You are on page 1of 7

Applied Thermal Engineering 29 (2009) 2379–2385

Contents lists available at ScienceDirect

Applied Thermal Engineering


journal homepage: www.elsevier.com/locate/apthermeng

Thermodynamic modeling of direct injection methanol fueled engines


Yuan Shen b, Joshua Bedford a, Indrek S. Wichman a,*
a
Department of Mechanical Engineering, 2555 Engineering Building, Michigan State University, East Lansing, MI 48824-1226, USA
b
Anstalt für Verbrennungskraftmaschinen (AVL), 47519 Halyard Dr., Plymouth, MI 48170, USA

a r t i c l e i n f o a b s t r a c t

Article history: In-cylinder pressure is an important parameter that is used to investigate the combustion process in
Received 5 July 2007 internal combustion (IC) engines. In this paper, a thermodynamic model of IC engine combustion is pre-
Accepted 5 December 2008 sented and examined. A heat release function and an empirical conversion efficiency factor are intro-
Available online 16 December 2008
duced to solve the model. The pressure traces obtained by solving the thermodynamic model are
compared with measured pressure data for a fully instrumented laboratory IC spark ignition (SI) engine.
Keywords: Derived scaling parameters for time to peak pressure, peak pressure, and maximum rate of pressure rise
IC engines
(among others) are developed and compared with the numerical simulations. The models examined here
Simplified thermodynamic model
Analysis
may serve as pedagogic tools and, when suitably refined, as preliminary design tools.
Experiment Ó 2008 Elsevier Ltd. All rights reserved.
Pressure trace
HCCI engines

1. Introduction der pressure trace in determining the rates and magnitudes of


other affiliated processes.
In an IC engine the cylinder pressure is an important parameter In this article, we present an analysis of the in-cylinder pressure
for analyzing the progress and intensity of the combustion process. variations and examine scaling relationships that correlate well
Even without combustion (as in idealized engine cycles such as the with the experimental data for an IC test engine. A heat-release
Otto and combined cycles [1,2]) the pressure trace is required in model is examined, producing mathematical parameters for scal-
order to calculate the work output. In fact, given the engine turn- ing pressure traces between (and during) important IC engine com-
over rate (RPM) the pressure trace forms the basis of the engine cy- bustion signposts, such as the instants of ignition, Top Dead Center
cle power computation. Examples of these calculations are (TDC), location of peak pressure and burnout. This article is a
provided in thermodynamics textbooks [1,2] and in classical ‘‘en- lengthier version of Ref. [11].
gines” textbooks [3–5]. Although the models examined in these
books are simple, they yield order-of-magnitude estimates of sat- 2. Thermodynamic analysis
isfactory accuracy. Various sophisticated models with empirically
determined parameters and heat release (i.e., combustion rate) A thermodynamic analysis is applied to a control volume in the
profiles have been developed using these methods. engine cylinder. The analysis employs the following four assump-
There are many pressure-based combustion studies in the liter- tions: (1) The mixture inside the cylinder is an ideal gas, thus the
ature. Rassweiler and Withrow [6] investigated the pressure relationship among the pressure (p), volume (V), and temperature
change in an engine cylinder finding that mass burned and the (T) is given by the equation of state, ln p + ln V = ln m + ln R + ln T. (2)
pressure change were proportional. Daw and Kahl [7] used peak Conditions in the engine cylinder are homogeneous and character-
SI engine cylinder pressure as the primary parameter to study cy- ized by unique values of the pressure (p), temperature (T), volume
cle-to-cycle variations during engine operation. Brunt et al. [8] (V) and heat release (Q). (3) The intake and exhaust valves are both
analyzed knock in a gasoline engine. Shen et al. [9] analyzed closed during combustion so the mixture mass (m) is constant. (4)
combustion in a direct injection methanol engine using cylinder The ideal gas constant R is assumed constant throughout the en-
pressure data. Ludwig et al. [10] introduced a method to describe gine cycle. Then, the above equation of state may be written as
in-cylinder diesel combustion by monitoring the pressure trace.
These works all ascribed primary importance to the engine cylin-
dp=p þ dV=V ¼ dT=T: ð1Þ
According to the first law of thermodynamics, the differential form
of the energy conservation equation in the engine cylinder is
* Corresponding author. Tel.: +1 517 353 9180; fax: +1 517 353 1750. dU = dQ  dW. Using dU = mcvdT and dW = pdV gives mcvdT = dQ 
E-mail address: wichman@egr.msu.edu (I.S. Wichman). pdV. Dividing this equation by pV = mRT = m(cp  cv)T = mcv(c  1)T,

1359-4311/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2008.12.002
2380 Y. Shen et al. / Applied Thermal Engineering 29 (2009) 2379–2385

Nomenclature

a, b, c dimensionless parameters used in Section 4 D angular coordinate centered at TDC, D = h  2p (–)


cv, cp specific heats at constant volume, pressure (e/m  T) f mathematically defined dimensionless function in Sec-
f(h), g(h) coefficient functions in Eq. (2) tion 3.1, f = ln[(n + x)/n] (–)
K nondimensional heat transfer to charge, K = (c  1)Qin/ h angular coordinate (–)
p1V1 (–) n positive dimensionless parameter in Eq. (5) (–)
m mass of gas in piston cylinder (m) s normalized angular coordinate used in Section 3.1,
p pressure (F/L2) s = (h  hi)/(hf  hi) (–)

p dimensionless pressure, p ¼ p=p1 (–) / dummy variable of integration in Eq. (5) (–)
Q(h) heat release during combustion of fuel (e) v positive dimensionless parameter in Eq. (5) (–)
Qin total heat energy added during cycle (e)
r compression ratio, r = VBDC/VTDC (–) Subscripts
R gas constant (e/m  T) f burnout value (‘‘final”)
s dummy variable of integration (–) i ignition value
T temperature (T) m value at maximum pressure
U internal energy (e) 1 inlet (BDC) value (‘‘initial”)
V volume (L3)

V dimensionless volume, V  ¼ V=V 1 (–)
Acronyms
W work (e)
BDC Bottom Dead Center
x(h) heat release progress variable (–) CAD crank angle degrees (same as h)
z dummy variable of integration (–) IC internal combustion
RPM revolutions per minute
Greek symbols TDC Top Dead Center
a positive dimensionless parameter in Eq. (5) (–)
c specific heat ratio, c = cp/cv (–)

then writing the differentials with respect to the crank angle degree in Eq. (3) and then employs Euler’s solution for linear first-order or-
h (CAD) and using Eq. (1) for the term containing the factor T1 dT/ =dh þ f ðhÞp
dinary differential equations dp  ¼ gðhÞ in the form
dh yields [Ref. [5], p. 388] the differential equation for the engine  Z h Z s    Z h 
cylinder pressure trace p(h), ðhÞ ¼
p const: þ gðsÞ exp f ðzÞdz ds exp  f ðzÞdz ;

dp=dh þ ðc=VÞðdV=dhÞp ¼ ððc  1Þ=VÞdQ =dh: ð2Þ


where s and z are dummy variables of integration. A brief calculation
The engine geometry dictates the cylinder volume. Thus, V(h) and ðhi Þ ¼ p
including the application of the initial condition p i yields
dV(h)/dh are known, given functions of h. Eq. (2) then becomes dp/ Z h
dh +f(h)p = g(h), where f(h) = (c/V)(dV/dh) is a known coefficient func- pi K dx
ðhÞ ¼
p cþ Vð/Þc1 d/; ð4Þ
tion and g(h) = ((c  1)/V)dQ/dh is the forcing function, which is ðV=V i Þ Vc hi d/
known when the heat release trace Q(h) is known.
where / is the dummy variable of integration and K = (c  1)Qin/
The function Q(h), however, is known only once the in-cylinder
p1V1 is the dimensionless heat release. Various possible heating
combustion problem is solved, a difficult (and so far impossible)
functions are illustrated in Fig. 1, showing the rise from zero heat
task. Consequently, in our analysis we specify this function, either
release at ignition to full heat release at burnout, approximately
a priori or a posteriori, in order to proceed with our analysis.
h  385 CAD in Fig. 1.
We develop two models in this paper. In the first model, Model
(i), the heat release Q(h) is deduced by comparing the predictions
of Eq. (2) with the experimental p(h) profiles, producing an accu- 3. Empirical heat release function
rate and reasonably sophisticated Q(h) expression. Model (i) pro-
duces a version of Eq. (2) that must be solved numerically. Model 3.1. Theoretical model
(ii) produces a version of Eq. (2) that can be solved analytically.
This analytical solution is used to derive mathematical scaling In order to deduce an empirically accurate Q(h) profile, six phys-
parameters for use in engine analysis. Without such a simplifica- ically justified restrictions on the heat release function are pre-
tion, the mathematical analysis is too difficult and scaling param- sented below in the form of bullet points (1)–(3) [9]:
eters cannot be analytically derived. This is true for Model (i).
For both Models (i) and (ii), we write Q(h) = Qinx(h) where x(h) is a (1) At h = hi: x = 0 and (dx/dh)i > 0. These conditions denote igni-
heat release progress variable defined such that x(hi) = 0 at the start of tion or start of combustion.
combustion and x(hf) = 1 at its completion. Here, hi is the ‘ignition’ (2) When hi < h < hf: 0 < x < 1 and (dx/dh) > 0. These conditions
crank angle while hf is the burnout or ‘final’ crank angle. All of the heat are commensurate with flame propagation and the continu-
Qin has been released when x(hf) = 1. In terms of x(h) Eq. (2) becomes ation of combustion.
(3) At h = hf: x = 1 and (dx/dh)f = 0. These conditions identify the
dp=dh þ ðc=VÞðdV=dhÞp ¼ ððc  1Þ=VÞQ in dx=dh: ð3Þ
termination of combustion.
Subject to the assumptions (1)–(4) Eq. (3) is exact. Eq. (3) is nondi-
mensionalized using p ¼ p=p1 ; V ¼ V=V 1 , where p1, V1 are inlet or By normalizing h to values ranging between zero at ignition and
Bottom Dead Center (BDC) values. For future reference p2 and V2 unity at burnout according to the mathematical definition
are TDC values. The mathematical solution of the nondimensional s = (h  hi)/(hf  hi) it has been shown [9,12] that conditions
Eq. (3) uses the nondimensionalizations for p  and V given above (1)–(3) can be satisfied by the following equation:
Y. Shen et al. / Applied Thermal Engineering 29 (2009) 2379–2385 2381

Heat Release Pofiles


100

80

Mass fraction burned (%)


60

40

exponential function
20 linear function
sine function
cosine fuction
3-line function
0
355 360 365 370 375 380 385 390
CAD

Fig. 1. Heat release profiles showing the various possible engine heat release rates during the combustion segment of the engine cycle.

dx=ds ¼ aðn þ xÞð1  sÞn ; ð5Þ


Table 1
in which the parameters a, n, v are taken to be positive. Eq. (5) sat- Engine parameters for the 4-cylinder naturally aspirated methanol-fueled direct-
injection engine.
isfies conditions (1) and (3) completely: at h = hi where x = 0 and
s = 0 we have (dx/ds)i = an > 0, while at h = hf we have (dx/ds)f = 0. Bore 7.95 cm
This analysis intends to satisfy conditions (2) in-between, thus pro- Stroke 9.99 cm
Connecting rod length 19.8 cm
ducing a plausible solution for the burned fraction x(h). For conve- Compression ratio 19.3:1
nience, Eq. (5) is split into two parts by introducing the function f Intake valve opened 12° BTDC
in such a way that dx/df = n + x and df/ds = a(1  s)v with a power Intake valve closed 8° ABDC
law dependence on time. Note that the product ( dx/df)(df/ Exhaust valve opened 38° BBDC
Exhaust valve closed 20° ATDC
ds) = dx/ds is simply Eq. (5). The initial (s = 0) and final (s = 1) con-
ditions of the functional relationship for x(s) are derived by solving
the first of these two differential equations, viz, ln[(n + x)/n] = f
Table 2
giving
Operating conditions for the methanol-fueled direct-injection engine at 1500 RPM.

x ¼ nðexp f  1Þ: ð6Þ Case k IMEP (kPa) Injection Ignition


hBTDC hBTDC
The second differential equation yields f = a[1  (1  s)v + 1]/ 1 1.5 700 60 10
(v + 1). At x = 1 this relationship gives immediately ff = a/(v + 1). 2 1.5 525 50 20
Substituting into Eq. (6) gives n = [exp ff  1]1. With the initial 3 1.75 600 50 10
and final conditions at s = 0 and s = 1 thus specified, Eq. (6) pro- 4 1.75 400 70 12.5
duces the analytical solution
xðhÞ ¼ ðexp f  1Þ=ðexp ff  1Þ; ð7Þ
According to [9], in the dynamic stage of combustion in these
where f and ff are specified beneath Eq. (6). The parameters a and v engine experiments the initial or ignition state of combustion (‘i’
are calculated for Model (i) of Section 2 by the regression of the where s = 0) and the terminal or final state of combustion (‘f’
experimentally measured cylinder pressure data. where s = 1) can be determined by introducing the polytropic pres-
sure model and the state-vector-space. Based on the determination
3.2. Filtering of parameters by comparison with experiments of these two critical states, the heat release function between these
two states is generated from a regression calculation. The resulting
A direct injection methanol fueled engine was run at the oper- parameters for the heat release function for the four cases of Table
ating speed of 1500 RPM. The parameters for this engine are shown 2 are shown in Table 3. Cases 1 and 3 are fast burns. Cases 2 and 4
in Table 1. Note that ‘ATDC’ and ‘BBDC’ mean ‘after TDC’ and ‘before are slow burns.
BDC,’ respectively.
The engine experiments at the MSU Automotive Research
Experiment Station examined several hundred operating condi- Table 3
The parameters of heat release functions for the methanol-fueled direct-injection
tions. For the current work, four sets of operating conditions, se-
engine under the operating conditions of Table 2.
lected for pressure diagnostics, are listed in Table 2. The air/fuel
ratios are written in terms of the air excess coefficient, k, with Case 1 (fast) Case 2 (slow) Case 3 (fast) Case 4 (slow)
respect to the stoichiometric mass-based air to fuel ratio, which hi 355 347 355 355
equals 6.435 for methanol. The selection was made to cover the hf 383 380 382 403
full scope of measured IMEP, represented by the highest and a 26.3 31.8 25.4 11.8
v 3.896 6.034 3.379 2.671
lowest of two air excess coefficients, k = 1.5 and k = 1.75.
2382 Y. Shen et al. / Applied Thermal Engineering 29 (2009) 2379–2385

Heat transfer to the engine cylinder walls prevents all of the Table 4
heat generated by combustion from being converted to mechanical The effectiveness of fuel utilization for the 4-cylinder naturally-asperated methanol-
fueled direct-injection engine under the operating conditions of Table 2.
work. The effectiveness of fuel utilization for these cases was stud-
ied in [9]. The results are shown in Table 4. The effectiveness of fuel Case 1 Case 2 Case 3 Case 4
utilization is defined as the percentage of the available chemical g 58.4% 43.4% 55.4% 40.4%
energy that is converted to cylinder pressure.
The two fast burns 1 and 3 have higher effectiveness of fuel uti-
lization than the two slow burns 2 and 4. Physically, a shorter time
Table 5
is required for the fast burns to finish combustion. Less heat is lost The comparison of calculated pressure from the theoretical model with experimental
to the cylinder walls, which leads to a higher effectiveness of fuel data.
utilization.
Location CAD (exp) Pressure (calculated-atm) Pressure (exp-atm)
Besides the heat release function Q(h), the volume of the com-
bustion chamber V(h) is also a function of the CAD. Several models Initial 355 42.6 42.6
TDC 360 55.5 47.9
of the combustion chamber volume are available, e.g. [5]. For Mod- Maximum 371 68.2 78.4
el (i) (see Section 2) the volume of the combustion chamber at a Final 382 56.2 50
specific crank angle h is defined as
2
VðhÞ ¼ V C f1 þ 0:5ðrC  1Þ½R þ 1  cos h  ðR2  sin hÞ1=2 g: ð8Þ
ties that no numerical analysis, whether Model (i) or even a much
Here VC is the clearance volume, rC is the compression ratio, and R is more sophisticated full numerical 3D simulation – no matter how
the connecting rod length to crank radius ratio, see Table 1. sophisticated – can ever produce.
After writing expressions for Q(h) = Qinx(h) and V(h), Eq. (4) can For this analysis, we shall employ the following simplifications.
be solved to find p(h). (1) The simplest possible volume/crank angle relationship will be
used, namely
3.3. Discussion
VðhÞ ¼ V TDC ½1 þ ðr  1Þð1  cos hÞ=2;
It is shown in Fig. 2a–d that the simple thermodynamic analysis where r = VBDC/VTDC is the compression ratio (compare this V(h)
with empirically deduced Q(h) = Qinx(h) gives good qualitative function with the far more sophisticated Eq. (8)); (2) The simplest
agreement with the experimental data. The agreement is accept- heat release function will be used,
able both in general curve shape and the location of the pressure
maxima. This is not unexpected because the heat release parame- Q ¼ Q in xðhÞ; xðhÞ ¼ ðh  hi Þ=ðhf  hi Þ;
ters were chosen with the specific objective of providing the best so that dQ/dh = Qindx/dh =Qin(hf  hi)1. We note that in the terminol-
possible agreement with the measured values. ogy of Section 3.1 we have, in this simple case, x(h) = s(h). In Eq. (4)
Fig. 2a–d shows curves representing cylinder pressures without we will begin the calculation of p at ignition (h = hi) because prior to
wall heat losses. The curves without heat losses are labeled ‘‘ideal.” ignition we have K = 0 and p ¼V  c as the polytropic compression
These curves were calculated numerically from Eq. (4) using Eqs. relation for the fuel/air mixture (the nondimensionalizations are
(7) and (8). Note that there are no compression or expansion the same as in Section 2). The use of these expressions in Eq. (4)
strokes in Fig. 2a–d. Only pressure traces during the combustion yields
period are shown. After applying the factor for the effectiveness  c " #
Z hf
of fuel utilization (Table 4), the pressures obtained by the simple pðhÞ Vðhi Þ K c1
¼  1þ ðVð/ÞÞ d/ : ð9Þ
thermodynamic model and the experimental measurements are pi VðhÞ h f  h i hi
much closer. For Cases 1–4, the relative errors of the simple ther-
modynamic analysis are 7.4%, 21.5%, 13.5%, and 25.3%, respec- Our analysis exploits the fact that ignition, burning, and
tively. These differences exist because: (1) Even though the gas quenching all occur very close to h = 2p (TDC) with at most an
in the IC engine cylinder is very nearly ideal1 the gas constant R approximately p/6 deviation from TDC.2 We expand the volume
is not constant when conditions change in the engine cylinder. In function near TDC, after defining the independent variable
addition, the specific heat ratio c is also not constant. In fact, R D = h  2p. Thus, after integration and rearrangement for small val-
and c are functions of temperature with significant variations in ues of Df, Di, and D we obtain the following analytical formula for
the range of 10–20% [13]. The majority of the error is believed to p(h) from Eq. (9):
result from these variations. (2) The pressure measured is not the  
pðhÞ cðr  1Þ 2
real pressure in the cylinder because of the thermal shock to the ¼ 1þ ðDi  D2 Þ þ OðD4i ; D4 ; D2i D2 Þ
pi 4
pressure transducer [12,14,15]. (3) Real-engine processes like leak- " ! #
age, blowby and friction contribute to the differences. ðc  1ÞQ in ðc  1Þðr  1ÞðD3  D3i Þ
 1þ D  Di þ þ  :
pi V ci ðhf  hi Þ 12
4. Mathematical analysis of simplified cases ð10Þ

Important knowledge can be gained by analytically solving Eq. By defining three nondimensional parameters a ¼ cðr  1Þ=
(4) for simplified cases, as described for Model (ii) of Section 2. 4; b ¼ ðc  1ÞQ in =½pi V ci ðhf  hi Þ and c ¼ ðc  1Þðr  1Þ=12, we can
Analytical solutions produce scaling parameters and proportionali- rewrite Eq. (10) as

1
The critical temperature and pressure for O2 and N2 are 154 K, 50 atm and 126 K,
33 atm, respectively. Thus, the reduced temperature ranges from approximately 2.6
2
during compression to 12 during combustion, assuming a peak combustion Note that ±p/6 = ±30°, which lies beyond the bounds of the ignition and burnout
temperature of 2000 K. The reduced pressure ranges from 0.025 at intake to a peak crank angles (referred to TDC). Also, note that with respect to the entire cycle (2p
near 0.5. Putting all of these values in a compressibility chart shows that the radians) the ratio (p/6)/2p = 1/12 = 0.0833  1. Thus, it is mathematically valid to
compressibility factor is 0.99 or better. The ideal gas approximation is an excellent state that in these expressions the heat release occurs very close to TDC because the
approximation for the equation of state. deviation is smaller than 10% to either side.
Y. Shen et al. / Applied Thermal Engineering 29 (2009) 2379–2385 2383

a 140
b 140
120 Ideal 120
Ideal
Experiment
100 Experiment 100
p(atm)

80 80

p(atm)
60 60
calculation
calculation
40 40

20 20

0 0
357 361 365 369 373 377 381 CAD 347 351 355 359 363 367 371 375 CAD 379
Ca se 1 C ase 2

c 140
d 140
Ideal
120 120

100 calculation 100


p(atm)

p(atm)
80 80
Ideal
Experiment Experiment
60 60

40 40

20 20
calculation

0 0
357 361 365 369 373 377 CAD 381 360 364 368 372 376 380 384 388 392 396 400 CAD
C ase 3 C ase 4

Fig. 2. (a–d) Comparison of pressure profiles in the thermodynamic model and in the experiment. The profiles labeled ‘ideal’ do not incorporate heat loss effects whereas
those labeled ‘calculation’ do account for heat losses through the development of an accurate x(h) satisfying conditions (1)–(3) of Section 3.1.

pðhÞ in order to avoid Dm > Df and the undesirable case of a pressure


¼ ½1 þ aðD2i  D2 Þ þ     ½1 þ bðD  Di þ cðD3  D3i ÞÞ þ   :
pi peak after burnout. When we use4 a  6 and b  3 along with
ð11Þ Di  p/36 (see Fig. 1) we obtain Dm  0.20944 radians, which is
approximately halfway between TDC and the burnout crank angle
Expanding Eq. (11) and neglecting the O(D3) higher order terms,3 Df  0.41888 radians shown in Fig. 1 (recall that D = CAD  2p radi-
we obtain ans5). Thus, the pressure peak is expected to occur approximately
pðhÞ halfway between TDC and burnout.
¼ ð1 þ aD2i Þð1  bDi Þ þ bð1 þ aD2i ÞD  að1  bDi ÞD2 : ð12Þ
pi
4.1. Ignition before TDC
To find the maximum p(h)/pi, we differentiate Eq. (12) with
respect to D. The maximum value of p(h) occurs when the crank For this case, the cylinder pressure at TDC can be calculated
angle takes the value Dm = b(1 + aDi2)/2a(1  bDi), whereby from Eq. (12) at D = 0, viz.
2
pm b ð1 þ aD2i Þ2 pTDC
¼ ð1 þ aD2i Þð1  bDi Þ þ ; ð13Þ ¼ ð1 þ aD2i Þð1  bDi Þ: ð15Þ
pi 4að1  bDi Þ pi

where pm = p(hm). The value of the cylinder pressure pf = p(hf) at the A qualitative rendering of these calculations is shown in Fig. 3.
end of combustion is obtained from Eq. (12) evaluated at D = Df, viz. For the pressure trace of Case 3 of Table 3, we estimated a = 6.08
and b = 2.8. These values are obtained by imposing the following
pf
¼ ð1 þ aD2i Þð1  bDi Þ þ bð1 þ aD2i ÞDf  að1  bDi ÞD2f : ð14Þ conditions: identical value of the CAD for the start of combustion,
pi
hi; identical value of the CAD of the peak pressure, hm; identical va-
For reference and for comparison, we also calculate the pressure lue of the starting pressure pi. Shown in Table 5 are the initial pres-
at TDC. We note that for a post-TDC pressure peak the value of Dm sures (identical CAD and pressure values), the TDC pressures, the
must exceed zero. This pressure peak should occur before D = Df peak pressures and the final pressures at hf. The CAD at finish of
combustion for both calculated and experimental pressure traces
3 agree very well at 382°.
Since the next terms in the expansion of the first factor of Eq. (11) are
OðD4 ; Di2 D2 ; D4 Þ the product in Eq. (11) can be calculated to OðD3 ; D2i D; Di D2 ; D3i Þ
accuracy by excluding the product ðacD2i  D2 ÞðD3  D3i Þ. The formula for the pressure
2 4
becomes p=pi ¼ 1 þ bðD  DI Þ þ aðD2i  D2 Þ þ cðD3  D3i Þ þ ab ðDD2i  D3  D3i þ Di D2 Þ. The use of these values is justified in Section 4.1 below.
5
From this formula can be evaluated pressure, pressure change rate and rate of Note that Figs. 1 and 2 use crank angle degrees (CAD) not radians. Of course,
pressure change rate, etc. 360 CAD = 2p radians.
2384 Y. Shen et al. / Applied Thermal Engineering 29 (2009) 2379–2385

5. Conclusions

This work has demonstrated that the use of a simplified ther-


(
pi b 2 1 + aΔ i
2
)
2
modynamics-based model to describe the combustion part of the
4a(1 − bΔ i ) heat addition and work strokes of the IC engine cycle can produce
accurate pressure traces p(h). This procedure requires deriving a
realistic heat release function such as Eq. (7) that can be used to
pi (1 + aΔ i )(1− b Δ i )− pi
2 evaluate p(h) through integration (Eq. (4)). One explanation, per-
. haps, for the accuracy is that the heat release function is integrated,
. which is a smoothing operation.
A benefit of the simplified approach is the ability to derive the
isentropic expansion
characteristic parameters shown in Figs. 3 and 4, describing the loca-
isentropic compression tion of the pressure peak, the amplification of the pressure by com-
bustion, and other quantities such as the maximum pressure
Δi Δf
increase rate. The latter occurs at D = a(b2Di  1)/[3(ab2  c)]: since
Di < 0 for IC engines and since usually ab2 > c, this formula predicts
θi θf
that the pressure rise rate maximum occurs before TDC. Recent re-
Fig. 3. Pressure trace in the spark ignition engine from the analysis of Model (ii) in
search on HCCI engines [16] has shown pressure profiles that are
Section 4. The value of Dm at the pressure maximum is given under Eq. (12). similar to those developed here (in versions of Fig. 4 without the
post-TDC ignition delay, i.e. Di ? 0), and it appears from preliminary
comparisons that an approximately 2-fold pressure rise is realistic.
Comparisons of experiments and models such as these may serve
to provide more accurate estimates for the heat release function.
( )
pi 1 + aΔ i (1 − bΔ i ) +
2 (
b 2 1 + aΔ i
2 2
)
− pi In summary, we have demonstrated the following in this work:
4a(1 − bΔ i )
(1) A simple thermodynamic model is able to qualitatively and quan-
titatively describe in-cylinder pressure versus crank-angle pressure
traces p(h). (2) The simple model can be used to derive several non-
dimensional parameters of engine operation, including crank angle
at pressure maximum and pressure rise ratio. (3) The simple model
pi aΔi
2 can describe features of the pressure versus crank angle for ignition
after TDC, as in HCCI engines, for which ignition time is a key feature
even though it is difficult to control. Plots of in-cylinder pressure for
isentropic
expansion HCCI engines are shown in [16] for example. The interested reader
isentropic compression may compare the functional dependences of those profiles with
the global profiles produced here.
There is room for additional research using this simplified ap-
π proach, as more physical processes are included. The principal va-
θ
θ i = 2π + Δi θf = 2π + Δ f lue of this study is likely not in design because Models (i) and (ii) of
Section 2 are simple, reductive, illustrative, and elementary
Fig. 4. Qualitative analysis of pressure trace with ignition after TDC. As Di ? 0+ the
whereas the actual problem is complicated, inclusive, real, and
point of ignition approaches TDC from the right. The location of Dm can be
determined from the mathematical pressure/crank angle relationships in Section 4. practical. The value of this research resides rather in the several
nondimensional correlations that were produced in Section 4 and
in the pedagogical value of the modeling approach, which can
From Table 5 the calculated TDC pressure is higher than the serve as excellent instruction of the interconnectedness of engine
experimental TDC pressure, whereas the calculated peak pressure processes. The most complex of all processes is the heat addition
is lower than the experimental value. The probable reason is that (combustion) segment of the engine cycle.
with the linear heat release function, the rate of heat release near
the middle of the combustion stage is slower than that of real heat Acknowledgements
release as shown in Fig. 1. This generates a lower peak cylinder
pressure even with the higher TDC pressure. The actual heat re- This research arose out of a graduate combustion course project
lease is lower at the start of combustion and higher during the of author J.B. The heat release function of Section 3.1 was formu-
middle of the combustion stage. lated in the Ph.D. thesis of Y.S. under the direction of Professor
H.J. Schock in close collaboration with the late Prof. A.K. Oppen-
4.2. Ignition after TDC heim of U.C. Berkeley. The use of the data obtained by Steven
George at the MSU Automotive Research Experiment Station is
For the case of ignition after TDC, the cylinder pressure at TDC greatly appreciated.
can be calculated from the polytropic compression equation,
 c  c References
pTDC Vi ðr  1Þ 2 cðr  1Þ 2
¼ ¼ 1þ Di ffi 1 þ Di [1] G.J. Van Wylen, R. Sonntag, Fundamentals of Classical Thermodynamics, John
pi V TDC 4 4
Wiley & Sons, 1985.
¼1þ aD2i : ð16Þ [2] Y.A. Cengel, M.A. Boles, Thermodynamics: An Engineering Approach, McGraw-
Hill, 1998.
A diagram of the pressure trace with ignition after TDC is shown in [3] C.F. Taylor, The Internal-Combustion Engine in Theory and Practice, The M.I.T.
Fig. 4 along with the associated pressure and crank angle scalings. Press, 1985.
[4] C.R. Ferguson, Internal Combustion Engines: Applied Thermosciences, John
In this simplified model the ignition can be made to coincide with Wiley & Sons, 1986.
TDC. [5] J.B. Heywood, Internal Combustion Engine Fundamentals, McGraw-Hill, 1988.
Y. Shen et al. / Applied Thermal Engineering 29 (2009) 2379–2385 2385

[6] G.M. Rassweiler, L. Withrow, Motion pictures of engine flames correlated with [12] Y. Shen, H.J. Schock, T.-H.J. Sum, A.K. Oppenheim, Dynamic stage of
pressure cards, SAE Paper 380139, 1938. combustion in a direct injection methanol fueled engine, SAE Paper 2002-
[7] C.S. Daw, W.K. Kahl, Interpretation of engine cycle-to-cycle variation by 01-0998, 2002.
chaotic time series analysis, SAE Paper 902103, 2003. [13] V.P. Carey, Statistical Thermodynamics and Microscale Thermophysics,
[8] F.J.M. Brunt, C.R. Pond, J. Biundo, Gasoline engine knock analysis using cylinder Cambridge University Press, Cambridge, UK, 1999.
pressure data, SAE Paper 980896, 1996. [14] K. Roth, A. Sobiesiak, L. Robertson, S. Yates, Performance of optical fiber and
[9] Y. Shen, Development of a basis for control of combustion in an internal piezoelectric pressure transducers in CNG fuelled engine, in: Combustion
combustion engine, Ph.D. Thesis, Michigan State University, 2003. Institute/Canadian Section (CI/CS) Spring Technical Meeting, Canada, 2002.
[10] C. Ludwig, S. Leonhardt, M. Ayoubi, R. Isermann, Measurement and monitoring [15] T. Tousignant, J. Tjong, G.T. Reader, Experience using different in-cylinder
of pressure curves in diesel engines, in: Proceedings of the American Control pressure measurement methods in an IC engine, in: Combustion Institute/
Conference, San Francisco, California, June 1993. Canadian Section (CI/CS) Spring Technical Meeting, Canada, 2002.
[11] Y. Shen, J. Bedford, I.S. Wichman, Thermodynamic analysis of direct injection [16] A. Pires Da Cruz, Three-Dimensional Modeling of Self-Ignition in HCCI and
methanol fueled engines, in: Proceedings of the Central States Section of the Conventional Diesel Engines, Combustion Science and Technology 176 (2004)
Combustion Institute, May 21–23, Cleveland, OH, 2006. 867–887.

You might also like