Scholarly article on topic 'Effect of flue gas impurities on CO2 capture performance from flue gas at coal-fired power stations by vacuum swing adsorption'

Effect of flue gas impurities on CO2 capture performance from flue gas at coal-fired power stations by vacuum swing adsorption Academic research paper on "Chemical engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Energy Procedia
OECD Field of science
Keywords
{"Carbon dioxide capture" / "Flue gas" / "Vacuum swing adsorption" / Impurity}

Abstract of research paper on Chemical engineering, author of scientific article — Jun Zhang, Penny Xiao, Gang Li, Paul A. Webley

Abstract CO2 capture from major stationary emission sites has been studied widely with the increasing realization of the negative impact of greenhouse gas emissions on climate change. In terms of capture technology, solvent scrubbing, membrane processes and adsorption processes are the major contenders with the latter making significant progress over the last decade due to both improved adsorbent and process design and operation. As is well known, capturing CO2 from flue gases at coal-fired power stations by pressure/vacuum swing adsorption is complicated by the existence of significant amounts of water, SOx, NOx and other impurities, which are detrimental to most commercial CO2 selective-adsorbents. Conventional adsorption-based CO2 capture processes rely on using a pre-treatment stage to remove water, SOx and NOx, which adds considerably to the overall cost. In contrast, we report here an adsorption process developed in our laboratory which directly tackles the untreated flue gas without a separate pre-treatment stage by using a propriety multiple-layered bed comprising different adsorbents. The species CO2, H2O, SOx and NOx are processed in the same column within different function layers optimized according to adsorption properties and process conditions. A fully programmable logic controller (PLC) automated three-column pilot plant was built to perform the study with real-time control and data acquisition conducted through Human Machine Interface/Supervisory Control and Data Acquisition (HMI/SCADA) system. Through running continuous experiments, the effects of impurities on process performance such as CO2 purity, recovery and process power are investigated and reported. This is the first in-depth report of the performance of adsorption based capture plants in the presence of impurities found in real flue gas streams.

Academic research paper on topic "Effect of flue gas impurities on CO2 capture performance from flue gas at coal-fired power stations by vacuum swing adsorption"

Available online at www.sciencedirect.com

ELSEVIER

ScienceDirect

Energy Procedía

Energy Procedía 1 (2009) 1115-1122

www.elsevier.com/locate/procedia

GHGT-9

Effect of Flue Gas Impurities on CO2 Capture Performance from Flue Gas at Coal-fired Power Stations by Vacuum Swing

Adsorption

CO2 capture from major stationary emission sites has been studied widely with the increasing realization of the negative impact of greenhouse gas emissions on climate change. In terms of capture technology, solvent scrubbing, membrane processes and adsorption processes are the major contenders with the latter making significant progress over the last decade due to both improved adsorbent and process design and operation. As is well known, capturing CO2 from flue gases at coal-fired power stations by pressure/vacuum swing adsorption is complicated by the existence of significant amounts of water, SOx, NOx and other impurities, which are detrimental to most commercial CO2 selective-adsorbents. Conventional adsorption-based CO2 capture processes rely on using a pre-treatment stage to remove water, SOx and NOx, which adds considerably to the overall cost. In contrast, we report here an adsorption process developed in our laboratory which directly tackles the untreated flue gas without a separate pre-treatment stage by using a propriety multiple-layered bed comprising different adsorbents. The species CO2, H2O, SOx and NOx are processed in the same column within different function layers optimized according to adsorption properties and process conditions. A fully programmable logic controller (PLC) automated three-column pilot plant was built to perform the study with real-time control and data acquisition conducted through Human Machine Interface/Supervisory Control and Data Acquisition (HMI/SCADA) system. Through running continuous experiments, the effects of impurities on process performance such as CO2 purity, recovery and process power are investigated and reported. This is the first in-depth report of the performance of adsorption based capture plants in the presence of impurities found in real flue gas streams.

© 2009 Elsevier Ltd. All rightt reserved. Keywords: carbon dioxide capture, flue gas, vacuum swing adsorption, impurity

1. Introduction

With the increasing recognition of the importance/feasibility of CO2 capture and storage (CCS) in mitigating global warming, great efforts have been made in the academic, engineering, and political sphere to further our

* Corresponding author. Tel.: +61-3-9915-3445; fax: +61-3-9915-5686. E-mail address: paul.webley@eng.mcnash.edu.au.

Jun Zhang, Penny Xiac, Gang Li, Paul A. Webley

Department of Chemical Engineering, Monash University, Wellington Road, Clayton, Victoria, 3800, Australia CO2CRC - Cooperative Research Centre for Greenhouse Gas Technologies

Abstract

doi:10.1016/j.egypro.2009.01.147

understanding of details and role of this technology [1-5]. In addition to cryogenic process, solvent scrubbing and membrane technology, pressure swing adsorption has attracted much attention, with recent focus on new adsorbent materials and new cycle developments [6-8]. Pressure/vacuum swing adsorption, as a widely utilized technology in the gas separation/purification industry, shows great potential in CO2 capture from coal fired power stations (postcombustion capture) because of its promise of low energy consumption and system simplicity. Extensive investigations have been undertaken in applying adsorption technology to post-combustion CO2 capture. These efforts include materials preparation and characterization, cycle design, process optimization, the effect of impurities, the results of pilot experiments and so on [9-12]. The presence of several impurity gases (SOx/NOx/H2O) greatly complicates the CO2 separation processes. SO2 is reported to have reactions with basic sites on adsorbent materials. NO can be oxidized to NO2 /N2O3/N2O5 and eventually react with the adsorbent [13-15]. The pressure-swing reversible adsorption capacity for these species is very small [9,10,12]. Much research has been done on the adsorption/desorption of SOx/NOx on activated carbon fibre, activated carbon, ZSM and other zeolites, especially with dry gas. To protect the main CO2 adsorption layer, pre-layers need to be placed in the column. The concept of multiple layers is well established and applied in the industrial gas separation/purification area [16-17]. The interactions of CO2/H2O/SOx/NOx in typical flue gas with adsorbents are quite complex. Systematic work needs to be done to gain a better understanding of these interactions and their effect on the process performance.

In this study, we focus on the effect of water vapour on the pressure/vacuum swing adsorption process, especially the influence of water vapor in the product purge gas. In addition, we report the adsorption behaviour of SOx/NOx/CO2/H2O/O2/N2 multiple mixtures in a simple breakthrough column as a starting point of a more systematic R&D project currently in progress. Such experiments not only help our understanding of the fundamental of impurity adsorption on adsorbent performance but also provide valuable process design and equipment selection reference data. The adsorbents chosen in this study are CDX (an alumina/zeolite blend), alumina and NaX zeolite as these adsorbents will be either the pre-layer for water adsorption or the main CO2 adsorption layer in the packed bed. This study follows on from previous work conducted through the Cooperative Research Centre for Greenhouse Gas Technologies (CO2CRC) at Monash University in which a pilot scale VSA process was operated with dry flue gas and a single-column system was operated with wet flue cycle [7, 8, 18]. The objective of the current study was to understand the adsorption and desorption of water in CO2 capture process from flue gas emitted from coal-fired power stations. Engineering solutions developed during this study formed the basis for designs of adsorption-based demonstration plants to be operated at an actual power station in 2009.

2. Experimental Work

Systematic experiments investigating the effects of flue gas impurities were conducted. Water isotherms were measured on an IGA (Intelligent Gravimetric analysis) apparatus and CO2 isotherms were measured using a modified Micromeritics ASAP2010 volumetric gas adsorption analyzer. SO2 and NO isotherms on a dry basis have been reported and were not measured in our laboratories. However, the adsorption behaviour of these gases under wet conditions was studied. A simple breakthrough type of apparatus (column diameter 30mm with a length of 280mm) was used to measure simple breakthrough of SO2/NO/CO2 under wet conditions. In each experiment, the CO2/H2O/NO/N2/O2/SO2 mixture gas was passed through the column and after breakthrough, wet argon gas (wet gas mimicking the real process condition) was used to desorb the column. A HORIBA multiple gas analyzer was used for analysis and data were transmitted to a computer for logging.

To investigate the effect of water on a realistic multiple-bed VSA process, full cycle experiments with wet gas were conducted in our 3-bed VSA pilot plant. Details of the apparatus have been reported earlier [8]. To equip the apparatus for operation with wet fed gas, a feed gas bubbler was added. Blended CO2/air mixtures were saturated through the bubbler and fed to the adsorption columns. Process control and data acquisition were performed by GEFANUC PLC and Citect SCADA/HMI. The role of water for different adsorbent layering and different cycles were examined and are discussed below.

3. Results and discussions

3.1 Water Isotherms on Alumina F-255 and zeolite NaX

Water isotherms were measured on Alumina F-200(BASF) and zeolite NaX(UOP) and CO2/N2 isotherms were also measured on NaX. The water isotherms on both alumina and zeolite exhibit the well-known Type IV isotherm (Figure 1) and are represented by the BET model in our simulation work. The adsorption of CO2/N2 (Figure 2) are represented with dual-site Langmuir model.

Pressure, kPa.a

Figure. 1. Water isotherms on 13X and F-200 Alumina

Pressure,kPa.a Pressure,kPa.a

Fig. 2. CO2 and N2 isotherms on 13X zeolite

Figure 1 shows that both 13X and F-200 Alumina have good water adsorption capacity with F-200 alumina showing superior at higher humidity levels. This is directly relevant to flue gas conditions in which high relative humidities at elevated temperatures are not uncommon. The CO2 and N2 isotherms on NaX zeolite show good CO2/N2 selectivity even up to 90°C.

3.2 Cyclic Adsorption Experiments with wet flue gas 1 Single layered Adsorbent 3.2.1 NaX Zeolite

The most widely used adsorbent for CO2 capture is zeolite NaX. As a starting point in our studies, the beds were packed with NaX and the VSA process was tested under wet flue gas conditions. A 6-step conventional VSA cycle was run [7] to cyclic steady state using a variety of feed conditions. It was of particular interest to investigate the movement of the water adsorption zone in the beds. It is well known that water adsorption excludes CO2 from the adsorption sites and hence could potentially remove all separation capacity of the adsorbent. In all cases the feed concentration of CO2 was 13%. When the water percentage is low (bubbler set at a temperature of 20°C), the water front can be stabilized within the bed and the process still is able to capture CO2 at relatively high purity (70%) and recovery although an increase in power is inevitable due to the need to evacuate water. However, when the water vapor percentage is high (35~40°C saturated), the water front steadily migrates through the column and breakthrough can be clearly seen by observing the corresponding temperature profiles. Thermocouples T1 to T7 are located in the bed from the bottom (feed) to the top (CO2 free product), separated by approximately 10 cm. Figures 3 and 7 shows these temperatures one and seven days after startup. On day one, a temperature swing of 15°C is

observed at locations T3 to T7. This indicates CO2 adsorption and desorption at these locations. The smaller temperature swing at T1 and T2 corresponds to water adsorption and desorption. On day 7 the temperature change within the main section of the bed bed has decreased to less than 5°C indicating that the water front has moved slowly through the column and reduced the CO2 adsorption capacity. Figure 5 shows a 20-hr trend change for T5 at day 7. It clearly shows the temperature swing attenuation over the 20 hour period due to migration of water to this location and subsequent CO2 adsorption reduction. Eventually, the water front moved to T7 (top of the column) and reached breakthrough. From a power consumption perspective more energy is consumed in the vacuum pump in the presence of water vapour and the specific power consumption for CO2 capture is increased. Purity and recovery drop significantly with propagation of the water front until no separation of CO2 is possible.

Time, second

Time,second

T2 T3 T4 T5

Figure. 5. Temperature profile at T5 on day 7 over a 20 hour period

Figure. 6. 9-step cycle temperature swing profile

The mechanism of water migration is related to the water isotherm and the relative extents of feed and countercurrent purge. At low water humidity, water adsorption is within the favourable "Langmuir" zone of the isotherm (see Figure 1) and the water front does not move significantly into the column. Sufficient countercurrent vacuum flow is produced to regenerate the adsorbent and hence arrest the migration of the water zone. A simple model to predict the position of the zone has been developed by Wilson et al. [19]. At higher levels of water in the feed, the water front moves further into the column and a proportionally greater ratio of column now becomes saturated with water greatly reducing CO2 capacity and hence vacuum countercurrent flow. In addition, the water isotherm is now unfavourable leading to a spreading pattern during adsorption. The vacuum level achieved in our experiments (6-7 kPa) was not low enough to desorb sufficient CO2 gas to stabilize the water front. Lower vacuum level and higher column temperature would be among possible solutions to prevent the breakthrough of water as such conditions create more countercurrent CO2 flow and desorb additional water from the adsorbent. Prediction and control of the movement of water as a function of water concentration and process conditions is therefore key to successful operation of the process.

3.3 Multiple layer adsorbent column experiments

3.3.1 NaX-CDX multiple layered bed

CDX Alumina, (a commercially available blend of NaY and Alumina), is a potential candidate for handling SOx/NOx, as well as water. The adsorbent beds were layered with CDX at the feed end followed by NaX in a 30/70 split by weight. The conventional 6-step cycle without purge was used as before with a high level of water fed to the system (35~40°C saturated). After 5 days of continuous operation, the water front had migrated to the center of the column. The product CO2 purity dropped from 66.68% to 46% and the recovery decreased from 65% to 20%. Eventual water breakthrough occurred as before.

3.3.2 NaX-F255, 6-step and 9-step experiments with water

A third set of experiments were performed with an alumina/NaX multi-layered bed. In these experiments F-200 alumina and NaX (40/60 split) was loaded into the columns and a more powerful oil vacuum pump was used to provide deeper vacuum capability. A 6-step cycle (as above) and a 9-step cycle was used and a vacuum level of 3-4kPa.a was achieved. For high humidity feed levels, both cycles sustained more than one week running time (>3,000 cycles) at which stage the water front was found to be stabilized at around the middle of column without breakthrough. The results are summarized in Table 1. As shown in Table 1, product purity from the 9-step cycle with product purge (vs the 6-step cycle with no purge) does not change substantially although only a small purge flow was used (16% of the product flow). Higher recovery was achieved in the 9-step cycle due to extended evacuation time. The specific power consumption for the 9-step cycle is surprisingly lower than the 6-step cycle. The reason is related to the change in CO2 concentration in the bed after product purge and subsequent change in the pumpdown profile during evacuation. Additional product purge would ultimately increase the required pumpdown power.

Careful examination of Figure 6 shows that the temperature rise at thermocouples T1 to T4 is small for the first 200s indicating that the water zone has penetrated to this point. Beyond T4 the temperature rise is larger indicating CO2 adsorption is still taking place. A temperature "spike" is observed at approximately 450s - this is the result of the high CO2 concentration product purge which re-adsorbs to the bed during the purge step. The purge gas used is "wet" resulting as it does from the CO2 product removed in a previous cycle. The 6-step cycle and 9-step cycle experiments indicated that with proper purge, product purity can be increased and the water front can be controlled with proper layering, though the purge gas needs to be optimized in terms of purge ratio and purge pressure.

3.4 Simulations

To further guide understanding of the process, numerical simulations were conducted using our in-house simulator MINSA (Todd et al.[20]). Different cycles and conditions were simulated. The results are summarized in Table2.

Table 1. 6-step and 9-step experiments results

Temperature 35 C 35 C 45 C

Cycle 6-step 9-step with 9-step with

without purge purge

Adsorbents F-200, F-200, F-200,

zeolite 13X zeolite 13X zeolite 13X

Feed 1.4 1.4 1.4

pressure,kPa.a

Vacuum 3 3 3

pressure,kPa.a

Inlet CO2-13% CO2-13% CO2-13%

compositions H2O-0.05% H2O-0.05% H2O-0.07%

CO2 in waste 3.85% 4.25% 3.36%

Product purity 89.63% 98.90% 98.90

Recovery 74.9% 78.5% 82.7%

Power, 8.43996 6.57668 7.62

kW/TPD

Table 2 Simulation performances

6-step cycle without purge 9-step cycle with purge

Feed gas 41.5 35~40

temperature, °C

Feed CO2 purity 15%~16% 15~16%

Product purity 86.1% 90%

Recovery 60% 62%

Evacuation time, 100 120

second

Power, kW/TPD 12.39 10.41

00 1.20

Fig. 7 CO2 concentration profile

0.70 0.80 0.90 1.00

Fig.8. H2O concentration profile

These simulations greatly assist the elucidation of profile data since detailed concentrations can be calculated, such as CO2 and water axial concentration profiles at different steps (shown in Figures 7 and 8). From those results, it is found that with 40% F-200 in the adsorber column, the water front can be stabilized within the alumina layer at 3kPa and produces reasonable purity and recovery even at a high humidity level.

The simulations performed in this study agree reasonably with experimental results although exact numerical agreement is extremely difficult to achieve given the large number of input parameters and the strong dependence of the results on detailed competitive adsorption of the species. In addition, the simulation is based on adiabatic operation while the laboratory data are generated at non-isothermal conditions. A significant amount of heat transfer is observed especially through the connecting flanges. In addition, greater pressure drop (as large as 2kPa) on the vacuum line is found in the experiment than predicted by simulation due to various flow elements - flow meter, control valve, dust filer and solenoid valve which all contribute. As a result, the simulation overpredicts CO2 product purity which is extremely sensitive to pressure drop in the system.

3.5 SO2/NO breakthrough experiments

In this study, breakthrough experiments (20°C) of SO2 and NO were performed on both zeolite 13X (5 LPM, 90 ppm SO2, 45ppm NO in wet CO2) and CDX (10 LPM, 50ppm SO2, 250 ppm NO). These are candidate adsorbents for the main layer and prelayer as discussed above. The SO2/NO concentrations used reflect expected values in the flue gas after the flue gas has passed through a water wash tower. As can be seen from Figure 9(a), SO2 breakthrough time was very short (a few seconds) surprisingly and NO breakthrough is much slower. Similar roll-up phenomena as reported in the literature was observed as the SO2 concentration rapidly increased to a much higher

level than in the feed gas [11,13]. The reason is that more competitive water adsorption is taking place and relatively weaker adsorbed SO2 was desorbed in the breakthrough. Such an effect also occurs for O2 as shown in Figure 9(b) though CO2 adsorption was not much significantly affected. The desorption curves (Figure. 10) indicated slow desorption of SO2 and a quick desorption of NO and O2. CO2 desorption is delayed due to its high uptake and nonlinear isotherm. SO2 concentration initially drops quickly but is then maintained for a long time due to its great capacity and adsorption bonding. However, a NO desorption roll-up spike was observed, which may be caused by the complex interactions of SO2/O2/H2O/NO/NO2. From Figure. 10, when combining the NO and O2 desorption together, it is found that NO concentration sudden increases - this may be because of the sudden drop of O2 concentration, which greatly reduces the oxidization of NO to NO2, however, more detailed experiments need to be done to investigate this mechanism. For CDX alumina, a combination of alumina and NaY (more hydrophobic than 13X), similar breakthrough roll-up effects for SO2 and O2 were observed. During desorption, NO roll-up effect was also observed. Overall, SO2/NO can be adsorbed and desorbed under ambient conditions though the partial desorption of SO2 is very slow.

450 400 - -

Time, second Time, second

Figure 9. 13X mixture gas breakthrough curve

0 10000 20000 30000 40000 0 10000 20000 30000 40000

Time, second Time, second

Fig. 10. 13X desorption curve

400 Time,second

n, o 12

Hz e 8

400 Time,second

Fig. 11. CDX Alumina breakthrough curve

4. Conclusions

This study confirms the applicability of multiple layer columns to tackle flue gases with CO2/H2O/O2/N2/SO2/NO.

Through continuous full cycle experiments and simulation, it is found that the water front can be managed by using a combination layer of alumina and 13X, and reasonably good performance was observed at steady state. Single layered adsorbents, such as 13X, can not stabilize the water front under high humidity feed conditions. Vacuum level is also found be one of the most important factors in controlling the counter current flow and equilibrium desorption of water. A vacuum level lower than 3kPa.a is preferred to promote higher CO2 purity and recovery. The movement and position of the water front was inferred by examination of the thermal profiles in the column during operation. It was seen that wet product gas purge has little effect on the purge process as the humidity level for this stream is low. With proper management of purge ratio, good product purity and recovery are achievable.

In SO2/NO mixture gas breakthrough experiments, it was found that H2O produces roll-up effects for SO2 and O2 for both 13X and CDX Alumina. Such effects may also contribute to the quick breakthrough of SO2 in the adsorbents studied. NO desorption is closely related with O2 concentration in the gas mixture. Also it is found that SO2 desorption is slow due to its great capacity and chemical adsorption and NO can be desorbed fairly quickly. However, more fundamental work needs to be done to better understand the mechanism of those interactions and more cycle work needs to be done to test the long term running effects of those gases on process performances and adsorbent structures.

Acknowledgement

The authors thank Cooperative Research Centre for Greenhouse Gas Technologies (CO2CRC) and Monash University, Australia for the financial support of the project.

Reference:

[1] N. Stern, Stern Review on the Economics of Climate Change; Cambridge University Press: Cambridge, 2006.

[2] D. Gielen, The future role of CO2 Capture and Storage Results of the IEA-ETP Model; International Energy Agency: Paris, 2003.

[3] IEA, The Prospects for CO2 Capture and Storage; International Energy Agency: Paris, 2004.

[4] R. Steeneveldt, B. Berger, T.A. Torp, CO2 capture and storage closing the knowing-doing gap, Chemcial Engineering Research and Design,84(A9)(2006) 739-763.

[5] J. Davison, Performance and costs of power plants with capture and storage of CO2, Energy,32(2006) 1163-1176

[6] T. Yokoyama, Japanese R&D on Large-Scale CO2 Capture, ECI Conference on Separation Technology, 2004, Australia

[7] J. Zhang, Jun; P.A. Webley, Cycle Development and Design for CO2 Capture from Flue Gas by Vacuum Swing Adsorption. Environmental Science & Technology, 42(2) (2008) 563-569.

[8] J. Zhang, Jun; P.A. Webley,P. Xiao, Effect of process parameters on power requirements of vacuum swing adsorption technology for CO2 capture from flue gas. Energy Conversion and Management, 49(2)(2008) 346-356.

[9] M.K.Ram Reddy, Z.P. Xu, G.Q.(Max) Lu, J.C.Diniz da Costa, Effect of SOx adsorption on layered doule hydroxides for CO2 capture, Ind. Eng. Chem. Res., 47(2008) 7357-7360.

[10] A. Sultana, D. D. Habermacher, C. E. A. Kirschhock, J.A. Martens, Adsortive separation of NOx in presence of SOx from gas mixtures simulating lean burn engine exhaust by pressure swing process on Na-Y zeolite, Applied Catalysis B: Environmental 48(2004) 65 -76.

[11] Y-W. Lee, H-J. Kim, J-W. Park, B-U. Choi, D-K. Choi, J-W. Park, Adsorption and reaction behavior for the simutaneous adsorption of NO-NO2 and SO2 on activated carbon impregnated with KOH, Carbon, 41(2003) 1881 -1888

[12] N. Oka, J. Izumi, Japan Patent No. P2003-53132, 2003

[13] S. A. Rouf,; M. Eic, Adsorption of SO2 from wet mixtures on hydrophobic zeolites. Adsorption, 4(1)(1998) 25-33.

[14 M. Mello, M. Eic, Adsorption of sulphur dioxide from pseudo binary mixtures on hydrophobic zeolites: modelling of breakthrough curves, Adsorption, 8(2002) 279-289.

[15] I. Perdana, D. Creaser, O. Ohrman, J. Hedlund, A comparison of NOx adsorption on Na, H and BaZSM-5 films, Applied Catalysis B: Environmental, 72(2007) 82 -91

[16] S. U. Rege, R.T.Yang, K. Qian, M. A. Buzanowski, Air-prepurifiction by pressure swing adsorption using single/layered beds, Chemical Engineering Science,56(2001) 2745-2759.

[17] S. Cavenati, C. A. Grande, A. E. Rodrigues, Separation of CH4/CO2/N2 mixtures by layered pressure swing adsorption for upgrade of natural gas, Chemical Engineering Science, 61(2006) 3893 -3906

[18] G. Li, P. Xiao, P.A. Webley, J. Zhang, R. Singh, M.Marshall, Capture of CO2 from high humidity flue gas by vacuum swing adsorption by zeolite 13X. Adsorption 14, 415, (2008)

[19] Simon J. Wilson,; Chris C. K.Beh, P.A. Webley, R.S. Todd, The Effects of a Readily Adsorbed Trace Component (Water) in a Bulk Separation PSA Process: The Case of Oxygen VSA. Industrial & Engineering Chemistry Research , 40(12) (2001) 2702-2713.

[20] Todd, Richard S.; He, Jianming; Webley, Paul A.; Beh, Christopher; Wilson, Simon; Lloyd, Matthew A. Fast Finite-Volume Method for PSA/VSA Cycle Simulation-Experimental Validation. Industrial & Engineering Chemistry Research , 40(14) (2001) 3217-3224.