(4l) Digitalization in Chemical Engineering: Accelerating Scientific Discovery and Enabling Smarter Manufacturing

Research Interests: digital twins, computational optimization, data science, machine learning, pharmaceutical modeling

As a systems engineer, I strive to develop new digital modeling, optimization, and data analysis methods and software that enable scientific and engineering breakthroughs in diverse domains in medicine, sustainable energy, and beyond. This poster outlines my past, present, and future research directions in process systems engineering as a future faculty member.

The Era of Digital Twins:

New paradigms in data science and intelligent decision-making frameworks are driving the next phase of the scientific revolution¹. Similarly, as engineering and manufacturing processes have become data-rich and digitally connected, methods for sanitizing, filtering, and leveraging historical data are pivotal to optimizing performance, robustifying operations, and increasing safety. New digital twin modeling paradigms provide a coherent research framework to realize these promises of digitalization.

The national academies define a digital twin as the following²:

“A digital twin is a set of virtual information constructs that mimics the structure, context, and behavior of a natural, engineered, or social system (or system-of-systems), is dynamically updated with data from its physical twin, has a predictive capability, and informs decisions that realize value. The bidirectional interaction between the virtual and the physical is central to the digital twin.”

My lab will develop new methods and accessible software tools to address key gaps in digital twin development, thereby addressing key research gaps recently identified by the national academies². We will strategically target societally pressing multi-scale modeling challenges in key areas such as pharmaceutical manufacturing scale-up and sustainable energy. Generally, there is no one-size-fits-all approach for developing digital twins, however, we aim to produce computationally efficient algorithms and models that will help balance model fidelity and computational tractability to create fit-for-purpose digital twins.

Research Area 1: Pharmaceutical Manufacturing

As the pharmaceutical industry modernizes, switching from the traditional batch manufacturing mode to continuous manufacturing has the potential to reduce environmental impact, improve operator and consumer safety, and decrease manufacturing costs, among others. However, realizing these benefits requires new computational methods and software to build predictive fit-for-purpose digital twins. This is especially important as regulators are moving towards model-based analyses when certifying pharmaceutical manufacturing processes.

Recently, we developed PharmaPy³, a pythonic tool for pharmaceutical processing modeling. PharmaPy empowers researchers to model pharmaceutical processes with an open-source model library of tailored unit operations in Python. PharmaPy supports parameter estimation, uncertainty quantification, and model selection⁴ for the technoeconomic analysis of both scale and operating mode⁵, as well as feasibility and design space analysis under uncertainty⁶, among others^7,8,9,10. My group will extend these foundational capabilities in PhramaPy to develop and validate fit-for-purpose digital twins. We will also leverage the generalized nature of PhramaPy to explore new applications in biological systems, energetics manufacturing, and beyond.

Research Area 2: Optimal Experiments for Creating Digital Twins

Choosing the correct model and estimating model parameters with potentially limited resources or data is an important first step in developing a digital twin. Also, automated methods such as self-driving labs¹ require identification of experiments that maximize potential for scientific discovery. Optimal experimental design saves physical resources and time developing predictive digital twins. With this in mind, we have accelerated optimal design through the robustification and redesign of Pyomo.DoE¹¹, which leverages information content of candidate experiments to inform experimental campaigns using science-based/model-based decision making. Pyomo.DoE uses science-based design of experiments (SBDoE), which includes a physical model of the system as the basis for information content of potential experiments. This tool directly addresses the iterative process of refining a digital twin using data from the physical system. We bolstered computational efficiency and capabilities of Pyomo.DoE to allow for more complex models with a more intuitive user interface. This computational improvement particularly reduces development time for digital twins, moving the bottleneck to other tasks. As an independent researcher, my group will combine topological metrics and SBDoE methods to design programmable materials, pharmaceuticals, or energetics. This area also encourages collaboration for future projects where new modeling or algorithmic paradigms need to be developed with limited budget or data acquisition bottlenecks in mind to make fit-for-purpose digital twins a reality.

Research Area 3: Advanced Data Analytics

Topological and geometric metrics provide a new paradigm to analyze complex spatio-temporal data arising from molecular dynamics, chemical sensing, and image analysis¹². Computing these topological data analysis (TDA) metrics on large-scale data sets is sometimes intractable as data size can exceed dynamic memory. Also, keeping track of all discrete components of a data structure becomes computationally expensive. We developed high-throughput and scalable algorithms to compute these metrics quickly (two orders of magnitude faster than state-of-the-art tools) and without having to read the entire data file into memory¹². This is a large step in leveraging topological metrics in the loop of digital twin development. There are opportunities for real-time classification and control in manufacturing processes utilizing topological metrics because of these efficient algorithms. My group will continue to develop new algorithms and create open-source software^3,12,13,14 while pursuing new applications in real-time quality control in pharmaceuticals as well as optimization of material structures for tailored physical properties.

Research Integration and Translation:

Through these research areas, my group will create novel computational methods, algorithms, and software necessary for the emerging era of digital twins to revolutionize chemical engineering. I will start by leveraging my expertise in pharmaceutical manufacturing, crystallization, and sustainable energy systems, while simultaneously pursuing exciting new applications in systems biology, real-time control, electrification, and smart manufacturing. Moreover, I plan to leverage my unique expertise in systems engineering and digitalization to help lead high-impact, multidisciplinary collaborations, which is increasingly important in the shift toward team-oriented science.

Teaching Interests:

I am qualified to teach all undergraduate chemical engineering courses. However, my experience and expertise in numerical methods, mathematics, and process systems engineering make me particularly suited for process design, process control, mathematics, data science, and computational problem-solving. I have extensive experience TA-ing pharmaceutical modeling courses and would be excited to develop and teach courses in optimization, pharmaceutical modeling, and topological data analysis in chemical engineering.

AIChE Presentations this year:

Flexible Technoeconomic Analysis Tools for Evaluating Emerging Power Generation Technologies in Hourly Electricity Markets Using IDAES and Pyomo

10B: Modeling, Control and Optimization of Energy Systems I

Room 33A-(Upper level) - 1:12 PM-1:33 PM

Pyomo.Doe 2.0: Improved Usability and Computational Efficiency for Science-Based Design of Experiments (SBDoE)

10: Software Tools and Implementations for Process Systems Engineering

Room 33A-(Upper level) - 5:18 PM-5:36 PM

Inverse Design of Materials with Globally Optimal Topology and Geometry Through Mixed Integer Linear Programming (MILP)

Machine Learning for Soft and Hard Materials II

Sapphire Ballroom D-(Hilton Bayfront) - 8:36 AM-8:54 AM

References:

1. Abolhasani, M., and Kumacheva, E.(2023) The rise of self-driving labs in chemical and materials science, Nature Synthesis, 2, pp.483-492.
2. National Academies of Sciences, Engineering, and Medicine.(2024) Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. https://doi.org/10.17226/26894.
3. Casas-Orozco, D., Laky, D., Wang, V., Abdi, M., Feng, X., Wood, E., Laird, C., Reklaitis, G.V., and Nagy, Z.K.(2021) PharmaPy: An object-oriented tool for the development of hybrid pharmaceutical flowsheets, Comput. Chem. Eng., 153, https://doi.org/10.1016/j.compchemeng.2021.107408
4. Casas-Orozco, D., Laky, D., Mackey, J., Reklaitis, G., and Nagy, Z.(2023) Reaction kinetics determination and uncertainty analysis for the synthesis of the cancer drug lomustine, Chemical Engineering Science, 275, https://doi.org/10.1016/j.ces.2023.118591
5. Casas-Orozco, D., Laky, D., Wang, V., Abdi, M., Feng, X., Wood, E., Reklaitis, G.V., and Nagy, Z.K.(2023) Techno-economic analysis of dynamic, end-to-end optimal pharmaceutical campaign manufacturing using PharmaPy, AIChE Journal, 69(9), https://doi.org/10.1002/aic.18142
6. Laky, D., Casas-Orozco, D., Rossi, F., Mackey, J.S., Reklaitis, G.V., and Nagy, Z.K.(2022) Determination of probabilistic design spaces in the hybrid manufacture of an active pharmaceutical ingredient using PharmaPy, Computer Aided Chemical Engineering, 49, pp.2131-2136.
7. Laky, D.J., Casas-Orozco, D., Destro, F., Barolo, M., Reklaitis, G.V., and Nagy, Z.K.(2022) Integrated synthesis, crystallization, filtration, and drying of active pharmaceutical ingredients: a model-based digital design framework for process optimization and control, Optimization of Pharmaceutical Processes, pp.253-287.
8. Laky, D.J., Casas-Orozco, D., Laird, C.D., Reklaitis, G.V., and Nagy, Z.K.(2022) Simulation-optimization framework for the digital design of pharmaceutical processes using Pyomo and PharmaPy, I&ECR, 61(43), pp. 16128-16140.
9. Barhate, Y., Laky, D.J., Casas-Orozco, D., Reklaitis, G.V., and Nagy, Z.K.(2024) Rule-based decision framework for the digital synthesis of optimal pharmaceutical processes. ESCAPE34/PSE24
10. Hur, I., Casas-Orozco, D., Laky, D.J., Destro, F., and Nagy, Z.K.(2024) Digital design of an integrated purification system for continuous pharmaceutical manufacturing, Chemical Engineering Science, 285, 119534
11. Wang, J. and Dowling, A.W.(2022) Pyomo.DOE: An open-source package for model-based design of experiments in Python, AIChE Journal, 68(12), https://doi.org/10.1002/aic.17813
12. Laky, D.J., and Zavala, V.M.(2024) A fast and scalable computational topology framework for the Euler characteristic, Digital Discovery, 3, pp. 392-409
13. Laky, D.J., Cortes, N.P., Eslick, J.C., Noring, A.A., Susarla, N., Okoli, C., Zamarripa, M.A., Allan, D.A., Brewer, J.H., Iyengar, A.K.S., Wang, M., Burgard, A.P., Miller, D.C., and Dowling, A.W.(2024) Market optimization and technoeconomic analysis of hydrogen-electricity coproduction systems, Submitted
14. Laky, D., Xu, S., Rodriguez, J.S., Vaidyaraman, S., García Muñoz, S., and Laird, C.(2019) An optimization-based framework to define the probabilistic design space of pharmaceutical processes with model uncertainty. Processes, 7 (2)