Hydrothermal damaging of molecular sieve and how to prevent it

Transcription

1 Hydrothermal damaging of molecular sieve and how to prevent it Peter B. Chr. Meyer Natural Gas Marketing Manager CECA / ATOFINA Paris La Défense France Paper presented at the Gas Processors Association Europe February 2003, Paris

2 Hydrothermal damaging of molecular sieve and how to prevent it Peter B. Chr. Meyer Natural Gas Marketing Manager CECA / ATOFINA Paris La Défense France ABSTRACT Natural gas treating units utilising molecular sieve technologies are usually optimised relative to the available adsorption time and the required regeneration time. The total cycle time is usually such that at the end of the adsorption a limited time is available for adequate regeneration of the adsorbent. This leads in many cases to the section of bed at the inlet of the adsorber being subject immediately to high temperatures from the start of the regeneration without any heating ramp. Water condensation can occur in certain layers of the molecular sieve which are not heated up sufficiently causing boiling of molecular sieves in liquid water at high temperature. The consequence might be attack of the binder (dust formation) and hydrothermal damaging (loss of crystalline structure and pore closure effect of the molecular sieves) which leads to a poor adsorption behaviour and a short life time. To prevent this kind of damaging a good molecular sieve formulation (binder and zeolite) is necessary. Further improvements can be done by optimising the regeneration conditions. This article discusses with an example of an existing unit how to prevent the hydrothermal damaging by changing the regeneration conditions based on simulations of the unit with the proprietary simulation program SIMATEP. Not to be copied without written authorization of 1

3 Hydrothermal damaging of molecular sieve and how to prevent it Introduction This article discusses how to make molecular sieve last longer by optimising the design package: product and operating conditions. In a previous paper presented on the GPA Europe conference in February 2002 [1] different causes for damaging of molecular sieves leading to a shorter life time of the adsorbent are discussed. Except the problem of different kind of liquid carry-over (Amines, glycols, caustic soda ) and in consequence the chemical attack and/or the fouling of the molecular sieves, or problems due to non perfect molecular sieve formulation (zeolite + binder) the operating conditions, especially the regeneration conditions, may cause severe damages. Especially the hydrothermal damaging of zeolites is discussed in this paper and how to prevent it. 1) Natural gas treating units : regeneration conditions in general Natural gas treating units utilising molecular sieve technologies are usually optimised relative to the available adsorption time and the required regeneration time. Compared to other MS units Natural Gas treating units (drying or purification like H2S, CO2 or mercaptan removal) have a relatively high flow rate of gas to be treated per adsorber. The adsorbent bed has usually a very huge diameter and a bed height of usually not more than 2-3 times the diameter. Thus, in order to limit the pressure drop to a reasonable value helping to economize compressor power and energy consumption. The adsorption is usually done from top to bottom, the regeneration in the inverse direction. The adsorption time is minimized to reduce the vessel size and the amount of adsorbent used. One of the design limits is the time available for an adequate regeneration of the adsorbent. The time is determined by the energy needed for the regeneration which is brought in the system by the flow rate and the temperature of the regeneration gas. Energy is needed to heat up the vessel, the adsorbent, the support material and to remove the adsorbed molecules. Depending on the type of molecular sieve and the regeneration pressure, the inlet regeneration temperature can vary between 200 C and 300 C (392 F to 572 F). Very often the regeneration gas flow rate should be minimized as the regeneration gas is recycled to the feed of the adsorber(s) and in consequence can play an important role for the design of the bed (quantity of adsorbent needed). As the regeneration gas is usually cooled down only with an air cooler the water content can be significantly higher than the water content of the feed gas. The optimisation of the energy to be put in is an optimisation between flow rate and temperature depending on the physical and chemical resistance of the molecular sieve. While a temperature of Not to be copied without written authorization of 2

4 300 C (572 F) for a regeneration at high pressure (far above 30 bar, 435 psi) could be possible for a 4A molecular sieve, the life time of a 3A molecular sieve would be shortened extremely using such a temperature. In this case, the choice would be rather to use a higher flow rate and a lower temperature. If the available regeneration time is very short, there is no time to heat up the adsorbent by heating up the regeneration gas with a ramp. The regeneration gas will flow in the adsorber almost immediately with the maximum regeneration temperature. 2) Hydrothermal damaging of molecular sieve [2] Heating up the adsorber without using a heating ramp leads to a strong temperature difference in the vessel. At the bottom, the molecular sieve will be very hot and will desorb rapidly the adsorbed water while the layers at the top of the adsorber will be still at adsorption temperature. The water desorbed in the bottom layer will condense in the top layer. This phenomena is called refluxing or retro condensation. Retro condensation in this paper does not mean condensation of hydrocarbons what may happen for associated gas at hydrocarbon dew point. The heating going on will heat up the liquid water and boil the molecular sieves in liquid water. Hydrothermal damaging will appear in consequence. Liquid water means in this article the appearance of condensed water due to oversaturation of the gas. The binder of the zeolite might be attacked or weakened leading to dust formation or formation of agglomerates. Usually the binder is more sensitive to chemical attacks like Amine carry-over for example. How to make molecular sieves stand this type of attack is described in previous articles [3]. Hydrothermal damaging is different depending on the type of molecular sieve: - X type molecular sieves (10 A or 13X): a loss of crystalline structure occurs with a roughly proportional loss in adsorption capacity - A type molecular sieve (3A, 4A, 5A): the external crystal surface is attacked resulting in a pore closure effect that affects the kinetics of adsorption. Nevertheless, the substantial adsorption capacity persists. In general, 3 A molecular sieves (potassium A type) are more sensitive to hydrothermal damaging than 4A (sodium type) and 4A type more than 5A (calcium A type). The different behavior of A type zeolites, compared to X zeolites, can be demonstrated in a laboratory with a simple test: an A type zeolite is hydrothermally degraded by steaming. By abrasion the outer layer of crystals is removed (the layer that is supposed to be affected by "pore closure"). Tests then show the zeolites have recovered their original adsorption kinetics. Not to be copied without written authorization of 3

5 The higher the temperature of regeneration, the heavier the damaging of the molecular sieves. In an industrial unit it is important, too, to limit the quantity of water appearing in liquid phase (condensing water due to oversaturation of the gas phase) as this decreases the teperature where hydrothermal destruction may occur with the water acting as stabilizer for intermediates formed by the dissolution of the zeolites. [4] The SIMATEP simulation program determines for a given unit the quantity of liquid water, the section of the molecular sieves and the temperature when it appears for a given regeneration procedure. 3) SIMATEP Simulation program Based on an approach of heat and mass transfer using different types of adsorption isotherms this program has been developed in the R&D center of the TotalFinaElf-group in collaboration with French engineering schools and universities. The program simulates the conditions at different steps of the adsorption and regeneration process. Information like temperature and water profile in the bed, appearance of free water can be achieved. The following case study gives an example how to optimise a unit (prevent damaging of molecular sieves) and which results may be obtained by SIMATEP in order to lengthen the lifetime of molecular sieves. 4) CASE Study: Natural Gas Drying unit (Middle East) Description of the unit: The unit consists of three adsorbers, two adsorbers in adsorption and one in regeneration. Each vessel contains 35 t (approx lbs) of 4A molecular sieve. Flow rate per vessel Nm3/h (approx. 210 MMSCFD) Temperature of 19 C (67 F) and 32 bar (464 psi), saturated with water Adsorption time 32 hours Regeneration conditions (initial): flow rate Nm3/h (approx. 29 MMSCFD), pressure 32 bar, maximum heating temperature 265 C (510 F). Heating time 450 minutes, cooling time approx 100 minutes. No intermediate heating step, almost no heating ramp (determined by the heater). There is almost no stand-by time during the regeneration (remaining time used for valve operation). Not to be copied without written authorization of 4

6 The plant people noted molecular sieve damaging and a short lifetime of the installed molecular sieves. Analysis of the installed molecular sieve showed hydrothermal destruction of the upper layers of the adsorbers. For replacement of the molecular sieves by CECA SILIPORITE molecular sieves the plant was simulated with SIMATEP and different regeneration procedure where proposed and tested. The simulation of the existing regeneration procedure gives the following results for the appearance of water. The diagrams show the temperature in the adsorber and the quantity of water per bed volume appearing at different bed length in function of the elapsed regeneration time. Diagram 1: Original regeneration procedure, temperature profiles Bed Temperature profiles 200 T C bed length elapsed time min Not to be copied without written authorization of 5

7 Diagram 2: Original regeneration procedure, retro condensed water 1 Retrocondensed Water in the bed during heating moles of water/m3 ms bed 1.2 bed length elapsed time The liquid water phenomena lasts from approximately the 20 th till to the 90 th minute. In order to compare the different regeneration procedures and the improvement to prevent hydrothermal destruction the time where the maximum peak of water appears was chosen. The diagram below shows the temperature profile in the adsorber and the maximum quantity of liquid water in the adsorber in function of the bed height. Not to be copied without written authorization of 6

8 Diagram 3: Original regeneration procedure, maximum water with temperature profile Instantaneous condensed water in the bed - minute 68 moles of liquid water/m3 of MS Liquid water Temperature bed length (m) Temperature C The instantaneous maximum water appears approximately at the 68 th min of the regeneration time and the zone where the liquid water appears is at a temperature between 170 and 200 C. The level of this temperature is very important. An improvement of the regeneration procedure would be to prevent the contact of a large amount of water at a high temperature with the molecular sieve. A regeneration procedure with an intermediate heating step and the same maximum temperature and same regeneration flow rate was proposed. Heating plateau at 130 C (266 F) for 90 minutes before heating up to 265 C. Total heating time 510 minutes. The temperature profiles are the following: Not to be copied without written authorization of 7

9 Diagram 4: Intermediate regeneration procedure, temperature profile 300 Bed Temperature ramp up T C Série7 Série6 Série4 bed length elapsed time min Diagram 5: Intermediate regeneration procedure, retro condensed water Retrocondensed w ater in the bed during heating bed length moles of water/m3 ms bed elapsed time The liquid water phenomena lasts from approximately the 25 th to 150 th minutes of the heating time but the maximum instantaneous water peak is lower (6 moles/m3 versus 0.9 moles/m3) and the temperature where it appears is lower, too ( C). This is shown on the diagram 6 below. Not to be copied without written authorization of 8

10 Diagram 6: Intermediate regeneration procedure, maximum water with temperature profile 0.9 Instantaneous condensed water in the bed - m inute Liquid water Tem perature The final regeneration procedure proposed was the following one: flow rate of Nm3/h (increase of 18% versus original one), maximum heating temperature 250 C (decrease of 6% versus original one) and a heating ramp at 130 C during 90 minutes. Total heating time 510 minutes. The corresponding diagrams are attached below. Diagram 7: Final regeneration procedure, retro condensed water Retrocondensed water in the bed during heating moles of water/m3 ms bed bed length elapsed time Not to be copied without written authorization of 9

11 Diagram 8: Final regeneration procedure, maximum water with temperature profile Istantaneous condensed water in the bed - minute 102 moles of liquid water/m3 of MS Liquid water Temperature bed length (m) Temperature C The liquid water phenomenon lasts from approximately the 25 th to 125 th minute of the regeneration time. The instantaneous water peak is lower again, 0.53 versus 6/0.9. The temperature where the liquid water appears is approximately C versus C/ C. Results after the change of the regeneration conditions: Peak water retro-condensation has been reduced by 42% Temperature range when water retro-condensation peak occurs has been reduced by 42% Water retro-condensation lasts longer but quantity and temperature range (when the phenomena occur) have been significantly reduced Due to higher gas velocity most of the droplets will be carried out of the bed through higher flow rate more energy is given to the system in the same time and therefore either soak temperature or soaking time can be reduced (less thermal stress to the MS for the first option and less overall regeneration time for the second option) The customer recently replaced the installed molecular sieve again by CECA SILIPORITE molecular sieve after a life time of 4 years instead of 2-3 years. Not to be copied without written authorization of 10

12 5) Conclusion In order to prevent hydrothermal damaging of molecular sieves it is not only important to choose the right formulation of the molecular sieve (binder and zeolite) but the operating conditions especially the regeneration conditions should be carefully determined. The proprietary product SIMATEP of CECA allows these detailed studies. 6) Acknowledgments I would like to thank you all my colleagues from CECA / ATOFINA for the critical review of this paper and their help putting it together. 7) References cited 1. Archiaston Musamma / Sutopo, Eliminate the molecular sieve problem and extent the life time to nine years, GPA Europe conference, London Feb 21 st, Guido Dona, customer information, Hydrothermal Stability of Molecular Sieves, 3. R. Le Bec, R. Voirin, D. Plee, S. Brunello, "New type of Molecular Sieve With longer life for Natural Gas Drying", B 3 presentation at LNG 12, 4-7 May 1998, Perth (Australia). 4. M. Suckow, W. Lutz, J. Kornatowski, M. Rozwadowksi and M. Wark, Calculation of the hydrothermal long-term stability of zeolites in gas-desulphurization and gas-drying processes, Gas Separation & Purification, Vol 6 No.2, , Not to be copied without written authorization of 11