Академический Документы
Профессиональный Документы
Культура Документы
www.elsevier.com/locate/apthermeng
a
Escuela Técnica Superior de Ingenieros Industriales, University of Castilla—La Mancha, Edificio Politécnico,
Avda. Camilo José Cela, s/n. 13071 Ciudad Real, Spain
b
Grupo GASURE, Universidad de Antioquia, Colombia
Abstract
The effect of considering the gaseous fuel as ideal gas during the diagnostic procedure from experimental cylinder pressure signal from
a Diesel engine is analysed. After the chemical characterization of the fuel and the estimation of the thermodynamic properties, the com-
pressibility factor of the fuel was quantified under engine operating conditions, the prediction capability of four cubic state equations was
tested, and finally their application was extended to the whole gaseous mixture throughout the engine cycle. One of the tested equations
was finally selected and used to compare the main results of the combustion diagnostic procedure (mean cylinder temperature and heat
release) with those obtained with the ideal gas state equation, under two selected operating modes. Although the obtained differences are
relatively small, the deviations in the gas temperature around the top dead centre may be important in case that these results are used for
predictions of heat radiation and pollutant formation.
Ó 2006 Elsevier Ltd. All rights reserved.
1359-4311/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2006.01.001
M. Lapuerta et al. / Applied Thermal Engineering 26 (2006) 1492–1499 1493
Nomenclature
Fig. 1. Procedure for the combustion diagnostic in a Diesel engine and the selection of state equation.
tion should not be neglected. As some authors have already The deviations of the gas behaviour with respect to that
reported [7,8], the selection of an appropriate state equa- of an ideal gas are mainly due to the proximity of the ther-
tion is essential for accurately solving the thermal and calo- modynamic conditions to those of saturation of some com-
ric variables, and consequently for an accurate diagnostic ponents, and to the acentric behaviour of other
method, and it is even more important if any elucidation components [9]. In fact, the Redlich–Kwong and Soave
about highly temperature dependant phenomena (heat equations have often been used in literature for calculating
radiation, nitric oxides or soot formation mechanisms, the phase equilibrium of the diesel fuel injected into the
etc.) is pretended. combustion chamber [10–12].
1494 M. Lapuerta et al. / Applied Thermal Engineering 26 (2006) 1492–1499
The objective of this work is to quantify the errors asso- in the p–V diagram. For the other equations, further
ciated with the use of the ideal gas state equation and the corrections were made to improve predictions for acen-
selection of the most appropriate one. With this purpose, tric hydrocarbons, as recommended in literature [9]. The
an additional procedure was integrated in the diagnostic use of the latter equations permitted to evaluate the
model, as sketched in the lower part of Fig. 1. The instan- effect of such acentricity or molecular nonsphericity,
taneous computation of the reduced thermodynamic vari- which in the case of the considered fuel is not negligible.
ables and the acentric factor permits to calculate the 2. Specific enthalpy correlations with temperature for the
compressibility factor through the generalized correlations three considered components and for the gas mixture.
proposed by Lee and Kesler and based on the three-param- The diesel fuel used for the computations is a commer-
eter corresponding states principle [13], and to compare it cial automotive one supplied by Repsol-YPF, attaining
with that provided by the state equation used in the model, the present European standards (EN-590), with a sum-
in order to select the most appropriate one by minimizing marized formula C15.3H27.3. Its chemical composition
errors. is also summarized in Table 2, and its specific enthalpy
was obtained by adding to the standard formation
2. Thermodynamic diagnostic model enthalpy the integral of the specific heat function at con-
stant pressure, whose correlation coefficients were esti-
Although the first principle is solved while intake and mated as an average of those of its components [3].Air
exhaust valves remain closed, the combustion chamber is is considered as dry standard air. Finally, no fix compo-
an open system, due to fuel injection and blow-by. The sition can be taken for the stoichiometric products, due
enclosed gas is a mixture of three components: gaseous fuel to the dependence of dissociation with temperature and
(fg), air (a) and burnt products under stoichiometric condi- pressure. From the resulting compositions of a 12 spe-
tions (b). As diesel combustion is lean, the excess of air cies equilibrium model (N2, O2, CO2, H2O, CO, H2,
mixes with combustion products during and after combus- NO, OH, N, H, O, Ar) at different equilibrium temper-
tion, but is not consumed [4]. This conception of the gas atures and pressures, a single correlation with tempera-
mixture permits to instantaneously identify the gas compo- ture was proposed for stoichiometric products.
sition and to estimate the thermodynamic properties of the However, the effect of pressure was not considered in
gas, as an average of those of the considered components. the correlations for this component, because such effect
Arrows numbers in Fig. 1 indicate the following is only significant at very high temperature, probably
equations: above the range in which the mean cylinder temperature
Table 1
State equations
Equation Formula a b
2
27R T 2c
Van der Waals RT
p ¼ vb va2 a¼ 64pc b ¼ RT
8p
c
c
0:42748R2 T 2:5
Redlich–Kwong RT
p ¼ vb a
ðv2 þbvÞT 0:5 a¼ pc
c
b ¼ 0:08664RT
p
c
c
0:42748R2 T 2c 2
Soave RT
p ¼ vb a
v2 þbv a¼ pc ½1 þ fwð1 T 0:5
r Þ b ¼ 0:08664RT
p
c
c
0:45724R2 T 2c 2
Peng–Robinson RT
p ¼ vb a
v2 þ2bvb2 a¼ pc ½1 þ fwð1 T 0:5
r Þ b ¼ 0:0778RT
p
c
c
2
fw = 0.37464 + 1.54226w 0.2699w
M. Lapuerta et al. / Applied Thermal Engineering 26 (2006) 1492–1499 1495
Table 5
Engine operating conditions
Operation mode Engine speed (rpm) Torque (N m)
A 1526 46
B 1743 77
C 1853 110
AFL 1526 205
BFL 1743 225
Fig. 5. Deviations of the compressibility factor along the engine cycle for
CFL 1853 230
different engine operation modes.
1.025
Peng Robinson
T c ¼ X a T ca þ X fg T cfg þ X b T cb ð4aÞ 1.02
0.995
-100 -80 -60 -40 -20 0 20 40 60 80 100
Crankangle
1.01
0.995
-100 -80 -60 -40 -20 0 20 40 60 80 100
Crankangle
of long-chain hydrocarbons such as those of a diesel These correlations were derived from a modification of
fuel. Again, mole fraction averaging is recommended the Benedict–Webb–Rubin state equation, with experi-
for each component and for the gas mixture [14] mentally fitted constants, and is widely recognized as
one of the most accurate methods for simulating the ther-
w ¼ X a wa þ X fg wfg þ X b wb ð5Þ
modynamic behaviour of hydrocarbon-containing sys-
Table 3 shows the critical parameters and acentric factor tems. As the reduced pressure decreases or the reduced
of each component. temperature increases Z0 tends to unity and Z1 to zero,
3. Determination of the reduced pressure and temperature and thus the effect of acentricity becomes less significant.
of the gas mixture
T p Finally, comparisons were made between the compress-
Tr ¼ ; pr ¼ ð6Þ ibility factors of the gas mixture obtained from Eq. (7) and
Tc pc
from the state equations of Table 1, along the engine cycle.
4. The compressibility factor is obtained for the gas mix- From this comparison, the best state equation was selected,
ture as a function of the reduced thermodynamic vari- and used for comparisons of the diagnostic results with
ables and the acentric factor those obtained with the ideal gas state equation.
Z ¼ Z 0 ðT r ; pr Þ þ wZ 1 ðT r ; pr Þ ð7Þ
4. Experimental work
0 1
where Z and Z were obtained from the generalized
correlations proposed by Lee and Kesler based on the As an example of application of the proposed method,
three-parameter corresponding states principle [13]. the results of the combustion process diagnostic of a turbo-
B
2000 FL (Ideal) B
B (Ideal)
FL (Soave)
1800
In cylinder temperature (°C)
1600
B
1400 B (Soave) (Soave)
1200 B
(Ideal)
1000
B
FL (Ideal)
800 B
FL (Soave)
600
400
-100 -80 -60 -40 -20 0 20 40 60 80 100
Crankangle
35
30
Temperature difference (°C)
25 B
FL (Ideal-Soave)
20
15
B
(Ideal-Soave)
10
0
-100 -80 -60 -40 -20 0 20 40 60 80 100
Crankangle
Fig. 7. Evolution of mean cylinder temperature obtained from ideal state equation and from Soave equation (above) and differences between them
(below).
1498 M. Lapuerta et al. / Applied Thermal Engineering 26 (2006) 1492–1499
charged 0.55 l/cylinder direct injection Diesel engine, sup- 5. Results and discussion
plied by Nissan (see Table 4), similar to those commonly
used in European passenger cars, were used as input. The The evolution of the compressibility factor along the
engine cylinder was instrumented with a water-cooled pie- engine cycle was firstly studied for different engine modes
zoelectric pressure transducer Kistler 6061B and the injec- by plotting the trajectories of the thermodynamic reduced
tor needle was equipped with a displacement sensor. variables on graphics with constant compressibility factor
This engine was tested in six different operating condi- curves in the background. As the acentric factor is modified
tions (see Table 5). Three of them were selected among with time in every case, two different plots were needed to
the collection of steady stages which reproduce the separate the generalized compressibility factor (Fig. 4)
sequence of operating conditions that the vehicles equipped from the deviations due to acentricity (Fig. 5). These fig-
with this type of engines must follow during the transient ures show that the errors in the calculation of the gas tem-
cycle established in the European Emission Directive 70/ perature can reach values higher than 2%, and that the
220, amendment 2001/C 240 E/01. These modes constitute effect of the acentricity could even enhance such errors in
more than 30% in time of the engine typical operation, and case that the fuel concentrations were significant.
are typically selected for combustion and emissions studies This information is quantified for the cases of Mode B
under steady conditions. The other three modes are the and Mode BFL, as an example, in Fig. 6, where the com-
full-load modes corresponding to the same engine speeds. pressibility factor was obtained along the engine crank-
Although the latter are no so usual during vehicle opera- angle from the Lee–Kesler correlations, together with Eq.
tion, they provide extreme pressure, temperature and fuel (7), and from the different state equations. As it can be
concentration conditions, which is helpful to study the observed, all the tested cubic equations improved notice-
range of applicability of the state equations. ably their predictions with respect to the ideal gas equation,
1800
1600
1400 B
FL (Ideal)
1200 B
Heat Released (J)
FL (Soave)
1000
B (Ideal)
800
600 B
(Soave)
400
200
-200
-100 -80 -60 -40 -20 0 20 40 60 80 100
Crankangle
40
35
Heat Release difference (J)
30
B
FL (Ideal-Soave)
25
20 B
(Ideal-Soave)
15
10
0
-100 -80 -60 -40 -20 0 20 40 60 80 100
Crankangle
Fig. 8. Evolution of the heat release obtained from ideal state equation and from Soave equation (above) and differences between them (below).
M. Lapuerta et al. / Applied Thermal Engineering 26 (2006) 1492–1499 1499
but such improvements were even more significant in the results are used to estimate or simulate highly temperature-
case of the acentric equations. This comparison permitted dependent phenomena.
to select Soave equation as the most appropriate one.
Finally, this equation was incorporated in the diagnostic Acknowledgements
model, which permitted to quantify the differences between
the resulting mean temperature and heat release obtained The authors wish to acknowledge the Spanish Ministry
with this equation and with the ideal gas state equation of Science and Technology for the financial support (Pro-
(Figs. 7 and 8). Prior to the analysis of such differences, ject DIMCI, ref: DPI2001-2704-C02), the Spanish Ministry
it can be noticed that the heat release rate and the temper- of Education, Culture and Sports, for the postdoctoral
ature peak are higher at full load, as expected. However, sponsorship of John R. Agudelo at the University of Cas-
compression temperature at partial load is higher than that tilla–La Mancha, Repsol-YPF for the supply and chemical
at full load, due to the increased exhaust gas recirculation. analysis of the diesel fuel, and Marı́a Rodrı́guez de la Ru-
As observed in the figures, differences in mean tempera- bia for her help.
ture are very small if compared with absolute values (max-
imum differences from 20 to 34 °C out of 1600 °C References
variations, which corresponds to deviations between 1.3%
and 2.1%) and heat release (maximum differences from 14 [1] R.B. Krieger, G.L. Borman, The computation of apparent heat
to 38 J, which corresponds to deviations between 1.8% release for internal combustion engines, ASME paper 66WA/DGP-4,
1966.
and 2.1%). In any case, overestimations of the gas temper- [2] D.N. Assanis, J.B. Heywood, Development and use of computer
ature and of the initial heat release rate are achieved if the simulation of the turbocompounded diesel system for engine perfor-
ideal gas state equation is used. mance and component heat transfer studies, SAE paper 860329, 1986.
However, in the case of the cylinder temperature, such dif- [3] H.M. Cheung, J.B. Heywood, Evaluation of a one-zone burn-rate
ferences, occurring very close to the temperature peaks, may analysis procedure using production SI engine pressure data, SAE
paper 932749, 1993.
provoke important errors in the estimation of heat radiation, [4] M. Lapuerta, O. Armas, J.J. Hernández, Diagnostic of DI Diesel
nitric oxide formation and soot formation, because these are combustion from in-cylinder pressure signal by estimation of mean
highly temperature-dependent phenomena. thermodynamic properties of the gas, Applied Thermal Engineering
19 (5) (1999) 513.
6. Conclusion [5] M. Lapuerta, O. Armas, V. Bermúdez, Sensitivity of Diesel engine
thermodynamic cycle calculation to measurement errors and esti-
mated parameters, Applied Thermal Engineering 20 (9) (2000) 843–
The use of ideal gas state equation is generalized in com- 861.
bustion diagnostic procedures aiming to obtain informa- [6] M. Lapuerta, O. Armas, S. Molina, Study of the compression cycle of
tion about the cylinder gas temperature and the heat a reciprocating engine through the polytropic coefficient, Applied
release rate from the experimentally obtained cylinder pres- Thermal Engineering 23 (3) (2003) 313–323.
[7] S.N. Danov, A.K. Gupta, Influence of imperfections in working
sure signal. However, other state equations provide media on Diesel engine indicator process, Journal of Engineering for
improvements in the temperature estimation, as well as in Gas Turbines and Power, ASME 123 (2001) 231–239.
the heat release estimation through the first principle of [8] H. Möhlenkamp, Zur Genauigkeit der Brenngesetzrechnung eines
thermodynamics. Detailed instantaneous estimation of Dieselmotors mit nichtunterteiltem Brennraum, MTZ Motortechni-
sche Zeitschrift 37 (7–8) (1976) 285–291.
the gas mixture thermodynamic properties is needed for
[9] R.C. Reid, J.M. Prausnitz, B.E. Poling, The Properties of Gases and
comparing the state equations behaviour. Among the Liquids, McGraw-Hill Book Company, USA, 1987.
tested state equations, the cubic and acentric ones showed [10] G.S. Zhu, S.K. Aggarwal, Transient supercritical droplet evaporation
the best performance because the combustion chamber with emphasis on the effects of equation of state, International
conditions are very similar to the critical conditions of Journal of Heat and Mass Transfer 43 (2000) 1157–1171.
the fuel and because of the acentricity of the fuel is high. [11] J.Y. Koo, J.B. Ko, Numerical study of droplet evaporation and
combustion at high pressure and high temperature, International
Such improvements are even more significant at high load, Journal of Automotive Technology 6 (6) (2005) 563–570.
as the fuel concentration is higher. Taking the compress- [12] H. Shin, K.W. Lee, W.S. Yoon, J. Chae, High pressure vaporization
ibility factor obtained from the generalized correlations of burning droplet with flash vapor–liquid equilibrium calculation,
proposed by Lee and Kesler as reference, the Soave equa- International Communications in Heat Mass Transfer 30 (4) (2003)
tion was proved to provide the best predictions. When 465–474.
[13] B.I. Lee, M.G. Kesler, A generalized thermodynamic correlation
the obtained mean temperature and heat release were com- based on three-parameter corresponding states, AIChE Journal 21 (3)
pared with those obtained with ideal gas state equation, (1975) 510–524.
small relative differences were obtained. However, such dif- [14] K. Wark, Advanced Thermodynamics for Engineers, McGraw-Hill,
ferences may become important in case that the diagnostic 1995.