Theoretical Development of a Simplified Electronic Fuel Injection System for Stationary Spark Ignition Engines ADRIA IRIMESCU Mechanical Engineering Faculty Politehnica University of Timisoara Bld Mihai Viteazul 1, 3222 Timisoara ROMAIA iamotors@yahoo.com Abstract: - The problem of emissions control is an important issue only for cogeneration systems over a certain power level. For natural gas or biogas fueled engines that power electric generators and feature exhaust gas heat recovery in micro-cogeneration installations, the quantity of pollutants expelled into the atmosphere is considered to be insignificant. evertheless, a tendency to reduce emissions for small power installations is noticed. Also, given the advantages of distributed power generation compared to centralized systems, small size applications will increase in numbers. Reducing carbon monoxide (CO), unburned hydrocarbons (HC) and nitrous oxide emission ( x ) can be achieved within the same catalyst bed by employing a three way catalytic converter. This emissions control system requires that the engine is operated as close to stoichiometric air-fuel ratio as possible in order to obtain maximum reduction efficiency. While carburetors ensure good engine operational characteristics, an electronic fuel injection system is required for running the engine at stoichiometric air-fuel ratio. Different control strategies are investigated and parameters that have to be monitored are evaluated for different situations such as cold start, idle, part load and full load operation. A simplified electronic fuel injection system is proposed based on these case studies. Key-Words: - spark ignition engines, electronic fuel injection, closed loop, control strategy 1 Introduction Reducing air pollution has become a major issue for most decision makers around the world. Until recently, the efforts of reducing emissions were concentrated on exhaust gas treatment for reducing particulate emissions, carbon monoxide (CO), nitrous oxides ( x ) and unburned hydrocarbons (HC). Even if it is not a toxic gas, carbon dioxide (CO 2 ) contributes to global warming through the green house effect it produces. For this reason, a reduction of CO 2 emissions is sought after as much as possible. To this end, the European Union adopted a plan to reduce green house gas emissions by 2 %, improve energy efficiency by 2 % and increase the share of renewable energy by 2 %, all by the year 22, compared to 199 [1]. Being the result of complete combustion of carbon, CO 2 emissions can only be reduced by increasing energy efficiency or through the use of renewable energy sources [2]. Cogeneration is a good method of increasing overall thermal efficiency, given that the heat produced is obtained by recovering waste heat contained in the flue gas. Even if the overall efficiency of a cogeneration power plant is around 85 %, transport and distribution losses can amount to 1-2 % for electric power and 5 % for heat delivered over long distances. As a comparison, cogeneration units powered by internal combustion engines and featuring waste heat recovery, can reach up to 9 % overall thermal efficiency, and transport losses are only 1 % for electric power and 2 % for heat delivery in a distributed system, resulting in overall thermal efficiency of 87 % [3], [4]. Micro-cogeneration systems powered by spark ignition (SI) engines, usually feature a simple fuelling system that employs a carburetor. In order to reduce CO, HC, and x emissions using a three way catalytic converter (figure 1), running the engine on a stoichiometric air-fuel ratio is required for obtaining high conversion efficiency. Fig. 1. Exhaust system with oxygen sensor and catalytic converter [5] The electronic fuel injection (EFI) system needed for maintaining relative air-fuel ratio as close to unity as possible (λ = 1) requires an array of sensors that monitor ISS: 1792-5924 / ISS: 1792-594 231 ISB: 978-96-474-237-
different operating parameters. Therefore, developing a control strategy that ensures appropriate running conditions for the engine as well as the emissions control system, with a minimum set of sensors, can significantly reduce the overall costs of implementing electronically controlled fuelling. 2 System Layout and Control Strategy Basically, all EFI systems rely on measuring air flow to the engine and adjusting fuel flow accordingly. Several adjustments are also performed for different working parameters, such as air temperature, load, coolant temperature and so on. The quantity of air inducted into the engine can be measured using a volumetric or mass air flow (MAF) meter (figure 2). As an alternative, air flow can be calculated be measuring engine speed (), manifold absolute pressure (MAP) and intake air temperature (IAT) (figure 3). Automotive injection systems feature more sensors and actuators, but the simplest architectures are presented in figures 2 and 3. Sensors MAF TPS Coolant temperature Oxygen sensor Adjustment ECU Fuel flow Basic injection time Injection pulses Injector Fig. 2. EFI with MAF sensor After determining the air flow to the engine by measuring it directly with a MAF sensor or through calculations using the signals received from the MAP, IAT and engine speed () sensors, a basic injection time is set by the ECU. This injection time represents the time the injector is kept open during one or two rotations of the crank shaft. Injection pulses are coordinated with the signal received from the engine speed sensor. Other sensors like the coolant temperature, throttle positioning signal (TPS) and especially the oxygen sensor, are used to adjust the basic injection time. The air-fuel mixture is enriched when the engine is cold so that proper running conditions are ensured. TPS is used to evaluate engine load. When the throttle is completely or partially closed, the engine is running at idle or partial load and injection time is adjusted so that λ ~ 1. At wide open throttle, the engine needs to deliver maximum power and therefore the air-fuel mixture is enriched so that λ ~.85.95. Load evaluation is also used to adjust the ignition timing, in conjunction with the signal from the engine speed sensor. The most important correction is needed to maintain the air-fuel ratio close to stoichiometric value so that good fuel conversion efficiency is achieved, along with maximum catalytic conversion efficiency (figure 4). To this end, an oxygen sensor is used. As its name suggests, this sensor measures the concentration of oxygen (O 2 ) in the exhaust gas stream. If the air-fuel mixture is rich, very little O 2 is present in the exhaust and the signal delivered is high, with values around ~ 8 mv. On the contrary, if the mixture is lean, more O 2 is found in the exhaust and the signal is low, with a voltage value above ~ 1 mv. An important aspect is that the minimum working temperature of oxygen sensors is ~ C [6]. Sensors MAP IAT TPS Coolant temperature Adjustment ECU Fuel flow Basic injection time Injection pulses Oxygen sensor Injector Fig. 3. EFI with MAP and IAT sensor Fig. 4. Emissions concentrations for spark ignition engines exhaust [7] ISS: 1792-5924 / ISS: 1792-594 232 ISB: 978-96-474-237-
Figure 5 shows an example of a control strategy for an engine used in automotive applications, where load and engine speed vary across a wide range. The main parameters evaluated by the ECU are engine speed (), throttle plate angle (TPS), air flow (measured by the MAF sensor or calculated using, manifold pressure MAP and air temperature IAT), coolant temperature (T c ) and oxygen sensor voltage (U O2 ). A basic setting for the injection time (t inj ) is calculated based on the air flow value. Therefore, for this basic setting,, MAP and IAT need to be monitored. If the throttle is fully open, the engine is operated at full load and closed loop control is aborted. During this so called open loop control mode, t inj is calculated based on air flow only. At part load, if the engine is warm (T c is within its nominal range), closed loop control is possible by evaluating the signal given by the oxygen sensor (U O2 ). arrow band oxygen sensors give a voltage of ~ 45 mv when the concentration of O 2 in the exhaust gas stream is at its value for stoichiometric combustion. If the reading is higher than this value, U O2 > 45 mv, the engine is running rich and the injection time needs to be reduced. On the contrary, if the voltage is low, U O2 < 45 mv, more fuel is needed to achieve a stoichiometric mixture and the injection time is increased. In this way, the injection time is constantly adjusted so that stoichiometric air-fuel ratio is achieved. The actual electric pulses controlling the solenoids that open the injectors are generated based on the injection time (t inj ), as the pulse length, and the reading from the crank shaft positioning sensor (CPS) for pulse frequency. The ECU determines top dead centre (TDC) and generates one injection pulse for every crank shaft rotation or once every two rotations of the engine. 3 Simplified Architecture and Proposed Control Strategy Stationary SI engines usually feature a carburetor and a simple electronic ignition system. The starting point of developing a simplified EFI architecture is that these engines operate at constant speed (15 or 3 rev/min for 5 Hz and 18 or 36 rev/min for 6 Hz). Therefore, ensuring optimum ignition is easier and existing ignition systems can be considered as more than appropriate. In the following case study, an engine speed of 3 rev/min will be considered, specific for small size SI engines used with electrical generators supplying power at 5 Hz. Sensors MAP ECU Fuel flow, TPS, MAP, IAT, T c, U O2 Oxygen sensor Adjustment Basic injection time Injection pulses Ignition signal t inj f (, MAP, IAT) Injector TPS Full load T c nominal U O2 > 45 mv 45 mv Open loop Open loop inj t inj t inj t inj Fig. 5. Closed loop control for an automotive EFI Fig. 6. Simplified EFI architecture The proposed simplified architecture relies on the signal generated by the ignition system for evaluating engine speed and injection pulses generation. In this way, there is no need for a dedicated engine speed sensor. Because pressure sensors are less expensive than MAF sensors, a MAP signal is considered for evaluating engine load. Compared to engines for automotive applications, stationary aggregates are run up to ~ 8 % of maximum power so that air-fuel mixture enrichment is not required. For this reason, the TPS signal is not necessary for a simplified EFI architecture. Using the MAP signal for evaluating engine load ensures high enough precision when load changes rapidly, as there is practically no delay between the TPS and MAP signals (figure 7). ISS: 1792-5924 / ISS: 1792-594 233 ISB: 978-96-474-237-
The control of air flow to the engine can be achieved by employing a simple speed governor that keeps engine speed at 3 rev/min and controls throttle opening as electrical loads varies. An IAT sensor would be required for calculating air density. However, as stoichiometric operation is required at all times, a basic injection time calculated based only on the MAP signal should be enough, and this basic injection time can be continuously adjusted based on the signal provided by the oxygen sensor. TPS [deg] MAP [mbar] tinj [ms] UO2 [mv] 9 7 5 3 1 12 1 8 6 2 15 1 5 1 8 6 2.2.4.6.8 1 1.2 1.4 1.6 1.8 2 Time [s] Fig. 7. Measured TPS, MAP, injection time and oxygen sensor voltage for an automotive SI engine during sudden acceleration As stationary engines are operated at nominal temperature for extended periods, precise correlation of the injection pulses with crank shaft rotation is not necessary, as is the case for engines using a cam shaft position sensor. Therefore, the only sensors strictly necessary for EFI systems used on stationary SI engines are the MAP and oxygen sensor, resulting in a much simplified architecture (figure 6). Cold start presents a specific challenge for fuel systems. Carburetors feature an additional air control valve (so called choke valve) that is closed when the engine needs to be started in low ambient temperature conditions. In this way, the air-fuel mixture is enriched so that enough fuel vapor enters the combustion chamber. EFI systems increase the injection time during cranking and also during engine warm-up (figure 8). When cranking the engine at ~ 12 rev/min, the injection time (t inj ) is 3 4 times higher compared to normal operation. After the engine is started, the air-fuel mixture is briefly enriched so that stable operation is ensured. When the coolant temperature reaches a certain minimum threshold, the ECU adjusts the injection time so that the engine runs on a stoichiometric air-fuel mixture. [rev/min] MAP [mbar] tinj [ms] UO2 [mv] 16 12 8 12 1 8 6 2 2 15 1 5 1 8 6 2 Ignition Rich Idle operation Stoichiometric 5 1 15 2 25 3 35 4 Time [s] Fig. 8. Cold start and warm-up parameters for an automotive EFI system The signal provided by the oxygen sensor shows an interesting behavior with the engine stopped. Figure 8 shows engine speed, MAP, injection time and U O2 for an automotive EFI system during cold start and warm-up. Before the engine is started, the oxygen sensor provides a signal specific for lean air-fuel mixtures. This is possible because the sensor is warmed by an electric ISS: 1792-5924 / ISS: 1792-594 234 ISB: 978-96-474-237-
heater before cranking the engine. As the exhaust contains only air, U O2 shows low values. After the engine is started, the oxygen sensor gives a much higher reading, specific for rich mixtures. Therefore, relative air-fuel ratio can be evaluated by analyzing U O2 voltage just seconds after the engine is started, if an electrically heated oxygen sensor is employed. Such analysis of the U O2 signal are used in some EFI systems to determine the content of alcohol in ethanol-blends used to fuel SI engines [8]. In a simplified EFI system for stationary SI engines, analyzing U O2 can provide the basis for evaluating engine working temperature. An oxygen sensor that is not electrically heated requires the engine to be operated for a few minutes before reaching minimum working temperature. A cold sensor gives a constant voltage of ~ 45 mv, no matter what the oxygen concentration in the exhaust. Therefore, a software containing an analysis as the one shown in figure 9 could by used to evaluate engine temperature. thus increasing air flow to the engine. As a consequence, the air-fuel mixture is leaned out and the engine speed tends to drop. In response, the ECU increases injection time until the condition > 3 rev/min is reached. If load decreases, the engine will briefly run rich and increases. The ECU responds by decreasing the injected fuel quantity. Of course, such a control strategy would probably be too slow to ensure good engine response when load changes rapidly, but would nonetheless allow the engine to be operated until the fault is remedied. Experimental investigations on this matter will provide an answer to the effectiveness of such a control strategy. Engine cranking < 2 rev/min U O2 ormal operation max(u O2 ) > 6 mv or min(u O2 ) < 3 mv Engine warm closed loop Fig. 1. Control strategy during engine cranking Engine cold open loop Fig. 9. Algorithm for evaluating engine temperature based on the U O2 signal Cold starts are a major challenge for EFI systems, as the engine needs to be easily started even in harsh ambient conditions [9]. Using an analysis of the signal given by the ignition system, engine speed can be evaluated. SI engines need to be cranked at ~ 12 2 rev/min to be started. Therefore, if the frequency of ignition pulses are in the range of 1 2 Hz for a single cylinder four-stroke engine, the ECU would assume that the engine is cranked to be started, and the air-fuel mixture can be enriched accordingly (figure 1). Analyzing this signal could also be used as a mean of setting the basic injection time in extraordinary situations such as the case of MAP sensor failure (figure 11). If the engine slow down < 3 rev/min, the ECU assumes that load is increased and the throttle is opened, 3 rev/min Fig. 11. Control strategy in the case of MAP sensor failure Readings from more than one sensors, even if they are mostly redundant, are used the ECU for selfdiagnosis and to evaluate sensors working parameters. Therefore, using only one sensor to control injection timing is not a practical solution, even if theoretically possible. Also, an important aspect of ECU software is ISS: 1792-5924 / ISS: 1792-594 235 ISB: 978-96-474-237-
that injection time is adjusted within certain limits. This is used by the ECU for self-diagnosis. When correction limits are exceeded, a fault code is generated and intervention by qualified personnel is required. As the nature of the simplified EFI system requires a much wider range for injection time adjustment, another procedure for self-diagnosis must be developed. This matter will be undertaken by the author in future work. An advantage of much wider limits for injection time adjustment is that multi-fuel operation is possible. As a result of using special control strategies, a simplified EFI system architecture (figure 6) can be used on stationary SI engines, employing an algorithm such as the one shown in figure 12., MAP, U O2 t inj f (, MAP, IAT) < 2 rev/min Evaluate T c T c nominal MAP failure U O2 > 45 mv 45 mv Open loop based control inj t inj t inj t inj Fig. 12. Control strategy for a simplified EFI architecture 4 Conclusion A control strategy for simplified EFI systems was developed to be used with stationary SI engines. This will ensure high efficiency with reduced emissions by using a three way catalytic converter that requires sotichiometric operation of the engine at all times. Initial theoretical evaluations of the algorithms used to determine injection timing, show that good control characteristics are possible with a small number of sensors. This ensures minimum costs, a very important aspect especially with small size SI engines used in micro-cogeneration installations. 5 Acknowledgment This work was partially supported by the strategic grant POSDRU/89/1.5/S/57649, Project ID 57649 (PERFORM-ERA), co-financed by the European Social Fund Investing in People, within the Sectoral Operational Programme Human Resources Development 27-213. References: [1] S. R. Schill, EU adopts 2-2-2 plan, includes emissions trading, Biomass Magazine Online, January 29. [2] A. Irimescu, L. Călin, A. JădăneanŃ, Aspects Concerning the Efficient Use of Biogas Obtained from Sewage Water Discarded by the Beer Industry, Journal of Environmental Protection and Ecology, Vol. 1, o. 4, 29, pp. 1137-1145. [3] M. Kanoglu, I. Dincer, Performance assessment of cogeneration plants, Energy Conversion and Management, Vol. 5, o. 1, 29, pp. 76-81. [4] A. Irimescu, D. Lelea, Thermodynamic analysis of gas turbine powered cogeneration systems, Journal of Scientific & Industrial Research, Vol. 69, o. 7, 21, pp. 548-553. [5] R. Bosch, Gasoline Engine Management, 2 nd Edition, Bentley Publishers, 26. [6] J. B. Heywood, Internal Combustion Engines Fundamentals, McGraw Hill, 1989. [7] R. Bosch, Gasoline Fuel-Injection System K-Jetronic Technical Instructions, Bentley Publishers, 21. [8] C. M. Engler-Pinto, L. de adai, Volumetric Efficiency and Air-Fuel Ratio Analysis For Flex Fuel Engines, Tecnologia da mobilidade 17 th SAE Conference Brasil, September, Sao Paolo. Brasil, 28. [9] C. Hu, X. Song,. Liu, W. Li, Investigation on Cold Starting and Warming up of Gasoline Engines with EFI, Small Engine Technology Conference, 3 October 1 ovember, iigata, Japan, 27. ISS: 1792-5924 / ISS: 1792-594 236 ISB: 978-96-474-237-