Thermodynamic optimization of the

turbofan cycle
Yousef S.H. Najjar
Mechanical Engineering Department, Jordan University of Science and Technology, Irbid, Jordan, and
Sharaf F. Al-Sharif
King Abdulaziz University, Jeddah, Saudi Arabia

Abstract
Purpose – To develop and find the effect of combination of four cycle design variables that minimizes the specific fuel consumption (sfc) of a turbofan
engine.
Design/methodology/approach – After choosing the four variables, namely bypass ratio (B), fan pressure ratio, overall pressure ratio, and turbine
inlet temperature (T04), first the sfc was minimized without a minimum thrust constraint. Then, a minimum specific thrust constraint was introduced.
Findings – The unconstrained-specific thrust is a two-dimensional optimization problem, whereas the specific thrust constrained problem was found
to be a three-dimensional one.
Research limitations/implications – The variables B and ı̈ are limiting factors to further improvement, as set by their maximum practical values,
whereas the other two variables are to be optimized.
Practical implications – A very useful work, in which numerical optimization program was developed, for a turbofan cycle and could be extended to
other cycles.
Originality/value – This paper offers a great help to those intending to optimize certain cycles with a number of variables.

Keywords Optimization techniques, Fuel consumption

Paper type Research paper

Nomenclature hc ¼ polytropic efficiency of the compressor and fan

A ¼ area, m2 ht ¼ polytropic efficiency of the turbine
As ¼ specific area (area/unit mass flow rate), m2 s/kg hm ¼ mechanical efficiency of the shaft
alt. ¼ altitude, m hb ¼ combustion efficiency of the burner
B ¼ bypass ratio hj ¼ isentropic efficiency of the jet nozzles
C ¼ velocity magnitude, m/s pin ¼ intake pressure ratio
cp ¼ constant pressure specific heat, J/kg K pf ¼ fan pressure ratio
F ¼ thrust, N pc ¼ overall pressure ratio
Fs ¼ specific thrust (thrust/total air mass flow rate), pb ¼ burner pressure ratio
Ns/kg pt ¼ turbine pressure ratio
f ¼ fuel air ratio of core flow (mass flow rate of fuel/ tin ¼ intake temperature ratio
mass flow rate of core air flow) tf ¼ fan temperature ratio
Hc ¼ enthalpy of combustion of air, MJ/kg tc ¼ overall compression temperature ratio
M ¼ mach number tt ¼ turbine temperature ratio
m_ ¼ mass flow rate, kg/s
n ¼ polytropic index of compression Subscripts
P ¼ pressure, bar, Pa 0 ¼ stagnation state
R ¼ ideal gas constant, J/kg K 1, 2, . . . 8 ¼ station numbering
sfc ¼ specific fuel consumption (mass flow rate of a ¼ air
fuel/unit thrust produced), g/kN s b ¼ burner
T ¼ temperature, K c ¼ compressor, combustion, critical, cold
(bypass flow)
Greek symbols f ¼ fan
g ¼ specific heat ratio g ¼ combustion gas
hi ¼ isentropic efficiency of the intake h ¼ hot (core flow)
i ¼ inlet
The current issue and full text archive of this journal is available at in ¼ inlet
www.emeraldinsight.com/1748-8842.htm ISA ¼ international standard atmosphere
j ¼ jet nozzle
m ¼ mechanical
Aircraft Engineering and Aerospace Technology: An International Journal p ¼ constant pressure
78/6 (2006) 467– 480
q Emerald Group Publishing Limited [ISSN 1748-8842]
s ¼ specific
[DOI 10.1108/00022660610707139] t ¼ turbine

467
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

Introduction Theoretical analysis and methodology

After the intended application of the aircraft is decided, the In what follows the model of the cycle to be used in this work
serious stages of engine design may commence. These include will be derived, based mainly on the development in Cohen
most importantly thermodynamic cycle design, aerodynamic et al. (1987) with some of the notation adopted from
design of turbomachinery, mechanical design, and off-design Mattingly (1996) and Kerrebrock (1992).
performance prediction. A schematic diagram of the turbofan engine is shown in
This work deals specifically with thermodynamic cycle Figure 1.
design. This consists of detailed calculations on the Input:
thermodynamic changes that the working fluids (air and .
Properties of working fluids: cpa ; cpg ; ga ; gg ; R; H c
combustion gases) experience as they flow through the engine .
Component performance parameters: hi,hc,hm,hb,ht,hj,pb
components, taking into account component losses as
.
Operating conditions: alt:; M a
realistically as possible (Mattingly, 1996; Kerrebrock, 1992).
.
Cycle design parameters: B; pf ; pc ; T 04
The purpose of these calculations is to evaluate the The thrust produced by the engine is the sum of the
performance of a turbofan engine, for different momentum and pressure thrust components of each nozzle:
combinations of design choices and operating conditions
(Cohen et al., 1987; Mattingly, 1996; Kerrebrock, 1992). _ h ðC 7 2 Ca Þ þ m
F¼m _ c ðC 8 2 Ca Þ þ A7 ðP 7 2 P a Þ
It is important to note that the characteristic behavior of a
specific engine under different operating conditions depends þ A8 ðP 8 2 P a Þ ð1Þ
on its geometry (blade angles and dimensions). Since,
geometry is not accounted for in cycle design, the trends If any of the nozzles is unchoked, its exhaust gas expands to
and plots that result do not represent the behavior of a specific ambient pressure, and its pressure thrust is then zero.
engine, but different engines each operating at its own Dividing equation (1) by m _ a ; and defining the bypass ratio B
“design-point” and having the geometry that will produce that as:
combination of cycle design variables (Kurzke, 1999). m
_c
In other words, the aim of cycle design is to select the values B; ;
m
_h
of the cycle design variables that are favored mainly by
thermodynamic considerations. After that the and noting that:
turbomachinery that will actualize these values of the cycle
design variables when operating at the design-point are ma ¼ mh þ mc ;
developed. And in turn after that, the behavior of the engine
at off-design conditions is investigated. equation (1) is rewritten in terms of specific thrust as:
This reveals the importance of cycle design, especially when F 1 B A7
considering the enormous cost of developing an aircraft gas Fs ¼ ¼ ðC 7 2 C a Þ þ ðC 8 2 Ca Þ þ ðP 7 2 P a Þ
_a Bþ1
m Bþ1 m
_a
turbine engine, and the long life cycle that successful engines
have (Cohen et al., 1987). The cycle must be properly A8
designed because it is the foundation for all subsequent work. þ ðP 8 2 P a Þ:
m
_a
Traditionally, cycle design involves extensive parametric
variations (Powel, 1991). The performance parameters are It is more convenient to write the third and fourth terms in
studied for a large number of combinations of design choices. terms of specific area As, the area per unit mass flow through
The task quickly becomes difficult as the number of cycle each nozzle. Thus:
design variables increase, as in the case of the turbofan engine.
It becomes difficult to present and visualize the trends for 1 B m
_ h A7
Fs ¼ ðC 7 2 Ca Þ þ ðC 8 2 C a Þ þ ðP 7 2 P a Þ
multi-dimensional variations. The problem becomes even Bþ1 Bþ1 m
_h m_a
more complicated when design constraints are introduced. m
_ c A8
The alternative to that is to employ a numerical þ ðP 8 2 P a Þ
m
_c m_a
optimization algorithm that searches for the optimum of a
selected figure of merit automatically. A program that
1 B
performs such an optimization would be a very useful and Fs ¼ ½ðC7 2 C a Þ þ As7 ðP 7 2 P a Þ
þ ½ðC 8 2 C a Þ
powerful tool for cycle design. Design constraints may also be Bþ1 Bþ1 ð2Þ
incorporated in the optimization program to find a truly þ As8 ðP 8 2 P a Þ

feasible optimum.
After the optimum combination of cycle design variables is
found, parametric studies may be carried out in the vicinity of
Figure 1 Schematic diagram of the turbofan engine with station
the optimum to determine its sensitivity to each of the cycle
numbering
design variables. In this case the region and number of
combinations that must be examined in the parametric study
is greatly reduced.
In this work a numerical optimization program is developed
and used to find the combination of four cycle design
variables that minimizes the specific fuel consumption (sfc) of
a turbofan cycle subject to a number of simple constraints.

468
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

It is now necessary to find expressions for the unknown

n =ðng 21Þ
variables C 7 ; C8 ; P 7 ; P 8 ; As7 ; As8 in terms of input variables. pt ¼ tt g
Starting with the core nozzle, the definition of stagnation
temperature gives: An expression for T7 is found from knowledge of the pressure
C2 ratio and the definition of the isentropic efficiency of nozzle:
cpg T 07 ¼ cpg T 7 þ 7
2 T 06 2 T 7
hj ¼ ;
Rearranging: T 06 2 T 07
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  
C7 ¼ 2cpg ðT 07 2 T 7 Þ T0
T 7 ¼ T 06 1 2 hj 1 2 7 ;
T 06
The duct is assumed to be adiabatic so that T07 ¼ T06, and:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi "   #
C7 ¼ 2cpg ðT 06 2 T 7 Þ ð3Þ T7 P 7 ðgg 21Þ=gg
¼ 1 2 hj 1 2
T 06 P 06
An expression for T06 is found by making an energy balance
over the gas-generator on a unit mass basis: At this point the calculations will depend on whether the
1 cpa B nozzle is choked or unchoked. For the core nozzle, the critical
ð1 þ f Þcpg ðT 04 2 T 06 Þ ¼ ðT 02 2 T 01 Þ
Bþ1 hm B þ 1 pressure ratio is given by:
P 06 1
1 cpa ¼h   igg =ðgg 21Þ
ðT 03 2 T 01 Þ
þ Pc g 21
B þ 1 hm 1 2 h1j gg þ1
g

    
cpa 1 T 02 T 03 If ðP 06 =P a Þ $ ðP 06 =P c Þ; the nozzle is choked and:
T 04 2 T 06 ¼ T 01 B 21 þ 21
hm cpg 1þf T 01 T 01 P 06
ð4Þ P7 ¼ Pc ¼ ;
P 06 =P c
Making the following definitions: 2T 06
T 01 P 01 T7 ¼ Tc ¼
tin ; ; pin ; gg þ 1
Ta Pa
Substituting fors 06 and T7 in equation (3):
Tffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
T 02 P 02  ffi
tf ; ; pf ; 2
T 01 P 01 C 7 ¼ 2C pg T 04 tt 1 2 ðchokedÞ
gg þ 1
T 03 P 03
tc ; ; pc ;
T 01 P 01 The density of exhaust gas is given by:
P7
T 06 P 06 r7 ¼ ;
tt ; ; pt ; RT 7
T 04 P 04
knowledge of which is employed in the continuity equation to
equation (4) can be rewritten: find the specific area:
cpa 1 Ta A7 1
tt ¼ 1 2 tin ½Bðtf 2 1Þ þ ðtc 2 1Þ
; ð5Þ As7 ¼ ¼
hm cpg 1 þ f T 04 _ h r7 C 7
m
which, except for the fuel air ratio f, is in terms of known or Otherwise, if ðP 06 =P a Þ , ðP 06 =P c Þ the nozzle is unchoked,
calculable quantities. Specifically: P 7 ¼ P a ; and T7 is given
( by:" #)
C 2a  
tin ¼ 1 þ ; P 7 ðgg 21Þ=gg
2cpa T a T 7 ¼ T 06 1 2 hj 1 2
P 06
tf ¼ pf
ðna 21Þ=na
; ( "  ðgg 21Þ=gg # )
1
T 7 ¼ T 04 tt 1 2 hj 1 2
tc ¼ pðn
c
a 21Þ=na pin pc pb pt

The fuel air ratio will be dealt with shortly. Substitution into equation (3) yields:
( "  ðgg 21Þ=gg # )1=2
T06 is now given as: 1
C 7 ¼ 2cpg T 04 tt hj 1 2 ðunchokedÞ
T 06 ¼ T 04 tt pin pc pb pt

Similarly, P06 is given by:

Similarly, for the bypass nozzle, the definition of stagnation
P 06 ¼ P a pin pc pb pt temperature gives:
C2
cpa T 08 ¼ cpa T 8 þ 8
in which: 2
 ðga 21Þ=ga from which:
C2a qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pin ¼ 1 þ hi ; C8 ¼ 2cpg ðT 08 2 T 8 Þ
2cpa T a

and: Again the duct is assumed to be adiabatic so that T08 ¼ T02, and:

469
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi objective function, we may now direct our attention to the

C8 ¼ 2cpa ðT 02 2 T 8 Þ ð6Þ optimization method.

T 02 ¼ T a tin tf
Optimization technique
As before, the nozzle pressure ratio must be checked against the
The method of optimization that is used is called the
critical pressure ratio, which is given by:
P 02 1 conjugate gradient method. In general, optimization
¼h   iga =ðga 21Þ techniques can be grouped into two main categories:
Pc g 21
1 2 h1j gaa þ1 methods that use gradient information, and direct search
methods. As its name implies, the conjugate gradient method
If ðP 02 =P a Þ $ ðP 02 =P c Þ; the nozzle is choked: falls into the first category.
P 02 In all gradient computing methods the optimization
P8 ¼ Pc ¼ ;
P 02 =P c problem is subdivided into two sub-problems:
1 determining a suitable search direction; and
2T 02 2 taking the optimum step size in that direction.
T8 ¼ Tc ¼
ga þ 1
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The different members of the family vary in the way they
 ffi address the first sub-problem.
2
C 8 ¼ 2C pa T a tin tf 1 2 ðchokedÞ The simplest of these methods is the method of steepest
ga þ 1
descents. In this method the local gradient is evaluated in
P8 each step, and the search direction is taken (in the
r8 ¼ ; minimization problem) as the negative of the gradient,
RT 8
which is, by definition of the gradient, the direction of
A8 1 steepest descent. This method is usually the least efficient of
As8 ¼ ¼
_ c r8 C 8
m the family, especially when the scales of the design variables
are not similar such that the objective function has a narrow,
If ðP 02 =P a Þ , ðP 02 =P c Þ; the nozzle is unchoked, P 8 ¼ P a ; and stretched contour map. This results from two facts:
T8 is given by: ( " #) 1 the local gradient of a function (i.e. its negative) does not
 
P 8 ðga 21Þ=ga generally point to the minimum; and
T 8 ¼ T 02 1 2 hj 1 2
P 02 2 at the minimum along some search direction the local
( " gradient is perpendicular to the search direction.
 ðga 21Þ=ga # )
Pa What this means is that the algorithm will zigzag its way along
T 8 ¼ T a tin tf 1 2 hj 1 2
P a pin pf small mutually perpendicular steps, even if it is relatively close
to the minimum (Arora, 1989).
Substitution into equation (6) and some algebraic manipulation The conjugate gradient method results from the idea of
yields: searching along “non-interfering” directions (Arora, 1989).
( "  ðga 21Þ=ga # )1=2
1 Simply put, minimization along one direction should not
C 8 ¼ 2cpa T a tin tf hj 1 2 ðunchokedÞ interfere or ruin previous minimizations. Based on this idea a
pin pf
sequence of arguments are made which lead to the derivation
of the method.
The expressions for the unknown variables in equation (2), which The two situations that are considered can be listed as
we had set out to find, are now complete, and the specific thrust follows:
may be evaluated. It remains now to address the sfc. 1 The sfc is to be minimized with no constraint on specific
The sfc is given by: thrust. The optimization will be run for a combination of
different operating conditions and maximum B
m
_f fm _h 1 f constraints given in the Table I.
sfc ¼ ¼ ¼ ;
F F sm_ a B þ 1 Fs 2 sfc is to be minimized subject to a constraint of minimum
specific thrust.
The fuel air ratio f is found by making an energy balance over
the combustor: The combination of operating conditions can be seen in
Table I.
_ h cpa ðT 03 2 298Þ þ f m
m _ h hb H c ¼ ðm
_ h þ fm
_ h Þcpg ðT 04 2 298Þ

where Hc is the enthalpy of combustion at 258C. Solving for f Discussion of results

and employing some previously defined quantities, the above
equation becomes: Minimizing sfc with no constraint on specific thrust
The results of the optimization runs for the previously
cpg ðT 04 2 298Þ 2 cpa ðT a tin tc 2 298Þ mentioned conditions are summarized in Table II. Each case
f ¼
hb H c 2 cpg ðT 04 2 298Þ is given a number in the table for easy referral. A parametric
variation of T04 and pc around the optimum in each of the
It is implemented in TK-Solver 3.0 in the rule function cases was performed, and the results were plotted in the form
SpecFuelCons and auxiliary rule functions Nozzle, of carpet plots of sfc versus Fs in Figures 2-5. The effect of pf
IsNozzleChoked and NozzleThrust. Having formulated the is considered separately in Figures 6-8, where case 8 is taken

470
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

Table I Combinations of operating conditions and bypass ratio are kept constant, sfc decreases 9.3 percent from 1.93 to
constraints 1.75.
.
The optimum pf increases as Mach number increases (all
Case Altitude Mach Bmax other variables kept constant). This is seen, for example,
1 9,000 0.8 5 when cases 8 and 11 are compared, in which optimum pf
2 9,000 0.8 10 increases from 1.75 to 1.8 (þ2.8 percent) when M is
3 9,000 0.8 15
increased from 0.9 to 0.8 (þ12.5 percent).
.
The optimum T04 increases as B is increased. This is
4 9,000 0.9 5
exemplified by cases 8 and 9 where increasing B 50 percent
5 9,000 0.9 10
from 10 to 15 increases optimum T04 by 14.4 percent from
6 9,000 0.9 15
1,600 to 1,830 K.
7 11,000 0.8 5 .
The optimum T04 increases as pc is increased. This is seen
8 11,000 0.8 10 in the carpet plots where the optimum T04 consistently
9 11,000 0.8 15 shifts to the right when pc is increased. This may be easier
10 11,000 0.9 5 to observe on the B ¼ 15 carpets because of their higher
11 11,000 0.9 10 curvature, but it is applicable to all cases.
12 11,000 0.9 15 .
Increasing B significantly lowers specific thrust, as can be
seen when comparing cases 8 and 9. Increasing B from 10
to 15 (þ50 percent) decreases Fs of the optimum cycle
as the reference cycle. Finally, a sensitivity plot for the from 137.1 to 124 N s/kg (2 10.6 percent).
reference case is shown in Figure 9. .
The effect of increasing T04 in increasing specific thrust is
Throughout this work, a constraint will be called “active” if stronger at lower bypass ratios. For example, in Figure 1,
the value of the constrained variable in its respective optimum when moving from T04 ¼ 1,400 to 1,550 K (an increase of
cycle is found to be equal to its limiting value (whether 10.7 percent) on the pc ¼ 45 line of the B ¼ 5 carpet, Fs
maximum or minimum). Thus, if in an optimization run, B is increases from 153.8 to 192.7 N s/kg. This represents a
constrained to a maximum of five, for example, and the value 25.3 percent increase in Fs. While a comparable
of optimum B found by the optimization process is also five, 10.5 percent increase in T04 from 1,900 to 2,100 K on
then the constraint on B is called an active constraint. If, on the pc ¼ 45 line of the B ¼ 15 carpet produces a thrust
the other hand, the optimum value of the constrained variable increase of only 18.3 percent (from 120 to 142 N s/kg.)
does not reach its limiting value, the constraint will be called The reason for that is that T04 represents the energy input
“passive”. to the core flow. As the bypass flow increases, the core
The concept of active and passive constraints helps in flow becomes relatively less important, and a larger
determining the real limiting factors to further improvement increase in T04 is needed to obtain the same specific thrust
under the conditions considered. increase.
Impact of the bypass ratio constraint
General observations An outstanding observation in Table II is that the B and pc
A number of points are observed in Table II and the constraints are always active constraints. This means that (in
corresponding carpet plots of Figures 2-5. These may be this situation, where no minimum Fs constraint is placed) they
listed as follows: are limiting factors. That is to say, sfc can be further lowered
.
The constraints on bypass ratio and overall pressure ratio if these constraints were not present, or if they could be
are active constraints in all cases. extended. However, while it is true that the sfc calculated
. The optimum pf is a function of bypass ratio, and from cycle analysis appears to improve continuously as B is
decreases as B is increased. For example, in cases 7 and 8, increased (provided T04 can also be increased), this does not
where B is doubled from 5 to 10 while all other variables hold true when installation effects (inlet and nozzle drag) are

Table II Results of optimization at different operating conditions and maximum B constraint (with no constraint on Fs)
Operating cond. Constraints Optimum cycle sfc Fs
Case No. Altitude Mach B T04 (K) pc B pf T04 (K) pc (g/kNs) (N s/kg)
1 9,000 0.8 5 2,000 40 5 1.94 1,450 40 18.1 173.4
2 9,000 0.8 10 2,000 40 10 1.76 1,700 40 16.75 141.6
3 9,000 0.8 15 2,000 40 15 1.7 1,950 40 16.15 129
4 9,000 0.9 5 2,000 40 5 1.97 1,500 40 19.4 169.5
5 9,000 0.9 10 2,000 40 10 1.8 1,775 40 18.1 139.3
6 9,000 0.9 15 2,000 40 15 1.7 2,000 40 17.46 122.1
7 11,000 0.8 5 2,000 40 5 1.93 1,360 40 17.5 167.5
8 11,000 0.8 10 2,000 40 10 1.75 1,600 40 16.23 137.1
9 11,000 0.8 15 2,000 40 15 1.68 1,830 40 15.65 124
10 11,000 0.9 5 2,000 40 5 1.96 1,410 40 18.76 163.8
11 11,000 0.9 10 2,000 40 10 1.8 1,675 40 17.5 134.4
12 11,000 0.9 15 2,000 40 15 1.72 1,925 40 16.92 122.3

471
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

Figure 2 Parametric variation of T04 and pc around optima at 9,000 m, 0.8 Mach

Figure 3 Parametric variation of T04 and pc around optima at 9,000 m, 0.9 Mach

472
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

Figure 4 Parametric variation of T04 and pc around optima at 11,000 m, 0.8 Mach

Figure 5 Parametric variation of T04 and pc around optima at 9,000 m, 0.9 Mach

473
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

Figure 6 sfc versus T04 at different pf

Figure 7 sfc versus B at different pf

474
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

Figure 8 sfc versus pc at different pf

Figure 9 Sensitivity of the sfc of the reference cycle to deviations from optimum values the design variables

475
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

accounted for. To account for this difference some references The value of pc,max has been assumed to be 40 throughout
such as Mattingly (1996) and Kerrebrock (1992) differentiate this work.
between the two quantities installed sfc (tsfc), which includes An important point to make here is that, when pc is
installation losses, and uninstalled sfc (sfc), from cycle increased beyond current limits, the value of polytropic
calculations. The installed sfc depends on how the engine is efficiency assumed in the model becomes questionable. It has
installed in the nacelle, and is related to the uninstalled sfc been assumed that polytropic efficiencies of the
through the relation: turbomachinery are constant at 0.9, a value that reflects
current technology up to the current limit of pressure ratio
sfc (Press et al., 1988). Even if technological advances allow the
tsfc ¼
1 2 finlet 2 fnoz pressure ratio of the turbomachinery to increase, it becomes
harder to maintain high polytropic efficiencies. Therefore,
where finlet ; fnoz are the inlet and nozzle loss coefficients, trying to draw conclusions from plots that go too far beyond
respectively, and are dimensionless measures of their drag. current limits is not reliable.
Wilson (1984) points out that due to increasing nozzle losses, Several references such as Mattingly (1996, 1999),
the installed specific fuel consumption (tsfc) reaches a Kerrebrock (1992) and Kurzke (1999) point out that the
minimum at about B ¼ 20. This of course, cannot be validity of the results of parametric cycle analysis depends on
verified by cycle calculations, because cycle calculations do the realism with which the variation of efficiency with pressure
not account for installation effects. The point of this ratio is accounted for. Mattingly (1999) gives correlations for
elaboration is to stress that the continuously improving turbine and compressor polytropic efficiency variation with
trend of sfc with increasing B, as predicted by cycle the respective pressure ratios that represent current
calculations alone can be misleading if installation effects technology levels for industrial gas turbines. However, it is
are not kept in mind. mentioned that the correlations over estimate the losses for
In addition to that, as the bypass ratio is increased, the multispool aircraft engines. According to Press et al. (1988),
specific thrust of the engine decreases, as can be seen clearly polytropic efficiencies of 0.9 up to the current limit of
in Figures 1-4. This means that for a given thrust requirement pressure ratio can be achieved for aircraft engines.
(not specific thrust), if B is increased the engine diameter The observation in Table II that the B and pc constraints
must be increased to increase the air mass flow rate. are always active constraints, suggests that the problem of
Consequently, the weight and (external) aerodynamic drag minimizing sfc without considering a minimum specific thrust
of the engine also increase. Thus, even if the sfc decreases, the is actually a two-dimensional optimization problem. After
load on the engine, the thrust it must provide to overcome the setting B and pc to their maximum practical values, the
drag, will also increase. This means that if the actual amount optimum T04 and pf combination must be found.
of fuel consumed (e.g. kg) is considered, instead of the sfc
Illustrating the two-dimensional nature of the problem
(e.g. kg/kN), the optimum B is further lowered (Wilson,
Figure 6 shows a plot of sfc versus T04 for different pf , with B
1984). One study, mentioned in Wilson (1984), which
and pc set to their maximum values for the reference case.
considered this point found that there was not much fuel
Two points are visible in this figure:
efficiency to be gained beyond a bypass ratio of about eight. 1 for each pf there is an optimum T04 that minimizes sfc;
Ultimately, for any accurate conclusions to be made in this
and
regard, the engine must be studied with proper consideration 2 of these pairs of (pf , T04,opt), there is one pair (pf,opt,
of installation effects and weight implications. T04,opt) that gives the minimum sfc among the set, in this
case: 1.75,1600 K, respectively.
Impact of the pressure ratio constraint
It is seen in Figures 1-4 that, above a certain T04, increasing Since, it has been shown that increasing B or pc would
pc improves sfc. For example, in Figure 1 in the B ¼ 5 carpet improve sfc, and since they are set to their maximum, it
at T04 ¼ 1,400 K, there is little improvement in sfc when follows that this pair (pf,opt, T04,opt) is the global minimum of
moving from pc ¼ 40 to pc ¼ 45. Below this temperature sfc the problem as it is currently defined.
actually deteriorates as can be seen on the 1,350 K line, where This is further shown in Figures 7 and 8. Figure 7 shows a
the sfc increases from about 18.7 to slightly more than 19 g/ plot of sfc versus bypass ratio at different pf , while T04 and pc
kN s when moving from pc ¼ 40 to 45. But above a certain are held at 1,600 K and 40, respectively. The plot shows that
temperature, about 1,400 K in this case, sfc decreases with there is an optimum B for each pf , and that as pf decreases
increasing pc. the optimum B increases. When the maximum B is marked, it
Is this trend continuous? Figure 8 suggests that if the other can be seen that pf of 1.75 gives the lowest sfc. If pf is
cycle design variables are held constant, there will be a value increased or decreased sfc increases, but if B is increased
of pc after which sfc will start to increase. However, if T04 is beyond ten, then sfc can be decreased.
allowed to increase the trend is practically continuous. The same concept is seen in Figure 8, which is a plot of sfc
The maximum overall pressure ratio is limited by the versus pc at different pf , while holding B and T04 at 10,
temperature limit of the compressor materials, which is 1,600 K, respectively. Again when the maximum pc is
currently 920 K (Press et al., 1988). If standard air at 288.2 K, marked, it becomes apparent that pf ¼ 1.75 gives the lowest
and a polytropic efficiency of 0.9 are assumed, this sfc, and that sfc can be lowered by increasing pc beyond its
corresponds to a pressure ratio of: current limit.
  Sensitivity of sfc to deviations from optimum cycle
920 0:9£3:5 Figure 9 shows a sensitivity plot for the reference optimum
pc ¼ ¼ 38:7
288:2 case. The plot was generated by holding the cycle design

476
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

variables at their optimum values (B ¼ 10, pc ¼ 40, conditions. To explain this figure, the table shows that in the
pf ¼ 1.75, T04 ¼ 1,600 K), and then systematically changing first case when Fs,min is set to 100 N s/kg, the constraint is
one of the variables while holding the others constant. passive, and the specific thrust of the resulting optimum cycle
The plot shows that the sfc of the cycle is most sensitive to is found to be 137 N s/kg. This practically means that for
T04 and pf , with deviation in T04 being more detrimental these conditions this minimum specific thrust is “guaranteed”
below the optimum and pf above. As expected, sfc improves but if Fs,min is increased above this value, it becomes an active
when pc and B increase, but the plot shows sfc to improve constraint and a penalty on sfc is incurred. Another
slightly and then to deteriorate when B is increased. This is outstanding feature is that B is no longer a limiting factor
because when B is increased the optimum pf changes except at low Fs,min values. This means that with the
(decreases), but pf is held constant so sfc decreases slightly introduction of the minimum Fs constraint the problem has
until B reaches an optimum and then starts to increase. This become three-dimensional (pc is still an active constraint).
can also be seen in Figure 6. Table III is constructed from the The table also shows that when B steps out as a limiting
results in Figure 9 to summarize the effect of a 5 percent factor (moving down the table), T04 steps in. After that, T04
decrease in each of the design variables on the sfc of the continues to be a limiting factor until the optimum B becomes
optimum cycle. low enough to allow the required Fs,min to be achieved with a
Generally, the plot shows that if T04 and pf can be lower T04. This occurs somewhere between 400 and 450 N s/
controlled within ^5% of their optimum value, an sfc within kg (between B ¼ 3 and 1.5).
about 2.65 percent of the minimum can be achieved. It also Comparison with the graphical method
shows that B can be decreased to 80 percent of its maximum A graphical method for cycle optimization involving extensive
value (2 20 percent) with only a 5 percent penalty in sfc. The parametric variations is described in Cohen et al. (1987). This
penalty may be even smaller if T04 and pf are optimized for method involves finding the pairs of (T04,pf,opt) at fixed B and
the new B. Similarly, pc may be decreased 20 percent of its pc, plotting sfc versus Fs for these pairs, repeating for several
maximum with a penalty of about 2.8 percent in sfc. B, and finally repeating the whole process altogether at
different pc. The envelope curve for the family of different B
Minimizing sfc for a given specific thrust requirement curves at constant pc gives the plot of optimum variation of
The problem becomes more meaningful when a minimum sfc with Fs at that pc.
specific thrust constraint is introduced. As mentioned Since, Table IV shows that pc is always a limiting factor,
previously, if the specific thrust of an engine decreases the this only needs to be done at the maximum pc. Figure 10
engine must ingest more air to produce a given required shows the result of this parametric variation performed at
thrust. This means the engine diameter must be increased, pc ¼ 40, with B ranging from 0.1 to 10. Each constant B
which introduces a number of penalties. Most importantly, curve was obtained by varying T04 from 900 to 2,000 K,
the weight and aerodynamic drag will increase. Other factors finding the optimum pf at each T04, and plotting the
include ground clearance and landing gear length, and corresponding sfc versus Fs. Superimposed on this plot is the
transportation difficulty for engines above 3 m in diameter plot of optimum sfc versus Fs obtained from Table IV, shown
(Wilson, 1984). as a dashed line, which incidentally happens to be the
A minimum specific thrust requirement for a given envelope curve that the graphical method seeks to find.
application may be obtained by knowledge of the required The obvious advantage of the numerical optimization
thrust and forward speed, and by specifying a maximum approach is the saving in calculation and plotting effort. But
allowed engine diameter. more importantly, the identity of the cycle is difficult to
Table IV summarizes the results of a series of optimization determine from the graph. The graph may outline the trend of
runs for progressively increasing Fs,min. The optimization runs optimum variation, but the corresponding cycle design
were carried out for an altitude of 11 km and a flight Mach variables B, T04 and pf cannot be read directly (unless
number of 0.8. The constraints were fixed at Bmax ¼ 10, constant parameter lines are drawn, but that adds to the
pc,max ¼ 40, T04,max ¼ 2,000 K, except for Fs,min which effort).
progressively increases from 100 to 580 N s/kg. The table
lists the optimum cycle design variables, and the status of the Introducing additional constraints with: single stage fan
constraints in each case. An “A” in a constraint status column Another advantage of the numerical optimization approach
is the ease of incorporating practical design constraints.
denotes an active constraint, while a “P” denotes an inactive
Powel (1991) mentions that for a single stage fan pf , 1.9.
or passive constraint.
This constraint was added to generate Table V, which
The table shows that, except in the first case, the minimum
shows that when this constraint is added pf becomes the
Fs constraint becomes a limiting factor in minimizing sfc.
limiting factor instead of T04, and the optimum B quickly
Specifically, this begins above an Fs,min of 137 N s/kg for these
decreases as Fs,min increases. The impact of this constraint
on sfc can be visualized in Figure 11, in which the dashed
Table III Sensitivity of optimum cycle to 5 percent decrease in design line is the plot of (constrained) optimum sfc versus Fs
variables from Table V.
Design variable Dsfc (percent)
Conclusions
pf þ 1.10
pc þ 0.65 Consideration of the problem of minimizing sfc without a
B þ 0.78 constraint for minimum Fs has revealed a number of points.
T04 þ 2.65 Table I showed that the maximum B and pc are limiting
factors for all cases, which means that the problem is

477
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

Table IV Results of optimization with a progressively increasing minimum Fs constraint

Constraints Optimum cycle Constraint status
Fs,min B pf T04 pc B pf T04 pc Fs sfc Fs,min B pf T04 pc
100 10 – 2,000 40 10 1.75 1,600 40 137 16.17 P A – P A
150 10 – 2,000 40 10 1.85 1,683 40 150 16.2 A A – P A
200 10 – 2,000 40 9.7 2.3 2,000 40 200 16.8 A P – A A
250 10 – 2,000 40 7 2.84 2,000 40 250 18 A P – A A
300 10 – 2,000 40 5.2 3.54 2,000 40 300 19.3 A P – A A
350 10 – 2,000 40 4 4.44 2,000 40 350 20.7 A P – A A
400 10 – 2,000 40 3 5.6 2,000 40 400 22.14 A P – A A
450 10 – 2,000 40 1.5 6.1 1,658 40 450 23.63 A P – P A
500 10 – 2,000 40 0.5 6.08 1,407 40 500 25.12 A P – P A
550 10 – 2,000 40 0.2 6 1,333 40 550 26.4 A P – P A
580 10 – 2,000 40 0.1 6 1,333 40 580 27.15 A P – P A
Notes: A, active constraint; P, passive constraint

essentially a two-dimensional optimization problem. optimum cycle is relatively most sensitive to T04 and pf . A
Generally, sfc continues to improve as B and pc are 5 percent decrease from optimum value in T04 and pf was
increased, provided T04 and pf are optimized. Practical found to incur a penalty of 2.65 and 1.1 percent. This is
considerations, however, limit the potential improvements. compared to a 0.652 and 0.776 percent penalty incurred by a
Some general trends were observed, which might be comparable decrease in pc and B, respectively.
summarized as follows: The nature of the problem changes when a minimum Fs
.
optimum T04 increases when either B or pc is increased; constraint is introduced. B no longer becomes a limiting
.
optimum pf decreases as B is increased; and factor, and the problem becomes three-dimensional (B, pf ,
.
increasing B significantly decreases Fs. T04). The overall pressure ratio, however, remains to be
limiting factor in all the cases studied.
The sensitivity analysis of the reference optimum cycle Using numerical optimization had a number of advantages.
showed that sfc is not very sensitive to small deviations from First, it allowed a better (and quicker) understanding of the
optimum design values. It further revealed that the sfc of the problem by revealing key features such as trends and limiting

Figure 10 Optimum variation of sfc with Fs at pc ¼ 40

478
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

Table V Results of optimization with a progressively increasing minimum Fs constraint and an additional constraint of pf # 1.9
Constraints Optimum cycle Constraint status
Fs,min B pf T04 pc B pf T04 pc Fs sfc Fs,min B pf T04 pc
100 10 1.9 2,000 40 10 1.75 1,600 40 136.5 16.17 P A P P A
150 10 1.9 2,000 40 10 1.85 1,683 40 150 16.2 A A P P A
200 10 1.9 2,000 40 5.45 1.9 1,573 40 200 18.26 A P A P A
250 10 1.9 2,000 40 2.1 1.9 1,303 40 250 20.55 A P A P A
300 10 1.9 2,000 40 1 1.9 1,213 40 300 22.13 A P A P A
350 10 1.9 2,000 40 0.5 1.9 1,167 40 350 23.28 A P A P A
400 10 1.9 2,000 40 0.185 1.9 1,142 40 400 24.14 A P A P A
420 10 1.9 2,000 40 0.1 1.9 1,137 40 420 24.4 A P A P A
430 10 1.9 2,000 40 0.063 1.9 1,134 40 430 24.56 A P A P A
Notes: A, active constraint; P, passive constraint

Figure 11 Optimum variation of sfc with Fs with a constraint of pfmax ¼ 1.9

criteria. Second, it significantly narrowed down the region of Mattingly, J. (1996), Elements of Gas Turbine for Propulsion,
interest for parametric study. Third, it allowed design McGraw-Hill, Singapore, International edition.
constraints to be easily incorporated in the study. Mattingly, J. (1999), “Need info for BSc project”, November
23, 1999, Technical correspondence, E-mail: Jack@
References aircraftenginedesign.com
Powel, D.T. (1991), “Propulsion systems for twenty first
Arora, J.S. (1989), Introduction to Optimum Design, 1st ed.,
century commercial transports”, Proc. Instn. Mech. Engrs, J.
McGraw-Hill, New York, NY.
of Eng. for Gas Turbine and Power, Vol. 205, p. 13.
Cohen, H., Rogers, G.F.C. and Saravanamuttoo, H.I.H.
Press, W.H., Flannery, B.P., Teukolsky, S.A. and Vetterling,
(1987), Gas Turbine Theory, 3rd ed., Longman, London.
Kerrebrock, J. (1992), Aircraft Engines and Gas Turbines, 2nd W.T. (1988), Numerical Recipes in C, 1st ed., Cambridge
ed., MIT Press, Cambridge, MA. University Press, Cambridge.
Kurzke, J. (1999), “Gas turbine cycle design methodology: a Wilson, D.G. (1984), The Design of High Efficiency
comparison of parameter variation with numerical Turbomachinery and Gas Turbines, 1st ed., MIT Press,
optimization”, Transactions of the ASME, Vol. 121, p. 6. Cambridge, MA.

479
 Thermodynamic optimization of the turbofan cycle Aircraft Engineering and Aerospace Technology: An International Journal
Yousef S.H. Najjar and Sharaf F. Al-Sharif Volume 78 · Number 6 · 2006 · 467 –480

Further reading Published 123 papers in international refereed journals and

conferences; granted a patent by British Patent Office (1988);
Vanderplaats, G.N. (1984), Numerical Optimization Techniques
two patent publications; lectured in 24 international
for Engineering Design, McGraw-Hill, New York, NY.
conferences; member of the “Editorial Advisory Board” for
the International Journals of: Energy and Environment, and
About the authors Applied Thermal Engineering. Awards: The 1995 Award for
excellence for an outstanding paper in J. Aircraft Eng. and
Yousef S.H. Najjar, Founding Director of the Aerospace Technology; Fellowship of ASME-USA (1999);
Energy Center, Fellow ASME (USA), Fellow Fellowship of Institute of Energy-UK (1990); Professional
the Institute of Energy (UK), PE, C.Eng.
Engineer; Chartered Engineer. Specialization: Energy-
Professor of Mechanical Engineering. BSc,
Thermal Power including Gas Turbines: Fuels,
Mech. Eng. (Power), Cairo University (1969);
Combustion, Turbomachines and Advanced Energy
MSc and PhD, Mech. Eng. (Thermal Power),
Cranfield Institute of Technology (UK) 1976 Systems; Internal Combustion Engines and Autotronics.
and 1979, respectively. Industrial experience: Chief Power Initiated a course on “Autotronics”. Authored three books
Engineer-Irbid District Electricity Company, Jordan (1969- and manuals. Latest research: “Autotronics” and “Fuel cell –
1975); Specialized industrial training with General Electric gas turbine hybrid power”. Founded Pioneering Labs for:
(GEC) and related power industries (UK) (1973-1974). “automotive diagnosis” “energy audit” and “autotronics”.
Academic experience: Yarmouk University (1980-1986): The Yousef S.H. Najjar is the corresponding author and can be
founding chairman of the Mech. Eng. Dept., 1980-1982; contacted at: [email protected]
Member University Council (1985-1986). King Abdulaziz
University-Jeddah (1986-2001): participated effectively in Sharaf F. Al-Sharif is a Researcher at King Abdulaziz
two funded research projects and ABET accreditation. University, Jeddah, Saudi Arabia.

To purchase reprints of this article please e-mail: [email protected]

Or visit our web site for further details: www.emeraldinsight.com/reprints

480