• Author / Uploaded
• kaaskopdawie5755

• Categories
• Procesamiento de gas natural
• Gas natural
• Ion
• Mercurio (Elemento)
• Rieles

Citation preview

DESIGN CONSIDERATIONS FOR MERCURY GUARD BEDS Laurance Reid Gas Conditioning Conference February 26 – March 1, 2017 – Norman, Oklahoma USA

Reece C. McHenry Black & Veatch 11401 Lamar Ave Overland Park, Kansas USA 66211-1508 +1 913-458-9285 [email protected] Dan McCartney McCartney Gas Advisors, LLC P.O. Box 27089 Shawnee Mission, Kansas USA 66225-7089 +1 913-593-3912 [email protected]

ABSTRACT This paper discusses the mechanisms of mercury attack on aluminum equipment. In order to prevent this attack, the best practices of the gas conditioning industry are outlined, with an emphasis on mercury guard beds. Types of sorbent, theory of operation, and relative cost for mercury guard beds are reviewed. This information is combined into a holistic analysis that uses feed gas composition to determine order of the unit operations within a gas conditioning process. Finally, there is a brief explanation about sorbent loading and the options for material disposal.

111

DESIGN CONSIDERATIONS FOR MERCURY GUARD BEDS Reece C. McHenry, Black & Veatch, Overland Park, KS USA Dan McCartney, McCartney Gas Advisors, LLC, Shawnee Mission, KS USA

Introduction Gas Plant Value Chain The value chain for natural gas and other light hydrocarbons begins at the wellhead. Once a well is opened for flow, rudimentary separations are performed on the lease before comingling the hydrocarbons and subsequent transportation via pipeline to a gas plant. Bulk water removal occurs at the plant boundary, and then follows the block flow diagram shown in Figure 1. The stream of hydrocarbons is sent to Gas Conditioning, which can consist of several unit operations; the most common being Acid Gas Removal Unit (AGRU), Dehydration Unit, and Mercury Removal Unit (MRU). Respectively, these processes remove carbon dioxide and sulfur species, water, and mercury. The treated gas is then passed through a Brazed Aluminum Heat Exchanger (BAHX) to cool the gas and condense hydrocarbons for natural gas liquids (NGL) recovery. For a liquefied natural gas (LNG) plant, the remaining gas is mostly methane and ethane, which passes through another BAHX to achieve liquefaction around -260 °F.

LNG BAHX

PIPELINE

Gas Conditioning

NGL Recovery BAHX

AGRU Dehydration MRU

NGL

Figure 1 - Gas Plant Block Flow Diagram

The two BAHX featured prominently in Figure 1 emphasize the importance of this piece of equipment. Without the ability to lower the temperature of the mixed hydrocarbon stream, the value chain is broken. Cryogenic Heat Exchangers The cryogenic heat exchanger industry is dominated by two types of exchangers, which are classified according to method of construction; Brazed Aluminum Heat Exchanger (BAHX) or Coil Wound Heat Exchanger (CWHX). A BAHX consists of thin sheets of corrugated aluminum which separate and contain the fluids that are exchanging energy. The sheets are joined together by brazing the entire bundle in a single operation, hence the name. A CWHX consists of aluminum tubing interwoven in complex patterns and separated by baffles, all contained within a

112

vessel. Heat transfer occurs between the fluid surrounding the tubes and the fluid within the tubes. In general, a BAHX is much more compact than a CWHX, and consequently has less external surface area to maintain at cryogenic temperatures. Because a BAHX is more widely used within the gas processing industry, it will be the focus of this paper. Aluminum is chosen as the material of construction for cryogenic applications because it has a high thermal conductivity, lacks a ductile-to-brittle transition temperature, and has a relatively low cost. However, the reaction of aluminum with mercury can be catastrophic in the BAHX, leading to header, weldment, and pinhole leaks. Photographs of actual exchanger failures can be seen in the paper by Willard [1]. Embrittlement & Amalgamation During the manufacture of a BAHX, the equipment is exposed to the presence of air, which allows for the development of a very thin but strong aluminum oxide (Al2O3) layer. This layer provides protection for the pure aluminum underneath. As a reference, the surface tension of liquid mercury is approximately 6.7 times greater than the surface tension of water [2], which means that this thin and mostly continuous layer of oxidation can prevent direct contact between the liquid mercury and the BAHX [3]. There are three methods of breaching this passivation: abrasion by solid particles, corrosion from additives, and oxide fatigue [4]. Abrasion can stem from any particulate matter entering the BAHX and destructively impinging on the aluminum oxide to expose the pure aluminum underneath. Similarly, corrosion from wellhead additives can selectively reduce the aluminum oxide layer to deplete the protection. However, the unavoidable mechanism is that of oxide fatigue, whereby thermal and mechanical strain induces micro-fractures in the aluminum oxide layer [4]. Thermal and mechanical strain arises from the difference in the coefficient of thermal expansion between aluminum and aluminum oxide. The aluminum is approximately 3 times more expansive than the aluminum oxide, so thermal cycling causes the aluminum to expand and contract in all dimensions more than the oxide layer on top of it [4]. This disparity induces stress which develops into fractures over many repeated cycles [4]. Once the passivation of aluminum oxide is breached, there are two possible mechanisms through which mercury can begin attacking the aluminum directly: Liquid Metal Embrittlement (LME) and Amalgamation Corrosion (AMC). It should be noted that amalgamation, the reaction of mercury with aluminum, occurs in both mechanisms, despite the naming convention. LME is the process whereby liquid mercury in contact with aluminum propagates a fracture along granular aluminum boundaries [4]. The exact mechanism of propagation is debated, but statistically occurs at weldments containing magnesium and along intergranular cracks [4]. While intergranular fractures may not seem detrimental, when mechanical weakening is coupled with operation at 1,015 psi, equipment failure becomes much more probable. AMC is the process whereby liquid mercury in contact with aluminum catalyzes the subsequent reaction of aluminum with water to form a weak and porous hydrated aluminum oxide [4]. Note that in the reactions presented below, the mercury reacts and is then free to attack a new molecule of aluminum, continuing the corrosion as long as water is present.

113

Hg + Al → Hg (Al) Hg (Al) + H2O → Al2O3 • H2O (porous) + H2 + Hg

(Equation 1) (Equation 2)

The process is mass transfer limited and also self-defeating [4]; eventually, the porous hydrated aluminum oxide will become too thick and obstruct additional water from reaching the site of the amalgamation reaction. Once this process is complete, the mechanical weakening of the aluminum (now hydrated aluminum oxide) may be severe enough to permit a leak. See Figure 2 for a drawing of this process.

MERCURY

AL2O3 FORMATION OF HYDRATED AL2O3 (ALONG CAVITY FACE)

ALUMINUM

Figure 2 – Schematic Depiction of AMC

Given these considerations, it is apparent that the BAHX can be protected by providing equipment and procedures that mitigate the loss or damage to the aluminum oxide layer. First, a strainer should be installed upstream of the BAHX to prevent solid particles from entering the BAHX so abrasion does not occur. This strainer is already an industry standard in order to prevent plugging of the narrow channels within the BAHX; however, the benefits still apply here. Second, upstream chemicals should be evaluated for undesired interactions before being added to the hydrocarbon stream. Third, the use of magnesium in the construction of the BAHX should be kept to a minimum, to reduce the probability of BAHX failure. This consideration is the responsibility of the BAHX manufacturer [5]. Fourth, careful temperature control must be maintained during thermal cycling that will occur during compressor trips or during commissioning and start-up [5, 6, 7]. The most critical element of success is the prevention of mercury from entering the BAHX. To this end, technology has been commercialized that captures mercury in a packed bed referred to as the Mercury Removal Unit (MRU). While the concentration of mercury in natural gas varies by geographic region, it has been determined that essentially all natural gas contains elemental mercury [8]. Thus, the application of mercury removal is important to all gas processors.

114

Technology Sorbents All MRU sorbents consist of two parts, called the support and the trap. The support is an inert material that provides structure and area for the trap. Additional desired properties of the support include high crush resistance and low cost. The trap is the reactive material that directly interacts with mercury. Desired properties of the trap include high selectivity for mercury, high loading capacity, and resistance to contaminants and upsets. There are a wide range of mercury sorbents available, and the advantages and limitations of each can be categorized according to support and trap material, as shown in Table 1. Table 1 – Product Space for Mercury Sorbents (C = commercial product; S = studied)

TRAP

SUPPORT Activated Carbon

Activated Alumina

Molecular Sieve

Sulfur

C [9]

C [10]

S [11, 20]

Metal Sulfide

S [9]

C [12]

S [13]

Silver

S [14, 15]

S [16]

C [17]

Ionic Liquid

S [18]

C [18]

S [18]

Commercialization of sulfur activated carbon marked the advent of the mercury removal industry. While this sorbent was cost effective and able to remove mercury from natural gas, it was suspected that saturated hydrocarbon service would leach the loosely held sulfur into the liquid phase. This concern prompted the development of metal sulfide as a trap material, because it contained a strong bond between metal and sulfur. Unfortunately, metal sulfides are generally pyrophoric, and must be loaded and unloaded under carefully controlled conditions [19]. This disadvantage spurred the development of sorbents that contained metal oxides, which are not pyrophoric and can be sulfided in-situ to form metal sulfide traps. Metal oxides are not shown in Table 1 because the sulfide form, rather than the oxide form, is responsible for the mercury trapping. In the early development stages all mercury sorbents were sacrificial, with each bed being loaded with sorbent, adsorbing mercury, and being discarded at the end of life. However, vendors already supplying regenerable molecular sieve for water removal were able to create a regenerable sorbent containing silver on zeolite. While this allows for the sorbent to be re-used multiple times before replacement, many new challenges became apparent, as will be discussed later in this paper. Recent innovation focuses on the development of unique materials for mercury removal. One example is impregnation of sulfur onto carbon molecular sieve, which has high surface area and nearly uniform pore size [11, 20]. Another example is the use of an ionic liquid as a trapping material. These liquids are composed purely of ions at room temperature [18], and are attractive for high removal efficiency, but are still being developed.

115

Material disposal is also an important factor that influences the choice of sorbent, and will be considered in the Operational Considerations section of this paper. Sorbents: Sulfur Activated Carbon The most prevalent technology in use for mercury removal from natural gas is sulfur activated carbon. Activation can be classified as either thermal or chemical, and a detailed description on each is outside the scope of this paper; however, it is sufficient to note that the coal type and activation process affect the ultimate mercury capture. For example, during chemical activation with elemental sulfur, the impregnation temperature determines the allotrope of sulfur formed, which determines reactivity and distribution [21]. The sulfur species is held within the pore space or weakly interacts with the carbon matrix [21]. Extensive studies on the method of mercury capture by sulfur activated carbon have determined that the process is a combination of chemisorption and physisorption [21, 22, 23]. Chemisorption is the primary mechanism of capture, which proceeds by formation of mercuric sulfide. Hg + S → Hg S

(Equation 3)

Since this is a chemical reaction, increasing the temperature of the natural gas increases the amount of mercury removed from the effluent. In a similar manner, increasing the amount of sulfur impregnated on the activated carbon increases the ability of the sorbent to chemisorb mercury, to an extent [21, 22]. Physisorption also occurs to a much smaller degree and is a function of the available surface area that is not covered by sulfur [21]. It is hypothesized that physisorption is controlled by defects within the carbon matrix, such as oxygen [24, 25]. While sulfur can be leached from the carbon support by liquid hydrocarbons, it is not soluble in liquid water and remains within the carbon pore space [10, 26]. The decrease in mercury sorption when liquid water is present occurs because the liquid imposes a mass transfer barrier. The presence of water does not cause mechanical weakening of the carbon support. For some form factors, such as pellets, water is not problematic as long as the binder (commonly clay) has been chosen correctly. Sorbents: Metal Sulfide or Metal Oxide Alumina Activated alumina consists of a homogeneous network of aluminum oxide, where a certain proportion of surface aluminum atoms have been substituted with other metal oxide molecules. This metal is then reacted with sulfur to create a strong bond that prevents the sulfur leaching into liquid hydrocarbons. An example would be the use of copper oxide or copper sulfide. However, the fact that the sulfur-metal bond is strong does not prevent deposition of liquid water or liquid hydrocarbon onto the sorbent. The establishment of a liquid barrier onto the surface of the material will inhibit mass transfer from the gas phase to the solid surface. Because the most common use of alumina is in desiccant service, it should be expected that alumina will tend to promote the sorption of water from the gas phase. A disadvantage of the metal sulfide however, is that it can be oxidized, which leads to the loss of reactivity with mercury. The reaction of the sulfide to the sulfate prevents the mercury ion from bonding with the sulfur species because of the steric hindrance created by the additional oxygen molecules.

116

The substitution of molecules other than aluminum into the network structure will reduce the mechanical strength of the material. As with activated carbon, mechanical weakening in activated alumina can occur in liquid water service if the binder is not carefully chosen. Another possible risk that must be accounted for is pyrophoric nature of the metal sulfide. Although hazardous, the industry has a long history of dealing with similar materials such as Iron Sponge, and using proper precautions, such as training and procedures, these materials can be safely handled [27]. Sorbents: Metal Molecular Sieve Molecular sieve consists of a homogeneous network of aluminum and silicon connected in a tetrahedral arrangement by oxygen atoms. At regular intervals, cations are present to balance the charge that would otherwise exist within the alumina-silica matrix. These materials are characterized by the diameter of the space that exists at the center of the crystal unit. For molecular sieves such as 3A, 4A, 5A, and 13X, the cations are different ratios of potassium, sodium, and calcium. However, transition metals such as silver or gold can also be inserted into this matrix. This metal can then form an amalgam with mercury, similar to the process shown in Equation 1. Similar to the activated alumina, the molecular sieve will preferentially absorb water in the pore space of the crystal network, per normal operation [17]. Since the metal is present on the exterior of the structure, the mercury is free to interact with little resistance to mass transfer. An additional benefit of this technology is that heat can be applied to regenerate the sorbent, without significant damage to mercury removal capacity [17]. As with previously discussed sorbents, the presence of liquid hydrocarbon covering the sorbent will still impose a barrier to mass transfer and prevent efficient mercury capture. There is no unique property of molecular sieve that compensates for two-phase service. It should also be mentioned that precious metals, such as silver, are reactive with sulfur compounds, which decreases the ability to capture mercury. Sorbents: Other Offerings An ionic liquid (IL) can be an excellent trap because the ions can be designed to remove mercury without being affected by liquid water or other contaminants. Although referred to as liquids, these fluids do not easily vaporize. An IL consists of at least one cationic species and at least one anionic species, which allow the solution to remain in the liquid phase. Because mercury within hydrocarbon streams is predominantly elemental [8], the IL is designed with an oxidizing functional group on the anion that strips away valence electrons [28] resulting in a positively charged mercury. The mercury species then partitions into the hydrophobic IL to interact with functional groups attached to the cation, the most preferable being sulfur species [29]. The mercury oxidation can be seen in Equation 4, and the mercury trapping in Equation 5; both equations use bracketed ions to represent the bulk IL. Hg0 + [C+][A-] → Hg2+ + [C+][A-] Hg2+ + [C+][A-] → [Hg2+][A-][C+]

(Equation 4) (Equation 5)

117

It has been determined that the presence of sulfur on either the anion or cation of the IL is essential to the retention of mercury, whereas the oxidation occurring at the anion merely enhances the effect. Since the cation usually contains the sulfur, it needs to be protected against ion exchange, which would deplete the liquid. For example, pyridinium can be synthesized with an 8-carbon hydrophobic tail in order to prevent ion exchange between mercury and other unreacted cations within the IL; attaching a sulfur moiety on the end of the tail retains the mercury within the IL [29]. Because research is still ongoing with regards to scale-up and the possibility of emulsions, side reactions, and poisoning, there are many pitfalls surrounding IL technology that must first be navigated [30]. For example, one patent specifies that the most preferable operating conditions should not exceed 104 °F and 17.4 psi [18], which are not amenable to gas conditioning operations. It should also be mentioned that the fluids are generally combustible and highly toxic [31]. Sorbents: Comparison The following information is a survey of the commercial offerings for mercury removal from natural gas streams. Table 2 – Average Properties of Commercial Adsorbents (AC = activated carbon; AA = activated alumina)

Support

Trap

Form

AC AC AC AA AA AA

Sulfur Sulfur Metal Sulfide Sulfur Metal Sulfide Metal Sulfide

Granular Pellet Pellet Granular Granular Tri-lobe

Diameter (in) 0.08 – 0.18 0.04 – 0.16 0.06 – 0.16 0.06 – 0.20 0.06 – 0.16 0.06

Bulk Density (lb / ft3) 36 35 32 50 – 51 32 – 53 36

Table 2 contains a range of sorbents that have been grouped based on the support material, trap material, and form factor described within product literature. This may be an oversimplification in some instances that skews average properties shown; as such, this table should be considered informative only. The maximum operating temperature of each sorbent was also queried, but the answers were inconsistent and varied widely (160 – 392 °F). The reasoning for the temperature limit was described as either physical degradation of the materials, or altering of the adsorption profile. For the sorbents discussed (except “Metal Molecular Sieve” and “Other Offerings”) the following physical degradations were identified as possible: non-chemisorbed sulfur on activated carbon; chemisorbed sulfur on activated carbon; aggregate sulfur breakup; breaking of metal sulfide bond. The most likely bond to break was presumed to be the sulfur-sulfur bond in the aggregate material. Although it varies with the method of preparation, the most likely allotrope of sulfur as an aggregate on activated carbon is monoclinic β-S, which melts at 247.3 °F [32]. This seems to indicate that margin has been applied to commercial products because of poor characterization or feed gas variability. The degradation of the strong metal sulfide bond is not a concern at

118

temperatures experienced in a natural gas plant. Evidence that elevated temperatures negatively impacted mercury uptake was not found in the literature. The final point of sorbent comparison is that of material cost. Multiple commercial vendors were consulted regarding designs and budgetary quotes for an MRU; however, quotes for precious metal on molecular sieve and ionic liquid on various media were not obtained. The unit considered was designed to treat 70 MMSCFD of natural gas containing no water, and was located after a dehydration bed. Given these parameters, the sorbent cost was normalized, and the offerings fell into a range from 0.28 – 1.46 $ / lb of sorbent / year, with 0.80 $ / lb of sorbent / year being the median. As the normalization implies, the essential parameters are years of service and the equilibrium capacity of the sorbent (which determines the necessary mass of sorbent). Note also that the bed support media (ceramic) is generally inexpensive, and costs approximately one order of magnitude less than the sorbent. A last characteristic to consider is the cost of the vessel; for this analysis the vendors were restricted to a maximum of 8 psi differential across the bed. Since all vessels in this position would have the same thickness and material of construction, the analysis became a comparison of (L/D), which ranged from 1.0 – 2.5, with 1.5 being the median. Selection Criteria Based on the information presented, it should be clear that the feed gas contaminants dictate the possible MRU sorbents, which in turn dictate the position of the MRU. There is no single sorbent that performs well in all services, as each one has advantages and disadvantages. The problematic feed gas species consist of hydrogen sulfide, oxygen, and water. The possibility of liquid phase hydrocarbon is a significant point of consideration as well. It is known that activated carbon, activated alumina, and metal sulfides catalyze the Claus reaction, shown in Equation 6. Therefore, it is critical to determine the scenarios where the reactants are also introduced [33]. In Equation 6, the wide variety of sulfur species that can form are denoted by an “x.” 2 H2S + O2 + catalyst → 2 H2O + (2⁄ ) Sx

(Equation 6)

If oxygen is present in the natural gas, the AGRU will remove only a minimal amount of the oxygen. This is because oxygen has a low solubility in amine solutions and slow reactivity to form heat stable amine salts. Thus, it is possible that 10 ppm levels of oxygen could reach the MRU bed downstream of the AGRU, even if it is placed immediately before the BAHX. In a similar manner, an upset in the AGRU could slip hydrogen sulfide through the dehydration unit and on to the MRU, where it could react with oxygen in the feed gas to form water. Since the removal of water is critical to the functioning of the BAHX, even water formation at 10 ppm could be problematic. Thus, prior to entering the BAHX, the hydrogen sulfide is removed to 1.0 ppm, and water is removed to 0.1 ppm. Arrangements: MRU following Dehydration As shown in Figure 3, the MRU is a packed bed that appears immediately after the Dehydration Unit and before the protective strainer of the BAHX. 119

AGRU CONTACTOR

PIPELINE BAHX

EXCHANGER FILTER

FILTER

AMINE LOOP

DEHY

MRU

Figure 3 – MRU following Dehydration

As mentioned in the Selection Criteria section, this position provides the cleanest feed gas to the MRU for processing, thus it would appear that any sorbent can be used. As discussed above, the possibility of the Claus reaction occurring should be considered. Another point of consideration comes from operational experience; when dealing with a metal sulfide it can take a longer amount of time to dry-out (compared to sulfur activated carbon) before placing the MRU into service [34]. This configuration offers no method to remove water from the hydrocarbon stream before entering the BAHX, so the MRU be must be absolutely dry before beginning liquefaction. Arrangements: MRU before the AGRU Contactor As shown in Figure 4, the MRU is a packed bed that appears immediately after the Feed / Effluent Exchanger, and before the AGRU contactor.

Figure 4 – MRU before AGRU Contactor

The primary benefit of this arrangement is that mercury is removed as one of the first operations within the plant. This minimizes the distribution of mercury within the plant piping and vessels to the greatest extent possible. Additionally, the possibility of undesired environmental releases is also minimized, because mercury cannot partition into waste streams leaving the AGRU and Dehydration Unit. In the worst case scenario, the presence of hydrogen sulfide and oxygen in the feed gas could catalyze in the MRU via the Claus reaction and form water. However, because the MRU is followed by a contactor with an aqueous amine solution, the generation of water here is not a concern. The primary consideration with locating the MRU at this position is the possibility of process upsets slugging liquid onto the bed. Regardless of the chosen sorbent, a mixed phase on the bed will inhibit mass transfer and thus mercury uptake. The design of the Feed / Effluent

120

Exchanger should consider this possibility and be designed to heat the feed stream an appropriate margin above the dew point. Arrangements: MRU combined with Dehydration As shown in Figure 5, the MRU is a packed bed that appears in combination with the Dehydration Unit. For this arrangement, the trap must be able to withstand the regeneration gas conditions used to remove water from molecular sieve (approximately 550 °F). These conditions disqualify the use of sulfur as the trap, as it will be liquefied and carried away (the vapor pressure is negligible) [35]. The only option for a trap becomes the use of a precious metal, such as silver. While this option is attractive because it removes the MRU vessel, there are several additional design factors that must be considered.

Figure 5 – MRU combined with Dehydration

The regeneration gas stream, shown as a dashed line in Figure 5, will be mercury-rich and water saturated. Any piping or vessels used to transport this vapor will be contaminated by the presence of mercury. Thermodynamics indicate that the mercury will partition into the aqueous phase, which allows a heat exchanger and separation vessel to generate a dry and low-mercury spent regeneration gas. There are multiple options to consider for the fate of this gas: re-liquefy, combust, dilute. If this gas will be liquefied, it can be sent to the front of the plant and passed though the MRU again. If this gas will be consumed as fuel, emissions permits may require the use of an additional MRU before combustion. This bed can be specified as a sacrificial guard bed, and sized using the regeneration gas flowrate. If the gas will be mixed with pipeline gas it will generate a less saleable product. This is because the mercury can harm precious metal catalysts and other aluminum equipment that users must safeguard against. The other stream produced from the regeneration operation will be the mercury-rich aqueous phase. Design of an additional vessel containing another sorbent or mass separating agent can be used to generate a pure mercury stream. This fluid can then be periodically collected and shipped for disposal as outlined in the section on Operational Considerations. Both the dehydration material and the mercury sorbent will require disposal after many regenerations due to the effects of hydrothermal aging. All of the spent material in the bed will need to be removed, managed, and disposed as though it contains trace mercury. On the basis of sorbent cost alone, precious metal materials are the most expensive.

121

Arrangements: Unsatisfactory Arrangements Note that three other arrangements are possible, from a purely theoretical viewpoint. However, consideration of process conditions reveals that these arrangements are unacceptable. Figures will not be presented for these, in order to discourage erroneous usage of these arrangements. First, the MRU could be placed immediately after the Inlet Gas Coalescer; this arrangement is not satisfactory because fluids leaving this vessel may be at saturation conditions. Thus, any process upset may lead to condensation within the MRU, requiring the sorbent to be replaced. Second, the MRU could be placed immediately before the water-saturated vapor is passed to the Dehydration Unit. This arrangement is not satisfactory for any sorbent, for the same reasons as placement after the Inlet Gas Coalescer. Third, the MRU could be placed after the BAHX strainer. This is not satisfactory because it would allow any particulate matter from the MRU to enter directly into the BAHX and cause abrasive damage to the aluminum oxide layer.

Operational Considerations Bed Screens Sacrificial guard beds are expected to operate isothermally, so there is generally no allowance made for the expansion or contraction of the vessel. A regenerable bed however will experience significant thermal stress during regeneration, and the bed screens must have a system in place to accommodate this movement. Failure to mitigate bed screen movement might allow slip of both particulates and untreated fluids into the BAHX. Loading Conditions Frequently, little consideration is given to the actual loading and unloading of sorbent materials; however, these operations are crucial to the overall success of the packed bed, and should not be overlooked. “Sock loading” is the most common method used to load beds. A sealed canvas sack or drum containing the bed support material or sorbent material is suspended above the vessel and then dumped into a cloth tube that conveys it into the vessel. The purpose of the cloth tube (colloquially known as the “sock”) is to control the flow of material into the vessel. The sorbent gradually distributes according to angle of repose on top of the support mesh screen and other support media. It is good practice to check a sample of the sorbent for size and bulk density, to verify that the Bill of Material is correct before material is loaded into the vessel. BAHX Inlet Strainer During loading operations, care must be taken not to load the bed materials too quickly. Doing so will generate an excessive amount of small diameter particles (referred to as “fines”) via mechanical agitation between particles. In the same way, operation of the bed can generate fines if the momentum of the gas stream entering the vessel is not distributed correctly. Particles can be perturbed within the bed and abrade each other or the vessel wall. Once fines have been created, they will either be retained within the bed or pass through it. Particles that are retained can plug the pore space of the packed bed and begin the process of creating flow maldistribution and increased pressure differential. This could be misinterpreted as 122

a sign of mercury loading and prompt a re-fill due to the plugging, but not due to the bed being spent. If the particles pass through the bed, the carryover will enter the BAHX, and possibly abrade the protective aluminum oxide film. As discussed previously in the context of AMC, an exposed aluminum surface allows mercury amalgamation to rapidly occur when the BAHX is warmed. Given this knowledge, it is considered best practice to utilize a strainer immediately before the BAHX, regardless of the upstream treating arrangement [5, 6, 7]. The Gas Processors Association minimum is 80 Tyler Mesh (0.0070 in) with finer mesh at the discretion of the BAHX manufacturer. As a final note, it is common for the BAHX manufacturer to offer designs which are “mercury tolerant.” This consideration means that the designer has minimized the area within the BAHX where mercury can accumulate. This can be done via computational fluid analysis, in order to determine where sloping and other geometries are necessary. Material Disposal The disposal of mercury containing materials must be in accordance with Title 40 of the Code of Federal Regulations, most notably the Resource Conservation and Recovery Act (RCRA). Based on the level of mercury in the material, the mercury can be recovered and recycled or stabilized and landfilled. The stabilization option is regulated according to the limitations given in the Toxicity Characteristic Leaching Procedure (TCLP). Because the spent sorbent is a hazardous waste material, it is packaged within two barriers and shipped to a waste processor. For mercury recycling, the vendor will subject the spent sorbent to high vacuum retort in order to vaporize the mercury. This vapor can then be condensed and distilled to generate high purity mercury for use in LCD screens, CFL bulbs, UV lamps, neon lights, medical equipment, or other applications. More commonly, the guard bed material undergoes stabilization; the spent sorbent is mixed with powdered sulfur polymer cement at elevated temperatures in order to ensure all mercury has been reacted to the form mercury (II) sulfide [36, 37]. This mixture is then liquefied and molded, followed by sealing in a container, and shipping to a landfill for indefinite storage.

References 1. S. Willard, “Field Repair of Brazed Aluminum Plate-fin Heat Exchangers,” in Gas Processors Association, San Antonio, 2015. 2. J. J. Jasper, “The Surface Tension of Pure Liquid Compounds,” J. Phys. Chem. Ref. Data, vol. 1, no. 4, 1972. 3. S. M. Wilhelm, A. McArthur and R. D. Kane, “Methods to Combat Liquid Metal Embrittlement in Cryogenic Aluminum Heat Exchangers,” in Seventy Third GPA Annual Convention, 1994. 4. S. M. Wilhelm, “Risk Analysis for Operation of Aluminum Heat Exchangers Contaminated by Mercury,” Process Safety Progress, vol. 28, no. 3, pp. 259-266, 2009. 5. API Standard 662, “Plate Heat Exchangers for General Refinery Service – Part 2: Brazed Aluminum Plate-fin Heat Exchangers,” American Petroleum Institute, 2011. 6. Aluminum Plate-fin Heat Exchanger Manufacturers’ Association, “The Standards of the Brazed Aluminum Plate-fin Heat Exchanger Manufacturers’ Association,” ALPEMA, 2010.

123

7. GPA-TB-M-001, “GPA Technical Bulletin – Brazed Aluminum Heat Exchangers,” Gas Processors Association, 2015. 8. L. Phannenstiel, C. McKinley and J. Sorensen, “Mercury in Natural Gas,” in Operational Section of Transmission Conference, Las Vegas, 1976. 9. K. Reddy, A. Shoaibi and C. Srinivasakannan, “Sulfur-Leaching Facts from SulfurImpregnated Porous Carbons in the Mercury Removal Process,” Energy & Fuels, vol. 29, pp. 4488-4491, 2015. 10. A. Costandi, “Process for Removing Hydrogen Sulfide and Mercury from Gases”. US Patent 4,786,483, 22 November 1988. 11. R. Ambrosini, R. Anderson, L. Fornoff and K. Manchanda, “Selective Adsorption of Mercury from Gas Streams”. US Patent 4,101,631, 18 July 1978. 12. Porocel Data Sheet, “DYNOCEL HG – SULFIDED,” January 2016. 13. A. Sugier and F. Villa, “Process for Removing Mercury from a Gas of Liquid by Adsorption on a Copper Sulfide Containing Solid Mass”. US Patent 4,094,777, 13 June 1978. 14. D. Karatza, M. Prisciandaro, A. Lancia and D. Musmarra, “Silver Impregnated Carbon for Adsorption and Desorption of Elemental Mercury Vapors,” Journal of Environmental Sciences, vol. 23, no. 9, pp. 1578-1584, 2011. 15. M. Manes, “Silver Impregnated Carbon”. US Patent 3,374,608, 26 March 1968. 16. J. Park and L. Winstrom, “Process for Reducing the Concentration of Mercury in Hydrogen Gas”. US Patent 3,257,776, 28 June 1966. 17. T. Y. Yan, “A Novel Process for Hg Removal from Gases,” Industrial & Engineering Chemical Research, vol. 33, no. 2, pp. 3010-3014, 1994. 18. R. Rogers, J. Holbrey and H. Rodriguez, “Process for Removing Metals from Hydrocarbons”. US Patent 20120121485 A1, 17 May 2012. 19. Carnegie Mellon Environmental Health & Safety, “Pyrophoric Handling Policy,” June 2010. [Online]. Available: www.cmu.edu.ehs. [Accessed 1 August 2016]. 20. T. Matviya, R. Gebhard and M. Greenbank, “Mercury Adsorbent Carbon Molecular Sieves and Process for Removing Mercury Vapor from Gas Streams”. US Patent 4,708,853, 24 November 1987. 21. W. Liu, R. Vidic and T. Brown, “Optimization of Fixed-bed Application of Activated Carbon-Based Sorbents for Gas-Phase Mercury Removal,” Environmental Science & Technology, vol. 32, no. 4, pp. 531-538, 1998. 22. W. Jozewicz, S. Krishnan and B. Gullet, “Bench-scale Investigation of Mechanisms of Elemental Mercury Capture by Activated Carbon,” in Second International Conference on Managing Hazardous Air Pollutants, Washington, D.C., 1993. 23. R. K. Sinha and P. L. Walker Jr., “Removal of Mercury by Sulfurized Carbons,” Carbon, vol. 10, pp. 754-756, 1972. 24. W. Feng, E. Borguet and R. Vidic, “Sulfurization of Carbon Surface for Vapor Phase Mercury Removal – I: Effect of Temperature and Sulfurization Protocol,” Carbon, vol. 44, pp. 2990-2997, 2006. 25. W. Feng, E. Borguet and R. Vidic, “Sulfurization of Carbon Surface for Vapor Phase Mercury Removal – II: Sulfur Forms and Mercury Uptake,” Carbon, vol. 44, pp. 2998-3004, 2006. 26. J. Boulegue, “Solubility of Elemental Sulfur in Water at 298 K,” Phosphorus and Sulfur and the Related Elements, vol. 5, no. 1, pp. 127-128, 1978. 27. C. R. Perry, “A New Look at Iron Sponge Treatment of Sour Gas,” in Laurance Reid Gas Conditioning Conference, Norman, 1970.

124

28. L. Ji, S. Thiel and N. Pinto, “Room Temperature Ionic Liquids for Mercury Capture from Flue Gas,” Industrial and Engineering Chemical Research, vol. 47, no. 21, pp. 8396-8400, 2008. 29. N. Papaiconomou, J.M. Lee, J. Salminen, M. Von Stosch and J. Prausnitz, “Selective Extraction of Copper, Mercury, Silver and Palladium Ions from Water using Hydrophobic Ionic Liquids,” Lawrence Berkeley National Laboratory, p. 25, 2008. 30. A. Hassan, J. Holbrey, et al. “Mercury Removal from Gas Streams using New Solid Adsorbents,” in International Gas Union Research Conference, 2011. 31. T. P. Thuy Pham, C.W. Cho and Y.S. Yun, “Environmental Fate and Toxicity of Ionic Liquids: A Review,” Water Research, vol. 44, no. 2, pp. 352-372, 2010. 32. B. Meyer, “Elemental Sulfur,” Chemical Reviews, vol. 76, no. 3, pp. 367-388, 1976. 33. M. Steijns and P. Mars, “The Adsorption of Sulfur by Microporous Materials,” Journal of Colloid and Interface Science, vol. 57, no. 1, pp. 175-180, 1976. 34. J. Mock, R. Ramani and D. Messersmith, “Experiences in the Operation of Dehydration and Mercury Removal Systems in LNG Trains,” in Laurance Reid Gas Conditioning Conference, Norman, 2008. 35. Freeport Sulfur Company, “The Sulphur Data Book,” Port Sulphur, LA: McGraw-Hill, 1954. 36. D. Melamed, M. Fuhrmann, P. Kalb and B. Patel, “Sulfur Polymer Cement Stabilization of Elemental Mercury Mixed Waste,” Brookhaven National Laboratory, Contract No. DEAC02-98CH10886, April 1998. 37. J. Boyle, B. Lawrence and S. Schreffler, “Method and Apparatus for Generating Mercury (II) Sulfide from Elemental Mercury”. US Patent 7,691,361 B1, 6 April 2010.

125

126