Patent Application
20090071333
Recent advances in process and adsorbent technology have enabled traditional large-scale pressure swing adsorption (PSA) systems to be scaled down to much smaller systems that operate in rapid cycles of very short duration. These small, rapid-cycle PSA systems may be utilized, for example, in portable medical oxygen concentrators that recover oxygen from ambient air. As the market for these concentrators grows, there is an incentive to develop increasingly smaller, lighter, and more portable units for the benefit of patients on oxygen therapy.
The impact of feed gas impurities on the adsorbent is a generic problem in many PSA systems, and the impact is especially serious in the small adsorbent beds required in small rapid-cycle PSA systems. For example, the water and carbon dioxide impurities in air can cause a significant decline in the performance of small PSA air separation systems by progressive deactivation of the adsorbent due to adsorbed impurities that are incompletely removed during regeneration steps of the PSA cycle. Because of this progressive deactivation, oxygen recovery will decline over time and adsorbent replacement may be required on a regular basis. Alternatively, the adsorbent beds may have to be oversized to account for progressive adsorbent deactivation. Both of these situations are undesirable because they increase the cost and weight of the oxygen concentrator system.
There is a need in the art for improved methods to remove impurities, particularly water, in the design and operation of small, portable, rapid-cycle PSA oxygen concentrators. This need is addressed by the embodiments of the invention described below and defined by the claims that follow.
A first embodiment of the invention includes a pressure swing adsorption process for the production of oxygen comprising
The adsorbent material in the first layer may comprise activated alumina; the activated alumina may have an average particle diameter between about 0.3 mm and about 0.7 mm. The adsorbent material in the second layer may be selective for the adsorption of argon from a mixture comprising argon and oxygen. The concentration of oxygen in the product gas withdrawn from the product end of the adsorber vessel may be at least 85 volume %. The pressurized feed gas may be air.
The depth of the first layer may be between about 10% and about 40% of the total depth of the first and second layers, and the depth of the first layer may be between about 0.7 and about 13 cm. The adsorber vessel may be cylindrical and the ratio of the total depth of the first and second layers to the inside diameter of the adsorber vessel may be between about 1.8 and about 6.0.
The pressure swing adsorption process may be operated in a repeating cycle comprising at least a feed step wherein the pressurized feed gas is introduced into the feed end of the adsorber vessel and the product gas enriched in oxygen is withdrawn from the product end of the adsorber vessel, a depressurization step in which gas is withdrawn from the feed end of the adsorber vessel to regenerate the adsorbent material in the first and second layers, and a repressurization step in which the adsorber vessel is pressurized by introducing one or more repressurization gases into the adsorber vessel, and wherein the duration of the feed step is between about 0.75 and about 45 seconds. The total duration of the cycle may be between about 6 and about 100 seconds. The flow rate of the product gas enriched in oxygen may be between about 0.1 and about 3.0 standard liters per minute.
The ratio of the weight in grams of the adsorbent material in the first layer to the flow rate of the product gas in standard liters per minute at 93% oxygen purity in the product gas may be less than about 50 g/slpm. The amount of oxygen recovered in the product gas at 93% oxygen purity in the product gas may be greater than about 35% of the amount of oxygen in the pressurized feed gas.
The adsorbent material in the second layer may comprise one or more adsorbents selected from the group consisting of X-type zeolite, A-type zeolite, Y-type zeolite, chabazite, mordenite, and clinoptilolite. This adsorbent material may be a lithium-exchanged low silica X-type zeolite in which at least about 85% of the active site cations are lithium.
Another embodiment of the invention relates to a pressure swing adsorption process for the production of oxygen comprising
The adsorbent material in the first layer may comprise activated alumina; the activated alumina may have an average particle diameter between about 0.3 mm and about 0.7 mm. The adsorbent material in the second layer may be selective for the adsorption of argon from a mixture comprising argon and oxygen. The concentration of oxygen in the product gas withdrawn from the product end of the adsorber vessel may be at least 93 volume %. The pressurized feed gas may be air.
The depth of the first layer may be between about 10% and about 40% of the total depth of the first and second layers; the depth of the first layer may be between about 0.7 and about 13 cm. The adsorber vessel may be cylindrical and the ratio of the total depth of the first and second layers to the inside diameter of the adsorber vessel is between about 1.8 and about 6.0.
The pressure swing adsorption process may be operated in a repeating cycle comprising at least a feed step wherein the pressurized feed gas is introduced into the feed end of the adsorber vessel and the product gas enriched in oxygen is withdrawn from the product end of the adsorber vessel, a depressurization step in which gas is withdrawn from the feed end of the adsorber vessel to regenerate the adsorbent material in the first and second layers, and a repressurization step in which the adsorber vessel is pressurized by introducing one or more repressurization gases into the adsorber vessel, and wherein the duration of the feed step is between about 0.75 and about 45 seconds.
The total duration of the cycle may be between about 6 and about 100 seconds. The flow rate of the product gas enriched in oxygen may be between about 0.1 and about 3.0 standard liters per minute. The amount of oxygen recovered in the product gas at 93% oxygen purity in the product may be greater than about 35% of the amount of oxygen in the pressurized feed gas. The adsorbent material in the second layer may comprise one or more adsorbents selected from the group consisting of X-type zeolite, A-type zeolite, Y-type zeolite, chabazite, mordenite, and clinoptilolite. This adsorbent material may be a lithium-exchanged low silica X-type zeolite in which at least about 85% of the active site cations are lithium.
Recent advances in process and adsorbent technology allow the designs of traditional large-scale pressure swing adsorption (PSA) processes to be scaled down to much smaller systems. These smaller systems are especially useful in transportable devices such as, for example, medical oxygen concentrators for recovering oxygen from air. As the medical oxygen concentrator market develops, there is a growing need for smaller, lighter, and more transportable devices for the benefit of patients requiring oxygen therapy.
The zeolite adsorbents commonly used as the nitrogen-selective adsorbents in oxygen PSA systems are sensitive to contaminants present in ambient air, specifically water and carbon dioxide, with water being the most serious and controlling contaminant. The nitrogen-selective zeolite adsorbents have a high affinity for these impurities, and rapid deactivation can occur when the impurities are not adequately removed during the regeneration steps of the PSA process. Numerous techniques have been used in the art to remove these impurities from the feed gas. In single or multiple bed systems, it is common to layer adsorbents in a vessel wherein a pretreatment layer of impurity-selective adsorbent is used at the feed inlet followed by one or more layers of nitrogen-selective adsorbent. The purpose of the impurity-selective pretreatment adsorbent is to reduce or remove water and/or carbon dioxide to protect the downstream adsorbent from progressive deactivation.
The impact of impurities on the performance of the nitrogen-selective adsorbent is much greater in the small PSA systems used for portable oxygen concentrators than in larger industrial PSA systems. If the impurities are not removed properly in small PSA systems, the impurities can progress through the nitrogen adsorbent beds and cause a slow decline in the performance of the PSA system over a long period of time. Although the contaminants may be removed by the pretreatment layer during the PSA feed step, inadequate regeneration of this layer during the purge step can occur and lead to the slow deactivation of the nitrogen adsorbent. Solutions to this problem are provided by the embodiments of the invention described below.
The generic term “pressure swing adsorption” (PSA) as used herein applies to all adsorptive separation systems operating between a maximum and a minimum pressure. The maximum pressure typically is superatmospheric, and the minimum pressure may be super-atmospheric or sub-atmospheric. When the minimum pressure is sub-atmospheric and the maximum pressure is superatmospheric, the system typically is described as a pressure vacuum swing adsorption (PVSA) system. When the maximum pressure is at or below atmospheric pressure and the minimum pressure is below atmospheric pressure, the system is typically described as a vacuum swing adsorption (VSA) system.
The indefinite articles “a” and “an” as used herein mean one or more when applied to any feature in embodiments of the present invention described in the specification and claims. The use of “a” and “an” does not limit the meaning to a single feature unless such a limit is specifically stated. The definite article “the” preceding singular or plural nouns or noun phrases denotes a particular specified feature or particular specified features and may have a singular or plural connotation depending upon the context in which it is used. The adjective “any” means one, some, or all indiscriminately of whatever quantity. The term “and/or” placed between a first entity and a second entity means one of (1) the first entity, (2) the second entity, and (3) the first entity and the second entity.
Modern portable oxygen concentrators utilize PSA systems and are battery-powered to allow ambulatory patients to move about independently for reasonable periods of time without requiring connection to a power source. Light weight is critical for the successful development and use of these oxygen concentrators, and important design factors to achieve this include advanced adsorbent materials, small scale compressor technology, improved battery chemistry, lightweight materials of construction, new valve technology, scaled-down electronic components, and improved conserver devices. In addition, the proper choice of PSA cycles and adsorbents can significantly improve oxygen recovery, thereby reducing the weight of the adsorbent and the batteries required to operate the system.
For any PSA process, recovery improvements can be realized by utilizing a rapid cycles with adsorbent materials having favorable adsorption capacity and kinetic properties. In rapid cycle processes, adsorption kinetics is an important factor in reducing the size of adsorbent beds. As described above, an adsorbent bed may comprise a pretreatment zone in which feed contaminants of varying concentrations are removed and a main adsorbent zone which the main separation takes place. In PSA oxygen concentrators, the contaminants typically include water, CO2, amines, sulfur oxides, nitrogen oxides, and trace hydrocarbons. The main separation is effected by adsorbing nitrogen on a nitrogen-selective adsorbent.
Because nitrogen-selective adsorbents have a high adsorption affinity for these contaminants, the adsorbed contaminants are difficult to remove once they are adsorbed. This adversely impacts the efficiency of the nitrogen/oxygen separation in an oxygen PSA system in which contaminants are removed by a pretreatment adsorbent that is regenerated by purging. The embodiments of the present invention are directed towards reducing the quantity of adsorbent in the pretreatment layer while maintaining the performance of the nitrogen-selective adsorbent under varied ambient operating conditions. The importance of proper feed gas pretreatment is illustrated in
Water vapor is the critical feed contaminant in PSA systems for recovering oxygen from ambient air. Nitrogen-selective adsorbents such as X-type zeolites and low silica zeolites containing lithium strongly adsorb water and require high activation energy to remove adsorbed water in regeneration. Water contamination on zeolites used in PSA air separation causes significant reduction in the nitrogen capacity as seen in
A key parameter used to describe the operation of a PSA system is the superficial contact time of the gas in the adsorbent bed. This parameter is defined as
where L is the bed length and vo is the superficial velocity of the feed gas through the bed based on the empty bed volume. The superficial contact time may be defined for all adsorbent in the bed including a pretreatment layer, or alternatively may be defined for the pretreatment layer only. A minimum superficial contact time is required to select an adsorbent for contaminant removal.
Under typical ambient conditions (for example, 10-20% relative humidity in the ambient air feed), operating a zeolite bed without a pretreatment adsorbent in an oxygen PSA system will result in a noticeable decline in system performance in a short period of time. This was illustrated in an experiment carried out with a single-bed oxygen PVSA system using a full bed of a nitrogen-selective LiX adsorbent without a pretreatment layer. A single bed of UOP Oxysiv-MDX adsorbent was cycled in a four-step process (feed repressurization, feed/make product, evacuation, purge). The bed ID was 0.88 inch, the bed height was 2.47 inch, the total cycle time was 19 seconds, and the product rate was 43-48 sccm with a bed feed superficial velocity of about 0.38 ft sec−1. The results of this experiment are given in
Process conditions for a typical portable oxygen concentrator design may include cycle differential pressures between about 0.4 atma and about 1.7 atma in PVSA and about 1 atma and about 6 atma in PSA processes. To achieve an oxygen recovery of 65% (i.e., the percentage of oxygen in the feed gas recovered as product), a feed flow rate in the range of about 2 slpm to about 40 slpm is required for the production of 0.25 to 5.0 slpm of 93% purity oxygen. The operating temperature of the oxygen concentrator typically is ˜70° F., but can range from 0° F. to 100° F. depending on the location of the concentrator. Altitude can range from sea level to 6000 ft above sea level. Standard conditions are defined as 21.1° C. and 1 atm.
For effective contaminant handling in the adsorber beds, a pretreatment adsorbent with favorable equilibrium properties and mass transfer properties is required. Various adsorbents are available to perform the task of reducing or removing the feed contaminants.
The adsorption kinetics of the pretreatment adsorbent and the nitrogen-selective zeolite can be quantified by a mass transfer coefficient, ki, where k is the rate constant for sorbate i using an appropriate mass transfer model. This parameter can be determined by fitting experimental breakthrough or cycle data. Fitting cycle data accounts for a complete combination of all mechanisms of mass transfer resistance which are present in the actual process, and a more accurate model of the process kinetics is determined from mass transfer parameters obtained from cyclic data.
An experimental single-bed PSA apparatus was constructed for evaluating the mass transfer parameter for water adsorption on a given adsorbent. The apparatus was capable of experimental process operation in which the bed pressures and feed flow rates can be varied. To determine a representative mass transfer coefficient, k, the apparatus was operated at selected pressures and feed velocities to match those of an actual or planned full-scale process.
A standard test procedure was used to evaluate the mass transfer characteristics of an adsorbent. The bed pressure was cycled from about 0.3 atm to about 1.2 atm, the oxygen product purity was maintained at 93%, and the feed and evacuation gas superficial velocities were about 0.39 ft sec−1. It was necessary to change the cycle times slightly and to change the product flow rates to achieve these targets. The feed gas humidity, pressure, temperature, and flow rates were determined by direct measurement. The product flow rate and concentration were measured at cyclic steady state. Using all of the collected process data, a computer simulator model was developed to determine the mass transfer coefficient, k, for the tested adsorbent. This computer model, SIMPAC, is a process simulator which solves energy, mass, and momentum balances for a cycle having one or more adsorbent beds and a multicomponent feed gas. The process simulator can utilize a range of mass transfer and equilibrium models. The use and validation of SIMPAC is described in U.S. Pat. No. 5,258,060, which is incorporated herein by reference. In the selected mass transfer model, k is the rate constant from the well-known linear driving force model with partial pressure driving force:
Where
Single component isotherms were used to describe the equilibrium properties, axial dispersion was determined to be negligible, and a natural convection heat-transfer model was used in the non-isothermal energy balance. In determining the mass transfer behavior of water adsorption on the identified materials, a bed having two adsorbent layers was used. The first layer adsorbs only water and carbon dioxide, while the second layer has affinity for all of the components in the feed gas. The second layer is a well-characterized material for which all of the pure component isotherms and the mass transfer coefficients are known. In addition to the cyclic experiments, the materials were removed from the adsorbent columns in well-maintained sections after the experiments were complete and were analyzed for water content by thermogravimetric analysis (TGA) or preferably thermogravimetric analysis with infrared detection (TGA-IR) of the desorbing gas. A profile of the adsorbed water was obtained from this direct measurement and was matched to the computer simulation results. The k parameter was therefore determined.
Alcan AA-300 and AA-400 and UOP aluminas were screened to various particle sizes and tested using the procedure described above. Bed heights were between 2.4 and 3.2 inch, and inside bed diameters were 0.88 inch. The pretreatment bed height was 1 cm and feed linear velocities were about 0.4 ft sec−1. As described above, mass transfer parameters determined for these materials are shown in Table 1.
The single bed experiments were extended to determine the overall effect of the pretreatment kinetic parameter on key properties of the process. Table 2 illustrates the impact of the pretreatment kinetics on the overall recovery and bed size factor (BSF). Adsorbents used in the main portion of the adsorbent bed are UOP Oxysiv MDX, UOP Oxysiv-7 and pilot scale LiLSX materials. This comparison of performance in systems having the same main bed adsorbent shows distinguishable differences where a pretreatment material having high k values are used. For example, we can compare case 1 with case 7 where the same Oxysiv-MDX is used and the bed split is 30/70. By using a pretreatment material having a larger kvalue (200 sec−1 versus 30 sec−1), the recovery improves from 29% to 45% and the bed size factor in case 7 is 73% of that in case 1.
The mass transfer properties of the pretreatment adsorbent were also used to predict the performance of a four-bed process previously described in patent application EP1598103A2 where cycle times were 6.0-8.0 seconds and individual step times were 0.75 to 1.0 seconds. This four bed process was run both in simulation and experimentally to illustrate the previously unrecognized relationship between the contaminant kinetics in the pretreatment layer and the overall product recovery and bed size factor in a portable system. Table 3 summarizes these experimental results.
In the fast cycle process, the amount of water removed in the pretreatment layer strongly influences the effectiveness of the nitrogen removal since part of the main bed adsorbent becomes irreversibly contaminated. Minimizing this main bed contamination is important in maintaining the desired performance. As stated earlier, both capacity and adsorption kinetics are important in the removal of water from the feed gas. The pretreatment adsorbent must have a fairly low activation energy (heat of adsorption) and high adsorption kinetics. Since the heat of adsorption for water on any adsorbent is not negligible, the thermal profile within the adsorbent bed becomes a contributing factor in the effectiveness of the contaminant removal and regeneration. In systems where water has a low heat of desorption relative to the nitrogen selective adsorbent in the main adsorbent bed, it is beneficial to run the system at near-isothermal conditions.
While no process can be run as purely isothermal, a system at near-isothermal conditions is defined as a system where there is a high degree of heat transfer from the adsorption process to the ambient surroundings. As shown in prior art, for various reasons a temperature effect described as a “cold zone” is observed near the interface of layered beds where the temperature profile of the beds dips very low relative to the feed inlet temperature. With improved heat transfer, this temperature dip can be minimized. For example, the degree of heat transfer from the adsorbent bed to the column wall is described by a single heat transfer parameter, hw, where it is shown that higher values of hw yield narrower bed temperature profiles. A large drop in bed temperature causes a higher energy requirement for regeneration of the zone where the “dip” occurs. In small portable adsorption systems, increased vacuum energy is costly in the form of increased compressor capacity and hence higher power and weight.
A solution to this problem is to use a layered adsorbent bed wherein the energy required to regenerate the pretreatment adsorbent is minimized and wherein the heats of adsorption and regeneration are easily transferred from or to the adsorbent bed. The effects of this improvement are shown in
Pressure drop effects are important in selecting and optimizing a pretreatment adsorbent. Since smaller particles will have better mass transfer properties and higher k values, they are preferred in rapid cycle systems. As adsorbent particles are decreased in size, however, there are significant issues with pressure drop and handling which make particles below a certain size unfeasible in packed beds.
Simulations were made using the 4-bed process described in Example 1. Ambient conditions of 1 atm, 73° F., and 25% relative humidity were assumed. Beds of Alcan AA400G alumina pretreatment layer with highly exchanged LiLSX main bed layer were used in a 25/75 ratio (pretreatment layer/main layer). The total cycle time was 8 seconds and a heat transfer coefficient of 0.87 BTU lb−1 hr−1° F.−1 was used. The simulations were made for various values of the pretreatment adsorbent particle size and water mass transfer coefficient, kw. The value of kw was varied according to the relation
where the effective diffusivity, Deff, was assumed to be constant for all particle sizes. Specific adiabatic power was determined for each case for comparison.
The results are presented in
A single bed experiment was run using a 4-step process analogous the process described above. The adsorbent column was loaded with LiLSX having an average particle diameter of 0.8 mm and an Alcoa AL H152 pretreatment adsorbent with an average particle diameter of 2.0 mm. The cycle time was varied from 85-105 seconds with feed time varied between 25 and 45 seconds. The feed linear velocity ranged from 0.2 to 0.4 ft/sec. The adsorbent column length was 17 inches and 30% of the total length was the pretreatment layer. Oxygen product purity was 90% and remained steady for about 300 hours before the experiment was completed. The column heat transfer coefficient (HTC) was about 0.15 BTU lb−1 hr−1° F.−1.
The experiments and Examples presented above illustrate the operation of a fast cycle PSA process in which each adsorber vessel has a first layer of adsorbent material at the feed end to remove water from a feed gas that contains at least water, nitrogen, and oxygen. A second layer of adsorbent material is used to preferentially adsorb nitrogen from the dried feed gas to provide the oxygen product. Based on the results of these experiments and Examples, it was observed that the most efficient PSA performance may be obtained for certain combinations of a physical property of the adsorbent material in the first layer, i.e., the heat of adsorption of water on the adsorbent material in a certain range of adsorbed water loadings, and a PSA operating parameter relative to the first layer, i.e., the mass transfer coefficient of water in the first layer of adsorbent material.
The selected heat of adsorption of water on the adsorbent material in the first layer is equal to or less than about 14 kcal/mole at loadings less than about 0.05 mmol adsorbed water per gram of adsorbent, and the selected mass transfer coefficient of water in the first layer of adsorbent is in the range of about 125 to about 400 sec−1. The parameters heat of adsorption vs. loading for water-selective adsorbents are plotted in
This application is a Continuation-in-Part of U.S. Ser. No. 11/542,948 that was filed on Oct. 4, 2006 and which is wholly incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 11542948 | Oct 2006 | US |
| Child | 12275752 | US |
