Patent Application
20210077942
The present invention relates to a process to respond to feed flow variations by changing the process cycle and thereby increasing the productivity and capacity of the system significantly over constant process systems.
Typically pressure swing adsorption (PSA) systems have an optimal design condition that is the peak performance achievable for the system. Under steady conditions, this design is acceptable most of the time. Occasionally variances occur and processes to handle the variances range from restricting the flow, to reducing the number of beds (effective physical size) of the system. Other methodologies have been suggested to address the issue of variable feed flow, composition, and temperature for PSA processes. Traditionally these methodologies are targeted toward bringing the feed stream within optimal operating parameters for the system.
U.S. Pat. No. 5,258,056. describes a turndown methodology to produce substantially less product in response to declining customer product demand. This is done by reducing the number of beds online and by taking substantially less feed flow.
U.S. Pat. No. 7,641,716 describes a throttling methodology to maintain a constant feed. This consists of valves located before the system to keep the flow rate at the optimal rate to achieve peak performance for the system.
U.S. Pat. No. 6,030,435 describes regulating the feed flow temperature in order to keep the temperature of the system at the optimal temperature for peak performance of the PSA process.
All these methodologies involve changing the feed stream rather than changing the process. The present invention offers a different approach for regulating pressure swing adsorption (PSA) systems by changing the process cycle and thereby increasing the productivity and capacity of the system significantly over constant process systems.
The present invention generally relates to a process for responding to feed flow variations by changing the process cycle and thereby increasing the productivity and capacity of the system significantly over constant process systems. This increases the flexibility a PSA system for customers that do not require a constant or uniform product flow rate and/or for processes and applications that experience feed streams that vary in flow, temperature, and/or composition.
The legend for
The present invention relates to a control method to respond to feed flow variations by changing adoption a new process cycle and thereby increasing the productivity and capacity of the system significantly over constant process systems.
There are two specific cases presented as to why this is necessary and the benefits that it imparts. This first is control of low kinetic difference systems. In these systems adsorbent rate selectivity is typically low (less than 100). As a consequence, the timing for adsorption during a cycle has a very narrow window that is sufficient to adsorb the contaminant, but not substantially adsorb the product. This is an issue because the state of the art all teaches that process cycles can be altered with cycle timing in order to respond to changing feed conditions (flow, pressure, temperature, composition, etc.). If the timing of a process cycle for one of these low selectivity kinetic processes (LSKPs) is increased, the adsorption of the product increases and the recovery does not increase as is taught in the prior art. Additionally if the adsorption time is shortened, the amount of contaminant adsorbed decreases and the amount of feed stream that can be processed while maintaining product purity decreases. Since almost all feed streams have variations, controlling LSKPs becomes critical to having a viable commercial system.
In order to control LSKPs different cycles are used to handle different flow conditions. The cycles are typically chosen to have the best performance over a feed flow regime and are used to handle the feed flow variations. The design point would be the cycle that is chosen to best suit the application based on highest recovery and lowest capital (which is synonymous with highest feed flow potential). In the state of the art, these considerations would be accounted for, the optimal cycle would be chosen, and that cycle would be used for the life of the system. Here it is demonstrated that the optimal cycle can be changed to accommodate expected or unexpected variations in the feed stream, leading to a more flexible system and ability to design a system for multiple feed stream conditions.
Another problem that is extremely similar is for typical PSA processes. Typically, PSA processes attempt to control the feed stream and adjust it to fit the optimal design or reduce the number of beds online to meet a reduced flow (which also changes the process cycle). These cycles are usually deemed turndown modes and an excellent example would be H2PSA systems. When the systems are originally designed, the maximum flow rate and the target recovery are used to design a system to meet those objectives. This makes sense at the time because the feed stream ahead of the H2PSA is well controlled by other processes. However, if the plant wished to expand capacity, a new H2PSA system would need to be built or hardware modifications are needed for the new cycles to accommodate the additional flow as the old one cannot handle the flow according to state of the art process cycles. The present invention takes a different approach in that the cycle/process is modified in order to fit the feed stream variability. Specifically, a lower recovery cycle can be chosen to increase the total production of the system and utilize the increased feed stream capacity, without the requirement to deploy additional capital. This has substantial benefits for customers that are able to take an unregulated flow of product or are looking to increase the flow of the product. Additionally, by being able to increase processing capability by slightly lowering recovery, the system can capitalize on opportunities where flow requests exceed design conditions. These can happen during specific instances when a customer's primary supplier of hydrogen goes down and the secondary producer wishes to meet the increase in demand on their system. When flow is lower than design conditions, adding back or even increasing equalization steps allows for higher system recovery by increasing the void recovery, and thereby reduction operating costs. This methodology adds considerable economic benefit over current designs by processing up to 60% or more flow than the design condition and increasing production by as much as 25% or more over state-of-the-art process cycles.
These two applications of the method for process control by varying process cycles can be summarized as applying to systems that are poly bed, in one embodiment 4 or more, in another embodiment from 4 beds up to 25 beds, with at least one equalization header and preferably two or more equalization headers. The design basis cycle is the cycle which is used to typically run the system at the design condition (feed flow and feed composition specification). This poly bed system is then enabled by the process methodology to respond to variable feed temperatures, flow rates, and compositions beyond the typical conditions the system was designed for or could be designed for using the state or the art teachings. The trade-off is increased processing capacity for reduced recovery, which is substantially different than prior art methodologies wherein the goal is not to regulate the flow, but to adjust to the flow. The design cycle typically has at least 1 equalization step and in another embodiment 2 or more.
An example of a cycle that could be switched to from a 4-1-3-1 design cycle of
The effect of the cycle changes (4-1-2-2 and 4-2-2-1 vs 4-1-3-1) on the pressure trace is shown in
Since there are significant feed variations to accommodate when selecting changes to the existing cycles, a methodology was developed to correlate flow, pressure, temperature and composition variations in terms of a single number. The reasoning behind the generation of the single number is the adsorption isotherms being used in the process and the effect of the feed flow variations on them. Essentially the working capacity of the bed can be inferred by the use of the LRC isotherm but is equally applicable to other multicomponent isotherms that account for temperature effects as well. A logarithmic extrapolation between the inlet and outlet conditions that the bed experiences at the top of pressurization and the bottom of pressurization can be used to generate the starting points. Assuming that the end composition is always best represented by the product purity at the top pressure and bottom pressure, and that the feed is best represented by the feed inlet at the top pressure and bottom pressure, we can then solve for the working capacity of the bed at all conditions.
q is the component being evaluated
T—temperature in Kelvin of the gas and adsorbent
P—pressure in Pascal of the gas
Pq—partial pressure in Pascal of the gas q
Wq—amount of component q adsorbed
A1, A2, A3, A4, X—fitting parameters, subscripts denote which gas the parameters correspond to
m is the number of components in the feed stream
Where:
q is the component being evaluated
Wq—amount of component q adsorbed as defined by a multicomponent temperature dependent isotherm, preferably the LRC isotherm
Pf—original feed pressure
Pfn—new feed pressure
Pv—original vent pressure
Pvn—new vent pressure
Tf—original feed temperature
Tfn—new feed temperature
Tv—original vent temperature
Tvn—new vent temperature
F—original feed flow rate
Fn—new feed flow rate
If the Required Processing Power is above 1 that means that more intensified cycles are required (meaning less equalizations and more time feeding and evacuating the beds). If this Required Processing Power is below 1, that means there's more time available for higher recovery by increasing adsorption feed time or the number of beds for instance. By definition a RPP of 1 will correspond to the maximum processing power of a cycle under conditions that produce the most product at the desired purity.
For processes that contain a vacuum step, it is almost always most beneficial to have the vacuum equipment fully utilized as taught by U.S. Pat. No. 5,702,504 to Schaub et al. There then exists a minimum number of vacuum steps that is taught here which is that at least one bed is undergoing vacuum at substantially all times of the cycle (momentary isolation from valve switching could occur). Additionally, vacuum is best performed on one bed at a time and the teaching here is that the maximum number of steps for a vacuum containing PSA process is the same as the minimum which is one bed on vacuum at substantially all times.
A component that is more readily adsorbable means that it can have:
A more rigorous method for calculating these effects and the optimal process cycle for a set of feed conditions is the modeling detailed in the modeling description.
In one embodiment the invention relates to a method for maximizing product production under variable feed conditions in a PSA system adapted for separating a pressurized feed supply gas containing at least one more readily adsorbable component from at least one less readily adsorbable product gas component to produce a stream of product gas enriched with said less readily adsorbable component and a stream of offgas that is enriched in said more readily adsorbable component, wherein said PSA system comprises feed gas, product gas make step, a product pressurization step, a high pressure equalization step, product make step that overlaps with feeding the bed, at least one equalization up step and one equalization down step, and a blow down step to depressurize the bed, wherein when the Required Processing Power of said PSA system is greater than 1, the PSA process cycle is modified by making at least one of the following cycle changes:
In another embodiment, when the Required Processing Power of said PSA system is greater than 1, the PSA process cycle is modified by:
In another embodiment, when the Required Processing Power of said PSA system is less than 1, the PSA process cycle is modified by making at least one of the following cycle changes:
In another embodiment, when the Required Processing Power of said PSA system is less than 1, the PSA process cycle is modified by:
In one embodiment the product gas is methane and the more readily adsorbable component is N2 and/or CO2.
In another embodiment the product gas is helium and the more readily adsorbable component is N2 and/or CO2 and/or methane and/or other hydrocarbons.
In another embodiment the product gas is hydrogen and the more readily adsorbable component is N2 and/or CO2 and/or methane and/or other hydrocarbons.
In yet another embodiment the product gas is N2 and the more readily adsorbable component is O2.
The adsorbent beds of the invention typically contain contains zeolitic material and other optional adsorbents depending on the separation desired.
In one embodiment the adsorption bed contains adsorbent materials used in H2 PSA, the product gas is H2 and the more readily adsorbable component is selected from one or more of CO, CO2, CH4, N2, Ar, and hydrocarbon.
In another embodiment the adsorbent is selected from at least one of activated carbon, Zeolite, 5A, CaX, LiX.
In one embodiment according to the invention where PSA system comprises a 4131 design cycle, and the Required Processing Power for said system is greater than 1, the design cycle is modified to a 4122 cycle according to the following cycle chart:
In another embodiment wherein the PSA system comprises a 4122 design cycle, when the Required Processing Power for said system is greater than 1, the design cycle is changed to a 4221 cycle according to the following cycle chart:
In another embodiment where the PSA system comprises a 4221 design cycle, and the Required Processing Power for said system is less than or equal to 1, the design cycle is modified to a 4122 cycle according to the following cycle chart provided that the Required Processing Power for the 4122 cycle is also less than 1:
In another embodiment where the PSA system comprises a 4122 design cycle, when the Required Processing Power for said system is less than or equal to 1, the design cycle is modified to a 4131 cycle according to the following cycle chart provided that the Required Processing Power for the 4131 cycle is less than or equal to 1:
The pilot system is a pressure swing adsorption system that operates by exploiting the difference in adsorption capacity of an adsorbent for the gas of interest over a specific pressure range. When the vessel containing the adsorbent is pressurized, the adsorbent will selectively adsorb the contaminant from the gas stream and thus remove it from the product stream that exits through the other end of the vessel. When vessel is depressurized, the contaminant will desorb, and the adsorbent will be ready to process the feed stream again. This process is made into a semi-continuous batch process by having 1 vessel or more than 1 vessel available to process the gas at the majority of all times. With more than 1 vessel to process gas, additional options are available to further increase efficiency by retaining pressurized gas in dead volume spaces (piping or the heads of the vessels) and the process then has the ability to generate a continuous stream of product.
The conceptual process flow diagram is presented in
The pilot system employs multiple PSA vessels to achieve the desired nitrogen rejection and hydrocarbon recovery target. The current pilot PSA design consists of 4-6 vessels with process steps consisting of 1 bed on feed and 1 bed on blowdown at a time. There are 2-3 equalization steps as well as product pressurization and purge steps. The pilot system was designed to process up to 17kscfd and capable of using 1 to 4 inch diameter beds. During the initial construction of the pilot test system the bed size was selected to be 1 inch due to the adsorbent performance and with considerations of adsorbent manufacturing. The height was based on maximum available height in the container. The remaining components of the design were based on similar 6 bed PSA pilot plant already in operation. Full range control valves were used for all valves. The system was constructed entirely of stainless steel grade 316. Additionally, a pretreatment system of 304 stainless steel was designed and built as H2S compatible in order to remove all condensed liquids and sulfur before entering the PSA portion of the system.
The material used in the pilot testing was created as follows: 23.00 lbs. of zeolite 4A powder supplied by Jianlong (as 4A-D) on a dry weight basis (29.50 lbs. wet weight) was placed in a WAM MLH50 plow mixer. With the mixer agitating, 2.16 lbs of MR-2404 (a solventless silicone containing silicone resin from Dow Corning) was pumped in at rate of 0.07 lb/min. After the MR-2404 addition was completed, 9.2 lbs of water was added at a rate of 0.3 lb/min under constant stirring in the plow mixer. At the end of the water addition, plow mixing was continued for an additional 5 minutes. The plow mixed powder product labeled hereinafter “the formulation” was transferred to a tilted rotating drum mixer having internal working volume of ˜75 L and agitated therein at a speed of 24 rpm. Mixing of the formulation was continued while beads were gradually formed which had a porosity, as measured using a Micromeritics Autopore IV Hg porosimeter on the calcined product, in the 30-35% range. The beads were subjected to a screening operation to determine the yield and harvest those particles in the 8×16 U.S. mesh size range. The product beads were air dried overnight prior to calcination using a shallow tray method at temperatures up to 595° C. The shallow tray calcination method used a General Signal Company Blue-M electric oven equipped with a dry air purge. ˜500 g. dry wt. of the 8×16 U.S. mesh adsorbent was spread out in a stainless steel mesh tray to provide a thin layer. A purge of 200 SCFH of dry air was fed to the oven during calcination. The temperature was set to 90° C., followed by a 6 hour dwell time. The temperature was then increased to 200° C. gradually over the course of a 6 hour period, and further increased to 300° C. over a 2 hour period and finally increased to 595° C. over a 3 hour period and held there for 1 hour before cooling to 450° C. after which the adsorbent was removed, immediately bottled in a sealed bottle and placed in a dry nitrogen purged drybox. The calcined beads were rescreened to harvest those particles in the 8×16 U.S. mesh range.
Characterization of the modified 4A samples calcined at 595° C. was performed using a thermogravimetric method as described earlier in “ANRU TGA Testing”. The nitrogen uptake rate as performed in the test was determined to be ˜0.2 weight %/minute as measured using the TGA method disclosed herein. When the product beads in Example 1 were calcined up to 575° C., the nitrogen uptake rate as performed in the test was determined to be ˜0.7 weight %/minute as measured using the TGA method disclosed herein. Subsequently, when the product beads in Example 1 were calcined up to 555° C., the nitrogen uptake rate as performed in the test was determined to be ˜1.2 weight %/minute as measured using the TGA method disclosed herein.
Routine characterization of modified 4A samples was performed using a thermogravimetric method using a TA Instruments Q500 system installed in a glove box to minimize the impact of air leaks. Nitrogen and oxygen gases supplied to the instrument were high purity. The balance purge gas and gas 1 was nitrogen and a gas 2 corresponds to oxygen. For all experiments, a balance purge of 5 cc/minute was used and the gas directly over the sample was set to 95 cc/minute (nitrogen or oxygen). A sampling frequency of 0.5 sec/point was used for all adsorption steps. Alumina pans were used for all studies and the sample size after activation was in the range 100 to 120 mg.
The TGA method involves both an in-situ activation step followed by adsorption tests using oxygen and nitrogen at 25° C. The sample activation was performed by heating the sample under nitrogen purge at 2° C. per minute to 150° C., maintaining isothermal for 60 minutes, heating at 5° C./minute to 350° C., holding at 350° C. for 120 minutes, then cooling to 25° C. The nitrogen equilibrium capacity at atmospheric pressure and 25° C. is reported as the weight gain on cooling under nitrogen relative to the minimum weight at 350° C. (the activated sample weight). An assessment of relative rate for different samples and preparation is captured by switching from nitrogen to oxygen. A transient weight gain is observed followed by a drop attributable to oxygen uptake followed by nitrogen leaving. A corresponding switch from oxygen back to nitrogen results in a transient weight loss followed by a weight gain attributable to oxygen loss followed by nitrogen pickup. Values reported as “nitrogen uptake rate” correspond to the maximum slope observed in the nitrogen uptake portion and is equivalent also to the peak in the derivative weight with respect to time for the same step. Values are reported in weight %/minute.
The results from the breakthrough test and parameters obtained from the modeling were used with the methodology described by Mehrotra, et al. in Arithmetic Approach for Complex PSA Cycle Scheduling, Adsorption, 2010, pp. 113-126, vol. 16, Springer Science+Business Media which details the basis for modeling PSA processes. These simulations were performed using Process Builder, from PSE.
A LSKP could be designed to handle a feed flow stream from a well head during flowback after hydraulic fracturing of the well. The state-of-the-art design condition would be based on the maximum amount of value delivered by recovering the most methane available. This design would call for a 4-1-3-1 cycle that could handle 5 MMscfd at a 35% N2 feed content and a 20% N2 product content. For flow rates above 5 MMscfd and 35% N2 feed content, the extra gas would be passed to the vent. For flow rates below 5 MMscfd and or 35% N2 feed content, the product gas would contain less than 20% N2 but the product flow rate would be substantially the same.
Using the proposed methodology, for feed streams above 5 MMscfd and 35% N2 in the feed, a switch to a 4-1-2-2 cycle would enable the system to process up to 7 MMscfd and up to 45% N2 in the feed stream while producing up to 35% more product than the equivalent feed stream with the 4-1-3-1 cycle. Additionally, switching to a 4-2-2-1 cycle would allow processing up to 10 MMscfd and up to 70% N2 in the feed stream while producing up to 45% more product that the equivalent feed stream with the 4-1-3-1 cycle and venting methodology taught in the state of the art. These values are shown in table 1 as demonstrated by modeling and pilot results. Additionally, shown in table 1 is that just choosing a 4-2-2-1 cycle or a 4-1-2-2 cycle as the design basis for the system, has substantially lower recovery for the point at which the most value can be generated by the system. Thus, while the 4-1-3-1 cycle is still the best choice for the design basis for the system, it is not the only cycle that should be employed during the operation of the system.
The methodology for switching between cycles can be extrapolated from those proposed by Baksh et al. and described previously.
Table 1 shows the demonstration of the three different cycle examples (4133, 4122, 4221). The feed of 105% for the 4-1-3 cycle represents the maximum possible product production of the cycle with any feed flow, but not the highest recovery. The process is restricted because it is unable to make higher product production at the desired purity. The ability to handle higher feed flow rates while maintaining a constant product purity (20% N2 in the product) can be seen in the table with the other cycles.
As noted earlier, hydrogen PSA (H2PSA) systems can also benefit substantially from the adoption of new cycles to increase the product flow of the system, beyond the original design basis, or design basis taught in the state of the art. In this instance a 12-3-4 cycle was chosen as the design for comparison. In the event that the feed flow to the system is increased, the 12-3-4 cycle cannot handle the flow and still meet the purity target required. Initially the cycle time can be reduced for the cycle until the system limitations are met or exceeded (cycle time, bed fluidization etc.). Once this occurs, the full limit of the system is reached using state of the art methodology.
Using the proposed methodology, table 2 was constructed showing the effect of modifications to the process cycle. These effects are a demonstration of the selection process, but other factors should be considered when switching to a different cycle, such as frequency of the cycle changes and the effect on the production, as well as cycle compatibility based on the teachings of Baksh et al.
Table 2 shows increasing feed processing capability and increasing produced product at reduced overall recovery. Highest product potential is the maximum production that could be obtained by the cycle at the required product purity (99.999% H2) as additional feed gas would need to be vented. These values are given as a general approximation and should be seen as a demonstration of the overall trend, rather than exact feed flows a different cycle is used for. A copy of the model used is provided with PSE process builder software. Lowest Feed Flow Optimal (LOFF) is the lowest feed flow point at which this cycle has the highest product recovery among all the cycles tested. Highest Feed Flow Optimal (HOFF) is the highest feed flow point at which this cycle has the highest product recovery among all the cycles tested or is no longer able to produce more product at purity beyond this flow rate. H2 Recovery is the recovery of the product from the feed relative to the recovery from the 12-3-4 cycle at its HOFF.
Conventional PSA system handles variable feed composition and flow by adjusting cycle time without changing the cycle and cycle steps. Within one cycle, cycle step and sequence, such as adsorption feed, equalization, purge, provide purge, blow down are fixed. Control valves are sized accordingly. Therefore, system processing range is limited for the feed and contaminant composition. With the proposed new control method, allowing and adopting new cycles to address wider feed flow and composition provides additional operational freedom compares to conventional PSA system.
The methodology for switching between cycles can be extrapolated from those proposed by Baksh et al. as described previously.
This application claims the benefit of U.S. Provisional Application Ser. No. 62/649,798, filed on Mar. 29, 2018, which is incorporated herein by reference.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/US2019/024848 | 3/29/2020 | WO | 00 |
| Number | Date | Country | |
|---|---|---|---|
| 62649798 | Mar 2018 | US |
