Patent application title: OPTIMIZED FUEL MANAGEMENT SYSTEM FOR DIRECT INJECTION ETHANOL ENHANCEMENT OF GASOLINE ENGINES
Leslie Bromberg (Sharon, MA, US)
Daniel R. Cohn (Cambridge, MA, US)
John B. Heywood (Newtonville, MA, US)
John B. Heywood (Newtonville, MA, US)
Massachusetts Institute of Technology
IPC8 Class: AF02D1908FI
Class name: Internal-combustion engines charge forming device (e.g., pollution control) diverse fuel supply
Publication date: 2012-11-15
Patent application number: 20120285429
Fuel management system for enhanced operation of a spark ignition
gasoline engine. Injectors inject an anti-knock agent such as ethanol
directly into a cylinder. It is preferred that the direct injection occur
after the inlet valve is closed. It is also preferred that stoichiometric
operation with a three way catalyst be used to minimize emissions. In
addition, it is also preferred that the anti-knock agents have a heat of
vaporization per unit of combustion energy that is at least three times
that of gasoline.
6. A spark ignition engine where two fuels are introduced into the engine where the first fuel has a greater alcohol concentration than the second fuel; and where the first fuel is produced by onboard separation of fuel from a third fuel which is a gasoline-alcohol mixture; and where the ratio of the first fuel that is used in the engine to the second fuel that is used in the engine increases with increasing torque.
7. The spark ignition engine of claim 6 where a membrane is employed in the onboard separation.
8. The engine system of claim 7 where the membrane is an inorganic membrane.
9. The engine system of claim 7 where the membrane is an organic membrane.
10. The engine system of claim 7 where the membrane is a transfusion membrane.
11. The engine system of claim 7 where the membrane is a porous membrane.
12. The engine system of claims 6 or 7 where the alcohol in the gasoline-alcohol mixture is ethanol.
13. The engine system of claims 6 or 7 where the first fuel is stored in a tank to which additional alcohol can be added.
14. The engine system of claims 6 or 7 where ratio of the first fuel to the second fuel is varied so as to prevent knock at the torque is increased.
15. The spark ignition engine of claims 6 or 7 where the engine is operated with a substantially stoichiometric fuel to air ratio.
16. The spark ignition engine of claims 6 or 7 where a closed loop control system with a knock detector is used to determine the ratio of amount of the first fuel to second fuel used in the engine.
17. The spark ignition engine of claims 6 or 7 where an open loop control system with a look up table is used to determine the ratio of the first fuel to the second fuel.
18. The spark ignition engine of claims 6 or 7 where a control system is used to minimize the amount of second fuel that is used to prevent knock.
19. The spark ignition engine of claims 6 or 7 where the first fuel is injected so as to increase its vaporization cooling of the fuel-air mixture in at least one engine cylinder.
20. The spark ignition engine of claim 19 where the first fuel is directly injected into the engine.
21. The spark ignition engine of claims 6 or 7 where the second fuel contains gasoline.
22. The spark ignition engine of claim 7 where the second fuel is produced by onboard separation from the third fuel.
23. A spark ignition engine where two fuels are introduced into the engine where the first foci has a greater knock resistance than the second fuel and where the first fuel is introduced into the engine so as to produce a non-uniform distribution of the first fuel with a greater concentration in the end gas region; and where the first and second fuel are produced by on board separation of a third fuel.
24. The spark ignition engine of claim 23 where a membrane is employed in the onboard separation.
25. The engine system of claim 23 where the third fuel is a gasoline-alcohol mixture and the first fuel has a greater alcohol concentration than the second fuel.
26. The engine system of claim 23 where ratio of the amount of first fuel used in the engine to the amount of the second fuel used in the engine is increased so as to prevent knock as the torque is increased.
27. The spark ignition engine of claim 23 where a closed loop control system with a knock detector is used to determine the ratio of amount of the first fuel to second fuel used in the engine.
28. The spark ignition engine of claim 23 where the first fuel is injected so as to increase its vaporization cooling of the fuel-air mixture in at least one engine cylinder.
29. The spark ignition engine of claim 23 where the first fuel is directly injected into the engine.
30. The spark ignition engine of claim 23 where the first fuel is introduced into the engine so as to have a greater concentration in the periphery of the engine cylinders.
 This application is a continuation of U.S. patent application Ser.
No. 11/100,026 filed Apr. 6, 2005, which is a continuation-in-part of
U.S. patent application Ser. No. 10/991,774 filed Nov. 18, 2004, the
contents of both of which are incorporated herein by reference.
BACKGROUND OF THE INVENTION
 This invention relates to an optimized fuel management system for use with spark ignition gasoline engines in which an anti-knock agent which is a fuel is directly injected into a cylinder of the engine.
 There are a number of important additional approaches for optimizing direct injection ethanol enhanced knock suppression so as to maximize the increase in engine efficiency and to minimize emissions of air pollutants beyond the technology disclosed in parent application Ser. No. 10/991,774 set out above. There are also additional approaches to protect the engine and exhaust system during high load operation by ethanol rich operation; and to minimize cost, ethanol fuel use and ethanol fuel storage requirements. This disclosure describes these approaches.
 These approaches are based in part on more refined calculations of the effects of variable ethanol octane enhancement using a new computer model that we have developed. The model determines the effect of direct injection of ethanol on the occurrence of knock for different times of injection and mixtures with port fuel injected gasoline. It determines the beneficial effect of evaporative cooling of the direct ethanol injection upon knock suppression.
SUMMARY OF THE INVENTION
 In one aspect, the invention is a fuel management system fear operation of a spark ignition gasoline engine including a gasoline engine and a source of an anti-knock agent which is a fuel. The use of the anti-knock agent provides gasoline savings both by facilitating increased engine efficiency over a drive cycle and by substitution for gasoline as a fuel. An injector is provided for direct injection of the anti-knock agent into a cylinder of the engine and a fuel management control system controls injection of the anti-knock agent into the cylinder to control knock. The injection of the antiknock agent can be initiated by a signal from a knock sensor. It can also be initiated when the engine torque is above a selected value or fraction of the maximum torque where the value or fraction of the maximum torque is a function of the engine speed. In a preferred embodiment, the injector injects the anti-knock agent after inlet valve/valves are closed. It is preferred that the anti-knock agent have a heat of vaporization that is at least twice that of gasoline or a heat of vaporization per unit of combustion energy that is at least three times that of gasoline. A preferred anti-knock agent is ethanol. In a preferred embodiment of this aspect of the invention, part of the fuel is port injected and the port injected fuel is gasoline. The directly injected ethanol can be mixed with gasoline or with methanol. It is also preferred that the engine be capable of operating at a manifold pressure at least twice that pressure at which knock would occur if the engine were to be operated with naturally aspirated gasoline. A suitable maximum ethanol fraction during a drive cycle when knock suppression is desired is between 30% and 100% by energy, it is also preferred that the compression ratio be at least 10. With the higher manifold pressure, the engine can be downsized by a factor of two and the efficiency under driving conditions increased by 30%.
 It is preferred that the engine is operated at a substantially stoichiometric air/fuel ratio during part or all of the time that the anti-knock agent such as ethanol is injected. In this case, a three-way catalyst can be used to reduce the exhaust emissions from the engine. The fuel management system may operate in open or closed loop modes.
 In some embodiments, non-uniform ethanol injection is employed. Ethanol injection may be delayed relative to bottom dead center when non-uniform ethanol distribution is desired.
 Many other embodiments of the invention are set forth in detail in the remainder of this application.
BRIEF DESCRIPTION OF THE DRAWINGS
 FIG. 1 is a graph of ethanol fraction (by energy) required to avoid knock as a function of inlet manifold pressure. The ethanol fraction is shown for various values of β, the ratio of the change in temperature in the air cylinder charge due to turbocharging (and aftercooling if used) to the adiabatic temperature increase of the air due to the turbocharger.
 FIG. 2a is a graph of cylinder pressure as a function of crank angle for a three bar manifold pressure.
 FIG. 2b is a graph of charge temperature as a function of crank angle for a three bar manifold pressure.
 FIG. 3 is a schematic diagram of an embodiment of the fuel management system disclosed herein for maintaining stoichiometric conditions with metering/control of ethanol, gasoline, and air flows into an engine.
 FIGS. 4a and 4b are schematic illustrations relating to the separation of ethanol from ethanol/gasoline blends.
 FIG. 5 is a cross-sectional view of a flexible fuel tank for a vehicle using ethanol boosting of a gasoline engine.
DESCRIPTION OF THE PREFERRED EMBODIMENT
 Ethanol has a heat of vaporization that is more than twice that of gasoline, a heat of combustion per kg which is about 60% of that of gasoline, and a heat of vaporization per unit of combustion energy that is close to four times that of gasoline. Thus the evaporative cooling of the cylinder air/fuel charge can be very large with appropriate direct injection of this antiknock agent. The computer model referenced below shows that evaporative cooling can have a very beneficial effect on knock suppression. It indicates that the beneficial effect can be maximized by injection of the ethanol after the inlet valve that admits the air and gasoline into the cylinder is closed. This late injection of the ethanol enables significantly higher pressure operation without knock and thus higher efficiency engine operation than would be the case with early injection. It is thus preferred to the conventional approach of early injection which is used because it provides good mixing. The model also provides information that can be used for open loop (i.e., a control system that uses predetermined information rather than feedback) fuel management control algorithms.
 The increase in gasoline engine efficiency that can be obtained from direct injection of ethanol is maximized by having the capability for highest possible knock suppression enhancement. This capability allows the highest possible amount of torque when needed and thereby facilitates the largest engine downsizing for a given compression ratio.
 Maximum knock suppression is obtained with 100% or close to 100% use of direct injection of ethanol. A small amount of port injection of gasoline may be useful in order to obtain combustion stability by providing a more homogeneous mixture. Port fuel injection of gasoline also removes the need for a second direct fuel system or a more complicated system which uses one set of injectors for both fuels. This can be useful in minimizing costs.
 The maximum fraction of ethanol used during a drive cycle will depend upon the engine system design and the desired level of maximum torque at different engine speeds. A representative range for the maximum ethanol fraction by energy is between 20% and 100%.
 In order to obtain the highest possible octane enhancement while still maintaining combustion stability, it may be useful for 100% of the fuel to come from ethanol with a fraction being port injected, as an alternative to a small fraction of the port-fueled gasoline.
 The initial determination of the knock suppression by direct injection of ethanol into a gasoline engine has been refined by the development of a computer model for the onset of knock under various conditions. The computer modeling provides more accurate information for use in fuel management control. It also shows the potential for larger octane enhancements than our earlier projections. Larger octane enhancements can increase the efficiency gain through greater downsizing and higher compression ratio operation. They can also reduce the amount of ethanol use for a given efficiency increase.
 The computer model combines physical models of the ethanol vaporization effects and the effects of piston motion of the ethanol/gasoline/air mixtures with a state of the art calculational code for combustion kinetics. The calculational code for combustion kinetics was the engine module in the CHEMKIN 4.0 code [R. J. Kee, F. M. Rupley, J. A. Miller, M. E. Coltrin, J. F. Grear, E. Meeks, H. K. Moffat, A. E. Lutz, G. Dixon-Lewis, M. D. Smooke. J. Warnatz, G. H. Evans, R. S. Larson, R. E. Mitchell, L. R. Petzold, W. C. Reynolds, M. Caracotsios, W. E. Stewart, P. Glarborg, C. Wang, O. Adigun, W. G. Houf, C. P. Chou, S. F. Miller, P. Ho, and D. J. Young, CHEMKIN Release 4.0, Reaction Design, Inc., San Diego, Calif. (2004)]. The CHEMKIN code is a software tool for solving complex chemical kinetics problems. This new model uses chemical rates information based upon the Primary Reference gasoline Fuel (PRF) mechanism from Curran et al. [Curran, H. J., Gaffuri, P., Pitz, W. J., and Westbrook, C. K. "A Comprehensive Modeling Study of iso-Octane Oxidation," Combustion and Flame 129:253-280 (2002) to represent onset of autoignition.
 The compression on the fuel/air mixture end-gas was modeled using the artifact of an engine compression ratio of 21 to represent the conditions of the end gas in an engine with an actual compression ratio of 10. The end gas is defined as the un-combusted air/fuel mixture remaining after 75% (by mass) of the fuel has combusted. It is the end gas that is most prone to autoignition (knock). The larger compression ratio includes the effect of the increase in pressure in the cylinder due to the energy released in the combustion of 75% of the fuel that is not in the end gas region. The effect of direct ethanol vaporization on temperature was modeled by consideration of the effects of the latent heat of vaporization on temperature depending upon the time of the injection.
 The effect of temperature increase due to turbocharging was also included. The increase in temperature with turbocharging was calculated using an adiabatic compression model of air. It is assumed that thermal transfer in the piping or in an intercooler results in a smaller temperature increase. The effect is modeled by assuming that the increase in temperature of the air charge into the cylinder ΔTcharge is ΔTcharge=βΔTturbo were ΔTturbo is the temperature increase after the compressor due to boosting and beta is a constant. Values of β of 0.3, 0.4 and 0.6 have been used in the modeling. It is assumed that the temperature of the charge would be 380 K for a naturally aspirated engine with port fuel injection gasoline.
 FIG. 1 shows the predictions of the above-referenced computer model for the minimum ethanol fraction required to prevent knock as a function of the pressure in the inlet manifold, for various values of β. In FIG. 1 it is assumed that the direct injection of the ethanol is late (i.e. after the inlet valve that admits air and gasoline to the cylinder is closed) and a 87 octane PRF (Primary Reference Fuel) to represent regular gasoline. The corresponding calculations for the manifold temperature are shown in Table 1 for the case of a pressure in the inlet manifold of up to 3 bar for an engine with a conventional compression ratio of 10. The temperature of the charge varies with the amount of ethanol directly injected and is self-consistently calculated in Table 1 and FIG. 1. The engine speed used in these calculations is 1000 rpm.
TABLE-US-00001 TABLE 1 Computer model calculations of temperature and ethanol fraction required for knock prevention for an inlet manifold pressure of 3 bar for an engine with a compression ratio of 10, for various values of β (ratio of change of the cylinder air charge temperature due to turbocharging to the adiabatic temperature increase due to turbocharging ΔTcharge = βΔturbo). The engine speed is 1000 rpm. β 0.3 0.4 0.6 T_charge init K 380 380 380 Delta T turbo K 180 180 180 Delta T after intercooler K 54 72 108 Delta T due to DI ethanol and gasoline K -103 -111 -132 T_init equivalent charge K 331 341 356 Gasoline octane 87 87 87 Ethanol fraction (by energy) needed to prevent knock 74% 82% 97%
 Direct fuel injection is normally performed early, before the inlet valve is closed in order to obtain good mixing of the fuel and air. However, our computer calculations indicate a substantial benefit from injection after the inlet valve is closed.
 The amount of air is constant in the case of injection after the inlet valve has closed. Therefore the temperature change is calculated using the heat capacity of air at constant volume (cv). The case of early injection where the valve that admits air and fuel to the cylinder is still open is modeled with a constant-pressure heat capacity (cp). The constant volume case results in a larger evaporation induced decrease in charge temperature than in the case for constant pressure, by approximately 30%. The better evaporative cooling can allow operation at higher manifold pressure (corresponding to a greater octane enhancement) without knock that would be the case of early injection by a difference of more than 1 bar. The increase in the evaporative cooling effect at constant volume relative to that at constant pressure is substantially higher for the case of direct injection of fuels such as ethanol and methanol than is the case for direct injection of gasoline.
 Typical results from the calculations are shown in FIG. 2. The figure shows the pressure (a) and the temperature (b) of the cylinder charge as a function of crank angle, for a manifold pressure of 3 bar and a value of β=0.4 Two values of the ethanol fraction are chosen, one that results in autoignition, and produces engine knock (0.82 ethanol fraction by fuel energy), and the other one without autoignition, i.e., no knock (0.83 ethanol fraction). Autoignition is a threshold phenomenon, and in this case occurs between ethanol fractions of 0.82 and 0.83. For an ethanol energy fraction of 0.83, the pressure and temperature rise at 360° (top dead center) is due largely to the compression of the air fuel mixture by the piston. When the ethanol energy fraction is reduced to 0.82, the temperature and pressure spikes as a result of autoignition. Although the autoignition in FIG. 2 occurs substantially after 360 degrees, the autoignition timing is very sensitive to the autoignition temperature (5 crank angle degrees change in autoignition timing for a change in the initial temperature of 1 K, or a change in the ethanol energy fraction of 1%).
 The effect of evaporative cooling from the antiknock agent (in this case, ethanol) is shown in Table 2, where three cases are compared. The first one is with port fuel injection of ethanol, in this case the vaporization of the ethanol on the walls of the manifold has a negligible impact on the temperature of the charge to the cylinder because the walls of the manifold are cooled rather than the air charge. The second case assumes direct injection, but with the inlet valve open, with evaporation at constant pressure, where the cooling of the charge admits additional air to the cylinder. The third case assumes, as in the previous discussions, late injection after the inlet valve has closed. It is assumed stoichiometric operation, that the baseline temperature is 380 K, and that there is cooling in the manifold after the turbocharger with β=0.4.
TABLE-US-00002 TABLE 2 Knock-free operation of ethanol port fuel injection (assuming no charge cooling), and of direct injection before and after the inlet valve is closed. Compression ratio of 10, baseline charge temperature of 380K, intercooler/cooling post turbo with β = -0.4, stoichiometric operation, gasoline with 87 RON. Engine speed is 1000 rpm. No Evaporative cooling Evaporative Before After Cooling Valve Closing Valve Closing Ethanol fraction 0.95 0.95 0.95 (by energy) Max manifold pressure (bar) 1.05 2.4 4.0 Cylinder pressure after 1.05 2.4 3.0 cooling (bar) Cylinder charge temperature 383 360 355 after cooling (K)
 The results indicate the strong effect of the cooling. The maximum manifold pressure that prevents knock (without spark retard), with 0.95 ethanol fraction by energy in the case of port fuel injection is 1.05 bar. With direct injection of the ethanol, the maximum knock-free manifold and cylinder pressures are 2.4 bar, with a temperature decrease of the charge of ˜75K. The final case, with injection after inlet valve closing, allows a manifold pressure of 4 bar, a cylinder pressure (after cooling) of 3 bar, and a charge temperature decrease of ˜120 K. It should be noted that the torque of the late injection case after the valve has closed is actually higher than that of the early injection case, even though the early injection case allows for additional air (at constant pressure). For comparison, the model is also used to calculate the manifold pressure at which knock would occur for port fuel injection of 87 octane gasoline alone. This pressure is ˜0.8 bar assuming spark timing at MBT (Maximum Brake Torque). Conventional gasoline engines operate at 1 bar by retarding the timing at high torque regions where knock would otherwise occur. Thus the model indicates that evaporative cooling effect of direct injection of ethanol after the inlet has closed can be significantly greater than that the higher octane number rating of ethanol relative to gasoline.
 A manifold pressure of 4 bar is very aggressive. Table 2 is indicative of the dramatically improved performance of the system with direct injection after the inlet valve has closed. The improved performance in this case can be traded for increased compression ratio or reduced use of the anti-knock agent.
 It should be noted that, as mentioned above, the calculations of autoignition (knock) are conservative, as autoignition for the case shown in FIG. 2 occurs relatively late in the cycle, and it is possible that the fuel has been combusted before it autoignites. Also it should be noted that the calculations in FIG. 2 break down after autoignition, as the pressure trace would be different from that assumed. Figures similar to FIG. 2 are used to determine conditions where autoignition would not occur, and those conditions are then used to provide the information for FIG. 1. The initial temperatures of the cases shown in FIG. 2 are 341 K for 0.82 ethanol fraction, and 340 K for 0.83 ethanol fraction, a difference of 1 K (the difference due to the cooling effect of the ethanol).
 Because of the large heat of vaporization, there could be enough charge cooling with early injection so that the rate of vaporization of ethanol is substantially decreased. By instead injecting into the hot gases, which is the case with injection after the inlet valve has closed, the temperature at the end of full vaporization of the ethanol is substantially increased with respect to early injection, increasing the evaporation rate and minimizing wall wetting.
 The optimum timing of the injection for best mixing and a near homogeneous charge is soon after the inlet valve closes, provided that the charge is sufficiently warm for antiknock agent vaporization. If, on the other hand, a non-uniform mixture is desired in order to minimize ethanol requirements and improve ignition stability, then the injection should occur later than in the case where the best achievable mixing is the goal.
 Late injection of the ethanol after the inlet valve has closed can be optimized through the use of diesel-like injection schemes, such as injectors with multiple sprays. It is important to inject the fuel relatively quickly, and at velocities which minimize any cylinder wall wetting, which as described below could result in the removal of the lubrication oils from the cylinder liner. Multiple sprays from a nozzle that has multiple holes results in a distributed pattern of sprays, with relatively low injection velocities. This is particularly important thr ethanol, because of the higher volume throughputs (as compared with gasoline) of ethanol for equal energy content.
 Injection after the valve has closed may require that a modest fraction of the fuel 25%) be port injected in order to achieve the desired combustion stability. A tumble-like or swirl motion can be introduced to achieve the desired combustion stability. The port injected fuel can be either gasoline or ethanol.
 Use of the computer model for operation with gasoline alone gives results that are consistent with the observed occurrence of knock in gasoline engine vehicles, thereby buttressing the credibility of the projections for ethanol. The computer model indicates that for knock-free gasoline operation alone with a compression ratio of 10, knock imposes a severe constraint upon the allowed manifold pressure for a naturally aspirated gasoline engine and very limited (i.e., less than 1.2 bar) manifold pressure can be achieved even with direct injection of gasoline unless spark retard and/or rich operation is used. These changes, however, can reduce efficiency and increase emissions.
 FIG. 1 shows that knock can be prevented at manifold pressures greater than 2 bar with direct injection of an ethanol fraction of between 40 and 80% in an engine with a compression ratio of 10. The manifold pressure can be at least 2.5 bar without engine knock. A pressure of 3 bar would allow the engine to be downsized to ˜1/3 of the naturally aspirated gasoline engine, while still producing the same maximum torque and power. The large boosting indicated by the calculations above may require a multiple-stage turbocharger. In addition to a multiple stage turbocharger, the turbocharger may be of the twin-scroll turbo type to optimize the turbocharging and decrease the pressure fluctuations in the inlet manifold generated by a small number of cylinders.
 With an increase in allowed manifold pressure in an engine by more than a factor of the engine could be downsized by a factor of 2 (that is, the cylinder volume is decreased by a factor of 2 or more) and the compression ratio could be held constant or raised. For example, the performance of an eight cylinder engine is achieved by a four cylinder engine.
 The occurrence of knock at a given value of torque depends upon engine speed. In addition to providing substantially more maximum torque and power, direct injection of ethanol can be used to provide a significant improvement in torque at low engine speeds (less than 1500 rpm) by decreasing or eliminating the spark retard. Spark retard is generally used with gasoline engines to prevent knock at low engine speeds where autoignition occurs at lower values of torque than is the case at high engine speeds.
 FIG. 1 can also be used to determine the ethanol fraction required to prevent knock at different levels of torque and horsepower, which scale with manifold pressure in a given size engine. This information can be used in an open loop control system.
 The efficiency of a gasoline engine under driving conditions using direct ethanol injection enhancement can be at least 20% and preferably at least 30% greater than that of a naturally aspirated gasoline engine with a compression ratio of 10. This increase results from the substantial engine boosting and downsizing to give the same power, and also the high compression ratio operation (compression ratio of 11 or greater) that is enabled by a large octane enhancement. With more aggressive downsizing of more than 50% (where the same engine performance is obtained with less than one-half the displacement), the increase in efficiency could exceed 30%.
 Greater downsizing and higher efficiency may also be obtained by decreasing the octane requirement of the engine by using variable valve timing (VVT). Thus, at conditions of high torque, variable valve timing can be used to decrease the compression ratio by appropriately changing the opening/closing of the inlet and exhaust valves. The loss in efficiency at high torque has a small impact on the overall fuel economy because the engine seldom operates in these conditions.
 VVT can also be used to better scavenge the exhaust gases [B. Lecointe and G. Monnier, "Downsizing a Gasoline Engine Using Turbocharging with Direct Injection" SAE paper 2003-01-0542]. Decreasing the exhaust gas decreases the air/fuel temperature. Keeping both the inlet and exhaust valves open, while the pressure in the inlet manifold is higher than in the exhaust, can be used to remove the exhaust gases from the combustion chamber. This effect, coupled with slightly rich operation in-cylinder, can result in increased knock avoidance while the exhaust is still stoichiometric. Cooled EGR and spark timing adjustment can also be used to increase knock avoidance.
 Any delay in delivering high engine torque at low engine speeds can decrease drivability of the vehicle. Under these conditions, because of the substantial engine downsizing, the vehicle would have insufficient acceleration at low engine speeds until the turbo produces high pressures. This delay can be removed through the use of direct injection of ethanol by reduction of the spark retard or ethanol/gasoline with rich operation and also with the use of variable valve timing.
 Another approach would be to use an electrically assisted turbo charger. Units that can generate the required boosting for short periods of time are available. The devices offer very fast response time, although they have substantial power requirements.
 A multiple scroll turbocharger can be used to decrease the pressure fluctuations in the manifold that could result from the decreased number of cylinders in a downsized engine.
 The temperature of the air downstream from the turbocharger is increased by the compression process. Use of an intercooler can prevent this temperature increase from increasing the engine's octane requirement. In addition, in order to maximize the power available from the engine for a given turbocharging, cooling of the air charge results in increased mass of air into the cylinder, and thus higher power.
 In order to minimize emissions, the engine should be operated substantially all of the time, or most of the time, with a stoichiometric air/fuel ratio in order that a 3-way exhaust catalyst treatment can be used, FIG. 3 shows a 3-way exhaust treatment catalyst 10 and air, gasoline and ethanol control needed to maintain the substantially stoichiometric ratio of fuel to air that is needed for its effective operation. The system uses an oxygen sensor 12 as an input to an electronic control unit (ECU) 14. The ECU 14 controls the amount of air into a turbocharger 16, the amount of gasoline and the amount of ethanol so as to insure stoichiometric operation. During transients, open-loop algorithms from a stored engine map (not shown) are used to determine air, gasoline and ethanol flows for keeping substantially stoichiometric combustion in a cylinder of the engine 18.
 Thus when variable ethanol octane enhancement is employed, the fuel management system needs to adjust the amounts of air, gasoline and ethanol such that the fuel/air ratio is substantially equal to 1. The additional control is needed because, if the air; gasoline ratio determined by the fuel management were not be corrected during the injection of ethanol, the mixture would no longer be stoichiometric. In contrast to the lean boost approach of Stokes et al. [J. Stokes, T. H. Lake and R. J. Osborne, "A Gasoline Engine Concept for Improved Fuel Economy--The Lean Boost System," SAE paper 2000-01-2902] stoichiometric operation with a 3-way catalyst results in very low tailpipe emissions.
 There are certain regions in the engine operating map where the ECU 14 may operate open loop, that is, the control is determined by comparison to an engine map lookup table rather than by feedback from a sensed parameter which in this case is engine knock (closed loop). As mentioned previously, open loop operation during transients may be advantageous.
 Another situation where open loop control can be advantageous would be under high load, where fuel rich conditions (where the fuel/air ratio is greater than stoichiometric) may be required to decrease the temperature of the combustion and thus protect the engine and the exhaust system (especially during prolonged operation). The conventional approach in gasoline engine vehicles is to use increased fuel/air ratio, that is, operating at rich conditions. The presence of ethanol on-board allows for two alternatives. The first is the use of ethanol fuel fractions beyond what is required to control knock, thus reducing the combustion temperature by a greater amount than could be obtained by gasoline alone due to the higher cooling effect of evaporation in direct ethanol injection, even while at stoichiometric conditions. The second one is, as in conventional applications, the use of increased fueling in rich operation (which could result in relative air/fuel mass ratios as low as 0.75 where a stoichiometric mixture has a relative air/fuel ratio of 1). The control system can choose between two fuels, ethanol and gasoline. Increased use of ethanol may be better than use of gasoline, with emissions that are less damaging to the environment than gasoline and decreased amount of rich operation to achieve the temperature control needed. Open loop operation with both gasoline and ethanol may require substantial modification of the engine's "lookup table."
 Thus, a method of operating an engine is, under conditions of partial load, to operate closed loop with the use of only gasoline. As the engine load increases, the engine control system may change to open loop operation, using a lookup table.
 The closed loop control of the engine can be such that a knock sensor (not shown) determines the fraction required of ethanol, while the oxygen sensor 12 determines the total amount of fuel. A variation of this scheme is to operate the knock control open loop, using a lookup table to determine the ethanol to gasoline ratio, but a closed loop to determine the total amount of fuel.
 In order to evaporative emission of the ethanol (which has a relatively low boiling point), solvents can be added to the ethanol to minimize the effect. An alternative means is to place an absorptive canister between the ethanol tank and the atmosphere that captures the ethanol and releases it when the engine is operational.
 Because of the large cooling effect from ethanol, it has been known for some time that startup of a cold engine is difficult (for example, during the first 30 seconds). With the multiple fuels, it is possible to start up the engine without ethanol addition. Gasoline vaporizes easier than ethanol, and conventional operation with port-fuel or direct injected gasoline would result in easier engine start up. A greater fraction of gasoline than would be ordinarily used can be used to facilitate start-up operation at times during the first 30 seconds of engine operation.
 Increased efficiency due to engine downsizing made possible through the use of 100% or close to 100% ethanol at the highest values of torque has the undesirable effect of requiring higher ethanol fractions. Hence the use of non-uniform ethanol distribution to minimize the use of ethanol at these values of torque becomes more attractive when achievement of the maximum efficiency gain is desired.
 Below a certain value of torque or boost pressure it can be advantageous to use a non-uniform ethanol distribution in order to reduce the amount of ethanol that is used. Above certain torque or turbocharger or supercharger boost pressures, non-uniform charge would not be used since the engine is operating mostly on ethanol and ethanol non-uniformity cannot be used for minimizing ethanol consumption. This is especially important if the desired fraction is higher than 50%.
 The capability to minimize the use of ethanol by non-uniform ethanol distribution in the cylinder can be realized by certain ethanol injection geometries. Ethanol can be injected in the periphery of a swirling charge. In order to minimize wall wetting by the ethanol, it would be convenient to achieve the injection in a manner such that the ethanol injection matches the swirling motion of the charge. The injection direction is thus positioned at an angle with respect to the main axis of the cylinder, injecting the ethanol with an angular direction component. Charge stratification in the case of swirl can be maintained by temperature stratification, with the cooler (and denser) regions in the periphery, which correspond to the end-gas zone.
 An alternative or additional method to provide ethanol non-uniform distribution in the cylinder is to inject the ethanol relatively late with respect to bottom dead center. Thus the time for transport and diffusion of the ethanol is minimized. However, sufficient time should be allowed for full vaporization of the ethanol. As the temperatures are higher after Bottom-Dead-Center (BDC), the vaporization time is reduced, and it is less likely that the ethanol would wet the cylinder walls, improved vaporization of the ethanol can also be achieved by using injectors that produce small droplets. The injector could be a single spray pattern injector with a relatively narrow directed jet. This type of jet would optimize the deposition of the ethanol in the desired region.
 Creating a non-uniform ethanol distribution in the cylinder (in the outer regions of the cylinder) has two advantages. The first one is the increased cooling effect of the region that has the propensity to autoignite (knock), the end gas region. The second is that the central region is not cooled, improving ignition and initial flame propagation. It is preferable to keep the central region hot, as having a fast flame speed early in the flame propagation has antiknock advantages, by reducing the burn time and the time for precombustion chemistry of the end gas. Minimizing the burn time decreases the propensity to knock, as there is no knock if the end gas is burned before it can autoignite. Thus it is possible to have good ignition properties of the air/fuel mixture, even under conditions where the gasoline is evenly spread throughout the cylinder.
 Stratified operation can result in locally increased charge cooling. This is because the injected ethanol cools only a small fraction of the charge, and thus, for a given amount of ethanol, the local decrease in temperature is larger with stratified operation than the average decrease of temperature with uniform ethanol distribution. Late injection can aid in the formation of a non-uniform air/ethanol mixture as mixing time is limited. Since a fraction of the gasoline is port-fuel injected, it can be assumed that this fuel is homogeneously distributed in the cylinder, but ethanol is preferentially in the cooler edges (the end-gas). Thus, although overall the air/fuel charge is stoichiometric, locally near the spark it is lean while in the region of the end gas it is rich. Both of these conditions are advantageous, since the ignition occurs in a region with higher temperature (although slightly lean), while the outside is rich and cool, both of which are knock-suppressors.
 In the case of swirl or tumble stratified air fuel charges with hot air/gasoline in the center and colder air/ethanol or air/ethanol/gasoline mixtures in the end gas, it is advantageous to place the spark in the region of the hot air/gasoline mixture (substantially near the center of the combustion chamber).
 Ethanol consumption can be minimized if the gasoline is also directly injected. In this case, the heat of vaporization of gasoline is also useful in decreasing the temperature of the charge in the cylinder. The gasoline can be injected using a separate set of injectors. This would provide the most flexibility. However, it may be difficult to fit two sets of injectors per cylinder in the limited space in the cylinder head. An alternative means is to provide a single set of injectors for injection of both the ethanol and the gasoline. Two options are possible, one in which there is a single nozzle and valve (and the gasoline and ethanol are co-injected), and one in which each fuel has a separate nozzle and valve.
 Using direct injection of both the gasoline and the ethanol has the disadvantage of increased cost. In addition to a sophisticated injector or injectors, a second high pressure fuel pump is also needed. The ethanol and the gasoline also need to have parallel common plenums.
 When a single nozzle is used, the ethanol and the gasoline are distributed in the same manner in the cylinder. In the case with a single nozzle and single valve, the fuels need to be mixed prior to the valve/nozzle part of the injector. This could be done either outside of the injector or in the injector body. The volume between the mixing point and the nozzle should be minimized to allow for fast response of the fuel mixture.
 A slight modification of the above embodiment involves an injector that has two valves but a single nozzle. This minimizes the need for a second valve outside the injector for controlling the gasoline/ethanol mixture, in addition to minimizing the volume between the mixing point and the valves.
 It is possible to use a separate nozzle/valve for each fuel in a single injector. In this case, the gasoline and the ethanol can be deposited in different regions of the cylinder. An additional advantage would be to provide different spray patterns for the ethanol and for the gasoline. This would provide the most flexible system (comparable to two independent injectors), with possibilities of simultaneous or asynchronous injection of varying fractions of ethanol/gasoline, as well as being able to deposit the ethanol and the gasoline in the desired location of the charge, for optimal non-uniform distribution of ethanol in the cylinder. Optimal distribution means knock avoidance with minimal consumption of ethanol, while maintaining engine drivability. Optimal non-uniform ethanol distribution can be obtained by centrally depositing the gasoline and by preferentially depositing the ethanol in the periphery of the cylinder, where the end gas will be. This can be accomplished more easily with direct injection as opposed to achieving non-uniform distribution of the gasoline through non-uniform spraying in the inlet manifold. Because the heat of vaporization of the gasoline is substantially lower than for ethanol (a factor of 4 smaller on an energy basis), the cooling effect in the region near the spark is smaller, affecting less the initial flame propagation. In addition, it may be beneficial to retard the injection of the ethanol with respect to the gasoline.
 When the ethanol has been exhausted, the engine can operate in a `lower performance gasoline only` mode with turbocharger boost decrease (e.g. by a wastegate) and elimination or avoidance of operation at maximum torque levels. These conditions could be limiting, and in some cases a means of operating the vehicle at higher loads would be desired. This could be accomplished by using gasoline in the ethanol system with gasoline direct injection (GDI), while at the same time port-fuel injecting a fraction of the gasoline. Under these conditions the engine will operate at higher loads and higher torques, but still far below what ethanol could achieve. Only the cooling effect of the direct injection fuel is obtained, since the directly injected fuel has the same octane number as the port-injection fuel (gasoline in both cases).
 If the ratio of ethanol in the ethanol fuel tank to gasoline in the gasoline fuel tank is lower than a predetermined value (because of the lack or availability of ethanol or for some other reason), it is possible to change the engine operation condition such that the ethanol/gasoline consumption ratio over a drive cycle is decreased. This is done for reducing the maximum ethanol fraction at a given engine speed that can be used in the engine. The allowed level of turbocharging and the maximum pressure, torque and horsepower would be correspondingly reduced to prevent knock, in this way, a continuous tradeoff between the ethanol/gasoline consumption ratio and the maximum torque and horsepower can be accomplished.
 By proper expert system evaluation of the recent ethanol/gasoline usage and amounts of gasoline and ethanol it is possible to provide means to minimize the need of the `low performance, gasoline only` mode. The usage of the antiknock agent can be restricted when the amount left in the tank is below a predetermined level, such that the main fuel will be exhausted prior to or simultaneously with the ethanol. It would be desirable to place a switch so that the operator could override the limitations, in those conditions where the desired vehicle operation will not be limited by the exhaustion of the antiknock agent.
 Over a drive cycle, the amount of ethanol (by energy) required to enhance the octane number sufficiently to increase efficiency by at least 25% would be less than 15% of the fuel (ethanol+gasoline energy) without ethanol stratification and less than 5% with ethanol stratification.
 Onboard separation of ethanol from diesel by fractional distillation has been demonstrated for use in ethanol exhaust aftertreatment catalysts ["Fuel-Borne Reductants for NOx Aftertreatment: Preliminary EtOH SCR Study", John Thomas, Mike Kass, Sam Lewis, John Storey, Ron Graves, Bruce Bunting, Alexander Panov, Paul Park, presented at the 2003 DEER (Diesel Engine Emissions Reduction] Workshop, Newport R.I. August 2003]. This approach could be employed for onboard separation of ethanol from a gasoline mixture. However, use of membrane separation can be simpler and less expensive. Although there is information about the use of membranes for the separation of ethanol from water, to our knowledge there is no available information on the membrane separation of ethanol from gasoline. Because the ethanol molecule is on the order of 4 Angstroms and the typical hydrocarbon fuel molecules are much lamer, it is possible to use membranes bar the separation. Both organic and inorganic membranes could be used. Since it is not necessary to obtain high purity ethanol, the process is relatively simple and requires low pressure.
 Both porous and transfusion membranes can be used because ethanol with two carbon atoms has significantly different properties than most other gasoline compounds which have five to ten carbon atoms. The other antiknock agents contemplated for use in this invention also have a small number of carbons relative to gasoline. For example, methanol has one carbon. The membrane approach can be significantly simpler than the distillation or absorption/desorption approaches (see Ilyama et al, U.S. Pat. No. 6,332,448) that have been suggested for separation of various gasoline/diesel fuels where there is much less of a difference in the number of carbon atoms.
 The location of the membrane could be in the region of high pressure in the fuel line (downstream horn the pump), or upstream from it. If it is located downstream, the separation occurs only when the engine is operational and the pump is on, while it is upstream the separation is continuous. The pressure of the fuel downstream from the pump is a few bars (characteristic of port fuel injection). This is to be differentiated from the pressure of the ethanol system, which is directly injected and thus requires much higher pressures.
 The separated ethanol is transported to a separate tank where it is stored. If there is too much ethanol, three options are available: 1) additional separation is stopped; 2) some ethanol is used in the engine, even if not required 3) ethanol is returned to the main gasoline tank.
 The tank should be reachable, in order to be able to introduce additional ethanol when required, as when towing, in high temperatures, or when doing extensive climbing, conditions that require operation at high torque and which if for extended periods of time would consume ethanol at a rate higher than what can be extracted from the fuel.
 Extraction of ethanol from the gasoline can have the unintended effect of reducing the octane of the rest of the fuel. Thus, it is likely that somewhat increased use of injected ethanol would be required to prevent knock. Even in the case without non-uniform distribution of the ethanol, under normal driving conditions the system can be designed so that the amount of ethanol extracted from the fuel matches the required ethanol.
 It may also be advantageous to separate the ethanol from a gasoline/ethanol mixture at the fueling station. As with onboard separation, this approach also allows use of the present fuel transportation infrastructure. The potential advantages could be greater flexibility in choice of a fuel separation system and lower cost relative to onboard separation. It may be of particular interest during the introductory phase of ethanol boosted engine vehicles.
 It can be useful to have the capability to adjust the volume of the ethanol tank, thus varying the maximum amount of ethanol in the ethanol tank. This capability would make it possible to drive longer distances between ethanol refueling and to operate on different gasoline/ethanol ratios over a drive cycle, depending on the availability and cost of ethanol and gasoline. In some cases, it may be advantageous to use more ethanol than is needed to provide the desired octane enhancement (e.g., to meet alternative fuel or CO2 reduction goals). It is desirable to have this capability without increasing the overall fuel tank size. A single fuel tank with a membrane or plate separating variable amounts of gasoline and ethanol can be used to accomplish this goal.
 The tank can be configured to have a horizontal or vertical moveable/deformable walls that are substantially impervious and separate the regions that are filled with gasoline and ethanol. Separate filling ports and fuel lines are incorporated for each region as shown in FIGS. 4a and b. The separation between the gasoline and ethanol (or other anti-knock agent) does not have to be perfect since a small amount of leakage of one fuel into the other will not adversely affect operation of the vehicle. The wall can be moved in response to the amount of either fuel in the tank. This process is automatic in the case of a separating membrane, and the latter can be more impervious to leaks from one fuel to the other.
 Ethanol is denser than gasoline. The movable/deformable wall can be placed such that the ethanol is located either on top of the gasoline or below the gasoline. However, since it is expected that less ethanol is required than gasoline, the preferred embodiment has the ethanol above the gasoline, as shown in FIG. 5.
 If the ethanol is stored so that it is separate from the gasoline, it can be mixed with various additives to insure the desired operation of the ethanol injection system. In addition, it is possible to use gasoline-ethanol mixtures, such as E85 (which contains 15% by volume of gasoline). The lubricity additives include fatty acids, organic amine salts (amine salts of acid phosphates and polyethyleneoxy acid phosphates), alkyl and aryl acid phosphates and dialkyl alkyl phosphonates.
 The modeling calculations show that for direct injection of alcohols, the larger impact of knock suppression is not the intrinsic knock-resistance of the fuel antiknock agent but rather its high heat of vaporization. In order to evaluate alternatives to ethanol, Table 3 shows the properties of proposed fuel antiknock/alternative fuels. Although some of these additives have higher octane numbers than gasoline, some of them have a much larger effect on the cylinder charge temperature (Table 3 assumes injection after the inlet valve has closed). Some of these additives (mostly the ethers) have a comparable charge temperature effect to that of gasoline direct injection, and thus are of less interest. The alcohols have optimal properties for the application, with temperature changes that are a factor of 3 or more larger than the temperature change due to gasoline direct injection (for 100% or near 100% operation with the additive). For ethanol, the change in temperature is a factor of more than 4 larger than that of gasoline, and for methanol the change is about 9 times larger. The temperature decrease of the air increases with the amount of oxygen in the fuel (in terms of the O/C ratio). Thus, it is highest for methanol, with an O/C ratio of 1, second for ethanol (O/C=2), and so on.
TABLE-US-00003 TABLE 3 Antiknock properties of various fuels (calculated from data obtained in SAE standard J 1297 Alternative Automotive Fuels, September 2002) Equiv. Latent Vapori- Latent Net heat of zation heat of heat of vapori- energy/ Stoic vapori- (R + Combustion zation heat of air/fuel zation ΔT air Fuel type Chemical formula RON MON M)/2 MJ/kg MJ/kg combustion ratio MJ/kg air X Gasoline 42.8 0.30 0.007 14.6 0.020 -28 Ethyl t-Butyl Ether CH3CH2--O--C(CH3)3 118 102 110 36.3 0.31 0.009 12.1 0.026 -35 t-Amyl Methyl Ether C2H5 C (CH3)2--O--CH3 111 98 105 36.3 0.32 0.009 12.1 0.027 -36 Toluene C7H8 111 95 103 40.5 0.36 0.009 13.5 0.027 -37 Methyl t-Butil Ether CH3--O--C(CH3)3 116 103 110 35.2 0.32 0.009 11.7 0.028 -37 Diisopropyl Ether (CH3)2CH--O--CH(CH3)2 110 97 103 38.2 0.34 0.009 12.1 0.028 -39 t-Butyl Alcohol (CH3)3 C--OH 103 91 97 32.9 0.60 0.018 11.1 0.054 -74 Isopropanol (CH3)2CHOH 118 98 108 30.4 0.74 0.024 10.4 0.071 -97 Methanol with 50% methanol/TBA 114 96 105 26.5 0.88 0.033 8.8 0.100 -137 cosolvent Ethanol CH3CH2OH 129 102 115 26.7 0.91 0.034 9 0.102 -138 Methanol CH3OH 133 105 119 20.0 1.16 0.058 5.4 0.181 -246
 Also shown in Table 3 are the ratios of the heat of vaporization to the heat of combustion, a measure of the potential effects when used as antiknock agents. This parameter gives a measure of the amount of evaporative cooling for a given level of torque. The last entry, ΔTair, measures the decrease in air temperature for a stoichiometric mixture with injection after the inlet valve closes. Although the effect clearly is maximized by the use of methanol, other considerations may make ethanol the preferred choice. Methanol is toxic and corrosive.
 Hydrous ethanol (with a small amount of water) has the advantage of lower cost than pure (neat) ethanol. Removing the last 10% to 15% water from ethanol has significant expense and consumes considerable energy. Manufacturing facilities typically produce ethanol with about 10% water by volume unless there is a need for essentially pure (anhydrous) ethanol. It could be advantageous to use ethanol with a water concentration of 5% to 15% by volume.
 By using a closed loop approach to identify engine knock, flexible gasoline grades (with different octane ratings) and flexible knock-prevention fuel grades can be used. An open loop system would require measurement of the quality of the antiknock additive. Similarly, an open loop system would require determining the quality of the fuel (octane number). Closed loop operation allows the use of less expensive gasoline, when available, thus partially compensating for the more expensive anti-knock agent. It is also possible to use different antiknock feel according to its availability, such as ethanol in the regions that produce and process corn, and methanol in those that have methanol production capabilities. Thus, the least expensive grade of gasoline available and the least expensive antiknock fuel can be used, allowing a decrease of the cost of operating the vehicle as well as increasing the availability of the antiknock fuel.
 Although the above discussion has featured ethanol as an exemplary anti-knock agent, the same approach can be applied to other high octane fuel and fuel additives with high vaporization energies such as methanol (with higher vaporization energy per unit fuel), and other anti-knock agents such as isopropanol, tertiary butyl alcohol, or ethers such as methyl tertiary butyl ether (MTBE), ethyl tertiary butyl ether (ETBE), or tertiary amyl methyl ether (TAME). It may be advantageous to use various mixtures of these fuels and additives with each other and with ethanol.
 Particularly during the introduction phase of the present invention, the ethanol fueling could be performed by the use of containers, such as one-gallon containers. To facilitate ease of fueling an expandable pipe and funnel can be built into the ethanol fuel tank of the vehicle.
 The ethanol in these containers would be denatured so as to prevent human consumption as an alcoholic beverage and could contain the additives described above. Ethanol sold for fuel, such as in Brazil, is denatured by a small fraction of gasoline (2%) among other denaturing agents (methanol, isopropanol and others).
 Recycling of the container could take place at certain specific locations such as gasoline stations
 Using a signal from a knock sensor to determine when and how much ethanol or other anti-knock agent must be used at various times in a drive cycle to prevent knock, the heel management system can be employed to minimize the amount of ethanol or other anti-knock agent that is consumed over the drive cycle. If sufficient ethanol or other anti-knock agent is available, the fuel management system can also be used to employ more ethanol than would be needed to prevent knock. This would allow greater gasoline savings (the gasoline savings component from substitution of ethanol for gasoline would increase) and carbon dioxide reduction. In this case it may be desirable to operate at an anti-knock agent fraction which is either varied or constant during the drive cycle.
 The contents of all of the references cited in this specification are incorporated by reference herein in their entirety.
 It is recognized that modifications and variations of the inventions disclosed herein will be apparent to those of ordinary skill in the art and all such modifications and variations are included within the scope of the appended claims.
Patent applications by Daniel R. Cohn, Cambridge, MA US
Patent applications by John B. Heywood, Newtonville, MA US
Patent applications by Leslie Bromberg, Sharon, MA US
Patent applications by Massachusetts Institute of Technology
Patent applications in class Diverse fuel supply
Patent applications in all subclasses Diverse fuel supply