Author ORCID Identifier

https://orcid.org/0000-0002-2368-3380

Semester

Fall

Date of Graduation

2025

Document Type

Dissertation (Campus Access)

Degree Type

PhD

College

Statler College of Engineering and Mineral Resources

Department

Chemical and Biomedical Engineering

Committee Chair

Fernando V. Lima`

Committee Co-Chair

Oishi Sanyal

Committee Member

Brent Bishop

Committee Member

Yuhe Tian

Committee Member

David Mebane

Abstract

Historical data shows a close relationship between the presence of CO2 in the atmosphere and the global average temperatures. Elevated CO2 levels have been linked to prolonged droughts, floods, wildfires, and ocean acidification, all of which origin from disruptions to the naturally occurring greenhouse effect. Because CO2 emissions are intrinsic to energy generation and industrial processes, carbon dioxide removal (CDR) technologies have emerged as critical tools to address both current and legacy emissions. Among these, Direct Air Capture (DAC) offers the unique advantage of removing CO2 directly from ambient air, regardless of the geographical location of the emission source. Advances in membrane separation performance under low driving forces, have recently renewed interest in applying membranes for DAC (m-DAC), offering a potentially energy-efficient and modular pathway for atmospheric CO2 removal. This dissertation employs process systems engineering methodologies to evaluate m-DAC through an integrated operability and machine learning framework. Using AVEVA Process Simulation, the m-DAC process was first modeled as a multistage hollow-fiber membrane system. Operability analysis was then applied to examine how intrinsic membrane properties, surface area, and feed loading affect CO2 capture and energy efficiency. By identifying feasible and optimal operating regions, operability provides a systematic way to assess process viability early in the design stage, before detailed equipment or material specifications are available, therefore reducing development time and avoiding infeasible design regions. Subsequently, inverse design techniques were implemented within the same framework to identify combinations of CO2 permeance and CO2/N2 selectivity that meet specific capture targets, with machine learning algorithms assisting in mapping and searching for enhanced membrane properties. The results revealed that system performance is primarily governed by the first membrane stage. Additionally, the feed studies showed that smaller-scale configurations promoted higher capture efficiency. From the inverse design studies, permeances between 6,000–7,000 GPU and selectivities around 2000 achieved permeate streams containing approximately 5% CO2, representing a 120-fold enrichment over ambient air (420 ppm). Preliminary techno-economic analysis (TEA) of this configuration also provided estimates of a minimum capture cost. Building on these findings, the investigation was extended to facilitated- transport membranes (FTMs), which exhibit superior performance under low CO2 partial pressures. The simulation framework was modified to include the equilibrium constant (Keq) and CO2 diffusion coefficient (DCO2) as variable inputs, enabling further operability and sensitivity analyses. The optimal parameter combinations derived from this study were implemented in single-stage simulations, achieving capture capacities of approximately 1.2 tons CO2 per day and permeate concentrations of 1–2%. Finally, a TEA of a coupled m-DAC and steel-slag mineralization process was performed to assess the economic and environmental viability of utilizing captured CO2 to produce carbonated slag, a precursor for cement and concrete. The integrated system achieved annual CO2 mineralization of 940-970 tons. While baseline conditions were not economically viable, breakeven analysis over a 45-year lifetime indicated feasibility at product (carbonated slag) prices of $99/ton and $116/ton, respectively for 1% and 2% CO2 feed conditions. Carbon intensity assessments further demonstrated that achieving net-zero or carbon-negative operation requires electricity supplied predominantly from renewable sources. In conclusion, this work demonstrates how integrating operability with process simulation and data-driven tools can accelerate the discovery-to-deployment cycle of emerging technologies. By bridging experimental membrane development with computational process design, this framework not only advances m-DAC but also provides an adaptable methodology to expedite the early-stage evaluation and scale-up of other energy and carbon-intensive processes.

Download

Share

COinS