Keihin Technical Review Vol.6 (2017) Technical Paper Development of Bi-Fuel Systems for Satisfying Fuel Properties Takayuki SHIMATSU *1 Key Words:, NGV, Bi-fuel add-on system, Fuel properties 1. Introduction With the reduction of CO2 emissions from vehicles becoming a major issue worldwide, (Compressed Natural Gas) with less CO2 emissions from combustion, has come to draw much attention. From the viewpoint of the balance between low fuel costs and energy supply, NGVs (Natural Gas Vehicles), which use as fuel, have become widespread in the Asian and South American regions. However, the overall proportion of NGVs in the world is still at a low level, for reasons such as insufficiency of filling stations, short driving range, and the necessity of large storage tanks. Fig. 1 shows the global sales outlook for passenger lightduty vehicles up to 2050 (1). PLDV annual sales (millions) Improve case 250 200 NGVs (incl., and LPG) 150 100 50 0 2000 2010 2020 2030 2040 2050 Plug-in hybrid diesel Plug-in hybrid gasoline Electricity FCEV Diesel /LPG hybrid Diesel hybrid Fig. 1 Sales forecasts for passenger light-duty vehicles Such market trends potentially demand that both OEMs and suppliers need to construct NGV systems at minimum development costs. In terms of this demand, the bi-fuel system, which uses gasoline as the base fuel along with switch-selectable fuel, has been given as a representative of NGV configurations. Because such a system requires only minor modifications to existing gasoline-powered vehicles, it is advantageous in reducing the cost of product development for a small lot of vehicles, and also in ease of utilization of the vehicles with the choice of running by gasoline in regions with insufficient infrastructure of gas filling stations. Since the bi-fuel system requires calibration for two different types of fuel, it is necessary to understand the basic characteristics of each fuel and to perform the control properly. However, the composition of depends on the area of production. Variation in gas composition influences the fuel air equivalence ratio, thus causing impact on exhaust emissions and power output, it is necessary to adjust the fuel injection quantity according to the gas composition. In this paper, comparative analyses of fuel properties have been conducted on power output characteristics, fuel air equivalence ratio characteristics and ignition timing characteristics, in order to specify the topics and countermeasures Technical Papers Received 26 May 2017, use of copyrighted material with permission from IAV GmbH, original publication in, Presentation Book of the 11th Conference on Gas-Powered Vehicles: Gaseous-Fuel Drives and Climate Protection Targets: The Right Path, September 15-16, 2016, Potsdam. Copyright 2016 IAV GmbH. *1 System Development Department, R&D Operations -21-
Development of Bi-Fuel Systems for Satisfying Fuel Properties of the bi-fuel system. Furthermore, learning control algorithms for variations in fuel composition have been provided, and proved through vehicle tests. 2. System Structure Table 1 Test engine specification Engine Type In line 4 cylinder NA Displacement 1600cc Fuel Type / Bi-Fuel ECU System Structure 2nd ECU add-on system Fig. 2 shows the development concept of a bifuel add-on system. With utilizing the ECU (1st ECU) for the gasoline system of the base vehicle, an add-on ECU (2nd ECU) that controls the system is installed. The 2nd ECU calculates the fuel injection quantity using information of the gasoline fuel injector pulse from the 1st ECU. Appropriate ignition timing according to operating conditions of the engine for fuel is calculated by the 2nd ECU, and sent to the 1st ECU via CAN bus, and the 1st ECU performs the ignition output. As mentioned, the 2nd ECU generally performs various controls based on gasoline fuel control information. However, when increasing fuel by gasoline fuel control in the acceleration phase, increasing the amount of fuel based on such information does not always result in the appropriate fuel control. Additionally, in regard to ignition timing, using the same ignition timing as that of gasoline fuel causes reduction in engine output. In order to formulate appropriate control for the 2nd ECU, it is necessary to understand the differences in properties between gasoline and fuels. 3. Fuel Properties Characteristics measurement at an engine test bench were conducted using an engine equipped with a switch-selectable bi-fuel system using gasoline fuel and fuel. Table 1 shows the specifications of the engine used in the measurement. 3.1. Power Output Properties fuel is a gaseous fuel, thus there is an influence on volumetric efficiency of intake air amount to the engine cylinders. Fig. 3 shows the degree of decrease in volumetric efficiency of intake air. The theoretical air fuel ratio (mass ratio) for methane - the main component of fuel - is 17.2, when the different densities of air and methane are into consideration, the theoretical air fuel ratio (volume ratio) becomes 9.5. That is, the methane volume is 9.5% regarding the air volume of 90.5%, thus the volumetric efficiency of methane by volume is relatively low. CH 4 9.5% 1st ECU Fuel Control Intake Air Control GF-INJ Pulse 2nd ECU Fuel Changeover Control Gas Fuel Control GAS-INJ -INJ η v 100% η v 90.5% Compensation for Fuel Composition Throttle Valve Ignition Control Ignition Plug Ignition Control via CAN Ignition Control Theoretical Air Fuel Ratio (Air/CH 4 in mass ratio) 17.2 Volume Ratio 9.5 : 1.0 Fig. 2 Add-on ECU structure Fig. 3 Volumetric efficiency for methane fuel -22-
Keihin Technical Review Vol.6 (2017) Additionally, the theoretical air fuel ratio (mass ratio) of gasoline is 14.7, the lower heating value of gasoline is 44.9 MJ/kg, and the lower heating value of methane is 50.0 MJ/kg. To illustrate this differently, if mixing 1 kg of air at the theoretical air fuel ratio, the lower heating values of gasoline and methane are 3.05 MJ/1 kg-air, and 2.91 MJ/1 kg-air, respectively, i.e., the lower heating value of methane is smaller. As a consequence, engine output tends to decrease using fuel, making techniques such as ignition timing adjustment described later necessary to compensate for this power reduction. 3.2. Fuel Air Equivalence Ratio Properties Combustion characteristics of gasoline and are different with respect to the equivalence ratio of the air fuel mixture. Fig. 4 shows the minimum ignition energy (2) related to the equivalence ratio difference between heptane, the representative component of gasoline fuel, and methane. It was found that heptane has a minimum ignition energy value (refer to 1) on the larger equivalence ratio side, and methane has a minimum value (refer to 2) slightly on the lower equivalence ratio side in comparison with the theoretical equivalence ratio. It indicates that gasoline fuel has a stable ignition region on the rich side, and fuel has a stable ignition region slightly on the lean side, from stoichiometry. The relationship between the equivalence ratio and power output characteristics for each fuel has been identified through measurements. Fig. 5 shows IMEP (Indicated Mean Efficient Pressure) versus the equivalence ratios for gasoline and, in which the vertical axis is non-dimensional IMEP at an equivalence ratio of φ = 1.0 for each fuel. It was found that power output of gasoline fuel becomes higher on the larger equivalence ratio side and power output of fuel becomes lower as the equivalence ratio increases. This indicates that, in contrast to the fact that burning velocity of gasoline rises for rich mixtures, burning velocity of fuel, after the maximum value at an equivalence ratio of φ = 1.1, descends at higher equivalence ratios (3). Therefore, from ignitability and engine power output points of view, gasoline fuel has a stable combustion region on the rich side, and fuel has a stable combustion region around stoichiometry. Generally, the gasoline fuel system carries out enrichment control/injection at the time of acceleration aiming at a stable combustion and Technical Papers Minimum ignition energy (mj) 4.0 2.0 1.0 Theoretical Equivalence Ratio 0.4 methane ethane propane 0.2 butane hexane heptane 0.1 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 3.0 3.3 3.6 Equivalence ratio φ (Source: E. Murase et. al Transactions of the Japan Society of Mechanical Engineers, Series B, Vol. 79, No. 805 (2013)) IMEP norm@stoich [-] 1.050 0 0.950 0.900 0.850 0.800 0.84 0.88 0.92 0.96 1 1.04 1.08 1.12 1.16 Equivalence Ratio φ 3000rpm IMEP 0.2MPa * IMEP is normalized by = 1.0 φ Fig. 4 Minimum ignition energy against equivalence ratio Fig. 5 Engine power output against equivalence ratio -23-
Development of Bi-Fuel Systems for Satisfying Fuel Properties prompt acceleration. However, performing enrichment control also when using operation based on the enrichment injection information from the 1st ECU, may cause unstable combustion and hesitant acceleration. It is required that fuel control, including acceleration, is performed in consideration of such characteristics by the 2nd ECU when using. 3.3. Ignition Timing Properties To analyze the combustion characteristics of fuel, internal pressure of the engine cylinders for gasoline fuel and fuel have been measured respectively. Fig. 6 shows the cylinder internal pressure behavior with respect to crank angle, and Fig. 7 shows the cylinder internal pressure behavior with respect to cylinder volume (P-V diagram). Engine speed at 3000 rpm, full load conditions, and ignition timing unified to that of gasoline operation, were set in the experiments. The resultant cylinder internal pressure is non-dimensional P-Max. As mentioned above, the heating value of Pcyl (-) 1.40 0.80 0.60 0.40 0.20 0.00-180 Fig. 6 Ignition Timing BTDC 30 deg -120-60 0 60 120 180 CA (deg) Cylinder pressure - crank angle Table 2 Phase Ign. ~ MFB 10% MFB 50% C.A. MFB 10% ~ 80% Crank angle for each phase 30.3 deg. ATDC 12.0 deg. 21.0 deg. 27.9 deg. ATDC 11.5 deg. 32.0 deg. fuel is lower, thus the maximum cylinder internal pressure (P-max) is 88% in comparison with gasoline. In reference to the P-V diagram, it is understood that a lower cylinder internal pressure causes the area surrounded by P-V (i.e., effective workload) to decrease. To analyze the combustion state in more detail, the combustion mass ratio was calculated from the cylinder internal pressure. Fig. 8 shows the combustion mass ratio with respect to the crank angle, and Table 2 shows crank angle between the states in Fig. 8. After ignition at BTDC 30 deg, comparing crank angle (1) where the combustion ratio MFB (mass fraction burned) is 10%, that for gasoline is 30.3 deg, and that for is shorter at 27.9 deg. When the MFB reaches 50% (2), the crank angle for gasoline is ATDC 12.0 deg, and for it is ATDC 11.5 deg, almost at equal levels. From MFB 10% to MFB 80% (3), the crank angle for gasoline is 21.0 deg, and for it is 32.0 deg and longer, which is reverse to 1. There exists the tendency that the burning velocity of is fast immediately after ignition in Pcyl (-) 0.80 0.60 0.40 0.20 0.00 0.06 Ignition Timing 0.13 0.25 0.50 vol (-) MFB (%) 120 100 80 60 40 20 0-90 Ignition Timing BTDC 30 deg 0 MFB 50% MFB 10% CA (deg) MFB 80% 90 180 * Ignition Timing: BTDC (deg) Fig. 7 P-V diagram Fig. 8 Mass fraction of burned fuel -24-
Keihin Technical Review Vol.6 (2017) phase 1, but the burning velocity is slower in the second half of combustion in phase 3. This is due to the chemical stability of methane, which is one of the reasons why the engine power output is lower. Furthermore, combustion characteristics of fuel have been analyzed by changing the ignition timing. fuel has a high octane number, and is superior in knocking resistance in comparison to gasoline. Fig. 9 shows the cylinder internal pressure behavior with respect to the crank angle, and Fig. 10 shows P-V diagrams for each ignition timing, in which the cylinder internal pressure is a non-dimensional P-Max when gasoline is burned at the theoretical air fuel ratio. As ignition timing is shifted from the retard side to the advance side, the P-max tends to increase. Although it is impossible with gasoline to advance the ignition timing from BTDC 30 deg due to the occurrence of knocking, for, it is possible to advance ignition timing, and obtain a P-max exceeding gasoline. However, as shown in Fig. 9, ignition timing advance makes the Pcyl (-) 1.4 BTDC 15 BTDC 18 BTDC 46 deg 1.2 BTDC 24 BTDC 30 1.0 BTDC 34 BTDC 40 BTDC 30 deg 0.8 BTDC 46 0.6 (IG. timing (deg.)) 0.4 0.2 0.0 BTDC 15 deg -180-120 -60 0 60 120 180 CA (deg) Fig. 9 Cylinder pressure - crank angle at each ignition timing period from ignition to P-max longer, a decrease of the area surrounded by P-V (i.e. effective workload) is observed. To confirm the influence on engine power output from cylinder internal pressure behavior due to an ignition timing change, measurements of the IMEP have been performed. Fig. 11 shows the P-max and the IMEP with respect to ignition timing. Although the P-max increases as ignition timing is advanced, the IMEP reaches the maximum value at a point where the crank angle is advanced 4 deg from the target ignition timing set for gasoline, and descends from that point. Because negative work occurs due to a pressure increase during the compression stroke, it is necessary to set the ignition timing based on not only P-max but also P-V characteristics. Since knocking tends to occur at low engine speeds, ignition timing for gasoline engines is set towards retard side. In this area, by utilizing the characteristics of a possible ignition timing advance for fuel, further improvement of power output is expected by adjusting the ignition timing. Pmax (-) 1.40 0.80 0.60 0.40 0.20 0.00 10.0 15.0 ( Pmax) Pmax IMEP IG. Timing ( IMEP) MAX. IMEP of 1.30 1.25 1.15 1.10 1.05 Advance 0.95 0.90 20.0 25.0 30.0 35.0 40.0 45.0 50.0 Ignition Timing BTDC (deg) IMEP (MPa) Technical Papers * P max : Maximum Pressure normalized by P max Pcyl (-) 1.40 0.80 0.60 0.40 0.20 0.00 0.06 BTDC 46 deg BTDC 15 BTDC 18 BTDC 24 BTDC 30 BTDC 34 BTDC 40 BTDC 46 (IG. timing (deg.)) BTDC 30 deg 0.13 BTDC 15 deg 0.25 0.50 vol (-) Fig. 11 Comparison of Pmax and IMEP against ignition timing 4. Compensation for Composition fuel composition varies depending on its production area. Taking the example of Thailand, Fig. 10 P-V diagram at each ignition timing the fuel composition varies widely within the same -25-
Development of Bi-Fuel Systems for Satisfying Fuel Properties country, and the value of Wobbe Index (WI) is allowed from 37 to 52 according to regulations. The WI is a value representing the amount of heat from the injection nozzle, and it is calculated from the gross heating value of gas divided by the square root of specific gravity, as shown in the following formula. WI = H S H: gross heating value of gas (MJ/Nm 3 ) S: specific gravity of gas (-) The variation rate of the WI is proportional to the injection mass of combustible component, meaning that the WI variation from 37 to 52 requires a correction of about 40% of injection quantity. The composition variation influences the fuel air equivalence ratio, which causes an increase of exhaust gas emission and a reduction of power output. Such composition variation also influences the fuel air ratio feedback control. At the time of restarting fuel injection after a fuel cut during deceleration, discontinuing fuel air ratio feedback in full load operation, and changing over between gasoline and, the fuel air ratio correction is insufficient, thus resulting in deterioration of exhaust emissions and operational performance. Machine learning algorithms have been developed for compensation of various fuels. The concept of compensation for composition is based on: executing learning control in the area where the difference in fuel composition is stable; compensating for the difference of fuel injection mass using an air-fuel ratio correction coefficient; and using independent learning values from that of gasoline fuel. By executing learning in the stable area, it is possible to avoid disturbances such as transient operation, or purge control of gasoline vapor, to perform accurate learning control. Fuel Cut Control Air-Fuel Ratio Feedback Control Fuel-Purge Action Fig. 13 Fig. 12 ON OFF ON OFF ON OFF Air-Fuel Ratio Correction Factor Difference of Fuel Composition Fuel Composition Learning Compensation Amount for Fuel Composition Execution area of compensation for fuel composition Air-fuel ratio correction factor First, the stable learning areas have been investigated. Fig. 12 shows a working example for learning immediately after lean gas is filled. In the fuel cut control during deceleration, air-fuel ratio feedback control and purge control are discontinued. After restart of injection, feedback control starts and after that, purge control begins. Learning control is executed in the area that purge control is discontinued and the air-fuel ratio feedback control is in operation. Next, stability of the air-fuel ratio at the restart of fuel injection after fuel cut has been confirmed. Fig. 13 shows the behavior of air fuel feedback correction coefficient when filling fuel of a different composition. Since fuel does not attach onto the wall surfaces like gasoline fuel, the air-fuel ratio correction factor changes so as to compensate for the difference of fuel injection mass, and tends to stabilize instantly. This indicates that, after restarting Working example of learning 1.25 Fuel Lean 1.2 1.15 1.1 1.05 1 0.95 0.9 0.85 Fuel Rich 0.8 0.75 0 0.5 1 1.5 2 2.5 3 Time (sec) Behavior of air fuel feedback coefficient 1.15 1.1 1.05 1 0.95 0.9 0.85-26-
Keihin Technical Review Vol.6 (2017) injection from fuel cut, the air-fuel ratio correction factor can accurately reflect the difference of fuel injection mass due to different composition so that this is possible to be adopted to learning control. Then, in order to confirm the behavior of learning control while the vehicle is running and the accuracy of compensation learning, testing on a chassis Compensation amount of fuel composition 20% compensable NEDC Mode speed (km/h) dynamometer has been conducted using the EUDC cycle as a representative. Two patterns were used to observe the degree of convergence of the learning value, i.e., for each test by filling 20% lean fuel from the initial state calibrated with rich fuel, and Fig. 15 0 10 20 Time (min) Behavior at switching rich gas from lean gas filling 20% rich fuel from the initial state calibrated with lean fuel. In order to confirm whether behavior changes due to outside air temperature, the outside air temperature was set at 0 C, 25 C and 45 C. Fig. 14 shows the behavior when switching to lean fuel from rich fuel, and Fig. 15 shows the behavior when switching to rich fuel from lean fuel. It has been confirmed that, at any outside air temperature for any pattern, the difference of fuel injection mass were accurately observed. Approximately 5 minutes 5. Conclusion (1) A 2nd ECU add-on bi-fuel system has been constructed for small quantity production NGVs. By utilizing the existing gasoline system to the maximum and understanding the difference of fuel characteristics for control, it is possible to reach a level comparable to a gasoline system with minimum development effort and cost. after the start of running, 15% from the expected value of 20% has been learned, about 10 minutes later, the expected learning value has been reached, and then, the learning value stays stable. (2) fuel causes lower engine power output due to a relatively low volumetric efficiency. It is necessary to recover the output reduction by combining optimization of air fuel ratio with Technical Papers ignition control. (3) has a stable combustion on the rich fuel Compensation amount of fuel composition 20% compensable NEDC Mode speed (km/h) side, but has a stable combustion around stoichiometry. The fuel control should target to stoichiometry, without depending on a gasoline increase in a transient state. (4) fuel has the characteristics that burning velocity becomes slower in the second half 0 10 20 Time (min) Target range 45 C 25 C 0 C NEDC Mode of combustion. It is possible to make use the superior knocking resistance of, so as to increase power output by advancing ignition timing. Fig. 14 Behavior at switching lean gas from rich gas (5) It has been confirmed that, in areas where the difference in fuel composition is observed to be -27-
Development of Bi-Fuel Systems for Satisfying Fuel Properties stable, learning control using an air-fuel ratio feedback correction coefficient can accurately compensate for the difference in fuel injection mass due to variation in composition. References (1) IEA-MMo2012 (2) E. Murase, O. Moriue, H. Hashimoto and I. Matsuzaki, A Study on Equivalence-Ratio Dependence of Minimum Ignition Energy Based on Initial Burning Velocity, Transactions of the Japan Society of Mechanical Engineers, Series B, Vol. 79, No. 805 (2013-9) (3) Y. Yamamoto and T. Tachibana, Comparison of Combustion Characteristics of Methane-Air and -Air Mixtures, Journal of the Japan Petroleum Institute, Vol. 40, No. 4, 1997 Author T. SHIMATSU As environmental concerns are rising day by day, fuel could be considered as an effective solution. Keihin s NGV development team will conduct further research so as to provide an energy efficient and eco-friendly system. I would like to thank everyone who joined the development and assisted me in organizing this paper. (SHIMATSU) -28-