Current Work

I work on the thermal management of electronic devices. For my PhD, I am working on designing highly optimal cold plate geometries. Cold plates are single-phase liquid cooling solutions used in high heat flux applications such as in data centers, electric vehicles, radar systems, etc. Although cold plates can be used in various applications, I am more focussed on the design of optimal cold plates for data center applications. This work leverages machine-learning methods to supplant traditional intuition-based design methods that require manual searching through a complex design space (boundary conditions and available surface morphologies), thereby enabling more optimal cold plate performance.

Till now, I have

• Compiled an extensive database of fully developed Nusselt number and friction factor values of different constant cross section flow geometries (~800 distinct geometries).
• Developed machine learning based surrogate correlation for predicting the fully developed Nusselt number and friction factor of different constant cross section flow geometries.
• Used the surrogate correlations to find novel cross sections with desired properties.

Currently I am generating training data for heat transfer and pressure drop characteristics of three dimensionally varying flow geometries using automated numerical simulations.

Publications

• S.S. Pai, D. Visaria, J.A. Weibel, “A machine-learning-based surrogate model for internal flow Nusselt number and friction factor in various channel cross sections” The IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems (ITherm), June 2021.
• S. Ozguc, S.S. Pai, J.A. Weibel, L. Pan, “Experimental Demonstration of an Additively Manufactured Vapor Chamber Heat Spreader”, The IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems (ITherm), Las Vegas, USA, May 2019
• A. Patel, S. S. Pai, H. R. Rajamohan, M. Bongarala, R. Samyak, “Finding novel links in COVID-19 graph using graph embedding techniques” Springer Communications in Computer and Information Science (under review)

Past Work

Undergrad Thesis

• Towards Efficient Computation of Rarefied Flows using Field Inversion and Machine Learning Techniques [pdf] Advisor: Prof. Balaji Srinivasan, Department of Mechanical Engineering, IIT Madras

  This work used machine learning methods to bypass the need for expensive high-fidelity computations which are needed for accurate flow description of rarefied flows.

  • Successfully implemented the technique of ‘Field Inversion and Machine Learning’ on different cases of heat transfer and used it to model the structure of an acoustic shock wave.
  • The Maximum Likelihood Estimation approach was used to develop an Artificial Neural Network model to predict the flow of rarefied gas around a spherical body.

Research Internship (at The Ohio State University, summer 2018)

• Device Fabrication and Electrical Characterization of Vessel Function Within Bifurcating Microfluidic Channels (not sure if I can share the presentation on a public place. Contains images of work done in Prof. Prakash’s lab).
  Guides: Prof. Shaurya Prakash, Prof. Jonathan Song, Department of Mechanical and Aerospace Engineering

  This work focused on developing the trans endothelial electrical resistance measurement technique to study the endothelial barrier function in bifurcated vessels, the breakdown of which is linked to various diseases like multiple sclerosis, cancer, etc.

  • Fabricated microfluidic devices using soft lithography and plasma oxidation and used them to make in vitro models of blood vessels using HUVECs. Also learnt the basics of cell culture.
  • Used Electrochemical Impedance Spectroscopy (EIS) to study the vessel barrier function and suggested a new design for the microfluidic device to improve the quality of the EIS results.

Research Internship (at Purdue University, summer 2017)

• Characterisation of 3D Printed Vapour Chamber Heat Spreaders[one-slider]
  Guides: Prof. Justin A. Weibel, Prof. Suresh V. Garimella, School of Mechanical Engineering

  Metal 3D printing holds promise in thermal management application by allowing integrated manufacturing of vapor chamber heat spreaders with the processor package. I worked on building the charging facility needed for studying the performance of these vapor chambers.

  • Designed and fabricated a Fluid Charging Station for evacuating 3D printed vapor chambers and filling them with a precise volume of degassed fluid, and wrote down its User’s Manual.
  • Designed the vapor chamber heat spreader which was to be 3D printed.