MASS FRACTION BURNED ALGORITHM BASED ON THE pv PRODUCT

Transcription

1 Journal of KONES Internal Combustion Engines 2002 No. 3 4 ISSN MASS FRACTION BURNED ALGORITHM BASED ON THE V PRODUCT Krzysztof Z. Mendera, Andrzej Syra, Michał Smereka Technical University of Częstochowa Institute of Internal Combustion Engines and Control Engineering Al. Armii Krajowej 21, Częstochowa, mendera@imc.cz.czest.l Abstract A novel simlified algorithm of mass fraction burned (MFB) calculations, based directly on firing engine cylinder ressure was roosed and comared with other MFB algorithms. 1. Introduction Engine in-cylinder ressure signal may be also used for real-time calculations of the mass fraction burned (MFB). This arameter may be alied for closed-loo control of IC engines. Calculation of MFB - a quantity describing the advancement of combustion rocess - can be carried out in several different ways. The most oular method is the Rassweiler and Withrow aroach, which relates MFB to the ressure rise due to combustion. R&W and other common algorithms were resented in [2]. Rassweiler and Withrow rocedure Pressure Ratio Management mb ( i) 0 MFB = = N m ( total) b mot i MFB MPR = 1 = 0 Δ Δ c c m m Proceduce alied by Isermann and Muller MFB V EOC n V n EOC V n SOC SOC n SOC VSOC The drawback of these methods is the neccessity of olytroic index calculations. The results of them may be misleading because stochastic variation of the in-cylinder thermodynamic arameters. 185

2 2. Fitting window length A ractical solution to the evaluation of olytroic indices (for comression and exansion) is to erform it in some crank angle interval (window). The comression index fit determination is carried out over a given crank angle window (tyically 20 ), just before ignition. The exansion index fit is evaluated in a crank angle window close to the end of combustion (but after EOC). Fig. 1. Influence of fitting window length on olytroic indices The size of fitting windows imlies changes of olytroic exonents (see Fig.1) and because of stochastic variations of cylinder thermodynamic arameters (and thus the olytroic indices) during the analysed engine cycle, the selection of window length should be referably done after reviewing the results of indices calculations. 3. Influence of fitting window length on MFB rofiles The influence of window length on MFB is illustrated in Fig.2 for three window lengths: 5, 20 and 50 crank angle degrees 1 (CAD) as the difference between MFB values for given windows of 20 and 50 o (calculated according to Rassweiler and Withrow (marked as R&W in the figure), rocedure of Pressure Ratio Management (Mat) and algorithm alied by Isermann and Müller (I&M)), and those for 5 CAD window. The figure indicates that the resented algorithms have different sensitivity to the fitting window length: - R&W rocedure is ractically nonsensitive to the window length; - PRM has moderate sensitivity to window length changes; - I&M algorithm is the most recetive to fitting window length and thus to the changes of olytroic indices. 1 In resented examle, the given window lengths corresonded to average olytoic indices of 1.246, 1.24 and 1.21 resectively. 186

3 It should be noted that the resented comarison has rather rough character in other engine conditions, the variations may be significantly different. Fig. 2. Influence of window length (the difference between MFB value for 20 and 50 window, marked 1 and 2 resectively, and that of 5 window 3.1. Algorithm based on the roduct of instantaneous ressure and volume of working medium (V algorithm) The algorithm roosed in the current work is based on the First Law of Thermodynamics for closed system: Δ Q ch = ΔU s + ΔL + ΔQ After substitution the equations for sensible internal energy work and heat transfer (according to Woschni), the FLT becomes: cv ΔQch = Δ R cv = Δ R T 0,8 ( V ) ( V ) + ΔV + const ( T T ) ( V ) + ΔV + const T 0,53 0,8 0,53 t V mr This equation roves that heat released is a function of cylinder ressure and volume. Thus, assuming that the heat released at the end of i-th crank interval is roortional to the mass of charge burned, the MFB may be aroximated as: ΔQch mbi MFB = = N N mbtota ΔQ i i ch i V V wall wall = 187

4 The rofile of the ressure and volume (V) roduct changes is shown in Fig.3 and the MFB ratio calculated according to roosed algorithm is resented in Fig.4. Fig. 3. The V roduct rofiles for 25 consecutive cycles Fig. 4. MFB rofiles evaluated according to V algorithm 4. Comarison of MFB algorithms Comarisons of MFB algorithms were erformed using the exerimental results of incylinder ressure measurements of 1-cylinder sark-ignition engine (based on S320 diesel engine; engine secifications are given in table 1). Engine tests were run at constant seed for various ignition timings and air-fuel ratios. Cylinder ressure, crank angle osition and ignition reference data were recorded using data acquisition system consisting of Kistler iezoelectric transducer tye 6061 (installed in a secial 188

5 engine head extension inserted under the head), DAS-12 data acquisition board controlled with LCT rogram [1] and PC class comuter. The system (caable for recording u to 400 consecutive engine cycles) was aced by Kistler tye 6061 shaft encoder (CAM) at 360 oints er engine revolution. No filtering techniques were alied on the cylinder ressure signal data. All other data including emission, inlet manifold ressure, exhaust temerature, air and fuel flows etc. were also recorded. The results resented here are based on the data acquired in 25 consecutive engine cycles. The number of analysed cycles was chosen as a comromise between an analysis accuracy and comutational effectiveness - the calculations were erformed using Excel sreadsheet [3]. Table 1 Secification of S320 single-cylinder engine adated for sark-ignition oeration Tye S320 Bore, mm 120 mm Stroke, mm 160 mm Comression ratio 9 Fuel system Single-oint ort injection Seed, rev/min 1000 Ignition Timing variable A/F ratio variable Four MFB algorithms, resented before, were used to evaluate the mass fraction burned rofile of the engine combustion rocesses in each cycle Comarison for constant A/F ratio Measurements were made at at the engine running with constant air-fuel ratio (λ 1.2) and various ignition timings: 24, 17 and 14 degrees before TDC. The results of engine measurements are illustrated in Fig.5. Fig. 5. Averaged ressure traces for 3 ignition timings: 24 o (1), 17 o (2) and 14 o (3) b. TDC (λ=1.2) 189

6 Based on recorded data the MFB analysis (according to 4 resented before algorithms) was erformed and its results (mean for 25 consecutive cycles) are deicted in Fig.6. Fig. 6. MFB algorithms comarison for various ignition timing (see location of 50% MFB) Figure 6 indicates that the differences in MFB rofiles are as follows: carnk angle osition of 50%MFB varies u to 3 crank angle degrees; taking R&W rofile as a reference, the V MFB rofile is slightly (by about 1 o i.e. measuremnt resolutuion) advanced (this advancement grows with the ignition timing advancement), the other are retarded by 1 2 o Comarison for constant ignition timing Data recorded for the engine with constant ignition timing of 17 CAD before TDC and three air-fuel ratios: 17, 14.3 and 11.1 kg air/kg fuel. Mean values for 25 consecutive cycles are illustrated in Fig.7. and the results of MFB analysis - in Fig.8. Fig. 7. Averaged ressure traces for different A/F ratios:17 (1), 14.2 (2) and 11.1 (3) (sark timing 17 before TDC) 190

7 Fig. 8. MFB algorithm comarison for various A/F ratios (the art corresonding to 50%MFB was magnified) Figure 8, showing the mass fraction burned rofiles for different air/fuel ratios with an ignition timng of 17 CA degrees, indicates that the differences between various MFB rofiles are less than 2 CAD and that the MFB rofile based on the V roduct is the best aroximation of classical Rassweiler and Withrow rocedure at the crucial (for engine control) location of 50% MFB. 5. Conclusions Diagnostic methods based on the in-cylinder ressure signal may be also used for real-time control of internal combustion engine. The overall energy conversion during combustion can be evaluated based on ressure signal and described by the mass fraction burned (MFB) - a quantity aroximating the advancement of combustion reaction and being the main engine combustion descritor. Quantitative analysis of this arameter may be in near future useful for closed-loo control of IC engines. Actually, the wide alications of ressure-based techniques are still limited due to high cost of the ressure sensors and the intrusive character of their alication but it is exected that the difficulties will be overcome. In site of the rogressive ressure sensor develoment, it should be remembered that in-cylinder ressure ick-u is an alien body in engine combustion chamber and has serious drawbacks: it is difficult to fix in comact chamber of modern IC engine and forms some cavity due to the intrusive measurement technique, a significant baseline drift and high cost, etc. That is why, the MFB evaluation algorithm based on less intrusive method is lanned. Commonly used methods of MFB rofile determination require the knowledge of olitroic indices which may be troublesome and ambigious. The novel MFB algorithm resented in this study is free of this drawback: no extra comutations are neccesary: ressure signal and CAD sensor inuts are sufficient for MFB determination (and 191

8 thus controlling engine oeration). The method roosed here alies ressure and volume data from an engine cycle to evaluate MFB rofile. Crank osition and so, cylinder volume is yet normally determined by an electromagnetic ick-u from a toothed wheel mounted on the crankshaft. Comarisons erformed in this study for different air/fuel ratios and ignition timings indicate that the MFB rofile based on V roduct is the best aroximation of classic Rassweiler and Withrow rocedure. References 1. Gruca M.: LCTxr - rogram do rejestracji i analizy harmonicznej sygnałów. Politechnika Częstochowska Mendera K.Z., Syra A, Smereka M.: Mass Fraction Burned Analysis. Paer submitted for Kones Mendera K.Z.: PawiMBF - rogram do analizy obiegu silnika salinowego (MS Excel). Politechnika Częstochowska,