Well-Mixed Dynamic Adsorber Model for Designing Cyclic Adsorption Processes
Year: 2026
Type: preprint
Abstract: The design and optimization of adsorption-based gas separations relies on process models that predict key performance indicators (KPIs) as functions of adsorbent properties and operating conditions. The one-dimensional axially-dispersed plug-flow (1-D ADPF) model is the established benchmark, but its computational cost limits its use in high-throughput tasks such as large-scale adsorbent screening and multi-objective process optimization. Existing simplified models achieve orders-of-magnitude speedups, yet the assumptions of a well-mixed adsorber and instantaneous equilibrium introduce large systematic errors in predicted KPIs. In this work, we present a dynamic well-mixed approximation of the 1-D ADPF model that retains the computational efficiency of prior well-mixed formulations while substantially improving predictive accuracy. The adsorber is represented by a well-mixed volume (or two in series for special cases). The formulation accounts for non-isothermal operation, finite heat and mass transfer rates, and linear pressure drop. The model is validated against experimental data and the 1-D ADPF simulations–including the solution of multi-objective optimization problems–across three representative applications: (I) pressure-vacuum swing adsorption (PVSA) for post combustion carbon capture; (II) temperature-vacuum swing adsorption (TVSA) for direct air capture; and (III) vacuum swing adsorption (VSA) process with light reflux for oxygen production from air. Across the first two cases, the well-mixed dynamic adsorption model predicts product purity and recovery, as well as specific energy consumption and productivity, with median deviations below ±0.35% while reducing simulation times by more than two orders of magnitude. Larger deviations were observed for the third case targeted towards light product recovery, but the well-mixed model correctly captures the overall trends in KPI trade-offs. The sensitivity of model accuracy to column geometry, mass transfer coefficient, and degree of bed utilization is systematically assessed, and practical validity boundaries are identified. These results establish a fast and accurate framework for adsorption process screening and optimization.
Source: ChemRxiv
Institution: Imperial College London
Language: English
Subfield: Mechanical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7203832871