PRE-REFORMING OF DIESEL FUEL AS FUEL PROCESSING TECHNOLOGY FOR HIGH TEMPERATURE FUEL CELLS

Transcription

1 PRE-REFORMING OF DIESEL FUEL AS FUEL PROCESSING TECHNOLOGY FOR HIGH TEMPERATURE FUEL CELLS Nils Kleinohl*, J. B. Hansen, P. Nehter, H. Modarresi, A. Bauschulte, J. vom Schloß, K. Lucka OWI OEL-WAERME-INSTITUT GmbH, Affiliated Institute RWTH Aachen, Kaiserstrasse 100, D Herzogenrath HALDOR TOPSØE A/S, Nymøllevej 55, DK-2800 Lyngby ThyssenKrupp Marine Systems AG / Howaldtswerke-Deutsche Werft GmbH, Werftstr. 112/114, D Kiel * n.kleinohl@owi-aachen.de, OWI Oel-Waerme-Institut GmbH ABSTRACT Pre-reforming is a well-known technique in large scale petro-chemical plants and is used in that case to prepare a homogeneous feedstock for a downstream steam reformer. This process delivers a hydrogen and methane rich product gas which could also be used as fuel gas for a SOFC system. For this purpose a test rig was built which helps to understand pre-forming of diesel fuel at atmospheric pressure in terms of lifetime and possible catalyst deactivation indicators. A long-term experiment was carried out successfully for 3200 hours with one catalyst bed. It is shown that in the course of this run the product gas of the fuel processor was highly suitable for a SOFC. No traces of hydrocarbons other than methane could be measured until complete deactivation of catalyst bed. 1. INTRODUCTION For electrical applications which are grid independent a generator system with an internal combustion engine operated with natural gas, gasoline or diesel is often used today. Although mobile or portable systems could use batteries for electrical energy storage they have to be recharged either from the grid or by a generator system. Large power consuming mobile electrical systems should ideally operate several days or weeks without connection to the grid. However, battery capacity is generally limited by space especially on ocean going ships which restricts the operational time of the battery. To increase the overall electric efficiency and minimise emissions for auxiliary power units (APU) for grid independent applications new technologies must be developed for commercial usage. One possible technology will be a fuel cell system. The overall efficiency of this kind of system can be larger than a generator system with an internal combustion engine as Nehter et al. [1] describes. It is clear that a fuel cell requires hydrogen to operate, but storage capacity for hydrogen is still limited by space and safety so that another energy carrier is favourable for grid independent applications. One option for energy storage is using an existing infrastructure for liquid hydrocarbons which can be stored easily with known technology. Expanding an APU with a fuel processor is than essential because hydrogen has to be formed for utilisation of diesel fuel with a fuel cell. In the near future emission limits for ocean-going ships will change. As a consequence they have to use diesel fuel for their main engines within territorial waters of the European Union. This makes it easy feasible to use also diesel fuel as energy source for the generator system. To increase the electrical efficiency of onboard power delivery this research project was started with the aim to develop a fuel cell based APU system which will be fed with diesel fuel as mentioned by Leites et al. [2]. The focus of the present system is to save energy costs for a ship owner by increase of the electrical efficiency. Furthermore, lower emissions during operation are achieved by applying a fuel cell system compared to internal combustion engines. For that purpose an APU will be developed which will consist of a fuel cell, a fuel processor and an off gas burner with integrated heat recovery. As fuel cell a solid oxide fuel cell (SOFC) was chosen because it has the highest exposure APU system with high electrical efficiency. Furthermore system design is simple in comparison to other fuel cell types, because gas cleaning downstream the fuel processor is not required. A SOFC can handle high amounts of carbon monoxide which is generally poisonous for all types of proton exchange membrane fuel cells. The efficiency of the SOFC is influenced by internal reforming of methane

2 and the water-gas-shift reaction. By using the internal heat production of the fuel cell both mentioned mechanisms can provide additional hydrogen. Consequently with the right ratio of hydrogen and methane in the inlet stream heat management of the SOFC is more efficient and the overall electrical efficiency increases. As fuel processor different possibilities were evaluated and it turned out that for a fuel cell system on a ship all kinds of steam reforming are suitable. When fuel processes like steam reforming, pre-reforming and partial oxidation were evaluated by their exergetic efficiency the process of pre-reforming has efficiency advantages over steam-reforming and partial oxidation in combination with SOFC [1]. Especially the methane content of the product gas and the low temperature of the process of pre-reforming of diesel fuel make it to a possible solution for fuel processing in a SOFC system on an ocean-going ship. Different authors like Krummrich et al. [3], Steinfeld et al. [4] and Peters et al. [5] have worked on this fuel processing technology with the result that it was possible to transfer knowledge of the process from pressurised into atmospheric conditions at laboratory scale. Running the fuel processor with a nonpressurised reactor design leads to a lower energy consumption of supply components and less expensive construction of the reactor. Scope of the present research is the extension of the lifetime for a nickel based catalyst bed for pre-reforming of diesel fuel. Recently Nehter et al. [6] has reached a maximum catalyst lifetime between 500 hours and 1000 hours time-on-stream. By addition of synthetic anode off gas recirculation this should be exceeded. 2. FUEL PROCESSOR BASED ON PRE-REFORMING PROCESS Pre-reforming of hydrocarbons is an established process in large petro-chemical plants and is widely used to increase process as well as catalyst efficiency in steam reforming reactors, for e.g. hydrogen or ammonium production. After this process the feed for a downstream steam reformer is homogenous and free of catalyst poison. Pre-reforming reactors are built in such a way that the conditions inside are nearly adiabatic which leads to significant energy savings compared to a heated steam reformer. Also less expensive reactor materials can be applied because the overall temperature level is 200 C lower compared to a steam reformer which has bed temperature above 750 C. Feedstock for such plants is commonly a mixture of light hydrocarbons like natural gas or naphtha. For these fluids knowledge was gathered over past 60 years. Most mechanisms for long time operation are understood and gathered by Rostrup-Nielsen [7]. In large scale petro-chemical plants process pressure of the prereformer is in the range between 5 bar and 40 bar. Inlet temperatures of catalyst bed depend on feedstock which is generally in a range between 350 C and 550 C. Steam-to-carbon ratio (S/C) varies between 0.3 and 4.0 depending on feedstock and desired product gas composition. The pre-reforming reactor is generally placed upstream of a steam reformer and filled with a highly active catalyst. Mainly nickel based catalysts are being used for pre-reforming but also precious metal catalysts can be applied. The product gas consists of hydrogen, methane and oxides of carbon as described by Boon et al. [8]. These characteristics are ideal for a fuel processor for a fuel cell system where water is easily available, which makes it suitable for application on an ocean-going ship. Typically inside a pre-reforming reactor three main reaction paths are established. These are the steam reforming reaction (1), the water-gas-shift (2) and methanation (3). The first reaction (1) is endothermic and both other main reactions (2) and (3) are exothermic as mentioned by Boon et al. [8] In the irreversible reaction (1) for temperatures above 347 C all higher hydrocarbons are converted into C 1 components with no intermediate products. A thermodynamic equilibrium of reactions (2) and (3) will be established as described by Rostrup-Nielsen [7]. At elevated pressures compared to atmospheric conditions the equilibrium for methanation (3) is shifted to a higher methane yield. With increase of the catalyst bed temperatures the content of carbon monoxide in the product gas rises and methane content is lowered. Rostrup-Nielsen [7] figures out that when steam-to-carbon ratio is increased the methane content in the

3 product gas is lowered when feedstock are hydrocarbons other than methane. This relation could also be observed by Kleinohl et al [9] for pre-reforming conditions and diesel fuel feedstock. Advantages of pre-reforming over steam reforming as fuel processing process are the overall lower bed temperatures of approximately 550 C, which decreases the tendency for formation of carbon deposition in the catalyst bed. Additionally, due to lower temperatures commonly used materials for the reactor can be applied. Downstream of the fuel processor the product gas can be heated up to 700 C without any formation of carbon deposition as described by Krummrich et al. [3]. This is highly suitable for operation of a SOFC which has a desired inlet temperature above 650 C. One limitation depends on gum formation at the beginning of the catalyst bed when bed temperatures reach below 350 C but this can be solved by increase of the overall process temperature to values between 400 C to 550 C. Several authors have shown that pre-reforming can be used as a fuel processor for fuel cell systems. Peters et al. [5] showed that most of these systems worked with natural gas like, although Krummrich et al. [3], Steinfeld et al. [4] also used diesel-similar fuel and showed the feasibility of this process. It was shown in a previous work by Nehter et al. [6] that the life-time of the catalyst bed could be extended with additional hydrogen or recirculation of synthetic anode off gas of a fuel cell introduced to the inlet of the reactor. This was considered for experiments carried out in this work. A test rig was designed which can apply a synthetic anode off gas of a SOFC to the feed of the pre-reforming process. 3. EXPERIMENTAL SETUP A test rig for detailed research of pre-reforming at atmospheric pressure was built. Kleinohl et al. [9] showed a proof-of-concept for this type of fuel processing technology. This test rig consists of a tubular reactor for pre-reforming with a mixing zone and periphery for dosing of gaseous and liquid mass flows. For saturated steam a Hovapor LF6000 steam generator is used which can evaporate water with a constant mass flow up to 100 gram per minute without any carrier gas which would dilute the reactants. For generation of synthetic anode off gas the test rig is equipped with mass flow controllers for hydrogen, carbon monoxide and carbon dioxide. Also a mass flow controller for nitrogen is implemented for flushing the reactor and thermal conditioning of the test rig before the experiment start. All gaseous components are superheated before entering the mixing zone of the reactor at approximately 650 Cvia a radial mounted inlet tube. Liquid diesel fuel is dosed via a metering piston pump on an electrical heated porous medium at the bottom of the mixing zone before entering the actual reformer. The temperature of the surface was measured by a thermocouple (TC) and controlled to 400 C. Downstream of the reactor a product gas cooler is attached which achieves exhaust gas temperatures below 40 C. This allows taking condensate samples at a drain. The product gas is burned in a natural gas supported combustor and the exhaust is led to a chimney. The arrangement of all periphery components can be seen in Figure 1. The tubular reactor is insulated with two layers of mineral wool ( 0.25 W/ (mk)). Directly onto the reactor wall, under the insulation, two heating zones with electrical heaters are applied in order to compensate the heat loss during operation. To achieve a constant wall temperature for both parts of the reactor independent controllers are used. For monitoring of the catalyst bed a total number of ten thermocouples are arranged inside the reactor. Seven of them are on the main axis and three are positioned near the reactor wall. With these measurements not only the temperature profile along the main axis is monitored but also the temperature distribution through catalyst bed could be observed. The detailed position of all TC can be seen in Figure 1, right lower part. For online gas analysis a Rosemount NGA2000 for measuring of dry gas concentrations is attached to the test rig behind the gas cooler. With this device the amount of hydrogen, methane, carbon monoxide and carbon dioxide can be measured by thermal heat conductivity (THC). For quantification of the amount of hydrocarbons in product gas samples for gas analysis can also be taken. The combination of gas chromatography (GC) and a flame ionization detector (FID) with the installed separation column in the GC allows identifying hydrocarbons up to C 5 with single digit ppm analysis resolution.

4 Figure 1: Test rig periphery and details of the pre-reforming reactor 4. RESULTS Different experiments were carried out to show that pre-reforming of diesel fuel can be used as fuel processing technology, while enlarging the lifetime of the catalyst bed through addition of synthetic anode off gas recirculation. As fuel standard European road diesel is used whereas for steam generation water for laboratory usage with specifications according to ISO 3696, type 2, was provided. Chemical properties of used diesel fuel are listed in Table 1. Table 1: Chemical properties of diesel fuel for long-term experimental Property Method of testing Result Unit Density DIN EN ISO kg/m³ Lower heating value DIN J/g FAME content DIN EN %(m/m) Sulphur content DIN EN ISO mg/kg Aromatics DIN EN %(m/m) As catalyst a commercially available pre-reforming catalyst based on nickel produced by Haldor Topsøe A/S was used. The experiment parameters were derived from the results shown in work of Nehter et al. [6]. For this experiment the thermal power input by diesel fuel was 0.9 kw and the oxygen-to-carbon ratio was adjusted to 3.0. For extension of catalyst lifetime synthetic anode off gas of a SOFC fed by pre-reforming product gas was added to steam before inlet of the mixing zone. It was assumed that the SOFC will utilize 50 % of the fuel gas. Furthermore, 50 % of the off gas of the anode would be recirculated into the pre-forming reactor. These mass flows were added to the overall steam mass flow. The overall global hourly space velocity (GHSV) was adjusted to 825 h -1 for complete experimental time. In Figure 2, the evolution of the main species hydrogen (H 2 ), methane (CH 4 ), carbon monoxide (CO) and carbon dioxide (CO 2 ) are plotted over time. The experiment was run with constant mass flows for all gaseous

5 and liquid fluids. There were several interruptions of the experiment due to standard maintenance of the test rig and these are clearly visible by the large spikes in Figure 2. Every 24 hours gas samples for GC/FID measurements were taken during the experiment, whereas condensate samples were taken every 50 to 100 hours. Both were used for detection of higher hydrocarbons other than methane in the product gas. No regeneration procedure of catalyst bed was necessary during the first 3200 hours. The experiment was stopped after 3200 hours because on top of a condensate sample a phase of liquid hydrocarbons occurred. This indicated significant deactivation of the catalyst bed dry gas composition in % v/v 50 H2 40 CO 2 30 CH CO runtime in hours Figure 2: Dry gas composition of product gas, evolution over time-on-stream As can be seen in the Figure 2 the hydrogen concentration, as a function of time-on-stream, increases by approximately 10 % relative to the start concentration of the experiment. Additionally, the methane concentration decreases in the same relation over time. This can be an indication of catalyst deactivation. However, a detailed analysis of the catalyst should be performed to justify this statement. The temperature measurement on the reactor main axis showed that over time the axial temperature profile flattens with all other parameters being constant. In Figure 3 six discrete temperature profiles are displayed. The temperature drop, due to the steam reforming reaction in the first 5% of the catalyst bed, decreases over time. At the beginning of the experiment after 50 hours time-on-stream the temperature dropped from 472 C at position 1 to 452 C at position 2 after 5% of catalyst bed length. After 3000 hours time-on-stream this temperature drop changed in position and range. At position 1 a value of 458 C and at position C was detected. Because of endothermic nature of the steam reforming, this shows that the activity of the catalyst lowered for the steam reforming reaction over experiment time. Also the temperature increase between 20 % and 75 % of the catalyst bed flattens over time. In this part of the reactor methanation and water-gas-shift reactions are the dominant reaction paths. The lower temperature increase indicates lower catalyst activity for the mentioned reactions.

6 Figure 3: Temperature profiles on main axis at discrete timestamps of time-on-stream Measurements with GC/FID system of gaseous samples resulted for the whole experiment time in detection of no hydrocarbons other than methane up to C 5. In Figure 4 condensate samples from different time-on-stream timestamps are displayed. All samples except the last one at 3208 hours time-on-stream were similar to the four left samples shown in Figure 4. No liquid hydrocarbons are visible in all samples taken during the long-term experiment except the last one. The liquid hydrocarbons on top of the sample after 3208 hours in Figure 4 indicate a catalyst deactivation. Therefore the long-term experiment was stopped after this result. The catalyst is currently being analysed on a detailed level. The error estimate with respect to lifetime is 94 hours because of the timespan between condensate samples, with a single experimental run. 50 hours 504 hours 991 hours 3114 hours 3208 hours Figure 4: Condensate samples until 3208 hours time-on-stream

7 All of these measurements show a highly suitable pre-reforming product gas which can be utilised for a fuel cell system based on a SOFC. Most of the knowledge about pre-reforming of hydrocarbons could be transferred and the lifetime of the catalyst bed could be significantly increased from less than 1000 hours time-on-stream described by Nehter el al. [6] to 3200 hours time-on-stream. 5. CONCLUSIONS Pre-reforming of hydrocarbons, especially short-chained ones (C 1 -C 5 ), is a well-known technique for largescale petrochemical plants to increase efficiency of production of hydrogen and ammonium compared to a stand-alone steam reformer. In comparison to steam reforming with a temperature level of around 750 C, pre-reforming has a lower temperature level between 450 C and 550 C. Pre-reforming is known as a highly efficient process to produce a homogeneous feedstock for a downstream fuel cell or steam reformer. In addition, there are further advantages like superheating of the product gas to around 700 C downstream of the pre-reforming reactor which will diminish the formation of carbon deposition. A test rig for detailed research of this kind of fuel processor was built. It is a laboratory scale tubular reactor where the main influences of operational parameters on lifetime of nickel based catalysts can be investigated. A long-term experiment with European standard road diesel fuel was run successfully for 3200 hours at this test rig without regeneration of the catalyst bed. The addition of synthetic anode off gas recirculation increased the catalyst lifetime significantly from less than 1000 hours to 3200 hours time-on-stream. Dry gas composition of the product gas was monitored online and condensate samples for observation of phase development were taken. Gas samples for GC/FID analysis were taken on regular basis. The measured species in the product gas stream were hydrogen, methane, carbon monoxide and carbon dioxide whereas the measurements with GC/FID showed no traces of hydrocarbons other than methane up to C 5 molecules within experiment time. Change of concentrations over time was less than 10 % relative to the start of the long-term experiment. Quality of the product gas of pre-reforming process was highly suitable for a SOFC for complete experimental run of 3200 hours. Currently no reason could be identified which would cause fast degradation of a downstream SOFC due to higher hydrocarbons or carbon deposition. Although additional and detailed research is still needed the present results demonstrated that a fuel processor based on the process of pre-reforming can feed a fuel cell system with a SOFC by use of diesel fuel. ACKNOWLEDGEMENT This research and development project Nationales Innovationsprogramm Wasserstoff- und Brennstoffzellentechnologie (NIP): SchIBZ Schiffs Integration Brennstoffzelle is funded by the German Federal Ministry of Transport, Building and Urban Development (BMVBS) (Bundesministerium für Verkehr, Bau und Stadtentwicklung) under contract 03BI206A. REFERENCES 1. P. Nehter, J. Bøgild Hansen, P. Koch Larsen, 2011, A techno-economic comparison of fuel processors utilizing diesel for solid oxide fuel cell auxiliary power units, Journal of Power Sources, vol. 196, n 17, pp K. Leites, A. Bauschulte, M. Dragon, S. Krummrich, P. Nehter, 2012, SchIBZ - Design Of Different Diesel Based Fuel Cell Systems for Seagoing Vessels and Their Evaluation, ECS Transactions, vol. 42, pp S. Krummrich, B. Tuinstra, G. Kraaij, J. Roes, H. Olgun, 2006, Diesel fuel processing for fuel cells - DESIRE, Journal of Power Sources, vol. 160, n 1, pp G. Steinfeld, R. Sanderson, H. Ghezel-Ayagh, S. Abens, Spring 2000, Distillate fuel processing for marine fuel cell applications, AICHE Meeting, Atlanta, USA 5. R. Peters, E. Riensche, P. Cremer, 2000, Pre-reforming of natural gas in solid oxide fuel-cell systems, Journal of Power Sources, vol. 86, pp P. Nehter, H. Modarresi, N. Kleinohl, J. Bøgild Hansen, A. Bauschulte, J. vom Schloss, K. Lucka, 2012, Adiabatic prereforming of ultra-low sulfur diesel: Potential for marine SOFC-systems and experimental results, European Fuel Cell Forum, Lucerne, Switzerland

8 7. J.R. Rostrup-Nielsen, Catalysis: Science and Technology, Volume 5, 1984, J.R. Anderson, M. Boudart, 1st ed., Springer-Verlag, Berlin. 8. J. Boon, E. van Dijk, July 2008, Adiabatic Diesel Pre-reforming: Literature Survey, ECN Hydrogen and Clean Fossil Fuels, ECN-E N. Kleinohl, A. Bauschulte, K. Lucka, 2011, Pre-reforming of liquid hydrocarbons at atmospheric pressure, European Fuel Cell Forum, Lucerne, Switzerland