex process, and employed for manufacturing of diversified products in various industrial sectors including aerospace, automobile, chemical, jewellery, etc. due to its capabilities to produce metallic products with higher accuracy and surface finish. These industrial sectors are always in demand of quality castings (e.g., desired mechanical properties as well as defect free castings). Mechanical properties are measured using destructive tests, and results are compared with desired range of mechanical properties. The mechanical properties (i.e., ultimate tensile strength, yield strength and percentage elongation) are further driven by various parameters associated with process, design and material related to sub-processes of investment casting process such as wax pattern making, shell making, dewaxing, melting as well as pouring.

Furthermore, these sub-processes and relevant parameters are generated by equipment associated with each sub-process. These parameters are inclusive but not limited to wax pattern making room temperature and humidity, viscosity and acidic nature of ceramic slurry, shell making room temperature and humidity, duration of shell making, shell weight, preparation time (melt) of alloys, chemical composition of alloys, etc. Though, these parameters are usually recorded using manual intervention, and has relatively less reliability especially when used for futuristic analytics. Therefore, further analytics (prediction of mechanical properties and/or identification of critical parameters affecting occurrence of defects) for improvement of quality especially mechanical properties of investment castings is relatively difficult to be employed.

One of recent technologies such as Industrial Internet of Things (IIoT) can be integrated with the investment casting foundry setup for acquiring parameters associated with sub-processes. Relevant sensors as well as microcontrollers can be embedded with equipment for acquiring parameters, and parameters can be streamed to cloud-based server using internet the further analytics.

Present work is mainly focused on exploring capabilities of Support Vector Machine (SVM) for predicting mechanical properties using data acquired from IIoT enabled investment casting foundries. This work also highlights an implementation of IIoT in investment casting foundries for acquiring the data that can be used in prognosis. Several other techniques including Artificial Neural Network (ANN), and Multivariate Regression Analytis (MRA) have also been explored for providing prediction of mechanical properties however SVM has been integrated with IIoT enabled investment casting process for providing prediction of mechanical properties. Specific cloud-based system highlighting implementation of SVM for prediction of mechanical properties is also developed, and demonstrated on data acquired from industrial investment casting foundry.

It was concluded that SVM can be easily integrated with IIoT enabled investment casting, and can be used for prognosis of mechanical properties however capabilities of SVM for identification of critical parameters and their specific range of values affecting mechanical properties of investment castings is yet to be explored, and will be considered for futuristic research.

Presenting Author: Dr. Amit Sata Marwadi University

Presenting Author Biography: Dr. Amit Sata is working as Professor in Mechanical Engineering at Marwadi University. He is also heading Innovation and Entrepreneurship Cell at Marwadi University. He is graduated in Mechanical Engineering, and is an alumina of Indian Institute of Technology Bombay where he earned postgraduation (M Tech) as well as Doctorate of Philosophy (PhD). He held short visiting appointment in Romania at University of Pitesti (2018) under ERASMUS+ program.

Dr Sata mainly works in the direction of extending an application of Industrial Internet of Things (IIoT) to the domain of manufacturing engineering specifically to metal casting. He is currently guiding six PhD students, and has guided more than 35 students at graduate as well as postgraduate level in the domain of metal casting. He has published more than 45 technical papers and delivered more than 50 talks in India. Dr Sata has been granted with 14 IPRs out of filed 26 patents, and has commercialized 07 IPRs.

Dr Sata is heading the two labs: SMART Foundry and MUJCAL – Association for Foundry Technology at Marwadi University. Idea of extending an application of Industrial Internet of Things (IIoT) was appreciated by government of India in 2015, and funded more than 35 lacs for a project SMART Foundry 2020 to implement the concept of IIoT to metal casting. This implementation is already in demonstration at Marwadi University. This research project has also provided an opportunity to initiate innovative start up Udhyog 4.0 LLP that mainly focuses on transforming existing manufacturing enterprise into SMART manufacturing setup. Udhyog 4.0 LLP has already initiated to transform existing investment casting foundry (Turbo Cast India Private Limited – Rajkot) into SMART foundry. This start up is incubated by Marwadi University, and has been funded by government of Gujarat with 20 lacs under Startup Gujarat Scheme. While MUJCAL – Association for Foundry Technology is mainly focused on imparting technical skills to students of mechanical engineering. MUJCAL- Association for foundry technology is funded by 100 lacs by Centre for Entrepreneurship Development – Government of Gujarat, and is a collaborative effort Jyoti CNC Automation Limited – Rajkot & Marwadi University. MUJCAL has dedicated space over 3200 square feet at campus of Marwadi University, and equipped with more than 40 equipment.

Dr Sata is committed to I^6 (Ideation, Information, Inspiration, Immersion, Innovation, Implementation) philosophy, and implementation of advanced technologies in the domain of manufacturing.

Authors:

Dr. Amit Sata Marwadi University
Nikunj Maheta Marwadi University
Himanshu Thaker Piping Technology & Products, Inc.