You are on page 1of 8

Real-Time Combustion Phase

Detection Using Centrai


Normaiized Difference Pressure
in CRDI Diesei Engines
Jongsuk Lim To develop eco-friendly diesel engines, accurate combustion phase control is important
e-mail: konamixi@hanyang.ac.kr due to its significant effects on harmful emissions and fuel efficiency. In order to accu-
rately control the combustion phase, the detection of the combustion phase should pre-
Seungsuk Oh cede control system design. Currently, combustion phase detection is done by the
e-mail: seungsukoh@gmail.com location of 50% mass fraction burned (MFB50), because of its close correlation with
emissions and fuel efficiency. However, this method is not easily implemented in real-
Jeasung Chung time applications because the calculation of MFB50 requires a large amount of
e-mail: jeasung6@hanyang.ac.kr in-cylinder pressure data and an excessive computational load. For this reason, a com-
bustion phase indicator with a simple algorithm is required for real-time combustion con-
Myoungho Sunwoo^ trol. In this study, we propose a new combustion phase indicator, called the "Central
e-maii: msunwoo@hanyang.ac,kr normalized difference pressures (CNDP)." The CNDP indicates the center of the two
crank angles where the normalized difference pressure between firing pressure and
Department ot Automotive Engineering, motoring pressure (NDP) reaches 90% of the maximum value before peak (NDPi,p90),
Hanyang University, and 70% of the maximum value after peak (NDPap70). The NDPhp90 and NDPap70 are
222 Wangsimni-ro, Seongdong-gu, highly correlated with MFB50 and the correlation is enhanced as the center between the
Seoui 133-791, Korea two points obtained. The CNDP is represented by a fixed quadratic polynomial with
MFB50 that robust to changes in various engine operating conditions such as engine
speed, main injection timing, injected fuel quantity, fuel-rail pressure, exhaust gas recir-
culation (EGR) rate and boost pressure. Furthermore, in performance evaluation, the
CNDP requires remarkably fewer in-cylinder pressure data samples, calculation steps
and less computation time compared to MFB50. These results show great potential for
the CNDP to be a substitute for the MFB50 since the proposed combustion phase detec-
tion algorithm can be used effectively for real-time combustion phase detection and con-
trol. [DOI: 10.1115/1.4006582]

Keywords: in-cylinder pressure, combustion phase detection, combustion phase control,


difference pressure, diesel engine

1 Introduction close relationship with emissions and fuel efficiency [9,10]. How-
ever, this heat release analysis requires not only a large amount of
In a self-ignited internal combustion engine, the change of
cylinder pressure data from the SOC to the end of combustion
operating conditions (e.g. engine speed, main injection timing,
(EOC), but also complex calculations. Consequently, implement-
fuel-rail pressure, boost pressure, exhaust gas recirculation (EGR)
ing the heat release analysis algorithm is difficult in a conven-
rate and fuel quality) causes a combustion phase variation. This
tional engine management system in real-time [11-13]. Thus, a
variation directly relates not only to a fluctuation of engine power
simple and accurate combustion phase detection algorithm is
and the degradation of fuel efficiency, but also the generation of
required to be embedded in a real-time engine controller.
harmful emissions. Therefore, control of the combustion phase is
required to maintain the optimal combustion state and dispersion In this study, we propose a new alternative combustion phase
reduction of particulate and nitrogen oxide emissions in a diesel indicator to MFB50, the central normalized difference pressures
engine [ 1 ^ ] . To achieve this control, the primary challenge is to (CNDP). The CNDP is defined as the center of two crank angles,
detect the combustion phase accurately. One method to do this where the normalized difference pressure between firing pressure
control is by using in-cylinder pressure, which has been widely and motoring pressure (NDP) reaches 90% of the maximum
used, since pressure directly reflects the combustion state in the value before its peak (NDPbp90), and 70% of the maximum
cylinder [1-3,5,6]. value after the peak (NDPap70). The two locations, NDPbp90 and
NDPap70, are highly correlated with MFB50, and the correlation
A well-known method for combustion phase detection using is enhanced as the center between the two points is obtained. As
in-cylinder pressure is heat release. This method can extract the a result of correlation analysis, the CNDP is represented by a
start of combustion (SOC) as well as a specific combustion phase, fixed quadratic polynomial function of MFB50, irrespective of
such as 10%, 50% and 90% of fuel mass fraction burned (MFB) changes in engine operating conditions such as engine speed,
[7,8]. The location of 50% mass fraction burned (MFB50) is main injection timing, injected quantity, fuel-rail pressure, EGR
known as an effective combustion phase indicator because of its rate and boost pressure. Furthermore, the CNDP is capable of
real-time combustion phase detection since the proposed indica-
'Corresponding author. tor requires only 23.9% of the computation time in comparison
Contributed by IC Engine Division of ASME for publication in the JOURNAL OF with MFB50. In addition, the CNDP is more robust than MFB50
ENGINEERING FOR GAS TURBINES AND POWER. Manuscript received December 21, 2011; with regards to pressure measurement problems such as a pres-
final manuscript received April 4, 2012; published online June 11, 2012. Assoc. Edi-
tor: Song-Chamg Kong. sure referencing error.

Journal of Engineering for Gas Turbines and Power AUGUST2012, Vol. 134 / 082801-1
Copyright © 2012 by ASME
This paper consists of four sections. In Sec. 2, previous research This approximation method shows a close match to the heat release
for combustion phase detection is reviewed. In Sec. 3, in order to trace in a spark ignition engine. Nevertheless, in CRDI diesel engines,
analyze the relationship between the NDP and MFB50, an the PRF can give different results from the actual heat release trace
experimental investigation is carried out with a CRDI diesel because of having a higher compression ratio, self-ignited combustion
engine. After analyzing the experimental results, we propose a and a variation of fuel injection scheduling strategies [11].
new combustion phase indicator. Section 4 includes the reliability
and performance of the CNDP compared to other indicators such
2.4 Difference Pressure Analysis. The difference pressure
as MFB50 and DPI50. Finally, conclusions about the newly pro-
Cdiff) also has been used for combustion phase detection. As
posed algorithm are discussed in Sec. 5.
described in Fig. 1 and Eq. (7), P¿iff is defined as the difference
between the cylinder pressure (Pcy\) and the motoring pressure (fmot)-
2 Related Works
In order to control the combustion phase accurately, many (7)
researchers have studied methods of combustion phase detection
using heat release analysis [14], the approximate heat release Based on the P¿iff, Herden [23] and Leonhardt [13] suggested
model [15,16], the pressure ratio management method [17-20], combustion phase detection methods which are the location of
and difference pressure analysis. The details are as follows. 50% difference pressure integral (DPI50) and the center of gravity
(COG) of the P¿iff for combustion phase detection, respectively.
However, these methods demand a wide range of pressure data.
2.1 Heat Release Analysis. Based on the first law of thermo-
Another approach to using the difference pressure is the loca-
dynamics, the heat release equation is represented as follows [14]:
tion of difference pressure 10 bar (DPIO) introduced by Lee et al.
[2,24], and the normalized difference pressure proposed by Choe
(1) et al. [25]. Although these methods demonstrate great detection
accuracy requiring only a simple algorithm, the indicator is lim-
ited to detecting the SOC.
where y, Ö, P, and V denote the specific heat ratio, crank angle,
cylinder pressure and volume, respectively. By using Eq. (1), an
expression for the percent of fuel mass fraction burned (MFB) is 3 New Combustion Phase Detection Method
derived as Eq. (2): In this section, we explain a new combustion phase detection
method and the central normalized difference pressures (CNDP).
- 6(9soc) The CNDP is able to play the role of MFB50 with less cylinder
MFB = X 100% (2) pressure data and computational load. The experimental setup and
the details will be described in the following sections.
where ösoc and 0EOC are the crank angle position at SOC and EOC,
respectively. The MFB is able to indicate how much fuel was 3.1 Experimental Setup. In order to extract a new combus-
burned; therefore, MFB calculated by using heat release analysis has tion phase indicator, we analyzed the relationship between the
been widely used as a combustion phase indicator. However, heat normalized difference pressure and the heat release. For the sake
release analysis has the disadvantages of requiring a large amount of of relationship analysis, engine experiments were carried out
cylinder presstire and a high computational load [ 11-13]. under steady-state operating conditions. The test engine was a
2.2 liter in-line four-cylinder CRDI diesel engine equipped with a
2.2 Approximate Heat Release Model. In order to reduce piezo-resistive type pressure sensor in each cylinder, made by
the computational load in the heat release calculation, Rassweiler Beru. The detailed engine specifications are listed in Table 1. In
et al. [15] and McCuiston et al. [16] presented simplified heat order to calculate and analyze the indicators, rapid control proto-
typing (RCP) equipment, the dSPACE MicroAutoBox, was used.
release models, Eqs. (3) and (4), respectively.

Piöf"v(d)-i 3.2 Correlation Between Heat Release and Normalized


MFB(0) (3) Difference Pressure. In this section, we describe the correlation
between NDP and MFB. As illustrated in Fig. 2, NDP can be
p{e)v{e)" -
MFB(e) (4)
)" - /'(0soc)V(0soc)"
g pressure

f\
3ring press
where n is the polytropic coefficient. These approximate models rence près
are simple to apply and give good agreement with heat release
analysis. However, they have drawbacks in that the polytropic
coefficient n is fixed and it is difficult to determine the EOC
[21,22]. \ —

\\
40

2.3 Pressure Ratio Management Method. The ratio of


firing pressure (Fcyi) and motoring pressure (P^od is the pressure
ratio (PR) and defined as Eq. (5) by Matekunas et al. [17-20]. 20 /
PR(0) = (5) 10
/
/'mot(ö)
0 r II — -—^ J ^1 .¿.-S:aie—

The MFB is approximated to the pressure ratio fraction (PRF)


-10
using the PR as follows [17-20]: -150 -100 -50 0 50 100 150

Crank Angle [deg CA ATDC]


(6)
Fig. 1 Concept of difference pressure

082801-2 / Vol. 134, AUGUST 2012 Transactions of the ASME


Table 1 Specification of the test engine

Descriptioti Specificatioti In
Engine type L-type, DOHC (9)
Number of cylinders 4
Bore/Stroke (mm) 85.4/96 In Eq. (9), a constant C^ot is precalculated using the cylinder volume
Displacement volume (cc) 2,196 at reference points, ö^f^i and öref,2 in the compression process. From
Compression ratio 16.0:1 Eqs. (8) and (9), we can readily obtain F^ot in real-time. Thereafter,
the locations of NDPbp, NDPap and MFB were obtained as their
index changed from 10% to 100%, at 10% intervals. From the loca-
tions, we obtained a quadratic polynomial curve fit between each
120 location of NDP and MFB. Figures 3 and 4 show all of the RMSE
NE p and R'^ values in a curve fit. As a result, NDPbp was found to be asso-
100 ciated with the early to middle phase of combustion, MFB20 to
MraSO, while the NDPap was found to have a close correlation with
80 the middle phase of combustion, MFB50 to MFB70. Furthermore,
N we concentrated on analysis of the relationship of NDPbp and NDP^p
g 60 with MFB50, as hsted in Table 3. As a result, the R^s of NDPbp90
Q.
and NDPap70 indicate a correlation of 0.9746 and 0.9554, respec-
§ 40 \ tively. It reveals very close relationship with MFB50.

20 - *"-
3.3 Definition of Central Normaiized Difference Pressure.
0 As mentioned in Sec. 3.2, NDPbp90 and NDPap70 show a close cor-
relation with MFB50 so that the MFB50 can be well estimated from
-40
-20 -20 0 20 40 60 80 NDPbp90 or NDPap70. However, the estimation error may be larger
Crank Angle [deg CA ATDC] depending on the indicator, NDPbp90 or NDPap70, according to
engine operating conditions. As depicted in Figs. 3 and 4, NDPbp90
Fig. 2 The definition of NDPbp and is highly related to early to middle combustion phases (MFB20 to
MFB50), but shows a less correlation with middle to later combus-
tion phases (MFB50 to MFB70). In contrast, NDPap70 indicates a
divided into two regions, before and after peak locations of NDP close relationship with middle to later combustion phase, but has a
it). In this study, we define the regions before and after weaker correlation with early to middle phase. For example, if the
p as NDPbp and NDP¡,p, respectively. In addition, NDPbpN rate of heat release near MFB50 shows a similar trend with the rate
denotes the location of N% of NDPbp, and NDP^pAi is the location of heat release at the early to middle combustion phase, the estimated
N% of NDPap. In order to analyze the correlation between the MFB50 derived from NDPbp90 shows a smaller error than that calcu-
locations of NDP and MFB, the root mean square error (RMSE) lated from NDPap70. On the other hand, when the rate of heat release
and the coefficient of detennination (R^) were obtained from the near MFB50 has a similar trend with the middle to late combustion
experimental indicator values and their curve fit. The value of phase, the estimation error between the calculated MFB50 from heat
RMSE was close to zero and the value of R^ was close to one, release and the estimated MFB50 from NDPap70 was less than that
which represents a close correlation between the locations of NDP from NDPbp90. For this reason, the NDPbp90 and NDPap70 have lim-
and MFB. As shown in Table 2, the correlation analysis was per- itations in detecting an acctirate combustion phase when they are
formed at 177 different steady state engine operating conditions used as a single indicator depending on the combustion characteris-
such as engine speed, main injection timing, fuel-rail pressure, tic. In order to improve the correlation and enhance estimation per-
EGR rate, and boost pressure changes. formance as a single indicator, we propose a central NDP (CNDP),
In order to analyze the correlations between NDP and MFB, we the center of the crank angles for NDPbp90 and NDPap70. In compar-
calculated both NDP and MFB from the measured pressure data ison with each single location, the CNDP etihances the correlation
in the 177 different conditions at each cylinder. In this step, the with MFB50, the RMSE of 0.3217 and the R^ of 0.9924 values.
motoring pressure, P^ot, for the NDP calculation was obtained by Figure 5 shows the relationship between the CNDP and MFB50. As
using a simple equation as follows: shown in Fig. 5, the CNDP can be represented with a simple quad-
ratic polynomial function of MFB50 regardless of the 177 operating
condition changes in all the cylinders.
/'mot(Ö) = (8)

This equation assumes that the compression process is a poly- 4 Performance Evaluation of CNDP
tropic process. Based on this assumption, the polytropic coeffi- In order to guarantee the performance of a closed-loop combus-
cient is computed at each cycle using Eq. (9). tion phase control, the combustion indicator should not only

Table 2 Engine operating conditions

Operating conditions Specifications

Engine speed (rpm) 1000 1500 2000 2500


Ace. pedal position (%) 15,20 20,25 30,35 35,40
BMEP (bar) 3.4, 5.4 3.8,5.8 5.6,7.2 6.0,7.8
Main injection timing (deg CA) -10BTDC-6ATDC -10BTDC-6ATDC -10BTDC-6ATDC -10BTDC-6ATDC
Fuel-rail pressure (bar) 400-1000 400-1000 600-1200 800-1400
Desired air mass (mg/str) 280-600 280-600 300-700 -
Boost pressure (bar) 1-1.2 1-1.4 1.2-1.8 1.4-2.4

Journal of Engineering for Gas Turbines and Power AUGUST2012, Vol. 134 / 082801-3
RMSE [NDP^ vs. MFB]

10 20 30 40 50 60 70 80 90 100
MFB [%] NDP,. [%]

100
90
80
70 'TÊÊÊÊ
60
WEM
40
30
20
10
10 20 30 40 50 60 70 80 90 100
MFB [%] 0 0

Fig. 3 RMSE and R^ between MFB and

/s. MFB]

100
90
20
80
15 70
UJ
g 60
m 10 I 50|
40!
5
oJ 30
100 80 100 20
60 80
40 40 10
20 20
MFBP/o] " 0 10 20 30 40 50 60 70 80 90 100

R^ [NDP vs. MFB]

1- - r

20 - 20 10 20 30 40 50 60 70 80 90 100
MFB[%] 0 0

Fig. 4 RMSE and R^ between MFB and NDPa

detect the accurate combustion phase but also be sufficiently ro- 4.1 The Effect of Pressure Referencing Error on
bust to pressure referencing error. In this section, we analyzed the Indicators. Piezo-electric type pressure sensors are widely used
robustness to pressure referencing error of CNDP compared to to obtain in-cylinder pressure because of their advantages such as
that of MFB50, and DPI50. In addition, for the sake of feasibility high measurement precision, fast response, light weight and
of realization in real-time, we evaluated the computational load of robustness under strenuous environmental conditions. However,
CNDP. the signal of a piezo-electric type pressure sensor can carry a bias

082801-4 / Vol. 134, AUGUST 2012 Transactions of the ASME


14
:3uadratic polynomial fitting curve L Z\ CNDP
O
A
:ylinder#1
3yiinder#2
12 • • • MFB50
I^H DPI50
{ •
X Dyiinder#3
+ :yiinder #4
1
"w i •

1 1 1
W
g" 1
1 1
Á
^ •" 6
W

LL•
20 -

ndicatc
15

10 CNDP = 0.01AÍFB50\ +•0.647 /F550 +12.22 0

j>
5 10 15 20 25 30 1 2 3 4 !
MFB50 [deg CA ATDC] Pressure referencing error [bar]

Fig. 5 Relationship between ÍVIFB50 and CNDP Fig. 7 The effect of pressure referencing error

Cylinder pressure with peggine em>r


70
— Pegging error +5 bar 1200

60 ft
— Pe gging error; +3 bar
gging error: +2 bar
50 —• Pe gging error: +1 bar .
— Wi hout pegging error
1 ;
5.40

r
30 i

20

10

Î" '"
-200 -180 -160 -140 -120 -100 - 8 0 - 6 0 - 4 0 - 2 0 20
Crank angle [deg CA ATDC]

Fig. 6 Cyiinder pressure offset for evaluation of robustness to Fig. 8 Tiie comparison of totai number of pressure sampies
pegging error (four-cylinder engine, 0.5 deg resolution)

error at abrupt temperature changes in the sensor housing and the


sensing element. Since combustion phase indicators are merely 3000
based on the phase of the pressure signal, the bias error leads to an
erroneous detection of the combustion phase. Thus, the bias error
2500
should be compensated by a known reference pressure called pres-
sure referencing [3,7,8,26,27]. However, the pressure referencing
process may also involve errors depending on pressure referenc-
ing method. For this reason, in order to guarantee accurate com-
bustion phase detection, the combustion phase indicator requires
robustness to the pressure referencing error.
For the sake of evaluation of the indicator robustness to the
pressure referencing error, we assume that the pressure referenced
one bar at 180 deg CA BTDC is an actual pressure and we added
a pressure offset from one to five bar to the actual pressure that
we assumed, as depicted in Fig. 6. From the pressures with offset,
we calculated MFB50, DPI50 and CNDP. As a result, the indica-
tor errors of MFB50 increase sharply as the pressure referencing
error increases, and the error of DPI50 shows a relatively smaller
CNCIP MFB50
error than the error of MFB50 as shown in Fig. 7. The error of
Indicators
CNDP shows the smallest error; consequently, CNDP has an
advantage of robustness to the pressure referencing error com- Fig. 9 The comparison of total number of additions (four-
pared to MFB50, and DPI50. cyiinder engine, 0.5 deg resolution)

Journal of Engineering tor Gas Turbines and Power AUGUST2012,Vol. 134 / 082801-5
2500 tions (shown in Table 2), MFB demands cylinder pressure data
from —30 to 90 deg ATDC for heat release calculations. In con-
trast, since CNDP is normalized from NDPpeak, and NDPpeak,
2000 NDPbp90 and NDPap70 exist in the range of from TDC to 50 deg
ATDC, CNDP can be calculated with cylinder pressure data from
10 deg BTDC to 60 deg ATDC, considering advance or retard in
.B 1500^
transient operation. This is only 41.7% of the data acquisition of
MFB50 or DPI50 as shown in Fig. 8. In addition, the CNDP can
be calculated with remarkably fewer calculations steps, 80.5%
reduction of additions and 73.0% reduction of multiplications of
1000
E MFB50, as depicted in Figs. 9 and 10.
With the aim of furthering the realistic evaluation of the compu-
tational load in a real-time application, the computation times
500 for each indicator were measured with a Freescale MPC5554
microcontroller designed for engine management systems. The
MPC5554 provides improved signal processing capabilities using
a Signal Processing Engine-Auxiliary Processing Unit (SPE-
CNDP MFB50 DP 150
Indicators
APU). A key feature of the SPE-APU is Single Instruction Multiple
Data (SIMD) functionality, which executes two independent oper-
Fig. 10 The comparison of total number of multiplication ations simultaneously in a 64-bit register for fast floating point
(four-cylinder engine, 0.5 deg resolution) operation. In Fig. 11, the evaluation results are depicted at each
system clock of the MPC5554. At the highest system clock,
MFB50 demands 15.42 ms of computation time when the cylinder
pressure data is sampled in 0.5 deg resolution for a four-cylinder
engine during one cycle. In comparison, the calculation of the
CNDP takes only 3.68 ms, which is about 76.1% less computation
time than that of MFB50. This remarkably reduced amount of data
acquisition, calculations and computation time shows the capabil-
ity of the proposed indicator to be embedded in a real-time engine
controller.

5 Conclusions
In order to archive a closed loop combustion phase control, the
combustion phase should be accurately detected in real-time. How-
ever, conventional detection methods are difficult to implement in a
real-time engine controller due to their requirements of excessive
pressure data acquisition and long computation times. In this study,
we suggest a new combustion phase detection indicator, CNDP,
which has great potential to represent the middle phase of combus-
32 64
MCU Clock rate [MHz] tion with adequate accuracy. In addition, the CNDP is simply calcu-
lated with less pressure data acquisition than that of MFB50. Its
Fig. 11 The comparison of computation time (four-cylinder close correlation with MFB50 is validated by correlation analysis
engine, 0.5 deg resolution) over various engine operating conditions. Moreover, the performance
of the CNDP as a combustion phase indicator is compared and eval-
uated with MFB50. The results are summarized as follows:
4.2 Analysis of Computational Load. In order to validate
the realization of the proposed combustion phase detection algo- (1) As a result of correlation analysis, the CNDP showed a
rithm in real-time, we compared the computational load of three close correlation with MFB50, which was verified to be the
indicators, MFB50, DPI50 and CNDP. As previously mentioned, lowest RMSE (0.3217) and the highest R^ (0.9924) values
MFB50 and DPI50 demand a large amount of cylinder pressure of their quadratic polynomial curve fit, regardless of
data from SOC to EOC because heat release and difference pres- whether the various engine operating conditions were
sure integration are normalized by the values of heat release and changed. Based on the results of the correlation analysis,
difference pressure integration at EOC, respectively. Considering we determined the CNDP to be a viable combustion phase
the most advanced SOC and the most retarded EOC in all condi- indicator.

Table 3 RMSE and R^ between the locations of NDP and MFB50

The index of NDPbp

Description 10 20 30 40 50 60 70 80 90 100

MFB50 RMSE 3.2705 2.8568 2.3397 1.7955 1.3481 1.1110 0.9021 0.7363 0.5770 0.7573
0.0283 0.0922 0.4000 0.7020 0.8512 0.9025 0.9378 0.9590 0.9746 0.9575

The index of NDPap

Description 10 20 30 40 50 60 70 80 90

MFB50 RMSE 17.2983 5.1873 2.3657 1.4931 1.0775 0.8742 0.8117 0.7905 0.7873
0.3121 0.6785 0.8550 0.9155 0.9431 0.9550 0.9554 0.9536 0.9513

082801-6 / Vol. 134, AUGUST 2012 Transactions of the ASME


(2) In the performance evaluation, the CNDP demonstrates 6= crank angle, deg
robustness to the pressure referencing error in comparison ÖEOC = crank angle at end of combustion, deg
with that of MFB50. A further remarkable performance ösoc = crank angle at start of combustion, deg
improvement of the CNDP is the benefit of 58.3% less data öref = reference crank angle
acquisition and 71.6% fewer calculations required.
(3) The proposed algorithm .showed a short computation time, References
only 23.8% of MFB50 and about 64.7% of DPI50. This [I] Park, S., Yoon, P., and Sunwoo, M., 2001, "Cylinder Pressure-Based Spark
faster computation not only enables real-time combustion Advance Control for SI Engines." JSME Int. J. Ser. B, 44, pp. 305-312.
control and monitoring but also permits the algorithm to be [2] Yoon, M., Lee, K., and Sunwoo, M., 2007, "A Method for Combustion Phasing
Control Using Cylinder Pressure Measurement in a CRDI Diesel Engine,"
embedded into a commercial mass produced engine con- Mechatronics, 17, pp. 469^79.
troller. Consequently, the proposed combustion phase indi- [3] Lee, K., Yoon, M., and Sunwoo, M., 2008, "A Study on Pegging Methods for
cator, the CNDP, is expected to be used for real-time Noisy Cylinder Pressure Signal," Control Eng. Practice, 16, pp. 922-929.
engine combustion phase detection and control in place of [4] Hadler, J., Rudolph, F., Dorenkamp, R., Stehr, H., Dusterdiek, T., Gilzendeger,
J., Mannigel, D., Kranzusch, S., Veldten, B., Rosters, M., and Specht, A., 2008,
conventional indicators. Even though, CNDP is highly "Volkswagen's New 2.0L TDI Engine Fulfils the Most Stringent Emission
related to MFB50, it is limited in that CNDP has only been Standards," 29th Internationales Wiener Motoren Symposium, Vienna, Austria,
proved for a tested engine. Therefore, it is required that GWV Publishers. Wiesbaden, Germany.
these results are validated with different engines, such as [5] Schiefer, D., Maennel, R., and Nardoni, W., 2003, "Advantages of Diesel
engines with a different shape of combustion chamber, a Engine Control Using In-Cylinder Pressure Information for Closed Loop Con-
trol," SAE Paper No. 2003-01-0364.
different swirl intensity, and different wall heat transfer 16] Powell, J. D., 1993, "Engine Control Using Cylinder Pressure: Past, Present,
coefficient. and Future," J. Dyn. Syst. Measure. Control, 115, p. 343.
[7] Brunt, M. F. J., Rai, H., and Emtage, A. L., 1998, "The Calculation of Heat
Release Energy from Engine Cylinder Pressure Data," SAE Paper No. 981052.
[8] Brunt, M. F. J., and Platts, K. C, 1999, "Calculation of Heat Release in Direct
Acknowledgment Injection Diesel Engines," SAE Paper No. 1999-01-0187.
[9] Northrop, W. F.. Bohac. S. V., and Assanis, D. N., 2009, "Premixed Low Tem-
This work was financially supported by the Ministry of Educa- perature Combustion of Biodiesel and Blends in a High Speed Compression
tion, Science and Technology through the BrainKorea 21 Program Ignition Engine," SAE Paper No. 2009-01-0133.
(201000000000173), the Ministry of Knowledge Economy [10] Corti, E., and Forte, C , 2010, "A Statistical Approach to Spark Advance
(MKE) and Korea Institute for Advancement in Technology Mapping," ASME }. Eng. Gas Turbines Power, 132, p. 082803.
[11] Asad, U., and Zheng, M., 2008, "Fast Heat Release Characterization of a Diesel
through the Workforce Development Program in Strategic Tech- Engine," Int. J. Therm. Sei., 47, pp. 1688-1700.
nology, Energy Resource R&D program (2006ETR11P091C) [12] Mittal, M., Zhu, G., and Schock, H., 2009, "Fast Mass-Fraction-Bumed Calcu-
under the MKE, the National Research Foundation of Korea lation Using the Net Pressure Method for Real-Time Applications," Proc. Inst.
(NRF) grant funded by the Korea government (MEST) (No. 2011- Mech. Eng. Part D, 223, pp. 389-394.
0017495), and the Industrial Strategy Technology Development [13] Leonhardt, S., Müller, N., and Isermann, R., 1999, "Methods for Engine Super-
vision and Control Based on Cylinder Presstire Information," IEEE/ASME
Program of MKE (No. 10039673). Trans. Mechatron., 4, pp. 235-245.
[14] Heywood, J. B., 1988, Internal Combustion Engine Fundamentals, McGraw-
Hill, New York.
Nomenclature [15] Rassweiler, G. M., and Withrow, L., 1938, "Motion Pictures of Engine Flames
ATDC = after top dead center Correlated With Pressure Cards," SAE Trans., 42, p. 185.
[16] McCuiston, F., Jr., Lavoie, G., and Kauffman, C , 1977, "Validation of a Turbu-
CNDP = central normalized difference pressure lent Flame Propagation Model for a Spark Ignition Engine," SAE Paper No.
COG = center of gravity 770045.
DPI50 = the location of 50% difference pressure integral [17] Sellnau, M., Matekunas, F., Battiston, P., Chang, C , and Lancaster, D. R.,
DPIO = the location of difference pressure 10 bar 2000, "Cylinder-Pressure-Based Engine Control Using Pressure-Ratio-Manage-
ment and Low-Cost Non-Intrusive Cylinder Pressure," SAE Paper No. 2000-
EOC = end of combustion 01-0932.
EGR = exhaust gas recirculation [18] Matekunas, F., 1986, "Engine Combustion Control With Dilution Flow by Pres-
MFB = mass fraction burned sure Ratio Management," US Patent No. 4,624,229.
MFB N = the location ofN% mass fraction burned [19] Matekunas, F., 1986, "Engine Combustion Control With Fuel Balancing by
Pres.sure Ratio Management," US Patent 4,621,603.
n = polytropic coefficient
[20] Matekunas, F., 1986, "Engine Combustion Control With Ignition Timing by
NDP = normalized difference pressure Pressure Ratio Management," US Patent No. 4,622,939.
eak = peak value of NDP [21] Ball, J., Raine, R., and Stone, C , 1998, "Combustion Analysis and Cycle-by-
N = the location oîN % NDP after peak Cycle Variations in Spark Ignition Engine Combustion Part I : An Evaluation of
N = the location of iV % NDP before peak Combustion Analysis Routines by Reference to Model Data," Proc. Inst. Mech.
Eng. Part D, 212, pp. 381-399.
P = pressure, bar [22] Cheung, H. M., and Heywood, J. B., 1993, "Evaluation of a One-Zone Bum-
Fcyi = in-cylinder pressure or firing pressure Rate Analysis Procedure Using Production SI Engine Pressure Data," SAE
/"(Jiff = difference pressure Paper No. 932749.
[23] Herden, W.. and Matthias, K., 1994, "A New Combustion Pressure Sensor for
^mot = motoring pressure Advanced Engine Management," SAE Paper No. 940379.
Q = heat release, J [24] Lee, K., Yoon, M., Son, M., and Sunwoo, M., 2006, "Closed-Loop Control of
RMSE = root mean square error Start of Combustion Using Difference Pressure Management," Proc. Inst.
R^ = coefficient of determination Mech. Eng. Part D, 220, pp. 1615-1628.
[25] Choe, D., Lee, M., Lee, K., Sunwoo, M., Kook Kong, J., Chang, K., and Shin,
SIMD = single instruction multiple data K., 2007, "SOC Detection of Controlled Auto-Ignition Engine," SAE Paper No.
SOC = start of combustion 2007-01-3538.
SPE-APU = single processing engine auxiliary processing unit [26] Lee, K., Kwon, M., Sunwoo, M., and Yoon, M., 2007, "An In-Cylinder Pressure
TDC = top dead center Referencing Method Ba.sed on a Variable Polytropic Coefficient," SAE Paper
V = volume, mm^ No. 2007-01-3535.
[27] Pipitone, E., 2008, "A Comparison Between Combustion Phase Indicators for
y = specific heat ratio Optimal Spark Timing," ASME J. Eng. Gas Turbines Power, 130, p. 052808.

Journal of Engineering for Gas Turbines and Power AUGUST2012, Vol. 134 / 082801-7
Copyright of Journal of Engineering for Gas Turbines & Power is the property of American Society of
Mechanical Engineers and its content may not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder's express written permission. However, users may print, download, or email
articles for individual use.

You might also like