Projects — Joseph N. E. Lucero

“Science, my lad, is made up of mistakes,
but they are mistakes which it is useful to make,
because they lead little by little to the truth.”

— Jules Verne

Projects I’m currently working on

Integrating phase-field models into battery management systems

Battery management systems (BMS) are used to monitor the state of battery systems generally contain algorithms that allow it to estimate the internal state of the battery during operation. Often many of the algorithms deployed in the BMS are based on empirical models like equivalent-circuit models that are capable of predicting the voltage dynamics of a particular battery given an input of current. These predictions are then used in filtering algorithms to determine, for example, the state-of-charge of the battery system. While these empirical models have been quite successful and a multitude of estimation algorithms have been designed around them, they ultimately do not contain any physical insights into what is happening within the battery during operation. In contrast, physics-based electrochemical models connect both solid-state physics and electrochemistry in an attempt to capture the transport of lithium within the electrodes and the electrolyte bulk, as well as the generation of lithium or lithium ions via surface redox chemical reactions at the surface of the electrodes. My research aims to leverage these physics-based models and design next-generation BMS algorithms that leverage these models, with some additions to describe electrodes with phase-separating components, to give a better understanding of how lithium evolves during battery operation and use these to improve state of charge estimation.

“Comparing mass-preserving numerical methods for the lithium-ion battery single particle model”. JNE Lucero, L. Xu, S Onori. (in review)

“Thermodynamically-consistent corrections to the single particle model of lithium-ion batteries”. JNE Lucero, Y. Gao, S. Onori. (in preparation)

Modeling battery packs for electrified heavy-duty commercial vehicles along with the charging and grid infrastructure to support their electrification

The majority of freight movement in the US occurs within 250 miles of shipping ports. This type of freight is known as “drayage” and has been identified to have a significant contributor to carbon dioxide emissions in the shipping sector. As such, the heavy-duty commercial vehicles which are used for drayage have been identified to be prime candidates for electrification. In collaboration with Oak Ridge National Lab and the Ohio State University, we work to build a framework that can co-optimize the battery chemistry used in the vehicle, the sizing of the battery pack, and the charging infrastructure given the typical mission demands and the available energy from the grid in order to minimize the total cost of ownership for these electrified HDCVs. Ultimately, we hope to provide guidance on how to best electrify these vehicles and identify the corridors originating from the different multi-modal ports in the US which require the most infrastructure support.

“An experimentally validated electro-thermal EV battery pack model incorporating cycle-life aging and cell-to-cell variation”. JNE Lucero, VA Sujan, S Onori. IEEE Transactions on Transportation Electrification.

Read further: Journal Article | PDF

Physically Optimizing Inference

Data is scaling exponentially across the natural sciences, primarily driven by rapid technological improvements but also because of the growing power and increased ease of computation. In particular, Machine Learning has made impressive strides in a multitude of pattern-recognition tasks, We ask the question of whether the same methods can be used to automate construction of models for complex natural systems, particularly those found in biology. We identify here an important connection between Machine Learning and Physics that reveals design principles for experimental protocols which optimize the physical conditions of a system that maximizes the quality of the resulting inferred predictive model.

“Physically optimizing inference”. JNE Lucero, CY Chen, A Huang, B Sheldan, DA Sivak, M Thomson
(in prep.)

Published (previous) projects

Stochastic Thermodynamics of Gaussian Information Engines

A thought experiment proposed by James Clerk Maxwell proposed in the 19th century, which was later named Maxwell’s Demon in honor of him, illuminated the link between information and thermodynamics. It was only recently, due in part to advances in single molecule techniques, that strong experimental evidence has been built up that this link between information theory and thermodynamics is real: information acquired by a system can be converted into work. In this work we focused on realizing experimentally a Maxwell demon that is conceptually simple consisting of a fluctuating mass on a spring attached to a movable stage. The stage is moved in response to certain types of fluctuations in the position of the mass. As such, this demon (or well-designed device in modern terms) processes information about a continuous degree of freedom. I have thus far explored this system widely through computational simulations and developed a simple theoretical formula for quantifying the information processed using Control Theory methods.

“Bayesian information engine that optimally exploits noisy measurements”. TK Saha, JNE Lucero, J Ehrich, DA Sivak, J Bechhoefer. Phys Rev Lett 129, 130601 (2022).
Read further: Journal Article | PDF | arXiv article

“Maximal fluctuation exploitation in Gaussian information engines”. JNE Lucero, J Ehrich, J Bechhoefer, DA Sivak. Phys. Rev. E, 104, 044122 (2021)
Read further: Journal Article | PDF | arXiv article

“Maximizing the power and velocity of an information ratchet”. TK Saha, JNE Lucero, J Ehrich, DA Sivak, J Bechhoefer. Proc. Natl. Acad. Sci. USA, 118(20):e2023356118 (2021)
Read further: Journal Article | PNAS Commentary | arXiv article

“Stochastic Thermodynamics of Gaussian Information Engines” (Masters Thesis)
Read further: Summit | PDF

Principles of Maximal Energy Transduction in Stochastic Rotary Motors

Molecular machines are complexes found inside organisms are generally which are responsible for carrying out the functions that are required for your cells to stay alive are generally made up of many smaller subunits. For example, the molecular machine FoF1-ATP synthase is a rotary motor found lining the inner membrane of the mitochondria and is a key machine in the process of turning the sugars that you eat into ATP, the energy currency that your cells use. The FoF1-ATP synthase complex is composed of two subunits Fo and F1. Due to their size and surrounding environment these subunits are generally subjected to really strong fluctuations. Despite this we generally find this machine to operate very efficiently, precisely, and transduce energy between subsystems really well. To figure out how it does this we explored a simple dynamical model of this machine and found that the presence of fluctuations makes it so that subunits rigidly coupled to one another is, non-intuitively, not the optimal configuration if the goal is for the machine to transduce energy.

“Nonequilibrium Energy Transduction in Stochastic Strongly Coupled Rotary Motors”. E Lathouwers, JNE Lucero, and DA Sivak. J Phys Chem Lett, 11, 5273-5378 (2020).
Read further: Journal Article | PDF

“Energy and Information Transduction in Strongly-Coupled Systems” (Honors Thesis)
Read further: Summit | PDF

Optimal Control of Rotary Motors

FoF1-ATP synthase is a central molecule in all living things because it is a key component in the chain that converts the sugars that you eat to ATP, the energy currency of your cell. Since it is such an important component of this chain, it stands to reason that this machine should be very efficient and indeed experimentally we find that it is. In order to find out why it’s so efficient we looked at how one should externally drive this motor between its different configurations. We used a recently derived theoretical result to design a protocol that minimizes the amount of energy wasted by the machine as it is externally driven, but interestingly we found that imposing this minimization sometimes came at the cost decreased chemical output of the machine.

“Optimal control of rotary motors”. JNE Lucero, A Mehdizadeh, and DA Sivak. Phys Rev E, 99, 012119 (2019)
Read further: Journal Article | PDF

Other projects that I’ve been involved with

Inferring B-MTB Phylogenies Using
MIRU-VNTR Data

Creating a robust phylogeny for the Beijing lineage of Mycobacterium Tuberculosis (B-MTB) has been of increasing interest in the past decade, particularly because this lineage is the cause of the pulmonary disease Tuberculosis (TB). It is estimated that about a third of the world’s population is infected with this disease and thus it is one of the largest, albeit silent, pandemics that plague the world today. Understanding the evolutionary ancestry of B-MTB becomes important as the understanding the mechanism of how the current strains came to be can be used to build a predictive model of how the strains will evolve in the future. I designed and implemented a Maximum Likelihood method to infer the phylogenies of B-MTB that utilizes a newly available bioinformatic genetic marker known as MIRU-VNTR, given a stochastic models of the genomic evolution. This work was done under the supervision of Dr. Leoni d Chindel evitch.

Modeling of Applicators in “egs_brachy”

Brachytherapy is a form of radiation therapy that treats tumorous from short distance. Such treatments are generally achieved by placing radioactive sources on, or within, the tissue to be irradiated. The current standard for dosimetry is known as TG-43 and while it was an improvement over previous formalisms it is limited in that it is incapable of capturing certain patient-specific parameters. To remedy this there has been recent efforts to begin utilizing Monte Carlo simulations to do dosimetry. I was tasked to extend the default library of applicators (devices that enable remote-loading of radioactive sources near deep-seated tumors) in EGSnrc user code “egs_brachy” by computationally modeling a clinically utilized applicator targeting gynecological tumors. This work was done under the supervision of Dr. Rowan Thomson.

“A MC-based anthropomorphic test case for commissioning model-based dose calculation in interstitial breast 192-Ir HDR brachytherapy”. V. Peppa, RM Thomson, SA Enger, GP Fonseca, C Lee, JNE Lucero, F Mourtada, FA Siebert, J Vijande, P Papagiannis. Med. Phys. 1-13. (2023)