Long-term Performance of Activated Carbon in Cyclic Adsorption/Regeneration of VOCs: Experimental and Modeling Investigation of Fixed Bed Adsorber

• Author / Creator

  Sadeghi Ardekani, Amin

• Activated carbon (AC) has attracted tremendous interest in adsorption-based air treatment. Nonetheless, a major challenge associated with the use of ACs is the decline in adsorption capacity with time due to heel build-up (i.e., accumulation of non-desorbed species). Designing a reliable adsorption system requires a deeper understanding of the changes occurring during the long-term use of ACs. For this purpose, the effect of ACs' properties such as porosity and operational conditions such as purge gas flow rate on the long-term performance of ACs requires further investigation.
  The objective of the present work was two-fold: first, to study the simultaneous effect of purge gas flowrate and activated carbon's porosity during prolonged cyclic adsorption/regeneration of three different ACs. Secondly, develop a model that can predict the long-term performance of ACs during adsorption/regeneration of a representative volatile organic compound (VOC). This section itself comprised two main stages: 1) Modeling the impact of heel on AC's pore size distribution (PSD), adsorption isotherm, and capacity, and 2) verifying the model using cyclic adsorption-desorption of 1,2,4-trimethyl benzene (TMB). The model predicts the cyclic adsorption capacity of AC by applying the Dubinin-Radushkevich-Langmuir (D-R-L) isotherm based on AC's limiting pore volume and adsorbate-adsorbent affinity coefficient.
  For the long-term experimental study, six scenarios were investigated by varying the dry air purge gas flow rates 0.5 and 5 SLPM and the porosity of adsorbent used (44%, 60%, and 86% microporosity). The cyclic adsorption/regeneration experiment results indicated that the cumulative heel and the adsorption capacity followed ascending and descending trends with cycle number, respectively. Initially, the porosity and micropore volume of the adsorbents played a more important role in their performance. However, at higher cycle numbers, the effect of purge gas flow rate was more determinant in the performance of ACs. In the first five cycles, the two adsorbents with the highest micropore volume, G-70R, and B101412, showed similar heel build-up formation rates while B100772 with lower micropore volume (0.43 (cm^3)/g as opposed to 0.50(cm^3)/g) had slightly lower heel build-up. Alternatively, at the 20th cycle, purge gas flow rate had a clear effect on the performance and cumulative heel build-up of all three ACs regardless of their porosity. For all three adsorbents used in this study, samples regenerated with 0.5 SLPM all had an average cumulative heel of 31 %. Those regenerated with 5 SLPM Had a cumulative heel build-up average of 21%. The presence of mesopores and a hierarchal pore structure certainly helped reduce heel build-up in the micropores. DTG analysis of the samples showed that with an increase in purge gas flow rate, the nature of heel build-up starts to change and transform into heavier chemically formed compounds.
  In the second part, two machine learning (ML) algorithms, multivariate linear regression (MLR) and Decision tree, were applied to predict Micropore volume reduction because of volatile organic compounds (VOCs) cyclic heel build-up on activated carbons (ACs). A dataset of 100 experimental tests of cyclic adsorption/regeneration of different VOCs on ACs with distinct properties was used. It was observed that micropore volume reduction could be predicted with acceptable accuracy with an R2 of 0.85 ± 0.08 using the MLR algorithm by considering the adsorbent characteristics, adsorbate properties, and regeneration conditions. The micropores prediction results were then combined with several mathematical equations to predict the pore size distribution of a used activated carbon. To verify the model, its results were tested against nine samples with various stages of heel build-up. The micropore and PSD were predicted with a mean relative absolute error (MRAE) of 3.5%, 10.8%, and 12.0% for G-70R, B101412, and B100772, respectively. The PSD prediction model was then utilized in conjunction with the DRL isotherm prediction model, and the adsorption capacity of samples at five concentrations of 0, 50, 100, 500, and 1000 ppm were predicted for each adsorbent. The prediction of adsorption capacity on the virgin G-70R, B101412, and B100772 had a MRAE of 0.6%, 8.9%, and 2.7, respectively while for the corresponding used samples the MRAE was 13.2%, 10.1%, and 10.0%. The results of this study are beneficial in improving the long-term performance of activated carbons and making them last longer.

• Subjects / Keywords

  • Adsorption
  • Activated Carbon
  • Heel
  • Long-term
  • pore size distribution
  • predictive model
  • isotherm

• Graduation date

  Spring 2022

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/r3-be2g-ba28

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.