An Investigation into the combustion characteristics of a ...



An Investigation into the Combustion Characteristics of a SI Engine to Verify Simulation ModelsJack LuLetter of TransmittalAbstractTable of ContentsAcknowledgementsIntroductionBackgroundToday’s automotive industry heavily relies upon the use of engine simulation software to develop and design both race and conventional engines. Sophisticated one dimensional engine Computational Fluid Dynamics (CFD) packages such as RICARDO WAVE? cost millions of dollars but are capable of producing results within an error of 1-3% of dynamometer (DYNO) results1. Like any other CFD or Finite Element Analysis (FEA) software, the results produced are only as good as the data input.UWAM has been utilising WAVE? to design their powertrain package since 2004 and over the past three years have defined their model to within 10% of their DYNO torque curve. The significance of acquiring such an accurate model for UWAM is so that the effects of powertrain hardware can be simulated during the design stage. Without an accurate engine simulation, multiple prototypes have to be manufactured and tested. In some cases large modifications to the engine itself can deem the resultant hardware ineffective and hence ultimately cost the team a large amount of time and finances.Recent work produced by Lu, 2007, and advances in engine research within UWAM have enabled one to retrieve pressure information from within a cylinder of the Honda CBR600R. When this data is logged alongside with crank angle it is capable of producing a Pressure-Volume (P-V) plot that is undoubtedly invaluable to the understanding of the thermodynamic system of the engine.Literature ReviewPast UWAM theses and engineering projects extensively cover intake design and gas exchange for a Formula SAE car particularly naming Kitsios (2002), Inkster (2004), Kawka (2005) and Rogozinski (2006). Kitsios centred his work upon the theoretical fundamentals of gas dynamics and produced a basis for engine simulations upon which Rogozinski built his CFD simulations and experimental analysis upon. The core of Inkster and Kawka’s work revolved around the design of a variable runner intake plenum and proved that educated decisions can be made from full engine simulation models. Paget (2003) evaluated and utilised the Stanford Engine Simulation Software to design for and optimise engine valve train. His work included the use of an in-cylinder pressure transducer but high levels of noise with secondary cyclic signals superimposed upon the main signal were evident. These were concluded to be caused by the existence resonance occurring within the 30mm long bore between the piezoelectric sensor and the cylinder.‘Design and Simulation of Four Stroke Engines’ by Prof. Gordon P. Blair covers engine simulation techniques in high detail with extensive references to the fundamentals of thermodynamics. Experimental research and case studies are provided to verify his models along with a great deal of advice on increasing engine efficiency and performance.‘Internal Combustion Engine Fundamentals’ by John B. Heywood is an extensive review of the vast and complex mass of technical material that now exists on spark-ignition and compressionignition engines. Heywood comprehensively covers all aspects of gas dynamics related to the internal combustion engine by applying the laws of chemistry and thermodynamics. A great deal of Heywood’s work is backed up by experimental results and illustrations.‘Measuring Absolute-Cylinder Pressure and Pressure Drop Across Intake Valves of Firing Engines’ by Paulius V. Puzinauskas, Joseph C. Eves and Nohr F. Tillman is a technical paper describing a technique which can accurately measure firing-cylinder full-load absolute pressure during intake events, thereby providing useful cylinder-pressure data for valve-timing optimisation.‘Spark Ignition Engines – Combustion Characteristics, Thermodynamics, and the Cylinder-Pressure Card’ by Frederic A. Matekunas is a research paper covering the thermodynamics theory behind combustion and discusses about the factors that are important to the timing of the burn for maximum brake torque operation. Combustion Process within the Four Stroke CycleAn internal combustion engine gains its energy from the chemical energy released during the combustion of the fuel/air mixture and therefore the combustion process dictates engine power, efficiency and emissions. The combustion process of a four stroke spark ignition (SI) engine can be divided into four distinct phases: spark ignition, early flame development, flame propagation and flame termination. The four phases lie between the compression and power strokes of the four stroke engine cycle seen in REF _Ref200183170 \h Figure 1. During the intake stroke the piston falls from top dead centre (TDC) increasing the cylinder volume while the intake valve is open. A fresh charge of fuel/air mix is inducted through the intake valve and into the cylinder mixing with the residual gas that remains in the cylinder from the previous cycle. During the compression stroke all valves are closed and the cylinder volume decreases as the piston moves up from bottom dead centre (BDC) compressing the gas mix. The combustion process is initiated by the spark plug towards the end of the compression stroke under normal operating conditions and continues through to the early portion of the power stroke. At this point a turbulent flame develops and propagates through the fuel/air/residual gas mix away from the spark plug and towards the chamber walls before extinguishing. Upon the start of the power stroke the cylinder pressure increases significantly and work is transferred to the piston pushing it down towards BDC ultimately increasing cylinder volume. The exhaust valve opens before BDC and the exhaust stroke expels the exhaust gases from the rising piston leaving some residual gases behind.Figure 1 The 4 stroke engine cycleSpark IgnitionThe ignition within an SI engine is provided by the discharge of the spark plug that is generally controlled by an electronic control unit (ECU). The spark ignition initiates the combustion process and therefore controls the burn.Early Flame DevelopmentThe early fame development (EFD) stage comprises of the flame development process from the spark discharge which initiates the combustion process to a point where a small but measurable fraction of the charge has burned or fuel energy released. In industry it is common to indicate the end of the EFD stage when 10% of the charge mass has been burnt. Other figures such as 1% and 5% have been used also.Flame PropagationThe flame propagation stage comprises of the rapid burning of the charge. During this stage each element of fuel/air burns and its density decreases by a factor of four. The expansion of the combustion product gas compresses the mixture ahead of the flame and displaces it towards the chamber walls. At the same time the already burnt gas behind the propagating flame is compressed and displaced towards the spark plug. Elements of the unburnt gas are of different temperatures and pressures just prior to combustion and are at different states after combustion and their condition is determined by the conservation of mass and energy.Flame TerminationThe flame termination stage comprises of the propagating flame reaching the chamber walls and extinguishing. At this point the combustion process has ended and a large portion if not all of the fuel energy has been released to produce work onto the piston. The amount of fuel energy released is dependent upon the efficiency of the expansion in burn. Variables Effecting CombustionCombustion PhasingCombustion events can be phased by advancing or retarding spark before top dead centre (BTDC). The phasing of the combustion event influences the magnitude and location of peak cylinder pressure by changing the rate of pressure rise within the chamber. REF _Ref200183272 \h Figure 2 illustrates the effects of combustion phasing by spark advance upon cylinder pressure.Figure 2 Combustion phasing by advancing spark timingBy phasing the combustion so that the 50% mass burned point is closer to TDC allows complete combustion at TDC and therefore increases the compression stroke work transfer from piston to cylinder gases resulting in higher cylinder peak pressure. Ultimately this leads to increased work transfer from the cylinder gas to piston upon the power stroke increasing the brake torque output. Matekunas, 1984, introduces the idea of “phase loss” defined as the loss in efficiency as the 50% mass burned point is moved away from TDC. The optimum phasing that provides maximum brake torque (MBT) is known as the MBT point and any timing advanced or retarded from this point increases the “phasing loss” and produces lower torque. MBT phasing often produces a peak pressure location within the range of 14-17deg ATDC. Matekunas (1984)Cylinder TurbulenceThe combustion process in a SI engine occurs in a turbulent flow field. This flow field is produced by the high shear flows generated by the intake jet and flow pattern. In turbulent flows, the rate of transfer and mixing are several times greater than the rates due to molecular diffusion [Heywood (1988)]. One method of adding turbulence within the combustion chamber is known as squish action and this is caused by the geometry of the combustion chamber as the piston rises and compresses the gas. Squish characteristics in SI engines are relatively moderate compared to that of a compression-ignition engine. Another method of promoting turbulence is through swirl and tumble caused by the intake geometry.Swirl and TumbleThe terms ‘swirling’ and ‘tumbling' are used to describe the rotating of flow within the cylinder. Swirl is defined as the controlled rotary motion of the charge about the cylinders axis whereas tumble ( REF _Ref200183382 \h Figure 3) is in cylinder flow at right angles to the cylinder axis. They are created by providing an initial angular momentum to the charge as it enters the cylinder through the intake ports. Swirl and tumble can assist in speeding up the combustion process within SI engines and hence achieve higher thermal efficiency. Figure 3 Tumbling of the intake charge within the cylinderMeasuring Swirl and TumbleSwirl ratioCompression RatioThe compression ratio (CR) is defined as the ratio of maximum volume (when the piston is at BDC) to minimum volume (when the piston is at TDC). At BDC the volume comprises of the swept volume Vs and the clearance volume Vc whereas at TDC the minimum volume at which combustion occurs consists of only the clearance volume Vc.CR=maximum volumeminimum volume=VbdcVtdc=Vs+VcVsIn the APPENDIX Blair (1999) proves that the highest thermal efficiency is achieved at the highest compression ratio but if the compression ratio is too high, engine operation will exhibit abnormal combustion which is an undesirable outcome.Abnormal CombustionNormal combustion is initiated by the discharge of the spark plug and develops a flame that propagates to the chamber walls before extinguishing but there can be several factors to cause abnormal co mbustion. These factors are fuel composition, engine design and operating parameters and combustion chamber deposits (Heywood, 1988). The two most common forms of abnormal combustion are identified as knock and surface ignition. Both of these reduce the combustion efficiency and through persistence will destroy engine components by exceeding the engines pressure design limits. REF _Ref200183451 \h Figure 4 illustrates the difference normal and abnormal combustion as seen from a pressure trace.Figure 4 Pressure trace involving abnormal combustionKnockKnock is described as the sharp metallic noise caused by the auto-ignition of the fuel/air/residual gas mix ahead of the propagating flame. During combustion the propagating flame compresses and displaces the end gas ahead of the flame towards the chamber wall. This causes its pressure, temperature and density to increase undergoing the chemical reactions prior to normal combustion. When pressures and temperatures become excessive the end gas burns very rapidly releasing a large amount of its energy at a rate five to twenty five times normal combustion causing high frequency pressure oscillations within the chamber that exceed engine design limits. These detonations are initiated by high pressures and temperatures and therefore can be avoided by reducing the compression ratio, using a higher rating octane fuel, appropriate calibration of the engines ignition timing and careful design of the engines cooling system. Surface IgnitionSurface ignition is the uncontrolled ignition of the fuel/air/residual gas mix from overheated valves, walls, spark plug or glowing deposits. There are two types of identifiable surface ignition: pre-ignition and post-ignition. Pre-ignition can be identified from a pressure trace as the combustion event is initiated before the targeted spark ignition time and causes the most severe effects as the spark no longer controls the combustion process. Post-ignition occurs after the spark ignition but can be difficult to distinguish from knock as they both portray the same characteristics under a pressure trace.Cyclic VariationsIt is evident from observation of cylinder pressure versus crank angle measurements over consecutive cycles within a sample that cyclic variation exists. For a motored pressure trace cyclic variations are negligible and pressure measurements tend to follow closely to the polytropic relationshippVn=constant. Therefore the pressure development is distinctively related to the combustion process which is dependent upon different variables. These cyclic variations are caused by variations in charge motion and mixture motion at the time of the spark, the amount of fuel/air within the cylinder and the fuel/air ratio, and the mixing of the fresh mixture with the residual gases remaining in the cylinder. Along with cylinder cyclic variations there also exists cylinder to cylinder variance in multicylinder engines which are caused by the same reasons. It is important note that due to cyclic variations, the optimum combustion phasing will then be different for each variation of combustion and that the MBT spark advance experimentally found from engine tuning methods described by Bleechmore (2006) is set for the average cycle. Any cycles faster than the average cycle effectively advances spark timing away from the MBT point and any cycles slow than the average cycle effectively retards spark timing away from the MBT bustion ExperimentationAim and MethodologyIt is necessary to measure performance data in order to tune and validate simulation models. The experiments discussed and outlined within this section include the extraction of multiple pressure traces across the range of engine speeds from 3000RPM to 14000RPM, cylinder volume measurements and calculations, and the measurement of port flow coefficients. The data obtained from these experiments are then analysed to further develop the simulation models.Recommended Data MeasurementsAll testing should be performed at wide opened throttle (WOT) to maximise the volumetric efficiency within each RPM. Ideally the following list of performance measurements at WOT should be available when tuning engine models:Brake Power/TorqueMotoring Friction PowerAir flow, Fuel Flow, Air-Fuel RatioIMEP, BSFC, volumetric efficiency or Mass Air Flow (MAF)Intake and Exhaust manifold Temperature and Pressure (Time averaged)Cylinder Pressure and/or Combustion RateDynamic Intake and Exhaust Port PressuresMean Temperatures at Exhaust Ports and Tertiary pipeIt is common to have inaccurate experimental data and measured data should be validated. According to the GT Power manual [REF], one method of validating measured data is to calculate BSFC from volumetric efficiency, fuel-air ratio and brake power/torque.UWAM Dynamometer and Data Acquisition ApparatusThe engine test bed employed during testing consists of an engine, a dynamometer (dyno) and dyno control system, sensor instrumentation, data logger, exhaust extraction and cooling system for both engine and dyno. The dyno control system allows the operator to hold a fairly constant engine speed by applying a torque to the engine via eddy currents for steady state tuning while the data logger records information from the test session.Previous UWAM engine research allows for the majority of the recommended data measurements listed in Section REF _Ref200183491 \r \h 7.1.1 to be logged by the MoTeC M800 ECU. The experiments covered by this investigation utilises the Kistler Type 6005 piezoelectric high pressure transducer with a pressure range of up to 1000bar and an operating temperature range of -50oC to 200oC. Via a 10-32 to BNC gold wire, the pressure transducer is connected to a charge amplifier that converts the charge into a voltage. This signal, along with those produced by the cam pulse generator and crank angle sensor, are then transmitted to the PCI-DAS 4020/12 high speed DAQ card and logged by the TracerDAQ Pro software. Acquiring the Pressure TraceTracerDAQ Pro logs all inputs within the PCI-DAS 4020/12 high speed DAQ card over a time domain. Triggered sampling is possible but did not produce a reading per crank angle which shows a flaw in the programs ability to trigger at high speeds. Therefore a specified sampling frequency must be specified by the user.Sampling FrequencyTo sample the pressure trace accurately, a significant amount of readings were needed to be taken. The minimum amount of readings per cycle that is needed to develop an accurate pressure trace is 360, one reading per degree. Inaccuracies then occur because the real world steady state operation of the engine is not truly in steady state as the engine dyno can only hold engine speeds to within +/- 50RPM accurately due to dyno stability issues discussed by Bleechmore (2006) and therefore sampling the trace with the minimum frequency will eventually lead to inaccuracies as the engine speed phases out with the sampling frequency. Secondly upon logging the data with the minimum frequency, each reading could be taken upon a non-5V point from the crank encoder and therefore yield further inaccuracies therefore a higher resolution is required. REF _Ref200183580 \h Table 1 below outlines the required frequencies that allow sampling to 4 readings per degree providing a high resolution sample. Sampling at a higher frequency than those listed below will result in large files with more data to process than necessary.Engine Speed(RPM)Minimum Freq (Hz)Required Freq (Hz)420 (Motored)25201008025001500060000300018000720003500210008400040002400096000450027000108000500030000120000600036000144000700042000168000800048000192000900054000216000100006000024000011000660002640001200072000288000130007800031200014000840003360001500090000360000Table 1 Required sampling frequencies to log a pressure trace accuratelyReferencing TDCReferencing TDC to the pulses produced by the crank angle sensor is achieved by locating the SYNC signal produced by the cam pulse generator which occurs at a certain number of degrees BTDC depending upon the camshaft timing. MoTeC refers to this signal as the Crank Reference Index Position (CRIP), which is a user input variable found by employing a timing light that is connected to the ignition circuit under cylinder 1 and matching the TDC timing markers on the crankshaft and cover. The unit of CRIP is in deg BTDC and therefore is added onto the SYNC signal to provide the position of TDC. This method of locating TDC is dependent upon the user’s CRIP input and therefore may not locate “true TDC”.A more accurate method of locating and validating “true TDC” can be found through analysing a motored trace without the presence of ignition and fuel. To eliminate losses of suck air within the combustion chamber the engine is cranked via the starter motor at WOT and because the pressures experienced within the cylinder are highly dependent upon heat losses, the motored trace must also be acquired at operating temperatures (ET 80oC). Due to the factor of time and heat loss, TDC does not occur at peak pressure as expected in a theoretical adiabatic process. In fact peak pressure occurs at 1o BTDC due to the heat losses within the compressed charge. The change in volume between 1o BTDC and TDC is insignificant to a noticeable increase in pressure but within this time heat is transferred out of the compressed charge hence slightly dropping the pressure. [REF]Signal to Pressure ConversionThe Kistler Type 6005 pressure transducer provides a gauge reading and does not provide absolute pressures. Therefore the readings taken upon from the experiments need to be referenced to a pressure that is known during the cycle. The intake Manifold Air Pressure (MAP) sensor provides absolute readings after calibrating the sensor within MoTeC and upon the IVC event, we can assume that the pressures within the cylinder is equivalent that of the manifold air pressure as the cylinder is filled with air. Therefore the pressure data is corrected by matching these values to provide absolute pressures.Calculating Cylinder Volume from Crank AngleEngine analysis requires acquired engine data referenced to cylinder volume and this is achieved by data acquisition at known crank angles. Cylinder volume is directly related to crank angle through a slider-crank mechanism. REF _Ref200183740 \h Figure 5 shows the piston restrained to move only in the y-axis while the crank is restricted to a rotational degree of freedom. Figure 5 Schematic representation of a piston within cylinderWhereVc is the clearance volume (clearance + combustion dome)Vg is the gasket volumex is the strokeb is the bore diameterL is the con rod lengtha is the crank radiusθ is the crank angle from TDCThe stroke x is calculated byx=a+L-L2-a2sin2θ+acosθTherefore the total cylinder volume at crank angle θ is the sum of the clearance volume Vc and the swept volume Vs, which is the multiplication of the stroke and the bore diameter. Vcyl=Vc+Vg+xb=Vc+Vg+π?b24?a+L-L2-a2sin2θ+acosθEQUATION is not valid for all engines as it does not take into account the wrist pin offset that some engines use to reduce the effects of piston slap. For the Honda CBR600RR, the engine of interest in this investigation, this method of cylinder volume calculation is valid as there is no offset.The clearance volume Vc is experimentally measured by filling the combustion dome with a Shellsol solution illustrated by FIGURE. The volume in which it retains is measured and added to the volume occupied by the piston clearance of 1mm with a 67mm bore diameter to establish the clearance volume. The combustion dome volume within the experimental engine is found to be 7.6ml across all four cylinders and therefore acquires a clearance volume Vc of 11,125.6mm3.The gasket volume is simply the cylindrical volume within the cylinder that exists due to the gasket that separates the head and block. This is measured by multiplying the cross sectional area of the cylinder bore by the gasket thickness, which is found to be 0.72mm on the experimental engine and therefore producing a gasket volume of 2,538.5mm3. It is important to note that measurements are taken from a used gasket as new gaskets are thicker prior being fitted.The dimensions of the engine components are listed in REF _Ref200183852 \h Table 2 producing a sinusoidal relationship between crank angle and cylinder volume seen in REF _Ref200183837 \h Figure 6. The mathematical model produces a pressure ratio of 11.96:1 and therefore can be deemed valid as the measured compression of the engine was recorded to be 12:1 from a compression gauge. These values agree with the specified compression ratio within the Honda CBR600RR service ponentMeasured DimensionsCrank radius (a)21.25mmBore diameter (b)67mmCon rod Length (L)94mmTable 2 Parameters measured from engine componentsFigure 6 Cylinder volume vs. Crank AnglePressure Trace Data ValidationlogP logVPort Flow TestingDefinition and BackgroundThe air flow characteristics within the combustion chamber are largely affected by the intake and exhaust ports and, in turn, these air flow characteristics have a large impact on combustion. There are two methods of measuring a port’s efficiency; steady-state flow bench testing and computational fluid dynamics simulations. Of the two, this investigation employed the use of a flow bench, which consists of the cylinder head of interest being mounted onto a pipe that replaces the block and acts as a dummy cylinder. Upstream of the port is a pressure box that is connected to a flow straightener, a flow meter and a fan to draw/force air through the system. The flow bench employed is similar to that of seen in REF _Ref200183936 \h Figure 7.Figure 7 Typical flow bench layout (RICARDO)The test consists of lifting the valve in increments of 1mm from very no lift (0mm) to maximum lift (10mm) and forcing air through the ports. A pressure measurement is then taken within the box and a pressure drop is recorded across the port and valve geometry.Flow Coefficient and Discharge CoefficientThe flow coefficient and discharge coefficient are both a measure of the ports efficiency when compared to a theoretical unrestrictive port. These coefficients are used interchangeably as the reference area differentiates the two. The flow coefficient uses the circle of the valve throat (defined by the inner seat diameter) whereas the discharge coefficient uses the valve curtain area. RICARDO. EQUATIONS are the mathematical definitions of the flow and discharge coefficients where Aeff is the effective throat area, D is the reference valve diameter and L is the valve lift. Valve lift is typically specified by the non-dimensional L/D format.Cf=4?AeffπD2Cd=AeffπDLWithAeff=mγRT1γp1p2p11γ2γ-11-p2p1γ-1γWheremmass flow rateγis the ratio of specific heatsRT1p1pressurep2pressure Test Results and DiscussionAny pressure loss due to temperature change should occur normally as the testing described above is typically performed at atmospheric temperature and doesn’t account for charge heating or bustion Characteristics & ResultsCombustion can be analysed from within an SI engine using a pressure trace acquired from within the combustion chamber with relation its crank angle. When a sufficient number of cycles are recorded, the data is capable of producing combustion characteristics such as the pressure plot, P-V diagram, indicated mean effective pressure (IMEP), friction mean effective pressure (FMEP), mass fraction burned (MFB), burn duration and the coefficient of variance (COV). These characteristics are vital for describing the combustion process and its efficiency.PressureDefinition and BackgroundCylinder pressure is the total pressure within the combustion chamber and is commonly logged with crank angle when analysing combustion. Without ignition the combustion does not occur and the pressure recorded describes the motoring pressure within the cylinder seen in FIGURE. This is the pressure that the cylinder experiences from its change in volume. When ignition occurs, the charge mass burns and the cylinder pressure significantly increases causing a higher transfer of work onto the piston. Note that ignition begins BTDC and peak pressure occurs ATDC as seen in FIGURE.Parameters of interest include magnitude and crank angle of maximum pressure, and magnitude and crank angle of the maximum pressure rise. The rate of pressure rise is calculated using the simple numerical differentiation in EQUATION. dpdθ=pi+1-pi-1θi+1-θi-1Ricardo [REF] states that for maximum efficiency the pressure rise rate should be 2.3 bar/degree.Test ResultsPressure-VolumeDefinition and BackgroundThe thermal cycle of an SI four stroke engine can be illustrated by mapping the pressure-volume (P-V) data extracted from a pressure trace. As the cylinder volume is a function of the crank angle, it is possible to relate cylinder pressure to cylinder volume and hence construct a P-V diagram as seen in REF _Ref200184016 \h Figure 8. Typical valve events such as intake valve open (IVO), intake valve close (IVC), exhaust valve open (EVO) and exhaust valve close (EVC) are shown in the diagram along with direction indicators to clarify the process. Figure 8 Typical Pressure-Volume diagram for a four stroke SI engineThe area under the curve is the indicated work per cycle as given by EQUATION where p is cylinder pressure and V is cylinder volume.W/cycle=p?dVFrom REF _Ref200184016 \h Figure 8 it can be seen that there are three distinctive areas known as Area A, Area B and Area C. The integral over the exhaust and intake strokes (Area B + Area C) is the indicated work done on the gas by the piston known as pumping indicated work whereas integral the over the compression and power strokes (Area A + Area C) is the indicated work done onto the piston by the gas known as gross indicated work. The work generated throughout the entire cycle is then known as the net indicated work. Note that work out of the system is negative and work into the system is positive.Test ResultsIndicated Mean Effective PressureDefinition and BackgroundWhile the cylinder pressure and volume varies throughout the engine cycle, an imaginary constant pressure difference can be substituted over the volume change to obtain the same net work (Spencer, 2004). This pressure difference is known as the indicated mean effective pressure (IMEP) and is used to assess combustion performance independent of the size of bore and stroke, speed and number of cylinders in the engine. FIGURE shows a rectangle with a height that represents the pressure difference that is IMEP and contains an equal area representing the identical work done by the complex cycle shape. In accordance to the definition of net and gross work, Elmqvist-M?ller (2006) defines the net IMEP (N.IMEP) and gross IMEP (G.IMEP) in EQUATION and EQUATION respectively.N.IMEP=WnVd=1Vd?0o720op?dVG.IMEP=WgVd=1Vd?360o720op?dVWhere Wn is the net indicated work, Vd is the swept cylinder volume and p is the cylinder pressure. The work obtained is integrated between crank angles where 0o is TDC upon the intake stroke and 720o is TDC upon the end of the exhaust stroke. The difference between net and gross IMEP is known as pump mean effective pressure (PMEP) or pump loss given by EQUATION. PMEP is the measure of work done by the engine expressed in units of pressure and therefore the relationship between N.IMEP, G.IMEP and PMEP is seen in EQUATION.N.IMEP=G.IMEP+PMEPBrunt (1980) [14 in loughbrough thesis] outlines that errors in IMEP calculations are mainly caused by thermal shock, crank angle phasing errors and transducer sensitivity. Minor errors are caused by coarse crank angle resolution, incorrect con rod length, signal noise and integration period error. Test ResultsFriction Mean Effective PressureDefinition and BackgroundThe friction mean effective pressure (FMEP) is the measure of frictional losses that contribute to the lower brake torque experienced at the crankshaft output expressed in units of pressure. The sum of N.IMEP and FMEP then result in the brake mean effective pressure (BMEP) measured at the crankshaft output defined by EQUATION where τ is the brake torque and Vs is the swept volume. BMEP=τ?4πVsBMEP=N.IMEP-FMEPFMEP is mathematically defined by the Chen Flynn (1965) model seen in EQUATION. This experimentally derived model states that the total engine friction is a function of peak cylinder pressure, mean piston speed and mean piston speed [GT Power].FMEP=C+PF?Pmax+MPSF?Speedmp+MPSSF?Speedmp2WhereC is the constant part of FMEPPF is the peak cylinder pressure factorPmax is the maximum cylinder pressureMPSF is the mean piston speed factorMPSSF is the mean piston speed squared factorSpeedmp is the mean piston speedThe two most common methods of measuring engine friction are motoring dyno testing and comparing indicated torque (calculated from cylinder pressure) to brake torque. Motoring dyno testing is recommended out of the two methods due to the difficulty of accurately measuring cylinder pressure across the entire engine cycle and across the multiple cylinders. Unless cylinder pressure measurement is taken from an average from several individual cylinders over several engine cycles, cylinder-to-cylinder and cyclic variations can strongly effect the IMEP measured when comparing to BMEP. Test ResultsMass Fraction BurnedDefinition and BackgroundThe mass fraction burned (MFB) in an engine cylinder is a normalised quantity between a scale of 0 and 1. It describes the chemical energy release as a function of crank angle as it measures charge mass that has been burned during the combustion event. MFB plots are ‘S’ shaped as seen in FIGURE and measures the fraction of charge mass which has burned within the cylinder at a given crank angle. Additionally combustion duration and ignition delay are determined from MFB curves. The ignition delay is the duration in crank angles between the start of combustion and typically 1, 2 or 5% MFB and the burn duration of a cycle is simply calculated by the crank angle duration from xb=0.1 and xb=0.9.The MFB is most commonly estimated by the Rassweiler and Withrow method publicated in 1938 [REF] that is essentially based upon the assumption that during engine combustion, the pressure rise Δp consists of a pressure rise due to combustion Δpc and a pressure change due to a volume change Δpv. ?p=?pc+?pvAt periods where there is no combustion, pressures at the start and end of interval Δθ are related by the polytropic equationpiVin=pjVjnHence the pressure change due to a change in volume is given by?pv=pj-pi=piViVjn-1And the pressure change due to combustion is given by?pc=pj-piViVjnThe subscript i and j denotes the start and end of the interval respectively. The exponent n is referred to as the polytropic expansion/compression constant and is found from the linear gradient constrained within the compression and power strokes of the logP-logV diagram. The polytropic constant is commonly found to be 1.3 (±0.05) for both compression and expansion processes for conventional fuels. Heywood (1988). Since the combustion process does not occur at constant volume, a pressure rise during combustion must be referred to a reference volume, such as that of the volume at TDC.?pc*=?pc?ViVTDCAssuming that the pressure rise due to combustion is proportional to the mass of charge burned within the interval Δθ then the MFB xb at the end of the ith interval is given byxb(i)=mb(i)mb(total)=0i?pc*0N?pc*Where mb is the mass burned, 0 denotes the start of combustion and N is the total number of crank intervals at the end of combustion. This method takes into assumption that the pressure rise due to combustion is proportional to the amount of fuel chemical energy released rather than the mass of mixture burned. There are other methods of measuring MFB such as the Isermann and Muller approximation seen in APPENDIX but the benefits of the Rassweiler and Withrow method is that no additional data (besides pressure and crank angle) is needed. A functional form often used in engine simulation to represent the mass fraction burned versus crank angle curve is the Wiebe function seen in EQUATION.xb=1-exp-aθ-θ0?θm+1Where θ is the crank angle, θ0 is the start of combustion, Δθ is the total combustion duration from xb=0 to xb=1, and a and m are adjustable parameters to change the shape of the curve to fit.Test ResultsCoefficient of VarianceDefinition and BackgroundCycle by cycle variability can be measured by three means: pressure related parameters, burn-rate related parameters and flame front positioning parameters. Pressure related parameters are the easiest to determine and an important measure of cyclic variability that is derived from pressure data is known as the coefficient of variance (COVimep). The COVimep is expressed as a percentage and is defined by EQUATION where σimep is the standard deviation in IMEP and imep is the mean IMEP.COVimep=σimepimep?100σimep=1N-1iNpi-p2The COVimep is a measure of the cyclic variability in the indicated work per cycle and it is noted by Heywood (1988) that vehicle drivability problems arise when COVimep exceeds 10 percent.Test ResultsMeasuring SwirlMeasuring TumbleRICARDO WAVEPort ModellingNow that the data is collected experimentally, the pressure loss within the ports due to the flow over the port surface and changes in flow momentum from the port geometry can be modelled into WAVE. Either coefficients discussed in section REF _Ref199067541 \r \h 7.7.2 can be entered into the flow coefficients profile but WAVE preferably uses Cd as internally, if Cf is used, WAVE converts the values to Cd using the valve reference diameter specified by the user within the valve lift editor. To accurately model the effects of reverse flow, the reverse flow coefficient is recorded, calculated and entered by the same means of achieving the forward flow coefficients with the flow direction reversed.To eliminate the effects of “duplicating” the pressure loss within the model, the ducts and junctions used to simulate the ports should not contribute any additional pressure losses and therefore all friction and pressure loss multipliers should be set to 0 and discharge coefficients set to 1. Any duct bend angles should be set to 0 since the pressure loss due to the ports geometries have been accounted for from the port flow tests. Flow CoefficientsSwirl CoefficientsCombustion ModellingSI WiebeFriction ModelWAVE’s friction model is defined by the modified Chen Flynn (1965) correlation seen in EQUATION. Where Acf, Bcf, Ccf, Qcf are the Chen Flynn coefficients inputted by the user, Pmax is the peak cylinder pressure, Sfact is the speed factor, RPM is the engine speed and stroke is the cylinder stroke.FMEP=Acf+i=1ncylBcfPmaxi+Ccf?Sfacti+Qcf?Sfacti2WithSfact=RPM?stroke2 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download