3.5.4.4. Natural Gas Cleaning Marit Mork and Jon Steinar Gudmundsson, Department of Petroleum Engineering and Applied Geophysics, Norwegian University of Science and Technology, Trondheim, Norway. Keywords Natural gas treatment, Gas cleaning, Carbon dioxide, Hydrogen sulfide, Acid gas, Adsorption, Absorption, Amine, Potassium Carbonate, Chemical solvent, Physical Solvent, Gas permeation, Membrane, Desulfurization, Dehydration. Contents 1 Introduction 2 Removal of Carbon Dioxide and Hydrogen Sulfide 2.1 Solid Bed Adsorption 2.2 Chemical Solvents 2.2.1 Amine Processes 2.2.2 Potassium Carbonate Process 2.3. Physical Solvents 2.4 Combined Processes 2.5 Gas Permeation 2.6. Other Methods 3. Direct Conversion of Hydrogen Sulfide to Sulfur 4 Dehydration of Natural Gas 4.1 Absorption 4.2. Adsorption 4.3 Other Methods Glossary Absorption process: a separation process where gas is dissolved in a liquid. Acid gas: compounds in natural gas that forms weak acids with water. Usually, the term acid gas. removal is used for removal of hydrogen sulfide and carbon dioxide in natural gas. Adsorbent: solid material used in adsorption processes. Adsorber: a column or a tower where gas is contacted with the adsorbent. Adsorption process: a physical or chemical process where some gas molecules are attached to the surface of a solid material by molecular attraction in order to separate some gas components from the gas stream. Alkaline salt: a salt that forms a base when dissolved in water Amine: an alkaline liquid used in chemical absorption processes Aromatic compounds: hydrocarbons that are easily absorbed in glycols Batch process: a process where new material is loaded for each run, consequently a non- continuously process Bauxite: rock largely composed of a mixture of aluminium oxides. Aluminium is produced from bauxite Bulk removal: removal of the main portion of the component v-Carbonate solution: an alkaline aqueous solution of potassium carbonate (KHCO:) or other carbonates. Catalyst: a material that affects the rate of reaction but emerges from the reaction unchanged. Condensate: light oil usually produced together with natural gas, Condenser: a cooler that cools gas and condenses vapor to liquid. Contactor: a column or a tower where liquid is contacted with the gas being processed in an absorption process. Cryogenic process: cooling of natural gas down to low temperatures resulting in light hydrocarbons in liquid state. Crystallization: a process where components in a liquid solution are converted into a solid by growth on seed particles. Dehydration: the process of removing water from gas or liquids. Desulfurization process: a process where hydrogen sulfide and other sulfur components are removed from natural gas. Dew point: the temperature at a given pressure at which gas or vapor condenses to liquid. Diffusivity: the rate of diffusion, which is the net flow of molecules transported from a region of high concentration to a region of low concentration. Distillation: separation of two or more liquids having different boiling points. The process involves heating of the mixture resulting in conversion of liquid into vapor and cooling to liquefy a part of the vapor. Exothermic reaction: a chemical reaction where heat is generated. Flaring: due to safety precautions, if the pressure in a process becomes too high, some of the gas is burned off controlled. Flashing: a process where liquid is stabilized by reducing the pressure and/or increasing the temperature so that light hydrocarbons evaporate, Gas permeation: a process where gas is transported through a membrane by diffusion, Glycol: an organic compound in the alcohol family often used as a solvent in dehydration processes, Incineration: oxidation of hydrogen sulfide and other sulfur compounds to sulfur dioxide, LNG production: production of Liquefied Natural Gas. Natural gas is cooled in a cryogenic plant resulting in liquefied methane, ethane and propane. Molecular sieve: a crystalline porous solid material with pores of exactly equal size containing polar charged active sites. Partial pressure: the pressure that each gas in a mixture of gases would exert if it alone occupied the volume of the mixture at the same temperature. Pipeline specifications: regulations for the upper concentration of impurities that can be transported with the natural gas in pipeline, ppm: parts per million, a measure of the portion of a component in the gas. Propane plus fractions: a mixture of hydrocarbons including propane and heavier hydrocarbons. Reboiler: a heat exchanger providing heat to regenerate an amine or a glycol. Regeneration: removal of the acid gases from the rich solvent after absorption. Sales gas specification: regulations giving the upper allowable concentration of impurities in sales gas. Scrubber: a separator to remove traces of liquid from gas. Standard m? (Sm’): gas volume in cubic meters at standard conditions (usually 1 bar and 15 °C) Stripper: a regeneration column where the reaction between acid gases or water and solvent is reversed by providing sufficient heat. Sweet gas: natural gas containing little or no hydrogen sulfide or other sulfur compounds. Synthesis gas production: a process where carbon monoxide and hydrogen is produced from methane and oxygen. Synthesis gas is used in the Fisher-Tropsch process for production of ‘methanol, waxes and diesel fuel. 2 822Tail-gas treating: a process for removal of traces of sulfur after other removal processes to meet specifications. Summary In the petroleum industry, gas cleaning involves removal of unwanted components from natural gas. Natural gas consists of hydrocarbons and impurities such as hydrogen sulfide, carbon dioxide, water and other sulfur compounds. The presence of these impurities can represent a health risk and an environmental risk, and thus specifications for the natural gas quality are established. Also, impurities cause problems during processing and transport of gas, having the result that corrosion increase, gas hydrates may form, and water is likely to condense. ‘One or more separation processes accomplish gas cleaning. To separate out hydrogen sulfide, carbon dioxide and water, these components can be solved in a liquid, attached to a solid surface, or they can diffuse through a membrane. These processes are called absorption, adsorption and gas permeation, respectively. Water and carbon dioxide can also be removed by cooling processes. In absorption and adsorption processes, liquid solvent and solid particles are usually regenerated. This means that impurities are removed, usually by heating, before the liquid or solid start attracting impure compounds again. However, concentrated hydrogen sulfide from this process can not be discharged to the atmosphere. Therefore, hydrogen sulfide is usually chemically converted into elemental sulfur, which is useful for other purposes. Many industrial methods are developed to purify the gas. In absorption processes, carbon dioxide and hydrogen sulfide are solved in different amines while water is absorbed by glycols. In a bed of solid particles, impurities are adsorbed on the particle surface when natural gas flows trough the bed. Particles of bauxite, zeolite, zinc oxide and iron oxide and other materials are utilized for adsorption. A relatively new industrial method for purification is by use of membrane, With membranes, a very Jow content of impurities can be obtained. This article will present an overview of methods and principles for natural gas purification. 1 Introduction Natural gas consists of light hydrocarbons and small amounts of heavier hydrocarbons and water vapor. In addition, natural gas contains impurities such as carbon dioxide, hydrogen sulfide and other sulfur compounds. Other components, not that important, are ammonia (NHs), hydrogen cyanide (HCN), carbonyl sulfide (COS), carbon disulfide (CS2), mercaptans (RSH), nitrogen (No) and sulfur dioxide (SO2). The gas quality is highly dependent on the concentrations of these contaminants. Moreover, the composition of the natural gas varies from one location to another. Gas cleaning usually involves removal of carbon dioxide, hydrogen sulfide and water. Well-known principles for removal are absorption or adsorption of these compounds, and lately, technology applying membranes are developed. After separation of hydrogen sulfide from natural gas, hydrogen sulfide is usually converted into pure sulfur, since emission of hydrogen sulfide to air is not allowed The degree of gas cleaning depends on the governmental regulations for emissions to air. However, the gas is required to meet the sales gas specifications. In the sales gas specifications, concentrations of carbon dioxide up to 2 percent, of hydrogen sulfide up to 4 ppm and a maximum water content of 0.1 g m®, are usually allowed. ew wwSome degree of gas cleaning is also important prior to processing and transport of natural gas. Hydrogen sulfide and carbon dioxide form weak acids in combination with water, which inorcase corrosion in process equipment and gas pipelines. Thus, the term “acid gas” is often used for natural gas with impurities. In addition, hydrogen sulfide is toxic and highly flammable. In contrast, carbon dioxide is non-flammable, but displaces oxygen that may cause suffocation. Also, water vepor and condensed water cause problems, however more practical problems. During transport the gas is cooled, and condensed water results. A removal of water vapor prevents accumulation of liquid Water in gas pipelines. Water accumulation can lead to the formation of natural gas hydrates, a solid material that may block pipelines (See Natural Gas Hydrates). As a result, pipeline specifications 2 Removal of Carbon Dioxide and Hydrogen Sulfide Gas cleaning is usually concentrated on removing sulfur compounds and carbon dioxide. Often both carbon dioxide and hydrogen sulfide are removed simultaneously and sometimes one of the compounds, usually hydrogen sulfide, is moved selectively. Due to the environmental and health clfects of hydrogen sulfide, further treatment of hydrogen sulfide is necessary before depositing. ‘Such conversion processes will be presented in section 3. Various processes are developed for acid gas removal, based on either chemical or physical Principles where the compounds are either absorbed in a physical or chemical solvent or edsorbed on a solid surface. Usually, for low concentrations of hydrogen sulfide in the feed stream, the adsorption processes are favorable to obtain a gas with very low hydrogen sulfide content. However, for removal of larger Smounts, adsorption is uneconomical and absorption processes are more suitable. At even higher concentrations, the physical solvent absorption processes are used. However, by using these processes, additional second cleaning of the gas is required because the natural gas is not satisfactorily purified in the first step. 2.1 Solid Bed Adsorption In an adsorption process, a fixed bed of solid particles is applied to remove mainly hydrogen sulfide but also some carbon dioxide and water by means of ionic bonding or chemical reactions, When a very high purity of the natural gas is required, one of the adsorption processes is used, In these Processes, the gas flows through the fixed bed while the acid gases are adsorbed on the solid Particles. When the bed is saturated with acid gases, the bed needs to be replaced or regenerated by heating. As a result, these processes are usually operated as a batch system, Three differeat sola bed adsorption processes are common; molecular sieves, iron sponge and zine oxide. Molecular Sieves: In the molecular sieves process polar molecules such as hydrogen sulfide and wwater can be removed by bonding to polar charged active sites on the surface of solid particles, This adsorption process can be used when very low vapor concentrations are required, and it is usually operated at moderate pressure at about 30 bar. meat ean ening heavy hydrocarbons and aromatic compounds to enter the pores. Small non-polar molecules such as carbon dioxide will also enter and become trapped inside the pores. ‘This 324obstructs the access of water and hydrogen sulfide, resulting in a decrease of the effectiveness of the adsorption. The molecular sieve bed does not suffer from any degradation, and therefore, it is regenerated by flowing hot sweet gas through the bed. The regeneration process is operated close to 315 °C for a Jong enough period of time to remove all adsorbed material, usually one hour or more. Hydrogen sulfide and carbon dioxide may form COS, but by proper design of the molecular sieve process this reaction is not promoted. Tron Sponge: The iron sponge process is applied to gases with low hydrogen sulfide content (less than 300 ppm), and is operated at low to moderate pressure up to 83 bar. The hydrogen sulfide is removed from natural gas by chemical reaction with ferric oxide: 2Fe,0, + 6H,S > 2Fe,S, + 6H,0 qa) This reaction requires the presence of water with a pH level of 8-10 and a temperature below 43 °C, The solid bed consists of wood chips with ferric oxide at the surface. During the process, hydrocarbons and sulfur tend to coat the surface area and inhibit the reaction. Solid particles are therefore regenerated regularly by introducing oxygen, however only 60 percent of the previous bed life can be expected. The oxidation of ferric sulfur is strongly exothermic so precautions must be taken. As a result, the oxygen gas must be introduced slowly so that the heat of reaction can dis- sipate, or the regeneration step is simply avoided by operating the process in batch mode where the solid particles are replaced. ‘Zine Oxide: The zine oxide process is similar to the iron sponge process, In a solid bed of granular ine oxide the acid gas reacts with the zinc oxide and forms water and zine sulfide: Zn0+ H,S — ZnS + H,O @) To promote the reaction rate, the temperature is normally kept above 121 °C. The catalyst must be removed and can at that time contain up to 20 weight percent of sulfur. Due to disposal problems with the spent catalyst, this process has seen decreasing use in the industry. However, a development of the process into a desulfurization process called Puraspec, allows purification of the gas at temperatures down to 0 °C and recovery of the metal in the metal oxide-based adsorbents. 2.2 Chemical Solvents In the chemical solvent processes, the acid gases are absorbed by and react with an aqueous solution of a weak base. The chemical reactions involved are reversible, so by changing pressure and temperature the base solution can be regenerated. Most processes use either an amine solution or a carbonate solution for purification of natural gas. 2.2.1 Amine Processes The amine processes are found to be particularly applicable when acid gas partial pressures are low, and/or when a low level of the acid gas compounds are required in the natural gas. 3.5.4.4figure | pdf: Flow diagram of an amine process 925Figure 1 shows a typical amine process. The sour gas enters an inlet separator where water and liquid hydrocarbons are removed. Then the gas flow upward in the contactor counter-current to the amine solution and hydrogen sulfide and carbon dioxide is absorbed by the solution. The rich solvent leaving the bottom of the absorber is then flashed to remove dissolved hydrocarbon gases, condensate and a small amount of the acid gases. The rich amine then heat exchanges with lean amine to recover some of the heat from the lean amine stream to decrease the duty in the reboiler. Thereafter, the heated amine enters the stripping tower where the bonds between amines and acid ‘gases are broken with heat from the reboiler. The amine solution is cooled and recycled back to the top of the absorber. Degradation of amine will occur in these processes by slow oxidation of the solution. The oxidation products often generate corrosion, but having an inert gas pocket at the top of the amine storage tank can reduce the oxidation process itself. Another problem with amine processes is the formation of foam. The formation may be due to the presence of suspended solids, condensed hydrocarbons, amine degradation products or foreign material from corrosion inhibitors or contaminants in the water. This problem can partly be avoided by good design and operation as well as by adding antifoaming agents. Various amines are available in the amine process, and they are categorized as primary, secondary and tertiary amines according to the number of organic groups bonded to the central nitrogen in the molecule. Monoethanolamine (MEA, primary), diethanolamine (DEA, secondary), diglycolamine (GA, primary), diisopropanolamine (DIPA, secondary) and methyldiethanolamine (MDEA) are all used to sweeten natural gas. The primary amines are the strongest bases, and thus form the strongest chemical bonds with carbon dioxide and hydrogen sulfide. Table 1 shows some physical properties of different amines, 3.5.4.4table1.doc: Physical properties of amines The chemical reactions take place in the contactor, where the acid gas components react with the amine and form a salt. These reactions are reversible so that amine can be regenerated. The basic reactions with a primary amine are shown below. With hydrogen sulfide: 2RNH, + HS © (RNH;),S (3) (RNH,),S +H,S © 2RNH,HS (4) where R is an organic group in all amines. These reactions are direct and fast in contrast to reactions with carbon dioxide where carbon dioxide first must form carbonic acid with water before reacting with amine. The formation of carbonic acid is a slow reaction. When reacting with amine, two Products may result. The formation of carbamate is shown in (5), and the formation of carbonate and bicarbonate is shown in (6) and (7), respectively: 2RNH, +CO, RNHCOONH,R (5) 2RNH, +CO, + HO © (RNH,),CO, (6) (RNH,),CO, +CO, + H,O «= 2RNH,HCO, (7) At high pressure and/or low temperature, all these reactions proceed to the right, and at low pressure and/or high temperature, these reactions proceed to the left. In this way, the absorption process and the regeneration process are controlled. The selection of amine will affect the reaction rate since the 326primary amines are more reactive than the secondary and the tertiary. Below, some applications of the various amines are presented. Monoethanolamine (MEA) is very reactive and can remove both carbon dioxide and hydrogen sulfide simultaneously but only non-selectively. With MEA it is even possible to reduce the hydrogen sulfide content in the gas to 1 ppm. It is usually used in concentrations of 10-15 percent to avoid extensive corrosion. MEA can be regenerated except when COS and CS> are present. These sulfur compounds react irreversibly and results in a considerable loss of amine. Thus, MEA is used for intensive purification of gas containing no COS or CS2. For many years, MEA was the most common amine, however more efficient ones have now replaced it for most applications. Diethanolamine (DEA) is weaker than MEA and the solution strength can range up to 35 percent. DEA has become the most common solvent for processing very sour natural gas at relatively high pressure. It can be used in the presence of COS and CSp, and hence the amine solution can be regenerated without losses. In addition, it is not that corrosive as MEA. Diglycolamine (DGA) is similar to MEA but the acid gas pick-up per volume of amine is much larger because concentrations up to 50-70 percent can be used, due to the low volatility of DGA. This results in a reduced circulation rate and improved economics of the process, Degradation products from COS and CS, can be regenerated. DGA systems can be operated in cold climate due to a low freezing point of the solution. The diisopropanolamine system (DIPA) can be used selectively to move either hydrogen sulfide or carbon dioxide, and the concentration of the solvent is usually between 30 and 40 percent. By lowering the pressure the DIPA will preferentially remove hydrogen sulfide, but when increasing the pressure, also carbon dioxide will be removed together with the hydrogen sulfide. Substantial amounts of COS can also be removed easily, and the process is non-corrosive. This process, which is used in the Shell ADIP process, is therefore mainly used for processing of refinery gas containing cos. Methyldiethanolamine (MDEA) also allows a selective absorption of hydrogen sulfide in the presence of carbon dioxide. It is commonly used for natural gases requiring a highly selective solvent but not the removal of COS and mercaptans. Removal of carbon dioxide can be favored in the presence of additives. Due to very low volatility, concentrations up to 60 percent can be used. MDEA has become the most frequently used amine for acid gas treating and is more expensive than most of the other amines. As with DGA systems, MDEA can also be operated in cold climate due to a low freezing point. New amine based solvents have also been introduced to the market, the so-called Formulated Solvents. These solvents are proprietary and are created by blending different components with standard amines. Another group is the Sterically Hindered Amines, which are formed to accommodate a specific process requirement. 2.2.2. Potassium Carbonate Process The potassium carbonate process is similar to the amine processes but the solutions are usually more concentrated. By using the alkaline salt hot potassium carbonate (KxCOs), carbon dioxide and hydrogen sulfide are removed in an absorption process. This process requires that carbon dioxide is present in the natural gas. One difficulty is to remove enough hydrogen sulfide to satisfy the gas Bedquality requirements. The potassium carbonate process is somewhat effective in removing COS and CS The solution concentration for a potassium carbonate system is limited by the solubility of the Product potassium bicarbonate (KHCOs) in the rich stream and therefore the whole system is operated at high pressure to increase the solubility. The hot potassium carbonate solution is extremely corrosive and to decrease this, inhibitors are added. Several proprietary processes have been developed based on this technology together with a catalyst. These catalysts increase the reaction rates and decrease the corrosion in the system, 2.3. Physical Solvents In a physical solvent process the acid gases are solved in instead of reacting with a solvent. The solubility of hydrogen sulfide and carbon dioxide vary with partial pressure and temperature, and therefore these processes are sensitive to changes in these operational parameters. The sour gas is flowing counter-current to the solvent in an absorber. To regenerate the rich solvent, it is usually flashed down to a pressure near atmospheric. This causes the acid gas partial pressure to decrease, and the gases evolve to the vapor phase and are removed from the solvent. In addition to flashing, some processes require stripping and some heat to regenerate the solvent. Some processes reboil the solvent in the regeneration step. Organic solvents are used to absorb the gas. They have usually a high affinity for heavier hydrocarbons, and this may result in a loss of these components because they are solved together with the acid gases and can not be recovered. Consequently, a physical solvent process should be avoided for gases with a high content of propane plus fractions. Usually, physical solvent processes are used when the partial pressures of the acid gases are above 3.5 bar. They are also appropriate when only bulk removal of the gases is required or when a selective removal of hydrogen sulfide is needed. Most physical solvent processes have been licensed by the company that developed them. The main processes are the Selexol, Rectisol, Sulfinol, Fluor Soivent and Purisol processes. Selexol: The Selexol process is developed by Allied Chemical Company and uses polyethylene glycol dimethylether as a solvent. The process removes sulfur compounds selectively, but the level of carbon dioxide can also be reduced by approximately 85 percent. Therefore it is also used for bulk removal of carbon dioxide. For such bulk removal, only one absorption step is required. The regeneration of the solvent is done in several steps where the solvent is flashed at successively reduced pressure. For selective removal of hydrogen sulfide, two such cycles are necessary. The operating temperature in this process varies from ambient to -18 °C. Water and heavier hydrocarbons are highly soluble in polyethylene glycol diethylether. Therefore, the Selexol process also offers dehydration of the gas and removal of heavier hydrocarbons. Dehydration of natural gas is discussed in section Dehydration of Natural Gas. Reetisol: In the Rectisol process developed by The German Lurgi Company and Linde A.G., methanol is used as the solvent to absorb the acid gases. Due to the high vapor pressure of methanol, the absorption step is carried out at low temperature (-30 to -80 °C). This also makes the Process complex with the need for cryogenic process stages. Heavier hydrocarbons are absorbed easily in methanol, and to reduce this, water can be added to the solvent. However, this process 328offers considerably lower energy consumption than the amine processes. Another advantage is that all impurities are removed in a single stage, even water can be removed to a very low level. The absorber, where the acid compounds are solved in the methanol, is split in two. The methanol is fed at the top of the lower stage of the absorber at about ~75 °C and the temperature increases down the absorber due to the heat of absorption. At the outlet of the lower section, the temperature is about -20 °C. In this section, almost all the hydrogen sulfide together with other sulfur components and the bulk of carbon dioxide are removed. In the upper section, the remaining carbon dioxide and organic sulfur is removed by very well purified methanol with an inlet temperature of -60 °C. The methanol from the upper and lower section is then regenerated by two successive pressure reductions down to 1 bara and 0.2 bara, respectively. The Rectisol process is mainly applied to purify the gas before synthesis gas production and LNG (liquefied natural gas) production. Fluor Solvent: In the Fluor Solvent process owned by Fluor Daniel, Inc., propylene carbonate is applied to absorb carbon dioxide and hydrogen sulfide. Propylene carbonate is a non-corrosive liquid with low viscosity. Also, this process offers selective removal of hydrogen sulfide, and can remove Cr, fractions, COS, CS2, $O2 and water from the natural gas as well. The Fluor Solvent process is usually used only for bulk removal of carbon dioxide because it requires uneconomical design and operation to obtain pipeline quality. It is operated well below ambient temperature to increase the solubility of the components in the gas, and thus a refrigeration unit for the solvent is usually necessary. Purisol: The Purisol process, also developed by The German Lurgi Company and Linde A.G., is based on N-metyl-2-pyrrolidone (NMP) as a physical solvent. This process can be used for selective sweetening of natural gas. 2.4 Combined Processes By combining the advantages of the physical solvent processes with the chemical solvent processes, success has been obtained. Many processes are now developing, and two have been proven successfully. These are the Sulfinol process where a blend of a physical and a chemical solvent result in better performance, and the Hi-Pure process which is a combination of a potassium carbonate process and an amine process. Sulfinol: The Sulfinol process developed by Shell, is a combination of a chemical and physical solvent process. The Sulfinol solution consists of a mixture of the physical solvent sulfolane (tetrahydrothiophene dioxide), DIPA or MDEA (see Chemical Solvents) and water. The proportions of the mixture may vary with the properties of the feed gas. Sulfolane is an excellent solvent for sulfur compounds, COS and CS2, while carbon dioxide, hydrocarbons and aromatics do not solve easily in sulfolane. The chemical solvent acts as a secondary treatment to remove both hydrogen sulfide and carbon dioxide. As a whole, the Sulfinol process solvent offers a high solvent loading of hydrogen sulfide especially at high partial pressure of hydrogen sulfide. This process is usually employed for gas with a hydrogen sulfide and carbon dioxide ratio greater than 1:1 or where it is not required to remove carbon dioxide to the same extent as hydrogen sulfide. Hi-Pure Process: In the Hi-Pure Process, a potassium carbonate process is combined with an amine process. In the first step, the natural gas is contacted with the potassium carbonate solution and secondly with the amine solution, With this process, low levels of both carbon dioxide and hydrogen sulfide in the outlet gas stream can be obtained. 3292.5 Gas Permeation Gas permeation is a relatively new method for removal of acid gases. By use of membranes, carbon dioxide, hydrogen sulfide and water can be separated from the feed stream. The technology is still under development and has some advantages over typical amine processes (see Amine Processes). The system is usually simple and small, well suitable for an offshore platform, as well as the ‘membrane system requires little or no need for human supervision and maintenance. The membrane technology is based on gas diffusion through a permeable membrane where the driving force is the difference in pressure over the membrane. Through thin polymeric films, different gases permeate with different rates due to differences in diffusivity and solubility of the gas components. The membrane surface consists of thousands of capillaries. The polymeric membranes can be classified either as glassy or as rubbery. Glassy membranes are generally more Permeable towards smaller molecular size gases while rubbery membranes allow heavier gases to penetrate more easily, In the separation process, feed entering the high-pressure side loses the fast-permeating components such as carbon dioxide to the low-pressure side of the membrane. An illustration of a membrane system is given in figure 2. The pressure on the low-pressure side is usually 10-20 percent of the pressure on the other side, which means that the permeated natural gas is only available at the lowest pressure. The slow-permeating gas components are enriched in the residual gas stream. The membrane process is considered a non-equilibrium process, because if the system was allowed to g0 to equilibrium, permeation would continue until the pressure and concentrations of the gases on both sides were equal. 3.5.4.4figure2.ipg : A simplified membrane system for gas separation A separation factor indicates the membrane’s ability to separate two gas components. The separation factor for carbon dioxide over methane is from 10 to 50 depending on the type of membrane material, feed composition, pressure and temperature. This means that for equal partial pressures of the two components and a separation factor greater than 1, carbon dioxide permeates faster through the membrane than methane. The separation factor for hydrogen sulfide over methane is higher, from 40 to 60. A membrane designed to remove carbon dioxide to pipeline specifications will also remove most of the hydrogen sulfide and some water from the feed gas, A commercial membrane unit can reduce the hydrogen sulfide content to around 100 ppm. However, this means that additional processes are necessary to remove the rest of the hydrogen sulfide since gas specifications require that the content of hydrogen sulfide in natural gas must be only a few pm (see Introduction). A membrane can also remove substantial amount of water vapor saturated i the natural gas. A separation factor as high as 500 with reference to methane can be obtained, resulting in efficient removal of water vapor to low ppm levels even at low partial pressures, but also resulting in a relatively high methane loss. Other methods for removal of water vapor from natural gas are presented in Dehydration of Natural Gas, Usually, membranes are good for bulk removal of acid gases. At low concentrations of the acid gases, membrane technology is inferior to other processes. This is because, at low concentrations the partial pressure of the components are small and the pressure driving force of the process is hhence reduced. In such situations, the membrane technology is successfully combined with chemical solvent systems (see Chemical Solvents). 10 3362.6 Other Methods Low-temperature distillation is another method for separating carbon dioxide present at high concentration levels. In the Ryan-Holmes process, a light hydrocarbon is added to the feed to avoid crystallization of the carbon dioxide at low temperatures. In the first column, methane and carbon dioxide are separated from the LPG (liquefied petroleum gas) fraction. Secondly, carbon dioxide is separated from methane and the light hydrocarbon additive. Methane is then separated from the additive that is recycled, Another distillation process is the Controlled Freezing Zone (CFZ) process, which allows crystallization of carbon dioxide. Carbon dioxide is produced at the bottom of the distillation column and methane at the top. This process has not been demonstrated industrially. 3 Direct Conversion of Hydrogen Sulfide to Sulfur After separation of hydrogen sulfide from the natural gas by use of methods described in Removal of Carbon Dioxide and Hydrogen Sulfide, the resulting hydrogen sulfide containing acid gas can be incinerated, or converted to sulfur so that hydrogen sulfide release to the atmosphere is avoided. One possibility is to flare the hydrogen sulfide and produce SOz, but there are also regulations for release of SO; to the atmosphere. Another possibility is direct conversion of hydrogen sulfide in the produced natural gas flow to sulfur by chemical reactions. These processes are generally based cither on reactions of hydrogen sulfide with oxygen or SOs, giving elemental sulfur and water. For large flow rates of natural gas, it is more common first to absorb the hydrogen sulfide by using chemical or physical solvents, and secondly use the direct conversion processes on the gas liberated from the regeneration step. In this section, some of the various direct conversion processes will be presented briefly. Claus Process: The Claus process is designed to treat high concentrations of hydrogen sulfide. The chemical reactions involve partial oxidation of hydrogen sulfide to sulfur dioxide (SO) and a catalytic reaction of hydrogen sulfide and SO. to produce elemental sulfur: H,S +150 > SO, + H,0 (8) 2H,S + SO, -> 38 +2H,0 (9) The first stage converts 1/3 of the hydrogen sulfide to SOx, by burning the acid gas flow with air. This exothermic reaction provides SO2 for the next step in the process. Multiple catalytic stages are used to achieve a high conversion of hydrogen sulfide, Between each stage, a condenser condenses the sulfur vapor and separates it from the main flow. By means of three multiple stages a conversion of 97 percent can be obtained. The product gas from the Claus process is sent to a "tail-gas treating" to remove the rest of the sulfur components before it is released to the atmosphere. The Sulfreen, Clauspol, Superclaus and Cold Bed Absorption processes provide normally a total conversion of at least 99.5 percent. Several modified Claus processes are developed to increase the effectiveness and improve the economy of the process. SCOT Process: The SCOT process can be used in addition to the Claus process. In the presence of hydrogen, remaining sulfur compounds and free sulfur from the Claus process are converted into hydrogen sulfide over a cobalt/molybdenum catalyst in a reduction unit. The reaction product is cooled and hydrogen sulfide is absorbed by an amine unit and recycled to the Claus unit, An overall yield of 99.99 percent sulfur recovery can be obtained with the SCOT process. au 33LOCAT Process: In the LOCAT process a dilute solution of iron ions in water reacts with hydrogen sulfide and produces elemental sulfur: H,S + 2Fe* > 2H* +S+2Fe™ (10) 0.50, + H,0+2Fe* —» 00H)" +2Fe* qi) The iron solution in reduced ferrous form is continuously regenerated by bubbling air through the solution as shown in equation 11. The sulfur precipitates out of the solution and is removed from the reactor by gravity, centrifuging or melting with a portion of the reagent. Due to side reactions, a small portion of the iron agent is lost with precipitated sulfur. The LOCAT process is used both for tail-gas treating and directly as a gas-sweetening unit. Other iron-based catalytic processes cartied out in the liquid phase are the Sulfint and Sulferox processes. Stretford Process: In the Stretford process hydrogen sulfide is oxidized by washing the gas flow with an aqueous solution of sodium carbonate, sodium vanadate and anthraquinone disulfonic acid, letting the solvent flow counter-current to the acid gas in an absorber. The reduced solution is flashed to remove hydrocarbons and is then re-oxidized and pumped to the absorber again, The sulfur is carried in a thickening slurry and transported to alternative sulfur recovery methods. Other vanadium-based methods for conversion of hydrogen sulfide are the Unisulf and Sulfolin processes, IFP Process: In the IFP process, a mixture of hydrogen sulfide and SOz flows counter-current to a liguid catalyst in a packed tower. The products are water and sulfur, The overall reaction is: 2H,S +SO, > 2H,0+3S (2) This process is operated above the melting point of sulfur so sulfur is recovered as a liquid in the bottom of the tower. Some of it is bumed to produce SOz required to remove more hydrogen sulfide. Analyzer equipment controls the important ratio of hydrogen sulfide to SO>, 4 Dehydration of Natural Gas The presence of water raises problems for the gas processing and transport of gas. To avoid formation of hydrates and hence blocking of gas pipelines and process equipment, dehydration of the gas is necessary. How hydrates are formed, is presented in Natural Gas Hydrates. Dehydration will also prevent condensation of water, which can cause severe corrosion. If in addition, carbon dioxide or hydrogen sulfide is present in the gas flow, acids will form in the presence of the condensed water. Dehydration of natural gas is the removal of water that is associated with the gases in vapour form. This operation will lower the dew point of the gas; the temperature at which water condenses from the gas flow. Different methods are available; dehydration by absorption, adsorption, gas permeation and refrigeration, 4.1 Absorption 12 332Dehydration by absorption is usually applied for removal of large amounts of water. Several glycols can be used as an absorbing liquid, but trietylene glycol (TEG) is almost always used because it can be regenerated with a high purity and is less expensive than tetraethylene glycol (TREG), which could have been an alternative. In the absorption process, the water vapor in the gas flow is dissolved in a glycol, while gas and liquid flow counter-current in a glycol contactor. By flowing counter-current, the wet gas will first contact the richest glycol, which contains most water in solution. As flowing up the absorber, the leaner glycol is able to absorb additional amounts of water vapor from the gas. This counter-current flow enables gas to transfer a significant amount of water to the glycol and still approach equilibrium with the leanest glycol concentration. A flow diagram of the absorption dehydration process is shown in figure 3. 3.5.4.4figure3 pdf: A flow diagram for a glycol dehydration process A glycol contactor consists of typically six to eight trays depending on the required dew point, and the temperature is usually kept below 38 °C to avoid excessive loss of glycol, and to reduce the amount of water in the gas phase. The glycol will also absorb heavy hydrocarbons in the gas, and therefore the gas must pass trough a separate inlet gas scrubber before entering the absorber. For large volumes, filter separators can be used as inlet scrubbers to reduce the glycol contamination by heavy hydrocarbons, corrosion inhibitors and treating chemicals, and thus increase the life of the glycol charge. By removing contaminants in the scrubber, also foaming in the contactor and fouling of surfaces on the inside of equipment are avoided. After the absorber, the dry gas is heat exchanged with the incoming glycol in order to cool the glycol and hence obtain an increased absorption capacity and decrease the tendency of glycol to flash in the absorber. The rich glycol from the absorber is regenerated in a reboiler unit where the water is stripped from the glycol, but first the glycol goes through a separation process. The glycol solution is heated and sent to a low-pressure three-phase separator where any hydrocarbon liquids and the entrained gases are removed. The glycol is then filtrated in a cartridge filter to trap any solid particles followed by a charcoal filter to retain chemical impurities. The filters help minimize foaming and sludge build-up in the reboiler. Afterwards, the rich glycol is heated, by heat exchanging with lean glycol to absorber, before flowing to a still column mounted on the reboiler. In the still column operating at near atmospheric pressure, the glycol falls through a packing and is heated by the vapors being boiled off the liquids in the reboiler. The temperature in the reboiler is usually between 170 and 204 °C to provide the heat necessary for the still column to operate Because of the large difference in boiling point between water and glycol, a sharp separation can be achieved with a relatively short still column, The falling glycol with water gets hotter and hotter, and water and some glycol will flash from the liquid. Thus, as the glycol falls through the packing it becomes leaner in water. If a very lean glycol is required, stripping gas may be used. By injecting @ small amount of wet gas from the inlet flow into the reboiler where the pressure is lower and temperature is higher, the gas becomes highly unsaturated and can be used to strip liquid water out of the glycol. After reboiling, the lean glycol is transported to a surge tank that allows expansion of the glycol before the glycol is pumped back to the absorber to complete the cycle. This process leads to a water content in the dehydrated gas usually between 20 and 35 g per 1000 standard m? depending on the operational temperature and pressure. Some of the glycol is lost with the water vapor and some may react and form heavy molecules that are separated out, and thus the glycol stream must be recharged continuously.4.2 Adsorption Solid bed dehydration is commonly used for dehydrating gases in cryogenic processes, The initial cost for such a dehydration unit generally exceeds that of a glycol unit, but on the other hand, solid bed dehydration results in a very low dew point of the gas, which is required for cryogenic processes such as liquefied natural gas (LNG) production. It is also applied when large changes in flow rates occur. In contrast, the solid adsorbent is sensitive to poisoning with liquids and other impurities in the gas, and the pressure drop over such a unit is high. Different adsorbents can be used for dehydration of natural gas. By using bauxite, a residual water content of about 1 ppm can be achieved. Heavy hydrocarbons are also adsorbed but can not be desorbed during regeneration, so these components must be avoided. Adsorption on a silica gel may result in water content of about 10 ppm, and the silica gel is easily regenerated. An expensive method is to use molecular sieves (zeolites) to adsorb water. With this method it is possible to adsorb the water molecules while excluding hydrocarbons resulting in a water content down to 0.03 ppm vol. The use of molecular sieves for removal of hydrogen sulfide and water is also presented in Solid Bed Adsorption, To dry the gas on a continuous basis, multiple solid beds are used in cyclic operation. Three cycles ‘must altematively be performed in each dehydrator; an adsorbing cycle, a regeneration cycle and a cooling cycle. A typical two-bed cycle might have eight hours of adsorption, six hours of heating and two hours of cooling. Before the adsorption cycle, the gas passes through an inlet separator where liquids, entrained mist and solid particles are removed. In the adsorption tower, the wet gas flows downward and light hydrocarbons first saturate the bed. Water is then starting to replace the hydrocarbons in the upper layers of the bed, and continues to displace the previously adsorbed hydrocarbons down the column, The tower must be switched from the adsorption cycle to the regeneration cycle before the whole bed is saturated with water. To regenerate the solid adsorbents the tower is both heated and cooled. First, usually a slipstream of dry process gas is heated to between 230 °C and 315 °C in a high-temperature heater, and routed to the tower to remove the previously adsorbed water on the solid adsorbents. Due to the high temperature, the water will evaporate and be absorbed by the natural gas flowing downward in the tower. Further downstream, the gas is then cooled in a regeneration gas cooler where water will condense. The water is eventually separated out, and the cool gas is recycled and dehydrated. A hot bed will not sufficiently dehydrate the gas. To retum the solid bed to normal operating temperatures, the bed is also cooled (to between 38 °C and 50 °C) by flowing cold gas, usually upward, through the tower. 4.3 Other Methods Partial dehydration of natural gas can be done by cooling and compression of the gas. When the temperature decreases in saturated hot gas, the water vapor will condense. This cooling process ‘ust cool the gas down to the lowest possible temperature at the prevailing pressure of the system, to prevent further condensation of water. 14 33YDehydration by permeation, i.e. using membranes can prove to be more economical and compact than glycol dehydration for some systems. The use of membranes for removal of water as well as carbon dioxide and hydrogen sulfide is discussed in Gas Permeation. Bibliography Amold K. and Stewart M. (1988). Surface Production Operations, Volume 2: Design of Gas- Handling Systems and Facilities. Houston, Texas: Gulf Publishing Company. (This book gives an introduction to acid gas treating and dehydration and design of such systems, as well as presenting processing of natural gas and equipment.] Astarita G., Savage D.W. and Bisio A. (1983). Gas Treating with Chemical Solvents, John Wiley & Sons, Inc. [This book presents theory, methods for analysing of gas treating systems and applications. Chemical solvent processes are emphasised.] Gas Processors Suppliers Association (1998). Engineering Data Book, Volume I and Il, Tulsa, Oklahoma: Gas Processors Suppliers Association. [This data book presents equipment and methods for processing and purification of natural gas. Along with many tables and charts, this data book also contains methods of calculation.] Kohl A. and Nielsen R. (1997). Gas Purification, Houston, Texas: Gulf Publishing Company. [In a comprehensive way, this book covers all methods for purification of natural gas.] Scott K. and Hughes R.(1996). Industrial Membrane Separation Technology, Glasgow: Blackie Academic & Professional. [This book covers the theory and applications of membrane technology in general. 15 335-Monoethanol- Diethanol- Digiyeo! Dilsopropanol- Methyldiethanol- Formula HOGHINH; — (HOGH);NH HOG H):NH2 (HOCHQ.NH— (HOC;H)2NCHs Molecular weight 61.08, 105.14 105.14 133.19 119.16 Boiling point @ 1 atm, °C 170 269 221 248.7 247 Freezing point, *C 105 280 1125 2 2B Density @ 15.6°C, kg mi? 1016 1089 1087 : 1040 Critical pressure, KPa (abs) 5985 3273 3m. 3770 : Critical temperature, *C 350 492.1 402.6 3992, ‘Thermal conductivity @ 20°C, 0.256 0220 0.209 0275 womscy! Specific heat @ 15.6°C, 255@20C 251 239 289@30C 2.24 Wag °c)" Viscosity, mPa s M1@2C —-350@ 20 H@1GC —870 @ 30. L3d0* mt st (ar 90 we ec solution) BoeCompressor Feed: t-—> Residue: Gas A+B Gas A Membrane Permeate: Gas B Figu red 339yuri ysey eal 86 usey caval. pinby 224 4 Sug P| veaanes ro2K(6 we, ‘eto, 339J.Leviand: Fysikalsk absorpsjon i platetarn Eksempler pa gassrensing og mulige prosesser Begreper og enheter Absorpsjon i flere trinn og i motstram Problemstilling og valg av verdier Matematisk modell Grafisk illustrasjon og losning Analytisk lesning, numerisk losning Platevirkningsgrad Plateavstand og tarnheyde Driftsomrade for plater Gasshastighet og tarndiameter Sammenligning platetarn og pakket tarnFysikalsk absorpsjon i platetarn Med gassrensing forstis vanligvis & fjerne en eller flere uonskede komponenter fra en hovedmengde ass. Eksempler System Konsentrasjon Rensekrav CO, ieksos 310% - * naturgass (Norge) <10% <3% "6 "LNG <3% <500 ppm H,S i naturgass (Norge) 30 ppm(?) <4 ppm SO, i eksos 0.1-0.3 % 90 % fjerning NO, ieksos 50- 1500 ppm - Vann i naturgass 600 ppm 30 ppm VOC fra tanker <100% - Siden VOC kan utgjore opptil 100 % av det som skal fjernes faller det muligens utenom definisjonen av gassrensing, men prinsippene som benyttes for 4 fjerne utslippene er dels de samme. Mulige renseprosesser Absorpsjon: Uonsket komponent overfores til en veeske " ” " "fast overflate Adsorpsjo1 Membranseparasjon:". "—_anrikes gjennom membran Destillasjon: Gassen mA kjoles til vaeske som sé destilleres Kjemisk reaksjon: Aktuelt for noen komponenter, feks. H,S +0, = SO, Adsorpsjon og adsorpsjon har det til felles at separasjonen oppnis ved stofftilsats, som normalt mA regeneres. Metodene ovenfor kan kombineres, det vanligste ekspemplet er absorpsjon med kjemisk reaksjon i vaskefasen. Det normale er at den uonskede komponent ma behandles videre med tanke pa avhending, Unntak for eksemplene ovenfor er vann og VOC. Valg av separasjonsprosess avhenger bade av system og av mengder og konsentrasjoner. Dette skal ikke behandles her. I det videre behandles bare absorpsjon, som oftest vil veere det naturlige valget Det omvendte av absorpsjon, dvs overfaring av en komponent fra en veeske til en gass kalles desorpsjon eller stripping. 242Innledende begreper ‘Ved absorpsjon - som ved endel annen separasjon - er folgende forhold avgjorende for prosessutformingen: + Kapasitet: Dvs hvor mye varske trenges for absorpsjonen. Kapasiteten er forst og fremst gitt av likevekten, men for praktisk kapasitet ogs4 av hastigheten av masseoverforingen. + Selektivitet: Betydningen er selvinnlysende, og formelt kan selektivitet defineres som mengde fjemet av onsket komponent/ mengde fjernet totalt. Strommer (mengde pr tidsenhet): Her benyttes kmol/s eller kmol/h for separasjonsberegningene, mens en for de hydrodynamiske forhold ma benytte masse, dvs kg/s eller kg/h. Mol og kmol er et mal for antall molekyler. Sammenseining: Molbrok benyttes. Enklest forklart som 1/100 av mol%, Molbrak ganger totalstrom i kmol/s gir stram av komponent i kmol/s. Molbrak i vesken gis symbol x og i gassen symbol y. Likevekt; Faselikevekt mellom komponent i gassen og i vesken, definert som forholdet mellom molbrak i gassen av komponent "i" og molbrak av samme komponent i vaesken, dvs y/x; . Likevekten er gitt av molekylenes egenskaper, hvilket er arsak til at kmol og molbrok er de foretrukne enhetene. Likevekten ma kjennes (dvs kunne beregnes) for at absorpsjon kan beregnes. Det finnes en rekke modeller som kan veere aktuelle avhengig av type molekyler og temperatur- og trykkforhold i systemet. For lave molbroker vil mengde last vere proporsjonal med trykket, og likevekten uttrykkes da oftest ved en Henrys lovs konstant H ( P er totaltrykket) ys P=H,x; Hvis forholdene er slik at H ikke er konstant ved en gitt temperatur er det bedre & benytte mer fundamentale modeller enn 4 finne formler for H, iallfall om en selv skal modellere likevekten. Loseligheten (x) oker med P, og synker oftest med temperaturen. Laseligheten er lav uansett, Fysikalsk absorpsjon: Det vi har nar gassen loser seg i vesken uten & reagere videre, eller uten at vi behover 4 ta hensyn til reaksjonen. Absorpsjon med kjemisk reaksjon: For & bedre kapasitet og selektivitet kan en tilsette kjemikalier til vaesken slik at den komponent som absorberes bindes kjemisk. Dette kan ogsa ake hastigheten for absorpsjonen, Pga regenereringen av vaesken mA ikke den kjemiske binding vere s& sterk at den ikke kan loses igjen rimelig greit, f.eks. ved 4 oke temperaturen. Flertrinns absorpsjon i motstrem Separasjon ved bruk av flere trinn i motstrom er det som benyttes ved de fleste tekniske separasjoner. For & illustrere prinsippet for absorpsjon er det valgt her & se pa det motsatte, nemlig stripping, dvs fjerning av en last komponent. Systemet vart er CO, last i vann, slik vi har det i en flaske soda, selters, Farris, og vi onsker A fjeme denne CO,-en, Hvis vi heller seltersen/sodaen opp i en enorm beholder hvor der er Luft, og rister ved hjelp av et eller annet slags maskineri, vil vi f& likevekt mellom CO, i gassen (luften) og i vesken. Siden der er s4 mye gass i forhold til vaesken vil da molbraken av CO, bli lav i gassen og dermed i veesken. Vi kan derfor i prinsippet klare oss med ett likevektstrinn, men p& bekostning av & benytte en aldeles uckonomisk stor mengde med strippegass. For absorpsjon ville det bli omvendt, vi kunne benytte en 343sveert stor mengde vaeske relativt til gasssen, og klare oss med ett trinn, {a oss i stedet elle sodaen over i en beholder f.eks. dobbelt sA stor som flasken, riste og fa likevekt. Det vil da fortsatt vase relativt hay molbrok av CO, i begge faser. S& heller vi sodaen fra forste beholder over ien like stor neste beholder, rister pA nytt og far ny ikevekt. Vi har né kanskje halvert molbrakene i gass og vaeske sammenlignet med det farste trinnet. SA gjor vi det samme en gang til, heller sodaen fra trinn 2 over i en ny lik beholder, rister og far ny likevekt med kanskje a halvparten av molbroken i trinn 2. Slik kunne vi holde pa til vi hadde fjemet onsket mengde CO, fra sodaen. I en figur vil det se slik ut: Reedy. Ys OG 2 = Asx, % asx, Af 1 2 3 Figur 1. Trinnvis fjemning av CO, fra soda med porsjoner av strippegass Vi vil ved denne fremgangsmaten, med stadig ny lufimengde (strippegass), redusere Iuftmengden betydelig, ford vi ikke behaver & bruke s& my luft at all luften har den endelige lave likevektsverdien av CO;, En slik fremgangsméte betegnes krysstrom, siden sodaen passerer gjennom apparaturen og stadig "krysses" av ny luft. For absorpsjon ville dette tilsvare & benytte frisk veeske pa hvert trinn, Vi ser at vi kan redusere luftmengden ytterligere ved 4 la luften g& motsatt vei av vesken, Tar vi Feks.luften i beholder 2 i figur 1 ovenfor (med molbrok y,) og ersattr luften i beholder 1 (som har rolbrok y,) med denne luften, kan lufien fra beholder 2 ake sittinnhold av CO, yiterligere. Den kan ikke ta opp si mye CO; som isk lut, men siden vi nf lar samme luft g4 gjennom apparaturen @ reduseres lufiforbruket drastisk. Dette er motstram, og et indikert pA figur 2 for de samme tre kolbene som i figur 1. 4 2 3 Figur 2. Stripping ved bruk av motstram. For absorpsjon blir det ingen forskell nd, gass og vaeske gér i motstrom enten det er stripping eller absorpsjon. BoyAbsorpsjon i tarn/kolonne Platetarn Ved en teknisk gjennomfaring ryster vi selvfolgelig ikke i mange sma kolber, men bygger trinnene sammen i én apparatur med innlop for gass og veske i motsatt ende, Denne apparaturen setter vi pi hoykant, s4 vaesken kan renne nedover og gassen stige oppover, og innretter oss sé vi far en rekke steder i apparaturen hvor gass og varske blandes og separeres igjen. En slik apparatur kalles tam eller kolonne, og trinnene inne i apparaturen kalles plater eller trinn. Figuren under viser oppbyggingen i prinsipp. Figur 3. Absorpsjonstar med N = 5 plater. Detalj av klokkeplater til hayre.(Lydersen 1983) Allerede her kan vi innse at antall plater vi benytter har med graden av separasjon 4 gjore, mens diameteren har med hvor mye gass og veske vi hindterer. Platene kan vere aller enklest plater med hull, eller det kan vere skalte ventilplater, eller Klokkeplater. De forste typene er vist pa figur 4, klokkeplater er visti figur 3.. Valve Cy Tray floor {fire eg en top for maxioum opening TZ catming space Figur 4. Plate med hull, og ventiler i ventilplater (Lydersen 1983) ysPakket tarn Deter ikke egentlig nodvendig at motstroms-separasjonen skjer pA distinkte steder i apparaturen, Vi kan i stedet la kontakten gass-veeske skje kontinuerlig gjennom témet. Vi ma da sorge for at det hele tiden er god kontakt mellom fasene, Det skjer ved a velge en sdkalt pakning, som har stor overflate kombinert med apen struktur for & holde trykktapet lavt. Pakningen kan bestd av lose biter (tete eller hullvegegede sylindre, sadler o.a. ) av vanligst dimension 25-75 mm, eller av sammenhengende, sékalt strukturert pakning, Denne kan vere kurrugerte og hullede tynne plater brettet sammen, eller kan vere flettet tynn staltr8d i matter, Eksempler er visti figur 5 under. MELAPAK in segmented form, Rashigringer b. Berl sedler Intelox-sadler c. Pail-ringer ‘Super-Intalox sadler Figur 5. Eksempler pa fyllegemer i los pakning og p& strukturert pakning Pakkede tm blir behandlet mer senere i kurset, og blir med ett unntak ikke omtalt mer her. Vanlige problemstillinger ved beregning Ved beregning av pakrevet apparatur for en gassrensing er folgende ofte gitt: + Mengde gass inn + Konsentrasjon av den komponent som skal fjernes * Krave tl rensing, dvs konsentrasjon i gassen etter rensing. Om den ikke er gitt mA vi velge en verdi Det som gjenstar for en kan foreta den egentlige timberegning er & fastlegge mengde 0g renhet av veesken til timet, 3MGSiden versken oftest ma regenereres vil konsentrasjonen av det stoff som absorberes sjelden verre helt 0 i inngaende absorpsjonsv'ske. Regenereringen er del av prosessutformingen, og vi kan til en viss grad velge konsentrasjon etter regenereringen, Denne konsentrasjonen ma iallfall vere godt under det som tilsvarer likevekt med konsentrasjonen i utgaende gass, typisk 50-70 % av likevektsverdien, Nedre grense for mengde vaeske fs ved & betrakte et tirn som har uendelig mange trinn, Endringen over hvert trinn blir.da uhyre liten, og nederst vil det varre likevekt mellom gass inn og veske ut. For praktiske forhold velges gjeme en konsentrasjon i vesken ut som tilsvarer 50-80 % av det som tilsvarer likevekt med inngéende gass. En materialbalanse over absorbert komponent gir da mengde veeske som trenges. Strommene (mengde og sammensetning) ut og inn av timet er da kjent, og vi kan gd las pa 4 finne antall trinn som gir den enskede separasjon. Matematisk modell for plate-kolonne Det antas i utgangspunktet folgende for forholdene pa hvert trinn i tamet: + Fullstendig blanding pa trinnet, dvs samme konsentrasjon overalt i vaesken og overalt i gassen pa selve trinnet. + Likevekt mellom utgéende gass og utgiende veeske. Ligningene som beskriver det som skjer pa trinnet er da + Materialbalanser, 1 for hver komponent + Likevekter, 1 for hver komponent + Energibalanse, | totalt Molbroksummer, 1 for hver ukjent stram Vanligvis setter en opp balanser fortlopende fra en ende av témet, som vist pa fig 6: Figur 6, Fortlopende balanser over enden av tamet Ligningene er ulinezere og koblede, og ma generelt loses numerisk, Hvis det bare er én komponent som absorberes, og temperaturen ca ikke endrer seg, vil energibalansen og molbroksummene falle bort ( regner med bare én molbrak, s& er den andre gitt ved x,=I-x,) . En kan da illustrere det som skjer i et enkelt diagram med molbraker i henh.vis vaeske 3208 gass langs aksene. Likevekten blir én kurve i diagrammet, Materialbalansen (kun I na) vil ogsa gi kun én kurve, som kalles driftslinjen. Selve diagrammet kalles et McCabe-Thiele diagram. Materialbalansen rundt toppen to.m. trinn n regnet ovenfia blir med ord : Absorberbar komponent inn med gass + inn med vazske = absorberbar komponent ut i gass + ut | veske, og med symboler (komponentindeks er unodvendig nar bare én komponent betraktes, sh indeksen er for trinnene): YaGn t Koby = 1G, + XL som lost m.h.p y, gir driftslinjen som y = f(x) nar vi betrakter x og y som kontinuerlige variable. Likevekten kan vi skrive Ya = KiXq Som ogsé gir en kurve i diagrammet nar vi lar x og y variere fritt y (upg) Menll yske Figur 7. MeCabe-Thiele diagram for absorpsjon av én Komponent, (Qverst i timet er molbrokene lavest, og er derfor nederst i diagrammet) Hvis det absorberes lite s& gassmengden og veeskemengden endres lite gjennom thmet blir driftslinien rettlinjet, og driftslinjen kan da trekkes direkte mellom avre og nedre ende av tamet i diagrammet. Vinkelkoeffisenten for driftstinen er forholdet mellom vaeske- og gassstram (L4G), for ovrig enten Log Ger konstante eller ei, som det ses fia ligningen ovenfor. Ft MeCabe-Thiele diagram er velegnet ti &illustrere begrensningene vi hari valg av renhet (x) og mengde L, av absorpsjonsveeske, som vist under. Mengde L er gitt av vinkelkoeffisienten i diagrammet, eller en kan finne den fra molbrok ut og en materialbalanse over hele timet y deo n Sx Me ee Figur 8. Valg av renhet av inngdende vaeske (venstre), og av molbrok ut (hoyre) SUPSiden materialbalansen tas suksessivt over et trinn til, er det den som knytter forbindelsen fra trinn til trinn, Likevekten gjelder mellom molbroker ut pa selve trinnet, og knytter forbindelsen over selve ttinnet. Dette gjor at en kan finne antall trinn grafisk ved 4 trekke forbindelser mellom drifislinjen og likevekten, som vist ved trappekurven i figur 9: y Yo Figur 9. Grafisk losning av antail trinn Generelt vil en ved en slik grafisk losning ikke ende opp med et helt antall trinn. Nar sammensetning inn og ut og mengder er spesifisert kan en ikke vente at det skal tilsvare et helt antall trinn, Det en da gjor er & velge nazrmeste hele antall trinn, og sé far en litt justerte sammensetninger, En grafisk losning er begrenset til at vi har kun én komponent og at tamet er isotermt. Men det ses at for slike forhold far en som svar antall trinn. Ved en numerisk losning ma gjette pa forhand, feks. fra en grafisk forenkling, hvor mange trinn en trenger, mens en av de utgaende molbrokene ma vaere fri. Ligningene loses sé for det tippede antall trinn, og en ser om en fir den onskede separasjon. Hvis ikke md en endre antall trinn og/eller en annen valgt storrelse. Analytisk lesning for antall trinn Hvis likevekten ogsd er rettlinjet (som kan vere tilfelle nér vi har sma molbraker giennom hele tamet) kan en ff en analytisk losning for antall trinn N: inden EL Vi ~ Vo intea =I Yu - Yo I denne formelen betyr y* den y-verdi som svarer til likevekt med den virkelige x-verdi, alts& det som her ellers kalles Kx. Det ses at det bare er sammensetning og likevektsverdier som inngér, men siden typisk xy velges og dermed verske-gass forholdet, inngér det ogs& som vi matte vente Storrelsene som inngar i formelen er illustrert i MeCabe-Thiele diagrammet nedenfor. < ~5Figur 10. Verdier som inngar i analytisk losning for antall trinn En slik analytisk losning kan bl.a. brukes for & illustrere trender, ogs4 der det vil vere noe tilnermet, Eksempel {Lass som eksempel ta utvasking av SO, fra eksos med sjovann, og anta trykk I atm = 1 bar, femperatur 10 °C, 0g 0.2 mol % SO, inn, som er det samme som yy., = 0.002. Vide antar vi x,=0, e som it ¥4=0. Likevekten er y = 20x for disse betingelsene. Velges % til 70 % av likevekt med yyy fs x,=0.7°10*som igjen gir YyF1410° . 3) 90% av SO, fjemes, som gir y,=0.210°, og antall trinn N=4,4 b) 99% av SO, flemes, som gir y,=0.0210°, og antall trinn N=9.8 Platevirkningsgrad Modellen for kolonneberegninger kan virke grei, og er det op forsividt, men vi har ikke enn& Siekket om antagelsene holder. Hvis vi tenker pé strommen tvers over en plate pa kanskje et par meter i diameter virker det ikke helt rimelig at det skal verre fullstendig blanding i vaesken. Mindre intuitivt er det kanskje om det er likevekt eller ei, men for absorpsjon vil likevekt snarere vaere unntak enn regelen, A regne med de virkelige forhold blir for komplekst, og vi jor her som sA ofte ellers: Vi Dvatianten ket mellom modell og virkelighe i en platevirkningsgrad. Denne kommer i minst 2 varianter: * Total platevirkningsgrad : Antal vitkelige plater nadvendig/ teoretiske plater, hvor en ‘eoretisk plate er en som oppfarer seg i henhold til modellen, * Murphree platevirlaingsgrad som defineres for hver enkel plate, som endring i dampsammensetning over platen dividert p& endringen ved likevelt. Piatevirkningsgrader kan en enten finne i boker eller brosiyrer, eller en kan benytte empiriske Korrelasjoner for & ansid dem. Uansett innfores et element av usikkerhet, men selvfolgelig mindre enn om de neglisjeres,s& gjenoppsté, med bedre blanding og platevirkningsprader typisk 60-80 % nar viskositeten er rimelig lav.) Beregninger basert pa virkelig hastighet for masseoverforing vil derfor i utgangspunktet vere riktigere. Men om vi né antar at vi har en rimelig korrekt virkningsgrad, vil altsé beregning av teoretiske trinn kombinert med platevirkningsgrad gi antall virkelige plater vi trenger, eller den absorpsjon vi oppnar for det valgte antall trinn, Plateaystand og tirnhayde Plateavstand vil stor sett velges, dvs ikke beregnes. Som utgangspunt benytter en av vaesken pa en plate ikke skal rives med oppover til neste. Om det skjer eller ei er avhengig av gasshastigheten. Hvis trykktap spiller stor rolle, som ved vakuum, vil en velge en relativt lav gasshastighet, og kan ha en noe lavere plateavstand enn ved mer normal drift. Foravrig vil hensyn til rengjoring og vedlikehold, dvs adkomst, vere et viktig hensyn, Av denne grunn er det vanlig & velge en ‘minsteavstand pa ca 60 cm, og som regel er det heller ingen grunn til & velge starre. Men selv om 60 cm er en slags normalavstand, kan en ha plateavstander innen 15 cm -1 meter. Hvis vi velger plateavstand og vet antall plater, vil alts (antall plater - 1) x plateavstand gi hoyden av den del av kolonnen hvor platene er. Dessuten ma en ha litt rom under for jevnt utlap, og noe rom averst for veskefordeler og (oftest) dripefanger. For kolonner med diameter over I m bar en regne ca 1 m under platene og 1-1.5 m over dem. Hoyden av en fylt/pakket kolonne; HETP En mulig mate for 4 finne hayden av en fylt kolonne vil vaere 4 benytte verdier for hvor stor pakningshayde som tilsvarer ett teoretisk trinn. Dette kalles HETP: Height Equivalent to a Theoretical Plate eller pa norsk hoyde ekvivalent med en teoretisk plate. Verdier for HETP kan finnes i litteraturen inkludert brosjyrer, men siet da lite om hvordan HETP vil variere fra et system til et annet. Grovt sett blir tamhoyden ca den samme for fylt tirn med lase fyllegemer og platetim, og noe lavere med strukturert pakning. Driftsomrade for plater Det er 4penbart en grense for hvor mye gass og vaeske en kan presse gjennom/over en plate, men det er ogsd nedre grenser for mengdene. Dette kan illustreres.grafisk som vist i figur 11 for en huliplate. Pa aksene er avmerket linewr hastighet av henh.v. gass og vaeske beregnet pA tomt tam. I figuren er entraining medrivning, flooding er oppstuvning, weeping at varske drypper gjenmom hullene (i stedet for & renne over platen), og coning at det er sé liten veskestram at gassen strommer gjennom varsken i "kjegler", uten & blande seg med veesken. Til hoyre i diagrammet er ogsa vist et ved for mye veske kan nedlopene for vaeske mellom platene (downcomer) bli begrensende. Fbcessive enveanns.— NS TM LE: i rome get 5 e ea 308 305 ar L10W0 FLow ma Figur 11. Driftsomrade for hullplater. (Coulsen and Richardson 1989), Gasshastighet Ved dimensjonering av kolonner onsker en & ligge i ovre del av mulig driftsomride i figur 11, og tar DensPunk’ i hastigheten som gir oppstuvning (flooding), se figur 11. Det falgende et kopiemt ha Lydersen (1983): In Perry,!° p, 18-7, Fi imc tg, ius tension, N/m, p, and p, = liquid and vapour density, and r= the constant given in Figure 2-41, ives the flooding vapour velocity (241) t 1 aEI0g 910 me 2 ia area 460 : 0, 230 on] a ax] 0.06] 0.04] ao "0.02 0.08 0.06 0.1 0204 8 LO Figure 241. The constant C; in equation (2.41) as a function of tray spacing and the liquid to vapour ratio, L/Gkg/kg, times the square root of vapour (o liquid densi, Zocor 74>Equation (2.41) is valid for low- of non-foaming mixtures under the following conditions: the weir height is less than 15 per cent of the tray spacing, the sieve- plate perforations are 6 mm or less, and the ratio of the perforated area (sieve trays) or slot area (bubble cap trays) or valve area (fully open valves, valve tray) to the total flow area between the trays is 0.1 of greater; this area being the total column area minus downcomer area, The highest plate efficiency is obtained with a vapour velocity between 0.4 and 0.85 times the flooding velocity. Reasonable design values for low and non- foaming mixtures are in the range 0.6-0.7 times flooding velocity. ‘The cross-section of the column can be estimated from the equation Vaagh, 2.42) Where V = the superficial vapour velocity, m/s, referred to the total column area ‘minus the downcomer area, p = fraction of flooding velocity, usually between 0.7 and 0385, and ¥, = looding velocity from equation (241) Table 2.4. Constant a in equation (2.42) Service Non-foaming, regular systems 1.00 Fluorine systems, ¢. BF, Freon 090 Moderate foaming, eg. oil absorbers, amine and glycol regenerators. 0.85 Heavy foaming, eg amine and glycol absorbers 075 ‘Severe foaming, e.g methylethyiketone units 0.60 Foam-stable systems, ¢, caustic regenerators 03-06 G og den beregnede lineaerhastighet gi nedvendig len den totale gassmengde G er kjent, tamtverrsnitt for gasstrommen. Areal for nedlepsrorene for vaesken Mellom trinnene er der ett nedlopsrar for vaesken. Veeskehastigheten i nedlopsrorene velges ut fra folgende formler, tatt fra Lydersen (1983) ‘The downcomer area may be obtained from the one of the following equations that gives the smallest value of the design liquid velocity in the downcomer,* ¥,= 0.174 (243) ¥,=0.007a/p,—p, 244) %= 0040 pa, (245) where V, = liquid velocity in downcomer referred to the total downcomer area, m/s, a= system factor (Table 24), (with foam-stable systems use the lowest value, 4=0.3), py ~p,= liquid density minus vapour density, kg/m?, and t, = tray spacing, m. Siden den totale vaeskemengden er kjent, vil den kombinert med linezerhastigheten fra formlene gi det nadvendig tamtverrsnitt som mA avsettes til nedlopsrorene. BYULPTarntverrsnitt og -diameter Det totale tamtverrsnitt er summen av det som trenges for gasstrommen og det som trenges for nedlopsrorene. Diameteren av t&rnet finnes da fra sammenhengen mellom diameter og areal av en sitkel, Platetarn eller pakket tirn? De folgende tabeller og diagrammet er tatt fra en meget fersk artikkel av Bravo (1997), Han sammenligner i hovedsak forbedrede plater med strukturert pakning, som er blitt mye timeligere de siste arene. Men i det siste diagrammet er ogs4 konvensjonelle plater med, sml fig. 3 og 4 her. I igrammet vises kapasitet, og en ser at for lave vaeske/gassforhold vil pakkede tin gi storst kapasitet, dvs minst diameter. Omvendt gir platetran storst kapasitet hvis veeske/gassforholdet er stort. S4 er forste valg mellom plater og pakninger vil baseres pa kapasitet som vist her. Deretter ma en ta for seg andre forhold som spiller inn, som skumming, trykktap, korrosjon,. tilstopning pea urenheter og eventuelle andre forhold (se tabell 2 og 4), og se om en vil sta fast pa valget, eventuelt med modifikasjoner, eller om en ma velge en darligere kapasitet for & minimere drifisproblemer. Supplier Structured Packing Metal Grid Packing Giaseh nes Gempak Series Crd & EF 25 Grd Series Jaeger Products MaxPak Series Kech Engineering Ca Flexipae Series Fleughd Series Norton Co. | tntalox Suucturea Packing Seies Nutr Engineenng Ca Mantz Series Snap Grd Sens ‘Sulzer Chenwech ‘Mellapak Series T Watagra Sores Rie nce Eu ances Disadvantages High capacity atiow flow parameters ow capacity a high flow parameters High volumetric eficiency atiow pressures Low elfciency at high pressures Very low pressure drop Sensitive to fouling Low entrainment igh volametic ost Go0¢ performance in foaming systems Low resistance ta corrosion Grid Packings igh eapaciny Poor mass-transter efficieney Mechavicaly resistant ood heatanster performance 349BOCA COC UE uu oe Supplier High-performence-tray tade name Ghseh ne SuperraciMinvaive Trays, Nye/Minivalve Toys Koch Engineering Co Max-Fac Ways, BiFrac Trays Norton Co, Teton Tays Nuter Engineering Co, MG Trays OP, ne. MO Teays, ECMO Trays, EMO Trays 1 Consep Uys ete: Some proprietary hgh-pvormance vay designs cevloes by operating come ‘nansivay nthe recent past These cle Shas Calming Seton (C1, MF, Si PUTT rho Cir cucu ss Advantages Disadvantages. High capacity a high low parameters Low capacity atlow flow parameters Good fouling resistance Lower efficiency in some cases Improved entrainment characterises Higher pressure drop than packings Lower volumetic costs than packings Greater entrainment than packings Good cotrsion resistance Sensitive to foaming £ = a0 s — s = Conventional Fayed — High Pertormance Tray = = Swuctured Packing (€ticiency) — Structured Packing (Capaciey) (Capacity Parameter R ° oo a Lorrain fay Figure 6. Maximum useful capacity of column internals.Litteratur Boker Coulsen J.M., Richardson J.F.: Chemical Engineering, Vol 2, 3rd Ed., Pergamon Press 1989, Geankoplis C.J.: Transport Processes and Unit Operations, 3rd Ed., Prentice Hall 1993. Humphrey J.L., Keller II G.E., Separation Process Technology, McGraw-Hill 1997. Lydersen A.L.: Mass Transfer in Engineering Practice, John Wiley 1983. McCabe W.L., Smith J.C., Harriott P.: Unit Operations of Chemical Engineering, 5th Ed.,McGraw- Hill 1993. e Perry R.H., Chilton C.H.: Chemical Engineers’ Handbook, 7th Ed, McGraw-Hill 1997. Kompendier Erga O.: Tower Packings for Distillation, Absorption and Extraction Columns, Inst f. kiemiteknikk, NTNU, 1993 Loviand J.: Separasjonsteknikk, Inst.f. kjemiteknikk, NTNU, 1997 Roald A.: Kjemiteknikk Il, eget forlag, 1997. Artikler Bravo J.L.: Select Structured Packings or Trays?, Chem.Eng.Progr., 36, July 1997 e McNulty K.J.: Effective Design for Absorption and Stripping, Chem.Eng., 92, Nov 1994