Sunteți pe pagina 1din 67

Computational Modeling

of

The Performance of a Variable Geometry Toroidal Engine

The V.G.T. Engine

by G . A . Karim and O. M. Bade Shrestha Department of Mechanical and Manufacturing Engineering University of Calgary

Confidential Report Submitted to VGT Technologies Inc. -Calgary- Canada

July 2000

Contents Abstract Introduction Modeling Engine Performance General Results and Discussion Results of a Parametric Study A Comparison With S.I. Piston Engines Concluding Remarks References Acknowledgements Appendix I

1 2 6 19 29 39 43 44 45 46

The report describes the results of work carried out for VGT Technologies Inc. of Calgary to predict some aspects of the performance of a novel variable geometry toroidal engine, VGT. A computer model developed previously by the authors for predicting the performance of a spark ignition reciprocating engine was suitably modified and applied to a version of the engine having two pistons when operating with methane as a fuel, over a range of operating and design conditions. As a result some of the unique operational features of the engine are described. However, no proper comparison with the performance of a corresponding conventional piston engine was attempted at this juncture.

i- Introduction A toroidal revolving piston engine, described as the TRP engine, was invented by Mr. R. Pekau of Calgary. The engine, which is also described as a Variable Geometry Toroidal Engine (VGT), was put forward as a possible unconventional rotary engine that could have some positive operational features that may be worthwhile to evaluate and assess realistically in comparison to the conventional piston engine. Clearly, the proper evaluation of such a novel and untried device and establishing some of its major design and operational features are much too difficult , costly and time consuming to achieve at this time with the resources and state of knowledge available. The present contribution, while fully acknowledging these serious limitations, adopts after suitable modifications a performance predictive model, previously developed at the University of Calgary for spark ignition engines operation. The suitably modified model predicts the performance of the VGT engine for a very limited number of cases with a typical set of design features and operating parameters. Obviously, to have tried to obtain a more comprehensive and balanced assessment of the operational advantages and limitations of the VGT engine, particularly in comparison to the conventional piston engine, is a formidable task that is well beyond the mandate and objectives of this project. Nevertheless, the predictive model developed and used to obtain engine performance represents to our knowledge a most effective and realistic approach that is distinctly superior to all previous approaches used. The VGT engine, at this stage of conceptual development, has many design and operational features that can be varied quite widely. Accordingly, only a very brief description of the workings of the engine will be made here. A more detailed

description of the engine and the different possible variations in its working and design features may be obtained from the appropriate unpublished reports available, including the patent application made by VGT Technologies Inc.

The TRP/VGT rotary engine is made up primarily of two major parts. A cylinder and a combustion chamber. The first is a toroidal or donut-shaped circular cylinder with a cross-sectional area that does not have to be circular. (For this study, the precise definition of the geometry of the engine is somewhat of a lesser significance). Within this toroidal cylinder two or more equally spaced pistons are connected to a power disc revolving around the central axis. The compression and expansion paths are thus curved. The force of the expanding gas acting on the rotating pistons is transmitted to a central shaft to provide a constant-length torque output acting between the pistons and output shaft. As shown schematically in Fig. 1, a moving sealing mechanism separates the cylinder volume between the two pistons into two separate segments for compression and expansion. A separate constant volume chamber provided for combustion receives the compressed gas from the compression segment and after fuel injection, mixing, ignition and flame propagation delivers the hot products of combustion via timed valving into the expansion segment of the cylinder. An intake port located into the shell of the cylinder, delivers suitably pressurized air via timed valving over a period of time, into the segment of the cylinder. The piston on its rotation covers the intake port to trap the working charge, immediately preceding the commencement of compression. Similarly, an exhaust port is located in the toroidal cylinder such that it intersects the expansion path allowing the unused gas to escape. The piston with one face forming the sealing mechanism of the compression segment is described as the compression piston. Also, the piston that forms the face of the expansion segment is described as the expansion piston. The other pistons when used, with a single combustion chamber, are described as idling pistons that retain the idle part of the cylinder.

Fig.1 - Schematic Representation of a VGT Engine With Two Pistons

Table I A Listing of Some of The Main Variables of The VGT Engine I. Design Variables 1. Number of pistons 2. Compression ratio 3. Geometry and size a. Size and shape of intake port b. Size and shape of exhaust port c. Size and shape of combustion chamber intake valve d. Size and shape of combustion chamber exhaust valve e. Mean diameter of toroidal cylinder f. Diameter of cylinder g. Size and shape of combustion chamber h. Size of residual gas in cylinder 4. Crank angle (CA) and timing a. CA to begin opening of intake port b. CA to start closing of intake port c. Duration for opening completely of intake port d. Duration for closing completely of intake port e. CA to begin opening of combustion chamber intake valve f. CA to start closing of combustion chamber intake valve g. Duration for opening completely of combustion chamber intake valve h. Duration for closing completely of combustion chamber intake valve i. CA to begin opening of combustion chamber exhaust valve j. CA to start closing of combustion chamber exhaust valve k. Duration for opening completely of combustion chamber exhaust valve l. Duration for closing completely of combustion chamber exhaust valve m. CA to begin opening of exhaust port n. CA to start closing of exhaust port o. Duration for opening completely of exhaust port p. Duration for closing completely of exhaust port 5. Crank angle to begin fuel introduction 6. Duration of fuel introduction II. Operating Variables 1. 2. 3. 4. Intake pressure Exhaust pressure Intake temperature Exhaust temperature 5. Spark timing 6. Equivalence ratio 7. Revolutions per minute

III. Combustion model variables 1. Combustion duration 2. Ignition lag 3. Energy release paten

ii- Modeling Engine Performance Assumptions The version of the VGT engine considered in this report as stated earlier, is assumed to consist of two separate sections: a main toroidal cylinder and a separate combustion chamber. The volume of these two parts are treated as separate and connected by a timed valve-orifice for the purpose of this modeling. The toroidal cylinder and combustion chamber which are shown schematically in Fig. 1, are to be described throughout this report as the cylinder and chamber , respectively. As in the case of the conventional spark ignition piston engine, the chamber charge during combustion and following the passage of a timed spark is to be divided into two time varying zones, burned products and unburnt reactants. The following are the main assumptions made: The fresh intake mixture and the residual gases from the previous cycle are mixed homogeneously and uniformly. The fuel when introduced is mixed with the charge homogeneously and uniformly. The two combustion zones are homogeneous throughout and have their uniform properties. The pressure at any time is uniform throughout the cylinder and chamber. The flame thickness is negligible, but heat transfer to the outside is accounted for. The gases behave as ideal. Leakage from the cylinder and chamber is negligible. The discharge coefficient of the ports and valves is taken for simplicity to be equal.

For the cases considered and displayed in this report, the following are the values of the major design and operating parameters used:

There are only two pistons in the toroidal cylinder located diagonally opposite to each other. The mean diameter of the toroidal cylinder is 0.60 m The bore and diameter of the piston are 50 mm. The volume of the residual gas in the cylinder at the end of compression is 0.60 mL. The area of the intake port is equal to the cross sectional area of the toroidal cylinder. The area of the combustion chamber intake valve is equal half the cross sectional area of the toroidal chamber. The area of the combustion chamber exhaust valve is equal to 0.20 of the cross sectional area of the toroidal cylinder. The crank angle at the beginning of opening the intake port is 40 degrees and will become fully open at 50 degrees. The crank angle at the beginning of opening the exhaust port is 310 degrees and will be fully open at 320 degrees. The crank angle at the beginning of the intake valve of the combustion chamber is taken as 100 degrees and becomes fully open at 120 degrees other timings could have been similarly employed instead. Also the crank angle at the beginning of the opening of the combustion chamber exhaust valve is 187 degrees and will become fully open 20 degrees later at 207 degrees. The intake valve closes at 173' The crank angle at the closing of the combustion chamber exhaust valve is at 280 degrees. The fuel used was considered to have been methane representing natural gas operation. The mode of flame propagation and consequent combustion energy release are assumed to be similar to those developed for spark ignition applications with methane. 7

The mode of heat transfer can be assumed to be similar to that encountered in the corresponding spark ignition engines.

The Mathematical Formulas and Equations Full description of the model developed for spark ignition engine operation which formed the basis for modeling the VGT engine can be seen in Appendix I. Relatively minor changes were made to it, mainly to consider the specifics of the the configuration and working pattern of the VGT engine. Accordingly, the following are the main relevant equations employed: Equation of State: PV = mRT

(1)

where m is the mass of the charge, R, is the gas constant, P, is the chamber or cylinder pressure , V, is the cylinder or chamber volume and, T, is the mean temperature.

Mass Conservation: (2)

where subscript, u, indicates the unburned mixture and subscript, b, is the burned mixture. is the crank angle.

Mass Conservation:

For the combustion chamber at any time during combustion: (3) 8

Vcc, is total volume of the fixed volume combustion chamber.

For the toroidal cylinder: The total volume of the toroidal cylinder and the rate of its change can be expressed as: (4)

and

(5)

where, D, is the cylinder bore and the diameter of the piston. R, is the mean radius of the toroidal cylinder. respectively. Indicated Work: At any instant the rate of work done by the whole charge considered to be the system: , and are the crank angle and the reference crank angle,

(6) Energy Conservation: The energy equation for the whole closed system of the charge at any instant of time corresponding to an angle, (7) where: (8) and dQ/d@ is the rate of net effective heat transfer of the charge.

The Amount of Residual Gas: It is assumed that at the end of the exhaust stroke, residual gas fills the whole clearance volume of the cylinder at the exhaust temperature and pressure. The amount of residual gas is given as:

(9)

where, nr, is the number of moles of residuals, R, is the universal gas constant, Vc, is the clearance volume, Pc, and ,Tc, are the exhaust pressure and temperature. Similarly it is also assumed that at the end of the combustion chamber exhaust valve closing, a corresponding residual gas fills the whole combustion chamber volume at the corresponding pressure and temperature. The amount of the residual gas is calculated, similarly.

The Intake Stroke: When the fresh charge enters the chamber, it is assumed that it will mix homogeneously with the residual gas remaining in the cylinder from the previous cycle. The resulting charge temperature and the amount of fresh charge can be found by applying the First Law and the conservation of mass equations. Since the intake of the fresh charge and and the exhaust of gas are made through ports, it is necessary to boost the intake pressure either by supercharging or turbocharging so as to scavenge the cylinder clearance volume. The amount of fresh charge and the amount of gas pushed through the exhaust port, as the pressure inside the cylinder rises due to the increased introduction of the fresh charge from the higher intake pressure, are calculated using an equation similar to the one used for the flow through an orifice as: 10

(10) and

(11)

where ,mfr, and mex are the mass of the fresh charge introduced into the cylinder and the mass of the gas exhausted from the cylinder, respectively. Cdis, is the discharge coefficient of the corresponding intake or exhaust port. Ain and Aex are the areas of the intake and exhaust ports, respectively. Pin, Pex and P are the intake, exhaust and cylinder pressures and P is the mixture density at the throat of the orifice.

The Compression and Expansion Processes: The compression and expansion processes in this VGT engine occur into two distinct phases: one while all the valves and ports are closed and another while a valve or a port is opened during compression and expansion. When all the valves and ports are closed:

(12)

A combination of equations 1 and 12:

(13)

and (14)

11

Compression and Expansion Processes with Mass Exchange: The compression and expansion of the charge when a valve of the combustion chamber is opened are accompanied by changes in the mass during the process which are calculated using equations similar to those of equations 10 and 11, while using the corresponding values. Then the pressure and temperature are updated continually with the new mass values. The changes in pressure and temperature due to compression or expansion are calculated using similar equations to those of equations 12, 13, and 14.

The Combustion Process: The combustion chamber of this VGT engine which is located independently outside the toroidal cylinder is of constant volume. It is assumed arbitrary that the diameter, of the cylindrical chamber is 70% that of the toroidal cylinder diameter. During flame propagation, when the charge is assumed to be divided into two varying in size zones, burned and unburned, the energy equation can be written as :

(15)

Differentiating the equations of state with respect to crank anglee;

(16)

Applying the First Law to a unit mass of mu,:

(17)

12

where Qu is heat transfer to the unburned mixture and VU is the specific volume of the unburned mixture. Equations 17 can be rewritten as:

(18)

i.e. (19)

From equations 15, 16 and 19:

(20)

(22)

Heat Transfer: The heat transfer from the two zones to the outside wall can be estimated using formulations developed for spark ignition engine applications (Woschni, 1). However, alternative formulations can be employed if available and considered more representative for this case.

13

(23)

To simplify the treatment and due to a lack of knowledge of the separate heat transfer characteristics of the two zones, T is taken here for the purpose of heat transfer estimates to be the mean mass weighted temperature:

(24)

X , which is the heat transfer coefficient may be taken to be similar to that of Woschni(1,). The total heat transfer through the whole cylinder walls are assumed to be equal the sum of the heat transfers from the burned and unburned zones.

(25) This equation may be written as follows:

(26) where

(27)

Combining equations 26 and 27, the burned and unburned heat transfer areas, Ab and AU , may be found. 14

Fig. 2 - Comparison Between Assumed And Experimentally Derived Mass Burning Rates (900 RPM, Cr =8.5, Ost = 15BTDC, O =1.0 and To =300K with methane)

(28)

(29)

Thus, the heat transfer through the burned and unburned zones are:

(30) (31)

In view of the relatively simple nature of the model developed, this approach is deemed quite adequate. The composition of the products of flame propagation is calculated at the prevailing temperature while accounting comprehensively for thermodynamic dissociation (3).

Combustion Energy Release: The energy released by combustion is assumed to be similar to that of the internal combustion spark ignition engine but taking place at constant volume and finite time. The examination of a very large number of energy release diagrams in an internal combustion spark ignition engine fueled with methane when obtained from the processing of experimentally gathered pressure-time data for a wide range of operating conditions, indicated that their shape tends to be essentially similar (4,5). As a typical example, the derived mass burning rate, which is approximately proportional to the energy release diagram obtained in a CFR engine, is compared with the corresponding values obtained while using approximating functions is presented in Fig.2. On this

16

basis it would appear that the effective combustion energy release pattern may be characterized well on the basis of two main elements: the combustion duration and the maximum value of the burning rate. The former determines how long the combustion process goes on, while the latter indicates how fast the combustion process progresses. The area enclosed within the energy release diagram represents the total net energy released by combustion. Moreover, it can be observed immediately following passage of the spark, no significant energy release appears, but after a very short ignition lag, a flame kernel is developed that grows. The energy release rate falls to zero as the unburnt reactant mass is used up. In this work these rate diagrams are suggested to be approximated by triangles with bases corresponding to the combustion period while their areas are the known total energy released. Therefore, for the triangular case (6), the instantaneous mass burning rate, Am, can be represented by:

(32) where

(33)

(34)

where 0e.i, 0e.c. and Bmax are the crank angles at the end of combustion period and at a location associated with maximum burning rate. This latter will vary from one case to another but on the basis of our own observations in engines, it may be assumed to take

17

place between 1/2 to 2/3 of the length of the combustion period. In the VGT rotary engine, where the combustion chamber is separate and becomes isolated from the toroidal cylinder during the period from closing of the combustion chamber intake valve and the opening of the combustion chamber exhaust valve, it is assumed that the combustion is completed during this time for this phase of modeling procedure. However, other alternative suitable assumptions can be equally employed. Generally, the ignition lag period for spark ignition internal combustion engines is much smaller than the corresponding combustion duration and the mass burnt during the lag is taken as 1% to 5% of the total charge. Therefore, for this work it is assumed that the mass burned during the lag is 5%, as suggested by Al-Himyary,(5). The length of the combustion duration, as mentioned earlier was taken for convenience to be equal the time between closing the intake valve and opening of the exhaust valve of the chamber. The location of the maximum burning rate is taken as at the mid-point of the combustion duration.

These assumptions and the modeling approach followed are adequate for providing a general overview of the performance of the VGT engine within the time and resources available for the project at this stage. Obviously, a more elaborate approach that would vary widely the values of the various design and operational parameters of the engine could have been similarly applied. However, this would have required significantly more resources especially when considering the very large number of variables associated with the VGT engine.

18

iii- General Results and Discussion The modeling of the VGT engine performed, as was indicated earlier, had to make necessarily numerous important assumptions. Most of these were made largely on adhoc basis, to make the task manageable and because of the flexible definition of the design specifications at this stage of conception and development. These assumptions, despite the fact that it was unknown beforehand how this unconventional engine would respond to any changes in the key design and performance variables, are quite reasonable and consistent with the expected good design features for the engine. Of course, the predicted performance characteristics displayed in the computed results of our modeling are very much dependent on these assumptions and values selected. Moreover, it would be expected that superior performance than what the results would show, can be readily be obtained when some optimization procedures are adopted. These in view of the large number of variables and lack of previous experience with the such an engine is just simply impractical, too time consuming and cannot be pursued to any practical extent at this stage. Of course also, no attempt was made at this stage to predict the levels of emissions of the VGT engine nor the onset of knock. Such predictions for this totally unconventional and untried engine cannot be usefully managed at this stage. It has been assumed when calculating the results to be shown that combustion takes place only in the combustion chamber and the products of combustion are expanded into the toroidal cylinder with fuel addition taking place in the combustion chamber only. Moreover, it is obvious at this very early stage of evaluation of the VGT engine, that no reliable comments can be made as to the frictional and motoring losses that the engine would incur. Accordingly, the apparent power output displayed throughout is the indicated value without acounting also for any of the compression work required for the incoming air. The variations of the volume of the toroidal cylinder with crank angle is shown in

19

Fig.3. The cylinder volume changes with a constant rate as the velocity of the stroke does not change. Similarly, the volume of the combustion chamber is shown to remain unchanged. The typical variations of the cylinder and chamber pressures with crank angle are shown in Fig.s. 4, 5 and 6. The cylinder pressure increases as the compression process begins just after the closer of the intake port and continues to increase until the end of the compression stroke as shown in Fig.4. At the end of compression when the combustion chamber intake valve just closes, some of the charge remains in the cylinder residual volume at high pressure and temperature. This trapped residual mass would not take part in the combustion and would eventually be flushed off to the exhaust stroke of the previous piston cycle as the piston crosses the disc valve arrangement. During the expansion stroke, the cylinder pressure just before opening the combustion chamber exhaust valve, remains equal to the intake pressure of the following piston cycle as the pressure at the back side of the piston is always exposed to the intake pressure of the following piston cycle until it crosses the disk valve arrangement, as shown in Fig.3. As the combustion chamber exhaust valve opens the cylinder pressure increases drastically and begins to expand until the valve is closed. The cylinder pressure drops quickly at the end of the expansion stroke as the exhaust port is opened. Similarly, Fig.5 shows the typical temporal variations in the chamber pressure. Initially, as the chamber intake valve is opened, the pressure increases during the compression stroke. This increase in pressure continues until its intake valve is closed when the chamber becomes isolated from the toroidal cylinder. The pressure increases very rapidly following ignition and subsequent combustion. Once the exhaust valve of the chamber is opened, the pressure begins to decrease as the hot products of

20

Fig.3 Variation of the Cylinder Volume With Crank Angle at a Compression Ratio of 8.5

Fig.4 Typical Cylinder Pressure Variation With Crank Angle

Fig.5 Typical Chamber Pressure Variation With Crank Angle

Fig.6 Variations of Cylinder and Chamber Pressures with Crank Angle

combustion escape into the cylinder. It can be seen that this drop in pressure continues until the closure of the chamber exhaust valve, when the chamber begins again to be isolated from the cylinder Fig.6 shows the combined temporal variations of the two pressures. The corresponding temporal variations of the temperature in the cylinder and the chamber are show in Figs. 7 to 10. The mean temperature of the cylinder gas which is shown in Fig.7, begins to increase as the intake port closes and compression commences. This continues until the end of compression. However, just before opening the chamber exhaust valve the mean temperature in the cylinder will be very small and equal to the intake temperature of the following piston cycle. As the chamber begins to discharge its hot contents, the mean temperature of the cylinder gas increases rapidly and then begins to decrease as the expansion proceeds to the exhaust stage. Similarly, the corresponding variation in the temperature of the chamber gas is shown in Fig.8. It can be seen that initially the temperature in the chamber is higher than the temperature of the gas in the cylinder. The chamber temperature decreases as new colder gas is introduced from the cylinder and the required amount of fuel injected into the chamber. This is continued until the closure of the chamber valve when combustion begins following the passage of a spark. As the combustion proceeds, the temperature increases very rapidly to very high peak values. Once the chamber exhaust valve begins to open and the hot gases to expand into the cylinder, the temperature continues to fall until the exhaust valve is closed. In Fig. 9, the temporal variations in the mean temperature of the gases in the cylinder and the chamber are shown in Fig.10, the temporal variations in the burned and unburned regions of the chamber during combustion are additionally shown. A typical variation with time of the gas masses in the cylinder and the combustion chamber are shown in Fig. 11. It can be seen that the mass remains unchanged after the

23

Fig.7- Variation of The Mean Gas Temperature Inside The Cylinder With Crank Angle

Fig.8- Variation of The Mean Gas Temperature Inside the Combustion Chamber With Crank Angle

Fig.9- Variations of The Gas Temperature Inside Both The Cylinder And The Chamber

Fig.10- Temporal Variations of The Burned And Unburned Gas Mixtures During Combustion Inside The Chamber, The Gas Mean Temperatures Inside The Chamber And Cylinder Are Also Shown

intake port closes until the chamber valve is opened and the gas begins to move into the chamber. As the cylinder pressure increases during the compression stroke, the mass flow into the chamber increases. This continues until the chamber valve is closed isolating the chamber from the cylinder. The trapped mass in the chamber begins to drop as the chamber exhaust valve is opened and the gas begins to escape into the cylinder. Further expansion of the gas into the cylinder takes place at constant mass until the exhaust port is opened. Figure 12 shows the corresponding values of the Mach number during the intake and exhaust processes through the two ports and the two valves. As can be seen, the value remains below unity except at the beginning of the combustion chamber exhaust valve opening when the valve is choked until the back pressure builds up sufficiently. Further, the exchange of mass between the cylinder and chamber occurs at relatively low Mach numbers. However, during the intake and exhaust processes, the Mach number remains high. A typical temporal variation in the indicated work output is shown in Fig.13. The work is negative during compression, constant during the constant volume combustion and increases very rapidly during expansion. Similarly, Fig.14 shows the work output from the two pistons in a cycle. The corresponding variation in indicated torque is also shown. It can be seen that both the total work output and torque vary significantly over the cycle. However, the constant torque arm of VGT engine does tend to moderate this torque variation somewhat in comparison to that in the reciprocating engine.

26

Fig.11- Variations of The Masses Inside The Cylinder And Combustion Chamber With Crank Angle

Fig.12- Variations With Crank Angle in The Mach Number of The Different Gas Flows During The Intake And Exhaust Processes

Fig.13- Variations With Crank Angle of The Indicated Work Output

Fig.14 Temporal Variations Over a Full Cycle of The Indicated Work Output And Torque

iv- Results of a Parametric Study

As was indicated earlier, the number of variables involved in the design and operation of the VGT engine are quite numerous. Also, to obtain a realistic view of the performance features of the engine, not only the effects of changes in each influencing parameter on engine performance need to be established,but also the consequences of the simultaneous changes in various groups of these variables. Only this way, some degree of optimization of performance can be achieved. Clearly, this is an impossible task to undertake at this stage, even when only predictive approaches are being considered. Accordingly, in view of the long computational time and effort associated with calculating a single cycle of the VGT engine, only a small number of cases were considered in the parametric study conducted. The values of some of the key design and operational variables were selectively changed in turn, one at a time and the consequent changes to engine performance noted. This limited investigation served to provide an indication of the relative importance of the variables considered, at least for the typical conditions selected. Figure 15 shows the variation in indicated power output with changes in the engine rotational speed for a set of operating and design configurations. It can be seen that over the range considered the power output continued to increase virtually linearly with speed, reflecting the increase in number of working cycles per unit time without affecting adversely significantly the breathing capacity of the engine nor its combustion rates. This increase in power, can be seen in Fig. 16, to have taken place in spite of the accompanying drop in the mean indicated effective pressure and torque output. Figure 16 also shows the associated drop with the increase in speed of the ratio of the fresh intake mass to the total trapped mass in the cylinder, (fresh mass ratio). It can be seen that for these conditions, as the engine speed increased the mass of the chamber

29

mixture and the ratio of the chamber mass to the total mass trapped in the cylinder, (mass ratio), were affected only little. It was stated earlier that the engine needs to be supplied with a positively externally charged air. Hence, it would be expected that for any engine and set of operating conditions, the variations in the intake supply pressure will have a significant effect on engine performance. Figure 17 shows the indicated power output increasing effectively linearly with the increase in the ratio of the air supply pressure to that of the ambient (or exhaust) pressure. The increase in the value of this pressure ratio, which is described here as the supercharging factor, increases linearly the mass of the mixture in the combustion chamber and the associated mass of fuel consumed per cycle. It can be seen in Fig. 18 that both the mass ratio and the fresh mass ratio remain unchanged. The indicated mean effective pressure and output torque, similar to the power output, increase also linearly with the supercharging ratio. Another key influencing variable for any engine and set of operating conditions is the amount of unutilized volume or residual volume to be inevitably left in the toroidal cylinder at the end of the compression stroke. Figure 19 shows the indicated output dropping linearly as this volume is increased for a set of operating conditions. Figure 20 shows that although the fresh mass ratio is unaffected by this increase in the residual volume, both the chamber mixture mass and the the mass ratio are reduced and hence reducing the power output. This is clearly one of the important features of the engine design that needs appropriate remedial measures. In the VGT engine, unlike in the reciprocating engine, the compression ratio may be defined in more than one way. This is mainly because the minimum volume in the VGT engine is that of the fixed combustion chamber. When an increase in the compression ratio is assumed to take place through reducing the combustion chamber volume while keeping the intake volume constant, the engine performance would continually suffer, as in the case of increasing the unutilized volume. Figures 21 and 22 show continued 30

decreases in the power output, I.M.E.P. and mass ratio with the increase in this compression ratio. The equivalence ratio ( i.e. the fuel to air mass ratio relative to the corresponding stoichiometric ratio ) is an important variable that affects engine performance significantly. Figure 23 shows the variation of the indicated power output for a fixed spark timing engine, with changes in equivalence ratio with methane as the fuel,. It can be seen that maximum power is obtained for these conditions expectedly with a mixture that is slightly richer than stoichiometric. The power output drops quite rapidly as the mixture is leaned. As can be seen in Fig.24 for this constant velocity engine; the I.M.E.P. followed similar trends to those of the power output. The corresponding mass ratio, fresh mass ratio and the chamber mass were unaffected. Figures 25 and 26 show the drop in the indicated power output and I.M.E.P. with increasing the intake temperature. For these operating conditions with methane, when the mass ratios are kept constant, the drop in power with increasing temperature tends to be somewhat moderate. Some calculations were also made to obtain an indication of the effects of changes in size on engine performance. The results showed that moderate changes in the sizes of ports and valves do not affect the power output substantially, except when various engine speeds were employed. As an example, Fig.27 shows the extent of variations in power output with engine speed for two sizes of ports and valves. The full size indicates that the sizes of the intake and exhaust ports and the combustion chamber intake and exhaust valves were of the same size as used for calculating the previous results. The half size indicates that the corresponding area of ports and valves were halved. As can be seen, reducing the size of these elements affected the power output adversely especially at high engine speeds reflecting the reduction in the time available to carry out the filling and exhausting processes. No other changes in the values of the design parmeters were attempted. 31

Fig.15- Variation of the Indicated Power Output With Engine Speed

Fig.16- Variation of I.M.E.P., Chamber Mass and Mass Ratios With Engine Speed For The Same Conditions as in Fig.15.

Fig.17 - Variation of The Indicated Power Output With The Intake Pressure Ratio, (Supercharging Factor)

Fig.18 - Variation of I.M.E.P., Chamber Mass and Mass Ratios With The Intake Pressure Ratio, (Supercharging Factor ), For The Same Conditions as in Fig.17.

Fig.19 - Variation of The Indicated Power Output With Changes in The Cylinder Residual Volume

Fig.20 - Variations in The I.M.E.P., Chamber Mass and Mass Ratios With Changes in The Cylinder Residual Volume For The Same Conditions of Fig.19.

Fig.21 - Variation in The Indicated Power Output With Compression Ratio.

Fig.22 - Variation of I.M.E.P., Chamber Mass And Mass Ratios With Compression Ratio For The Same Conditions of Fig.21.

Fig.23 - Variation in The Indicated Power Output With Equivalence Ratio.

Fig.24 - Variation in I.M.E.P., Chamber Mass And Mass Ratios With Equivalence Ratio For The Same Conditions of Fig.23.

Fig.25 - Variation of The Indicated Power Output With Intake Temperature.

Fig.26 - Variation of I.M.E.P.,Chamber Mass and Mass Ratios With Intake Temperature For The Same Conditions as of Fig.25.

Fig.27 - Variation of The Indicated Power Output With Engine Speed When The Sizes of Ports And Valves Were Reduced by a Half.

v- A Comparison With S.I. Piston Engines Obviously, much care is needed to be exercised when trying to compare the computed VGT performance with that of a corresponding conventional spark ignition reciprocating engine. The VGT, as was indicated earlier, has far too many influencing key variables with the compression, combustion and expansion processes performed differently and under previously untried conditions. Such a comparison would in all likelihood show the performance of the conventional engine in a better light since no attempt has been made throughout to select the values of the key operational and design variables of the VGT engine optimally. Accordingly, only a limited and somewhat inadequate form of comparison between the two power outputs was attempted. The performance of the two types of engine was modeled for similar size and under the same operating conditions. Their work outputs were then compared on the basis of the same working mass. Figure 28 shows the specific work output of the conventional spark ignited piston engine while operating at 300 rev/min and a constant spark timing of 20 degrees BTDC, compared to the the specific work output of the rotary engine when operating at the same speed and compression ratio on the basis of the working mass can be either that of the combustion chamber or the mass inside the toroidal chamber. On this basis, the specific work output of the VGT engine is lower than the corresponding values of the conventional engine. A contributary factor to this observed difference is the fact that some of the compressed mass in the VGT engine is trapped in the cylinder and cannot take part in the combustion. Thus, it will not contribute to the observed output. It is shown that when the volume of this unutilized residual volume is reduced arbitrarily to half its value, significant increases in the specific output can be seen. Figure 29 shows that the specific work output of the rotary engine is throughout lower than that of the corresponding piston engine and reduces when the engine speed 39

is increased over a wide range. This is since there is proportionally less time to complete the combustion and carry-out the filling and scavenging processes. The specific work output on the basis of the cylinder mass expectedly is lower than that on the basis of the chamber mass. However, it is very interesting to note that there might be some optimum conditions that can lead to superior performance as shown for the case with an operational speed of 900 rev/min. Moreover, as shown in Fig. 30, the specific indicated power is also compared with that of a corresponding spark ignition engine operating on the four stroke cycle. It can be seen that at low compression ratios the power output of the rotary engine may appear to be greater than that of the conventional engine which is direct reflection of the fact that the rotary engine has twice the number of power strokes in comparison to the reciprocating engine for the same rotational engine speed.

40

Fig.28 - A Typical Comparison of The Specific Indicated work Output of The VGT Engine For Different Unutilized Residual Volume With a Reciprocating Engine at Different Compression Ratios. (Equiv ratio 0.99, ambient pressure 87 kPa, 294 K and 300 rev/min-methane operation )

Fig.29- A Typical Comparison of The Specfic Indicated Work Output With Compression Ratio of The VGT Engine on The Basis of The Chamber Mass And The. Cylinder Mass With a Conventional Piston Engine at The Same Speed (Equiv ratio 0.99,87 kPa, 294 K at 300 rev/min-methane operation )

Conventional reciprocating engine at 300 rpm --Rotary engine per chamber mass at 300 rpm --Rotary engine per cylinder mass at 300 rpm

Fig. 30 - A Typical Comparison of The Specific Indicated Power Output Variation With Compression Ratio For The VGT Engine And a Conventional Piston Engine at Three Different Speeds. (Equiv ratio 0.99, ambient pressure of 87 kPa, temperature 294 and engine speed of 300 rev/min)

vi- Concluding Remarks

The performance of the VGT engine has been shown to be dependent on far too many inter-related design and operational variables. The predicted results presented in this report were obtained using a suitably modified for the purpose, computer model developed previously for predicting the performance of a spark ignition engine. These results, which relate to the performance of a specific version of the VGT engine having two pistons, showed clearly the engine to be operational and capable of producing power over a range of operating and design conditions that were selected largely on adhoc basis. However, it is concluded that in view of the VGTs unique design and its previously untried operational features, a fair and proper comparison with the corresponding performance of a conventional engine cannot be made meaningfully at this juncture.

vii - References [1] Woschni G. A Universally Applicable Equation for the Instantaneous Heat Transfer Coefficient in the Internal Combustion Engines. SAE Trans., 76(670931):3065-3083, 1967. [2] Heywood J.B. Internal Combustion Engine Fundamentals. McGraw Hill Book Co., 1988. [3] Strehlow R. Fundamentals of combustion. McGraw HilI Co., 1988. [4] Gao J. Knock Modeling in S.I. Engines. PhD thesis, Department of Mechanical Engineering, University of Calgary, 1993. [5] Al-Himyary T.J. A Diagnostic Two-Zone Combustion Model for Spark Ignition Engines Bused on Pressure Time Data. PhD thesis, Department of Mechanical Engineering, University of Calgary, 1988. [6] Bade Shrestha S. O. and Karim G. A. A Predictive Model for Gas Fueled Spark Ignition Engine Applications. SAE, (1999-01-3482), October 1999.

44

viii - Acknowledgements

The contribution of the many past associates of the authors to the development of the spark ignition engine modeling program is gratefully acknowledged.

45

Appendix I

A Description of the Predictive Model Modified for Application to the VGT Engine

Published as an S.A.E. Paper No.1999-01-3482

G. A. Karim and O. M. Bade Sherstha

46

SAE TECHNICAL

PAPER SERIES

1999-01-3482

A Predictive Model for Gas Fueled Spark Ignition Engine Applications


S. O. Bade Shrestha and G. A. Karim
University of Calgary

Reprinted From: Modeling and Diagnostics in S.I. Engines (SP-1481)

The Engineering Society For Advancing Mobility Land Sea Air and Space

INTERNATIONAL

International Fall Fuels and Lubricants Meeting and Exposition Toronto, Ontario Canada October 25-28, 1999
Tel: (724) 776-4841 Fax: (724) 776-5760

400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A.

1999-01-3482 A PREDICTIVE MODEL FOR GAS FUELED SPARK IGNITION ENGINE APPLICATIONS S. O. BADE SHRESTHA AND G. A. KARIM The University of Calgary Dept. of Mechanical Engineering Calgary, AB, T2N 1N4 Phone (403)-220-5775, Fax (403)-282-8406 Email:karim@enme.ucalgary.ca

Abstract A predictive procedure for establishing the performance parameters of spark ignition engines fueled with a range of gaseous fuels and their mixtures is described. The incidence of knock and its relative intensity are also accounted for. The two-zone model incorporates a procedure for deriving an estimate of the effective duration of combustion and the associated mass burning rate for various operating conditions and gaseous fuels. The preignition chemical reaction activity of the unburned end gas zone and its consequences on cylinder pressure development is evaluated while using detailed chemical kinetics. The onset of autoignition and knock is established via a parameter that monitors the incremental pressure increase solely due to the preignition reaction activity per unit of mean effective combustion pressure. This knock parameter corresponds also to the specific energy release within the end gas due to the preignition reaction activity relative to the total energy released by combustion per unit of initial cylinder volume. For normal knock-free combustion the value of this parameter remains throughout very low, while for knocking combustion its value exceeds a certain acceptable limit. The more intense the knocking the earlier this value is reached. It is to be shown that this relatively simple model can be used to predict many aspects of engine performance parameters including the incidence of knock and account for cyclic variations as well as to be used in analytical optimization procedures. Experimental results obtained in a CFR engine for different fuels that include methane, hydrogen, ethane, propane and their mixtures yield satisfactory agreement with the corresponding predicted values. 1 INTRODUCTION The modeling of the complex combustion processes in the internal combustion engine has been a major area of research and development over the years. Continued 1

development of digital computers with their increasing capabilities and reduced relative costs have increased the interest to develop a comprehensive, yet relatively simple simulation models that can be used in more elaborate optimization approaches. This was mainly driven by the need to obtain higher fuel economies and lower emissions which make the optimization of engine performance experimentally a lengthy and expensive process. The many types of numerical simulation models that have been developed over the years can be divided broadly according to whether they are diagnostic or predictive. Diagnostic models usually employ engine experimental data, such as pressure-time records, to provide through analysis further information about the various processes of the engine. Predictive models attempt to simulate the engine combustion processes and produce key combustion and performance parameters such as pressure-time and mean temperaturetime histories that can be employed in association with other predictive sub-models for knock, cyclic variations, emissions etc. The simple zero-dimensional modeling approach assumes usually that the mixture throughout the cylinderremains homogeneous and for uniform properties as time is varied independently is clearly highly approximate. A simple treatment of the combustion processes is to consider their effect as equivalent to a heat addition or heat release phenomenon established usually on the basis of the first law and some correspondingresponding experimental data. On the other hand, computational fluid dynamics (CFD) models which are based on the numerical calculation of mass, momentum, energy and species conservation equations in either one, two or three dimensions can provide a great amount of predicted data of the flow, temper-

ature and concentration fields for the operating engine. Such approaches are becoming increasingly successful but require very large computing capacity even with a simplistic description of the complex chemical reactions taking place during combustion. The employment of multi-dimensional CFD models with full chemistry remains at present not feasible, especially when knock incidence and optimization procedures are to be modeled. In the present contribution, a relatively simple twozone quasi-dimensional simulation model is employed with full consideration of the combustion reactions. The model can simulate the temporal pressure and temperature variations for each of the two zones for any given operating conditions. It can predict the occurrence of knock and emissions and may be employed in the optimization of engine processes when required. 2 THE MODEL The model to be described [1-3] is two zone and quasidimensional. It has been developed mainly for predicting the performance of gas fuelled spark ignition engines including the determination of whether knock can take place and an estimate of its intensity. The homogeneous charge of the cylinder is assumed during the combustion period to be divided into two zones: burned products and unburned reactants that comprise the end gas region. Changes in the condition of the unburned reactants up to the time of spark ignition can be evaluated from a knowledge of the fuel and air flow rates while accounting for the presence of residual gases and heat transfer. The representation of the thermal aspects of the combustion process by an equivalent energy release pattern is a simplifying approach often adopted in engine performance modelling. In principle, temporal variations of the energy release rate and the corresponding length of the effective time period for combustion can be determined through either measuring or calculating the turbulent flame propagation rate right from the formation of the flame at the spark plug to the completion of its travel to the cylinder walls. At present, modeling accurately the unsteady three dimensional character of turbulent combustion in the engine is both very difficult and uncertain [4]. Therefore, merely to produce the pressure-time development with reasonable accuracy, several types of appropriate functions for the energy release rate pattern have been considered in the literature [5] based on formulations obtained following examination of a large number of experimental data for a variety of op2

erating conditions. Purely to maintain a simplicity of approach, a triangular function [3,6] was adopted in the results shown in the present contribution. It is assumed that turbulent flame propagation is completed and the combustion energy release would take place over a certain effective period of time described as the combustion duration. It starts just beyond the spark ignition timing after a short ignition lag period when a significant amount of energy begins to be released due to flame kernel development and ends with the termination of turbulent flame propagation. Formulations for the combustion period variations with some of the key operating parameters were obtained from an analysis of a mass of experimental data in a CFR engine operating on methane as well as binary mixtures of methane with hydrogen, ethane and propane [1,3,7]. The resulting energy release rate variation with time diagram with the combustion duration as a base would have its area represent the known total effective energy released during the cycle by combustion. This simple approach gives engine pressure temporal development and hence power output, that is in satisfactory agreement with experiment. In order to monitor for the likelihood of the onset of knock, the reactivity of the end gas region of the charge is evaluated throughout employing sufficiently detailed description of the reaction kinetics of the charge. The formation and growth with time of the many associated reactive species are considered and the energetic consequences of such reactions [2] are accounted for. A detailed chemical kinetic scheme [8,9] was employed that can describe the preignition and combustion reactions of common gaseous fuels such as propane, ethane, methane and hydrogen and their mixtures for conditions relevant to those normally encountered within the operating range of engines. The scheme employed, consisted of 155 elementary reaction steps and the following 39 chemical species: CH, CHCO, CH 2 , CH 2 CO, CH 2 O, CH 3 , CH 3 CHO, CH 3 CO, CH 3 O, CH 3 O 2 , CH 4 , C 2 H, C 2 H 2 C 2 H 3 , C 2 H 4 , C 2 H 5 , C 2 H 6 , C 3 H 4 , C 3 H 5 , C 3 H 6 , I C 3 H 7 , NC 3 H 7 , C 3 H 8 , CO, CO2, H, H2, H2O, H2O2, HCO, HO2, N2, O, O2, OH, N2O, NO, N, NO2. The corresponding thermochemical data were obtained mainly from JANAF tables [10]. The scheme, when necessary, can include additional reactions and species accounting for the detailed reactions of nitrogen and/or fuel mixtures that may contain more fuel components. The results of the present contribution relate to methane, ethane, propane, hydrogen and their mixtures.

A knock detection parameter, K, is incorporated in the model to establish whether knock may take place or not. It is based on the calculated variations with time of the integrated amount of energy released within the temporally changing end gas due to the preignition reactions while employing the full kinetic scheme per unit of the instantaneous cylinder volume, relative to the corresponding total energy released normally through flame propagation over the whole cycle per unit of cylinder swept volume [7,11,12]. This would also represent the increase in cylinder pressure due to the preignition reaction energy release of the instantaneous size of end gas relative to the mean effective combustion pressure. As the preignition reaction activity of the end gas becomes significantly intense to lead to autoignition and hence knock, the value of this parameter increases beyond a critical value despite the continued reduction in the size of the end gas. This value was found experimentally to be approximately constant when associated with the onset of mild knocking. The composition of the products of flame propagation is calculated at the prevailing temperature while accounting for thermodynamic dissociation [13]. Heat transfer from the two combustion zones to the outside engine walls may be accounted for using experimentally derived formulations [14]. The solution of the set of relevant simultaneous equations employing numerical methods yields values of the main properties of the two combustion zones and their variations with time. From the knowledge of the calculated cylinder pressure and volume variations with time, the corresponding indicated power output and efficiency can be evaluated. Moreover, the net changes in the mean concentrations of each species in the reactive end gas charge can be established. These are functions of the rates of all the simultaneous reaction steps involved [8,9,15] in the detailed chemical kinetic scheme describing the oxidation of the fuel-air mixture. The model is also capable of predicting other engine performance parameters such as the temporal variations in cylinder pressure and the mean temperatures of the two zones for different fuel compositions, speeds, intake temperatures, intake pressures and spark timings as well as accounting for the effects of changes in some design parameters such as compression ratio, engine size and valves timing [7,12]. Good agreement between the simulated data and the corresponding experimental values was found. The model could also predict satisfactorily the onset of knock due to the autoignition of the end gas and is used in other 3

modeling approaches such as optimization procedures [7] and accounting for operational limits and increased cyclic variations [16]. 3 MATHEMATICAL TREATMENT 3.1 Assumptions The cylinder charge is assumed during combustion to be divided into two zones: burned products and unburned reactants which comprise the end gas region, as shown in Figure 1. The following are the main as-

Figure 1: Schematic diagram of the two-zone model. Assumptions: The two zones are homogenous and have uniform properties. The pressure at any time is uniform throughout the cylinder. Flame thickness is negligible. Gases behave as ideal. Leakage from the cylinder is negligible. Equation of State: PV = mRT (1)

where m is the total mass of the charge, R is the gas constant, P is cylinder pressure, V is the cylinder volume and T is mean temperature. Mass Conservation: (2)

where subscript u indicates unburned mixture and subscript b is for the burned mixture. 6 is the crank angle. Volume Conservation: (3) The total volume of the engine cylinder and the rate of its change can be expressed respectively as:

(4) and (5) where D= The internal diameter of the cylinder S= Stroke length CR= Compression ratio a= Crank Radius L= Length of the connecting rod Indicated Work: The rate of work done by the whole charge considered to be the system can be expressed by the mean cylinder pressure and the corresponding rate change of the system volume.

where nr is the number of moles of residuals, E is the universal gas constant, Vc is clearance volume, Pe and Te are exhaust pressure and temperature. Intake stroke: At the beginning of the intake stroke the pressure of the residuals is normally higher than that of the intake manifold and the fresh charge would not enter into the cylinder until the pressure inside the cylinder has dropped sufficiently through expansion following the piston outward movement. The residual gas is assumed to expand from the exhaust pressure to the intake pressure, polytropically. When the number of moles of residuals, nr exhaust and intake pressures, etc. are known or assumed, then the cylinder volume at which the fresh charge begins to enter the cylinder can be estimated. When the fresh charge enters the cylinder, it is assumed to mix homogeneously with the residual gas. The resulting charge temperature and the amount of fresh charge can be estimated by applying the First Law and the conservation of mass equations. Since the intake valve does not necessarily close exactly and, promptly at bottom dead centre and open at top dead centre and because of pressure losses through the intake valves and scavenging process, a correction factor may be incorporated into the volumetric efficiency which is defined as the mass flow rate of the fresh fuelair mixture divided by that could have ideally been introduced via the volume displaced by the cylinder at intake conditions.

Energy Conservation: The energy equation for the whole closed system charge at any instant of time, corresponding to a crank angle, 8, is: (7) where

(8) and dQ/dO is the rate of heat transfer of the charge. The amount of residual gas: It is assumed that at the end of the exhaust stroke residual gas fills the whole clearance volume at the exhaust pressure and temperature. The amount of residual gas is given as:

where mo is the mass of fresh charge introduced into the cylinder per cycle, p+ is the mixture density at intake conditions and Vd is the swept volume. Compression and Expansion Strokes: The compression process unless otherwise stated is assumed to start at bottom dead centre (BDC) following valve closure, when the charge mass has reached its final value. During early stages of compression, the temperature tends to be relatively low and for simplicity reactions within the charge may be neglected. These reactions can be accounted for when deemed necessary. Combustion of the charge is initiated by a spark, when the reaction activity in the unburned zone is always considered. The first law of thermodynamics when applied to the charge as a single zone both during compression before flame initiation and during expansion following the completion of flame propagation results: (11) 4

A combination of Equations 1 and 11: (12) and (13) Combustion Period: During flame propagation, when the charge is assumed to be divided into two zones, burned and unburned, the energy equation can be written as:

The heat transfer from the two zones to the outside wall can be estimated using experimentally based formulations, such as that of Woschni [14]. Alternative formulations, of course, could be equally used. (22) For simplicity and due to the relatively poor knowledge of the separate heat transfer characteristics of the two zones, T is taken here for the heat transfer to be the mean weighted temperature of the charge: (23)

(14) Differentiating the equation of state with respect to crank angle 8:

X which is the heat transfer coefficient and may be taken according to Woschni as:

(24) A is the area of the cylinder surface. C1 is 6.18 for the scavenging period while for compression and ex-3 pansion strokes it is equal to 2.28. C2 is 3.24 x 10 (m/sK) for the whole cycle. D is the cylinder diameter (m). V1, T1 and P1 represent the known state of the working gas related to inlet valve closure. P0 is the gas pressure of the corresponding motored engine in the absence of combustion. V, T and P represent the instantaneous known state of the working gas in 3 m , K and kPa, respectively. Sp is the average linear piston speed in m/s. The total heat transfer through the whole cylinder walls are assumed to equal to the sum of the heat transfers from the burned and unburned zones. (25) (18) From Equations 14, 15 and 18: (19) i. e. (28) and (20) Hence: (30) (29) where: (26) (27)

(15) Applying the first law to a unit of mass of m,: (16) where QU is heat transfer to the unburned mixture and vu is the specific volume of the unburned mixture. Equation 16 can be rewritten as: (17) i.e.:

(21) 5

This relatively simple submodel of heat transfer is adequate for the predictive model of the cycle.

3.2 Energy Release Pattern The energy release diagrams in a spark ignition engine obtained from the processing of experimentally gathered pressure-time diagrams for a wide range of operating conditions indicated that their shape tends to be essentially similar [1,12]. As a typical example, the derived mass burning rate which is approximately proportional to the energy release rate obtained experimentally with methane in the CFR engine is compared with the corresponding values obtained while using approximating functions is presented in Figure 2. The effective combustion energy release rate pattern may be characterized on the basis of two main elements: combustion duration and maximum value of the burning rate. The period determines how long the combustion process goes on, while the rate indicates how fast the combustion process progresses. The area enclosed within the energy released diagram rep resents the total net energy released by combustion. Moreover, it can be observed that significant energy release appears only after a very short ignition lag. The rate falls to zero as the unburned reactant mass is used up.

but it was found that they change only very little the calculated pressure development with time and the major operating characteristics especially power output. Therefore, for the triangular pattern, the instantaneous mass burning rate, dm, can be represented by:

where (33)

(34) where 6e.i.l &.=. and 8,,, are the crank angles at the end of ignition lag, the end of combustion period and at a location associated with maximum mass burning rate. This latter value will vary from one case to another but on the basis of experimental observations it may be assumed generally to take place at around 2/3 of the combustion period. Other viable locations, ofcourse, can be easily used instead whenever necessary. However, these tend to represent usually only moderately small shifts from this chosen value. Combustion Duration: The variations of the combustion duration with equivalence ratio tends to have a similar trend to that shown typically in Figure 3. It is consistent with the typical variations of the turbulent flame speed with equivalence ratio for homogeneous fuel air mixtures, since the duration is proportional to the inverse of the averaged flame propagation rate [7].

Figure 2: Comparison between calculated and experimentally derived mass burning rates. Operating parameters 900 RPM, CR =8.5:1,0st = 15 BTDC, 4 = 1.0, T 0 = 300 K. The ignition lag and the combustion duration obviously are functions of a number of operating and design parameters that will be dealt with in the following sections. In this work these rate diagrams, merely for simplicity, were approximated by triangles with bases corresponding to the combustion period while their area are the known total energy released. Alternative variations on such an assumed simple shape can be made 6

Figure 3: Typical variations of experimentally derived combustion duration versus equivalence ratio in a CFR spark ignition engine with methane at 900 RPM, eSt = 10 BTDC, CR =8.5:1, To = 300 K and P0 = 87 kPa.

The operational equivalence ratio limits in an engine correspond to extremely long combustion periods. These limits have been examined experimentally and methods for their adequate estimation for any specific engine have been proposed [17,18]. The minimum value of the combustion period occurs normally around the stoichiometric ratio and may be estimated or measured experimentally [19] for the purpose of modelling. An appropriate combustion period correlation with changes in equivalence ratio can be formulated for any set of engine operating conditions. As an example, the following empirical formula may represent the curve of Figure 3: (35) where (36) and (37) A0, is the combustion duration in degrees for an equivalence ratio of 4. The corresponding lean and rich operational limits equivalence ratios are 41 and &, respectively. The equivalence ratio for minimum combustion time is &in. A and B are fitting constants that can be derived from experimental data or estimated for the engine and operating conditions employed (see Equations 40 and 41). 41 and & can be also found or estimated for any operating conditions. Karim et. al. [17,20] measured the apparent lean and rich operational limits for several fuels in a CFR spark ignition engine for different operating conditions and showed that the lean limit to be approximately a linear function of the mean calculated mixture temperature at the moment of spark discharge (Karim and Wierzba [18]): (38) where A1 and B1 are constants, which were found for methane operation in a CFR engine to be -1/50000 and 0.071, respectively. Tst is the average calculated mixture temperature (K) at the moment of spark discharge. Similarly, the rich operational limit was also found to be approximately a linear function of the difference of the mixture temperature at the moment of spark discharge (Tst) and the mixture temperature at intake conditions (To) (Karim and Wierzba [18]): (39) where Ar and Br are constants found by fitting experimental data for methane operation in a CFR engine to be l/10000 and 0.1, respectively. 7

Most hydrocarbon fuels tend to have generally similar flame propagation characteristics in a spark ignition engine [21]. Thus, this combustion duration calculation method proposed can be extended from methane applications to other gaseous fuels such as propane and ethane. The values of the constants of Equations 38 and 39 for these fuels then can be estimated from relevant experimental data [18,21]. Since the fastest flame propagation and hence, the minimum combustion duration normally occurs around the stoichiometric mixture, then a simplification can be employed to establish the values of A and B in terms of the known 41, 4r, &in and Ae,,,i, by taking dA&,/dqSl,,, = 0. i. e.: (40)

(41) where A&,,, is the minimum combustion duration. There are two-main factors affecting the combustion duration in a spark ignition engine. The first factor is the cylinder geometry and its size, which decides the effective flame propagation distance, dc. The other factor is the effective mean flame propagation speed at the prevailing conditions, Sf: (42) In a spark ignition engine, due to the motion of the piston and the changing temperature and pressure, both dc and Sf would vary during combustion. Various simplifications may be introduced to find a suitable correlation for estimating the minimum combustion duration for a certain engine and operating condition particularly when lean mixture operation is involved. For example, assuming that the flame propagation 1/3 distance dc is proportional to Vst is a reasonable approximation, where Vst is the combustion chamber volume at the time of spark discharge and depends on the spark timing and the compression ratio. Thus, an advanced spark timing and/or a low compression ratio will lead to a longer flame propagation path. Through a correlation of experimental data obtained in a CFR engine, the following equation was suggested [22]: (43) where Sf is the apparent flame propagation speed which is the other important factor influencing the value of the combustion duration which for any equivalence ratio and engine speed would depend on mixture temperature and pressure. Since the flow in a

spark ignition engine is strongly turbulent, the flame propagation is affected by the engine speed and nature of the turbulence intensity and scale. Correlations of the laminar flame propagation obviously cannot be applied directly in engine modelling mainly because of the variable geometry of the combustion chamber and the intense turbulent nature of the flow in a running engine. A general effective approach is to use these correlations after incorporating some appropriate corrections to fit the observed experimental engine data [1,23]. Al-Himyary and Karim [24] developed a generalized expression for the maximum flame propagation speed for methane based on the correlation of extensive experimental data obtained by others under different operating conditions. The maximum burning speed of methane, Sf,max, was found to be: (44) where P and T are the instantaneous mixture pressure (atm) and temperature (K), respectively. For a fixed engine speed, it is reasonable to assume that the minimum combustion duration of Equation 43 may be expressed as: (45) where P and T are the instantaneous mean pressure and temperature of the mixture during combustion which are unknown before the combustion process calculation in a predictive model. But, what is known are the P and T values at the moment of spark discharge. Thus, these values when employed in Equation 45 will need some correction to make it usable for predicting the combustion duration as: (46) where C is a proportionality constant. Accounting for the effects of turbulence on the length of the combustion duration may be through considering that the turbulent characteristic velocity is usually a function of the mean piston linear speed Sp [5]. Hirst et. al. [25] found the following simple relationship: (47) which may be employed to provide the following expression for Oc,min: (48) 8

where C is a constant assumed to be dependent on the cylinder geometry and the spark plug location. Other correlative approaches of experimentally observed data obviously also can be derived. An example of yet a simpler method to calculate the value of A&,min will be outlined further on in this work. In any case, the value of Atic,min in the key relationship of Equation 40 does not need to be known necessarily very accurately. Moreover, for lean operation even an assumed negligibly small value for A&,,;, at the stoichiometric equivalence ratio can produce reasonably workable approximation for the value of A& as a function of 4 to be used in simple predictive models, such as the one being described here. For any given set of operating conditions, from the computations during the compression stroke, the mean values of pressure and temperature at the moment of spark discharge are known. Equation 48 gives the value of the minimum combustion duration, while the lean and rich operational limits can be found using the procedure described earlier. The constants A and B then can be calculated. By substituting these into Equation 35, the combustion duration can be predicted. This method for estimating the combustion duration may be applied for various operating conditions. The constant, C, may be determined by measuring the combustion duration only once for the engine of interest. A typical comparison between the estimated combustion duration values calculated in accordance with this procedure and the corresponding experimental values obtained in a CFR engine operating on methane and air is shown in Figure 4 where good agreement can be seen. Ignition Lag: The term ignition lag is chosen for spark ignition engine operation to distinguish it from the ignition delay in diesel engines. Usually, the amount of the burned gas mass during the ignition lag has been considered to range from 1% to 5% of total charge. Hires [26] took 1% of total mass as the end of the ignition lag, while Hong [27] considered 2% at the end point. Al-Himyary [1] suggested 5% mass consump tion at the end of ignition lag. In the present work, Al-Himyarys criterion for ignition lag is used so that the predicted results can be compared with the mass of his diagnostic results. The length of the ignition lag is governed mainly by the chemical reaction activity and the diffusive transport processes and the effects of several operational parameters such as the equivalence ratio, piston speed, mixture temperature and pressure on the ignition lag have been investigated [1, 28,29]. A similar procedure as that used for estimating the combustion duration

lag with equivalence ratio can also be represented by a similar exponential function as that of Equation 35 obtained for the combustion duration: (49) where Ae;, is the ignition lag in crank angle (Degr.) and Aig Big are constants. Also in a similar approach to the model for the combustion period: (50)

(51) The value of the minimum ignition lag, Ae;,,,i,, can be found also by fitting experimental data to the following expression to produce acceptable agreement with the corresponding experimental observations: (52) where AOig,min is the minimum ignition lag and C is a constant. Hence, the ignition lag can be found using Equation 49. Figure 6 shows a typical comparison between the estimated ignition lag using the approach described and the corresponding results of experimental analysis [30]. It is to be remembered

Figure 4: A comparison between estimated combustion duration versus equivalence ratio and experimental data from a CFR spark ignition engine with methane at 900 RPM, CR =8.5:1 and To = 300 K for two spark timings. was adopted here to find a correlation for the variation of the ignition lag with operating conditions. Figure 5 shows a typical plot of experimentally derived values of the ignition lag with equivalence ratio [30] displaying very short durations in comparison to the corresponding combustion duration period and excessively long lags at the lean and rich limits. There is also a min-

Figure 5: Typical Variations of experimentally derived ignition lag with equivalence ratio in a CFR spark ignition engine with methane at 900 RPM, P0 = 87 kPa and T0 = 300 K. imum ignition lag that corresponds with conditions for the fastest flame propagation for any operational conditions. Accordingly, the variations of the ignition 9

Figure 6: A comparison between the estimated ignition lag and experimental data from a CFR spark ignited engine for methane operation at 900 RPM, CR =8.5:1, To = 300 K and P0 = 87 kPa for two spark timings. that since the value of the ignition lag tends to be always quite small relative to the length of the combustion duration, a less precise estimate of the length

of the ignition lag would not produce significant errors in estimating engine output parameters. Combustion Duration and Ignition Lag for Fuel Mixtures: The effect of the addition if common gaseous fuels such as hydrogen, ethane or propane to methane on the effective flame burning speed and performance of an engine have been investigated [30-32]. Generally, there appears to be no reliable simple model capable of predicting the burning velocity of fuel mixtures in terms of the corresponding values of their fuel components on their own in air within a wide range of temperature, pressure and composition, especially in relation to spark ignition engine applications where fully turbulent conditions prevail throughout. Hence, any correlations for the combustion duration and ignition delay of fuel mixtures have to be based empirically on experimental data obtained directly from measurements in the engine. Furthermore, the combustion duration is inversely proportional to the flame propagation speed which can be viewed in the case of fuel mixtures as a mean flame speed where the contribution of each fuel component of the mixture would be proportional to its own flame speed under similar conditions and its concentration in the mixture. Consequently, a simple expression for the combustion duration of a fuel mixture based on experimental observations we made in a CFR engine can be given approximately as: (53) where yi is the molar fraction of the fuel, i, in the fuel mixture and AOk is the corresponding combustion period or ignition lag for the engine when operating with the fuel component i on its own under the same conditions. A&, is the combustion period or ignition lag for the mixture. To estimate the corresponding operational limits of a mixture of fuels LeChateliers type rule was shown to be applicable [7,13]. A typical comparison between such estimated combustion duration and ignition lag with the corresponding experimentally derived data [7] for mixtures of hydrogen and methane in the CFR engine are shown in Figure 7. Both the combustion duration and ignition lag as expected were shortened as the amount of hydrogen in the fuel mixtures was increased because of the faster combustion characteristics of hydrogen in comparison to methane. Similarly, the estimated combustion durations and ignition lags for propanemethane and ethane-methane mixtures in Figure 8 are presented along with the corresponding experimental data. In these cases, the values of the combustion duration and ignition lag did not change substantially as 10

Figure 7: Predicted variations of the combustion duration (top) and the ignition lag (bottom) for two spark timings for mixtures of CH4 and H2 at 900 RPM, CR =8.5:1,4 = 1.0, P0 = 87 kPa, T0 = 300 K. The corresponding experimental values are shown. the amount of propane or ethane was changed in the fuel mixtures reflecting the similar combustion characteristics of these fuels with methane. Thus, the model developed to estimate the mean effective combustion duration and ignition lag for methane operation may be extended for applications involving other gaseous fuels and their mixtures. Due to the nature of the strong turbulent flow encountered within a running spark ignition engines and the associated cyclic variations, the real value of the combustion duration tends to be necessarily fluctuating and statistical in nature. Hence, the accurate prediction of the combustion duration in a theoretical model, which is not, only difficult but, also unavailable at present, can be shown as strictly unnecessary for predicting engine performance parameters with gaseous fuel operation. A Further Simplification to Calculating the Combustion Duration: Some further simplification for estimation of the combustion duration can be also used, especially in cases where much of the needed experimental information were not available, since small changes in the value of the combustion duration will

duration at other operating conditions.

Figure 9: Comparison of the calculated combustion duration for methane operation using the model and the simplified method suggested for various equivalence ratios at CR =8.5:1, 4 = 1.0, P0 = 87 kPa, To = 294 K and 8,t = 27.5 BTC. The corresponding experimental values were also shown. As a typical example, the crank angle at maximum pressure for the best power output with MBT of 27.5 degree BTC at the operating condition of 8.5:1 compression ratio, initial room temperature and atmospheric pressure in a spark ignition CFR engine with the fuel methane was found to be around 10 degrees ATC [30,33]. Then the calculation of the combustion duration using this simplified approach was carried out and the results were presented along with the corresponding experimental values [30,33] in Figure 9. In such calculations the ignition lag was not accounted for, since its value tends to be very small relative to the values of the corresponding combustion duration at these optimum operating conditions. As it can be seen that the difference between the combustion durations calculated from the model and those when using this simplified approach was small and appeared to be less than 3 % for the case considered. Hence, both methods produced good agreement with the corresponding experimental values indicating an error of less than 9 % for the full model and 13 % for the simplified approach except for the very lean and rich near limit operations. Moreover, even the substitution of the values of the operational limits of the engine (when such data were unavailable) with estimated values corresponding to the flammability limits of the same fuel at the same operating conditions produced calculated results of engine performance parameters that were satisfactory. 11

Figure 8: Comparison of the calculated combustion duration (top) and ignition lag (bottom) with experimental data for mixtures of ethane-methane or propane-methane at 900 RPM, CR =8.5:1, 4 = 1.0, P0 = 87 kPa, To = 300 K and eSt = 15 BTC for ethane-methane and est = 10 BTC for propane-methane mixtures. Our own experimental points are shown. not affect considerably the calculated values of the performance parameters of the engine. Moreover, the combustion duration in a spark ignition engine tends to be fluctuating cyclically and statistical in nature. For an example, the fastest burning rate or the minimum combustion duration in a spark ignition engine occurs around the stoichiometric equivalence ratio at MBT timing [5,7,30]. For these optimum operating conditions, generally the maximum pressure occurs just after the top dead centre. The position of the maximum pressure can also be viewed as the location where the pressure rise due to the combustion of the fuel and the pressure change due to the volume change are equal. Accordingly, this information can be used to calculate the combustion duration at this operating condition using the mass burning rate profile described earlier. This will be the minimum combustion duration, Aeolmin, at that operating condition which can be substituted to calculate subsequently the combustion

4 KNOCK MODELLING Knock may be considered to take place when the energy released due to the preignition reaction activity within the end gas becomes sufficiently significant and intense that results in autoignition of a portion of the mixture yet to be consumed by the propagating flame. The knock criterion (K) [7,12] used in the present model is based on the assumption that the calculated accumulated amount of energy release due solely to end gas preignition reactions activity per unit of the instantaneous cylinder volume, relative to the total energy released normally through flame propagation over the whole cycle per unit of cylinder swept volume exceeds for knock certain acceptable levels; i.e. Knock Criterion(K) = (54)

set of reaction equations:

where o;jf and 02ijb are stoichiometric coefficients of the ith species appearing in the reactants and products of the jth reaction , respectively. Xi is the formula for a species i taking part in the reaction. The net rate of production and consumption of each species i of a reactive end gas charge depends on the rates of all the reaction steps involved [15,34] in the detailed chemical kinetic scheme for the oxidation of fuel. Therefore, the change of chemical species concentration over a certain element of time can be calculated from the following set of equations: (59)

This is equivalent to the incremental pressure increase due solely to the preignition reaction activity of the current mass of the end gas per unit of mean effective combustion pressure. The energy released by the self reactions of the end gas is defined as: (55) where h is the specific enthalpy of the mixture and mu is the instantaneous end-gas mass determined from the two zone model. Subscripts tst and t indicate the values at the instant of spark discharge and at any time, respectively. The energy released due to combustion by normal flame propagation can be simplified as: (56) where h0 is the effective specific heating value of the charge and m0 is its initial mass. Consequently, Equation 54 may be simplified as: (57) In order to predict the energy release by the end gas reactions, the progress of any reaction activity needs to be followed using a detailed chemical kinetics model which in the current application is a formulation of 155 reaction steps and 39 species. A chemical reaction scheme involving i species and j reaction steps can be represented by the following 12

where ci is the concentration of ith species, p is the mixture density, Rjf and Rjb are the forward and backward reaction rates of the jth reaction, respectively:

(60) Kjf and Kjb are the forward and backward rate constants for the jth reaction and can be calculated according to the following:

(61)

Kjc is the corresponding equilibrium constant of the reaction step and Ajf, Ajb, Bjf and Bjb are constants for the forward and backward rate constants of the jth reaction equation. Ejf and Ejb are the activation energies of the forward and backward reactions of the jth reaction equation, respectively. Kjf and Kjb are related through the corresponding equilibrium constant, Kjc which is based on the concentrations for the jth, reaction equation. When the above set of simultaneous equations is solved numerically, the concentration of each species

in the unburned zone at any instant of time can be found. The physical properties of the species, the reaction scheme used and its corresponding constants for methane, ethane, propane and hydrogen are listed in references [7,9]. The knock modelling proceeds to monitor the reaction activity and the changes in the concentrations of each of the reacting species using the above group of equations right from the time of spark discharge, or even earlier when necessary. This is carried out for each time increment while accounting for the local temperature, pressure and volume of the end gas throughout, right up to the point of the total consumption of the unburned charge by the propagating flame at the end of the combustion period. When the end gas chemical reaction activity is sufficiently intense, the associated energy release may also cause some further temperature and pressure rises. Normally the effects of any reaction activity within the unburned gas on the cylinder pressure is relatively small compared with that due to piston motion and normal combustion when no autoignition takes place. The effect of the reaction activity on temperature is assumed to be uniform across the end gas zone. Since the rates of these reactions are very sensitive to temperature changes, their effect on the end gas temperature temporal development should be accounted for. However, these influences cannot be predicted well with the thermodynamic two-zone combustion model alone. Accordingly, a simple procedure is employed where at each time interval, the reaction rates of the end gas were calculated at the predicted Tub and P from the two-zones model. Then, the resulting temperature change dT'ub of the end gas zone due to these reactions over the short time interval was evaluated at constant cylinder pressure. If the resulting calculated temperature change dT'ub is detectably larger than the temperature change dTub, due to energy release from flame propagation, then it was used as the new dTub for the next interval of considering the chemical reaction rate instead of the value derived from the twozone combustion progression model. This way an account favouring earlier, and hence safer, prediction of autoignition in the end gas is made. Figure 10 shows a typical case where the variations of the calculated unburned gas mean cylinder charge temperature with crank angle for both methane operation when accounting for the reactions of the end gas region leading to knock and when only changes due to flame propagation are considered. With a significant end gas reaction activity, Tub increases very quickly, which speeds up the preignition reactions and causes 13

Figure 10: Variations of the calculated unburned temperature of the end gas versus crank angle for methane operation at 900 RPM, CR = 16, 4 = 0.88,8,t = 10 BTDC and To = 311 K at experimentally established knocking state for a CFR engine. autoignition with great intensity. When operating conditions are far from those producing knock, then the changes in the value of Tub due to reaction activity would have been insignificant and the unburned zone temperature would be essentially the same as that obtained with the two-zone progression of combustion model. The knock criterion of Equation 57 may be used to test for the onset of knock and may indicate its intensity. Since the spark timing is usually set so that much of the combustion process takes place around top dead centre so as to produce the largest power output from the engine, then V0/Vt may be set to approximately equal to (CR - 1) which simplifies Equation 57 further to: (62) Thus, it can be seen the instantaneous value of K at any instant of time, t, is dependent on the relative mass of the end gas, the relative energy release via preignition reactions and relative cylinder volume via the compression ratio. Hence, as an example for a 20 % relative energy release via preignition reactions and 25% end gas relative mass at 11:1 compression ratio, the value of the K is 0.50. Figures 11 and 12 show two typical variations with time of the end-gas relative energy release, dimensionless mass and the knock criterion K for an experimentally known knock free and borderline knock operation. At the beginning of the combustion process, since the reaction rates of the end gas are very slow,

Figure 11: Typical variations of K, mu/mo and (hSt ht)/h0, with crank angle for an effectively knock free operation with methane at 900 RPM, CR = 11, edt = 18 BTDC, 4 = 1.0, PO = 87 kPa, To = 308 K. The thick line shows the variations of K before autoignition of the end-gas.

the end gas at the time of autoignition and the instantaneous cylinder volume at the time. If autoignition takes place at virtually the end of the combustion duration, the amount of energy released will be much too small to produce a detectable knock. However, if under more severe operating conditions when a large portion of the charge participates in the autoignition well ahead of flame arrival, knock will be detectable and may be sufficiently severe to render such operation unacceptable (see Figures 11 and 12). Through a comparison with experimental observations at different operating conditions and compression ratios, it was found that whenever the value of the knock criterion during a cycle exceeded a certain threshold, then these conditions will be also associated with a knocking state. The bigger the difference between the maximum value of K in a cycle and the threshold value, the earlier is the autoignition and the greater is the intensity of knock, as shown typically in Figure 13. For knock free operation, the changing value of this criterion remains throughout below the threshold value while it would indicate sufficiently

Figure 12: Typical variations of K, mu/m0 and (hSt ht)/h0 with crank angle for borderline knock operation with methane at 900 RPM, CR = 11, B,t = 23 BTDC, 4 = 1.0, PO = 87 kPa, To = 308 K. The thick line shows the variations of K before autoignition of the end-gas. K values will be small. As the combustion process progresses, the end-gas temperature and pressure are increased, causing an increase in the reactivity of the end gas. Therefore, the end gas energy release may be so intense that will lead to autoignition. On the other hand with time, flame propagation decreases the ratio of mu/mo in consequence of the rate function of Equation 32. As a result, the K value may increase to a maximum and then drops to zero as the end gas becomes fully consumed by the flame propagation. A very rapid rise in the value of K to relatively high values is associated with the onset of autoignition in the mixture. For any compression ratio, the maximum tolerated value of K would depend primarily on the amount of self reacted energy released, moderated by the size of 14 Figure 13: Variations of the knock criterion K with crank angle for methane operation at 900 RPM, CR = 11, q5 = 1.0, To = 308 K, and PO = 87 kPa at three experimentally conditions showing no knock, borderline knock and strong knock operations. The thick lines indicate the variations before autoignition of the end-gas. well in general for practical purposes through its value whether knock is encountered and its relative intensity for any set of operating conditions. The model when applied to a wide range of operating conditions that were known to produce experimentally light knock in a CFR engine in our laboratory confirmed the presence of an approximately constant threshold value for K at borderline knock as shown typically in Figure 14. The maximum values of the knock criterion then was found to be approximately constant at 1.50 for these borderline knock oper-

Figure 14: Variations of the calculated K with crank angle for some operating conditions associated with knock limit operations in a CFR engine using methane-hydrogen mixtures at 900 RPM, 4 = 1.0, P0 = 87 kPa, To = 305 K. The thick lines show the variations of K before auto-ignition of the end-gas. ations [7,12]. A slightly more intense knock increased the value rapidly and significantly. It was confirmed further that the conditions to produce the knock limit may be determined while employing the model by assigning a value of around 1.5 for the maximum value of the criterion. 5 SOME RESULTS AND DISCUSSION The model was shown to be capable of predicting some of the mean engine performance parameters such as the temporal variations in cylinder pressure and the two zone mean temperatures for different speeds, intake temperature, intake pressure, spark timing and fuel. The effects of changes in some design parameters such as bore diameter and compression ratio were also considered [7,12]. A typical calculated temporal pressure variation while operating on methane in CFR engine is shown in Figure 15 with our corresponding experimental data. Similarly, Figure 16 shows the changes in the calculated indicated power output versus equivalence ratios for two different compression ratios along with the corresponding experimental values. Very good agreement can be seen between the predicted and the corresponding experimental values. Further typical comparison involving fuel mixtures of methane-hydrogen is shown in Figure 17. It can be seen that the presence of very high concentrations of hydrogen in the fuel mixtures improved slightly 15

Figure 15: A typical variations of pressure-time diagram in the cylinder in CFR engine while operating on methane at equivalence ratio of 0.99, compression ratio of 8.5:1, spark timing of 20 Degr. BTC and initial temperature of 294 K.

Figure 16: A typical variations of indicated power output with equivalence ratio for different compression ratios while operating on methane at 900 rpm, spark timing of 15 Degr. BTC and initial temperature of 294 K. Experimental points are also shown. the indicated power output mainly for very lean and rich mixtures reflecting the faster combustion of hydrogen in comparison to methane. The overall effect of the high concentration of hydrogen in methane at other equivalence ratios tends to be affected somewhat adversely by the lower heating value of hydrogen on molar basis in comparison to that of methane [31] and the need to optimize spark timing with changes in both equivalence ratio and hydrogen concentration. Similarly, Figure 18 shows typically a comparison of results obtained for methane-propane and methaneethane mixtures, It can be seen that the increasing addition of ethane or propane to methane resulted in only relatively small improvement in the indicated power output at constant spark timing. This was expected because of both ethane and propane provide somewhat faster combustion rates and higher heat-

Figure 17: The indicated power output variations with equivalence ratios when operating on methane-hydrogen mixtures for a compression ratio of 8.5:1, spark timing of 20 deg. BTC and initial mixture temperature of 298K at 900 rpm. The corresponding experimental data are also shown. ing values on volume basis than those for methane. Moreover, the addition of propane tends to produce slightly higher output than the corresponding values for ethane addition for the same mixtures and operating conditions, since propane tends to have slightly faster combustion rates and has a higher heating value on volume basis than ethane. These trends are also consistent with those reported experimentally by others(e.g. [30]). It is also evident that the agreement between the results of the model and the corresponding experimental data tends to be for practical purposes very satisfactory. 6 CONCLUSION The simulation model described when combined with detailed account of the chemical kinetics of the mixture yet to be burned, can provide a satisfactory description of the main features of the combustion process in a spark ignition engine including accounting for the onset of knock. This predictive model which requires relatively very little empirical input was shown to be capable of generating important performance parameters, such as the temporal variations of cylinder pressure, the two zone mean temperatures, power output and efficiency for different values of intake temperature, intake pressure, spark timing and speed as well as the effects of changes in some design parameters such as compression ratio, bore diameter for a spark ignition engine while operating on common gaseous fuels and their mixtures. The predicted values showed good agreement with the corresponding experimental data. The model can also be used to test for the occur16

Figure 18: The indicated power output variations when operating on methane-ethane and methane-propane mixtures for a compression ratio of 8.5:1, equivalence ratio of 1.0, spark timing of 15 deg. BTC for methane-ethane mixtures and 10 deg. BTC for methane-propane mixtures and initial mixture temperature of 298K at 900 rpm. The corresponding experimental data are also shown. rence of knock for any set of operating conditions. When knocking is encountered the value of the knock criterion, K, is shown to build up to a sufficiently high value that exceeds an acceptable limit while under normal operating conditions, its value remains comparatively low throughout. For the CFR engine tested, the knock free limit operation was found to be associated with a maximum value of K that is less than 1.5. The corresponding calculated value can provide an indication of the intensity of the resulting knock when occurs. The larger the value of K, the bigger the potential intensity of knock. 7 ACKNOWLEDGEMENTS The financial assistance of NSERC and province of Alberta Graduate Fellowship is gratefully acknowledged. The contribution to this work of Drs. T. J. AlHimyary, J. Gao and A. A. Attar is appreciated. References [1] Al-Himyary T.J. A Diagnostic Two-Zone Combustion Model for Spark Ignition Engines Based on Pressure Time Data. PhD thesis, Department of Mechanical Engineering, University of Calgary, 1988. [2] Karim G.A., Gao J., and Attar A. A Predictive Approach to Spark Ignition Engine Performance Fueled with Common Gaseous Fuels and Their Mixtures. In th Proceeding of the 17 Annual Fall Technical Conference of the ASME Internal Combustion Engine Division, volume 25-2, pages 59-64, September 1995.

[3] Karim G.A. and Gao J. A Predictive Model for Knock in Gas Fueled Spark Ignition Engine. In Proceeding of the International Congress on Computational Methods in Engineering (ICCME), volume 2, pages 429-437, May, Iran 1993. [4] R. R. Maly. State of the Art and Future Needs in S. I. Engine Combustion. In Twenty-Fifth Symposium (International) on Combustion, pages 111-124, 1994. [5] Heywood J.B. Internal Combustion Engine Fundamentals: McGraw Hill Book Co., 1988. [6] Alizadeh Attar A. and Karim G.A. An Analytical Approach for the Optimization of a Gas Fueled SI Engine Performance Including the Consideration of Knock. In Proc. of ASME, International Combustion Engine Division, volume 28-2, pages 65-71, April 1997. [7] Attar A. A. Optimization and Knock Modelling of a Gas Fueled Spark Ignition Engine. PhD thesis, Department of Mechanical Engineering, University of Calgary, 1997. [8] Zhou G. Analytical Studies of Methane Combustion and the Production of Hydrogen and/or Synthesis Gas by the Uncatalyzed Partial Oxidation of Methane. PhD thesis, Department of Mechanical Engineering, University of Calgary, 1993. [9] Liu Z. Combustion in Gas Fueled C.I. Engines. PhD thesis, Department of Mechanical Engineering, University of Calgary, 1995. [10] JANAF. Thermochemical Tables. National Bureau of Standards, Washington D.C., USA, 1985. [11] Karim G.A. and Gao J. A Predictive Model for Knock in Spark Ignition Engines. SAE, (922366), 1992. [12] Gao J. Knock Modeling in S.I. Engines. PhD thesis, Department of Mechanical Engineering, University of Calgary, 1993. [13] Strehlow R. Fundamentals of combustion. McGraw Hill Co., 1988. [14] Woschni G. A Universally Applicable Equation for the Instantaneous Heat Transfer Coefficient in the Internal Combustion Engines. SAE Trans., 76(670931):30653083, 1967. [15] Westbrook C.K. and Pitz W.J. Complex Chemical Reaction Systems Mathematical Modeling and Simulation. J. Warnatz and W. Jager(eds.) Springer-Verlag, Heidelberg, West Germany, 1986. [16] Bade Shrestha S.O. A Predictive Model for Gas Fueled Spark Ignition Engine Applications, PhD thesis, Department of Mechanical Engineering, University of Calgary, 1999. [17] Badr O., Elsayed N., and Karim G.A. An Investigation of the Lean Operational Limits of Gas Fueled Spark Ignition Engines. Transactions of the ASME, J. of Energy Resources Technology, 118:159-163, 1996. 17

[18] Karim G.A and Wierzba I. Experimental and Analytical Studies of the Lean Operational Limits in Methane Fueled Spark Ignition and Compression Ignition Engines. SAE, (891637), 1989. [19] Karim G.A. and Gao J. Prediction of the Performance of Spark Ignition Engine Including Knock. SAE, (932823), 1993. [20] Karim G.A. and Klat S.R. Knock and Autoignition Characteristics of Some Gaseous Fuels and Their Mixtures. J. of the Institute of Fuel, 39:109 -119, March 1966. [21] Zabetakis M. Flammability Characteristics of Combustible Gases and Vapours. In US Bureau of Mines, Dept. of Interior, Bulletin 627, 1965. [22] Al-Himyary T.J. Research Report. The University of Calgary, 1988. [23] Benson R.S., Annand W.J.D., and Baruah P.C. A Simulation Model Including Intake and Exhaust Systems for a Single Cylinder Four Stroke Spark Ignition Engine. Int. J. Mech. Sci., 17:97-124, 1975. [24] Karim G.A. and Al-Himyary T.J. A Diagnostic TwoZone Combustion Model for Spark Ignition Engines Based on Pressure-Time Data. SAE, (880199), 1988. [25] Hirst S.L. and Kirsch L.J. The Application of a Bydrocarbon Autoignition Model in Simulating Knock and other Engine Combustion Phenomena. Plenum Press, 1980. [26] Hires S.D., Tabaczyinski R-J., and Novak J.M. The Prediction of Ignition Delay and Combustion Intervals for a Homogeneous Charge Spark Ignition Engine. SAE, (780232), 1978. [27] Hong C.W. A Combustion Correlation for Spark Ignition Engines Simulation Under Steady and Transient Conditions. SAE, (901602), 1990. [28] Dimpelfeld P.M. and Foster D.E. The Prediction of Autoignition in a Spark Ignition Engine. SAE, (841337), 1984. [29] Pischinger S. and Heywood J. B. A Model for Flame Kernel Development in a Spark Ignition Engine. In Twenty - Third Symposium (International) on Combustion, page 1033, 1990. [30] Al-Alousi Y. Examination of the Combustion Processes and the Performance of a Spark Ignition Engine, Using a Data-Acquisition System PhD thesis, Department of Mechanical Engineering, University of Calgary, 1982. [31] Bade Shrestha S. O. and Karim G. A. Hydrogen as an Additive to Methane for Spark Ignition Engine Applications. In The Thirty Second Intersociety Energy Convention Engineering Conference Proceedings, volume 2, pages 910-915, 1997.

[32] Kido H., Huang S., Tanoue K., and Nitta T. Improving the Combustion Performance of Lean Hydrocarbon Mixtures by Hydrogen Addition. SAE of Japan, (9430653), 1994. [33] Karim G.A. and Al-Alousi Y.H. Some Considerations of Cyclic Variations in Spark Ignition Engines Fueled with Gaseous Fuels. SAE, (840232), 1984. [34] Karim G.A. and Zhou G. An Analytical Examination of Various Criteria for Defining Autoignition within Heated Methane-Air Mixtures. In Transactions of the ASME, J. of Energy Resources Technology, volume 116-3, pages 175-180, September 1994.

18

S-ar putea să vă placă și