Two Wheeler Engine Control Unit Development, Challenges and Solutions

M20100155

Two Wheeler Engine Control Unit – Development, Challenges and Solutions

Aravind Aithal, Ramakrishna Donakonda

Robert Bosch Engineering and business Solutions Limited, Bangalore, India

ABSTRACT

In motorcycles, Engine Management systems (EMS) are used mostly in high end models. However, increasing demands for improved fuel efficiency and tighter legislations for reduced emissions have made it inevitable to adopt EMS, instead of the carburetors even in small gasoline engines motorcycles which are already quite fuel-efficient.

Today’s system cost is significantly high and comparable with passenger car applications and success in emerging markets due to fun-to-drive aspects is likely to be limited to premium segment. So there is an obvious requirement for minimized system cost along with smart system solutions. For reducing the number of sensors and cost, highly accurate methods have been developed to eliminate sensors like barometric sensor and cam timing sensor. An engine management system has been developed for the unique problems of one cylinder, high performance engines.

INTRODUCTION

In automobiles, hydrocarbons, carbon monoxide and oxides of nitrogen are created during the combustion process and are emitted into the atmosphere from the tail pipe. Over the last 30 years, reductions in tailpipe exhaust emissions of more than 90% have been demanded of and have been achieved by the automobile industry. The need to control these emissions gave rise to the computerization of the automobile.

To achieve improved combustion efficiency and performance, as well as the reduction of exhaust emissions, an EMS is used is to control air fuel ratio (AFR) and spark timing in response to various operating conditions like engine speed, temperature, load, altitude, and battery voltages. The EMS comprises of an Electronic control unit (ECU), related sensors and actuators.

In countries like India a large number of two and three wheelers are used and the numbers continue to increase, leading to high levels of pollution in urban areas. Figure 1 and figure 2 explain the legislation, emission limits and test cycles in different countries for two wheelers.

Figure 1 – Two wheeler emission legislations

Figure 2 – Emission limits and driving cycles

Future legislations for motorcycles with small engines will make it necessary for manufacturers to follow the same path in engine development as car manufacturers have already taken.

The figure 3 shows the evolution of motorcycle control technologies -present and future.

Figure 3 - Different stages of motor cycle control technologies

Adopting EMS for 2 wheeler application is quite challenging. There are many technical challenges and challenges related to cost. Approach of directly taking over a passenger car EMS to a small engine application will result in severe disadvantages due to the large differences in physical behavior between the engine types, different technologies, market requirements, functional differences and system complexities. Hence EMS of passenger cars cannot be directly applied to small engines.

When considering the small engine motorcycle market relative to the automobile market, one key difference immediately becomes apparent, namely; cost. The need for cost-effective emission solutions in the small vehicle industry is therefore widely recognized and as outlined in the following section, a number of alternative low-cost strategies are currently being pursued.

One of the main problems to solve has been accurate load detection at different throttle openings. Depending on the inlet pressure for load determination is difficult due to its very non-linear behavior. On the other hand, using only throttle potentiometer information will result in poor load signal calculation at part loads which often makes it necessary to run rich at low loads.

This paper describes an investigation into an EMS applied to small gasoline-fuelled, spark-ignited internal-combustion engines used in two and three wheelers, to meet the challenges of fuel economy, emissions and also capable of some diagnostics to assist in troubleshooting faults as the system becomes more complex. The principles of the advanced algorithms and their effects are discussed in the paper. The paper will also show emissions, and engine operating data that verify system performance. This paper also discusses the system and the software developed which is able to combine good response, improved cold start, good idle stability, improved transient behavior and better drivability for motorcycle applications and is also cost-efficient, compact, and state of art for the defined segment.

SMALL ENGINE EMS

The engine management system developed for small engine applications as shown in figure 4

Figure 4 - System schematic - 4-stroke electronic injection system

Test Engine - The engine used for this work was a 1-cylinder 125 cc. Specifications are given in figure 5

Figure 5 - Test engine and motorcycle specification

A one-cylinder engine has high pressure oscillations in the intake manifold compared to a car engine. This makes it difficult to accurately estimate the air mass in the combustion chamber. The oscillations also result in sensitivity to wall wetting effects, which in turn affects the drivability when the mixture is set toward stoichiometric ratio. Fuel film compensation algorithms are used in the EMS to handle these phenomena.

For measuring the engine control data, the following methods are investigated from a small engine point of view

· Intake air mass calculation

· Atmospheric pressure estimation

· Phase detection

· Dynamic correction

INTAKE AIR MASS CALCULATION

Study was made as to the applicability of the conventional intake air mass measurement methods - speed density, throttle speed and the method combining both of these - with 4-stroke single-cylinder engines generally used in small engine applications.

Speed density method

In the 4-cylinder engine, there is a relatively large volume called surge volume connected to the manifold for each cylinder in the throttle valve downstream. This gives averaged intake manifold pressure automatically and in case of steady operation, an intake air mass can be calculated from the pressure value independent of sampling timing as well as from engine speed (speed density method). On the other hand, the volume of the throttle valve downstream in the single-cylinder engine is far smaller than that of the above-mentioned multi-cylinder engine and thus will be subjected to great fluctuations in one cycle of engine operation as figure 6 shows. If these pressure fluctuations are to be smoothened, signal may be delayed, which would be inappropriate for motorcycles requiring high response. Furthermore even in a steady state, not much measurement accuracy can be achieved because of small change in intake manifold pressure between medium and heavy load, thereby making it difficult to adopt this method.

Compared to multi-cylinder passenger engines, small motorcycle engines have a relative small intake manifold volume between throttle body and intake valve strongly pulsating air mass flow, highly dynamic behavior as shown in the figure 6

Figure 6 - Difference of intake manifold behavior between 4 cylinder engine and single cylinder engine [1]

The screenshot of the measurement of the above behavior on actual engine is shown in figure 7

Figure 7 - Intake manifold pressure of single cylinder engine

The relationship between intake air mass and manifold pressure in multi cylinder engines at different engine speeds is shown in figure 8

Figure 8 - Averaged intake manifold pressure with respect to intake air mass

In single cylinder engines, the relationship between intake air mass and manifold pressure is shown in figure 9. A non linear behavior is observed at high engine load due to pressure pulsations in the intake manifold. The difficulty to correlate this pressure pulsation with the actual airflow to the engine is the main reason that a manifold air pressure (MAP) sensor is often not used as primary sensor for air charge determination in single cylinder applications.

Figure 9 - Air charge calculation based on speed density method

Throttle speed method - This method, which is employed in racing cars etc., produces a high response. However, because of a very sharp initial rise in intake air mass against the change at narrow opening and non-linear character (figure 10), it would be difficult for this method to provide the intake air mass measuring accuracy that could satisfy today’s rigorous emission regulations.

Figure 10 - Air charge calculation based on throttle speed method

Combined speed density and throttle speed method - To compensate for drawbacks of these two methods, a combined method could be considered – the speed density method at low load range and the throttle speed method at the remaining operation range. This method has potential for sufficient accuracy; however complications in switching over between the two methods exist. For a fixed engine speed, the figure 11 shows the relation between the throttle opening and estimated air charge in the combustion chamber for different load ranges. Using this approach, the deviation between the estimated and actual air quantity is within a tolerance band of AFR. This variation is shown in upper part of figure 11.

Figure 11 - Variation of estimated air charge w.r.t. throttle angle using combined method. The error in the calculation is measured by AFR variation.

ATMOSPHERIC PRESSURE ESTIMATION

During engine stop state, the atmospheric pressure can easily be measured because it is equal to the intake manifold pressure. While the engine is running, during the suction phase there is a large drop in intake manifold pressure, however since the intake manifold volume is very small for small engines, it will result in quick recovery towards the ambient pressure once the suction phase is over even when the throttle is not wide open. As shown in figure 7, a point free from increase in pressure (saturation point) is detected and is set for the atmospheric pressure and during full load; the filtered intake manifold pressure can be used as the primary source of ambient pressure.

Figure 12 shows the test results for the atmospheric pressure estimation during positive gradients. Measurements on a separately installed barometer are given for comparison plotted as error level. Results prove that this method gives a fairly accurate estimation of the atmospheric pressure.

Figure 12 - Tolerances of atmospheric pressure estimation based on an up hill test trip

PHASE DETECTION

Early engine control used to provide ignition once in each engine revolution without distinction of compression and exhaust stroke. In modern engines, the fuel injection and ignition occur at the correct phase of the engine cycle. For distinction of the stroke in 4-stroke multi cylinder engines, a cam position sensor is in common use. In single cylinder engines however it is not often used for cost reasons. To detect phase, the existing methods (without a cam position sensor) have disadvantages like start emissions (depending on initial piston position at start). To tackle this, the following method of phase detection was implemented which uses variations in the intake manifold pressure. It was also found to be a very reliable method of phase detection.

As described so far, the intake manifold pressure in single-cylinder engines are deeply affected by the intake stroke taking place during the intake valve opening. From this characteristic, it may be considered that observing the intake manifold pressure affords an understanding of the intake stroke. P0 and P1 in figure 13 indicate the crank timing pressures at the BDC. At the BDC in one cycle, the intake valve is open at one end and closed at the other, obviously making a great change in the intake manifold pressure. Thus, comparing the intake manifold pressures sampled in this timing should make the distinction of engine stroke possible.

Figure 13 - Wave form for engine phase distinction

The BDC is detected by crank pulse and the intake manifold pressure is sampled and stored in memory every 360°CA. The value thus memorized is retained until the next sampling. The difference between the new and the previous pressure data is checked as to whether it is above the set value. When a difference with a higher value than set is repeated more than several times, the program confirms the engine stroke and switches over from one-revolution to one-cycle control.

DYNAMIC CORRECTION

Today's fuel injection systems usually inject the fuel late at closed inlet valves. This is a compromise which may result in either that too much fuel already has been injected at decelerations due to early injection timing or that too little fuel injected at fast accelerations, which will result in engine misfire. This may be acceptable on a multi cylinder engine, but absolutely not on a one-cylinder engine that will lose its power during a whole cycle. This makes the engine very sensitive to engine stall at fast throttle opening from idling (throttle blipping). Immediate response control is required in order to ensure that no combustion misfire may occur even during the fastest throttle opening. Engine performance is strongly dependent on gas dynamic phenomena in intake and exhaust systems.

Another major problem in driving an engine with extreme transients, such as in motorcycle driving, is that the effect from changes in the wall wetting makes it necessary to enrich the overall air fuel mixture. This can result in drivability problems due to over fueling and fouled sparkplugs. This is not a problem with the wall wetting compensation algorithm used. Figure 14 shows a typical behavior in a fuel injection system which has a target steady state lambda at 1.0. Without the wall wetting function, a lambda variation of more than ± 20% is possible and hence results in drivability problems and the need of more overall enrichment or enleanment, which will produce more exhaust emissions.

Figure 14 Transients without wall wetting compensation

Figure 15 shows the results when the wall wetting control is activated and calibrated with a steady state lambda target of 1.0. It can be seen that big improvements in lambda behavior under very fast load and speed changes are realized. This functionality makes it possible to achieve a vast reduction in engine out exhaust emissions in real driving conditions and should therefore make it possible to reduce costs on the catalyst. Figure 15 Transients with wall wetting compensation