Thermally Driven Gas Separation System

Position – Graduate Research Assistant

Host – Sustainable Thermal Systems Laboratory 

PI – Dr. Srinivas Garimella

Duration – August 2011 – December 2015


Natural gas has become increasingly important as a fuel source with lower environmental impact; therefore, there is a growing need for scalable natural gas purification systems with small footprints. Current industrial purification systems are based on absorption, membrane separation, or adsorption techniques; however, each of these technologies requires large capital costs or suffers from scalability issues. Adsorption-based separation techniques are categorized into pressure-swing adsorption (PSA) and temperature-swing adsorption (TSA). Among adsorption-based gas purification techniques, PSA has typically been preferred over TSA due to the ease of operation and reliability. TSA processes have not commonly been used for industrial gas separation due to the typically low thermal conductivity of the adsorbent bed, which poses challenges for desorption of impurities and regeneration of the adsorbent. However, the high heat and mass transfer coefficients possible with microchannels offer the potential for using the TSA process for gas purification and enhance the performance of existing PSA processes.


  •  Identify and simulate heat and mass transfer processes in each of the stages in a temperature swing adsorption process cycle that uses adsorbent-coated microchannels as the bed.
  • Develop an experimentally-validated model for a pressure enhanced temperature swing adsorption gas purification process.
  • Generate a process performance map and compare with existing processes to demonstrate industrial feasibility.

Process Description 

Complete purification process cycle showing integral steps

The temperature swing adsorption process studied in this work uses the same microchannels for transport of the gas and the heat transfer fluid, resulting in a four stage process as shown.

1.      Adsorption stage – CO2 is removed from CH4 by passing the feed gas through the adsorbent microchannel.

2.      Desorption stage – Hot heat transfer fluid (HTF) enters the microchannel, heats the adsorbent, and desorbs the adsorbed gases. These desorbed gases are carried away with the hot liquid stream.

3.      Cooling stage – After impurities are removed from the adsorbent, cold liquid is sent through the microchannel, which lowers the adsorbent temperature and prepares it for the next cycle.

4.      Purge stage – After the adsorbent is cooled, the cold liquid is driven out of the microchannel by a purge gas, bled from the product stream. The purge gas is expected to displace the cold liquid in the microchannel and dry it by shearing the liquid attached to the microchannel walls.


Major Results

Interaction of Gases and Liquids in Microchannels

Interface velocities, void fractions, and film thicknesses were determined using high speed recording and image analysis software that detects the liquid-gas interface locations and compared with the computational models. The important findings from the study are:

  • Mean and median error of 11% and 10%, respectively for displacement of gas. 90% of the predicted interface velocities within 20% of the experimentally observed values.
  • Mean and median errors of 20% and 15%, respectively. This decrease in model accuracy for the displacement of liquid attributed to liquid-gas interface deformation and the presence of several different flow patterns at the liquid-gas interface as shown in Figure 2.
  • Liquid films as large as 30 µm thick observed.
  • One of the first ever experimental and analytical investigations of sequential two-phase displacement phenomena that reinforced the need of an efficient purge stage in the overall purification cycle. The presence of residual liquid films on the microchannel walls could hinder mass transfer from the microchannel to the adsorbent wall.
Displacement of liquid – Steady films at low interface velocity
Displacement of liquid -Wavy films at high interface velocity
Displacement of liquid -Slug flow at high interface velocity
Displacement of gas – Clean displacement without any residual flow regime

Design, Modeling and Optimization of Modified TSA process

  • First, a pressure swing adsorption process using microchannels was simulated. Based on these results, a TSA process using separate microchannels for feed and HTF was modeled. The use of microchannels resulted in the process capacity enhancement by up to 25 times. The results from these feasibility studies suggested the use of the same microchannel for both gas and HTF and the separation process becomes that shown in the schematic above.Phenomena-based computational simulation of the cyclic steady state of the purification system and further optimization of the geometry showed that a cycle time of 203 s can yield an excellent purification performance . This overall cycle time of the TSA process is smaller than the PSA processes documented in the literature (600 – 1200 s).
  • The process capacity up to two orders of magnitude greater than those reported for the adsorbent bed-based PSA systems by Kapoor and Yang (1989) and Olajossy et al. (2003) with a competitive set of product purity and CH4 recovery factors (Figure a). With a 203 s cycle time, and an initial mole fraction of 70% CH4, a range of product purities from 87% to 99% CH4 possible, simultaneously recovering up to 83% CH4 from the feed stream (Figure b).
  • The specific energy utilization predicted for the present work as low as 0.68 kWh kg-product-1 and lower than the actual energy requirement for MEA absorption systems and the reversible electrical input for cryogenic distillation installations (Figure c).
  • Product purities up to 99.9% possible through appropriate timing of the process stages and by maintaining the necessary thermodynamic equilibrium between the stages.
  • Detailed information on this work can be found in:

Pahinkar, D. G., S. Garimella and T. R. Robbins (2017), “Feasibility of Temperature Swing Adsorption in Adsorbent-Coated Microchannels for Natural Gas Purification,” Industrial & Engineering Chemistry Research 56(18) pp. 5403-5416 DOI:

Pahinkar, D. G., Garimella and T. R. Robbins (2015), “Feasibility of Using Adsorbent-Coated Microchannels for Pressure Swing Adsorption: Parametric Studies on Depressurization,” Industrial & Engineering Chemistry Research Vol. 54(41) pp. 10103-10114.

Pahinkar, D. G. and S. Garimella (2018), “A Novel Temperature Swing Adsorption Process for Natural Gas Purification: Part I, Model Development,” Separation and Purification Technology 203 pp. 124-142 DOI:

Pahinkar, D. G. and S. Garimella (2018), “A Novel Temperature Swing Adsorption Process for Natural Gas Purification, Part II: Performance Assessment,” Separation and Purification Technology 204 pp. 81-89 DOI:

(a) Process capacity vs. Product purity (b) Methane Recovery vs. Purity (c) Process Capacity vs. Energy Requirement for the TSA-based process.

Experimental Validation of Gas Separation

In this part of the overall research, gas separation in adsorbent-coated microchannels was studied experimentally and analytically.

  • For preliminary validation, adsorption experiments conducted on PLOT columns over a range of imposed ΔP and L.
  • Trace water vapor adsorption affected the repeatability of the adsorption tests. Installation of gas dryers in flow paths resulted in reliable and uniform adsorption time and ΔT data.
  • The model predictions for customized adsorbent-coated microchannels with known packing properties agree very well with the data with an AAD of 4% for adsorption time. Local fabrication variability results in a 25% AAD for ΔT and 20% for ΔTMax. A sample model validation for single data point for heat transfer and mass transfer is shown adjacent figures.
  • Detailed information on this aspect of the project can be found here –

Pahinkar, D. G., Garimella (2017), Experimental and Computational Investigation of Gas Separation in Adsorbent Coated Microchannels. Chemical Engineering Science, Volume 173, 14 December 2017, Pages 588-606. DOI –

Heat Transfer model validation for sample data point.
Mass Transfer model validation for a single data point.

This work was sponsored by: ExxxonMobil Upstream Research Company, Houston, TX