HYDROGEN SEPARATION FROM REFORMATE PRODUCED BY AN ON-BOARD METHANOL REFORMER FOR SPFC VEHICLES

Transcription

1 ECN-C HYDROGEN SEPARATION FROM REFORMATE PRODUCED BY AN ON-BOARD METHANOL REFORMER FOR SPFC VEHICLES Development and evaluation of a metal membrane unit P.P.A.C. Pex, M.J. den Exter, H.M. van Veen R.A.J. Dams, S.C. Moore; Wellman Defence CJB Ltd. J.A. Dalmon, S. Miachon; CNRS Institut de Recherches sur la Catalyse R. Hughes, K. Hou; University of Salford S. Järås, J. Agrell, M. Boutonnet; KTH Stockholm DECEMBER 2004

2 Acknowledgement/Preface The work described in this final publishable report has been performed in the Fifth Framework Programme of the European Commission (Non Nuclear Energy Programme - Joule III) under contract JOE3-CT from February 1, 1998 to July 31, The financial contribution of the European Commission is greatly acknowledged. Abstract The objective of the project was the development and evaluation of a prototype hydrogen separation unit based on composite ceramic-pd/ag membranes. The unit should be performance-effective, size-effective and cost-effective as gas clean up unit by means of separation of hydrogen from reformate produced by an on-board methanol reformer for 20 kw SPFC vehicles. The main research guideline in the project was the integrated system development for a methanol fuelled SPFC vehicle. The system has been specified in the context of its application within the total fuel processing system for the propulsion of the vehicle. A flowsheet model in Aspen + has been developed consisting of two vaporising units for water and methanol, the methanol reformer, the membrane unit (in counterflow mode), the SPFC and a burner-unit that supplies heat to the reformer from the catalytic burning of the membrane s retentate. The model has been successfully validated. Two flow sheets have been produced and the effects of the sensitivity of the process variables have been investigated. The best process option appeared to be one, which operates with some type of sweep system. In the first half of the project 6 different manufacturing techniques were elaborated. At the midterm evaluation it was decided that all effort should be concentrated on the electroless plating technique and a dedicated optimisation programme was performed comprising sequential electroless plating upscaling, coplating as cost effective alternative, performance improvement by alloy modification and studying surface effect during operation. 24 tubes with a length of 1 metre were manufactured comprising 1 m 2 of membrane surface area. The membranes were tested and showed high flux and high selectivity performance. The developed membranes could withstand 50 temperature cycles between 200 C and 400 C. A CFD model was made and was used to validate a chemical engineering calculation model with which the basic design concept of the separator was determined. The basic design concept was elaborated in a technical design for a 21-tube module. As an intermediate step a module comprising 4 membrane tubes was designed, made and used in a test with simulated reformate. It appeared that it was very difficult to obtain leak tight sealing. Also the mechanical integrity of the tubes gave problems during sealing. An evaluation of all achievements against the set of requirements has been performed. Concluded was that a separator unit based on metal membranes would be a feasible option for gas clean up under the condition that several characteristics were further improved. Major drawbacks would be the startup procedure in which very fast heating of the unit is required and the need to use sweep gas in the separator, which makes the system for on-board application probably too costly. 2 ECN-C

3 CONTENTS LIST OF FIGURES 4 1. PARTNERSHIP 5 2. OBJECTIVES OF THE PROJECT 6 3. TECHNICAL DESCRIPTION OF THE PROJECT 7 4. RESULTS AND CONCLUSIONS 9 5. EXPLOITATION PLANS AND ANTICIPATED BENEFITS 15 ECN-C

4 LIST OF FIGURES Figure 3.1 Task interrelation scheme... 8 Figure 4.1 General flow sheet of the system... 9 Figure 4.2 SEM cross section of fractured sample of Pd/23%Ag membrane layer on ceramic support Figure 4.3 Hydrogen permeance of Pd/23%Ag membrane made with Electroless Plating. Membranes maintained selectivity H 2 /N 2 > Figure 4.4 Set of 24 membrane tubes with a length of 1 meter for the separator unit Figure 4.5 Results from Computational Fluid Dynamics calculations for module optimisation showing depletion of hydrogen of the feed along the membrane surface Figure 4.6 Schematic and picture of the separator test unit ECN-C

5 1. PARTNERSHIP ECN Energy research Centre of the Netherlands Dept. Energy Efficiency in Industry Separation Technology Group Address: Westerduinweg 3 P.O. Box ZG Petten The Netherlands Contact person: Ir. P.P.A.C. Pex Telephone: Fax: Pex@ecn.nl Wellman Wellman Defence CJB Ltd. Address: Airport Service Road PO3 5PG Portsmouth, Hampshire United Kingdom Contact person: Mr. R. Dams Telephone: Fax: Bob@cjbdev.demon.co.uk IRC CNRS Institut de Recherches sur la Catalyse Address: Avenue Albert Einstein Villeurbanne France Contact person: Dr. J.A. Dalmon Telephone: Fax: Dalmon@catalyse.univ-lyon1.fr Salford University of Salford; Chemical Engineering unit Address: Contact person: Prof. R. Hughes Telephone: Fax: r.hughes@chemistry.salford.ac.uk KTH KTH Stockholm; Chemical Engineering and Technology Address: Technikringen 42 S Stockholm Sweden Contact person: Prof. S. Järås Telephone: Fax: svenj@ket.kth.se ECN-C

6 2. OBJECTIVES OF THE PROJECT The objective of the project was the development and evaluation of a prototype hydrogen separation unit based on composite ceramic-pd/ag membranes. The unit must be performanceeffective, size-effective and cost-effective as gas clean up unit by means of separation of hydrogen from reformate produced by an on-board methanol reformer for 20 kw SPFC vehicles. The project concerns the development of the module as well as the membrane material itself based on the requirements for this particular application. For a compact module design it was important to have a high membrane surface area to volume ratio, a reliable sealing technique and a geometry, which ensures minimal concentration polarisation. The membrane manufacturing procedure had to give low cost membranes whereas their durability (startup/shut-down cycles) and longevity (lifetime) should be sufficient. In their performance the membranes must have negligible surface catalytic effect, high flux and a good response to transient behaviour. All above-mentioned issues were investigated experimentally in order to achieve an advanced state of the art. In the evaluation of the feasibility of the separation unit, system integration studies are carried out for setting boundary conditions and requirements, to determine optimum process designs and to calculate cycle efficiencies. Flow sheeting packages are used and the system performance should be validated experimentally. The viability in supplying pure hydrogen gas from the separation unit using reformate gas to an SPFC stack should be demonstrated. The most important measurable technical objectives were: A prototype hydrogen separation unit that is: operational for methanol fuelled 20 kw SPFC vehicles and will have, if fully developed, a unit volume less than 15 litres, optimised for gas treating operation by Computational Fluid Dynamics (e.g. high surface to volume ratio), tested under static and dynamic conditions with synthetic reformate, cost-effective: the cost target in the final application will be 10 Euro/kW based on the assumption that mass production is technically feasible. A tubular composite Pd/alloy membrane on a defect-free low cost ceramic support which: is able to separate hydrogen from reformate with a CO content below 10 ppm, has a high flux and sufficient dynamic response, is stable during thermal cycling in hydrogen (equivalent to 4,000 start-up/shut-down cycles), has a long life-time (equivalent to 300,000 vehicle kilometers). Results of system integration studies done by flow sheeting as well as experimental testing of a real vehicle application setup. An estimation of the total system viability and cost-effectivity. 6 ECN-C

7 3. TECHNICAL DESCRIPTION OF THE PROJECT Solid Polymer Fuel Cells (SPFC) fuel cells operate on hydrogen as fuel gas. Due to the fact that the fuel electrode contains platinum or other materials, which are able to split the hydrogen molecule, it is not allowed to have contaminants such as CO or H 2 S in the fuel gas. These contaminants adsorb on the active sites for hydrogen splitting and thus poison the electrode. The current available SPFC fuel cells can work very well with fuel gas streams containing about 50 to 100 ppm CO. However, best is to operate the system very well below this level. In producing hydrogen rich fuel gas streams by reforming hydrocarbons such as methanol, nafta or gasoline inherently significant amounts CO will be formed. The fuel gas stream must be cleaned up for this and several technologies are being developed for this purpose such as selective catalytic oxidation and vacuum swing adsorption. Also membranes can be used by which very pure hydrogen is separated from the reformate gas as permeate and the contaminants are concentrated then in the retentate stream. The main research guideline in the project was the integrated system development for a methanol fuelled SPFC vehicle. The boundary conditions from this system development gave a clear focus to the development of the hydrogen separation unit as gas clean up step. Final goal was to have this unit tested in combination with a methanol reformer, fuel cell and off-gas combustion unit. The system development in task 1was started with interactive activities of flow sheeting and experiments with single membrane tubes (task 3) to define realistic requirements and boundary conditions for this particular application. Module development in task 4 started with Chemical Engineering and Computational Fluid Dynamics calculations followed by a basic conceptual and technical design of the unit. Important was the optimal arrangement and diameter of the membrane tubes in this unit. Parallel to the system development, the most important part of the project was to develop membranes (task 2), which can withstand the operating circumstances better than the current state of the art membranes. For this membrane research low-cost, smooth and defect-free ceramic gas separation membrane supports were used. Since it was not clear which route was the most promising one at the start of the project, six membrane manufacturing routes were pursued. University of Salford developed Pd/alloy membrane layers by magnetron sputtering and chemical vapour deposition. ECN employed electroless plating as a Pd/alloy layer deposition technique. IRC used a laser technique to deposit Pd/alloy films to the membrane supports and IRC and KTH developed membranes according to the 'pore plugging' principle. At the midterm review there was an evaluation of the progress in all six membrane manufacturing routes. It was decided to continue only with the electroless plating technique. A dedicated optimisation programme was elaborated and all membrane researchers then focussed each to one of the research aspects of the chosen membrane preparation procedure. These research aspects concerned performance improvement, manufacturing cost reduction and further durability and life-time improvement. After engineering, manufacturing and basic testing of the separation unit containing optimised membranes, the unit should be integrated in the benchscale test system at Wellman. In the next schedule the task interrelation and the involvement of all partners in the different tasks are given. The task leaders are underlined in Figure 3.1. Due to company restructuring of the BMW mother company Rover had to withdraw from the project after the first year and all outstanding tasks were taken over by Wellman. ECN-C

8 Metal Membrane Unit as Fuel Gas Clean Up for SPFC Vehicles (COCLUP/HYSEP) Task 6 Project Coordination ECN Task1 System Integration Study Rover, ECN, WCJB Task 2 Membrane Research IRC, ECN, U.Salford,KTH Task 3 Single Component Testing ECN, WCJB, IRC Task 4 Separation Unit ECN, WCJB, Rover Figure 3.1 Task interrelation scheme Task 5 Integrated System Testing WCJB, Rover, ECN 8 ECN-C

9 4. RESULTS AND CONCLUSIONS System integration study In an early stage of the project system requirements have been prepared by Rover in order to guide and focus the technical developments. The system has been specified in the context of its application within the total fuel processing system for the propulsion of the vehicle. It is assumed that the vehicle application will take the form of a parallel power system, where the fuel cell power output is employed in parallel with a traction battery such that the latter will provide the main start up and transient loads, leaving the fuel cell to provide the main steady state power demands and to keep the battery in a suitable state of charge. See Figure 4.1. Methanol Water Reformer Pd/Ag Membrane Fuel Cell Auxiliary Battery Catalytic Burner Inverter Motor Figure 4.1 General flow sheet of the system For several system requirements specifications to the separator unit were set such as maximum physical dimension, location, weight, resistance against vibration and cost. A flowsheet model in Aspen + has been developed consisting of two vaporising units for water and methanol, the methanol reformer, the membrane unit (in counterflow mode), the SPFC and a burner-unit that supplies heat to the reformer from the catalytic burning of the membrane s retentate. After proper validation of the flow sheet steady-state calculations have been performed for which the reformer is set at 250 C and at a pressure of 21 bar, assuming chemical equilibrium. The SPFC is set at 80 C and 3 or 1.5 bar (depending on possible SPFC operating pressures). With calculations it was determined that a membrane surface area of 0.5 m 2 would be sufficient to supply the required hydrogen to the 20 kw SPFC. It appeared that membrane surface areas larger than 1 m 2 did not increase the hydrogen recovery. Calculations, using a sweepgas, at a reformer pressure of 21 bar reveal that the hydrogen recovery of the membrane should be around 80% in order to get a net heat production of zero. Membrane development and membrane characterisation The membrane development was performed by starting with six different manufacturing techniques: Magnetron sputtering, metal organic chemical vapour deposition, electroless plating, laser deposition, pore plugging by in-situ reduction and pore plugging with microemulsions. Up to the mid-term of the project these techniques were actually competing with each other and at the mid-term an evaluation of the progress in all six membrane manufacturing routes was made. The criteria against which the evaluation was done were performance, stability, production cost, ease of production, raw materials cost, impact on environment, ECN-C

10 possible improvements on short term. The relative overall ranking of the evaluation is given in the next table. Membrane manufacturing technique Ranking Electroless plating 1 Pore plugging interfacial reaction 0.77 Pore plugging micro emulsion technique 0.72 Magnetron sputtering 0.63 Laser deposition 0.52 Chemical vapour deposition 0.46 It was decided to continue only with the electroless plating technique. A dedicated optimisation programme was elaborated and all membrane researchers then focussed to certain research aspects of the chosen membrane. These research aspects concerned performance improvement, manufacturing cost reduction and further durability and life-time improvement. By optimising the electroless plating technique it was possible to manufacture membrane layers (Pd/23%Ag) with a total thickness of 3 to 5 µm on ceramic supports. See the Scanning Electron Micrograph in Figure 4.2. In single gas permeance tests the membranes showed a very high hydrogen flux and nitrogen permeance was measured with the majority of the samples. If the detection limit of the equipment is taken as the measured nitrogen flux, then the permselectivity is > Also tests with simulated reformate gave very high selectivities (process selectivities > 500). Figure 4.2 SEM cross section of fractured sample of Pd/23%Ag membrane layer on ceramic support The alloying of the membrane layers was done by sequential plating of palladium followed by silver. The necessary amount of silver to obtain a Pd/23%Ag layer was calculated from the amount of palladium, which was deposited in the first step. The operating window for depositing the right amount of silver was determined experimentally. After deposition it was necessary to make the alloy by heat treatment. In Figure 4.3 it can be seen how the alloying treatment is proceeding because it causes an increase in hydrogen flux. 10 ECN-C

11 Hydrogen Permeance at 350 o C after sintering at 't' hours at 500 o C Permeance [*10-6 mol/m 2.s.Pa] 3,00 2,50 2,00 1,50 1,00 0,50 0,00 0,50 1,00 1,50 2,00 2,50 3,00 3,50 4,00 4,50 Pressure difference [bar] t=9 T=350 t=18 T=350 t=27 T=350 t=36 T=350 t=72 T=350 t=99 T=350 Figure 4.3 Hydrogen permeance of Pd/23%Ag membrane made with Electroless Plating. Membranes maintained selectivity H 2 /N 2 > 1000 Using the above described equipment 24 membrane tubes with a length of 1 meter and an outer diameter of 14 mm have been prepared. This comprises about 1 m 2 of membrane area. He-tests show that the prepared and sintered membranes are gas-tight. In Figure 4.4 a photograph is given of this set of membranes. ECN-C

12 Figure 4.4 Set of 24 membrane tubes with a length of 1 meter for the separator unit Next to the development of the sequential plating technique an attempt was made to use coplating as cost effective alternative. However in all experimental work during the project it was not possible to either cope with the instability of the plating bath or when using stable plating baths to achieve the right alloy composition. Further performance improvement has been studied by alloy modification (ternary alloys) and by studying the influence of surface effects. The surface effects comprise both segregation of the alloy (silver tends to migrate to the surface of the membrane layer) and the effect of CO and water adsorption. The adsorption of CO causes a flux decline of about a factor 2, which has to be compensated by an increased membrane surface area in the module. Water has shown to give hardly any effect on the flux. Separation unit development and integated system A Computational Fluid Dynamics (CFD) model was made (see Figure 4.5) and was used to validate a mathcad calculation model with which the basic design concept of the separator was determined. 12 ECN-C

13 Feed flow Membrane H 2 Sweep gas flow Figure 4.5 Results from Computational Fluid Dynamics calculations for module optimisation showing depletion of hydrogen of the feed along the membrane surface. The basic design concept was elaborated in a first technical design in which technical limitations of the membrane separator became apparent. Via an iterative procedure a second concept and a second technical design were made. Before a complete engineered design of a 21 tube module was made (see Figure 4.6) it was decided to check the practical performance of the design. In order to do this a module comprising 4 tubes was designed, made and used in a test with simulated reformate. It appeared that it was very difficult to obtain leak tight sealing. Also the mechanical integrity of the tubes gave problems during sealing. During the first test it appeared that the membrane layer peeled off the substrate which was not observed in the lab testing. An evaluation of all achievements against the set requirements has been performed. It was concluded that a separator unit based on metal membranes would be a feasible option for gas clean up under the condition that several characteristics were further improved. Major drawbacks would be the startup procedure in which very fast heating of the unit is required and the need to use sweep gas in the separator which makes the system for on-board application probably too costly. ECN-C

14 Figure 4.6 Schematic and picture of the separator test unit. 14 ECN-C

15 5. EXPLOITATION PLANS AND ANTICIPATED BENEFITS For a CO removal unit operation on-board of a vehicle the following options can be considered: selective CO methanation, CO conversion (water gas shift), adsorption (PSA), membrane separation and selective CO oxidation. All options are still being investigated in a number of research groups all over the world. Currently the CO selective (catalytic) oxidation seems to be the most preferred technology because of its simplicity and operating window. However there are several drawbacks such as durability of the catalyst and necessary auxiliary (control) equipment that could be a barrier for final application. Although the development of the membrane technology for this application has not progressed as far as the CO selective oxidation technology it is considered as the best runner up technology. Because of its expected performance and its simplicity it would be a feasible option for gas clean up under the condition that several characteristics will be further improved. Major drawbacks are the startup procedure in which very fast heating of the unit is required and the need to use sweep gas in the separator which makes the system for on-board application probably too costly. The above mentioned drawbacks will not play a role when the membrane separator is used as part of a fuel processing system for stationary SPFC application. Another very feasible application is the application in on-site hydrogen production systems. One of the possible scenarios in a future hydrogen fuel infrastructure would be that hydrogen would be produced at fuelling stations. As the separator unit can run then constantly this would be a unit operation having a very high position on the short list of feasible options. It is most likely that the results of this project will be used for further research and development projects aimed at application as indicated above. Typical photographs, diagrams or figures to illustrate the potential applications of the results of the project can be found in the above text. ECN-C