Combustion in Spark-Ignition Engines PDF

Combustion in spark-ignition engines 8.1 Introduction The combustion process is the most important phase in the operating cycle of an internal combustion engine. In this phase, the chemical energy of the fuel is released in order to produce the high pressure and high temperature gasses that expa...

Combustion in spark-ignition engines 8.1 Introduction The combustion process is the most important phase in the operating cycle of an internal combustion engine. In this phase, the chemical energy of the fuel is released in order to produce the high pressure and high temperature gasses that expand and transfer work to the piston and then to the crankshaft. A good combustion process has to be fast and highly repeatable, so that it does not happen in di↵erent ways between cycles. Combustion obviously produces pollutant emissions, which must be reduced at the minimum. In addition to that, fuels also have to respect some regulations and have some properties, like anti-knock properties and volatility properties for SI engines. The region in which combustion takes place is called flame. It is the confined region in which the fuel oxidizing reactions and chemical energy release occur. The flame is usually less than a mm thick. In SI engines, the fuel-air mixture is well mixed before combustion and the turbulence inside of the cylinder guarantees the mixing. the combustion is started by a spark. The reaction process that propagates the premixed flame is in the form of a thin wrinkled sheet through the mixture. The energy release starts relatively slowly, increases to a maximum value and then decreases at the end of the combustion process. The in-cylinder pressure behaves accordingly. 8.1.1 Combustion fundamentals in SI engines We need an high energy source in order to ignite the air-fuel mixture. The energy is released in a small volume of air, causing its temperature to grow until the energy release rate surpasses the heat transfer to the walls, creating a self-sustained reaction. The ignition system has a primary circuit at 12÷48 V and a secondary circuit that can go up to 50000 V. We use an inductor (transformer) between the two circuits in order to increase the voltage. The device that releases the energy is called spark plug and it is made by two electrodes: one is a cathode, grounded to the engine and one is an anode, eclectically insulated. The minimum energy required is the one needed to ignite a specific air-fuel mixture. If more energy is added, we will have some plasma generated in the cylinder and then sustained flame propagation. If less energy is given, we will have flame quenching and the reaction will not 83  Figure 8.1: Flame propagation in the cylinder be completed. Values of minimum energy range from 0.017mJ (hydrogen) to 0.28mJ for methane. Gasoline is around 0.20mJ. The leaner the mixture, the more energy is required to ignite it. Turbulence increases flame speed, but also increases heat transfer to the walls. Conventional ignition systems provide around 0.40mJ of energy in order to be clear of losses and flow velocity losses. 8.1.2 Flame propagation The speed at which the premixed flame propagates is called flame speed or burning speed. If the flow inside of the cylinder is laminar (no turbulence), the speed is called laminar flame speed SL , which is a characteristic property of the unburned gasses ahead of the flame. It depends from the composition of the charge, the properties of the fuel and the mixture temperature and pressure. Usually, SL < 1m/s. As we can see from Figure 8.1, the zones in the cylinder can be divided in Unburned mixture zone; Pre-heat zone: the zone in front of the flame which receives heat from the combustion going on ahead of it; Conduction-zone: a zone in which the conduction of heat happens in a very strong way, sharply raising the temperature of the unburned mixture; Visible flame zone: we have combustion and a visible flame; 84  Burned gasses zone The flame speed scales with the temperature gradient in the conduction zone. A premixed flame in stoichiometric or rich proportions has the highest laminar speed. In real engines, the laminar conditions are never (or almost never) present. The flame front is, in fact, turbulent. The speed of propagation of this flame is called the turbulent flame speed ST. The turbulent flame is actually a wrinkled laminar flame. The higher the turbulence, the faster the flame propagates and becomes wrinkled. Notice that ST is 3 to 30 times greater than SL. If the flame were not turbulent, engines could not operate in a satisfactory way since the combustion would be too slow. The turbulence in the engine scales with the piston speed and, therefore, the combustion rate scales with the piston speed as well. 8.2 SI engine combustion process When a spark ignition engine is motored, it has a certain pressure in the cylinders, called motored pressure. In a firing engine instead, we have that the pressure does not rise for a small time after the discharge because the energy released by the flame is too small. As the flame grows, the pres- sure rises above the motored pres- sure. This is called firing pres- sure. The pressure reaches a maxi- mum after TDC, before the cylin- der charge is fully burned. The flame development and propagation varies from cycle to cycle since the flame growth depends from the lo- cal mixture composition and turbu- lence. In particular, the composi- tion and motion in the vicinity of the spark plug are the most impor- Figure 8.2: Pressures during combus- tant factors. The fraction of vol- tion ume occupied by the burned gasses rises more steeply than the fraction of mass burned, since the density of the burned gasses is much lower than the one of the unburned ones. 8.2.1 Phases of combustion We can divide the combustion in four phases: 1. Spark ignition; 85  (a) Spark advance vs pressure (b) Spark advance vs relative torque Figure 8.3: E↵ect of spark advance 2. Early flame development; 3. Flame propagation; 4. Flame termination. The combustion event must be properly phased relative to TDC position to obtain the maximum power or torque. The flame development and propa- gation lasts between 40 and 60 CA. The combustion starts before the end of compression and lasts after the peak cylinder pressure occurs (during the expansion stroke). As we vary the spark ignition timing (spark advance) the torque produced varies as well. If we initiate the combustion in advance before TDC, the work transferred to the piston will increase. The limit is that the combustion must still happen very close to TDC or we will have a loss of work. We also need to make sure that the combustion does not stop too late into the expansion stroke or we will have a reduction of work. The optimum timing gives the MBT (maximum brake torque). Sometimes we like to retard the combustion with respect to MBT in order to improve other areas like the post oxidation of HC and CO in the cylinder and exhaust port. There are empirical rules used to correlate the mass burning profile and the maximum cylinder pressure with the MBT timing or the 50% mass fraction burned (MFB50). 8.2.2 Abnormal combustion Several factors like fuel composition, engine design and operating parame- ters, might prevent the normal combustion process giving rise to an abnor- 86 mal combustion. The two types of abnormal combustion that we know of are the knock and the spontaneous ignition. The knock is a phenomenon in which a portion of the fuel (usually at the end of the cylinder, close to the piston) self-ignites because of an excessive pressure or temperature. This phenomenon causes very powerful pressure oscillations inside of the cylinder that produce a sharp metallic noise called knock (hence the name of this phenomenon). Spontaneous ignition is similar to the knock but in this case the whole mixture self ignites. 8.3 Thermodynamics of SI engine combustion In a cylinder, close to TDC, the work is exchanged between gasses and pis- ton. Before TDC the work goes from piston to gas while after TDC the work goes in the opposite way. Heat transfer, instead, occurs from the burned gasses (hot) to the chamber walls. We can consider the pressure inside of the cylinder to be constant (as- suming the combustion zone has a negligible volume). The variation of pressure is only with time. The temperature, instead, is not uniform neither with respect to the cylinder chamber nor with respect to time. As the mixture burns, it expands (density reduces of a factor of four) thus compressing the gasses ahead of the flame (unburned) towards the walls of the chamber. The gasses Figure 8.4: Schematic view of the that have just burned, also com- cylinder upper part press the already burned ones back towards the spark plug. The un- burned gasses that burn at a di↵erent time from the others, have di↵erent pressures and temperatures and therefore end up with a di↵erent state at the end of combustion. This means that the thermodynamic state and com- position of the burned gasses is non-uniform. We have two models that can describe what happens inside of the cylin- der: the fully mixed model and the unmixed model. In the first one, we assume that each element that burns mixes immediately with the already burned mixture, thus resulting in the burned gasses having an uniform av- erage temperature. In the second model we assume the exact opposite: no mixing occurs after combustion. Each burned element is compressed isen- tropically after it burns. The real situation lies in between the two models. It is worth noticing that the di↵erence in temperature between a portion of 87 charge that burns first and a portion which burns last can be as big as 400K in peak temperature. 8.4 Analysis of cylinder pressure data The pressure inside of the cylinder changes with the crank angle as a result of cylinder volume change, combustion, heat transfer to the walls flow in and out of the crevice regions and leakage. Combustion rate information can be obtained from accurate pressure measurements, provided that mod- els for the other phenomena a↵ecting cylinder volume can be developed with a certain degree of approximation. Cylinder pressure is measured with a piezoelectric transducers, which con- tain a quartz crystal. One end of the crystal is exposed to the cylinder pressure (through a diaphragm). As the pressure increases, the crystal is compressed and it produces an electric charge which is proportional to the pressure. A charge amplifier is used to produce a voltage proportional to the charge that can be elaborated. We need to make some calculations and take some actions in order to measure the pressure in this way: 1. Since the transducer cannot provide an absolute pressure, but only a variation from the previous instant, the correct reference pressure has to be adopted (we usually use a piezoresistive transducer installed in the intake manifold or runner) in order to convert the pressure obtained in absolute values; 2. the pressure versus crank angle phasing needs to be accurate within 0.2 CA; 3. the clearance volume has to be estimated with sufficient accuracy. 8.4.1 Maximum pressure in the cylinder For a motored cycle, the maximum pressure is not reached when we are at the maximum compression because of the heat transfer to the walls. In fact, the charge receives heat only in the first part of compression and then gives it in the second part. As a result, the maximum pressure occurs earlier with respect to the minimum volume. The angle between the maximum pressure and TDC is called loss angle and its value can be considered to be approximately 0.6 CA. The loss angle is a↵ected by the engine speed. We can consider also a pV chart in logarithmic scale. The compression process becomes a straight line of slope around 1.3. We can clearly see the start and end of combustion in the graph as they are the points in which the line becomes curved. Compression and expansion can be considered as polytropic transformations, for which pV m = const. 88  Figure 8.5: pV diagram in logarithmic scale This is valid only if leakage is neglected. 8.4.2 Heat release approach The e↵ects of heat transfer, crevices and leakage can be taken into account by using a heat release approach based on the first law of thermodynamics. The chamber content is treated a as single zone. We can compute the first law for open systems as Qch Qht = dUs + W + ⌃hi dmi The piston work is equal to pdV and the mass-related term takes the flow of mass in and out of the cylinder. We have that dUs = mcv (T )dT + u(T )dm. the mean temperature T is evaluated through the ideal gas law and it is close to the mass averaged cylinder temperature during combustion. If we neglect the crevices, we can write Qch Qht = mcv (T )dT + pdV We can use the ideal gas law in order to to write a heat release rate cv cp Qch Qht = V dp + pdV R R We can define the heat release rate, referred to crank variation as dQch dQht 1 dp 1 dV HRR = HRR = = V + p d✓ d✓ 1 d✓ 1 d✓ 89 We can obtain the net or apparent heat release rate by integrating the terms on the right hand side of the equation just described (net is referred to the crank, apparent is referred to what happens in the cylinder). This HRR is often interpreted as the fraction of mass burned or the energy release rate versus the crank angle profile. The heat transfer becomes more prominent as the combustion process ends an the temperature peaks. Another factor that we can easily compute is the heat transfer rate (convective) to the chamber walls: dQht Q˙ht = = Ah(T Tw ) dt 8.5 Combustion process characterization We can use the mass fraction burned or the energy release fraction curve in order to characterize the di↵erent stages of spark ignition combustion. In the flame development process, we have that a small but measurable fraction of mass is burned (in fact, it goes from spark discharge to the moment in which we have this mass burned). It is primarily influenced by mixture state and motion close to the spark plug. In the flame propagation phase, we have that the majority of the charge is burned. this stage is influenced by the conditions throughout the combustion chamber. In the final stage, the last part of the charge is burned. 8.5.1 Parameters used to characterize combustion We use three main parameters to characterize the combustion: 1. Flame development angle, ✓d : the crank angle interval between the spark discharge and the time when a small but significant frac- tion of the cylinder mass has burned or fuel chemical energy has been released. Usually this fraction is 10% (0-MFB10); 2. Rapid burning angle, ✓b : the crank angle interval required to burn the bulk of the charge. It is the interval between the end of the development stage and the end of the flame propagation process (MFB10-MFB90); 3. Overall burning angle, ✓o : the duration of the burning process, the sum of ✓d and ✓b The selection of MFB10 and MFB90 with respect to the crank angle is arbitrary. We can convert crank angle in to time and vice versa by using ✓ = 6n · t (n in rpm, ✓ in degrees and t in seconds). The MFB50 (✓50% ) is a useful parameter in the characterization of the combustion as it represents its barycenter. For a wide range of engines, at MBT, MFB50 occurs around 90  Figure 8.6: Burned mass fraction 5 to 7 CA degrees after TDC. We can use a formula to determine the mass fraction burned with respect to the crank angle and this formula is the Wiebe function: " ✓ ◆ # ✓ ✓0 m+1 xb = 1 exp a ✓ 8.6 Rate of reaction and burning speed After the spark ignition, we have the two phases that characterize combus- tion the most. We have the early flame development, which is related to the propagation of the combustion outward from the charge trapped in the spark plug gap and the flame propagation, which is related to the com- pletion of the chemical reactions of the masses that are ignited. We have two speeds in these two phases, one per each phase. The first one is the rate of reaction, wr , measured in s 1 , which is the measure of the rate at which the compound is burning. The second one is the flame propagation speed (laminar burning speed), SL , which is the velocity of the flame front through the mixture under laminar conditions. It is measured in m/s. We can define the burning speed as the derivative of function of the space inside of the cylinder (x = g(t, T )): @x SL = |T =Ti @t where Ti is the ignition temperature. 91  Figure 8.7: Reaction rate for a fixed concentration 8.6.1 Rate of reaction All chemical reactions happen at a defined rate and depend on the condi- tions of the system. Concentration, temperature, radiation e↵ects and the presence of a catalyst (or inhibitor) are the main factors that a↵ect a re- action. The rate of reaction can be expressed as the rate of decrease of the concentration of the reactants or the rate of increase of the products. In our case, we have many reactions happening during the combustion. For simplicity, we will only consider the following reaction 2CO + O2 () 2CO2 For the law of mass action we can say that the rate of disappearance of a chemical species is proportional to the product of the concentrations of the reacting chemical species each of them raised to a power equal to their stoichiometric coefficient. Since the reaction can happen in both directions, we can define a forward reaction rate ✏+ w! = k e RT [CO]2 [O ] + + 2 and a backwards reaction rate ✏ w =k e RT [CO2 ]2 The reaction rate is defined as wr = w! + w The reaction rate is also a↵ected by temperature for a given concentration. The air-fuel ratio has to be properly chosen so that we obtain the best reaction rate possible. In particular, a rich mixture gives a lower reaction rate. A lean mixture provides a low reaction rate. The optimal concentration (which also gives the highest w!+ ) is usually a slightly rich one, with = (0.85 ÷ 0.9) 92 8.7 Laminar burning speed For the determination of the laminar burning speed, we will consider no radiation heat transfer, no convection heat transfer. The ambient will be open (not a combustion chamber), therefore the combustion happens at constant pressure. We can define the laminar burning speed as dmb /dt SL = AL ⇢ u We are basically defining a mass rate (of burned gasses), which follows the law ṁ = ⇢Aw. The mass flow rate is at the numerator, while AL is the area of the flame front at laminar combustion. The flame is as thin as 0.1mm. As we can see from Figure 8.8, the laminar burning speed is maximum at a slightly rich composition. The presence of exhaust gasses in the cylinder and mixed with the fresh charge, causes a reduction of the laminar burning speed. This is be- cause each element of unburned gas in the mixture reduces the heat- ing value per unit mass of mix- ture and thus reduces the adiabatic flame temperature. Notice that the flame front area is defined as a spherical surface having as radius that corresponds to the advance of the flame front. Experimental mea- surements show that the laminar burning speed is never above 1m/s, which is to slow to have a good op- eration of the engine. In reality, tur- Figure 8.8: Laminar speed depen- bulent motion inside of the cylinder dence on composition make the current much faster. This is why we need to define a turbu- lent burning speed, ST. We can find that speed by considering both radiation and heat transfer by convection. We can find that the turbulent speed is more than 10m/s. 8.8 Charge motion within the cylinder The motion of the piston pulls air inside of the cylinder and the intake flow velocity scales with the piston speed. The gas that flows from the valve is an 93 annular conical jet and the radial and axial velocities in the jet are several times the mean piston speed. These jets produce turbulence, which is a rotational motion characterized by high fluctuating vorticity. Turbulence increases momentum, heat and mass transfer. There will be some amount of dissipation associated with this turbulence. We would like to have some degree of turbulence in our cylinders and we can adopt some devices in order to promote it. Let us now analyze the main types of motion that we can find in a cylinder. (PSAV08 61-71) 8.8.1 Swirl Swirl is defined as an organized rotation of the charge about the cylinder axis. We can create swirl by bringing the intake flow into the cylinder with an initial angular momentum. This swirl will be reduced by engine friction, but it lasts during the whole process. If the piston head has a bowl-in-piston, the swirl motion will be substantially modified during the compression stroke. This type of turbulence is useful in Diesel engines in order to properly mix air and fuel, but it is also beneficial in SI engines since it increases the combustion velocity. In order to obtain swirl, we can use a shrouded valve. The intake valve has a part of it that is closed by a deviator tile that stops the flow from going through a side of the valve. The flow is thus compelled to only go in one side of the valve, which gives some angular momentum to it. Other solutions to have swirl can be to have pipes and valves placed tangentially to the cylinder walls, thus making sure that the flow has angular momentum. Lastly, another solution can be to use a throttle valve in one of the two intake valves (two intake and two exhaust valves per cylinder configuration). The valve can be partially closed or open. When it is closed, the flow will have to go mainly through the valve that has no throttle valve, so it will have a large angular momentum. 8.8.2 Tumble Tumble is the rotation of the charge around an axis orthogonal to the cylin- der axis. Usually, we can generate tumble by positioning the intake ports in a way that they bring the charge into the cylinder through the upper portion of the open area between valve head an valve seat. This impacts the transverse motion of the entering charge. The piston takes the flow down and towards the opposite side of the cylinder, then the flow goes around the piston and finally comes back upwards. The engine volumetric efficiency will be lower by this practice since the flow resistance will increase in the intake system. We use tumble to achieve faster burn rates. If we wanted tho have a rotation in the opposite direction, we would have to use a port that di- rects the flow vertically downward through the valve. This is called reverse tumble. Usually we use this tumble in direct injection gasoline engines to 94  Figure 8.9: Schematic view of the turbulent flame control the motion of the evolving fuel-air mixture cloud. 8.8.3 Squish Squish is the radial or transverse motion that occurs at the end of the compression stroke (and in the early part of the expansion stroke) when a portion of the piston face and cylinder head approach each other closely. If the piston crown and cylinder head are flat, we only have axial fluid movement. If, instead, the cylinder head or piston crown are not flat but shaped (pent-roof, bowl-in-piston) they also generate transverse or radial flows due to the di↵erence in axial distance between the cylinder head and piston crown surfaces across the cylinder cross-section. 8.8.4 Eddies The largest eddies that we have in our flow are limited by the system bound- aries. The smallest eddies are limited by the molecular di↵usion. Eddies are responsible for most of the turbulence during intake. 8.9 Flame structure and turbulent burning speed In Figure 8.9, we can see the structure of the turbulent flame. The flame is a wrinkled thin reaction sheet laminar flame. We can identify an average leading and trailing boundary, with the wrinkles in between. The flame moves towards the unburned charge and we can define its speed considering the mid-line of the two boundaries. The local flame speed is SL but the whole flame front moves at ST. The scale of the wrinkles is about (1÷2)mm. We can define the turbulent speed as we did for the laminar one, this time 95  Figure 8.10: Expansion of the flame front considering the turbulent flame front area: dmb /dt ST = S L = AT ⇢u We can also say that ST AT = SL AL. The turbulent speed increases at higher engine speeds. 8.9.1 Flame structure The leading surface of the inflamed zone is close to spherical. In presence of high swirl, the leading surface is distorted. 8.10 Relations about flame propagation The speed that we experimentally measure in SI engines is even higher than the turbulent one. If we remove the open ambient hypothesis, we can better approximate the real engine speed. The motion of the flame is further complicated by the fact that the burned charge expands and compresses and pushes away the unburned charge. Therefore, the flame motion is the sum of two movements. The first one is the rate at which the flame moves into the unburned part of the charge (ST = Sb , burning speed). The second motion is the rate at which the flame front is pushed forward by the expansion of the burned gasses ahead of the flame front, called transport speed (or mean gas speed) ug. This means that the burning speed is in reality u b = Sb + u g In Figure 8.10, we can see the expansion of the flame front from a time t (left) to a time t + dt (right). The real ub is lower than the theoretical one at the end of the combustion process because the wall e↵ects are significant. The charge temperature is, in fact, lowered nearby the walls and therefore, the ub tends to zero an then 96 the flame deadens. In fact, when the flame front reaches the walls, it is extinguished. Notice that fractions of charge that burn later, are already compressed when the flame front arrives. They will therefore ignite at an higher temperature with respect to the one that the charge that burned first had. 8.10.1 Flame speed dependence on engine speed The burning rate is strongly influenced by the engine speed. The crank-angle degrees duration increases slowly with increasing engine speed. We can see that the flame development angle and flame propagation angle both increase of around 1.6 times for an engine speed that increases four times. As a reminder, the crank angle of combustion is depen- dent from the speed as ✓c = 6n t The combustion time decreases (less than linearly) as the engine speed increases (linearly), since we have Figure 8.11: Dependence of the com- more turbulence in the cylinders. bustion speed on engine speed The engine speed increases more than how the time decreases, thus making the combustion angle increase slightly with the speed. 8.10.2 Flame speed dependence on air-fuel ratio The pattern of the mean expansion speed of the burned gas ub looks similar to the one of the rate of expansion as a function of ↵. The maximum is in the same range of ↵ that gives us the maximum wr. Both flame development and burning angles show a minimum for slightly rich mixtures. Burning angles increase significantly for lean mixtures. If we have burned gas in the mixture (EGR, exhaust gasses,...) we will have a slower flame front. The temperature at the end of the combustion is at the highest for slightly rich mixtures and it is a function of the air-fuel ratio (it sharply drops for leaner mixtures). The burning speed can be defined as s k T w r Tb Ti SB = Ti Tu Tb Tu and we can see that the only variable parameter (for a given fuel) is the burning temperature Tb. The flame can propagate for a shorter range of ↵, while it can burn at larger ranges (but, as we said, it cannot propagate). 97  Figure 8.12: Burning speed with respect to ↵ In Figure 8.12 we can see just that. If we go below the limit for a lean mixture, we will have misfiring. Below the misfiring speeds, we will have no propagation at all. The limit value is usually for ⇡ (60÷80)%. Depending on the desired crank angle duration for combustion, we can set the limits for the ↵ value. 8.11 Dependence of efficiencies on the air-fuel ra- tio 8.11.1 Internal thermodynamic efficiency The amplitude of the combustion angle directly a↵ects the internal ther- modynamic efficiency ⌘✓i. This is because the angle influences the time of combustion and the heat transfer through the walls increases with the combustion angle (it lasts longer). The internal thermodynamic efficiency follows a pattern similar to ub , with a peak for a slightly rich mixture and low to zero values for very rich or very lean mixtures. 8.11.2 All the efficiencies We can plot the behaviour of all of the efficiencies on a single graph, as well as the the imep and tfmep. We will keep v constant as well as the tfmep. We can write the imep as a function of ↵: QLHV QLHV imep = ⌘a f ⌘✓i = ⌘✓i ⌘a f,st v ↵V ↵stV 98  Figure 8.13: Efficiencies as function of ↵ In this expression, everything is constant except for the internal thermody- namic efficiency. This means that the imep has a peak when the internal thermodynamic efficiency is maximum. We can consider the indicated fuel conversion efficiency ⌘f,i = ⌘✓i ⌘a f and write QLHV imep = ⌘f,i v ↵V Since we need to make sure that imep > tf mep, the e↵ective range of imep is actually reduced. In Figure 8.13, we can see the limit: where the imep curve meets the tfmep one. 8.11.3 Bmep and fuel conversion efficiency As we did for the imep, we can plot the bmep as a function of air fuel-ratio (PSAV08 98). We can see that it follows the imep path and has a maximum for a slightly rich mixture. If we instead plot the fuel conversion efficiency as a function of the air-fuel ratio, we can see that it has a maximum for a lean mixture. Since the brake specific fuel consumption is 1 bsf c = ⌘f QLHV we can see that we have less fuel consumption for a lean mixture. In the past (before the 70’s oil crisis) the mixture was rich, in order to increase the bmep. After the oil crisis, the mixture became lean. Nowadays, we use a value near the stoichiometric one, in order to help the three way catalyst to better reduce the emissions. 99  Figure 8.14: Oscillation around the stoichiometric value of 8.11.4 Three way catalyst It is called this way because there are three reactions happening inside of it. Two are the oxidation of HC and CO in CO2 and H2 O and the other one is the reduction of the N Ox into N2 and O2. For the first two we would like to have a lot of oxygen in the ambient ( > 1) while for the reduction we would like to have a low oxygen environment ( < 1). This is why we stay around stoichiometric values: we oscillate around it so that for one part of the cycle the oxidations are favoured, while for the other part of the cycle the reduction is favoured. If the catalyst is warmed up, its efficiency, defined as pollutants out ⌘T W C = 1 pollutants in is actually around 99%. The temperature at which the efficiency is at least 50% for all species is the light-o↵ temperature of the TWC. We usually have < 1 when we start the engine and > 1 when we are at WOT. 8.12 Cyclic variations in combustion We now need to consider the fact that we have substantial variation from cycle to cycle in the combustion process (because we know of the changes in pressure, which are related to combustion). In a multi-cylinder engine, we have variations also between cylinders in a single cycle. These variations are caused by variations in mixture motion at the time of spark, variation of the amount of air and fuel and variation in the mixing of the fresh charge with the residual gasses, especially in the vicinity of the spark plug. These variation are important since the spark timing is based on the average cycle, which means that some cycles will have an advanced spark while others will have a retarded one, with consequent losses of power. Another reason of their importance is that we need to know the extremes of the cyclic variations, since they are the limits of engine operation. At high load, the fastest burning cycles (over-advanced spark) are the most likely to have some knock. Therefore, the fastest burning cycles determine the fuel characteristics for the SI engines (octane requirement) and limit the 100 compression ratio of the engine. The slowest burning cycles (retarded) are most likely to have incomplete burn. These cycles are, therefore, setting the practical lean operating limit of the engine or limit the amount of EGR. 8.12.1 Partial burning and misfire The faster burning cycles have higher values of maximum pressure with respect to slower burning cycles. Their peak of pressure arrives earlier, closer to TDC. As the mixture becomes leaner, the magnitude of the cycle to cycle variation increases. Eventually, some cycles will have such a slow combustion that they will not even complete it. In case we have this kind of cycles, we are in the partial burning regime. If the mixture becomes leaner, we will have misfire: the mixture in some cycles will not even ignite. This operation is undesirable under every aspect. We can measure the cycle to cycle variability in a few ways: Pressure-related parameters: maximum cylinder pressure, crank angle at which the maximum pressure occurs, maximum rate of pressure rise,... Burn rate-related parameters: maximum heat release rate, flame de- velopment angle, rapid burning crank angle,... Flame front position parameters When it comes to pressure-related parameters, they are easy to determine but their relationship with cycle variations are actually difficult to under- stand. one important measure of cycle variability is the coefficient of varia- tion of the imep. It is the standard deviation in imep divided by the average value of of the imep: imep CoVimep = · 100 imep We have driveability problems when we exceed 2% to 5%. Cylinder pres- sure data are often averaged over many cycles to obtain the mean cylinder pressure (ensemble-average) at each crank angle. The primary use of this average pressure versus crank-angle data is in calculating the average indi- cated mean e↵ective pressure. Since combustion parameters are not linearly related to the cylinder pressure, analysis of the average pressure data will not necessarily yield accurate values of average combustion parameters. The error will be most significant when the combustion variability is largest. It is best to determine mean combustion parameters by averaging their values obtained from a substantial number of individual cycle analysis results. The number of cycles, which must be averaged to obtain the desired accuracy depends on the extent of the combustion variability. For example, while 40 to 100 cycles may define imep to within a few percent when combustion 101  Figure 8.15: Control characteristic of an SI engine is highly repeatable, several hundred cycles of data may be required when cyclic combustion variations are large. 8.13 Control characteristic of SI engines The control characteristic of an SI engine is obtained by plotting the fuel conversion efficiency (or the bsfc) against the bmep at constant speed. It is fundamental for the engine performance analysis. In the ideal engine, the control characteristic corresponds to ⌘f = ⌘f,0 = const, which means that the brake power is proportional to the volumetric efficiency and to the air-fuel ratio (curve a). Since the air-fuel ratio is kept usually the same, the volumetric efficiency is the parameter that we use to control the en- gine’s power. We use a throttle valve or the intake system valves (timing and lift) in order to control the volumetric efficiency and, therefore, the power. At constant engine speed, if we decrease the load, the friction losses become more important, which means that the mechanical efficiency goes down (curve b in Figure 8.15) The actual control curve is c, since the fuel conversion efficiency is also influenced by thermal phenomena in the cylin- ders. We can also plot the bsfc as a function of the brake power as we can see in Figure 8.16. Lastly, we can make a geometrical analysis on the normalized ⌘f vs normalized bmep graph (normalized with respect to the maximum value). We can say that 102  (a) bsfc map (b) bsfc as a function of pb Figure 8.16: Control characteristic in terms of bsfc Figure 8.17: ⌘f vsbmep 103  Figure 8.18: Ideal characteristic of a SI engine ⌘f ⌘f,Pb,max mf,max ˙ tan = Pb = Pb,max ṁf If < 45 , we will get a fuel consumption increase but a power reduction! The limit case is = 45 , for which we have a power reduction for the same fuel. 8.13.1 WOT performance characteristic of a SI engine Power, torque and fuel consumption are the most important parameters of an ICE when it comes to performance analysis. The maximum brake power defines the engine’s full potential. The maximum brake torque over the full speed range, indicates the engine’s capability to obtain a high air flow over the full speed range and use that air e↵ectively. The maximum engine op- erating performance (full load) is determined at a test bench in wide-open throttle condition. The ideal torque-speed characteristic has an hyperbolic trend, for which the torque of the engine goes to infinity as the speed of the engine goes to zero and vice versa. In reality, the maximum torque cannot be exceeded and a torque-limiting device kicks in as the required torque be- comes greater than the maximum one. In the range of speeds in which the torque is the maximum one, the power varies linearly with the speed while, after that range, the power is constant. Having a torque that is limited with respect to the ideal one should not be a cause of concern since having a torque greater than the one that the wheels can transfer through grip is 104  (a) WOT engine characteristic (b) WOT engine characteristic with tuning Figure 8.19: Characteristic of the WOT SI engine useless. The air fuel efficiency is not a↵ected by the engine speed as a first approxi- mation. If we consider a more precise analysis, tough, the losses through the exhaust valve increase with the speed, resulting in more mass of residual gas being left in the clearance volume after the combustion. This means having the next combustion happen with a more diluted charge, resulting in lower exhaust temperatures. This improves the air-fuel efficiency since the risk of dissociation is lower. Power and gearbox The maximum torque is achieved for the speed at which a line starting from the origin of a PICE n diagram and tangent to the PICE curve. For the same power, if we go after nmax , we will have a much larger bsfc. The power-speed range of an engine is actually pretty limited. This is why we use a gearbox in order to be able to reach di↵erent speeds for the same torque and vice versa. 105 Figure 8.20: Power characteristic at WOT Abnormal combustion and fuel properties for SI engines In a normal combustion process of an SI engine, the combustion starts at one or more ignition points and propagates smoothly by means of a moving flame front. Combustion is virtually complete when the flame front reaches the last part of the charge. The time required for this process to happen depends form composition, combustion chamber shape and size and engine operating conditions. In an abnormal combustion process, we have some phenomena that make the combustion not smooth and may cause major engine damage if they are too severe. Even when they are not that severe, they still cause excessive engine noise. The two phenomena that cause abnormal combustion are knock and spontaneous ignition. 9.1 Definition of abnormal combustion phenomena Let us now define these two abnormal combustion phenomena, starting from the less severe one: knock. 9.1.1 Definition of knock Knock is the most common abnormal combustion phenomenon and its name comes from the metallic knocking noise that is generated when it happens. It consists in the spontaneous ignition of part of the charge, usually the one situated towards the end of the chamber (end gas). This self-ignition is mainly due to excessive pressure and temperature reached at the end of the chamber before the flame can reach it. The release of chemical energy of the fuel is sudden and powerful and it causes high frequency pressure oscillations inside of the cylinder which produce the characteristic metallic noise. Knock will not occur if the flame consumes the whole charge before the reactions in the end gas have the time to cause auto-ignition. 9.1.2 Definition of spontaneous ignition and engine damage Spontaneous ignition occurs in the surface of the charge, which is in contact with very hot engine components such as the valves or spark plug, com- bustion chamber deposits or by any part of the chamber. Because of this 106 excessive heat, the ignition of the charge occurs maybe before the right time or after the right time. The combustion is no longer controlled by the spark plug and this can lead to severe knock or loss of combustion smoothness. We can have spontaneous ignition followed by knock and vice versa. Spontaneous ignition can cause severe engine damage since it makes the combustion start earlier and increases the maximum pressure that we have at the end of combustion, because of a further compression due to the heat generated by the combustion. The engine chamber and piston might not be able to handle such pressure. Ignition before the right time is the most damaging surface-ignition phenomenon. In any case, we will have higher heat rejection because of increased pressure and temperature. The surfaces that cause self-ignition are the ones that are less cooled and where deposits build up and provide extra thermal insulation. Usually the ignition is caused by an exhaust valve covered by deposits coming from the fuel and the lubricant which penetrates into the combustion chamber. If we properly cool down valves and we adopt rounded metal edges we can reduce the risk of pre-ignition. 9.2 Knock Knock varies cycle by cycle since it is related to fuel properties and cylinder conditions. It also varies from cylinder to cylinder and it might not happen every cycle. The most used method for knock control (other than using a proper fuel) is the spark advance control. By choosing the proper spark advance, we can reduce the risk of knock. Notice that knock is more likely to appear at high engine loads since the parts of the chamber are more ther- mally loaded and might cause surface ignition. First studies on knock were conducted by Ricardo, who noticed that at high load some strange noises occurred. The onset of knock limits the en- gine performance since we have to make sure that we do not have too high temperature or pressure in the chamber (we also limit the compression ra- tio). Properties of the fuels are also another factor in knock onset. We measure the quality of fuels in terms of ant-knock properties with the oc- tane number, ON. 9.2.1 Pressure oscillations and audible noise In Figure 9.1, we can see some examples of cycles with knock of increasing severity. We can see that if knock starts, the pressure is no longer smooth, but oscillates. Where there is knock, the maximum pressure is higher than the regular one and the magnitude of the increase depends from the severity of the knock. The substantial local increase in gas pressure and temperature causes a strong pressure wave (at the speed of sound a) to propagate away from the end-gas region across the combustion chamber: the expansion wave 107  Figure 9.1: Cycles with and without knock that accompanies it, and the reflection of these waves by the chamber walls create the oscillatory pressure versus time records shown in Figs. b and c. Note that once knock occurs, the pressure distribution across the combustion chamber is no longer uniform. These pressure oscillations, for the fluid internal losses, are damped down in the course of time. Transducers located at di↵erent points in the chamber will record di↵erent pressure levels at a given time. The sound that we hear is the result of the intense pressure waves in the charge which force the cylinder walls to vibrate. 9.2.2 Transition from normal to knocking combustion There is no defined point of transition between normal and knocking con- ditions. In fact, as the spark timing is advanced progressively, the pressure and the temperature will increase. At a certain end gas temperature, auto- ignition starts to occur. The rate of energy release by the arts of the charge that auto-ignite is small when we are not in knocking conditions. If the spark timing gets even more advanced, we will have a strong increase in energy release rate. This energy released will increase significantly the pressure in these zones, creating the pressure waves. We define trace knock a condition in which 1% of engine cycles presents knock. 9.2.3 Knock detection The human ear is very good at detecting knock and we can roughly de- termine the octane requirements for a certain engine just by listening to its noise. In order to detect and control knock automatically, we need ac- celerometers that are a cheap but e↵ective devices which can perceive the vibrations due to knock. Other methods are optical probes (high intensity flash is registered when knock occurs) and ionization detectors. We can also use the usual piezoelectric transducer i order to determine the onset of knock precisely. 108 Intensity of knock The amplitude of the pressure fluctuation is a useful measure of the intensity of knock because it depends on the amount of end-gas which ignites sponta- neously and rapidly, and because engine damage due to knock results from the high gas pressures (and temperatures) in the end-gas region. Use of this measure of knock severity or intensity shows there is substantial variation in the extent of knock, cycle-by-cycle. Since the knock phenomenon produces a nonuniform state in the cylinder, and since the details of the knock process in each cycle and in each cylinder are di↵erent, a fundamental definition of knock intensity or severity is difficult. Cylinder pressure is measured by the pressure transducer. The low-frequency component of pressure change due to normal combustion is filtered out (with a band-pass filter) and the rate of pressure rise is averaged over many cycles during the pressure fluctuations following knock. This approach obviously provides only an average rela- tive measure of knock intensity. The maximum amplitude of the pressure oscillations gives a good indication of the severity of knock. 9.2.4 Impact of knock The overall e↵ects of engine knock are: overheating of the engine, with substantially higher heat transfer with the combustion chamber walls (up to 5 times higher); engine performance that worsen in the course of time; a long lasting of the knock can give rise to extensive engine damage (knock over extended periods) The impact of knock depends on its intensity and duration. Trace knock has no significant impact on the engine’s performance and durability. This is why we try to stay around trace knock conditions, so that we maximize the performance without having the risk of engine damage. We can have knock during acceleration (short duration, just an annoyance) and during constant-speed travelling (can lead to engine damage). The damage done to the engine is mainly due to fatigue caused by the fluctuating pressure. In fact, the pressure waves weaken the material and can cause pitting. We can have failure in the piston head gasket, piston ring sticking, piston melting and holing, cylinder head erosion,... 9.2.5 Factors a↵ecting knock Knock intensity grows with the compression ratio rc , the ambient conditions and the engine speed. The greater the spark advance, the greater the knock intensity. Usually knock occurs at wide-open throttle conditions (or close 109 to that condition). Fuel composition is another factor a↵ecting knock and knock intensity. The presence of steam in the combustion chamber (due to the water in the air that was inducted) reduces the risk of knock since it lowers the temperature inside of the combustion chamber. A shorter com- bustion time reduces the risk of knock as well, but if the pressure increases too much, this beneficial e↵ect will not be able to stop knock from happen- ing. 9.3 Knock fundamentals As we said previously, knock is generated by the auto-ignition of a por- tion of the charge before the flame front reaches it because of an excessive pressure or temperature. Auto-ignition does not happen in an uniform way throughout the end gas, but it originates in local regions where the air-fuel composition and temperature are such that the induction time (time re- quired for the oxidation reactions to start) is shorter. These regions that are easier to ignite are called exothermic centers. By using photographs of the combustion, we can clearly see this phenomenon happening. We can relate the frames from the videos of the combustion to the cylinder pres- sure development. In heavy knock conditions we can see the entire end gas region completely ignited with consequent high pressure waves. Instead, under trace knock, we can see that there are only small portions of end gas ignited and the pressure oscillations are almost negligible. 9.3.1 Mechanisms of auto-ignition The intermediate products that are formed before the flame starts (pre- flame products) are fundamental in determining the behaviour of the com- bustion. In particular, hydro-peroxides are important meta-stable inter- mediate products in the chain propagation process in the low temperature ignition process. They have the form ROOH, where R is an organic radical. Auto-ignition can be defined as rapid combustion reaction not caused by an external ignition source. By this definition, compression ignition is a form of auto-ignition. Auto-ignition of a gaseous fuel-air mixture occurs when the chemical energy released by the reaction as thermal energy is larger than the thermal energy lost to the surroundings by heat transfer; as a result the tem- perature of the mixture rapidly increases the rates of the reactions involved (due to their exponential temperature dependence). The state at which such spontaneous ignition occurs is often characterized by the self-ignition tem- perature, and the resulting self-accelerating event where the pressure and temperature increases rapidly is also termed a thermal explosion. 110  Figure 9.2: Graph of a two-stage ignition 9.4 Rapid compression machine A rapid compression machine (RCM) is an experimental device used to simulate a single compression stroke of a combustion engine to study the phenomenon of auto-ignition with compression rates similar to those occur- ring in the end-gas of an engine. The combustion cylinder of the machine is filled with a carefully prepared homogeneous mixture of air and fuel vapor at a given initial pressure and temperature. This mixture is then suddenly compressed to a predetermined volume, (usually in the range of 1/8 to 1/15 of the original volume). The compression time is about a few ms, i.e. about the same time between start of compression and the attainment of maxi- mum pressure in an engine running at 5500 rpm. For each run with the rapid-compression machine, pressure-time records are made. In addition, high-speed motion pictures of flame development are taken for many runs. The main advantage of a rapid compression machine is that the combustion phenomena can be studied in an exhaustive way. In fact, the homogeneous mixture is carefully prepared and the in-cylinder pressure and temperature are varied in the field of interest. The compression ratio can be changed as we wish. We need to make sure that the passage from initial to final conditions is as quick as possible, in order to reduce the chemical reactions to a minimum. The distribution of pressure with respect to time is re- ported in Figure 9.2. In the period ⌧ , called delay period, hydro-peroxides are produced. That pressure step that we can see in the delay period is the one associated with the production of the so-called cool flames. This intermediate step is not always present and it is related to fuel and to ini- tial conditions. Some fuels present a small step in pressure rise followed by a steep increase in pressure where we have the completion of the reac- tion. Other fuels (for example some benzene-based ones) Present a much 111 smoother combustion with the pressure gradually increasing. We can define a couple of combustion parameters: Reaction rate: defined as the maximum value of the derivative of the pressure with respect to time. This is the maximum slope of the pressure-time curve; Reaction time or delay period: also called induction time, it is the time between the end of the compression stroke and the end of the appreciable pressure rise due to reaction. It will be called reaction time; Pre-reactions: the relatively slow reactions that occur during the delay period. By considering several tests that have been performed with di↵erent types of hydrocarbons, we can conclude that: The auto-ignition process involves a period of relatively slow reaction followed by a period of rapid reaction. The relative lengths of the slow and rapid reaction periods, and the maximum rate of reaction, depend both on the composition of the fuel and on the test conditions; A given fuel tends to retain its characteristic type of pressure-time curve, (that is, either rapid or slow reaction rate) over wide ranges of initial temperature, compression ratio, and fuel-air ratio. For a given fuel, increasing the initial temperature decreases the re- action time. Increasing the compression ratio decreases the reaction time. The minimum reaction time occurs near the stoichiometric value of the air-fuel ratio; A fuel found to have high resistance to knock in engines has a longer reaction time and a slower reaction rate than a fuel with less resistance to knock; Auto-ignition reaction not always occur with great suddenness. Re- actions of the type of benzene would not cause the intense pressure waves characteristic of knock in engines. 9.5 Analysis of auto-ignition 9.5.1 Auto-ignition maps We can plot the temperature-pressure curves (values at the end of combus- tion) for a given ⌧. We can see that the reaction time increases when we decrease both temperature and pressure. If we decrease one and increase the 112  (a) Temperature vs pressure (b) Temperature vs pressure, two-stage reac- tion Figure 9.3: Auto-ignition maps other, the reaction time remains constant. At low pressures, the influence of a change of pressure on reaction time, at a given temperature, is large. At high pressures, on the other hand, changes in pressure have a smaller influence on reaction time, as evidenced by the near-horizontal nature of the curves. We can see that for two-stage combustions, the curves have a ”knee”. For isooctane, ignition in the low temperature regions is by a two- stage process: there is a first time interval before the cool flame appears and a second time interval before the appearance of the hot flame. Ignition in the high-temperature region is by a continuous one-stage (or single-stage) process. Cool flame phenomena vary significantly with hydrocarbon struc- ture. There are theories that state that some fuels have knock when they are using a two-stage combustion, some when they use an high temperature one-stage temperature and some fuels in which knock can appear in both situations. 9.5.2 Chemistry of auto-ignition The most widely accepted theory for the chemistry of auto-ignition of fuels is that of chain reaction. A reaction is called a chain reaction when chemical combinations involving active particles generate additional active particles as well as end products of the reaction. The chain-reaction theory postu- lates that certain conditions of pressure, temperature, container material, etc. may promote chain reactions and that other conditions may inhibit them, or act as chain breakers. In complex reacting systems such fuel-air mixture during combustion in engines, the ”reaction” is not a single- or even a few-step process; the actual chemical mechanism consists of a large number of simultaneous, interdependent reactions or chain reactions. There are initiating reactions in which highly reactive intermediate species, called radicals, are produced from stable molecules. Then we have propagation reactions in which radicals react with the reactants and form products plus 113  Figure 9.5: Continuation of the peroxide creation reaction other radicals which continue the chain reaction. Then we have a termi- nation in which the radicals which propagate the chain are removed. Some propagating reactions produce two reactive radical molecules for each rad- ical consumed. These are called chain-branching reactions. When, due to chain-branching, the number of radicals increases sufficiently rapidly, the re- action rate becomes extremely fast and a chain-branching explosion occurs. For our purposes, we can consider a reaction in which peroxides are formed. These peroxides can ex- plain why some fuels knock more with respect to others. In Figure 9.4, we can see the hydrocarbon rep- resented as having three hydrogen atoms and a free bond *. If the structure is flexible enough, the free bond of the oxygen atom O in the organic peroxide can capture one Figure 9.4: Formation of peroxides hydrogen atom present in the hy- drocarbon molecule. The structure as we said, must be flexible, such as the one of long-chain hydrocarbons like straight-chain paraffins (normal-heptane,...) The breaking of the O O bond in peroxides, gives rise to a large heat production. It is a branching reaction which is also strongly exothermic: the gas pressure in the end gas region rises substantially due to this rapid release of energy. The critical mass of peroxides is about (8 ÷ 10)ppm. With a concentration of this order, it is likely that knock will occur. The presence of antiknock agents hinders the peroxide formations and the breaking of bonds: therefore the antiknock agents do not a↵ect the value of ⌧1 but only ⌧2 ( second delay time in a two-stage combustion), lengthening it so as to prevent the formation of hot flames and the auto-ignition. 9.5.3 Considerations on the auto-ignition map Considering the auto-ignition map, it is possible to determine when knock will occur. There is a basic di↵erence between the compression processes in the engine and in the rapid compression machine: in the latter, compression 114 is halted early in the reaction process at a measurable temperature and pressure. In the case of a knocking engine, however, the end-gas is subjected to a continually rising pressure and temperature up to the completion of the auto-ignition process. In order to correlate the two processes, Livengood and Wu proposed the theory of conservation of the delay. The theory is based on the assumption that if the rapid compression machine process occurred in two steps, such that the compression pressure and corresponding temperature were suddenly raised during the delay period to the pressure and temperature of the next step, the fraction of delay consumed in the first step will be transferred to the second step without change. This applies for a combustion of n steps. Basically: t1 t2 tn ti + +... + =⌃ =1 ⌧1 ⌧2 ⌧n ⌧i If we reach 1 before the flame can arrive, we will have knock, while if we do not reach 1 before the flame front arrives, we will not have knock. We can also use an equation in Arrhenius form to express the delay time: ✓ ◆ B C ⌧ = Ap exp T 9.6 Conventional fuels from crude oil Crude oil (or petroleum) is the primary source or feedstock used today to produce transportation fuels, accounting for more than 95% of transport en- ergy. The gasoline (or petrol, UK) and diesel fuel produced from crude oil are the overwhelmingly dominant fuels used in spark-ignition and compression- ignition (or diesel) engines, respectively. Crude oil contains a large number of di↵erent hydrocarbons, and these compounds range from gases to liquids to waxes. In a refinery, the crude oil is physically separated by distilla- tion into various fractions. Portions of these fractions are then chemically processed into fuels and other products. The terms virgin or straight run indicate that no chemical processing has been done to the fraction. Di↵erent chemical processes can be typically used: alkylation, reforming, coking, and catalytic cracking. We can classify the products of crude oil in order of density and volatility: 1. gaseous products: gaseous hydrocarbons are usually associated with liquid petroleum, either dissolved in the liquid or standing above it in the earth. They are the first products that are separated during distillation process, in the form of natural gas or liquefied petroleum gas (LPG); 2. gasoline: it is the lightest petroleum-derived liquid, whose density range from 0.73 to 0.76 kg/dm3 obtained by distillation within 25 and 200 C, that is used primarily as a fuel in spark ignition engines; 115  3. kerosene: : it is the next fraction heavier than gasoline, with density lying between 0.77 and 0.83 kg/dm3 , obtained by distillation in the range of 170 to 260 C and primarily intended for use as fuel in gas turbines and jet engines; 4. diesel oils: they are petroleum fractions that lie between kerosene and lubricating oils and cover a wide range of density (0.815–0.855 kg/dm3 ) and of distillation temperature (180 ÷ 360) C. They are suitable for use in various types of diesel engines; 5. fuel oils: they cover a range of density and distillation temperature similar to that of diesel oils, but since they are designed for use in continuous burners, their composition does not require such accurate control as for diesel fuels. Composition of each of these fuels varies widely depending on the nature of the original crude oil and on the processes used in refining. In any case, the fuel consists of a mixture of hydrocarbon compounds having di↵erent molecular weight and chemi- cal structure and classified on the basis of the number and the position of carbon atoms in the molecules. On average, a refinery will refine 25 to 45% of the input crude oil into gasoline, 25 to 40% into diesel, jet fuel, and heating oil, 5 to 20% into heavy fuel oils, and the remaining 20% into other products. Fuels essentially consist of hydrocarbons composed of the elements C and H. Both gasoline and diesel fuel consist of hundreds of di↵erent hydrocarbon molecules, including a small fraction of organic compounds containing sulphur, nitrogen, etc. The real composition of crude oil widely di↵ers according to its source. 9.6.1 Volatility of liquid fuels Volatility is loosely defined as the tendency of a liquid to evaporate. This quality is of basic importance in spark-ignition engines, because it has a major influence on the fuel vapor-air ratio in intake manifolds and in the cylinders at the time of ignition. In spark-ignition engines, operation will be satisfactory only if the various cylinders receive approximately the same fuel-air ratio and if nearly all the fuel is evaporated before ignition, under all operating conditions, including starting and warming up a cold engine. The volatility of the available fuels influences engine design, particularly in re- gard to size and shape of inlet manifolds and the minimum temperatures of inlet air and inlet manifold required for satisfactory operation. For example, light, i.e., low-boiling, components are responsible for fast starting cold en- gines, good response, and low exhaust emissions during the warm-up period. Too many low-boiling components can lead to vapor bubble formation and greater evaporation loss in the summer. In wet, cold weather throttle valve icing can occur. Too many high-boiling components can lead to condensa- tion on the cylinder walls in cold operation and dilute the oil film and oil 116  Figure 9.6: Volatility curves supply. Too few components in the middle boiling range impairs drivability and may cause bucking during acceleration. After a hot engine is turned o↵ and quickly restarted, the demands on the fuel are precisely the opposite. Under unfavorable conditions parts of the fuel system can become so hot that a large portion of the fuel evaporates, which causes vapor bubbles in the fuel pump or vapor cushions in the fuel injection lines. The figure shows the opposite requirements for cold starting and hot driving. The relative volatility of fuels is ordinarily measured by means of the distillation curves. Typical distillation (or boiling) curves are shown in fig. 9.6. Between two fuels, the one that has the lower curve has the greater volatility. When doing a boiling analysis, the fuel specimen is evaporated and then condensed with variable heat output and a fixed temperature increase of 1 C/min. The resulting boiling curve contains a great deal of information for application engineering. Well balanced boiling behavior is an essential prerequisite for operating vehicles with spark-ignition engines under every condition. 9.7 Fuel factors Individual hydrocarbon compounds vary enormously in their ability to re- sist knock, depending on their molecular size and structure. Increasing the molecular chain length increases the knock risk, while having shorter chains connected side-by-side decreases it. Methyl groups in the chain, in the sec- ond spot from the end or center position, decrease the risk of knock. This is a practice valid especially for paraffins. A double bond has little e↵ect on 117  Figure 9.7: Types of components of the chain the knock tendency. Two or three double bonds are, instead, more e↵ective. Exceptions to this rule are acetylene C2 H2 , ethylene C2 H4 and propylene C3 H6. Napthenes have a great knock resistance. Aromatic rings are very e↵ective because they have three double bonds (PSAV09-80 for graph). 9.8 Motor fuels and knock rating By motor fuels we mean fuels suitable for general use in engines, and more particularly fuels for road vehicles. The Cooperative Fuel Research (CFR) Committee established four basic requirements necessary for a comparative scale of the knock tendency, namely: a standardized engine; a set of standard conditions; a standard measuring method for the intensity of the detonation; a pair of standard reference fuels The parameter that rates the anti-knock properties of fuels is the octane number. We can have two types of octane number. One is the research octane number (RON) and the motor octane number (MON). They di↵er from one another in terms of initial conditions like the temperature at inlet, speed of the engine and spark advance. The conditions for motor method are more severe, so usually MON is smaller than RON. 9.8.1 Standard reference fuels We have two standard reference fuels: the first one is isooctane (C8 H18 , 2,2,4-trimethylpentane), with a sti↵ structure and much less knock tendency than the standard fuel. The second one is the normal heptane (n C7 H16 ). The first one has an octane number ON = 100 while the second one has an octane number ON = 0. Notice that fuels can have both ON > 100 and ON < 0. We can also say that blends of these two fuels have ON 118 which depends from the percentage of the reference fuels inside of them. For instance, a 90% isooctane and 10% n-heptane will have ON of 90. This is how we study the octane number of vehicle fuels: we consider the amount of isooctane in the blend. The reference fuels have the same RON and MON by definition. The test engine is a robust four-stroke overhead-valve engine with an 82.6mm bore and 114.3 mm stroke. The compression ratio can be varied from 3 to 30 while the engine is operating, with a mechanism which raises or lowers the cylinder and cylinder head assembly relative to the crankcase. The spark plug is located in a side position to enhance knock onset. The engine is equipped with several fuel reservoirs, with a selector valve, so the engine can be operated on any of the three fuels: two reference fuels (usually blends of n-heptane and isooctane) and the fuel being rated. Tests are performed at wide-open throttle. With the fuel under test, the fuel-air ratio is adjusted for maximum knock. The compression ratio is then adjusted to produce knock of a standardized intensity. The level of knock obtained with the test fuel is bracketed by two blends of the reference fuels not more than two octane numbers apart (with one knocking more and one less than the test fuel). The octane number of the gasoline is then obtained by interpolation between the knock-meter scale readings for the two reference fuels and their octane numbers. For fuels below 100 ON, the primary reference fuels are blends of isooctane and n-heptane; the percent by volume of isooctane in the blend is the octane number. For fuels above 100 ON, the antiknock quality of the fuel was determined in terms of isooctane plus milliliters of the antiknock additive, tetraethyl lead, per U.S. gallon. 9.8.2 Knock rating of fuels There are several parameters that we can use as rating for fuels in addition the RON and MON. One of them is the fuel sensitivity, that represents the sensitivity of the fuel to initial conditions. It is f uel sensitivity = RON M ON Usually, paraffins have a low sensitivity while olefins and aromatics have high sensitivity. In general, gasolines containing high percentages of saturated hydrocarbons have low sensitivity, while cracked or reformed gasolines which contain large percentages of unsaturated hydrocarbons have large sensitivity. By EN regulation, all gasolines must have a research octane number of at least 95 with ”premium” fuels having ON > 98. In the US we have RON > 91 119 Road octane number We can define a new octane number called road octane number that is measured on a running vehicle. We can define it as z = a(RON ) + b(M ON ) + c where a,b and c are experimentally determined. Octane Index Another parameter that measure how much a fuel is keen to auto-ignite is the octane index, which is defined as OI = (1 K)RON + K · M ON In the US we use the anti-knock index as the main rating RON + M ON AI = 2 Basically it is the octane index with K = 0.5. 9.8.3 Knock suppression We need to make our fuel as knock-resistant as possible in order to be able to have high compression ratios and boost levels of turbocharged engines. We can perform di↵erent strategies to improve anti-knock properties on fuels: choosing the constituents of the hydrocarbons properly, also consider- ing their structure; adding anti-knock agents; blending the fuel with products with high ON In order to improve anti-knock properties, we usually start the fuel im- provement directly during the refining process, choosing mostly isoparaffins, which are paraffins with branched so that the chain is not very long. We never add aromatics, since they are carcinogens. Anti-knock agents In the past, the main anti-knock agents were derivative of lead: tetraethyl lead (C2 H5 )4 P b and tetramethyl lead (CH3 )4 P b. These elements suppress the formation of the peroxides, thus stopping the reactions from happening before the arrival of the flame front. Unfortunately, lead is poisonous for humans and for the environment. It is also inhibiting the functions of the 120 three way catalyst, since it binds with the molecules of the TWC, deacti- vating it. Because of this, lead was gradually removed from fuels. Nowadays, the use of oxygenates in fuels is more widespread. They can be obtained from non-petroleum sources and they have good anti-knock prop- erties. Some examples are methanol (CH3 OH), ethanol (C2 H5 OH),... 9.8.4 E↵ects of engine ageing The octane requirement of an engine-vehicle combination usually increases somewhat during use, primarily due to the buildup of combustion chamber deposits within the engine cylinder. While these deposits increase the engine compression ratio modestly, their largest e↵ect is to increase the tempera- ture of the outer surface of the combustion chamber of the surface of the combustion chamber in contact with the inducted mixture. This increases heat transfer to the fresh mixture during induction and decreases heat trans- fer from the unburned charge during compression. End gas temperatures are therefore higher, thus increasing the likelihood of knock. As the com- bustion chamber deposits stabilize (over some 15,000 km of driving), the engine octane requirement typically increases by about 5 octane numbers. 9.8.5 Commercial vehicle fuels Two (sometimes three) grades of (unleaded) gasoline with di↵erent octane ratings are usually available at refueling stations. In the United States the lower and higher rated fuels are labeled “regular” and “premium” with antiknock indexes of 87 and 93. In Europe the standard and premium fuels have RON values of 95 and 98. In order to encourage the use of biofuels, gasolines with up to 5, 10, or 15% ethanol can be marketed (amount depending on country or region). Both MBTE and ethanol have higher octane numbers than the base gasoline, so they improve the fuels knock resistance. The sulfur levels in these clean unleaded gasolines are being reduced from historical levels (some 300 parts per million, i.e. ppm, by mass) to levels approaching 10 ppm. Sulfur is a catalyst poison that degrades the e↵ectiveness of exhaust emission-control catalyst systems (PSAV09 99-101 for full regulations). 9.9 Knock control strategies There are many factors a↵ecting the onset of knock (from the spark advance to altitude) but, in order to reduce knock in a controlled way, we need to fo- cus on the parameters that we can control, like the spark advance. We could also control the air-fuel ratio, but we usually have it oscillating around the stoichiometric ratio in order to help the TWC to do its job. Usually what we do is to retard the spark ignition. At low load, knock will occur very far 121 from MBT, so we can actually keep the usual spark advance. When we get to higher loads, knock might happen even before MBT, so we need to retard the spark. The engine ECU receives knock data from the accelerometer in the piston in order to detect knock. If knock is present, we will retard the spark. If not, we can keep the current SA or advance it even more. The goal is to be in trace knock condition. At high load, we also have a 20% more rich mixture with respect to the stoichiometric value, in order to increase torque. For around 5% fuel rich mixtures, the octane number required is maximum. One other strategy used to suppress knock onset is the use of cooled ex- haust gas recycle (EGR). This is mostly used in turbocharged gasoline engines where knock significantly constrains engine compression ratio and boost level. The use of EGR at high load dilutes the engine in-cylinder fuel/air/burned gas mixture, which normally reduces the torque a given dis- placement engine produces. However, this knock suppression e↵ect allows a higher compression ratio and boost level to be used which more than o↵sets the charge dilution e↵ect. Direct fuel injection into the cylinder helps suppress knock. The octane requirement decreases by about 4 relative to the equivalent port fuel injec- tion value. As the fuel spray evaporates with direct injection, its heat of vaporization is drawn from the in-cylinder charge, thus cooling that charge. 122