NO X storage on heavy-duty diesel vehicles

Transcription

1 NO X storage on heavy-duty diesel vehicles Final report project P Swedish Energy Agency Ingemar Odenbrand Klaus Papadakis Department of Chemical Engineering, Lund Institute of Technology Lund University, February 2004

2 Contents Contents... 2 Summary... 3 NO X storage technology... 4 Bench scale investigation... 5 Build up of the engine rig... 6 Build up, test and control of the mechanical components of the exhaust system... 7 Size of the catalyst flow... 8 Radial distribution of injected hydrocarbons... 9 Degree of NO X reduction... 9 Catalyst model Statistical designs for optimization of stationary experiments A first ETC cycle test Conferences Publications

3 Summary The NO X storage and reduction technology was used to illustrate how it is possible to reduce the amounts of nitrogen oxides emitted from heavy-duty diesel vehicles. The system is complex and has been investigated in bench scale as well as in full scale on an 11 litre engine mounted in a rig. The method uses a catalyst system consisting of oxidation catalysts and NO X traps. NO X is stored in the traps under normal conditions and is reduced with the aid of repeated short injections of hydrocarbons (mostly diesel) each or every other minute. A catalyst model that describes the function of the NO X traps has been developed and adjusted to data from test runs. Statistical design was used to evaluate data from stationary experiments performed in order to give a basis for transient experiments. Results from the first transient cycle showed that 3.5 to 3.6 g NO X /kwh (54 %) was reduced with a fuel penalty of 5.8 %. 3

4 NO X storage technology The emission of NO X from heavy-duty vehicles is one of todays biggest environmental problems. NO X storage technology is one among other measures which is able to reduce the NO X emission of trucks. It is a technique in which NO X is stored on the catalyst surface under lean conditions and then reduced with the aid of repeated hydrocarbon injections every or every other minute. At Lund Institute of Technology/Department of Chemical Engineering an engine rig with such an exhaust system has been built up. This was done with the help of the Swedish Energy Agency s financing and in cooperation with the Division of Combustion Engines at Lund Institute of Technology (LTH) and the Department of Chemical Reaction Engineering at Chalmers University of Technology. Further, representatives of the truck industry, namely Volvo and Scania and also the catalyst manufacturer Johnson Matthey, participate in the project. The NO X storage technology was discovered in 1994 by Toyota and has then been further investigated by many international research groups. The reaction mechanism can simplified be described in the way that NO 2 is stored on a BaO-surface as Ba(NO 3 ) 2. NO is not stored and has first to be oxidized to NO 2 on Pt particles on the catalysts surface what proceeds under oxygen rich conditions. The NO X storage time can be one or two minutes. Then, hydrocarbons are injected under a short period (1-20 s) so that a λ value below one (oxygen deficient conditions) is reached. NO X is desorbed and reduced to nitrogen probably on Rh. The period when NO 2 is stored as Ba(NO 3 ) 2, is called NO X storage period, the period when it is reduced regeneration period. Together they form the NO X storage- and regeneration cycle with a respective cycle time. If the technology shall be applied on trucks the reducing agent can be diesel fuel because it is already present on the truck. In this project, diesel fuel is used to reach conditions as realistic as possible. It is injected with an injector which is placed 80 cm upstream the catalyst in the exhaust pipe. The fuel is injected against the exhaust flow to achieve a good mixing of the fuel with the exhaust. The period when the diesel fuel is injected is called injection time. Figure 1 shows the mechanism for the NO X storage technology. Figure 1. Mechanism for the NO X storage technology. 4

5 Bench scale investigation In the beginning of the project, there was a lot of literature about work with NO X traps in laboratory scale, i.e., most researchers worked with catalysts of just a few centimeters size. The question arose if results from such experiments could be applied in a full-scale setup with catalysts with a size of l. There are many significant differences between a small and a big setup. In laboratory experiments, special gas mixtures are used and not real exhaust which varies in temperature, flow and composition. These three parameters can have some influence on the catalysts function. Further, a small catalyst can be placed in an oven and the temperature can be controlled quite accurately. A big catalyst can only be heated up when warm exhaust flows through it. This means that a large catalyst can easily be cooled down too much so that it gets too cold to perform well. It can also be heated up too much. Also the reducing agent is not the same. In most laboratory investigations pure organic substances are used, e.g., propene. On a truck, however, diesel fuel is used because it is already present on the truck. The diesel fuel consists of hundreds of substances of which many do not react as good as propene. In full-scale, further auxiliary catalysts upstream the NO X trap are applied to pre oxidize the reducing agent. These catalysts influence also the NO X reduction degree. Thus, it is difficult to draw correct conclusions from a laboratory experiment to a full-scale experiment. Under 2000 an investigation was done on the conclusions that can be drawn from laboratory experiments and how they should be interpreted. A bench scale reactor was built up with an oven in which a small catalyst was placed. Now, different experiments could be done. Experiments with a synthetic gas mixture and with exhausts from a 6.7 l truck engine were performed. A huge number of experiments was done and certain parameters investigated, e.g., the influence of the catalyst size, the reducing agent and which kind of exhaust was used. It was even investigated how different auxiliary catalysts influence the NO X reduction degree in bench scale. In some experiments a three way- or an oxidation catalyst was placed upstream the NO X trap. In other experiments only a NO X trap was used. The results showed how the emissions of the different gases (CO, HC, N 2 O) were influenced as well as the NO X reduction. The experiments were planned and analyzed statistically. The results where thus complex and showed a strong influence of the parameters, e.g., propene reacted much better than diesel fuel so that a higher NO X reduction degree was achieved in laboratory scale with this reducing agent than in fullscale with diesel fuel as the reducing agent. One should thus always be careful with an interpretation of laboratory scale results. Further, it has been shown that always some kind of bypass should be used when injecting diesel fuel into the exhaust pipe which can heat up very fast what can destroy the catalysts. In bench scale, the temperature rises strongly because of the combustion of hydrocarbons while this is not the case in laboratory scale. Propene reacts faster than n-heptane or i- octane and reacts completely already at 375 C. N-heptane and i-octane react much slower and further into the monolith. The hydrocarbon emission can be easily underestimated in laboratory scale because propene has a lower ignition temperature as n- heptane. The NO 2 emission was higher in setups with auxiliary catalysts because the NO 2 content in the exhaust depends on the catalyst composition. The NO 2 fraction of the NO X content was up to 50 % or ppm. The NO 2 emission was reduced by a longer 5

6 reduction time. A longer NO X storage time increased thee NO 2 emission only in laboratory scale. This can have a positive influence on the NO X storage on the catalyst. Thus, combinations of oxidation catalysts and NO X traps can be advantageous in applications. The significance of combination systems is, however, overseen in laboratory experiments. NO X reduction normalized towards fuel penalty has been underestimated in the laboratory experiments at low temperatures and overestimated at higher temperatures. This depends probably on the lack of water in the laboratory experiments. NO X reduction normalized towards fuel penalty was in the laboratory experiments in the NO X trap-only setup approximately 18 % at 225 C and 45 % at 425 C. In the bench scale experiments the respective values were 17 and 75 % even for the combination catalyst system. In the laboratory experiments the combination catalyst system showed a much higher NO X reduction at 225 C (55 %) but only 20 % at 425 C. These values show that the performance of the combination catalyst system can not be understood from the laboratory experiments. In the experiments with the NO X trap-only -setup the normalized NO X reduction was 0 % below 300 C when n-heptane or i-octane were used but 30 % at 325 C and 85 % at 425 C. Build up of the engine rig In 2001 the engine rig was built up at LTH consisting of an 11 l Scania diesel engine which normally is used in heavy-duty trucks, NO X storage catalysts, oxidation catalysts, an exhaust analysis system and a control system for controlling the engine and the catalyst system. The exhaust system is equipped with a bypass to be able to reduce the flow through the catalyst. The setup of the exhaust system is shown in Figure 2. Oxidation catalyst Bypass NO X trap catalyst T G InjectorT T, G T λt T, G, λ G Flow Exhaust Pressure Governors (EPG) Figure 2. The setup of the exhaust system. G =Gas sampling point λ = Lambda sensor T = Temperature The bypass is controlled pneumatically with two different kinds of valves. There is a butterfly valve on each pipe which either can be open or closed. There is further an exhaust pressure governor (EPG), a small bypass which can be open or closed. If the butterfly valve is closed and the EPG open there is still a small flow through the catalysts. 6

7 The flow size is approximately 10 % of the total flow. Figure 3 shows the valve system which controls the mass flow through the bypass. Figure 3. Valve system of the bypass and exhaust pressure governor. Build up, test and control of the mechanical components of the exhaust system Under 2002 the exhaust system was built up and a number of experiments were done to achieve an understanding on how the system actually works. The bypass, which should bypass a large amount of the exhaust under the regeneration period was installed. This was necessary because of the high oxygen content in the diesel exhaust. Large amounts of reducing agent would have to be injected to reach rich conditions at full flow and this leads to a high fuel consumption because much fuel reacts with the oxygen in the exhaust. Further, the released reaction heat heats up the catalyst which then can be thermally damaged. With the installation of the bypass new questions arose which we had to answer. The first question was how big the bypassed mass flow actually should be. To be able to reach rich conditions with a reasonable amount of diesel fuel, the mass flow through the catalysts had to be reduced. The problem arose that it was difficult to control the valves reasonably fast and precisely for each new load point. This depends partly on the fact that both temperature and exhaust pressure have their part in how large the mass flow through the catalysts will be when the valve is closed partly. A functional control system for the bypass would also be very complex and very expensive. Thus, we decided to control the valves pneumatically so that they are always in the same position under the regeneration time independent of the engines load point. 7

8 Then, the size of the catalyst flow has been examined. As long as a λ value lower than 1 can be reached with a reasonable amount of hydrocarbons it is sufficient to know how large the flow through the catalysts is to be able to model it. This is more important than always having the same flow under the regeneration period. It has been shown in the first NO X reduction experiments (in spring 2002) that it was possible to reduce NO X with 0.8 g/s which at that time was the largest amount in our dosing system. From calculations it was known that up to 4 g/s should be needed at some load points. The size of the dosing system increased and the amount of 2.3 g/s was at this time (in autumn 2002) the biggest amount which could be injected. Now, we only had to find a valve position which fitted good for all load points, i.e., where rich conditions could be reached with approximately 2.3 g/s injected amount. To examine this, the catalyst valve has been opened in different positions with the help of a manual switch. The bypass valve was completely open. Then, approximately 20 g diesel fuel were injected in 10 s. It has been shown that λ<1 was reached at all load points when the butterfly valve on the catalyst side was closed and only the EPG was open. Now, it was clear that the mass flow through the catalyst can be reduced in a cheap and simple way by keeping the EPG in the bypass pipe always closed and the EPG in the catalyst pipe always open. The system can now be controlled by opening or closing the butterfly valves in both pipes completely. Size of the catalyst flow Another investigation aimed at the size of the reduced mass flow which flows through the catalyst. In the engine rig the input air flow is measured with the help of a thermal sensor in the air intake of the engine. Further, the engine s fuel consumption is measured. From the air flow and the fuel consumption the full exhaust flow can be calculated. The reduced flow through the catalysts was examined by measurements of the exhaust pressure loss over the catalyst monoliths under different driving conditions. It has been shown that the reduced flow rises from 4 to 12 % when the total exhaust flow increases in our working range. Figure 4 shows the reduced mass flow through the catalyst as a function of the total mass flow. This flow is used in catalyst simulations and calculations of the gas loads Reduced mass flow [%] Total mass flow [kg/h] 8

9 Figure 4. The reduced mass flow as a function of the total mass flow. Radial distribution of injected hydrocarbons Another important investigation aimed at the radial distribution of the injected hydrocarbons. The diameter of the catalyst is 28 cm and it is important that the hydrocarbons are evenly distributed over the whole diameter so that the whole catalyst is used. The hydrocarbons which are injected into the exhaust pipe combust in the oxidation catalysts. Here, heat evolutes and an increased temperature can be measured on the catalysts where the hydrocarbons were combusted. The investigation was done by placing three thermocouples at different radial positions into the first oxidation catalyst. The injection was done at different load points and the temperature increases were registered. Two different injectors were tested, a 1-hole injector which injected against the exhaust flow and a 6-hole injector which injected at different radial directions. The 6-hole injector showed an instable injection especially at low flows. It was thus not used anymore. The 1-hole injector showed a good distribution at all three measurement points and was used in the following. Figure 5 shows both injectors. Air- and fuel flow Ø 1 mm Ø 1 mm Ø 6 mm Ø 6 mm 1-hole injector Figure 5. The injectors. 6-hole injector Degree of NO X reduction The next step in the project was the investigation of the NO X reduction degree. This is a function of different variables. The catalyst temperature has a large effect on the NO X reduction degree. There are further parameters which can be controlled by the control system. Some of the most important are: the NO X storage time, the regeneration time, the injected amount of reducing agent, the injection rate and the period when the bypass is open and the exhaust flow through catalyst is reduced (flow bypass time). Further, it is important in the investigation at which parameters a high NO X reduction can be achieved and at the same time the fuel penalty kept low. 9

10 The first investigation was done to study the influence of the temperature on the NO X reduction. Many stationary experiments were done at different load points. The cycle time was 4 min, the injection time was 10 s and the injected amount 35 g. The time when the bypass was open was 10 s. The examined temperature intervall was between 300 and 600 C. The results are shown in Figure 6. The NO X reduction degree was between 20 and 60 %. The maximum NO X reduction degree was found to be at 450 C NO X reduction [%] Temperature in NO X trap [ C] Figure 6. NO X reduction degree as a function of the catalyst temperature. The following investigation was done to answer the questions on how the system parameters injection time, amount and rate and also the cycle time and the bypass time affect the NO X reduction degree. A large series of single experiments was done where all parameters were kept constant except of one which was varied. The following parameter values were used: The injection amount under one cycle was 20, 35 and 55 g, the injection time was 10 and 20 s, the flow bypass time 10, 15 and 20 s and the cycle time 160 and 240 s. The NO X reduction degree which was achieved was compared as a function of temperature. The result was that a shorter cycle time lead to an increased NO X reduction degree at all temperatures. The effect of the cycle time on the NO X reduction degree is shown in Figure 7. 10

11 NO X reduction [%] s cycle time 160 s cycle time Temperature in NO X trap [ C] Figure 7. NO X reduction degree for 160 and 240 s cycle time. An increased injection amount lead to an increased NO X reduction degree by % at temperatures below 500 C. Figure 8 shows the NO X reduction degree at injections of 20, 35, and 55 g diesel fuel under one cycle NO X reduction [%] g 35 g 55 g Temperature in NO X trap [ C] Figure 8. NO X reduction degree for 20, 35 and 55 g injected amount (in one cycle). 11

12 A similar effect had a prolonged flow bypass time. It has thus been shown that it is possible to raise the NO X reduction degree with the help of the system parameters when the temperature is below 500 C. Figure 9 shows the NO X reduction degree for different flow bypass times. The injection time was always 10 s NO X reduction [%] s bypass time 15 s bypass time 20 s bypass time Temperature in NO X trap [ C] Figure 9. NO X reduction degree for 10, 15 and 20 s flow bypass time. Catalyst model How it is possible to change the reduction degree with the help of parameter changes at temperatures below 500 C could be explained when the catalyst model was developed. In the experiments done to develop the catalyst model hydrocarbons were injected and the concentrations measured upstream and downstream the NO X traps. For measuring concentrations directly before and after the catalyst, the exhaust sample had to be diluted with air. The reason for this was that approximately 90 % of the exhaust flow were taken away under the regeneration period and led through the bypass. When a certain amount of hydrocarbons was injected into the reduced flow through the catalyst their concentration increased drastically. In our experiments the concentrations were so high that the measurement ranges of the instruments were exceeded. After dilution with air the experiments could be done as planned. The catalyst model consists of a part for mass and heat transport and a kinetic part. The latter was developed in laboratory experiments at Chalmers University of Technology and adapted to the conditions in the engine rig. Most of the steps in the reactions were taken away. The kinetic is now described by 9 equations for NO X adsorption, catalyst regeneration, NO X reduction, storage of oxygen and oxidation of hydrocarbons and carbon monoxide. The model shall be used on-line as a control model. This means that the time for the calculations should be as short as possible. Thus, the transport model has been simplified using the following assumptions: 12

13 The radial distribution of concentrations, temperatures and flows was even. Axial diffusion and heat exchange was neglected. The concentrations in the gas phase were always constant, i.e., only the concentrations on the catalyst surface were assumed to change. The diffusion resistance in the catalyst wash coat was neglected. Heat transport resistance in the wash coat was neglected. The monoliths were modeled as a series of 25 tank reactors. The catalyst model which was developed showed a good agreement with the experimental data under the NO X storage period. The oxidation catalysts were not part of the modeling and the agreement of the model with the experiments is low under the regeneration periods. Figure 10 shows a comparison between the catalyst model prediction of NO X concentrations downstream the catalysts and the experimental data at different temperatures. Figure 10. Comparison between the catalyst model prediction and experimental data at three different temperatures: (a) 330 C, (b) 430 C and (c) 530 C. With the help of the model it could be shown that nearly all NO X storage sites on the catalyst surface were used at low temperature. Figure 11 shows the predicted occupancies for NO X at different temperatures. At low temperature (a) nearly all NO X storage sites are used but under regeneration the occupancy is decreased by only 20 %. This means that a large amount of NO X can not be reduced at low temperature. At high temperature the situation is reversed (c). The fractional occupancy is only 20 % but all sites can be reduced. This was the reason why the NO X reduction degree could not be raised with the 13

14 help of the system parameters at high temperature. However, at low temperature the NO X reduction could be improved. Figure 11. Predicted site occupancies for NO X at three different temperatures: (a) 330 C, (b) 430 C and (c) 530 C. Statistical designs for optimization of stationary experiments The last investigation which was done with stationary experiments should show which system parameters should be applied to develop a simple dosing strategy to control the injection in an ETC cycle. The investigation should specify how much diesel fuel should be injected under which time and how long the bypass should be open and further how long the cycle time should be for a given temperature. To approach the aim with a low number of experiments, statistical experimental planning was used. Experiments were done at 4 different load points and approximately 10 single experiments were performed at each load point. The best results are shown in Table 1. To reach the best result, the cycle time should be 160 s at temperatures below 500 C, 80 s at 520 C and 65 s at 540 C. The injection time should be 5 s at temperatures above 500 C, 10 s at 460 C and 20 s at 370 C. The bypass time should be reduced from 40 s to 5 s between 370 C and 540 C. The injected amount of diesel fuel should be approximately 40 g below 500 C, 19 g at 520 C and 10 g at 540 C. With these parameters NO X reduction degrees between 51 and 63 % were reached. The fuel penalty was between 2.5 and 2.9 %. Only at the load point with temperature 370 C the fuel penalty was 5.0 %. 14

15 Table 1. Best parameters and results from the design experiments. Temperature in NOX trap [ C]: Cycle time [s]: Injection time [s]: Flow bypass time [s]: Injection rate [g/s]: Injected amount [g]: NOX reduction [%]: Fuel penalty [%]: A first ETC cycle test These parameters were used in a first temperature based dosing strategy. It is important to understand that the best experiments were only best within the limits of the designs. There can be even better system parameters. An examination of the original data files has shown that two of the designs (at 520 C and 460 C) had generally to low cycle times. The parameters which were thought to be better to achieve a high NO X reduction were changed and the dosing strategy developed. The parameters which are shown in Table 2 were used in the ETC tests. Load point Table 2. Parameters for thee first dosing strategy rpm 1000 Nm 540 C 1250 rpm 1000 Nm 520 C 1500 rpm 1000 Nm 460 C 1500 rpm 500 Nm 370 C Cycle time [s]: Injection time [s]: Bypass time [s]: Storage time [s]: Injection speed [g/s]: Injection amount [g]: The different values are temperature based. They were related with a linear function based on the temperature which is measured in the NO X trap. Because the reaction heat from the oxidation catalysts heats up the NO X trap approximately half a minute after the injection, the temperature in the NO X traps was read earliest one minute after the injection. Instead for using the cycle time, a temperature based NO X storage capacity is used. How much NO X there is in the catalyst is calculated with the help of a simple engine model. The dosing strategy has been tested in an ETC cycle. The reduction degree which was achieved was 54 % with 5.8 % fuel penalty which is an absolute reduction of g NO X /kwh. The NO X reduction is on a level with the best values which were demonstrated worldwide. However, a lower fuel penalty could be achieved if the engine was of, e.g., Euro IV class instead of Euro II which we use. 15

16 Conferences 23rd Task Leaders Meeting of the IEA Implementing Agreement, Energy Conservation and Emissions Regulation, September 9-12, 2001, Kauai, Hawaii, USA Program Conference: Energy System in Road Vehicles, November 14-15, 2001, Skövde, Sweden 10 th Nordic Symposium on Catalysis, "Bridging the Gaps in Catalysis", June 2-4, 2002, Marienlyst, Helsingør, Denmark 24th Task Leaders Meeting of the IEA Implementing Agreement, Energy Conservation and Emissions Regulation, June 23-26, 2002, Trondheim, Norway EUCHEM CONFERENCE on Environmental Catalysis, November 27 - December 1, 2002, Göteborg, Sweden Program Conference: Energy System in Road Vehicles, February 4-5, 2003, Södertälje, Sweden 2003 JSAE/SAE International Spring Fuels & Lubricants Meeting, May 19-22, 2003, Yokohama, Japan 25th Task Leaders Meeting of the IEA Implementing Agreement, Energy Conservation and Emissions Regulation, September 7-10, 2003, Faringdon, United Kingdom CAPoC 6, Sixth International Congress on Catalysis and Automotive Pollution Control, October 22-24, 2003, Brussels, Belgium Publications Bench Scale NO X Trap System for Diesel Applications, C. Künkel, and C.U.I. Odenbrand, Soc. Automot. Eng., [Spec. Publ.] SP (2001), SP-1644 (General Emissions and Gasoline Emission Control Systems), Catalytic Reduction of NO X on Heavy-Duty Trucks, C. Künkel, Doctoral Thesis, Department of Chemical Engineering II, Lund University, Institute of Technology, September Stationary NO X Storage and Reduction Experiments on a Heavy-Duty Diesel Engine Rig Using a Bypass System, K. Papadakis, C.U.I. Odenbrand, D. Creaser, SAE Technical Paper Series , Evaluation of a NO X Reduction System on an Engine Rig under Stationary Operation, K. Papadakis, C.U.I. Odenbrand, D. Creaser, Preprints CAPoC 6, Poster Sessions, Volume 2, p , Sixth International Congress on Catalysis and Automotive Pollution Control, October 22-24, 2003, Brussels, Belgium. Exhaust Gas Cleaning by Means of NO X storage and Reduction Technology on Heavy-Duty Diesel Vehicles, K. Papadakis, Licentiate Thesis, Department of Chemical Engineering, Lund University, Institute of Technology, December