Apr-2009

Modelling to optimise natural gas dehydration

Computer modelling can help processors to understand the severity of conditions in molecular sieves used for the dehydration of natural gas

Henry Rastelli and Julie Stiltner Shadden
UOP, A Honeywell Company

Article Summary

Raw natural gas contains many undesirable constituents and therefore must be treated to make it viable as a fuel source. Depending on the composition of the gas, various processing operations may be necessary to remove nitrogen (N2), carbon dioxide (CO2), sulphur species and water. Various liquid-phase processes have been developed (utilising glycols, amines and so on) to handle most of these contaminants.

However, when the gas stream is to be treated cryogenically, as in liquefied natural gas (LNG) or natural gas liquids (NGL) production, a molecular sieve-based adsorption system is typically required to reduce water and CO2 levels to the point where freeze-ups in the cryogenic plant do not occur. Zeolite molecular sieves are the preferred choice for deep dehydration and ultra-purification of natural gas, since natural gas components (methane, ethane and propane) are non-polar and weakly adsorbed in zeolites. Contaminants of natural gas are polar compounds (water, CO2 and hydrogen sulphide (H2S), for instance) and these are strongly adsorbed by zeolite molecular sieves.

Zeolite molecular sieves are ideal for gas processing applications where there is a requirement for extremely low dew points (-150°F for LNG and <-100°F for NGL plants). The sodium form of Linde Type A zeolite (also called 4A or NaA) is the typical choice for deep dehydration because the sodium cation produces a crystalline lattice that has a high selectivity and affinity with water, and the A type structure has a large open space and therefore a high capacity. Once the 4A zeolite is saturated with water, it must be thermally regenerated to restore it to its activated state so it can be used repeatedly for dehydration. This cyclic process is conducted in large fixed beds.

Although a very stable material at room temperature, 4A is susceptible to thermal and hydrothermal damage, so it must be regenerated carefully. Regeneration conditions play a significant role in the destabilisation of the 4A zeolite, referred to as the life of the zeolite. Zeolite beds are therefore designed with cycles to minimise the number of regeneration cycles in order to extend the molecular sieve’s life for as long as possible. However, as 4A ages in service, it loses capacity, resulting in shorter adsorption steps and consequently more regenerations in a given time period, which accelerates the ageing of the 4A. Eventually, the adsorbent bed must be replaced, typically within three to six years.

Since the regeneration step is so critical to the life of the molecular sieve, UOP’s development team felt it necessary to understand this process step in more detail, in the hope of minimising any detrimental effects on the adsorbent. Previously, regeneration designs were based on the amount of heat needed to strip water from the adsorbent. Usually, this heat was related to the temperature, time and mass flow rate of the regeneration fluid, with consideration given to good flow distribution. The impact of process limitations, either heating or stripping, was not fully appreciated because simulation capability was unavailable in the past to properly assess the comparative severity of either of these conditions. Therefore, knowing the thermal stability of zeolitic materials, the primary focus was put on keeping the regeneration time as short as possible.

Recently, much of UOP’s design and technical field efforts have centred on natural gas processing operations with high moisture concentrations, large temperature gradients and high-pressure regeneration fluids, all of which put more stress on the molecular sieve adsorbent. These common energy-efficient practices often trigger the formation and refluxing of free liquid water during regeneration.

To better understand the onset and magnitude of refluxing, we have developed a computer model of the adsorption and regeneration process. Based on UOP’s expertise in adsorption processes over the last 50 years, our model has been able to explain the commercial problems encountered by our customers, primarily in the last ten years. The model is adaptable to different regeneration schemes, has improved design capabilities, and can be used to optimise the adsorbent configuration for maximum performance and life. An example of an installation where extensive information is available will be used to show how the model compares with past commercial experience. While the model can predict undesirable performance, in many cases there is little the customer can do to avoid the refluxing condition without spending large sums of money. For these customers, UOP focused on enhancing the robustness of the 4A adsorbent to tolerate more severe operating conditions.

Most commercial molecular sieve products used in process applications are in the form of shaped particles (eg, extrudates or beads). To make these forms, the zeolite crystals are usually embedded in a binder. While only a minor component in the product, the binder plays a vital role in the performance of the product. The binder is needed to give the zeolite crystals their aggregate shape, but it cannot interfere with access of the process fluid to the zeolite crystals. The zeolite crystals themselves are microporous solids with pore diameters in the range 3–10 Å, depending on the zeolite and cation type. The binder particles are smaller than the zeolite crystals and serve as a bridge between crystals, “gluing” them together. Since the binder is amorphous in nature, it will introduce mesopores (10–300 Å diameter) and macropores (>300 Å diameter) into the overall structure. These larger pores are necessary to bring the adsorbate(s) to the zeolite crystals, where they can be adsorbed. However, too many large pores will lead to excessive co-adsorption of the heavier hydrocarbon components in the feed, catalytic effects upon regeneration and liquid entrainment.

Liquid carry-over from an under-designed or poorly operated inlet separator can cause problems ranging from channelling to destruction of the binder. The effect on the adsorbent ranges from coking to pore blockage and particle break-up, all leading to a significant reduction in service life. By developing a product with the appropriate binder, UOP can tailor a molecular sieve to a specific application, whether it is low catalytic activity or break-up resistance. Binder selection must also allow for the mechanical strength of the adsorbent particle while creating an open structure to allow easy transition of the adsorbate molecules from the bulk fluid phase to the zeolite crystals, remaining stable when subjected to numerous regenerations.

The problem of under-performing natural gas dehydrators has been approached on two fronts. We have developed a detailed model of the regeneration step in the dehydration process to identify the potential for severe service conditions. We have also developed, concomitantly, a more robust 4A product, Molsiv UI-94 adsorbent, which uses a unique binder system and is designed to be used where engineering solutions are not possible or adequate in resolving poor performance. By combining these two efforts, our model can also be used to determine the best possible adsorbent options to maximise performance.