The separation of a natural gas mixture containing 85 mol% CH4 and 15 mol% N2 by dual reflux pressure swing adsorption (DR PSA) was investigated using dynamic non-isothermal numerical simulations previously benchmarked against experiments. Two different adsorbents were employed: a commercially available activated carbon (AC1) and the recently developed ionic liquidic zeolite (ILZ), which has about twice the selectivity for methane over nitrogen than the activated carbon. Four DR PSA cycle configurations (PL-A, PH-A, PL-B and PH-B) were studied and iteratively optimized in terms of separation performance and energy consumption. The dynamic non-isothermal simulations, implemented in the software platform Aspen Adsorption, included a polytropic compressor model with temperature dependent heat capacities for more realistic predictions of the compressor duty. The heavy product to feed ratio, the feed step time and the light reflux ratio were selected as key optimization parameters for all four cycles at fixed bed pressure ratio, feed location and feed flow rate. The simulations showed that A-cycles were superior to B-cycles in terms of their separation performance for both adsorbent materials. Overall, the separation performance achieved by AC1 was lower than that by ILZ. At optimized conditions for AC1, PL-A and PH-A cycles achieved approximately 92 mol% CH4 in the heavy product and 4.5 mol% CH4 in the light product with corresponding energy requirement at 9.9 kJ mol−1 (feed) and 8.7 kJ mol−1 (feed), while using the ILZ adsorbent, both PL-A and PH-A cycles were able to deliver 0.9 mol% CH4 in the light product and 93.5 mol% CH4 in the heavy product. The cycle work of the PL-A and PH-A configuration using ILZ was 8.8 and 7.7 kJ mol−1 (feed), respectively. These simulation results show that DR PSA processes using ionic liquidic zeolites can meet strict natural gas separation requirements and that such processes can efficiently upgrade sub-quality natural gas.
Read full abstract