Technical field

The invention is related to a dual combustion fuel engine with feedback control of combustion according to the first part of claim 1 . Furthermore, the invention relates to a method operating a dual fuel combustion engine with feedback control of combustion and a feedback control of combustion device or arrangement for a dual fuel combustion engine.

Background of the invention

Road transportation is responsible for 17% of worldwide emissions of carbon dioxide, according to the report of the International Energy Agency, 201 1 a. Passenger light- duty vehicles account for 60% of that amount. The number of these vehicles is assumed to double between 2010 and 2035 due to their fast growing number in emerging countries, according to the report of the International Energy Agency, 201 1 b. Reducing the CO ₂ emissions of passenger light-duty vehicles is thus a major task toward achieving the long-term goal of reducing total carbon dioxide emissions. Using compressed natural gas (CNG) instead of gasoline or diesel is one attractive option to reduce the CO2 emissions of passenger and light-duty vehicles. CNG mainly consists of methane which has a higher hydrogen to carbon (H to C) ratio compared to gasoline or diesel. This higher H to C ratio of methane leads to a reduction of CO2 emissions in the order of 20 - 25%, for the same engine-efficiency, according to Semin, R. (2008), "A technical review of compressed natural gas as an alternative fuel for internal combustion engines" Am. J. Engg. & Applied Sci, 1 (4), 302-31 1 . Among all possible gas engines, the dual fuel engine is one of the most promising. In the dual fuel engine, the gas is injected into the intake manifold. The premixed air-gas mixture is then ignited with a small amount of directly injected diesel. This engine has the potential of achieving diesel-like efficiencies without the need for a lean de-NOX system, according to Serrano and Bertrand, 2012,

"Exploring the potential of dual fuel diesel-CNG combustion for passenger car engine", in IFAC 2012. Combustion in the gas/diesel dual fuel engine has been extensively investigated. It has been found, that even if the ratio of diesel fuel to total fuel mass is very small, the influence of the diesel injection on the whole combustion is large. More

specifically, the sensitivity of combustion phasing and combustion noise on start of injection is high. (Serrano and Bertrand). The start of combustion depends on the start of injection of the diesel and on the ignition delay of the diesel. The latter is mainly dependant on the chemical reaction kinetics of the diesel fuel. This process is very sensitive to small changes in pressure, temperature or cylinder charge composition.

Combustion processes which are prone to external disturbances can best be controlled by using feedback control based on the measured cylinder pressure.

Feedback control based on the measured cylinder pressure has been applied successfully to conventional diesel engines, as disclosed in WO 2009/143 858 A1 . It has also been applied to the control of HCCI engines, as by Olsson, J., Tunestal, P., Johansson, B., et al. (2001 ) in "Closed-loop control of an hcci engine" SAE

TRANSACTIONS, 1 10(3), 1076-1085, as an example. And it has also been applied to many other combustion types.

Description of the invention

According to a first aspect the invention has the object to provide a gas diesel dual fuel engine having a feedback control of combustion based on the measured cylinder pressure to control the combustion phasing and combustion noise. Furthermore, according to a second aspect, the invention has the object to provide a method of feedback control of combustion based on the measured cylinder pressure to control the combustion phasing and combustion noise in a gas diesel dual fuel engine.

According to a third aspect, the invention has the object to provide respective feedback control device or arrangement.

The objects of the invention according to the aforementioned aspects are solved by the subject-matters of claims 1 , 6 and 1 1 respectively. The main fuel is any fuel having good mixture characteristic with air and a low auto-ignition characteristic, preferably compressed natural gas, which is injected into the intake manifold. It should be noted that other low ignition gases like gasified gasoline, can be used instead. The second fuel is any fuel having a high auto-ignition characteristic, preferably diesel which is directly injected into the combustion chamber.

The goal of the control design is to minimize the diesel quantity using the center of combustion (COC) and, according to a preferred embodiment of the invention, the maximum pressure rise rate (dp _ma x) as parameters. The center of combustion is the point, where 50% of the fuel-energy is released. The maximum pressure rise rate is the maximum rise rate of the cylinder pressure evaluated over one engine cycle.

The controlled variables z are therefore

COC

dp (1 )

In this preferred embodiment of the invention, these two variables ensuring an efficient combustion with acceptable noise level. The center of combustion and maximum pressure rise rate are controlled by manipulating the start of diesel injection SOI and the duration of diesel injection DOI. The manipulated values are therefore

SOI DOL (2)

The center of combustion (COC) and maximum pressure rise rate (dp _max ) are obtained from the measured cylinder pressure (p _cy i). Preferably, each cylinder is controlled individually and independently from the other cylinders.

It should be noted that the start of injection value is considered as an angle of the combustion cycle of the respective cylinder and is considered to have any phase value, especially when the start of injection value is used to calculate an one to one and onto (bijective) function at least for a duration of injection value above a predetermined duration minimum. According to the easiest version of this function is a constant quotient (DOI = b ^* SOI+c). In a more advanced version it can be a bijective polynomial function comprising the terms DOI = a ^* SOI ² +b ^* SOI+c with the parameters a, b, and c selected that the function is bijective in the range of interest of SOI.

Furthermore, a function DOI=d1 ^* e ^d2*sol+d3 is a candidate for a bijective function, when the parameters d1 and d2 are selected appropriately.

The aforementioned elements as well as those claimed and described in the following exemplary embodiments, to be used according to the invention, are not subject to any particular conditions by way of exclusion in terms of their size, shape, use of material and technical design, with the result that the selection criteria known in the respective field of application can be used without restrictions. It should be noted that all means used to implement the invention are not limited to a particular design but can be provided as digital, analog or hybrid means having A/D and D/A conversion means at several points. Brief description of the drawings

Examples of the engine will hence forth be described in more detail by reference to the drawings, wherein are shown: is a diagram of the Center of Combustion (COC) as the axis of ordinate over the Start of Injection (After Top dead Center/ADTC) as the axis of abscissae, both in phase angles

Figure 2 shows the line (linearized) of the first Start of Injection in a diagram of the Duration of injection as the axis of ordinate in microseconds over the Start of Injection as the axis of abscissae, latter in phase angles

Figure 3 shows the controller structure of a preferred embodiment of the

invention including a limitation of the maximum pressure rise rate ure 4a to 4h show diagrams of the measurement results with the time in seconds as the axis of abscissae, wherein

is a diagram of the break mean effective pressure in bar 4b is a diagram of the intake manifold pressure in bar

4c is a diagram of the air/fuel ratio

4d is a diagram of the exhaust gas recirculation rate in percent

4e is a diagram of the measured center of combustion

4f is a diagram of the maximum pressure rise rate in bar/degree

4g is a diagram of the start of injection in degree

4h is a diagram of the duration of injection in microseconds

Figure 5a to 5c show diagrams of the controller design wherein

Figure 5a is a diagram of the frequency response of the controller with the gain as the axis of ordinate in decibel over the frequency as the axis of abscissae

Figure 5b is a diagram with the phase angle as the axis of ordinate over the

sampling frequency as the axis of abscissae

Figure 5c is a diagram of the disturbance rejection y(k) over the number of engine cycle

Figure 6a to 6e show diagrams of steady state measurement results with brake mean effective pressure values in bar as the axis of abscissae, wherein 6a is a diagram of the air/fuel ratio

6b is a diagram of the center of combustion

6c is a diagram of the maximum pressure rise rate in bar/degree

6d is a diagram of the diesel ratio in percent

6e is a diagram of the engine efficiency.

Embodiments for carrying out the invention

All embodiments described hereby are related to a 4-cylinder diesel engine 21 with a displacement volume of 2.0 litres. The engine 21 according to the described embodiment is equipped with a common rail diesel injection system a VTG- turbocharger and a high pressure EGR system. Table 1 gives additional details on the engine. Manufacturer Volkswagen

Type TDI 2.0 - 475 NE (CJDA)

Industrial Engine

Cylinders 4

Displ. Volume 1 .968 I

Bore 81 .0 mm

Stroke 95.5 mm

Compression Ratio 16.5

Injection System Bosch Common Rail

Diesel Injectors Piezo

max pressure 1800 bar

Table 1 : Data of the engine used for measurements

In comparison with the series production engine, the following modifications have been made:

• port-fuel injection system for natural gas

• installation of low-pressure EGR system, the available

high-pressure EGR system is not used for the

presented measurements

• cylinder-pressure sensors 22 in all four cylinders

Modelling

Combustion in an internal combustion engine is an inherently discrete system. It is assumed, that the combustion at time k + 1 only depends on the input at time k. The combustion properties _Z k+i , can thus be expressed as a function of the previous input v _k .

¾ ₊ Γ^) (3)

where: COC SOI

z ^~ \ dp and v=

DOL (4)

The influence of the combustion of the last cycle (z _k ) is neglected. All other influences such as intake pressure, air/fuel ratio, exhaust gas recirculation rate and others are neglected too.

Identification

The function f in equation (3) is identified using measurements. Figure 1 shows one part of the result of the measurements. The figure shows the dependency of the center of combustion on the duration of injection and start of injection. The center of combustion is always advanced, if the injection duration is increased. The

dependency of the center of combustion on the start of injection is more complicated. For late injection, the center of combustion is advanced, if the start of injection is advanced. But for early injection, the center of combustion is retarded, if the start of injection is advanced. For any injection duration, there exists a start of injection, which leads to the earliest possible center of combustion for that injection duration. A physical explanation for such a phenomena can be found in Eichmeier, J.,

Wagner, U., and Spicher, U. (201 1 ). ..Controlling gasoline low temperature

combustion by diesel micro pilot injection", ASME 201 1 .

Linearization

The system is linearized around an operating point z ₀ with corresponding input v ₀ . The function f(v) can be approximated by:

/(ν)-β(ν _η )·(ν-ν _η )+/(ν ) (5)

of_

B(v _Q )

dv

(6)

The deviation from the nominal operating point is described as: x = z - z ₀ . The deviation of the control from the nominal value is described by: u = v-v ₀ . The resulting system equation in state space form is given by: y _k = C ^■ x _k

(7)

where C = I. The discrete time state space model can be transformed into a discrete time transfer function:

Y Pll Pl2 1

= P(z ) =

P21 P22

(8)

where

The derivatives

■are obtained from the measured function f by finite differences. The dependency of the center of combustion on the start of injection which has been shown in figure 1 leads to a sign change in p depending on the linearization point. This sign change renders control a difficult task. The sign change, however can be eliminated by an appropriate input transformation.

Input Transformation

Figure 1 showed that for any duration of injection, there exists a start of injection (SOI*) which leads to the earliest possible center of combustion.

SOf=f _SOI (DOI) (10)

This is the combination of SOI and DOI which leads to the lowest possible diesel quantity for a desired center of combustion. A new input a is defined which is related to the input u by:

The function f _SO i can well be approximated by an affine function. Figure 2 shows the approximation. The minimal DOI is limited to 150 s. This minimal DOI value is introduced in order to avoid operation of the injectors at very short injection durations, where the cycle to cycle variation of injected diesel quantity is large.

The input transformation leads to a transformed system P (z):

Plz) SOI

U(z) dSOI

(12)

The transformed system P has no sign change (p1 > 0, p2 < 0), for all possible linearization points. Further on, the proposed input transformation leads to control of combustion with minimal diesel quantity. Controller Structure

Figure 3 shows the structure of the control system. Both errors ecoc,e _dP , derived by the comparison of the measured values COC and dp with the parameters r _C oc and r _dp , respectively, are filtered using a low-pass filter, 14 and 15 respectively. The parameters r _C oc and r _dp are the reference values of the center of combustion and the maximum pressure rise rate. The signal e _dP is then multiplied by p1 /p2 according to the transfer input function harmonization means 16 ensuring that the transfer function (a→ 8i) is equal to the transfer function (a→ 82). This enables the use of one single controller.

The input to the controller 12 is the maximum of 81 and 8 _2, according to the limiting means 1 1 . This leads to the following behaviour:

COC = rcoc ^an( i #ma < U _v

or

COO > rcoc and dp _max = r _dp The controller also includes the input transformation as described above.

Controller Design

The controller is a Pl-Controller. The discrete time transfer function of the controller is given by:

The controller parameters (k _p , T| ) as well as the time constant of the low-pass filters are chosen based on the following specifications:

• Stepwise disturbances are reduced to 10% within 10 engine cycles. · Phase margin φ > 60°.

• The phase margin is also maintained for large uncertainties in the gain of the plant (pi, p ₂ )

The frequency response and the disturbance rejection of the design can be seen in figure 4.

Experimental results

Transient

Figures 4a to 4h show results of transient engine operation at a constant engine speed of 2000 rpm. Load-steps are applied to validate the combustion controller. The load is increased after 20 seconds and again after 40 seconds and is reduced to its initial value after 80 seconds. The first four diagrams 4a to 4d show the brake mean effective pressure (load), the intake manifold pressure, the air-fuel ratio and the exhaust gas recirculation rate. All values change as a consequence of the loadsteps. These changes are disturbances on the center of combustion and maximum pressure rise rate. The resulting center of combustion, the maximum pressure rise rate, the start of injection and the duration of injection are shown by diagrams 4e to 4h. The results are given individually for all four cylinders. Figure 4e shows the center of combustion. At the beginning, the load is low and the center of combustion remains at its nominal value of 8°. The small load step after 20 seconds is hardly visible. After 40 seconds, the load is increased to 15 bar break-mean effective pressure. As a consequence, the center of combustion is retarded because combustion is limited by the maximum pressure rise rate of 4 barf. After 80 seconds, the load is reduced and the center of combustion is brought back to its nominal value. Figure 4f shows the maximum pressure rise rate over time. At the beginning, the maximum pressure rise rate is below its nominal value of 4 barf. After 20 seconds, the load is increased which also increases the maximum pressure rise rate, but it still remains below its nominal value. After 40 seconds, the increasing loads lead to an increase of the maximum pressure rise rate far above its nominal value. But the value is reduced by the controller to its nominal value within short time.

Between 40 and 80 seconds, the maximum pressure rise rate is controlled to its nominal value, as a consequence, the center of combustion is retarded.

Figure 4g shows the start of injection of the diesel injection. The small load step after 20 seconds requires only little change in the start of injection to maintain the center of combustion at its nominal value. The second step after 40 seconds requires a change of the center of combustion of more than 10°, to keep the maximum pressure rise rate at its nominal value. After 80 seconds, the engine load is set to

the same value as at the beginning. However, the start of injection is retarded compared to the value at the beginning. The reason is the increased engine temperature due to the previous operation at high load. This higher temperature leads to a faster auto-ignition of the diesel. Figure 4h shows the duration of the diesel injection. The duration of injection is directly coupled to the start of injection through the function f _D oi shown in figure 2.

Steady State

Figures 6a to 6e show measurements taken at constant operating points. The shown measurements are averaged over 60 seconds. Engine speed is 2000 rpm for all measurements, while the break mean effective pressure (load) is varied. Figure 5a shows the air-fuel ratio. The engine is operated lean at low loads and is operated stoichiometric above 9 bar brake mean effective pressure. This enables exhaust gas after treatment with a three way catalytic converter at medium to high loads.

Figure 6b shows the center of combustion for various loads. At low loads, the center of combustion is at its nominal value of 8°. At higher loads combustion is retarded, to keep the maximum pressure rise rate at its nominal value. The center of combustion is latest at 9 bar brake mean effective pressure. The reason for that is, that the engine is operated stoichiometrically with high rates of recirculated exhaust gases at this point, but the intake manifold pressure is still moderate ( ^« 1 bar). These conditions increase the ignition delay of the diesel and, accordingly, a larger diesel quantity is required. The larger diesel quantity leads to a faster combustion. This requires a later center of combustion to keep the maximum pressure gradient within the limit. At higher loads, the intake pressure is higher and the egr rates are smaller which promotes the diesel autoignition. Figure 6c shows the maximum pressures rise rate for various loads. At low loads, the maximum pressure rise rate is below its nominal value. At higher load, the

combustion is limited by the maximum pressure rise rate.

Figure 6d shows the ratio of the diesel fuel to the total fuel energy (r _D ). The value r _D is defined as: , ,

(15)

wherein m _D is the mass of diesel fuel, H _uD is the lower heating value of diesel, ITICNG is the mass of natural gas and H _UC NG is the lower heating value of natural gas.

The energetic ratio of diesel is around 15% for a brake mean effective pressure of 3 bars. With increasing loads, the energetic diesel ratio r _D is constantly decreasing to values below 2% for high loads.

Figure 6e shows the measured brake efficiency. The measured efficiency reaches maximum values around 40%.

List of reference signs

10 Controller

1 1 Limiting means

12 Device

14 low pass filter

15 second low pass filter

16 Transfer input function harmonization means 21 Engine

22 Center of combustion determination means

30 comparisation means

31 additional comparisation means