R&D on Hydrogen Manufacturing Process Using Inorganic Membrane

Transcription

1 2001.M1.1.1 R&D on Hydrogen Manufacturing Process Using Inorganic Membrane (Inorganic Membrane Reaction Technology Group) Ohi Research Room No.401 Hideto Koide, Yoshikazu Aoyagi Mizuho Research Room No.401 Toyosu No.401 Research Room Tomonori Takahashi, Hitoshi Sakai,Tatsuya Hishiki, Shigetaka Matsuda Tadashi Sasa, Akihiko Suzuki, Fumiki Tomioka, Yasuhiro Otake, Seiji Sato, Norihiro Okumura, Takenori Watabe, Minoru Mizusawa, Koki Hamada 1. Contents of Empirical Research In terms of the global environment and energy problems, etc., technologies for reducing environmental loads and for using energy at high efficiency have been gaining sharply in importance in recent years. At chemical plants, separations and reactions are the greatest energy-consuming processes, and there is strong demand for the development of energy-conserving, highly efficient processes. The membrane separation process is an energy-conserving process that enables separation with only slight energy consumption. In addition to separation of specific substances from multiple reaction byproducts, efficient reaction processes can be implemented by combining membrane separation with chemical reactions and thereby achieving simultaneous reaction and separation. When an organic membrane is used, however, the highest application temperature is about 200 C. At higher temperatures, metal or inorganic membrane must be used. Yet among the oil refining processes conducted to date, there are no e x amples of practical application of any stable membrane separation process at high temperature involving metal or inorganic membrane. Meanwhile, in the oil refining process, the need for converting heavy oils to lighter components has been mounting, and thus there is also increasing demand for hydrogen. At present, hydrogen is produced mainly by the steam reforming method, but then the reaction temperature reaches 800 C or higher. If a membrane separation process employing hydrogen separation membrane can be applied to this process, it will become practical to reduce reaction temperature, and an energy-conserving process can be expected. 1

2 In the present research, separation performance and mechanical strength of inorganic functional materials, which are emphasized when applying the membrane separation process at steam reforming plants, are confirmed; fundamental data are being collected, and basic structural designs are investigated. Basic structural designs are drawn up on the basis of functional verification tests of a small-scale steam reforming membrane reactor [S/C (steam/carbon) ] = 3 in which the outlet temperature is 600 C, hydrogen purity is 95% and reaction rate is 85%. Conceptual designing for a practical plant will be implemented this fiscal year. 1.1 Research on Membrane Reactor Technology Membrane reactors are onstructed and used in reaction tests. Evaluations focus mainly on the relationships between reaction rate and reaction temperature, reaction pressure and S/C (steam/carbon) ratio, and the possibilities for achieving lower temperatures, higher pressure and low S/C ratio, are investigated. 1.2 Research on Separation Pipe Elemental Technology Concerning the alumina porous ceramic material used as membrane separation base material and the junctions between ceramics and metals, basic data on mechanical strength are collected. 1.3 Research on Reaction Pipe Elemental Technology (1) Porous pipe strength test A 3-point bending test and rotary bending test are conducted on porous pipe, and the characteristic strength of porous pipe is determined. (2) Analysis of reaction pipe strength Respecting porous pipe joint structure, is analyzed and the produced by porous pipe under actual equipment load conditions is investigated. In order to analyze the strength of porous pipe and of reaction pipe as a whole with catalyst included, a reaction pipe strength analysis model is created. The basic tests required for modeling and the structural formability of structural elements are investigated. (3) Reactor analysis A reactor analysis code is created for analysis of reactions and separation phenomenon inside a membrane reactor in which a hydrogen separation membrane is applied. The impact of each parameter on hydrogen separation performance is investigated and the performance of a small hydrogen separation reactor is evaluated. 1.4 Conceptual Design (1) System preliminary investigation It is made possible to incorporate a code for analysis of hydrogen separation reactor into a plant general-purpose simulator where the system as a whole can be investigated; to make verifications, and to investigate the system as a whole. 2

3 (2) Conceptual reactor design An investigation is being made of the format and strength of a large-scale hydrogen separation reactor on the assumption of practical equipment based on system investigation results, and optimum structure is determined. (3) Conceptual plant design The system operating conditions and system composition of a hydrogen separation type hydrogen production unit are being investigated, and a conceptual plant design is being formulated. System renovations are evaluated for economy, and the economic benefits of introducing a hydrogen separation reactor are considered. 2. Results of Empirical Research and Analysis Thereof 2.1 Research on Membrane Reactor Technology (1) Hydrogen permeability Table 2.1 presents findings on the hydrogen permeability of Pd-Ag membranes measuring 20 µm and 5 µm in thickness. At 20 µm, permeability is m 3 /m 2 /sec/kpa, whereas at 5 µm, it is m 3 /m 2 /sec/kpa, indicating that the performance gained is roughly proportional to the thickness. Table 2.1 Hydrogen Permeability of Pd/Ag Membrane Pd/Ag Membrane (µm) 20 5 Hydrogen Permeability (m 3 /m 2 /sec/kpa) (2) Separation membrane module at 20 µm membrane thickness Figure 2.1 illustrates the dependency of methane conversion rate on S/C. A methane conversion rate exceeding the equilibrium conversion could be obtained. With a drop in S/C, the equilibrium conversion also drops sharply, but measured values do not decline so extensively. This is ascribed to a gain in hydrogen partial pressure due to the drop in S/C and to an increase in permeable hydrogen flow volume. Accordingly, it is found that the effect of the membrane reactor becomes larger as the S/C becomes smaller. (3) Separation membrane module at 5 µm membrane thickness Figure 2.2 presents the dependency of methane conversion rate on pressure. S/C was fixed at 3. In all the data, methane conversion rate exceeds the equilibrium conversion rate, and a value of 85% or more could be obtained under all conditions. In comparison to a membrane thickness of 20 µm, a high methane conversion rate could be obtained. Moreover, with an increase in reaction pressure, the equilibrium conversion rate drops, but the margin of decline in the measured methane conversion rate as compared to the theoretical value is small. This is ascribed to the promotion of greater hydrogen permeability due to an increase in hydrogen partial pressure as well as an increase in reaction pressure. 3

4 Conversion rate (%) Test value Theoretical value Reaction side pressure 5.3 kg/cm 2 Transmission side pressure 0 kg/cm 2 Methane conversion rate (%) Test value (%) Theoretical value (%) Reaction pressure (kg/cm 2 ) Figure 2.1 S/C vs Methane Conversion Rate (20 µm Membrane Thickness) Figure 2.2 Reaction Pressure vs Methane Conversion Rate (5 µm Membrane Thickness) 2.2 Research on Separation Pipe Elemental Technology [Research on Ceramic Composition Material (Junction)] A junction pseudo test piece was trial produced as shown in Figure 2.3 and 4-point bending strength was evaluated at room temperature and at high temperature, and repeated fatigue test was conducted at room temperature. A single type of ceramic was used in which the pore diameter is 0.6 µm; open porosity is 25%; apparent density is 2.91g/cm 3 and Young s modulus is 162 GPa. With a hydrogen separation membrane connected to metal flange, a durability test was conducted at 500 C in pseudo-mixed gas. Alumina: material A Permalloy Pd-Ag membrane Ag wax Figure 2.3 Drawing of Test Sample Format Weibull curves of strength are presented in Figure 2.4. At room temperature, the test sample fractures in the alumina, but at high temperature, it ruptures vertically at the boundary between alumina and wax. At 600 C, it is believed that the boundary anchor effect weakened and there was a drop in strength. In the repeated fatigue test, no drop in residual strength was noted even after cycles. 4

5 The durability test was conducted for a total of 740 hours, but no drops in density due to the production of pinholes before and after the durability test were noted. Weibull Curves of Junction Test Sample 4-Point Bending Strength Junction component 600C Junction component RT Porous body RT Fracture probability (%) Stength (MPa) Figure 2.4 Weibull Curves of Junction Strength 2.3 Research on Reaction Pipe Elemental Technology (1) Analysis of reaction pipe strength (a) Junction analysis In the joint structure between porous pipe and junction component, it is assumed that great is produced at the junction component because of the differences in thermal expansion of the various materials. In the present research, a numerical simulation was conducted on joint structure in which the joint component configuration (total of 11 types) and the materials (Permalloy, Kovar) were changed; the strengths of porous pipe and junction component were then evaluated. Figure 2.5 shows an analysis model. Figure 2.6 is a drawing of sample element division. As the analytical model, a two-dimensional model of axial symmetry was assumed in which the top edge is fixed and the model is suspended. The load applied to the analytical model is taken as the weight of the porous pipe and junction component, external pressure and thermal load. 5

6 Permalloy Porous pipe Figure 2.5 Analytical Model of Joint Structure Figure 2.6 Drawing of Sample Element Division For the analysis of junction component, a total of 11 types of analytical model were used, 6 types in which the junction component length L and width T were changed and 5 types in which the junction component shape was changed, covering cases in which only maximum differential pressure was applied to the analytical model shown in Figure 2.6, using Permalloy as the junction component material, and cases in which the catalyst pressure determined from analysis of reaction pipe bending was applied in addition to maximum differential pressure. In the course of the analysis it was found that the least amount of was manifested in the model configuration shown in Figure 2.6. In order to reduce still further, an investigation was made in which the junction component material was changed to Kovar, which has a lower thermal expansion differential with porous alumina than does Permalloy. The results of analysis at running time (600 C maximum temperature) and at temperature rise (400 C) are given in Table 2.2 and Table 2.3, respectively. It is at the temperature of 400 C during temperature rise that the difference in the linear expansion coefficients of Kovar and porous alumina becomes greatest. Table 2.2 Site Porous pipe Junction (Kovar) Junction Stress Analysis Results (While Running: Maximum temp. at 600 C) Analysis Conditions Maximum of each component (MPa) External pressure Heat load R direction θ direction Z direction Equivalent of Mises Maximum main A A A B B A B B A A A B B A B B

7 Table 2.3 Site Porous pipe Junction Stress Analysis Results (While Temperature is Rising : 400 C) Analysis Conditions External pressure Heat load R direction Maximum of each component (MPa) θ direction Z direction Equivalent of Mises Maximum main A A B A Junction A A (Kovar) B A Table 2.2 reveals that the maximum principal of porous pipe was less than the average 3-point bend strength of MPa, and that the maximum equivalent of junction component Mises was less than the corresponding yield of MPa. Hence it was discovered that the joint structure in which Kovar is used as the junction component is efficacious at running time. Table 2.3 shows that the maximum principal of porous pipe was lower than its strength, but that the maximum equivalent of junction component Mises was between the yield of Kovar at MPa and the tensile strength at 327 MPa, and thus that the junction component underwent plastic deformation. These findings indicate that the joint structure that uses Kovar as the junction component is superior to the joint structure using Permalloy junction component. The Kovar junction component did not fracture during temperature rise and breakdown results were obtained. (b) Reaction pipe bend test A reaction pipe test unit model half the size of practical equipment was created and subjected to a 3-point bending test. A cross-section photo taken during assembly of a reaction pipe test unit is shown in Figure 2.7(a). An external view of the reaction pipe mounted on test equipment is presented in Figure 2.7(b). The reaction pipe test piece is depicted in Figure 2.8. In the 3-point bending test, a 20 kn load is applied and distortion is measured at the center of the separation pipe. The measured distortion is used in the analysis of reaction pipe. (a) Cross-section photo during assembly (b) External view of pipe installed to test equipment Figure 2.7 Photos of Reaction Pipe Test Unit 7

8 Figure 2.10 shows the empirical values of axial distortion in separation pipe. The figure indicates that the distortion produced in separation pipe when reaction pipe is bent cannot be ignored. This distortion also becomes a large compression at the compression side of the reaction pipe being bent and small at the extension side, but it does not reach zero. This is ascribed to the fact that catalyst can convey force when it is compressed but cannot easily convey it when it is extended. (c) Reaction pipe analysis Using the distortion measured with separated pipe in bending test of reaction pipe test unit, the material constant of catalyst was identified by analysis. The analysis model is shown in Figure 2.8. Loads P1 and P2 in Figure 2.8 were each applied. In view of the test unit s symmetry, a 1/4 model was taken as the analytical model. In Figure 2.8 the catalyst is divided into 10 layers vertically in the load direction, and at each layer a different material constant can be inputted. The steps in the analysis are given in Figure 2.9. Load P 1 Load Catalyst Separation pipe Reaction pipe Assumption of catalyst (10 layer) material constant Finite element analysis of reaction pipe Does calculated values correspond to measured values? Catalyst material constant has been identified. Figure 2.8 Reaction Pipe Analysis Model Figure 2.9 Steps in Reaction Pipe Analysis Figure 2.10 presents axial distortion of separation pipe, calculated using Table 2.5. The figure shows that calculated values and test values exhibit the same trend, indicating that the material constants of catalyst in Table 2.4 are roughly appropriate. Next, using catalyst material constants and porous pipe in separation pipe, catalyst pressure was calculated for when it is assumed that a 20 C linear temperature differential has been applied parallel to the reaction pipe cross section. Moreover, the maximum principal of MPa produced in the porous pipe was lower than MPa, the average 3-point bend strength of porous pipe. It was also discovered that porous pipe does not fracture during running even when this maximum principal of MPa is added to each maximum principal in Table

9 Test Calculated value value Case of P 1 Case of P 2 Axis distortion Distortion gage position Figure 2.10 Reaction Pipe Bend Test Results vs Computed Values Table 2.4 Catalyst Material Constants No.of catalyst layer Young s modulus (GPa) Poisson ratio (load side) Conceptual Design (1) System preliminary investigation With respect to naptha and offgas raw material, system calculations were made under various conditions. In the system, a reformer already existing is combined upstream with a hydrogen reactor in which hydrogen separation reactor analysis code is built in. System flow is illustrated in Figure 2.11 and a typical e x ample of investigation results is presented in Table 2.5. In both system calculations and independent calculations the material balances were roughly the same, and it was confirmed that incorporating the analysis code into plant general-purpose simulator proceeds smoothly. The differences in the two approaches were due to the fact that system calculation takes place at isotemperature while independent calculation occurs at thermal insulation. 9

10 Combustion exhausts Reformer Raw material gas Hydrogen separation reactor (Membrane reactor) Refined H 2 gas Combustion air Burner Non-permeable gas Figure 2.11 Flow of System Preliminary Investigation (2) Reaction pipe conceptual design From preliminary system conditions, a conceptual design was completed of a hydrogen separation reactor in the shape of conventional heating furnace and heat exchanger, and the following problems became apparent. (a) With the heating furnace type system, radiation is not effectively used because heating furnace fuel is of low calorification volume. (b) With the heat exchanger type system, there are no suitable materials for high-temperature design. From final system conditions, a conceptual design was completed of a hydrogen separation reactor having the structure of a convective heat transfer type heating furnace which brings out the superior points of both heating furnace type and heat exchanger type system. Specifications are given in Table 2.6 and the structure is shown in Figure The reactor is comprised of a convection heater, burner and heat recovery component. By separating the burner, it is possible to burn low-calorie gas stably. Due to low-calorie stable combustion, naptha is reheated and some high-pressure, non-permeable gas is used for atomizing. Heat-resistant, austenite stainless steel is used as the reaction pipe material. 10

11 Table 2.5 System Preliminary Investigation Conditions and Results System calculation Independent calculation Raw material Naptha S/C 5.4 Hydrogen permeation speed constant 2 x 10-2 Nm 3 /m 2 /sec/(kpa) 0.5 Reform chamber pressure MPa 2.74 Transmission chamber pressure MPa 0.29 Separation pipe category Monolith pipe Separation pipe size φ 30 mm L1,000 mm No.of separation pipes 1,000 Non-permeable gas flow volume Nm 3 /h 46,026 46,920 H2O % H2 % CO % CO2 % CH4 % Non-permeable gas outlet temperature C Permeable hydrogen volume Nm 3 /h 15,490 13,784 Combustion air flow volume Nm 3 /h 29,176 - Combustion air inlet temperature C Combustion exhausts temperature C 1,134 - Table 2.6 Hydrogen Separation Reactor Specifications Independent calculation Raw material Naptha Hydrogen separation pipe shape μm Single-port pipe Hydrogen permeable membrane thickness μm 5 No. of hydrogen separation pipes - 5,130 Hydrogen separation pipe length mm 1,000 No. of hydrogen separation pipes mm 30 Reaction pipe outer diameter/inner diameter mm 150/130 No. of separation pipes/reaction pipe - 7 Reaction pipe length m 9 No. of reaction pipes - 82 Catalyst capacity m Inlet flow volume Nm 3 /h 33,837 SV value h -1 5,511 Flow passage cross-section area m Reaction pipe average diameter mm 140 Heating transfer area m Absorbed duty kw 11,169 Heat flux kw/m

12 Figure 2.13 Flow of Hydrogen Production Unit With Hydrogen Separation Membrane 3. Results of Empirical Research (a) (b) (c) (d) (e) A Pd-Ag membrane 5 µm in thickness was fabricated, and a hydrogen permeation speed two times or greater than the target value was obtained. Using the aforesaid Pd membrane, a membrane reactor test was performed, and at a reaction temperature of 600 C and with S/C = 3, a methane conversion rate exceeding 90% could be obtained. An alumina/metal junction strength test and analysis test were performed. It was found that there is adequate strength in the design. A conceptual design was completed of a hydrogen separation reactor, and a reactor with the structure of a convective heat transfer type heating furnace was designed. The economy of a hydrogen separation type hydrogen production plant was evaluated, and it was confirmed that such a plant offers economic benefits. 12

13 4. Synopsis In the past, membrane reactors have been vigorously researched at universities and national research centers, and major results are being obtained. Against this backdrop, the present research project undertook to confirm reactions and to investigate mechanical features and economy from the standpoint of practicality for the sake of application at refineries. In this research, membrane reactor reactions were verified and the methods of strength testing and analysis were used to verify the possibility that ceramic materials can be used; this possibility was manifested. It was also established that ample profit can be gained in terms of economy. These results demonstrate that the membrane reactor can be made practical. It was also concluded that the project will impact upon the orientation of future R&D and that its objectives have been amply achieved. To make the project practical, the next step required is to investigate reaction characteristics and mechanical characteristics using bench-scale, trial-produced equipment. The aim of future planning is to collect data essential for modularization and to construct design parameters for cases in which ceramics are used. Copyright 2001 Petroleum Energy Center all rights reserved. 13