To realize large-scale CO 2 separation to prevent global warming, energy and space saving separation technologies are required. Flexible metal−organic frameworks (flexible MOFs) with gate-opening properties have been found to be promising adsorbents, but the unconventional sigmoidal shapes and hysteresis of the isotherms have hindered quantitative evaluation of their applications in separation processes. This paper is the first to evaluate ELM-11, a flexible MOF that exhibits sigmoidal isotherms with hysteresis, in a process model and quantify its advantages over a conventional adsorbent. Based on the experimental uptake data, isotherm models for adsorption and desorption were developed, and their parameters were estimated. The resulting isotherm models are incorporated into a rigorous dynamic model of partial differential algebraic equations (PDAEs) to simulate a vacuum pressure swing adsorption (VPSA) process to evaluate the process performance. Numerical challenges to solve the PDAEs, including sigmoidal and hysteresis isotherm models, were resolved with the proposed numerical approaches. Sensitivity analysis for feed pressure and temperature was performed to identify the optimal operating strategy. A comparison with a conventional adsorbent, zeolite 13X, showed that high selectivity and sigmoidal isotherms of ELM-11 give higher productivity and CO 2 product purity exceeding 99% without rinse and purge operations as well as lower power consumption for the compressor and vacuum pump.

Flexible metal–organic frameworks (flexible MOFs) are considered promising adsorbents for CO2 capture, some of which have sigmoidal isotherm shapes that allow adsorption and desorption operations within a narrow partial pressure range. Nevertheless, modeling of adsorption processes employing flexible MOFs remains a challenge due to the unique isotherm shapes and kinetics. In this work, a Bayesian estimation framework is applied sequentially to handle two experimental data sets: isotherm and breakthrough measurements. The computational challenge for estimating the isotherm and kinetic parameters from the isotherm measurements and breakthrough experiments is resolved by Markov chain and sequential Monte Carlo methods. The uncertainties of the model parameters are obtained as probability distributions.

No abstract