Experimental validation of an adsorbent-agnostic artificial neural network (ANN) framework for the design and optimization of cyclic adsorption processes

The efficacy of an adsorbent agnostic machine-learning surrogate model for rapid design and optimization of a Skarstrom cycle vacuum swing adsorption (VSA) process is experimentally validated. The surrogate model is trained to predict the process performance using adsorbent features that include hypothetical Langmuir adsorption isotherm parameters, particle density, porosity and bed voidage, and process variables such as pressure, step duration and feed velocity. The training data was generated from a detailed process model for 20,000 unique combinations of the training variables. The model shows high accuracy of R2adj>0.99 for predicting key performance parameters such as product purity, recovery and productivity. The ability of this surrogate to predict the experimental performance for the purification of O2 from the air on two adsorbents, namely 13X and LiX zeolites, was studied. Two separate multi-objective optimization studies, to maximize purity and recovery, and to maximize productivity and purity were performed. For these optimization studies, the volumetrically measured isotherms of N2 and O2 were used as inputs to the surrogate model. Note that these isotherms were not a part of the dataset used to train the model. Nine points were chosen from the Parteo curves and the corresponding decision variables were used as set-points in a two-column lab-scale rig. The average difference between the calculated and experimentally measured purity, recovery and productivity was 3%, 5% and 9%, respectively. This study provides the necessary confidence to use surrogate-based process models for adsorbent screening and adsorption process optimization.

Physics-based neural networks for simulation and synthesis of cyclic adsorption processes

A computationally faster and reliable modelling approach called a physics-based artificial neural network framework for adsorption and chromatography emulation (PANACHE) is developed. PANACHE uses deep neural networks for cycle synthesis and simulation of cyclic adsorption processes. The proposed approach focuses on learning the underlying governing partial differential equations in the form of a physics-constrained loss function to simulate adsorption processes accurately. The methodology developed herein does not require any system-specific inputs such as isotherm parameters. Accordingly, unique neural network models were built to fully predict the column dynamics of different constituent steps based on unique boundary conditions that are typically encountered in adsorption processes. The trained neural network model for each constituent step aims to predict the entire spatiotemporal solutions of different state variables by obeying the underlying physical laws. The proposed approach is tested by constructing and simulating four different vacuum swing adsorption cycles for post-combustion CO2 capture without retraining the neural network models. For each cycle, 50 simulations, each corresponding to a unique set of operating conditions, are carried out until the cyclic-steady state. The results demonstrated that the purity and recovery calculated from the neural network-based simulations are within 2.5% of the detailed model's predictions. PANACHE reduced computational times by 100 times while maintaining similar accuracy of the detailed model simulations.

Physicochemical characterization of the performance of acidified modified eggshell cyclic adsorption of CO2

The CO2 calcium based adsorbent was prepared by using citric acid monohydrate, L(+)-tartaric acid, glacial acetic acid and L(-)-malic acid modified eggshell. The cyclic calcination/carbonation reaction of the adsorbent before and after modification was carried out by thermogravimetric analyzer to investigate the attenuation characteristics of the cyclic CO2 absorption performance of the calcium-based adsorbent after modification. XRD, N2 adsorption analysis and other methods were used for physical and chemical characterization. The phase composition, morphology, specific surface area and porosity of the acidified calcium based adsorbent were investigated. The results showed that: during 20 cycles of calcination/carbonation after 900°C pre-calcination, the maximum carbonation conversion of CIES900 modified with citric acid is significantly higher than that of the previous adsorbent, and its maximum carbonation conversion is 90.7%. The L(-)-malic acid modified adsorbent MAES900 significantly enhanced the carbonation cycle stability of the adsorbent before modification, and the cycle stability reached 92.9%.

The Experimental Study of the Kinetics of the Cyclic Adsorption Process of Air Enrichment with Oxygen

For mass transfer cyclic processes in the “adsorptive - porous adsorbent” system when air is enriched with oxygen by the method of short-cycle heatless adsorption, a new method has been implemented for determining the coefficients of mass transfer and mass conductivity of processes in systems with a solid phase from kinetic curves. It has been experimentally proved that during the adsorption separation of atmospheric air, the rate of cyclic “adsorption - desorption” processes can be limited by both internal and external diffusion resistance. The mass conductivity coefficients are determined depending on the mass content of the distributed adsorptive (O2, N2) by a method that does not require the implementation of the intradiffusion kinetic regime. The analysis of the kinetics of the process of air enrichment with oxygen is carried out; the coefficients of mass transfer and mass conductivity, which can be used in kinetic calculations and numerical study of the properties and modes of the cyclic adsorption process of atmospheric air separation and oxygen concentration, are determined.

Optimization of Cyclic Adsorption Processes and Gas Mixture Separation Plants

The analysis of the cyclic adsorption process and the installation of separation of gas mixtures by the method of pressure swing adsorption as an object of optimization have been carried out. The study found operating (control) variables (the duration of the adsorption stage, the pressure at the compressor outlet, the coefficient of reverse flow for the regeneration of the adsorbent, the program of changes in the opening time of the inlet and outlet valves of the installation); undefined parameters (composition, temperature and pressure of the initial gas-air mixture); output variables of the installation (concentration of oxygen, nitrogen in the product gas flow, the productivity of the installation, the degree of extraction and reduced costs for the production of oxygen with a given purity of 40 - 90 and higher vol.%). A one-stage problem of optimization of the regimes of a stationary periodic process of adsorption separation of atmospheric air and oxygen concentration was formulated and solved by the method of short-cycle adsorption under conditions of partial uncertainty of the initial information in the presence of restrictions on the purity of the product gas, the productivity of the installation and the rate of gas flow in the “frontal layer” of the adsorbent. An iterative algorithm for its solution is proposed.

Rapid Ammonia Carriers for SCR Systems Using MOFs [M2(adc)2(dabco)] (M = Co, Ni, Cu, Zn)

Ammonia is one of the most common reductants for the automotive selective catalytic reduction (SCR) system owing to its high NO2 reduction (deNOx) efficiency. However, ammonia carriers for the SCR system have sluggishly evolved to achieve rapid ammonia dosing. In this study, the MOFs [M2(adc)2(dabco)] (M = Co, Ni, Cu, Zn) were synthesized and characterized as ammonia carriers. Among the four obtained MOFs, Ni2(adc)2(dabco) possessed the highest surface area, 772 m2/g, highest ammonia uptake capacity, 12.1 mmol/g, and stable cyclic adsorption-desorption performance. All the obtained MOFs demonstrated physisorption of ammonia and rapid kinetics of ammonia adsorption and desorption. Compared with halide ammonia carrier MgCl2, the obtained MOFs showed four times faster adsorption kinetics to reach 90% of the ammonia uptake capacity. For the ammonia desorption, the Ni2(adc)2(dabco) provided 6 mmol/g ammonia dosing when temperature reached 125 °C in the first 10 min, which was six times of the ammonia dosing from Mg(NH3)6Cl2. The results offer a solution to shorten the buffering time for ammonia dosing in the SCR system.