Title of Invention

PREMIXED CHARGE COMPRESSION IGNITION ENGINE WITH OPTIMAL COMBUSTION CONTROL

Abstract

A premixed charge compression ignition engine, and a control system, is provided which effectively initiates combustion by compression ignition and maintains stable combustion while achieving extremely low oxides of nitrogen emissions, good overall efficiency and acceptable combustion noise and cylinder pressures. The present engine and control system effectively controls the combustion history, that is, the time at which combustion occurs, the rate of combustion, the duration of combustion and/or the completeness of combustion, by controlling the operation of certain control variables providing temperature control, pressure control, control of the mixture's autoignition properties and equivalence ratio control. The combustion control system provides active feedback control of the combustion event and includes a sensor, e.g. pressure sensor, for detecting an engine operating condition indicative of the combustion history, e. g. the start of combustion, and generating an associated engine operating condition signal. A processor receives the signal and generates control signals based on the engine operating condition signal for controlling various engine components to control the temperature, pressure, equivalence ratio and\or auto ignition properties so as to variably control the combustion history of future combustion events to achieve stable, low emission combustion in each cylinder and combustion balancing between the cylinders.

Full Text

This invention relates generally to a compression ignition engine arranged to internally burn a premixed charge of fuel and air using auto ignition to achieve reduced emissions while maintaining the desired fuel economy. For well over 75 years the internal combustion engine has been mankind"s primary source of motive power. It would be difficult to overstate its importance or the engineering effort expended in seeking its perfection. So mature and well understood is the art of internal combustion engine design that most so called "new" engine designs are merely designs made up of choices among a variety of known alternatives. For example, an improved output torque curve can easily be achieved by sacrificing engine fuel economy. Emissions abatement or improved reliability can also be achieved with an increase in cost. Still other objectives can be achieved such as increased power and reduced size and/or weight but normally at a sacrifice of both fuel efficiency and low cost. The challenge to contemporary designers has been significantly increased by the need to respond to governmentally mandated emissions abatement standards while mamtamuig or improvmg fuel efficiency. In view of the mature nature of engine design, it is extremely difficult to extract both improved engine performance and emissions abatement from further innovations of the basic engine designs commercially available today. Yet the need for such innovations has never been greater in view of the series of escalating emissions standards mandated for the future by the United States government and other countries. Attempts to meet these standards includes some designers looking for a completely new engine design. Traditionally, there have been two prunary forms of reciprocating piston or rotary internal combustion engines: diesel and spark ignition engmes. While these engine types have similar architecmre and mechanical workings, each has distiact operating properties which are vastly different from each other. Diesel and spark ignited engmes effectively control the start of combustion (SOC) using simple, yet distmct means. The diesel engine controls die SOC by the timing of fuel injection. In a spark ignited engme, the SOC is controlled by the spark timing. As a result, there are important differences in the advantages and disadvantages of diesel and spark-ignited engines. The major advantage that a spark-ignited natural gas, or gasoline, engine has over a diesel engine is the ability to achieve extremely low NOx and particulate emissions levels. The major advantage that diesel engines have over premixed charge spark ignited engmes (such as passenger car gasolme engmes and lean burn natural gas engines) is higher thermal efficiency. One key reason for the higher efficiency of diesel engmes is the ability to use higher compression ratios tiian premixed charge spark ignited engines (the compression ratio m premixed charge spark ignited engines has to be kept relatively low to avoid knock). A second key reason for the higher efficiency of diesel engines lies m the ability to control the diesel engine"s power output without a throttle. This elimmates the throttling losses of premixed charge spark igmted engines and results in significantly higher efficiency at part load for diesel engines. Typical diesel engines, however, cannot achieve the very low NOx and particulate emissions levels which are possible with premixed charge spark ignited engines. Due to the mixmg controlled nature of diesel combustion a large fraction of the fuel exists at a very fuel rich equivalence ratio which is known to lead to particulate emissions. Premixed charge spark ignited engmes, on the other hand, have nearly homogeneous ah fuel mixmres which tend to be either lean or close to stoichiomettic, resulting m very low particulate emissions. A second consideration is tiiat the mixing controlled combustion m diesel engmes occurs when the fuel and air exist at a near stoichiometric equivalence ratio which leads to high temperamres. The high temperatures, m mm, cause high NOx emissions. Lean bum premixed charge spark ignited engines, on the other hand, burn their fuel at much leaner equivalence ratios which results in significantly lower temperatures leading to much lower NOx emissions. Stoichiometric premixed charge spark ignited engmes, on the other hand, have high NOx emissions due to the high flame temperatures resulting from stoichiometric combustion. However, the virtually oxygen free exhaust allows the NOx emissions to be reduced to very low levels with a three-way catalyst. Relatively recently, some engine designers have directed their efforts to another type of engme which utilizes premixed charge compression ignition (PCCI) or homogeneous charge compression ignition (HCCI), hereinafter collectively referred to as PCCL Engmes operating on PCCI prmciples rely on autoignition of a relatively well premixed fuel/ah mixmre to initiate combustion. Importantly, the fuel and air are mixed, in the intake port or the cylinder, long before ignition occurs. The extent of the mixture may be varied dependmg on the combustion characteristics desired. Some engines are designed and/or operated to ensure the fuel and air are mixed into a homogeneous, or nearly homogeneous, state. Also, an engme may be specifically designed and/or operated to create a somewhat less homogeneous charge having a small degree of stratification. In both instances, the mixture exists in a premixed state well before ignition occurs and is compressed until the mixture autoignites. Importantly, PCCI combustion is characterized in that: 1) the vast majority of the fuel is sufficiently premixed witii the an- to form a combustibie mixture (hroughcat the charge by the time of igm"tion and throughout combustion; and 2) combustion is mitiated by compression ignition. Unlike a diesel engme, the tuning of the fuel delivery, for example the timing of injection, in a PCCI engme does not strongly affect the timing of ignition. The early delivery of fuel in a PCCI engine results m a premixed charge which is very well mixed, and preferably nearly homogeneous, thus reducmg emissions, unlike the stratified charge combustion of a diesel which generates higher emissions. Preferably, PCCI combustion is characterized m that most of the mixture is significantly leaner than stoichiometric or heavily diluted to advantageously reduce emissions, unlike the typical diesel engine cycle in which a large portion, or all, of the mixture exists in a rich state during combustion. An engme operating on PCCI combustion principles has the potential for providmg the excellent fael economy of the diesel engine while providing NOx and particulate emissions levels tiiat are much lower than that of current spark-ignited or diesel engme. For example, U.S. Patent No. 4,768,481 to Wood discloses a process and engine that is intended to use a homogeneous mixture of fuel and air which is spontaneously ignited. A controlled rate of combustion is said to be obtained by addmg exhaust products to the au:-fiiel mixture, A combustion chamber is connected to the engine cylinder and fuel gas is supplied to the chamber via a check valve. A glow plug is positioned between the combustion chamber and the cylinder. The mixture entering the combustion is heated by the glow plug and by the hot walls of the combustion chamber. The mixture ignites due Co Cbe increase in tempersiure arid (he increase in pressure resalting from, compression. The Wood patent is specifically directed to a two-stroke engine, but generally mentions that the technology could be applied to a four-stroke engine. However, this reference fails to discuss how the exhaust gas recirculation and glow plug would be controlled to optimize the start of combustion and to mamtain the optimal start, and duration, of combustion, as load and ambient conditions change. A practical embodiment of this engine is unlikely to be capable of effectively controlling and maintainmg PCCI combustion without additional controls. U.S. Patent No. 5,535,716 issued to Sato et al., discloses a compression ignition type engine which greatly reduces NOx emissions by mtroducing an evaporated fuel/air mixture mto the combustion chamber during the intake event and early in the compression event for self-ignited combustion later in die compression event. The amount of NOx emissions produced by this engine is about one-thirtieth of that produced by a diesel engine. These principles are also set forth m SAE Technical Paper No. 960081, Aoyama, T. et al., "An Experimental Smdy on Premixed-Charge Compression Ignition Gasoline Engine", February 26, 1996. However, these references do not specifically discuss controlling the timing of the start of combustion and the rate of combustion. Moreover, the engine disclosed in diese references only uses the heat generated by compression to ignite the charge, without the use of any preheating. Also, these references do not suggest the controls, nor the maimer of operating the controls, necessary to maintam stable combustion. Also, these references only disclose the use of gasoline. U.S. Patent No. 5,467,757 issued to Yanagihara et al., discloses a direct mjection compression-ignition type engine in which fuel is injected into a combustion chamber during the intake stroke or compression stroke, before 60 degrees BTDC of the compression stroke, so as to reduce the amount of soot and NOx generated to substantially zero. These advantages are achieved by considerably enlarging the mean particle size of the injected fuel from the mean particle size used in conventional combustion processes to prevent the early vaporization of injected fuel after injection and by making the injection timing considerably earlier than conventional injection timing to ensure a uniformed fusion of the injected fuel in the combustion chamber. However, this reference nowhere suggests a manner of actively controlling the combustion history, such as the timing of die start of combustion and/or the duration of combustion. Researchers have used various other names to refer to PCCI combustion. For example, Onishi, et al. (SAE Technical Paper No. 790501, February 26-March 2, 1979) called it "ATAC", which stands for "Active Thermo-Atmosphere Combustion." Noguchi, et al. (SAE Technical Paper No. 790840, September 10-13, 1979) called it "TS", which stands for "Toyota-Soken", and Najt, et al. (SAE Paper No. 830264, 1983) called it "CIHC", which stands for "compression-ignited homogeneous charge." Onishi, et al., worked with two-stroke engines. They found that PCCI combustion (ATAC) could be made to occur in a two-stroke engine at low load over a wide speed range. Combustion stability was much better than in the standard engine and there were significant improvements in fuel economy and exhaust emissions. Schlieren photography of the combustion was carriea out with results quite similar to those obtained in their combustion smdies. It was found that combustion was initiated at many points in the combustion chamber. However, there were small time differences between die start of combustion of these many points. Also, the combustion reactions were found to require a relatively long time compared to conventional spark-ignited flame propagation. To attain PCCI combustion, the following conditions were found to be important. The quantity of mixmre and the air/fuel ratio supplied to the cylinder must be uniform from cycle to cycle. The scavenging "directivity" and velocity must have cyclic regularity to ensure the correct condition of the residual gases remaining in the cylinder. The temperature of the combustion chamber walls must be suitable. The scavenging passage inlet must be located at the bottom of the crankcase. It was found that at very light loads, PCCI was not successful because charge temperatores were too low. At very high loads, PCCI was not successful because the residual gas quantity was too low. In between these regions, PCCI combustion was successful. Noguchi also obtained PCCI combustion in a two-stroke engine. Very stable combustion was observed, with low emissions of hydrocarbons (HC) and improved fuel consumption. Operation in PCCI mode was possible between 800 and 3200 rpm and air/fiiel ratios between 11 and 22. Delivery ratios of up to 0.5 could be achieved at idle conditions. They observed diat combustion could start at lower temperamres and pressures than those required for conventional diesel combustion. The combustion behavior was different from that of conventional spark-ignited combustion. Ignition occurred at numerous points around the center of the combustion chamber and the flame spread rapidly in all directions. The combustion duration was shorter than that of conventional combustion. It was proven that ignition kernels were not generated from contaminants deposited on the combustion chamber walls (generally presumed to be die cause of "run-on" phenomena in conventional gasoline engines). To gain a better understanding of die combustion, they set up an experimental apparams for detecting radicals in the combustion chamber. It was found that the radicals showed higher peaks of luminous intensity diat disappeared at an earlier time than with conventional spark-ignited combustion. In the case of conventional spark-ignition combustion, all the radicals such as OH, CH, C^, H, and CHO, HO2, O were observed at almost the same crank angle. However, with PCCI combustion, CHO, HOj and O radicals were detected first, followed by HC, C2, and H radicals, and finally die OH radical. Najt, et al. were able to achieve PCCI combustion in a four-stroke engine. They used a CFR single-cylinder engine with a shrouded intake valve. Several compression ratios were tried, and it was found that, although higher ratios would allow combustion at lower charge gas temperatures, they also resulted in excessively fast heat release rates. While a compression ratio of 7.5:1 was satisfactory, a compression ratio of 10:1 was not. Intake temperatures were in the range of 480 °K to 800 °K. Their average energy release rates were considerably higher than those measured by Onishi and Noguchi. SAE Paper No. 960742, entitled "Improving the Exhaust Emissions of Two-Stroke Engines by Applying the Activated Radical Combustion", Ishibashi, Y. et al., 1996, is noted as disclosing yet another study of PCCI combustion in a two-stroke engine. Although Onishi et al., Noguchi et al., Naijt et al. and Ishibashi, et al. have made significant progress in understanding PCCI combustion, these references fail to suggest a practical PCCI engine having a control system capable of maintaining stable, efficient PCCI combustion with low emissions by controlling the time at which combustion occurs, the duration of combustion, the rate of combustion and/or the completeness of combustion. Specifically, these references do not suggest a PCCI engine and control system capable of effectively controlling the start of combustion. Moreover, these references do not suggest a system capable of actively enhancing the engine startability and achieving combustion balancmg between the cylinders in a multi-cylmder engine. SAE Technical Paper No. 892068, entitled "Homogeneous-Charge Compression Ignition (HCCI) Engines", Thring, R., September 25, 1989, invesfigated PCCI operation of a four stroke engine. The paper found that PCCI required high exhaust gas recirculation (EGR) rates and high intake temperamres. It was shown that PCCI combustion produces fuel economy results comparable to a direct injection diesel engine and, that under favorable conditions, i.e. equivalence ratio of .5 and EGR rate of 23%, produces very low cyclic irregularity. This study also concluded that before PCCI can be made practical, it will be necessary to operate an engine in the PCCI mode without the need to supply large amounts of heat energy to die intake. The paper suggests two possibilities: the use of heated surfaces in the combustion chamber and the use of multi¬stage turbocharging without intercoolers. However, although this paper suggests further investigating the effects of EGR and intake temperature on the timing of the start of combustion, this paper fails to disclose a system for effectively achieving active control of the start and duration of combustion. U.S. Patent No. 5,476,072 to Inventor discloses another example of a PCCI engine which includes a cylinder head design that prevents excessive stresses and structoral damage that PCCI engines mherently tend to cause. Specifically, the head includes a movable accumulator piston which moves to limit the peak cylinder pressure and temperamre. However, control over the movement of the piston is merely passive and, therefore, this engine is unlikely to effectively stabilize combustion. Moreover, this reference nowhere suggests controUmg the timing at which rapid combustion occurs, nor how such control could be accomplished. An October 1951 publication entitled "Operating directions - LOHMANN BICYCLE MOTOR" discloses a two-stroke engine operating on PCCI combustion principles. Compression ratio is continuously adjustable based on outside temperature, fuel, speed and load. However, this engine requires the operator control the compression ratio manually. Therefore, this engine could not provide effective active control of combustion to ensure efficient combustion with low emissions throughout all operatiug conditions. Also, manual adjustment of compression ratio alone, without automatic temperamre, equivalence ratio and\or auto ignition property control, will not result in stable, optunized combustion throughout all operating conditions. Conventional "dual fuel" engines operate on both a gaseous fiiel mixmre and diesel fuel. However, conventional dual fuel engines utilize the tuning of the injection of diesel fuel to control the SOC of the fuel/air mixture received from the intake duct. In order to achieve this result, dual fuel engines inject the diesel fuel at approximately top dead center. In addition, the quantity of diesel ftiel injected m a dual fuel engine is sufficient to ensure that the gaseous fuel in the combustion chamber ignites and bums virmally completely. As a result, dual fuel engines produce emissions similar to most conventional diesel and namral gas engines. In particular, in known dual fuel engines using diesel fuel and natural gas at high load, only a small amount of diesel fuel is required to start ignition and the emissions produced would be similar to a spark ignited namral gas engine. Under other conditions when substantial diesel fuel is injected, the emissions produced would be similar to a conventional diesel engine. Consequently, there is a need for an engine operating on PCCI principles which includes a combustion control system capable of effectively controlling the timing of the start of combustion or location of combustion, and the rate or duration of combustion during engine operation. SUMMARY OF THE INVENTION A general objective of the subject invention is to overcome the deficiencies of the prior art by providing a practical PCCI engine and a control system for effectively and efficiently operating the PCCI engine. Another object of the present invention is to provide a PCCI engine and control scheme for controlling the engine in a manner to optimally minimize emissions, especially oxides of nitrogen and particulate emissions, while maximizing efficiency. Yet anodier object of the present mvention is to provide a PCCI engine and control system for optimally controlling the combustion history of subsequent combustion events to effectively control the combustion event. Still anodier object of die present invention is to provide a PCCI engine and control system for effectively controlling PCCI combustion in such a manner to achieve acceptable cylinder pressure while minimizing combustion noise. A fiirther object of die present invention is to provide a PCCI engine and control system which operates to actively control the combustion history of fumre combustion events during engine operation by sensing an engine operating condition indicative of the combustion history. A still fiirdier object of the present invention is to provide a PCCI engine and control system which effectively controls various engine operating control variables to control the time at which the combustion event occurs during die compression and expansion events of the engine. Yet anodier object of the present invention is to provide a PCCI engine and control system which effectively ensures diat combustion occurs at an appropriate crank angle during the engine cycle to ensure stable combustion, low emissions, acceptable pressure levels and optimum efficiency. Another object of the present invention is to provide a PCCI engine and control system which effectively controls the temperature, pressure, equivalence ratio and/or air/fuel mixture autoignition properties to precisely control die timing of the start of combustion. A still further object of the present invention is to provide a PCCI engine and control system which effectively achieves continuous, stable PCCI combustion while achieving acceptable cylinder prtssures and the desired brake mean effective pressure. Yet another object of the present invention is to provide a PCCI engine and control system which effectively controls die commencement of combustion and the combustion rate so as to ensure that substantially all of the combustion process occurs within an optimal crank angle limit, i.e. 20 degrees BTDC through 35 degrees ATDC, while minimizing emissions and maximizing efficiency. Another object of the present invention is to provide a PCCI engine which can be easily started. Still another object of the present invention is to provide a multi-cylinder PCCI engine and control system which effectively minimizes variations in the combustion events of die cylinders. Yet another object of the present invention is to provide a multi-cylinder PCCI engine and control system which effectively controls the start of combustion to achieve stable, low emission, efficient combustion throughout exposure to changes in engine load and ambient conditions. Another object of the present invention is to provide a control system for a PCCI engine which effectively detects or senses the start of combustion to provide feedback control and tiien controls the operating conditions of the engine to optimize the start of combustion. Still anodier object of the present invention is to provide a PCCI engine and control system which effectively minimizes the unburned hydrocarbon and carbon monoxide emissions. The above objects and others are achieved by providing a premixed charge compression ignition internal combustion engine, comprising an engine body, a combustion chamber formed in the engine body and combustion history control system for controlling a combustion history of future combustion events to reduce emissions and optimize efficiency. The combustion history control system includes at least one of a temperature control system for varying the temperature of the mixture of fuel and air, a pressure control system for varying the pressure of the mixture, an equivalence ratio control system for varying an equivalence ratio of the mixture and a mixture autoignition property control system for varying an autoignition property of the mixture. The engine ftirther includes an operating condition detectmg device for detecting an engine operating condition indicative of the combustion history and generating an engine operating condition signal mdicative of the engine operating condition, and a processor for receiving the engine operating condition signal, determining a combustion history value based on the engine operating condition signal, and generating one or more control signals based on the combustion history value. The one or more control signals are used to control at least one of the temperature control system, the pressure control system, the equivalence ratio control system and the mixture autoignition property control system to variably control the combustion history of future combustion events. The engine operating condition detecting device may include a start of combustion sensor for sensing the start of combustion and generating a start of combustion signal. Also, the combustion history value may be determmed based on the start of combustion signal. The engine operating condition detecting device may be a cylinder pressure sensor. BRIEF DESCRIPTION OF THE DRAWINGS Fig. la is a schematic diagram of one embodiment of the present invention showing a single cylinder of the engine of Fig. lb and associated control system; Fig. lb is a schematic diagram of a multi-cylinder engine of the present invention; Fig. 2 is a graph showing cylmder pressure and heat release rate as a function of crank angle for the PCCI engine of the present invention; Fig. 3 is a graph showing the apparent heat release rate as a function of crank angle for several different engine operating conditions; Fig. 4a is a graph showing knock intensity as a function of time for a given set of operating conditions; Fig. 4b is a graph showing gross mdicated mean effective pressure (GIMEP) as a function of time; Fig. 4c is a graph showmg peak pressure as a function of time for the same conditions of Figs. 4a and 4b; Fig. 5 is a graph showing apparent heat release rate as a function of crank angle and illustrating the increase in the heat release rate duration as the combustion or heat release location or timing is retarded; Fig. 6 is a graph showing cylinder pressure as a function of crank angle and illustrating the decrease in peak cylinder pressure as the heat release rate retards; FIG 7a. is a graph showing GIMEP as a function of mtake manifold temperamre for two different engine speed cases; Fig. 7b is a graph showmg the coefficient of variation of GIMEP as a function of intake manifold temperature for two different engme speed cases; Fig. 7c is a graph showing peak cylinder pressure as a function of intake manifold temperature for two different engine speeds; Fig. 7d is a graph showing the start of combustion as a function of intake manifold temperature for two different engine speeds; Fig. 7e is a graph showing heat release duration in crank angle degrees as a function of intake manifold temperamre for two different engine speeds; Fig. 7f is a graph showmg heat release duration in time as a function of intake manifold temperature for two different engine speeds; Fig. 7g is a graph showing gross mdicated thermal efficiency as a function of intake manifold temperature for two different engine speeds; Fig. 7h is a graph showing fuel specific hydrocarbons as a function of intake manifold temperature for two different engine speeds; Fig. 7i is a graph showing fuel specific carbon monoxide as a function of intake manifold temperature for two different engine speeds; Fig. 7j is a graph showing fuel specific oxides of nitrogen emissions as a function of intake manifold temperature for two different engine speeds; Fig. 7k is a graph showing noise as a function of intake manifold temperature for two different engine speeds; Fig. 8 is a graph showing apparent heat release rate as a function of crank angle for three different intake manifold temperatures; Fig. 9 is a graph showing botii die start of combustion and combustion duration as a function of wall temperature; Fig. 10 is a graph showing bodi die start and end of combustion as a function of crank angle for a given time period, and GIMEP for the same tune period, wherein a glow plug is cycled; Fig. 11 is a graph showing the apparent heat release rate as a function of crank angle for the glow plug transient of Fig. 10; Fig. 12 discloses one embodiment of an end cylinder compensating system of the present invention for providing cylinder-to-cylinder temperature control; Fig. 13 is a schematic diagram of a second embodiment of die end cylinder compensating device for providing cylinder-to-cylinder temperature control; Fig. 14 is a graph showing the effects of changing intake and exhaust valve opening and closing events on top dead center (TDC) temperauire; Fig. 15 is a graph showing the effects of changing mtake and exhaust valve opening and closing events, and variable compression ratio, on the residual mass fraction and temperature at top dead center; Fig. 16 is a graph showing both cylinder pressure and heat release as a function of crank angle for different exhaust valve lash settings; Fig. 17 is a graph showmg the effects of varying exhaust gas recirculation (EGR) on the location of the heat release rate relative to the crank angle and the effect of variations in EGR on the magnitude of the heat release rate; Fig. 18 is a graph showing the effect of varying the EGR rate on the tuning of the start of combustion; Fig. 19 is a schematic of an improved engine of the present invention having one cylinder operating under PCCI conditions to optimize the use of EGR; Fig. 20 is a graph showing the effects of changing compression ratio on the temperature at top dead center; Fig. 21 is a graph showing the start of combustion as a fiinction of intake manifold temperature and the effects of changing the compression ratio on the start of combustion and intake manifold temperamre; Fig. 22a is a partial cross sectional view of one cylinder of the PCCI engine of the present invention mcluding one embodiment of a compression ratio varying device; Fig. 22b is a partial cross sectional view of one cylinder of the PCCI engine of the present invention showing a second embodiment of a compression ratio varying device; Fig. 22c is a partial cross sectional view of one cylinder of the present PCCI engine showing a third embodiment of the compression ratio varying device; Fig. 22d is a partial cross sectional view of a single cylinder of the present PCCI engine showing a fourth embodiment of the compression ratio varying device of the present invention; Fig. 23 is a schematic diagram of an opposed piston PCCI engine of the present invention including a variable phase shifting mechanism for varying the compression ratio; Fig. 24 is a side view of the differential mechanism used in the variable phase shifting mechanism of Fig. 23; Fig. 25 is a graph showing compression ratio as a function of the degrees out of phase of two pistons in the opposed piston engine, for example, of FIG 23 illustrating various compression ratio settings; Fig. 26 is a graph showing cylinder volume as a function of crank angle of a reference piston in an opposed piston PCCI engine which shows that the compression ratio decreases as the pistons become more out of phase; Fig. 27 is a graph showing the effects of changing intake and exhaust valve opening and closing events, and varying the compression ratio, on the percent of baseline airflow rate and the TDC temperature; Fig. 28 is a graph showing the effects of changes and intake in exhaust valve opening and closing events, and varying the compression ratio, on die diesel equivalent brake specific fiiel consumption and TDC temperature; Fig. 29 is a graph showmg the effects of changes and intake in exhaust valve opening and closing events, and variations in compression ratio, on peak cylinder pressure and TDC temperature; Fig. 30 is a graph showing the effects of water injection on intake manifold temperature and temperamre at top dead center; Fig. 31a is a graph showing the combustion duration in crank angle degrees as a function of intake manifold pressure (IMP); Fig. 31b is a graph showing combustion duration in time as a function of IMP; Fig. 31c is a graph showing the effect of changes in IMP on the magnitude and timing or location of the heat release rate; Fig. 3 Id is a graph showing the start of combustion timing and crank angle degrees as a function of IMP; Fig. 31e is a graph showing fuel specific hydrocarbons as a function of IMP; Fig. 3If is a graph showing GIMEP as a function of IMP; Fig. 3 Ig is a graph showing gross indicated thermal efficiency as a fiinction of IMP; Fig. 31h is a graph showing fuel specific carbon monoxide as a function of IMP; Fig. 31i is a graph showing fuel specific oxides of nitrogen emissions as a function of IMP; Fig. 31j is a graph showing the coefficient of variation of GIMEP as a fiinction of IMP; Fig. 31k is a graph showing the peak cylinder pressure as a function of IMP; Fig. 311 is a graph showing noise as a function of IMP; Fig. 3 Im is a graph showing the effects of increasing IMP on peak cylinder pressure and GIMEP; Fig. 32 is a graph showing the effect of various trace species on a start of combustion and temperamre; Fig. 33 is a graph showing the effects of additional amounts of ozone on advancing the start of combustion; Fig. 34 is a graph showing the effect of varying the type of fuel used m the present PCCI engine on the start of combustion wherein the increase in temperature indicates the start of combustion; Fig. 35 is a graph showing the apparent heat release duration as a function of equivalence ratio; Fig. 36 is a graph showing the start of combustion in crank angle degrees as a function of equivalence ratio; Fig. 37 is a graph showing the effects of variations in equivalence ratio on the start of combustion wherein an increase in temperature indicates the start of combustion; Fig. 38 is a graph showing the effects of variations in the equivalence ratio on the magnitude and timing, or location, of the heat release rate; Fig. 39 is a graph showing the effects of equivalence ratio on the compressor pressure ratio and the compressor outlet temperamre; Fig. 40 is a graph showing the effects of varying the equivalence ratio on the brake specific fuel consumption; Fig. 41 is a graph showing die differences in pumping mean effective pressure and GIMEP for two differently sized turbine casings; Fig. 42 is a graph showing the diesel equivalent BSFC and BMEP for two differently sized turbine casings; Fig. 43 is a graph showing the turbine rotor speed and intake manifold pressure for two differently sized mrbine casings; Fig. 44 is a graph showing the fuel specific oxides of nitrogen emissions for PCCI combustion with various fuels in comparison to a typical compression ignition diesel engine; Fig. 45 is a graph showing emissions as a function of engine speed; Fig. 46 is a graph showing emissions as a function of temperature at bottom dead center; Fig. 47 is a graph showing fuel specific carbon monoxide as a function of end of combustion flame temperature; Figs. 48a-50b are partial cross sectional views of a single cylinder of the PCCI engine of the present mvention showing an alternative embodiment including various crevice minimizing feamres; and Fig. 51 is a graph showing the effects of various percentages of diesel pilot injections on the heat release rate location and shape. Fig. 52 is a schematic diagram of one embodiment of the present invention showing a flapper valve for controlling intake air flow; Fig. 53 is a graph showmg cumulative heat release and the determined SOC as a function of the crank angle using closed loop control in the PCCI engine of the present invention; Fig. 54 is a graph showing the GIMEP and intake manifold temperature as a function of the start of combustion; Fig. 55 is a graph showing pressure as a function of volume for a conventional PCCI combustion event; Fig. 56 is a graph showing pressure as a function of volume for one embodiment of the present invention including a constant pressure portion; Fig. 57 is a schematic diagram of another embodiment of the present PCCI engine; Fig. 58 is a schematic diagram of yet another embodiment of the present invention including a pumping cylinder to enhance engine thermal efficiency; Figs. 59A-59D illustrate the various positions of the power pistons and the pumping pistons of the embodiment of Fig. 58 at different points in a cycle; Fig. 60 is a schematic diagram of still another embodiment of the PCCI engine of the present invention mcludmg a pumping cylinder to enhance engine thermal efficiency; Fig. 61 is a graph showmg the effective flow area of the exhaust valves and intake valves of the engine of Fig. 60 as a function of crank angle; Figs. 62A-62E illustrate the various positions of the power pistons and the pumping pistons of the engine of Fig. 60 at different points in a cycle; Fig. 63 is a schematic diagram showing another embodiment of the present PCCI engme includmg a displacer for controlling SOC; Fig. 64 is a schematic diagram of the engine of Fig. 63 showing the displacer moving between positions; Fig. 65a-65d disclose an embodiment of the present invention including variable exhaust valve closing to control EGR; Fig. 66 is a schematic of one embodiment of controlling exhaust valve timing to control EGR; Fig. 67 is a detailed view of the hydraulic link system used in the embodiment shown in Fig. 66; and Fig. 68 is a schematic diagram of a second embodiment for controlling the timing of exhaust valve closing to control EGR. DETAILED DESCRIPTION OF THE INVENTION The present mvention is directed to an improved premixed charge compression ignition (PCCI) engine and control scheme for controlling the engme in a manner to optimally minimize emissions while maximizmg efficiency. For the purposes of this application, PCCI refers to any engine or combustion process in which: 1) the vast majority of the fiiel is sufficiently premixed with the air to form a combustible mixture throughout the charge by the time of ignition and throughout combustion; and 2) combustion is initiated by compression ignition. PCCI also refers to any compression ignition engine or combustion process in which the fuel and air are premixed long before ignition. As a result, the timing of injection of the fuel in the PCCI engine does not significantly affect the timing of ignition of the fuel/air mixture. Also, it should be understood that PCCI is meant to encompass homogeneous charge compression ignition (HCCI) engines and processes wherein the mixmre exists in a homogeneous, or nearly homogeneous state, at the start of combustion. In the present invention, the fuel/air mixture is thoroughly mixed to form a very lean homogeneous mixtore, or is mixed in. a manner to form a less homogeneous mixture with a desired air/fuel stratification, to ensure relatively even, low flame temperanires which result in extremely low oxides of nitrogen (NOx) emissions. It should be understood the some engines operate under PCCI conditions continuously while other enguies may operate under PCCI conditions for only a limited period of operation either by design or inadvertently. Applicants have recognized that the key to producing a commercially viable PCCI engine lies in the control of the combustion history of subsequent or fiimre combustion events in such a manner so as to result in extremely low NOx emissions combined with very good overall efficiency, combustion noise control and with acceptable cylinder pressure. The combustion history for a particular cycle may include the time at which combustion occurs (combustion timing), the rate of combustion (heat release rate), the duration of combustion and/or the completeness of combustion. Applicants have determined that the combustion history, and especially the combustion timing, is sensitive to, and varies depending on, a variety of factors including changes in load and ambient conditions. The engine and control system of the present invention operates to actively control the combustion history of future combustion events during engine operation to ensure the desired combustion and engine operation is maintained. In the preferred embodiment, the present engine and control system controls the combustion timing during the compression and expansion events of the engine. Figs, la and lb illustrates the PCCI engine and control system of the present invention, indicated generally at 10. Fig. la shows a single engine cylinder 12 of the multi-cylinder reciprocating piston type engine shown in Fig. lb. Of course, the PCCI control system of the present invention could be used to control PCCI combustion in an engine having only a single cylinder or any number of cylinders, for example, a four, six, eight or twelve cylinder internal combustion engine. In addition, although the present PCCI control system is primarily discussed with reference to a four stroke engine, the present control system could be applied to a two stroke engine. Also, the PCCI system of the present invention may be adapted for use on any internal combustion engine having compression, combustion and expansion events, including a rotary engine and a free piston engine. As shown in Fig. la, a piston 14 is reciprocally mounted in the cylinder to form a combustion chamber 13. The piston transmits forces generated by a combustion event into a conventional engine drive system. Referring to Figs, la and lb, an intake air system 23 including an intake manifold 15 supplies intake air, or an ak/fuel mixture to a respective intake port 26 associated with each cylinder 12. Likewise, an exhaust gas system 27 including an exhaust manifold 17 receives exhaust gases flowing from exhaust ports 31. One or more intake valves, such an intake valve 19 and one or more exhaust valves, such as exhaust valve 21, are moved between open and closed positions by a conventional valve control system, or a variable valve timing system, to control the flow of intake air or air/fuel mixture into, and exhaust gases out of, the cylinder, respectively. The PCCI system 10 includes a combustion sensor 16 for sensing or detecting an engine operating condition indicative of the combustion history and generatmg a corresponding signal 18. In the preferred embodiment, sensor 16 permits effective combustion control capability by detecting an engine operating condition or parameter directly related to, or indicative of, the time at which the combustion event occurs during the compression and\or expansion strokes, i.e. preferably the start of combustion (SOC). For example, a cylinder pressure sensor may be provided on any or all engine cylinders for sensmg, on a cycle-by-cycle basis, the SOC. In this case, the sensor 16 also provides odier engine condition data, such as the combustion rate, combustion duration, crank angle at which peak cylinder pressure occurs, combustion event or heat release location and end of combustion data, any one of which may be used instead of the start of combustion data. Any conventional means for detecting the start of combustion may be used, for example, by sensing the very rapid increase in the cylinder pressure. Other forms of sensors could be used including accelerometers, ion probes, optical diagnostics, strain gages and/or fast thermocouples in the cylinder head, liner or piston. Also, torque or RPM sensors could be used to detect changes m engine torque and RPM associated with each combustion event. Alternatively, or additionally, an emissions sensor could be used to detect emissions having a known correlation to the completeness of combustion. Sensor 16 provides feedback control to an electronic control unit 20 (ECU). ECU 20 receives signal 18, processes the signal and determines an actual combustion history value, i.e. stkrt of combustion value. The actual combustion history value is tiien compared to a predetermined desired combustion history value obtained, for example, from a look-up table. Based on the comparison of the actual combustion history value to the desired combustion history value, ECU 20 then generates a plurality of output signals, indicated at 22, for variably controlling respective components of the system so as to effectively ensure, in the preferred embodiment, that the SOC and completion of combustion occur between 20 degrees before top dead center (BTDC) durmg the compression stroke and 35 degrees after top dead center (ATDC) durmg the power stroke of the piston thereby minimizing NOx emissions while maximizing engine efficiency. The PCCI combusdoa control scheme is most preferably irapZemented in software contained m ECU 20 that includes a central processing unit such as a micro-controller, micro¬processor, or other suitable micro-computmg unit. Of course, the position of the piston can be determined by detecting the angular position of the crankshaft or sensing the actual position of the piston. As discussed herein, PCCI system 10 may include various components for optimizing the combustion event. The objectives of the present system, i.e. low oxides of nitrogen (NOx) emissions, high efficiency, etc, may be achieved usmg any one of the various control components, or any combination of the components. In particular, as shown in Fig. lb, a compressor 24 may be provided along an mtake air system 23 upstream of intake manifold 15 for varying the boost intake pressure. Compressor 24 may be driven by any conventional means, such as an exhaust gas driven turbine 25. A bypass circuit 33 including a waste gate valve 43 may be provided in a conventional manner. A second compressor or supercharger 58 may be provided upstream of compressor 24. Supercharger 58 is mechanically driven by the engine drive system. A charge air cooler 28 may also be provided downstream of compressor 24. Also, an intake air heater 30 (such as a burner, heat exchanger or an electric heater) may be provided, for example, after cooler 28 as shown in Fig. lb, or alternatively, upstream of compressor 24. Also, an individual heater 29 may be provided in the intake port 26 associated with each cylinder 12 to provide quicker control of the intake manifold temperaUire for each cylinder to enhaaceboth individual cylinder combustion control and balancing of combustion between the cylinders. Compressor 24, cooler 28 and heater 30 each include control devices for varying the effect of the particular component on the pressure/temperamre of the intake ak or mixture. For example, a bypass valve or waste gate 43 could be used to regulate die amount of exhaust gas supplied from the associated exhaust system, which is connected to an exhaust duct 31, to mrbine 25 thereby varying the intake pressure as desired. Similarly, a control valve could be provided m the cooling fluid flow path supplied to cooler 28 to permit variable control of the cooling effect of cooler 28. Likewise, various types of variable controls could be used to vary the heating effect of heater 30. Output signals 22 from ECU 20 are supplied to the various control devices to control compressor 24, cooler 28 and heater 30 so as to variably control the pressure and temperature of the intake air or mixmre preferably on a cycle-by-cycle basis. In addition, the PCCI system 10 may include a plurality of fiiel supplies 32 and 34 for supplying fuels having different autoignition properties (for example, different octane or methane ratings, or activation energy levels) into the intake air flow. Fuel control valves 39 and 41 are used to control the amount of each fuel supply 32, 34 delivered, respectively. For example, fuel may be supplied along the intake air path between cooler 28 and air heater 30 as shown in Fig. lb. Of course, fuel could be introduced at various locations along the intake of the engine, such as upstream of the cooler, e.g. upstream of the compressor. Alternatively, the fuel could be injected, by for example an injector 35, into the respective intake duct 26 associated with each cylinder, as shown in Fig. la. The present PCCI system 10 also importantly includes a variable compression ratio means 38 for varying the compression ratio so as to advantageously advance or retard the combustion event as desired. For example, variable compression ratio means 38 may be in the form of a control mechanism for varying the shape of die combustion chamber or height of the piston to vary die effective compression ratio. The effective compression ratio could also be varied by varying the timing of closmg of mtake valve 19 as discussed more fully hereinbelow. The variations m the timmg of opening and closing of the intake and exhaust valves may be accomplished usmg any conventional variable valve timing actuator system capable of receiving signals from ECU 20 and effectively varying the opening and/or closing of the valves in accordance with the principles set forth hereinbelow. In addition, in-cylinder diluent injection may be accomplished usmg an mjector 40 for injecting a gas or liquid, e.g. air, nitrogen, carbon dioxide, exhaust gas, water, etc., into the cylinder to vary the temperamre and the temperamre distribution in the cylinder so as to control the combustion event. Similarly, a dUuent may be injected into intake duct 26 using, for example, an injector 42. The present PCCI system may also include a fuel injector 36 for injecting fuel 37, e.g. diesel fuel, directly into the combustion chamber. Fuel 37 would be injected either early in the compression event, preferably approximately between 180 degrees and 60 degrees BTDC, as described below, or later in the compression event near TDC. By injecting the fuel 37 early in the compression event, it is much more thoroughly mixed with the fuel/air mixmre received from the mtake duct than would be the case for a diesel engine, thus ensuring a more desirable combustion process, in particular the fiiel will burn at a leaner equivalence ratio which results in much lower NOx emissions. The start or initiation of the combustion (SOC) of the fuel/air mixmre received from the intake duct may be varied by controlling the quantity of fuel 37 injected. For instance, an earlier combustion event may be achieved by increasing the quantity of fuel 37 while the timing of the combustion event may be delayed by decreasing die quantity of fuel 37 injected. By injecting the fuel 37 later in die compression stroke, that is near TDC, conventional diesel fuel injection systems can be used. This approach could be combined with the introduction of one or more additional types of fuel in the intake manifold to achieve a PCCI mode of operation. In particular, die fuel injected into the intake manifold would have a higher excess an- ratio. The excess air ratio is the acmal air-fuel ratio of die engine divided by the air-fuel ratio at stoichiometric conditions. For the very lean excess air ratio, combustion along a flame front is impossible. However, autoignition is possible thereby allowing combustion of a mixture that would be too lean to bum in a typical spark-ignited engine. Applicants have determined that PCCI combustion does not initiate at, and propagate out from, a single location. On die contrary, die results show diat combustion includes multiple ignition sites distributed diroughout the combustion chamber. For efficient, low emission PCCI combustion, it is important to have combustion occur during an appropriate crank angle range during the engine cycle. If combustion starts too early, cylinder pressures will be excessively high and efficiency will suffer. If combustion starts too late, then combustion will be incomplete resulting in poor HC emissions, poor efficiency, high carbon monoxide (CO) emissions, and poor stability. Applicants have determined that die timing of die SOC and the combustion rate, and dierefore combustion duration, in a PCCI engine primarily depend on the temperature history; the pressure history; fuel autoignition properties, e.g. octane/methane rating or activation energy, and trapped cylinder charge ak composition (oxygen content, EGR, humidity, equivalence ratio etc.). The present invention presents a strucmred approach to affecting diese variables in such a way diat the start of combustion and/or die combustion rate (heat release rate) can be controlled through various combmations of feamres discussed more fully hereinbelow. The various control feamres for controUmg the start of combustion and the combustion rate are controlled/varied to ensure optimum combustion throughout engine operating conditions so as to achieve low NOx emissions and high efficiency. Application of these control features will cause combustion to occur withm a preferred crank angle range relative to the top dead center position of the engine piston. Specifically, applicants have recognized that substantially all of the combustion event should occur between 20 crank angle degrees BTDC and 35 crank angle degrees ATDC. Also, combustion would be initiated, preferably between 20 crank angle degrees BTDC and 10 crank angle degrees ATDC, and ideally, approxunately between 10 degrees BTDC and 5 degrees ATDC. In addition, the duration of the combustion event will typically correspond to a crank angle in the range of 5 - 30 crank angle degrees. Preferably, however, one or more of the control features listed below will be controlled to prolong the duration of combustion to approximately 30-40 degrees to achieve desirable peak cylinder pressures and reduced noise. Thus, optimal control of one or more of the following features will effectively control the start of combustion and/or the rate of combustion such that substantially all of die combustion event occurs between 20 crank angle degrees BTDC and 35 crank angle degrees ATDC. Of course, there may be conditions under which the start of combustion occurs outside the above-stated crank angle range and/or die duration of combustion m die PCCI engine occurs over a broader crank angle range, or may extend beyond die limit described above. Applicants have shown diat stable, efficient PCCI combustion can be achieved widi most of die heat release occurring after TDC. For example, as shown in Fig. 2, die centroid of heat release may be positioned at 5° ATDC. Applicant have determined diat, at light load and lean conditions, as shown in Fig. 3, heat release duration may be in die range of approximately 21.5-25 crank angle degrees. As shown in Figs. 4a, 4b and 4c, applicants have determined that with an engine running close to its misfire limit, the SOC and end of combustion (EOC) progressively retard and heat release duration lengthens. Gross indicated mean effective pressure (GIMEP) passes through a maximum as the SOC retards to after TDC. Meanwhile, the knock intensity and peak cylinder pressure (PCP) decrease substantially close to die misfire limit, while GIMEP remams acceptable. As shown in Fig. 5, the peak heat release rate also decreases and the heat release duration increases as the misfire limit is approached. Meanwhile, as shown in Fig. 6, the peak cylinder pressure decreases as the heat release rate retards. Clearly, the engine cannot sustain this reaction process without providing the appropriate controls discussed herem. Applicants have determined that the best operating pomt occurs widi the SOC occurring a few degrees after TDC. Certainly, improvmg the PCP-GIMEP tradeoff for PCCI combustion requires a SOC after TDC. As a result, it is clear that variable, active control is necessary to maintam the SOC and duration of combustion at the desired location and at the desired length, respectively, to achieve effective, efficient PCCI combustion. Variation in the SOC, between sequential combustion events in a single cylinder engine and between cylinders in a multi-cylinder engine, is due to the sensitivity of PCCI combustion to the pressure and temperature history leading up to the particular combustion event. Very small variations in the compression ratio, the amount of trapped residual, wall temperatures, etc. have a significant effect on the pressure and temperature history. The present PCCI engine and method of operating the engine include control variables/features capable of compensating for, and controlling, these variations to achieve optimum PCCI combustion. Generally, the control variables, which can be used to effectively control the commencement of combustion and the combustion rate so as to ensure that substantially all of the combustion process occurs within the optimal crank angle limit, i.e. 20 degrees BTDC through 35 degrees ATDC while minimizing emissions and maximizing efficiency, may be classified in four categories of control: temperature control; pressure control; control of the mixture"s autoignition characteristic; and equivalence ratio control. Temperature Control The temperature of the in-cylinder air/fuel mixture (m-cylinder temperature) plays an important role in determining the start of combustion. The in-cylinder temperature may be varied to control the start of combustion by varying certain key control features, such as compression ratio (CR), intake manifold temperature (IMT), exhaust gas recirculation (EGR), residual mass fraction (RMF), heat transfer and temperature stratification. Applicants have determined that intake manifold temperamre (IMT) has a significant effect on propane-fiieled PCCI combustion. During two of Applicants" studies, engine speed, equivalence ratio ((j)) and mtake manifold pressure (IMP) were held constant while IMT was swept through the practical operating range. The lowest IMT was limited by unstable operation and the highest IMT was limited by maximum allowable peak cylinder pressure (PCP). The conditions of the first and second studies, respectively, included engine speed = 1200 rpm and 2000 rpm; equivalence ratio = .30 and .24; and IMP = 3.3 bar and 4.1 bar. As shown in Figs. 7a and 7b, increasing IMT resulted in increased GIMEP and a decreased coefficient of variation (CoV) of GIMEP. Also, increasmg IMT increased the PCP as shown in Fig. 7c, while advancing the SOC and decreasing combustion duration (Figs. 7d-7f). Increasmg IMT also increased gross indicated thermal efficiency (Fig. 7g) and the estimated noise (Fig. 7k). With respect to emissions, increasing IMT decreased FSHC emissions (Fig. 7h), decreased fuel specific carbon monoxide (FSCO) emissions (Fig. 7i), but had no observable effect on FSNOx (Fig. 7j). In summary. Applicants have determined that small changes in IMT have large effects on many aspects of propane-fueled PCCI combustion. By varying the mtake temperamre, the combustion event can be advanced or retarded. Increasing the intake temperature will advance the start of combustion; decreasing the intake temperamre will retard the start of combustion, as shown graphically in Fig. 8. This temperature control may be accomplished using heat exchangers or burners. For example, a charge air cooler may be positioned along the intake manifold. A burner or heater m combination with a cooler offers exceptional intake temperamre control. The exhaust products of the burner may be directly mixed with the intake air, the burner could use the intake air directly for its an: supply, or the heat generated by the burner could be added to the intake air through a heat exchanger. The heat exchanger may use waste heat in engine coolant or exhaust gases to heat the intake an:. Also, rapid control of IMT can be achieved by using a charge air cooler bypass. For example. Fig. 52 discloses a system 65 to control intake air temperature in an engine 66 at many operating conditions. One or more valves, such as a flapper valve 67, are used to control the quantity of hot intake air and quantity of cooled intake air mixed downstream of the flapper valve. An intercooler 68, positioned in one of the intake air flow paths upstream of the flapper valve, provides the cooling effect. Also, an electric heater may be provided to permit optunum control at other conditions. This system permits effective control over the SOC by controlling the intake an: temperamre. The flapper valve may alternatively be positioned upstream of the intercooler at junction 69, or two valves may be used with each valve positioned in a respective one of the cooled intake passage 71 (either upstream or downstream of the intercooler) and the bypass intake passage 73. Any arrangement of one or more valves capable of variably controlling the flow through one or both of the passages may be provided. A regenerator (similar to that used in a Stirling engine) could be used to recover and transfer exhaust heat into the intake air through a heat exchanger thereby controlling the intake temperature. In addition, IMT could be varied by injecting fuel into the manifold in different phases, e.g. as a liquid or a gas. The change in the heat required for vaporization of a liquid fiiel would reduce IMT. Of course, different types of fiiels would have different effects on IMT. Applicants have also determined how residual and intake temperature, boost and combustion chamber and port wall heat transfer, affect in-cylinder bulk temperature throughout intake and compression, and also the effect on spatial temperature distribution at TDC. Specifically, Applicants compared die intake and compression events for an engine running on an air and propane mixture. Applicants determined that the temperature at the SOC is also determined in part by the reheating of the intake charge by existing heat energy. For the purposes of this application, reheat is defined as T(average in-cylinder @ intake valve closing (IVC)) - T(average intake manifold), that is, the difference between intake manifold temperature, i.e. temperature assigned at the inlet to the port and the in-cylinder bulk temperamre at IVC. Applicants determined that reheat starts in the port and continues in-cylmder. Moreover, 56% of reheat was due to wall heat transfer and 44% due to mixing and boost for the condition exammed. Clearly, heat transfer is very important in determining reheat. One smdy that elucidates the importance of the wall temperamres on the in-cylmder heat transfer is the following. In comparmg the fu-ing cylinder to the misfiring cylinder, it was noted that the misfiring cylinder"s reheat was 63% of the firing case (27 vs 43 K). Lower wall temperatures for a misfiring cylinder compared to a firing cylinder are the mam reason for its lower in-cylmder temperamres. The firing cylmder had a TDC m-cylmder temperamre 46 K higher than a misfiring cylinder, compared to a 16 K higher temperamre at IVC. If compression were done adiabatically for each case, the temperamre difference at TDC would have been ~ 35 K given the initial 16 K difference. Therefore, ~ 11 K (46-35 K) temperamre loss from IVC to TDC is due to cooler misfirmg wall temperatures. Interestingly, although walls heat the in-cylmder gases for the majority of the mtake and compression event, relatively fast rates of heat transfer out of the gas near TDC compression can result in cooler m-cylmder contents than if there were no heat transfer at all. Also, mass flow rate decreased 7.5% due to heat transfer when comparing a normally firing cylinder with wall heat transfer to a firing cylinder with adiabatic walls, primarily due to the density effect. Referring to Fig. 9, with respect to the effect of wall temperamres, i.e. piston temperamre, head temperamre, and liner temperamre, on the SOC, Applicants have determined that as wall temperamres are increased, SOC becomes more advanced. The increased surface temperamres cause lower heat transfer to the combustion chamber surfaces thereby advancing combustion. Applicants have shown that with wall temperamre varying from 255 to 933 K and all other parameters kept constant (IMT = 342 K, reheat = 43 K, (j) = 0.24), the mixture did not ignite with a wall temperamre below 400 K. From about 400 K to 550 K combustion duration mcreases as a larger percent of the fuel bums. Above 550 K all the fuel burns and the combustion duration decreases with increasing temperamre. Varying in-cylinder surface temperamres can be achieved by varying the cooling effect of the engine coolant and/or the lubricating oil on the cylmder/ piston assembly. Although cylinder wall temperamre may be difficult to use as a lever for effectively controlling SOC, cylinder wall temperamres are one of the parameters considered when controlling SOC, particularly for starting or transient operation. Applicants have shown that there is a region of operating conditions where there are two stable solutions: one without combustion and cool walls, and one with combustion and hot walls. Also, varying the surface to volume ratio in the combustion chamber can change the heat transfer and, therefore, can be used to control the combustion. Applicant"s studies show that the optimum operation for a PCCI engine is one where timing (SOC and HOC) are very retarded (as retarded as possible while still burning most of the fuel) compared to typical heat release rates tiiat have been reported for PCCI engines. When operating in such a retarded mode, this results in lower peak cylmder pressures, better ISFC, lower noise and lower heat transfer. Applicant has determined that stable PCCI operation is easily achieved for advanced timings. Applicant has also determined that by using closed loop control, combustion can be significantly retarded past the stable region. As discussed herein, the closed loop control may be used to control one or more of a variety of variables, for example IMT, equivalence ratio, EGR rate, etc. This is possible because the SOC is strongly controlled by temperamre. The temperature, in mrn, is very sensitive to in-cyltnder heat transfer, which is very sensitive to the wall temperatures. Retarding the timmg lowers the wall temperatures. As timing retards enough, the engine can no longer be operated without active controls. A key concept is that the time constant associated with the thermal inertia of the walls is responsible for the rate at which an instability progresses, and therefore dictates the time constant required for the active control. The time constant of the wall temperature is on die order of many seconds, allowing control of SOC by adjusting one or more of a variety of variables including IMT equivalence ratio, EGR rate, etc. For example. Applicant has determined that IMT can be actively adjusted quickly enough to advantageously operate the engine at a retarded timing which would otherwise result in an unstable condition without the active IMT control. Docket: 0270-2486 By con^paring a normally firing cylinder with wall heat transfer to a firing cylinder with adiabatic walls, wall heat transfer is seen to be the major contributor to spatial temperature distribution at TDC. Spatial temperature distribution is defined as the manner in which the temperature varies throughout a region, be it in the port, or in the cylinder at a particular crank angle. By varying the in-cylinder temperature distribution, the start of combustion and/or the overall combustion rate can be positively affected. One way to vary in-cylinder temperature distribution is to use split intake ports arranged so that some of the incoming air/fuel mixture is warmer/colder than the rest of the incoming mixture. Another method is to introduce hot spots in the cylinder or to use a glow plug 44 (Fig. la). Also, in-cylinder temperature distribution may be controlled by varying the temperature of the combustion chamber walls (e.g. the wall temperature of the cylinder Imer, piston and/or engine head) by varying, for example, the temperature of the engme coolant, the temperamre of the engine oil or die rate of cooling of the combustion chamber walls. As shown in Fig. lb, the temperature of the engine coolant may be varied by controlling the flow through a coolant heat exchanger 46 positioned in the engine coolant circuit 47 by varying the flow tiirough a bypass circuit 48 using a bypass valve 50. It was determined that wall heat transfer has similar impact on spatial temperature distribution for both firing and misfiring cylinders. Similarly, applicants also determined how residual temperature and wall heat transfer affect m-cylinder temperature distribution throughout intake and compression. The determination included three studies of the intake and compression events of an air and propane mixture. These studies revealed that, during most of intake and compression, hot residual is the main source of spatial temperamre variation. However, near TDC compression, residual history is of minor miportance compared to heat transfer with the walls in setting up temperature variations in the combustion chamber. As a result, it is believed that to promote a combustion event that uses more of the fuel that is available, fuel may be introduced m such a way that at SOC, fuel and air exist in proper proportion in regions where the temperature field is adequate to sustain combustion. Two areas where the temperamre field is inadequate to sustain combustion are in the crevices and adjacent cooled surfaces. It is therefore desirable to keep the fiiel away fi-om both the crevices and cooled surfaces. Clearly, heat transfer into the in-cylinder mixmre increases the temperature of the in-cylmder mixture thus advancmg SOC. Applicants have shown that a glow plug can be used to effectively control the SOC to a small degree. As shown in Fig. 10, once the glow plug is turned off, die SOC and HOC retard slightly. Also, GIMEP decreases significantly since less ftiel is being burned. The decrease in the amount of fiiel being burned also results in a decrease in the heat release rate as shown in Fig. 11. Between cycles #1 and #100, the glow plug was mmed off and remained off until a time between cycles #300 and #400, at which pomt it was mrned back on. Perhaps most importantly, when the glow plug is turned off, the start of rapid combustion is significantly delayed without an increase in duration, which in combination with die decrease in heat release rate, causes the cumulative heat release to decrease. Thus, glow plug 44 (Fig. lb) could be used to positively control combustion to a limited degree. In any practical reciprocating engine, heat will be lost from the combustion chamber durmg the compression process. The heat loss depends upon many factors, but primarily upon engine speed and the temperature difference between inside and the outside of the cylinder. This heat transfer during die compression process becomes a problem for diesel engines during cold ambient starts as combustion can be difficult to initiate and sustain in cylinders where the combustion chamber surfaces are cold. Typically, die cylinders located at the ends of each bank of cylmders run the coldest and are die least likely to fire. It is quite common under such conditions for the charge in die end cylinders to fail to combust due to excessive heat exchange widi die colder cylinder walls. With diesel engines, however, once all die cylinders warm up, combustion is quite consistent and much less dependent on combustion chamber surface temperatures. With PCCI, the combustion process is initiated by obtaining a certain pressure and temperature "history". Thus, as discussed hereinabove, the PCCI combustion process is strongly dependent upon, and sensitive to, the surface temperatures of the combustion chamber. The present PCCI engine may include an end cylinder compensating means for achieving desired combustion chamber surface temperatures in the end cylinders to ensure better cylinder-to-cylinder temperature control thereby increasing the likelihood of stable combustion and very low NOx emissions. The end cylinder compensating means may include a system for reducing the effective cooling of specific cylinders, such as reducing piston cooling nozzle flow; mcreasing coolant temperature; or reducing coolant flow rate. Specifically, referring to Fig. 12, the end cylinder compensating means may include an oil flow control system 70 including oil flow control valves 72 positioned in branch flow passages 74 delivering cooling oil to piston cooling nozzles 76 from an oil pump 78. Thus, control valves 72 can be controlled to vary the flow of cooling oil to the piston assemblies to vary the temperature of the piston and thus favorably influence the in-cylinder temperature. Alternatively, flow restrictions could be used instead of valves 72, or the nozzles 76 associated with the end cylinders may be designed witii a smaller effective flow area than the remaining nozzles to permanently reduce the flow to these piston cooling nozzles. In addition, if more than one nozzle 76 is provided as shown in Fig. la, die number of nozzles operating could be varied by controlling the respective control valves associated witii each nozzle. Referring to Fig. 13, end cylmder compensating means may include an engine coolant flow control system 80 including a coolant pump 81 and coolant flow control valves or restrictions 82 positioned in branch passages 84 leading to tiie end cylinders 86 Docket: 0270-2486 of the engine 88. The valves 82 are operated to reduce the flow of cold coolant delivered from a radiator 90. Also, control valves 92, positioned in hot coolant return passages 94, are used to control the flow of higher temperature coolant, bj^assing radiator 90, and delivered directly to the end cylinders. These systems all function to control the flow of coolant to the end cylinders to compensate for the fact that they are cooled more by the ambient surroundings so that the total cooling to each end cylinder is equal to each of the other cylinders. These systems can be used to assist in cylinder warm-up to improve engine startability and to provide enhanced control of cylinder combustion and cylinder-to-cylinder balancmg. The end cylmder compensating means may, alternatively, or additionally, include end cylmders havmg an effective compression ratio nominally greater than the other cylinders to offset the extra heat loss. This compression ratio could be designed mto the end cylinders so that the end cylinder compression temperatore is equal to the middle cylmders. This approach is advantageous from a performance perspective smce end cylinder combustion chamber surface temperamres would be enhanced for both start-up as well as warmed-up operation. This compression ratio difference may alternatively be accomplished through the camshaft valve lobe phasing. In this scenario, the end cylinders would have intake valve closmg (IVC) near bottom dead center (BDC) so tiiat the effective compression ratio (CR) is approximately equal to die geometric CR. The middle cylinders could then have a retarded IVC which would produce a lower nommal effective CR than the end cylinders. The effect of varying the compression ratio on PCCI combustion is discussed more fully hereinbelow. One of the biggest challenges with premixed charge, compression ignition (PCCI) engine technology is in the placement of the heat release profile. Start of combustion with standard diesel or spark ignition engines is controlled with injection timing or spark timing. With PCCI engines, the start of combustion is dictated by the in-cylinder temperatures and pressures. As SOC timings near TDC (and after) are approached on the PCCI engine, the sensitivity to small geometric and/or operational variations in temperatures, pressures, etc. increase dramatically. As retarded heat release profiles are sought for PCCI engmes (to minimize peak cylmder pressures and improve efficiency), the risk of misfure or partial burn increases dramatically. This is due to the fact that the cylinder temperamres decrease after top dead center due to the expansion of the charge. If autoignition has not yet occurred by TDC, autoignition will not likely occur much after top dead center. This problem is fiirther aggravated if one cylinder begins to misfire. The misfiring cylinder cools down making it even more likely that the misfiring will continue. In a multi-cylinder engine variations inevitably exist between cylinders with respect to compression ratio, wall temperatures, reheat and residual mass fraction. This variability makes it quite difficult to operate a PCCI engme with the desired retarded combustion timing while maintaining optunum combustion without having individual cylinders (which happen to be runnmg slightly cool) begin to misfne. Applicants have determined that manipulating valve events can have a significant effect on the temperamre at TDC and therefore is an effective tool for controllmg the start of combustion as suggested by analytical results shown in Fig. 14. Specifically, referrmg to Table I, varymg valve events has the foliowmg effects: TABLE I modiHed event baseline effect of advancing valve timing relative to baseline effect of retarding valve timing relative to baseline EVC -357° traps hot residual which advances SOC exhaust blown back into intake which advances SOC EVO 135° no effect no effect IVC -167° Miller cycle - lowers effective CR which retards SOC at these particular conditions, retarding slightly unproves breathing; retarding further reduces effective CR which retards SOC IVO 341° allow hot exhaust to flow into intake which advances SOC restricts flow from intake manifold which has minimal effect on SOC As shown in Fig. 15, exhaust valve closing (EVC) plays a significant role in determining the amount of combustion products that remain in, or are made available to, the combustion chamber from one combustion event to the next event, i.e. the residual mass fraction (RMF). The residual exists at a higher temperature than the incoming charge and therefore heats the charge for the next combustion event. Thus, the timing of exhaust valve closing can be used to adjust the in-cylinder temperature and therefore controlling the SOC. In order to "heat up" a cold cylinder (e.g. one that is beginning to mis&e) die residual mass fraction can be increased in die mdividual cylinder by an early exhaust valve closing event. These hot residuals will increase the reheat of the incoming charge and tend to advance the start of combustion thereby, for example, restoring a misfiring cylinder. As shown in Fig. 15, advancing EVC traps hot residual in die cylinder while retarding EVC allows hot exhaust to be blown back into the cylinder (in diis case, Docket: 0270-2486 exhaust manifold pressure (EMP) > IMP). The baseline EVC is the optimum of these two effects: trapping the mmimum amount of residual and resulting in the lowest TDC temperature. Similarly, advancing IVO allows some of die hot residual in the cylmder to be blown back in to the mtake, again because EMP > IMP, causing the TDC temperature to increase. Lowering compression ratio, discussed more fully hereinbelow, by, for example, advancmg IVC, will also mcrease residual in the cylinder, but to a lesser extent. Adjusting the timing of exhaust valve closing may also be used to effectively compensate for the small geometric and operational variations between the cylinders to permit the engine to be "tuned" cylinder-to-cylinder. Any other means for effectively increasing or decreasmg the RMF may be used to advance or retard the SOC, respectively. One method of implementing this strategy has been successfully tested on a multi-cylinder PCCI engine. This technique involved the increase of the exhaust valve lash setting. Opening up the lash effectively closes the exhaust valve early and advances the start of combustion as desired. Applicants have determined that reducing the exhaust valve event by 10 degrees leads to slightly higher surface temperamres and 22 degree warmer inlet temperatures. Given the dramatic effect that 22 degree IMT swings have on combustion (Figs. 7c-7f), this method would indicate a potential for mning the multi-cylinder engine with valve lash adjustments. As shown in Fig. 16, shortening the duration diat an exhaust valve is open by increasmg the lash does mdeed advance combustion. Ultimately, cylinder-to-cylmder variations can be controlled passively by any means which can adjust the static exhaust valve closing. It could also be controlled actively if it is coupled with some diagnostic measurements. If control exists on all cylinders then diis technique could also be used to effect the overall start of combustion widiin the engine. Controlling SOC in a PCCI engine during a transient can be challenging. In another embodiment, a throttle (or odier restriction) is placed in the exhaust system. In a situation where SOC must be quickly advanced, the throttle is closed, resulting ui an increased residual mass fraction and, therefore, in higher m-cyliuder temperamres thereby advancing SOC. Of course, this effect could also be achieved with variable exhaust valve timing. Applicant has shown that restricting the exhaust m this way will advance SOC. However, this control scheme also adversely affects BSFC. In this embodiment, the exhaust is only restricted for a short period during a transient until another control mechanism, e.g. an intake air heater, has a chance to control SOC in a way that does not adversely affect BSFC as much. The advantage of this method is diat it is very fast and easy to unplement. Effective methods and mechanisms for introducing EGR into a non-throtded engine are currently difficult to unplement. The present embodunent introduces a simple method of efficiendy introducing EGR through late exhaust valve closing. By closing the exhaust valve or valves later in a given cycle, the EGR is essentially "internal" EGR since the method increases the residual mass fraction (RMF). The present embodiment could easily be applied to any four stroke engine requiring EGR but is especially effective when applied to a PCCI engine. A PCCI engine requires much higher intake temperatures at low load as compared to high load operation. The present embodiment would allow the engine to operate with larger amounts of internal EGR at low load. This operation would increase the charge temperamre similar to mcreasmg the intake temperature. A closed loop control may be used to adjust the amount of internal EGR to control the start of combustion (or combustion history) of any or all cylinders. In order to control die amount of internal EGR, the closmg of the exhaust valve or valves is delayed in a variable manner as controlled by, for example, the electronic control module of die engine. A comparison of a normal exhaust event with a "late exhaust valve closing" event is helpful in understanding die present embodiment. In a normal exhaust event, most of die combustion products are expelled during die exhaust stroke. At diis pomt, the exhaust valve closes and the intake valve opens allowing fresh an: or an air/fuel mixture to fill most of the combustion chamber as the piston moves downwardly. In the late exhaust valve closing event of die present embodiment, the exhaust valve stays open through part of the intake stroke of the piston. As a result, the engme draws m bodi fresh air (or mixture) and combustion products. By varymg the delay of closing, the amount of internal EGR can be adjusted. Additional mternal EGR could be achieved by delaymg the openmg of the intake valve. The advantage of the present embodiment over early exhaust valve closing is that early exhaust valve closing results in very poor PMEP, while Applicants" studies indicate the late exhaust valve closing does not significantly affect PMEP. Another advantage of the late exhaust valve closing of the present embodiment is that the exhaust valve need not be closed very quickly. That is, the exhaust valve must only be kept open for a longer period of time. Referring to Figs. 65a-65d, the method of the present invention is illustrated and explained as follows. At the beginning of the exhaust stroke of die piston 500, an exhaust valve 502 is open while an intake valve 504 is closed. As shown in Fig. 65b, at die top dead center position of the exhaust stroke of piston 500, bodi exhaust and mtake valves 502, 504, respectively, are open. As piston 500 begins to move downwardly tiirough die intake stroke as shown in Fig. 65c, a fresh charge of mtake air (or fuel/ah mixmre) is drawn mto die combustion chamber dirough the opening of intake valve 504 while combustion products are drawn mto the combustion chamber through die openmg of exhaust valve 502. Referring to Fig. 65d, later in die intake stroke, the exhaust valve 502 closes while a fresh mixmre of ftiel and air is drawn dirough die opening of intake valve 504. Referring to Figs. 66 and 67, a first embodiment of die mechanism for achieving die delayed exhaust valve closing is illustrated. In diis embodiment, die engine includes intake and exhaust cam lobes much like a "typical" engine. This embodiment prevents the exhaust valve from closing during the retraction part of the cam. Specifically, a cam lobe 510, push rod 512 and rocker lever 514 are used to operate exhaust valve 502. In addition, a hydraulic link 516 is positioned along the valve tram between cam lobe 510 and rocker lever 514, such as at the connection between push rod 512 and rocker lever 514. During engine operation that does not require late exhaust valve closing, the hydraulic link 516 is operating in a "collapsed" state. When late (or delayed) exhaust valve closing is desired, hydraulic link 516 fills with oil as cam lobe 510 allows the retraction of push rod 512. As a result, exhaust valve 502 remains in the open position. At a predetermined time during the cycle as determined by the ECM, when exhaust valve closing is desired, oil is allowed to bleed out of hydraulic link 516. The oil is pushed through an orifice which controls die seating velocity of exhaust valve 502. Fig. 67 illustrates the details of hydraulic link 516. The hydraulic link system includes a solenoid valve 518 positioned along a drain passage 520 extending from a hydraulic link chamber 522 within which push rod 512 is positioned. Of course, alternatively a plunger for positioning in chamber 522, that is connected to push rod 512, may be used. Solenoid 518 permits die controlled flow of fluid from chamber 522 through an orifice to a low pressure drain. The hydraulic link system also includes a solenoid valve 524 positioned along a pressurized oil supply passage 526 for controlling the flow of pressurized oil to chamber 522. During operation, for early (i.e. normal) exhaust valve closing, solenoid valve 524 remains closed while solenoid valve 518 remains open. For late exhaust valve closing, solenoid valve 524 opens while solenoid valve 518 closes during or after the opening of exhaust valve 502. Once the cam causes push rod 512 to retract, the oil flowing into chamber 522 from solenoid 524 causes chamber 522 to fill with oil, preventing the exhaust valve from closing. When the exhaust valve closing is desired, solenoid 518 opens to bleed down chamber 522. Fig. 68 discloses a second possible embodiment for achieving late exhaust valve closmg. This system also includes a push rod 600 for operating a rocker lever 602 which operates exhaust valve 502. Also, a spring 604 is used to maintam an operating connection between rocker lever 602 and push rod 600. In this embodiment, however, a chamber 606 is positioned for receiving the upper end of exhaust valve 502 or, alternatively, a plunger coimected to exhaust valve 502. An oil supply passage 608, which includes a check valve 610, is coimected to chamber 606 for supplying pressurized oil to the chamber. A second passage 612 is connected to chamber 606 and includes a solenoid valve 614 positioned along the passage for controlling the flow through passage 612 from chamber 606. During operation, exhaust valve 502 is opened by rocker lever 602. As exhaust valve 502 opens, hydraulic chamber 606 is filled with oil flowing from passage 608. Check valve 610 maintains the oil in chamber 606 even when rocker lever 602 retracts. Solenoid valve 614 opens to allow exhaust valve 502 to return to the closed position under the bias force of a valve spring (not shown). If early closmg of the exhaust valve is desired, solenoid valve 614 remams open. A restriction 616 may be positioned downstream of solenoid valve 614 in passage 612 to control the seating velocity of exhaust valve 502. Alternatively, the bleed down passage 612 can be positioned in such a way that it slows down the seating velocity of the valve to achieve a gentie seating into the closed position. Another method of controlling in-cylinder temperature by controlling the residual mass fraction (RMF) is to compress a pocket of residual gas from the previous cycle in a chamber positioned separate from the mcoming charge. The proportion of trapped residual to fresh charge can be manipulated by the size of such a chamber. The mass of hot exhaust could be as large as (1/2)(1/CR) and therefore = 1/30 of the chamber mass if all the IDC volume is in such a chamber. The structure of such a chamber will have to be managed to make at least a portion of the hot gas survive the compression process without completely mixing with the incoming charge. If the trapped exhaust is mixed very early in the compression process, the high temperature required to initiate the fast reactions will not be reached. The timing of flows into and out of such a chamber may help manage the tuning of the beginning of rapid energy release in cylinder. Additional sources of local heat input may be able to supply such a fast reaction initiation. This might be a heated glow plug or a thermally isolated mass. The residual mass fraction is also sensitive to the exhaust manifold back pressure (EMP). By increasing EMP relative to IMP, the residual mass fraction can be increased thus increasing the temperamre of the charge which, in mm, advances combustion. Applicants have determined that raising EMP does have the expected result of advancing SOC. However, applicant also showed that SOC advanced only by about 4° with a 3 bar increase in EMP for a four cycle engme. Applicants have determined tiiat the increase in temperature is nearly linear with increase in EMP, with all other things being held constant. For a 1 bar increase in EMP, temperamre at TDC increased about 10 K. Therefore, considering the practical range of EMP, controlling EMP seems to be a relatively weak lever in controlling SOC on a four cycle engine. Moreover, a very substantial BSFC penalty is paid when using EMP to increase TDC temperamre in a four cycle engine. The BSFC would be significantly higher dian usmg either exhaust valve closing or variable compression ratio. Although the effect of increasing EMP is the same as advancmg EVC, i.e. trapping more hot residual mass in the cylinder, the BSFC is much higher because, when EMP is mcreased, the piston has to work against that pressure through the entire exhaust stroke. If the engine has mrbomachinery, then further complications would arise with trying to use EMP to control SOC. However, using an exhaust restriction may still be viable on a two-cycle engine. Another important way to control intake temperature is by using hot exhaust gas recirculation (EGR). As shown in Fig. lb, a high pressure EGR circuit 54 may be used to direct hot exhaust gas from upstream of mrbine 25 mto the mtake system 23. EGR circuit 54 includes a high pressure EGR control valve 60 for controlling the recirculation of exhaust gas. A low pressure EGR circuit 62 and control valve 64 may be used to direct a flow of low pressure EGR from downstream of turbine 25 into the intake system 23. Applicants have shown that EGR is especially effective m increasing the intake manifold temperamre when introduced upstream of the compressor 24 (assuming the effect of adding EGR is not canceled by additional charge air cooling). Exhaust gas recurculation (EGR) has more utility in PCCI engines because the exhaust gas of such an engine will contain less particulates and thus the exhaust gas can be recnculated to the ideal upstream location (intake of compressor of turbocharger). The intake of the compressor is the best location because the pressure differential is almost always favorable. The fresh intake air and hot EGR mixture will get compressed by the compressor thereby providing heating and mixing. By introducing the EGR upstream of the compressor and increasing the compressor inlet temperature, the result is a much higher compressor outlet temperature than if the EGR is introduced after the compressor. Introducing EGR into the intake of the compressor is very difficult in normal diesel engines because the particulates in the exhaust gases of the engine "gum up" the compressor. In a PCCI engine, however, the virtually particulate free exhaust could be introduced upstream of the compressor without significant problems. Also, as shown in Figs. 16, 17 and 18, applicants have determined that, regardless of die technique used to introduce exhaust products, e.g. EGR, RMF, etc., by adding exhaust products while maintaining the temperature of the charge by, for example, injecting a cooling diluent, such as air and/or water, the combustion rate can be slowed thus increasmg the combustion duration, retarding combustion and decreasing the amount of heat release. Referring to Fig. 19, an improved engine 100 is shown which benefits from the PCCI engine and control system of die present invention by operating a limited number of a plurality of cylinders in a PCCI mode while operating the remamder of the cylinders m a diesel mode. Specifically, for example, five cylinders 102 in a six cylinder engine may be operated in the diesel mode while one cylinder 104 is operated in a PCCI mode. This engine also includes an EGR system 106 associated only with die PCCI cylinder 104 and separate from an exhaust system 108 associated with the diesel cylinders 102. The pressure of the piston in the PCCI cylinder 104 is used to force the exhaust gas into the intake system. The EGR system 106 includes an EGR cooler 110 utilizing, for example, engine coolant, which cools PCCI exhaust gas before recirculating the gas to the upstream side of a compressor 105. Of course, the exhaust gas could be delivered to die intake manifold 112 serving only diesel cylinders 102. A well known problem confronted in die use of EGR in diesel engines is the excessive amounts of particulates and NOx present in diesel engine exhaust gas. The improved engine 100 permits a diesel engine to benefit firom EGR while substantially avoiding die drawbacks associated with heavy particulate diesel exhaust diereby providing a less complex and cosdy system. For example, as discussed hereinabove, die PCCI EGR from cylinder 104 could more easily be introduced upstream of die compressor without foulmg the compressor. Also, the low NOx emissions of die PCCI EGR reduce die formation nitric acid diereby reducing corrosion in the engine. Applicants have shown that the engine of Fig. 19 lowers die brake specific NOx emissions while only negligibly increasing the brake specific fuel consumption. Perhaps one of die most effective control feamres for varying the temperature at TDC and dierefore die SOC is variable control of die compression ratio (CR) of a cylmder. By varying the effective or the geometric compression ratio, both the temperature and the pressure histories can be controlled. Increasing the compression ratio advances the combustion event. Decreasing die compression ratio retards it. For certain purposes, the compression ratio may range from 24:1 (to promote cold starting) to 12:1 (to permit control over the start of combustion and limit the peak combustion pressures). The range of compression ratios would depend on, among other factors, the type of fuel used (more specifically its autoignition properties), for example, natural gas or propane. Applicants have determined the effect of compression ratio on PCCI combustion. For example, referring to Fig. 20, applicants have shown that varying the compression ratio is a large lever in changing in-cylinder temperature and therefore SOC. As shown in Fig. 21, applicants have shown that variations in compression ratio significantly affects the location of the SOC relative to TDC. The compression ratio can be varied by varying the geometric compression ratio, i.e. using a control mechanism to vary the physical dimensions/shape of the combustion chamber. The present mvention includes a compression ratio varying device 38 for varying the geometric or the effective volume of the combustion chamber during engine operation to achieve a desired SOC. The compression ratio varying device may be a mechanical device for causing compression heating of the charge near TDC by changmg the geometric volume of the combustion chamber. As shown in Figs. 22a-22d, the compression ratio varying device may include a movable auxiliary piston or plunger which moves to extend into the combustion chamber at a crank angle near TDC to decrease the combustion chamber volume thereby increasing die compression ratio and heatmg die charge sufficiently to allow ignition to start. The key function of die plunger is to displace some of die charge near TDC. Therefore, die shape and location of the plunger in die combustion chamber will not be critical to its function, except to die extent diat the plunger affects the crevice volume. The size of the plunger will be based on die desired compression ratio control range and may be estimated by the foliowmg example: Swept volume (displacement) per cylinder = 1,000 cc = 1 liter TDC clearance volume = 100 cc Compression ratio = (1000 cc + 100 cc)/100.0 cc = 11.0 If the plunger volume = 30 cc, then die effective compression ratio widi plunger fully extended = (1000 cc + 100 cc)/(100 cc - 30 cc) = 15.7. For a given set of conditions, the modified compression ratio should be sufficient to allow a large enough increase in temperature and pressure to cause compression ignition for a fuel/air mixture diat would not ignite without die plunger. Of course, the engine"s compression ratio and die size of die plunger are easily changed during the design stage of die engme. Also, different fiiels and intake temperatures could require different plunger sizes and compression ratios. As shown m Fig. 22a, die plunger 150 may be positioned in a bore 152 in die cylinder head 154 and operated by a cam 156 rotated in predetermined tuned relationship to the movement of the engine piston 158. A retraction spring 160 biases die plunger toward cam 156 to mcrease die size of combustion chamber 162. This particular arrangement is advantageous in diat cam driven plunger 150 can put work back into the camshaft as die plunger retracts. Also, some of the work diat plunger 150 does on the charge can be extracted by the engine piston, as long as plunger 150 does not retract until late in the expansion stroke, or after the expansion stroke. Alternatively, referring to Fig. 22b, a plunger 170 may be hydraulically operated by a pressurized supply of fluid, e.g. fuel, delivered to a chamber 174 by a hydraulic circuit 172 connected to, for example, a jerk pump or common rail system. Fig. 22c, illustrates another hydraulically actuated embodunent m which a plunger 180 is assisted by a spring 182, positioned in a chamber 184 formed adjacent one end of plunger 180, to allow energy to be stored in the spring. In this system a retaining mechanism, e.g. hydraulic, electromagnetic or mechanical, (not shown) maintains the plunger in the unextended position. When the piston is near TDC, a hydraulic fluid supply system 186 forces plunger 180 down (at this pomt the retaining system no longer holds the plunger). This downward motion is heavily assisted by spring 182. After combustion, plunger 180 moves back up recompressing spring 182 thereby remming energy to the spring. To optimize diis energy extraction process, the hydraulic chamber 184 bleeds down at a rate controlled by a valve 188. Fig. 22d illustrates yet another embodiment in which a spring 190, biasing a plunger 192 into the extended position, is strong enough to overcome the gas pressure in the combustion chamber before combustion. Near TDC, a bleed down valve 194, connecting a chamber 196 is opened and the spring 190 pushes plunger 192 into the extended position in the combustion chamber 162 causing the charge to ignite and the pressure in the combustion chamber 162 to increase. As a result, plunger 192 is pushed back up agamst sprmg 190. If needed, a high pressure supply 200 supplies hydraulic fluid to chamber 196 to ensure plunger 192 moves back up into the retracted position. If the gas pressure is sufficient to move the plunger back up into the retracted position, a low pressure hydraulic fill supply 202, including a one-way valve 204, may be used to fill the chamber 196 below plunger 192. The compression ratio may also be varied by providing an opposed piston engine design having variable phase shifting to permit the compression ratio to be varied during operation by changing the phase of rotation between two crankshafts. The opposed piston engme may be of die type disclosed m U.S. Patent No. 4,010,611 or of die interconnected cylinder type widi variable phasing as disclosed in U.S. Patent No. 4,955,328, die entire contents of both of these references being hereby incorporated by reference. Alternatively, referring to Fig. 23, the compression ratio could be varied using a phase shifting mechanism 210 including a conventional differential assembly 211 connected between an input shaft portion 212 of one of die crankshafts 214, 216 associated with respective pistons 218, 220 and an output shaft portion 222 of die same crankshaft 214 to permit die portions of the crankshaft to be rotatively shifted relative to one anodier. Crankshafts 214 and 216 are connected via a convendonal gear assembly 223 for transferring power to a driven shaft 225. As shown in Fig. 24, the differential 211 includes a ring gear 224 mounted on one end of input shaft pordon 212, an arm 226 extending from ring gear 224 and a gear assembly 227 mounted on the opposing ends of shafts portions 212, 222. A rotator mechanism 228, including a pinion gear 230, is operatively connected to ring gear 224 to rotate the ring gear when a change in die phasing between the crankshafts is desired. As long as ring gear 224 remains stationary, shafts portions 212, 222 remain m phase. When ring gear 224 is rotated by rotating pinion gear 230, arm 226 rotates causing a change in die phasmg between shaft portions 212, 222. The rotator mechanism 228 would, dierefore, be used to adjust die relative phasmg of die input shaft to die output shaft, thereby adjusting the phasing of the two crankshafts and the compression ratio. In addition, two crankshafts per cylmder could be used to eliminate the inherent side dirust imparted by the crankarm in the smgle crankshaft design. The effect of the maximum possible compression ratio on the sensitivity to CR on phasing should be noted. It might be advantageous to have a geometry where the pistons interfere with each other at "zero" phasing. Of course, diis set up would operate widi non-zero phasing all the time. Applicants have determined how die change in phasing of an opposed piston engine changes the compression ratio. This effort includes three studies as shown in Fig. 25. In the furst, when the two pistons were in phase, i.e. both pistons reach TDC at the same time, the compression ratio was 25:1. In the second, when the pistons were in phase they would come together and just touch at TDC. With a flat top piston there would be no volume between the pistons and, assuming no crevice volume, the compression ratio would become infinite. The third case assumes negative interference so that the pistons would come in contact while out of phase to some degree. For this case, the overlap was about 10% of die stroke causing die pistons to contact at 46° out of phase. Of course, engine geometry (bore, stroke, connecting rod lengdi) will also effect CR versus phasing; diese values were kept constant in diis study. These results indicate that the compression ratio could be varied over a very large range using an opposed piston arrangement with variable phasing. Also, die slope of die change in compression ratio widi phasing depends on die amount of clearance or negative clearance between die pistons at TDC widi 0° phasing. Thus, in a practical application, it would be desirable to strike a balance between the range of phasing needed to cover the desired range of compression ratio and die precision with which the phasing needs to be controlled, i.e. the slope of the curve in Fig. 25 should be optimized. Thus, ideally, die slope of the curve would be steep enough that the desired range of compression ratio could be achieved widiin a limited amount of phasing, and not so steep that the phasing needs to be too precise. Referring to Fig. 26, it is very clear diat as the pistons become more and more out of phase that the compression ratio decreases. It is also clear that there is very litde change in shape of die cylinder volume versus crank angle curve for phasing angles less dian about 120°. As a result, the variation in phasing can be used to control compression ratio over a large range without any affect in the cylinder volume versus crank angle. An opposed piston system with variable phasing clearly provides the desired flexibility to achieve a broad range of compression ratio values. The effective compression ratio may be varied with variable valve timing. Specifically, as shown in Table I, advancing of the intake valve closing lowers the effective CR while significant retarding of the IVC also reduces effective CR. However, changing valve events can have a very large effect on the breathing of an engine, and thus the air/fuel ratio, m comparison to varying the geometric compression ratio (assuming that the fuel flow rate is held constant). The steepest change in airflow with TDC temperature is when IVC is changed. As IVC becomes earlier, the TDC temperature is lowered, but airflow is severely restricted possibly undesirably changing the equivalence ratio. In this case, an increase in boost accompanying earlier IVC could be used to maintain a constant air flow rate. Sunilarly with EVC, as EVC is changed the amount of residual trapped in cylinder changes, and therefore breathing is affected. The slope of the IVC Ime is roughly twice that of EVC and IVO while varying the geometric compression ratio does not have an effect on airflow. In terms of changing TDC temperature without effecting airflow, variable geometric compression ratio appears to be the most effective of the control features. Referring to Fig. 28, changing any of the valve events or compression ratio has a definite effect on BSFC. In order to get the best BSFC, increasing compression ratio would be a better choice than changing exhaust valve closing when a higher temperature is needed. A very large BSFC penalty would be paid if EVC is advanced to increase the temperature at TDC. If a lower temperature is needed, advancing IVC is the best method whereas varying the geometric compression ratio could also be an option since it results in only somewhat higher BSFC. Applicants have also determmed, as expected, that changing the effective compression ratio has a large effect on peak cylmder pressure, as shown in Fig. 29. IVC has an almost identical curve as VCR, confirming the fact that changing IVC really changes the effective compression ratio. Because the heat release starts 5° ATDC in this case, the cylinder pressure trace appears to be "double-humped": the first peak at TDC is due to compression; the second peak after TDC is due to combustion. The appearance of two slopes of VCR and IVC lines is due to die absolute peak cylinder pressure occurring either on the combustion hump (CR 18). In order to increase the temperature at TDC from the baseline without undesirably affecting peak cylmder pressure, changing EVC or IVO would be the best strategy. However, this strategy may result in an undesirable uicrease m BSFC (Fig. 28) and may also change the engine breathing (Fig. 27). Applicants have also determined that very high compression ratios are needed for combustion at low intake temperatures. For example, it has been found that at intake temperatures of 0, 20, and 40°F, no combustion occurs when the corresponding compression ratios are below 35, 33, and 30, respectively. At warmed up conditions, the desired compression ratio is approxunately 15, which means that a change of approximately 20 compression ratios would be needed to cover these conditions. Due to die very high compression ratios requured under these conditions, peak cylinder pressures are also high and in some cases greater than 200 bar. As a result, mtake an heaters and/or some other method of starting m cold conditions may be more practical than usmg variable compression ratio alone. Also, mamtaining a lower compression ratio will allow a higher GIMEP to be achieved before hittmg the peak cylinder pressure limit. Anodier mediod of controlling the temperature is to mtroduce water into the intake manifold or directly into the cylinder. Applicants have shown that when the nitrogen in the intake air is completely replaced with water, the water will likely result in a lower flame temperature (205 K lower) due to dissociation. Also, in applicants smdy, the ignition delay increased slightly (by 0.04 msec) and the peak reaction rate dropped by about 50%. Also, when water was added into the intake manifold, e.g. water fumigation, the chemical effect, although small, is to slightly retard the SOC. However, liquid water injection into the intake manifold effectively cools the intake manifold due to the vaporization of the liquid to steam. As a result, IMT and TDC temperatures are significantly decreased as shown in Fig. 30. The impact of water injection on temperature at TDC is mostly due to the decrease in IMT, not due to the change in the ratio of specific heats. The effect on IMT should be viewed as an upper limit. It should be noted that applicants have shown that PCCI can be maintained without adverse thermal effects on the piston 14 (Fig. la). Even though PCCI combustion can create knock intensity levels 10-20 times higher than the safe level experienced in spark-ignited engines, both aluminum and steel pistons do not reach excessive temperature levels. In applicants" preferred embodiment, the temperatures resulting from autoignition in PCCI combustion are much lower than the temperatures experienced in spark-ignited engines since, in applicants" preferred embodiment, PCCI combustion operates under such lean conditions. Pressure Control The SOC may also be controlled by controlling the pressure in the combustion chamber. One way of controlling m-cylinder pressure is to use a compression ratio varying device to vary the pressure in the combustion chamber. Although varying die compression ratio ultimately varies both the pressure and temperature of the charge, the pressure is direcdy changed. An increase in the compression ratio will tend to increase the pressure at TDC, and a decrease in compression ratio will decrease pressure at TDC. Applicants have shown that increasmg the in-cylinder pressure advances die start of combustion and decreasing the in-cylinder pressure retards the SOC. Any of the compression ratio varying devices discussed hereinabove with respect to temperature control may be used. A second way of controlling the in-cylinder pressure is to vary to die intake manifold, or boost, pressure (IMP). The timing of the SOC has been shown to be a function of pressure. Applicants have determined the effects of varying IMP on combustion and engine operation. The engine conditions for one engine study were 1200 RPM, 355.7K smaller gains in GIMEP cause larger rises in PCP. This effect is due to the earlier SOC tiiat occurs as IMP increases. One study varied the pressure at BDC of the compression stroke. The study was performed using a compression ratio of 14.5:1, an engine speed of 1200 rpm, a BDC compression temperature of 389 K, an equivalence ratio of .3285, and no heat transfer. The fuel used was propane and the pressure at BDC was varied while all other parameters were held constant. This smdy clearly revealed that as pressure at BDC increases, the SOC becomes earlier. In addition, for BDC pressures less than 1.75 bar, less than 10% of the fiiel energy was released, while for BDC pressures greater than P = 1.75 bar, virmally all of the fuel energy was released. This indicates that the combustion is highly sensitive to changes in pressure. At very low pressures, very little of the fuel burns, leading to high FSHC emissions. Smce none of the fuel is burning at these low pressures, no carbon monoxide is produced. As the pressure increases (while maintaining IMT constant), a higher percentage of die fuel is burned, which leads to decreased production of carbon monoxide and lower FSHC. Above a certain critical pressure, all of the fuel burns completely, leading to extremely low FSHC and FSCO emissions. Also, a very small change in BDC pressure leads to a very large change in peak cycle temperature (PCT). The results of the simulation indicate that at low peak cycle pressures (PCP), the fuel does not bum. Hence, the pressure peaks at the isentropic compression. As pressure is increased, a higher percentage of the fuel energy is released, causing the cylinder pressure to rise above the isentropic compression pressure. As pressure increases further, all of the fuel energy is being released and further increases in pressure raise the PCP due to isentropic effects. Clearly, varying IMP can be an effective way of controlling the SOC and the duration of combustion. Increasing the IMP tends to advance SOC while decreasing the duration of heat release. Likewise, decreasmg the IMP tends to retard SOC while increasing the duration of heat release. In a typical application, for a constant torque condition, the fuel flow rate would remain virtually constant, and the boost pressure would be increased to advance the start of combustion or decrease the boost to retard the start of combustion. For example, an air compressor, a turbocharger, a supercharger such as driven by an engine power take-off, or an electrically powered compressor, could be used. For a given power level, and, therefore, for a given fuel flow rate, there t)q)ically exists a preferred mtake pressure and temperahire. At very low loads, it may be deshable to control the intake manifold pressure with a throttle 53 (Fig. la) m the same way that the mtake pressure is controlled on a current production spark ignited engine. Throttle 53 would also be used when operatmg a multi-mode PCCI engme m a spark ignited mode as described hereinbelow. Of course, a throttle could alternatively be located at other locations m the intake system, such as in the mtake manifold. As previously discussed, Applicant has demonstrated the ability to control SOC in a PCCI engine usmg variable IMT. It is becoming more clear that a lower compression ratio (CR) is desirable at high load and a higher CR deshable at low load. One embodiment of the present mvention uses passive variable compression ratio (VCR) to vary the CR based on load conditions. Any conventional passive VCR system may be used. The advantage of usmg passive VCR versus active VCR is one of simplicity and lower cost. In this embodiment, SOC is controlled through some means like variable IMT or the addition of ozone. The passive VCR system automatically moves to a high compression ratio at low loads (when cylmder pressures are low) and to a lower compression ratio at high load (when cylinder pressures are high). The advantage of this system is also a wider operating range. The lower CR permits operation at higher BMEP without exceeding cylinder pressure limits. The higher CR allows the engine to be started more easily and run at lower loads, while still maintaining start of combustion at an optimum timing. Others have suggested using passive VCR to limit cylinder pressure, but not in conjunction with active SOC control ui a PCCI engine. In the past, engines using PCCI combustion have been unable to demonstrate gross indicated mean effective pressures that are as high as diose produced by current diesel engines. Since the combustion duration of PCCI engines is usually very short, the PCCI combustion process is essentially a constant volume process for low engine speeds ( In another embodiment of the present invention, a constant pressure portion is added to the combustion process. The constant pressure heat addition occurs below the peak cylinder pressure limit of the engine. The constant pressure portion of the heat addition is accomplished by direct injection of fuel at a controlled rate, similar to the way modern diesel engines operate. The direct injected fuel may be the same fuel as the premixed fiiel or a different fuel, such as diesel, gasoline, natural gas, etc. This would result in the log(P) vs. log(V) plot shown in Fig. 56. PCCI combustion produces essentially no NOx emissions. The combustion due to direct injection of the fiiel would produce NOx emissions at a rate diat is similar to die diffusion bum portion of combustion in a diesel engine with high EGR rates. The fuel that is directly mjected would have an extremely short ignition delay since combustion chamber temperatures would already be high due to the PCCI combustion. This would also result in lower NOx emissions. Bodi the PCCI combustion and the combustion due to burning fuel at constant pressure, produce power, but the PCCI combustion produces little or no NOx emissions. Hence, the FSNOx and the brake specific NOx emissions of such a cycle could be significantiy lower dian for diesel operation. Similarly the soot should be lower for the proposed cycle of the present embodiment than for a diesel, since PCCI combustion does not produce soot. The cycle of the present embodiment may be accomplished by mixing fuel with air to achieve an approximately homogeneous state using, for example, a carburetor, a throttle body, a port fuel injector, or a direct injector injecting fiiel at or before about intake valve closing (IVC). The charge is compressed imtil autoignition occurs (PCCI combustion). When the cylinder pressure begins to drop, direct injection of fuel begins. A sensor may be used to detect the pressure drop m die chamber, or a predictive method, such as a look¬up table, may be used. The injection rate is controlled to maintain constant cylinder pressure at or below the peak cylinder pressure limit of the engine. Direct injection of fuel ends early enough to allow low soot production. Expansion continues in a nearly isentropic manner until the exhaust valve opens near bottom dead center. The percent of energy from the PCCI combustion could be varied to minimize NOx and soot emission and maximize efficiency and GIMEP. This method could be particularly effective to improve the engine"s transient response. Fig. 57 illustrates the general engine arrangement of the present embodiment including a sensor 120, controller 122, direct fiiel injector 124, port fuel injector 126, carburetor 128 and spark plug 131. The engine can be started in a variety of ways including using direct fuel injection and compression ignition; direct fuel injection and spark ignition; port fuel injection and spark ignition; or direcdy started in the PCCI mode. The controller controls one or more of the amount of fiiel injected, the rate at which the fuel is mjected and the timing of the direct injection event. Also, the controller controls the percentage of fuel burned in the PCCI combustion process. Since die volume of the charge is increasing during the constant pressure portion of combustion, the bulk temperature of the charge must increase. This temperature increase would aid in lowering the CO and HC emissions that usually result from PCCI combustion. Air/Fuel Mixture Autoignition Properties Another strategy for controlling the start and duration of combustion is to vary the air/fuel mixmre autoignition properties. The autoignition properties of the air/fuel mixture may be controlled by injectmg a gas, e.g. air, oxygen, nitrogen, ozone, carbon dioxide, exhaust gas, etc., mto the air or air/fuel mixture either in the intake system, e.g. preferably in the port using, for example, iojector 42, or in the cylinder directly using, for example, injector 40, thereby providing control over the start of combustion and the combustion rate. Applicants have examined the effect of adding reactive species to the air/fuel mixmre on the combustion process. One study was performed using an equivalence ratio of 0.3, a temperamre at BDC of 389 K, pressure atBDC of 3 bar, and propane as the fuel. The compression ratio was 14.5, and the engine speed was 1800 RPM. The engine geometry used was for a Cummins C series engine. The nitrogen, oxygen, and fuel mole fractions were held constant at 0.771, 0.216, and 0.0123, respectively, for all cases. The mole fraction for the reactive species added was 0.000411 for all cases. The reactive species examined were H2, H2O2, OH, CO, O, HO2, H, and O3. Fig. 32 shows the temperature versus crank angle. Although CO and H2 advanced the SOC by less dian 0.5 crank angle degrees, all other species significantly advanced the SOC, with O3 (ozone) causmg the largest change in the SOC. Therefore, small concentrations of most common radicals will cause significant changes in the SOC. Thus, applicants have determined that the addition of very small quantities of ozone advances the SOC by significant amounts. Applicants have also shown that virtually all of the ozone will be consumed by the combustion process and that the change in the SOC wUl diminish as the amount of ozone added increases. Specifically, Fig. 33 illustrates the effects of additional ozone on advancing the SOC. The increase in temperature indicates the start of the combustion event. Applicant"s studies have further shown that ozone can be used to advance SOC in PCCI, and thus HCCI, engines fueled with natoral gas, propane and diesel fuels. Because ozone was effective in advancing SOC for such a wide range of fuels, Applicant expects similar effects for gasoline and alcohols, as well as other hydrocarbon fiiels, including oxygenated fiiels. Applicant also observed only a small effect when adding O3 to the intake of a conventional diesel engine. Given the significant effect additional ozone has on the SOC, ozone can be used in several ways to advantageously control the combustion in a PCCI engine. First, by adding different amounts of O3 to the intake ports, one, several, or all cylinders could have their SOC adjusted. Second, adding O3 to the intake could be used as a cold starting aid for PCCI and diesel engines. Third, adding O3 to the exhaust of an engine would allow a catalyst to light earlier thus possibly significantiy reducing cold start emissions on catalyst-equipped spark ignited engines, diesel engines and PCCI engines. O3 could be produced "on board" through a simple electro-chemical reaction. Ozone generators are commercially available. Also, the ignition delay of a diesel engine could be reduced by adding O3 to the intake. This would reduce the premixed burn fraction which would then lower NOx emissions and reduced noise. Applicant has shown that ozone (O3) can be used to improve die autoignition of a PCCI engine operating on propane. It is speculated that adding O3 to the intake of a lean burn spark-ignition (SI) engine could significantly extend the lean limit. This effect would Docket: 0270-2486 be very advantageous because leaner operation reduces NOx emissions. By speeding up lean SI combustion through the addition of O3, the BSFC NOx tradeoff would be improved. This BSFC NOx tradeoff is also improved because of the higher effective expansion ratio. Applicant believes that the addition of O3 would increase the lean limit because, near the lean limit, the combustion event in the flame front can only just heat the reactants enough to cause the reactants to ignite. The lowering of the autoignition temperature with ozone should extend the lean limit. Applicants have shown that increasing the oxygen concentration advances the SOC. Since oxygen enrichment of the intake charge slightly advances the SOC and die percentage of ozone at the outlet of an ozone generator increases as the inlet oxygen concentration increases, ozone can be used in conjunction with oxygen enrichment to further advance the SOC. Oxygen enrichment may be accomplished by the use of selective membranes or by odier means. Also, ozone could be used in conjunction with a catalyst to reduce UHC emissions from a PCCI engine or conventional spark ignition or diesel engine. In this case, ozone is to be added to the warm exhaust causing the temperature of the exhaust to increase, thereby enhancing catalyst efficiency. Furthermore, if the ozone is added before the turbine inlet, turbine inlet temperature may increase, thus easing air-handling issues. Given these results, ozone can be effectively used as a starting aid for diesel engines. An engine can be started by injecting some fuel very early (diis is easily done with a common rail system). Ozone is added to the intake or produced within the cylinder during the intake and/or compression process to allow the fuel that was injected early (about 60° BTDC - or even earlier) to ignite much more easily. More fuel can then be injected near TDC (into the hot combustion product) where it will easily ignite. By injecting a reasonably small amount of fuel very early, it doesn"t matter if it ignites too Docket: 0270-2486 early because there isn"t enough fuel to result in excessively high cylinder pressures. As a result, no special SOC control system is needed. Using ozone to aid in cold starting of diesels requires a pilot injection of diesel fuel. For an automotive spark ignition engme, ozone could be added to the exhaust or to the cylinder just before exhaust valve opening to increase the temperature of the exhaust. Ozone decays to O2 and O; the 0 increases the reaction rates of exothermic reactions thereby increasing exhaust temperamre. This would aid in catalyst warm-up, thereby lowering unburned hydrocarbons at cold-start conditions. Ozone may be added to the exhaust of an engine to increase the concentration of oxygen O atoms resulting in lower NOx emissions. Note the presence of the 0 atom in the following extended Zeldovich mechanism. O + N2 NO -H N N + O2 NO + O N + OH NO + H The free oxygen atoms play a critical role in the extended Zeldovich mechanism of NO formation. Ozone will aid in the transition from a PCCI mode to an SI mode and vice versa. Example: an engine is operating at light load in the PCCI mode. Ozone is being used to control the SOC. The amount of ozone added is equivalent to a significant change in IMT. To switch to SI mode, the ozone is shut off. As a result, the IMT is too low to support autoignition. The equivalence ratio is increased to that sufficient to support stable spark ignition operation, probably via port or direct fuel injection. The spark plug is tiien fu:ed at the appropriate point. Although mcreasmg the oxygen concentration advances the SOC, however, applicants have determined that oxygen enrichment from 20.7 percent to 21.65 percent will advance the SOC by less than one crank angle degree, and oxygen enrichment from 20.7 percent to 23.7 percent will advance the SOC by less than 1.5 crank angle degrees. Therefore, combustion may be controlled to a lunited degree by modifying the oxygen concentration of the intake air. This may be done by adding oxygen (or an oxygen rich gas mixmre) to the mtake or by selectively removing nitrogen from the intake air (using a membrane for example). Applicants have also shown that increasing the percent of nitrogen in the intake charge from 78.6 percent to 80.6 percent resulted in the retardation of the SOC by less than 2 crank angle degrees at 1800 rpm. It was also noted the same percentage increase of N2 m the fresh charge lowers the FSNOx from . 144 to .048 grams of NOx per Kg of fuel. Another method of varying the effect of oxygen on the combustion process is to dilute the mixture with EGR. In one study, an engine EGR system was plumbed from the exhaust manifold to the compressor inlet. Because the EGR is mixed in upstream of the aftercooler, and in the present study, the aftercooler exit temperamre was controlled and held fixed, the EGR should not have significantly effected the temperamre at SOC. During this study, fuel rate and intake manifold temperature were held constant. As the EGR rate was increased, exhaust manifold pressure decreased, which in turn decreased ak flow on this turbocharged engme. The fuel rate was held constant, so the fresh equivalence ratio increased. In spite of the increased equivalence ratio, SOC retarded as the EGR rate increased, most likely due to die diluent effect of the EGR. As expected, SOC retarded as the EGR rate increased. However, as EGR rate increased, CO and HC emissions also increased. Also, as EGR rate increased, the spread in SOC between cylinders increased. In a similar smdy, the SOC was held constant by adjusting IMT. As the EGR rate was increased, exhaust manifold pressure decreased, which in turn decreased air flow. The fuel rate was held constant thus causing the equivalence ratio to increase. In addition, as the EGR rate increased from about 7 to 13% EGR, there was a sharp rise in the cylinder to cylinder variation in SOC. Ultimately, a higher IMT was required to maintain constant SOC as the EGR rate increased, in spite of an mcrease in equivalence ratio. This requirement was due to the diluent effect of increased EGR on the intake air. Another technique for modifying the autoignition properties of the air/fuel mixture to control SOC and the duration of combustion is to vary the octane, methane or cetane number of the charge by, for example, by providing two or more fuels have different octane, methane or cetane numbers. The fuel supply can be either selectively switched between the fuels or the fuels can be mixed. This techmque makes it possible to retard or advance the combustion event. For example, a fuel which tends to autoignite more readily (lower octane or methane number, or higher cetane number) could be controllably mixed with a fuel that tends to autoignite less readily (or a fiiel that ignites at a high temperature and a fuel that ignites at a low temperature could be used) to enable direct control over the timing of igmtion and rate of combustion by changing the ratio of the fuels that are present in the combustion chamber during the combustion event. As shown in Fig. 34, propane, octane and heptane have significantly different effects on the SOC. The same effect may be achieved by using a fuel additive, such as a controlled amount of propane, ethane, or other hydrocarbons, such as engine lubricating oil, that change the autoignition properties of the fuel to advance or retard the start of combustion. Of course, any method that changes the fuel"s octane/methane number or the activation energy of the fuel can be used to advance/retard combustion. Applicants have determined that there is a significant sensitivity of start of combustion to octane number. This effect was independent of intake manifold temperature. Moreover, in one study, the start of combustion was retarded approximately 7° for an increase in octane number from 80 to 100. Achieving dynamic control over individual cylinder combustion in a multi-cylinder PCCI engine will be critical to achieving improved combustion. Since many of the gases/fluids discussed hereinabove, e.g. fiiel, ozone, oil, water, etc. have now been shown to significantly affect the SOC and/or rate of combustion, these additives can be used to advantageously balance combustion between the cylinders in a multi-cylinder engine running on PCCI principles. For example, by injecting a liquid or gas diluent, such as a less reactive fiiel, water, uncooled or cooled exhaust products, air and/or nitrogen eidier into the intake air or directly into the charge in the cylinder, the SOC can be retarded. Also, by injecting, for example, a more reactive fuel, ozone, oil and/or oxygen into the charge die SOC can be advanced. Fig. lb illustrates one system for balancing combustion between cylinders of a multi-cylinder engine. The system uses port injection of fuel on the engme with two types of supplies per cylinder - supply 32 to inject liquid fuel and supply 34 to inject gaseous fuel. Although die supplies 32 and 34 are shown feeding mto a single passage for delivery to the intake port, die supplies may include separate delivery passages connected to the intake port at different locations. Liquid fuel will decrease intake charge temperature by die heat of vaporization of the liquid fuel. The temperamre at TDC compression, and dierefore SOC, can be controlled by varying the amount of liquid versus gaseous fuel. Also, it should not matter if die liquid vaporizes in the port or during compression. The gaseous and liquid fuel can be die same fuel in different states, e.g. propane, or different fuels, e.g. gaseous namral gas and liquid gasolme, such as indolene. It is important that die port injection system have good separation between cylinders and sequential (tuned to intake event) injection is likely to be required. During operation, a cylinder diat is "going out" would be given more gaseous fuel and a cylinder that is "too hot" would be given more liquid. This method can be used to achieve about a 20 degree temperamre difference. One of die supplies could be lubricating oil or ozone while the other supply could be a fuel having a high resistance to ignition, e.g. a high octane number, to permit the SOC to be effectively controlled by varying the amount of oil or ozone added to die mixture. Also, by using the engme"s lubricating oil supply, or using ozone created by the engine during operation, an additional supply of fuel/additive can be avoided. Equivalence Ratio Anodier control variable that applicants have shown can be effectively used to control the SOC and combustion duration or heat release rate is the equivalence ratio (j) of the fuel/air mixture. Equivalence ratio is equal to fuel/air ratio divided by stoichiometric fuel/ah- ratio (if (j) 1, fuel excess). Combustion needs to be slowed down in a PCCI engine because fast combustion leads to high noise, lowered efficiency and high peak cylmder pressure. If different temperatures and/or equivalence ratios can be achieved throughout the charge of an/fiiel at or near point of ignition, the resultmg rate of combustion will possibly be slowed down thus advantageously lengthening the duration of combustion. The equivalence ratio could be increased by increasing the fuel flow to the cylinder without a corresponding increase in intake air flow, or by decreasing the intake air flow. The equivalence ratio could be lowered by decreasing the fuel flow to the cylinder widiout a corresponding decrease in air flow, or increasing the air flow rate. Variations in the quantity of fuel delivered to a cylinder is varied by controUmg the operation of fuel control valves 39, 41, and/or fuel injectors 35, 36 m a known manner. The variations in the air flow rate could be achieved by, for example, variably controlling compressor 24 to vary boost pressure. To test die lower limit for equivalence ratio, applicants conducted engine studies to determine whether acceptable PCCI combustion could be obtained with an extremely lean mixture. The results indicate that very stable combustion can be achieved at an extremely lean equivalence ratio of .05 while obtaining a heat release duration of approximately 30 degrees. Also, as shown in Figs. 35 and 36, the results indicated that the start of combustion advances and die apparent heat release duration decreases as equivalence ratio increases, i.e. air/fuel mixture becomes richer. Applicants have clearly shown, as mdicated in Fig. 37, where cylinder temperature mcreases indicate the heat release event. Moreover, referring to Fig. 38, the apparent heat release duration becomes longer as equivalence ratio decreases, i.e. air/fiiel mixture becomes leaner. Also, applicant have shown that for a four stroke engine that both peak cylinder pressure and GIMEP increase as equivalence ratio becomes richer. With respect to a two-stroke engine, applicant have determined that as equivalence ratio increases, GIMEP increases. Studies were also conducted to investigate whether the equivalence ratio affects the amount of fiiel burned in PCCI combustion. The results indicated that as equivalence ratio becomes richer, die percentage of fuel energy showing up as apparent heat released mcreases at first and then levels off near 80%. This number can never reach 100% because of heat transfer. Witii respect to emissions, as equivalence ratio becomes richer, fuel specific hydrocarbon emissions decrease. In addition, as equivalence ratio became richer, average noise levels increased, and GIMEP increased. As equivalence ratio becomes richer, the average knock intensity increases. As equivalence ratio became richer, die cycle-to-cycle combustion variation, as measured by the coefficient of variation (COY) of GIMEP, generally decreased. In fact, the COV"s of GIMEP, for the conditions of the study, stayed below the combustion stability lunit (in this case defined as 5%), where a GOV above the limit indicates unacceptable stability. Smdies were performed to determine the effect that variations in equivalence ratio have on tiiermal efficiency in PCCI combustion. An equivalence ratio smdy was performed while matching the following parameters: speed, IMT, IMP, engine oil temperature, and engine water temperamre. Equivalence ratio was increased by holding air flow constant and increasing fuel flow to the engine. As fuel flow increased and equivalence ratio became richer, gross indicated thermal efficiency increased at first and finally leveled off. Engine work output increased witii respect to increased fuel flow as more fuel was burned. At the leaner equivalence ratios, a significant amount of fuel is left unburned. At the richer equivalence ratios, tiie percentage of fiiel that is being burned levels off as noted hereinabove, and the gross indicated tiiermal efficiency levels off because the increase in engine output is being offset by the additional fuel input. In addition, an engine study was conducted with the engine cycle running from bottom dead center of tiie compression stroke to BDC of the expansion stroke. The study was conducted using a compression ratio of 14.5:1, an engine speed of 1200 RPM, a BDC compression temperature of 389 K, pressure at BDC of 4.31 bar, and no heat transfer. The fuel used was propane. The equivalence ratio was varied while all other parameters were held constant. It was discovered that the percent of energy released slowly tapered off as the equivalence ratio drops below 0.15. This data indicates that for a given temperatore and pressure, there is a lower lunit to the equivalence ratio of a mixmre tiiat will burn completely. Also, it was shown that FSCO emissions are very high at equivalence ratios below 0.15. This data indicates tiiat only a small amount of tiie fuel burns to completion at these low equivalence ratios for tiiis temperature and pressure. In addition, tiie FSHC decrease slightiy as the equivalence ratio is varied from 0.05 to 0.4. Thus, most of the fiiel reacts regardless of the equivalence ratio. It was also shown that SOC occurs earlier as equivalence ratio mcreases. The study showed that peak cylmder temperature gradually increases as equivalence ratio is increased showing the increased amount of energy available to be released. Peak cylinder pressure (PCP) gradually increases as equivalence ratio is increased showing the increased amount of energy available to be released. At equivalence ratios greater than or equal to 0.18, virtually all of the available fuel energy is released, leading to a nearly linear increase in PCP as equivalence ratio increases. Applicants have determined that it may be possible, although not necessarily desirable, to maintain PCCI combustion at very rich equivalence ratios, e.g. .5, if IMP and IMT are sufficiently low to prevent the peak cylinder pressure limit from being exceeded. It will be difficult to start an engine at the low boost and IMT levels needed for maintaining low cylinder pressures at such rich equivalence ratios. The very advanced heat release, loud knock, and combustion roughness make running at this condition undesirable. A lower CR for retarding SOC may improve these aspects. Also, by varying the level of charge stratification, the temperature and equivalence ratio distribution can be altered to permit control of the combustion rate and/or the start of combustion. An auxiliary combustion chamber concept may be a mechanism for achieving die desired stratification, thereby enablLug better control over the start of combustion. For example, conventional auxiliary combustion chamber designs typically used on small engines having indirect injection (IDI), and large spark ignited engines usmg namral gas fuel, could be used. In order to operate under the desired lean conditions for optimal PCCI combustion, substantial air flow must be provided to the intake manifold. A turbocharger could provide the needed air flow for a multi-cylinder PCCI engine. Applicants" original target was to reach an equivalence ratio of 0.40 or leainer. Referring to Fig. 39, applicants have shown that operating at leaner than an equivalence ratio of 0.29 would violate the compressor pressure ratio limit of the available mrbocharger. Applicant determined that mrbine pressure ratios are very high at lean equivalence ratios. As a result, the exhaust manifold pressure is very high which causes a large BSFC penalty. Because of the relatively cool exhaust temperatures produced by PCCI combustion, very small turbine cases are needed which result in high exhaust manifold pressures. Applicants have determined tiiat it would be desirable to operate under slightly leaner conditions than the origmal target. At an equivalence ratio less than .4, a smaller turbme casmg was used to decrease the compressor pressure ratio and exhaust manifold pressure ratio, but a high BSFC penalty is paid, as shown in Fig. 40. Figs. 41 and 42 illustrate the higher PMEP losses with the smaller mrbine casing and the higher BSFC. Also, with the smaller turbine casing, the rotor speed is much higher and, in fact, near the limit on rotor speed as seen in Fig. 43 (rotor speed limit 120-125k rpm range). Applicants discovered that there is a lower limit on the size of the mrbine casing used due to the losses incurred with the high back pressure and with reaching the rotor speed lunit. In order to avoid tiiis problem with the high back pressure and rotor speed limiting airflow, one possible solution is to use a mechanically driven supercharger in conjunction with a turbocharger. The supercharger would be upstream of the compressor so that the turbine bears less of a burden for producing boost. Some BSFC penalty would be incurred for the shaft work absorbed by the supercharger; however, the BSFC penalty is less than the very high penalty incurred with the very small mrbine. Because the supercharger is driven mechanically from the shaft, there should be no trouble getting the desired air flow. The mrbine then can be sized somewhat larger, and should not approach the speed limit and should not have extremely high back pressure. Applicants have also determined the effect of engine speed on SOC. The time of autoignition depends on the temperamre and pressure histories. By changing the engine speed, these histories are changed. It is possible to advance the combustion event by reducing the engine speed, and to retard the combustion event by increasing the engine speed. Specifically, a 75% mcrease m engine speed, from 1000 to 1750 resulted in a 1.5% increase in the start of combustion pressure and a 2.8% increase in the start of combustion temperature. In addition, a 75% increase in engine speed decreased the heat release rate duration by 0.81 ms (only a 23% decrease) which corresponds to an increase in heat release duration of 1.7 crank angle degrees (only an 8% increase). Given this minimal impact of engine speed on the SOC and heat release, and the inability to effectively vary engine speed in many practical engine applications, engine speed is not viewed as an effective combustion control variable. However, one example where engine speed could be used to provide some control over combustion is in an application where the engine drives a generator or alternator. As discussed hereinabove, the foregoing control variables are used to control the SOC and the duration of combustion to achieve optimum PCCI combustion. One key consequence of efficient, optimum combustion is reduced emissions. Applicants have shown that a PCCI engine can achieve NOx emission levels that are well below any other NOx emission levels ever demonstrated by applicants using diesel and natural gas engines, and well below future emissions standards as shown in Fig. 44. Applicants have also determined the effect of the control variables and other factors on emissions of a PCCI engine. Engine speed has little effect on the quantity of NOx emissions. Although a 75% increase in engine speed approximately tripled the FSNOx, the levels of NOx emissions produced were still extremely low. Also, as equivalence ratio becomes richer, fuel specific NOx generally increases, but still remains at extremely low levels. Referring to Fig. 45, applicants have determined tiiat engine speed appears to affect FSCO and FSHC emissions more significantly. As shown, below a certain critical speed, virtually all of the fuel bums, FSHC are low and FSCO is low. Just above the critical speed, the fuel partially burns, resulting m higher FSCO emissions. As engine speed continues to increase, the percentage of the fuel that burns continues to drop, resulting in lower FSCO emissions. These emissions also vary as the temperamre at BDC varies. Referring to Fig. 46, at very low temperatures, very little of the fuel bums, leading to high FSHC emissions. Since none of the fuel is burning at these low temperatures, no carbon monoxide is produced. As the temperature increases, a higher percentage of the fuel is burned, which leads to increased production of carbon monoxide and lower FSHC. Finally, above a certain critical temperature, all of the fuel burns completely, leading to extremely low FSHC and FSCO emissions. In fact, as shown in Fig. 47, applicants have shown tiiat all data points with end of combustion flame temperatures above 1600 K had acceptable CO emissions. It has been shown that both high temperature and the hydroxyl radical (OH) are critical for the desired oxidation of CO. Importandy, as equivalence ratio becomes richer, fuel specific CO decreases, while the concentration of COj in the exhaust increases. In one study, all points taken with an equivalence ratio As equivalence ratio becomes richer, fuel specific HC decreases. Clearly, unburned hydrocarbons (UHC) are a key concern for PCCI engines since reducing unburned hydrocarbons is essential to the commercial feasibility of a PCCI engine. Applicants have determined that UHC, and CO, is formed in small crevices positioned in the components forming the combustion chamber, i.e above die top ring of the piston between the piston and the liner; between die cylinder head and the cylinder liner; and around die components mounted in the cylinder head. The crevices prevent the volume of mixture in the crevice from reaching a sufficiently high temperature necessary for burning of die HC and oxidation of the CO. For example, applicants have shown diat similar pistons widi different crevice volumes have different UHC levels. Applicant"s studies have shown that reducing the crevice above the top piston ring reduces the HC and CO emissions from a PCCI engine. For two pistons that have die same compression ratio and virmally identical geometry, die piston widi die ring closer to the top of the piston has significanfly lower HC and CO emissions. Specifically, Applicant"s studies show diat when the crevice above die top ring is reduced from 5.6% of the clearance volume to 1.6% of the clearance volume, FSHC is reduced by about 25% and FSCO emissions reduced by about 40%. The reduction in emissions occurs for a different reason in a PCCI engine than it does in a spark-ignition engine. In a spark ignition engine, die flame front cannot propagate into a small crevice. Thus, die fuel in die small crevice will not bum in a spark ignition engine. In a PCCI engine, the fuel in the crevice is too cool to autoignite. Furthermore, since bulk gas temperatures are low in a PCCI engine, fuel diat outgases from die crevice during the expansion stroke encounters cool bulk gas. Thus, fuel that was originally in die crevice is rarely at a temperature diat is sufficiently high to autoignite or burn completely. Hence, HC and CO emissions remain high. Therefore, a combustion chamber geometry diat minimizes the crevice volumes is highly desirable. The present PCCI engine may include one of several designs to minimize UHC. The present crevice minimizing designs result in a low crevice volume; keep die fuel away from any existing crevices; or cause die mixture in die crevice volume to burn appropriately. The designs shown in Figs. 48a and 48b are most easily implemented in a ported two stroke cycle engine. Referring to Fig. 48a, in one embodiment, die engine has a single piece head and liner combination 300, aldiough a two-piece system could be used. Just above die top ring 302 (at TDC), die bore 304 increases to eliminate die crevice around the top land 306 of piston 308. There are no crevices in the cylinder head, as it is a single piece without valves, gaskets, etc. Referring to Fig. 48b, a second embodiment of the crevice minimizing design may similarly include a one-piece head and liner 310. However, in this embodiment, the piston 312 has a very aggressive cutback 314 forming the top land to enlarge the crevice volume 316 between die top land and die luier. The crevice volume 316 is now so large tiiat it will no longer quench combustion in this area tiius allowing fuel in this volume to burn resulting in reduced UHC. Fig. 49 illustrates yet another embodiment including a cup or chamber 320 formed in the cylinder head 322 of the engine. The fuel injector 324 is positioned to mject fuel directly into cup 320 early in the compression stroke. Because air is pushed into cup 320, the fuel does not exit the cup. After compression ignition occurs, the products can pass through the relatively large passage or throat 326 between the cup 320 and die main cylinder 328. The fuel is well mixed because of the torbulence of the air entering the cup. Because there are no crevices in the cup and because the fiiel does not leave the cup until after combustion is completed, UHC are extremely low. The cup could easily be coated with a thermal barrier coating to reduce heat losses. Figs. 50a and 50b illustrate a cup design for a four stroke engine. The exhaust and intake valves 330 are arranged around a cup 332 in die head 334. Cup 332 may be positioned directly above die combustion chamber 336 as shown in Fig. 50a or offset to allow more room for the valves 330 as shown in Fig. 50b. Another possibility is to include a small auxiliary valve in the cup to allow the products to exit the cup more efficiently. This valve could open after die main exhaust valve opens so that die auxiliary exhaust valve in the cup would not open against a high pressure. In this case, the auxiliary exhaust valve could be electronically operated. The timing of opening and closing of diis valve could be used to vary die residual mass fraction which would allow control over die SOC using this auxiliary valve. Also, an opposed piston engine, as discussed hereinabove, may be used to substantially reduce the crevice volume by avoiding a cylinder head and the associated crevices. Now referring to Fig. la, another embodiment of the present invention for reducmg emissions is disclosed. Specifically, this embodiment controls UHC and CO by heating the upper portion of the cylinder liner 49 to cause oxidation of the charge in the crevices. A heater 51 is mcorporated into the upper part of the liner. The heater could be any type of heater capable of effectively producmg heat, such as an electrical resistance heater. The heater heats the gas in the crevice above the top ring when the piston nears TDC. This heating will cause the gas to be less dense resultmg m a smaller mass of charge remaining m the crevice. The charge leavmg die crevice wUl be at a higher temperamre due to the heating thus mcreasing the tendency of the charge to react and form CO2 instead of CO and UHC. Also, a glow plug may be used to heat the combustion gases to reduce emissions by enabling a larger portion of the crevice volume to be burned. It has been determined by applicants that a glow plug will have only a slight impact on the SOC. Since SOC changes only slightly when the glow plug is turned on, it does not appear that the glow plug is mitiating combustion. It is more likely that when the glow plug, which was located in a spacer plate, is turned on, it gradually warms up the gas in the crevice volume. This increase m temperatore is sufficient to speed up die onset of rapid combustion, and burn more of the fuel than would have been burned without the glow plug on, resulting m a slight mcrease in GIMEP. The present engine, as shown in Figs, la and lb, may also be operated as a multi-mode engine which changes modes of operation based on the operating conditions or needs of the particular application. For instance, the engine may be operated on diesel fuel only as a conventional diesel engine, as a modified diesel engine with diesel fiiel being injected earlier in the compression event than the conventional diesel engine, as a spark-ignited engine usmg spark plug 56 (Fig. la) or as a PCCI engine. This type of spark ignited/compression ignited DI (direct injection) variable compression ratio engine provides a combination of low emissions, high power density, and ease of starting. This engine operates in the following different modes depending on the current operating conditions/needs of the engine. 1) Medium compression ratio ( — 10:1), early injection (fiiel injected during intake stroke or very early in the compression stroke) nearly homogeneous: a) Overall lean mixture, spark ignited - allows low NOx, high brake mean effective pressure (BMEP) operation, as well as medium BMEP operation. b) Stoichiometric mixture, spark ignited - allows high NOx high BMEP transient operation, as well as low NOx operation with a 3 way catalyst. 2) High compression ratio (~ 15:1), early injection, nearly homogeneous, very lean (({) 3) High compression ratio (~ 15:1), late injection, stratified charge: a) Spark ignited - allows medium NOx, medium BMEP unthrottied operation, and low BMEP operation. b) Compression ignition - allows medium NOx medium and low BMEP unthrottied operation. 4) Low compression ratio (~8:1), early injection, nearly homogeneous, spark ignited: a) Lean burn - allows very high BMEP operation. b) Stoichiometric - allows very high BMEP operation. 5) Medium compression ratio (-10:1), late injection, stratified charge, spark ignited - allows medium NOx, medium and low BMEP, and high BMEP operation. 6) Very high compression ratio (-20:1), lean bum, early injection, nearly homogeneous, compression ignition - allows the engine to be started in PCCI mode. The key here is to take full advantage of die variable compression ratio. Starting of the engine can be achieved witii spark ignition at a lower compression ratio and then transitioning to high compression ratio, lean PCCI operation for low NOx. For less severe (e.g. not as cold) conditions, engine starting could be achieved directly with very high compression ratio PCCI operation. At low and medium loads, the engine can operate in a PCCI mode as the compression ratio is adjusted to keep the start of combustion near the optimum crank angle. For high load requirements, the air/fuel ratio can be enriched, compression ratio lowered, and the engine can be spark ignited. In order to handle sudden transients, the engine may go into one of the late injection modes where richer air/fiiel ratios are possible without engine damage. In die multi-mode engine, ECU 20 (Fig. 1) functions widi a control strategy for controlling the various controlled features of the engine to effectively switch between, and operate in, the different modes in order to achieve a variety of objectives. For example, the multi-mode engine achieves low NOx emissions in the PCCI mode while enhancing startability by providing a high compression ratio or spark ignition. In addition, the engine can achieve a high cylinder pressure at high BMEP by switching to a lower compression ratio spark-ignited mode. The multi-mode engine also permits stable combustion to occur after switching to late injection which results in a stratified charge by rapidly adjusting the compression ratio. Also, fuel consumption can be effectively controlled using high compression, PCCI operation and stratified charge operation requiring no throttling which have excellent thermal efficiency. This operation also improves transient response by going from PCCI to late injection, stratified charge to suddenly enrich the mixture. This multi-mode engine can also effectively minimize knock, and therefore knock damage, by effectively operating at lean PCCI or stratified charge or low compression ratio, lean bum or stoichiometric conditions. Of course, the engine operates to effectively control the start of combustion during PCCI operation by varymg, for example, as discussed hereinabove, the temperature and/or the equivalence ratio and/or the pressure and/or the air/fuel mixture autoignition properties. This engine could mn on a variety of fuels like gasoline or diesel fuel. Another operating mode is dual injection in which an early injection is used to create a lean charge for PCCI operation. A second, late injection then adds a small amount of stratified fiiel which can be either spark or compression ignited to help ignite the remaining fuel. This mode is similar to diesel pilot operation but would only be used during transition between the different modes of operation or during engine starting. Applicants have studied the effects of diesel pilot operation on emissions. Fig. 51 shows a comparison of the normalized heat release rate versus crank angle for the three different diesel pilot mjection quantities into a PCCI engine operating on propane. A micro-pilot injection of . 1 % resulted in good heat release placement with no measurable increase in FSNOx. A diesel pilot of an amount estimated to supply 3.6 % of the fiiel energy resulted in a heat release curve having substantially the same shape as the previous case. The SOC is slightly more advanced than that of the . 1 % case despite a lower IMT and constant equivalence ratio. Also, FSNOx emissions have increased over die .1% case from zero to 3.9 g/kg. The final curves illustrates the heat release for a case with ~ 18 % of die fuel energy coming from the diesel pilot. The heat release rate curve is shaped the same as the classic diesel heat release rate curve with a premixed burn spike and a diffusion bum region. Also, the FSNOx (15.3 g/kg) and FSHC (478 g/kg) are significantly higher than in the cases with smaller diesel pilots. With respect to diesel pilot injection, as the percentage of fuel energy from die pilot increases, the start of combustion (SOC) becomes more advanced, despite the lowering of IMT and a constant equivalence ratio. This earlier SOC is caused by the diesel fuel autoigniting earlier than die propane. As the percentage of pilot increases, the heat released by the pilot during the compression stroke increases, leading to higher temperatures earlier in the cycle. Higher temperatures increase die chemical reaction rates of reactions involving propane, leading to earlier autoignition of the propane. Therefore, extremely low NOx levels and good heat release placement can be achieved when using a very small diesel pilot or micropilot, preferably less than 4% of the total fuel energy. Applicants have also studied the control of noise associated with PCCI combustion. Level of noise generated by PCCI combustion is related to die knock intensity. Thus, as knock intensity is decreased, noises decreases. As shown in Figs. 4a, 4c and 6, lowering cylinder pressure, for example, by retarding die SOC, substantially decreases the knock intensity and, therefore, noise. The present engine and control system permits continuous PCCI combustion widi minimal noise by avoiding excessive peak cylinder pressures while maintaining the requu-ed cylinder pressure necessary for efficient, low emission PCCI combustion and the desired power output. The control system of the present invention operates to actively and variably control the mixture"s temperature, pressure, autoignition characteristic and equivalence ratio to ensure that the combustion event occurs between 20 crank angle degrees BTDC and 35 crank angle degrees ATDC. The control system achieves this function by using combustion sensor 16, e.g. pressure sensor, to signal the start of combustion or the location of the heat release event for each cycle. Also, ECU 20, v/hich receives the signals from sensor 16, determines whether the SOC is occurring within a predetermined crank angle range and determines whether the duration of combustion is within a predetermined desired crank angle range. One conventional way for the ECU to determine the optimum SOC would be to use a look-up table. If the SOC and/or die duration of combustion are outside the predetermined crank angle range, then ECU 20 determines the appropriate control variable or variables to adjust, and generates and sends the appropriate signal 22 to the chosen control mechanism or mechanisms, e.g. an cooler 28, heater 30, glow plug 44, fuel control valves 39, 41, variable compression ratio device 38, etc., as discussed hereinabove. The control variables are varied as required to maintain the timing of the start of PCCI combustion preferably between 20 crank angle degrees BTDC and 10 crank angle degrees ATDC, and to maintain the duration of combustion in the range of 5-30 crank angle degrees. Applicants have determmed that, m order to mitiate and maintain PCCI combustion upon start-up in a cold engine, the conditions in the cylinders, e.g. temperature and/or pressure, must be actively influenced. For example, the mtake air temperature could be raised using heater 30 and/or a glow plug 44, and/or die in-cylinder walls heated using a cylinder wall heater 51 and/or an engine coolant/lubricating oil heater. Also, the in-cylmder pressure and temperature could be increased using variable compression ratio device 38. Another effective control feature for enhancing startability is to add small amounts of ozone to the mtake air supply usmg mjector 42, or into die cylinder using injector 40. Alternatively, or additionally, one of the fuel supplies could have a high autoignition property, e.g. low octane number. Also, the engine may be operated in a non-PCCI, for example, as a spark-ignition, dual ftiel or diesel engine, during starting of the engine. One or a combination of these controls are varied, in accordance with the principles discussed hereiaabove with respect to each control feature, to cause PCCI combustion to occur. As the engine starts, the ECU will monitor the start of combustion and duration of combustion by receiving combustion data, e.g. pressure signals, from sensor 16 throughout engine operation. Once the engme is warmed up, the SOC and duration of combustion will vary due to the sensitivity of PCCI combustion to the temperature and pressure history. Small variations in the numerous factors affecting temperature and pressure history, such as combustion chamber wall temperature, IMT, equivalence ratio, IMP, etc. result in significant variation in the SOC and the duration of combustion. During operation, the control system of the present invention will vary one or more of the control variables, that is, lemperamre, pressure, air/fuel mixture autoignition properties and/or equivalence ratio, using the various control mechanisms discussed hereinabove, in such a manner to maintam the SOC and duration of combustion in the desired ranges. For example, applicants have shown that SOC can be advanced fi:om 5° ATDC to 0.5° BTDC by increasmg the IMT from 184°F to 195°F, as shown in Fig. 8. Applicants have also shown that increasing CR, which raises the in-cylinder temperatures, can be used to advance SOC. For example. Fig. 21 shows that increasing CR from 14:1 to 22:1 advanced the SOC from 2° ATDC to 13° BTDC when the equivalence ratio was 0.35 and IMT was 380 K. In addition, applicants have shown that increasing RMF to raise the temperature of the charge also can be used to advance SOC. When RMF was increased by adjusting exhaust valve lash from 0.025" to 0.046", the SOC advanced from 6.4° ATDC to 1.7°ATDC, as shown m Fig. 16. Heat transfer to the charge, whether from active heating elements or hot surfaces such as the combustion chamber walls, has also been shown to advance SOC. Applicants have also shown that, with a glow plug installed in the combustion chamber, the SOC retarded from 0.6° ATDC to 1.5° ATDC after the glow plug had been mmed off, as shown in Fig. 11. Applicants have determmed, as shown in Fig. 9, that increasing combustion chamber wall temperatores from 400K to 933K can advance the SOC from 7° ATDC to 14° BTDC. With respect to pressure control, increasing IMP serves to advance the SOC. Fig. 31c, for example, shows that increasing IMP on the single cylinder engine from 52 psia to 57 psia caused the SOC to advance from 3.7° ATDC to 1.5° BTDC. Any method of affecting cylinder pressure, such as varying compression ratio or changing valve timing, both illustrated above, can be used to control SOC. With respect to equivalence ratio, applicants have determined, as shown in Fig. 38, show that increasing equivalence ratio from 0.30 to 0.33 by increasing fuel flow to the engine advanced the SOC from 5.5° ATDC to 2.0° ATDC. Also, varying the autoignition properties of the air/fuel mixture by the addition of reactive species or even diluent can affect SOC. Applicants have shown that for the case shown m Fig. 33, increasing the amount of ozone added to the charge from 0 to 36 g/kg of fuel had the effect of advancing SOC from 1° ATDC to 12.5° BTDC. In one smdy where diesel fuel was used m a pilot injection to initiate SOC in an air-propane mbcture, the amount of pilot used affected SOC. For example, when pilot quantity was increased from approximately 0.1% to 18% of the total fuel energy, the SOC advanced from 2° ATDC to 10° BTDC. In one smdy, EGR was used as a diluent to retard SOC while holding IMT constant with an aftercooler. As shown m Fig. 17, when EGR rate was increased from 2.9% to 8.0%, die SOC retarded from 1.2° ATDC to 4.2° ATDC. Applicants have shown that increasing the air/fiiel mixture"s resistance to autoignition by increasmg octane number, for example, can be used to retard SOC. Also, applicants have shown that when octane number was increased from 80 to 100, the SOC retarded from 14° BTDC to 7° BTDC for a case where IMT plus reheat was 31 IK. Of course, any of these control variables could be adjusted in the opposite direction from die above examples to achieve the opposite effect on SOC if necessary. For example, rather than increasing IMT to advance SOC, IMT could be decreased to retard SOC. Also, the magnitudes of such variations would be increased or decreased as necessary to maintain the desired SOC. Applicants have shown that the combustion or heat release duration can be affected by varying different parameters. As SOC is retarded, die heat release duration increases. For example, Fig. 8 shows that as SOC is retarded, by reducing IMT from 195 degrees F to 184 degrees F, the duration increases from approximately 6 degrees to approximately 24 degrees. Similarly, increasing the equivalence ratio decreases die heat release duration. Applicants also believe that increasing die degree of temperature and equivalence ratio stratification of die charge increases the heat release duration. However, given the difficulty of measuring the degree of temperature or equivalence ratio stratification more work is needed to quantify the level of stratification. Of course, given the relationship between SOC and duration, any control strategy that retards SOC should also increase die duration. By maintaining die SOC and the duration of combustion in the desired ranges while controlling the equivalence ratio to ensure lean burn conditions, the control system mininlizes NOx emissions. Also, the present engine design, also reduces UHC and CO emissions by minimizing die crevices in the cylinder thereby minimizing die unbumed gases as shown in Figures 48a-50b. During operation, balancing the combustion processes between the cylinders of the engine of Fig. lb can be accomplished by varying any of the control variables used to control die SOC, as discussed hereinabove. The ECU 20 compares the SOC and duration of combustion data provided by sensor 16 for each cylinder. When die data indicates that die SOC and/or duration of combustion of one or more cylinders is occurrkg outside a predetermined crank angle range, the ECU will determine die appropriate control variable or variables most effective for the given operating conditions and generates a control signal for controlling the control variable to cause the SOC and/or duration of combustion to adjust so as to fall within die desired range. Applicants have determined that cylinder balancing is best achieved by controlling equivalence ratio, adding ozone to the mixture, controlling individual heaters associated with each cylinder intake port, varying compression ratio using device 38 or variable valve timing, adding oil via pilot injection or port fuel injection, port injection of water and/or any of the methods discussed hereinabove for varying EGR or RMF. Any of diese or other forms of combustion control could be used alone, or in a variety of combinations, to enhance combustion balancing control. For example, die combustion control provided by die multiple fuel/additive system described heremabove could be enhanced by providing variable valve timing and/or combustion chamber surface temperature cooling, e.g. engine coolant, or piston cooling nozzle control. Also, one or more glow plugs 44 (Fig. la) may be used as an inexpensive, easy mediod of achieving at least partial control over combustion balancing between the cylinders. It may also be possible to control the EGR rate for each cylinder in order to balance combustion quality. Fig. 58 illustrates another embodiment of the present invention which uses a pumping cylinder 129 to enhance the diermal efficiency of a compression ignition engme. This embodiment works especially weU for a PCCI engme. The mduction of air or an air-fuel mixture into the power cylinders 130 occurs in two stages. The first stage of the induction is at roughly atmospheric pressure through die reed valve or valves 132. The second stage occurs through die heat exchanger 134 and is at a higher pressure. The pumping cylinder 129 allows a nearly constant volume heat transfer through heat exchanger 134. The pumping cylinder 129 operates at low lemperamre and pressure and hence it can be double-acting. The pumping piston could be driven directly or indirectly by the crankshaft. Also, since the pumping cylinder 129 is at low temperature and pressure, it can be made of inexpensive, lightweight materials. Moreover, volumetric efficiency of the pumping cylinder is very high due to the low temperatures in the pumping cylinder 129. The heat exchanger receives a fresh charge of intake air on one side and engine coolant or engine oil on the other side. A bypass valve 136 around heat exchanger 134 can be used to control the temperature of the charge at intake valve closing, thereby controUmg the start of combustion (SOC) for an PCCI engine. Note that the mass flow rate through the turbine 138 is larger than the mass flow rate through the compressor 140. This would minimize air-handling challenges associated with PCCI engines. Figs. 59a-59e illustrate the piston motions for the engine of Fig. 58 with six power cylinders 130. Although six power cylinders and one pumping cylinder 129 are shown, more or less power cylinders could be used with a single pumping cylinder. This would require that Miller cycling (early intake valve closing) be used to achieve a nearly constant volume heat addition through the heat exchanger. Miller cycling would improve the thermal efficiency smce the expansion ratio would be greater than the effective compression ratio. This assumes that the pumping cylinder 129 has the same displacement as the other cylinders. As can be seen in Figs. 59a-59e, the piston motions allow for convenient cycle arrangement for a four stroke engine. If the nearly constant volume heat transfer was not used, the mass of charge mducted would be smaller for the same temperature at intake valve closmg. Since more mass is mducted, the GIMEP will be higher for the same equivalence ratio. The pumping cylinder 129 does virtually no work since the difference in pressure across the pumping cylinder"s double-acting piston is nearly zero. In order to achieve high GIMEP levels, modern diesel engines utilize turbo charging. Supercharging can also be used. However, a turbocharger is only about 50% efficient overall: "Hoverall ~ "Hcompressor ^ "Hshaft ^ "Hturbine " 0.7X0.95X0.7 Tlovcrall ^ 47% Similarly, a conventional supercharger is about 50% efficient overall: "Hoverall ~ "Hmechanical ^ "Hcompressor = 0.9X0.6 = 54% Furthermore, the supercharger"s power ultimately comes from the engine, which has approximately 85% mechanical efficiency. Superchargers that utilize sonic effects can have higher overall efficiencies. Hence, there is a conflict between the need to have high GIMEP and high thermal efficiency. High exhaust temperatures and pressures are required to drive a turbocharger. But high exhaust temperatures and pressures indicate that much exhaust energy remams untapped. This exhaust energy will be inefficiently utilized by the turbocharger. Furthermore, high exhaust manifold pressures reduce the BMEP for a constant intake pressure. Similarly, driving a supercharger reduces shaft work in that the supercharger must ultimately be driven by the engine. This may not hold true for a supercharger utilizing sonic effects. If it were possible to reduce the quantity of air required for an internal combustion engine, the overall efficiency could be improved. This could be done by operating an engine with an equivalence ratio Spark ignition, diesel and PCCI engines could be operated in this manner. Although 50% EGR and an equivalence ratio of 0.5 are given as examples, other values are possible. The present erabodunent, as shown in Fig. 60, includes an engine having eight power cylmders J and one double-acting pumping cylinder G although other combinations are possible. The engine also includes a turbocharger T although the same concept applies for a supercharged engine. The pumping cylmder G is double-actmg, and is driven dkecdy or indirectly by the crankshaft. Since the pumping cylinder operates at relatively low temperamres and pressures, it can be made of inexpensive, lightweight material. Also, since the difference in pressure across the pumping piston is small, the work required to drive it is small. Two exhaust valves are required to implement such an engine cycle. Fig. 61 illustrates the valve events for one cylinder. One exhaust valve directs exhaust to be used as EGR. The other directs exhaust to the exhaust manifold. Figs. 62a-62e illustrate the motion of the pistons in the engine of the present embodiment. Good mixing of the fresh air and the EGR is important for most applications. Lowermg the mass flow rate of air through the compressor is especially important for PCCI engines since they operate so lean and have low exhaust temperamres. As can be seen from Figs. 62a-62e, each cylinder operates with approximately 50% EGR. Hence, each cylinder must operate lean of stoichiomefric in order to have sufficient oxygen available to consume all of the fuel. If high GIMEP was required for a short period of time, valve F would be opened, valve D would be shut (see Fig. 60), and valve B would be opened. Variable valve timing may also be used to affect the EGR rate as discussed herein. In Fig. 60, heat exchanger E could be used in conjunction with valve K to adjust the charge temperature. Check valve or reed valve C prevents the flow of fresh air to the exhaust. The pumping cylinder G is provided with appropriate inlet H and oudet valves I. Heat exchanger E may be a charge air cooler, or alternatively, use engine coolant or engine oil as the second fluid flowing through it. This engine scheme could be used for a lean bum SI engine, if desired. Note that because 50% EGR is used, 50% less air flows through the compressor. This results in less power being required to drive the compressor as compared to a conventional engine arrangement. The use of 50% EGR would also lower NOx emissions of a diesel or SI version of tiiis engine arrangement. The pumping cylinder G need not have the same bore as the power cylinders. By changing the pumping cylinder bore or the number of power cylinders served by the pumping cylinder. Miller cycling of this engine arrangement is possible and may result in higher thermal efficiency. Also, some valve overlap of the pumping cylinder mlet valves H may be necessary. Moreover, a manifold may be needed at the compressor outlet. One novel feature of the engine cycle of the present embodiment shown in Fig. 60 is that pistons force the EGR and fresh intake together regardless of their respective pressures. In a conventional engine, it is difficult to have EGR if intake pressure is higher than exhaust pi^essure. For the cycle shown in Fig. 60, the pistons displace a constant volume regardless of intake or exhaust pressures. The tise of pistons to displace constant volumes of EGR and fresh intake results in higher volumetric efficiency of the power cylinders than for a conventional engine. Hence, more mass enters the power cylinder dian would be the case for a conventional engme. This effect raises the GIMEP capability of the engine cycle shown in Fig. 60. Applicant"s smdies show that the NOx emissions of PCCI engines that are operated with no EGR rise dramatically when equivalence ratio is increased to above about 0.55. The use of large quantities of EGR, as is possible with the engine cycle shown in Fig. 60, may lower the NOx emissions of PCCI engines that are operating at equivalence ratios greater than 0.5. Another novel feature of the engine cycle shown in Fig. 60 is that the component arrangement allows lower exhaust pressures for a given intake pressure than would be the case for a conventional engine arrangement. This allows the cycle to have lower "pumping losses" during the intake and exhaust strokes resulting in a higher BMEP than a conventional engine. Another advantage of the cycle shown in Fig. 60 is realized when the engine is operated using a Miller cycle. If a heat exchanger was used to transfer heat from the engine oil, coolant, or exhaust to the fresh intake, thermal efficiency would be improved. Furthermore, for a PCCI engine, peak cylinder pressure is reduced since the temperature at IVC is mcreased. This increased temperature at IVC allows a lower compression ratio to be used and autoignition still achieved at the desired crank angle. The lower compression ratio results in lower peak cylinder pressures and less work required to compress the charge in the cylinder. Furthermore, the use of the pistons to displace die EGR and the fresh intake forces virtually the same mass mto the power cylinder regardless of the temperature of the charge. In a conventional engine, if the intake temperature increases, the mass of the fresh charge inducted decreases proportionally. Hence, for a conventional engine, heating die intake charge reduces power density. Smce Miller cycling would be used in this case, the expansion ratio is greater than die compression ratio. This allows more work to be extracted from the charge, and tiius improves diermal efficiency. Another embodiment of tiie present invention includes a displacer 350 for controlling die start of combustion (SOC) of a PCCI engine. A displacer is similar to a piston and is typically used in Stirling engines to move the working fluid from a cool region to a hot region. The displacer is operated in a maimer sunilar to that described in U.S. Patent No. 856,102 issued to Ossian Ringbom in 1907, the entire contents of which is hereby incorporated by reference. A useful reference related to this patent is Ringbom Stirling Engines by James Senet (1993), ISBN 0-19-507798-9. Operation is as follows. Fig. 63 shows the positions of the engine piston 352 and displacer 350 at BDC of the compression stroke. The displacer includes a displacer piston 354, a displacer rod 356 and a stop 358. The stop 358 is positioned in a gas spring chamber 360 which functions as a gas spring at a pressure Pj Near BDC, PCYL > Ps ^^^ the net force on the displacer at this position is F^^ET = PCYLAR - Ps^s, where Aj is the area of stop 358, AR is the area of displacer rod 356, and PCYL is the pressure in cylinder 362. Since at that point, PCYL^R PjAj and the displacer is forced downward very rapidly displacing the fresh charge from beneath the displacer through the heater and to the hot region near the glow plug, as shown in Fig. 64. That is, when the top of stop 358 of the displacer rod 356 is exposed to Pj (the pressure of the gas spring), the net force on displacer 350, assuming friction is neglected, is F^ET = PCYLAR - PSAR. Since PCYLAR > PSAR, the net force on the displacer is downward. Since PjAj > PSAR, the displacer accelerates more rapidly downward after the displacer has begun to move. Prior to the displacer moving downward, the fresh charge is below the temperamre required for autoignition. After the displacer moves downward, the fresh charge is heated by the heater, the glow plug, and the hot gases and metal m the vicinity of the heater and glow plug. The fresh charge is heated sufficientiy to autoignite within a suitable amount of time. Autoignition of the fresh charge raises the temperature and pressure within the cylinder and the fuel energy is rapidly released. After, or while, the fuel energy is released, the engine piston 352 travels downward toward BDC. Shortly before, or after, the exhaust valve opens, PCYL-^D The pressure inside die chamber 360 that contains the gas spring may be varied by a controller as needed to maintam the SOC at the desired point. The amount of heat supplied by the heater and/or glow plug may also be controlled by the controller. The controller includes some means of determining the SOC. The controller may also include some means of sensing displacer position. With proper design, it should be possible to have a SOC that is significantly after TDC. As Applicant"s smdies have shown, with PCCI engines, this will increase combustion duration. A retarded SOC and longer combustion durations will lower the peak cylinder pressures, which is desirable. The engine of die present embodiment may be an internal combustion engine operating in a PCCI mode. Applicant has demonstrated stable PCCI operation. However, in some instances, it may be desirable, or necessary, to operate the engine in an unstable mode. First, at a steady state condition, increasing IMT advances SOC. Using closed loop IMT control based on a SOC signal derived from the cumulative heat release trace (based on cylinder pressure) as shown in Fig. 53, Applicant was able to operate die engine under steady state conditions that are not inherently stable. Consider Fig. 54 showing a plot of GIMEP and IMT vs. start of combustion. Note that initially when SOC occurs well before TDC, decreasing IMT retards SOC as expected. However, as SOC becomes greater than TDC, the curve of IMT has a positive slope. This means that a higher IMT is required to achieve more retarded combustion. However, at any steady state condition, an increase in IMT will cause combustion to advance resulting in an unstable condition. Note that the maximum GIMEP occurs close to the point where the engine transitions from stable to unstable operation. It is therefore desirable to operate in the unstable region at times. For example, we expect, that for some operating conditions, the optimum timing will be in the unstable region. In many conditions, the optimum point will be very close to the unstable region, and excursions into the unstable region will be required as engine load and speed change. The reason for the unstable behavior is explained as follows: - As discussed earlier, increasing IMT advances SOC and decreasing IMT retards SOC. - As SOC retards, heat transfer goes down. This is partially due to the more retarded heat release placement itself and partially (for very retarded cases) due to a reduction in the percentage of ftiel burned. The reduced heat transfer results in lower wall temperatures which retards SOC. - For retarded timings, the effect of the lower wall temperatores is so strong that steady state retarded conditions require higher IMT as SOC retards, even though for any steady state condition increasing IMT will advance SOC. Therefore, this is an unstable condition. Note the difference between steady state and stable. In this case, steady state means that all the forces that move SOC are balanced so that SOC does not move. Stable means that a perturbation in one of those forces wUl push SOC back to its original value. Unstable means that the permrbation will cause SOC to move away from its origmal value. INDUSTRIAL APPLICABILITY The present PCCI engine and control system may be used in any stationary or nonstationary power plant, including any automotive, industrial, marine or military appilication. The present PCCI engine and control system is especially advantageous m any power generation application where low emissions are desirable. WE CLAIM: 1. An internal combustion engine operable in a premixed charge compression ignition mode, comprising: an engine body; a combustion chamber formed in the engine body for containing a mixture of fuel and air an intake air system for delivering intake air, having at least one of air and a mixture of air and fuel to said combustion chamber, combustion history control system for controlling a combustion history of future, combustion events to optimize engine operation, characterized in that said combustion history control system comprises a temperature control system for ultimately varying the temperature of the mixture of fuel and air, said temperature control system having an intake air temperature control system for controlling the temperature of the, intake air in. the intake system, said intake air temperature control system having a cooler positioned in said intake air system, a bypass passage for bypassing flow around said cooler and at least one flow valve positioned in said intake air system and adjustable to control flow through said cooler and said bypass passage; and a processor adapted to move said at least one flow valve based on engine operating conditions to control air flow so as to optimize engine operation. 2. The engine as claimed in claim 1 ,wherein it comprises an engine operating condition detecting device for detecting an engine operating condition indicative of the combustion history and generating an engine operating condition signal indicative of said engine operating condition; and wherein said processor is adapted to receive said engine operating condition signal, determine a combustion history value based on said engine operating condition signal, and generate one or more control signals based on said combustion history value, said one or more control signals controlling a position of said at least one flow valve to control the combustion history of future combustion events. 3. The engine as claimed in claim 1, wherein said at least one flow valve has only one valve positioned downstream of said cooler. 4. The engine as claimed in claim 3, wherein said single valve is a flapper type valve. 5. The engine as claimed in claim 2 wherein said combustion history control system further has at least one of a pressure control system for ultimately varying the pressure of the mixture, an equivalence ratio control system for varying an equivalence ratio of the mixture and a mixture autoignition property control system for varying an autoignition property of the mixture. 6. The engine as claimed in claim 5, wherein said processor compares said combustion history value to a predetermined desired combustion history value and generates said one or more control signals based on the comparison of said combustion history value to said predetermined desired combustion history value, said at least one control signal controlling said at least one of said temperature control system, said pressure control system, said equivalence ratio control system and said mixture autoignition property control system to cause said combustion history value to approach said predetermined desired combustion history value. 7. The engine as claimed in claim 6, wherein said engine operating condition detecting device has a combustion sensor for sensing a combustion characteristic and generating a combustion characteristic signal, said combustion history value being determined based on said combustion characteristic Signal. 8. The engine as claimed in claim 7, wherein said combustion sensor is a combustion timing sensor for sensing a timing of a combustion event. 9. The engine as claimed in claim 8, wherein said combustion timing sensor is a start of combustion sensor for sensing a start of combustion. 10. The engine as claimed in claim 7, wherein said combustion sensor is a cylinder pressure Sensor. 11. The engine as claimed in claim 1, wherein said intake air temperature control system having a heater positioned in said intake air system. 12. The engine as claimed in claim 1, wherein it comprises an exhaust gas system for directing exhaust gas from said combustion chamber, wherein said temperature control system has an exhaust gas recirculation system for delivering exhaust gas from said exhaust gas system to said intake air system to control intake air temperature. 13. The engine as claimed in claim 1, wherein it comprises of a fuel supply system connected to the engine for directing fuel into at least one of said intake air system and said combustion chamber, said fuel supply system having a first fuel supply and a second fuel supply, said first and said second fuel supplies having different auto ignition properties. 14. The engine as claimed in claim 1, wherein said temperature control system has a residual mass fraction control system for varying a residual mass fraction of combustion products in the combustion chamber. 15. The engine as claimed in claim 14, wherein said residual mass control system has a variable valve timing control system. 16. The engine as claimed in claim 5, wherein it comprises of an exhaust system for directing exhaust gas from said combustion chamber, said exhaust system having an exhaust port associated with said combustion chamber and formed in said engine body, and an exhaust valve operable to open and close said exhaust port in timed relation to a crankshaft angle, wherein said variable valve timing control system has a variable exhaust valve timing control means associated with said combustion chamber for controlling a time duration during which said exhaust valve is open during an intake stroke of the engine so as to cause exhaust gas to flow from said exhaust system through said exhaust port into said combustion chamber to increase an amount of residual exhaust gas in said combustion chamber. 17. The engine as claimed in claim 16, wherein said variable exhaust valve timing control means functions to retard the timing of closing of said exhaust port by said exhaust valve to cause said exhaust valve to close during the intake stroke so as to cause exhaust gas to flow from said exhaust system through said exhaust port into said combustion chamber to increase an amount of residual exhaust gas in said combustion chamber. 18. The engine as claimed in claim 1, wherein said temperature control system has a compression ratio varying system for varying at least one of an effective compression ratio and a geometric compression ratio. 19. The engine as claimed in claim 18, wherein it comprises of an intake port formed in said engine body and an intake valve operable to open and close said intake port, wherein said compression ratio varying system varies the timing of closing of said intake port by said intake valve to vary the effective compression ratio. 20. The engine as claimed in Claim 5, wherein it comprises a rotatable crankshaft and a piston operatively connected to said crankshaft for reciprocal movement through a top dead center position, wherein said one or more control signals controls at least one of said temperature control system, said pressure control system, said equivalence ratio control system and said mixture auto ignition property control system to variably control the combustion history of fiiture combustion events to cause a start of rapid combustion to occur between 20 crank angle degrees before said top dead center position and 10 crank angle degrees after said top dead center position. 21. The engine as claimed in claim 16, wherein said variable exhaust valve timing control means having a hydraulic link system having a hydraulic chamber, a collapsable hydraulic link positioned in said hydraulic chamber and one of a plunger, a valve stem and a push rod mounted for reciprocal movement in said hydraulic chamber. 22. The engine as claimed in claim 21 , wherein said hydraulic link system further has an oil supply passage connected to said hydraulic chamber for supplying hydraulic fluid to said .hydraulic chamber to create a hydraulic link, an oil drain passage connected to said hydraulic chamber for draining hydraulic fluid from said hydraulic chamber to collapse said hydraulic link and a least one control valve for controlling flow through at least one of said oil supply passage and said oil drain passage. 23. The engine as claimed in claim 22, wherein said at least one control valve, has a supply control valve positioned in said supply passage and a second control valve positioned in said drain passage. 24. The engine as claimed in 22, wherein said at least one control valve has a solenoid valve positioned in said drain passage and a check valve positioned in said supply passage. 25. The engine as claimed in claim 1, wherein it comprises an exhaust system for directing exhaust gas from said combustion chamber and a throttle valve positioned in said exhaust system to limit exhaust gas flow thereby increasing a residual mass fraction of combustion products. 26. A premixed charge compression ignition internal combustion engine, comprising; an engine body; a combustion chamber forme""" the engine body; an intake air system for supplying intake air, having at least one of air and a mixture of air and fuel, to said combustion chamber; an exhaust system for directing exhaust gas from said combustion chamber, said exhaust system having an exhaust port associated with said combustion chamber and formed in said engine body, and an exhaust valve operable to open and close said exhaust port in timed relation to a combustion event; combustion history control system for controlling a combustion history future combustion events in said combustion chamber to optimize engine operation, characterized in that said combustion history control system comprises a temperature control system for ultimately varying the temperature of the mixture of fuel and air, said temperature control system having a residual mass fraction control system for varying a residual mass fraction of combustion products in said combustion chamber, said residual mass fraction control system having a variable exhaust valve timing control means associated with said combustion chamber for controlling a time duration during which said exhaust valve is open during an intake stroke of the engine based on engine operating conditions so as to cause exhaust gas to flow from said exhaust port into said combustion chamber to increase an amount of residual exhaust gas in said combustion chamber after a combustion event so as to optimize engine operation. 27. The engine as claimed in claim 26, wherein said variable exhaust valve timing control means retards the timing of closing of said exhaust port by said exhaust valve based on engine operating conditions to cause said exhaust valve to close during the intake stroke so as to cause exhaust gas to flow from said exhaust system through said exhaust port into said combustion chamber to increase an amount of residual exhaust gas in said combustion chamber after a combustion event. 28. The engine as claimed in claim 26, wherein it comprises an engine operating condition detecting device for detecting an engine operating conditions indicative of the combustion history and generating an engine operating condition signal indicative of said engine operating condition, and a processor for receiving said engine operating condition signal, determining a combustion history value based on said engine operating condition signal, and generating one or more control signals based on said combustion history value, said one or more control signals controlling said variable exhaust valve timing control means to control the combustion history of future combustion events. 29. The engine as claimed in claim 26, wherein said combustion history control system has at least one of a pressure control system for ultimately varying the pressure of the mixture, an equivalence ratio control system for varying an equivalence ratio of the mixture and a mixture auto ignition property control system for varying an auto ignition property of the mixture. 30. The engine as claimed in claim 26, wherein said temperature control system intake air temperature control system for controlling the temperature of the intake air in the intake system, said intake air temperature control system has a cooler positioned in said intake air system, a bypass passage bypassing flow around said cooler and a least one flow valve positioned in said intake air system to control flow through said cooler and said bypass passage. 31. The engine as claimed in claim 30, wherein said at least one flow valve has only one valve positioned downstream of said cooler. 32. The engine as claimed in claim 31, wherein said single valve is a flapper type valve. 33. The engine as claimed in claim 32, wherein the engine has a valve train for operating said exhaust valve, said variable exhaust valve timing control means having a hydraulic link system positioned in said valve train. 34. The engine as claimed in claim 33, wherein said hydraulic link system has a hydraulic chamber, a collapsible hydraulic link positioned in said hydraulic chamber and one of a plunger, a valve stem and a push rod mounted for reciprocal movement in said hydraulic chamber. 35. The engine as claimed in claim 34 wherein said hydraulic link system has a supply passage connected to said hydraulic chamber for supplying hydraulic fluid to said .hydraulic chamber to create a hydraulic link, a drain passage connected to said hydraulic chamber for draining hydraulic fluid from said hydraulic chamber to collapse said hydraulic link and a least one control valve for controlling flow through at least one of said supply passage and said drain passage. 36. The engine as claimed in claim 35, wherein said at least one control valve has supply control valve positioned in said supply passage and a second control valve positioned in said drain passage. 37. The engine as claimed in claim 35, wherein said at least one control valve has a solenoid valve positioned in said drain passage and a check valve positioned in said supply passage. 38. The engine as claimed in claim 28 wherein it comprises a plurality of cylinders formed in the engine body, each of said plurality of cylinders defining a respective combustion chamber, having a cylinder combustion balancing means for variably adjusting operating conditions in said combustion chamber of each of said plurality of cylinders to achieve a similar combustion history for combustion events of said plurality of cylinders to balance combustion between said plurality of cylinders. 39. The engine as claimed in claim 26, wherein a plurality of cylinders formed in the engine body, each of said plurality of cylinders defining a respective combustion chamber, having a cylinder combustion balancing means for variably adjusting operating conditions in said combustion chamber of each of said plurality of cylinders to achieve a similar combustion history for combustion events of said plurality of cylinders to balance combustion between said plurality of cylinders. 40. The engine as claimed in claim 39, wherein said cylinder combustion balancing means has said variable exhaust valve timing control means, and variable exhaust valve timing .control means being associated with each of said plurality of cylinders for receiving a variable valve timing control signal from said processor and controlling a time duration during which said exhaust valve opens said exhaust port to vary an amount of residual exhaust gas remaining in said combustion chamber after a combustion event so as to balance combustion events between said plurality of cylinders. 41. The engine as claimed in claim 39, wherein said intake system has a respective intake port associated with each of said plurality of cylinders and formed in said engine body, and a respective intake valve operable to open and close said respective intake port at predetermined times during operation of the engine, wherein said cylinder combustion balancing means has a variable intake valve timing control means associated with each of said plurality of cylinders for receiving a variable valve timing control signal from said processor and varying the time period during which said intake valve is open so as to balance combustion events between said plurality of cylinders. 42. The engine as claimed in Claim 39, wherein said cylinder combustion balancing means has an injector mounted on said engine body for injecting a balancing fluid into one of said intake system and said combustion chamber, said balancing fluid changing the auto ignition property of the mixture. 43. The engine as claimed in claim 39, wherein said cylinder combustion balancing means has a variable compression ratio means associated with each of said plurality of cylinders for varying a compression ratio of each of said plurality of cylinders to maintain combustion balance between said plurality of cylinders. 44. The engine as claimed in claim 43, wherein said variable compression ratio means has a variable valve timing control system for varying an effective compression ratio. 45. The engine as claimed in claim 39, wherein said cylinder combustion balancing means has a fuel supply system connected to the engine for directing fuel into the engine, wherein said fuel supply system has a first fuel supply and a second fuel supply, said first and said second fuel supplies having different ignition properties. 46. The engine as claimed in claim 26, wherein said temperature control system has an intake air temperature control means having an exhaust gas recirculation system for delivering exhaust gas from said exhaust gas system to said intake air system to control intake air temperature. 47. The engine, as claimed in claim 26, wherein said residual mass fraction control means has a throttle valve positioned in said exhaust system to limit exhaust gas flow thereby increasing the residual mass fraction of combustion products. 48. A premixed charge compression ignition internal combustion engine, comprising: an engine body; a plurality of cylinders formed in the engine body; a respective piston mounted in each of said plurality of cylinders for reciprocal movement through successive exhaust and intake strokes, each respective piston defining a , combustion chamber for containing a mixture of fuel and air; an intake air system for supplying intake air to said plurality of cylinders; an exhaust system for directing exhaust gas from said plurality of cylinders; combustion history control system for controlling a combustion history of future combustion events in said combustion chamber of each of said plurality of cylinders to optimize engine operation characterized in that combustion history control system comprises a temperature control system for ultimately varying the temperature of the mixture of fuel and air, said temperature control system having a mass fraction control system for varying a residual mass fraction of combustion products in said combustion chamber, said residual mass fraction control system having a throttle valve positioned in said exhaust system to limit exhaust gas flow thereby increasing the residual mass fraction of combustion products; and a processor for controlling said throttle valve based on engine operating conditions to optimize combustion. 49. The engine as claimed in claim 48 wherein an engine operating condition detecting device for detecting an engine operating condition indicative of the combustion history and generating an engine operating condition signal indicative of said engine operating condition; and wherein said processor for receiving said engine operating condition signal, determining a combustion history value based on said engine operating condition signal, and generating one or more control signals based on said combustion history value, said one or more control signals controlling said throttle valve to variably control exhaust gas flow thereby controlling the residual mass fraction and the combustion history of future combustion events. 50. The engine as claimed in claim 48, wherein said residual mass fraction control system has a variable exhaust valve timing control means associated with each of said plurality of cylinders for controlling a time duration during which said exhaust valve opens said exhaust port during the intake stroke of said respective piston so as to cause exhaust gas to flow from said exhaust system through said exhaust port into said combustion chamber to increase an amount of residual exhaust gas in said combustion chamber after a combustion event. 51. A method for controlling a timing of combustion in a combustion chamber of an internal combustion engine using a timing of combustion control system, the engine operable in a premixed charge compression ignition mode during which the engine is capable of operating in a stable mode wherein a combustion timing will return to a steady state timing value upon the occurrence of a perturbation without the use of the timing of combustion control system permitting continuous stable engine operation and capable of operation in an unstable mode wherein the combustion timing one of advances and retards upon the occurrence of a perturbation without returning to the steady state timing value potentially causing termination of engine operation unless the timing of combustion control system is used, comprising: providing the timing of combustion control system for advancing and retarding the timing of combustion of future combustion events in the combustion chamber, said combustion history control system having at least one of a temperature control system for ultimately varying the temperature of a mixture of fuel and air for combustion in the combustion chamber, a pressure control system for ultimately varying the pressure of the mixture, an equivalence ratio control system for varying an equivalence ratio of the mixture , and a mixture auto ignition property control system for varying an auto ignition property of the mixture; operating the engine with the timing of combustion retarded sufficiently to cause engine operation in the unstable mode; controlling at least one of said temperature control system, said pressure control system, said equivalence ratio control system and said mixture auto ignition property control system while the engine is operating in the unstable mode to compensate for the perturbations to cause timing of combustion to remain at approximately the steady state timing value thereby maintaining steady engine operation in the unstable mode 52. The method as claimed in claim 51, wherein it comprises the steps of: detecting an engine operating condition indicative of the timing of combustion; generating an engine operating condition signal indicative of said engine operating condition; processing said engine operating condition signal and determining an actual timing of combustion value based on said engine operating condition signal; generating at least one control signal based on said actual timing of combustion value; providing said at least one control signal to at least one of said temperature control system, said pressure control system, said equivalence ratio control system and said mixture auto ignition property control system to actively variably control the timing of combustion of future combustion events during operation in the unstable mode. 53. The method as claimed in claim 52,wherein it comprises the steps of comparing said actual timing of combustion value to a predetermined desired timing of combustion value and generating said at least one control signal based on the comparison of said actual timing of combustion value to said desired timing of combustion value, said at least one control signal controlling said at least one of said temperature control system, said pressure control system, said equivalence ratio control system and said mixture auto ignition property control system to cause said actual timing of combustion value to approach said predetermined desired timing of combustion value. 54. The method as claimed in claim 53, wherein a timing of combustion sensor for sensing the timing of combustion and generating a timing of combustion signal, said actual timing of combustion value being determined based on said timing of combustion signal. 55. The method as claimed in claim 54, wherein said timing of combustion sensor is a start of combustion sensor adapted to detect the start of combustion. 56. The method as claimed in claim 51, wherein the steps of providing a rotatable crankshaft and a piston operatively connected to said crankshaft for reciprocal movement through a top dead center position, said unstable operation occurring when a start of rapid combustion occurs after said top dead center position. 57. The method as claimed in claim 54, wherein said timing of combustion sensor is a cylinder pressure sensor.

Documents:

0370-mas-1999 abstract.pdf

0370-mas-1999 claims-duplicate.pdf

0370-mas-1999 claims.pdf

0370-mas-1999 correspondence-others.pdf

0370-mas-1999 correspondence-po.pdf

0370-mas-1999 description (complete)-duplicate.pdf

0370-mas-1999 description (complete).pdf

0370-mas-1999 drawings-duplicate.pdf

0370-mas-1999 drawings.pdf

0370-mas-1999 form-13.pdf

0370-mas-1999 form-19.pdf

0370-mas-1999 form-2.pdf

0370-mas-1999 form-26.pdf

0370-mas-1999 form-3.pdf

0370-mas-1999 form-4.pdf

0370-mas-1999 form-6.pdf

0370-mas-1999 others.pdf

0370-mas-1999 petition.pdf